INTRODUCTION

Farnesol is the dephosphorylated form of farnesyl pyrophosphate, the last precursor common to all branches of the mevalonate pathway (1). The metabolic and biologic importance of farnesol has been recently demonstrated by several reports that identified the isoprenol as a natural nonsterol regulatory component of 3-hydroxy-3-methylglutaryl-CoA reductase activity (2-4) and an inhibitor of neoplastic cell growth (5, 6). Farnesol is catabolized into farnesal, farnesoic acid, and prenyl dicarboxylic acids (7, 8). However, it can also be "re-phosphorylated" into farnesyl pyrophosphate and used for protein isoprenylation (9). The observation that shorter (C₁₀, geraniol) and longer (C₂₀, geranylgeraniol) isoprenols, which are metabolically and structurally related to farnesol, are devoid of biological activity (2, 3) suggest the existence of farnesol-specific cellular targets or binding sites. It has been proposed that farnesol inhibits the cytosol to membrane translocation of protein kinase C (PKC,¹ Ref. 10). An effect on PKC has also been observed with farnesylamine, a closely related structural analogue of farnesol (11). However, a direct effect on PKC is unlikely as farnesol does not affect PKC cellular localization in cell lines derived from normal tissue (12). Other studies have reported the existence of a farnesol-specific, orphan nuclear receptor in vertebrate cells, the farnesoid X-activated receptor, but the role of this receptor in cell signaling pathways still needs to be defined (13, 14).

We have shown that mevalonate (MVA) availability is an important determinant of vascular tone in animal and human arteries (15, 16). Decreased vascular MVA availability following treatment with lovastatin, a 3-hydroxy-3-methylglutaryl-CoA reductase inhibitor, was associated with an increase in the response to vasoconstrictors, whereas addition of MVA to the arteries inhibited vasoconstriction. These findings, together with the recently characterized metabolic importance of farnesol, led us to hypothesize that farnesol itself has vasoactive properties. In evaluating the functional properties of various farnesyl analogues in the vascular tissue (17), we indeed confirmed this hypothesis and observed that farnesol is a potent inhibitor of vasoconstriction which affects vascular tone in both animals and humans. The effect is rapid, dose-dependent, reversible, and specific of farnesol as geraniol and geranylgeraniol are inactive. The study further indicated that both GTP-binding protein-dependent contractions and those induced by KCl are inhibited by farnesol. We concluded that farnesol inhibits post-receptor and post-GTP-binding protein events and perhaps Ca²⁺ channels. However, the precise mechanism of action of farnesol on modulating vasoconstriction remained elusive. In the present study, we have explored further the vasoactive properties of farnesol and document that farnesol 1) inhibits Ca²⁺ signaling in arteries and vascular smooth muscle cells and 2) possesses Ca²⁺ channel blocker properties.

EXPERIMENTAL PROCEDURES

C₁₅ (farnesol), C₁₀ (geraniol), and C₂₀ (geranylgeraniol) isoprenols were purchased from Aldrich (trans-trans-farnesol, catalog number 27,754-1) and Sigma (trans-geraniol, product number G 5135, and all-trans-geranylgeraniol, product G 3278). Stock solutions of the isoprenols were prepared in ethanol. Fura-2/AM was purchased from Molecular Probes (Eugene, OR). All other chemicals were from Sigma except where specified.

Arteries (internal diameter of approximately 200 µm) were isolated from the mesenteric arterial bed of ~300-g male Wistar rats (Charles River Breeding Laboratories, Inc., Boston, MA). The vessels were mounted on an isometric myograph at 37 °C in Hepes-buffered (pH 7.40) salt solution (HBSS) containing (in mM): NaCl 130, KCl 4.7, MgSO₄ 1.17, Hepes 15, CaCl₂ 1.25, glucose 5. After stretching for optimal tension recording and maximum stimulation ("wake-up" procedure consisting of three consecutive stimulations with a high (100 mM) potassium salt solution and 10⁶ M NE) (16-18), the arteries were incubated for 3 h at room temperature and for 1 h at 37 °C in HBSS containing 50 µM fura-2/AM (16). Intravascular [Ca²⁺]_i was then determined using fluorescence ratios (340/380 nm excitation, 510 nm emission) as described previously (16, 19). The experiments were conducted with a dual wavelength spectrofluorometer (Spex Industries, Inc., Edison, NJ), a Nikon microscope, and a 25 × water-oil immersion Zeiss lens. Background fluorescence was determined before loading with fura-2 and subtracted from all fluorescence readings. Based upon our having shown previously that farnesol inhibits NE- and KCl-induced contraction (17), determinations of [Ca²⁺]_i were conducted in basal conditions, and then after challenge with either a high K⁺ (100 mM KCl) depolarizing solution or NE. Active tension (mN/mm) was recorded simultaneously.

