Chem.-Biol. Interactions, 81 (1992) 243-269 Elsevier Scientific Publishers Ireland Ltd.

243

SPECIFIC INTERACTIONS OF MERCURY CHLORIDE WITH MEMBRANES AND OTHER LIGANDS AS REVEALED BY MERCURY-NMR

MARIELLE DELNOMDEDIEU a,b, ALAIN BOUDOUt), DINU GEORGESCAULDa and ERICK J. DUFOURC a aCentre de Recherche Paul Pascal, CNRS, Av. A. Sehweitzer, 33600 Pessac and I~Laboratoire d'Ecologie Fondamentale et Ecotoxicologie, URA CNRS 1356, Universitd de Bordeaux L 33405 Talence (France)

(Received June 17th, 1991) (Revision received November 11th, 1991) (Accepted November 20th, 1991)

SUMMARY

High resolution mercury nuclear magnetic resonance (199Hg-NMR) experiments have been performed in order to monitor mercury chemical speciation when HgC12 is added to water solutions and follow mercury binding properties towards biomembranes or other ligands. Variations of 199Hg chemical shifts by several hundred ppm depending upon pH and/or pCl changes or upon ligand or membrane addition afforded to determine the thermodynamic parameters which describe the equilibria between the various species in solution. By comparison to an external reference, the decrease in concentration of mercury species in solution allowed to estimate the amount as well as the thermodynamic paramaters of unlabile mercury-ligand or mercury-membrane complexes. Hence, some buffer molecules can be classified in a scale of increasing complexing power towards Hg(II): EGTA > Tris > HEPES. In contrast, MOPS, Borax, phosphates and acetates show little complexation properties for mercury, in our experimental conditions. Evidence for complexation with phosphatidylethanolamine (PE), phosphatidylserine (PS) and human erythrocyte membranes has been found. Hg(II) does not form complexes with egg phosphatidylcholine membranes. Interaction with PE and PS model membranes can be described by the presence of two mercury sites, one labile, the other unlabile, in the NMR time scale. In the labile site Hg(PE) and Hg(PS)2 would be formed whereas in the unlabile site Correspondence to: Erick J. Dufourc, Centre de Recherche Paul Pascal, CNRS, Av. A. Schweitzer, 33600 Pessac, France. Abbreviations: ~, logarithmic cumulative growth constant; Borax, sodium tetraborate; 5, chemical shift; EGTA, ethylene glycol tetraacetic acid; EPC, egg phosphatidylcholine; HEPES, hydroxy ethyl piperazine ethane sulfonic acid; MOPS, morpholino propane sulfonic acid; NMR, nuclear magnetic resonance; pC1, -log [C1-]; PE, phosphatidylethanolamine; PS, phosphatidylserine; Ri, lipid-tomercury molar ratio; RBC, red blood cell; Tris, tris hydroxy amino methane.

0009-2797/92/$05.00 © 1992 Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland

244

Hg(II) would establish bridges between three PE or PS molecules. Calculated thermodynamic data clearly indicate that PE is a better complexing agent than PS. Evidence is also found that complexation with lipids uses at first the HgC12 species. Interestingly, mercury complexation with ligands or membranes can be completely reversed by addition of decimolar NaC1 solutions. Minute mechanisms for mercury complexation with the primary amine of PE or PS membrane head groups are discussed.

Key words: Mercury-NMR -- Mercury chloride -- Model membranes -- Mercury complexes -- Equilibrium constants -- Chemical speciation

INTRODUCTION

Cell membranes are the first target of exposure to toxicants, thus interactions with the cell membrane represent some of the earliest signs of toxicity. Basically, the contamination of cells by mercury compounds may be described as a sequence of processes starting with interactions between the metal and some ligands of the cell membrane, followed by transport across this biological barrier and reactions with different cytoplasmic and/or nuclear components [1 -4]. The membrane is therefore the biological barrier which controls all bioaccumulation processes as well as the accessibility to the intracellular binding sites. Concerning the binding of mercury to various cellular ligands, available data stress that protein sulfhydryl groups [1], amino groups of the base moiety of nucleic acids [5,6] and amino acid nitrogens [7] are potential binding sites for Hg(II) and methyl mercury. Recently, it was observed that Hg(II) strongly perturbs membranes composed of phospholipids bearing a primary amine on their polar heads. It was thus proposed that PS and PE represent a new class of binding sites for Hg(II) in cell membranes, offering a new mechanism for toxicological effects of the metal at the membrane level [8,9-11]. However, little is known about the complexation of mercury with lipids. In particular, the stoichiometry, the equilibrium constants of the various lipid-mercury complexes, as well as the mechanisms for complexation are missing. Similarly, the influence of the physicochemical characteristics of the water medium (pH, pC1) and the presence of other ligands (which may form complexes with mercury) on the toxicological properties of inorganic mercury compounds have been scarcelY investigated. The aim of the present study is to better characterize the interaction of mercury with membrane lipids, in terms of stoichiometry and equilibrium constants. The influence of the medium will also be investigated. As it will be shown below, a general model for complexation will be presented and mechanisms for mercury-lipid binding proposed. In order to undertake this work a particularly well adapted and non perturbant technique will be utilized, that is, NMR of the isotope 199 of mercury. Interactions will therefore be monitored at the molecular level and from the viewpoint

245 of the perturbating agent. Few NMR experiments mentioned the l~gHg nucleus except for general chemical shift studies of the II-B group nuclei of the periodic table [12,13]. To our knowledge, this nucleus has never been employed for biological or toxicological investigations. Hence, it appeared necessary to determine its capabilities and limitations as being a good reporter in the frame of our subject, prior to using it as a molecular probe of mercury-membrane interactions. As a first and compulsory step, NMR investigations of mercury chloride in solution were undertaken. It is well known [14,15] that when HgC12 is dissolved in water, there exist complex equilibria among a wide variety of mercury chemical species (Hg 2+, HgOH +, Hg(OH)2, HgC1 +, HgC12, HgCl.(, HgCl~~ -, HgOHC1) depending upon pH and chloride concentration (pC1). NMR of mercury will be shown herein to be capable of differentiation between some of these chemical species and hence will be used to follow mercury chemical speciation [16]. Membrane reconstitution is very often performed in buffer solutions. As a second and logical step, various molecules used for biological buffer preparations were chosen in order to check their possible complexation properties towards mercury: tris hydroxy amino methane (Tris), hydroxypiperazine ethane sulfonic acid (HEPES), sodium tetraborate (Borax) and morpholino propane sulfonic acid (MOPS). In order to compare the observed effects, a well-known mercury complexing agent has been tested as well: ethyleneglycol tetraacetic acid (EGTA). Finally, in order to compare with previous results [8], interactions of Hg(II) with membranes made of phosphatidylethanolamine (PE), phosphatidylserine

(PS) and egg phosphatidylcholine(EPC) have been monitored by I:~:'Hg-NMR. First results on human erythrocyte ghosts will also be presented. MATERIALS AND METHODS HgC12 was obtained from Prolabo (Paris, France), MOPS, Tris, HEPES, Borax, EGTA and NaC1 from Sigma (St Louis, MO). All reagents were divalent cation free and deionized water was utilized (Millipore MQ system, USA). Lyophilized PS and phosphatidylethanolamine PE were purchased from Lipid Products (Nutfield, U.K.) and used without further purification. EPC was prepared according to the method of Singleton et al. [17]. In order to correlate with previous studies [8] an acetate buffer was used from pH = 4 . 0 - 6.0 (74 mM CH3COOH, NaOH) and a phosphate buffer from pH = 6 . 0 - 9 . 0 (50 raM, KH2PO4).

