- Email: [email protected]
Crystalline sheets of tropomyosin
J. Mol. Biol.
(1984) 174, 231-238
Crystalline
Sheets of Tropomyosin
In the presence of spermine tropomyosin forms sheets having two-dimensional crystallinity and tactoids. The most common form of sheet has cmm symmetry with a = 80 nm and b = 5 nm. The structure of this sheet has been solved in projection to a nominal resolution of 15 nm by combining data from electron diffraction and electron microscopy. Analysis of this pattern and that of rarely observed sheets having p2 symmetry (a = 40 nm, b = 5 nm and y = SO”) indicated that the cmm structure was formed by superposition of two p2 sheets. The tropomyosin molecules in each p2 sheet were arranged in rows directed along the p2 (0, 1) lattice lines, with all the molecules in one row having the same polarity and lying antiparallel to the molecules in adjacent rows. These rows associated in pairs, possibly by the supercoiling of the molecules in one row about those in the neighbouring row.
Tropomyosin is a rod-shaped molecule which lies in the grooves of the actin helix in the thin filaments of muscle and many non-muscle systems. In conjunction with troponin, tropomyosin plays a key role in the regulation of contraction of skeletal and cardiac muscle, although its function in smooth muscle and nonmuscle systems is less clear. Muscle tropomyosin is about 41 nm long and is constructed from two parallel E-helical chains arranged, in register, in a coiled-coil configuration (Caspar et al., 1969; Cohen et al., 1972; Stewart, 1975a). A number of tropomyosin aggregates have been studied in order to obtain structural information and also to investigate the interactions of tropomyosin with itself and with other muscle proteins. Tactoids having one-dimensional order are formed with divalent metal cations (Caspar et al., 1969) and molecular positions have been established in magnesium paracrystals, which are the commonest form (Ohtsuki, 1974; Stewart, 19756,1981). Crystals in which the molecules are arranged in an open kite-like meshwork have been examined to about 2 nm resolution and molecmar positions established (Phillips et al., 1978,1981). A sheetlike crystal produced by CaCl, at pH 6 has also been reported but has not been analysed at high resolution (Longley, 1977). The present paper describes the production of thin sheet-like crystals of tropomyosin using spermine instead of metal ions as precipitant. Electron diffraction and electron microscopy indicates and show structure to better than 1.5 nm that these sheets have cmm symmetry resolution. Solutions of 1 to 5 g/l rabbit tropomyosin in 25 mM-Tris . HCl (pH 8) produced an immediate precipitate when treated with spermine up to about 200 m;M. No precipitate was observed at higher spermine concentrations, although one was immediately produced if the solution was diluted with 25 mw-Tris . HCl (pH 8) or with water. Negatively stained preparations of tropomyosin-spermine precipitates showed the presence of both tactoids and thin sheets. Generally the tactoids were more numerous than the sheets and often resembled those produced 0022-2836/84/0902314Pi
pjO3.00/0
231
0
1984 Academic
Press Inc.
(London)
Ltd.
-‘Y:!
