Adsorption of gentamicin onto a bilayer lipid membrane

153

Bioelectrochemisiryand Bioenergetics, 23 (1990) 153-160 A section of J. Electroanal. Chem, and constituting Vol. 298 (1990)

Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands

Adsorption of gentamicin onto a bilayer lipid membrane P. Butko *, 2. Salamon ** and H.T. Tien *** Membrane BiophysicsLaboratory, Department of Physiology, Michigan Sfare Uniuersity#East Lunsing, MI 48824 (U.S.A.)

(Received 13 May 1989; in revised form 15 November 1989)

ABSTRACT Addition of the arninoglycoside antibiotic, gentamicin (GM), to one side of a biiayer lipid membrane (BLM) results in a potential difference across the membrane. Evidence is presented that the membrane potential is caused by the adsorption of GM, bearing four positive charges, on the BLM surface. The experimental results are subjected to a quantitative analysis based on the double-layer theory and the Langmuir adsorption isotherm. The adsorption is saturated (i.e., the BLM is fully covered) at the bulk GM concentration of about 80 pmol/l. At this point, the calculated GM-induced increase in the BLM surface charge density is u = 0.0054 C m-‘, which is equivalent to one positive charge per 50 lipids or one molecule of GM per 200 lipids.

‘INTRODUCTION

Gentamicin (GM) is an aminoglycoside antibiotic. Its formula is shown in Fig. 1. GM shares many structural and functional features with other antibiotics containing streptamin or its derivatives, such as streptomycin, neomycin and kanamycin. Aminoglycosides are generally hydrophilic basic compounds active against a broad spectrum of both Gram-positive and Gram-negative bacteria [l]. Despite their wide use, the molecular mechanism of their action has not been firmly established. There is substantial evidence that aminoglycosides bind to ribosomes and interfere with protein synthesis [2]. In order to be able to do this, they must, obviously, first cross the cell membrane. Anand et al. [3] and Dubin et al. [4] found that E. co/i cells accumulated radioactively labeled streptomycin in two distinct phases. The quick On leave from the Institute of Animal Physiology, Slovak Academy of Sciences. Kosice, Czechoslovakia. ** On leave from the Institute of Physics, Teclmical University, Poznan, Poland. *** To whom correspondence should be addressed. l

0302-4598/90/$03.50 Q 1990 - Blsevier Sequoia S.A.

154

Fig. 1. The goss structure of gentamicin C,, [7].

first phase, completed in less than one minute, was followed by the second phase after a lag of ten minutes, with a half-rise time of approximately 15 min. The slow secondary uptake, but not the fast primary one, is eliminated by the inhibition of protein synthesis, the blocking of electron transport, and the uncoupling of oxidative phosphorylation [3,5]. These results suggest that the transport of aminoglycosides across the cell membrane may be an active process requiring energy and, possibly, the presence of nonlipid membrane constituents. Nevertheless, the effects of aminoglycoside antibiotics on membrane lipids have also been documented. Specifically, aminoglycosides competitively bind to Ca’+ binding sites on a model phospholipid membrane 161; they increase surface pressure in lipid monomolecular ‘: films [7]; and they affect the electrophoretic mobility of the phospholipid liposomes [S]. Here, we describe for the first time the interaction of gentamicin with a bilayer lipid membrane (BLM) as investigated by means of sensitive direct electrical measurements [g-12]. Our results indicate that, when added to one side of a BLM, the GM molecules adsorb onto the BLM surface. This results in a membrane potential difference. A similar phenomenon has been observed previously with many other ions (UO:‘) by Sukharev et al. [13]. EXPERIMENTAL

BLMs were created by injecting a small amount of membrane-forming solution [lW lecithin in dodecane/butanol (3): (l)] into an orifice (diameter 1.2 mm) in the wall of a Teflon cup bathed by 0.01 M KC1 and 0.5 M sucrose. Spontaneous &inning of the membrane up to the black stage was followed optically through a low-power microscope, as well as electrically by observing an increase in the membrane capacitance. The bathing solutions on both sides of the BLM were continuously stirred. Electrical contact between the solutions and the measuring circuit occurred through a pair of Ag/AgCl electrodes with saturated KC1 bridges. Open-circuit vdtagc, current, and capacitance were measured with a Keithley 616 digital electrometer, Keithley 6108 electrometer and an ICE Electronics I-6 low level capacitance meter, respectively. Gentamicin (Sigma Chemicals) was used in the form of gentamicin sulfate (purity ca. 650 pg gentamicin per mg solid). Molar concentrations were calculated using molecular mass 500. Ionic strength and pH

