Molecular Actions of Propofol on Human 5-HT3A

8/14/2019 Molecular Actions of Propofol on Human 5-HT3A

1/12


2/12

Propofol is clinically the most extensively used IVanesthetic. Compared with other IV anesthetics, suchas barbiturates or etomidate, the molecular structureof propofol is simple and symmetric (Fig. 1), consist-ing of a plain benzene ring and a phenolic hydroxy-group located between two identical alkyl residues.Thus, any systematic study of IV anesthetic interac-tions may start with this clinically important drug. Forreasons not yet understood, propofol, on the molecu-

lar level, suppresses ion flux through excitatory ionchannels but it enhances ion flow through inhibitoryion channels14 Propofol has the advantage over in-haled anesthetics of causing less emesis, if any at all.5

The 5-HT3 receptor is a ligand-gated excitatory ionchannel6,7 which is located in the periphery andwithin the central nervous system. It plays a majorrole in the modulation of nausea and vomiting,8,9

consistent with the fact that 5-HT3 receptor antago-nists are commonly used antiemetic drugs.10 Theamino acid sequences of the pentameric 5-HT3 recep-tors show homologies with excitatory nicotinic acetyl-

choline receptor channels,11

and, to some extent, evenwith inhibitory -aminobutyric acid type A (GABAA)receptors and glycine receptors. 5-HT3A receptorsare the only functional homopentameric 5-HT3 re-ceptors, and because of their well defined stoichi-ometry, they provide a useful model studying themolecular actions of drugs such as general anesthet-ics and cannabinoids.12,13

A previous study has shown that propofol sup-presses 5-HT3 receptors (mouse) by more than oneaction, reducing the peak current amplitude and ac-celerating the desensitization process.14 In the present

study, we used an experimental approach that al-lowed us to differentiate kinetically between different

actions of propofol on human 5-HT3A receptors. Usingthe patch-clamp technique (excised outside-out patchmode) and a fast solution exchange system, we aimed(a) to test whether the human 5-HT3A receptor is alsoa sensitive target for propofol; (b) to identify separatecomponents of action by, first, applying propofol todifferent conformations of the 5-HT3 receptor and,second, by comparing its actions with structurallyrelated derivatives (Fig. 1); and, thus, (c) to shed light

on the different responses to propofol by excitatoryand inhibitory ligand-gated ion channels.

METHODS

Cell Culture

A human embryonic kidney (HEK)-293 cell linecontaining stably transfected human 5-HT3A receptorswas used.12 HEK-293 cells were grown as monolayeron culture plates (Nunc) in DMEM Nutrient Mix F12medium containing 10% heat inactivated fetal calfserum, penicillin (100 IU/mL), streptomycin (100

g/mL), geneticin (0.75 mg/mL), and glutamine (292g/mL). Cells were cultured at 37C in humidifiedatmosphere (5% CO2). For patch clamp experiments,cells were subcultured in monodishes (Nunc, 35 mmdiameter) 711 days before an experiment.

Drugs and Solutions5-HT (the creatinine sulfate salt) was obtained

from Sigma (Munchen, Germany). Propofol (free sub-stance, without vehicle) was obtained from RBI (MA,USA), propofol derivatives, 5-hydroxyindole, and

benzene from Sigma (Munchen, Germany). 5-HT so-

lutions were prepared daily from 25 mM aqueousstock solutions stored at 20C. Propofol and

Figure 1. (a) Structures of phenols and

benzene with their respective experi-mental log P values, taken from thePhysProp Database (Syracuse ResearchCorporation; http://www.syrres.com/esc/physdemo.htm), see also Tetko etal.16 (b) Structure of 5-hydroxyindole(left) and 5-HT (right).

Vol. 106, No. 3, March 2008 2008 International Anesthesia Research Society 847


3/12

2-isopropylphenol were prepared daily from ethanolstock (1 M) by dilution (e.g., 1:10,000 resulting in 100M) in extracellular buffer and stirring for 3 h in glass

bottles. The highest resulting ethanol concentrations atthe highest drug concentrations were 2.2 mM (propo-fol) and 6.6 mM (2-isopropylphenol); they had noeffect on 5-HT control currents. Ethanol concentra-tions were proportionally lower at lower drug concen-trations. Benzene, phenol, and 5-hydroxyindole were

soluble in the extracellular buffer without any furthervehicle. All concentrations of the lipophilic drugspropofol and 2-isopropylphenol (log P 3.79 and 2.88,respectively) were measured by high performanceliquid chromatography [HPLC,15]. For this purpose,samples of drug solution were collected from thetubes directed to the patches of the application system.It was found, for example, that near the IC50 value anexpected concentration of 25 M propofol resulted in(mean standard deviations) 14 2.5 M (n 40)propofol and 30 M 2-isopropylphenol resulted in21 4.7 M (n 11).

ElectrophysiologyIn previous studies, we described human 5-HT3A

receptors as activated by 5-HT in a concentration-dependent manner (EC50 values, 59 M) and that theresulting currents were sensitive to the 5-HT3 receptorantagonist, ondansetron. For the experiments in thisstudy, we used identical conditions, including electro-physiological methods, solution application systems,12,13

and a concentration of 30 M 5-HT (unless specifiedotherwise) because it elicits a reproducible, nearlymaximal (85%) signal.

Before starting patch-clamp recordings, the culturemedium was replaced by extracellular solution ofthe following composition: NaCl 150 mM; KCl 5.6mM; CaCl2 1.8 mM; MgCl2 1 mM; HEPES 10 mM;d-Glucose 20 mM; pH 7.4. d-Glucose was omitted inthe extracellular solution used for superfusion of theexcised patches. Patch pipettes were filled with intra-cellular solution containing: KCl 140 mM; EGTA 10mM; MgCl2 5 mM; HEPES 10 mM; pH 7.4 and hadresistances of 24 M. Experiments were performedat room temperature (20C26C).

