TLR4 and AT1R mediate blood-brain barrier disruption, neuroinflammation, and autonomic dysfunction in spontaneously hypertensive rats

Abstract

Angiotensin II (AngII) is implicated in neuroinflammation, blood-brain barrier (BBB) disruption, and autonomic dysfunction in hypertension. We have previously shown that exogenous AngII stimulates Toll-like receptor 4 (TLR4) via AngII type 1 receptor (AT1R), inducing activation of hypothalamic microglia ex vivo, and that AngII-AT1R signaling is necessary for the loss of BBB integrity in spontaneously hypertensive rats (SHRs). Herein, we hypothesized that microglial TLR4 and AT1R signaling interactions represent a crucial mechanistic link between AngII-mediated neuroinflammation and BBB disruption, thereby contributing to sympathoexcitation in SHRs. Male SHRs were treated with TAK-242 (TLR4 inhibitor; 2 weeks), Losartan (AT1R inhibitor; 4 weeks), or vehicle, and age-matched to control Wistar Kyoto rats (WKYs). TLR4 and AT1R inhibitions normalized increased TLR4, interleukin-6, and tumor necrosis factor-α protein densities in SHR cardioregulatory nuclei (hypothalamic paraventricular nucleus [PVN], rostral ventrolateral medulla [RVLM], and nucleus tractus solitarius [NTS]), and abolished enhanced microglial activation. PVN, RVLM, and NTS BBB permeability analyses revealed complete restoration after TAK-242 treatment, whereas SHRs presented with elevated dye leakage. Mean arterial pressure was normalized in Losartan-treated SHRs, and attenuated with TLR4 inhibition. In conscious assessments, TLR4 blockade rescued SHR baroreflex sensitivity to vasoactive drugs, and reduced the SHR pressor response to ganglionic blockade to normal levels. These data suggest that TLR4 activation plays a substantial role in mediating a feed-forward pro-hypertensive cycle involving BBB disruption, neuroinflammation, and autonomic dysfunction, and that TLR4-specific therapeutic interventions may represent viable alternatives in the treatment of hypertension.

Graphical Abstract

1. INTRODUCTION

According to the 2019 Heart Disease and Stroke Statistics, approximately 46% of Americans have hypertension¹. Recent estimates suggest that up to 20% of patients remain resistant to currently available anti-hypertensive medications², highlighting the need to identify new therapeutic targets for the effective treatment of hypertension. A common observation among these patients is an elevation in sympathetic nervous system (SNS) activity^3–7. Within the central nervous system (CNS), multiple cardioregulatory nuclei govern sympathetic outflow, including the hypothalamic paraventricular nucleus (PVN), rostral ventrolateral medulla (RVLM), and nucleus of the tractus solitarius (NTS). In brief, whereas the PVN enhances sympathetic activity via projections to the intermediolateral spinal column and RVLM, the primary nucleus responsible for net sympathetic outflow^{8, 9}, the NTS suppresses RVLM activity by way of the caudal ventrolateral medulla¹⁰. Previous studies within these nuclei show that altered renin-angiotensin system (RAS) mechanisms, particularly dysregulation of angiotensin II (AngII) signaling, contribute to increased sympathetic activity in hypertension¹¹. As AngII’s type 1 receptor (AT1R) is expressed by multiple CNS cell types, AngII may influence neuronal activity directly, by binding neuronal AT1R, or indirectly, by influencing the activity of endothelial cells, perivascular macrophages¹², astrocytes¹³, and microglia¹⁴.

AngII dysregulation is strongly associated with low-grade neuroinflammation, microglial activation, and blood-brain barrier (BBB) disruption in CNS cardioregulatory nuclei^{11, 15, 16}. We have shown that AngII-induced BBB breakdown facilitates entry of circulating AngII into the PVN, RVLM, and NTS of spontaneously hypertensive rats (SHRs), targeting neurons and microglia therein. Prior work consistently reports increased microglial activation and elevated levels of pro-inflammatory cytokines (PICs; i.e., tumor necrosis factor [TNF]-α, interleukin [IL]-6, and IL-1β) in the PVN of hypertensive animals in response to AngII, with several studies demonstrating clear links amid microglia, PIC signaling, and elevated sympathetic activity in SHRs^{17, 18}, L-N^G-nitro-l-arginine methyl ester (L-NAME)- and AngII-infusion mice¹⁹, and AngII-infusion rats²⁰.

Microglia are direct regulators of the neuroinflammatory response. These innate immune cells constitutively express AT1R and Toll-like receptor 4 (TLR4)^{14, 21}, a pattern-recognition receptor implicated in multiple neuroinflammatory and cardiovascular diseases, including hypertension^{15, 22, 23}. Upon stimulation by pathogen-associated molecular patterns (i.e., lipopolysaccharide [LPS]) or damage-associated molecular patterns²⁴, TLR4 initiates pro-inflammatory signaling cascades, causing microglial activation and triggering downstream PIC production. Using an ex vivo preparation of PVN-containing hypothalamic slices, we have demonstrated an interaction between AT1R-dependent signaling cascade(s) and TLR4 following exogenously applied AngII, such that TLR4 becomes stimulated, leading to microglial activation and reactive oxygen species (ROS) generation¹⁴. While various in vivo and in vitro studies have shown that AngII upregulates TLR4 expression, stimulates microglia, enhances PIC production, and disrupts the BBB^{14, 15, 25, 26}, the extent to which in vivo AT1R-TLR4 signaling interactions contribute to the etiology and preservation of a hypertensive state remains unclear.