155

were adjusted by varying the KC1 concentration and by the addition of a small amount of concentrated HCi or KOH, respectively. Changes in the surface charge of the BLM upon the adsorption of GM were calculated from the Gouy-Chapman equation. RESULTS AND DISCUSSION

In our previous work [14], we used GM as an agent promoting the do&ing of an enzyme at the BLM surface. There we found that the addition of GM to one side of a BLM results in a potential difference across the membrane, the GM side being positive. The present work has been aime9 at elucidating the nature of this membrane potential. Generally, membrane potential can originate from: (1) transmembrane equilibration of the electrochemical potentials of mobile ions (diffusion potential, Donnan potential), (2) an electrical field exerted from fixed charges and electrical polarization of the membrane surfaces (adsorption potential, surface potential), (3) a hydrostatic pressure difference (electrokinetic streaming potential), (4) a transmembrane temperature gradient, and (5) redox reactions at the membrane interfaces [15,16]. Trivial as it may seem, it is not easy to differentiate experimentally between the types of potentials. In our case, certain conditions of the experiments (namely, the absence of redox species, a temperature gradient, and a hydrostatic pressure gradient) allowed us to disregard the last three kinds of potentials. The Nemst equation, which describes the diffusion potential, and the Gouy equation, which describes the surface potential, both give the same linear dependency between the potential and the logarithm of the ion concentration [9,X]. Considering the diffusional origin of the GM-induced potential, one has to take into account the following: (1) GM creates potential in the absence of a transmembrane concentration gradient of any other ion (K+, Cl-). Therefore, even if it acted as a K- or Cl- ionophore, no potential should be generated, since there is no driving force for an ion movement. (2) GM does not decrease the electrical resistance of the BLM, which ionophores do very drastically [9,11]. (3) GM itself is not likely to cross the BLM, due to its relatively large molecular size and a negligible oil-water partition coefficient [7]. Thus, it is not probable that the GM-induced potential is connected to a diffusional process. Our experimental results give support to the adsorptional origin of the potential. Figure 2 provides three important pieces of information: (1) the effect of GM is nearly saturated at a concentration of about 80 pmol/l; (2) this effect is describable by the Langmuir isotherm (see irtset); and (3) GM does not change the electrical resistances of the BLM (current i and voltage E change in parallel). The particular curve presented was obtained on a BLM containing organic electron conductor tetracyano-p-quinodimethane (TCNQ), as described in earlier studies [14,16]. Therefore, the resistances, calculated from Ohm’s law, are only about 3 Ma cm2. Data obtained on BLMs made from pure lecithin (with resistances of 100 Ma cm*) were, unfortunately, more scattered, but they showed the same concentration dependency.

156

- 220 - 180

2

B

- 140

30 -

- 100

0

0.02

Oo

20

40

0.10

0.12

a.14

[CM]%W

80

100

-

60

-

20

l

.

I

0.04

0.06

0.08

0.16

0.18

0.20

0.2'2-20

[GM]lmM

Fig. 2. The gentamicin-induced potential AE and current I as functions of the gentamicin concentration on one side of the BLM. Bathing solution: 0.01 M KC1 and 0.5 M sucrose, pH 5.8. All the experimental points of this representative data were obtained on the same single BLM.

We concluded that, within the range of resistance changes from 1 to 100 MO cm2, we did not find any relationship between the magnitude of the GM-induced potential and the membrane resistance. The data of Fig. 2 show that the measured potential fits the Langmuir isotherm; the data were also used for subsequent quantitative analysis. The isotherm can be written in the form l/c, = l/c,

+ l/( kc,c)

(1)

where c,, c, and c are the surface concentration of the adsorbed molecules, the maximum surface concentration when the surface is fully covered, and the bulk concentration, respectively, and k is the equilibrium constant of the adsorption. Let us assume that the measured potential AE originates solely from the GM cations adsorbed on the surface of the BLM: AE=Kc,

(2)

where K is a proportionality AE = Kc,

when

l/c+0

constant. The combination of eqns. (1) and (2) yields (3)

Since GM has four positive charges, full coverage of the BLM surface necessarily results in an increase in the net surface charge density (I at the BLM/water interface

157

by the value of u = 4 x 1.6 X 10s6 x c, C m-‘. Thus, AE = Ku,/(6.4 x 10-‘9)