Patch pipettes were manufactured from Borosili-

cate glass capillaries (Kwik-Fil, World PrecisionInstruments, USA) using a pipette puller (List L/M-3P-A, List Electronic, Darmstadt, Germany). The sealresistances (excised outside-out patches) were 16 G.For current measurements, we used a patch-clampamplifier (EPC-7, List) in combination with an exter-nal low pass filter set at 1 kHz (Frequency Devices,MA). Data were digitally recorded at a sampling rateof 2 kHz with a Digidata 1200 (Axon Instruments,Foster City, CA) interface. Clampex-6 software (Axon)was used for the recording protocols. Five-hundredmilliseconds before 5-HT exposure, the membrane

potential was stepped from a holding value of 0 mV to100 mV. These conditions were chosen to optimize

the stability of the excised patches and the reproduc-ibility of results.

Fast Solution Exchange System

A multitube perfusion system with an exchangerate (solution exchange) below 2 ms (RSC 200, Bio-logic, France) was used.12 The drug application sys-tems were equipped with five separate tubes and inertmaterials, such as Teflon tubing and glass, to avoid

loss of hydrophobic drugs.15

Drug Application ModesThree different protocols of drug application were

used12:

1. Equilibrium application: Continuous exposure tothe drug 60 s before and during the applicationof 5-HT;

2. Open channel application: No drug applicationprior to the 5-HT pulse, only simultaneous ap-plication of drug and 5-HT;

3.Closed channel application:

Pre-exposure to thedrug 60 s before but not during the 5-HTapplication.

Limitations of Data AquisitionAlthough more than one measurement could be

performed on a typical patch, the interval betweenmeasurements had to be at least 1 min because of theslow kinetics of recovery from desensitization. Tocorrect for small rundown of currents when presentand to discard inconsistent data, the measurement ofeach data point involved a three-fold repetition of thesequence: control, drug test, recovery (control for

next drug test), lasting a minimum of six min. Thelimited but varying life times of such patches (nor-mally ranging from 5 to 25 min, in a few instances 1 h)resulted in inhomogeneous data sets with regard tothe number of different data points measured on eachpatch. However, each data point was measured on atleast three different patches. As the recording of entireconcentrationresponse curves on single patches wasnot possible, mean IC50 values and their standarddeviations could not be calculated. Instead, IC50 val-ues and estimates of their standard errors were calcu-lated from fitting a single concentrationresponse

curve to the entire data set (Graph Pad Prism 3.02).Highly hydrophobic substances have a tendency toabsorb in the tubing,15 and, therefore, drug concentra-tions could not be randomized. Most experimentswere started with the lowest hydrophobic drug con-centrations and continued with increasingly largerconcentrations. Otherwise, the superfusion systemwould have had to be extensively rinsed and re-checked for remaining drug after each experiment,requiring unrealistic experimental time for each ex-periment. Instead, a second, very different superfu-sion system and set-up was used to check whether

there was any indication that the sequence (history) inwhich experiments were performed had any impact

848 Propofol Derivatives and 5-HT3A Receptors ANESTHESIA & ANALGESIA


4/12

on the outcome. Whenever evidence for dependenceon history was found, the underlying reasons wereinvestigated. Only in artifactual situations was a de-pendence on history found, otherwise there was noevidence or suggestion that the sequence in whichexperiments were performed matters.

As this was an exploratory study, the number ofexperiments needed to reach statistical significance

could not be estimated in advance with a poweranalysis.

Partition Coefficients and Predictions of Potency from

Meyer-Overton CorrelationsThe following Meyer-Overton correlations1 have

been used to estimate anesthetic potencies arisingonly from nonspecific, hydrophobic actions. For

Figure 2. (a, c, e; left panels) Effects of propofol, 2-isopropylphenol, and phenol (equilibrium application) on 5-HT (30M)-induced currents (original traces from three patches). Note that all three drugs caused suppression of the peak current;however, propofol speeded current decay, whereas 2-isopropylphenol and phenol slowed current decay (statistical analysisin Fig. 5). (b, d, f; right panels) Time courses of wash-in of drugs. Currents induced by 5-HT (30 M) in the presence of drug

are plotted against the time duration for which the 5-HT3A receptors were pre-exposed to the respective drug. Data points(circles) represent percent of the control currents after 60 s wash with propofol-free buffer (means sd, n 4 10 differentexperiments). A fast and slow inhibitory process was observed for propofol, and a fast potentiating and a slow inhibitoryeffect was detected for 2-isopropylphenol. Phenol also showed a slow inhibitory effect, whereas potentiation was immediate.Wash-in curves for propofol and 2-isopropylphenol were fit biexponentially: i imax1 * exp(t/IN-1) (100-imax1 Plateau) *exp(t/IN-2) Plateau. The wash-in curve for phenol was fit monoexponentially: i imax * exp(t/IN) Plateau.



5/12

ligand-gated ion channels pooled for a wide rangeof anesthetics (Fig. 12b in Ref. 1), the parameterswere: log (IC50) 1.568 to 0.9898 * log (Poctanol/water);for ligand-gated ion channels pooled for inhaled an-esthetics only (Fig. 12d in Ref. 1): log (IC50) 1.178to 0.8975 * log (Poctanol/water); for voltage-gated ionchannels pooled for IV anesthetics only (Fig. 12c,published in Ref. 1): log (IC50) 0.8064 to 0.9306 *log (Poctanol/water). Experimental octanol/water parti-

tion coefficients, Poctanol/water, are taken from thePhysProp Database (Syracuse Research Corporation;http://www.syrres.com/esc/physdemo.htm), see alsoTetko et al.16

Data Analysis and Statistics

Analysis of the original current traces (baselineadjustment, determination of peak currents and cur-rent kinetics) was performed with Clampfit 8 software(Axon Instruments, Foster City, CA). Graph Pad Prism3.03 software (Graph Pad, CA) was used to creategraphics. The concentrationresponse curves were fit-

ted by the Hill equation, i [ICn

50/(cn

ICn

50)], i isthe remaining peak current as fraction of the maximal(control) current, c is the drug concentration, n is theHill coefficient, and IC50 is the drug concentrationcausing half-maximal effect. Potencies are expressedas IC50 values plus standard error as calculated byGraph Pad Prism 3.03. Differences between single datapoints were tested for significance with either pairedor unpaired t-tests (Excel and Prism 3.03). Differenceswere considered significant when P values for the re-spective test were 0.05. Values are reported as means standard deviations (sd), unless stated otherwise.