A10 and A7r5 cells were purchased from the American Type Culture Collection (Rockville, MD) and used between passages 7 and 25. A10 cells were cultured in RPMI medium 1640 with 10% fetal calf serum, streptomycin (30 µg/ml), and penicillin (30 units/ml). A7r5 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and antibiotics. Both cell lines were cultured at 37 °C under a 5% CO₂ atmosphere. All cell culture reagents were obtained from Life Technologies, Inc.

A10 cells were grown to confluency on glass slides in 5% CO₂-buffered RPMI medium 1640 containing 10% fetal calf serum and antibiotics. Intracellular [Ca²⁺]_i was determined in adherent cells, after a 30-min loading with 2 µM fura-2/AM in HBSS as described previously (20). Intracellular [Ca²⁺]_i was measured at base line first and then after the addition of either one of the following agonists: KCl, arginine-vasopressin (AVP), 12,13-phorbol myristate acetate (PMA), and thapsigargin.

Inward barium (I_Ba) currents were studied in single cells by the patch-clamp technique in the perforated- (nystatin) or cell-attached configuration (21-23). In the perforated patch experiments, a List patch-clamp amplifier (model EPC 7) was used for current amplification and data acquisition; command potentials were controlled with commercial software programs using a CED1401 interface (Cambridge Electronic Design Ltd., Cambridge, UK). Currents were recorded from holding potential 80 mV during linear voltage ramps from 100 to +100 mV at 0.67 V/s or 400 ms-step pulses to 0 mV (pulse frequency = 0.2 Hz, Ref. 24). Analysis of the obtained currents was performed using CED Patch and Voltage Clamp Software Version 6.08 (Cambridge Electronic Design). Ba²⁺ was used as charge carrier; K⁺ currents were blocked by Cs⁺. The bath solution contained (in mM) NaCl 125, BaCl₂ 10.8, MgCl₂ 1, CsCl 5.4, glucose 10, and Hepes 10 (pH 7.4 at 24 °C). The patch pipette was filled with a solution containing (in mM) CsOH 75, CsCl 55, MgCl₂ 5, aspartic acid 75, Hepes 10 (pH 7.4 at 22 °C). Nystatin (Biochrom KG, Berlin, Germany) was dissolved in Me₂SO and diluted into the pipette solution to give a final concentration ranging from 50 to 100 µg/ml. The resistance of the pipettes was 2 to 4 M. Series resistance in perforated patch recordings was <20 M.

In cell-attached patch experiments, the pipette solution contained (in mM) BaCl₂ 110, Hepes 5, buffered to pH 7.4 with tetraethylammonium hydroxide. Patch pipettes were made of borosilicate glass and coated with wax to minimize capacitative transients and noise; pipette resistance was 7-10 M. The recordings were made with a Biologic model RK 300 amplifier, filtered (3 dB, 5-pole Tchebicheff filter) at 1 kHz and sampled at 5 kHz. Bath solution contained (in mM) potassium aspartate 140, MgCl₂ 5, EGTA 5, Hepes 10, buffered to pH 7.4 with KOH.

The cells were continuously perfused during all recordings. Farnesol and other drugs were applied by changing the bath solution. Final concentrations of ethanol, the diluent, were 0.1% (v/v). Experiments were carried out at room temperature (20-24 °C).

Rat mesenteric arteries were mounted on an isometric myograph at room temperature, stretched for optimal tension recording, and challenged with KCl and NE for maximum stimulation. The arteries were then exposed for ~15 min to a Ca²⁺-free cytoplasmic salt solution (CSS, Refs. 25 and 26) containing 130 mM potassium propionate, 4 mM MgCl₂, 4 mM Na₂ATP, 2 mM Tris maleate, 10 mM creatine phosphate, 0.1 mg/ml creatine phosphokinase, and 4 mM EGTA (pH 6.8 at 22 °C). The bath was emptied, and the arteries were covered with 10 µl of CSS containing 2 mM EGTA and 1,000 units/ml Staphylococcus aureus -toxin. After a 20-min incubation, the permeabilized arteries were washed 3 times with 4 mM EGTA-CSS, bathed in 2 mM EGTA-CSS, and exposed to cumulative addition of Ca²⁺ using a concentrated (0.1 M) CaCl₂ solution. The bath was maintained at 20-22 °C during tension recording, and results were calculated as % of the maximum response obtained during wake up with KCl and NE. Free Ca²⁺ concentrations were calculated as described previously (25-27).

