INTRODUCTION

Top Abstract Introduction Conclusion References

CLONIDINE IS A PARTIAL AGONIST of ₂-adrenergic receptors (₂AR) with highest affinity (3 nM) for the _2AAR, one of three paralogous gene products (11, 22). It is also a weak agonist of ₁ARs. Until very recently, the ₂AR agonist activity of clonidine had been thought to account fully for the central and peripheral components of its antihypertensive activity. However, this notion is now being revisited given recent evidence that clonidine and several "second generation" centrally acting hypotensive drugs also bind to cellular components other than ₂ARs. These binding sites, henceforth called nonadrenergic imidazoline binding sites (NAIBS), have been classified into two broad categories: I₁ and I₂. This distinction is based on their relative affinity for a series of drugs with an imidazol(idin)e, imidazole, or guanidinium structure (5).

The I₂ class of NAIBS is associated with mitochondrial monoamineoxidases (18, 24), and their affinity for clonidine is very low [10-100 µM (5)]. On that basis, their contribution to the clinical activity of this prototypical hypotensive agent can probably be discounted, and these sites will not be discussed further in this review.

I₁ sites are present at relatively low level in the brain [binding absent in the nucleus of the solitary tract, binding capacity ranging from <10% of ₂ARs in the cortex to ~25% in the brain stem (5)]. The chemical identity of these binding sites remains unknown. The affinity of clonidine for I₁ sites (4-8 nM; see Ref. 5) is comparable to its affinity for ₂ARs (22). The I₁ affinity of two recently introduced central hypotensive drugs, rilminidine and moxonidine (38), exceeds that for ₂ARs by a factor of 3 to 30 depending on the study (22). These observations have spurred efforts to determine to what extent binding to I₁ sites might contribute to the hypotensive effect of clonidine. A related issue is whether the new agents rilminidine and moxonidine owe their distinct pharmacological profile (lesser degree of sedative side effects) to the fact that they are more specific "I₁ receptor agonists" than clonidine. From here on, the theory that clonidine and related agents produce their hypotensive effect partly or totally via I₁ receptors will be called the "imidazoline theory." The classic theory according to which the pharmacological activity of these agents derives predominantly from their ₂AR agonist activity will be called the "₂ theory."

In a separate editorial, Drs. Ernsberger and Haxhiu review the evidence suggesting that I₁ NAIBS might be some type of G protein-coupled receptor or a molecule associated with these receptors (5, 28). The remainder of this article reviews the evidence for and against a role of I₁ receptors in the antihypertensive activity of clonidine and its analogs.

	EVIDENCE SUPPORTING THE 2-THEORY

This theory is supported by four main categories of evidence.

Effect of systemically administered clonidine and related hypotensive agents is blocked by selective ₂AR antagonists. The central hypotensive effect of clonidine is antagonized by all ₂AR antagonist drugs that are sufficiently lipophilic to penetrate the blood-brain barrier (37). Many of these antagonists and especially the most potent (e.g., idazoxan, methoxyidazoxan, efaroxan) have an imidazoline structure and bind to I₁ sites. Therefore, proponents of the imidazoline theory have argued that these drugs antagonize the effect of clonidine because they are also (idazoxan) or primarily (efaroxan) "imidazoline receptor antagonists." Unfortunately, this interpretation is incompatible with the results of the classic experiments performed with the ₂AR antagonists of the Rauwolscia alkaloid family (yohimbine and rauwolscine; e.g., see Ref. 34). Indeed these very effective blockers of clonidine actions have very low affinity for I₁ sites [e.g., yohimbine: inhibition constant (K_i) = 5 µM for I₁ sites (5), 43 nM for rat _2AARs (11), and ~3 nM for _2AARs in species others than rodents (19)]. Moreover the effect of clonidine and related drugs (bromoxidine, rilminidine) can also be blocked by SKF-86466 (19, 36), an antagonist with even higher ₂AR selectivity than the Rauwolscia alkaloids [K_i of SKF-86466 for I₁ sites: 100 µM (15); K_i for ₂AR: 40-50 nM (14, 15)]. These newer studies and the older ones based on the use of Rauwolscia alkaloids provide convincing evidence that the central sympatholytic effect of systemically administered clonidine and some analogs can be blocked by selective ₂AR antagonists devoid of affinity for I₁ sites. This evidence leaves room for only a limited number of possible interpretations. One is that the contribution of the putative I₁ receptors to the hypotensive activity of ₂AR agonists such as clonidine, bromoxidine, and rilminidine is negligible when these agents are administered systemically. The second interpretation is that I₁ receptor binding produces no effect unless ₂ARs are simultaneously activated.

