The Gatekeepers in the Mouse Ophthalmic Artery: Endothelium-Dependent Mechanisms of Cholinergic Vasodilation

Cholinergic regulation of arterial luminal diameter involves intricate network of intercellular communication between the endothelial and smooth muscle cells that is highly dependent on the molecular mediators released by the endothelium. Albeit the well-recognized contribution of nitric oxide (NO) towards vasodilation, the identity of compensatory mechanisms that maintain vasomotor tone when NO synthesis is deranged remain largely unknown in the ophthalmic artery. This is the first study to identify the vasodilatory signalling mechanisms of the ophthalmic artery employing wild type mice. Acetylcholine (ACh)-induced vasodilation was only partially attenuated when NO synthesis was inhibited. Intriguingly, the combined blocking of cytochrome P450 oxygenase (CYP450) and lipoxygenase (LOX), as well as CYP450 and gap junctions, abolished vasodilation; demonstrating that the key compensatory mechanisms comprise arachidonic acid metabolites which, work in concert with gap junctions for downstream signal transmission. Furthermore, the voltage-gated potassium ion channel, Kv1.6, was functionally relevant in mediating vasodilation. Its localization was found exclusively in the smooth muscle. In conclusion, ACh-induced vasodilation of mouse ophthalmic artery is mediated in part by NO and predominantly via arachidonic acid metabolites, with active involvement of gap junctions. Particularly, the Kv1.6 channel represents an attractive therapeutic target in ophthalmopathologies when NO synthesis is compromised.


Role of NO, and PGI 2 in ACh-induced vasodilation.
Cumulative administration of ACh (10 −9 -10 −4 M) evoked concentration-dependent vasodilatory responses (76.44 ± 9.45%) that were markedly attenuated (49.18 ± 10.48%, P < 0.01) following incubation with the non-isoform-selective NOS inhibitor, L-NAME (10 −4 M) ( Fig. 2A). To test whether the NO receptor, sGC was involved in vasodilation, the ophthalmic arteries were stimulated with ACh (10 −9 -10 −4 M) before and after addition of the sGC inhibitor, ODQ (10 −5 M). Responses to acetylcholine were markedly reduced (ACh reference: 83.61 ± 8.67% vs ODQ: 48.92 ± 8.71%, P < 0.001) after ODQ treatment (Fig. 2B), indicative of NO involvement. Conversely, responses to acetylcholine in the ophthalmic arteries cannot be ascribed to PGI 2 because exposure of the arteries to the non-isoform-selective COX inhibitor, indomethacin (10 −5 M), did not significantly affect the vasodilation (ACh reference: 76.63 ± 5.93% vs indomethacin: 69.60 ± 9.84%), as shown in Supplementary Fig. S1. Additionally, combined incubation with L-NAME and indomethacin (48.91 ± 9.01%) did not alter the vasodilatory response compared to inhibition with L-NAME alone (49.18 ± 10.48%) ( Fig. 2A,C), suggesting that COX metabolites did not contribute to the cholinergic vasodilation in the mouse ophthalmic artery. However, the residual dilatory response observed after the blocking of both NOS and COX with L-NAME and indomethacin, respectively, was abolished by the addition of 30 mM K + solution (potassium chloride, KCl), as depicted in Fig. 2C. At this concentration, KCl acts as a partial depolarizing agent and antagonizes the action of EDHF. Therefore, this finding suggests the important involvement of EDHF in ACh-mediated vasodilation in this vascular bed. Vasodilatory responses to ACh were markedly attenuated in the endothelium-denuded vessels, whereas vasodilation responses to SNP were retained in both endothelium-denuded and -intact vessels. Values are expressed as mean ± standard error of the mean (s.e.m) (n = 6 per group; ***P < 0.0001, endothelium-denuded versus endothelium-intact).
