I. Foreword: A Brief History of a Really "Hot" Aspect of Pharmacology

Top Next References

Natural products afford a window of opportunity to study important biology. If the natural product is used or abused by human beings, finding its biological target(s) is all the more significant. Hot pepper is eaten on a daily basis by an estimated one-quarter of the world's population and represents an aspect of pharmacology intimately familiar to most readers of this review. Recall your culinary experiences in, let us say, a Mexican restaurant. Food flavored with jalapeno pepper bites, it induces profuse perspiration, and a subsequent diarrhea is not uncommon (when pepper bites again). However, these symptoms become less intense in regular patrons of the restaurant. In more scientific terms, hot pepper is pungent, affects thermoregulation, activates autonomic reflexes, and is poorly absorbed. However, all these acute effects undergo tachyphylaxis upon repeated applications.

Hot pepper is a native of the Americas (cf. Andrews, 1984; Naj, 1992). Aztecs called it chili (this is how hot pepper is mentioned in the pre-Columbian Aztec manuscripts, known as tlacuilos), a name that has stuck in Latin America (information from the Internet: Encyclo'Pepr'edia, http://thepeppershop.com/index.html, and Chili Gazette, http://mexicanfood.tqn.com/msubhis.html). But the Old World adopted a different name, red pepper, instead, erroneously linking chili pepper (Capsicum annuum) to a similarly hot-tasting plant, black pepper (Piper nigrum) (Garrison, 1929). Following the discovery of the New World, the habit of the natives to eat their food hot was first noted by Diego Alvarez Chanca, a physician to the fleet of Columbus (cf. De Ybarra, 1906). In Europe, a depiction of the chili pepper plant appeared for the first time in the magnificent book of Fuchs (1542). Hot pepper was cultivated in monastery gardens in Moravia as early as 1566 (cf. Köhler, 1883). The Latin name of the plant, Capsicum, was given by the French botanist de Tournefort for unclear reasons (cf. Naj, 1992). A popular theory holds that the name Capsicum was derived from the Greek kapto, meaning "to bite" (cf. Maga, 1975), which appropriately describes the main characteristic of the fruit. Others argue that the name Capsicum is derived from capsa, the Latin word for box, referring to the fact that the pepper pod is hollow, divided into compartments containing the seeds (cf. Naj, 1992).

The active ingredient in hot pepper was first isolated by Thresh (1846) more than a century ago. Thresh named this compound capsaicin and predicted that the structures of capsaicin and vanillin were closely related. Despite this early discovery, it was not until 1919 that the exact chemical structure of capsaicin (Fig. 1) was determined (Nelson, 1919). The complete synthesis of capsaicin took another decade (Spath and Darling, 1930). In 1912, Wilbur Scoville (Scoville, 1912) introduced his Organoleptic Test to quantitate the pungency of hot peppers (that is, their capsaicin content) and, although HPLC has long supplanted the use of the human tongue (cf. Suzuki and Iwai, 1984), for quantitation the Scoville Unit remains the measure of hotness. If a pepper has 50,000 Scoville Heat Units, this means that its alcoholic extract needs to be diluted 1:50,000 for it to cease to be hot-tasting on the human tongue. The hottest pepper is the Mexican habanero, which boasts 350,000 Scoville Units (cf. Naj, 1992). The oral test of Scoville, abandoned by the food industry, survives in laboratories, where it is still used to detect capsaicin-like compounds in plant extracts.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Typical vanilloid structures: RTX, capsaicin, isovelleral, and scutigeral, representing the four known chemical classes of naturally occurring vanilloids, resiniferanoids, capsaicinoids, unsaturated dialdehydes, and triprenyl phenols, respectively.

With regard to the varied, nonculinary uses of hot pepper, there is apparently little new under the sun. For example, Incas burned dried chili peppers to combat the invading Spaniards by temporarily blinding them (cf. Naj, 1992). Four centuries later, the first U.S. patent was issued for the use of capsaicin for martial (tear gas) purposes (U.S. patent 1,659,158, 1928). Nowadays, capsaicin-containing sprays, fondly called Sergeant Pepper, are used by law-enforcement officials in the United States to subdue violent criminals, and in California are marketed as a "cop-in-a-can" for self-defense (Hyder, 1996).

The analgesic use of capsaicin is not novel either (cf. Whittet, 1968; Lembeck, 1987; Dasgupta and Fowler, 1997). Native Americans rubbed their gums with pepper pods to relieve toothache (cf. Naj, 1992). This practice later also gained popularity in European folk medicine, as was noted by the Hungarian botanist-turned-clergyman, Hangay (1887). As early as 1850, the Dublin Free Press recommended the use of alcoholic hot pepper extract on sore teeth for instant relief, recognizing for the first time the therapeutic potential of capsaicin (Turnbull, 1850). A fascinating aspect of the rich history of the analgesic use of capsaicin is that eunuchs serving the Chinese Emperors were castrated after their scrotums had been repeatedly rubbed with hot pepper extracts (cf. Anderson, 1990). As a curiosity, it is also notable that Native Americans used hot pepper extracts topically as an aphrodisiac, a practice adopted by early settlers to the dismay of their priests (Chili Gazette, http://mexicanfood.tqn.com/msubhis.html).

In 1640, Sir John Parkinson noted in his famous Theatricum Botanica that dogs detest hot peppers (cf. Naj, 1992). Chickens, by contrast, can be fed dried pepper powder to turn the yolks of their eggs orange-red. These observations anticipate the modern use of capsaicin to make bird food squirrel-free (cf. Rouhi, 1996). It remains a mystery why is it that the same hot taste, which repels a variety of mammals from rats to squirrels to dogs, is found pleasurable by so many humans. Psychologists speculated that eating hot peppers may be a form of masochism (Rozin and Schiller, 1980). This theory, however, is at variance with the well known geographic pattern of hot pepper consumption, namely, people living in tropical climates prefer their food hotter than those residing in temperate climates (Moore, 1970; Rozin, 1978). This pattern gave rise to the theory that eating hot, spicy food helps combat the warm climate via gustatory sweating (Haxton, 1948; Lee, 1954). This model has gained recent reinforcement by the cloning of a capsaicin receptor, which seems to be operated by both heat and capsaicin (Caterina et al., 1997). Nonetheless, the human liking of, or aversion toward, the taste of hot pepper is probably far more complex than this relatively simple physiological model implies. For instance, Russians like their vodka hot (vodka peperovka), whereas efforts by the Pabst Brewing Company to introduce a pepper-flavored beer were unsuccessful (cf. Naj, 1992).

