INTRODUCTION

Top Abstract Introduction References

THIS REVIEW CONSIDERS the data for and against the following hypotheses. First, I₁-imidazoline-binding sites are functional receptors according to accepted criteria for the identification of receptors. Second, these receptors can contribute to vasodepressor responses to imidazoline agonists within the rostral ventrolateral medulla (RVLM) region of the medulla. Third, the well-known vasodepressor action of ₂-adrenergic receptors is mediated in regions other than the RVLM. Fourth, recent reports that are seemingly inconsistent with the I₁-imidazoline receptor hypothesis require reconsideration and reinterpretation. We suggest procedural guidelines for in vivo studies of I₁-imidazoline receptor function and propose a local circuit model for the action of imidazolines within the RVLM.

	I. ARE I1 RECEPTORS FUNCTIONAL RECEPTORS?

Any claim for a novel receptor should be appraised skeptically by rigorous criteria (39). First, radioligand-binding studies characterizing I₁-binding sites show properties comparable to known receptors. The binding of radiolabeled clonidine analogs to I₁ sites is rapid, specific, saturable, reversible, and of high affinity (9, 11-14). Furthermore, the ligand recognition profile of I₁ sites is unique. Although many compounds bound by I₁ sites are also bound by ₁- or ₂-adrenergic receptors (₂ARs), I₁ sites do not recognize phenylethylamine agonists or nonimidazoline antagonists such as rauwolscine and SKF-86466. Conversely, several agonists and antagonists interact preferentially with I₁ sites relative to ₂-adrenergic receptor (₂AR) sites. The other I receptor subtype, the I₂ site, is mitochondrial and might be identical to monoamine oxidase (48). The I₂ site is not a receptor and will not be reviewed here.

Physiological responses have been linked to I₁ receptors. The role of an I₁ receptor is established when specific ₂-antagonists fail to abolish the actions of imidazolines, whereas I₁ antagonists are effective. Central control of intraocular pressure by the RVLM has been linked to I₁ receptors (7). Adrenal chromaffin cells and the PC12 cell line lack ₂AR (13, 18, 41), so I₁ receptors likely mediate the effects of imidazolines on these cells, including induction of phenylethanolamine N-methyltranserase (PNMT) mRNA (18), release of prostaglandins (13) and choline phosphate (42), and generation of diacylglyceride (DAG; 41, 42). These actions of imidazolines in chromaffin-derived cells are receptor mediated, since relative potencies correlate with binding affinities (18) and I₁ antagonists, but not ₂-antagonists, block these cellular responses (13, 41).

Specific anatomic and histological distribution of I₁- binding sites appropriate for their proposed functions has been shown using autoradiography (13). Labeling with p-[¹²⁵I]iodoclonidine persisted in RVLM in the presence of epinephrine, showing the presence of I₁ sites, whereas the intense labeling in the nucleus of the solitary tract (NTS) was abolished, indicating a high density of ₂AR but not I₁ sites in NTS. Similarly, in the pons, the locus ceruleus expressed mainly ₂ while I₁ were present in ventral tegmentum. This distribution is consistent with the proposed role of I₁ receptors in central cardiovascular control (9, 12, 27). At the subcellular level, I₁ sites are localized to plasma membrane fractions in RVLM (16), human platelets (35), and PC12 cells (13). The plasma membrane localization of I₁-binding sites is appropriate for a receptor.

A 1:1 correlation of binding affinity with function potency is the most important criterion for identifying a receptor (39). Correlations with the binding affinity of I₁ sites have been found for vasodepressor potency on microinjection into RVLM (12, 17). In contrast, ₂AR affinity was unrelated to vasodepressor potency. Intracisternal doses of ₂- and I₁-agonists sufficient to lower blood pressure in conscious SHR rats (6) and effective doses for control of human hypertension (10) both correlate with I₁ but not ₂-affinity. The absence of a relationship between ₂-potency and vasodepressor action in rats or humans can best be explained by the existence of an additional receptor. Binding affinity at I₁ sites also predicts the relative induction by I₁ agonists of mRNA for PNMT, the synthetic enzyme for epinephrine, in adrenal chromaffin cells (18).