Scientific RepoRts | 6:20322 | DOI: 10.1038/srep20322 Contribution of EDHF-mediated vasodilator responses to ACh. To assess the EDHF mechanisms involved in mediating ACh-induced vasodilation in the mouse OA, various pharmacological agents were employed to inhibit different factors implicated as the putative EDHF. Acetylcholine was previously reported to induce generation of vasoactive amounts of H 2 O 2 both NOS-dependently and -independently 27,28 . To evaluate the contribution of H 2 O 2 to endothelium-dependent dilatation, responses to ACh before and after incubation with catalase (1000 units/ml) were tested. Catalase, when applied either alone or in combination with both L-NAME and indomethacin, evoked negligible inhibitory effect on vasodilation responses elicited by ACh ( Supplementary Fig. S2).

Effect of potassium ion channel blockers on endothelium-dependent vasodilation.
To further characterize the EDHF-related dilatation, specifically the contribution fostered by K Ca channels, all three channel subtypes, SK Ca , IK Ca and BK Ca , were inhibited with combination of Apa and ChTX. Marked inhibition of cholinergic responses was observed with this combination blocking (L-NAME and indomethacin: 56.96 ± 7.40% vs Apa and ChTX: 4.07 ± 1.65%, P < 0.0001), as shown in Fig. 4A. To further validate this finding, each K Ca channel subtype was blocked in combination with their respective, highly specific inhibitors consisting of Apa for SK Ca , TRAM-34 for IK Ca and IbTX for BK Ca. Remarkably, this combination blocking demonstrated that the vasodilator The non-subtype-selective NOS inhibitor, L-NAME (10 −4 M, n = 5) partially attenuated vasodilation to ACh (**P < 0.01, ACh reference vs L-NAME). (B) The sGC inhibitor, ODQ (10 −4 M, n = 6) evoked partial yet significant attenuation of the dilatory responses to ACh (***P < 0.001, ACh reference vs ODQ). (C) The residual dilatory responses in the presence of both L-NAME and indomethacin were abolished by 30 mM of potassium (K + ) solution (***P < 0.001, ACh reference vs L-NAME + Indomethacin; ***P < 0.001, L-NAME + Indomethacin vs L-NAME + Indomethacin + KCl). Values are expressed as mean ± s.e.m. response to ACh remained unchanged (L-NAME and indomethacin: 63.26 ± 9.07% vs Apa and TRAM-34 and IbTX: 66.06 ± 11.34%, P > 0.05) (Fig. 4B). Each of these three K Ca channels was inhibited individually with Apa, TRAM-34, IbTX and ChTX and our findings showed that only the inhibition with ChTX displayed significant attenuation of the vasodilation ( Supplementary Fig. S3), while the other blockers comprising of Apa, TRAM-34 and IbTX did not contribute to significant blunting of vasodilation in the OA (Supplementary Fig. S4-S6, respectively). Based on these results, we suspected that ChTX, which had also been reported to block several K v channel subtypes, blocked one or more of the K v channel subunits in the mouse ophthalmic artery.
Hence, to confirm this hypothesis, a number of agents that block K v channels were employed to assess the nature of the K v channel involved in mediating ACh-elicited vasodilation. In order to determine which of these ChTX-sensitive K v channel subtype(s) is particularly involved in mediating the ACh-induced relaxations, vessels were incubated with MgTX, a selective blocker of the K v 1.3 and K v 1.6 channels in the presence of NOS and COX inhibitors. Incubation with MgTX completely blunted ACh-induced dilations (L-NAME and indomethacin: 64.67 ± 4.93% vs MgTX: 1.25 ± 1.25%, P < 0.001) (Fig. 5A), underscoring the probable role(s) of ChTX-and MgTX-sensitive K v 1.3 and K v 1.6 channels in the mouse ophthalmic arteries. Following incubation with MTX, psora-4 and β -DTX which, specifically blocks the K v 1.2, K v 1.3 and combinations of K v 1.1 and K v 1.2 channels, respectively, the ACh -induced vasodilatation remained unaffected ((Supplementary Figs S7-9) In contrast, incubation with α -DTX, a potent blocker of K v 1.1, K v 1.2 and K v 1.6 channels evoked complete attenuation of the vasodilatory responses to ACh, as demonstrated in Fig. 5B (L-NAME and indomethacin: 92.73 ± 3.48% vs α -DTX: 4.96 ± 2.34%, P < 0.001). Taken together, the results of the K v channel inhibitions is attributed to the involvement of the K v 1.6 channel that is sensitive to the blocking effects of ChTX, MgTX as well as α -DTX.