	II. Introduction

Top Previous Next References

If one compares the two previous capsaicin reviews in this Journal (Buck and Burks, 1986; Holzer in 1991), one can notice the steady evolution of ideas, for instance, the identification of ultrapotent capsaicin analogs (cf. Szallasi and Blumberg, 1990a) or the development of the first competitive capsaicin antagonist, capsazepine (Bevan et al., 1991). The past years, however, have witnessed unprecedented advances that have revolutionized this field. Holzer (1991) described the vanilloid receptor (VR)² as a capsaicin-operated conductance, the expression of which is virtually restricted to a distinct subpopulation of primary sensory neurons. This is apparently no longer true. In 1997, a VR, termed VR1, was cloned (Caterina et al., 1997). The emerging concept is that this VR1 functions as a molecular integrator of painful chemical and physical stimuli including noxious heat (>48°C) and low pH (Tominaga et al., 1998). Apparently, it is not capsaicin but heat that has the capability of opening the channel pore of VR1, whereas capsaicin and protons only serve to lower the heat threshold of the receptor. Consequently, even room temperature is able to gate VR1 in the presence of mildly acidic conditions and/or capsaicin (Tominaga et al., 1998). Moreover, the expression of VRs is not restricted to sensory neurons. In addition to several brain nuclei (Ács et al., 1996a; Sasamura et al., 1998), nonneuronal tissues, such as mast cells (Bíró et al., 1998a) and glial cells (Bíró et al., 1998b), may express VRs. There is mounting evidence, both biological and electrophysiological, to suggest heterogeneity within VRs (cf. Szallasi and Blumberg, 1996; Szallasi, 1997). As yet, it is not known whether or not all VR subtypes recognize capsaicin. Therefore, we chose to use in this review the broader term vanilloid-sensitive neuron over the traditional term capsaicin-sensitive neuron.

It has also been found that ligands with little resemblance to the vanillyl group in capsaicin can also act as vanilloids (Szallasi et al., 1996a, 1998a, 1999a), implying that the term VR is somewhat of a misnomer. Terpenoid unsaturated dialdehydes (Szallasi et al., 1996a, 1998a) and triprenyl phenols (Szallasi et al., 1999a) (Fig. 1) are two chemical classes of such newly discovered "vanilloids". At present, resiniferatoxin (RTX; see Fig. 1 for structure), an ultrapotent capsacin analog (Szallasi and Blumberg, 1989a, 1990a), is undergoing clinical trials, where it proves clearly superior to capsaicin (Chancellor, 1997; Cruz et al., 1997a; Lazzeri et al., 1997; Cruz, 1998). At a recent meeting on urinary incontinence (1st International Consultation on Incontinence; Monaco, 1998), a proposal was accepted that a Vanilloid Club should be formed. From cloning to clinic, the vanilloid field is now vibrating with a frenzy of activity, which makes the life of review writers most complicated. For instance, rumor has it that additional VR isoforms have been cloned and an endogenous vanilloid has been isolated.

In this review, we are deliberately focusing on recent developments in the field. The reader can refer to either the above review by Holzer (1991) or the subsequently published book on capsaicin edited by John N. Wood (1993) for detailed background information. Excellent, short reviews are also available from several authors, e.g., Andy Dray (1992), Gábor Jancsó and colleagues (1994), Carlo A. Maggi (1991), or János Szolcsányi (1996), among others. In any case, we have tried to provide limited background information to the degree that it is necessary to provide the context for understanding recent developments.

The literature on vanilloids is vast. Using the subject word capsaicin, a Medline search yielded 2892 publications that have appeared since 1991, the year when the comprehensive capsaicin review by Holzer was published. On average, one paper dealing with capsaicin actions has thus been published every day over the past 7 years. Approximately one tenth of these papers (245 publications since 1991) deal with some aspect of the therapeutic application of capsaicin in humans. Because a complete overview of this vanilloid literature would be overwhelming, this review tries to provide a selective coverage of the literature, with the focus on breakthrough discoveries, emerging concepts, and provocative new ideas.

Capsaicin excites a subset of primary sensory neurons with somata in dorsal root ganglion (DRG) or trigeminal ganglion (Table 1). As a general rule, these vanilloid-sensitive neurons are peptidergic, small diameter (>50 µm) neurons, giving rise to thin, unmyelinated (C) fibers (cf. Holzer, 1991). Among sensory neuropeptides, the tachykinin Substance P (SP) shows the best correlation with vanilloid sensitivity (cf. Buck and Burks, 1986; Holzer, 1991). However, on the one hand, not all small diameter DRG neurons respond to capsaicin, and, on the other hand, large diameter vanilloid-sensitive neurons, predominantly of the A-type, are also known to exist (Table 1). Indeed, in certain tissues such as the tooth pulp, A-fibers predominate among the vanilloid-sensitive nerve population (Ikeda et al., 1997). Some small diameter sensory neurons are polymodal nociceptors, whereas neurons with A-fibers may function as mechanoheat-sensitive nociceptors (cf. Meyer et al., 1994). In other words, vanilloid-sensitive neurons are heterogeneous morphologically, neurochemically, and functionally, and they encompass several subclasses of DRG neurons (Table 1) (cf. Holzer, 1988, 1991; Szolcsányi, 1996; Holzer and Maggi, 1998). Because sensitivity to vanilloids is the only known trait that all of these neurons seem to share, they are best described as vanilloid-sensitive neurons (cf. Szallasi and Blumberg, 1990a,b; Szallasi, 1996).

According to a recent exciting finding by Seybold and colleagues (Stucky et al., 1998), vanilloid sensitivity is a plastic property of DRG neurons. For instance, the algogenic substance bradykinin is able to recruit intermediate-size neurons (240-320 µm²), normally unresponsive to capsaicin, to respond to vanilloids (Stucky et al., 1998). Consequently, the number of nociceptors that innervate inflamed tissues increases. As we will see later, this novel mechanism may play an important role in the development of inflammatory hyperalgesia via spatial summation on spinal neurons.

