Abstract
The pharmacological properties of the α2-adrenergic receptors regulating the release of norepinephrine were investigated in human neocortex. Slices were preincubated with [3H]norepinephrine, superfused under blockade of transmitter reuptake, and stimulated electrically. First, the autoinhibitory circuit of [3H]norepinephrine release was analyzed quantitatively by estimation of theKd of norepinephrine at the α2-autoreceptor (10−7.99 M), the concentration of the endogenous transmitter causing this autoinhibition at a stimulation frequency of 3 Hz (10−7.61 M), and the maximum inhibition obtainable through the autoreceptor (83%). Second, antagonist pKb values of nine antagonists were determined by using their pEC50 values (negative logarithms of antagonist concentrations that increased the electrically evoked overflow of tritium by 50%) against the release-inhibiting effect of the endogenous transmitter. When compared with binding or functional data from the literature, the pKbvalues correlated best with the antagonist affinities at α2A binding sites. In contrast, the correlations with α2B, α2C, and α2D sites were not as good. It is concluded that in human neocortex prejunctional autoreceptors are α2A.
In 1992, Raiteri and colleagues concluded that the presynaptic α2-autoreceptors that modulate norepinephrine release in the human neocortex are distinct from α2B and α2C and are either α2A or α2D. This conclusion was mainly based on three findings: the potent release-inhibiting effect of oxymetazoline; the potent antagonism by yohimbine as opposed to the weak antagonism by prazosin and 2-{2-[4-(o-methoxyphenyl)piperazin-1-yl]ethyl}-4,4-dimethyl-1,3(2H,4H)-isoquinolinedione (ARC 239) of the release-inhibiting effect of clonidine; and the marked release-enhancing effect of yohimbine as opposed to the lack of a release-enhancing effect of prazosin and ARC 239 when these antagonists were given alone. All observations were in accord with the affinities of oxymetazoline, yohimbine, prazosin, and ARC 239 for α2A- and α2D-adrenoceptors but were not compatible with their affinities for α2B- and α2C-adrenoceptors (Raiteri et al., 1992).
It is now thought that the α2A- and α2D-adrenoceptors are species orthologs, of which only one occurs in a given species, and that humans possess the α2A version, whereas rodents possess the α2D version (see Bylund, 1995). Human neocortical α2-autoreceptors therefore should be α2A. If so, they would obey the rule that the main mammalian α2-autoreceptors belong to the α2A/D branch of the adrenoceptor tree (Trendelenburg et al., 1993, 1999; see Docherty, 1998). Recent evidence indicates, however, that noradrenergic neurons in addition may possess α2C-autoreceptors (Trendelenburg et al., 1997; see Docherty, 1998; Ho et al., 1998; Altman et al., 1999).
Because the identification of a receptor type by means of an agonist (oxymetazoline in the work of Raiteri et al., 1992) is ambiguous and because these authors used only three antagonists, we reinvestigated the subtype to which the α2-autoreceptors belong in the human neocortex. For this purpose, we quantified the release-enhancing effect of nine antagonists, including prazosin and ARC 239. The concentration-response data were evaluated by fitting a logistic function, which yielded the maximal enhancement and the EC50 of the antagonist. The effect of exogenous norepinephrine under autoinhibition-free as well as autoinhibition conditions was also studied. The evaluation of the norepinephrine concentration-response curves suggested proportionality between α2-autoreceptor occupation and response and allowed the conversion of the EC50 values of the antagonists into their dissociation constants,Kb values, at the autoreceptors. Thus, the subclassification of the α2-autoreceptor in human neocortex was possible, based on functionally definedKb values of antagonists in comparison with corresponding values from the literature. The conversion of EC50 to Kbjustified the use of the functional parameterKb as an estimate of the antagonist dissociation constant. Thus, the bias or shift between EC50 values and dissociation constants of binding experiments could be bridged.
Materials and Methods
Fresh neocortical tissue was obtained from patients during surgical access to subcortical tumors. The procedure was approved by the local Ethics Committee. The patients (n = 42) were of either sex and were between 21 and 86 years old. After premedication with midazolam or chlordiazepoxide, patients were anesthetized with thiopental, fentanyl, or flunitrazepam. Pancuronium was given for muscle relaxation. The tissue was immersed in ice-cold medium (see below) and processed immediately.
