J. Membrane Biol. 9, 127-140 (1972) 9 by Springer-Verlag New York Inc. 1972

Water in Biological Membranes: Adsorption Isotherms and Circular Dichroism as a Function of Hydration M. J. T. Schneider and A. S. Schneider* Polymer Department, The Weizmann Institute of Science, Rehovot, Israel Received 20 January 1972

Summary. Adsorption isotherms of water on human red cell membranes have been measured and the results used to determine the Brunauer-Emmett-Teller monolayer coverage, saturation hydration, and relevant heats and entropies of hydration. The tightly bound monolayer corresponds to a complete covering of only one membrane surface and the saturation hydration is greater than values for globular proteins and phospholipids. A hysteresis loop is observed in the adsorption-desorption cycle. Circular dichroism spectra of membrane films have been measured as aLfunction of hydration in the ultraviolet region of the spectrum. The spectra resemble those of membrane suspensions with reduced scattering. The circular dichroism did not change upon dehydration of the membrane films from 92 to 0 % relative humidities, implying a stable membrane protein structure independent of both water content and lipid phase.

Of the various factors that determine biological membrane structure and function, the role of water has been among the least studied and is today only poorly understood. That water is an important component of membrane structure is generally agreed upon [17, 27]. Hydrophobic interactions among membrane components would require water to stabilize them. By coating polar surfaces and lining hydrophilic pores, water may enable ions to penetrate a primarily nonpolar core. Hydrophobic-hydrophilic changes in membranes, e.g., changes in membrane water content, have been postulated to be important in certain aspects of membrane function suc,h as nerve conduction [37]. However, the actual knowledge of the quantity of water in biomembranes, its location and mode of interaction is still very limited. In this paper we report on measurements of adsorption isotherms of water on human red cell membranes to obtain quantitative data on the water content of membranes, thermodynamic information on the water* Present address for reprint requests: Laboratory of Neurobiology, Building 36, Room 1D-02, National Institute of Mental Health, Bethesda, Maryland 20014.

128 membrane

M . J . T . Schneider and A. S. Schneider: interaction,

and

to check for hysteresis in the adsorption-

d e s o r p t i o n cycle. I n a d d i t i o n , w e h a v e m e a s u r e d c i r c u l a r d i c h r o i s m ( C D ) of membrane

f i l m s a s a f u n c t i o n of h y d r a t i o n t o d e t e r m i n e t h e effects of

water on membrane protein structure.

Materials and Methods Ghosts were prepared from human blood by hypotonic hemolysis in phosphate buffer, p H 7.4 as previously described, using successive washes of 300, 150, 50, 30 and 25 mosm [33]. Buffers and equipment were sterilized to prevent bacterial contamination. To avoid high salt concentrations in the dried samples, the ghosts were given one final wash with distilled water (about 8:1 v/v) immediately prior to preparing gravimetric samples. This reduces the salt concentration to 3 mosm or about 0.7 % of the dry weight of the membranes, at which values the residual salt hydration is estimated to be one to two orders of magnitude smaller than the measured membrane hydration. Samples of 5 or 6 ml of concentrated ghost suspension, which yielded about 100 mg dry weight, were poured into each glass weighing jar. The jars were placed in desiccators in the presence of saturated salt solutions to control relative humidity. The desiccators were evacuated to approximately 40 mm mercury and placed in a water bath at 20 q- 1 ~ or in the cold room at 4 _ 2 ~ Sample jars were removed daily from the desiccator, covered and weighed immediately. Measurements at 20 ~ were done on a Mettler automatic balance, accurate to 0.05 mg. The 4 ~ samples were weighed in the cold room on an analytical balance, accurate to 0.5 mg. A n empty jar was placed in each desiccator and weighed with the samples as a control for water adsorption on glass. Initial equilibrium was reached in 1 to 3 weeks depending on the relative humidity, and after that the samples generally took less than a day to equilibrate at a new relative humidity. A series of measurements was generally done by equilibrating initially at 92% relative humidity, then changing the salt solution to obtain progressively lower humidities. At the completion of the desorption isotherm, the sample was dried over concentrated H2SO 4 to constant dry weight, and then resorption was measured by reversing the order of the procedure. A few samples were more rapidly equilibrated initially at various re/ative humidities lower than 92 % to ensure that aging was not affecting the hydration curves. Measurements on the same sample were reproducible to within 0.5 % hydration. In a few cases desorption measurements were repeated after the completion of one desorption-resorption cycle and indicated that the curve is reproducible for a given sample. F o r different samples, the variation in the hydration value was generally less than 1% for humidities below about 80 %. In all, a total of 9 samples were measured. Percent hydration is with respect to fully dried weight, in units of g HzO/100 g dry membranes. Relative humidity values used for different saturated salt solutions are given in Table 1. In some cases there is variation (-t-1%) among values reported by different authors. These uncertainties may affect the shape of the hydration curve, especially in the high humidity region, where the isotherms are steep. Relative humidity is reported either as per cent or as a fractional partial pressure, as indicated in the text. Films for CD work were prepared by spreading a few drops of the appropriately diluted distilled-water-washed ghost suspension on one inside face of a 1-cm quartz CD cell. The film was allowed to equilibrate in an evacuated desiccator at the various relative