Central fibers of vanilloid-sensitive neurons enter the central nervous system (CNS), where they synapse on second-order neurons of the dorsal horn of the spinal cord (for DRG neurons) or the spinal nucleus of the trigeminal tract (for trigeminal ganglion neurons), respectively (cf. Yaksh and Malmberg, 1994). Generally speaking, vanilloid-sensitive neurons transmit noxious information (usually perceived as itching or pain) to the CNS (afferent function), whereas peripherial terminals of these neurons are sites of release for a variety of proinflammatory neuropeptides (efferent function; these neuropeptides are summarized in Table 1). These neuropeptides are believed to play an important role in initiating the cascade of neurogenic inflammation (Fig. 2) (cf. Foreman, 1987; Geppetti and Holzer, 1996). In most experimental paradigms, capsaicin was found to activate both afferent and efferent functions, leading to the adaptation of the "axon reflex" model (cf. Bayliss, 1901; Bruce, 1910; Lewis, 1927; Celander and Folkow, 1953; Lisney and Bharali, 1989) by capsaicin researchers (cf. Holzer, 1988; Maggi and Meli, 1988; Burnstock, 1990; Lynn and Cotsell, 1991). However, it has recently been shown that capsaicin is capable of releasing sensory neuropeptides such as SP, somatostatin, and calcitonin gene-related peptide (CGRP) from the peripheral endings of sensitive nerves in the presence of lignocaine, tetrodotoxin, -conotoxin, or agatoxin, suggesting a direct mechanism for peptide release not mediated by axon reflex (Szolcsányi et al., 1998a).

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Schematic illustration of the role of peripheral vanilloidsensitive nerve endings in evoking neurogenic inflammatory and allergic-hypersensitivity reactions. Reproduced with permission from Szallasi and Blumberg, 1993b. Vanilloid-sensitive nerves may be stimulated to release prestored proinflammatory neuropeptides by both exogenous and endogenous stimuli. Some of these agents, like bradykinin, have their own receptors; others may act on VRs. Protons are unique in that they have their own receptors (called acid-sensitive ion channels or ASICs) but they act also on VRs. The competitive VR antagonist capsazepine ameliorates carrageenan-induced inflammation in vivo, implying a role for an endogenous vanilloid in initiating the inflammatory cascade. Generally speaking, the tachykinin SP released from vanilloid-sensitive nerves causes smooth muscle cells to contract (e.g., bronchospasm) and opens endothelial gaps (plasma extravasation) by interacting at NK-1Rs. Also, SP can stimulate mucus secretion and activate various inflammatory cells. The predominant effect of CGRP is vasodilation. There are several important positive feedback mechanisms involved in neurogenic inflammation. For example, SP released from vanilloid-sensitive nerves activates mast cells. Mast cells liberate histamine, which, in turn, stimulates vanilloid-sensitive nerves to release more SP. It is easy to visualize how the defunctionalization of sensory nerves by vanilloids may prevent, or at least ameliorate, neurogenic inflammatory symptoms. For further details, see text.

In addition to capsaicin, vanilloid-sensitive neurons are also activated by a variety of chemical and physical (both heat and pressure) stimuli (cf. Maggi, 1991; Lundberg, 1993). Some of these compounds, like histamine and bradykinin, have their own receptors. Others (for example, xylene and mustard oil) are believed to work in a non-receptor-mediated fashion (N. Jancsó et al., 1968), possibly by perturbing membranes. Protons (low pH) are unique in that they have their own receptors (known as acid-sensitive ion channels or ASICs) (Waldmann et al., 1997), but, at the same time, they also act on VRs (capsaicin) (Bevan and Yeats, 1991; Petersen and LaMotte, 1993; Liu and Simon, 1994; Martenson et al., 1994; Kress et al., 1996a; Caterina et al., 1997; Tominaga et al., 1998). Finally, there are compounds (e.g., sesquiterpene unsaturated dialdehydes; Szallasi et al., 1996a) and, more surprisingly, physical stimuli (e.g., noxious heat; Caterina et al., 1997) that appear to stimulate sensory neurons in a VR-mediated fashion. Again, it needs to be emphasized that VR is "a target" but not "the target" for noxious heat, because the overlap between heat- and capsaicin-sensitive DRG neurons is only partial (Cesare and McNaughton, 1996; Kirschstein et al., 1997; Reichling and Levine, 1997).

Among irritant compounds acting on primary sensory neurons, capsaicin and related vanilloids are unique in that the initial stimulation by vanilloids is followed by a lasting refractory state, traditionally termed desensitization (N. Jancsó and A. Jancsó, 1949; N. Jancsó, 1955; 1968; cf. Szolcsányi, 1984; G. Jancsó, 1994). As discussed later, this desensitization is a complex process and has a clear therapeutic potential.

Although capsaicin was regarded as a "remarkably selective tool for primary sensory neurons" (Holzer, 1991), it was clear from the beginning that not all capsaicin actions can be attributed to the activation of primary sensory neurons. The lack of probes to detect VRs has led to considerable confusion as to what constitutes vanilloid-sensitive targets. This confusion persists to the present. There was little debate that there are vanilloid-sensitive neurons in nodose ganglia (Szolcsányi and Barthó, 1978, 1982; Marsh et al., 1987; Raybould and Taché, 1989; Waddell and Lawson, 1989; Sharkey et al., 1991; Carobi, 1996). The actual existence of these neurons was recently confirmed by both [³H]RTX binding (Fig. 3) (Szallasi et al., 1995a) and VR1 mRNA in situ hybridization studies (Helliwell et al., 1998a). It was also generally accepted that intrinsic sensitive neurons located in the hypothalamus can mediate the well known effects of vanilloid administration on temperature regulation (Jancsó-Gábor et al. 1970; Szolcsányi et al., 1971). Recent experiments, with both RTX binding (Fig. 4) (Ács et al., 1996a) and reverse transcription-polymerase chain reaction using primers based on the VR1 sequence (Sasamura et al., 1998), have confirmed the existence of these VR-expressing hypothalamic neurons. Other capsaicin actions turned out to be mediated by targets unrelated to VRs (cf. Holzer, 1991; Szallasi, 1994). Nonetheless, the recognition that nonneuronal tissues may express VRs (Bíró et al. 1998a,b) implies that capsaicin effects previously considered to be nonspecific may, in fact, be mediated by VRs.

