INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

ENDOTHELIAL CELLS play a central role in the regulation of vascular tone through the release of soluble mediators such as prostacyclin (PGI₂), nitric oxide (NO), and endothelium-derived hyperpolarizing factor (EDHF). In the cerebral circulation, PGI₂ and NO have, by far, received the greatest attention. Hence, little is known about the relative extent of EDHF contributions to the control of cerebrovascular tone. Even if one includes the much larger body of information derived from studies on peripheral vascular tissue, the chemical identity of EDHF remains controversial. Possibilities include products of arachidonate metabolism (epoxides and anandamide), potassium ions, cAMP, and reactive oxygen species (2, 4, 15, 19, 28). Additional evidence points to gap junctions between endothelial cells and/or smooth muscle cells as important participants in EDHF-mediated vascular relaxation (12, 33).

Gap junctions are sites of electrical coupling between cells of the vascular wall and play a role in the coordination of vascular responses (e.g., Refs. 10 and 11). Gap junctions are composed of proteins called connexins (Cx). At least 14 homologous connexin isoforms have been identified in rats. Cx37, Cx40, Cx43, and Cx45 are expressed in cerebral vessels (22-24, 38). The two cells linked by a gap junction contribute equally to its structure. Thus, in each cell, six connexins are bound together in a radial configuration, forming a central aqueous channel. That hexameric structure, which spans the plasma membrane, is termed a connexon. A gap junction is created when connexons from adjacent cells dock. That docking mechanism can be disrupted by inhibitory peptides, such as Gap 27, which interacts with a region of the second extracellular loops of Cx43 and Cx37 (5). Gap 27 has proven to be a useful tool in studying the role of gap junctions in EDHF-type relaxations (12).

It was recently been reported (15, 37) that the degree to which a putative EDHF participates in agonist-induced cerebrovasodilation may be modulated by estrogen. Specifically, the findings from those studies indicated that purinergic (P₂Y) receptor-mediated vasodilation [which is, to a significant degree, endothelium NO synthase (NOS) dependent, but EDHF independent, in "estrogen-normal" females] is transformed to an EDHF-dependent response in estrogen-depleted, i.e., ovariectomized (Ovx), females. Additional evidence from those reports suggested that the apparent EDHF contributing to that response was not derived from activation of an epoxygenase.

The present study was, therefore, undertaken to test the hypothesis that the ovariectomy-associated switchover to EDHF dependency in the pial arteriolar response to the P₂Y₁ agonist ADP (see Ref. 37) relates to a gap junctional mechanism. To that end, we compared in vivo pial arteriolar reactivities to topical applications of ADP in intact and Ovx female rats with respect to 1) the endothelial dependence of the ADP response [using a light dye (L/D) endothelial injury procedure]; and 2) in the absence and/or presence of NO and PGI₂ synthesis inhibition, the effect of gap junctional disruption on the ADP response by using topical applications of gap junction inhibitory peptides, or a Cx43 antisense oligonucleotide. Cx43 was targeted because it has been shown to be particularly well expressed in rat pial arterioles (24).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The experimental protocol was approved by the Institutional Animal Care and Use Committee. Two groups of age-matched female Sprague-Dawley rats (300-400 g) were used: Ovx and intact female rats. Ovariectomies were performed by the supplier (Charles River; Wilmington, MA) at 4-6 wk before the study. Pial arteriolar reactivities were evaluated by using a closed window and intravital microscopy system as previously described (36). After anesthesia induction with halothane and paralysis (curare), the rats were tracheotomized and mechanically ventilated. Bilateral femoral arterial and venous catheter insertion was performed under continuous anesthesia with 0.8% halothane plus 70% N₂-30% O₂ for arterial blood gas measurement and drug infusion. After catheterization was completed, the rat was placed in a head holder, and a craniotomy (10 mm in diameter) was performed over the midline of the skull. After the underlying dura was removed, a cranial window (11 mm diameter) equipped with three ports (inflow, outflow, and intracranial pressure monitoring) was fixed to the skull. Halothane was discontinued, and a 10 µg/kg fentanyl bolus was given intravenously. Anesthesia was maintained during the study with fentanyl (25 µg · kg¹ · h¹ iv) and 70% N₂-30% O₂. The space under the window was filled with artificial cerebrospinal fluid (aCSF, pH 7.35) that was equilibrated with a gas consisting of 20% O₂-5% CO₂ with a balance of N₂. The aCSF solution was suffused at 1.0 ml/min and maintained at 37°C. Body temperature was maintained at 37°C with a servocontrol heating pad, and mean arterial pressure and intracranial pressure were monitored continuously during the experiment.