Based on recent evidence, we hypothesized that interactions between microglial AngII-AT1R and -TLR4 signaling cascades represent a critical mechanistic link between AngII-mediated neuroinflammation and BBB disruption within the PVN, RVLM, and NTS, ultimately contributing to the maintenance of an inflammatory and sympathoexcitatory state in cases of chronic hypertension. As such, this study aimed to evaluate TLR4 as an alternative target in the treatment of hypertension by examining the efficacy of systemic TLR4 blockade in mitigating indices of neuroinflammation and autonomic dysfunction relative to the widely used AT1R inhibitor, Losartan. In addition, we explored the potential for repurposing TAK-242 (resatorvid), a specific TLR4 inhibitor originally tested in clinical trial for the treatment of sepsis (NCT00633477) that binds Cys747 in the receptor’s intracellular domain²⁷ and is able to cross the intact BBB^{28, 29}. Using SHRs, we examined the impact of TAK-242 administration on BP changes, TLR4 protein expression, PIC production, microglial activation status, BBB integrity, baroreflex sensitivity, and indirect sympathetic activity.

2. MATERIALS AND METHODS

2.1. Animals and experimental groups

All procedures were approved by the Institutional Animal Care and Use Committee (protocol 2017–2883) and were performed in accordance with the Guide for the Care and Use of Laboratory Animals, as recommended by the US National Institutes of Health. Male Wistar Kyoto rats (WKY) and SHRs (Charles River Laboratories, USA) were housed in temperature- and humidity-controlled rooms (22±1°C; 50±5%) under a 12–12h light-dark cycle with standard rat chow and water ad libitum. Animals were 7–8 weeks old (175–225g) at the start of experiments. SHRs were randomly divided into control or experimental groups. For AT1R blockade, SHRs were treated daily with Losartan (AT1R antagonist; TCI America, USA; 20mg/kg BW¹⁶; oral gavage; SHR-Los) or vehicle for 4 weeks, and age-matched with WKYs (n=6/group). For TLR4 blockade, SHRs were treated daily with TAK-242 (TLR4 antagonist; Apex Bio, USA; MedChem Express, USA; 2mg/kg BW³⁰; i.p.; SHR-TAK; n=17) for 2 weeks, and age-matched with control SHRs (n=13) and WKYs (n=11). The dose of TAK-242 was chosen based on previous work supporting suppression of cardiac and renal inflammatory cytokines levels (TNF-a, IL-1β and MCP-1) and prevention of blood pressure increases in a model of Aldosterone-induced hypertension³¹.

2.2. Indirect blood pressure assessment

Indirect blood pressure (BP) measurements were performed using a volume-pressure recording tail-cuff system (CODA-6, Kent Scientific Corporation, USA)³². Animals were acclimated to the BP recording system for three days prior to the start of experiments. For data collection, 5 acclimation and 20 regular cycles were run, the latter of which were averaged to determine mean arterial pressure (MAP) values, as previously reported³³. BP measurements were taken on a weekly basis for animals in the Losartan cohort. For TAK-242-treated animals and age-matched controls, BP was assessed on alternating days.

2.3. Surgical procedures

Animals were anesthetized with isoflurane (induction: 5%, maintenance: 2%) and/or a cocktail of Ketamine (100 mg/kg BW)-Xylazine (15 mg/kg BW; i.p.). Anesthesia levels were evaluated regularly throughout procedures by checking for absent tail- and toe-pinch reflexes.

2.3.1. Baroreceptor reflex sensitivity and indirect sympathetic activity:

The abdominal aorta was cannulated through the left femoral artery to allow for continuous BP and heart rate (HR) recordings. The left femoral vein was cannulated for delivery of drugs during autonomic function assessments. Cannulae were tunneled subcutaneously and exteriorized through the mid-scapular region of the back, and lidocaine was applied to surgical sites. Animals were given Carprofen (2.5 mg/kg BW; s.q.) for pain. The following day, cardiac parameters were recorded from unrestrained conscious rats by a pressure transducer (model SP 844, Memscap AS, Norway) system with computer data acquisition (Bridge Amp/PowerLab 4/35, ADInstruments, Australia). After 30 min of baseline recording, i.v. doses of phenylephrine (Phe; 20 μg/kg BW; MilliporeSigma, USA) and sodium nitroprusside (SNP; 25 μg/kg BW; Spectrum Chemical, USA) were randomly administered, followed by the ganglionic blocker, hexamethonium bromide (20 mg/kg BW; Sigma, USA)^{34, 35}. Drug doses were separated by 10–15 min to allow cardiac parameters to return to baseline. Baroreceptor reflex sensitivity was determined by heart rate compensation for a given change in MAP (ΔHR [BPM]/ΔMAP [mmHg]), as evaluated at the maximal response. Indirect SNS activity was indexed as the magnitude of the depressor response to hexamethonium bromide, expressed as ΔMAP, relative to the 1 min period immediately prior to injection.

2.3.2. BBB permeability:

BBB permeability surgeries were performed as previously described¹⁶. In short, a fluorescent dye cocktail of rhodamine B isothiocyanate-dextran (RHO70; 70kDa, 10 mg/mL; Sigma-Aldrich, USA) and fluorescein isothiocyanate-dextran (FITC10; 10kDa, 10 mg/mL; Sigma-Aldrich, USA) dissolved in sterile saline was injected through the left carotid artery (3 μL/g BW) and allowed to circulate for 30 min. Brains were extracted, post-fixed in 4% formaldehyde (PFA; 48 h), cryoprotected in 30% sucrose (72 h), and stored at −80°C until sectioning.

2.4. Immunohistochemistry

Brains were removed following transcardial perfusion with 150 mL 0.01M phosphate-buffered saline (PBS) and 350 mL 4% PFA. Whole brains were post-fixed in 4% PFA for 24 h, cryoprotected in a 30% sucrose solution for 72 h, and stored at −80°C prior to sectioning. Serial hypothalamic slices of 30 μm (containing the PVN) and medullary sections of 40 μm (containing the RVLM and/or NTS) were collected (Microm cryostat HM 525). Slices were stored in cryoprotectant solution (200 mL glycerol [RNAse-Free; Sigma, USA], 300 mL ethylene glycol [Aldrich, USA], 450 mL dH2O, 75 mL 0.3 M PBS) at −20°C until processing.

