Medicinal plant rosemary relaxes blood vessels by activating vascular smooth muscle KCNQ channels

Abstract

The evergreen plant rosemary (Salvia rosmarinus) has been employed medicinally for centuries as a memory aid, analgesic, spasmolytic, vasorelaxant and antihypertensive, with recent preclinical and clinical evidence rationalizing some applications. Voltage-gated potassium (Kv) channels in the KCNQ (Kv7) subfamily are highly influential in the nervous system, muscle and epithelia. KCNQ4 and KCNQ5 regulate vascular smooth muscle excitability and contractility and are implicated as antihypertensive drug targets. Here, we found that rosemary extract potentiates homomeric and heteromeric KCNQ4 and KCNQ5 activity, resulting in membrane hyperpolarization. Two rosemary diterpenes, carnosol and carnosic acid, underlie the effects and, like rosemary, are efficacious KCNQ-dependent vasorelaxants, quantified by myography in rat mesenteric arteries. Sex- and estrous cycle stage-dependence of the vasorelaxation matches sex- and estrous cycle stage-dependent KCNQ expression. The results uncover a molecular mechanism underlying rosemary vasorelaxant effects and identify new chemical spaces for KCNQ-dependent vasorelaxants.

Graphical Abstarct

Extract of Rosemary Aerial Parts (RAP) is a highly efficacious opener of vascular KCNQ channel isoforms, including KCNQ4 and KCNQ5 (pictured). Correspondingly, RAP extract, and constituents carnosol and carnosic acid (pictured), are effective and sex-specific vasorelaxants, via their KCNQ4 and KCNQ5 opening action, which hyperpolarizes vascular smooth muscle cells.

Introduction

Voltage-gated potassium (Kv) channels in the KCNQ (Kv7) subfamily serve a diverse number of essential roles in physiology, ranging from action potential repolarization in the heart to K⁺ recycling in the gastric epithelium and inner ear (KCNQ1), neuronal action potential gatekeeping (KCNQ2/3) and regulation of vascular tone (KCNQ4/5)^1,2. In addition to heteromultimerization within the same subfamily (e.g., KCNQ2/3, KCNQ4/5), the functional repertoire of KCNQ channels (as subsequently also discovered for other Kv channels) is expanded by co-assembly with single transmembrane segment KCNE regulatory subunits³.

In vascular smooth muscle, KCNQ4 and KCNQ5 can heteromultimerize^4,5, together with KCNE4^6,7, to form functional channels that control the resting membrane potential and facilitate vasodilatations elicited by certain receptors, e.g., β adrenoceptors and calcitonin gene-related peptide receptors^5,7–12. Nearly all the work involving vascular KCNQ channels has been performed in male rodents; however, vascular KCNQ channel expression and function were recently revealed to be estrogen-dependent, manifesting as the potency of certain KCNQ-specific activators shifting by >10-fold when comparing female rats from different estrus phases (high versus low estrogen levels)^13,14.

Although KCNQ4 is downregulated and KCNQ channel function attenuated in male rodent models of hypertension^15–18, the vascular KCNQ channels remain interesting as a potential therapeutic target in the treatment of hypertension. Recently, we showed that certain plants used traditionally as botanical folk medicines can activate KCNQ channels, particularly KCNQ5, which resulted in a vasodilatation in arteries from male rodents^19–21. We identified the KCNQ-active small molecules in these plants, which had previously been considered to have anti-hypertensive properties, thereby revealing the antihypertensive mechanism of commonly used botanical medicines and identifying novel KCNQ channel activators.

Another plant considered to act as a vasorelaxant with antihypertensive properties is the evergreen plant rosemary (Salvia rosmarinus), with recent preclinical and clinical evidence rationalizing some applications^22,23. Here, we examined the sensitivity of KCNQ4 and KCNQ5 homomers and heteromers to whole extract and constituent compounds of rosemary (Salvia rosmarinus). We also investigated whether rosemary and its constituents regulate vascular tone in a KCNQ-dependent manner in arteries from both male and female rodents. Our findings support a role for KCNQ channels in underlying the historical use of rosemary to regulate vascular tone.

Materials and Methods

Collection and preparation of plant extracts

Salvia rosmarinus leaves and flowers were collected from the last author’s garden by his children, and frozen until the day of extraction. Leaves and flowers were pulverized in 50 ml tubes using a bead mill with porcelain beads in batches (Omni International, Kennesaw, GA, United States), then the homogenates were resuspended in 80% methanol/20% water (100 ml per 5 g solid) and incubated for 48 hours at room temperature, with occasional inversion to resuspend the particulate matter. We then filtered the extracts through Whatman filter paper #1 (Whatman, Maidstone, UK), removed the methanol using evaporation in a fume hood for 24–48 hours at room temperature, centrifuged extracts for 10 minutes at 15 °C, 4000 RCF to remove the remaining particulate matter, followed by storage (−20 °C). On the day of electrophysiological recording, we thawed the extracts and diluted them 1:50 in bath solution (see below) immediately before use.