RESULTS

Propofol (0.5160 M) did not cause an activationof human 5-HT3A receptors in the absence of 5-HTwhen it was applied to patches which before hadshown a pronounced response to 5-HT. Propofolinhibited 5-HT-induced currents reversibly with anIC50 18 1 M (Figs. 2a and 3) when applied underequilibrium condition (see Methods). Next, we exam-ined the kinetics of washing in the propofol effect.Pre-exposing the patch to propofol for various durations(32 ms to 60 s) before applying 5-HT in the presence of

propofol resulted in a wash-in time course of propofolthat was characterized by a fast (IN-1 35 ms) and aslow (IN-2 4.8 s) process (Fig. 2b).

In an attempt to further characterize the slow andthe fast wash-in component of the propofol effect,2-isopropylphenol and phenol were examined. Bothsubstances are straight derivatives of propofol, inwhich the balance of polar and hydrophobic proper-ties is shifted towards lesser hydrophobicity. In theabsence of 5-HT, neither 2-isopropylphenol (0.011mM) nor phenol (0.13 mM) caused direct activationof 5-HT3A receptors in patches which in a preceding

control had shown pronounced response to 5-HT.When applied continuously (equilibrium condition),

both drugs inhibited 5-HT-induced currents as can beseen in the traces of Figure 2c and e. As a thirdsimilarity with propofol, wash-in experiments (drug

application for varying durations before 5-HT andduring 5-HT exposure) revealed that fast and slowprocesses (Fig. 2d and f) contribute to the effects of2-isopropylphenol (IN-1 64 ms and IN-2 6.6 s) andphenol (IN-1 10 ms, IN-2 20.4 s).

The concentrationresponse curves of all three drugscould be fitted to Hill-equations (Fig. 3 and Table 1). TheIC50 values for propofol and 2-isopropylphenol wereabout the same (18 1, 17 3.2 M, respectively),whereas the slope of the concentrationresponse curvefor propofol (nHill 2 0.2) was about two timessteeper than that of 2-isopropylphenol (nHill 1 0.2,

Table 1). Phenol was considerably less potent (IC50 1.6 0.2 mM).

However, propofol (6 M) increased the speed ofthe current desensitization (e.g., by 51% at 18 M) asdid many other anesthetics before,12 whereas both2-isopropylphenol (20 M) and phenol (1 mM)caused an extensive slowing of current decay (e.g., by214% at 21 M 2-isopropylphenol and by 188% at 1mM phenol). This can be seen in the traces of Figure 2and it is substantiated by the subsequent statisticalanalysis (see below).

As these observations may suggest different pro-cesses, an attempt was made to separate these (Fig. 4).Fast actions were separated by coapplying propofoland the phenol derivatives simultaneously with 5-HT

but not before (open channel application). In thismanner, their effects on peak currents could be inves-tigated within a time frame of approximately 20 ms,whereas their effects on current desensitization could

be observed for a period of 50 ms to several 100 ms.Slow drug actions were separated by exposing drugsto the patches for 60 s exclusively before 5-HT wasapplied (closed channel application). The combined

effects of open channel and closed channel application(fast and slow actions of propofol) were studied by

Figure 3. Concentrationresponse curves for three phenolson 5-HT (30 M)-induced peak-currents. The bottoms andtops of the fits were set constant at 0% and 100%, respec-tively. Drugs were applied in equilibrium (means sd, n 410 different patches).



6/12

applying drugs 60 s before and also during the acti-vation of the 5-HT3A receptor by 5-HT (equilibriumapplication).

The fast actions (open-channel) of 2-isopropylphenoland phenol (Fig. 4), in contrast to propofol (whichcaused significant peak amplitude reduction and ac-celeration of the desensitization at concentrations 6M, Fig. 5), consisted of an increase of the peak

amplitude and a pronounced slowing of the desensi-tization process at drug concentrations 20 M and 1

mM, respectively (Fig. 5). The slow action (closed-channel application) of 2-isopropylphenol and phenolleft the time constants of desensitization unchanged(Fig. 5), whereas both substances suppressed the peakamplitudes in a concentration-dependent manner(IC50 values 14 1.1 M and 1.04 0.1 mM,respectively).

However, whenever the fast process was visible

(both in the open-channel and in the equilibriumapplication), the desensitization time constants changed

Figure 4. Effects of three phenols on 5-HT (30 M)-induced currents, using three different drug application modes (originalcurrent traces). Traces for propofol were obtained from two patches (separate patch for the closed channel application); thosefor 2-isopropylphenol and phenol from single patches, respectively.

Table 1. Parameters of the Hill Equations Derived from the ConcentrationResponse Curves (Inhibition of Peak-Currents Induced by30 M 5-HT) of Phenols

IC50

(M)/Hill coeff.