&I. YTEWAK’I
by magnesium or barium (Caspar et al., 1969). Overall the spermine preparations showed greater polymorphism than that observed with divalent metal ions (Caspar et al., 1969). The same general distribution was seen with either cardiac ol-tropomyosin or with a mixture of CI- and p-tropomyosins derived from rabbit skeletal muscle, and no influence of the oxidation state of the tropomyosin thiol residues nor of temperature over the range 0 to 25°C was observed. Specimens containing a higher proport,ion of sheets were obtained by adding spermine to 10 KIM at 4”C, centrifuging the sample and allowing the supernatant’ to slowly come to room temperature. Figure 1 shows the general appearance of the sheets, which appeared to be extremely thin and were up to several micrometres across. Optical diffraction of electron micrographs (Fig. 2) indicated the presence of two-dimensional order and, at higher magnification, clear series of striations could be seen at an angle of about 10” to the axis of the sheet (Fig. 3). Low-angle electron diffraction patterns (Fig. 4) were easily obtained. The optical and electron diffraction patterns were characterized by a series of row lines spaced about li5nm apart which were finely sampled at every 1/40nm. The spots in the first row line were clearly staggered relative to those on the meridian and second row line, and the pattern indexed on an 80 nm x 5 nm rectangular unit cell. All reflections with (h + k) odd were absent,
FIG. 1. Low magnification micrograph of a thin crystaliine sheet of tropomy.osin formed with spermine negatively stained with many1 acetate. The bar represents 250 nm. Frozen rabbit hearts were used to prepare cc-tropomyosin as described by Stewart (1981). Skeletal tropomyosin (a mixture of a- and P-tropomyosins) was prepared from rabbit back and thigh muscles as described by Stewart & McLachlan (1976). Preparations were homogeneous as assessed by sodium dodecyl sulphate/ polyacrylamide gel electrophoresis. As required. the tropomyosin cysteine residues were oxidized with dithio bis-nitrobenzoate (Lehrer, 1975) or reduced with dithiothreitol, Specimens were applied to thinly (approx. 10 nm) carbon-coated 400 mesh grids, negatively stained with 1% (w/v) aqueous many1 acetate and examined at 80 kV in a Philips EM300, EM301 or EM400 eiectron microscope. Micrographs were calibrated using negativeiy stained tropomyosin magnesium pamcrystals. taking t.heir axial repeat as 39.5 nm (Caspar et al., 1969).
LETTERS
FIG. 2. Optical diffraction 1/dnm.
pattern
TO THE
EDITOR
from Fig. 1, showing 2-dimensional
233
order. The bar represents
FIG. 3. High magnification image of a uranyl acetate negatively stained sheet. In addition to the fine periodicity perpendicular to the axis, there are quite distinct striations at about lo” to the axis. These are most easily seen by viewing the micrograph obliquely from the side. The bar represents 100 nm.
indicating a centric space group, and the obvious mirror symmetry of the pattern about the meridian and equator established the space group as cmm in projection. Negatively stained specimens had structure well preserved to high resolution and on electron diffraction patterns one could often see out to the (0, 60), (1, 49),
FIG. 4. Electron diffraction pattern obtained from a negativeiy stained tropomgosin sheet using a camera length of 3.7 m. There is a clear series of POW lines spaced at 1/.5nm and each row line is sampled every li40nm. Note that the spots on row line 1 are staggered relative to the meridian and the second row line, indicating that the structure is centric. The bar represent,s 1/5nm.
(2, 46) and (3, 17) reflections, which correspond t,o resomtions of 1.33, 1.55: I.43 and 1.57 nm, respectiveiy. A broad meridional arc at 0.54 nm. typiea! of that seen in high-angle X-ray diffraction patterns of a-helical proteins (see Fraser & NacRae, 1973): was often seen with shorter camera, lengths and rarely a fine meridional arc at 0.15 nm could also be seen. However, these high-angle reflections were not sampled clearly by the cmm lattice. There was also some diffuse scattering on the equator at about 2.5 nm. Specimens negatively stained with many1 acetate were remarkably resistant to radiation damage and, even after recording several electron micrographs, generally showed electron diffraetion spots to about 1.5 nm, with the overall distribution of intensity little altered. Phases to a nominal resolution of 1.5 nm were obtained by Fourier transformation of 1024 x 2048 pixel areas of electron micrographs which had been digitized on an approximately 0.3 nm raster. These were combined with amplitudes derived from digitized electron diffraction patterns (using soft,ware kindly provided by Dr Richard Henderson) t’o produce, by Fourier inversion; the reconstructed projection of the structure shown in Figure 5. Rarely, sheets were observed which did not show c?nm symmetry but rather showed a series of striations at about 10” to t,he long axis of t,he sheet (Fig. 6). Generally; the areas showing this pattern were considerably smaller than the areas showing cmm symmetry and they were usua,lly found as part of a sheet which had the cmm pattern over most of its area. Optical diffraction patterns derived from t,hese areas (Fig” 6) appeared to have spots on the same la,ttice positions as the cmm sheets but the intensity distribution was different and, in particular, the
LETTERS
TO THE
EDITOR
235
FIG. 5, Reconstructed image of a tropomyosin sheet negatively stained with uranyl acetate, obtained by combining amplitude data derived from electron diffraction with phases obtained by computer processing of electron micrographs. The bar represents 50 nm.