Channel subunit cRNA preparation and Xenopus laevis oocyte injection

We generated cRNA transcripts encoding human KCNQ4, KCNQ5 and KCNE4 by in vitro transcription using the mMessage mMachine kit (Thermo Fisher Scientific, Waltham, MA, USA) according to manufacturer’s instructions, after vector linearization, from cDNA sub-cloned into expression vectors (pTLNx and pXOOM) incorporating Xenopus laevis β-globin 5’ and 3’ UTRs flanking the coding region to enhance translation and cRNA stability. We injected defolliculated stage V and VI Xenopus laevis oocytes (Xenoocyte, Dexter, MI, USA) with KCNQ cRNAs (2–10 ng) and incubated the oocytes at 16 °C in ND96 oocyte storage solution containing penicillin and streptomycin, with daily washing, for 2–4 days prior to two-electrode voltage-clamp (TEVC) recording.

Chemicals

We purchased carnosic acid from Combi-Blocks (San Diego, CA) or Sigma-Aldrich (St. Louis, MO). Carnosol was from Sigma-Aldrich or prepared from carnosic acid using the method of Han et al.²⁴. Anhydrous acetone was supplied by Sigma-Aldrich. Water was 18.2 mΩ-cm purity from a Barnstead NANOpure Diamond^™ system. Trifluoroacetic acid was obtained from EMD Millipore. Melting points (mp) were determined on a Electrothermal MEL-TEMP 3.0 apparatus (Barnstead International, Dubuque, IA) and are not corrected. We purchased other rosemary compounds and chemicals from Sigma Aldrich.

Two-electrode voltage clamp (TEVC)

We conducted TEVC at room temperature with an OC-725C amplifier (Warner Instruments, Hamden, CT, USA) and pClamp10 software (Molecular Devices, Sunnyvale, CA, USA) 2–4 days after cRNA injection. Oocytes, perfused in a small-volume oocyte bath (Warner), were viewed with a dissection microscope for cellular electrophysiology. We studied effects of Salvia rosmarinus extracts and of compounds, solubilized directly in bath solution (in mM): 96 NaCl, 4 KCl, 1 MgCl₂, 1 CaCl_2, 10 HEPES (pH 7.6). We introduced extracts or compounds into the oocyte recording bath by gravity perfusion at a constant flow of 1 ml per minute for 3 minutes prior to recording. Pipettes (1–2 MΩ resistance) were filled with 3 M KCl. We recorded currents in response to voltage pulses between −120 mV or −80 mV and +40 mV at 10 mV intervals from a holding potential of −80 mV, to yield current-voltage relationships and examine activation kinetics. We analyzed data using Clampfit (Molecular Devices) and GraphPad Prism software (GraphPad, San Diego, CA, USA), stating values as mean ± SEM. We plotted raw or normalized tail currents versus prepulse voltage and fitted with a single Boltzmann function:

$$\mathit{g}=\frac{({\mathit{A}}_1-{\mathit{A}}_2)}{\left\{1+\mathit{e}\mathit{x}\mathit{p}[{\mathit{V}}_{\frac{1}{2}}-\mathit{V}/\mathit{V}\mathit{s}]\right\}\mathit{y}+{\mathit{A}}_2}$$ Eq. 1

where $\mathit{g}$ is the normalized tail conductance, ${\mathrm{A}}_1$ is the initial value at −∞, ${\mathrm{A}}_2$ is the final value at +∞, ${\mathrm{V}}_{\frac{1}{2}}$ is the half-maximal voltage of activation and ${\mathrm{V}}_{\mathrm{s}}$ the slope factor. We fitted activation and deactivation kinetics with single exponential functions.

Mesenteric artery myography

Rats were group-housed with regular 12-hour light/dark cycles, in clear plastic containers with ad libitum access to food and water and underwent at least one week of habituation. In accordance with the methods of killing animals described in annex IV of the EU Directive 2010/63EU on the protection of animals used for scientific purposes, male and female Wistar rats, 12 weeks old (Janvier Labs, France), were made unconscious by a single, percussive blow to the head. Immediately after the onset of unconsciousness, cervical dislocation was performed. For female rats, after euthanasia, 50 μL physiological saline solution (PSS) containing (in mM): 121 NaCl, 2.8 KCl, 1.6 CaCl₂, 25 NaHCO₃, 1.2 KH₂HPO₄, 1.2 MgSO₄, 0.03 EDTA, and 5.5 glucose dissolved in water filtered to 18.2 mΩ-cm purity via Q-gard 1 (QGARD00R1, Merk, Germany), was flushed 4–5 times within the vaginal canal. The cell containing suspension was plated on a glass coverslip then imaged at x20 magnification via Axiovert A.1 microscope (Zeiss). Images were recorded, and processed via Axiocam 705 mono (Zeiss) and ZEN blue (Ver 3.5.093, Zeiss). Changes in the number of leukocytes, small nucleated epithelial cells and large, anucleated cornified epithelial cells allowed for the determination of estrous cycle as described previously²⁵. Female rats were then categorized into either proestrus/estrus (P/E), or diestrus/metestrus (D/M).