Cortical slices, 350 μm thick and perpendicular to the surface, were incubated with 0.1 μM (−)-[3H]norepinephrine in 4 ml of medium for 45 min at 37°C and then superfused with [3H]norepinephrine-free medium at 0.4 ml/min. For electrical stimulation, rectangular pulses of 2-ms width and a voltage drop of 11 V across the electrodes of each superfusion chamber were used, yielding a current strength of approximately 76 mA (Stimulator I; Hugo Sachs Elektronik, Hugstetten, Germany). Four stimulation periods were applied (S1 to S4); they began after t = 75, 110, 145, and 180 min (t = 0 being the start of superfusion). To evoke release free of autoinhibition, each stimulation period consisted of two trains of 4 pulses/100 Hz, with a train interval of 2 min [pseudo-one-pulse conditions; Singer, 1988; Allgaier et al., 1995]. To evoke autoinhibited release, each stimulation period consisted of 90 pulses/3 Hz. Successive 5-min samples of the superfusate were collected from t = 60 min onward. Unlabeled norepinephrine (tested under pseudo-one-pulse conditions as well as at 90 pulses/3 Hz) or α-adrenoceptor antagonists (tested at 90 pulses/3 Hz only) were added at increasing concentrations 15 min before S2, S3, and S4. At the end of experiments, tissues were dissolved, and tritium was determined in superfusate samples and tissues.
The medium used for tissue collection, incubation, and superfusion contained 118 mM NaCl, 1.8 mM KCl, 1.3 mM CaCl2, 1.2 mM MgSO4, 25 mM NaHCO3, 1.2 mM KH2PO4, 11 mM glucose, and 0.57 mM ascorbic acid. It was saturated with a mixture of 95% O2 and 5% CO2. The superfusion medium also contained 1 μM desipramine or, in experiments with >1 μM exogenous norepinephrine, 10 μM desipramine plus 10 μM (+)-oxaprotiline.
The outflow of tritium was calculated as a fraction of the tritium content of the slice at the onset of the respective collection period (fractional rate; min−1). The overflow elicited by electrical stimulation was calculated as the difference: total tritium outflow during the collection period in which stimulation was applied and during the two collection periods thereafter minus estimated basal outflow; basal outflow was assumed to decline linearly from the collection period before to the collection period 10 to 15 min after onset of stimulation. The evoked overflow was then expressed as a percentage of the tritium content of the slice at the time of stimulation. For further evaluation, ratios were calculated for the overflow evoked by S2, S3, and S4 and the overflow evoked by S1. Moreover, effects of exogenous norepinephrine and of the antagonists were calculated for each single slice as a percentage of control, using the corresponding mean average control S2/S1, S3/S1, and S4/S1 ratios (solvent-treated slices, no agonist, no antagonist) as the reference. Drug effects on the basal efflux of tritium were evaluated similarly, based on values immediately before stimulation periods (b1, etc.).
Concentration-response data were evaluated as follows. In the case of the inhibitory effect of exogenous norepinephrine under autoinhibition-free conditions, a logistic function was fitted to the “percentage of control” data to yield the maximal effect of norepinephrine Imax observed, its IC50, and the slope parameter c (eq. 7 of Feuerstein and Limberger, 1999). In the case of the effect of exogenous norepinephrine under autoinhibition conditions, a special function was fitted to the data that describes the combined effects of exogenous and endogenous norepinephrine, assumes the proportionality of receptor occupation and the effect of norepinephrine, and yields the maximal obtainable effect of norepinephrine under autoinhibition-free conditions Imax derived, the dissociation constant Kd of the norepinephrine-autoreceptor complex, and the concentration of released transmitter norepinephrine at the autoreceptors in the absence of exogenous norepinephrine [NEtr] (biophase concentration; eq. 14 of Feuerstein and Limberger, 1999). In the case of the facilitatory effect of the antagonists, we fitted a logistic function to the percentage of control data to obtain theEmax of the antagonist and its EC50 [eq. 7 of Feuerstein and Limberger, 1999, adapted for enhancement instead of inhibition, i.e., Sx/S1 = 1 +Emax × 10−p[B]/(10−pEC50 + 10−p[B]), where p[B] is the negative logarithm of the applied antagonist concentration and pEC50 is the negative logarithm of EC50]. The conversion of antagonist EC50 to antagonist dissociation constantKb was then based on theKd and Imax derived of norepinephrine, determined independently as described above.