View larger version (91K): [in this window] [in a new window]   Fig. 3.   Visualization by [3H]RTX autoradiography of vanilloidsensitive neurons in the rat. Note the intense labeling in dorsal root (b, upper five samples), trigeminal (b, lower two samples), and nodose (c) ganglia, containing cell bodies of vanilloid-sensitive neurons. Central fibers of DRG, trigeminal ganglion, and nodose ganglion neurons terminate in the dorsal horn of the spinal cord (a, small arrowheads), the spinal trigeminal nucleus of the medulla oblongata (a, big arrowhead), and the area postrema/nucleus of the solitary tract (a, open arrow), respectively. Peripheral axons of DRG and nodose ganglion neurons traverse in the sciatic (d) and vagus (c) nerves, respectively. Following ligation, there is a marked accumulation of specific binding sites proximal to the ligature in the vagus (c) or sciatic (d) nerves, suggesting that VRs are transported intraaxonally to the periphery in a form capable of ligand binding. Reprinted with permission from Szallasi, 1995.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   VRs in the human brain, as detected by specific [3H]RTX binding. For comparison, in the first column we show the density of specific RTX binding sites in the dorsal horn of the human spinal cord (SC), the central termination site for vanilloid-sensitive neurons. In three nuclei (PA, preoptic area; LC, locus ceruleus; MH, medial hypothalamus) the density of RTX binding sites approaches one-third of that in the dorsal horn. The presence of VRs in these nuclei is expected, as they were reported to respond to capsaicin in vivo. A low density (or affinity) but reproducible RTX binding was detected in additional two areas, the reticular formation (RF) and the ventral nucleus of the thalamus (VT). A very low level of binding was found in the midbrain central gray matter (CG). Finally, no specific binding could be detected in the somatosensory cortex (SSC). Reproduced with permission from Ács et al., 1996a.

From the beginning, three lines of strong evidence pointed to the existence of a specific capsaicin recognition site. First, capsaicin-like activity required fairly strict structure-activity relations (Szolcsányi and Jancsó-Gábor, 1975, 1976; Szolcsányi, 1982; Hayes et al., 1984; Walpole and Wrigglesworth, 1993). Second, capsaicin sensitivity seemed to be restricted to well-defined neuronal tissues (cf. Buck and Burks, 1986; Holzer, 1991). Third, susceptibility to capsaicin showed striking species-related differences. Most authorities agreed that capsaicin sensitivity occurred only in mammals, and even mammalian species differed considerably in their responsiveness to capsaicin (cf. Buck and Burks, 1986; Holzer, 1991). Furthermore, corroborative evidence for the existence of a capsaicin receptor was the finding that the inorganic dye ruthenium red was able to block the capsaicin-induced responses (cf. Amann and Maggi, 1991).

If sufficiently high capsaicin doses are used, the above tissue and species specificities of capsaicin actions are, however, lost. For example, Ritter and Dinh (1993) demonstrated capsaicin-evoked silver staining along almost the entire neural axis of the rat, including the retina. In keeping with this, Szikszay and London (1988) described enhanced glucose use by capsaicin in a wide variety of CNS structures in the rat. Capsaicin was found to inhibit various enzymes (Shimomura et al., 1989; Yagi, 1990; Teel, 1991; Wolvetang et al., 1996), induce pseudochannel formation in lipid bilayers (Feigin et al., 1995), alter membrane fluidity (Aranda et al., 1995), and block K⁺ channels (Dubois, 1982; Petersen et al., 1987; Bleakman et al., 1990; Castle, 1992; Baker and Ritchie, 1994; Kuenzi and Dale, 1996), just to name a few characteristic, non-VR-mediated capsaicin actions.

Early efforts to demonstrate specific binding sites for [³H]dihydrocapsaicin (Szebeni et al., 1978) or photoaffinity-labeled capsaicin-like molecules (James et al., 1988) failed due to a combination of the high lipophilicity and relatively low affinity of these molecules.

	III. Direct Evidence for a Vanilloid (Capsaicin) Receptor

Top Previous Next References

It has been known since the dawn of recorded history that the latex of the Moroccan cactus-like plant, Euphorbia resinifera, contains an extremely irritant component (Fig. 5) (cf. Appendino and Szallasi, 1997). But it was not until 1975 that the compound responsible for this irritancy was isolated and named RTX (Hergenhahn et al., 1975). RTX (see structure in Fig. 1) combines the structural features of two classes of natural irritants, phorbol esters and capsaicinoids. Although RTX did bind to the phorbol ester receptor, protein kinase C (PKC) (see in Section XIII.B), this low-affinity interaction could not explain its extreme pungency. In a series of experiments beginning in 1989 and continuing to date, we have identified RTX as an ultrapotent capsaicin analog with a unique spectrum of biological activities (Table 2). Because capsaicin and RTX analogs share a (homo)vanillyl group as a structural motif essential for bioactivity but differ dramatically in the rest of the molecule (Fig. 1), they are collectively termed vanilloids. Consequently, the primary target for these compounds appears to be best described as the VR.

View larger version (140K): [in this window] [in a new window]   Fig. 5.   The collection of the latex of an euphorbia plant as depicted in Codex Ayasofia (3103, f. 136.1.3, Freer Gallery of Art, Smithsonian Institution, Washington, DC), an Arabic version of De Materia Medica by Dioscorides. During antiquity, the dried latex, called euphorbium, was used as a vesicant (skin irritation) and sternutative (nose irritation) agent. It was also given as a general remedy for various snakebites and poisons.

In several assays, RTX is several thousandfold more potent than capsaicin (Table 2) (cf. Szallasi and Blumberg, 1990a, 1993b, 1996). This ultrapotency predicted the existence of high-affinity specific RTX binding sites (VRs). Despite multiple obstacles, in 1990 we finally managed to demonstrate specific [³H]RTX binding by rat DRG membranes (Szallasi and Blumberg, 1990b), providing the first unequivocal evidence for the existence of VRs (Fig. 6). This initial binding assay was, however, plagued by a very high nonspecific binding, which prevented the detection of VRs in the spinal cord or peripheral tissues.

View larger version (25K): [in this window] [in a new window]   Fig. 6.   Scatchard plots of specific [3H]RTX binding to rat DRG membranes. This binding is of high affinity (Kd is 20 pM) and follows positive cooperative behavior, as reflected in the convex Scatchard plot. The cooperativity index is 2, indicating the existence of at least two interacting binding sites. Capsaicin and capsazepine, both at a concentration of 3 µM, reduce the apparent affinity of RTX binding, having little or no effects on cooperativity or maximal receptor density. This behavior is consistent with a competitive binding mechanism.

