Alcohol. Vol. 8. pp. 143-150. t~ PnrgamonPress pie. 1991. Printed in the U.S.A.
t1741-83:19/91 S3.00 + .00
Interfacial Dehydration by Alcohols: Hydrogen Bonding of Alcohols to Phospholipids J A N G - S H I N G C H I O U , * C H I H - C H U KUO,i" S H E N G H. L I N , ~ H I R O S H I K A M A Y A * A N D I S S A K U U E D A *l
*Department of Anesthesia, UniversiO' of Utah School of Medicine and Anesthesia Service, VA Medical Center, Salt Lake City, UT 84148 and fDepartment of Chemisto', Arizona State Universio,, Tempe, AZ 85287 R e c e i v e d 1 1 July 1990: A c c e p t e d 27 S e p t e m b e r 1990
CHIOU. J.-S.. C.-C. KUO, S. H. LIN. H. KAMAYA AND I. UEDA. Interfacialdehydration by alcohols: Hydrogen bonding of alcohols to phospholipids. ALCOHOL 8(2) 143-150, 1991.--The interaction between alcohols ¢ethanol and n-butanol) and dipalmitoylphosphatidylcholine (DPPC) in carbon tetrachloride was studied by Fourier transform infrared spectroscopy (FTIR). Upon addition of the alcohols, the P = O stretching band of DPPC at 1260 cm - ~ shifted to lower frequency (red-shift). The red-shift indicates that the P= O vibration became slower possibly because the heavier alcohol molecules replaced the water molecules hydrogen bonded to the PO, moiety. The formation constants between the PO_, group and ethanol or n-butanol (n-butanol data in parenthesis) were 19.0 M-~ (7. I M-~) when estimated from the spectral change. A new absorbance peak appeared at 3265 c m - t (3275 cm-~) representing the DPPC-alcohol complex. The formation constant of this complex was also 19.0 M-~ (7.1 M-~I. The identical formation constant suggests that the DPPC-alcohol complex was formed at the PO., moiety of DPPC with hydrogen bonding to the alcohol OH. At higher alcohol concentrations, the absorbance peak of DPPC-alcohol complex shifted to 3225 cm ~ 13235 cm ~). Apparently, the lower frequency shift at higher alcohol concentration occurred by the formation of alcohol multimers (dimer, trimer, and tetrameri interacting with DPPC. Alcohol-phospholipid interaction
Interface
Hydrogen-bonding
Dehydration
Anesthesia theory
ers the problem of distribution of the drug in the body, it is easy to understand that these expressions do not represent the true potency because the amount of ethanol bound at the action site is not identical with the blood concentration. As early as 1939, Ferguson (6) argued that the true potency must be compared by the amount of drugs bound at the action site rather than the bulk concentration. When expressed by the chemical potential, however, these two values become identical at an equilibrium condition. He proposed to express the potency by "thermodynamic" activity. For volatile compounds, the thermodynamic activity is expressed by the fractional saturation in the gas phase. This is the ratio between the partial pressure of the vapor and the saturated vapor pressure of the pure liquid. This is analogous to the relative humidity that is used to express the dryness in meteorology. One feels mugginess by the relative humidity and not by the actual water content in the air. When expressed by the fractional vapor pressure, all volatile compounds induce anesthesia at about 3% saturation (6,16). Thus, all anesthetics (including alcohols) are equipotent when expressed by the thermodynamic activity (6,16). With ethanol, anesthesia ensues when the end-expiratory ' partial pressure of ethanol reaches 35 tort. Because the activity
THIS study deals with anesthetic effects of alcohols. The general consensus, however, appears to account alcohols as a very weak anesthetic. One of the referees of this article commented that alcohols are not an anesthetic. On the contrary, clinician anesthesiologists know that ethanol is a potent anesthetic when given intravenously. The patients are rapidly anesthetized without the excitement stage. It is wrong to assume that only a handful of compounds have anesthetic potency. Most hydrophobic molecules that perturb lipid membranes are anesthetics, including gases, hydrocarbons, alcohols, ethers, ketones, other organic solvents, and their halogenated counterparts (24). The anesthetic action of xenon is well known. Nitrogen gas is a potent anesthetic at 30 atmospheric pressure. Carbon dioxide is an anesthetic at 0.3 atmospheric pressure. These gases are not clinically applicable because of the high cost of xenon, high pressure required for nitrogen, and pH disturbances associated with carbon dioxide administration (24). The " w e a k " anesthetic potency of alcohols )s another misconception. Potency is evaluated by the amount of drugs required to obtain a desired effect. With ethanol, the potency is often expressed by the blood concentration. However, when one consid-
(Requests for reprints should be addressed to Dr. Issaku Ueda, Anesthesia Service I12A, VA Medical Center, Salt Lake City, LIT 84148.
143
144
CHIOU, KUO, LIN, KAMAYA AND UEDA
coefficient of ethanol in blood is not identical with that in water, the alcohol concentrations in the model system cannot be meaningfully compared with the blood content that induces intoxication. Another problem is the lethal dose (LD50) of alcohols and narcotics. This is because the cause of death is not standardized. Death by ethanol or narcotic overdose occurs usually by the respiratory depression. When the respiration is mechanically supported, death does not occur at the lethal dose. This is evident in clinical anesthesia, where respiration is stopped at the beginning of anesthesia by muscle relaxants and is mechanically controlled. In this situation, the patients tolerate the lethal doses. For instance, the clinical dose of morphine is in the range of 8 to 12 mg. For cardiac operations, however, anesthesiologists use 300 mg and more, which grossly exceeds the lethal dose. The clinical concentrations of all anesthetics are close to lethal doses if the respiration were not supported. Thus, the actions of short-chain alcohols and inhalation anesthetics are similar. They interact with lipid bilayers and decrease the temperature of the main transition between the solid-gel and liquid-crystal phases. The chain-melting or disordering effects correlate extremely well to the anesthetic potency (24). Because membrane order is a property represented by the conformation of lipid tails of phospholipid molecules, the alcohol actions are often,attributed to their effect on the hydrophobic core of the membrane (8,17). Nevertheless, alcohols are dipolar molecules with hydrophilic OH moiety and are expected to reside at the watermembrane interface (1-3, 11. 13, 14). Eyring et al. (5) proposed that anesthetics destruct the clusters of water molecules electrostricted at the charged surface of macromolecules. Di Paolo and Sandorfy (4) have shown that inhalation anesthetics hinder the formation of hydrogen bonds, and the anesthetic potency correlates to the hydrogen-bond-breaking activity. By quantum chemical analyses, Sandorfy and co-workers (9,10) showed that anesthetics can form competitive proton donor-acceptor complex so that H,O-X dimer can be replaced with anesthetic-X dimer. They propose that the "breaking" and "formation" of hydrogen bonds may be central to the mechanism of anesthesia (22). Because alcohols contain hydrophilic OH moiety, they are expected to interact with the hydrophilic moiety of DPPC and break the hydrogen bonds between the membrane surface and water molecules by the competitive mechanism. Klemm (15) also proposed that the alcohol action is caused by the interfacial dehydration based on his biochemical finding on the ethanol-induced depletion of brain sialic acid. We (2, 11-14, 23-25, 27) suggested that anesthesia is an interfacial phenomenon, replacing the surface-bound water with anesthetics. Tsai et al. (23) measured the effect of inhalation anesthetics on water-DMPC (dimyristoylphosphatidylcholine) interaction in a water-in-oil reversed micellar system using FTIR spectroscopy. They reported that the primary hydration site of DMPC was the phosphate moiety, and up to 18 water molecules were restricted at the polar head group. The anesthetics replaced part of the water molecules hydrogen-bonded to the head group, but did not affect the choline signal. The present study measured the interaction between alcohols (ethanol and n-butanol) and DPPC in carbon tetrachloride solution. Hydrogen-bonded complexes between alcohols and DPPC were identified and the formation constants of such complexes were estimated. From the changes in the frequency and shape of IR bands, we determined the binding site of alcohols on the lipid molecule. METHOD