FIG. 6. A rare micrograph showing a sheet in which only a single set of striations was present. Negatively stained with uranyl acetate. The bar represents 100 nm. Inset: optical diffraction pattern (note that this pattern is oriented to correspond to the micrograph and so the meridian is horizontal. Thus: the pattern is rotated by 90” compared to Figs 2 and 4). The most striking difference between this pattern and that shown in Figs 2 and 4 is the intensity of the reflections which would index as (1 i 3) and ( - 1, 3) on the emm lattice. In this pattern one of these is very strong (large arrow on right) and the other is weak (small arrow on left) whereas these reflections are of equal intensity in the centric patterns.
FIG. 7. Schematic representation of a probable molecular arrangement in tropomyosin crystalline sheets. Two sheets having p2 symmetry overlap to produce the cmm pattern commonly observed. The p2 unit cell is indicated on the lower single layer while the cvzm unit ccl! is shown where the two layers overlap. Individual tropomyosin molecules are represented by arrows to indicate their polarity. The pairing of adjacent rows in each single sheet is consistent with the pattern shown in Fig. 6. The a.ntiparallel pairs of molecules may coil around one another rather than lie parallel as shown here. To aid interpretation the vert.ical scale of this Figure has been expanded 3-fold.
centric symmet,ry was lacking. This was most clearly seen with the intensity of the reflections which would have indexed as (1, 3) and ( - 1, 3) on the cmm lattice. In the cent,ric pattern these were of equal intensity whereas in these sheets one reflection was virtually absent while the other was the strongest In the pattern. Generally, the degree of st’ructural preservation of these sheets was much inferior to the cmm sheets and there was little reliable information past about 2.5 nm, with most of the image power being concentrated in the (1, 3) reflection (indexed on cmm). Computed Fourier transforms were consistent with a.11the reflections being real and so the pattern was indexed as p2 with a = 40 nm; b = 5 nm and y = 80”. The patterns observed in the negatively stained sheets st,rongly suggested that the cmm pattern was produced by t,he superposit,ion of an even number (probably 2 in the simplest case) of p2 sheets having alternating hand. Figure 7 illustrates how this could come about although it is st’ressed that this should be considered as a highly schematic representation. Tropomyosin is about 41 nm long and 2 nm in diameter (Caspar et al., 1969) and so the 40 nm x 5 nm unit cell for a single p2 sheet would be consistent with the presence of two molecules aligned antiparallel along the a-axis. The relative axial position of the molecules was not immediately clear from the electron micrographs of single p2 sheets, but data from the cm.m sheets support the positions shown in Figure 7. By analogy with tropomyosin magnesium paracrystals (Ohtsuki; 1974; St.ewart & NcLachlan, 1976; Stewart,
LETTERS
TO THE EDITOR
237
1981) one would expect there to be white stain-excluding bands across the sheets where a local increa,se in protein density results from the overlap of ends of consecutive molecules. In the reconstructed image of the cmm sheets (Fig. 5) there was a pair of very closely spaced (about 2 nm) white bands every 40 nm, which indicated that the molecular ends were probably all located very close to the same position axially in these sheets. Furthermore, these lines were rows of white dots spaced 5 nm apart across the sheet. This would be consistent with the overlapping ends in the p2 sheets lying directly over one another in the composite cmm sheet as illustrated in Figure 7. A cmm structure constituted from two overlapping p2 sheets would also be consistent with the strong (1, 3) reflections in the cmm diffraction patterns (Figs 2 and 4) which would correspond to the strong (0, 1) reflections from the two p2 sheets. Unfortunately, it was not possible to decompose the cmm structure into its constituent sheets because the spots deriving from each sheet occupied the same positions in the Fourier transform. Tropomyosin