Results are given as arithmetic means or estimates with 95% confidence intervals (CI95) in parentheses to indicate statistical probability (Altman, 1991). n is the number of brain slices.
Purchased drugs were (−)-[ring-2,5,6-3H]norepinephrine, specific activity 40.5 Ci/mmol (DuPont, Dreieich, Germany); (−)-norepinephrine hydrogen tartrate, desipramine HCl, corynanthine HCl, (±)-idazoxan HCl (Sigma, Deisenhofen, Germany); (+)-oxaprotiline HCl, phentolamine HCl (Novartis, Basel, Switzerland); spiroxatrine, (±)-2-(2,6-dimethoxyphenoxyethyl)aminomethyl-1,4-benzodioxane HCl (WB4101) (Biotrend, Köln, Germany); rauwolscine HCl (Roth, Karlsruhe, Germany); prazosin HCl (Pfizer, Karlsruhe); 6-chloro-9-((3-methyl-2-butenyl)oxy)-3-methyl-1H-2,3,4,5-tetrahydro-3-benzazepine maleate (SKF104078) (Smith Kline Beecham, Palo Alto, CA); and ARC 239 (Thomae, Biberach, Germany). Drugs were dissolved in distilled water, except for WB4101 (1 mM HCl) and spiroxatrine (10 mM HCl).
Results
The basal outflow of tritium (b1) from slices superfused with medium containing 1 μM desipramine was 0.0034 (0.0033, 0.0035) min−1 (n= 389), the overflow of tritium elicited by two trains of 4 pulses/100 Hz (S1) averaged 0.69 (0.64, 0.74) % of tissue tritium (n = 132), and the overflow elicited by 90 pulses/3 Hz (S1) was 3.07 (2.92, 3.23) % of tissue tritium (n = 257). When the medium contained 10 μM desipramine and 10 μM (+)-oxaprotiline, the basal outflow of tritium as well as the evoked overflow were similar; 10 μM desipramine and 10 μM (+)-oxaprotiline were used in experiments with high concentrations of unlabeled norepinephrine (>1 μM) to ensure blockade of the uptake of the unlabeled amine.
Stimulation by two trains of 4 pulses/100 Hz led to autoinhibition-free release of [3H]norepinephrine, as shown previously (Allgaier et al., 1995) and confirmed here by the lack of an overflow-enhancing effect of 1 μM rauwolscine, added after S1 (n = 9 versus 6 controls). Stimulation by 90 pulses/3 Hz, in contrast, led to autoinhibited release, as shown by the effects of the antagonists (see below).
Effect of Exogenous Norepinephrine.
Unlabeled norepinephrine, when added before S2, S3, and S4 at increasing concentrations, progressively reduced the electrically evoked overflow of tritium, both under autoinhibition-free conditions (Fig.1A) and under conditions in which autoinhibition developed (Fig. 1B). Norepinephrine did not change the basal efflux of tritium. The concentration-response curve in Fig. 1A was obtained by logistic curve fitting (eq. 7, Feuerstein and Limberger, 1999). The curve in Fig. 1B was obtained by fitting a function that takes the effect of transmitter norepinephrine into consideration and is based on the assumption of proportionality between receptor occupation by norepinephrine and effect (eq. 14, Feuerstein and Limberger, 1999). The two concentration-response curves obviously differ: under autoinhibition conditions (Fig. 1B), the curve is shifted to the right and the maximal observed degree of inhibition is smaller. The parameters estimated from the curves are shown in Table1. The parameters obtained under autoinhibition-free conditions by logistic curve fitting can be read easily from Fig. 1A: Imax observed is the asymptotic maximal inhibition, which the experimental curve approaches at high concentrations of exogenous norepinephrine, and pIC50 is the abscissa of the point of inflection (Fig. 1A). The parameters derived from the data obtained under autoinhibition conditions, in contrast, cannot be read immediately from inspection of Fig. 1B: Imax derived is not the asymptotic maximal inhibition that the experimental curve approaches at high concentrations of exogenous norepinephrine and is superimposed on a background of ongoing autoinhibition, but it is the maximal inhibition obtainable with norepinephrine, whether released or exogenous, against an autoinhibition-free background. Moreover,Kd is not the concentration of norepinephrine causing 50% of the asymptotic maximal inhibition of the experimental curve but is the dissociation constant of the norepinephrine-autoreceptor complex, calculated on the basis of proportionality between receptor occupation and effect, as mentioned.