Synthetic dipalmitoylphosphatidylcholine ( 1,2,-dihexadecanoyl-
sn-glycero-3-phosphorylcholine, DPPC) and n-butanol were obtained from Sigma, ethanol from U.S. Industrial Chemicals (Anaheim, CA), and spectroscopic grade carbon tetrachloride from EM Science (Cherry Hill, NJ). DPPC was kept in a desiccator at reduced pressure until use. DPPC was dispersed in CCI 4 by irradiation in a cup-horn of a Branson ultrasonic disrupter (Danbury, CT) at 0.01 or 0.02 M. After the baseline spectra were obtained, alcohols were added to the DPPC-CCI4 mixture with a microsyringe under nitrogen gas, and sealed. Each sample was mixed by a vortex mixer. A Perkin-Elmer (Norwalk, CT) model 1750 FTIR spectrophotometer interfaced with a Perkin-Elmer model 7300 computer was used for the analysis. The cell window (FT 04-035) was a fixed thickness sodium chloride with 1.28 mm pathlength (SpectraTech, Stamford, CT). The cell was calibrated against the benzene band at 845 c m - J. A TGS detector was used for all experiments. Each sample was scanned over the frequency range 400-4000 c m - ] and the spectra were accumulated for 20 scans. The spectra of the solvents were subtracted from the spectra of the solutions. For DPPC-alcohol mixture, the difference spectrum was obtained after subtraction of the component spectrum from the spectrum of the sample mixture. All spectra were acquired at 22°C. RESULTS AND DISCUSSION
Alcohols in CCI~ Figures 1-3 show the IR spectra of n-butanol (A) and ethanol (B) in carbon tetrachloride. When the concentrations of the alcohols were below 0.006 M, only the monomer spectrum was observed (Fig. 1). Dimers began to appear when the alcohol concentration exceeded 0.008 M (Fig. 2). Trimers and tetramers appeared when the concentration exceeded 0.03 M. With the alcohol concentrations above 0.1 M (Fig. 3), all four species were present. According to the increase in the alcohol concentration. the absorbance intensity of monomers decreased and those of dimers to tetramers increased. The IR band (Fig. IA) at 3635 c m - ~ represents the nonbonded free O-H stretching vibration of n-butanol. For ethanol it was 3634 c m - t (Fig. 1B). The shoulder at 3515 c m - ~ is that of the self-associated dimers H-O. • .H-O for both alcohols (Fig. 2). The vibrational bands of trimers are at about 3443 c m - ~ (n-butanol) and 3450 c m - ~ (ethanol) (Fig. 3). The tetramer bands appeared at 3345 cm i (n-butanol) and 3343 c m - ~ (ethanol) (Fig. 3). Two ranges of the alcohol concentrations were studied. The low concentration range covers 0.001 to 0.006 M where only n-butanol monomers are present. The high concentration range covers 0.2 to 0.3 M, where monomers, dimers, trimers and tetramers coexist. AlcohoI-DPPC Mixture Figures 4--6 show the spectra of the alcohoI-DPPC mixture. In Fig. 4, spectrum (a) is 0.01 M DPPC with 0.006 M n-butanol (Fig. 4A) and ethanol (Fig. 4B). The peaks at 3635 c m - ~ (n-butanol, Fig. 4A) and 3634 c m - ] (ethanol, Fig. 4B) are the nonbonded free O-H stretching vibration of the alcohols. Spectrum (b) is the 0.01 M DPPC control in the absence of the alcohols. The broad peak at 3373 c m - ~ is assigned to the O-H stretching band of the water molecules adhering to the hydrophilic parts of DPPC. These water molecules are tightly bound to DPPC despite storage in a desiccator at reduced pressure. The difference spectra of the alcohols (c) were obtained by subtracting the DPPC spectrum (b)
PHOSPHOLIPID-ALCOHOL INTERACTION
145
0.040
0.040
>I,(/) Z ILl I-Z
0.025
0.025
__.__J -0.010
-0.010
I
4000
3550
__5
3100
!
4000
WAVENUMBER
3550
3100
WAVENUMBER
A
B
FIG. 1. IR spectra of n-butanol (A) and ethanol (B) in carbon te~achlofide. The alcohol concentrations are from the top downward; 0.006.0.005.0.004. 0.003, and 0.002 M.
from (a). Spectrum (c) shows DPPC-alcohol interaction. A peak, representing the alcohol-DPPC complex, appeared at about 3275 c m - ~ (n-butanol, Fig. 4A) and 3265 c m - ~ (ethanol, Fig. 4B) in the difference spectrum. At the same time, the intensity of the O-H stretching bands of free alcohols at 3635 c m - ~ (n-butanol) and 3634 c m - ! (ethanol) decreased. Figure 5 shows the difference spectrum of DPPC when interacted with the alcohols. Spectrum (a) was obtained by subtracting the free alcohol 0.006 M spectra from the spectra of the alcohol-DPPC mixture. Spectrum (b) is the 0.01 M DPPC control. By subtracting (b) from (a), the interaction between DPPC and the alcohols is visualized (c). The negative peaks at 3635 c m - ~ (n-butanol, Fig. 5A) and 3634 c m - ! (ethanol, Fig. 5B) signify that the free alcohol concentration is decreased by binding to DPPC. Figure 6 shows the difference spectra at various alcohol concentrations (0.003 M to 0.006 M) and at 0.01 M DPPC. When the alcohol concentration was increased from 0.003 M to 0.006 M, the intensity of the free O-H stretching bands at 3635 c m - ! (free n-butanol, Fig. 6A) and 3634 c m - ~ (free ethanol, Fig. 6B)
decreased, whereas the intensity of 3275 c m - ~ band (n-butanolDPPC complex, Fig. 6A) and 3265 c m - ! band (ethanol-DPPC complex, Fig. 6B) increased. To calculate the association constants of the DPPC-alcohol complexes, we used dilute solutions of the alcohols (0.002--0,006 M) to limit the interaction between the alcohol monomers and the functional groups of DPPC. Let A be the alcohol and L be DPPC, then at low alcohol concentrations, we have Kc [A] + [L] "-s [ALl
CAL KC = CA CL
Z LU I--
0.08
z
-0.02
(2)
where K c is the formation constant of the alcohol-lipid complex, and CAt., CA, and C , are the concentrations of the complex, alcohol, and lipid, respectively. The observed absorbance per unit cell length for this system, A, is expressed as
0.14
>l--
( 1)
and
0.14
t _3
>I--
zLU
-0.02
I
4000
3550 WAVENUMBER
A
SL__
0.08
I-Z
310q
!
3550
4000
WAVENUMBER
B
FIG. 2. Same as Fig. I. A: n-butanol, and B: ethanol. The alcohol concentrations are from the top downward; 0.015, 0.010, and 0.008 M.
3100
146
CHIOU, K U O , LIN, K A M A Y A AND U E D A
0.70
0.70
>,I,-
>I-0')
Z
0.40
Z III pZ
tl.I I-Z
0.40
_J
J
-0.10
3550
!
-0.10
I
4000
4000
31 O0
3550
3100
WAVENUMBER
WAVENUMBER
A
B
FIG. 3. Same as Fig. 1. A: n-butanol, and B: ethanol. The alcohol concentrations are from the top downward; 0.15.0.13, 0.10 M.
A = E A C A + e L C L "~ E A L C A L
(3) AAc-
where EAL, EA, and e L are the respective absorption coefficients o f the complex, alcohol and lipid. By designating AAc as the change in the absorbance o f the association band relative to those o f pure A and L in CC14, Equation 3 is written as ~u~ c
=
A -- E A C A -- e L C L
=
C A L (CA L -- eA -- (EL)
=
A EALCAL
(4)
When C A < C L, we have. approximately,
CAL
KcCACL
(5)
1 + KcC L and
(6)
Equation 6 shows that when C L is kept constant, AA c versus CA is linear (Fig. 7). In Fig. 7, we have plotted the intensities o f O-H and P = O bands against the alcohol concentration at 0.01 M and 0.02 M DPPC. Accoraing to the increase in the alcohol concentration, the intensity o f the DPPC-alcohol association band at 3275 c m - ~ (n-butanol, Fig. 7A) and 3265 c m - ~ (ethanol, Fig. 7B) increased :.'hile the intensities of the O-H stretching band at 3635 c m - ! (n-butanol) and 3634 c m - t (ethanol) decreased. The association band arises from O-H. - . P = O hydrogen bonding. From the slope and the intercept o f the plot between the intensity (AA c) versus the alcohol concentrations, K c is determined. For n-butanol, K c = 7.1 M - j, and for ethanol, K c = 1 9 . 0 M - I . At high alcohol concentrations (0.2 M) with various DPPC concentrations (0.005-0.04 M) the difference spectra showed a peak at about 3235 c m - t (n-butanol) and 3225 c m - ! (ethanol).
0.14
0.14
>I-t~ Z I,LI I-Z
>I.Z
AEALKcCACL 1 + KcCL
0.08
pZ
-0.02
-0.02
I
4000
3500 WAVENUMBER
A
0.08
3000
!