paracrystals formed with barium or magnesium (Caspar et-al., 1969) have a distinctive banding pattern along their axis that derives from attachment of uranyl ions to carboxyl residues, which are arranged with a periodicity of about 193 residues through the tropomyosin sequence (Stewart & McLachlan, 1976), and which gives rise to a particularly strong 14th order in their optical and electron diffraction patterns. By comparison, the cmm pattern shows a very weak 14th order, although the 13th is quite strong. This may indicate that the negative zones in adjacent molecules are staggered in this structure whereas they are opposed in the barium and magnesium paracrystals. Alternatively, the highly charged spermine molecules might bind so strongly to the negatively charged residues as to inhibit, at least partially, the binding of uranyl ions. The p2 sheets showed very little periodicity along the molecules. The (0, 1) reflection was very strong compared to the (0, 2) in the p2 sheets and the corresponding (1, 3) reflection in the cmm patterns was stronger than the (2, 6). This produced the distinct alternating dark and light lines along the (1, 0) planes (Fig. 6) in images of the p2 sheets and the distinct striations at about 10” to the axis in the cmm patterns. While the (0, 1) p2 reflection and the corresponding (1, 3) reflections in the cmm patterns are not forbidden by the symmetry of the pattern, one would generally not expect them to be very strong if alternating rows of tropomyosin molecules were evenly spaced. Evenly spaced molecules should give rise to striations about 2 to 2.5 nm apart whereas the striations in both the p2 (Fig. 6) and cmm (Figs 3 and 5) sheets are separated by 5 nm. This suggests that the molecules were paired in some way so that the structure consisted of rows containing two closely packed molecules separated by more than the spacing between molecules in the row as illustrated in Figure 7. It could well be that the pair molecules in each row twist round each other but the resolution of the images of these sheets was not sufficient to decide this point unequivocally. Pairing of coiled coils may be a general feature of the assembly of this sort of molecule into higher structures, since analogous structural units of about 4 nm diameter (i.e. corresponding to a pair of coiled-coils) also appear to be formed by the coiled-coil tails of myosin molecules in the backbone of muscle thick filaments (Wray, 1979; Pepe et al., 1981; Stewartf et al., 1981).
“38
iV1. STEWART
I am most grateful to my colleagues, and in particular to Hugh Huxiey, Aaron Klug, Tonv” Crowther and Richard Henderson, for their assistance, comments and criticisms and to Pat Edwards for technical assistance. I am also indebted to the Department of Experimental Pathology, John Curtin School of Medical Research, Canberra, for providing facilities during the initial stage of this study. AMedical Research Council Laboratory of Molecular Biology Hills Road, Cambridge CB2 2&H, England
MURRAY
STEWART
Received 7 October 1983 REFEREXCES Caspar, D. L. D., Cohen, C. & Longley, W. (1969). J. Mol. Biol. 4X, 87-107. Cohen, C., Caspar, D. L. D., Johnson, J. P., Nauss, K., Margossian, S. S. & Parry, D. A. D. (1972). Cold Spring Harbor Symp. Quant. Biol. 37, 287-297. Fraser, R. D. B. & MacRae, T. P. (1973). Conformation in Fibrous Proteins, Academic Press, London. Lehrer, S. S. (1975). Proc. Nat. Acad. Sci., U.S.A. 72, 3327-3331. Longley, W. (1977). J. Mol. Biol. 115, 381-387. Ohtsuki, I. (1974). J. Biochem. (Tokyo), 57, 753-765. Pepe, F. A., $shton, F. T.; Dowben, P. & Stewart, M. (1981). J. Mol. Biol. 145, 421-440. Phillips, G. N., Lattman, E. E., Cummins, P., Lee, K. & Cohen, C. (1978). Xature (London), 278, 413-417. Phillips, G. N., Fillers, J. P. & Cohen, C. (1981). Biophys. J. 36; 485-500. St’ewart, M. (1975a). FEBS Letters, 53; 5-7. St’ewart, M. (1975b). Proc. Roy. Sot. ser. B, 190, 257-266. Stewart, M. (1981). J. Mol. Biol. 148, 411-425. Stewart, M. & McLachlan, A. D. (1976). J. Mol. Biol. 103, 251-269. Stewart, M., Ashton, F. T., Lieberson, R. & Pepe, F. A. (1981). J. Mol. Riol. 153, 381-392. Wray. J. S. (1979). Nature (London), 277, 37-40.