Slices were washed three times in PBS to remove cryoprotectant solution and incubated for 1 h with 10% normal donkey serum (Jackson ImmunoResearch, USA). Sections were incubated in PBST (0.01M PBS, 0.1% Triton, 0.04% NaN3) with 5% normal donkey serum and primary antibodies against TLR4 (mouse monoclonal IgG2bκ, 1:250, Novus Biologicals, USA, NB100–56567, lot CJU03–11 and IMG-5031A), ionized calcium-binding adaptor molecule 1 (IBA1; microglial marker [rabbit polyclonal, 1:1000, Wako Chemicals, USA, 019–19742, lot WDK2121 and WDF6884; goat polyclonal, 1:500, Abcam, UK, ab5076, lot GR3195324–1; or rabbit polyclonal, 1:2000, EnCor Biotechnology Inc., USA, RPCA-IBA1, lot 040119]), TNF-α (mouse monoclonal IgG1, 1:100, Santa Cruz Biotechnology, USA, sc-52746, lot C0119), and/or IL-6 (mouse monoclonal IgG2a, 1:100, Santa Cruz Biotechnology, USA, sc-32296, lot I2818). Hypothalamic slices were also incubated with (Arg8)-vasopressin (VP; guinea pig polyclonal, 1:5000, Peninsula Laboratories, USA, T-5048.0050, lot A17901) and brainstem sections were incubated with tyrosine hydroxylase (TH; rabbit polyclonal [1:2000, EnCor Biotechnology Inc., USA, RPCA-TH, lot 040199] or mouse monoclonal IgG2a [1:1000, Santa Cruz Biotechnology, USA, sc-25270, lot G1917 and G1918]) for use as anatomical markers. In addition, antigen-retrieval (5 min incubation in PBS containing 1% w/v sodium dodecyl sulfate [SDS]) was performed prior to serum blocking in tissue sections used for microglial morphological analyses. Slices were washed in PBS and incubated with respective Alexa Fluor® AffiniPure Donkey IgG (H+L) secondary antibodies from Jackson ImmunoResearch: 488 anti-mouse (1:250), 594 anti-rabbit (1:250), 594 anti-guinea pig (1:250), or 647 anti-rabbit (1:50). Negative control sections were run without primary antibodies. Sections were washed and mounted with Vectashield Antifade Mounting Medium or VECTASHIELD® Antifade Mounting Medium with DAPI (Vector Laboratories, USA).

2.5. Immunofluorescence imaging and analysis

Full-thickness confocal z-stacks (1 μm intervals) of PVN-, RVLM-, and NTS-containing brain sections were acquired with a Nikon Eclipse TE2000-E inverted microscope coupled to a Nikon A1 confocal laser and analyzed using ImageJ software (NIH; https://imagej.nih.gov/ij/index.html).

2.5.1. Protein density quantification:

Confocal z-stacks were taken with a 20x objective in the PVN, RVLM, and NTS (1 image/unilateral nucleus/slice, 4–6 slices/animal). TLR4, IL-6, and TNF-α immunofluorescence signals (% area), expressed as % change from WKY, were quantified from maximum projection images, as previously described¹⁶, using ImageJ.

2.5.2. Microglia morphological analysis:

Microglial activation status was examined with a skeletal analysis method adapted from Morrison and Filosa³⁶ using maximum projection images of IBA1 fluorescence at 60x-magnification in the PVN, RVLM, and NTS (3 images/unilateral nucleus/slice, 4–6 slices/animal). Briefly, threshold adjustments and noise reductions were applied to increase visualization of cell processes, images were converted to binary, skeletonized, and analyzed with the AnalyzeSkeleton plugin (http://github.com/fiji/AnalyzeSkeleton). The number of end points and total branch length were used as morphological parameters reflective of relative branching complexity among groups.

2.5.3. BBB permeability assessment:

Dye-injected brains were cut to 40 μm-thick sections containing the PVN and 50 μm-thick sections containing the RVLM and NTS. Slices were counterstained with DAPI (VECTASHIELD® Antifade Mounting Medium with DAPI, Vector Laboratories, USA) and/or TOTO-3 Iodide (1:50,000, ThermoFisher Scientific, USA) and imaged with a 20x objective (2 images/slice [bilateral], 4–6 slices/animal). Extravasation of the low molecular weight (MW) dye (FITC10_EV) was used as an index of BBB integrity. Quantification of FITC10_EV was achieved by subtracting colocalized FITC10 and RHO70 pixels from a maximum projection of the FITC10 channel, and measuring the percent area of FITC10_EV from the newly generated image, as previously described¹⁶.

2.6. Statistical Analyses

Data are reported as mean±SEM unless otherwise indicated. Groups were compared by one-way or two-way ANOVA with Tukey post-hoc tests. Analyses were run using GraphPad Prism 7 (GraphPad Software, Inc., USA) with statistical significance considered at p<0.05.

3. RESULTS

3.1. AT1R and TLR4 inhibitors decrease MAP in SHRs.

Following 4 weeks of AT1R blockade, SHR-Los MAP (103.6±2.4 mmHg), as measured via indirect tail-cuff, was normalized to similar levels as observed in WKYs (100.0±3.0 mmHg, p=0.473), whereas SHR MAP (154.9±1.7 mmHg, p<0.0001) remained significantly elevated (n=6/group; Fig. 1A). Treatment with TAK-242 attenuated the MAP increases in SHR-TAK (n=17) that were observed in SHRs (n=12) beginning on day three (142.6±3.01 vs. 161±1.5 mmHg, p<0.0001), and persisted throughout the rest of the treatment period (day 14: 129.5±2.8 vs. 153.8±2.1 mmHg; p<0.0001). SHR-TAK MAP, while reduced, was significantly greater than MAP levels in WKYs at the end of treatment (109.2±2.9 mmHg, p<0.0001; Fig. 1B). The trends observed in the TAK-242 cohort were confirmed in the cohort subjected to direct MAP recordings in unrestrained conscious animals at the end of treatment (WKY: 109.2±6.1 mmHg [n=5]; SHR: 169.4±13.5 mmHg [p<0.0001 vs. WKY, n=6]; SHR-TAK: 150.8±9.0 mmHg [p=0.003 vs. WKY, p<0.0001 vs. SHR, n=10]; Fig. 1C).