4000
3500
3000
WAVENUMBER
B
FIG. 4. Difference 1R spectra of alcohol-DPPC mixtures after subtracting the DPPC spectrum. A: n-butanol, and B: ethanol. (a) 0.006 M alcohol and 0.01 M DPPC. (b) 0.01 M DPPC control without alcohol. (c) The difference spectrum after subtracting the DPPC spectrum from that of the alcoholDPPC mixture. The peaks at 3275 c m - t (n-butanol) and 3265 cm- ~ (ethanol) represent the alcohol-DPPC complex.
PHOSPHOLIPID-ALCOHOL INTERACTION
147
0.14
0.14
>I--
>I'-
0.08
Z LLI I.Z
Z
0.08
I,Z
-0.02 4000
-0.02 3500
3000
4000
3500
WAVENUMBER
3000
WAVENUMBER
A
B
FIG. 5. Difference IR spectra of alcohoI-DPPC mixtures after subtracting the alcohol and DPPC spectra. A: n-butanol, and B: ethanol. (a) The difference spectrum of DPPC after subtracting the alcohol spectrum from the alcohol-DPPC mixture in Fig. 4. (b) 0.01 M DPPC. (c) The difference spectrum (al minus (b). The depression corresponding to the alcohol peak n-butanol 3635 cm- ~ and ethanol 3634 cm- ~) in Fig. 4 indicates the decrease in the unbound alcohol by forming alcohol-DPPC complex. These frequencies were lower than the alcohoI-DPPC association peak at 3275 c m - t (n-butanol) and 3265 c m - i (ethanol) of the low alcohol concentrations shown in Figs. 4-6. This is because only monomers exist at low alcohol concentrations to interact with DPPC, while multimers exist at high alcohol concentrations and participate in forming alcohol-DPPC complex. At a given DPPC concentration, the absorbance intensity of the association band was greater in the high alcohol concentrations than in the low alcohol concentrations. This finding indicates the existences of multimers interacting with DPPC. The binding constant of the alcohol OH group with DPPC, KoH, was derived in the same way as Equation 6 for the C a < C L case. KoHCACL CAL = l + KoHC L
(7)
and
z'kA°H =
Bhzding Site There are three possible hydrogen-bonding sites in phosphatidylcholines; the charged P = O - and the (CH3)3-N + groups in the polar head moiety and the C = O groups at the glycerol skeleton. Red-shift in the frequency is expected when these groups are hydrogen bonded with alcohol molecules. P = O - site. Figure 8 shows the region of the P = O stretching band. The spectrum at the highest frequency range (the left
0.021
>,. I,-
>,I.-,,
LLI
(8)
where ZL4OH is the change in the absorbance of O-H band relative to 3635 c m - ~ (n-butanol) or 3634 c m - ~ (ethanol) due to complex formation. From the slope and the intercept of the plot between the intensity (ZL4OH) versus alcohol concentration in Fig. 7, KOH was determined. For n-butanol, KOH = 7.1 M - ~, and for ethanol, KOH = 19.0 M - i .
0.021
z
A~KoHCACL I -I- KoHC L
z
0.016
u,I I'Z
I'Zm
I
-0.010
4000
3500 WAVENUMBER
A
3000
0.016
-0.010
I
4000
3500
3000
WAVENUMBER
B
FIG. 6. Difference IR spectra after subtracting the spectra of alcohols and DPPC from the alcohol-DPPC mixtures at various alcohol concentrations. A: n-butanol, and B: ethanol, The alcohol concentrations are from the top downward; 0.006, 0.005, and 0.003 M.
148
CHIOU, KUO, LIN, KAMAYA AND UEDA
0.06
0.06 I
>I.Z LLI I-Z
0.00
0.00
_=
-0.06
-0.06
4
0
BUTANOL CONCENTRATION (mM)
4 ETHANOL CONCENTRATION (raM)
A
B
FIG. 7. Plot of the intensity (AAc, AAoH and AAr) versus alcohol concentrations (CA). A: n-butanol, and B: ethanol. Open symbols; 0.0l M DPPC. Filled symbols; 0.02 M DPPC. Circles; Newly formed alcohol-DPPC complex at 3275 cm- z (n-butanol) and 3265 cm - J (ethanol), squares; alcohol O-H stretching at 3635 cm- ~ (n-butanol) and 3634 cm- ~ (ethanol), and triangles; P = O stretching at 1260 cm- ~. most line) is the P = O stretching band of 0.01 M DPPC in CCI.~ without the alcohols. When the alcohols were added, the band shifted toward lower frequencies. The effects of the alcohols are shown at two concentration ranges: low (0.005 and 0.006 M) and high (0.2 and 0.3 M). This red-shift (low-frequency shift) characterizes the formation of the P = O " • 'H-O bond. The P = O group of DPPC links to the O-H group of the alcohol. Even at very dilute n-butanol (Fig. 8A) or ethanol (Fig. 8B) concentrations of 0.005 and 0.006 M, we observed red-shift of the P = O asymmetric stretching band from 1260 cm - J of the control to 1257 c m - t. At high alcohol concentrations (0.2 and 0.3 M), the frequency peak was further shifted to 1245 c m - ~, and split into five peaks. When CA < CL, the binding constant, Kp, of the P = O group with the O-H of the alcohol is derived according to Equation 6.
CAL = KpCAC~L 1 + KpC L
(9)
and
~,Ap = AEALKpCACL
(lO)
I ÷ KpCL where AAp is the change in the absorbance intensity of the P = O band relative to 1260 c m - ~. From the slope and intercept of the plot between ZXApversus alcohol concentrations in Fig. 7, Kr, was estimated. For n-butanol, Kp = 7.1 M - ~, and for ethanol, Kp = 19.0 M - ~. The value matches the formation constant of the new peak representing the alcohol-DPPC complex. The result indicates that the alcohoI-DPPC complex is formed mainly at the P = O site. In the absence of alcohols, the P = O band of DPPC, which is centered around 1260 c m - ~ with shoulders on the low frequency side, agrees with the reported values (23,26). In the presence of dilute alcohol concentrations, this band shifted toward lower fre-
0.7
0.7
0.3
i
0.3
7:
0.1
1300
I
1250 WAVENUMBER
A
1200
0.1
I
1300
1250
,
1200
WAVENUMBER
B
FIG, 8. The alcohol effects on the P = O stretching in 0.01 M DPPC suspension. A: n-butanol, and B: ethanol. For the clarity of the figure, low (0.005 and 0.006 M) and high (0.2 and 0.3 M) concentrations of n-butanol (A) and ethanol (B) are shown. From the left, DPPC control, 0.005, 0.006, 0.2,
and 0.3 M alcohols. The spectra of 0.005 and 0.006 M appear almost overlapped. At high concentrations, the single peak split into five peaks.
PHOSPHOLIPID-ALCOHOL INTERACTION
149
1.0
1.0
0.5
0.5
>I-or) Z LU I,Z
0.0 1900
i 1775
J
0.0 1650
,
1900
WAVENUMBER
1775
1650
WAVENUMBER
A
B
FIG. 9. The alcohol effects on the C = O stretching band in 0.01 M DPPC suspension. A: n-butanol, and B: ethanol. The spectra are from the top downward: DPPC control, alcohol 0.006, 0.2. and 0.3 M. The 0.006 M spectrum overlapped with the control.
quency but maintained its original shape. When the alcohol concentration was increased, five peaks appeared between 1230 to 1260 c m - , in addition to the red-shift. All five peaks are spaced approximately 5 c m - ~ apart. Although the origin of these peaks are not known at present, they are the result of the interaction between the alcohol multimers and DPPC. Two possible sources of these peaks are considered: 1. Because the phosphoryl group is a strong proton acceptor, it can form very strong hydrogen bonds with O-H groups. The IR band of this group frequently occurs as a doublet due to rotational isomerism or due to the coupling effects with C-H stretching modes (19,20). The O-H bending mode, on the other hand, is sensitive to coupling effects. It was shown (18) that the O-H bending band was displaced to 1244 c m - ~ by ethanol due to coupling with a torsional mode of the C-H2 group. Any interactions between P = O doublet and this 1244 c m - t mode could produce multiple peaks. 2. When the bond strength in the alcohol-DPPC complexes is
increased, the O-H stretching band splits into two peaks. With further increases in bond strength, a new band .begins to appear near 1900 c m - ' (7). It is not unfeasible that these three bands produce separate effect on the P = O band to form the multiple bands. The detailed mechanism, however, remains to be elucidated. C = 0 Site. Figure 9 shows the stretching vibrational region of the C = O group. At the alcohol concentrations of 0.005 and 0.006 M, the spectra completely overlapped with the 0.01 M DPPC control. When the alcohol concentrations were increased to 0.2 and 0,3 M, small changes in the band intensity were observed. A small shift to the high frequency region may be caused by the release of a part of the water molecules that are strongly bound to the P = O moiety of DPPC (26). Future studies are needed to clarify the mechanism. (CH3)3-N ÷ Site. Figure 10 shows the effect of the alcohols on the (CH3)3-N* band in 0.01 M DPPC. There is essentially no change in the (CH3)3-N ÷ band upon addition of the alcohols up
0.35
0.35
>I'Z
>I-"
0.15
Z uJ
I,Z
0.15
Z
0.05 1000
950 WAVENUMBER
A
900
0.05 1000
950
900
WAVENUMBER
B
FIG. 10. The alcohol effects on the (CH3)3-N+ stretching band in 0.01 M DPPC. The control DPPC spectrum is overlapped with the difference spectra of alcohol-DPPC mixture. The alcohol concentrations are 0.006.0.2 and 0.3 M.