Edited
by M. F. Moody
(1984) 174, 231-238
Crystalline
Sheets of Tropomyosin
In the presence of spermine tropomyosin forms sheets having two-dimensional crystallinity and tactoids. The most common form of sheet has cmm symmetry with a = 80 nm and b = 5 nm. The structure of this sheet has been solved in projection to a nominal resolution of 15 nm by combining data from electron diffraction and electron microscopy. Analysis of this pattern and that of rarely observed sheets having p2 symmetry (a = 40 nm, b = 5 nm and y = SO”) indicated that the cmm structure was formed by superposition of two p2 sheets. The tropomyosin molecules in each p2 sheet were arranged in rows directed along the p2 (0, 1) lattice lines, with all the molecules in one row having the same polarity and lying antiparallel to the molecules in adjacent rows. These rows associated in pairs, possibly by the supercoiling of the molecules in one row about those in the neighbouring row.
Tropomyosin is a rod-shaped molecule which lies in the grooves of the actin helix in the thin filaments of muscle and many non-muscle systems. In conjunction with troponin, tropomyosin plays a key role in the regulation of contraction of skeletal and cardiac muscle, although its function in smooth muscle and nonmuscle systems is less clear. Muscle tropomyosin is about 41 nm long and is constructed from two parallel E-helical chains arranged, in register, in a coiled-coil configuration (Caspar et al., 1969; Cohen et al., 1972; Stewart, 1975a). A number of tropomyosin aggregates have been studied in order to obtain structural information and also to investigate the interactions of tropomyosin with itself and with other muscle proteins. Tactoids having one-dimensional order are formed with divalent metal cations (Caspar et al., 1969) and molecular positions have been established in magnesium paracrystals, which are the commonest form (Ohtsuki, 1974; Stewart, 19756,1981). Crystals in which the molecules are arranged in an open kite-like meshwork have been examined to about 2 nm resolution and molecmar positions established (Phillips et al., 1978,1981). A sheetlike crystal produced by CaCl, at pH 6 has also been reported but has not been analysed at high resolution (Longley, 1977). The present paper describes the production of thin sheet-like crystals of tropomyosin using spermine instead of metal ions as precipitant. Electron diffraction and electron microscopy indicates and show structure to better than 1.5 nm that these sheets have cmm symmetry resolution. Solutions of 1 to 5 g/l rabbit tropomyosin in 25 mM-Tris . HCl (pH 8) produced an immediate precipitate when treated with spermine up to about 200 m;M. No precipitate was observed at higher spermine concentrations, although one was immediately produced if the solution was diluted with 25 mw-Tris . HCl (pH 8) or with water. Negatively stained preparations of tropomyosin-spermine precipitates showed the presence of both tactoids and thin sheets. Generally the tactoids were more numerous than the sheets and often resembled those produced 0022-2836/84/0902314Pi
pjO3.00/0
231
0
1984 Academic
Press Inc.
(London)
Ltd.
-‘Y:!
