- Email: [email protected]
Conformational switching in the flagellar filament of Salmonella typhimurium
J. Mol. Biol. (1992) 226, 447-454
Conformational
Switching in the Flagellar Filament of Salmonefla typhimurium Shlomo Trachtenbergt
Department of Membrane and Ultrastructure The Hebrew University-Hadassah Medical Jerusalem 91010, Israel
Research School
and David J. DeRosier Rosen&e1 Basic Medical Sciences Research Center Brandeis University, Waltham, MA 02254-9110, U.S.A. (Received 23 December 1991; accepted 17 March
1992)
The flagellar filament of the mutant Salmonella typhimurium strain SJW814 is straight, and has a right-handed twist like the filament of SJW1655. Three-dimensional reconstructions from electron micrographs of ice-embedded filaments reveal a flagellin subunit that has the same domain organization as that of SJW1655. Both show slight changes from the domain organization of the subunits from SJW1660, which possesses a straight, left-handed filament. This points to the possible roIe of changes in subunit conformation in the left-toright-handed structural transition in filaments. Comparison of the left and right-handed filaments shows that the subunit’s orientation and intersubunit bonding appear to change. The orientation of the subunit in the SJW814 filament is intermediate between that of SJFVl655 and SJW1660. Its intermediate orientation may explain why the filaments of S,J\Yl655 and SJW1660 are locked in one conformation, whereas the filament of SJW814 (aan be induced to switch by, for example, changes in pH and ionic strength. Keywords: cryoelectron microscopy; bacterial flagella; helical three-dimensionai reconstruction; bacterial motility; switchable protein -
1. Introduction
protofilament is 0.5 A (1 A = 0.1 nm) longer (Trachtenberg & DeRosier, 1991). The native filament contains a mixture of R and L states, but within each protofilament all subunits are in the same state. The coexistence of protofilaments in different states in the same filament produces stress, which is relieved by deformation of the straight filament into a superhelix (Calladine, 1975, 1976, 1978). In comparing the structures of an R with an L filament, we found a change in subunit packing and in subunit conformation that we supposed resulted from the switch of R to L (Trachtenberg & DeRosier, 1987, 1991). We obtained three-dimensional reconstruction of filaments from a pair of Salmonella typhimurium mutants (SJW1655 and SJWl660) derived from the same parent strain (SJW1103 serotype i). These mutants are locked in the straight R (SJW1655) and L (SJW1660) conformations and cannot be induced to supercoil, even
The filament, the propeller of the rotating bacterial flagellum, is a self-assembling helical structure made of one protein species, tlagellin. The subunits in the filament are organized into 11 protofilament,s. To function as a propeller, the filament must exist as a rigid superhelix. It can do so because the flagellin monomers are switchable and can coexist in two stable conformational states (Asakura, 1970). If all monomers are in one of the two conformational states, e.g. the R state, the filament is st.raight and the protofilaments have a right-handed twist. If all subunits are in the other conformational state (L), the filament is also straight, but the twist of the protofilaments is lefthanded and the intersubunit spacing along the t Author to whom all correspondence should be ad(Irrxsrd 447
0 1992 Academic Press Limited
S. Trachtenberg
448
and D. J. DeRosier
at low pH and/or high ionic strength (Hyman & Trachtenberg, 1991). Here we explore the generality of our structural results by analyzing the structure of a second, straight, right-handed S. typhimurium filament from strain SJW814 serotype 1,2. The symmetry of the SJW814 filament (O’Brien L%Bennett, 1972) is similar to that of the right-handed filament from SJW1655 but the filament of SJW814, unlike that of SJW1655 and SJW1660, can be induced to supercoil at 05 M-citric acid (pH 40). Shirakihara & Wakabayashi (1979) published a three-dimensional reconstruction of these filaments, but their maps seemed to lack some of the features found in those of our frozen-hydrated preparations. We have therefore carried out a study on these filaments embedded in ice.
2. Materials and Methods (a) Bacterial strains, growing conditim.s and Jilament preparation
Bacterial strains were kindly provided by R. Kamiya. Cells were grown in @S% nutrient broth. Filaments were detached from the cells by vortex mixing and purified by differential centrifugation as described by Trachtenberg & DeRosier (1987). (b) Electron microscopy Electron microscopy was carried out in the European Molecular Biology Laboratory (EMBL), Heidelberg. The microscope used was a Philips EM 400 equipped with an anticontaminator, as described by Homo et al. (1984), a low-dose kit and a Gatan cryoholder and transfer unit. Images were recorded on Kodak SO-163 film developed for 11 min in full-strength D-19 developer. Specimen preparation, image processing (DeRosier & Moore, 1970) and statistical analysis were as described by Trachtenberg 6 DeRosier (1987).
3. Results (a) Filament
symmetry
An image of a vitrified filament and its diffraction pattern are shown in Figure 1. The n = 11, -5 and 6 layer-lines dominate the diffraction pattern. The intensity of the n = 6 layer-lines is weaker than the n = -5 layer-line. This is typical of the righthanded filaments.