The methodological means to overcome the problems created by the high nonspecific RTX binding was furnished by an accidental observation of ours. In search of endogenous ligands for VRs, we assayed a variety of body fluids and tissue extracts and found that the acute phase serum protein, ₁-acid glycoprotein (also known as orosomucoid), binds RTX (Szallasi et al., 1992). ₁-Acid glycoprotein is an important drug-binding protein (Paxton, 1983; Kremer et al., 1988; Maruyama et al., 1990). We showed that RTX binds to the well known drug-recognition domain on ₁-acid glycoprotein that it shares with warfarin (Szallasi et al., 1992). Plasma binding of RTX to ₁-acid glycoprotein may have clear consequences on pharmacokinetics upon systemic administration that we will discuss later. RTX binding to ₁-acid glycoprotein, however, differs in two very important aspects from VR binding. First, it is not markedly influenced by temperature, and second, it is of an at least 1000-fold lower affinity (Szallasi et al., 1992). At 0°C, the dissociation of receptor-bound RTX is unmeasurably slow (Szallasi and Blumberg, 1993a). By contrast, ₁-acid glycoprotein readily binds unbound RTX in the aqueous phase (Szallasi et al., 1992). Because nonspecifically bound RTX in the membrane lipids is in equilibrium with the unbound RTX in the aqueous phase, ₁-acid glycoprotein is able to extract nonspecifically bound RTX from the membranes without compromising specific binding (Szallasi et al., 1992). By centrifuging the membranes, it is then easy to separate the membrane-, mostly receptor-bound RTX from the ₁-acid glycoprotein-bound form that remains in the supernatant.

The introduction of ₁-acid glycoprotein to the VR binding assay resulted in a dramatic improvement in the ratio of specific binding from 50% of the total at concentrations close to the K_d to 90% of the total binding or even higher. Using this improved assay, we are now able to detect specific RTX binding sites in the dorsal horn of the spinal cord (Szallasi et al., 1993a,b; Ács and Blumberg, 1994; Ács et al., 1994a,b; Szallasi and Goso, 1994), in various peripheral tissues [e.g., urinary bladder (Szallasi et al., 1993c; Ács et al., 1994a), urethra (Parlani et al., 1993), nasal mucosa (Rinder et al., 1996), airways (Szallasi et al., 1993b, 1995b), colon (Goso et al., 1993a)] as well as in several brain nuclei (Fig. 4 and Table 3) (Ács et al., 1996a). Furthermore, we developed an autoradiographic approach to visualize VRs in several species, including human (Figs. 3 and 7) (cf. Szallasi, 1995).

View larger version (133K): [in this window] [in a new window]   Fig. 7.   Autoradiographic visualization by [3H]RTX binding of VRs in porcine (A, small arrowheads) and human (B) dorsal horn of the spinal cord, the central termination site for vanilloid-sensitive neurons. The labeling is highly specific, because it is completely missing in the presence of nonradioactive RTX (C). Reprinted with permission from Szallasi et al., 1994a.

Capsazepine (Fig. 8), the first and, as yet, only commercially available competitive VR antagonist comes from an extensive program at the former Sandoz, now Novartis, Institute for Medical Research, London, to explore structure-activity requirements for vanilloid-like activity (cf. Walpole and Wrigglesworth, 1993). Capsazepine inhibits vanilloid responses in vitro with Schild plots consistent with a competitive mechanism (Bevan et al., 1991, 1992). Moreover, capsazepine competes for specific RTX binding sites in a competitive manner (Szallasi et al., 1993b). In rat trigeminal ganglion neurons, capsazepine inhibits vanilloid-evoked currents (Liu and Simon, 1994). In a variety of bioassays, capsazepine is effective against both capsaicin (Dickenson and Dray, 1991; Urbán and Dray, 1991; Belvisi et al., 1992; Perkins and Campbell, 1992; Maggi et al., 1993; Santicioli et al., 1993; Seno and Dray, 1993; Ueda et al., 1993; Lee and Lundberg, 1994; Lalloo et al., 1995; Fox et al., 1995) and RTX (Ellis and Undem, 1994; Walpole et al., 1994; Ács et al., 1996b, 1997; Wardle et al., 1996, 1997) but, surprisingly, not against olvanil (Davey et al., 1994). The utility of capsazepine is, however, limited by its moderate potency. At micromolar concentrations, which are necessary to inhibit capsaicin-evoked responses in most tissues, capsazepine also blocks voltage-gated calcium channels (Docherty et al., 1997) as well as nicotinic acetylcholine receptors (Liu and Simon, 1997). Furthermore, recent evidence suggests that capsazepine-insensitive VRs also exist (Liu et al., 1998). Therefore one should be very careful when interpreting results obtained with capsazepine, because positive effects are not necessarily mediated by VRs, nor do negative data rule out the involvement of VRs. One more reason for being cautious when working with capsazepine was furnished by the observation that, at least in the rabbit, it may act as a weak vanilloid agonist (Wang and Håkanson, 1993).

View larger version (17K): [in this window] [in a new window]   Fig. 8.   Selected vanilloid structures. Capsazepine is a competitive VR antagonist. Eugenol in an analgesic used in dental practice. Olvanil is a nonpungent, orally active capsaicin analog. Compound 57 is the most potent capsaicinoid for inducing Ca2+-uptake by DRG neurons in culture. PPAHV binds to VRs in a noncooperative fashion. Also, it gates two pharmacologically distinct conductances in rat DRG neurons, one that is inhibited by capsazepine and another that does not recognize this antagonist. PDDHV induces Ca2+-uptake by DRG neurons with a potency of 15 nM; however, it fails to inhibit [3H]RTX binding to these cells up to a concentration of 10 µM. This finding implies that the RTX binding domain on VRs is distinct from the site mediating calcium influx.

Repeated efforts to clone a VR using RTX-like photoaffinity probes resulted in the identification of several, relatively low-affinity RTX-binding proteins, none of which showed the expected tissue distribution for VRs, nor did they show a VR-like activity in functional assays (Ninkina et al., 1994; Davies et al., 1997). David Julius' group at the University of California in San Francisco chose, therefore, a different approach. They transfected eukaryotic cells with pools of a rat cDNA library and used calcium imaging to identify those cells that responded to capsaicin (Caterina et al., 1997). Once a positive pool was found, it was divided into smaller pools (a procedure known as sib-selection) until they had isolated a single cDNA encoding the capsaicin-gated channel. They named this receptor VR1.