Expression of mutated _2AAR in mice eliminates hypotensive response to ₂-adrenergic agonists. One of the most persuasive new pieces of evidence that ₂AR activation is responsible for the hypotensive activity of clonidine and analogs comes from a gene substitution experiment in the mouse (21). Substitution of a single amino acid of the _2AAR (aspartate to asparagine in position 79) produced a strain of mice in which _2AAR expression is 90% downregulated and in which the remaining receptors cannot activate potassium currents nor inhibit calcium currents. Remarkably, these mice (D79N) have lost the hypotensive response to systemic injection of two ₂AR agonists with an imidazoline structure [dexmedetomidine, bromoxidine (21)] and to clonidine itself (L. Limbird, personal communication). Conversely, a mouse strain in which the _2CAR has been knocked out responds normally to dexmedetomidine (20). Finally, in an _2BAR knockout mouse strain, the early hypertensive component of dexmedetomidine is eliminated, whereas the later hypotensive response is increased (20). The most tempting interpretation of these genetic studies is as follows. First, the sustained hypotensive effect of ₂AR agonists with imidazoline structure is mediated predominantly (perhaps exclusively) by their agonist activity at _2AARs. Second, the initial hypertensive effect of these drugs is due to their agonist activity at _2BARs (location yet to be determined). Finally, stimulation of _2BARs tends to attenuate the hypotensive effect produced by activation of _2AARs. The above interpretation is based on the implicit assumption that the D79N point mutation of the _2AAR has no effect on I₁ sites. This point needs verification. Also, the D79N mouse experiments remain compatible with the possibility that I₁ receptor stimulation produces no effect unless _2AARs are simultaneously activated.

₂ARs are present on brain stem neurons that control sympathetic tone. ₂ARs are present, albeit in widely different density, at multiple sites throughout the brain and spinal cord (27, 32, 35). Consistent with the key role of _2AARs in mediating the cardiovascular effects of clonidine (21), high levels of this receptor subtype are found in all brain stem nuclei involved in autonomic regulations (32). In particular, immunoreactivity for _2AARs is present in most bulbospinal cells that contribute an excitatory input to vasomotor sympathetic preganglionic neurons, including the C1 and A5 catecholaminergic neurons of the ventrolateral medulla and the serotonergic cells of the medullary raphe (8, 10, 29). This immunohistochemical evidence is congruent with electrophysiological data. Indeed, one or both of the two classically described effects of ₂AR stimulation (augmented inwardly rectifying K conductance and decrease in high voltage-activated calcium current) have been found in C1, A5, and raphe serotonergic cells (Ref. 17 and unpublished data by Li, Bayliss, and Guyenet). In addition, postsynaptic ₂ARs coupled to an increase in K conductance are also present on sympathetic preganglionic neurons (see Ref. 8). More generally, almost all electrophysiological studies have found that the effect of clonidine on central nervous system neurons can be mimicked by the application of a catecholamine in the presence of ₁AR blockade. Exceptions exist (e.g., Ref. 31), but they are generally associated with concentrations of clonidine (threshold 1 µM) that are much higher than the therapeutic plasma concentrations of this drug (<10 nM) or its K_i for _2AARs (~5 nM). Conceivably, such high-dose effect could contribute to the hypotension produced by microinjection of high concentrations of clonidine in the ventrolateral medulla "in vivo."

In short, bona fide ₂ARs are present on sympathetic preganglionic neurons and on their main supraspinal excitatory inputs. Activation of these receptors by clonidine produces cumulative levels of inhibition of the sympathetic network, which may account for the particular sensitivity of the sympathetic outflow to this drug. To the best of our knowledge, there is no electrophysiological evidence that low concentrations of clonidine (<1 µM) produce effects that are different from those of norepinephrine applied in the presence of an ₁AR antagonist.