Steps leading from occupation of I₁ receptors to induction of cellular signals have been identified. Cell biological studies of I₁ receptor signaling have focused on chromaffin cells and a tumor cell line derived from them, the PC12 pheochromocytoma. These cells express I₁ receptors, but lack ₂AR (13, 41). The I₁-agonist moxonidine elicits release of prostaglandin E₂ (PGE₂) from PC12 cells (13). This effect is attenuated by BDF-6143, an effective blocker of moxonidine-induced vasodepression. Cimetidine, which behaves as an I₁-agonist by eliciting vasodepression in RVLM (12), also elicits PGE₂ release that can be antagonized by BDF-6143 (13). Release of the PGE₂ precursor arachidonic acid is elicited by low concentrations of I₁ agonist, but not by specific ₂-agonists such as guanabenz, and can be blocked by I₁ receptors, but not by ₂AR antagonists (15).

The stimulation of I₁ receptors in PC12 cells elicits accumulation of the second-messenger DAG from phospholipid precursors and increased total cellular DAG mass (41). DAG generation was dose dependent and competitively inhibited by efaroxan (41), a potent blocker of moxonidine's cardiovascular effects (8, 27). Because previous studies ruled out activation of phosphatidylinositol-selective phospholipase C (PI-PLC) (38) and phospholipase D (41) by I₁ receptors, phosphatidylcholine selective-phospholipase C (PC-PLC) was implicated. We directly tested the role of PC-PLC in the actions of I₁ receptors in PC12 cells and in RVLM (42). In PC12 cells, moxonidine elicited production of both reaction products of PC-PLC: DAG and phosphocholine. Both products were prevented from appearing by D609, a specific PC-PLC inhibitor. In SHR, complete prevention of the vasodepressor response to intravenous moxonidine was achieved by microinjection of D609 in RVLM (Fig. 1). These data implicate PC-PLC in RVLM in vasodepressor actions of imidazolines (42).

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Blood pressure response to intravenous moxonidine following pretreatment by rostral ventrolateral medulla (RVLM) microinjection with imidazoline and nonimidazoline 2-adrenergic receptor (2AR) antagonists or a selective phosphatidylcholine-phospholipase C (PC-PLC) inhibitor. Spontaneously hypertensive rats (27, 42) received bilateral RVLM microinjections of either vehicle, the I1/2-antagonist efaroxan, the 2-antagonist SKF-86466, or the PC-PLC inhibitor D609. Ten minutes later, moxonidine was given intravenously at 40 µg/kg. Efaroxan or D609 prevented the action of systemic moxonidine, whereas SKF-86466 was completely ineffective.

Therefore, I₁ receptors fulfill each of the essential criteria for identification as functional receptors (39). In particular, the vasodepressor response to imidazolines is consistent with each of the major criteria of specificity, function, location, correlation, and cell signaling. Furthermore, the action of various agonists, antagonists, and inhibitors in isolated cells correlates well with their in vivo effects on blood pressure within the RVLM.

	II. DO I1 RECEPTORS CONTRIBUTE TO VASODEPRESSOR RESPONSES IN RVLM?

Clearly, ₂AR agonists with no affinity for I₁ receptors (i.e., guanabenz) act centrally to lower blood pressure. We propose that ₂AR agonists and I₁ agonists act in separate brain regions, with ₂AR predominant in NTS and I₁ receptors in RVLM. The competing concepts in this debate are therefore not "the I₁ receptor hypothesis" versus "the ₂AR hypothesis," but instead "the NTS ₂/RVLM I₁ hypothesis" versus "the RVLM ₂ hypothesis."

The RVLM is the site of vasodepressor actions of imidazolines. Localization of the site of action for imidazolines to the RVLM is supported by three types of experiments. First, brain stem transections rule out sides above the medulla or in the spinal cord (22). Humans with cervical spinal cord transection fail to show a vasodepressor response to clonidine (30). Second, the RVLM is exquisitely sensitive to microinjected imidazolines (5, 20). Third and most important, microinjection of imidazoline antagonists into the RVLM abolishes the vasodepressor response to systemic imidazolines (22, 27, 37). Thus, although the agonist continues to circulate and have actions elsewhere in the brain and throughout the periphery, blockade of receptors only within the RVLM is sufficient to eliminate any fall in blood pressure.