Figure 1. (a) An 814 filament preserved in vitreous ice. The outer flagellin domains produced striations along the 5, 6 and 11-start directions in the filament. Bar represents 1000 A. (b) A computed diffraction pattern in which the ?E= 0, - 5,6 and 11 layer-lines are marked. Note that the n = 6 is weaker than the n = 5 layer-line, a characteristic of right-handed filaments.
(c) Averaging
that the j2amen.t
is right-handed
Table 1 shows the average phase differences, A6, between the symmetrical major peaks of each layerline. The differences should be 180” and 0” for layerlines of odd and even orders of n, respectively. Since layer-line n = -5 (identified because A8 1: 180’) lies below the n = 6 layer-line (A0 1: 0), the lowest layer-line must be n = + 11 (i.e. right-handed) and not n = - 11 (left-handed) (see Trachtenberg t DeRosier (1991) for a complete discussion of this point).
filaments
Table 2 lists the phase residuals between the individual filament sides and their average. Three rounds of averaging were carried out. The phase differences are between about 30” and 50”, indicating good agreement between particles. The radial scaling is listed as well. Averaged phases and amplitudes of G,,, are shown in Figure 2(a) and the amplitudes of gn,, in Figure 2(b). (G,,, and gn,, are defined by Klug et al. (1958).)
Table 1 Verijkation
(b) Confirming
the layer data from diflerent
that the Jilament is right-handed n=+ll
n=-5
n=6
00025 188+24
00176 183f20
00201 5+6
An unequivocal way to fix the handedness of flagellar filaments is to determine the axial positions of the n = -5 and 12= 6 layerlines. The difference in phase (IA@ between major layer-line peaks symmetrical about the meridian should be 0” for even Uessel orders and 180” for odd Be-1 orders. In the R form, the odd layer-line (n = - 5) is closer to the equator than the even one (n = 6), as opposed to the L form in which these layer-lines switch places.
Conformation&
449
Switching
Table 2 Agreement between individual Jilaments and their average Filament identification 04F.N 04F.F 04E.N 04E.F 15Al.N 15Al.F 15Bl.N 15Bl.F IfiD2.N 15D2.F 17B.N 17B.F 17C.N 17C.F 17E.N 17E.F
Phase residual 41.1 424 41.0 41.6 353 380 47.6 509 385 337 52.2 460 48.4 47.2 424 463
Radial scaling 1.0 1.0 1.0 1.0 1.0 1.0 0.95 O-95 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
814 .. .. 1655 *-+
35*0030.0025*0020*00I5*00I o*oo5.00o*oo-
The phase residual (R,) is a weighted average of the squared differences in phase and a measure of the degree to which an individual filament represents a common structure (Amos & Klug, 1975). Residuals of 30 to 50” indicate good agreement. The radial scaling factor is used to adjust the radial position of reflections along layer-lines relative to the average.
(d) The average density map and its reliability A transverse section through the density map is shown in Figure 3(a) and (b). The lowest contour level was chosen to ensure the connectivity between density peaks in the full map. Figure 3(c) and (d) are “reliability” maps (Trachtenberg & DeRosier, 1987) corresponding to 95 y. and 99% confidence
-- -I8.00I 6.00l4-ooI 2.00I o-008*006*00-
.- -_ 35.00 30.00 25.00 20.00
r--
..__... ..” n=6 .._,___,_,:
15.00 IO*00 5.00 o-00 i
0 Radius
(8)
(b)
-360”
a-1 (0)
Figure 2. (a) Averaged layer-line data extracted from diffraction patterns similar to those shown in Fig. 1. The continuous line corresponds to relative amplitude and the dotted line is the phase plotted against reciprocal space radius. The scale of the phase is from 0 to 360”. (b) Radial density functions for each layer-line. The Figure is a composite of data from filaments of 3 S. typhimurium strains: a left-handed filament from SJW1660 () and 2 right-handed ones from SJW1655 (- - -) and SJW814 (.....). Note that the peaks’ amplitudes and radial positions are similar, but there are differences. The differences reflect the conformational changes in the 3 different filaments.
levels of Student’s t-test, respectively. The white region corresponds to significant peaks (protein) and the black to significant holes (ice). Note that features in the maps are significant even in the
S. Trachtenberg and D. J. DeRosier
Figure 3. (‘ross-sections through the 3-dimensional maps of the filament. (a) h contour map showing the positive peaks corresponding to protein. (b) A gray-scale map. I’roteiu is white and ice is black. (c) and (d) Reliability maps reI)resenting the results of Student’s t-tests for each individual pixel using 95O;, and 9Y0,, confidence levels. resprctivrly. White regions correspond to significant positive peaks. Black regions correspond to significant nrgatire peaks. The features that are not significantly different from the background density are gray. Sotr that there are significant featurrs even at small radii (r = 25 A). Bar represents 100 L%.
central parts (at radii of 25 A). A representation of the radially averaged signal, variance and signal-tonoise (S/N) ratio is shown in Figure 4(a), (b) and (c), respectively. Three-dimensional representations of the density map are shown in Figure 5.