&I. YTEWAK’I
by magnesium or barium (Caspar et al., 1969). Overall the spermine preparations showed greater polymorphism than that observed with divalent metal ions (Caspar et al., 1969). The same general distribution was seen with either cardiac ol-tropomyosin or with a mixture of CI- and p-tropomyosins derived from rabbit skeletal muscle, and no influence of the oxidation state of the tropomyosin thiol residues nor of temperature over the range 0 to 25°C was observed. Specimens containing a higher proport,ion of sheets were obtained by adding spermine to 10 KIM at 4”C, centrifuging the sample and allowing the supernatant’ to slowly come to room temperature. Figure 1 shows the general appearance of the sheets, which appeared to be extremely thin and were up to several micrometres across. Optical diffraction of electron micrographs (Fig. 2) indicated the presence of two-dimensional order and, at higher magnification, clear series of striations could be seen at an angle of about 10” to the axis of the sheet (Fig. 3). Low-angle electron diffraction patterns (Fig. 4) were easily obtained. The optical and electron diffraction patterns were characterized by a series of row lines spaced about li5nm apart which were finely sampled at every 1/40nm. The spots in the first row line were clearly staggered relative to those on the meridian and second row line, and the pattern indexed on an 80 nm x 5 nm rectangular unit cell. All reflections with (h + k) odd were absent,
FIG. 1. Low magnification micrograph of a thin crystaliine sheet of tropomy.osin formed with spermine negatively stained with many1 acetate. The bar represents 250 nm. Frozen rabbit hearts were used to prepare cc-tropomyosin as described by Stewart (1981). Skeletal tropomyosin (a mixture of a- and P-tropomyosins) was prepared from rabbit back and thigh muscles as described by Stewart & McLachlan (1976). Preparations were homogeneous as assessed by sodium dodecyl sulphate/ polyacrylamide gel electrophoresis. As required. the tropomyosin cysteine residues were oxidized with dithio bis-nitrobenzoate (Lehrer, 1975) or reduced with dithiothreitol, Specimens were applied to thinly (approx. 10 nm) carbon-coated 400 mesh grids, negatively stained with 1% (w/v) aqueous many1 acetate and examined at 80 kV in a Philips EM300, EM301 or EM400 eiectron microscope. Micrographs were calibrated using negativeiy stained tropomyosin magnesium pamcrystals. taking t.heir axial repeat as 39.5 nm (Caspar et al., 1969).
LETTERS
FIG. 2. Optical diffraction 1/dnm.
pattern
TO THE
EDITOR
from Fig. 1, showing 2-dimensional
233
order. The bar represents
FIG. 3. High magnification image of a uranyl acetate negatively stained sheet. In addition to the fine periodicity perpendicular to the axis, there are quite distinct striations at about lo” to the axis. These are most easily seen by viewing the micrograph obliquely from the side. The bar represents 100 nm.
indicating a centric space group, and the obvious mirror symmetry of the pattern about the meridian and equator established the space group as cmm in projection. Negatively stained specimens had structure well preserved to high resolution and on electron diffraction patterns one could often see out to the (0, 60), (1, 49),
FIG. 4. Electron diffraction pattern obtained from a negativeiy stained tropomgosin sheet using a camera length of 3.7 m. There is a clear series of POW lines spaced at 1/.5nm and each row line is sampled every li40nm. Note that the spots on row line 1 are staggered relative to the meridian and the second row line, indicating that the structure is centric. The bar represent,s 1/5nm.
(2, 46) and (3, 17) reflections, which correspond t,o resomtions of 1.33, 1.55: I.43 and 1.57 nm, respectiveiy. A broad meridional arc at 0.54 nm. typiea! of that seen in high-angle X-ray diffraction patterns of a-helical proteins (see Fraser & NacRae, 1973): was often seen with shorter camera, lengths and rarely a fine meridional arc at 0.15 nm could also be seen. However, these high-angle reflections were not sampled clearly by the cmm lattice. There was also some diffuse scattering on the equator at about 2.5 nm. Specimens negatively stained with many1 acetate were remarkably resistant to radiation damage and, even after recording several electron micrographs, generally showed electron diffraetion spots to about 1.5 nm, with the overall distribution of intensity little altered. Phases to a nominal resolution of 1.5 nm were obtained by Fourier transformation of 1024 x 2048 pixel areas of electron micrographs which had been digitized on an approximately 0.3 nm raster. These were combined with amplitudes derived from digitized electron diffraction patterns (using soft,ware kindly provided by Dr Richard Henderson) t’o produce, by Fourier inversion; the reconstructed projection of the structure shown in Figure 5. Rarely, sheets were observed which did not show c?nm symmetry but rather showed a series of striations at about 10” to t,he long axis of t,he sheet (Fig. 6). Generally; the areas showing this pattern were considerably smaller than the areas showing cmm symmetry and they were usua,lly found as part of a sheet which had the cmm pattern over most of its area. Optical diffraction patterns derived from t,hese areas (Fig” 6) appeared to have spots on the same la,ttice positions as the cmm sheets but the intensity distribution was different and, in particular, the
LETTERS
TO THE
EDITOR
235
FIG. 5, Reconstructed image of a tropomyosin sheet negatively stained with uranyl acetate, obtained by combining amplitude data derived from electron diffraction with phases obtained by computer processing of electron micrographs. The bar represents 50 nm.