150
CHIOU, KUO, LIN, K A M A Y A AND U E D A
to 0.3 M to the solution o f 0.01 M DPPC. The (CH3)3-N + group o f DPPC cannot form hydrogen-bonds because there is no donor proton in the (CH3)3-N ÷ group to form hydrogen bonds with the oxygen atom in alcohols. In addition, the positive charge o f the (CH3)3-N ÷ group prevents its nitrogen atom to interact with the hydrogen atom o f alcohols. The interaction between alcohols and the choline moiety in DPPC is a charge-dipole interaction and is much weaker than hydrogenbonding. CONCLUSION In a hydrophobic environment, alcohols self-associate to form dimers when the concentration exceeds 0.008 M. Above 0.03 M, alcohols exist as a mixture o f monomer, dimer, trimer and tetramer. The binding constant between the alcohols and the P = O group in DPPC was identical to the formation constant of the newly formed complex observed at 3275 c m - i (n-butanol) and 3265 c m - ~ (ethanol)..This finding indicates that the alcohol-binding site is the P = O group in DPPC, and is consistent with the previ-
ous studies in which the phosphate group was found to be the main functional group in DPPC, affected by anesthetics (3,13). The interaction between alcohols and DPPC is confined to the hydrophilic phosphate group o f DPPC and the alcohol. There is no interaction between alcohols and hydrocarbon tails o f the lipid. Membranes are supported by the hydrogen-bonded water matrix. When the supporting force is weakened, the whole membrane becomes disordered. Yet the motion o f the interfacial region becomes slower due to the binding of alcohol molecules. The alcohol-induced decrease in the " v i s c o s i t y " o f the membrane core is a consequence of the change in the state o f the membrane due to the loss of this support. The decrease in the microviscosity o f lipid membranes does not necessarily mean that alcohols directly interact with the hydrophobic core. The present data agree with the notion that anesthesia is induced by dehydration of the watermacromolecule interface (2, 11-15. 23-25, 27). ACKNOWLEDGEMENTS This study was supported by the VA Medical Research. and NIH grants GM25716 and GM27670.
REFERENCES 1. 'Brasseur, R.; Chatelin, P.; Goormaghtigh, E. Ruysschaert, J. M. A semi-empirical conformation analysis of the interaction of n-alkanols with dipalmitoylphosphatidylcholine. Biochim. Biophys. Acta 814: 227-236; 1985. 2. Chiou, J. S.; Ma, S. M.; Kamaya, H.; Ueda, I. Anesthesia cutoff phenomenon: lnterfacial hydrogen bonding. Science 248:583-585; 1990. 3. Colley, C. M.; Metcalfe, J. C. The localization of small molecules in lipid bilayer. FEBS Lett. 24:241-246; 1972. 4. Di Paolo, T.; Sandorfy. C. Hydrogen bond breaking potency of fluorocarbon anesthetics. J. Med. Chem. 17:809-814; 1974. 5. Eyring, H.; Woodbury. J. W.; D'Arrigo, J. S. A molecular mechanism of general anesthesia. Anesthesiology 38:414--425; 1973. 6. Ferguson, J. The use of chemical potentials as indices of toxicity. Proc. R. Soc. Lond. [Biol.] 127:387---404; 1939. 7. Hadzi, D. Infrared spectra of strongly hydrogen-bonded systems. Pure Appl. Chem. 11:435--453; 1965. 8. Hitzemann, R. J. Effects of non-electrolyte molecules with anesthetic activity on the physical properties of DMPC multilammelar liposomes. B iochim. Biophys. Acta 983:205-211; 1989. 9. Hobza, P.; Mulder, F.; Sandorfy, C. Quantum chemical and statistical thermodynamic investigations of anesthetic activity. 1. The interaction between chloroform, fluoroform, cyclopropane, and an O-H'"O hydrogen bond. J. Am. Chem. Soc. 103:1360-1366; 1981. 10. Hobza, P.; Mulder, F.; Sandorfy, C. Quantum chemical and statistical thermodynamic investigations of anesthetic activity. 2. The interaction between chloroform, fluoroform, and a N-H..-O = C hydrogen bond. J. Am. Chem. Soc. 104:925-928; 1982. 11. Kamaya, H.; Kaneshina, S.; Ueda, I. Partition equilibrium of inhalation anesthetics and alcohols between water and membranes of phospholipids with varying acyl chain-lengths. Biochim. Biophys. Acta 646:135-142; 1981. 12. Kamaya, H.; Ueda, I.; Eyring, H. General anesthesia and interfacial water. In: Fink, B. R., ed. Molecular mechanisms of anesthesia, vol. 2. Progress in anesthesiology. New York: Raven Press; 1980:429433. 13. Kaneshina, S.; Kamaya, H.; Ueda, I. Transfer of anesthetics and alcohols into ionic surfactant micelles in relation to depression of Krafft point and critical micelle concentration, and interfacial action of anesthetics. J. Colloid Interface Sci. 83:589-598; 1982. 14. Kaneshina, S.; Kamaya, H.; Ueda, I. Benzyl alcohol penetration into micelles, dielectric constant of the binding site, partition coefficient
15. 16. 17. 18.
19. 20.
21. 22.
23.
24. 25.
26. 27.
and high-pressure squeeze-out. Biochim. Biophys. Acta 777:75-83. 1984. Klemm, W. R. Dehydration: A new alcohol theory. Alcohol 7:4959; 1990. Mullins, L. J. Some physical mechanism in narcosis. Chem. Rev. 54:289-323; 1954. Orme, F. W.: Mononne, M. M.; Macey, R. I. Modification of the erythrocyte membrane dielectric constant by alcohols. J. Membr. Biol. 104:57--68; 1988. Perchard, J. P.; Josien. M. L. Etude des spectres vibrationnels de douze especes isotopiques d'ethanols monomers. J. Chim. Phys. 65: 1835-1875; 1968. Raevsky, O. A. Conformational analysis of organophosphorus compounds by means of IR spectroscopy and dipole moment methods. J. Mol. Struct. 19:275-287: 1973. Raevsky, O. A.; Vereshchagin, A. N.; Mumzhriva, N. G.; Zyablikova. T. A.; Alexandrova, N. A.; Arbuzov. A. E. A conformational study of c~-chloroethylphosphono-dichloridate by IR. NMR, dipole moments and Kerr constants. J. Mol. Struct. 36:299-308; 1977. Rowe, E. S.; Fernandes. A.; Khalifah, R. G. Alcohol interactions with lipids: a carbon-13 nuclear magnetic resonance study using butanol labeled at C-I. Biochim. Biophys. Acta 905:151-161; 1987. Trudeau, T.; Dumas, J. M.: Dupuis, P.; Guerin, M.; Sandorfy, C. Intermolecular interactions and anesthesia: Infrared spectroscopic studies. Top. Curr. Chem. 93:91-125; 1980. Tsai, Y. S.; Ma. S. M.; Kamaya, H.; Ueda, I. Fourier transform infrared studies on phospholipid hydration: Phosphate-oriented hydrogen bonding and its attenuation by volatile anesthetics. Mol. Pharmacol. 31:623-630; 1987. Ueda, I4 Kamaya. H. Molecular mechanisms of anesthesia. Anesthesiology 63:929-945; 1984. Ueda. 1.; Tseng, H. S.; Kaminoh, Y.; Ma, S. M.; Kamaya, H.; Lin, S. H. Anesthetics release unfreezable and bound water in partially hydrated phospholipid lamellar systems and elevate phase transition temperature. Mol. Pharmacol. 29:582-588; 1986. Wong, P. T. T.; Mantsch, H. H. High-pressure infrared spectroscopic evidence of water binding sites in 1,2-diacyl phospholipids. Chem. Phys. Lipids 46:213-224; 1988. Yokono, S.; Shieh, D. D.; Ueda, I. Interfacial preference of anesthetic action upon the phase transition of phospholipid bilayers and partition equilibrium of inhalation anesthetics between membrane and deuterium oxide. Biochim. Biophys. Acta 645:237-242; 1981.