Sample preparation Some experiments were carried out in the presence of an external reference to allow the determination of Hg(II) concentration in solution ([Hg(II)aq]). It was constituted of a small glass tube (4 mm diameter over 20 mm long) filled with 100 ~l of a concentrated solution of mercury chloride (700 mM HgC12, pH = 5.50, pCl = 0.00) and subsequently sealed under inert atmosphere, pH and pC1 conditions were chosen such that the NMR line coming from this external reference, placed in the 10 mm diameter sample tube, did not interfere with signals coming from sample resonance lines.

246

Samples used to study mercury chemical speciation in solution were prepared without the external reference. Initial mercury chloride concentration ranged between 6 and 500 mM, and pH and pCl varied in the ranges from 3.00 to 9.00 and 4.00 to -0.79, respectively. The effect of ligands and chelates was investigated for different concentration ranges by addition of the corresponding amount of dry powder to one ml of a 100 mM Hg(II) solution. Experiments performed with EGTA and Tris included the external reference. In order to study mercury-membrane interactions, a 100 mM solution of mercury chloride was prepared (pH -- 5.68, pC1 = 2.22). Lyophilized phospholipids were added to one ml of this solution and homogenized above the phase transition temperature of the phospholipid, to obtain multilamellar vesicles. The external reference was included in samples, p H and pCl values were measured prior and after completion of the NMR experiment. Isolated human erythrocyte membranes were prepared in two steps adapted from standard methods [18]: (i) wash with an isosmotic solution (154 mM NaC1) and centrifugation (3000 rev./min, 5 rain) and (ii) hemolysis with a phosphate buffer (pH -- 5.40) without NaC1 and centrifugation (20 000 rev./min, 20 min). The second step was renewed until the supernantant was free of hemoglobin. A Mallassez cell counter was used before hemolysis to determine the average number of red blood cells per ml of buffer solution. pH and pC1 were measured on a Schott Gerate TR 155 instrument. A chloride selective electrode (Orion, Switzerland) was used for pC1 determination. Mercury concentrations were controlled by Atomic Absorption Spectrometry Without Flame (Varian AA 475).

NMR spectroscopy High resolution 199Hg-NMR experiments were performed on a Bruker MSL 200 spectrometer, operating at 35.7 MHz in the unlocked mode and at room temperature (20°C). Signals were recorded with a spectral window of 100 kHz, ~r/2 pulses of 6 t~s and a recycle time of 5 s. Acquisition was performed with a one pulse sequence and full phase cycling of both receiver and transmitter. 199Hg chemical shifts, (~obs, a r e expressed relative to neat (CH3)2Hg (0 ppm), as proposed by Garth Kidd and Goodfellow [13]. 5obs was measured to ±2 ppm. HgCl2 solutions yielded single Lorentzian lines. Linewidth at half intensity was about 100 Hz, without sample rotation, and spin-lattice relaxation of the order of 0.5-1 s. Standard Bruker programs allowed to determine NMR line areas for samples and for the external reference, thus affording measurements of [Hg(II)~q]. Further data treatment was accomplished on a VAX 8600 computer. RESULTS

Mercury chemical speciation in aqueous solution Three series of experiments were at first designed in order to determine and analyse the influence of physicochemical characteristics of the medium (pH and pC1, mercury concentration and composition of the buffer solution) on the Hg(II)

247 chemical speciation and on the results obtained by 199Hg-NMR. Investigations on Hg(II) solutions (20 mM) in a large pC1 range were thus performed at pH = 3.0, 5.4 and 9.0. The maximum pC1 value for each pH value corresponds to that measured without NaC1 addition, that is, 3.7 for pH = 3.0; 2.74 for pH = 5.4 and 1.4 for pH -- 9.0. Basically, when pH increases, the pCl value decreases due to an increase of the dissociation degree of HgC12. 199Hg-NMR spectra recorded at pH = 5.4, for a decreasing pC1 from 2.74 (no NaCl added) to - 0.79 (NaC1 added) are presented on Fig. 1. In these experimental conditions, a single NMR line is observed with a good signal to noise ratio, for about 2500 scans. One notices that a pC1 variation from 2.74 to - 0.79 induces a progressive shift upfield of approx. 400 ppm, without any line broadening. As soon as pC1 is lower than 0, it is noteworthy that 5ob~remains unchanged. Similar chemical shifts were recorded at pH = 3.0 (pC1 from 3.7 to -0.79) and pH = 9.0 (pC1 from 1.4 to -0.79), at comparable pC1 values. For pH = 9.0, a brown-red precipitate is observed at pCl values higher than 1.4. According to Hahne et Kroontje [14] in these conditions, the main species formed is Hg(OH)2, but it immediately precipitates for concentrations greater than 0.337 mM. Unfortunately, it is not possible to obtain a high resolution NMR line for such a sample: the corresponding solid state NMR spectrum is very wide, because of the very large 199Hg chemical shift anisotropy [13]. Earlier reports [12] concluded to a mercury concentration dependence of the 199Hg-NMR chemical shift. Consequently, in order to establish the possible involvement of the Hg(II) concentration in the NMR line shifts observed, experiments were also performed at different mercury, Hg(II), concentrations (6 mM, 100 mM, 500 mM) with a decreasing pC1 range from 2.74 to -0.79 (pH = 5.4). When the Hg(II) concentration of the sample was sufficient to obtain a good enough signal-to-noise ratio, NMR spectra recorded for the different mercury concentrations were similar to those presented on Fig. 1, for a same pC1 value. Furthermore, it was noted that a NMR line could be detected with a Hg(II) concentration as low as 5 mM, with a large number of scans. Even if there is no mention about a possible complexation between the Hg nucleus and (CHsCO0-) molecules in Hahne and Kroontje's mercury chemical speciation diagrams [14], other data available in the literature [19] mentioned this fact, by giving high cumulative growth constants for 'Hg(II)-(CH3CO0-).' chemical species. Therefore, in order to insure of the absence of mercury complexation with this buffer solution, especially with (CH3COO)n molecules, and to correlate with previous studies [8], the influence of three different aqueous buffered media was investigated. Samples (pH -- 6.0, 100 mM HgC12) were prepared separately in distilled water (pH adjusted with NaOH), in acetate buffer and in phosphate buffer, as described previously. Depending on the buffered solution employed, very weak ~obs variations were observed and only at pC1 values higher than 2.5. For pC1 ranging from 2.0 to - 0.79 chemical shifts recorded for the three buffered solutions were similar (data not shown).

Mercury with possible complexing agents Some of the 199Hg-NMR spectra recorded for Hg(II) in the presence of in-

248

pCl

2.7'4

2.0

1.5

1.2

1.0

-0.79 ! -1500

! -1000

PPM

Fig. 1. mgHg-NMR spectra of a HgC12 solution (20 mM) as a function of pCl = -log[C1-], pH = 5.4, acetate buffer, 74 mM. Experimental parameters: spectral window, 100 kHz; r/2 pulse, 6 ~s; recycle time, 5 s; approx. 2500 acquisitions. Chemical shift scale is expressed relative to neat Hg(CH~)2 (0 ppm).