Propofol 2-Isopropylphenol Phenol

Equilibrium (human) (fast and slow effects) 18 1/2.0 0.2 17 3.2/1.0 0.2 1620 223/1.1 0.2Equilibrium (mouse) 14 1/1.5 0.2a

Open channel (fast effect) 28 2.6/1.2 0.1 Potentiation PotentiationClosed channel (slow effect) 22 2.5/1.2 0.2 14 1.1/1.1 0.1 1035 108/1.1 0.1

Octanol/water partition coefficient 6170 760 29 Three different drug-application modes were used.aThe data for equilibrium (mouse) are from Barann et al.14 For octanol/water partition coefficients see Methods.



7/12

with concentration in the same manner (Fig. 5): theyaccelerated with increasing propofol concentration(6 M) and they were successively slowed by in-creasing concentrations of either 2-isopropylphenol(20 M) or phenol (1 mM). During closed-channelapplication (slow action), desensitization time con-stants were not affected by any of the three drugs (Fig.5). No significant effects of propofol, 2-isopropylphenol,and phenol on the current activation kinetics were

observed in any application mode.An additional series of experiments tested for eachapplication mode separately whether the drug effectsdepended on the agonist (5-HT) concentration (Fig. 6).We compared the effects of propofol and its deriva-tives when currents were elicited either by a nearlysaturating concentration of 5-HT (30 M) or a concen-tration well below the EC50 for 5-HT (59 M) buthigh enough (3 M) to evoke currents amenable toreproducible curvefitting. The slow (closed channelapplication) inhibitions of current amplitudes byany of the three substances, propofol (14 M), 2-

isopropylphenol (21

M), or phenol (1 mM), were notsignificantly different when the 5-HT concentration

was changed (Fig. 6, bottom row), in contrast to thefast (open channel application) actions of the threedrugs. Here, in the case of propofol, 3 M 5-HT-evoked currents were much more potently suppressed(by 58%) than 30 M 5-HT-evoked currents (by 14%),whereas 2-isopropylphenol or phenol produced muchlarger current increases at 3 M 5-HT (by 206% and197%, respectively) than at 30 M 5-HT (by 11% and13%, respectively (Fig. 6, middle row).

In the equilibrium application, the inhibition by 14M propofol was larger than in either of the otherapplication modes (open channel and closed channel),irrespective of the 5-HT concentration (Fig. 6, leftpanels). Although increases in current amplitudes by2-isopropylphenol and phenol were still present forcurrent peaks evoked with the lesser 5-HT concentra-tion in the equilibrium application (Fig. 6, uppertrace), they were reduced compared with the openchannel application. However, at the higher 5-HT (30M) concentration in the equilibrium application, inthe presence of either 2-isopropylphenol (21 M) or

phenol (1 mM), inhibition of current amplitudes (by37% and 30%, respectively) was observed (Fig. 6,

Figure 5. Effects of phenols on the current decay-kinetics of 5-HT (30 M)-induced currents using three different drugapplication modes (note the different scaling of the y-axes). Results are expressed as percentage of the time constants off ofthe control currents in the absence of drug (100%, indicated by dashed lines); the mean for all off values of the respectivecontrol currents was 117 73 ms, n 43, which is a typical value for human 5-HT3A receptors, see discussion. Data are shown

as means sd, n 310 different patches. *Significant difference from the control value in the absence of drug (P 0.05,paired t-test).



8/12

upper row), in contrast to the current increases in theopen channel application.

When the slow and the fast effects on currentamplitudes were multiplied for each drug, then theproduct obtained both for the lesser 3 M 5-HTconcentration (Fig. 6, upper row, dotted line on theleft) and the higher 30 M 5-HT concentration (Fig. 6,upper row, dotted line on the right) came very close tothe effect measured in the equilibrium application(Fig. 6, upper row, respective bars through whichdotted line runs).

A phenolic group is also part of the agonist of the5-HT3 receptor. When ethylamine is removed from5-HT, 5-hydroxindole is obtained, which no longer

directly activates currents through 5-HT3A receptors(data not shown). When this aromatic alcohol with its

phenolic OH group (1 mM) was applied to the 5-HT3receptor in the open-channel mode, it also lead to a

slowing of the current desensitization time constant(Fig. 7, left panel). On the other hand, benzene (3 mM)no longer showed a slowing of desensitization butrather an acceleration (Fig. 7, right panel).

DISCUSSION

Comparison with Previous Study

The present study confirms that human 5-HT3Areceptors are also inhibited by 30 M 5-HT are alsoinhibited by propofol with a potency similar to thatrecorded for murine 5-HT3 receptors.

14

The slope of the concentrationresponse curve forpropofol (at 30 M 5-HT) is steeper for the human

Figure 6. Effects of three phenols on 3 M and 30 M 5-HT-induced currents, using three different drug application modes.Any of the three substances propofol, 2-isopropylphenol or phenol caused suppression of current in the closed channelapplication, which did not depend on the concentration of 5-HT. Note the different current scales in the open channel or inthe equilibrium applications, since propofol caused inhibition while isopropylphenol or phenol produced current potentia-tion. Significance (P 0.05, unpaired t-test) refers to the difference between relative drug effects on current amplitudeselicited by either 3 M or 30 M 5-HT (each reduction normalized to the respective 3 and 30 M 5-HT control in the absenceof drug).



9/12

receptor (Table 1). Other differences between humanand murine receptors are the biphasic wash-in of thepropofol effect reported here and the fact that the fastwash-in time constant (35 ms) is almost a factor of 3faster than the current desensitization time constant,off, of the 5-HT3A receptor [this study: off 117 73ms, n 43 patches; entire data pool from our studies:off 112 100 ms, n 1760 patches

17], whereas theslow wash-in time constant is considerably slowerthan off. This latter observation goes against thehypothesis proposed in an earlier paper14 that propo-fol triggers an unmodified desensitization mechanismin the absence of 5-HT with the same kinetics as does5-HT.