Next, to evaluate the potential functional relevance of other potassium channels, arteries were treated with blockers of Kir and K ATP channels, and Na + /K + -ATPase. Neither glibenclamide nor ouabain had any significant effect on the ACh-mediated vasodilation ((Supplementary Figs S10-11)), indicating the null involvement of the K ATP channel and Na + /K + -ATPase, respectively. Conversely, Fig. 6 shows that BaCl 2 , a blocker of Kir channels, caused 38.07% inhibition (P < 0.001) of the ACh-elicited vasodilation in the presence of L-NAME and indomethacin (L-NAME and indomethacin: 48.64 ± 8.35%).
Localization of the K v 1.6 channel in the mouse ophthalmic artery. The K v 1.6 channel plays a central role in the regulation of the ophthalmic blood flow as demonstrated by the functional experiments in the present study. In order to determine the localization of this K v channel subtype in the ophthalmic artery, immunostaining was carried out on the sagittal cryosections of ophthalmic artery. Localization of K v 1.6 was particularly restricted to the vascular smooth muscle cell layer but no expression was observed in the endothelial cells, as shown in Fig. 7A. The negative control of the same tissue in which the primary antibody was omitted, was not stained (Fig. 7B).

Discussion
This is the first functional study reporting on the EDHF mechanisms mediating agonist-induced vasodilator response in the mouse ophthalmic artery. There are several key findings, including some novel aspects, emerging from the current study. First, in addition to the well-established observations of the role of endothelium in various vascular beds and species, we endeavoured to investigate the role of endothelium in vasodilator response to ACh  Values are expressed as mean ± s.e.m [n = 5-6 per group; ***P < 0.001, L-NAME and Indomethacin vs L-NAME and Indomethacin and blocker(s)]. Absence of error bar indicates that the SEM was less than the size of the symbol. particularly in the mouse ophthalmic artery. The use of ACh instead of other agonists e.g. bradykinin is advocated in this study to circumvent desensitization of the endothelial receptors due to tachyphylaxis 10 . Moreover, our previous study has clearly demonstrated that in the murine ophthalmic artery, endothelium-dependent vasodilator responses were mediated by the M 3 muscarinic ACh receptor 25,26 . Mechanical denudation of endothelium abolished ACh-induced vasorelaxation, thereby demonstrating that the vascular endothelium plays an obligatory role in the cholinergic vasodilation of mouse ophthalmic artery. Our results also showed that the endothelium-dependent responses were partially mediated by a NOS-and sGC-dependent mechanism, supporting the involvement of NO. These findings are in contrast to the vasodilatory mechanism in the human ophthalmic artery, where the ACh-induced dilation is mediated predominantly by the NO pathway 12 . Conversely, the involvement of PGI 2 was discounted in the mouse ophthalmic artery because indomethacin exerted no inhibitory effect on the dilatory responses to ACh. These results imply that prostanoid-dependent signalling pathway do not account for the ACh-mediated vasodilatory response in the mouse ophthalmic artery.