FIG. 6. A rare micrograph showing a sheet in which only a single set of striations was present. Negatively stained with uranyl acetate. The bar represents 100 nm. Inset: optical diffraction pattern (note that this pattern is oriented to correspond to the micrograph and so the meridian is horizontal. Thus: the pattern is rotated by 90” compared to Figs 2 and 4). The most striking difference between this pattern and that shown in Figs 2 and 4 is the intensity of the reflections which would index as (1 i 3) and ( - 1, 3) on the emm lattice. In this pattern one of these is very strong (large arrow on right) and the other is weak (small arrow on left) whereas these reflections are of equal intensity in the centric patterns.
FIG. 7. Schematic representation of a probable molecular arrangement in tropomyosin crystalline sheets. Two sheets having p2 symmetry overlap to produce the cmm pattern commonly observed. The p2 unit cell is indicated on the lower single layer while the cvzm unit ccl! is shown where the two layers overlap. Individual tropomyosin molecules are represented by arrows to indicate their polarity. The pairing of adjacent rows in each single sheet is consistent with the pattern shown in Fig. 6. The a.ntiparallel pairs of molecules may coil around one another rather than lie parallel as shown here. To aid interpretation the vert.ical scale of this Figure has been expanded 3-fold.
centric symmet,ry was lacking. This was most clearly seen with the intensity of the reflections which would have indexed as (1, 3) and ( - 1, 3) on the cmm lattice. In the cent,ric pattern these were of equal intensity whereas in these sheets one reflection was virtually absent while the other was the strongest In the pattern. Generally, the degree of st’ructural preservation of these sheets was much inferior to the cmm sheets and there was little reliable information past about 2.5 nm, with most of the image power being concentrated in the (1, 3) reflection (indexed on cmm). Computed Fourier transforms were consistent with a.11the reflections being real and so the pattern was indexed as p2 with a = 40 nm; b = 5 nm and y = 80”. The patterns observed in the negatively stained sheets st,rongly suggested that the cmm pattern was produced by t,he superposit,ion of an even number (probably 2 in the simplest case) of p2 sheets having alternating hand. Figure 7 illustrates how this could come about although it is st’ressed that this should be considered as a highly schematic representation. Tropomyosin is about 41 nm long and 2 nm in diameter (Caspar et al., 1969) and so the 40 nm x 5 nm unit cell for a single p2 sheet would be consistent with the presence of two molecules aligned antiparallel along the a-axis. The relative axial position of the molecules was not immediately clear from the electron micrographs of single p2 sheets, but data from the cm.m sheets support the positions shown in Figure 7. By analogy with tropomyosin magnesium paracrystals (Ohtsuki; 1974; St.ewart & NcLachlan, 1976; Stewart,
LETTERS
TO THE EDITOR
237
1981) one would expect there to be white stain-excluding bands across the sheets where a local increa,se in protein density results from the overlap of ends of consecutive molecules. In the reconstructed image of the cmm sheets (Fig. 5) there was a pair of very closely spaced (about 2 nm) white bands every 40 nm, which indicated that the molecular ends were probably all located very close to the same position axially in these sheets. Furthermore, these lines were rows of white dots spaced 5 nm apart across the sheet. This would be consistent with the overlapping ends in the p2 sheets lying directly over one another in the composite cmm sheet as illustrated in Figure 7. A cmm structure constituted from two overlapping p2 sheets would also be consistent with the strong (1, 3) reflections in the cmm diffraction patterns (Figs 2 and 4) which would correspond to the strong (0, 1) reflections from the two p2 sheets. Unfortunately, it was not possible to decompose the cmm structure into its constituent sheets because the spots deriving from each sheet occupied the same positions in the Fourier transform. Tropomyosin paracrystals