Slow Action

Propofol and its derivatives clearly interact with

human 5-HT3A receptors, even when they are appliedto the receptor exclusively in the absence of any 5-HT.Although not with the same, but with much slower,kinetics than for the unmodified desensitization pro-cess hypothesized above, the compounds investigatedin this study may still shift activable 5-HT3A receptorsinto an anesthetic-related desensitized state. The frac-tion of channels inhibited by the slow action (closedchannel application) may be in this state. The actionsconsist of a reduction in the peak current of the testpulse for all three substances (Fig. 4), without anyeffect on the current decay time constant (Fig. 5), even

though the duration of superfusion by propofol or itsderivatives had lasted long enough so that the reduc-tion in peak current had reached a steady state.

The IC50 of phenol for slow action (closed channelapplication, Table 1) is very close to the one predicted

by the Meyer-Overton relation for nonspecific inhibi-tion by phenol (predicted, see Methods: 970 M;measured: 1035 M). For 2-isopropylphenol, the pre-diction under-estimates the actual potency by a factorof 3 (predicted: 38 M; measured: 14 M), whereas forpropofol it over-estimates the potency by a factor of 4(predicted: 5 M; measured: 22 M). This may be an

indication that more than added hydrophobicity isinvolved in the actions of compounds made more

hydrophobic by the addition of purely hydrophobicisopropyl-groups. By adding an isopropyl-group tophenol, the resulting 2-isopropylphenol becomesmore potent than predicted by lipophilicity. In con-trast, when the second isopropyl-group is added,turning 2-isopropylphenol into propofol, the IC50 re-mains unchanged, although hydrophobicity aloneshould have made propofol roughly nine times morepotent. This observation, together with the differencesin the slopes of the respective concentrationresponsecurves, suggests additional interactions to be involvedthat are antagonistic to hydrophobic actions. How-ever, as has been found before for many other anes-thetics,1 on average the three substances follow thecorrelation with hydrophobicity very well, as propofolscatters by approximately the same factor to the rightas 2-isopropylphenol scatters to the left.

Fast ActionExclusive exposure of 5-HT3A receptors to propofol

or its derivatives during the 5-HT pulses (open chan-nel application) sufficed to produce effects on theamplitudes of the elicited currents as well as on theirdesensitization time constants (Figs. 4 6). Propofolsuppressed the amplitudes of the currents and causedthem to desensitize more rapidly, whereas phenol and2-isopropylphenol, in contrast, increased their ampli-tudes and slowed desensitization in a concentration-dependent manner. Increases in peak amplitudes have

been observed for volatile anesthetics1821

and alco-hols,2224 whereas retardation of desensitization hasbeen reported for morphine.25

The effects on the amplitudes were more clearlyseen for lower 5-HT concentrations, whereas the ef-fects on the time constants of desensitization wereobvious at both 5-HT concentrations. These observa-tions may, at least partly, be explained by the kineticsof wash-in (Fig. 2), as the fast action may not havefully developed by the time the current peak isreached (about 20 ms at 30 M 5-HT), whereas desen-sitization can be observed for a duration of several 100

ms. Whether or not additional competition betweenthese drugs and 5-HT is involved, could only be

Figure 7. Effect of 5-hydroxyindole (1 mM) and benzene (3 mM) on 5-HT-induced currents in two outside-out patches.5-Hydroxyindole or benzene was exclusively applied together with the 1.2 s 5-HT pulse. Note that the traces for control andwash superimpose. 5-Hydroxyindole decreased the current decay kinetics almost two-fold (i.e., slow down), which remainedmonoexponential (time constant off). In contrast, benzene accelerated the current decay. Typical experiments from fourexperiments (each) are shown.



10/12

decided in an additional study, which would bedifficult and extensive, as the 5-HT activation curveitself is bell-shaped,26 which is typical for ligand-gatedion channels.27,28

Additivity of Fast and Slow Actions

The effects of propofol and its derivatives on thedesensitization time constants persisted as long asthey were administered together with 5-HT (Figs. 4

and 5), whether or not they had been given before the5-HT application as well (equilibrium or open channelapplication). Thus, the concentration-dependent accel-eration (propofol) or retardation (propofol deriva-tives) of the desensitization time constant requires thecoapplication with 5-HT, i.e., an open channel confor-mation during drug application.

There is additional evidence that fast and slowactions appear to be independent and additive: Whenthe effects of the open-channel application and theclosed-channel applications on current amplitudes aremultiplied, the resulting product (see dotted lines in

Fig. 6, upper panels) is close to the measured resultunder equilibrium condition (bars in the same figure).This observation supports the hypothesis of at leasttwo separate and superimposing processes postulatedin our previous paper.14

Comparison with Other Ligand-Gated Ion Channels

Comparing the concentrations at which propofolinteracts with other ion channels, one finds them all tolie within a similar range, as observed in this study(IC50 18 M; more potent at lower 5-HT concentra-tions, Fig. 6). IC50 values of 20 M,

29 between 4.6 and

23 M,30

and between 6 and 25 M31

have beenreported for human brain, skeletal muscle, and rat

brain of sodium channels, respectively. The voltage-dependent potassium channel from SH-SY5Y human

blastoma cell lines was suppressed with an IC50 44M.32 For the neuronal type 42 nicotinic acetylcho-line receptor channel, IC50 values of 19 M

33 or 4.5M34 have been found, whereas the muscle subtype was half-suppressed by 46 M.34 Depending onsubtype, GABAA receptors are enhanced by 2.6 M(132

35), 4.6 M (63235), 8.7 M (21

36), or 39.7M (122S

37) propofol. Enhancement is also ob-

served for the glycine receptor channel 1 with anEC50 12.5 M.38 All these IC50 and EC50 values

scatter around the value of 5 M predicted by theMeyer-Overton correlation for nonspecific interac-tions of propofol (see Methods).