Values are reported as mean ± S.E. Differences were assessed using paired or unpaired Student's t tests as appropriate, and a p value < 0.05 was assumed to indicate a significant difference.

RESULTS

Addition of KCl (100 mM) or NE (10⁵ M) to fura-2 loaded arteries induced a sharp [Ca²⁺]_i increase with slow (~300 s) return to base line (not shown). Exposure of arteries to farnesol (30 µM, 30-min incubation) reduced the peak [Ca²⁺]_i transients evoked by addition of NE to approximately 35% of control values (p < 0.0001, n = 10, Fig. 1). As noted in our previous report (17), NE-induced contractions were also reduced as follows: 2.40 ± 0.18 mN/mm versus 3.84 ± 0.24 mN/mm for farnesol and control, respectively (n = 10, p < 0.001). Under similar conditions, geraniol and geranylgeraniol were inactive on both [Ca²⁺]_i and tension development (not shown). Farnesol also decreased K⁺-evoked peak [Ca²⁺]_i transients, to approximately 35% of control values (Fig. 1, p < 0.001, n = 10). This was associated with a pronounced decrease in contraction, 1.20 ± 0.25 mN/mm versus 3.22 ± 0.22 mN/mm for farnesol and control, respectively (p < 0.001). Finally, farnesol significantly reduced basal [Ca²⁺]_i (Fig. 1, p < 0.001).

5

[View Larger Version of this Image (18K GIF file)]

The cells were loaded with fura-2/AM, and the Ca²⁺ response to KCl (25 mM) and AVP (20 nM) was studied after incubation with farnesol (25 µM, 30-min incubation) or ethanol (control). Farnesol did not affect basal [Ca²⁺]_i (35.8 ± 2.2 versus 31.7 ± 1.4 nM for farnesol and control, respectively, n = 25, p > 0.05, not significant). As illustrated in Fig. 2A, the response of control cells to KCl was characterized by a sharp rise in [Ca²⁺]_i (from 29.2 ± 4.8 to 72.2.0 ± 6.9 nM, mean ± S.E., n = 6) followed by a sustained plateau (55.0 ± 3.4 nM at 150 s). This response was totally inhibited by farnesol at 25 µM (both KCl^peak and KCl^150 s = 34.7 ± 8.2 nM, n = 5, p < 0.001 versus control values; Fig. 2A). The response to AVP was characterized by a large and rapid increase in [Ca²⁺]_i (AVP^peak = 293.6 ± 17.1 nM, n = 5) followed by a rapid decrease to a lower plateau (63.4 ± 3.3 nM at 400 s). The rise in [Ca²⁺]_i is believed to result from both Ca²⁺ influx and Ca²⁺ release from intracellular stores, whereas the plateau corresponds to the opening of store-regulated and voltage-sensitive plasma membrane Ca²⁺ channels (28). As shown in Fig. 2B, farnesol significantly reduced both phases of the response to AVP to 50.9 ± 6.1 (peak) and 73.7 ± 1.4% (plateau) of average control values (n = 5, p < 0.01).

[View Larger Version of this Image (15K GIF file)]

Fura-2-loaded A10 cells were incubated with or without farnesol (25 µM) for 30 min and challenged with either thapsigargin (TG) or with PMA. The addition of TG to control cells induced an immediate rise of free [Ca²⁺] (onset: 9 ± 2 s; rate: 5.1 ± 1.4 nmol × l¹ × s¹; n = 5) to a maximum (TG^peak) followed by a sustained plateau phase at ~400 s after TG addition (TG^400 s). This response is illustrated in Fig. 3A. In the absence of extracellular Ca²⁺, only the plateau phase was decreased (Fig. 3A), and no modification of the onset and the rate of the [Ca²⁺]_i transient was observed. Farnesol decreased TG^400 s but did not affect TG^peak (Fig. 3A); average TG^400 s values were 49.5 ± 3.5 and 103 ± 5.9 nM for farnesol and control, respectively (n = 5, p < 0.006), whereas TG^peak values were 83.7 ± 5.1 and 114.0 ± 5.6 nM (p = not significant). Onset (8 ± 1 s) and rate of the TG response (3 ± 1 nmol × liter¹ × s¹) were not modified by farnesol (p = not significant). In the absence of extracellular Ca²⁺, farnesol had no significant effect on either TG^peak (92.2 ± 4.4 versus 106.0 ± 3.3 nM for farnesol and control, respectively; n = 5, p = not significant) or TG^400s (45.8 ± 1.8 nM versus 55.8 ± 1.9 nM, p = not significant).