t1741-83:19/91 S3.00 + .00
Interfacial Dehydration by Alcohols: Hydrogen Bonding of Alcohols to Phospholipids J A N G - S H I N G C H I O U , * C H I H - C H U KUO,i" S H E N G H. L I N , ~ H I R O S H I K A M A Y A * A N D I S S A K U U E D A *l
*Department of Anesthesia, UniversiO' of Utah School of Medicine and Anesthesia Service, VA Medical Center, Salt Lake City, UT 84148 and fDepartment of Chemisto', Arizona State Universio,, Tempe, AZ 85287 R e c e i v e d 1 1 July 1990: A c c e p t e d 27 S e p t e m b e r 1990
CHIOU. J.-S.. C.-C. KUO, S. H. LIN. H. KAMAYA AND I. UEDA. Interfacialdehydration by alcohols: Hydrogen bonding of alcohols to phospholipids. ALCOHOL 8(2) 143-150, 1991.--The interaction between alcohols ¢ethanol and n-butanol) and dipalmitoylphosphatidylcholine (DPPC) in carbon tetrachloride was studied by Fourier transform infrared spectroscopy (FTIR). Upon addition of the alcohols, the P = O stretching band of DPPC at 1260 cm - ~ shifted to lower frequency (red-shift). The red-shift indicates that the P= O vibration became slower possibly because the heavier alcohol molecules replaced the water molecules hydrogen bonded to the PO, moiety. The formation constants between the PO_, group and ethanol or n-butanol (n-butanol data in parenthesis) were 19.0 M-~ (7. I M-~) when estimated from the spectral change. A new absorbance peak appeared at 3265 c m - t (3275 cm-~) representing the DPPC-alcohol complex. The formation constant of this complex was also 19.0 M-~ (7.1 M-~I. The identical formation constant suggests that the DPPC-alcohol complex was formed at the PO., moiety of DPPC with hydrogen bonding to the alcohol OH. At higher alcohol concentrations, the absorbance peak of DPPC-alcohol complex shifted to 3225 cm ~ 13235 cm ~). Apparently, the lower frequency shift at higher alcohol concentration occurred by the formation of alcohol multimers (dimer, trimer, and tetrameri interacting with DPPC. Alcohol-phospholipid interaction
Interface
Hydrogen-bonding
Dehydration
Anesthesia theory
ers the problem of distribution of the drug in the body, it is easy to understand that these expressions do not represent the true potency because the amount of ethanol bound at the action site is not identical with the blood concentration. As early as 1939, Ferguson (6) argued that the true potency must be compared by the amount of drugs bound at the action site rather than the bulk concentration. When expressed by the chemical potential, however, these two values become identical at an equilibrium condition. He proposed to express the potency by "thermodynamic" activity. For volatile compounds, the thermodynamic activity is expressed by the fractional saturation in the gas phase. This is the ratio between the partial pressure of the vapor and the saturated vapor pressure of the pure liquid. This is analogous to the relative humidity that is used to express the dryness in meteorology. One feels mugginess by the relative humidity and not by the actual water content in the air. When expressed by the fractional vapor pressure, all volatile compounds induce anesthesia at about 3% saturation (6,16). Thus, all anesthetics (including alcohols) are equipotent when expressed by the thermodynamic activity (6,16). With ethanol, anesthesia ensues when the end-expiratory ' partial pressure of ethanol reaches 35 tort. Because the activity
THIS study deals with anesthetic effects of alcohols. The general consensus, however, appears to account alcohols as a very weak anesthetic. One of the referees of this article commented that alcohols are not an anesthetic. On the contrary, clinician anesthesiologists know that ethanol is a potent anesthetic when given intravenously. The patients are rapidly anesthetized without the excitement stage. It is wrong to assume that only a handful of compounds have anesthetic potency. Most hydrophobic molecules that perturb lipid membranes are anesthetics, including gases, hydrocarbons, alcohols, ethers, ketones, other organic solvents, and their halogenated counterparts (24). The anesthetic action of xenon is well known. Nitrogen gas is a potent anesthetic at 30 atmospheric pressure. Carbon dioxide is an anesthetic at 0.3 atmospheric pressure. These gases are not clinically applicable because of the high cost of xenon, high pressure required for nitrogen, and pH disturbances associated with carbon dioxide administration (24). The " w e a k " anesthetic potency of alcohols )s another misconception. Potency is evaluated by the amount of drugs required to obtain a desired effect. With ethanol, the potency is often expressed by the blood concentration. However, when one consid-
(Requests for reprints should be addressed to Dr. Issaku Ueda, Anesthesia Service I12A, VA Medical Center, Salt Lake City, LIT 84148.
143
144
CHIOU, KUO, LIN, KAMAYA AND UEDA
coefficient of ethanol in blood is not identical with that in water, the alcohol concentrations in the model system cannot be meaningfully compared with the blood content that induces intoxication. Another problem is the lethal dose (LD50) of alcohols and narcotics. This is because the cause of death is not standardized. Death by ethanol or narcotic overdose occurs usually by the respiratory depression. When the respiration is mechanically supported, death does not occur at the lethal dose. This is evident in clinical anesthesia, where respiration is stopped at the beginning of anesthesia by muscle relaxants and is mechanically controlled. In this situation, the patients tolerate the lethal doses. For instance, the clinical dose of morphine is in the range of 8 to 12 mg. For cardiac operations, however, anesthesiologists use 300 mg and more, which grossly exceeds the lethal dose. The clinical concentrations of all anesthetics are close to lethal doses if the respiration were not supported. Thus, the actions of short-chain alcohols and inhalation anesthetics are similar. They interact with lipid bilayers and decrease the temperature of the main transition between the solid-gel and liquid-crystal phases. The chain-melting or disordering effects correlate extremely well to the anesthetic potency (24). Because membrane order is a property represented by the conformation of lipid tails of phospholipid molecules, the alcohol actions are often,attributed to their effect on the hydrophobic core of the membrane (8,17). Nevertheless, alcohols are dipolar molecules with hydrophilic OH moiety and are expected to reside at the watermembrane interface (1-3, 11. 13, 14). Eyring et al. (5) proposed that anesthetics destruct the clusters of water molecules electrostricted at the charged surface of macromolecules. Di Paolo and Sandorfy (4) have shown that inhalation anesthetics hinder the formation of hydrogen bonds, and the anesthetic potency correlates to the hydrogen-bond-breaking activity. By quantum chemical analyses, Sandorfy and co-workers (9,10) showed that anesthetics can form competitive proton donor-acceptor complex so that H,O-X dimer can be replaced with anesthetic-X dimer. They propose that the "breaking" and "formation" of hydrogen bonds may be central to the mechanism of anesthesia (22). Because alcohols contain hydrophilic OH moiety, they are expected to interact with the hydrophilic moiety of DPPC and break the hydrogen bonds between the membrane surface and water molecules by the competitive mechanism. Klemm (15) also proposed that the alcohol action is caused by the interfacial dehydration based on his biochemical finding on the ethanol-induced depletion of brain sialic acid. We (2, 11-14, 23-25, 27) suggested that anesthesia is an interfacial phenomenon, replacing the surface-bound water with anesthetics. Tsai et al. (23) measured the effect of inhalation anesthetics on water-DMPC (dimyristoylphosphatidylcholine) interaction in a water-in-oil reversed micellar system using FTIR spectroscopy. They reported that the primary hydration site of DMPC was the phosphate moiety, and up to 18 water molecules were restricted at the polar head group. The anesthetics replaced part of the water molecules hydrogen-bonded to the head group, but did not affect the choline signal. The present study measured the interaction between alcohols (ethanol and n-butanol) and DPPC in carbon tetrachloride solution. Hydrogen-bonded complexes between alcohols and DPPC were identified and the formation constants of such complexes were estimated. From the changes in the frequency and shape of IR bands, we determined the binding site of alcohols on the lipid molecule. METHOD