Vascular reactivity was assessed by measuring the diameters of pial arterioles (20-50 µm). The vessels were viewed by videomicroscopy and measurements made by using a calibrated videomicroscaler (see Ref. 36). In all experiments, the initial diameter measurements were made 1 h posthalothane and after 40 min drug-free aCSF suffusion. Hypercapnia (PCO₂ 70 mmHg) was then imposed to test pial arteriolar reactivity. Only those vessels displaying adequate responses to CO₂ (reactivity > 1.0% diameter increase/mmHg PCO₂ change) were selected for this study.

In advance of the experiments described below, we endeavored to optimize the L/D endothelial injury procedure (details given below). Thus, in addition to measuring hypercapnic responses, we also monitored pial arteriolar reactivities to suffusions of acetylcholine (ACh, 10 and 100 µM) and the NO donor S-nitroso-N-acetylpenicillamine (SNAP, 0.1 and 1.0 µM) before and after L/D exposure. Previous studies from our laboratory (34) established that ACh-induced pial arteriolar dilation in rats is entirely endothelium dependent, whereas the responses to SNAP and hypercapnia are unaffected by endothelial injury and reflect vascular smooth muscle function. Thus a finding of loss of ACh reactivity, but no changes in the responses to SNAP and CO₂, indicates adequate L/D exposure. However, because pial arteriolar responses to ACh are abolished in Ovx female rats (28), only intact females were used in the above evaluation.

The remaining rats were divided into four experimental subgroups, according to different treatments: 1) L/D endothelial injury; 2) nonselective NOS inhibitor N-nitro-L-arginine (L-NNA) coadministration with the cyclooxygenase inhibitor indomethacin (Indo); 3) suffusion of the gap junction inhibitory peptides Gap 27 and Gap 26; and 4) topical application of Cx43 antisense or missense oligonucleotide. In subgroup 1, following the initial measurement of CO₂ reactivity, ADP, at concentrations of 10 and 100 µM, was suffused into the space under the cranial window (5 min each concentration). After 10 min of drug-free aCSF suffusion, a baseline measurement was made, and 0.1 and 1.0 µM SNAP were added sequentially. At this time, the animals were subjected to L/D injury (see Ref. 34). For this procedure, mercury light was passed through a filter that allowed transmission only at 450-490 nm. The light passed coaxially through the microscope objective and was focused on an area of 600 µM in diameter, including the vessels being studied. Two percent fluorescein dye (0.8 ml/100 g body wt) was given intravenously, and the mercury light was turned off 55 s after initiation of the fluorescein dye infusion. The responses to ADP or SNAP were evaluated before and after the L/D application. In subgroup 2, after initial measurements of CO₂, ADP, and SNAP-induced responses, L-NNA (1 mM) was added to aCSF and suffused for 40 min, at which time Indo (10 mg/kg) was given intravenously. After another 20 min of continued L-NNA suffusion, the responses to ADP and SNAP were again evaluated. At this point, L/D injury was imposed, and the ADP- and SNAP-induced responses were measured once more. In subgroup 3a, Gap 27 (300 µM) was suffused 1 h before reevaluation of ADP and SNAP responses. This was followed by combined L-NNA and Indo (see above) and measurement of ADP and SNAP reactivities. Finally, L/D injury was performed, and the ADP and SNAP suffusion sequence was repeated. Gap 26 applications (subgroup 3b) were performed only in Ovx females. Responses to ADP and SNAP were measured first in the absence of any added inhibitors and then in the presence of Gap 26 (300 µM) followed by Gap 26 + Gap 27 (both 300 µM). The rationale for using both Gap 27 and Gap 26 derives from the fact that Gap 27 targets Cx43 and Cx37, whereas Gap 26 targets Cx40 and Cx37 (5). By studying these two inhibitory peptides separately, one may be able to better establish which of the three is involved in the ADP response. For example, a finding that Gap27, but not Gap 26, reduces the response would indicate Cx43 participation, whereas the converse would be suggestive of Cx40 involvement.