At the same time, the intestines were removed and placed in ice-cold PSS. Third-order mesenteric arteries were dissected in ice-cold PSS, cut to 2 mm in length, and mounted on 40 μm stainless steel wires in a myograph (Danish Myo Technology, Aarhus, Denmark) for isometric tension recordings. The chambers of the myograph contained PSS maintained at 37 °C and aerated with 95% O₂/5% CO₂. Changes in tension were recorded by PowerLab and Chart software (ADInstruments, Oxford, United Kingdom). The arteries were equilibrated for 30 minutes and normalized to passive force. Artery segments were either precontracted with 10 μM methoxamine (Sigma; Copenhagen, Denmark) or, or contracted with isotonic high K+ physiological salt solution (K⁺PSS) of the following composition (in mM): 63.5 NaCl, 60 KCl, 1.17 MgSO₄·7H₂0, 1.18 NaH₂PO₄, 25 NaHCO₃, 5 glucose and 1.25 CaCl₂ balanced to a pH of 7.4 by gassing with 95% O₂ and 5% CO₂. Arteries were contracted in the absence or presence of linopirdine (10 μM) (Sigma; Copenhagen, Denmark), before application of rosemary extracts or constituents as described in the text.

Statistics and Reproducibility

All values are expressed as mean ± SEM. Comparison of two groups was conducted using a t-test; all p values were two-sided. Electrophysiological data were confirmed in at least two batches of oocytes. Replicates are defined as numbers of oocytes or artery preparations; sample sizes are given in the figure legends.

Results

Rosemary extract opens homomeric and heteromeric vascular KCNQ isoforms

Rosemary, an evergreen flowering plant, exhibits clusters of white to pale blue flowers (Figure 1A) among needle-shaped, aromatic leaves (Figure 1B). We performed a methanolic extraction on rosemary aerial parts (leaves and flowers) and after evaporation of the methanol, tested the 1:50 aqueous rosemary extract by bath application to Xenopus laevis oocytes heterologously expressing human KCNQ4, KCNQ5 and heteromeric KCNQ4/5 channels (Figure 1C). The rosemary extract produced a striking increase in KCNQ5 constitutive current at highly hyperpolarized membrane potentials (especially negative to −80 mV, at which KCNQ5 is normally closed), with lesser, yet still consequential, hyperpolarization of midpoint voltage dependence of activation (V_0.5act) in KCNQ4 and KCNQ4/5 channels (Figure 1C, D). Thus, rosemary extract hyperpolarized the unclamped resting membrane potential of oocytes expressing KCNQ4, KCNQ5 or KCNQ4/5 by between −10 and −15 mV (Figure 1E).

Figure 1. Rosemary extract opens homomeric and heteromeric KCNQ4 and KCNQ5.

Error bars indicate SEM. n indicates number of oocytes. Statistical comparisons by one-way ANOVA. RAP, rosemary aerial parts.

A. Salvia rosmarinus flower cluster with leaves (image: GWA).

B. Salvia rosmarinus flower closeup (image: GWA).

C. Mean traces for KCNQ channels as indicated, expressed in oocytes in the absence (Control) or presence of 1:50 RAP extract. Scale bars lower left for each trace pair; voltage protocol upper inset; n = 5 per group.

D. Mean tail current (left) and normalized tail current (G/Gmax) (right) for traces as in C; n = 5 per group.

E. Mean unclamped oocyte membrane potential for oocytes as in C; n = 5 per group.

Rosemary compounds open homomeric and heteromeric vascular KCNQ isoforms

We next screened previously identified rosemary components (21) (Figure 2A) at 100 μM to determine active compounds for the KCNQ isoforms studied. Carnosol and carnosic acid each increased KCNQ4 peak current ~twofold (Figure 2B, C) without altering its voltage dependence of activation (Figure 2D). Each of these resulted in hyperpolarizations in E_M of oocytes expressing KCNQ4 (Figure 2E). Other rosemary components tested were not influential upon KCNQ4 current and E_M, except that ursolic acid was inhibitory (Figure 2A–E).

Figure 2. Rosemary extract compounds carnosol and carnosic acid open KCNQ4.