249

creasing concentration of EGTA are presented on Fig. 2. The left NMR line corresponds to the external reference signal and the right line was recorded for the sample. Values for 5obs, pH and pCl, linewidth at half height, Avl/2, and the effective mercury concentration in solution, determined by comparison with the area of the external reference, are summarized in Table I. For EGTA concentrations from 0 to 2 0 - 3 0 mM, one observes an important NMR line shift of about 75 ppm to higher frequencies. 5obs remains constant for higher EGTA concentrations (> 30-100 mM). It must be mentioned that for these latter samples a white precipitate was observed in the NMR tube. When EGTA concentration increases from 0 to 3 0 - 4 0 mM pH decreases by about 2.5 units and the pC1 value reaches 1.62 for 100 mM EGTA. This latter observation reflects an increase of the chloride ion concentration, [C1- ], in the medium. For higher EGTA concentration, both pH and pC1 values remain constant. Mercury concentration in solution ([Hg(II)aq]) is proportional to the area of the NMR high resolution signal. [Hg(II)aq] decreases progressively when EGTA concentration increases in the medium and seems to reach a plateau at about 60 raM, when EGTA concentration is as high as 30 - 40 mM. A line broadening is also observed with increasing EGTA content, but it is not considered as significant, due to small experimental variations in magnetic field homogeneity from sample to sample. A high chloride concentration effect was performed (pCl = -0.79) on a 100-mM Hg(II) sample in the absence and the presence of 100 mM EGTA. In both cases 5obs = - 1186 ppm with no line broadening, no signal decrease and no precipitate. The effect of a concentration range 0 - 2 0 0 mM of Tris on a 100-raM Hg(II) sample was also investigated in the presence of the external reference. Parameters obtained from 199Hg-NMR spectra, or measured directly on samples are presented in Table II. An increasing Tris concentration induces a progressive upfield lineshift. For concentrations higher than 150 mM 5ob~seems to reach a plateau and an additional increase of Tris does not change the line position. No observable precipitate is detected on this series of samples. For increasing amounts of Tris, the pH value rises weakly: at 150 mM, the variation is close to 0.6 unit. In the same range, pCl decreases from 2.22 for the control to 1.50 for 150 mM Tris. [Hg(II)aq] basically decreases when Tris concentration increases, but more progressively than observed for EGTA. At the same time, a progressive and significant line broadening is measured. This phenomenon begins for small concentration of Tris and for 200 mM, Avl/2 is more than one order of magnitude higher than that for the control. As in the case of EGTA, a high chloride concentration effect was performed on a control sample in the presence and the absence of 200 mM Tris. Again, for both samples, 5obs = -1186 ppm, without NMR signal decrease nor line broadening or precipitate. Experiments with HEPES, MOPS and Borax were carried out without the external reference. Therefore, quantitative data concerning Hg(II)aq are not available. A 0 - 3 5 mM range of H E P E S induces a progressive shift upfield from - 1585 ppm to - 1559 ppm which then levels off for higher concentrations. The line width was not affected by the addition of H E P E S a n d n o precipitate was observed in samples. A pC1 decrease from 2.22 (0 mM HEPES) to 1.9 for 150

250

! -1200

! -1400 PPM

Fig. 2. 199Hg-NMR spectra of a 100 mM Hg(II) solution in the absence and presence of various EGTA concentrations: (1) 0 raM; (2) 10 raM; (3) 20 raM; (4) 30 raM; (5) 60 raM; (6) 100 raM. The NMR line on the left is that of an external reference (70 mM Hg(II)). Experimental NMR parameters as in Fig. 1.

251

TABLE I EXPERIMENTAL PARAMETERS FOR THE Hg(II)-EGTA INTERACTION Accuracy of measurements: 5obs± 2 ppm; pH :e 0.05 ; pCl ± 0.03 ; remaining concentration of Hg(II) in solution, [Hg(II)aq], ± 10%; linewidth at half height, 5Vl/2 + 40 Hz. [EGTA] (mM)

8obs (ppm)

pH

pCl

[Hg(II)aq] (mM)

AVl/2 (Hz)

0 5 10 20 30 40 50 60 80 100

-1585 -1565 -1544 -1515 -1511 -1511 -1511 -1511 -1511 -1510

5.68 5.68 5.68 4.33 3.35 3.22 3.25 3.22 3.22 3.21

2.22 2.22 2.07 1.70 1.62 1.60 1.62 1.62 1.62 1.62

100 100 75 85 72 67 59 51 62 58

140 135 140 100 245 320 315 220 245 245

mM HEPES, is measured correlatively. Experiments in same experimental conditions were realized for MOPS (0-100 mM) and Borax (0 - 150 mM). Both did not induce significant variation of (~obs, n o precipitate and no line broadening were observed and pCl values remained unchanged for increasing concentration of Borax or MOPS. A high chloride concentration effect (pC1 = - 0.79) was performed on a control sample in the presence or absence of 150 mM HEPES, MOPS or Borax. ~obs recorded for all these samples was again -1186 ppm, similar to the control, without NMR signal decrease nor line broadening or precipitate in the sample.

TABLE II EXPERIMENTAL PARAMETERS FOR THE Hg(II)-TRIS INTERACTION Accuracy of measurements as in Table I. [Tris] (raM)

6ohs (ppm)

pH

pCl

[Hg(II)aq] (mM)

5v112 (Hz)

0 5 10 20 30 40 50 100 150 200

-1585 -1581 -1576 -1568 -1564 -1557 -1556 -1537 -1528 -1528

5.68 5.98 5.87 5.92 5.95 5.95 5.97 6.10 6.25 6.25

2.22 2.28 2.19 2.14 2.04 1.96 1.82 1.60 1.50 1.50

100 100 97 97 95 96 90 80 70 60

100 150 210 300 390 530 450 875 955 1400

252 TABLE III STRUCTURAL AND THERMODYNAMICAL DATA FOR MERCURY CHLORIDE S P E C I E S IN SOLUTION Species

/~a

~b

50c

HgCl 2 HgCl3HgCI 2 -

13.22 14.07 15.07

13.15 14.25 15.20

- 1605 - 1350 - 1180

aCumulative growth constants (logarithmic value) from Sillen and Martell [22]. bCumulative growth constants (logarithmic value) from the present work. cChemical shift (:e 5 ppm) for the isolated species.