In cats, clonidine microinjections were most effective caudal to the RVLM sympathoexcitatory area (20). Clonidine presumably stimulates neurons in this sympathoinhibitory region, which in turn inhibit sympathoexcitatory RVLM neurons. Another group localized an active site for clonidine caudal to these neurons (49). A third group found two active sites for clonidine microinjections in the RVLM region, one in the pressor zone and another more caudal (53). Thus in the cat clonidine acts in a region of RVLM apart from the tonically active sympathoexcitatory neurons, as depicted in our hypothetical model (see Fig. 4).

NTS ₂AR mediate vasodepressor responses to nonimidazolines. Activation of brain stem ₂AR lowers blood pressure, most likely in NTS. Microinjection of minute doses (0.02-0.3 nmol) of norepinephrine, epinephrine, or -methylnorepinephrine in NTS elicited vasodepression (reviewed in Ref. 9). Clonidine was much less effective, requiring 600-fold higher doses to produce similar falls in pressure (56). Another imidazoline, oxymetazoline, was completely ineffective. Microinjection of the selective I₁-agonists rilmenidine and moxonidine did not affect blood pressure at doses up to 40 nmol (22, 27). The ineffectiveness of clonidine, oxymetazoline, and rilmenidine in lowering blood pressure within NTS probably reflects low ₂-efficacy, as each is a partial agonist. BHT-920, a noncatecholamine full ₂-agonist, is as potent as -methylnorepinephrine in NTS (32). Conversely, yohimbine microinjected into NTS increases blood pressure in doses as low as 10 pmol (47). Thus ₂AR in NTS clearly regulate blood pressure tonically.

Role of I₁ receptors in vasodepressor actions in RVLM. The evidence implicating I₁ receptors in the vasodepressor actions of imidazolines has been reviewed (9, 17). The presence of I₁ receptors in RVLM is well established. The vasodepressor action of cirazoline (3), an ₂-antagonist and I₁-agonist, is incompatible with the "₂AR only" hypothesis. Participation of ₁AR in the action of cirazoline is ruled out because neither ₁-agonists nor ₁-antagonists affect blood pressure in cat RVLM (2, 4). The ineffectiveness of phenylethylamine -agonists in eliciting vasodepression in RVLM is also incompatible with a lone role for ₂AR. Microinjection of various nonimidazoline ₂-agonists into the cat RVLM failed to elicit vasodepression at doses up to 40 nmol (3, 4). Comparable experiments in rats have yielded inconsistent results. Several studies show little response to nonimidazoline ₂-agonists in RVLM (12, 55) or on the medullary ventral surface (40). A single study found -methylnorepinephrine to lower blood pressure in RVLM (23). However, 30-fold higher doses were needed than in NTS (9, 56).

If ₂AR in RVLM mediate vasodepression, then local drug responses in RVLM should resemble those in the NTS, a region with a proven ₂AR-mediated fall in blood pressure. In contrast, structure-activity relationships for vasodepressor responses differ radically between NTS and RVLM, with phenylethylamines more potent in NTS and imidazolines more potent in RVLM. Epinephrine, for example, lowers blood pressure when microinjected in NTS at 0.02 nmol unilaterally (56), whereas in RVLM a 50-fold higher dose bilaterally is less effective (12). In another RVLM microinjection study, 3 nmol of epinephrine or norepinephrine did not affect blood pressure (55). Conversely, oxymetazoline is completely inactive in NTS at doses up to 20 nmol (56), whereas in RVLM oxymetazoline is nearly as potent as clonidine (12). The relative effectiveness of different -agonists in NTS are fully consistent with ₂AR: epinephrine > norepinephrine > -methylnorepinephrine >> imidazoline partial agonists. ₂-Antagonists also have contrasting effects in NTS and RVLM. In NTS, antagonists potently elicit sustained pressor responses (31, 32, 47). In contrast, most ₂-antagonists lower pressure when microinjected into cat (2) or rat RVLM (12). One might argue that the catecholamines are inactive when microinjected into RVLM because they are subject to uptake and degradation, whereas the imidazolines are not. However, this cannot account for the high potency of these agents upon microinjection into NTS, an area with abundant uptake sites and degradative enzymes. The ₂AR-only hypothesis cannot account for differences between RVLM and NTS sites of injection, which are readily explained by the presence of I₁ receptors in RVLM but not in NTS.