Figure 1 – Progression of MAP elevation in SHRs is dependent upon AT1R and TLR4.

Final indirect tail-cuff mean arterial pressure (MAP; mmHg) measurements of Losartan cohort (n=6/group) (A) and TAK-242 cohort (n=11 WKY, 13 SHR, 17 SHR-TAK) (B) at conclusion of respective treatment periods; final direct MAP (mmHg) of TAK-242 sub-cohort (n=5 WKY, 6 SHR, 10 SHR-TAK; C). Data evaluated by one-way ANOVA with Tukey post-hoc analysis; data shown as mean±SEM; ****p<0.0001 vs. WKY; +++p<0.001 vs. SHR; ++++p<0.0001 vs. SHR.

3.2. TLR4 protein expression is upregulated through AT1R and TLR4.

To examine the role of AngII-AT1R signaling in regulating TLR4 protein expression, we performed immunofluorescence assays in the PVN, RVLM, and NTS (expressed as % change from WKY, Fig. 2). Consistent with previous reports^37–39, TLR4 protein density was increased in SHR PVN compared to WKY (+133.0±14.7%, p<0.0001), and normalized in SHR-Los (+27.8±10.6%) and SHR-TAK (+10.7±4.9%). TLR4 expression was elevated in SHR RVLM (+107.9±6.7%, p<0.0001) and normalized in SHR-Los (−1.05±4.6%) and SHR-TAK (+11.3±3.1%). Similarly, NTS TLR4 expression was higher in SHR (+101.6±6.9%, p<0.0001), with normalization in both treatment groups (SHR-Los: +11.0±7.4%; SHR-TAK: +1.5±4.7%), suggesting a feed-forward upregulation of central TLR4 in SHRs that relies upon activation of both AT1R- and TLR4-dependent signaling pathways.

Figure 2 – TLR4 protein expression within CNS cardioregulatory nuclei.

Example confocal maximum projection images of PVN vasopressin (VP; red) and TLR4 (white) from WKY, SHR, SHR-Los, and SHR-TAK (n=6/group) (A). Percent change in TLR4 staining (% area) compared to WKY in the PVN (B), RVLM (C), and NTS (D). Data evaluated by one-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; ****p<0.0001 vs. WKY; ++++p<0.0001 vs. SHR; scale bars: 100μm; 3V: third ventricle.

3.3. PICs are downregulated following AT1R or TLR4 blockade.

Using a semi-quantitative densitometry analysis, we examined IL-6 and TNF-α immunofluorescence in the PVN, RVLM, and NTS (expressed as % change from WKY). Protein expression of both cytokines was increased in SHR PVN (IL-6: +46.60±4.8%, p<0.0001; TNF-α: +57.97±8.9%, p<0.0001), RVLM (IL-6: +77.44±11.6%, p<0.0001; TNF-α: +58.01±7.9%, p<0.0001), and NTS (IL-6: +49.3±4.1%, p<0.0001; TNF-α: +56.5±4.3%, p<0.0001) versus WKYs. PVN cytokine expression was normalized with Losartan (IL-6: +0.85±5.0%; TNF-α: +5.99±3.9%) and TAK-242 treatment (IL-6: +1.30±2.9%; TNF-α: +5.70±2.6%; Fig. 3). NTS IL-6 and TNF-α were normalized in SHR-Los (IL-6: +8.8±6.1; TNF-α: +3.5±4.8%) and SHR-TAK (IL-6: +5.6±4.0%; TNF-α: +6.7±4.3%) (Fig. 4). Both treatments restored RVLM TNF-α expression to baseline (SHR-Los: +8.16±4.7%; SHR-TAK: +4.48±5.2%), and reduced IL-6 (SHR-Los: +32.37±3.9% [p=0.0011 vs. WKY, p=0.0013 vs. SHR]; SHR-TAK: +33.90±7.6% [p=0.0023 vs. WKY, p=0.0017 vs. SHR]; Fig. 5), suggesting that both AT1R and TLR4 signaling mechanisms are involved in regulating the pro-inflammatory profile of CNS cardioregulatory nuclei in SHRs.

Figure 3 – PVN pro-inflammatory cytokine density.

Example confocal maximum projection images of PVN TNF-α (A) and IL-6 (B) immunofluorescence (white) with respective anatomic marker, vasopressin (insets; red) in WKY, SHR, SHR-Los, and SHR-TAK (n=6/group). Box-plot diagrams of TNF-α (C) and IL-6 (D) protein density in the PVN, calculated as change in % area relative to WKY. Data evaluated by one-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; ****p<0.0001 vs. WKY; +++p<0.001 vs. SHR; ++++p<0.0001 vs. SHR; scale bars: 200μm; 3V: third ventricle.

Figure 4 – NTS pro-inflammatory cytokine density.

Box-plot diagrams of TNF-α (A) and IL-6 (B) protein density in the NTS of WKY, SHR, SHR-Los, and SHR-TAK (n=6/group), calculated as change in % area relative to WKY. Data evaluated by one-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; **p<0.01 vs. WKY; ****p<0.0001 v.s WKY; ++++p<0.0001 vs. SHR.

Figure 5 – RVLM pro-inflammatory cytokine density.

Example confocal maximum projection images of RVLM TNF-α (A) and IL-6 (B) immunofluorescence (white) with respective anatomic marker, tyrosine hydroxylase (insets; red) in WKY, SHR, SHR-Los, and SHR-TAK (n=6/group). Box-plot diagrams of TNF-α (C) and IL-6 (D) protein density in the RVLM, calculated as change in % area relative to WKY. Data evaluated by one-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; **p<0.01 vs WKY; ****p<0.0001 vs WKY; ++++p<0.0001 vs SHR; scale bars: 200μm.