Secondly, the predominant involvement of EDHF accounts for the residual dilatory response observed in the mouse ophthalmic artery in the presence of both L-NAME and indomethacin, whereby the abolishment of dilation by concomitant addition of depolarizing concentration of potassium solution was observed 29 . It is widely recognized that the EDHF phenomenon evokes vasodilatation in the presence of COX and NOS inhibitors 30,31 and its physiologic influence is deemed more prominent as the vessel diameter decreases. Since the smaller vessels have fundamental roles in vascular resistance, EDHF is suggested to be of major importance in the blood flow control in these vessels 32,33 . Consistent with this possibility, the involvement of these different factors implicated as EDHFs was tested and we found that endothelium-dependent vasodilation in the ophthalmic vasculature was mediated in part by CYP450 and predominantly by LOX metabolites, with a major involvement of the gap junctions. While the individual blockade of CYP450 and LOX only partially reduced vasodilation responses, combined blockade of CYP450 and LOX virtually abolished vasodilation suggesting that metabolites of both enzymes almost exclusively contribute to the EDHF-mediated responses in this vascular bed. The CYP450 pathway appears not to be completely dependent on the gap junctions because combined blockade of both CYP450 and gap junctions only resulted in additive response to the individual inhibitions. It is becoming increasingly well recognized that the arachidonic acid metabolites generated via the CYP450 pathway, most likely the four epoxyecosatrienoic  acid regioisomers (EETs), have been implicated in the augmentation of gap junctional communications and to regulate active communications between endothelial cells [34][35][36] . This is especially relevant because EETs are highly lipophilic transferable factors that cannot pass through the gap junctions, which comprise of aqueous pores, but rather may act as modulators to hyperpolarize the VSMC via the gap junctions [37][38][39] .
In contrast to the CYP450-mediated signalling, the LOX signalling mechanism(s) seemed to be highly dependent on the gap junctions, through which the downstream signals and/or molecule(s) that dilate the VSMC are transmitted, since combined blockade of LOX and gap junctions did not result in any further attenuation of the response. However, it should be remarked that the molecular weights of LOX-derived metabolites, namely 15-hydroxy-11, 12-epoxyeicosatrienoic acid (HEETA) and 11, 12, 15-trihydroxyeicosatrienoic acid (THETA,) are large and since the aqueous central pore of the gap junctions can only permit the passage of molecules < 1 kDa, it is unlikely that LOX and/or its metabolites are transferable via this channel to hyperpolarize the VSMC [40][41][42] . However, a plausible explanation can be that arachidonic acid metabolites generated via the LOX pathway may act as autocrine or intracellular modulators of gap junctions, as was previously proposed in the rat middle cerebral artery and rabbit arteries, where LOX metabolites directly stimulated the SK Ca channels, instead of the gap junctions as observed in our study, to hyperpolarize the VSMC [43][44][45] . The murine LOX share a highly conserved sequence similarity with the human's based on the phylogenetic classification and they belong to the same 12/15-LOX subfamily 46 . Intriguingly, 12/15-LOX was found to be associated with key regulation roles in pathologies of the central nervous system such as Parkinson's disease and Alzheimer's 47,48 . Looking at the pivotal roles of the LOX-derived metabolites in human pathologies and the high similarity between both mouse and human, these findings broadens the use of murine models for further in-depth investigations of the molecular mechanisms of LOX-related pathway in the next studies.
Despite the rapid progress made in the past decade in elucidating the physiological roles of the arachidonic acid metabolites in various biological systems, many important questions still remain unanswered. For example, in our study, the existence of putative receptor(s) of the downstream signalling cascade of CYP450, especially for the EETs, require further investigation to extend our current hypothesis beyond the present findings. Likewise, studies involving the CYP450 metabolites in cardioprotection are also seeking to identify the precise molecular receptor(s) target(s) of EETs for potential development of new therapeutic strategies 39,49 . Additionally, it is important to define the precise identity of the arachidonic acid metabolites generated via the CYP450 and LOX mechanisms responsible for the observed vasodilatory phenomenon, as emphasized by Thollon et al. 50 .