(4)

On the other hand, according to the double-layer theory AE = (2RT/F)

- (TI/(2RTd))“‘~o

(5)

where (I is the same surface charge density as in eqn. (4), E is the dielectric permittivity of water (taken as 7.12 x 10 -lo J-t C2 m-l), and I is the ionic strength (in our experiment 10 mmol/l = 10 mol m-‘). Comparing eqn. (4) with eqn. (5) we obtain K = (2RT,‘E) - ( H(2RTf1))“2

- 6.4

x lo-l9

(6) Substituting this value and the experimentally determined value of AE = 0.083 V (found from the inset of Fig. 2 as a reciprocal of the value of (AE)-’ at c-’ extrapolated to zero) into eqn. (3), we are able to calculate c,, i.e. the surface concentration of the adsorbed molecules when the BLM is fully covered. We obtained c, = 8.5 X 10” mol m- 2. With the BLM area 1.13 mm’, this is equivalent to 10” molecules of GM per BLM. A BLM of this size contains about 1.6 X 10’” lecithin molecules in one surface (taking the surface of one lecithin molecule 0.7 nm2) [17]. Thus, the fully covered BLM contains approximately one molecule of GM per 2K) lipids, which 1uads to one positive charge of GM per 50 lipids (0. = 0.0054 C mS2 = 0.02 e-/0.7 nm’). In order to check the above reasoning, which started from the Langmuir isotherm, we devised an experiment whose results are shown in Fig. 3. Here, the starting point is eqn. (5) the formuIa for the surface potential according to the double layer theory [18], as simplified for the case of high ionic strength P without a high surface charge density o. If the observed membrane potential is caused by the change in surface potential at the BLM/water interface, it must vary with ionic strength according to eqn. (S), i.e., BE as a function of 1-li2 should give a straight line with a slope of (2RT/F)n/(2Rn))1’2u. Figure 3 demonstrates that this is indeed the case for ionic strengths of 10 mmol/l and higher. The slope provides another, independent, means for dete rmining the surface charge density for the observed potential. The calculation yields u = 0.0026 C mS2 = 0.01 e-/0.7 run2 (one charge per 100 lipids or one GM molecule per 400 lipids), which is in accord with the results obtained from adsorption considerations. We think that our work has given substantial support to the premise that the membrane potential induced by GM is a result of the adsorption of the antibiotic polycations onto the BLM surface. Additional evidence can be extracted from Fig. 4, which shows the pH dependence of the potential induced at 80 PM GM. Curve 1 was obtained on BLMs formed from pure lecithin and curve 2 on those containing TCNQ. Here we see that at the higher pH, there is a significant difference between the two curves. At a lower pH, the difference is less and falls within the limits of the large standard deviation of lecithin data. This explains why we failed to observe the difference in the experiment of Fig. 2 (which was carried out at pH = 5.8). The role

1%

?‘

?

1OmM

0

092 (IONIC

I I

2.5mM

I

0.3 0.4 STRENGTH~“2/mM-1’2

I

0,s

I

0.6

Fig. 3. The gentamicin=induced potential as a. function of the ionic strength of the bathing solution, pH 8.4, GM concentration 40 pmol/l. Every point represents an average of at least ihree measurements on different ELMS with the standard deviation shown. Arrows indicate concentrations of WI.

of TCNQ in the process we studied apparently deserves more attention and warrants further experiments. Lecithin is a zwitterionic lipid, supposed to be neutral at pH from 3.5 up to 11 [17]. Watanabe [19] found that the isoelectric point of lecithin at the oil-water interface may be as low as 3.2. This would mean that with increasing pH, lecithin changes its net charge at this point from positive to neutral, due to the appearance of the negative charge of the phosphate group. And the more negative the phosphate group is, the more GM-binding sites appear at the BLM surface. This explains our observation shown in Fig. 4. At pH = 3.5, the net positive charge of the lipid polar headgroups prevents adsorption of the GM cations and no membrane potential is induced. Returning to early observations on the effect of streptomycin on living cells [2,3,5], there is little doubt nowadays that the first step of aminoglycoside action is the BLM/solution interface (the fast phase in the experiments of the Travis group [4]). In the present work, we have attempted to treat the adsorption of GM on a BLM (the best model of the biological membrane so far) quantitatively. Furthermore, we have experimentally proven that the adsorption of the antibiotic to the membrane induces a membrane potential. In an in vivo system, with the potential originally present, the antibiotic would cause serious disturbances in the cell-mem-

