Chemistry and Physics of L#~ids, 62 (1992) 153-163 Scientific Publishers Ireland Ltd.

153

Interactions of aminoglycoside antibiotics with phospholipids. A deuterium nuclear magnetic resonance study Andr6 Schanck a, Marie-Paule Mingeot-Leclercq b, Paul M. Tulkens b, Danielle Carrier c, Ian C.P. Smith d and Harold C. Jarrell d aLabaratoire de Chimie-Physique et de Cristailographie and Research Center for Advanced Materials, bLaboratoire de Chimie Physioioglque and International Institute of Cellular and Molecular Pathology, Universitd Catholique de Louvaln, (Belgium) CDepartment of Biochemistry, University of Ottawa, and dlnstitute for Biological Sciences, National Research Council of Canada, Ottawa,

(Canada) (Received February 17th, 1992; revision received May llth, 1992; accepted May 12th, 1992) The effect of ~,eral aminoglycmide(AG) antibiotic8 on aqueous muitilamellar dispersions of mixtures of phosphatid~linositoi (PI) and deuterated phmphafidylcholine (PC') has been studied by deuterium (2H) NMR. Isepandcin and amikachl gave rise to no significant changes in 2H-NMR fineshape relative to that of the lipid mixture without antibiotic. Both kanamycin A and B, which have a greater aflVmityfor Pl than the other two antibiotics examined in this study, induced temperature-dependent changes in 2H-NMR lineld~q~esand a~ciated spectral moments. The results are consistent with an antibiotic-induced lateral phase separation giving rise to PC_.-enfid~ domains free of drug and PI-AG domains. These effects are correlated with the inhibitory potency of aminoglycusides towar~ PC degradation.

Key words: aminoglycomde; deuterium; interaction; nuclear magnetic resonance; phosphatidylcholine; phosphatidylinositol

Aminoglycoside (AG) antibiotics are hydrophilic molecules consisting of an aminated cyclitol glycosylated with aminosugars. These broad spectrum antibiotics are highly potent and are active against most aerobic Gram negative organisms, staphylococci and Gram positive bacilli. Unfortunately toxic reactions in kidneys (nephrotoxicCorrespondence to: Andr~ Schanck, U ~ M C , B~timent Lavoisier, Place Louis Pasteur, 1, 13-1348 Louvain-la-Neuve,

Belgium. A/~re~iatio~ and symhols: AG, aminnglycoside; DPH, 1,6dipbenyl,l,3,5-1~matri~c; DPPC-~, a~l chain l~rdeuterated [sn-l,2-2H~dil~lmitoyl phosphatidylcholine; LLvQ, quadrupolar splitting of 2H-NMR spectra; IC50, aminoglycoside concentration causing 50% inhibition of phuspholipase Al activity towards PC degradation; MI, fast moment and M2, second moment of 2H-NMR spectra; PC, phusphatidylcholine; PI, phosphatidylinmitol; POPC-d2, l-palmitoyl-2[5-2H2]oleoyl phmphatidyteholine labelled at the 5 position of the oleoyi chain; SC_D, ord~ parameter; To phase transition temperature.

ity) and in ears (ototoxicity) affect between 8 and 30% of patients treated with aminogiycosides [1]. In spite of its reversibility, the nephrotoxicity of these antibiotics is of considerable concern due to the cost and medical burden involved. Aminogiycoside antibiotics are eliminated by urinary excretion, after fdtration through the glomerulus. A small but sitmificant proportion of the administered dose is taken up into proximal tubular cells [2-31 by adsorptive pinocytosis at the luminal side [4], after binding to the brush border membrane [5-6]. A number of studies at the cellular and molecular levels have implicated the plasma membranes, mitochondria and lysosomes, as primary sites of AG-inflicted structural and functional damage. Lysosomal alterations with early disturbances in the catabolism of phospholipids appear, however, to be the most plausible origin of aminoglycoside antibiotic nephrotoxicity. In this respect, a putative sequence of events has been

0009-3084/92/$05.00 © 1992 Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland

154

proposed by Tulkens [7] and Laurent and Tulkens [8]. This sequence envisages that endocytosis and intralysosomal accumulation of these antibiotics within renal proximal tubular cells lead to the storage of undigested phospholipids (phospholipidosis) by inhibition of lysosomal phospholipase activity. Phospholipidosis, after reaching a threshold, somehow triggers tubular cell necrosis. The molecular mechanism underlying the lysosomal phospholipidosis seems to result from the interaction of AG with the substrate which is organized in bilayers containing negatively charged lipids [9-14]. Indeed, it has been proposed that AG-induced inhibition of lysosomal phospholipase activity towards PC degradation first involves binding of the drug to acidic phospholipids such as phosphatidylinositol. Thus, studies on the interaction of this class of antibiotic with membrane phospholipids offer an approach towards elucidating the mechanisms underlying AG-induced nephrotoxicity. The results of such studies may also prove useful in the rational design of less toxic drugs. Conformational analysis has provided some theoretical insight into the interaction of aminoglycoside antibiotics with negatively charged phospholipids. Such studies revealed that the critical parameters in phospholipase inhibition include the extent of the binding of the drug to the negatively charged phospholipid in the bilayer, the energy of interaction between the drug and the phospholipid, the accessibility of the drug to aqueous phase and the orientation of the drug relative to the lipid-water interface. Indeed, the inhibitory potency of AG towards PC degradation decreases with increasing apparent drug accessibility to water and when the drug is positioned parallel to the fatty acid chain direction [12,13,15,16] Various experimental approaches have been used to obtain more ~ knowledge of the interactions between the drug and the lipid matrix. 31p-NMR spectroscopy and f l l ~ ~ depolarization of 1,6-diphenyl-l,3,5-hexatriene (DPH) have been used to probe the efi~t of AG on head group and acyl chain ~ , respectively [17]. G e n t , caused a: si_~t~dfiotmt~ t i o n in the phosphate head group ~ l i t y but no

change in the fluorescence depolarization of DPH. 15N-NMR was used for studying the interactions of AG's with PI and specific bindings of these antibiotics to PI have been observed [18]. In the present study, we have used 2H-NMR of acyl chain-labelled PC's to probe at the molecular level, the effect of AG-lipid interactions on the hydrophobic region of the PC's. 2H-NMR of lipid membranes has proven to be of wide utility in examining phospholipid structural and motional properties since the residual quadrupolar interaction is sensitive to both the motional rates and amplitudes of labelled sites [19-21]. Moreover, the deuterium nucleus is a non-perturbing probe and so is preferred over the bulky spin labels and over the fluorescent probes frequently used for determining order in membranes. Most membranes of eucaryotic cells, including the lysosomal membrane, contain between 10 and 30% negatively charged phospholipids [22-24]. In this study we have used a system confining a mixture of PC and PI, an acidic phospholipid. We compare the effect on phospholipid properties of four AG (isepamicin, amikacin, kanamycin A and kanamycin B, see Fig. 1 for structures and Table I for substituents) which exhibit large differences in their nephrotoxic potential, as evaluated in vitro by their potential to inhibit lysosomal phospholipase A1 and A2 activity towards PC degradation. Amikacin and isepamicin are aminohydroxyacid (butyryl or propionyl)N-1 derivatives of kanamycin A and gentamicin B, respectively. This substitution reduces only slightly the antibacterial activity of the drug towards sensitive

,, ,,H.\

"/.o2" R4NH_ ~,

~

/

~,

Fig. 1. Strufture of the aminog]yc~oside$ used ill thi$ ~ [ y .

155 TABLEI Su~fituen~umciated~theaminoglycosideantibioticsusedinthisstudy. AG

RI

R2

R3

R4

R5

Re

R7

R8

Isepamicin Amikacin Kanamycin A Kanamycin B

OH OH OH NI-I2

OH OH OH OH

OH OH OH OH

COR COR' H H

H CH2OH CH2OH CH2OH

CH 3 OH OH OH

OH H H H

CH 3 H H H

R = -CHOH-CH2-NH 2

R ' =-CHOH-(CH2)2-NH2

strains, but, from high dosage studies, the nephrotoxic potential of isepamicin appeared to be very low [25]. Combined biochemical and morphological studies in vivo also showed that amikacin and isepamicin have a low potential to cause phospholipidosis in animals [7,26] which is rationalized in vitro by the position and the degree of insertion of the antibiotic into acidic phospholipid bilayers [12,13,16,27]. In addition, kanamycin B was chosen for the knowledge of the relationship between structure/activity and stn~ture/toaicity performed on many derivatives of this drug [28-30] and because related compounds such as tobramycin and dibekacin are important pharmaceuticals which show excellent activity against Pseudomonas aeruginosa [31]. Materials aad Methods

dried under a stream of argon and the solvent completely removed under vacuum. The resulting residue was dispersed in deuterium-depleted water and lyophilized overnight. The solid was hydrated with acetate buffer (40 mM, pH 5.4) alone or conmining the antibiotic at the desired concentration. This pH value was chosen because the in vivo lysosomal pH of cells submitted to AG treatment is considered to be - 5 - 5 . 5 [33] and previous biochemical measurements have been performed in this medium [11]. The sample was then subjected to several freeze-thaw cycles and vortexed for several minutes.