In subgroup 4, 300 µl of a solution containing either 5 µM Cx43 antisense (5'-GTCACCCATGTCTGG-3') or 5 µM Cx43 missense (5'-CTGGACCAGTGCTCT-3') were injected into the space under the cranial window 24 h before analysis of pial arteriolar reactivity. The procedure for chronic placement of cranial windows in experiments utilizing topical applications of oligonucleotides was described in a previous paper from our laboratory (36). Six bases (3 at the 5'-end and 3 at the 3'-end) were phosphorothioated to minimize nuclease-mediated oligonucleotide breakdown. On the day of the study, pial arteriolar responses to CO₂, ADP, and SNAP were evaluated as described above.

Two animals each from the antisense- and missense-treated Ovx groups were analyzed for Cx43 protein expression in the pial tissue removed from under the cranial windows. The tissue was obtained on the day following oligonucleotide (or aCSF) application (see Ref. 36) and prepared for Western immunoblotting utilizing a procedure described in a recent publication from our laboratory (36). For the present assay, a monoclonal Cx43 primary antibody was used (2,000:1 dilution, BD Transduction Laboratories; Lexington, KY), along with a goat anti-mouse secondary antibody. In each gel, a Cx43-positive control (from rat cerebrum lysate-obtained from BD Transduction Laboratories) was run with the tissue samples from the missense- and antisense-treated animals.

Additional Ovx females were treated with oligonucleotides or oligonucleotide-free aCSF. Twenty-four hours later, the brains were perfusion fixed with 400 ml of cold 2% paraformaldehyde in phosphate-buffered saline, introduced transcardially. The brains were then cut into blocks and placed into 2% paraformaldehyde for several hours and then transferred to 70% ethanol. Those samples were stored at 4°C until commencement of paraffin embedding. Coronal sections (8 µm) were prepared from the paraffin-embedded blocks using a microtome (36). Cx43 was detected using an immunofluorescence procedure (36), where the primary antibody was the same as that described above, and the secondary antibody (see also above) was conjugated to the fluorophore Cy3. In adjacent slices from the same brains, immunofluorescence detection of Cx40 or Cx37 was performed by using rabbit polyclonal primary antibodies (250:1 dilutions, from Alpha Diagnostics; San Antonio, TX) and goat anti-rabbit secondary antibodies conjugated to Cy3 (Cx40) or fluorescein (Cx37).

In all experiments, arterial blood samples were taken at 30-min intervals for arterial PO₂ (Pa_O2), arterial PCO₂ (Pa_CO2), and pH analysis by using a Radiometer Copenhagen blood gas-pH analyzer (model ABL 520). Pa_O2 (100 mmHg), Pa_CO2 (30-40 mmHg), and pH (7.40) were maintained during the study.

All reagents were obtained from Sigma (St. Louis, MO) and dissolved in aCSF unless otherwise stated. Gap 27 peptide (SRPTEKTIFII) and Gap 26 peptide (VCYDQAFPISHIR) were synthesized by the Protein Research Lab at the University of Illinois at Chicago. The purities of the peptides were >95%. The oligonucleotides were obtained from Sigma Genosys (St. Louis, MO). Indo was dissolved in 1% bicarbonate solution. Values are presented as means ± SE. Comparisons of arteriolar diameter values within groups were made using one-way repeated-measures ANOVA combined with a post hoc Tukey analysis. Analyses of diameter changes between groups were made using Student's t-test. A P value <0.05 was considered as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In all experiments, the physiological variables were within normal limits. That is, Pa_CO2, pH, and mean arterial blood pressure (MABP) in these groups did not show any significant differences when initial and final values were compared during the experiments. The values measured in subgroups 1-3 are presented in Table 1. Not included in Table 1 are the Pa_O2 values, which were maintained above 100 mmHg in all the studied rats. Also provided in Table 1 are the initial pial arteriolar diameters in the first three subgroups. The arterial blood data for subgroup 4 on the day of the study (not shown) were all within the ranges seen in other subgroups. Initial pial arteriolar diameters in the intact antisense- and missense-treated females were 41.2 ± 1.3 and 37.4 ± 0.6 µm, respectively, whereas in the Ovx rats, those values were 36.3 ± 2.5 and 34.7 ± 3.7 µm, respectively.