4. Discussion (a) Surface
view: solid models
The solid models of the SJW814 (this work), SJW1655 (R) and SJW1660 (L) (Trachtenberg & DeRosier, 1991) filaments in Figure 5 display the domain organization of the subunits. The bottom
segment (300 A long) shows the outer surfacae of the model; the middle segment is the same but with itIs front. half removed; and the short (150 A) segment is the same but with the outer flagellin domains removed to expose the middle domain at a radius of 62 A. In the SJWl660 filament, the middle domains lie roughly along the 1 l-start line. This connectivity is absent in filaments from both SJW814 and SJWl655. Rather, in the latt,er two, the middle domains lie roughly along the five-start lines, although the inclination of this domain in t,he SJW814 filament is intermediate between that, of SJW1655 and that of SJW1660. The orientation of the outer domain in the SJW814 filament, is also
Conformational
(a) Signal
1 (b) Variance
147 Radius (8)
Fi gure 5. Three-dimensional
Switching
451
c) Signolhoise
Figure 4. (a) Radial dependence of the signal: the averaged density (p) of each pixel was squared and radially averaged. The square root is plotted as a function of radius. The signal, ((p2)) lf2. is lower at) the (*enter than at. the outside of t.he particle. (b) Radial dependence of the noise: the variance (kV2) was also radially averaged and its plotted as a function of radius. square root ((X2))“’ (c) Radial dependence of t,he signal-to-noise ratio: the ratios of the results from (a) and (b) are plotted as a function of radius.
representations of the filaments and their interiors. The Figure consists of 3 columns: the r part represents a surface presentation of a filament segment 300 A long. The middle part is a representation of t,he of the filament The top part is a filam lent sameA segment. but wit)h the front half removed, thus exposing the interior sepn rent. 150 A long, from which the features outside a radius of 62 A were removed. The center column represents t’hr The central panel shows the fihn rent of SJW1660 with those of SJW814 and SJW1656 at its left and right. respectively. Note that no strong connectivity along the I1 -start lines (i.e. vertically) is seen in the filamc :nts intet ?or of the filaments. ,JW814 and SJW16.55, whereas it is seen in that of SJW1660. The inner subunit domains are now exposed. In of SC the filaments are in the form of small knobs that connect along the 5-start lines. A st.rong feai >UW S,JW ‘814 and SJWl655, emei rging from the top segments of the Figure is the difference in inclination of the middle domain, seen behind the orJter knot IS. In the filament, of SJW1660 it almost coincides with the 1 l-start protofilaments; in that of SJW1655 it coinc ides wit,h the R-start lines. In that of SJW814 it follows the 5-start, lines; however, it does not coincide with them but forn ?S a st.ag.grred line. IOWR
452
S. Trachtenberg and D. J. DeRosier
(a
intermediate between that of SJW1655, which points radially out, and that of SJWl660, which is slewed to the right. The inner domains in the SJW814 filament connect along the five-start direction as do those of SJW1655. In the right-handed forms connectivity along the five-start direction dominates throughout. (b) The central channel
(b
The central hole in the filament is the putative channel through which flagellin subunits are exported to the growing tip of the filament. Namba et al. (1989) found a very large channel of about 69 A in diameter, whereas in our studies we found a much smaller channel of about 30 Bi in diameter (Fig. 3). One possibility is that the channel in our map is filled with subunits in the act of transporting, whereas that of Namba et al. (1989) is free of subunits. This is a plausible explanation because Namba et al. (1989) studied reconstituted filaments, whereas we used native filaments. The matter we find filling the channel has the symmetry of the filament. This is evident in the plots of g.,, (Fig. 2(b)), which show peaks at small radii, as well as in the reliability maps (Fig. 3(c), (d)), which show fingers of density projecting inward to small radii. It is possible that transported subunits would stick preferentially to sites in the channel, thus mimicking aspects of the filament’s symmetry. (c) Subunit rotations
(c
Figure 6. Subunit connectivity. The data in Fig. 5 have been reduced here to 1 cylindrical section at a radius of 62 A corresponding to the outer surface of the shell in the upper panel. The width of each panel in the Figure corresponds to the perimeter, p, of the filament at r = 61.81 A (p = 388.2 A). The height is 32497 A. The cylindrical section is unrolled into a flat sheet in the Figure. The 5 and 11-start lattice lines are indicated in circled numbers and marked by broken lines. Dotted horizontal and vertical lines mark the center of arbitrarily chosen subunits. Continuous lines mark the axis of the middle domain of the subunit. The inclination of this axis
The symmetry of the SJW814 filament is similar to that of SJW1655 and the pitches of the helical families are identical. The most prominent difference in the radial density g.,, for the n = 11 layerline is a strong inner peak in the SJW1660 diffraction pattern, which is weak in the curve corresponding to SJW1655 but equally strong in that of SJW814. On the other hand, the n = 1 layer-line is
off the vertical line (z axis) or rotation about a radial line (perpendicular to the panel) is indicated by $ (SJW814 section), Cp(SJW1655 section) and w (SJW1669 section). Note that the subunits’ middle domains in the SJW814 and SJW 1655 maps are aligned more or less along the B-start lattice line, but their orientation, as indicated by $ and 4, is different. In these right-handed filaments, there is no connectivity along the 11-start latttice lines, but there is connectivity along the 1l-start lines and not along the 5-start lines in SJW1669 filaments. Each rectangular
subunit domain can be identified by 4 binding sites marked clockwise A, B, C, D, starting from the upper right-hand corner. The connectivity in (b) is through sites B = D and in (a) through sites A = B. In (c), which can be used as a reference, the connectivity is through sites A 5 C. The R-L switch (C G A to B = A) requires a large rotation (depicted by the upper arch in (c)) in the SJW1669 protofilament. The same change in the SJWl669SJW814 filaments (C = A to B = D) requires a smaller rotation and is depicted by the lower arch in (c).