159

/

80 / LECITHIN/TCN$/ BLM 0' / 2/'

70 -

FQ "

A’

30-

20-

lo-

0

0

1 3

5

6

7

8

d

Fig. 4. Effect of pH on the magnitude of the gentamicin-induced potential AE (1) Lecithin membrane: every point represents an average of at least three measurements on different BLMs with the standard deviation shown. (2) Lecithin/TCNQ membrane: the single points obtained from measurements on three different BLMs. Bathing solution: 2.5 mM KCI and 125 rnM sucrose. GM concentration 80 ~mol/l.

brane potential. Here, a speculative question arises as to whether such depolarization of the cell membrane cannot in itself be damaging to the cell. In fact, Dubin et al. [4] observed that E. coli cells lost viability at for least 5 tin before the secondary uptake of the drug was registered. ACKNOWLEDGEMENT8

This work was supported by NIH grant GM-14871 and by ONR grant NOO01485-K-0399. Thanks are due to Debbie Benedict for her expert dras.-tingsand to Lynn Anderson for the final typing. REFERENCES 1 I.R. Hooper in H. Umezawa and I.R. Hooper (Eds.), Handbook of Experimental Pharmacology, Vol. 62: Aminoglycoside Antibiotics, Springer Verlag, Berlin,, 2982, pp. l-35. 2 N. Tanaka in ref. 1, pp. 221-266. 3 N. Anand and B.D. Davis, Nature (London), 185 (1960) 22; N. Anand, B.D. Davis and AK. Armitage, ibid., 185 (1960) 23. 4 D.T. Dubin, R. Hancock and B.D. Davis, B&him. Biophys. Acta, 74 (1963) 476. S LE. Bryan and H.M. Van der EIzen, Antimicrob. Agents Chemother., 9 (1976) 928.

160 6 A.P. Corrado, W.A. Prado, I. Pimenta de Morais in M. Rocha de Silva and G. Suarez-Kurtz (Eds.), Concepts of Membranes in Regulation and Excitation, Raven Press, New York, 1975, pp. 201-215. 7 D.E. AusIander, A. Fehneister and 3.J. Sciarrone, J. Pharmaceut. Sci., 64 (1975) 516. 8 A.M. Alexander, I. Ganda, ES. Harpur and J.B. Kayes, J. Antibiotics, 32 (1979) 504. 9 H.T. Tien, Bilayer Lipid Membranes: Theory and Practice, Marcel Dekker, New York, 1974. IO H.T. Tien in G. Dryhurst and K. Niki (Eds.), Redox Chemistry and Interfacial Behavior of Biological Molecules, Plenum Press, New York, 1988, pp. 529-556. 11 R. Blumenthal and R.D. Klausner in G. Paste and G.L. Nicolson @is.], Membrane Reconstitution, EIsevier, Amsterdam, 1982, Ch. 2. 12 I. Ivanov (Ed.), Thin Liquid Films, Marcel Dekker, New York, 1988. 13 I. Sukharev, L.V. Chemomordii, LG. Abidor and Yu.A. Chizmadzhev, Bioeltitrochem. Bioenerg., 9 (1982) 133. 14 Z. Saiamon, L.B. Vitello, J.E. Erman, P. Butko and H.T. Tien, Bioelectrochem. Bioenerg., 21 (1989) 213. 15 S. Ohki in D.A. Cadenhead and J.F. Danielli (Eds.). Progress in Surface and Membrane Science, Vol. 10, Academic Press, New York, 1976, pp. 117-252. 16 H.T. Tien, Etioelectrochem. Bioenerg, 15 (1986) 19. 17 H. Hauser and M.C. Phillips in D.A. Cadenhead and J.F. Danielli (Eds.), Progress in Surface and Membrane Science, Vol. 13, Academic Press, New York, 1979, pp. 297-413. 18 G. Eisenman, G. Szabo, S. Ciani, S. McLaughlin and S. Krasne in J.F. DanieIIi, M.D. Rosenberg and D.A. Cadenhead (Eds.), Progress in Surface and Membrane Science, Voi. 6, Academic Press, New York, 1973, pp. 139-241. 19 A. Watanabe in E. Matijevic and RJ. Good @Is.), Surface and Colloid Science, Vol. 13, Plenum Press, New York, 1984, pp. l-70.