Active Functional Groups Just as has been postulated for the interaction

between serotonin and its receptor site,39 the interac-tion between propofol or its phenol derivatives andthe 5-HT3 receptor may be a combination of hydro-phobic and specific polar interactions. Evidence for

hydrophobic contributions is the relatively close pre-diction of the IC50 values for closed channel inhibition

from Meyer-Overton correlations (see Methods). Can-didates for specific interactions are hydrogen bondsformed by the phenolic hydroxygroup, van der Waalsinteractions with the -electrons of the aromatic ringof the phenol and its derivatives, a lipophilic pocket oflimited dimensions, or a combination of some or all ofthese possibilities.

The phenolic OOH group seems to be involved inthe effect of increasing current amplitudes and slow-

ing desensitization as suggested by the failure ofbenzene to cause either effect (Fig. 7). The only differ-ence between phenol and benzene is theOOH group,which benzene is lacking. There is other circumstantialevidence that phenolic groups of a molecule may slowthe desensitization of 5-HT3 receptors. Morphine,which also contains a phenolic OOH group, causes apronounced slowing of 5-HT3A receptor current de-cay.25 5-Hydroxyindole, a 5-HT derivative lacking theethylamine group, has lost the intrinsic activity ofeliciting currents through 5-H3A receptor, but it doesslow current decay of 5-HT3 receptors [Ref. 40 and

Figure 7].When first synthesizing and exploring phenol de-

rivatives suitable for general anesthesia, James andGlen41 already noted not only the importance of theirlipophilic character and H-bond donor/acceptorproperties but they also realized that steric consider-ations mattered. They found that potency and kineticsappeared to be a function of both the lipophiliccharacter and the degree of steric hindrance exerted byortho substituents. The phenolic group, in conjunctionwith aliphatic substituents in the ortho position, has

been implicated in the direct activation of chloride

currents through GABAA receptors in the absence ofGABA.42 Additional evidence that steric consider-ations are important comes from the observation byKrasowski et al.37 that 2,6-di-sec-butylphenol did en-hance GABAA receptor, whereas 2,6-di-tert-butylphenoldid not. In addition, hydrophobic groups of a drugmay show specificity, for example, when they exceedthe molecular volume available in a hydrophobicpocket. Thus, fewer hydrophobic sites may be avail-able to propofol than to either 2-isopropylphenol or tophenol.

This may be why inhaled anesthetics with too large

a molecular volume no longer cause net enhancementof 5-HT3 receptors.43 Thus, steric limitations compro-

mising the simultaneous interaction of two or morefunctional groups and/or too large a hydrophobicvolume may be why propofol, in this and otherstudies, does not enhance 5-HT3A receptors,

22,44,45

while 2-isopropylphenol and phenol (this study) do.

Implications for Anesthetic MechanismsJames and Glen41 finally settled for propofol to be

tested in human trials, because of all the phenolderivatives they had tested, propofol had proved to be

the most potent hypnotic. In our study, we find thatpropofol is about equally as potent as 2-isopropylphenol in



11/12

suppressing 5-HT3A receptor currents in the equilib-rium application, although propofol is much morehydrophobic. This implies that propofol in compari-son with 2-isopropylphenol has lost some potential forinteractions that cause current suppression, eventhough propofol already lacks the counteracting abil-ity for enhancement, i.e., it causes neither increases incurrent amplitudes nor in the desensitization timeconstant.

There is other evidence for a lesser specificity ofpropofol compared with other IV drugs. The IC50 (18M) for propofol differs little from those listed in aprevious section for voltage-gated ion channels. Itdiffers by only a factor of two from that predicted byhydrophobicity for voltage-gated ion channels (seeMethods), and not by about an order of magnitude aswould be the case when average values for IV anes-thetic action on ligand-gated and on voltage-gated ionchannels are compared.1

In fact, in some sense, propofol is not unlike inhaledanesthetics, which are considered less specific in their

molecular actions than IV anesthetics, because here wefind propofol to be more potent than inhaled anesthet-ics by a factor of only 1.5 (see Methods), whereas onaverage, IV anesthetics are an order of magnitudemore potent on ligand-gated ion channels than in-haled anesthetics.1 When the various ion channelswere compared in a previous section, it was evidentthat they all respond to propofol within a factor of 10in the low micromolar concentration range, ratherthan in the low nanomolar or picomolar range, aswould be expected for a specific pharmacologicaldrug, supporting the empirical hypothesis that the less

specific drug may be the more useful general anes-thetic. Thus, both diethylether and propofol lackspecificity compared with other inhaled anestheticsand IV anesthetics, respectively, yet each has beenclinically one of the most successful anesthetics intheir respective class.

The fast time resolution resulting from the use ofoutside-out patches in conjunction with a fast solutionexchange system made it possible to discover anenhancing component of anesthetic action in somephenol derivatives, a component normally masked byan overall suppression in the equilibrium condition.

Thus, the different responses by excitatory and inhibi-tory ligand gated ion channels to propofol may simplyreflect an altered balance of inhibiting and enhancingeffects rather than a qualitative difference betweenthese two channel types. Kinetic (as an alternative tomathematical) separation of anesthetic interactions, asdemonstrated in this study, may prove to be a usefultool in identifying relevant functional groups andcharacterizing these interactions.

The fast time resolution has also shown that elec-trophysiological competition experiments are very dif-ficult to interpret when drug action takes longer than

the time required for the agonist to act. In our system,the peak of the current is reached typically after 20 ms,

whereas the fast component of drug action describedhere (Fig. 2) may be considerably slower. When thereis a part of the fast component that is present only inthe open channel conformation, then neither the openchannel nor the equilibrium application would allowsteady-state measurements of the current peak. In thiscase, it would be extremely difficult to experimentallyseparate kinetic from competitive effects. On the otherhand, the experiments involving drug action in the

absence of the agonist clearly demonstrate that there isat least a large component of drug action that is notcompetitive with the agonist. In vivo it would not bepossible to overcome this component by synapses thatare flooded with transmitter agonists. As to the fastcomponent, clearly, a synapse releasing a great deal ofagonist would be suppressed by propofol less than asynapse that released less 5-HT.