Membrane Water Adsorption and CD Spectra

129

Table 1. Saturated salt solutions used for maintaining constant relative humidity [1, 10, 13, 16, 23, 32, 34, 36, 41] 20 ~

4~

Relative humidity (%)

Salt

Relative humidity (%)

Salt

92 90.5 87 85 83 80 78.5 75.3 70 68.5 66 58.5 54 44 39 33 23 18.5 11 7

NaBrO 3 BaClz

92 87.5 85 81 79 76 74 68.5 63 58.5 43.5 34 24.8 23 14

BaC12 KC1 KBr NH4C1 NaNO 3 NaC1 KI CuCl 2 NaBr Mg(NO3) 2 NaI MgC12 KAcetate CaBr 2 LiCI

KzCrO 4

KC1 KBr (NH4)2SO 4 NHaC1 NaC1 KI CuC12 NaNO 2 NaBr Mg(NOa) z K2CO 3

:

NaI MgC12 KAcetate CaBr 2 LiC1 :NaOH

Where necessary, relative: humidity values were obtained by interpolation or extrapolation of temperature-dependent curves. In the case of KBr and (NH4)2SO4, the variation among authors was considerable and we measured the relative humidity of our saturated solutions by equilibrating them against sulfuric acid solutions. The resulting H2SO 4 concentrations were measured by titration, and corresponding relative humidities were obtained from ref. [22]. H2SO 4 solutions were also used to obtain low relative humidity values including 0 %.

humidities. The optical cells were then stoppered to maintain the humidity over the film, and spectra were measured on a Cary 60 spectropolarimeter equipped with a circular dichroism attachment.

Results W e present b e l o w a d s o r p t i o n i s o t h e r m s of w a t e r o n h u m a n e r y t h r o c y t e m e m b r a n e s at t e m p e r a t u r e s of 20 a n d 4 ~ together with the resulting B r u n a u e r - E m m e t t - T e l l e r or B E T [8] a d s o r p t i o n p a r a m e t e r s , a n d t h e r m o d y n a m i c d a t a derived f r o m such isotherms. I n addition, the circular d i c h r o i s m spectra of films of e r y t h r o c y t e ghosts are given as a f u n c t i o n of h y d r a t i o n f r o m 92 to 0 % relative humidity.

130

M.J.T. Schneider and A. S. Schneider: 35 F

j

I

50!

q

2a

[

rf

2C

g

ill/

:r:

/

f"

"g

l0

5

0

!O

20

50

40 50 60 70 • (% Relalive Humidity]

80

90

I00

Fig. 1. Sorption isotherms at 20 ~ for membrane preparation I. Upper curve desorption 9 lower curve adsorption.; repeat desorption 3~

j

3C

25

2O

m 15

i~"

-~176 :z:

z"

I0

00

I0

20

50

40

50

60

70

80

90

100

X (% Relalive Humidity)

Fig. 2. Sorption isotherms at 20 ~

for membrane preparation II.

Upper curve desorption o; lower curve adsorption ~; repeat desorption

Figs. 1 and 2 show the sorption isotherms at 20 ~ given by samples from two different ghost preparations. Hydration a, in units of g HzO/100 g dry weight membranes, is plotted as a function of relative humidity x. The familiar sigmoid shape curve observed for many sorbing systems is apparent, and it is interesting to note that the quantity of water adsorbed at relative humidities of 50 % and below is lower than values obtained for proteins,

Membrane Water Adsorption and CD Spectra

30

_~

131

r

25

~ 20

0

I

I

I

I0

20

30

I

I

I

I

I

I

I

40

50

60

70

80

90

I00

x(% Relati~,e Humidi~y)