Thirdly, our data strongly suggest the important involvement of the Kir and K v 1.6 channels in mediating endothelium-dependent dilation to ACh. Previous studies have shown that K + released from the endothelium can act as an EDHF by activating K Ca and stimulating Na + /K + -ATPase and Kir channel in guinea pig choroidal arterioles and rat hepatic arteries, respectively 51,52 . Therefore, we examined the possible role of potassium channels in endothelium-dependent vasodilation of the mouse ophthalmic artery. Consistent with the finding that K ATP channels are usually not involved in EDHF-mediated vasodilation 53 , our results indicated that the inhibitory effect of glibenclamide on K ATP channel had negligible influence on the vasodilatation of mouse ophthalmic artery induced by ACh. The application of ouabain also failed to inhibit dilation, suggesting the lack of Na + / K + -ATPase involvement in mediating responses to ACh. However, it is of interest that the blockade of Kir channels caused significant attenuation of vasodilation in the ophthalmic artery. The Kir, channel localized on the SMCs, is one the major targets of external K + ions, which activate the channel conductance to lower intracellular Ca 2+ and leads to vasodilatation 54,55 .
Accumulating evidences imply that the action of EDHF is generally inhibited by combination blockade of the SK Ca with Apa and, IK Ca and BK Ca with ChTX 32,56 . Correspondingly, our study demonstrated that the combined inhibition with these blockers virtually abolished cholinergic responses in the mouse ophthalmic artery. However, it is important to highlight here that the combined blockade of all three K Ca channels with their respective specific blockers, and not ChTX, had no significant effect on the ACh-mediated vasodilatory responses in the current study as hypothesized. Of note, an interesting phenomenon was observed in mice where the expression of IK Ca and SK Ca in the endothelial cells was relatively low as the size of the artery decreased 57 which, was in sharp contrast to the increased expression of both channel subtypes in the rat artery as the vessel size decreased 58,59 . On the basis of our results, these observations support the hypothesis of our study that the expression of the K Ca channels in the mouse ophthalmic artery may be low or null and are unlikely to account for the attenuation of the vasodilation when blocked with Apa and ChTX. Taken together, this confounding finding can be extrapolated to the multi-channel blocking properties of ChTX, which not only blocks the IK Ca and BK Ca channels, but also inhibits the Shaker-related voltage-gated K + channels K v 1.1, 1.2, 1.3, and 1.6 with high affinity 60,61 .
An ongoing, unresolved restriction to study the post-receptor mechanisms is the use of most characterized pharmacological blockers and inhibitors with unspecific nature that may be affecting another alternative EDHF signalling cascade with similar affinity, as demonstrated in the current study. Therefore, to confirm the involvement of the K v channels and in particular to dissect which of these is/are involved in the vasodilation of the mouse ophthalmic artery, several highly specific K v channel inhibitors were employed. Complete attenuation in vasodilation was observed in the presence of MgTX and α -DTX. MgTX is widely used as a potent inhibitor of the K v 1.3 in ion channel investigations 62 . However, a recent study by Bartok et al. provided critical evidence that MgTX is not a highly specific K v 1.3 inhibitor as had been assumed in many previous studies 63 and this toxin has also been shown to inhibit other K v channels, namely K v 1.1, 1.2 and 1.6, with high potency 61,64-67 . Our results support a possible participation of other channel subtype(s) considering the potential overlap in blocking selectivity exhibited by ChTX and MgTX, as summarized in the Venn diagram (Fig. 8). Therefore, with the use of several other toxins, this study unravelled that the K v 1.6 channel is functionally relevant in mediating vasodilatory responses. Additionally, immunostaining confirmed the localization of this voltage-gated channel subtype in the VSMC of the mouse ophthalmic artery. The novelty of the present investigation lies in the identification of the K v 1.6 channel in mediating vasodilation that has never been reported hitherto. It is well recognized that the altered K v 1.6 channel expression is associated with neurodegenerative diseases such as Amyotrophic Lateral Sclerosis (ALS) that affects the duration of action potential of motor neurons 68 . On the other hand, study by Carrisoza-Gaytan et al. emphasized the importance of the K v 1.6 channel in K + reabsorption in the thick ascending limb of the rat nephron 69 . This channel is also implicated in the pulmonary artery smooth muscle cells as one of the crucial hypoxia-sensitive K v channels that regulate membrane potential and intracellular Ca 2+ homeostasis during hypoxia [70][71][72] . As our understanding of the K v channels continues to evolve, it is therefore tempting to conjecture that the specific identification of the K v 1.6 channel in this study may represent an innovative molecular target in the ophthalmic circulation to enhance vasodilation in conditions of channelopathy, albeit the exact function of this channel subtype in the ophthalmic circulation warrants further investigation. Our current investigation provides a plausible hypothesis as to how ACh-induced vasodilatation may occur in the mouse ophthalmic artery and based on our results, the hypothesized signalling pathways involved in the vasodilator mechanisms are as depicted in Fig. 9.