[View Larger Version of this Image (18K GIF file)]

As illustrated in Fig. 3B, addition of PMA to control cells induced a sharp rise of [Ca²⁺]_i to a maximum (PMA^peak = 80.0 ± 5.7 nM from a base line of 30.6 ± 3.1 nM, n = 7) with a relatively long onset (35 ± 7 s) as compared with TG transients. The peak response was followed by a slow decline in [Ca²⁺]_i. In the absence of extracellular Ca²⁺, no response to PMA was observed (Fig. 3B). Incubation with farnesol inhibited the PMA response in a concentration-dependent manner (PMA^peak = 59.0 ± 4.7 and 39.4 ± 3.8 nM for farnesol 5 and 25 µM, respectively; n = 5, p < 0.02 and p < 0.001 versus control). For a farnesol concentration of 5 µM, the rate of the [Ca²⁺]_i transients was decreased (Fig. 3B) but not the onset (42 ± 9 s; p = not significant versus control). For a concentration of 25 µM, the response to PMA was totally abolished. This was similar in magnitude to the effect of farnesol on the Ca²⁺ response to KCl and to the effect of "0 mM" extracellular [Ca²⁺] on the PMA response.

Two vascular smooth muscle cell lines (A10 and A7r5) were studied. Barium currents were reversibly inhibited by dihydropyridines (1 µM nimodipine) and augmented by Bay K 8644 (not shown). Both cell lines showed current-voltage relationships typical for high voltage-activated L-type Ca²⁺ channels, with apparent threshold and reversal potentials of approximately 50 and +60 mV, respectively (Fig. 4, A and B, and see Ref. 29). Low voltage-activated T-type calcium channels were observed only in approximately 10% of cells in both cell lines. Farnesol (10 µM) reduced the peak current in A10 cells to 30 ± 5% (p < 0.01, n = 5) and in A7r5 to 26 ± 6% (p < 0.001, n = 10) of control values. The effect of farnesol was completely reversible after wash out (Fig. 4, A-C) and was observed over the whole voltage range (Fig. 4, A and B). Geraniol, applied to the bath solution at the same concentration (10 µM, Fig. 4C), did not modify Ca²⁺ channel currents (peak current was reduced to 99 ± 5% of the control values, n = 5, p = not significant). Geranylgeraniol (10 µM, Fig. 4C) had a modest but reversible effect (reduction to 89 ± 3%, n = 5, p < 0.02). No change in leak current was observed during application of the isoprenols suggesting that, at these concentrations, they had no "detergent-like" effect on cell plasma membrane. Half-maximal inhibition of the Ca²⁺ channel current was observed at 2.2 mM farnesol (Fig. 4D). The Hill coefficient was 0.82, indicating 1:1 binding of farnesol.

[View Larger Version of this Image (27K GIF file)]

In these experiments, the effect of farnesol on single L-type Ca²⁺ channel within the patch was determined. I_Ba were evoked every 10 s by either 400-ms pulses to 10 mV (from holding potential of 40 mV) or steady state recordings at 10 mV. I_Ba was augmented by application of 1 µM ()-Bay K 8644 to the bath solution and eliminated by addition of 1 µM nimodipine, thus indicating the presence of functional L-type Ca²⁺ channels (not shown).

Control conditions contained minimal (50 nM) ()-Bay K 8644 in the bath solution to promote mode-2 gating and prolong openings. Typical control recordings with both long and short transient openings are shown in Fig. 5 (left) with corresponding histogram analysis. After control currents were recorded, farnesol (25 µM) was perfused over the cells. As shown in Fig. 5 (middle), the open probability was dramatically reduced in the presence of farnesol. Maximum, steady state inhibition was reached after approximately 3 min. No change in leak current was observed during farnesol application. The effect of farnesol was reversible as evidenced by recovery of unitary currents after wash out within 3-5 min (not shown) and full response to ()-Bay K 8644 (Fig. 5, right).