Error bars indicate SEM. n indicates number of oocytes. Statistical comparisons by one-way ANOVA.

A. Structures (upper) and surface charge plots (lower) of rosemary compounds used in the screen.

B. Mean traces for KCNQ4 in the absence (Control) or presence of rosemary compounds (100 μM) as indicated. Scale bars lower left for each trace; voltage protocol upper inset; n = 5–6 per group.

C. Mean tail current for KCNQ4 traces as in B; n = 5–6 per group.

D. Mean normalized tail current (G/Gmax) for KCNQ4 traces as in B; n = 5–6 per group.

E. Mean unclamped oocyte membrane potential for KCNQ4-expressing oocytes as in B; n = 5–6 per group.

Carnosol (100 μM) induced a minor (−10 mV) shift in KCNQ5 V_0.5act (Figure 3A–D). As we recently reported²⁶, carnosic acid introduces a constitutive component to KCNQ5 current that, at −120 mV, retains >5% of maximal activity (Figure 3A–D). Carnosol and carnosic acid each induced a hyperpolarization in oocyte E_M (Figure 3E). Although other rosemary constituents (except quinic acid) also induced minor (<10 mV) hyperpolarizations in KCNQ5 V_0.5act (Figure 3A–D), they did not alter E_M (Figure 3E).

Figure 3. Rosemary extract compounds carnosol and carnosic acid open KCNQ5.

Error bars indicate SEM. n indicates number of oocytes. Statistical comparisons by one-way ANOVA.

A. Structures (upper) and surface charge plots (lower) of rosemary compounds used in the screen.

B. Mean traces for KCNQ5 in the absence (Control) or presence of rosemary compounds (100 μM) as indicated. Scale bars lower left for each trace; voltage protocol upper inset; n = 4–5 per group.

C. Mean tail current for KCNQ5 traces as in B; n = 4–5 per group.

D. Mean normalized tail current (G/Gmax) for KCNQ5 traces as in B; n = 4–5 per group.

E. Mean unclamped oocyte membrane potential for KCNQ5-expressing oocytes as in B; n = 4–5 per group.

KCNQ4 and KCNQ5 form heteromeric channels in vascular smooth muscle cells, which also incorporate the KCNE4 single transmembrane domain ancillary subunit ⁷. Here, we found that KCNQ4/5 heteromers responded to carnosol (100 μM) similarly to KCNQ4, i.e., with a largely voltage-independent current increase (Figure 4A, B), although there was a relatively small (−6.2mV) negative shift in midpoint voltage dependence of activation (Figure 4C). Carnosol also hyperpolarized the E_M of cells expressing KCNQ4/5 (Figure 4D). Regulation of KCNQ4/KCNQ5/KCNE4 channels by carnosol was similar, inducing a −6.9 mV shift in the midpoint voltage dependence of activation (Figure 4A–D). The KCNQ4/KCNQ5 and KCNQ4/KCNQ5/KCNE4 responses to carnosic acid (100 μM) were intermediate between that of homomeric KCNQ4 and KCNQ5, with increased peak current at depolarized potentials, modest hyperpolarization of V_0.5act, increased constitutive current at hyperpolarized potentials, and hyperpolarization of E_M. Of note, peak KCNQ4/KCNQ5 activity at depolarized potentials was increased more by carnosic acid when KCNE4 was present (Figure 4A–D). Concentration response studies revealed that the voltage dependences of activation of KCNQ4 and KCNQ4/KCNQ5 were much less sensitive to carnosol than that of KCNQ5 (Figure 4E); a similar pattern was seen for ΔE_M (Figure 4F). In contrast, KCNQ4/KCNQ5 exhibited similar carnosic acid sensitivity to that of KCNQ4 with respect to ΔV_0.5act (Figure 4G) and similar carnosic acid sensitivity to that of KCNQ5 with respect to ΔE_M (Figure 4H; Supplementary Figures 1, 2).

Figure 4. Carnosic acid and carnosol open heteromeric KCNQ4/5 channels.

Error bars indicate SEM. n indicates number of oocytes. Statistical comparisons by one-way ANOVA.

A. Mean traces for KCNQ4/KCNQ5 or KCNQ4/KCNQ5/KCNE4 heteromers in the absence (Control) or presence of carnosol (100 μM) as indicated. Scale bars lower left; voltage protocol inset; n = 5–10 per group.

B. Mean peak prepulse (left) and tail (right) current for KCNQ4/KCNQ5 or KCNQ4/KCNQ5/KCNE4 traces with carnosol concentrations as indicated; n = 5–10 per group.