The hypothesis that imidazolines act through I₁ receptors when microinjected into the RVLM has not been contradicted experimentally. Only antagonists active at I₁ receptors (efaroxan, idazoxan, or methoxyidazoxan) block the action of imidazolines microinjected into the RVLM, whereas other ₂-antagonists such as SKF-86466 are inactive (9, 12, 27). However, several investigators propose that when imidazolines are given systemically, only ₂AR participate (28). This hypothesis was recently tested. As shown in Fig. 1, microinjection of the I₁ antagonist efaroxan into RVLM completely prevented the hypotensive action of intravenous moxonidine. In contrast, blockade of RVLM ₂AR with SKF-86466 failed to attenuate this response. Similarly, microinjection of SKF-86466 into RVLM did not attenuate the effect of intravenous rilmenidine (Fig. 2A). In contrast, the ₂/I₁ antagonist idazoxan completely abolished rilmenidine's effect at a low dose. Similarly, Nosjean and Guyenet (34) showed that RVLM microinjection of rauwolscine lowered pressure, and subsequent intravenous clonidine elicited a further fall, so that the total depressor response was indistinguishable from the response to clonidine alone (Fig. 2B). In contrast to rauwolscine, idazoxan completely prevented the action of clonidine. These studies implicate RVLM I₁ receptors in actions of either local or systemic imidazolines. Furthermore, they rule out participation of ₂AR, at least in RVLM.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Actions of 2AR and I1 receptors agonists and antagonists in RVLM. A: vasodepression elicited by intravenous rilmenidine (recalculated from Ref. 22). As in Fig. 1, microinjections of vehicle, the I1/2-antagonist idazoxan, or the 2-antagonist SKF-86466 were made 5 min before intravenous injection of the selective I1-agonist rilmenidine (500 µg/kg). Baseline blood pressure ± SE is shown as continuous and dashed lines. Idazoxan (1 nmol) blocked rilmenidine, but SKF-86466 had no effect at up to 10 nmol. B: vasodepressor action of intravenous clonidine (20 µg/kg) with or without blockade of 2AR in RVLM by local microinjection of rauwolscine (16 nmol) or blockade of I1 receptors with idazoxan (2 nmol). Data are from Ref. 34. The fall in mean arterial pressure elicited by clonidine was not affected by pretreatment with rauwolscine in RVLM, but clonidine's action was completely blocked by the I1 receptors antagonist idazoxan. C: the firing activity of a putative adrenergic C1 neuron in RVLM, after increasing cumulative doses of intravenous clonidine, followed by yohimbine (data from Fig. 6B of Ref. 1). Note that yohimbine's stimulatory effect greatly exceeds clonidine's inhibitory action.

Although much of the support for a role of I₁ receptors in vasodepression derives from anesthetized preparations, several studies used conscious animals (6, 29). Intracisternal injection of the selective ₂-agonist BHT-920 in freely moving rats increased blood pressure and plasma catecholamines, opposite to clonidine (29). Rauwolscine did not abolish the action of clonidine (29).