[View Larger Version of this Image (26K GIF file)]

Effect of Farnesol on Ca²⁺-dependent Contraction in -Toxin-permeabilized Arteries

As shown in Fig. 6 ("basal" curves), farnesol (30 µM, 60-min incubation) did not inhibit significantly Ca²⁺-dependent contractions in permeabilized arteries. The Ca²⁺ concentrations necessary to induce half-maximum contraction (pCa₅₀) were 732 ± 102 and 724 ± 78 nM for control and farnesol-treated vessels, respectively (p = not significant, n = 4). The response to Ca²⁺ was also determined in the presence of GTPS, a non-hydrolyzable form of GTP, and in the presence of PMA, a PKC activator (30). As previously reported (31, 32), both compounds shifted the Ca²⁺ dose-response curve to the left, indicating an increased sensitivity of the artery to Ca²⁺ (Fig. 6, GTPS and PMA curves). However, the sensitizing effect of GTPS and PMA was not modified by farnesol (same figure).

Effect of farnesol on Ca²⁺-dependent contraction in -toxin-permeabilized arteries.

[View Larger Version of this Image (22K GIF file)]

DISCUSSION

This study was conducted to elucidate the molecular mechanisms by which farnesol, a natural endogenous intermediate of the mevalonate pathway, inhibits vasoconstriction. Because elevation of intracellular Ca²⁺ concentration in response to either membrane depolarization or vasoconstrictors is the main trigger for vascular smooth muscle contraction (33), we first characterized the effect of farnesol on Ca²⁺ signaling. Experiments were conducted in both intact arteries and isolated vascular smooth muscle cells loaded with fura-2, a fluorescent cytosolic Ca²⁺ indicator (34). For the experiments with vascular smooth muscle cells, the clonal A10 cell line was used (35). These cells do not contract and do not respond to NE. However, they do respond to KCl and to arginine-vasopressin by elevating [Ca²⁺ ]_i and were therefore chosen to examine the impact of farnesol on Ca²⁺ signaling at the cellular level. It must be mentioned that farnesol inhibits contractions induced by AVP in intact arteries (36), thus further justifying the use of the cell line in elucidating the mechanism of action of farnesol.

The results of these experiments indicate that farnesol decreases agonist and depolarization-dependent [Ca²⁺]_i transients in arteries and vascular smooth muscle cells. These findings suggest that inhibition of contraction is the consequence of reduced Ca²⁺ signaling in the presence of farnesol. Interestingly, farnesol also decreases basal arterial [Ca²⁺]_i; this may explain our previous observation of a vasodilatory action of farnesol on resting human arteries, i.e. in the absence of agonist (17). The greater impact of farnesol on the KCl response compared with the AVP response observed in our cell experiments suggests that the primary action of farnesol is on plasma membrane-dependent Ca²⁺ influx, i.e. voltage-dependent Ca²⁺ channels, and not on Ca²⁺ release from intracellular stores. However, the reduction by farnesol of the peak response to AVP indicates a possible effect of farnesol on intracellular stores as both release from the stores and Ca²⁺ influx overlap during this phase of the response.

The issue of the origin of the reduction in Ca²⁺ signaling was further explored in another series of experiments in which TG, a specific inhibitor of endoplasmic reticulum Ca²⁺-ATPase (37), and PMA, a PKC activator, were used. The data show that farnesol mimics the effect of extracellular Ca²⁺ removal on both TG- and PMA-induced Ca²⁺]_i transients. As to the TG response, farnesol decreases TG^400 s without affecting TG^peak. Studies by others have suggested that the peak response to TG reflects the Ca²⁺ leak from endoplasmic reticulum, inositol 1,4,5-trisphosphate-sensitive stores, and the plateau reflects activation of Ca²⁺ influx across plasma membrane (38). Indeed, in the absence of extracellular Ca²⁺, only the plateau phase was decreased (Fig. 3A). Although these results do not exclude the possibility that other cellular Ca²⁺ stores implicated in Ca²⁺ signaling such as the caffeine-sensitive, thapsigargin-insensitive Ca²⁺ stores (39) could be affected by farnesol, they strongly suggest that farnesol inhibits Ca²⁺ entry from extracellular space and does not impact on the endoplasmic reticulum Ca²⁺ stores. An inhibitory effect of farnesol on Ca²⁺ entry was further supported by the observation that farnesol abolishes the [Ca²⁺]_i response to PMA. In our experimental conditions, this response is the sole consequence of an activation of plasma membrane Ca²⁺ influx since it is totally inhibited by extracellular Ca²⁺ removal (Fig. 3B). It is actually possible that in A10 cells, the phorbol ester action on [Ca²⁺]_i is mediated by an activation of L-type Ca²⁺ channels as observed in A7r5 cells, another rat aortic smooth muscle cell line (40). This possibility, together with our observation that KCl-dependent contraction and KCl-dependent Ca²⁺ signaling are strongly inhibited by farnesol, led us to postulate that the primary targets of farnesol were plasma membrane, voltage-dependent Ca²⁺ channels present in vascular smooth muscle cells.