Synthetic dipalmitoylphosphatidylcholine ( 1,2,-dihexadecanoyl-
sn-glycero-3-phosphorylcholine, DPPC) and n-butanol were obtained from Sigma, ethanol from U.S. Industrial Chemicals (Anaheim, CA), and spectroscopic grade carbon tetrachloride from EM Science (Cherry Hill, NJ). DPPC was kept in a desiccator at reduced pressure until use. DPPC was dispersed in CCI 4 by irradiation in a cup-horn of a Branson ultrasonic disrupter (Danbury, CT) at 0.01 or 0.02 M. After the baseline spectra were obtained, alcohols were added to the DPPC-CCI4 mixture with a microsyringe under nitrogen gas, and sealed. Each sample was mixed by a vortex mixer. A Perkin-Elmer (Norwalk, CT) model 1750 FTIR spectrophotometer interfaced with a Perkin-Elmer model 7300 computer was used for the analysis. The cell window (FT 04-035) was a fixed thickness sodium chloride with 1.28 mm pathlength (SpectraTech, Stamford, CT). The cell was calibrated against the benzene band at 845 c m - J. A TGS detector was used for all experiments. Each sample was scanned over the frequency range 400-4000 c m - ] and the spectra were accumulated for 20 scans. The spectra of the solvents were subtracted from the spectra of the solutions. For DPPC-alcohol mixture, the difference spectrum was obtained after subtraction of the component spectrum from the spectrum of the sample mixture. All spectra were acquired at 22°C. RESULTS AND DISCUSSION
Alcohols in CCI~ Figures 1-3 show the IR spectra of n-butanol (A) and ethanol (B) in carbon tetrachloride. When the concentrations of the alcohols were below 0.006 M, only the monomer spectrum was observed (Fig. 1). Dimers began to appear when the alcohol concentration exceeded 0.008 M (Fig. 2). Trimers and tetramers appeared when the concentration exceeded 0.03 M. With the alcohol concentrations above 0.1 M (Fig. 3), all four species were present. According to the increase in the alcohol concentration. the absorbance intensity of monomers decreased and those of dimers to tetramers increased. The IR band (Fig. IA) at 3635 c m - ~ represents the nonbonded free O-H stretching vibration of n-butanol. For ethanol it was 3634 c m - t (Fig. 1B). The shoulder at 3515 c m - ~ is that of the self-associated dimers H-O. • .H-O for both alcohols (Fig. 2). The vibrational bands of trimers are at about 3443 c m - ~ (n-butanol) and 3450 c m - ~ (ethanol) (Fig. 3). The tetramer bands appeared at 3345 cm i (n-butanol) and 3343 c m - ~ (ethanol) (Fig. 3). Two ranges of the alcohol concentrations were studied. The low concentration range covers 0.001 to 0.006 M where only n-butanol monomers are present. The high concentration range covers 0.2 to 0.3 M, where monomers, dimers, trimers and tetramers coexist. AlcohoI-DPPC Mixture Figures 4--6 show the spectra of the alcohoI-DPPC mixture. In Fig. 4, spectrum (a) is 0.01 M DPPC with 0.006 M n-butanol (Fig. 4A) and ethanol (Fig. 4B). The peaks at 3635 c m - ~ (n-butanol, Fig. 4A) and 3634 c m - ] (ethanol, Fig. 4B) are the nonbonded free O-H stretching vibration of the alcohols. Spectrum (b) is the 0.01 M DPPC control in the absence of the alcohols. The broad peak at 3373 c m - ~ is assigned to the O-H stretching band of the water molecules adhering to the hydrophilic parts of DPPC. These water molecules are tightly bound to DPPC despite storage in a desiccator at reduced pressure. The difference spectra of the alcohols (c) were obtained by subtracting the DPPC spectrum (b)
PHOSPHOLIPID-ALCOHOL INTERACTION
145
0.040
0.040
>I,(/) Z ILl I-Z
0.025
0.025
__.__J -0.010
-0.010
I
4000
3550
__5
3100
!
4000
WAVENUMBER
3550
3100
WAVENUMBER
A
B
FIG. 1. IR spectra of n-butanol (A) and ethanol (B) in carbon te~achlofide. The alcohol concentrations are from the top downward; 0.006.0.005.0.004. 0.003, and 0.002 M.
from (a). Spectrum (c) shows DPPC-alcohol interaction. A peak, representing the alcohol-DPPC complex, appeared at about 3275 c m - ~ (n-butanol, Fig. 4A) and 3265 c m - ~ (ethanol, Fig. 4B) in the difference spectrum. At the same time, the intensity of the O-H stretching bands of free alcohols at 3635 c m - ~ (n-butanol) and 3634 c m - ! (ethanol) decreased. Figure 5 shows the difference spectrum of DPPC when interacted with the alcohols. Spectrum (a) was obtained by subtracting the free alcohol 0.006 M spectra from the spectra of the alcohol-DPPC mixture. Spectrum (b) is the 0.01 M DPPC control. By subtracting (b) from (a), the interaction between DPPC and the alcohols is visualized (c). The negative peaks at 3635 c m - ~ (n-butanol, Fig. 5A) and 3634 c m - ! (ethanol, Fig. 5B) signify that the free alcohol concentration is decreased by binding to DPPC. Figure 6 shows the difference spectra at various alcohol concentrations (0.003 M to 0.006 M) and at 0.01 M DPPC. When the alcohol concentration was increased from 0.003 M to 0.006 M, the intensity of the free O-H stretching bands at 3635 c m - ! (free n-butanol, Fig. 6A) and 3634 c m - ~ (free ethanol, Fig. 6B)
decreased, whereas the intensity of 3275 c m - ~ band (n-butanolDPPC complex, Fig. 6A) and 3265 c m - ! band (ethanol-DPPC complex, Fig. 6B) increased. To calculate the association constants of the DPPC-alcohol complexes, we used dilute solutions of the alcohols (0.002--0,006 M) to limit the interaction between the alcohol monomers and the functional groups of DPPC. Let A be the alcohol and L be DPPC, then at low alcohol concentrations, we have Kc [A] + [L] "-s [ALl
CAL KC = CA CL
Z LU I--
0.08
z
-0.02
(2)
where K c is the formation constant of the alcohol-lipid complex, and CAt., CA, and C , are the concentrations of the complex, alcohol, and lipid, respectively. The observed absorbance per unit cell length for this system, A, is expressed as
0.14
>l--
( 1)
and
0.14
t _3
>I--
zLU
-0.02
I
4000
3550 WAVENUMBER
A
SL__
0.08
I-Z
310q
!
3550
4000
WAVENUMBER
B
FIG. 2. Same as Fig. I. A: n-butanol, and B: ethanol. The alcohol concentrations are from the top downward; 0.015, 0.010, and 0.008 M.
3100
146
CHIOU, K U O , LIN, K A M A Y A AND U E D A
0.70
0.70
>,I,-
>I-0')
Z
0.40
Z III pZ
tl.I I-Z
0.40
_J
J
-0.10
3550
!
-0.10
I
4000
4000
31 O0
3550
3100
WAVENUMBER
WAVENUMBER
A
B
FIG. 3. Same as Fig. 1. A: n-butanol, and B: ethanol. The alcohol concentrations are from the top downward; 0.15.0.13, 0.10 M.
A = E A C A + e L C L "~ E A L C A L
(3) AAc-
where EAL, EA, and e L are the respective absorption coefficients o f the complex, alcohol and lipid. By designating AAc as the change in the absorbance o f the association band relative to those o f pure A and L in CC14, Equation 3 is written as ~u~ c
=
A -- E A C A -- e L C L
=
C A L (CA L -- eA -- (EL)
=
A EALCAL
(4)
When C A < C L, we have. approximately,
CAL
KcCACL
(5)
1 + KcC L and
(6)
Equation 6 shows that when C L is kept constant, AA c versus CA is linear (Fig. 7). In Fig. 7, we have plotted the intensities o f O-H and P = O bands against the alcohol concentration at 0.01 M and 0.02 M DPPC. Accoraing to the increase in the alcohol concentration, the intensity o f the DPPC-alcohol association band at 3275 c m - ~ (n-butanol, Fig. 7A) and 3265 c m - ~ (ethanol, Fig. 7B) increased :.'hile the intensities of the O-H stretching band at 3635 c m - ! (n-butanol) and 3634 c m - t (ethanol) decreased. The association band arises from O-H. - . P = O hydrogen bonding. From the slope and the intercept o f the plot between the intensity (AA c) versus the alcohol concentrations, K c is determined. For n-butanol, K c = 7.1 M - j, and for ethanol, K c = 1 9 . 0 M - I . At high alcohol concentrations (0.2 M) with various DPPC concentrations (0.005-0.04 M) the difference spectra showed a peak at about 3235 c m - t (n-butanol) and 3225 c m - ! (ethanol).
0.14
0.14
>I-t~ Z I,LI I-Z
>I.Z
AEALKcCACL 1 + KcCL
0.08
pZ
-0.02
-0.02
I
4000
3500 WAVENUMBER
A
0.08
3000
!
4000
3500
3000
WAVENUMBER
B
FIG. 4. Difference 1R spectra of alcohol-DPPC mixtures after subtracting the DPPC spectrum. A: n-butanol, and B: ethanol. (a) 0.006 M alcohol and 0.01 M DPPC. (b) 0.01 M DPPC control without alcohol. (c) The difference spectrum after subtracting the DPPC spectrum from that of the alcoholDPPC mixture. The peaks at 3275 c m - t (n-butanol) and 3265 cm- ~ (ethanol) represent the alcohol-DPPC complex.