Mercury with phospholipid model membranes Interactions between Hg(II) and multibilayers vesicles composed of PS, PE and EPC were performed with a 100-mM mercury concentration, at initial pH and pC1 values of 5.68 and 2.22, respectively, and in the presence of the external reference. 199Hg-NMR spectra recorded for PS concentrations ranging from 0 to 100 mM are presented on Fig. 3. The left line corresponds to the external reference signal and the right one to the sample. Data calculated from spectra or measured on samples are summarized in Table IV. As PS concentration increases, different phenomena are observed. (~obs is progressively shifted towards higher frequencies, to reach a maximum displacement of 144 ppm. Correlatively, a decrease of the NMR signal intensity and an important line broadening are observed. Higher PS concentrations were not feasible, considering the amount of phospholipid required in the one ml sample and the number of scans necessary to obtain a good enough signal-to-noise ratio for the NMR line. One remarks in Table IV that, as PS concentration increases, pH values decrease weakly by about 0.7 unit for 100 m M P S . At the opposite, it induces a very important pC1 variation: 1.10 for the 100 mM PS, compared to 2.22 for the control. This reflects a chloride ion concentration increase in the medium from 6 to 80 mM. Similar experiments were carried out with PE multibilayers. Data obtained for a PE concentration range from 0 to 52 mM are summarized in Table V. An increasing concentration of PE in a 100 mM Hg(II) solution induces, as observed for PS, a progressive shift upfield of about 146 ppm for 52 mM, without reaching a plateau value. On comparing to PS, a similar upfield shift is observed with twice less PE. An important pH decrease of about one unit for 52 mM of PE is measured, with a large pCl decrease from 2.22 for the control, to 1.01 for the 52 mM PE sample. Both variations were greater than those recorded in the presence of PS. For a same concentration, the linewidth was similar for PS and PE. It is also noteworthy that [Hg(II)aq] decreases very drastically with increasing concentrations of PE. A high chloride concentration effect performed on a control and on 52 mM PE or 100 mM PS samples yielded 6obs = - 1186 ppm for the three samples. Again no NMR signal decrease, no precipitate and no line broadening were detected.

253

0 mi

5 mM

12.5 m M

! 31 m M

62.5 m M

| 100 m i

|

!

-1200

-1400

PPM Fig. 3. ~:~!~Hg-NMRspectra of a 100 mM Hg(II) solution in the absence and the presence of various concentrations of PS, as indicated on the right hand side of spectra. The NMR line on the left is that of an external reference (70 mM Hg(II)). Experimental NMR parameters as in Fig. 1.

254 T A B L E IV E X P E R I M E N T A L P A R A M E T E R S FOR T H E Hg(II)-PS I N T E R A C T I O N

Accuracy of measurements is as in Table I. Ri

[PS] (raM)

5ohs (ppm)

pH

pCl

[Hg(II)aq] (raM)

/~vl/,~ (Hz)

0 0.010 0.050 0.125 0.310 0.625 1.000

0 1.0 5.0 12.5 31.0 62.5 100.0

-1587 -1584 -1565 -1543 -1513 - 1475 -1443

5.68 n.d. n.d. n.d. 5.46 5.24 4.97

2.22 2.22 1.72 1.73 1.44 1,27 1.10

100 74 74 73 45 37 28

170 170 170 220 360 330 830

In order to correlate these NMR results with previous studies performed by fluorescence polarization [8], the same experiments were also undertaken with EPC multibilayers vesicles. When EPC concentration increases, no significant shift of ~iobsis recorded in a 0 - 1 0 0 mM concentration range. The 199Hg-NMR signal intensity remains the same, with no line broadening (data not shown). A high chloride concentration effect gave the same results as reported for previous samples. Finally, a preliminary study was carried out with a more complex, but more biologically significant membrane system, that is, human erythrocyte membranes in a range from 0 to 16 • 1011 red blood cells (RBC) per ml of buffer. 199Hg-NMR spectra in the presence and the absence of these ghost suspensions are presented on Fig. 4. No external reference was used. When the concentration increases up to 16 • 1011 RBC/ml, a small but significant ~iobs shift towards high frequencies is recorded, together with a broadening of the NMR line. With this last sample, it was necessary to increase the number of scans to obtain a good enough signal-to-noise ratio. This reflects the decrease of [Hg(II)aq], similarly to what has been observed with model membranes of PS and PE. TABLE V E X P E R I M E N T A L P A R A M E T E R S F O R T H E Hg(II)-PE I N T E R A C T I O N

Accuracy of measurements as in Table I. Ri

[PE] (raM)

~tobs (ppm)

pH

pC1

[Hg(II)aq] (mM)

z~v112 (Hz)

0 0.052 0.104 0.251 0.520

0 5.0 10,4 25.1 52.0

-

5.68 n.d. n.d. n.d. 4.60

2.22 1.87 1.68 1.32 1.01

100 80 n.d. 62 22

170 n.d. 210 350 540

1587 1582 1573 1529 1441

255

,

I -1500

I -1600

PPM

Fig. 4. 199Hg-NMR spectra of a 100 mM Hg(II) solution in the absence and the presence of human erythrocyte membranes. (1) 0 RBC per ml; (2) 2.7 • 1011 RBC per ml; (3) 16 • l0 ll RBC per ml. Experimental NMR parameters as in Fig. 1.

DISCUSSION

Experimental results clearly indicate that 199Hg-NMR is a very powerful technique to monitor changes in the vicinity of the mercury nucleus, depending upon medium conditions, or related to a particular complexation with ligands or membranes. Available NMR parameters are the observed mercury chemical shift, ~tobs, the area and the width at half height of the NMR line. All these parameters report on mercury-ligand interactions, the term ligand being taken here as very general, i.e. it can represent hydroxide or chloride ions, chelates or membrane phospholipid head groups. Specifically, 5obsreports on the close electronic environment of the mercury nucleus, that is, on increases or decreases of electron density due to binding with ligands. The area under the line yields a direct estimate of mercury in solution and the linewidths some insight on the lifetime of metal-ligand complexes. In what follows, some of these NMR parameters together with measured values of pH and pC1 will be used to des-

~ribe, (i) the mercury chemical speciation in solution, (ii) the mercury binding with buffer molecules and (iii) the mechanisms of mercury-lipid interactions. Specific data such as metal-ligand equilibrium constants, stoichiometry of the various complexes and changes in the electronic distribution at the mercury nucleus will be extracted within the frame of a very simple thermodynamic model taking into account liquid-liquid and liquid-solid equilibria.

Mercury chemical speciation When HgC12 is dissolved in water, a plethora of mercury species can be observed depending upon hydroxide and/or chloride ion concentrations. The total mercury concentrations, CM, can then be written as [14]: CM = [Hg ~+] + [HgCl +] + [HgC12] + [HgCl3-] + [HgC12-] + [HgOH +] + [Hg{OH)2] + [HgOHCI]

(1)

The molar fraction of each species, f, can be expressed as a function of [OH - ] or/and [Cl - ]: fHgXn -- ~HgXn " f H g 2+ " IX/n

(2)

where n represents the number of X groups, [X] = [OH-/and/or [Cl-], fHg 2+ the molar fraction of Hg 2+ (fHg 2+ ---- [Hg2+]/CM) and BHgXn the cumulative growth constant of the HgX~ species. For instance,fngoHcl = ~ngOHCl " fHg2+ " [OH-] [Cl-], 6 (logarithmic) values can be obtained from the literature [14,19- 21] and used to calculate the fraction of each species as a function of pH or pC1. A FORTRAN program package (Delnomdedieu and Dufourc, unpublished) was therefore written to represent mercury chemical speciation diagrams for various medium conditions (pH, pC1) (Fig. 5). Figure 5A corresponds to f ffi f(pH) for pC1 values of - 0.79, 2.0 and 4.0, whereas Fig. 5B s h o w s f -- f(pCl) for pH values of 3.0, 5.4 and 7.0, from top to bottom, respectively. Arrows on the right hand side of the dotted vertical lines define the area investigated in our 199Hg-NMR experiments. If we now consider Fig. 5, together with results from Fig. 1, it is clear that when pC1 = 1 and pH -- 5.4 we have three mercury species in solution, i.e. HgC12, HgC13-, HgC12 -, and only one sharp NMR line. This indicates that all these chloro-complexes undergo a fast chemical exchange in the NMR time scale (< 10-4 s), in agreement with earlier reports [12]. This situation leads to the observation of a single Lorentzian NMR line whose chemical shift, 8obs, is governed by the equation: (~obs --