The initial experiments involved optimizing the L/D procedure. Thus the reactivities of selected arterioles were evaluated after different times of mercury light exposure. As shown in Fig. 1, it was found that 55-s L/D exposure was optimal for injuring the endothelium but not the smooth muscle. That is, that period of exposure eliminated ACh-induced vasodilation completely without affecting the responses to SNAP and hypercapnia. Thus the results described below were obtained after 55-s L/D exposure. In subgroup 1 experiments, initial applications of ADP (10 and 100 µM) induced similar dose-dependent vasodilations in the intact and Ovx females (15.7 ± 8.6% and 37.7 ± 4.7% in intact rats, 14.2 ± 1.9% and 32.1 ± 5.8% in Ovx rats, respectively). Subsequent L/D exposure was found to reduce the response to ADP (by 40-60%) in both groups, with no difference in the magnitude of those reductions seen when intact rats were compared with Ovx rats (Fig. 2).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Effect of mercury light plus fluorescein dye (L/D) endothelium injury on acetylcholine (ACh, A)-, hypercapnia [B, expressed as % diameter increase per mmHg increase in arterial PCO2 (PaCO2)]-, and S-nitroso-N-acetylpenicillamine (SNAP, C)-induced pial arteriolar dilation. Values are means ± SE; n = 4 rats in each group. * P < 0.05 vs. initial.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   ADP-induced pial arteriolar dilation in intact (A) and ovariectomized (Ovx) (B) female rats before and after L/D endothelium injury. Values are means ± SE; n = 4 rats in each group. * P < 0.05 vs. initial.

In subgroup 2 experiments, before coadministration of Indo and L-NNA, as above, ADP elicited statistically similar dose-dependent responses in intact and Ovx rats (15.7 ± 8.6% and 37.7 ± 4.7% in the intact rats, 13.9 ± 1.3% and 32.9 ± 4.9% in Ovx rats, respectively). Administration of L-NNA + Indo did not affect the ADP response in Ovx females but reduced the ADP response by 50-60% in intact rats (Fig. 3). However, in the Ovx group, L/D exposure, applied following L-NNA + Indo, was still accompanied by a significant 50-60% reduction in ADP-induced relaxation (Fig. 3). Suffusion of SNAP at 0.1 and 1.0 µM produced identical pial arteriolar responses in the intact and Ovx females. Treatment with L-NNA + Indo, in the absence or presence of L/D exposure, had no effect on SNAP-induced dilations in either group (Fig. 4).

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Pial arteriolar diameter increases-induced by ADP suffusions in intact (A) and ovariectomized (B) rats in the absence (initial) and in the presence of combined application of N-nitro-L-arginine (L-NNA, 1 mM) and indomethacin (Indo, 10 mg/kg iv) followed by addition of L/D endothelial injury. Values are means ± SE; n = 4 rats in each group. * P < 0.05 vs. initial.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Effects of combined administration of L-NNA (1 mM) and Indo (10 mg/kg iv) followed by L/D endothelial injury on SNAP (0.1 and 1.0 µM)-elicited pial arteriolar dilation. Results for intact (A) and Ovx (B) females are presented. Values are means ± SE; n = 4 rats in each group.

When Gap 27 was added to the suffusate in subgroup 3, a statistically significant decrease in baseline was found in the intact and Ovx groups compared with the initial value (20.7 ± 4.3% and 16.0 ± 1.6%, respectively). The ADP (10 and 100 µM) response was diminished (by 50%) only in Ovx rats (Fig. 5). In marked contrast, Gap 27 was completely without effect on the ADP response in intact female rats. Adding L-NNA + Indo on top of the Gap 27 produced no further decrease in ADP reactivity in Ovx females, but, in the intact females, a ~50% diminution was observed (Fig. 5). That attenuation of ADP-induced relaxation was statistically similar to that observed in subgroup 2 intact females. L/D injury imposed subsequent to initiation of L-NNA + Indo treatment produced no further effects on arteriolar reactivities in the intact group (data not shown). In Ovx females, addition of Gap 26 was accompanied by a significant reduction (8.5 ± 2.2%) in the baseline diameter (not shown). However, no changes in ADP reactivity were observed in Ovx rats in the presence of the Gap 26 peptide (Fig. 6). When Gap 27 was subsequently added to the suffusate, the ensuing ADP response was significantly reduced (by ~50%, Fig. 6). Neither Gap 27 (Fig. 7) nor Gap 26 (not shown) altered SNAP-induced vasodilations.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Effect of Gap 27 on ADP-induced pial arteriolar dilation in intact (A) and Ovx (B) female rats. Gap 27 (300 µM) was suffused first in the absence and then in the presence of L-NNA + Indo. Values are means ± SE; n = 4 rats in each group. * P < 0.05 vs. initial.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Effect of Gap 26 (300 µM) followed by the combination of Gap 26 + Gap 27 (both 300 µM) on ADP-induced pial arteriolar dilation in Ovx female rats. Values are means ± SE; n = 4 rats. * P < 0.05 vs. Gap 26 + Gap 27.