The rat VR1 cDNA contains an open reading frame of 2514 nucleotides. This cDNA encodes a protein of 838 amino acids with a molecular mass of 95 kDa. At the N terminus, VR1 has three ankyrin repeat domains (Fig. 9A). The carboxy terminus has no recognizable motifs. Predicted membrane topology of VR1 features six transmembrane domains and a possible pore-loop between the fifth and sixth membrane-spanning regions (Fig. 9A). There are three possible protein kinase A phophorylation sites on the VR1 that might play a role in receptor desensitization.

View larger version (41K): [in this window] [in a new window]   Fig. 9.   A, predicted membrane topology and domain structure of the first cloned VR, called VR1. VR1 has six complete transmembrane segments and a partial one, which is believed to be associated with the channel pore. As indicate ankyrin repeat domains. Outer (o) and inner (i) plasma membrane leaflets are also indicated. B, rat VR1 shows homology to the Drosophila TRP channel and a related protein in the nematode C. elegans. Several human ESTs, such as T12251 in heart, are also similar to VR1. Identical residues are in black boxes and conservative substitutions are in gray. Reprinted by permission from Nature (Caterina et al., 1997). Copyright (1997) Macmillan Magazines Ltd.

VR1 is a distant relative of the transient release potential (TRP) family of store-operated calcium channels (Montell and Rubin, 1989; Hardie and Minke, 1993; Wes et al., 1995; Clapham, 1996; Colbert et al., 1997; Roayaie et al., 1998). There is considerable homology between VR1 and the drosophila TRP protein in retina (Fig. 9B). This sequence similarity seems to be restricted to the pore-loop and the adjacent sixth transmembrane segment in VR1. Interestingly, VR1 also shows similarity to a Soares human retina cDNA (L. Hillier, N. Clark, T. Dubuque, K. Elliston, M. Hawkins, M. Holman, M. Hultman, T. Kucaba, M. Le, G. Lennon, M. Marra, J. Parsons, L. Rifkin, T. Rohlfing, F. Tan, E. Trevaskis, R. Waterston, A. Williamson, P. Wohldman and R. Wilson, unpublished observations, Washington University-Merck expressed sequence tags (EST) Project; Accession: AA047763). Because capsaicin causes a marked calcium accumulation in rat retina (Ritter and Dinh, 1993), it might be speculated that the retina has a site, related to VR1, that recognizes vanilloids. OSM-9, a novel protein with similarity to rat VR1, plays a role in olfaction, mechanosensation, and olfactory adaptation in Caenorhabditis elegans (Colbert et al., 1997). OSM-9, however, does not recognize capsaicin (Cornelia Bargmann, personal communication). These findings imply that 1) in contrast to previous beliefs, VR isoforms did occur early during evolution, but 2) the capsaicin recognition site is a recent addition to VR1.

It should be noted that a human EST in heart (T12251) displays striking similarity (68% amino acid identity) to the pore-loop and the adjacent sixth transmembrane segment in rat VR1 (Fig. 9B). Additional human EST clones are similar to other regions of the VR1 and could represent fragments of the human transcript. The presence of a VR1-like EST clone in heart is surprising but it is entirely in accord with the recent recognition of nonneuronal VRs (Bíró et al., 1998a,b). Capsaicin has long been known to influence cardiac functions (Fukuda and Fujiwara, 1969; Molnár et al., 1969). Some capsaicin actions on heart were attributed to an interaction at K⁺ channels (Castle, 1992), whereas others were explained by the liberation of neuropeptides, most notably CGRP, from the vanilloid-sensitive innervation of the heart (Franco-Cereceda et al. 1988, 1991; Ono et al., 1989). It is not impossible that capsaicin can also act directly on the heart via a cardiac VR.

When expressed in Xenopus oocytes, VR1 is similar in its electrophysiological properties to native vanilloid-operated channels in sensory ganglia (Caterina et al., 1997). As observed in cultured DRG neurons (Baccaglini and Hogan, 1983; Heyman and Rang, 1985; Forbes and Bevan, 1988; Winter et al., 1990; Vlachová and Vyklicky, 1993; Liu and Simon, 1994; Petersen et al., 1996), capsaicin-evoked currents readily disappear after agonist removal, whereas RTX-induced currents are much longer in duration and often persist even in the absence of the agonist. Hill coefficients (approximately 2) derived from the analysis of capsaicin-induced currents in oocytes injected with mRNAs encoding VR1 indicate the existence of more than one agonist binding site (Caterina et al., 1997). Again, this is in accord with the properties of native VRs in sensory neurons (Szallasi et al., 1993a; Oh et al., 1996). The implications of this finding are discussed in Section VI.J.

RTX is, however, only 20-fold more potent than capsaicin to activate the cloned VR1 (Caterina et al., 1997), which is at variance with the several thousandfold higher affinity of RTX in the binding assay (cf. Szallasi and Blumberg, 1990a, 1996; Szallasi, 1994). This apparent contradiction was first explained by postulating the existence of two distinct classes of VRs, the channel, which represents a low-affinity site for RTX, and a yet-to-be cloned high-affinity site, seen in the binding assay. We referred to these receptors as C-type and R-type VRs, respectively (cf. Ács et al., 1997; Bíró et al. 1997; Szallasi, 1997). However, preliminary binding experiments with human embryonic kidney (HEK)293 cells stably transfected with VR1 cDNA suggests that this is not the case: the same receptor protein seems to mediate both the high-affinity RTX binding and the lower affinity calcium uptake response (A. Szallasi, D. N. Cortright, P. M. Blumberg, and J. E. Krause, manuscript in preparation).