3.4. AT1R and TLR4 are necessary for microglial activation.

We quantified morphological changes of microglia using a skeletal analysis³⁶ of IBA1 immunostaining to index total branch number (end points) and branch length. Under normal physiological conditions, surveillance microglia present with a small soma and numerous fine, highly-ramified, motile processes. Conversely, the classically activated pro-inflammatory M1 microglial phenotype is associated with an enlarged cell soma and overall deramification, such that a relative reduction in branching complexity is reflective of increased activation towards an amoeboid phagocytotic state. Within the PVN, SHRs showed a significant reduction in end points (−36.1±3.6%, p<0.0001) and branch length (−26.8±4.7%, p<0.0001) compared to WKYs, indicating a significant increase in microglial activation in hypertensive animals relative to that observed in normotensive controls. These values were normalized in the PVN of SHR-Los (end points: 2.9±2.2%; branch length: 11.3±2.5%) and SHR-TAK (end points: 4.7±5.6%; branch length: 4.0±5.1%). The reduction in end points and branch length in SHR RVLM tissue (end points: −29.3±3.3%, p<0000.1; branch length: −26.9±2.9, p=0.0004) compared to WKY was restored in SHR-Los (end points: −4.3±1.5%; branch length: −2.9±4.8%) and SHR−TAK (end points: 3.8±4.8%; branch length: 8.7±3.7%) (Fig. 6), with a similar trend observed in the NTS (Fig. 7) of SHRs (end points: −53.1±2.0%, p<0.0001; branch length: −57.2±2.0%, p<0.0001), SHR-Los (end points: +1.7±1.7%; branch length: −7.1±1.8%), and SHR-TAK (end points: +6.8±3.2%; branch length: −3.2±3.0%). These findings confirm a regulatory role for AngII-AT1R signaling in promoting microglial activation, and support a substantial contribution of TLR4 stimulation in mediating said AngII-induced activation.

Figure 6 – Skeletal analysis of microglial morphology in the PVN and RVLM.

Example confocal maximum projection images of IBA1 (microglial marker; white) in the PVN (A) and RVLM (B) of WKY, SHR, SHR-Los, and SHR-TAK (n=6/group). Maximum projection images were converted to binary and skeletonized, as illustrated by inset skeletons of outlined microglia. Percent change in PVN end points (C), PVN branch length (D), RVLM end points (E) and RVLM branch length (F) relative to WKY. Data evaluated by two-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; ****p<0.0001 vs. WKY; ++++p<0.0001 vs. SHR; scale bars: 50μm.

Figure 7 – NTS microglial morphology.

Percent change in NTS end points (A) and branch length (B) relative to WKY in the NTS of WKY, SHR, SHR-Los, and SHR-TAK (n=6/group). Data evaluated by two-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; ****p<0.0001 vs. WKY; ++++p<0.0001 vs. SHR.

3.5. Blockade of TLR4 prevents BBB disruption.

To examine the role of TLR4 in BBB permeability alterations, we quantified the degree of extravasation of a low MW dextran-conjugated fluorescent dye (FITC10_EV) in the PVN, RVLM, and NTS of SHR-TAK (Fig. 8). As previously described¹⁶, both FITC10 and the simultaneously injected high molecular weight dye, RHO70, are maintained within cerebral vasculature when the BBB is intact. Conversely, BBB disruption results in FITC10 leakage from the vasculature to the parenchyma, reflected by an increased FITC10_EV % area. We found significant BBB disruption in the PVN (3.619±0.108% area, p<0.0001) and RVLM (3.62±0.11% area, p<0.0001) of control SHRs relative to WKYs (PVN: 1.56±0.05% area; RVLM: 1.57±0.06% area). Inhibition of TLR4 restored the barrier’s integrity in the PVN (1.52±0.05% area) and RVLM (1.51±0.05% area). In SHR NTS, FITC_EV was significantly greater than that observed in WKYs (3.588±0.09% area in SHR vs. 1.800±0.04% area in WKYs, p<0.0001). As in the PVN and RVLM, we found NTS FITC10_EV to be normalized in SHR-TAK (1.801±0.0513% area). Given our prior work demonstrating a reliance of PVN, RVLM, and NTS BBB disruption upon AngII-AT1R signaling in SHRs, these data indicate that activation of TLR4 is a potential mechanism by which AngII promotes BBB disruption.

Figure 8 – BBB permeability assessment in the PVN, RVLM, and NTS.

Example confocal maximum projection images showing FITC10 (green) and RHO70 (red) in the PVN, RVLM, and NTS (A) of WKY, SHR, and SHR-TAK (n=6/group), with corresponding images of extravasated FITC10 (FITC10_EV). Box-plot diagrams of FITC10_EV (% area) in the PVN (B), RVLM (C), and NTS (D). Data evaluated by one-way ANOVA with Tukey post-hoc analysis; shown as mean±SEM; ****p<0.0001 vs. WKY; ++++p<0.0001 vs. SHR; scale bars: 100μm; 3V: third ventricle; AP: area postrema; CC: central canal.

