LOW-ANGLE X-RAY DIFFRACTION AND ELECTRON-MICROSCOPE STUDIES OF ISOLATED CELL MEMBRANES

Transcription

1 J. Cell Sci. I, (1966) 287 Printed in Great Britain LOW-ANGLE X-RAY DIFFRACTION AND ELECTRON-MICROSCOPE STUDIES OF ISOLATED CELL MEMBRANES J. B. FINEAN, R. COLEMAN, W. G. GREEN* AND A. R. LIMBRICK Department of Medical Biochemistry and Pharmacology, University of Birmingham, Birmingham, 15 SUMMARY A sequence of low-angle X-ray diffraction patterns obtained during the controlled drying of a preparation of rat erythrocyte ghosts has been interpreted with the aid of corresponding electron micrographs and of a parallel study of myelin isolated from guinea-pig brain. A diffraction pattern that persists down to a level of % hydration of the sample is believed to arise from the close packing of native erythrocyte membranes. Each membrane is about 100 A thick and it is suggested that it consists of a predominantly continuous bimolecular layer of lipid, with non-lipid components associated with both surfaces. Further changes in diffraction pattern which accompany continued drying could be interpreted either as a change from a lamellar to a hexagonal structure or as the formation of a multiphase system. Evidence is put forward to support the latter interpretation. INTRODUCTION Sequences of well-defined, low-angle X-ray diffraction patterns have been obtained during controlled dehydration of preparations of a variety of isolated cell membranes (Finean, Coleman & Green, 1966). The early patterns were tentatively interpreted in terms of a lamellar system involving hydrated membranes, and the final patterns included reflexions that were suggested probably to represent separated lipid phases. In this paper, X-ray diffraction and electron-microscope data obtained from isolated erythrocyte membranes are presented in full and information obtained from parallel studies of isolated and sonicated myelin is put forward as justification for interpreting the earliest patterns in terms of native hydrated erythrocyte membranes. MATERIALS AND METHODS A preparation of erythrocyte ghosts was obtained by taking blood from the neck of a freshly killed rat (white albino) direct into approximately 100 volumes of o-ooi M bicarbonate at ph 7-5. The membranes were sedimented by centrifugation at 1200 g for 10 min and washed by redispersion in bicarbonate four times until the supernatant was haemoglobin-free. The membranes were dispersed finally in distilled water before being packed into a compact pellet by centrifugation at g for 1 h. Some samples Present address: Department of Biochemistry, University of Toronto, Ontario, Canada 19 Cell Sci. 1

2 288 J. B. Finean, R. Coleman, W. G. Green and A. R. Limbrick were fragmented with an MSE sonic vibrator for 3-4 min (with cooling) and resedimented in order to obtain a more compact membrane pellet. Myelin was prepared from guinea-pig brain by a modification (Evans & Finean, 1965) of a method described by Eichberg, Whittaker & Dawson (1964). Some samples were subsequently sonicated for 10 min (intermittently, with cooling) and further fractionated by ultracentrifugation to isolate a sample of membrane fragments sedimenting above g for 20 min. This sample was twice dispersed in water and packed into a compact pellet by centrifugation at g for 1 h. For X-ray diffraction studies a sample of the membrane pellet was suspended across the gap in a small brass ring (Fig. 1) which was mounted in a closed chamber Mica window Specimen chamber Fig. 1. Diagram of specimen holder. This is 1 cm in diameter. between thin mica windows. A stream of air brought to controlled humidity by bubbling through a calcium chloride solution of selected strength ( %) was passed through the chamber, the rate offlowand the strength of the calcium chloride solution being manipulated to produce a controlled dehydration and ultimately to maintain a constant level of hydration. As the sample dried it formed a flattened thread across the gap in the holder and this was normally oriented edge-on to the direction of the X-ray beam. A selection of specimen holders with jaws of a variety of shapes and depths facilitated the mounting of samples of different consistencies in the most favourable way for achieving good orientation of membranes. Sequences of low-angle diffraction patterns were recorded during the slow drying of samples using either a line-collimated (Finean, 1953) or a line-focused (Franks, 1955) X-ray beam from a Philips Fine Focus Tube and an automated film changer. Exposure times of min provided adequate intensities at resolutions up to 200 A. A point-focused X-ray beam (Franks, 1955) was used to examine selected stages of dehydration maintained for 8-12 h at controlled humidity. The fully dried samples were examined both edge-on and face-on to the X-ray beam. In all diffraction cameras sample-to-nhn distances were in the region of 100 mm. In some experiments the specimen holder was removed briefly from the humidity chamber between exposures over the critical period of dehydration and weighed on a