	IV. Anatomical Localization and Tissue Specificity of VRs

Top Previous Next References

As first shown by RTX autoradiography (cf. Szallasi, 1995), VRs are expressed along the entire length of vanilloid-sensitive sensory neurons, from the peripheral terminals to the axons to the somata to the central endings (Fig. 3 and Table 3). In corresponding areas (compare Figs. 3 and 10) the presence of VR1-like immunoreactivity has also been demonstrated (Guo et al., 1999; Tominaga et al., 1998). Nerve ligation studies suggest that VRs are transported from the cell bodies to the periphery in a form capable of ligand binding (Fig. 3) (Szallasi et al., 1995a). Peripheral tissues in which VRs were demonstrated by RTX binding to be present are listed in Section III.A and in Table 3. In the urinary bladder, VRs seem to be expressed exclusively on nerve endings rather than in bladder epithelium or the muscle layer, for denervation of the bladder leads to a complete depletion of specific RTX binding sites (Szallasi et al., 1993d). The majority (approximately 90%) of the RTX sites appear to be present on the pudendal nerve, with a relatively minor fraction on the hypogastric nerve (Szallasi et al., 1993d).

View larger version (130K): [in this window] [in a new window]   Fig. 10.   Visualization by immunohistochemistry of VR1 receptor-like immunoreactivity in central (A and C) and peripheral (E and G) terminals of vanilloid-sensitive neurons in the rat. Compare this figure to Fig. 3 and notice the colocalization of specific RTX binding sites and VR1 in the nucleus of the solitary tract (A, arrows), the spinal trigeminal nucleus (A, arrow heads), and the superficial dorsal horn of the spinal cord (C). VR1-like immunoreactivity is also present in the sciatic nerve (F), the paw skin (E), and the cornea (G). The specificity of the immunostaining is demonstrated by peptide blocking (B), and by the absence of labeling following unilateral dorsal rhizotomy (D). Reprinted with permission from the European Journal of Neuroscience (Guo et al., 1999).

Northern blot analysis confirmed the presence of VR1 transcripts in dorsal root and trigeminal ganglia (Caterina et al., 1997; Helliwell et al., 1998a). VR1-like immunoreactivity was detected in more than 50% of DRG neurons (Fig. 11), with the expression being most prevalent in small to medium sized neurons (Guo et al., 1999). VR1-like immunoreactivity was also observed in both the central (e.g., dorsal horn of the spinal cord and the caudal nucleus of the spinal trigeminal complex) and peripheral (e.g., skin and cornea) processes of primary afferent neurons (Fig. 10) (Guo et al., 1998). In DRG neurons, VR1-like immunoreactivity was associated with the Golgi complex and the plasma membrane (Guo et al., 1999). In the dorsal horn of the spinal cord, the VR1 protein was associated with "small clear vesicles" in preterminal axons and with the plasma membrane of nerve terminals (Guo et al., 1999). The latter finding is in line with an earlier observation of Szolcsányi (Szolcsányi et al., 1975), according to which capsaicin depletes "small clear vesicles" from nerve endings.

View larger version (101K): [in this window] [in a new window]   Fig. 11.   Colocalization of VR1-like immunoreactivity in the spinal dorsal horn (A, B, D, E, G, H, J, and K) and in DRG neurons (C, F, I, and L) with the lectin IB4 (A, B, and C), the P2X3 receptor (D, E, and F), and the neuropeptides SP (G, H, and I) and CGRP (J, K, and L). This figure suggests that vanilloid-sensitivity encompasses several subpopulations of DRG neurons, both peptidergic and nonpeptidergic. However, none of the markers tested (IB4, P2X3, SP, and CGRP) is present exclusively in VR1-expressing neurons. Note that the intensity of VR1 receptor-like immunostaining is strongest in the substantia gelatinosa (lamina I) and the inner layer of Rexed lamina II. Reproduced with permission from the European Journal of Neuroscience (Guo et al., 1999).

The distribution of VR1-like immunoreactivity in the spinal dorsal horn deserves particular attention. The labeling is strongest in the Lissauer zone (lamina I; Fig. 10C). VR1 protein is also abundant in the inner, but not in the outer, layer of lamina II (Fig. 10C) (Guo et al., 1999). [There is an apparent discrepancy between this finding and the even distribution of VR1-like immunoreactivity in the substantia gelatinosa reported by Julius and colleagues (Tominaga et al., 1998), the reason for which is not clear.] It should be noted that several important proteins involved in pain transmission are also enriched in the inner layer of lamina II (i.e., in close proximity to VR1). Notable examples include PKC isozyme  (PKC) (Malmberg et al., 1997a) as well as ATP-sensitive P2X₃ receptors (Vulchanova et al., 1998). P2X₃ receptors colocalize with VR1 both in DRGs, where 75% of the P2X₃ positive neurons also express VR1 (Fig. 11F), and in the inner layer of lamina II (Fig. 11, D and E) (Guo et al., 1999). By contrast, PKC is confined to a population of interneurons that reside in lamina II of the dorsal horn (Malmberg et al., 1997a). The possible role of PKC and P2X₃ receptors in nociception is discussed in Section VIII.

Surprisingly, the colocalization of VR1 with SP in DRG neurons is limited: for example, only 33% of L5 DRG neurons positive for VR1 contain SP-like immunoreactivity as well (Fig. 11I) (Guo et al., 1999). Moreover, much of the VR1 staining in the dorsal horn is concentrated in the inner layer of lamina II where both SP- (Fig. 11, G and H) and CGRP-like immunoreactivities are sparse (Fig. 11, J and K) (Guo et al., 1999). This limited colocalization between VR1 and SP is at variance with the profound effects that vanilloids have on SP expression (see in Section VIII.B.5).

The colocalization of VR1 with the lectin IB4 (Fig. 11, A-C) (Guo et al., 1999; Tominaga et al., 1998) may shed new light on the regulation of VR expression by trophic factors. Vanilloid-sensitive neurons require nerve growth factor (NGF) for survival during embryogenesis (Ruit et al., 1992), as evidenced by a severe deficit in nociception and thermoreception in mice with a null mutation in the gene encoding trkA, the signal transduction receptor for NGF (Smeyne et al., 1994). Mature DRG neurons, however, are able to survive in the absence of NGF (Yip et al., 1984; Lindsay, 1988). There is evidence that those neurons that bind IB4 also express receptors for glial cell-derived neurotrophic factor, abbreviated GDNF (Bennett et al., 1998). In 1997, Snider and coworkers (Molliver et al., 1997) showed in elegant experiments that IB4-positive DRG neurons switch from dependence on NGF to dependence upon GDNF during development. Subsequently, McMahon and colleagues (Bennett et al., 1998) demonstrated that IB4-positive neurons may be rescued by exogenous GDNF following nerve injury. Clearly, it would be interesting to explore the effects of GDNF on VR expression.