~

(~0 . j~

(3)

i

whereJ} and 8i° represent the fraction and the chemical shift of the isolated species (i = HgC12, HgCl3-, HgC12-), respectively. Equation 3 may be written in a more explicit way as (using Eqn. 2): (~obs = fHg2+ [~HgCIO " ~HgCI2° [CI- ] 2 4- (~OgCl~ " ~HgCI~ [CI- ] 3 ~HgC~24- • [C1- ] 4}

+

(~0HgCI2

(4)

257

pCI-----O.T9

pH=3.0

(A)

,H:I,~"~

HgCI42-

,.., 0.8 £

(B)

Rgcll'

0.4

HgC~0.0

!

.~...-':', ~..J,

!

I

pCl=2.0 ;

~,

,

~

/l.~C~-

)H=5.4

!

,i

}

v

|

!

ti

•

!

w

•

,

•

•

HgCI2

0.8

,~.

•

..J."T'_

|

0

0.4

HsCI~ 0.0

I

i

I

I

\

HsOHC:# . . . . . i ......

"/

pCl=4.0

1.

/

,

•.

pH=7.0

"~I'I'

o.4

HKC~-

ss(osh \

.s~,

4

2

/

/

°- ~P'-.

o.o 3

4

pH

5 = -

log[H3 O+]

6

7

3 pCI

1

0

-1

= - Iog[Cl']

Fig. 5. Mercury speciation diagrams built according to Eqn. 2 of text. (A) Molar fraction of mercury species as a function of pH for pCl values of -0.79, 2.0, 4.0, from top to bottom. (B) Molar fraction of mercury species as a function of pCl for pH values of 3.0, 5.4, 7.0, from top to bottom. Graph areas on right hand side of arrows show our domains of study.

258 This equation can now be used to fit the variation of the observed chemical shift, 5obs, when pC1 increases (Fig. 6). Estimates of 5~gCl2 and 6°gC1~- can be obtained from Fig. 1, for pCl values where these species are at approx. 100 % in the solution. These values were used as starting parameters together with literature ~i values [14, 19- 21] and ~t~cj~- was varied. Finally, optimization of the fit was performed by allowing small v'ariations of b~gCl2, ~t~gc142- and ~i for

!

,

.

!

.

|

'

pH=3.0

E

i

I

-1200

-1400

z

-1600 m , =,

l

i

!

3

I

!

$

t_

j

-1600

I

3

l

,

1

|

•

I

0

|

-1

i

j

-1200

i

t

2

pH=5.4

~ -1400

-!

I

2

I

1 pCI

I

i

0

i

-1

= - Iog[CI-]

Fig. 6. l'~gHg-chemicalshift variation in function of pC1 for pH = 3.0 and 5.4. The solid line represents the fit of experimentalpoints (black circles)accordingto Eqn. 4 of text.

259

each species. The fit is shown on Fig. 6, for two pH values, and yields a very good description of the ~obs variation as a function of pC1. Results are reported in Table III. Our Bi values agree remarkably well with those reported by Sillen and Martell [22] and used by Hahne and Kroontje [14] to build their speciation diagrams. It must be mentioned that the speciation diagrams we report in Fig. 5 are built with our optimized Bi values and literature/~i values. These diagrams are in very close agreement with Hahne and Kroontje graphs, except for the appearance in certain conditions of pH and pC1 of the species HgOHC1 that these authors did not mention in their work. Although this species is never dominant, we nevertheless included it in the analysis, for completeness. The progressive increase in chemical shift from - 1605 ppm for HgC12 to - 1350 ppm for HgC13and to -1180 ppm for HgC12- reflects nicely the addition of chloride ions to the Hg(II) coordination sphere, resulting in a progressive deshielding of the mercury nucleus. Experiments performed with increasing mercury chloride concentration, under fixed conditions of pH and pC1, showed little variation of 5ob~, apparently at variance to earlier reports by Krtiger et al. [12]. These authors reported a shielding effect on increasing mercury concentration. If one looks closely at their experimental conditions, they increased the Hg(II) concentration by addition of HgC12 powder to a fixed volume of HC1 solution. Hence, when the number of moles of Hg per mole of solvent increases, more chloromercuric species are formed leading to a decrease of chloride ions in solution and then to an increase of pCl. As we report herein, an increase in pC1 leads to a shielding effect. Therefore, we believe that results by Krfiger et al. are in agreement with ours, simply these authors did not take into account the pC1 values of the solution. We also report that 6obs is not very sensitive to the presence of phosphate or acetate molecules in the buffer solution. This is in contrast with the high ~ values reported by Morel et al. [19] for mercury complexation with (CH3CO0-) species. By including their ~i values for Hg(II)-acetate complexes into Eqn. 2 one would observe (not shown) the speciation diagrams of Fig. 5 to be dominated by these new complexes. Since bobs is not affected by the presence of acetates, we believe that the reported ~i values for Hg(II)-acetate complexes are overestimated.

Mercury complexation with EGTA, Tris and HEPES In the presence of increasing concentrations of EGTA, Tris and H E P E S molecules in the mercury solution, a shift of the observed 199Hg-NMR line to high frequencies was detected together with a decrease of pC1 and variations of pH. As discussed earlier, a decrease of pCl leads to a deshielding of the mercury nucleus. However, the bobs variation monitored in the presence of EGTA, Tris and H E P E S cannot be accounted for solely by pC1 and pH variations (Eqns. 4 and 2).Therefore, there must exist a new mercury chemical species, a mercuryligand complex, participating in the equilibrium in solution. The decrease in pC1, i.e. the release of chloride ions in solution, as the ligand concentration increases also dictates such a remark. This new equilibrium in solution may be characterized by a cumulative growth constant ~L = [HgLn/([Hg 2÷] " [L] n), where n

260

stands for the number of ligands L, [L] is the ligand concentration and [HgL.] that of the complex in solution. This new species will also possess its own mercury chemical shift, 8 °, which should be close to the 6ob~plateau value observed at high ligand concentration. Sharp Lorentzian lines, with no significant broadening, observed in the presence of EGTA and H E P E S indicate that this new species is in fast exchange with the other species in solution. Alternatively, the progressive line broadening measured with Tris addition could be interpreted in two ways.There could be a slowing down of the chemical exchange, in the NMR time scale, or a fast exchange between resonances with very different linewidths. Although one cannot differentiate between these two hypotheses, on the basis of available data, the monotonous increase of •Vl/2 v e r s u s Tris concentration seems to support the second hypothesis. Therefore, the case of fast exchange (Eqn. 3) will be considered herein. Equation 4 can thus be extended to (~obs = fHg 2 + [(~HgCl 0 0 . ~HgCI~ o [C1-]3 2 " f~HgCi2 . [C1- ] 2 + (~HgCI~ + ($Ogcl 2 . ~HgCl2_ . [ C 1 - ] 4