3 Isolated cell membranes 289 semi-micro balance so that the water content of the sample could be assessed in relation to the diffraction changes. Reversibility of changes was investigated by immersing the sample carrying part of the specimen holder in water for a minimum of 1 h and then re-examining in an atmosphere saturated with water vapour in the humidity cell. Fully dried samples were removed from the holders and supported on a mica strip in a cell which could be maintained at temperatures up to 200 C. In these experiments diffraction patterns were recorded with the sample heated to 80 C and C, and then after cooling for 4 and 16 h. Samples for electron microscopy were taken from the initial pellet and also at various stages of dehydration in the humidity cell used for the diffraction experiments. They were fixed either with 6-5 % glutaraldehyde in cacodylate buffer for 30 min, followed by washing and further fixation in 1 % osmium tetroxide in veronal-acetate buffer (ph 7) for 4 h, or with osmium tetroxide alone. The ethanol-dehydrated specimen was finally embedded in Araldite and thin sections were stained for 30 min in 25 % uranyl acetate in methanol before examination in the electron microscope. The dehydrated samples were sufficiently thin for sections to include the whole thickness of the ribbon. The effects of sonic treatment on the erythrocyte membranes were observed using negative-staining techniques in which non-fixed or osmium-tetroxide-fixed membranes were dispersed in a 1 % solution of sodium phosphotungstate at ph 7 and allowed to settle on a carbon-covered specimen grid. RESULTS X-ray diffraction Examples from the sequences of low-angle X-ray diffraction patterns recorded from samples of pellets of erythrocyte ghosts are reproduced in Fig. 3. The main series of patterns was obtained using a line-focused X-ray beam. Diffraction from most samples was initially extremely weak but as water was removed a well-defined diffraction pattern gradually developed which eventually featured 3 or 4 equally spaced reflexions which could readily be identified as the first 3 or 4 orders of diffraction from a lamellar repeat of A (Fig. 3 b). This fundamental periodicity did not change appreciably during the whole of the period of development. The relative intensities of the first three diffraction orders were approximately 100:34:13. In some experiments the state of hydration corresponding to the maximum development of this lamellar pattern was maintained for 8-12 h in order to record the pattern using a point-focused X-ray beam (Fig. 3, right-hand side). All reflexions showed the same orientation which indicated a precise orientation of crystallites with lamellae predominantly parallel to the broad face of the ribbon-like specimen. The water content of the sample at this stage of the experiment was of the order of % whereas the water content of the fresh pellet varied from 70% in the sonicated material to 90 % in pellets of intact ghosts. This corresponds to the removal of about 95 % of the water from the original pellet. Continued dehydration beyond this 19-2