PHOSPHOLIPID-ALCOHOL INTERACTION
147
0.14
0.14
>I--
>I'-
0.08
Z LLI I.Z
Z
0.08
I,Z
-0.02 4000
-0.02 3500
3000
4000
3500
WAVENUMBER
3000
WAVENUMBER
A
B
FIG. 5. Difference IR spectra of alcohoI-DPPC mixtures after subtracting the alcohol and DPPC spectra. A: n-butanol, and B: ethanol. (a) The difference spectrum of DPPC after subtracting the alcohol spectrum from the alcohol-DPPC mixture in Fig. 4. (b) 0.01 M DPPC. (c) The difference spectrum (al minus (b). The depression corresponding to the alcohol peak n-butanol 3635 cm- ~ and ethanol 3634 cm- ~) in Fig. 4 indicates the decrease in the unbound alcohol by forming alcohol-DPPC complex. These frequencies were lower than the alcohoI-DPPC association peak at 3275 c m - t (n-butanol) and 3265 c m - i (ethanol) of the low alcohol concentrations shown in Figs. 4-6. This is because only monomers exist at low alcohol concentrations to interact with DPPC, while multimers exist at high alcohol concentrations and participate in forming alcohol-DPPC complex. At a given DPPC concentration, the absorbance intensity of the association band was greater in the high alcohol concentrations than in the low alcohol concentrations. This finding indicates the existences of multimers interacting with DPPC. The binding constant of the alcohol OH group with DPPC, KoH, was derived in the same way as Equation 6 for the C a < C L case. KoHCACL CAL = l + KoHC L
(7)
and
z'kA°H =
Bhzding Site There are three possible hydrogen-bonding sites in phosphatidylcholines; the charged P = O - and the (CH3)3-N + groups in the polar head moiety and the C = O groups at the glycerol skeleton. Red-shift in the frequency is expected when these groups are hydrogen bonded with alcohol molecules. P = O - site. Figure 8 shows the region of the P = O stretching band. The spectrum at the highest frequency range (the left
0.021
>,. I,-
>,I.-,,
LLI
(8)
where ZL4OH is the change in the absorbance of O-H band relative to 3635 c m - ~ (n-butanol) or 3634 c m - ~ (ethanol) due to complex formation. From the slope and the intercept of the plot between the intensity (ZL4OH) versus alcohol concentration in Fig. 7, KOH was determined. For n-butanol, KOH = 7.1 M - ~, and for ethanol, KOH = 19.0 M - i .
0.021
z
A~KoHCACL I -I- KoHC L
z
0.016
u,I I'Z
I'Zm
I
-0.010
4000
3500 WAVENUMBER
A
3000
0.016
-0.010
I
4000
3500
3000
WAVENUMBER
B
FIG. 6. Difference IR spectra after subtracting the spectra of alcohols and DPPC from the alcohol-DPPC mixtures at various alcohol concentrations. A: n-butanol, and B: ethanol, The alcohol concentrations are from the top downward; 0.006, 0.005, and 0.003 M.
148
CHIOU, KUO, LIN, KAMAYA AND UEDA
0.06
0.06 I
>I.Z LLI I-Z
0.00
0.00
_=
-0.06
-0.06
4
0
BUTANOL CONCENTRATION (mM)
4 ETHANOL CONCENTRATION (raM)
A
B
FIG. 7. Plot of the intensity (AAc, AAoH and AAr) versus alcohol concentrations (CA). A: n-butanol, and B: ethanol. Open symbols; 0.0l M DPPC. Filled symbols; 0.02 M DPPC. Circles; Newly formed alcohol-DPPC complex at 3275 cm- z (n-butanol) and 3265 cm - J (ethanol), squares; alcohol O-H stretching at 3635 cm- ~ (n-butanol) and 3634 cm- ~ (ethanol), and triangles; P = O stretching at 1260 cm- ~. most line) is the P = O stretching band of 0.01 M DPPC in CCI.~ without the alcohols. When the alcohols were added, the band shifted toward lower frequencies. The effects of the alcohols are shown at two concentration ranges: low (0.005 and 0.006 M) and high (0.2 and 0.3 M). This red-shift (low-frequency shift) characterizes the formation of the P = O " • 'H-O bond. The P = O group of DPPC links to the O-H group of the alcohol. Even at very dilute n-butanol (Fig. 8A) or ethanol (Fig. 8B) concentrations of 0.005 and 0.006 M, we observed red-shift of the P = O asymmetric stretching band from 1260 cm - J of the control to 1257 c m - t. At high alcohol concentrations (0.2 and 0.3 M), the frequency peak was further shifted to 1245 c m - ~, and split into five peaks. When CA < CL, the binding constant, Kp, of the P = O group with the O-H of the alcohol is derived according to Equation 6.
CAL = KpCAC~L 1 + KpC L
(9)
and
~,Ap = AEALKpCACL
(lO)
I ÷ KpCL where AAp is the change in the absorbance intensity of the P = O band relative to 1260 c m - ~. From the slope and intercept of the plot between ZXApversus alcohol concentrations in Fig. 7, Kr, was estimated. For n-butanol, Kp = 7.1 M - ~, and for ethanol, Kp = 19.0 M - ~. The value matches the formation constant of the new peak representing the alcohol-DPPC complex. The result indicates that the alcohoI-DPPC complex is formed mainly at the P = O site. In the absence of alcohols, the P = O band of DPPC, which is centered around 1260 c m - ~ with shoulders on the low frequency side, agrees with the reported values (23,26). In the presence of dilute alcohol concentrations, this band shifted toward lower fre-
0.7
0.7
0.3
i
0.3
7:
0.1
1300
I
1250 WAVENUMBER
A
1200
0.1
I
1300
1250
,
1200
WAVENUMBER
B
FIG, 8. The alcohol effects on the P = O stretching in 0.01 M DPPC suspension. A: n-butanol, and B: ethanol. For the clarity of the figure, low (0.005 and 0.006 M) and high (0.2 and 0.3 M) concentrations of n-butanol (A) and ethanol (B) are shown. From the left, DPPC control, 0.005, 0.006, 0.2,
and 0.3 M alcohols. The spectra of 0.005 and 0.006 M appear almost overlapped. At high concentrations, the single peak split into five peaks.
PHOSPHOLIPID-ALCOHOL INTERACTION
149
1.0
1.0
0.5
0.5
>I-or) Z LU I,Z
0.0 1900
i 1775
J
0.0 1650
,
1900
WAVENUMBER
1775
1650
WAVENUMBER
A
B
FIG. 9. The alcohol effects on the C = O stretching band in 0.01 M DPPC suspension. A: n-butanol, and B: ethanol. The spectra are from the top downward: DPPC control, alcohol 0.006, 0.2. and 0.3 M. The 0.006 M spectrum overlapped with the control.
quency but maintained its original shape. When the alcohol concentration was increased, five peaks appeared between 1230 to 1260 c m - , in addition to the red-shift. All five peaks are spaced approximately 5 c m - ~ apart. Although the origin of these peaks are not known at present, they are the result of the interaction between the alcohol multimers and DPPC. Two possible sources of these peaks are considered: 1. Because the phosphoryl group is a strong proton acceptor, it can form very strong hydrogen bonds with O-H groups. The IR band of this group frequently occurs as a doublet due to rotational isomerism or due to the coupling effects with C-H stretching modes (19,20). The O-H bending mode, on the other hand, is sensitive to coupling effects. It was shown (18) that the O-H bending band was displaced to 1244 c m - ~ by ethanol due to coupling with a torsional mode of the C-H2 group. Any interactions between P = O doublet and this 1244 c m - t mode could produce multiple peaks. 2. When the bond strength in the alcohol-DPPC complexes is
increased, the O-H stretching band splits into two peaks. With further increases in bond strength, a new band .begins to appear near 1900 c m - ' (7). It is not unfeasible that these three bands produce separate effect on the P = O band to form the multiple bands. The detailed mechanism, however, remains to be elucidated. C = 0 Site. Figure 9 shows the stretching vibrational region of the C = O group. At the alcohol concentrations of 0.005 and 0.006 M, the spectra completely overlapped with the 0.01 M DPPC control. When the alcohol concentrations were increased to 0.2 and 0,3 M, small changes in the band intensity were observed. A small shift to the high frequency region may be caused by the release of a part of the water molecules that are strongly bound to the P = O moiety of DPPC (26). Future studies are needed to clarify the mechanism. (CH3)3-N ÷ Site. Figure 10 shows the effect of the alcohols on the (CH3)3-N* band in 0.01 M DPPC. There is essentially no change in the (CH3)3-N ÷ band upon addition of the alcohols up
0.35
0.35
>I'Z
>I-"
0.15
Z uJ
I,Z
0.15
Z
0.05 1000
950 WAVENUMBER
A
900
0.05 1000
950
900
WAVENUMBER
B
FIG. 10. The alcohol effects on the (CH3)3-N+ stretching band in 0.01 M DPPC. The control DPPC spectrum is overlapped with the difference spectra of alcohol-DPPC mixture. The alcohol concentrations are 0.006.0.2 and 0.3 M.