Alternatives to the I₁R hypothesis. Despite identification of NTS as the major locus for ₂AR regulation of blood pressure (9), the RVLM was proposed by one group of investigators (24). ₂AR have been detected in RVLM by membrane binding studies (10, 12, 14), autoradiography (13), in situ hybridization (26), and electrophysiology (Fig. 2C). However, if receptor density is used as a criterion, the NTS must be identified as the most likely location for ₂AR regulation of cardiovascular function. Of RVLM neurons projecting to the spinal cord, those expressing ₂AR also contain PNMT (43). PNMT neurons in RVLM reportedly have little role in regulating blood pressure (1, 46). Notably, the stimulatory effect of ₂-antagonists on the firing rate of PNMT neurons far exceeds the inhibitory effect of ₂-agonists (Fig. 2C). This contrasts with the vasodepressor action of most ₂-antagonists in RVLM (2, 12, 22, 34).

If RVLM ₂AR tonically regulate blood pressure, then microinjection of ₂-antagonists should elicit large pressor responses, as they do in NTS. However, only ₂-antagonists active at I₁ receptors, such as efaroxan and methoxyidazoxan, elicit pressor responses in RVLM (27, 43). Conversely, selective ₂-antagonists decrease pressure (2, 12, 22, 34). Cirazoline, an ₂-antagonist and I₁-agonist, also elicits a fall in pressure (3). Thus tonically active ₂AR are coupled to the control of blood pressure in NTS but not in RVLM.

Some tests of the role of I₁ receptors in vasodepressor responses have yielded negative or mixed results. Although clonidine's actions can often be blocked by ₂-antagonists, many of these agents are not specific (Table 1). Some ₂-antagonists also block I₁ receptors, particularly idazoxan, methoxyidazoxan, piperoxan, and tolazoline (11, 17). A common weakness of studies reporting negative results is systemic delivery of antagonists (Table 1). Partial blockade of vascular AR decreases the neural contribution to resting blood pressure. Thus systemic ₁AR blockade with prazosin (1 mg/kg) attenuates the hypotensive response to clonidine nearly as effectively as yohimbine (1 mg/kg) (50). This does not mean that clonidine's action is mediated by ₁AR, but rather that the antagonist acts downstream. Similarly, -antagonists can prevent the action of clonidine in humans or animals (19). ₂AR antagonists such as SKF-86466 (28, 51) or yohimbine (52) may induce similar perturbations when given intravenously. An additional weakness emerges when data are presented only as net change (28, 51, 52). After intravenous ₂AR antagonist, starting blood pressure is lower. For example, SKF-86466 lowered mean pressure and increased sympathetic activity 50% (51). Thus the physiological state was altered by systemic ₂AR blockade. Plotting absolute data rather than net change alters the picture (Fig. 3). Thus SKF-86466 plus clonidine lowers blood pressure to the same level as clonidine alone (Fig. 3A). A very high dose of SKF-86466 followed by clonidine still resulted in blood pressure substantially below the starting level. These data implicate both I₁ receptors and ₂AR in clonidine's actions.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Recalculated data from studies claiming negative results for involvement of I1 receptors in the cardiovascular actions of systemic imidazolines. A: effect of pretreatment with intravenous antagonists on the response to intravenous clonidine at 10 µg/kg (recalculated from Ref. 28). Shown is the lowest diastolic blood pressure reached after (left to right), no treatment, vehicle followed by clonidine, 1 and 3 mg/kg SKF-86466 followed by clonidine, and 1 mg/kg idazoxan followed by clonidine. Blood pressure remained substantially below baseline levels whatever the antagonist pretreatment. B: mean arterial blood pressure following treatment with increasing intravenous doses of moxonidine after injection of either vehicle or yohimbine into the cisterna magna (recalculated from Fig. 9A of Ref. 52). C: plasma norepinephrine levels from experiment shown in B (recalculated from Fig. 9B of Ref. 52).

One negative study was appropriately designed to test the I₁ receptor hypothesis. The ₂AR/5-HT_1a antagonist yohimbine was injected into the rabbit cisterna magna (52), thereby avoiding peripheral ₂-blockade (Fig. 3B). Moxonidine induced a dose-dependent but tiny fall in blood pressure. The authors' claim that brain stem ₂-blockade with yohimbine abolished the vasodepressor response to the I₁-agonist moxonidine is therefore unconvincing. Importantly, despite the lack of physiologically significant vasodepression, plasma catecholamines were decreased by moxonidine (Fig. 3C). This sympatholytic response was not attenuated by brain stem ₂AR blockade with yohimbine. The authors suggest that persistent sympathoinhibition is mediated by peripheral ₂-autoreceptors (52). However, the central locus of moxonidine-induced sympathoinhibition is well established (10). The simplest explanation for Fig. 3C is that moxonidine acts primarily through brain stem I₁ receptors.