4 290 J. B. Finean, R. Coleman, W. G. Green and A. R. Limbrick stage produced a relatively abrupt change in diffraction pattern. Diffractions in the region of 100 A and 50 A became broader and more intense and the 50-A diffraction quickly resolved into two sharp reflexions at about 55 A and 51 A (Fig. 3 c). The spacing corresponding to the peak of intensity of the ~ 100-A band gradually decreased but at no stage was it possible to resolve it into more than one line. The diffraction pattern recorded from the fully dehydrated sample (Fig. 3d) featured a broad and intense line at 85 A and sharper and less dense lines at 53 A and 43 A. Patterns recorded from the dried sample using the point-focused X-ray beam directed parallel to the broad face and perpendicular to the length of the ribbon showed ah 1 three bands to be similarly and well orientated (Fig. 3d), whilst examination perpendicular to the broad face of the ribbon gave much less intense diffraction featuring a poorly orientated band at about 83 A. When the dried sample was examined at high temperatures the 53-A band was eliminated at a temperature between 70 and 80 C and the 43-A band decreased significantly in spacing and became somewhat faint and diffuse at C. The 85-A band was relatively unaffected by heating the sample to these temperatures. On re-cooling the sample, a diffraction reappeared at about 50 A. The fully dried samples that were immersed in water for 1-2 h yielded diffraction patterns very similar to the lamellar pattern found to be characteristic of the intermediate stage of dehydration of the original sample. This pattern was also obtained from dried samples that were immersed in osmium tetroxide solution. The X-ray diffraction data from samples of sonicated erythrocyte ghosts were essentially the same as from the intact ghosts but with slightly poorer definition of the diffraction bands. Freshly isolated myelin pellets yielded diffraction patterns that were very much less intense and slightly less well defined than that recorded from the intact (fully hydrated) brain tissue, but the diffraction spacings were very similar (a in Figs. 5 and 4, respectively). Controlled dehydration of the sample led to a considerable intensification and sharpening of the reflexions before the introduction of an additional diffraction at about 63 A (Fig. 4c) indicated changes in structure (Elkes & Finnean, 1942). At this stage the sample contained approximately 30-40% water. The pattern of the fully dried sample featured diffractions at 63 A and 45 A (Fig. 5 d). Samples from pellets of myelin that had been sonicated before sedimentation gave only very faint diffraction effects but reflexions at approximately 80 A and 40 A appeared during the early stages of controlled dehydration (Fig. 6b). The diffraction bands were much broader than those obtained from the non-sonicated myelin and the introduction of an additional reflexion in the region of 60 A during the later stages of dehydration was seen only as an intensification and improved definition of the outer edge of the 80-A band (Fig. 6 c). Apart from the poor definition the sequence of diffraction changes associated with dehydration of the sonicated material appeared essentially similar to those observed with non-sonicated myelin and with intact tissue.

5 Isolated cell membranes 291 Electron microscopy Electron micrographs of sections from preparations made from the initial pellet of erythrocyte ghosts (Fig. 7) demonstrated a clear unit-membrane structure of overall dimension about 70 A in the membranes and there was little evidence of contamination with material that might be identified as haemoglobin. The membranes were, however, relatively widely separated. Electron micrographs of preparations at intermediate stages of dehydration showed that the structure had now condensed to form assemblies of closely packed membranes equivalent to small crystallites aligned predominantly parallel to the broad face of the ribbon (Fig. 10), but with a tendency towards a cylindrically symmetrical arrangement in thick samples. At a stage corresponding to the maximum development of the lamellar pattern in the diffraction series, the system appeared fully condensed, with regions of crystalline order of membranes interspersed with membrane-enclosed pockets of seemingly amorphous material (Fig. 11). The appearance was such as to suggest that some material, possibly haemoglobin, had gradually collected into pockets during dehydration. In the regions where membranes formed close-packed arrays the repeating layer appeared to include only one membrane thickness but the unit-membrane features remained separated by narrow bands of low density (Fig. 11, inset). The pair of dense lines to be included in the unit membrane have been identified by tracing the profiles of individual membranes and noting which lines remained paired when membranes parted to enclose pockets of contaminating material. In this way it was established that the layer of low density located at the centre of the unit membrane was wider than that which separated membranes. The dimension of the repeating layer was found to be approximately 100 A, the two dense lines being about 25 A thick and the low density layers 30 A and 20 A, respectively. Ordered structure was more difficult to demonstrate in preparations that had been fully dried before fixation but some lamellar structure was evident. Studies of negatively stained preparations by electron microscopy using both fixed and non-fixed material demonstrated that sonication effected a strikingly uniform fragmentation of the ghosts down to microsomal (~ 1000 A) dimensions (Figs. 8, 9). Electron micrographs of sections of freshly isolated myelin (Fig. 12) showed a wide spectrum of myelin fragments, ranging from vesicular profiles involving single layers to largely intact multilayered myelin sheaths. The condensation achieved by dehydration appeared to produce large areas of crystalline order (Fig. 14) and the late stages of drying were characterized by the formation of a multiplicity of layered systems, some of which showed fine layering of the type previously attributed to lipid phases (Finean, 1961). The samples of sonicated and fractionated myelin showed predominantly small vesicular profiles in which membrane detail was poorly defined (Fig. 13) and the regions of crystalline order in the dehydrated samples were very small.