Conformational
453
Switching
Figure 7. A stereo pair showing the superposition of the SJW814 subunit on the SJW1655 subunit. The 2 subunits in the right-handed filaments of SJW814 (orange) and SJW165.5 (pink) were aligned. The main peaks align quite well, althoigh the innermost peaks do not superpose exactly.
weak in the diffraction pattern of both right-handed filaments. These changes in the transforms and
radial densities reveal themselves in differences in the shapes of the subunits and in the orientation and bonding of the subunits. Oyfindrical
sections
(Fig. 6) at the radius. of the
middle domain demonstrate tation and clearer way. the middle vertical (the
connectivity
the changes in orien-
between
In the SJWl660
subunits
in
a
map the long axis of
domain (continuous line) lies - 9” off direction of the filament axis). In that
of SJWl655 the axis of the middle domain (cont.inuous line) lies along the five-start helical line (broken line); that is $J = -54” off vertical. In that of SJWS14, the middle domain’s axis (continuous line) does not coincide with the five-start helical direction
(broken
line);
rather,
it
is inclined
by
$ = - 24” off vertical. The change in orientation changes the intersubunit, bonding. Four contact sites on each subunit are marked A, B, C and D in a clockwise direction, start.ing
from the upper
right-hand
corner.
One can
use the left-handed form SJW1660 as a reference for both right-handed forms. The subunits in the S,JWI660 map bind along the protofilaments by sites A to C. In the transition to a right-handed filament, the contact shifts to lie along the five-start direction in which either D interacts with B (the map of SJWS14) or A interacts with D (the map of SJWl655). In previous work, we detected a change in subunit shape in the I, to R transitions. In the SJW 1660 map, the subunit has a dog-leg shape but
has a straight shape in that of SJW1655 (see Figs 8 and 9 of Trachtenberg compared the putative
& DeRosier, 1991). We again subunit of the SJW814 map
with that of SJW1655 (see Fig. 7), and the subunit shape only slightly deviates from straight. Thus, it seems that the subunit conformation in the SJW814 map is again intermediate but is more like that of SJW1655 than SJWl660. Our conclusion is that, compared to the subunits of SJWl660, the subunits of SJW814 show the same binding of changes as those of -SJW1655 but to a lesser degree. Perhaps the intermediate nature of the SJW814 filament correlates with its ability to switch at low pH, whereas the filaments of SJW1655 and SJW 1660 are locked. We are grateful to R. Kamiya for providing the bacterial strains and to Louise Seidel, Beth Finkelstein and Phyllis Freedman for typing and editing the manuscript. This research was supported by grant 8700039 from the United States-Israel Binational Science Foundation (to S.T.) and grant GM35433 from the National Institutes of Health (to D.J.D.).
References Amos, L. A. & Klug, A. (1975). Three-dimensional image reconstructions of the contractile tail of T4 bacteriophage. J. Mol. Biol. 99, 51-73. Asakura, S. (1970). Polymerization of flagellin and polymorphism of flagella. Adwun. Biophys. 1, 99-155. Callsdine, C. R. (1975). Construction of bacterial flagella. Nature (London), 338, 514-518. Calladine. C. R. (1976). Design requirements for the construction of bacterial flagella. J. Theoret. Biol. 57, 469-489. Calladine, C. R. (1978). Change of waveform in bacterial flagella: the role of mechanics at the molecular level. J. Mol. Biol. 118, 457-479. DeRosier, D. J. & Moore, P. B. (1970). Reconstruction of three-dimensional images from electron micrographs
8. Trachtenberg and D. J. DeRosier
454 of structures with helical symmetry.
J. Mol. Biol. 52,
355-369.
Homo,
Klug.
220, 79-88.
A., Crick, F. H. C. t Wyckoff, H. W. (1958). Diffraction by helical structures. Acta Crystallogr. 11, 199-213. Namba, K., Yamashita, I. & Vonderviszt, F. (1989). Structure of the core and central channel of bacterial 342, 648-654. flagella. Nature (London),
O’Brien, E. J. 6 Bennett, P. M. (1972). Structure of straight flagella from a mutant Salmonella. J. Mol. Biol. 70, 133-152. Shirakihara, Y. & Wakabayashi, T. (1979). Threedimensional image reconstruction of straight flagella from a mutant Salmonella typhimurium. J. Mol. Rid. 131,485-507. Trachtenberg, S. & DeRosier, D. J. (1987). Threedimensional structure of the frozen-hydrated flagellar filament. The left-handed filament of Salmonella typhimurium. J. Mol. Biol. 195, 581-601. Trachtenberg, S. & DeRosier, D. J. (1991). A molecular switch: subunit rotations involved in the righthanded to left-handed transitions of Salmonella typhimurium flagellar filaments. J. Mol. Rid. 220. 67-77.