3.6. Blockade of TLR4 rescues autonomic function.

We evaluated baroreceptor-heart rate reflex (baroreflex sensitivity) in response to TAK-242 treatment using bolus i.v. injections of vasoactive pressor (Phe) and depressor (SNP) drugs in conscious animals during continuous direct BP and HR recording^{34, 35} (Fig. 9). SHRs showed a reduced HR compensation for MAP changes compared to WKYs following Phe (−1.501±0.1 vs −2.609±0.08 ΔHR [BPM]/ΔMAP [mmHg]; p<0.0001) and SNP (−0.9463±0.2 vs −2.379±0.3 ΔHR/ΔMAP; p=0.0013). TAK-242 treatment rescued SHR responsiveness to both Phe (−2.250±0.1 ΔHR/ΔMAP; p=0.0002 vs SHR) and SNP (−2.181±0.2 ΔHR/ΔMAP; p=0.0017 vs SHR). Following baroreflex assessments, indirect sympathetic activity was evaluated as the magnitude of the depressor response to i.v. hexamethonium bromide³⁴ (Fig. 9). Compared to WKYs, SHRs exhibited a greater depressor response (−65.1±3.1 vs −46.7±4.2 ΔMAP, p=0.0396). Conversely, sympathetic activity was restored to normal values with TLR4 inhibition in SHR-TAK animals (−42.9±4.3 ΔMAP; p=0.0034 vs SHR). Together, these findings indicate that TLR4 activation is necessary for the development of baroreflex impairment and sympathoexcitation in SHRs.

Figure 9 – Baroreceptor reflex sensitivity assessment and indirect SNS activity.

Example tracings (A) of pulsatile arterial pressure (PAP; mmHg), mean arterial pressure (MAP; mmHg), and heart rate (HR; BPM), showing responses to injection of phenylephrine (gray arrows) and sodium nitroprusside (black arrows). Box-plot diagrams of pressor response to phenylephrine (B) and depressor response to sodium nitroprusside (C), calculated as ΔHR [BPM]/ΔMAP [mmHg]. Example tracings of MAP following ganglionic blockade with hexamethonium bromide injection (dashed line) (C), quantified as maximum ΔMAP from baseline (D) in WKY (n=5), SHR (n=6), and SHR-TAK (n=10). *p<0.05 vs. WKY; **p<0.01 vs. WKY; ****p<0.0001 vs. WKY; ++p<0.01 vs. SHR; ++++p<0.0001 vs. SHR.

4. DISCUSSION

In the present study, we investigated the relative efficacy of AT1R and TLR4 inhibitors in mitigating neuroinflammation, BBB disruption, and sympathoexcitation in hypertensive SHRs. Our results demonstrate that TLR4 inhibition combats TLR4 and PIC upregulation, abolishes microglial activation, and preserves BBB integrity in the PVN, RVLM, and NTS of SHRs. Likewise, AT1R blockade normalizes TLR4 expression and microglial activation in the PVN, RVLM, and NTS of SHRs, and reduces PIC expression in these nuclei to the same extent as TLR4 inhibition. Furthermore, TLR4 blockade attenuates the progression of MAP increases in SHRs and protects against autonomic dysfunction. These findings support a significant role for TLR4 activation in the maintenance of central pro-hypertensive pathophysiology and suggest that TLR4 represents a viable alternative target in the treatment of resistant hypertension.

A number of studies implicate TLR4 activity throughout the body in the pathophysiology of hypertension (see^{23, 40}). With regards to the CNS, the majority of studies have focused on alterations either within the PVN or following targeted PVN interventions. For instance, upregulation of PVN TLR4 mRNA and protein expression has been reported in multiple models of hypertension^{37–39, 41–43}. Specific blockade of PVN TLR4 with TAK-242 reduces local PICs, mitigates sympathetic activity, and lowers BP in salt-induced^{37, 41} and Goldblatt two kidney, one clip models of hypertension⁴³. PVN-specific AT1R inhibition similarly decreases TLR4-dependent TNF-α, IL-1β, and IL-6 levels in SHRs³⁸. Ogawa et al. have demonstrated that brainstem TLR4 stimulation via AT1R contributes to elevated sympathetic activity in chronic heart failure⁴⁴, and that silencing brain TLR4 RNA dampens sympathetic activity and ameliorates the cardiac remodeling observed in this disease pathology⁴⁵. However, to the best of our knowledge, the relative contribution of TLR4 activation to alterations within cardioregulatory nuclei of the brainstem has yet to be investigated in a genetic model of hypertension. Importantly, minimally invasive long-term systemic TLR4 blockade has yet to be investigated as a potential intervention strategy for patients presenting with resistant hypertension.

Consistent with the aforementioned studies, we found increases in TLR4, TNF-α, and IL-6 protein expressions in the PVN of SHRs that were dependent upon AT1R and TLR4 activation. Within the RVLM and NTS, we observed that TLR4 was likewise increased in an AT1R- and TLR4-dependent manner in SHRs, pointing to a feed-forward mechanism of TLR4 upregulation in each of the cardioregulatory nuclei examined. Despite the consistency of the trends observed across both excitatory and inhibitory cardioregulatory nuclei in treated versus untreated animals, it is likely that the specific mechanisms involved in creating and maintaining the pro-inflammatory milieu varies between nuclei. Whereas heightened TNF-α levels in the RVLM were normalized following AT1R or TLR4 blockade, IL-6 expression, while reduced relative to untreated SHRs, remained elevated compared to normotensive animals. Interestingly, while SHR TNF-α increased to a similar extent within all three nuclei, the magnitude of increased IL-6 expression was approximately 30% greater in the RVLM than in the PVN or NTS. The observation that AT1R and TLR4 blockades reduce RVLM IL-6 levels to +32.37% and +33.90% relative to WKYs, respectively, points to separate AT1R- and TLR4-independent mechanism(s) of IL-6 upregulation in the RVLM of SHRs that is absent in the PVN and NTS. Of note, we recently reported on PICs within CNS cardioregulatory nuclei of SHRs treated with a probiotic (kefir) and observed a similar trend – PVN IL-6 expression was normalized in SHRs following treatment, whereas RVLM IL-6 expression remained elevated relative to WKYs by a magnitude corresponding to the difference between RVLM and PVN IL-6 levels in the untreated SHRs³³. Based on the evidenced RAS-⁴⁶ and TLR4-⁴⁷ modulatory capabilities of kefir, it is possible that a similar mechanism is responsible for the persistence of elevated RVLM IL-6 expression in kefir-, Losartan-, and TAK-242-treated SHRs.