+ (~O. ~ L "

[ L ] n}

(5)

The variation of the NMR frequency versus the ligand concentration, 5obs = •L]), are the experimental data which were fitted using Eqn. 5 with the following adjustable parameters: 13L, 6 ° and n. Starting estimates for these parameters can hopefully be obtained by consideration of the following: /~L should be of the order of the growth constants of the mercury chloride species but no larger than that of HgC12- since high chloride concentration avoids complexation. Secondly, it is reasonable to think that 5° will be close to the plateau value obtained for high ligand concentration. The number of ligands in the complex, n, is more difficult to adjust since no starting estimate can be guessed from the data. Successive adjustments of these three parameters are finally performed to fit ~obs = if[L]). The progressive decrease of the observed 199Hg-NMR signal in the presence of EGTA and Tris can be interpreted, as previously described for Hg(OH)2, by the appearance of a very large NMR line corresponding to a species in slow exchange with other species in solution. This species could be a 'solid' as evidenced by the appearance of a white precipitate in the presence of high EGTA concentrations, or a colloidal suspension, invisible by eye, in the case of Tris. As already mentioned such a 'solid' species is not detectable with our high resolution NMR technique, and therefore does not participate to the mercury signal in solution, Hg{II)aq. This 'solid' species can be accounted for by the equilibrium equation [22]: HgL,, + bL KL ¢, HgLn + b (solid), KL =

[HgLn + b] [HgLn] [L]b

(6)

where [HgLn] and [L] represent the concentrations of the soluble complex and of the ligand in solution; b leads to the stoechiometry of the 'solid' species and

261

[HgLn÷b] stands for its concentration. This latter can be expressed as a fraction, a, of the initial mercury concentration in the sample, a can be obtained by subtracting the area under the NMR line (concentration of Hg(II) in solution) from that corresponding to the initial mercury concentration. The apparent formation constant, KL, can thus be written, in function of the available experimental parameters: KL =

-.......

ot

(1 - o~)fHgL [L] ~

(7)

where fHgLn represents the fraction of the Hg(II)-ligand complex in solution, according to Eqn. 2, in which X must be replaced by L and/~H~L, by/~L- Equation 7 can now be used to fit the mercury concentration in the solid complex (a = (100 mM - [Hg(II)aq]/100 mM) versus the ligand concentration [L]. The values obtained for ~L and n during the first fit (Eqn. 5, complex in solution) will then be used to determine by this second fit the parameters KL and b characterizing the 'solid' complex. In order to summarize the data analysis, equilibria parameters are extracted from the experimental data in two successive steps: (i) values characterizing the mercury-ligand complex in solution (5(~, /~L, n) will be obtained by fitting 5ob~ = if[L]) with Eqn. 5 whereas (ii) constants describing the 'solid' complex (KL, b) will be given from the analysis of a = if[L]) with Eqn. (7). Data collected in Tables I and II were subjected to such a treatment, pH and pCl variations monitored during experiments were taken into account for calculations. As an example, Fig. 7 displays experimental points for 5,,b.~and ~, versus Tris concentration, together with the fitting curves. All fitting values calculated for EGTA, Tris and H E P E S are summarized in Table VI. Among all ligands used in this study, EGTA has the strongest/3 L. This is in agreement with the well known chelating effect of EGTA towards mercury. Both EGTA and Tris molecules lead to an apparent stoichiometry for the complex in solution of approx. 2. However, 5o determined for the soluble species HgL2 indicates that EGTA induces a larger deshielding effect as compared to Tris, for a same number of ligand molecules. The apparent stoichiometry (n + b) of the 'solid' complex is close to 3 for both molecules. This could lead, due to the peculiar structure of EGTA (tetradentate, two nitrogen atoms), to the appearance of a network of 'solid' complexes giving rise to the observed precipitate at high ligand concentration. Limited informations are available for HEPES. The corresponding ~L is lower than that observed in the presence of EGTA or Tris and the stoichiometry of the complex in solution is close to 1, i.e. formation of Hg(HEPES). On comparing the three ligands investigated with respect to their ability to form mercury complexes in solution (~L) or to perturb the electronic environment at the mercury nucleus (5(~) during complexation, one can class them on a scale of increasing complexation efficiency as: EGTA > Tris > HEPES. In our experiments, MOPS and Borax molecules did not induce complexation, as detected by NMR. These buffer molecules can therefore be used with confidence when studying mercury interactions with other species. Care must be taken with EGTA, Tris or H E P E S if used for similar

262

s: -1520

e~ t-

c~

-1560

z yaw

-1600

I

i

I

i

i

I

0.4

g

0.0

0

40

80 [TRIS]

120

160

200

(raM)

Fig. 7. Top: 199Hg-chemical shift variation in function of Tris concentration. The solid line represents the fit of experimental points (black circles) according to Eqn. 5 of text. Bottom: fraction of bound mercury, ~ (see text), in function of Tris concentration. The solid line represents the fit of experimental points (black circles) according to Eqn. 7 of text.

studies since these molecules have been shown herein to complex Hg(II). Finally, the observation of an identical chemical shift, ~obs, in the presence or absence of EGTA, Tris and H E P E S under high chloride concentration (pC1 = -0.79) can easily be accounted for on the basis of our results. The ~HgCl 2 - value is 15.20,

263 TABLE VI STRUCTURAL AND THERMODYNAMICAL P A R A M E T E R S FOR MERCURY COMPLEXATION W I T H LIGANDS AND P H O S P H O L I P I D S Cumulative growth constants,/~L, and formation constants, KL, are expressed as logarithms. Accuracy in calculations: EL 4. 0.1, n 4- 0.2, (~L0 4- 5 ppm, K L ± 0.3, b 4- 0.2. ~tL° is expressed relative to neat Hg(CH3)2. L

EGTA

Tris

HEPES

PS

PE

~L

14.10 2.30 - 1509 0.4 0.4

12.25 2.20 - 1534 0.7 1..0

10.80 1.20 - 1558 n.d. n.d.

13.35 2.15 - 1505 1.4 0.6

12.70 1.10 - 1590 3.5 2.0

n d ° (ppm) KL b

i.e. much greater than any ~L, leading to the conclusion that Hg(II) affinity is higher for four chloride ions than with any other ligands present in the solution.