Edited by A. Klug
Conformational
Switching in the Flagellar Filament of Salmonefla typhimurium Shlomo Trachtenbergt
Department of Membrane and Ultrastructure The Hebrew University-Hadassah Medical Jerusalem 91010, Israel
Research School
and David J. DeRosier Rosen&e1 Basic Medical Sciences Research Center Brandeis University, Waltham, MA 02254-9110, U.S.A. (Received 23 December 1991; accepted 17 March
1992)
The flagellar filament of the mutant Salmonella typhimurium strain SJW814 is straight, and has a right-handed twist like the filament of SJW1655. Three-dimensional reconstructions from electron micrographs of ice-embedded filaments reveal a flagellin subunit that has the same domain organization as that of SJW1655. Both show slight changes from the domain organization of the subunits from SJW1660, which possesses a straight, left-handed filament. This points to the possible roIe of changes in subunit conformation in the left-toright-handed structural transition in filaments. Comparison of the left and right-handed filaments shows that the subunit’s orientation and intersubunit bonding appear to change. The orientation of the subunit in the SJW814 filament is intermediate between that of SJFVl655 and SJW1660. Its intermediate orientation may explain why the filaments of S,J\Yl655 and SJW1660 are locked in one conformation, whereas the filament of SJW814 (aan be induced to switch by, for example, changes in pH and ionic strength. Keywords: cryoelectron microscopy; bacterial flagella; helical three-dimensionai reconstruction; bacterial motility; switchable protein -
1. Introduction
protofilament is 0.5 A (1 A = 0.1 nm) longer (Trachtenberg & DeRosier, 1991). The native filament contains a mixture of R and L states, but within each protofilament all subunits are in the same state. The coexistence of protofilaments in different states in the same filament produces stress, which is relieved by deformation of the straight filament into a superhelix (Calladine, 1975, 1976, 1978). In comparing the structures of an R with an L filament, we found a change in subunit packing and in subunit conformation that we supposed resulted from the switch of R to L (Trachtenberg & DeRosier, 1987, 1991). We obtained three-dimensional reconstruction of filaments from a pair of Salmonella typhimurium mutants (SJW1655 and SJWl660) derived from the same parent strain (SJW1103 serotype i). These mutants are locked in the straight R (SJW1655) and L (SJW1660) conformations and cannot be induced to supercoil, even
The filament, the propeller of the rotating bacterial flagellum, is a self-assembling helical structure made of one protein species, tlagellin. The subunits in the filament are organized into 11 protofilament,s. To function as a propeller, the filament must exist as a rigid superhelix. It can do so because the flagellin monomers are switchable and can coexist in two stable conformational states (Asakura, 1970). If all monomers are in one of the two conformational states, e.g. the R state, the filament is st.raight and the protofilaments have a right-handed twist. If all subunits are in the other conformational state (L), the filament is also straight, but the twist of the protofilaments is lefthanded and the intersubunit spacing along the t Author to whom all correspondence should be ad(Irrxsrd 447
0 1992 Academic Press Limited
S. Trachtenberg
448
and D. J. DeRosier
at low pH and/or high ionic strength (Hyman & Trachtenberg, 1991). Here we explore the generality of our structural results by analyzing the structure of a second, straight, right-handed S. typhimurium filament from strain SJW814 serotype 1,2. The symmetry of the SJW814 filament (O’Brien L%Bennett, 1972) is similar to that of the right-handed filament from SJW1655 but the filament of SJW814, unlike that of SJW1655 and SJW1660, can be induced to supercoil at 05 M-citric acid (pH 40). Shirakihara & Wakabayashi (1979) published a three-dimensional reconstruction of these filaments, but their maps seemed to lack some of the features found in those of our frozen-hydrated preparations. We have therefore carried out a study on these filaments embedded in ice.
2. Materials and Methods (a) Bacterial strains, growing conditim.s and Jilament preparation
Bacterial strains were kindly provided by R. Kamiya. Cells were grown in @S% nutrient broth. Filaments were detached from the cells by vortex mixing and purified by differential centrifugation as described by Trachtenberg & DeRosier (1987). (b) Electron microscopy Electron microscopy was carried out in the European Molecular Biology Laboratory (EMBL), Heidelberg. The microscope used was a Philips EM 400 equipped with an anticontaminator, as described by Homo et al. (1984), a low-dose kit and a Gatan cryoholder and transfer unit. Images were recorded on Kodak SO-163 film developed for 11 min in full-strength D-19 developer. Specimen preparation, image processing (DeRosier & Moore, 1970) and statistical analysis were as described by Trachtenberg 6 DeRosier (1987).
3. Results (a) Filament
symmetry
An image of a vitrified filament and its diffraction pattern are shown in Figure 1. The n = 11, -5 and 6 layer-lines dominate the diffraction pattern. The intensity of the n = 6 layer-lines is weaker than the n = -5 layer-line. This is typical of the righthanded filaments.
Figure 1. (a) An 814 filament preserved in vitreous ice. The outer flagellin domains produced striations along the 5, 6 and 11-start directions in the filament. Bar represents 1000 A. (b) A computed diffraction pattern in which the ?E= 0, - 5,6 and 11 layer-lines are marked. Note that the n = 6 is weaker than the n = 5 layer-line, a characteristic of right-handed filaments.