Deuterium NMR 2H-NMR data were acquired at 30.7 MHz on a 'home-built' Fourier-transform spectrometer operated by a Nicolet 1280 computer. Spectra were recorded by means of a modified quadrupolar echo sequence [34] with quadrature detection of the echo signals. Typical parameters were: pulse spacing, 60 ~ts; 90° pulse length, 4.7 /~s (10 mm coil); spectral width -t- 100 kHz; recycle time 1 s (DPPC-d62) and 0.1 s (POPC-d2). The frequency of the spectrometer was carefully set at the center of the quadrupolar powder patterns. Samples were enclosed in a glass jacket where the temperature was regulated to ±0.5°C. The quadrupolar splitting AVQ is directly related to the order parameter Sc-v of the C - D bond [19]

PI (sodium salt from wheat germ) was obtained from Lipid Products (Nr Redhill, UK). Two different deuterated PC's were used for this study: acyl chain perdeuterated DPPC-d62 purchased from Avanti Polar Lipid and POPC-d2 labelled at the 5 position of the oleoyl chain prepared as described previously [32]. Isepamicin was obtained from Toyo Jozo (Shizuoka, Japan), amikacin from Bristol Benelux (Brussels, Belgium), kanamycin A was supplied by Continental Pharma (Brussels, Belgium) and kanamycin B by Meiji Seika Kaisha (Tokyo, Japan). Deuterium-depleted water used for all samples was purchased from Aldrich Chemical.

AVQ = (3/4) (AQSC_D)

Preparation of liposomes Chloroform solutions of phospholipids were

where the static deuterium quadrupolar coupling constant AQ=e2qQ/h is about 170 kHz for

156

aliphatie C - D bonds [35] and the order parameter ~C-Dis defined as:

3C-D = (1/2)




where O is the angle between the C - D bond and the axis of symmetry for acyl chain reorientation and the angular brackets signify an ensemble average. AVQ was obtained for POPC-d2 spectra from the oriented-sample spectra ('Depaked') calculated from corresponding powder spectra according to Bloom et al. [36]. When individual linewidth becomes more hnportank the residual quadrupolar splittia8 obtained from the 'depaked' spectrum is more accurate than that meamm~-on the powder spectrum. For DPPC-d62 spectra, spectral moments (Mn) were calculated from ,the symmetric 2H-NMR powder pattern a c c o ~ g to: Io

25°C

f(oJ) do

Mn= ~o f(0~)do~ where o~is the frequency with respe~ to the central I a m ~ r frequency 0~e,f(o)is the l ~ p e and n is the order of the s ~ momeat. In the ease of a multiply labetled lipid (e.g. Imrdeuterated acyl chain) in the fluid phase, the first moment is a measure of the average acyl ~ order parameter [20]. Ramits

DPPC-d62 and AG As a control, we monitored the interaction of isepamicin and kanamycfin A (40 tool%) with multilamellar dispersions of DPPC-d62 as a function of temperature. Typical 2H-NMR spectra above and below the # t o ~ t ~ . _ l i a e , p h M e transition temperature Tc (37°C) are ~ o w n in Fig. 2 for lipid in the absence and presence of kanamycin A. Addition of isepamicin had as 1 ~ influen~ on the speotntm a s : ~ A (a~ttat,.not-t~), The ~igl~t ~ i n l i ~ m ' ~ by

.

~

.

. . . kHz

! :to go. ~: 30. o. -3o.'@. ,9o. kHz

Fi& 2. 2H.NMR ~ . A, DPPC-de2; B, DPPC,,d¢2+ 40 tool% kanamycinA (7500FID). the antibiotic that can, be comidered as an impurity in the phmptmlipid, denees of the first and ~ ~ ~ N M R spectraof D ~ aloae and DPPC-d~2 mixed with 40 moP,6 ofthe two AG are iltestrated in Fig. 3. The small diffecenees are of the order of the experimental ac~airacy (-v 5%). DPPC-d6JPl and A G 2H-NMR spectra of equimolar Dtq~-de2/Pl mixtures c o n ~ IO moP,~ and 20 tool% AG were recorded as a function of t~perature. Some typical spectra are shown in Fig. 4. At low concentration (10 tool%) the AG had no effect on the spectral tiaeshapa (data not shown). At 20 mol% significant ~ are observed, especially at 15 and 25°C, namely the p m e m e of a more p r o n o e m ~ broad ¢mm0qxment. At 35°C, the effect is mailer bet the c h a t ~ in a m ~ of the powder ~ t t e m 0 o u o f ~ ) is a ~ t . These d i ~ in ~ behaVteer are manifest i n t h e , - ~ in, the t e m ~ Fig. 5.