When mice with mutant _2aAR were given selective ₂AR agonists medetomidine or bromonidine systemically, they failed to show vasodepressor responses (33). The authors also concluded that any functional role of I₁ receptors in the mouse was ruled out, because they incorrectly assumed that medetomidine and bromonidine are I₁ agonists. However, medetomidine has almost no I₁ affinity (11, 35), and bromonidine is 100-fold selective for _2aAR over I₁ receptors (17). Moreover, mouse brain stem I₁ receptors have never been studied, and thus the role of these receptors in mice remains open.

A local circuit model of imidazoline action in RVLM. Clonidine-induced vasodepression can be attenuated by the serotonin antagonist methysergide or lesions of serotonin neurons, the opiate antagonist naloxone, the GABA antagonist bicuculline, desipramine and other tricyclic antidepressants, and even ethanol (reviewed in Ref. 9). Blockade of clonidine by diverse antagonists implicates a neuronal circuit with multiple transmitters. GABA has been consistently implicated, since bicuculline completely prevents the clonidine's hypotensive action (21, 44), and clonidine enhances GABA release in medulla and hypothalamus but not in cerebellum (36). In vitro, clonidine inhibits RVLM pacemaker neurons through release of GABA (44, 45).

A model of a local circuit within RVLM and its output to sympathetic preganglionic neurons (SPGNs) is shown in Fig. 4. Stimulation of I₁ receptors does not directly inhibit pressor neurons in RVLM but rather activates inhibitory GABA interneurons. Besides GABA, opiates and serotonin are candidate transmitters for a local circuit (omitted for simplicity). Interneuron activation inhibits reticulospinal sympathoexcitatory neurons providing tonic control of SPGNs by release of glutamate. Activation of ₂AR, in contrast, would inhibit PNMT-containing C1 neurons. Few RVLM neurons are sensitive to iontophoresed clonidine (1), except PNMT neurons (24, 46). These PNMT neurons show large increases in activity in response to ₂-antagonists (Fig. 2C), whereas these same antagonists tend to lower blood pressure (34). Thus the firing activity of these neurons is unrelated or even inversely correlated with sympathetic activity. Because norepinephrine and epinephrine usually inhibit SPGNs (25), PNMT neurons of the C1 group are probably sympathoinhibitory (Fig. 4) as reviewed elsewhere (44). Admittedly, C1 neurons are inhibited by activation of baroreceptors, but this defines the inputs of these neurons, not their outputs. Nearby glutamate neurons provide the primary drive to SPGNs (24, 46). Future studies will reveal whether this model or the alternative proposed by the Charlottesville group (43) best predicts experimental outcomes.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Hypothetical model of a local circuit of interneurons in RVLM mediating the vasodepressor action of imidazolines. Stimulation of I1 receptors in RVLM region may activate inhibitory interneurons. Inhibitory -aminobutyric acid (GABA) inputs to cardiovascular neurons in RVLM are activated by I1 receptor agonists. Reduced activity of RVLM reticulospinal neurons reduces activity of SPGNs. In RVLM, 2AR are expressed almost entirely by C1 adrenergic neurons, where they serve a powerful autoinhibitory function (24). We propose that these C1 neurons are mainly sympathoinhibitory, owing to the predominantly inhibitory action of norepinephrine (NE) on SPGNs mediated by spinal 2AR (25). This inhibitory action of NE released from C1 terminals is partially offset by excitatory 1AR. Although 2AR are expressed on C1 neurons in RVLM and on sympathetic preganglionic neurons in the spinal cord, their predominant cardiovascular role is mediated in NTS (not shown). E, epinephrine; Glu, glutamate.