(c) Averaging
that the j2amen.t
is right-handed
Table 1 shows the average phase differences, A6, between the symmetrical major peaks of each layerline. The differences should be 180” and 0” for layerlines of odd and even orders of n, respectively. Since layer-line n = -5 (identified because A8 1: 180’) lies below the n = 6 layer-line (A0 1: 0), the lowest layer-line must be n = + 11 (i.e. right-handed) and not n = - 11 (left-handed) (see Trachtenberg t DeRosier (1991) for a complete discussion of this point).
filaments
Table 2 lists the phase residuals between the individual filament sides and their average. Three rounds of averaging were carried out. The phase differences are between about 30” and 50”, indicating good agreement between particles. The radial scaling is listed as well. Averaged phases and amplitudes of G,,, are shown in Figure 2(a) and the amplitudes of gn,, in Figure 2(b). (G,,, and gn,, are defined by Klug et al. (1958).)
Table 1 Verijkation
(b) Confirming
the layer data from diflerent
that the Jilament is right-handed n=+ll
n=-5
n=6
00025 188+24
00176 183f20
00201 5+6
An unequivocal way to fix the handedness of flagellar filaments is to determine the axial positions of the n = -5 and 12= 6 layerlines. The difference in phase (IA@ between major layer-line peaks symmetrical about the meridian should be 0” for even Uessel orders and 180” for odd Be-1 orders. In the R form, the odd layer-line (n = - 5) is closer to the equator than the even one (n = 6), as opposed to the L form in which these layer-lines switch places.
Conformation&
449
Switching
Table 2 Agreement between individual Jilaments and their average Filament identification 04F.N 04F.F 04E.N 04E.F 15Al.N 15Al.F 15Bl.N 15Bl.F IfiD2.N 15D2.F 17B.N 17B.F 17C.N 17C.F 17E.N 17E.F
Phase residual 41.1 424 41.0 41.6 353 380 47.6 509 385 337 52.2 460 48.4 47.2 424 463
Radial scaling 1.0 1.0 1.0 1.0 1.0 1.0 0.95 O-95 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
814 .. .. 1655 *-+
35*0030.0025*0020*00I5*00I o*oo5.00o*oo-
The phase residual (R,) is a weighted average of the squared differences in phase and a measure of the degree to which an individual filament represents a common structure (Amos & Klug, 1975). Residuals of 30 to 50” indicate good agreement. The radial scaling factor is used to adjust the radial position of reflections along layer-lines relative to the average.
(d) The average density map and its reliability A transverse section through the density map is shown in Figure 3(a) and (b). The lowest contour level was chosen to ensure the connectivity between density peaks in the full map. Figure 3(c) and (d) are “reliability” maps (Trachtenberg & DeRosier, 1987) corresponding to 95 y. and 99% confidence
-- -I8.00I 6.00l4-ooI 2.00I o-008*006*00-
.- -_ 35.00 30.00 25.00 20.00
r--
..__... ..” n=6 .._,___,_,:
15.00 IO*00 5.00 o-00 i
0 Radius
(8)
(b)
-360”
a-1 (0)
Figure 2. (a) Averaged layer-line data extracted from diffraction patterns similar to those shown in Fig. 1. The continuous line corresponds to relative amplitude and the dotted line is the phase plotted against reciprocal space radius. The scale of the phase is from 0 to 360”. (b) Radial density functions for each layer-line. The Figure is a composite of data from filaments of 3 S. typhimurium strains: a left-handed filament from SJW1660 () and 2 right-handed ones from SJW1655 (- - -) and SJW814 (.....). Note that the peaks’ amplitudes and radial positions are similar, but there are differences. The differences reflect the conformational changes in the 3 different filaments.
levels of Student’s t-test, respectively. The white region corresponds to significant peaks (protein) and the black to significant holes (ice). Note that features in the maps are significant even in the
S. Trachtenberg and D. J. DeRosier
Figure 3. (‘ross-sections through the 3-dimensional maps of the filament. (a) h contour map showing the positive peaks corresponding to protein. (b) A gray-scale map. I’roteiu is white and ice is black. (c) and (d) Reliability maps reI)resenting the results of Student’s t-tests for each individual pixel using 95O;, and 9Y0,, confidence levels. resprctivrly. White regions correspond to significant positive peaks. Black regions correspond to significant nrgatire peaks. The features that are not significantly different from the background density are gray. Sotr that there are significant featurrs even at small radii (r = 25 A). Bar represents 100 L%.
central parts (at radii of 25 A). A representation of the radially averaged signal, variance and signal-tonoise (S/N) ratio is shown in Figure 4(a), (b) and (c), respectively. Three-dimensional representations of the density map are shown in Figure 5.