C. Mean normalized tail current (G/Gmax) for KCNQ4/KCNQ5 or KCNQ4/KCNQ5/KCNE4 recordings as in B; n = 5–10 per group.

D. Mean unclamped oocyte membrane potential for KCNQ4/KCNQ5 or KCNQ4/KCNQ5/KCNE4-expressing oocytes as in B; n = 5–10 per group.

E. Dose responses for channels as indicated, for ΔV_0.5act versus [carnosol], n = 4–5.

F. Dose responses for channels as indicated, for ΔE_M versus [carnosol], n = 4–5.

G. Dose responses for channels as indicated, for ΔV_0.5act versus [carnosic acid], n = 4–5.

H. Dose responses for channels as indicated, for ΔE_M versus [carnosic acid], n = 4–5.

Rosemary, carnosol and carnosic acid are KCNQ- and sex-dependent vasorelaxants

Rosemary extract, carnosol and carnosic acid concentration-dependently relaxed ex vivo rat mesenteric arteries pre-contracted with methoxamine (10 μM) (Figure 5A–F). EC₅₀ values were essentially sex-independent, although there was a trend toward higher EC₅₀ values in female rat artery tissue for rosemary extract, carnosol and carnosic acid (Figure 5B, D, F). Additionally, the concentration-dependence in male arterial tissue exhibited a steeper slope than that of female arterial tissue, such that at lower concentrations the rosemary extract and compounds were more efficacious in female tissue (Figure 5A, C, E).

Figure 5. Rosemary extract, carnosic acid and carnosol are effective relaxants of pre-contracted arterial tone in mesenteric arteries from both male and female rats.

A. Mean concentration-effect curves for relaxations produced in segments of preconstricted male (black, n = 6) and female (red, n = 12) rat mesenteric artery by rosemary extract (RME). Significance values generated via a 2-way ANOVA with a post-hoc Bonferroni test.

B. Mean EC₅₀ values for RME (n = 6–12). Significance values generated via Student’s t-test.

C. Mean concentration-effect curves for relaxations produced in segments of preconstricted male (black, n = 7) and female (red, n = 11) rat mesenteric artery by carnosol. Significance values generated via a 2-way ANOVA with a post-hoc Bonferroni test.

D. Mean EC₅₀ values for carnosol (n = 7–11). Significance values generated via Student’s t-test.

E. Mean concentration-effect curves for relaxations produced in segments of preconstricted male (black, n = 6) and female (red, n = 12) rat mesenteric artery by carnosic acid. Significance values generated via a 2-way ANOVA with a post-hoc Bonferroni test.

F. Mean EC₅₀ values for carnosic acid (n = 6–12). Significance values generated via Student’s t-test.

We next examined the sensitivity of the vasorelaxant effects of rosemary extract and its constituents to the pan-KCNQ inhibitor linopirdine. Relaxation of precontracted mesenteric artery tissue from male rats by rosemary extract, carnosol or carnosic acid was KCNQ-dependent, i.e., efficacy was reduced by linopirdine (10 μM), especially for the individual compounds but less markedly for rosemary extract (Figure 6A, B). In contrast to these effects, quinic acid, a rosemary compound that showed negligible effects on KCNQ4 and KCNQ5 currents, and no effect on the E_M of oocytes expressing KCNQ4 or KCNQ5 (Figures 2, 3), was an ineffective vasorelaxant in arterial segments from male animals and expressed no linopirdine sensitivity (Supplementary Figure 3). When investigating precontracted mesenteric arteries from female animals, KCNQ-dependence of relaxation was only observed for carnosol (Figure 6A, C).

Figure 6: Sex dependence of KCNQ contribution to rosemary extract and constituents.

A. Representative traces of myography demonstrating rosemary extract (RME), carnosol and carnosoic acid mediated relaxation of pre-contracted tone (10 μM methoxamine) in mesenteric arteries from male rats pre-incubated in solvent control DMSO (black) or 10 μM pan-KCNQ inhibitor Linopirdine (red).

B, C. Mean data (left) and scatter plot with mean ± SEM of EC₅₀ values generated from raw data (right) for RME, carnosol and carnosic acid mediated relaxation in arteries from male (B; n = 5–6) and female (C; n = 11–12) adult Wistar rats. Statistics: 2-way ANOVA with a post-hoc Bonferroni test (left) and an unpaired Student’s t-test (right) were used to generate significance values (*P<0.05).