In conclusion, at least two separate inhibitory ac-tions on 5-HT3A receptors could be identified forpropofol, whereas the enhancing action seen for thetwo related smaller phenol derivatives could no

longer be detected. 5-HT-dependent and 5-HT-independent interactions could be distinguished forall three drugs. Propofol was less potent than expectedfrom its hydrophobic properties. Underlying mecha-nisms appear to involve the phenolic hydroxyl group,hydrophobic interactions, and steric restrictions.

ACKNOWLEDGMENTS

We thank Dr. J.P. Dilger for carefully reading this manu-script and discussions, Ms von dem Bussche for maintainingthe cell cultures, and Ms M. Meiboom for performingexperiments with 5-hydroxyindole.

REFERENCES

1. Urban BW, Bleckwenn M, Barann M. Interactions of anestheticswith their targets: non-specific, specific or both? Pharmacol Ther2006;111:72970

2. Belelli D, Callachan H, Hill-Venning C, Peters JA, LambertJJ. Interaction of positive allosteric modulators with human andDrosophila recombinant GABA receptors expressed in Xenopuslaevis oocytes. Br J Pharmacol 1996;118:56376

3. Orser BA, McAdam LC, Roder S, MacDonald JF. Generalanaesthetics and their effects on GABA(A) receptor desensitiza-tion. Toxicol Lett 1998;100101:21724

4. Pistis M, Belelli D, Peters JA, Lambert JJ. The interaction ofgeneral anaesthetics with recombinant GABAA and glycinereceptors expressed in Xenopus laevis oocytes: a comparativestudy. Br J Pharmacol 1997;122:170719

5. Borgeat A, Stirnemann HR. [Antiemetic effect of propofol].Anaesthesist 1998;47:91824

6. Kilpatrick GJ, Jones BJ, Tyer MB. The identification and distri-bution of 5-HT3 receptors in rat brain using radioligand bind-ing. Nature 1987;330:746 8

7. Costall B, Naylor RJ. 5-HT3 receptors. Curr Drug Targets CNSNeurol Disord 2004;3:2737

8. Carpenter DO. Neural mechanisms of emesis. Can J PhysiolPharmacol 1990;68:230 6

9. Minami M, Endo T, Hirafuji M, Hamaue N, Liu Y, Hiroshige T,Nemoto M, Saito H, Yoshioka M. Pharmacological aspects ofanticancer drug-induced emesis with emphasis on serotoninrelease and vagal nerve activity. Pharmacol Ther 2003;99:14965

10. Tramer MR, Reynolds DJ, Stoner NS, Moore RA, McQuay HJ.Efficacy of 5-HT3 receptor antagonists in radiotherapy-induced

nausea and vomiting: a quantitative systematic review. Eur JCancer 1998;34:1836 44



12/12

11. Hille B. Ligand-gated channels of fast chemical synapses. In:Hille B, ed. Ion channels of excitable membranes. Sunderland:Sinauer, 2001:16999

12. Barann M, Meder W, Dorner Z, Bruss M, Bonisch H, Gothert M,Urban BW. Recombinant human 5-HT3A receptors in outsideout patches of HEK 293 cells; basic properities and barbiturateeffects. Naunyn Schmiedebergs Arch Pharmacol 2000;362:25565

13. Barann M, Molderings G, Bruss M, Bonisch H, Urban BW,Gothert M. Direct inhibition by cannabinoids of human 5-HT3Areceptors: probable involvement of an allosteric modulatorysite. Br J Pharmacol 2002;137:58996

14. Barann M, Dilger JP, Bonisch H, Gothert M, Dybek A, UrbanBW. Inhibition of 5-HT3 receptors by propofol: equilibrium andkinetic measurements. Neuropharmacology 2000;39:106474

15. Barann M, Friederich P, Retzmann K, Dybek A, Urban BW. Lossof propofol during in vitro experiments. Anesthesiology 2000;93:3101

16. Tetko IV, Tanchuk VY, Kasheva TN, Villa AE. Internet softwarefor the calculation of the lipophilicity and aqueous solubility ofchemical compounds. J Chem Inf Comput Sci 2001;41:24652

17. Linden I. Molecular effects of general anesthetics on human5-HT3A receptors, their kinetics and relevance for postoperativeemesis (German). PhD Thesis. Faculty for Science and Math-ematics, University of Bonn, 2007;36

18. Zhang L, Oz M, Stewart RR, Peoples RW, Weight FF. Volatilegeneral anaesthetic actions on recombinant nACh alpha 7,5-HT3 and chimeric nACh alpha 7-5-HT3 receptors expressed in

Xenopus oocytes. Br J Pharmacol 1997;120:353519. Suzuki T, Koyama H, Sugimoto M, Uchida I, Mashimo T. The

diverse actions of volatile and gaseous anesthetics on human-cloned 5-hydroxytryptamine3 receptors expressed in Xenopusoocytes. Anesthesiology 2002;96:699704

20. Stevens RJ, Rusch D, Davies PA, Raines DE. Molecular proper-ties important for inhaled anesthetic action on human 5-HT3Areceptors. Anesth Analg 2005;100:1696703

21. Jenkins A, Franks NP, Lieb WR. Actions of general anaestheticson 5-HT3 receptors in N1E-115 neuroblastoma cells. Br J Phar-macol 1996;117:150715