Lack of evidence that selective I₁ ligands produce biological activity. One way to prove that I₁ binding sites have a regulatory role would be to demonstrate that specific I₁ ligands devoid of affinity for ₂ARs are biologically active. The biological chemistry has now been achieved but the first fully tested compound, AGN-192403, was found to have neither agonist nor antagonist activity in hemodynamic tests (22). AGN-192403 is an imidazoline derivative with moderately high affinity for I₁ binding sites (K_i: 42 ± 17 vs. 9 nM for the reference compound clonidine) but without significant affinity for any of the three subtypes of ₂ARs (K_i > 20 µM) or for ₁ARs. AGN-192403 produced no effect on blood pressure in monkeys at doses of up to 5 mg/kg, whereas clonidine produced its predictable hypotensive effect with a mean effective dose of 17 µg/kg in the same animals. AGN had no effect in rabbits when administered either intravenously or intracerebroventricularly to circumvent the blood-brain barrier, and it did not antagonize the hypotensive effect of clonidine or that of the purported I₁-selective hypotensive agent moxonidine. These results are consistent with three possibilities. First the high-affinity I₁ binding site(s) may not be a regulatory protein or proteins. Second, ligand binding to these proteins causes no change in neuronal activity. Third, I₁ receptor activation has no effect on the autonomic nervous system. However, a definitive conclusion must await the testing of additional selective I₁ ligands.

The discovery of the putative neurotransmitter agmatine is a serendipitous by-product of the search for endogenous ligands of NAIBS (16). Agmatine is present in most tissues including brain, and it possesses weak binding affinity for both ₂ARs and I₁ sites (25). It was briefly considered as a putative endogenous ligand of I₁ "receptors" but more complete investigations have not supported this concept (30).

	EVIDENCE FOR THE IMIDAZOLINE THEORY: THE RVLM PARADOX

The evidence supporting the imidazoline theory originates from pharmacological experiments in vivo in which drugs have been applied intracerebroventricularly or intraparenchymally, particularly within the rostral ventrolateral medulla (RVLM). This section will focus mainly on the RVLM work, since there is general agreement that this part of the medulla oblongata is the main site at which systemically administered clonidine and congeners produce their sympatholytic effect, at least in anesthetized animals (8, 23, 29). What constitutes the RVLM paradox is the apparent incompatibility between the evidence reviewed in the first section and the data of at least three groups of investigators, which suggest that the hypotension caused by the presence of clonidine and related drugs in RVLM owes nothing to their ₂AR agonist activity but is entirely due to their "I₁ agonist" activity (3, 5, 26, 33). This discrepancy is puzzling given that 75% of clonidine binding in RVLM is to ₂ARs [perhaps an even larger percentage in rats (2)]. The rest of this section is an attempt to resolve the paradox by examining the three types of evidence used by proponents of the view that clonidine does not work as an ₂AR agonist in RVLM.

First evidence: microinjection of norepinephrine into RVLM does not produce hypotension while clonidine does; therefore, clonidine cannot work as an ₂AR agonist. These negative data (e.g., Ref. 1; for review see Ref. 9) provide the weakest evidence in favor of the imidazoline theory. Norepinephrine (NE) is an agonist of ₁, ₂, and -receptors all of which are present in the ventrolateral medulla. Thus the effect produced by microinjecting norepinephrine into RVLM is the sum of opposing actions on many types of catecholaminergic receptors and neurons. There is little reason to expect as profound a hypotensive effect from NE, which exerts predominantly excitatory effects on neurons via the ₁AR, as from a selective ₂AR agonist such as clonidine. A catecholamine analog that can be more legitimately compared with clonidine than NE is -methylnorepinephrine, which is a relatively selective agonist for ₂ARs and has no affinity for I₁ binding sites. In fact, this compound does produce hypotension when injected into RVLM in adequate doses [1-10 nmol (4, 7)].

Second type of evidence: the hypotensive effect of clonidine cannot be reversed by microinjection into RVLM of SKF-86466; therefore, it cannot be due to ₂AR stimulation. The second type of evidence supporting the selective action of clonidine on I₁ receptors in RVLM is again based on negative findings. The evidence relies on comparisons between the ability of various ₂AR antagonists (with or without affinity for I₁ binding sites) to reverse or prevent the hypotensive effect of clonidine. The typical experiment (4, 26) reveals that the hypotensive effect of clonidine can be reversed by injection into RVLM of a given dose of idazoxan (1 nmol) but not by the same dose of SKF-86466. Idazoxan is a potent ₂AR antagonist that also binds to I₁ sites, and SKF-86466 is an ₂AR antagonist that does not bind to I₁ sites. It is then concluded that clonidine works via I₁ sites rather than as an ₂AR agonist. However, this conclusion is not justified by the data because the doses of SKF-86466 and idazoxan that were administered were not equiefficacious at blocking ₂ARs. This statement relies on classic receptor theory. The equation that describes the effect of a competitive antagonist on the response to an agonist is as follows where D is concentration of agonist, K_d is dissociation constant of agonist for receptor, and I is concentration of antagonist.