6 292 J. B. Finean, R. Coleman, W. G. Green and A. R. Limbrick DISCUSSION The comparison of X-ray diffraction data obtained during the controlled drying of isolated myelin with those for myelin in the intact nerve tissue clearly demonstrates the feasibility of reconstituting a dose-packed native membrane system from a preparation of isolated and dispersed membrane fragments. The diffraction bands arising from the single lamellar phase present down to very low water levels in the sample of isolated myelin differ only in definition from those from the intact (fully hydrated) brain tissue, and this difference can readily be accounted for in terms of a reduced size of crystallites in the diffracting specimen. The observations suggest that the bulk of the water contained in the membrane pellet is located between and about the membranes rather than within them, and that most of it can be removed by evaporation without modifying the membrane structure. The point at which dehydration does begin to affect the membrane structure is clearly indicated by well-characterized changes in diffraction pattern. It seems reasonable to assume that in this preparation, which is free of non-membranous contaminants, the water content of the sample immediately prior to the detection of diffraction changes represents the essential water of hydration of the myelin layer. This is a more direct and probably much more reliable estimate of the water content of myelin than any previously reported, and the value of 30-40% is a little lower than earlier estimates (Finean, 1957, 1961). The diffraction changes which accompany the dehydration of a sample of erythrocyte ghosts are essentially similar to those described for isolated myelin. There is a clearly denned point during dehydration at which changes in low-angle X-ray diffraction patterns indicate a modification of membrane structure, but prior to this it seems probable that the membranes retain the water of hydration essential to their structural integrity and in this sense they remain in the native state. The diffraction pattern observed before this point is reached arises from a single lamellar system which can be identified in corresponding electron micrographs as regions of close-packing of erythrocyte membranes. The pattern therefore relates to the native erythrocyte membranes. The repeat period in this lamellar system is A (from the diffraction pattern) and the electron micrographs indicate that this represents the thickness of one membrane. It includes a unit-membrane structure but in this condensed system such units remain separated by a narrow band of low density. The discrepancy between the periodicity calculated from the diffraction bands and the thickness of the repeating layer measured in the electron micrographs is attributed to the preparative techniques employed in electron microscopy. If the unit-membrane structure is assumed to include a bimolecular layer of lipid with protein and perhaps other components associated with both ionic surfaces then the electron-density profile of the repeating unit will be of the form shown in Fig. 2. Such a profile will readily account for the approximately regular decrease in diffraction intensities through the diffraction orders, but the intensity data are not sufficiently extensive or precise to confirm this as a unique solution or to indicate directly the magnitude of the variations in electron-density levels. The deep trough of electron density is considered to represent the hydrocarbon

7 Isolated cell membranes 293 region of the lipid layer and its width is tentatively equated with that for the equivalent feature in the myelin layer (Finean & Burge, 1963). This lipid phase includes approximately equimolar proportions of phospholipid and cholesterol, and both types of molecule are included in the schematic diagram of the membrane structure. The shallow trough in the electron-density curve would correspond to the narrow band of low density which separates unit-membrane features in the layered system observed in electron micrographs. The region is not necessarily of low density in the non-fixed preparation and in the absence of precise information as to the nature and structural Nonlipid Lipid Cholesterol Phospholipid 110 A Fig. 2. Suggested form of electron-density profile of the rat erythrocyte membrane together with schematic drawings (to scale) of phospholipid and cholesterol molecules. Stippled regions indicate possible locations and thicknesses of non-lipid layers. form of the non-lipid components of erythrocyte membranes the region allocated to them in the schematic structure is represented simply as a uniformly stippled area. The proportion of lipid calculated for such a structure is roughly in accord with preliminary chemical analyses of this erythrocyte membrane preparation, and it is hoped that more detailed analyses will provide data that can be related quantitatively to the electron-density curve. The water content (10-20%) of the membrane preparation immediately prior to the detection of structural changes in the membrane is taken to represent a very low level of hydration of the native erythrocyte membrane as isolated in these experiments. However, in view of the uncertain effects of the specimen heterogeneity demonstrated by the electron micrographs the exact relationship between the measured water content and the repeat period in the membrane crystallites cannot yet be assessed. Two alternative interpretations of the changes in low-angle diffraction patterns occurring during the later stages of dehydration have been considered. The values for the three spacings recorded from the fully dried sample are in the ratio of approximately 1: ^3 : ^4 as required for a two-dimensional hexagonal network (Luzzati & Husson,