157

N2xlO'9 (s-2} 20

MlXlO-4 I [

(s-Z)

15 10

10 7 5

~5

3~

3~

~o

4~'c

2's

Temperature

~o

3~

4~

,~'c

TemI~rature

Fig. 3. Temperature dependence of the first moment M I and the second moment M 2 of 2I-I-NMR spectra: 0, DPPC.~2; &, DPPCd~2 + 40 tool% kenamydn A; 0, DPPC-d62 + 40 mol% isepamicin.

15%

90. 60. 30. O..30,-60, .90. kHz

35%

25%

90. 60. 30.

.30, .60.-90. kHz

90, 60. 30. O. -30.-60..00, kHz

Fig. 4. 2H-NMR spectra of equimolar mixtures of DPPC-d62/PI: A, controls; B, DPPC-de~/PI+ 20 tool% kanamycin A (5000 FID).

158 H2xlO-9 (s -2) 20

HIXI0"4

\

(s -1) ,5

10

10

i'0

2'0 3'0 T~6ratuEe

40"C

10

2'0 3b Tenp6rature

4¢'C

Fi& 5. Temperaturedependenceof the firstmomentM! and the secondmomentM2 of 2H-NMItspectraof DPPC-de2/PIequimolar mixturesand afteradditionof 20 tool%AO~O , ~ ; ~ + ~ ; A, + ~ A, + kanamycinA ; ~ , B,

Mr) for the systems containing isepamicin and amikacin is essentially the same as that of thelipid mixture without antibiotic. In contrast, the orientational order of mixtures containing 20 mol% kanamycin A or B shows a markedly different temperature dependence relative to the lipid mixture in the absence of AG.

POPC-dJPI and AG POPC exhibits a gel to liquid crystalline transition temperature of - - 5 ° C [32] and therefore an cquimolar mixture of POPC/PI may he expected to have a lower T¢ than -5°(2. Figure 6 shows the temperature dependence of 2H spectra obtained for an equimolar POPC-d2/PI mixture. At -10°C the spectra remain representative of fast axially symmetric motion which is typical for liquid crystaUine phase lipid. ~ quadrupolar splittiugs ~ Q were detmniaed at each temperature fi spectra for the mol% A G and are presented in Fig. 7. The twideal ~ ~~demme

with increasing temperature. In the case of lipid containing AG, the quality of spectra below -5*(2 precluded accurate measurement of ApQ values and these data are not presented in Fig. 7. At -10°C the 2H spectral lineshal)es depend stroogly on which antibiotic is present (Fig. 8). It should be noted that at this temperature, the samples containing isepamicin and, to a smaller extent, amikacin give rise to spectra (Fig. 8A,B) that ~ that obtained for the lipid mixture alone (Fig. 613). Dixmdm In the present study, we use a deuterium labelled PC to monitor the interaction of the aminoglycoside antibiotics with the mixed PC/PI matrix. Sinoe previous ~ ~ that the AG,lip~

159

~vQ (kHz) 30

25

o'

5'

i

i

,5

i

2ooc

Temperature

Fig. 7. Temperature dependence of the quadrupolar splitting A~Q of 2H-NMR spectra of POPC-d2/PI equimolar mixture and after addition of 20 moW, AG. @, POPC-d2/PI; + isepamicin; A, + araikacin; A, + kanamycin A; II, + kanamycin B. The vertical bars indicate the SD of values determined from four spectra at each temperature.

i

60

!

30

1

!

|

0

-30

-60

kHz Fig. 6. 2H-NMR spectra of equimolar mixture of POPC-d2/PI; A, -15°C; B, =10°C; C, + 10eC; D, + 20°C; (100000 FID).