View larger version (17K): [in this window] [in a new window]   Fig. 7.   Effects of Gap 27 (300 µM) on SNAP-induced pial arteriolar dilation in intact (A) and Ovx (B) female rats. Gap 27 was suffused in the absence and presence of L-NNA + Indo. Values are means ± SE; n = 4 rats in each group. * P < 0.05 vs. initial.

Prolonged exposure of pial arterioles to Cx43 antisense in intact females was not associated with any reduction in the ADP response relative to the diameter change observed in the presence of the (control) missense oligonucleotide (Fig. 8A). On the other hand, when the same oligonucleotide applications were performed in Ovx females, the ADP-induced diameter increase was 50-60% lower in antisense relative to missense-treated females (Fig. 8B). Also, in the rats given missense, the 15% and 30% increases in diameter during exposure to ADP at 10 and 100 µM, respectively, was virtually the same as that seen in the preceding subgroups before any experimental manipulation. No differences in the SNAP responses were seen when comparing antisense- and missense-exposed intact or Ovx females (data not shown). The functional evidence of antisense selectivity and efficacy provided by the results summarized in Fig. 8 was corroborated on analysis of cranial window area pial-brain surface expression of Cx43 in the antisense- versus missense-treated Ovx females. Western immunoblot assay revealed a substantially lower Cx43 expression in pial tissue harvested from antisense- versus missense-treated rats (Fig. 9). A similarly diminished expression (antisense vs. missense) was indicated by immunofluorescence analysis of coronal sections of cortical surface tissue (Fig. 10). This was particularly emphasized by the appearance of a lesser expression in the pial vessels overlying the cortex and in the underlying surface tissue of the cerebral cortex (presumably the glia limitans). In one Ovx female given oligonucleotide-free aCSF, Cx43 expression (immunofluorescence analysis) was similar to that seen in the missense-treated animals (not shown). On the other hand, analysis of adjacent sections for expression of Cx40 or Cx37 revealed no obvious differences when comparing antisense- versus missense-treated rats (Fig. 10).

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Influence of 24-h exposure to connexin43 (Cx43) antisense or missense oligonucleotides on ADP-induced pial arteriolar dilations in intact (A) and Ovx (B) females. Values are means ± SE; n = 3-4 rats for each treatment. * P < 0.05 vs. missense.

View larger version (28K): [in this window] [in a new window]   Fig. 9.   Western immunoblot detection of Cx43 in pial tissue samples from antisense- and missense-treated Ovx rats and in a rat brain lysate (lane 1, supplied by the vendor for use as a positive control). Multiple bands seen in the pial tissue specimens (lanes 2 and 3) likely reflect phosphorylated (P) and unphosphorylated (NP) Cx43 (17).

View larger version (96K): [in this window] [in a new window]   Fig. 10.   Connexin expression in tissue under the cranial windows in Cx43 antisense (right) and missense (left)-treated Ovx females. Pial vessels are depicted by arrows. For Cx43 (top), a lesser immunofluorescence intensity in pial vessels, cortical surface tissue, and the overlying pial and arachnoid membranes are evident in the antisense-treated rats. No differences in Cx40 (middle) or Cx37 (bottom) immunofluorescence were evident when comparing Cx43 antisense- and missense-treated rats.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Results obtained in a recent report from our laboratory (37) indicated that, whereas the magnitude of the pial arteriolar vasodilating response to ADP was not diminished at 4-6 wk following ovariectomy, the NO dependency of that response disappeared and was replaced by an EDHF-like mechanism. The present study represents an extension of that work. Thus we found that not only is ADP reactivity maintained in Ovx females, but the endothelium dependency of the ADP response is retained as well. Moreover, the switchover from NO to a presumed EDHF dependency in Ovx females, to a large degree, appears to involve gap junctions and, at the least, Cx 43.