Effect of α-Adrenoceptor Antagonists.
As shown in Fig.2, all antagonists, when added before S2, S3, and S4 at increasing concentrations, increased the overflow of tritium evoked by 90 pulses/3 Hz, indicating autoinhibition of transmitter release. For each antagonist, the data were evaluated by logistic curve fitting [E/Emax = 10L/(10−pEC50 + 10L), where L is the used log concentrations (M) of the antagonists]. The individualE = Sx/S1values scattered considerably, and clear maxima were not reached for several antagonists (Fig. 2). For these reasons, probably, the iterative calculations used to estimate the confidence intervals of the logistic parameters Emax and pEC50 and an additional slope factor cdid not converge when all three were left unconstrained. The slope factor, c, therefore, was constrained to 1 (see functionE/Emax above). The parameter estimates are summarized in Table2.
Of all antagonists, only prazosin changed the basal outflow of tritium, causing acceleration by 64% at 3.2 μM and by 141% at 10 μM. The reason for the correction of a too low EC50 to a more real EC50-corr is given in the Discussion; the steps of the conversion EC50-corr →Kb can be comprehended by considering the diagram in Fig. 3, which uses the graph of Fig. 1A.
Discussion
The subclassification of α2-autoreceptors in human neocortex was achieved by using calculated dissociation constants, Kb values, of antagonist-autoreceptor complexes of nine α-adrenoceptor antagonists in comparison to binding or functional data on these antagonists from the literature. The antagonist Kbvalues were obtained from their concentrations causing half-maximal disinhibition, EC50 values. The following consideration was the rationale of this conversion, EC50 → Kb.
Usually the evaluation of the disinhibition of release to assess the affinity of release-enhancing antagonists is limited to the calculation of antagonist EC50 values that are notKb values (e.g., Limberger et al., 1995a,b). This restriction can be surmounted by the knowledge of the relationship between EC50 andKb. We first analyzed the interplay of exogenous and endogenous norepinephrine at the autoreceptors, using a previously developed model. The model assumes proportionality between receptor occupation and effect of norepinephrine or, in other words, assumes that the Kd of norepinephrine equals its concentration causing half-maximal inhibition, IC50, of [3H]norepinephrine release under autoinhibition-free conditions. It should be noted that the IC50 of exogenous norepinephrine under autoinhibition-free conditions, 10−8.07 M, in fact was almost identical to the calculatedKd, 10−7.99 M (Table 1). Thus, the present experiments have for the first time established the Kd of norepinephrine at a central human α2-autoreceptor in functional experiments. The above-mentioned assumption of proportionality as a prerequisite for IC50 =Kd is supported by the estimate near unity of the slope factor, c (0.98, Table 1). Because of the limited number of data points of Fig. 1A and the small S1 values (0.69% of tissue tritium) that increased the variation in the S2/S1 ratio, the CI95 of c, however, was large (0.65, 1.75). This precludes a low error probability of the statementc = 1 (see Agneter et al., 1997; Feuerstein and Limberger, 1999). Accordingly, the statement IC50= Kd would also have a rather large error probability if it was only based on this large CI95 of c. We tried, therefore, to increase the specificity of the estimate of c by reassessing this value from the data of Fig. 1B. In other words, the “logistic components” of eq. 14 of Feuerstein and Limberger (1999) were endowed with a slope factor c, and this amended function was then refitted to the data of Fig. 2B with fixed values for pKd = 7.99 and Imax derived = 0.83 (Table 1), i.e., with a reduction of the number of parameters to reach convergence. This seemed reasonable because the maximum α2-autoreceptor-mediated effect had to be the same in the presence and in the absence of autoinhibition and because the pKd(≠pIC50 of Fig. 1B) was nearly identical to the pIC50 of Fig. 1A. The refit yielded a value for p[NEtr] corresponding to that of Table 1 (not shown) and an additional c of 1.03 (0.64, 1.42). Now two similar estimates for c were available, and their mean with deviation could be calculated (using the approximate standard errors ofc, 0.17 and 0.18): 1.01 (0.77, 1.24). Thus the CI95 of this mean c became considerably smaller, which improved our evidence for assumingc = 1 or IC50 =Kd.