Therefore, direct measurement of the activity of voltage-dependent Ca²⁺ channels was performed, using the patch-clamp techniques. The data, established in two different cell lines (A10 and A7r5) indicate that farnesol blocks vascular smooth muscle L-type Ca²⁺ channels (IC₅₀ ~ 2 µM, Fig. 4). This effect is reversible and specific of the C₁₅ structure since geraniol, C₁₀, is inactive, and geranylgeraniol, C₂₀, has only a limited inhibitory action on channel activity (10% inhibition of peak Ca²⁺ currents for a concentration of 10 µM as compared with 70% with equimolar concentration of farnesol).

The specificity of farnesol over geranylgeraniol may be only apparent and due to differences in solubility. Indeed, partition coefficients (logP) are 4.62 and 6.59 for farnesol and geranylgeraniol, respectively, indicating a 100-fold difference between the two isoprenols (by comparison, logP for geraniol = 2.65). However, biological activity does not necessarily correlate with biophysical constants (41). Considering that hydrophobicity favors both membrane incorporation and membrane permeability (42), and assuming a membrane site of action for the isoprenols, one could expect a greater effect of geranylgeraniol over farnesol. Elucidation of the exact cellular site of action of farnesol together with precise determination of actual membrane concentration using radiolabeled compounds and purified membrane preparations will help clarify this issue in the future.

Whether Ca²⁺ channel inhibition occurs within a "physiological" range of farnesol concentrations can only be speculated at this point. To our knowledge, there is no report documenting extracellular or plasma farnesol concentrations, and there is only one report (43) that gives a range of farnesol tissue concentration (approximately 50-200 ng/g wet weight of rat liver). Assuming a tissue water content of 70% and a homogeneous distribution, the calculated concentration of farnesol in hepatic tissue would be 0.1 to 1.2 µM. This is similar to the low range of farnesol concentrations active on Ca²⁺ channels and suggests that, providing similar concentrations are also present in the vascular tissue, our findings are physiologically meaningful.

Our study has not specifically examined the mechanism of action of farnesol on Ca²⁺ channels. However, we believe that our experiments already exclude a number of possible mechanisms whereby farnesol might inhibit Ca²⁺ channels, in particular those that involve messenger molecules such as PKC, cyclic nucleotides, and GTP-binding (G) proteins.

As discussed below, our -toxin experiments (Fig. 6) indicate that farnesol does not affect PKC activity. Moreover, our experiments with intact vessels show that farnesol inhibits KCl-induced contraction, a response that is not attenuated by down-regulation of the enzyme (44). Thus, primary inhibition of PKC by farnesol with secondary inhibition of voltage-gated channels is unlikely. The participation of cyclic nucleotides, cAMP and cGMP, to the inhibitory effect of farnesol was not explored directly in our study but is also unlikely as both nucleotides have been shown to decrease the sensitivity (shift to the right) of the smooth muscle to Ca²⁺ when tested in -toxin-permeabilized (rat mesenteric) arteries (45). As shown in Fig. 6, in no instance (basal, PMA, or GTP-S curves) do we see a significant shift of the Ca²⁺ dose-response curve to the right.