region of the bilayer, we examined the interaction of isepamicin and kanamycin A with DPPC-d62 liposomes. Comparison of temperature dependent 2H spectra for DPPC-d62 with and without AG reveals that the antibiotic has little influence (see Fig. 2 for typical results): the plateau region that is shorter after addition of AG indicates a slightly modified distribution of the order parameters. The small differen¢~ in spectra suggests that the anti-

biotics have little effect on the orientational order parameters along the lipid acyl chain. A more quantitative measure of the effect of AG on acyl chain order may be obtained from the spectral moments M1 and M2 (which are related to the average and mean squared order parameter, respectively) of the acyl chain [37]. The temperature dependence of M! and M2 for lipid with and without antibiotic are shown in Fig. 3. The spectral moments are, to within experimental error, the same, indicating that the presence of antibiotic has no effect on the phase transition temperature of the lipid or on the amplitude of acyl chain fluctuations (order) even in the gel phase. These results indicate that the antibiotics have a minor influence on the motional and structural characteristics of PC molecules in a bilayer. Within the context of the present study, one may conclude that, in the mixed PC/PI systems, changes in 2H spectral characteristics of deuterated PC will reflect the effect of AG-PI interactions on the lipid matrix. 2H-NMR spectra of the DPPC-d62/PI mixture exhibit a very different temperature dependence (Fig. 4) relative to that of DPPC-de2 alone as a

160

i

60, 30

ii

i

0

II

-30

/

1

-60

kHz Fig. 8. 2H-NMR spectra of equimolar mixture of POI~-d2/PI at -10°C after ~ o n of 20 mo~: A, ~ l t , mi'kacin; C, kanamycin B; D, kanamycin A; (100000 FIE)).

result ofthemuchlower tramitioa teml~fatme of PI in the mixture. The mixed ~ 4W,e a g e t t o h'quid cry~_~tine,l~metmmitien w h i ~ o w m ~ o,~r mot~:than J ' ~ ~ i s ~ at - 15'(::. This is readily seen f m m ~ , ~ mnlm~ature dependence of M1 and M2 (Fig. 5). After

addition of antibiotic at 10 mol%, the 2H-NMR spectral parameters stay unchanged (data not shown). At 20 tool%, neither isepamicin nor amikacin modify significantly the 2H-NMR spectra relative to those of the lipid mixture without AG as evidenced by the spectral moments (Fig. 5): neither the average chain orientational order parameter (reflected in Ml) nor its temperature dependence is significantly influenced by the two antibiotics. The presence of kanamycin B and kanamycin A leads to a substantial change in the temperature dependence of the 2H spectra (Fig. 4B) and spectral moments (Fig. 5) relative to that of the lipid mixture alone. Order in the low temperature phase of PC/PI mixture is increased by these two antibiotics as manifest by the increase in the first moment and the temperature of the transition to a less ordered phase is increased. The influence on the order in the high temperature phase of the lipids is much less than on the low temperature phase, but in the same direction. This influence of AG on T¢ of mixed phospholipids also observed by differential scanning calorimetry has been reported for PC/PI liposomes incubated with gentamicin [38] and neomycin [39] but our NMR data are the first direct observations, at a molecular level, of different interactions between a series of AG and model membranes. The results in Fig. 5 may be interpreted as follows: in the presence of kanamycin A and B at 20 tool%, the PI binds antibiotic and forms AG-PI domains leaving the lipid matrix effectively enriched in PC; below the DPPC transition temperature, these PCenriched domains form a gel phase (less mobile acyl chains) detected by the broad component clearly observed at 15 and 25°C (Fig. 413). As a result the thermal behavior of the system, as monitored by 2H-NMR of the labelled PC, resembles more closely that of pure PC. This corresponds to the shift to higher temperature of the large increase in spectral moments seen in Fig. 5 for the systems with kanamyein A and B. Inspection of Fig. 5 suggests that, above T o the average acyl chain order parameter of each mixture is essentially the same (as evidenced by the values of the moments being similar to within the experimental error) indicating that the AG-PI is not influencing significantly the properties of the PC molecules in the liquid crystalline phase.

161 TABLE II Energy of interaction between Pl and AG and degree of inhibition of phospholipase A I activity. Aminoglycoside

Interaction with Pl a

Energy of interactionb (kJ/tool)

IC50c (ms/l)

Isepamicin Amikacin Kanamycin A Kanamycin B

Weak Weak Strong Strong

-23.4 -35.7 -46.8 -48.5

125 485 469 454 4-

[16] [12] [18] [30]

7 4 11 4

"Interaction betw~'--nAG and PI detected by 2H-NMR. bCalculated energy of intcraction with PI for different aminoglycosides. CAminoslycoside concentration causing 50% inhibition of phospholipase A I activity towards PC degradation [16].