In conclusion, the hallmark of this study was the identification of the major signalling cascades that mediate endothelium-dependent vasodilation in the mouse ophthalmic artery which, were previously uncharacterized. Although the findings emerging from this experimental study do not fully account for the precise molecular mechanisms underlying the observed vasodilation in vitro, the current elucidation of EDHF mechanisms in mouse ophthalmic artery assigns a pivotal platform for the use of mice to further explore the functional relevance of specific CYP450 and LOX metabolites in mediating ACh-induced vasodilation, as well as the existence of a potential putative, 'novel' receptor on the endothelial cells that mediates the efflux of K + remains to be determined in this vascular bed. This study also addressed the potential therapeutic target(s) for future translational applications in human ocular diseases. It will be interesting to determine whether the contribution of the specific potassium ion channels outlined here in the mice ophthalmic artery could also play similar roles in the human ophthalmic circulation, particularly in pathological conditions when NO synthesis is impaired.

Materials and Methods
Experimental animals. This study was approved by the Animal Care Committee of Rhineland-Palatinate, Germany, and animal care conformed to the institutional guidelines and 'The Association for Research in Vision and Ophthalmology' (ARVO) statement for the use of animals in ophthalmic and vision research. Mice were treated according to the EU Directive 2010/63/EU for animal experiments. Male C57BL/6J mice (The Jackson Laboratory, Bar Harbour, ME, USA) aged 3 to 7 months old were used for the experiments. Animals were housed under standard conditions (temperature 23 ± 2 °C, humidity range 55 ± 10% and 12 h light/dark cycles), and had access to standard mouse chow and water ad libitum.
Drugs. The following drugs were used in this experiment: N ω -nitro L-arginine methyl ester (L-NAME), indomethacin, acetylcholine hydrochloride (ACh), phenylephrine, 1H-(1, 2, 4) oxadiazole (4, 3-alpha) quinoxaline-1-one (ODQ), catalase, baicalein 18 alpha-glycyrrhetinic acid (18α -GA), ouabain, glibenclamide, barium chloride (BaCl 2 ), and psora-4 [5-(4-Phenylbutoxy)psoralen] (all purchased from Sigma-Aldrich Chemie  Following interaction with CaM (calmodulin), Ca 2+ activates eNOS and release of NO. NO causes relaxation by interacting with the haem group of the enzyme, sGC, which then mediates the formation of cyclic guanosine monophosphate (cGMP) and activation of protein kinase G (PKG) that relaxes the VSMC. The second vasodilator mechanism is via the arachidonic acid (AA) metabolites synthesized through the CYP450 oxygenase pathway: The increase in [Ca 2+ ] i elicits translocation of phospholipase A2 (PLA 2 ) to the membrane and its major hydrolysis product is AA, which can be metabolized by CYP450 oxygenase to EETs. The EETs function as messenger molecules that modulate gap junctions (GJ) to spread the conductance to the VSMC. EETs may also directly activate a channel/receptor on the VSMC to hyperpolarize and dilate the vessel. The third signalling pathway involves the AA metabolites generated via the LOX pathway: It is hypothesized that LOX and/or its metabolites, namely THETA and HEETA do not pass the GJ as EDHF per se but activate GJ to hyperpolarize the VSMC. The forth key players are the gap junctions. The fifth proposed mechanism involves the active participation of the Kir and K v 1.6 channels: It is hypothesized that both Kir and K v 1.6 channels on the VSMC are activated by the increase in extracellular K + resulting in hyperpolarization and vasodilation. The precise identity of the putative channel(s) on the endothelial cells that is activated and opened for K + efflux for hyperpolarization to occur is unknown. Question marks represent unknown receptors that are yet to be identified. Blockers and inhibitors are indicated in red boxes. Green arrows indicate the activation of the gap junctions and downstream receptor(s). Blue solid arrows show potential pathways for transfer of hyperpolarization from the endothelium to the smooth muscle cells. Blue quadrangular point arrow indicates the hypothesized transfer of hyperpolarization via an unknown receptor on the VSMC.