This relationship indicates that the effect of an antagonist depends on the ratio between its concentration (I) and its dissociation constant K_i for the receptor. The affinity of idazoxan for postsynaptic ₂ARs [~3 nM, (4)] is 10- to 15-fold higher than that of SKF-86466 [K_d: 35-50 nM (4, 14, 15)]. Because the dose of inhibitors used (I) was the same (1 nmol), the lack of effect of the SKF compound could simply reflect that its I/K_i ratio was 10- to 15-fold smaller than that of idazoxan. The same experimental problem undermines the interpretation of analogous experiments designed to test whether or not rilmenidine (6) and moxonidine (12) lower arterial pressure via ₂AR activation. In the latter study (12) where the dose of SKF-compound was three times that of idazoxan (still 3 or 4 times lower than an equiefficacious dose), this drug was in fact able to antagonize ~50% of the effect of moxonidine.

In conclusion, the antagonist studies described in this section remain theoretically compatible with the possibility that the hypotensive effects produced by microinjecting clonidine into RVLM could be due entirely to ₂AR stimulation.

Third line of evidence: the hypotensive potency of clonidine and related drugs injected into RVLM correlates better with their affinity for I₁ binding sites than for ₂ARs. The third type of result that supports the imidazoline theory is again a whole animal bioassay. A single dose of a series of agents is injected into the RVLM [catecholamines, ₂AR agonists with imidazoli(di)ne structure, and related agents such as imidazole acetic acid]. Their hypotensive "potency" is rank-ordered, and it is found that it correlates better with their K_i values for I₁ sites than for ₂ARs (e.g., Ref. 4; for additional references, see Ref. 9). The first problem with such a design is that agonist potency has been estimated by measuring the magnitude of the hypotensive response to a fixed dose of each agonist. Second, these experiments have ignored the impact of the widely different lipophilicity of the injected agents, the large differences in half-life between catecholamines, which are avidly taken up and metabolized, and that of stable ₂AR agonists such as those with an imidazoline structure. Finally, many of the agents used to draw the dose-effect relationships act on multiple receptors besides ₂ARs and the hypothetical I₁ receptors. For instance, one of the agents used in these experiments, imidazole acetic acid, is a powerful GABA_A receptor agonist (K_i = 50 nM; see Ref. 9), and catecholamines activate multiple receptors. In brief, innumerable factors contribute to distort the rank-ordering of potency of agonists, possibly in a way that favors stable imidazoline derivatives that have affinity for I₁ sites. Therefore, these correlations do not permit definite conclusions regarding the type of receptor responsible for the effect of clonidine and certainly they do not permit exclusion of a role for ₂ARs.

	CONCLUSION

Top Abstract Introduction Conclusion References

In conclusion, new evidence supports a largely predominant, if not exclusive, role of ₂ARs in the cardiovascular and neurophysiological effects of clonidine and its closest congeners. This statement applies only to doses or concentrations of these drugs that are low enough to be therapeutically relevant. An unresolved question is whether the imidazoline theory has the potential to account for the clinical differences between clonidine and the new generation of hypotensive agents (rilminidine, moxonidine; e.g., see Refs. 13, 36). Several animal studies show clear if subtle differences between clonidine and these newer agents, especially moxonidine (e.g., see Ref. 13). Possibly, these differences could be due to their affinity for I₁ NAIBS but the evidence is still circumstantial because of the unavailability of specific agonists or antagonists of the presumed I₁ receptors (13, 26). It is also possible that the relative affinity and activity of these various drugs for the three ₂AR subtypes could underlie their differences or that the new ₂AR agonists may also bind to a spectrum of additional G protein-linked receptors as is the case for virtually all central nervous system active drugs. In short, the imidazoline theory faces three major hurdles. The first is the search for the structure and function of the protein responsible for I₁ ligand binding. The second is the search for some specific electrophysiological effect of I₁ ligands. The third is the search for biologically active selective I₁ ligands. Regardless of the success of these endeavors, the study of NAIBS has already had at least one important unintended consequence, which is the serendipitous discovery that agmatine may be a central nervous system transmitter (16).