To further probe the effects of AG on PC/PI mixtures, the interaction of the antibiotics with POPC/PI mixture was examined. POPC has a Tc of -5°C [32] which allows examination of the POPC/PI system in the liquid crystalline phase over a larger temperature range than was convenient for the DPPC-d~2/PI system. 2H-NMR spectra of an equimolar mixture POPC-d2/PI at several temperatures are shown in Fig. 6. The spectra reflect lipid in the liquid crystalline phase above -10°C and hegin to broaden between -10 and -15°C suggesting the onset of the gel to liquid crystalline phase transition. The results for the DPPC-d62/PI mixture established that AG at 20 tool% had little effect on average acyl chain order in the liquid crystalline phase. The temperature dependence of the residual quadrupolar splitting (reflecting segmental order) at C5 of the POPC oleoyl chain in the presence and absence of antibiotic (20 tool%) is presented in Fig. 7. It is clear that, for POPC in the liquid crystalline phase, segmental order is not influenced by AG-PI interaction because, down to -5°C, the differences of A~,Qvalues are not si~mificant (see the SD in Fig. 7, our measurements at -5°C are accurate to ± 1.5 kHz). In addition, the temperature dependence of the sn-2 C5 segmental order is not influenced by the putative formation of AG-PI domains (in the presence of kanamycin A and B ) over a temperature range of some 25°C in~ the liquid crystalline phase. The results confLrm conclusions reached with the DPPC-de~JPI mixtui'e which indicated that AG-PI interaction does not

significantly perturb the hydrophobic region of the mixed-lipid bilayer as sensed by the PC component. As the temperature is lowered below -10°C, the spectral intensity of the POPC-d2/PI mixture decreases because of broadening. With the small sample sizes available for this study, it was not practical to examine AG-PI interaction below -10°C. A s a result, it was not possible to detect directly whether AG-PI domain formation would effectively restore the gel to liquid crystalline phase transition temperature of POPC-d2 in the AG-lipid mixture as seen for the DPPC-d62/PI system described above. However, inspection of 2H spectra of lipid containing 20 tool% antibiotic obtained at -10°C suggest that some effects of AGPI interaction are manifest. In the presence of isepamicin, the 2H spectrum (Fig. 8A) of POPCd2 has a lineshape and residual quadrupolar splitting which is comparable to that of lipid alone at -10°C (Fig. 613). Amikacin appears tO give effects qualitatively similar to those of isepamicin. However, 2H spectra of lipid in the presence of kanamycin A (Fig. 8D) and kanamycin B (Fig. 8C) at -10°C exhibit a lineshape which is comparable to that of the lipid mixture alone at -15°C (Fig. 6A). Qualitatively, the interaction of kanamycins A and B with the lipid mixture results in the PC acyl chain exhibiting spectral characteristics consistent with lipid undergoing a liquid crystalline to gel transition (if the 2H spectrum of POPC-d2/PI at -15°C is taken to be representative of such a transition). At 20 mol%, these two antibiotics lead to AG-PI domains and concomitantly to regions

162

enriched in POPC which, at -10°C are able to form gel phase lipid. Thus the effects of AG-PI interaction as monitored by the PC lipid is similar for both the saturated DPPC and unsaturated POPC containing lipid mixtures. The 2H-NMR results obtained for DPPC-d62/PI and POPC-d2/PI mixtures are consistent with the following description. The AG antibiotics interact only weakly with PC, as if they were impurities. Amikacin and isepamicin at concentrations up to 20 mol% show no or very weak interaction with PI and as a result do not perturb significantly phospholipid organization in PC/PI mixtures. In contrast, kanamycin A and kanamycin B bind to PI to form AG-PI domains in PC/PI mixtures. When the ratio of AG to Pl is sufficiently high (e.g. 20 mol%) PI-AG domain formation leads to domains enriched in PC. The latter effect is manifest in 2H-NMR spectra of labelled PC as a broad spectral component attributable to gel phase lipid at temperatures above the Tc of the PC/PI mixture alone. The 2H-NMR observations and conclusions from the present study on the influence of aminoglycoside antibiotics on PC/PI mixtures may be correlated to biochemical data and theoretical calculations which are summarized in Table II. The present study demonstrates that interaction of the lipid mixture with either isepamicin or amikacin is weak and teach to little effect on membrane organization. Inspection of Table II reveals that this is consistent with the lower inhibitory potential (higher IC50) towards phospholipase activity reported for these antibiotics [16] and with the lower energies of interaction between the AG and PI. 2H-NMR results suggest that kanamycin A and B interact with PI and can disrupt PC/PI membrane organization with the major effect being formation of PI-AG complexes and domains. This study is in good agreement with 15NNMR results showing that kanamycin A and B interact more strongly with PI than i ~ n and amikacin [18]. Our 2H-NMR study has thus evidenfed two important points. Firstly, the correlation between the present results and those of previous studies confirms that there is a strong relationship be-