The dissociation constants Kb of the antagonists were calculated by the use of their EC50 values, theKd (=IC50) of norepinephrine, Imax observed obtained under autoinhibition-free conditions, and the calculated endogenous concentration [NEtr]. In addition, the observed mean of the maximum disinhibition by the antagonists, 60% (Table 2), was considered as follows. A theoretical maximum disinhibition, Sxmax/S1, can be calculated on the basis of the values of Table 2. Sxmax is the stimulation-induced transmitter release that is not inhibited by the endogenous agonist [NEtr], or, in other words, Sxmax is the stimulation-induced transmitter release when the autoreceptor is completely blocked (when the concentration of the antagonist [B] → ∞). At S1, however, the stimulation-induced transmitter release is diminished to S1 = 1 −Imax observed × 10−pNEtr/(10−pKd+ 10−pNEtr) [compare eq. 9 of Feuerstein and Limberger (1999)]. Thus Sxmax/S1 = 1/(1 − 0.79 × 10−7.61/(10−8.07 + 10−7.61)) = 2.39. The theoretically expected value of 2.39, or an increase by 139%, is at variance with the observed mean of maximum disinhibition, which was 1.60, or an increase by 60% at the highest antagonist concentrations used (1–10 μM). This discrepancy may be due to the following condition. At the highest antagonist concentrations (up to 1000-fold of theKb values) additional, nonspecific effects of the antagonists, not related to the α2-autoreceptor under investigation, must be taken into account. These nonspecific effects are probably inhibitory, not stimulatory, in nature, i.e., they may diminish the increase in the evoked [3H]norepinephrine release because action potential-evoked, exocytotic release is a highly specific phenomenon that is dependent on the integrity of the neuronal environment. Decreasing nonspecific effects are much more likely than increasing nonspecific effects. As one example of perturbations by high antagonist concentrations of the release process local anesthetic inhibitory effects of yohimbine and rauwolscine at higher concentrations are well known (e.g., Goodall et al., 1984). Therefore, depressant, not stimulatory, effects of the antagonists at concentrations that are much higher than theirKb values may be assumed, and the evaluation of their concentration-disinhibition curves may be amended as follows. The depressions of the concentration-disinhibition curves of the antagonists at their highest concentrations, but not at rather low concentrations specific for α-adrenoceptors, correspond to the condition of an uncompetitive, use-dependent antagonism, as opposed to a noncompetitive antagonism (Segel, 1975; Jackisch et al., 1994). If fitted with the usual logistic function, e.g., eq. 5 of Feuerstein and Limberger (1999), an apparent EC50 is obtained that is too low. To obtain a real EC50,E/Emax = 10L/(10−pEC50 + 10L × Depr-Emax) should be used, where Depr-Emax is the relative extent to which the uncompetitive mechanism depressesEmax, instead ofE/Emax = 10L/(10−pEC50 + 10L). In our case, Depr-Emax is 2.39/1.60 = 1.49. With respect to the quantitatively dissimilar depressions by the nine antagonists, note that most of the CI95 values of the Emax values of Table 2 overlap, suggesting a roughly similar depression of the theoretical maximum disinhibition. Therefore, 1.49 may be roughly the factor by which the too low EC50 obtained withE/Emax = 10L/(10−pEC50 + 10L) may be corrected to get a more realistic EC50, according toE/Emax = 10L/(10−pEC50 + 10L × Depr-Emax) (Jackisch et al., 1994). This correction yields the pEC50-corr values of Table 2.
When the values pIC50, p[NEtr], Imax observed, EC50-corr are introduced into eq. 7 of Feuerstein and Limberger (1999), step 5 in Fig. 3, (1 + [EC50-corr]/Kb) = 8.37 or pKb = pEC50-corr + 0.87 is obtained. The correspondingKb values for the antagonists represent the first dissociation constants of antagonists at human cerebral autoreceptors obtained in functional experiments (Table 2).