8 294 J- B- Finean, R. Coleman, W. G. Green and A. R. Limbrick 1962), and the observed sequence of diffraction changes might be accounted for in terms of a gradual change from a lamellar to a hexagonal arrangement of structure. The lamellae show little change in thickness and as the reflexion corresponding to the [no] spacing of hexagonal network is about 51 A, which is close to one half of the lamellar thickness, the change to a hexagonal system can be envisaged as a break-up of the lamellae into cylindrical micelles of about 100 A diameter and a rearrangement to hexagonal packing. The long axes of the micelles would be assumed to be parallel to the original lamellae. The observed reversibility of the changes when the system is rehydrated would be readily understood on the basis of a lamellar-hexagonal change but the identical orientations of the three reflexions in the pattern from the dried sample and the appreciable variations in their widths would be difficult to reconcile with this explanation. The disappearance of the 53-A reflexion when the sample is heated to 80 C, together with the fading of the 43-A band, as contrasted with the stability of the 85-A reflexion above 120 C, suggest that these three reflexions may represent more than one diffracting phase. The variations in the definitions of the bands could be explained in this way as also the identical orientations, but the ready reversibility of the changes on rehydration is perhaps a little surprising. An interpretation in terms of the formation of multiple phases has previously been favoured for the diffraction changes accompanying the dehydration of peripheral nerve myelin but in this case the process was not reversible, although there was a return to a single diffracting system. So far the diffraction experiments have not provided any indication of an arrangement of subunits of macromolecular dimensions within the erythrocyte membranes. As the dehydrated samples tended to have a cylindrical symmetry, reflexions relating to such an organization might have been expected to be recorded simultaneously with the lamellar diffractions but none was detected. Even in the pattern recorded with the X-ray beam perpendicular to the broad face of the sample ribbon, the one faint reflexion observed could be considered to represent the strongest of the lamellar reflexions. The electron micrographs of preparations made from the fully dried erythrocyte sample gave evidence of a lamellar structure but as there were strong indications that fixation in aqueous solutions tended to rehydrate the specimen these data may be irrelevant to the problem. The demonstration in the present experiments that what has been identified as the pattern of the native hydrated membranes can also be obtained through the rehydration of fully dried samples, enhances the value of X-ray diffraction studies of preparations of erythrocyte ghosts previously reported by Husson & Luzzati (1963). These authors rehydrated frozen-dried preparations of human erythrocyte ghosts and detected a lamellar system with a periodicity of about 170 A. This value is probably directly comparable with the value of A obtained for rat erythrocyte ghosts in the present experiments, and indicates the possibility of species variations, for which we already have some evidence but have not as yet studied systematically. It seems reasonable to conclude from this study that low-angle X-ray diffraction patterns relating to the native membrane can readily be obtained through the con-