Within the CNS, TLR4 is primarily expressed by microglia, with relatively low levels detected in astrocytes and neurons^{21, 48}. The presence of AT1R in microglia allows for AngII-mediated microglial activation¹⁴, leading to increased ROS production and PIC synthesis. Moreover, the neuroinflammatory effects of AngII via microglial AT1R are evidenced factors underlying sympathoexcitation in neurogenic hypertension^{15, 20, 49}. Microglial inhibition in the PVN blunts the hypertensive response to AngII²⁰. Upon targeted microglial deletion in AngII- and L-NAME-induced hypertension, Shen et al. observed a significant drop in blood pressure, neuroinflammation, renal norepinephrine, and circulating arginine vasopressin¹⁹. We have previously demonstrated that the ability of AngII to increase PVN microglial density and ROS production is dependent upon the presence of functional TLR4¹⁴. Herein, we found microglial activation in SHRs to be normalized in the PVN, RVLM, and NTS following either AT1R or TLR4 inhibition. The reductions in end points and branch lengths observed in SHRs suggest that the majority of microglia in these nuclei are chronically activated, likely towards an M1/pro-inflammatory phenotype. The M1 state is well evidenced to be induced by pro-inflammatory factors and pathogens, with TLR4 stimulation considered one of the primary activation pathways. In addition to the cytotoxic and pro-inflammatory activity of these cells, M1 microglia are known to be associated with altered BBB permeability⁵⁰. Thus, despite the inability to differentiate between classically activated M1 microglia and alternatively activated M2 microglia based on morphology alone, it is reasonable to suspect that the deramification seen in SHR nuclei results from M1/pro-inflammatory microglial polarization. A recent study by Cohen et al. (2019) employed a similar skeletal analysis technique to examine microglia in the RVLM of 15-week-old SHRs, reporting decreased microglial density and branch length, with no differences in branch number or end points compared to WKYs⁵¹. One potential explanation for our observation of end point reductions is the difference in age, our studies being carried out at approximately 12 weeks of age. However, we would suggest that these differences are likely due to variations in staining and imaging techniques that allowed for enhanced visualization of fine microglial processes.

The potential contribution of astrocytic and/or neuronal TLR4 to the pro-hypertensive CNS milieu in SHRs cannot be disregarded. However, the evidenced effects of TLR4 activation in non-microglial cells of the CNS may indicate that microglial TLR4 stimulation is a primary mechanism. Interestingly, Liddelow et al. (2017) demonstrated that in vivo A1 activation of astrocytes (i.e., neurotoxic/pro-inflammatory activation) is absent in mice lacking microglia (Csf1r^−/− knock-out)⁵². Indeed, mixed reports appear regarding the expression of astrocytic TLR4 and its downstream signaling components in rodents. Whether these observations are due to differences in species and/or strain, astrocyte activation state, or regional differences in astrocyte gene profiles is unknown. However, the ability of microglia to induce A1 astrocyte activation through TNF-α, IL-1α, and complement component 1q secretion⁵² following LPS stimulation supports the notion of astrocyte reactivity as a secondary event in neuroinflammation.

The recognition of BBB disruption as a pathological phenomenon is increasing across an array of neurological disorders associated with neuroinflammation, including Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis. BBB breakdown is apparent in hypertensive models^{53, 54}, and our prior work with Losartan-treated SHRs demonstrated such disruption to be dependent upon AngII via AT1R¹⁶. A subsequent study by Buttler et al. (2017) found disruption within autonomic centers in SHRs to increase between 1 and 3 months of age – dye leakage was absent in pre-hypertensive 4-week-old animals and, in agreement with our work, apparent at 12 weeks⁵⁵. Herein we report marked disruption at 9–10 weeks of age in the PVN, RVLM, and NTS of SHRs, during the establishing phase of hypertension. That SHR-TAK exhibited full barrier integrity raises the question of protection versus restoration. Moreover, whether the normalization of BBB permeability following TLR4 inhibition was due to protection against initial disruptions or a result of BBB repair processes is unknown. Given the indications of a time-course association between dye leakage and BP elevation, we suggest the latter explanation, wherein the already developing hypertension points to a loss of BBB integrity by 7–8 weeks in SHRs. While either scenario illustrates a clear role for TLR4 activation in BBB disruption, the distinction between them is particularly important from a clinical standpoint due to the evidenced disruption in cases of chronic hypertension, as well as in other neuroinflammatory diseases. To that end, further investigation regarding the timeline of BBB disruption in hypertension and the potential for BBB restoration via TLR4 inhibition is undoubtedly warranted.

Although hypertension is not fully established in SHRs at 7–8 weeks, a number of studies show that key characteristics of resistant hypertension (i.e., increased sympathetic neurotransmission) manifest prior to elevations in blood pressure, thus contributing to disease pathogenesis, as opposed to developing subsequent to the establishment of hypertension. For example, in pre-hypertensive SHRs (4 weeks of age), while arterial blood pressure is not different from age- and weight-matched WKY controls, plasma levels of the sympathetic co-transmitter NPY is increased and dysregulation in cardiac adrenergic signaling is already apparent⁵⁶. Within the CNS, SHRs as young as 9–16 days post-natal present with increased sympathetic nervous activity, as reflected by amplified respiratory-sympathetic coupling⁵⁷. In humans, increased sympathetic activity has been shown in borderline hypertensive patients, and enhanced pressor responses in normotensive individuals have been shown to predict the development of hypertension later in life⁵⁸. Thus, it is likely that, rather than preventing the development of TLR4- and AT1R-dependent hypertensive pathologies outright, the pharmacological interventions employed were sufficient to normalize/attenuate the majority of parameters investigated. Taken in combination with the aforementioned studies, this would indicate that the characteristics of resistant hypertensive pathophysiology are already apparent by 7–8 weeks in SHRs. However, it is possible that the alterations in sympathetic regulation observed prior to blood pressure elevations similarly precede the development of neuroinflammation and blood-brain barrier disruption. In this regard, the question of prophylactic vs. therapeutic warrants further investigation, as does the question of relative treatment efficacy at various time points over the course of disease progression.