formed with barium or magnesium (Caspar et-al., 1969) have a distinctive banding pattern along their axis that derives from attachment of uranyl ions to carboxyl residues, which are arranged with a periodicity of about 193 residues through the tropomyosin sequence (Stewart & McLachlan, 1976), and which gives rise to a particularly strong 14th order in their optical and electron diffraction patterns. By comparison, the cmm pattern shows a very weak 14th order, although the 13th is quite strong. This may indicate that the negative zones in adjacent molecules are staggered in this structure whereas they are opposed in the barium and magnesium paracrystals. Alternatively, the highly charged spermine molecules might bind so strongly to the negatively charged residues as to inhibit, at least partially, the binding of uranyl ions. The p2 sheets showed very little periodicity along the molecules. The (0, 1) reflection was very strong compared to the (0, 2) in the p2 sheets and the corresponding (1, 3) reflection in the cmm patterns was stronger than the (2, 6). This produced the distinct alternating dark and light lines along the (1, 0) planes (Fig. 6) in images of the p2 sheets and the distinct striations at about 10” to the axis in the cmm patterns. While the (0, 1) p2 reflection and the corresponding (1, 3) reflections in the cmm patterns are not forbidden by the symmetry of the pattern, one would generally not expect them to be very strong if alternating rows of tropomyosin molecules were evenly spaced. Evenly spaced molecules should give rise to striations about 2 to 2.5 nm apart whereas the striations in both the p2 (Fig. 6) and cmm (Figs 3 and 5) sheets are separated by 5 nm. This suggests that the molecules were paired in some way so that the structure consisted of rows containing two closely packed molecules separated by more than the spacing between molecules in the row as illustrated in Figure 7. It could well be that the pair molecules in each row twist round each other but the resolution of the images of these sheets was not sufficient to decide this point unequivocally. Pairing of coiled coils may be a general feature of the assembly of this sort of molecule into higher structures, since analogous structural units of about 4 nm diameter (i.e. corresponding to a pair of coiled-coils) also appear to be formed by the coiled-coil tails of myosin molecules in the backbone of muscle thick filaments (Wray, 1979; Pepe et al., 1981; Stewartf et al., 1981).
“38
iV1. STEWART
I am most grateful to my colleagues, and in particular to Hugh Huxiey, Aaron Klug, Tonv” Crowther and Richard Henderson, for their assistance, comments and criticisms and to Pat Edwards for technical assistance. I am also indebted to the Department of Experimental Pathology, John Curtin School of Medical Research, Canberra, for providing facilities during the initial stage of this study. AMedical Research Council Laboratory of Molecular Biology Hills Road, Cambridge CB2 2&H, England
MURRAY
STEWART
Received 7 October 1983 REFEREXCES Caspar, D. L. D., Cohen, C. & Longley, W. (1969). J. Mol. Biol. 4X, 87-107. Cohen, C., Caspar, D. L. D., Johnson, J. P., Nauss, K., Margossian, S. S. & Parry, D. A. D. (1972). Cold Spring Harbor Symp. Quant. Biol. 37, 287-297. Fraser, R. D. B. & MacRae, T. P. (1973). Conformation in Fibrous Proteins, Academic Press, London. Lehrer, S. S. (1975). Proc. Nat. Acad. Sci., U.S.A. 72, 3327-3331. Longley, W. (1977). J. Mol. Biol. 115, 381-387. Ohtsuki, I. (1974). J. Biochem. (Tokyo), 57, 753-765. Pepe, F. A., $shton, F. T.; Dowben, P. & Stewart, M. (1981). J. Mol. Biol. 145, 421-440. Phillips, G. N., Lattman, E. E., Cummins, P., Lee, K. & Cohen, C. (1978). Xature (London), 278, 413-417. Phillips, G. N., Fillers, J. P. & Cohen, C. (1981). Biophys. J. 36; 485-500. St’ewart, M. (1975a). FEBS Letters, 53; 5-7. St’ewart, M. (1975b). Proc. Roy. Sot. ser. B, 190, 257-266. Stewart, M. (1981). J. Mol. Biol. 148, 411-425. Stewart, M. & McLachlan, A. D. (1976). J. Mol. Biol. 103, 251-269. Stewart, M., Ashton, F. T., Lieberson, R. & Pepe, F. A. (1981). J. Mol. Riol. 153, 381-392. Wray. J. S. (1979). Nature (London), 277, 37-40.
Edited
by M. F. Moody