In addition to DRG (Figs. 3 and 7) and trigeminal ganglion (Fig. 3) neurons (Szallasi and Blumberg, 1990b; Szallasi et al., 1993a,b, 1994a; Ács et al., 1994a), a subset of nodose ganglion neurons also expresses VRs (Fig. 3 and Table 3) (Ács et al., 1994a; Szallasi et al., 1995a). A strong autoradiographic signal is present in the area postrema/nucleus of the solitary tract region (Fig. 3), representing the central termination area for sensory neurons of the vagus nerve (Szallasi et al., 1995a). This area is also stained by an anti-VR1 antibody (Fig. 10A; Guo et al., 1999).

An unexpected result in the VR1 cloning article by Julius and colleagues (Caterina et al., 1997) is the failure of Northern blot hybridization to detect mRNAs encoding VR1 in nodose ganglia. However, using a probe corresponding to nucleotides 1513-2482 of the rat VR1 sequence, a strong in situ hybridization signal was detected in nodose ganglion neurons (Helliwell et al., 1998). To resolve the apparent contradiction between these studies, it should be noted that whereas Julius and coworkers (Caterina et al., 1997) used the entire VR1 sequence for Northern blot hybridization, Bevan and colleagues (Helliwell et al., 1998) generated by polymerase chain reaction a partial VR1 sequence only. Therefore, it is entirely possible that nodose ganglion neurons express a VR isoform that differs from VR1 in nucleotides not included in the probe used for in situ hybridization histochemistry. The existence of distinct VR isoforms in DRGs and nodose ganglia would be consistent with the different embryonic origin of these tissues (cf. Vogel, 1992). It would also be in accord with the observations that different neurotrophic factors regulate capsaicin sensitivity in DRG (NGF and possibly also GDNF) and nodose ganglion neurons (brain-derived neurotrophic factor), respectively (Winter et al., 1988; Winter, 1998).

A low to moderate level of specific RTX binding can be detected in various CNS areas not associated with primary sensory neurons (Ács et al., 1996a), suggesting the existence of intrinsic brain neurons with VRs (Fig. 4). Among these brain areas, the presence of VRs in the hypothalamus was expected based on the proposed role of this structure in the hypothermic action of vanilloids (Jancsó-Gábor et al., 1970; Szolcsányi et al., 1971). The presence of VR1 mRNA in the hypothalamus has recently been confirmed by reverse transcription-polymerase chain reaction (Sasamura et al., 1998). These vanilloid-sensitive terminals in the hypothalamus are probably glutaminergic because capsaicin evokes glutamate release from slices of hypothalamus (Sasamura et al., 1998). It is also easy to visualize how VRs in the reticular formation (Ács et al., 1996a) may mediate some vanilloid actions on autonomic regulation (Jancsó and Such, 1985; Koulchitsky et al., 1994; Seller et al., 1997). The biological role of specific RTX binding sites in other brain areas is unclear.

A link between the SP content of sensory neurons and their sensitivity to capsaicin was postulated as early as the mid-1960s (Gaparovic et al., 1964). In 1976, the high concentration of SP in the substantia nigra was first reported (Brownstein et al., 1976) and the search for capsaicin-sensitive neurons in the basal ganglia began. Five years later, an enhanced locomotor activity in rats following bilateral intranigral capsaicin injection and a concomittant decrease in the cataleptic action of fluphenazine was described (Dawbarn et al., 1981). In 1988, capsaicin microinjected into the substantia nigra or caudatus putamen was reported to induce peripheral vasodilatation and a subsequent hypothermic response (Hajós et al., 1988). These biological actions suggested the existence of intrinsic vanilloid-sensitive cells in the basal ganglia. The recent detection of VR1 mRNA in the striatum now firmly establishes the existence of such neurons (Sasamura et al., 1998). As yet, it is not known which subpopulation of nigral neurons express VRs. Neonatal capsaicin treatment does not deplete SP from the striatum, nor does it alter the expression of the opioid peptides dynorphin and met⁵-enkephalin (Sivam and Krause, 1992). Repeated attempts to elucidate the effects of capsaicin treatment on monoaminergic systems in the brain have yielded controversial results. For example, increased (Holzer et al., 1981) or decreased (Dawbarn et al., 1981) 5-hydroxytryptamine levels, or no change at all (G. Jancsó et al., 1981), were reported in the very same year. With the recent availability of VR1 detecting antibodies as well as probes for in situ hybridization histochemistry, the identity of the VR1-expressing subpopulation of basal ganglion neurons will be revealed shortly.

VR1 mRNA seems to be widely present in the brain. Relative levels are as follows: hypothalamus and cerebellum > cortex, striatum, and midbrain > olfactory bulb, pons, hippocampus, and thalamus (Sasamura et al., 1998). Although Northern blot analysis failed to show VR1 mRNA in the brain (Caterina et al., 1997), this discrepancy may be due to the difference in sensitivity between polymerase chain reaction and Northern blot hybridization. In support of this explanation is the detection of VR1 mRNA in the rat cortex in solution hybridization-nuclease protection experiments (D. N. Cortright and J. E. Krause, personal communication). VR1-positive cells may also be visualized in the brain by both in situ hybridization and immunostaining experiments (E. Mezey, A. Guo, D. N. Cortright, R. Elde, J. E. Krause, P. M. Blumberg, and A. Szallasi, manuscript in preparation).

It has been long noted that capsaicin exerts a variety of effects on nonneuronal tissues (cf. Buck and Burks, 1986; Holzer, 1991). These actions were considered nonspecific because they were at variance with the prevailing concept of capsaicin being selective for sensory neurons. In 1998, we showed that vanilloids induce calcium uptake by mast cells (Bíró et al., 1998a) and in a glioma cell line (Bíró et al., 1998b). The pharmacological characteristics of this calcium influx response by mast cells and glioma cells were very similar to those described in DRG neurons (Table 4), with the exception of the magnitude of the uptake, which was much smaller, implying a VR density lower than in sensory neurons. In keeping with this, no specific [³H]RTX binding by mast cells or glioma cells could be detected (Table 4). We will return to the biological relevance of such nonneuronal vanilloid interactions later. The novel concept of nonneuronal VRs has received support from the presence of EST clones homologous to VR1 in the heart and other nonneuronal tissues (Caterina et al., 1997).