9 Isolated cell membranes 295 trolled dehydration of preparations of isolated membranes and this opens up numerous possibilities for the study of their structural characteristics, both in the native and in experimentally modified states. We are grateful to Professor A. C. Frazer for his encouragement of these studies, and to the Wellcome Trust and the Rockefeller Foundation for providing equipment. REFERENCES EICHBERG Jr., J., WHITTAKER, V. P. & DAWSON, R. M. C. (1964). Distribution of lipids in subcellular particles of guinea-pig brain. Biochem. J. 92, ELKES, J. & FINEAN, J. B. (1949). The effect of drying upon the structure of myelin in the sciatic nerve of the frog. Discuss. Faraday Soc. no. 6, pp EVANS, M. J. & FINEAN, J. B. (1965). The lipid composition of myelin from brain and peripheral nerve. J. Neurochem. 12, FINEAN, J. B. (1953). A versatile X-ray camera for low-angle diffraction studies. J. scient. Instrum. 30, FINEAN, J. B. (1957). The role of water in the structure of peripheral nerve myelin. J. biophys. biochem. Cytol. 3, FINEAN, J. B. (1961). Electron microscope and X-ray diffraction studies of the effects of dehydration on the structure of nerve myelin. J. biophys. biochem. Cytol. 8, FINEAK, J. B. & BURCE, R. E. (1963). The determination of the Fourier transform of the myelin layer from a study of swelling phenomena. J. molec. Biol. 7, FINEAN, J. B., COLEMAN, R. & GREEN, W. G. (1966). Studies of isolated plasma membrane preparations. Arm. N.Y. Acad. Sci. U.S.A. (in the press). FRANKS, A. (1955)- An optically focusing X-ray diffraction camera. Proc. Phys. Soc. B 68, HussON, F. & LUZZATI, V. (1963). Structure of red-cell ghosts and the effect of saponin treatment. Nature, Lond. 197, 822. LUZZATI, V. & HUSSON, F. (1962). The structure of the liquid-crystalline phases of lipid-water systems. J. Cell Biol. 12, (Received 25 January 1966)

10 296 J. B. Finean, R. Coleman, W. G. Green and A. R. Limbrick Fig. 3. Examples from a sequence of low-angle X-ray diffraction patterns recorded during dehydration of a sample of rat erythrocyte ghosts. On the left patterns obtained using a line-focused X-ray beam; on the right equivalent patterns obtained with a point-focused X-ray beam are included for critical stages b and d. Figs Examples from sequences of low-angle X-ray diffraction patterns recorded during the dehydration of samples of white matter from guinea-pig cerebrum, myelin isolated from guinea-pig brain, and isolated myelin subjected to sonic vibrations, respectively. Line-focused X-ray beam. All patterns are reproduced on the same scale and in each case pattern a was obtained from the fully hydrated preparation.

11 Journal of Cell Science, Vol. 1, No. 3 o 37 A 40 A 4 5 J. B. FINEAN, R. COLEMAN, W. G. GREEN AND A. R. LIMBRICK (Facing p. 296)

12 Journal of Cell Science, Vol. i, No. 3 V 8 Figs Electron micrographs from rat erythrocyte ghosts. Fig. 7. Thin section from preparation of freshly isolated pellet. Uranyl-acetate stained, x Inset x Fig. 8. Intact ghosts fixed with buffered osmium tetroxide and examined by negative-staining technique, x Fig. 9. Sonicated ghosts fixed with buffered osmium tetroxide and examined by negative-staining technique, x J. B. FINEAN, R. COLEMAN, W. G. GREEN AND A. R. LIMBRICK

13 "journal of Cell Science, Vol. i, No. 3 Figs, io, 11. Electron micrographs from a uranyl-acetate-stained thin section of a sample of rat erythrocyte ghosts dried to a stage corresponding to the maximum development of the lamellar diffraction pattern before fixation and embedding. Fig. 10. x Fig. II. x Inset x Arrows indicate limits of structure identified as the unit membrane. J. B. FINEAN, R. COLEMAN, W. G. GREEN AND A. R. LIMBRICK

14 Journal of Cell Science, Vol. i, No. 3 Figs Electron micrographs from preparations of myelin isolated from guineapig brain. Fig. 12. Normal preparation, x Inset x Fig. 13. Sonicated preparation, x Fig. 14. Sample of normal preparation dried to equilibrium in air before preparation for electron microscopy, x J. B. FINEAN, R. COLEMAN, W. G. GREEN AND A. R. LIMBRICK