To further investigate the mechanism underlying the sex dependence, we identified the estrus cycle stage of female rats immediately before artery harvest, categorizing females as either estrus/proestrus (P/E) or diestrus/metestrus (D/M), as we recently found that vascular KCNQ channel expression and function are highly estrogen-dependent ^13,14. (Figure 7A). Strikingly, arteries harvested from female rats in D/M were ≥tenfold more sensitive to the vasorelaxant effects of rosemary extract, carnosol or carnosic acid, versus rats in P/E (Figure 7B–D). Vasorelaxation of arteries from female rats in D/M by rosemary extract, carnosol and carnosic acid were more linopirdine-sensitive than for arteries from P/E rats, suggesting a greater role for KCNQ channels in mediating vasorelaxation by rosemary and its constituents in D/M rats than P/E rats (Figure 8A–F). The data in Figures 7 & 8 indicate that KCNQ channels underlie the high sensitivity of female D/M rat arteries to vasorelaxation by rosemary and its constituents.

Figure 7: Estrous cycle-dependent sensitivity to rosemary extract and constituents.

A. Representative images of cervical histology of female Wistar rats in either proestrus/estrus (P/E; black box) or diestrus/metestrus (D/M; red box), magnification x20. Arrows indicate the following: black, small nucleated epithelial cells; white, large anucleated epithelial cells; red, leukocytes/ neutrophils.

B, C. Mean data (B) and scatter plot with mean +− SEM of EC50s generated from raw data (C) for rosemary extract (RME), carnsosol and carnosoic acid mediated relaxation of pre-contracted tone (10 μM methoxamine) in mesenteric arteries from female rats in either P/E (black) or D/M (red; n = 6).

Statistics: 2-way ANOVA with a post-hoc Bonferroni test (B) and an unpaired Student’s t-test (C) were used to generate significance values (*P<0.05).

Figure 8. Estrous cycle dependent sensitivity of rosemary extract and derivative compounds to KCNQ block.

A. Mean data (left) and scatter plot (right) with mean ± SEM of EC₅₀ values for rosemary extract (RME), carnosol and carnosic acid mediated relaxation of pre-contracted tone (10 μM methoxamine) in mesenteric arteries from female rats in proestrus/estrus (P/E) pre-incubated in either solvent control (DMSO; black) or Linopirdine (10 μM; red; n = 6).

B. Mean data (left) and scatter plot (right) with mean ± SEM of EC₅₀ values for rosemary extract (RME), carnosol and carnosic acid mediated relaxation of pre-contracted tone (10 μM methoxamine) in mesenteric arteries from female rats in diestrus/metestrus (D/M) pre-incubated in either solvent control (DMSO; black) or Linopirdine (10 μM; red; n = 5–6).

Statistics: 2-way ANOVA with a post-hoc Bonferroni test (left) and an unpaired Student’s t-test (right) were used to generate significance values (*=P<0.05).

We also tested the effect of pretreatment with rosemary or its constituents on peak contraction by KPSS of arteries harvested from male rats. Rosemary extract diluted to 1:100, but not 1:1000, was effective at reducing the peak contraction by KPSS (Figure 9A, B), while carnosol and carnosic acid were not (Figure 9B) suggesting that a high rosemary extract concentration has the ability to partially block L-type calcium channels, although this effect cannot be attributed to carnosol or carnosic acid.

Figure 9. Rosemary extract, but not carnosol or carnosic acid, impairs KCl mediated contraction.

A. Representative traces of high-KCl physiological salt solution (KPSS) mediated contraction in mesenteric arteries from male rats. After the second contraction, vessels were incubated in either 1:1,000 (black) or 1:100 (grey) rosemary extract (RME) prior to the 3^rd contraction to KPSS.

B. Mean data for the peak of the 2^nd KPSS mediated contraction (circles), and the 3^rd KPSS mediated contraction (squares) in the presence of RME (1:1,000, black / 1:100, grey), carnosol (10 μM, black / 30 μM, grey) and carnosic acid (10 μM, black / 30 μM, grey; n = 5). An unpaired Student’s t-test (right) was used to generate significance values (*P<0.05).

We also studied the KCNQ5-KCNE4 response to rosemary and constituents, to model what might occur with reduced KCNQ4 expression in arteries (i.e., a relatively greater contribution from channels lacking KCNQ4). Using 1 ng KCNE4 cRNA per oocyte with 10 ng KCNQ5 cRNA, we recorded 1–2 μA peak KCNQ5/KCNE4 currents (Figure 10A) that showed reduced sensitivity to rosemary extract, increased sensitivity to carnosol, and similar sensitivity to carnosic acid (Figure 10B–D) compared to homomeric KCNQ5 (Figure 3). Carnosol (100 μM) induced a greater shift in activation voltage dependence of KCNQ5/KCNE4 (Figure 10A–C) compared to KCNQ4/KCNQ5/KCNE4 (Figure 4A–C), but unlike the latter (Figure 4B), carnosol did not increase KCNQ5/KCNE4 peak current (Figure 10B). Effects of carnosol on E_M of cells expressing either channel were comparable (Figure 4D; Figure 10D). Carnosic acid (100 μM) effects on KCNQ4/KCNQ5/KCNE4 and KCNQ5/KCNE4 were similar (Figure 4A–D; Figure 10A–D).