twecn the various nephrotoxic potentials of the four AG's investigated and their affinity for PI, a negatively charged phospbolipid. Secondly, for the first time, we are able to propose a molecular mechanism which explains the inhibition of the activity of lysosomal phospholipases towards PC degradation induced by these antibiotics. Indeed, nephrotoxic aminoglycosides, such as kanamycin A and kanamycin B, could sequester PI and therefore reduce the amount of negative charges available for optimal lysosomal phospholipase activity toward PC [11-12] or interfere with the access of the enzyme to the surface of the lipid vesicle containing the substrate [14]. This inhibition of enzyme activity generates phospholipidosis which is at least partially responsible for renal cell necrosis. The formation of domains or the increase in heterogeneity of lateral organization of lipids is likely to have functional as well as structural significance. The presence of specific local domains is believed to be one of the first requirements for the fusion between membranes [40]. In addition, the activity of intrinsic membrane enzymes are known to depend on the types of lipids surrounding them [41]. It is thus reasonable to propose that the sequestration of PI is responsible for the inhibition of lysosomal phospholipases activity, a key event in the sequence of reactions leading to aminoglycoside nephrotoxicity.

Aekm~ts For the gift of aminoglycoside samples, we thank the companies: Bristol Benelux (Brussels, Belgium), Continental Pharma (lkuasels, Belgium), Meiji Seika Kaisha (Tokyo, Japan) and Toyo Jozo (Shisuoka, Japan). The f ~ support of A.S. by Shering-PIOugh International (NJ, USA) and Bristol Myers International (NY, USA) is gratefully acknowledged. This work has been supported by a grant to D.C. from the Natural Sciences aad Ea$iaeeria8 Reeeardt C o u a ~ cad by the National Relmreh Coeaefl, Ottawa. M~-P. M.-L is Cherehe~r ~ of the ~ Fonch National de la Recherche ~ . Sti~ii~in our laboratories also ~ ~ ffe~,~ the Belgian Fonds Nationalde la R ~ . S e l e n t i f i -

163

que (grant no. 1.5213.87), the Programme visant ~t renforcer le Potentiel Scientifique clans les Technologies Nouvelles (PREST) and the Belgian State Prime Minister'sOffice Science Policy Programming (InteruniversityAttractionPoles,Grant no. 7 bis). Refm 1 G. gahlmeter and J.l. Dahlager (1984) J. Antimicrob. Chemother. 13, 9-22. 2 M.J. Kuhar, L.L. Mak and P.S. Lietman (1979) Antimicrob. Agents Chemother. 15, 131-133. 3 A. Vandewalle, N. Farman, J.P. Morin, J,P. FiUastre, P.Y. I-latt and J.P. Bonvalet (1981) Kidney Int. 19, 529-539. 4 V.U. Collier, P.S. Lietman and W.E. Mitch (1979) J. PharmacoL Exp. Ther. 210, 247-251. 5 M. Just and E. I-laberman(1977) Naunyn-Schmiedebergs Arch. Pharmakol. 300, 67-76. 6 F.J. Silverblatt and C. Kuehn (1979) Kidney Int. 15, 335-345. 7 P.M. Tulkens (1986) Am. J. Med. 80 (6B), 105-114. 8 G. Laurent and P.M. Tulkens (1987) in: ISI Atlas of 8cieace/Phatmacology, Waverly Press, Baltimore, pp. 4~-44. 9 G. Lanrent, M.B. Carlier, B. Roilman, F. Van Hoof and P.M. Tulkens (1982) Biochem. Pharmacol. 31, 3861-3870. 10 M.B. Carlier, G. Laurent, P.J. Claes, H.J. Vanderhaeghe and P.M. Tulkens, (1983) Antimicrob. Agents Chemother. 23, 440-449. 11 M.P. Mingeot-Leclercq, G. Laurent and P.M. Tulkeus (1988) Blochem. Pharmacol. 37, 591-599. 12 M.P. Mingoot-Ledercq, J. Piret, R. Brasseur and P.M. Tulkeus, (1990) Biochem. PharmaooL 40, 489-497. 13 M.P. Min~eot-Leclercq, J.Piret, P.M.Tulkens and R. Brasseur (1990) Blochem. Pharmacol. 40, 499-506. 14 K.Y.Hostetler and E.J. Jeilison (1990)J. Pharmacol. Exp. Ther. 254, 188-191. 15 R. Brasseur, M.B. Carlier, G. Lament, PJ. Claes, H.J. Vanderhaeghe, P.M. Tulkens and J.M. Ruysschaert (1985) Biochem. Pharmaool. 34, 1035-1047. 16 P.M. Tulkeus, M.P. Mingeot-Lcelercq,G. Laurent and R. Brasseur (1990) in: R. Brasseur (Ed.), Molecular Description of Biological Membrane Components by Computeraided Conformational Analysis, CRC Press, Boca Raton, FL, Vol. II, pp. 63-93. 17 M.P. Mingeot-Leclerc~, A.N. Schanck, M.F. RonvauxDupal, M. Deicers, R. Brasseur, J.M. R~ysschaert and P.M. Tulkeus (1989) Blochem. Pharmacol. 38, 729-741.