Although the well-documented property of halothane to act as a gap junctional blocker might be viewed as a potential complicating factor in the present study, we are confident that this concern is unwarranted. Thus Burt and Spray (3) reported that halothane reduced gap junctional conductance only at concentrations exceeding 1 mM. That concentration of halothane in aqueous, biological fluids is achieved at an inhaled halothane fraction of nearly 3% of the gas mixture (see Ref. 20). In fact, in most studies employing halothane as a gap junctional blocker, halothane concentrations in the 2-3.5 mM range and even higher were needed for effective blockade. This includes the brain (see Ref. 29) as well as vascular tissue (e.g., Ref. 7). Because we used 0.8% halothane inhalation (in N₂O) for surgical preparation [yields ~0.25 mM halothane in body fluids (20)], the highest level of halothane achieved was not even close to the level associated with a documented loss of gap junctional communication. Moreover, halothane clearance from the brain occurs rather quickly when halothane inhalation is discontinued. Thus, by the time the first ADP reactivity measurement was made (~90 min posthalothane), it is even less likely that halothane had any influence on gap junctions and the results we observed.

In our earlier publication, the presumption of an EDHF contribution to ADP-induced pial arteriolar dilation was based on the observation that the response in Ovx females was not affected by combined NOS-cyclooxygenase inhibition but was substantially attenuated in the presence of Ca²⁺-activated K⁺ (K_Ca) channel blockade. Indeed, the hallmarks of an EDHF are that 1) it is not NO or a prostanoid, 2) it acts via K_Ca channels, 3) it requires an intact endothelium, and 4) it hyperpolarizes vascular smooth muscle cells. Because of the difficulties involved in measuring vascular smooth muscle membrane potentials in vivo in intact animals, the first three identifiers are commonly used to demonstrate EDHF participation. Thus the first two criteria were satisfied in our previous study (37), whereas the third criterion was demonstrated in the current investigation. That is, we applied the well-established L/D endothelial injury model (34) and confirmed its effectiveness by showing that the duration of L/D exposure used eliminated pial arteriolar responses to an established endothelium (and endothelial NOS)-dependent vasodilator (ACh) while not affecting responses to vasodilating agents or conditions that act directly on vascular smooth muscle. It is of some interest to note that, in the intact females, endothelial injury resulted in a similar magnitude reduction in ADP reactivity as occurred in the presence of L-NNA + Indo. Moreover, combined application of L/D and L-NNA + Indo produced no further diminution in the ADP response over that seen with either intervention alone. Because Indo does not affect ADP-induced pial arteriolar dilations (37), those results add to the evidence obtained in our recent study (37) indicating that the NO which contributes to the ADP response in intact females is endothelial NOS derived.

Whereas the present and previous studies have suggested that EDHF function is enhanced in cerebral vessels of Ovx females and that this alteration relates to chronic estrogen depletion (15, 37), neither the mechanisms involved nor the identity of the EDHF has been established. Candidate EDHFs include diffusible factors such as products of arachidonate metabolism (epoxides and anandamide) or carbon monoxide (9, 26, 35). The EDHF may also relate to a process whereby current passage between endothelium and vascular smooth muscle is facilitated. That process could involve K ions (9) and/or the formation of specialized sites of intercellular communication. With respect to the latter, gap junctions between endothelial and smooth muscle cells or between cells within the endothelial and smooth muscle layers have been proposed as playing a key role in EDHF-related vasodilation (30). Indeed, that process may even involve diffusible factors that utilize gap junctions to facilitate their actions (13).

Results to date imply that the EDHF sometimes manifests itself under circumstances where the NO-generating function is diminished (37). It has also been suggested that this relates to NO acting to inhibit EDHF production or function (1, 27). Because cerebrovascular endothelial NOS expression and activity are repressed in the Ovx female (28, 37), it is tempting to ascribe the nascent EDHF dependency in the pial arteriolar response to ADP in Ovx rats to such a mechanism. One potential target for NO inhibition of EDHF generation is cytochrome P-450 epoxygenase, the enzyme responsible for arachidonate conversion to K_Ca channel-activating epoxides (26). However, the likelihood of an increased epoxygenase activity (linked to diminished NO presence) playing any role in Ovx females was essentially obviated by recent findings showing no effect of an epoxygenase inhibitor on the pial arteriolar dilations elicited by ADP (37). Yet, these results did not eliminate epoxides from consideration altogether because it is possible that epoxides might be mobilized from preformed pools rather than via de novo synthesis (14). Nevertheless, irrespective of the EDHF source, there is no evidence to indicate that NO in any way affects EDHF function in cerebral vessels. This is particularly emphasized by the results of a recent study showing that manipulating NO levels did not alter agonist-induced EDHF-mediated vasodilating responses in the rat middle cerebral artery (31).