Fig. 3. Sorption isotherms at 4 ~ for membrane preparation 2[I. Upper curve desorption o; lower curve adsorption

lipids and polysaccharides. The existence of a hysteresis loop in the desorption-adsorption cycles of up to about 3 % hydration can be seen for both membrane preparations. The magnitude of the membrane hydration and hysteresis reproduces quite well for the two ghost preparations, but there is some variation in the shape of the hysteresis loop that may in part depend on sample preparation. Fig. 3 shows the isotherm at 4 ~ for the ,;ame ghost preparation as in Fig. 2. The amount of water adsorbed at 4 ~ is considerably greater than at 20 ~ for the same material. The shape of the hysteresis loop is also different at the two temperatures, the hysteresis being greater at low relative humidities and hydrations for the low temperature case. The desorption isotherms at 20 ~ and 4 ~ are compared in Fig. 4. Although the average curves appear to show a greater temperature effect at low humidity, converging at higher relative humidities, measurements on individual samples show that the membranes consistently bind more water at 4 ~ than at 20 ~ The above data give a good fit to the BET equation [8] u p t o a relative humidity of about 70 % at 20 ~ and 50 % at 4 ~ as shown in Fig. 5. The BET parameters for our red cell membranes are summarized iin Table 2 along with representative parameters for proteins, lipids and polysaccharides for comparison. For the membranes, the BET parameter al, which is considered to be the monolayer adsorption, has a relatively low value, as does the net hydration at relative humidifies of 50 % and below. Thus, it appears that some of the polar groups of the membrane componLents must

132

M.J.T. Schneider and A. S. Schneider: 55 30 z5 o 2~ oo_ j~ .~/~f

t~ 10 I.er ~

10

20

50 4-0 50 60 x (% Relalive Humidily)

70

Fig. 4. Desorption isotherms at 20 ~ (o lower curve) v s . 4 ~ (~ upper curve). Double arrows ~ show individual sample checks which were equilibrated first at 4 ~ and then at 20 ~ .25

/

,20

J

. . ~

,15"

/

/

-'"&

9I 0

+ . ~ ~..~/ . . . .

.05

0(~

I

1

I_

I

I

.2

.5

4

I_,

.5

I

I

I

6

.7

.8

x (RelGtive Humidityl

Fig. 5. BET plots using desorption isotherms. 20 ~ ~; 4 ~ o

be shielded from exposure to water, probably due to interactions with each other. The BET parameter c is a measure of the strength of binding to the primary binding sites. When c = o% the first adsorbed layer fills completely before the second begins. The magnitude of c given by membranes is quite similar to that of globular proteins, lecithin and polysaccharides (Table 2). It is quite small, and so the concept of an adsorbed monolayer is only a first approximation in these systems. However, as is commonly done, we

Membrane Water Adsorption and CD Spectra

133

Table 2. Water adsorption values and BET parameters for membranes and membrane components Adsorbent

g H20/100 g dry weight a i

Membranes 20 ~ Membranes 4 ~ Globular proteins [2, 9, 14, 19, 26] Fibrous proteins (collagen and gelatin) [9] Lipids (egg lecithin) [15] Polysaccharides [4, 5]

4.1 5.1 6-7 9 5.6 5-11

as0 %

as

7.2 9.2 11-12 17 9.5 10-16

73 ___115 77 ___20 45-65 62-85 44 -

c

7-8.3 9-11 9-14 17.6 7.7 10

can estimate the area which would be covered by the most tightly bound water and compare it with the surface area of the membranes. Assuming a value of 10.5 h e [7] for the area covered by one water molecule, the coverage of the first hydration layer is 1.4 • 108 cm z at 20 ~ Using values of 1.2 • 10 -12 g for the weight of one ghost [11] and 1.55 x 10 -6 c m 2 for its surface area [30, 40], we estimate that 100 g of membranes have a surface of 1.3 • 108 cm 2. Then the most tightly bound water is approximately enough to completely cover only one surface of the membrane. At high humidities, the membranes are more highly hydrated than lipids or globular proteins. Extrapolation of an a / x vs. x plot to 100% relative humidity yields a value of hydration at saturation, as = 70 to 80 g HzO/100 g dry weight. Although this value is only a rough estimate, a comparison of the values given in Table 2 clearly demonstrates the higher hydration of the membranes in this region. This may be related to the surface charge on the membranes which leads to long-range interaction with water, or to swelling or capillarity of the membrane matrix. Hysteresis is commonly observed in sorbing systems and the values of 1 to 3 % given by membranes are within the range observed for proteins and polysaccharides [2, 5, 14]. Egg lecithin did not show hyste,resis [15]. Hysteresis may be related to swelling of the sorbing matrix or to capillarity. It is interesting to compare thermodynamic parameters calculated from the membrane isotherms with values similarly obtained for membrane components. Although there is some question about the validity of applying thermodynamics to systems showing hysteresis, it has commonly been done for proteins and polysaccharides, and we expect the errors involved here to be relatively small. Differential and integral heats and entropies of adsorp-

134

M.J.T. Schneider and A. S. Schneider:

5 -r 4 E "6 3