Figure 10. KCNE4 sensitizes KCNQ5 to carnosol.

Error bars indicate SEM. n indicates number of oocytes. Statistical comparisons by one-way ANOVA.

A. Mean traces for KCNQ5/KCNE4 heteromers in the absence (Control) or presence of 1:50 rosemary aerial parts extract (RAP), carnosol (100 μM) or carnosic acid (100 μM) as indicated. Scale bars lower left; voltage protocol inset; n = 5–9 per group.

B. Mean peak prepulse current for KCNQ5/KCNE4 heteromers as in A; n = 5–9 per group.

C. Mean normalized tail current (G/Gmax) for KCNQ5/KCNE4 heteromers as in A; n = 5–9 per group.

D. Mean unclamped oocyte membrane potential for KCNQ5-expressing oocytes as in B; n = 5–9 per group.

Discussion

Rosemary was a favorite botanical medicine of the great early civilizations of Ancient Egypt, Greece, Israel and Rome, where it was used to treat a wide range of disorders, from memory loss to bacterial infection^27,28. An evergreen flowering plant native to the Mediterranean and Asia, rosemary now enjoys an extended range encompassing northern Europe, Africa and the Americas. The Celtic Druids are reported to have used rosemary for indications such as headaches as far back as the 13^th century, and it may have reached Southern England even before the Romans invaded^27,28. Interestingly, potential mechanisms for the efficacy of rosemary against headache may include vasorelaxation²⁹. Rosemary is a highly versatile botanical medicine, with controlled studies reinforcing the veracity of traditional usage for enhancement of memory and sleep, analgesia, anxiety, depression, epilepsy, amelioration of withdrawal symptoms, and also its antimicrobial, anti-inflammatory and antioxidant properties (the last two properties being primarily attributed to rosmarinic acid and carnosic acid)^28,30,31. Previously, carnosol and carnosic acid were shown to induce a vasorelaxation in rat thoracic aorta segments. Although the mechanism of vasodilatation was not fully investigated, the authors found that rosemary extract increased nitric oxide, thereby suggesting the relaxation was endothelial-dependent. Additionally, oral administration of rosemary extract tablets containing carnosol (0.97 mg), carnosic acid (8.6 mg) and rosmarinic acid (10.3 mg) improved endothelial dysfunction assessed by flow-mediated dilation in healthy male volunteers, together with decreasing serum plasminogen-activator-inhibitor type 1 (PAI-1) activity³². These data suggested that rosemary extract could be protective against atherosclerosis. Our study further supports a cardiovascular protective effect of rosemary extract, particularly the active substances carnosic acid and carnosol. Importantly, we have discovered the KCNQ-dependence and sex-dependence of this therapeutic effect.

Our TEVC experiments thoroughly characterized the effect of rosemary extract and several of the active ingredients on KCNQ4 and KCNQ5 currents, as homomers, heteromers and in the absence and presence of the ancillary KCNE4 subunit. We found that homomeric KCNQ5 is more sensitive to carnosol, in terms of potency of negative-shifting the V_{0.5 act}, than KCNQ4/KCNQ5, which is slightly more sensitive than KCNQ4 (Figure 4E). A similar pattern is observed for E_M of cells expressing the above channels, although in this case, KCNQ4 is more sensitive than KCNQ4/KCNQ5, but at 100 μM, carnosol is more efficacious at hyperpolarizing E_M of cells expressing KCNQ4/5 than KCNQ4 (Figure 4F). For carnosic acid, KCNQ4 and KCNQ4/5 show equal sensitivity, and KCNQ5 much lower sensitivity, in terms of V_{0.5 act} (Figure 4G). With respect to ΔE_M, cells expressing KCNQ4 are much more sensitive to carnosic acid than those expressing KCNQ5 or KCNQ4/KCNQ5 (Figure 4H).

Having determined carnosic acid and carnosol to be the predominant active substances in rosemary extract that can activate KCNQ4/KCNQ5/KCNE4 channels, we investigated whether this translated into a physiological effect on arterial segments. To the best of our knowledge, this is the first study to investigate the effect of rosemary extract, carnosol and carnosic acid on resistance arteries, where they have a prominent vasodilatory effect, which is mediated partially through activation of vascular KCNQ channels, as determined by inhibition of the relaxations by the KCNQ channel blocker linopirdine. It is important to note that even in the presence of linopirdine, rosemary extract, carnosol and carnosic acid still elicited a full relaxation. This suggests these active substances work in addition through other vasodilatory mechanisms that have not been explored in this study. At 1:100, we found rosemary extract partially inhibited KPSS contractions, suggesting an inhibition of the L-type calcium channels on the vascular smooth muscle cells, but we did not see the same for carnosic acid or carnosol, further highlighting the complexity of rosemary extract and its potential cardiovascular protective effects.