18 A. Schanck, R. Braueur, M.P. MinBeot-Leclercq and P.M. Tulkens (1992) Map. Reson. ~ 30, 11-15. 19 J. Seeli8 (1977) Q. Rev. Biophys. 10, 353--418. 20 J.H. Davis (1983) Bloehim. BiophyL Acta 737, 117-171. 21 I.C.P. Smith, ICW. Butler, A.P. Tulloch, J.H. Davis and M. Bloom (1979) FEBS Lett. 100, 57-61. 22 R.W. MeGilvery (1979) in Biochemistry: A Functional Approach, W.B. Sannders Co., PA, pp. 206-210. 23 D. T h i n ~ u x (1973) in: J.T. Dingle and H.B. Fell (Eds.), Lysosomes in Biology and Pathology, North Holland Publ. Co., Amsterdam, Vol. 3, pp. 278-299. 24 J.D. Esko and C.R.H. Raetz (1983) in: P.D. Bayer (Ed.) The E n z y m e s - Lipid Enzymology, Academic Press, New-York, Vol. 16, pp. 208-253. 25 K. Matsumoto, H. Fujii, H. Miyake, K. Shiraiwa, M. Miura, H. YAmnmotoand A; Saito (1985) Chemotherapy (Tokyo) 33, 47-89. 26 K. Matsumoto, P. Lambrieht, B.K. Kishore, S. Ibrahlm; B. Rolhnann, G. Laurent and P.M. Tulkens (1988) in: 28th Interscience Conf. Antimierob. Agents Chemother., Abstract no. 1503, Los Angeles, CA. 27 R. Brasseur, G. Laurent, J.M. Ruysschacrt and P.M. Tulkeus (1984) Blochem. Pharmacol. 33, 629-637. 28 A. Van Schepdacl, R. Busson, L. Verbist, H.J. Vanderhaeghe, M.P. Mingeot-Ledercq, R. Brasseur and P.M. Tulkens (1991) J. Med. Chem. 34, 1483-1492. 29 A. Van Schepdael, J. Deloourt, M. Mulier, R. Busson, M.P. Mingeot-Ledercq, P.M. Tulkens and P.J. Claes (1991) J. Med. Chem. 34, 1468-1475. 30 M.P. Mingeot-Leclercq,A. Van Schelxiael, R. Brasseur, R. Busson, H.J. Vanderhaeghe, P.J. Clacs and P.M. Tulkeus (1991) J. Med. Chem. 34, 1476-1482. 31 P.S. Lietman (1985) in: G.L. MandeU, R.G. Douglas and J.E. Bennet (Eds.), Principles and Practice of Infectious Diseases, John Wiley and Sons, New-York, pp. 192-206. 32 B. Perly, I.C.P. Smith arid H.C. Jarrell (1985) Biochemistry 24, 1055-1063. 33 S. Okhuma and B. Poole (1978) Proc. Natl. Acad. Sci. USA 75, 3327-3331. 34 J.H. Davis, K.R. Jeffrey, M. Bloom, M.I. Vali¢ and T.P. Higgs (1976) Chem. Phys. Lett. 42, 390-394. 35 J.L. Burnett and B.H. Muller (1971) J. Chem. Phys. 55, 5829-5831. 36 M. Bloom, J.H. Davis and A.L. MacKay (1981) Chem. Phys. Lett. 80, 198-202. 37 J.H. Davis (1979) Biophys. J. 27, 339-358. 38 L.S. Ramsammy and G.J. Kaioyanides (1987) Blochem. Pharmacol. 36, 1179-1181. 39 B.M. Wang, N.D. Weiner, M.G. Ganesan and J. Schacht (1984) Biochem. Pharmacol. 33, 3787-3791. 40 R.K. Scheule (1987) Biochim. Binphys. Acta 899, 185-195. 41 A. Carruthers and D.L. Melchior (1986) Trends Biol. Sci. 11, 331-335.