2.5 mM CaCl 2 , 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 25 mM NaHCO 3 , 11 mM glucose (Carl Roth GmbH, Karlsruhe, Germany). The ophthalmic arteries were carefully isolated and cleaned of surrounding connective tissues using fine-point tweezers under a dissecting microscope. Arterial segments were then placed in an organ bath with ice-cold Krebs-Henseleit buffer, cannulated onto two glass micropipettes and secured with 10-0 nylon monofilament suture. Vessels were pressurized via these micropipettes to 50 mm Hg under no-flow conditions using two reservoirs filled with Krebs-Henseleit buffer. The ophthalmic artery was equilibrated for 30-40 minutes before the commencement of the experiments. During this equilibration period, the vessels developed a stable spontaneous myogenic tone by constricting to ~86 to 81% of the initial arterial luminal diameter measured immediately after pressurization to 50 mmHg, as described elsewhere 24 . Video sequences were captured to a personal computer using a video camera mounted on an inverted microscope for off-line analysis. The organ bath was continuously circulated with Krebs solution maintained at 37 °C and pH 7.4 and, aerated with 95% O 2 and 5% CO 2 . A minimum 50% vasoconstriction from the resting diameter in response to 100 mM K + solution was used as a criterion to assess vessel viability 26 .
In some experiments, the endothelium was mechanically removed by rubbing the luminal surface of the arteries with a human hair, as described previously 26 . Next, arteries were preconstricted to 70-50% of the initial vessel diameter by titrating the α 1 -adrenoceptor agonist phenylephrine, and concentration-response curves to ACh (10 −9 -10 −4 M) was obtained by cumulative application of ACh to the circulating bath solution. The pre-treatment of the arteries with L-NAME slightly constricted the vessels and in this circumstance, the phenylephrine concentration was adjusted to reach a similar preconstriction level in all experiments. All reported drug concentrations refer to final molar concentrations in the organ bath.

Experimental Protocols. Protocol 1: The role of endothelium in acetylcholine-induced vasodilation. To test
whether ACh-induced responses were completely endothelium-dependent in preconstricted ophthalmic arteries of the C57BL/6J genotype, endothelium-intact and endothelium-denuded arteries were stimulated with ACh (10 −4 M) and with the endothelium-independent NO donor, sodium nitroprusside (SNP, 10 −4 M) to ensure that smooth muscle reactivity was not affected by endothelium removal 74 .

Protocol 2: Contribution of NO and cyclooxygenase (COX) metabolites to ACh-induced vasodilation.