Alternatively, farnesol may inhibit one of the G proteins known to stimulate L-type Ca²⁺ channels. A direct regulation of L-type Ca²⁺ channels by G_s has been demonstrated in cardiac cells (46). This pathway may also be functional in A7r5 cells (47). Similarly, G_i has been implicated in the adrenergic activation of dihydropyridine-sensitive Ca²⁺ currents in rat portal vein myocytes (48). However, G_s inhibition (with GDPS) does not reduce significantly basal Ca²⁺ channel activity (46, 49, 50), and G_i inhibition with specific antibodies reduces Ca²⁺ currents only after receptor activation (48). As shown in Figs. 4 and 5, farnesol inhibits A10 and A7r5 L-type currents in the absence of any specific activation of the G protein pathway. Finally, a G_q/G₁₁-mediated regulation of the dihydropyridine-sensitive Ca²⁺ channels was described recently by Mironneau and Macrez-Lepretre (51) in rat portal vein myocytes. However, the study suggests that the G_q/G₁₁ effect on the channels is secondary to Ca²⁺ release from intracellular stores; as shown in Fig. 3A (thapsigargin response), farnesol does not significantly affect this response. Thus, our results do not support the hypothesis of a G protein-mediated effect of farnesol on vascular smooth muscle L-type Ca²⁺ channels.

Altogether, and by exclusion, we believe that farnesol may act directly on the channels. The presence of farnesol inside cells and tissues actually raises the question of its site of action. The results of our perforated patch experiments, in which farnesol is added to the bath and the activity of Ca²⁺ channels located outside of the patch pipette is measured, point to an extracellular action of farnesol on the channels. It is possible although that after addition to the bath and because of its hydrophobicity farnesol crosses the plasma membrane and blocks Ca²⁺ channels from the intracellular side.

Cell-attached experiments were conducted to provide further insight on this issue. Their results indicate that farnesol inhibits the Ca²⁺ channels present within the patch, in tight seal configuration. One of the most likely explanations for the effect of farnesol in these conditions is that farnesol first diffuses intracellularly through the plasma membrane and then reaches its site of action within the patch from the cytosolic side of the membrane. Such a mechanism has been proposed for methoxyverapamil, a Ca²⁺ channel blocker of the phenylalkylamine class (52) Alternatively, a lateral diffusion of farnesol in the lipids of the plasma membrane followed by binding to an intra-membrane site of the channel could be postulated, as proposed for dihydropyridine-like Ca²⁺ channel blockers (53). Although the issue of the exact site of action of farnesol cannot be fully resolved with the present data, our results strongly suggest that intracellularly produced farnesol is capable of interacting with plasma membrane L-type Ca²⁺ channels and thus support the concept of farnesol being a natural, endogenous Ca²⁺ channel blocker.

Contraction also depends on the functional integrity of vascular smooth muscle Ca²⁺-sensitive pathways, which include enzymes directly activated by Ca²⁺ as well as systems controlling the sensitivity of the smooth muscle to Ca²⁺ (26, 54). The possibility that farnesol might inhibit these pathways was then examined in a last series of experiments conducted in -toxin-permeabilized arteries. In these preparations, Ca²⁺-sensitive pathways can be studied directly since intravascular [Ca²⁺]_i is controlled using EGTA-containing buffer solutions, and the regulation of [Ca²⁺]_i by Ca²⁺ channels and intracellular stores is bypassed. The data indicate that farnesol has no direct effect on the Ca²⁺-responsive elements implicated in arterial contraction, including those activated by GTP and PKC. The absence of effect of farnesol in -toxin-permeabilized arteries is not due to its inability to penetrate the tissue, because farnesol is a small molecular weight (M_r = 222.4) molecule and is likely to diffuse freely inside the arterial smooth muscle cells through the pores created by the toxin (54, 55). Therefore, our findings suggest that the vascular action of the isoprenol is solely the consequence of its inhibition of Ca²⁺ signaling.

In conclusion, we have shown that farnesol reduces Ca²⁺ signaling in vascular smooth muscle cells and arteries and inhibits voltage-dependent L-type Ca²⁺ currents. Our studies have excluded an effect of farnesol on the pathways implicated in smooth muscle contraction beyond Ca²⁺ signaling and point to the plasma membrane as a primary site of cellular action. These properties of farnesol are not shared by other metabolically related isoprenols and likely account for its inhibitory effect on vasoconstriction. Blocking Ca²⁺ signaling with specific compounds has been the focus of intensive pharmacological research and the goal of several therapeutic strategies including those applied to the treatment of hypertension and atherosclerosis. Since farnesol is an intermediate of the mevalonate pathway, our findings are consistent with farnesol acting as a natural, endogenous Ca²⁺ blocker and suggest that controlling intracellular levels of farnesol or farnesol analogues might be useful in the regulation of vascular tone in vivo.