Consistent with previous findings, we confirmed that 2 weeks of systemic TAK-242 reduced MAP in SHRs. Bomfim et al. observed reductions in MAP (approx. 20mmHg) in 15-week-old SHRs treated with an anti-TLR4 antibody (1μg/day, i.p.) for 15 days⁵⁹. Dange et al. reported a similar reduction following targeted bilateral PVN administration of VIPER (viral inhibitory peptide of TLR4; 40 μg/kg/day) for 14 days in 10–12-week-old SHRs³⁷. Despite the continued elevation of MAP in SHR-TAK versus WKY, the physiological relevance of the attenuated pressure should be considered. Moreover, while still presenting with elevated BP, the overwhelming majority of the other parameters examined in this study were normalized, which may suggest some threshold BP whose associated pathological changes are responsible for inciting neuroinflammation, BBB disruption, and autonomic dysfunction. Conversely, taking into account our prior work showing that a reduction of BP alone is insufficient to remedy these pathological alterations, it may be that their normalization occurs prior to BP reductions. As discussed above, a time-course evaluation of the pathological changes examined herein relative to the development of hypertension is necessary to address this point. Additionally, the question of long-term treatment with TAK-242 has yet to be investigated, and it would be of great interest to determine whether further MAP reductions occur beyond 2 weeks.

RAS-driven aberrations in baroreflex and autonomic function are well documented in hypertension. AngII increases neuronal activity within CNS cardiovascular nuclei, a process demonstrated to contribute to the maintenance of neurogenic hypertension^{15, 20, 49, 60}. Prior studies show that intracerebroventricular (i.c.v.) injection of LPS induces sympathetic hyperactivity through upregulation of PIC⁶¹, whereas PVN VIPER microinjection drastically reduces plasma norepinephrine in SHRs³⁷. In agreement with these studies, our results indicate that autonomic dysfunction was abolished in SHR-TAK, as indexed by baroreflex sensitivity and indirect SNS activity.

Whereas systemic TLR4 blockade precludes us from identifying a specific effector location responsible for the findings herein, the administration route and pharmacological agent employed in this study are clinically significant. TAK-242 has already received FDA approval, and the efficacy of less invasive treatment (i.e., i.p. versus i.c.v.) to facilitate improvements in the investigated parameters provide a clear basis for further studies regarding repurposing of TAK-242. Additionally, it should be considered that the ability of TAK-242 to cross the BBB, whether intact or disrupted, would result in non-specific inhibition if directly administered i.c.v.. Moreover, regardless of the primary site of action, the physiological impact of TAK-242 administration on autonomic function and within cardioregulatory nuclei is evident. It is imperative to note the studies reported herein were conducted using male SHRs. Previous work targeting pro-inflammatory and/or pro-oxidative pathways have revealed sexually dimorphic responses in both animal models and clinical trials⁶². As such, the efficacy of TAK-242 in combating those central hypertensive pathophysiological processes investigated in this work are, as yet, limited to males. Additional studies are necessary to determine the extent of TLR4’s contribution to the female hypertensive state.

4.1. Conclusions

In summary, the present work demonstrates a clear contribution of chronic TLR4 and AT1R activation in SHRs to 1) neuroinflammation via TLR4 upregulation, microglial activation, and pro-inflammatory cytokine production within the PVN, RVLM, and NTS; 2) BBB disruption in the PVN, RVLM, and NTS; 3) baroreflex desensitization; 4) sympathoexcitation; and 5) the progression of hypertension development. The alleviation of neuroinflammatory and sympathoexcitatory indices as investigated herein subsequent to long-term treatment with the TLR4 antagonist, TAK-242, support the further evaluation of this therapeutic as an alternative option in the treatment of resistant hypertension. Furthermore, the reliance of TLR4 protein upregulation upon AT1R activation, in combination with the consistent effects of AT1R and TLR4 inhibition on those parameters investigated, provides support for TLR4 activation as a mechanism of AngII-AT1R-dependent neuroinflammation and BBB disruption in multiple cardioregulatory nuclei, as well as autonomic dysfunction in hypertensive pathophysiology.

Highlights.

• TLR4 participates in a central feed-forward neuroinflammatory pro-hypertensive cycle

• TLR4 activation is required for BBB disruption in cardioregulatory nuclei

• TLR4 contributes to the development of autonomic dysfunction in hypertension

• TAK-242 is a promising therapeutic for the treatment of hypertension in males

ABBREVIATIONS

AngII

Angiotensin II

AT1R

Angiotensin II type 1 receptor

BBB

blood-brain barrier

BP

blood pressure

BPM

beats per minute

CNS

central nervous system

FITC10

fluorescein isothiocyanate-dextran

HR

heart rate

i.c.v.

intracerebroventricular

IBA1

ionized calcium-binding adaptor molecule 1

IL

interleukin

L-NAME

L-N^G-nitro-l-arginine methyl ester

LPS

lipopolysaccharide

MAP

mean arterial pressure

MW

molecular weight

NTS

nucleus tractus solitarius

PBS

phosphate buffered saline

PFA

paraformaldehyde

Phe

phenylephrine

PIC

pro-inflammatory cytokine

PVN

hypothalamic paraventricular nucleus

RAS

renin-angiotenisin system

RHO70

rhodamine B isothiocyanate-dextran

ROS

reactive oxygen species

RVLM

rostral ventrolateral medulla

SHR

spontaneously hypertensive rat

SNP

sodium nitroprusside

SNS

sympathetic nervous system

TH

tyrosine hydroxylase

TLR4

Toll-like receptor 4

TNF

tumor necrosis factor

VIPER

viral inhibitory peptide of TLR4

VP

vasopressin

WKY

Wistar Kyoto rat