To assess the role of NO and prostanoids in mediating ophthalmic artery vasodilation, responses of arteries to cumulative application of ACh (10 −9 -10 −4 M) were tested before and after incubation (30 min) with the non-isoform selective NOS inhibitor, L-NAME (10 −4 M) or COX inhibitor, indomethacin (10 −5 M). Similarly, responses of arteries to cumulative application of ACh were tested before and after treatment (30 min) with soluble guanylate cyclase (sGC) inhibitor, ODQ (10 −5 M). Arteries were preconstricted with phenylephrine after the incubation with blockers. Protocol 4: Contribution of calcium-activated potassium channels (K Ca ) and voltage-gated potassium channels (K v ) to ACh-induced vasodilation. To characterize the K Ca that mediate ACh-induced dilator reactivity, ophthalmic arteries were pre-treated with the combination of following agents: Apamin (10 −7 M), a specific blocker of the small conductance K Ca (SK Ca ) and ChTX (10 −7 M), an inhibitor of both intermediate conductance K Ca (IK Ca ) and big conductance K Ca (BK Ca ). Due to the limited specificity of ChTX, which also blocks some of the voltage-gated channels 34 , highly selective K Ca blocker combinations were employed, as follows: IbTX (10 −7 M), a selective BK Ca blocker (Maxi K Ca ) and TRAM-34 (10 −6 M), a specific blocker of the IK Ca 60 . The role of specific Shaker-related type 1 K v channels was evaluated employing blockers with varying sensitivity and specificity for the different K v channels: MgTX (10 −8 M), α -and β -DTX (5 × 10 −8 M) and MTX (5 × 10 −8 M).
Statistical analysis. Data are expressed as mean ± SEM, with n representing the number of animals per group. Changes in vascular responses to various reagents tested are presented as percentage of diameter change from the initial precontraction levels or the percent vasodilator responses as compared to maximal vasodilator response induced by ACh. Statistical comparisons of concentration-response curves were made using the two-way ANOVA for repeated measures followed by Bonferroni post-hoc test. Unpaired two-tailed t-test was used for single-dose responses. The level of significance α was set at 0.05. Graph Pad Prism 6 software (GraphPad Inc., San Diego, USA) was used for statistical analyses. Immunohistochemistry. To determine the localization of the K v 1.6 channel in the mouse ophthalmic artery, segments of the ophthalmic artery were subjected to immunohistochemistry. The blood vessels were carefully isolated, rinsed in cold Krebs-Henseleit buffer and cryopreserved in Tissue-TEK OCT media (Sakura FineTek Europe, Alphen aan den Rijn, Netherlands) and immediately frozen at − 20 °C in a freezer. Transverse cryosections of the arterial rings (8 μ m thick) were thaw mounted onto Superfrost Plus slides (Thermo Scientific, Gerhard Menzel GmbH, Braunschweig, Germany), air-dried and stored at − 20 °C until use. Prior to immunolabelling, the sections were fixed in 4% paraformaldehyde for 20 min, followed by permeabilization in PBS (0.05 M Na 2 HPO 4 , 0.14 M NaCl, pH 7.40) containing 0.1% Triton X-100 (TX). Sections were then blocked with PBS-TX containing 1% BSA and 10% normal goat serum for 30 min followed by overnight incubation with the primary antibody diluted at 1:50 at 4 °C. The rabbit polyclonal K v 1.6 antibody (APC-003, Alomone Labs, Jerusalem, Israel) was generated against a glutathione S-transferase (GST) fusion protein corresponding to residues 463-530 of the rat K v 1.6 protein. After overnight incubation, slides were rinsed in PBS and incubated with peroxidase conjugated polyclonal goat anti-rabbit IgG, H & L chain specific secondary antibody (Calbiochem, San Diego, CA, USA) at 1:200 for 1 h at room temperature. Negative control sections were incubated with blocking media and the secondary antibody. Sections were extensively rinsed to remove unbound antibody and the detection of antibody binding was carried out with Vector ® NovaRED ™ Substrate Kit for peroxidase (Vector Laboratories, Burlingame, CA, USA). Finally, slides were mounted and cover-slipped.