In rosemary, carnosic acid is the major phenolic constituent, and as a typical diphenol, it has potent antioxidant activity. As an antioxidant, carnosic acid can scavenge free radicals, thereby decreasing the consumption of nitric oxide due to depletion of the free radicals. In addition, other antioxidants, such as H₂S, can activate KCNQ channels³³ as well as K_ATP and BK_Ca channels^34,35, both of which are also expressed in vascular smooth muscle and play a functional role in controlling arterial relaxations when activated. It is therefore possible that, particularly in the case of carnosic acid, other K⁺ channels are contributing to the relaxation, being activated either directly (as described previously for H₂S³³) or indirectly through the nitric oxide-cGMP pathway.

In previous studies, it was demonstrated that synthetic KCNQ channel activators, and also the pan-KCNQ inhibitor linopirdine, are more effective in female rats in D/M versus those in P/E¹³, similar to the observation herein for rosemary and its constituents (Figure 8). We found reduced KCNQ4 plasma membrane expression in arteries from P/E rats versus D/M female rats, due to higher plasma estradiol levels in P/E rats, while levels of KCNQ1, KCNQ5 and the regulatory subunit KCNE4 were unchanged¹³. In the current study, we confirm that activators of vascular KCNQ channels have a more prominent effect in females in D/M compared to P/E. Not only do these findings support the estrus cycle-dependent changes in KCNQ channel expression in the arterial wall, but they also confirm the ability of rosemary extract and its active substances to induce a KCNQ channel-dependent relaxation, which is enhanced when the KCNQ4 channel is upregulated. Such findings also highlight the need to assess potential antihypertensive small molecules in female (as well as male) preclinical models at different estrus stages, since the potency of the compounds can change by >10-fold depending on the stage of the cycle, thereby leading to unwanted and potentially dangerous side-effects.

Clearly, the pharmacology is complex, and even more so with complex mixtures such as are found in the whole plant extract, but the greater relative potency in terms of ΔE_M of carnosic acid on KCNQ4 versus KCNQ4/KCNQ5 or KCNQ5, and of KCNQ4 versus KCNQ4/KCNQ5, might contribute to the lesser sensitivity of arteries from female rats in P/E versus D/M, given the relatively higher KCNQ4 in the latter.

In addition to estradiol levels inducing changes in KCNQ channel activity in females, we previously found that 5α-dihydrotestosterone, which is higher in adult male versus female rats, increases expression of the single-pass transmembrane regulatory subunit KCNE4 in mouse ventricles³⁶. KCNE4 co-assembles with KCNQ4 and KCNQ5 in male rat mesenteric arteries and knockdown of KCNE4 protein results in reduced KCNQ4 expression in vascular smooth muscle cells and a more depolarized arterial membrane potential⁷. Interestingly, in Kcne4 knockout mice, the vasorelaxation to KCNQ channel-specific activators was only compromised in male mice, but not in females when compared to their respective wild-type controls⁶. Thus, changes in KCNE4 expression have a more pronounced effect on KCNQ channel activity in arteries from males compared to females. We did not observe striking differences in carnosol or carnosic acid sensitivity on KCNQ4/KCNQ5 or KCNQ5 channels upon co-expression of KCNE4, but it is feasible that effects of KCNE4 co-expression on KCNQ4 plasma membrane expression density in vivo in arteries contribute to the vascular effects of rosemary, carnosol and carnosic acid, particularly in males.

With respect to other KCNQ isoforms, we recently found that KCNQ1 is insensitive to rosemary extract. KCNQ2 and KCNQ3 homomers are weakly sensitive to rosemary extract, but much less so than KCNQ5; further investigation revealed that while KCNQ2 is insensitive to carnosic acid, KCNQ3 is hypersensitive, with 100 μM carnosic acid eliciting a −62 mV shift in the midpoint voltage dependence of KCNQ3 activation²⁶. Interestingly, KCNQ2/3 heteromers are insensitive to carnosic acid and weakly sensitive to rosemary extract, but KCNQ3/5 is highly sensitive, being robustly opened at hyperpolarized potentials both by rosemary extract and by carnosic acid. In contrast to KCNQ4 and KCNQ5, KCNQ2 and KCNQ2/3 are each insensitive to carnosol (100 μM), while KCNQ3 is weakly sensitive²⁶. The KCNQ isoform selectivity of rosemary and its constituents may hold clues to development of future KCNQ isoform-selective therapeutics, and the molecular basis for their selectivity is currently being studied.

Data and Materials Availability:

All datasets are included in the manuscript and supplementary information. Raw data files are available upon reasonable request.