150
CHIOU, KUO, LIN, K A M A Y A AND U E D A
to 0.3 M to the solution o f 0.01 M DPPC. The (CH3)3-N + group o f DPPC cannot form hydrogen-bonds because there is no donor proton in the (CH3)3-N ÷ group to form hydrogen bonds with the oxygen atom in alcohols. In addition, the positive charge o f the (CH3)3-N ÷ group prevents its nitrogen atom to interact with the hydrogen atom o f alcohols. The interaction between alcohols and the choline moiety in DPPC is a charge-dipole interaction and is much weaker than hydrogenbonding. CONCLUSION In a hydrophobic environment, alcohols self-associate to form dimers when the concentration exceeds 0.008 M. Above 0.03 M, alcohols exist as a mixture o f monomer, dimer, trimer and tetramer. The binding constant between the alcohols and the P = O group in DPPC was identical to the formation constant of the newly formed complex observed at 3275 c m - i (n-butanol) and 3265 c m - ~ (ethanol)..This finding indicates that the alcohol-binding site is the P = O group in DPPC, and is consistent with the previ-
ous studies in which the phosphate group was found to be the main functional group in DPPC, affected by anesthetics (3,13). The interaction between alcohols and DPPC is confined to the hydrophilic phosphate group o f DPPC and the alcohol. There is no interaction between alcohols and hydrocarbon tails o f the lipid. Membranes are supported by the hydrogen-bonded water matrix. When the supporting force is weakened, the whole membrane becomes disordered. Yet the motion o f the interfacial region becomes slower due to the binding of alcohol molecules. The alcohol-induced decrease in the " v i s c o s i t y " o f the membrane core is a consequence of the change in the state o f the membrane due to the loss of this support. The decrease in the microviscosity o f lipid membranes does not necessarily mean that alcohols directly interact with the hydrophobic core. The present data agree with the notion that anesthesia is induced by dehydration of the watermacromolecule interface (2, 11-15. 23-25, 27). ACKNOWLEDGEMENTS This study was supported by the VA Medical Research. and NIH grants GM25716 and GM27670.
REFERENCES 1. 'Brasseur, R.; Chatelin, P.; Goormaghtigh, E. Ruysschaert, J. M. A semi-empirical conformation analysis of the interaction of n-alkanols with dipalmitoylphosphatidylcholine. Biochim. Biophys. Acta 814: 227-236; 1985. 2. Chiou, J. S.; Ma, S. M.; Kamaya, H.; Ueda, I. Anesthesia cutoff phenomenon: lnterfacial hydrogen bonding. Science 248:583-585; 1990. 3. Colley, C. M.; Metcalfe, J. C. The localization of small molecules in lipid bilayer. FEBS Lett. 24:241-246; 1972. 4. Di Paolo, T.; Sandorfy. C. Hydrogen bond breaking potency of fluorocarbon anesthetics. J. Med. Chem. 17:809-814; 1974. 5. Eyring, H.; Woodbury. J. W.; D'Arrigo, J. S. A molecular mechanism of general anesthesia. Anesthesiology 38:414--425; 1973. 6. Ferguson, J. The use of chemical potentials as indices of toxicity. Proc. R. Soc. Lond. [Biol.] 127:387---404; 1939. 7. Hadzi, D. Infrared spectra of strongly hydrogen-bonded systems. Pure Appl. Chem. 11:435--453; 1965. 8. Hitzemann, R. J. Effects of non-electrolyte molecules with anesthetic activity on the physical properties of DMPC multilammelar liposomes. B iochim. Biophys. Acta 983:205-211; 1989. 9. Hobza, P.; Mulder, F.; Sandorfy, C. Quantum chemical and statistical thermodynamic investigations of anesthetic activity. 1. The interaction between chloroform, fluoroform, cyclopropane, and an O-H'"O hydrogen bond. J. Am. Chem. Soc. 103:1360-1366; 1981. 10. Hobza, P.; Mulder, F.; Sandorfy, C. Quantum chemical and statistical thermodynamic investigations of anesthetic activity. 2. The interaction between chloroform, fluoroform, and a N-H..-O = C hydrogen bond. J. Am. Chem. Soc. 104:925-928; 1982. 11. Kamaya, H.; Kaneshina, S.; Ueda, I. Partition equilibrium of inhalation anesthetics and alcohols between water and membranes of phospholipids with varying acyl chain-lengths. Biochim. Biophys. Acta 646:135-142; 1981. 12. Kamaya, H.; Ueda, I.; Eyring, H. General anesthesia and interfacial water. In: Fink, B. R., ed. Molecular mechanisms of anesthesia, vol. 2. Progress in anesthesiology. New York: Raven Press; 1980:429433. 13. Kaneshina, S.; Kamaya, H.; Ueda, I. Transfer of anesthetics and alcohols into ionic surfactant micelles in relation to depression of Krafft point and critical micelle concentration, and interfacial action of anesthetics. J. Colloid Interface Sci. 83:589-598; 1982. 14. Kaneshina, S.; Kamaya, H.; Ueda, I. Benzyl alcohol penetration into micelles, dielectric constant of the binding site, partition coefficient
15. 16. 17. 18.
19. 20.
21. 22.
23.
24. 25.
26. 27.
and high-pressure squeeze-out. Biochim. Biophys. Acta 777:75-83. 1984. Klemm, W. R. Dehydration: A new alcohol theory. Alcohol 7:4959; 1990. Mullins, L. J. Some physical mechanism in narcosis. Chem. Rev. 54:289-323; 1954. Orme, F. W.: Mononne, M. M.; Macey, R. I. Modification of the erythrocyte membrane dielectric constant by alcohols. J. Membr. Biol. 104:57--68; 1988. Perchard, J. P.; Josien. M. L. Etude des spectres vibrationnels de douze especes isotopiques d'ethanols monomers. J. Chim. Phys. 65: 1835-1875; 1968. Raevsky, O. A. Conformational analysis of organophosphorus compounds by means of IR spectroscopy and dipole moment methods. J. Mol. Struct. 19:275-287: 1973. Raevsky, O. A.; Vereshchagin, A. N.; Mumzhriva, N. G.; Zyablikova. T. A.; Alexandrova, N. A.; Arbuzov. A. E. A conformational study of c~-chloroethylphosphono-dichloridate by IR. NMR, dipole moments and Kerr constants. J. Mol. Struct. 36:299-308; 1977. Rowe, E. S.; Fernandes. A.; Khalifah, R. G. Alcohol interactions with lipids: a carbon-13 nuclear magnetic resonance study using butanol labeled at C-I. Biochim. Biophys. Acta 905:151-161; 1987. Trudeau, T.; Dumas, J. M.: Dupuis, P.; Guerin, M.; Sandorfy, C. Intermolecular interactions and anesthesia: Infrared spectroscopic studies. Top. Curr. Chem. 93:91-125; 1980. Tsai, Y. S.; Ma. S. M.; Kamaya, H.; Ueda, I. Fourier transform infrared studies on phospholipid hydration: Phosphate-oriented hydrogen bonding and its attenuation by volatile anesthetics. Mol. Pharmacol. 31:623-630; 1987. Ueda, I4 Kamaya. H. Molecular mechanisms of anesthesia. Anesthesiology 63:929-945; 1984. Ueda. 1.; Tseng, H. S.; Kaminoh, Y.; Ma, S. M.; Kamaya, H.; Lin, S. H. Anesthetics release unfreezable and bound water in partially hydrated phospholipid lamellar systems and elevate phase transition temperature. Mol. Pharmacol. 29:582-588; 1986. Wong, P. T. T.; Mantsch, H. H. High-pressure infrared spectroscopic evidence of water binding sites in 1,2-diacyl phospholipids. Chem. Phys. Lipids 46:213-224; 1988. Yokono, S.; Shieh, D. D.; Ueda, I. Interfacial preference of anesthetic action upon the phase transition of phospholipid bilayers and partition equilibrium of inhalation anesthetics between membrane and deuterium oxide. Biochim. Biophys. Acta 645:237-242; 1981.