4. Discussion (a) Surface
view: solid models
The solid models of the SJW814 (this work), SJW1655 (R) and SJW1660 (L) (Trachtenberg & DeRosier, 1991) filaments in Figure 5 display the domain organization of the subunits. The bottom
segment (300 A long) shows the outer surfacae of the model; the middle segment is the same but with itIs front. half removed; and the short (150 A) segment is the same but with the outer flagellin domains removed to expose the middle domain at a radius of 62 A. In the SJWl660 filament, the middle domains lie roughly along the 1 l-start line. This connectivity is absent in filaments from both SJW814 and SJWl655. Rather, in the latt,er two, the middle domains lie roughly along the five-start lines, although the inclination of this domain in t,he SJW814 filament is intermediate between that, of SJW1655 and that of SJW1660. The orientation of the outer domain in the SJW814 filament, is also
Conformational
(a) Signal
1 (b) Variance
147 Radius (8)
Fi gure 5. Three-dimensional
Switching
451
c) Signolhoise
Figure 4. (a) Radial dependence of the signal: the averaged density (p) of each pixel was squared and radially averaged. The square root is plotted as a function of radius. The signal, ((p2)) lf2. is lower at) the (*enter than at. the outside of t.he particle. (b) Radial dependence of the noise: the variance (kV2) was also radially averaged and its plotted as a function of radius. square root ((X2))“’ (c) Radial dependence of t,he signal-to-noise ratio: the ratios of the results from (a) and (b) are plotted as a function of radius.
representations of the filaments and their interiors. The Figure consists of 3 columns: the r part represents a surface presentation of a filament segment 300 A long. The middle part is a representation of t,he of the filament The top part is a filam lent sameA segment. but wit)h the front half removed, thus exposing the interior sepn rent. 150 A long, from which the features outside a radius of 62 A were removed. The center column represents t’hr The central panel shows the fihn rent of SJW1660 with those of SJW814 and SJW1656 at its left and right. respectively. Note that no strong connectivity along the I1 -start lines (i.e. vertically) is seen in the filamc :nts intet ?or of the filaments. ,JW814 and SJW16.55, whereas it is seen in that of SJW1660. The inner subunit domains are now exposed. In of SC the filaments are in the form of small knobs that connect along the 5-start lines. A st.rong feai >UW S,JW ‘814 and SJWl655, emei rging from the top segments of the Figure is the difference in inclination of the middle domain, seen behind the orJter knot IS. In the filament, of SJW1660 it almost coincides with the 1 l-start protofilaments; in that of SJW1655 it coinc ides wit,h the R-start lines. In that of SJW814 it follows the 5-start, lines; however, it does not coincide with them but forn ?S a st.ag.grred line. IOWR
452
S. Trachtenberg and D. J. DeRosier
(a
intermediate between that of SJW1655, which points radially out, and that of SJWl660, which is slewed to the right. The inner domains in the SJW814 filament connect along the five-start direction as do those of SJW1655. In the right-handed forms connectivity along the five-start direction dominates throughout. (b) The central channel
(b
The central hole in the filament is the putative channel through which flagellin subunits are exported to the growing tip of the filament. Namba et al. (1989) found a very large channel of about 69 A in diameter, whereas in our studies we found a much smaller channel of about 30 Bi in diameter (Fig. 3). One possibility is that the channel in our map is filled with subunits in the act of transporting, whereas that of Namba et al. (1989) is free of subunits. This is a plausible explanation because Namba et al. (1989) studied reconstituted filaments, whereas we used native filaments. The matter we find filling the channel has the symmetry of the filament. This is evident in the plots of g.,, (Fig. 2(b)), which show peaks at small radii, as well as in the reliability maps (Fig. 3(c), (d)), which show fingers of density projecting inward to small radii. It is possible that transported subunits would stick preferentially to sites in the channel, thus mimicking aspects of the filament’s symmetry. (c) Subunit rotations
(c
Figure 6. Subunit connectivity. The data in Fig. 5 have been reduced here to 1 cylindrical section at a radius of 62 A corresponding to the outer surface of the shell in the upper panel. The width of each panel in the Figure corresponds to the perimeter, p, of the filament at r = 61.81 A (p = 388.2 A). The height is 32497 A. The cylindrical section is unrolled into a flat sheet in the Figure. The 5 and 11-start lattice lines are indicated in circled numbers and marked by broken lines. Dotted horizontal and vertical lines mark the center of arbitrarily chosen subunits. Continuous lines mark the axis of the middle domain of the subunit. The inclination of this axis
The symmetry of the SJW814 filament is similar to that of SJW1655 and the pitches of the helical families are identical. The most prominent difference in the radial density g.,, for the n = 11 layerline is a strong inner peak in the SJW1660 diffraction pattern, which is weak in the curve corresponding to SJW1655 but equally strong in that of SJW814. On the other hand, the n = 1 layer-line is
off the vertical line (z axis) or rotation about a radial line (perpendicular to the panel) is indicated by $ (SJW814 section), Cp(SJW1655 section) and w (SJW1669 section). Note that the subunits’ middle domains in the SJW814 and SJW 1655 maps are aligned more or less along the B-start lattice line, but their orientation, as indicated by $ and 4, is different. In these right-handed filaments, there is no connectivity along the 11-start latttice lines, but there is connectivity along the 1l-start lines and not along the 5-start lines in SJW1669 filaments. Each rectangular
subunit domain can be identified by 4 binding sites marked clockwise A, B, C, D, starting from the upper right-hand corner. The connectivity in (b) is through sites B = D and in (a) through sites A = B. In (c), which can be used as a reference, the connectivity is through sites A 5 C. The R-L switch (C G A to B = A) requires a large rotation (depicted by the upper arch in (c)) in the SJW1669 protofilament. The same change in the SJWl669SJW814 filaments (C = A to B = D) requires a smaller rotation and is depicted by the lower arch in (c).