22. Machu TK, Harris RA. Alcohols and anesthetics enhance thefunction of 5-hydroxytryptamine3 receptors expressed in Xeno-

pus laevis oocytes. J Pharmacol Exp Ther 1994;271:89890523. Lovinger DM, Sung KW, Zhou Q. Ethanol and trichloroethanol

alter gating of 5-HT3 receptor-channels in NCB-20 neuroblas-

toma cells. Neuropharmacology 2000;39:5617024. Barann M, Ruppert K, Gothert M, Bonisch H. Increasing effect ofethanol on 5-HT3 receptor-mediated 14C-guanidinium influx inN1E-115 neuroblastoma cells. Naunyn Schmiedebergs ArchPharmacol 1995;352:14956

25. Wittmann M, Peters I, Schaaf T, Wartenberg HC, Wirz S,Nadstawek J, Urban BW, Barann M. Effects of morphine onhuman 5-HT3A receptors. Anesth Analg 2006;103:74752

26. Barann M. Pharmacological studies on the cation channel of the5-HT3 receptor and the fast, voltage-dependent sodium channelin cultured neuronal cells (German). PhD Thesis. Faculty forScience and Mathematics, University of Bonn, 1993;70

27. Daly JW, Nishizawa Y, Edwards MW, Waters JA, Aronstam RS.Nicotinic receptor-elicited sodium flux in rat pheochromo-cytoma PC12 cells: effects of agonists, antagonists and non-competitive blockers. Neurochem Res 1991;16:489500

28. Forman SA, Firestone LL, Miller KW. Is agonist self-inhibition atthe nicotinic acetylcholine receptor a nonspecific action? Bio-chemistry 1987;26:280714

29. Frenkel C, Urban BW. Human brain sodium channels as one of themolecular target sites for the new intravenous anaesthetic propofol(2,6-diisopropylphenol). Eur J Pharmacol 1991;208:759

30. Haeseler G, Stormer M, Bufler J, Dengler R, Hecker H, PiepenbrockS, Leuwer M. Propofol blocks human skeletal muscle sodiumchannels in a voltage-dependent manner. Anesth Analg 2001;92:11928

31. Rehberg B, Duch DS. Suppression of central nervous systemsodium channels by propofol. Anesthesiology 1999;91:51220

32. Friederich P, Urban BW. Interaction of intravenous anestheticswith human neuronal potassium currents in relation to clinicalconcentrations. Anesthesiology 1999;91:185360

33. Flood P, Ramirez-Latorre J, Role L. Alpha 4 beta 2 neuronalnicotinic acetylcholine receptors in the central nervous system

are inhibited by isoflurane and propofol, but alpha 7-typenicotinic acetylcholine receptors are unaffected [see comments].Anesthesiology 1997;86:85965

34. Violet JM, Downie DL, Nakisa RC, Lieb WR, Franks NP.Differential sensitivities of mammalian neuronal and musclenicotinic acetylcholine receptors to general anesthetics [seecomments]. Anesthesiology 1997;86:86674

35. Krasowski MD, OShea SM, Rick CE, Whiting PJ, HadinghamKL, Czajkowski C, Harrison NL. Alpha subunit isoform influ-ences GABA(A) receptor modulation by propofol. Neurophar-macology 1997;36:9419

36. Krasowski MD, Koltchine VV, Rick CE, Ye Q, Finn SE, HarrisonNL. Propofol and other intravenous anesthetics have sites ofaction on the gamma-aminobutyric acid type A receptor distinctfrom that for isoflurane. Mol Pharmacol 1998;53:530 8

37. Krasowski MD, Jenkins A, Flood P, Kung AY, Hopfinger AJ,

Harrison NL. General Anesthetic Potencies of a Series of Propo-fol Analogs Correlate with Potency for Potentiation of gamma-Aminobutyric Acid (GABA) Current at the GABA(A) Receptor

but Not with Lipid Solubility. J Pharmacol Exp Ther 2001;297:33851

38. Ahrens J, Haeseler G, Leuwer M, Mohammadi B, Krampfl K,Dengler R, Bufler J. 2,6 Di-tert-butylphenol, a nonanestheticpropofol analog, modulates alpha1beta glycine receptor func-tion in a manner distinct from propofol. Anesth Analg 2004;99:916

39. Beene DL, Price KL, Lester HA, Dougherty DA, Lummis SCR.Tyrosine residues that control binding and gating in the5-hydroxytryptamine3 receptor revealed by unnatural aminoacid mutagenesis. J Neurosci 2004;24:9097105

40. Kooyman AR, van Hooft JA, Vijverberg HPM. 5-Hydroxyindole

slows desensitization of the 5-HT3 receptor-mediated ion cur-rent in N1E-115 neuroblastoma cells. Br J Pharmacol 2006;108:2879

41. James R, Glen JB. Synthesis, biological evaluation, and prelimi-nary structure-activity considerations of a series of alkylphenolsas intravenous anesthetic agents. J Med Chem 1980;23:13507

42. Mohammadi B, Haeseler G, Leuwer M, Dengler R, Krampfl K,Bufler J. Structural requirements of phenol derivatives for directactivation of chloride currents via GABA(A) receptors. Eur

J Pharmacol 2001;421:859143. Stevens RJ, Rusch D, Davies PA, Raines DE. Molecular proper-

ties important for inhaled anesthetic action on human 5-HT3Areceptors. Anaesth Analg 2006;100:1696703

44. Patten D, Foxon GR, Martin KF, Halliwell RF. An electrophys-iological study of the effects of propofol on native neuronalligand-gated ion channels. Clin Exp Pharmacol Physiol 2001;28:4518

45. Lopreato GF, Banerjee P, Mihic SJ. Amino acids in transmem- brane domain two influence anesthetic enhancement ofserotonin-3A receptor function. Brain Res Mol Brain Res2003;118:4551

V l 106 N 3 M h 2008 2008 I t ti l A th i R h S i t 857

Documents

Molecular Actions of Propofol on Human 5-HT3A