To subclassify the α2-autoreceptors, the autoreceptor pKb values were compared with dissociation constants at known subtypes by means of a correlation analysis (Table 3), as has become usual in the literature, e.g., Bylund et al. (1992). Note that use of the pEC50 values in the correlation analyses would have sufficed for the identification of the α2-autoreceptor subtype because the subsequent transformation to pKb values has no effect on the correlation coefficients. However, apart from obtaining accurate pKb values, we wanted to demonstrate that the potency of an antagonist in disrupting the autoinhibitory circuit of transmitter release is a direct measure of its dissociation constant at the receptor to be blocked.
The known subtypes of the literature were, first, prototypical native α2 radioligand binding sites (Table 3A); second, radioligand binding sites in COS cells transfected with α2-adrenoceptor genes (Table 3B); and third, previously subclassified α2-autoreceptors (Table 3C). Table 3 shows that the dissociation constants of the antagonists at the human neocortical autoreceptors correlate significantly and without exception with their dissociation constants at both α2A and α2Dbinding sites or receptors; correlations with α2B and α2C binding sites or receptors are not significant (exception: the α2C binding sites in opossum kidney cells; Table 3A). Moreover, Table 3 shows that the coefficients for the correlation with α2A are generally higher than for the correlation with α2D, the error probability is generally lower in the former than in the latter case, and the slopes of the regression lines for α2Aare generally closer to unity. We conclude that the α2-autoreceptors in human brain cortex are α2A.
Presynaptic α2-autoreceptors have also been subclassified in the human saphenous vein, kidney, and heart. In the saphenous vein, the receptors were suggested to be α2A (Molderings and Göthert, 1995), whereas in the kidney and heart they were initially classified as α2C (Trendelenburg et al., 1994; Rump et al., 1995). A reinvestigation of the kidney receptors, however, also yielded an α2A diagnosis (Trendelenburg et al., 1997). Overall, α2A (i.e., genetically α2A/D) autoreceptors seem to predominate in humans, as they do in various animal species (see the Introduction).
In summary, this paper shows that it is possible to analyze quantitatively the autoinhibitory circuit of [3H]norepinephrine release in human neocortex tissue, i.e., to estimate the biophase concentration of the transmitter in relation to its Kd, and to calculate true, unbiased dissociation constants of antagonists by evaluation of their disinhibition of [3H]norepinephrine release. The functionally obtained pKb values at the presynaptic α2-autoreceptors in human brain cortex correlated highly with pKb values at previously subclassified α2A sites, but did not correlate significantly or correlated much less well with pKb values at α2B, α2C, and α2D sites. It is concluded that the α2-autoreceptors in human neocortex are α2A.
Acknowledgment
We are very grateful to Prof. Dr. K. Starke for critical and constructive comments.
Footnotes
-
Send reprint requests to: Dr. T. J. Feuerstein, Sektion Klinische Neuropharmakologie der Neurologischen Universitätsklinik, Breisacherstrasse 64, D-79106 Freiburg, Germany. E-mail:feuer{at}ukl.uni-freiburg.de
-
↵1 This work was supported by the Deutsche Forschungsgemeinschaft (SFB 505, TP C4, C8).
-
↵2 Present address: Neurochirurgische Universitätsklinik, Breisacherstrasse 64, D-79106 Freiburg, Germany.
-
↵3 Present address: Pharmakologisches Institut der Universität Freiburg, Hermann-Herder-Strasse 5, D-79104 Freiburg, Germany.
- Abbreviations:
- ARC 239
- 2-{2-[4-(o-methoxyphenyl)piperazin-1-yl]ethyl}-4,4-dimethyl-1,3(2H,4H)-isoquinolinedione
- WB4101
- (±)-2-(2,6-dimethoxyphenoxyethyl)aminomethyl-1,4-benzodioxane HCl
- SKF104078
- 6-chloro-9-((3-methyl-2-butenyl)oxy)-3-methyl-1H-2,3,4,5-tetrahydro-3-benzazepine maleate
- CI95
- 95% confidence interval
- Received November 30, 1999.
- Accepted April 4, 2000.
- The American Society for Pharmacology and Experimental Therapeutics