Although attempts to identify the chemical nature of the EDHF have met with limited success, there is evidence to suggest that this mediator may affect relaxation via its transcellular diffusion through gap junctions between endothelial cells and/or vascular smooth muscle cells (6, 25). Results from the present study strongly suggest that gap junctions play a key role in the apparent nascent EDHF dependency of ADP-induced pial arteriolar relaxation in Ovx females. That is, application of the selective inhibitor of gap junctional communication Gap 27 had no effect on ADP-induced vasorelaxation in intact female rats but attenuated that response in Ovx females. In addition, the magnitude of the reduction in pial arteriolar reactivity to ADP was remarkably similar to the reduction seen in the presence of endothelial injury or, as shown previously (37), K_Ca channel blockade. These findings would, therefore, seem to imply that endothelium, K_Ca channel, and gap junctional dependency are related characteristics of a key EDHF-like mechanism supporting ADP-induced relaxation of pial arterioles in Ovx, chronically estrogen-deprived females.

Whereas current findings provide strong evidence favoring the appearance in Ovx females of an endothelium-associated gap junctional participation in the pial arteriolar response to ADP, no information was obtained that provides any clues as to the specific mechanisms involved in this "transformation." A number of publications have addressed the issue of sex steroid influence on gap junction protein expression and gap junctional communication. However, no consensus can be derived from those studies. Thus Hortovanyi et al. (19) reported no differences in smooth muscle Cx43 expression in aortic tissue harvested from Ovx rats with or without chronic 17-estradiol (E₂) supplementation. On the other hand, in uterine smooth muscle, estrogen has been reported to positively influence Cx43 expression (8, 21), although that estrogen effect may be counteracted by progesterone (16). In the brain, Ovx rats treated with E₂ exhibited regionally selective higher neuronal Cx43 mRNA levels compared with untreated Ovx females (32). The experimental findings that are perhaps the most intriguing, with respect to the data of the present study, were published by Herve et al. (18). These authors reported that E₂, via physically inserting itself into the plasma membrane, was capable of disrupting gap junctional communication. Such a process could provide an explanation for the advent of a gap junctional dependence of the ADP response in estrogen-depleted states. That is, Ovx females may experience a disinhibition of gap junctional function. This rather interesting possibility should be addressed in future studies.

Because both Cx43 and Cx37, as well as Cx40, are reportedly expressed in cerebral vessels (22-24, 38), we cannot ascertain from the Gap 27 data alone whether Cx43 (or Cx37) is the target of the Gap 27 action. That is because the Gap 27 sequence is found in the second extracellular loops of Cx43 and Cx37 (5). To examine whether Cx37, as well as Cx40 [which is known to be expressed in rat pial vessels (24)], may be involved, we applied the Gap 26 peptide, which targets a sequence found in the first extracellular loops of Cx40 and Cx37 (5). The finding that Gap 27, but not Gap 26, attenuated the ADP response in Ovx females strongly suggests that Cx43, but not Cx37 or Cx40, participates in ADP-induced dilations in these estrogen-depleted animals. To further substantiate that Cx43 played an important role in the ADP-induced dilation of pial arterioles in Ovx females, we applied an antisense oligonucleotide strategy already established in our laboratory (see Ref. 36). That topical application approach is designed to affect pial vascular protein expression, in particular (see Figs. 9 and 10). Thus we found that antisense administration was associated with a similar magnitude reduction in ADP reactivity in Ovx females as that seen in the presence of Gap 27. Furthermore, in the intact female in the presence of the antisense oligonucleotide (and identical to the results seen with Gap 27), no changes in ADP reactivity were observed. In summary, these findings, at the least, indicate that Cx43 plays a much greater role than Cx37 or Cx40 in the nascent EDHF dependency of ADP-induced arteriolar relaxation in the brain. Whether other connexins, like Cx45, play a similar role must await the results of additional experimentation.

In conclusion, the findings of the current study provided compelling evidence that gap junctional communication is involved in the switch from NO to EDHF dependency in the pial arteriolar response to ADP in Ovx rats. Furthermore, Cx43 was found to play a key role in this response. However, the mechanism(s) through which chronic alterations in sex steroids (estrogen in particular) act to yield such a change will require further study.