Conformational
453
Switching
Figure 7. A stereo pair showing the superposition of the SJW814 subunit on the SJW1655 subunit. The 2 subunits in the right-handed filaments of SJW814 (orange) and SJW165.5 (pink) were aligned. The main peaks align quite well, althoigh the innermost peaks do not superpose exactly.
weak in the diffraction pattern of both right-handed filaments. These changes in the transforms and
radial densities reveal themselves in differences in the shapes of the subunits and in the orientation and bonding of the subunits. Oyfindrical
sections
(Fig. 6) at the radius. of the
middle domain demonstrate tation and clearer way. the middle vertical (the
connectivity
the changes in orien-
between
In the SJWl660
subunits
in
a
map the long axis of
domain (continuous line) lies - 9” off direction of the filament axis). In that
of SJWl655 the axis of the middle domain (cont.inuous line) lies along the five-start helical line (broken line); that is $J = -54” off vertical. In that of SJWS14, the middle domain’s axis (continuous line) does not coincide with the five-start helical direction
(broken
line);
rather,
it
is inclined
by
$ = - 24” off vertical. The change in orientation changes the intersubunit, bonding. Four contact sites on each subunit are marked A, B, C and D in a clockwise direction, start.ing
from the upper
right-hand
corner.
One can
use the left-handed form SJW1660 as a reference for both right-handed forms. The subunits in the S,JWI660 map bind along the protofilaments by sites A to C. In the transition to a right-handed filament, the contact shifts to lie along the five-start direction in which either D interacts with B (the map of SJWS14) or A interacts with D (the map of SJWl655). In previous work, we detected a change in subunit shape in the I, to R transitions. In the SJW 1660 map, the subunit has a dog-leg shape but
has a straight shape in that of SJW1655 (see Figs 8 and 9 of Trachtenberg compared the putative
& DeRosier, 1991). We again subunit of the SJW814 map
with that of SJW1655 (see Fig. 7), and the subunit shape only slightly deviates from straight. Thus, it seems that the subunit conformation in the SJW814 map is again intermediate but is more like that of SJW1655 than SJWl660. Our conclusion is that, compared to the subunits of SJWl660, the subunits of SJW814 show the same binding of changes as those of -SJW1655 but to a lesser degree. Perhaps the intermediate nature of the SJW814 filament correlates with its ability to switch at low pH, whereas the filaments of SJW1655 and SJW 1660 are locked. We are grateful to R. Kamiya for providing the bacterial strains and to Louise Seidel, Beth Finkelstein and Phyllis Freedman for typing and editing the manuscript. This research was supported by grant 8700039 from the United States-Israel Binational Science Foundation (to S.T.) and grant GM35433 from the National Institutes of Health (to D.J.D.).
References Amos, L. A. & Klug, A. (1975). Three-dimensional image reconstructions of the contractile tail of T4 bacteriophage. J. Mol. Biol. 99, 51-73. Asakura, S. (1970). Polymerization of flagellin and polymorphism of flagella. Adwun. Biophys. 1, 99-155. Callsdine, C. R. (1975). Construction of bacterial flagella. Nature (London), 338, 514-518. Calladine. C. R. (1976). Design requirements for the construction of bacterial flagella. J. Theoret. Biol. 57, 469-489. Calladine, C. R. (1978). Change of waveform in bacterial flagella: the role of mechanics at the molecular level. J. Mol. Biol. 118, 457-479. DeRosier, D. J. & Moore, P. B. (1970). Reconstruction of three-dimensional images from electron micrographs
8. Trachtenberg and D. J. DeRosier
454 of structures with helical symmetry.
J. Mol. Biol. 52,
355-369.
Homo,
Klug.
220, 79-88.
A., Crick, F. H. C. t Wyckoff, H. W. (1958). Diffraction by helical structures. Acta Crystallogr. 11, 199-213. Namba, K., Yamashita, I. & Vonderviszt, F. (1989). Structure of the core and central channel of bacterial 342, 648-654. flagella. Nature (London),
O’Brien, E. J. 6 Bennett, P. M. (1972). Structure of straight flagella from a mutant Salmonella. J. Mol. Biol. 70, 133-152. Shirakihara, Y. & Wakabayashi, T. (1979). Threedimensional image reconstruction of straight flagella from a mutant Salmonella typhimurium. J. Mol. Rid. 131,485-507. Trachtenberg, S. & DeRosier, D. J. (1987). Threedimensional structure of the frozen-hydrated flagellar filament. The left-handed filament of Salmonella typhimurium. J. Mol. Biol. 195, 581-601. Trachtenberg, S. & DeRosier, D. J. (1991). A molecular switch: subunit rotations involved in the righthanded to left-handed transitions of Salmonella typhimurium flagellar filaments. J. Mol. Rid. 220. 67-77.
Edited by A. Klug