Mercury-lipid-interaction Addition of phospholipids greatly affects the recorded NMR line and the physicochemical characteristics of the medium. The general behaviour detected in the presence of ligands is also observed with PS and PE multibilayers and to some extent with human erythrocyte membranes. Increasing concentration of these membrane systems induces line shifts to higher frequencies which reflects a strong deshielding of the mercury nucleus. As previously described, the NMR signal area in solution lowers down, line width increases and an important release of chloride is detected in the medium. Consequently, the model developed to account for mercury binding to EGTA, Tris and HEPES can be applied in the case of phospholipids. However, the metal-ligand complexes in solution and 'solid' as previously introduced must be formally adapted to the specific structure of added membranes. A phospholipid membrane is not a water-soluble ligand. It has lamellar symmetry and must be considered as an organized medium. The 'solid' complex can now be defined as a strong binding of mercury to membrane sites. The mercury atom in this situation would be in slow exchange with the rest of the metal in solution, and give rise to a very large and anisotropic (due to the intrinsic anisotropy of the membrane) NMR signal, undetectable with our experimental conditions. Furthermore, taking only into account pH and pCl variations (Eqns. 3 and 4) does not allow to describe the observed ~iobs variation. Therefore, there must be a mercury-lipid species 'in solution', characterized by (~O, ~L and n according to our model (Eqn. 5) and participating to the Hg(II) signal in solution. From the NMR point of view the mercury bound to this species is considered to be in fast exchange with the other species, in the NMR time scale. As a consequence, our previous model describing the mercury-ligand interaction by two equilibria, one for species in solution, the other for the formation of solid-like complexes can be adapted to describe mercury-membrane interactions by the presence of two binding sites. The site where the mercury-lipid complex has the properties of a liquid, from the NMR viewpoint, will be called labile

264 whereas the other site in which mercury is strongly bound to the membrane will be called unlabile. The labile complex will be characterized by 5o, /~L and n coming from the fit of 5obs = if[lipid]) with Eqn. 5 whereas the unlabile complex will be described by K L and b resulting from the fit of a = f ([lipid]) with Eqn. (7). Experimental data collected in Tables IV and V were subjected to such an analysis. As an example, Fig. 8 displays experimental points for ~obsand a versus PS concentrations, together with fitting curves. All fit parameters calculated for PS and PE are reported in Table VI. Both lipids show comparable ~L values with however different stoichiometries, n, for the labile complex. The labile site for mercury is composed of two lipids with PS and only one in the case of PE. Since n values are different for PS and PE, it is difficult to compare ~L values. However, one can enter these values (/~L, n) in the program used to graphically represent speciation diagrams (vide supra) and calculate as a function of pC1 the fraction of different mercury species in solution. Figure 9 shows the result of such a calculation, for a PS (middle) or PE (bottom) concentration of 50 mM, with an initial Hg(II) concentration of 100 mM. Top panel shows the control. This figure clearly indicates that mercury-lipid complexes are formed by using at first Hg(II) from the HgC12 species. Furthermore, it can be noticed that the domain of existence of the Hg(PE) complex is greater than that for the Hg(PS)2 complex. These diagrams also account very nicely for the absence of complexation observed with high chloride concentrations (low pC1). Both complexes disappear when pC1 ___ 0.5, that is, when the chloride concentration in the solution reaches approx. 300 mM. This clearly indicates that mercury complexation with PS or PE membranes can be completely reversed by salty solutions. From the 6 ° values observed in the presence of PS and PE, it appears that 2 PS molecules deshield the mercury nucleus much more that one PE molecule. This is certainly in relation with the differences in electric charges of the two head groups. The net negative charge on the PS head group is therefore more efficient in decreasing the electron density at the nucleus than the zwitterionic PE molecule. However the binding appears to be stronger with PE than with PS. A steric and/or electronic reason could be invoked: to some extent the COOH group would disadvantage the interaction of mercury with PS molecules, by decreasing the accessibility to the amine group. This group has indeed been recently proposed as a common mercury binding site in PE and PS lipids [8]. This lack of accessibility could also be proposed to account for the absence of interaction with the nitrogen of EPC buried under three methyl groups. However, in this case, the net positive charge on the nitrogen also disfavours the interaction with Hg(II). For both unlabile complexes, the stoichiometry (n + b) is close to 3, i.e. formation of Hg(PS)3 and Hg(PE)3, with a KL value larger for PE than for PS. Our results allow to go further in the mechanistic interpretation of the mercury-lipid complexation. The binding could be described in two steps, different for each of the lipids. The first step would be an approach to the membrane surface and a light binding to two PS or one PE molecules. The second step could be described by a strong Hg(II) bridging of three lipid molecules. This could result in some networking of the membrane surface, hence to capping phenomena leading in turn to an increase in ordering of the lipid molecules. This agrees well with the

265

-1400

~ c._

-1500

z.

-1600

0.8 t_

t..

0 b. 0

0.4

° 0~ r~ L

0.0 0

20

40 [PS]

60

80

0

(mM)

Fig. 8. Top: 199Hg-chemical shift variation in function of PS concentration. The solid line represents the fit of experimental points (black circles) according to Eqn. 5 of text. Bottom: fraction of bound mercury, ~ (see text), in function of PS concentration. The solid line represents the fit of experimental points (black circles) according to Eqn. 7 of text.

/

HgCI 20.8

HgC! 2

/

O

i_ @

0.4

0.0

IIII

HgCI42"

Hg(PS}2

0.8

'

/

,

/ \ I

L

0.4

,\

/ t

HgCI

HgCI3

0.0

0.4

Y~-.

,gczi.

",."~

HgCI2 7 / 2

1

I

~ 0

t I

-1

pCI = - log[Cl'l Fig. 9. Mercury speciation diagrams, built according to Eqn. 2 of text, as a function of pCl. Top: Hg(II) speciation in solution. Middle: Hg(II) speciation in the presence of 50 mMPS. Bottom: Hg(II)

speciation in the presence of 50 mM PE.

267

previously reported increase, upon Hg(II) addition, of the polarization ratio of diphenylhexatriene embedded in such membranes [8]. Such capping phenomena have been recently evidenced when polylysines [23] or calcium ions [24] are added to membranes and shown, by solid state deuterium NMR, to considerably increase the lipid chain ordering. Preliminary results obtained for Hg(II)-interaction with erythrocyte membranes pointed out phenomena similar to those observed with phospholipids. This is in fact-not surprising if one considers that PS and PE represent 45% of the total phospholipid of the red blood cell membrane [25]. However, the rather small mercury deshielding recorded in the presence of ghosts, as compared to that monitored for PS and PE membranes, suggests that other binding sites, such as protein thiol or amino groups, could be involved in the mercury complexation. It would then be very interesting to determine by 199Hg-NMR the binding parameters of these sites in order to compare with those reported herein for certain membrane lipids. CONCLUSION

The 199Hg-NMR technique has been shown herein to be very sensitive to mercury chemical speciation in solution and to mercury binding to ligands or membrane lipids. Very large chemical shift variations or decrease of the mercury signal in solution have been used to characterize equilibria between species in solution and formation of solid-like mercury complexes within the frame of a very simple thermodynamic model. Data analysis according to this model affords the determination of equilibrium constants and stoichiometries. Some buffer molecules have been shown to strongly complex mercury species and must be avoided for quantitative studies with other ligands. Moreover, the calculated thermodynamic parameters allow to classify ligands in a scale of increasing complexing power towards mercury. Mercury interaction has also been evidenced with lipids bearing a primary amine on their polar heads. Equilibrium constants and stoichiometries of mercury-lipid complexes have been determined using the same thermodynamic model. Minute mechanisms for interaction have been put forward from thermodynamic data. PE lipids have been shown to better complex inorganic mercury than phosphatidylserine lipids. From the structure of the unlabile lipid-mercury complexes, the membrane ordering effect promoted by addition of inorganic mercury can be accounted for by capping phenomena on the polar head group. Very interestingly, reversal of the interaction can be performed by addition of a 0.3 M solution of NaC1 and nicely be accounted for by our thermodynamic model. It is clear that the methodology presented herein can also be extended to follow mercury interaction with proteins and characterize their Hg(II) binding sites. Other forms of mercury, e.g. mercury nitrate or organic mercury (CH3HgX) which plays a very important role in ecotoxicology, could also be subjected to such a molecular approach. Other metal toxicants, e.g. cadmium, might also be followed by the NMR technique. However, from the work presented herein, a necessary survey of their chemical speciation in solution and influences

268

of medium (pH, pC1, ligands) must at first be performed prior to the study of interactions with membranes. Interestingly, the non-perturbant NMR technique can be used to follow metal interactions with very complex membranes such as those of red blood cells. This offers new possibilities to investigate, at the molecular level, the toxic effects of pollutants on real systems. ACKNOWLEDGMENTS

This work was supported in part by the E.E.C. Research Program (DG XII) and the French Minist~re de la Recherche et de la Technologie. REFERENCES 1. A. Boudou and F. Ribeyre, Fundamental concepts in aquatic Ecotoxicology, in: Aquatic Ecotoxicology: Fundamental Concepts and Methodologies, CRC Press, Boca Raton, Vo]. I, 1989, pp. 35-75. 2 J. Gutknecht, Organic mercury (Hg 2+) transport through lipid bilayer membranes, J. Membr. Biol., 61 (1981) 61-66. 3 J.R. Lakowicz and C.J. Anderson, Permeability of lipid bilayers to methylmercury chloride: quantification by fluorescence quenching of a carbazole labeled phospholipid, Chem.-Biol. Interact., 30 (1980) 309- 323. 4 E. Bienvenue, A. Boudou, J.P. Desmaz~s, C. Gavach, D. Georgescauld, J. Sandeaux, R. Sandeaux and P. Seta, Transport of mercury accross bimolecular lipid membranes: effect of lipid composition, pH and chloride concentration. Chem.-Biol. Interact., 48 (1984) 91 - 101. 5 L.S. Kan and N.C. Li, Nuclear magnetic resonance studies of mercury (II) interaction with nucleotides in dimethyl sulfoxide, J. Am. Chem. Soc., 92 (1970) 4823-4827. 6 E. Buncel, A.R. Norris, W.J. Racz and S.E. Taylor, Metal ion-biomolecule interactions. Synthesis, spectroscopic and magnetic resonance investigations of methylmercury (II) complexes of the nucleotides guanosine andinosine, Inorg. Chem., 20 (1981) 98-103. 7 R.S. Reid and B. Podanyi, A proton NMR study of the glycine-mercury(II) system in aqueous solution, J. Inorg. Biochem., 32 (1988) 183-195. 8 M. Delnomdedieu, A. Boudou, J.P. Desmazes and D. Georgescauld, Interaction of mercury chloride with the primary amine group of model membranes containing phosphatidylserine and phosphatidylethanolamine, Biochim. Biophys. Acta, 986 (1989) 191-199. 9 M. Delnomdedieu, A. Boudou, J.P. Desmazes, J.F. Faucon and D. Georgescauld, Inorganic mercury-phospholipidic membrane interactions: fundamental role of headgroups bearing a primary amine. A fluorescence polarization study, in: J.P. Vernet (Ed.), Heavy Metals in the Environment, Vol. 1, CEP Consultants Ltd, Edinburgh, 1989, pp. 578-581. 10 A. Boudou, M. Delnomdedieu, D. Georgescauld, F. Ribeyre and E. Saouter, Fundamantal roles of biological barriers in mercury accumulation and transfer in freshwater ecosystems. Analysis at organism, organ cell and molecular levels. 56 (1991) 807- 821. 11 M. Delnomdedieu, Le mercure: mise en ~vidence et caract~risation au niveau mol~ulaire de ses interactions avec les phospholipides des membranes. Th~se de Doctorat, Universit~ de Bordeaux I, France, 1990. 12 H. Krfiger, O. Lutz, A. Nolle and A. Schwenk, Fourier transform nuclear magnetic resonance of 199Hg. Z. Physik A, 273 (1975) 325-330. 13 R. Garth Kidd and R.J. Goodfellow, Mercury 199'2°1, in: P.K. Harris and Mann (Eds.), NMR and the Periodic Table, Academic Press, London, 1978, pp. 266-278. 14 H.C.H. Hahne and W. Kroontje, The simultaneous effect of pH and chloride concentrations upon mercury as a pollutant. Soil Sci. Am. Proc., 37 (1973) 838-843. 15 C.F. Baes and R.E. Mesmer, in: J.O. Nriagu (Ed.), The Hydrolysis of Cations, Wiley and Sons, New York, 1976, pp. 310-312.

269

16 17 18 19

20 21

22 23

24

25

M. Delnomdedieu, D. Georgescauld, A. Boudou and E.J. Dufourc, Mercury-199 NMR: a tool to follow chemical speciation of mercury compounds, Bull. Magn. Reson., 11 (1990) 420. W.S. Singleton, M.S. Gray, M.L. Brown and J.L. White, Chromatographically homogeneous lecithin from egg phospholipids, J. Am. Oil Chem. Soc., 42 (1965) 53-56. J.T. Dodge, C. Mitchell and D.J. Hanahan, The preparation and chemical characteristics of hemoglobin free ghosts of human erythrocytes, Arch. Biochem. Biophys., ]00 (1963) 1] 9 - 130. F.M.M. Morel, R.E. Mc Duff and J.J. Morgan, Interactions and chemostasis in aquatic chemical systems: role of pH, pE, solubility and complexation, in: P.C. Singer (Ed.), Trace Metals and Metal-organic Interactions in Natural Waters, Ann. Arbor. Sci. Publ. Inc. (Ann Arbor), 1983, pp. 157-200. R.M. Smith and A.E. Martell, Critical Stability Constants. Inorganic Chemistry, Vol. 4, Plenum Press, New York, 1976. D. Dyrssen and M. Wedborg, Major and minor element, chemical speciation in estuarine waters, in: E. Olausson and I. Cato (Eds.), Chemistry and Biogeochemistry of Estuaries, John Wiley, Chichester, 1980. L.G. Sillen and A.E. Martell, Stability Constants of Metal-ion Complexes. Special publication No 17. The Chemical Society, Burlington House, London, 1964. G. Laroche, E.J. Dufourc, M. P~zolet and J. Dufourcq, Coupled changes between lipid order and polypeptide conformation at the membrane surface. A ZH-NMR and Raman study of polylysine-phosphatidic acid systems, Biochemistry, 29 (1990) 6460-6465. G. Laroche, E.J. Dufourc, J. Dufourcq and M. P~zolet, Structure and dyna:nics of dimyristoylphosphatidic acid/calcium complexes by ZH-NMR, infrared and Raman spectroscopies and small-angle X-ray diffraction, Biochemistry, 30 (1991) 3105-3114. G.B. Ansell, R.M.C. Dawson and J.M. Hawthorne (Eds.), Table 9, In: Form and Function of Phospholipids, Elsevier, Amsterdam, 1973, pp. 454.