- Email: [email protected]
The molecular structure of the silk fibers from Hymenoptera aculeata (bees, wasps, ants)
Journal of Structural Biology 192 (2015) 528–538
Contents lists available at ScienceDirect
Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi
The molecular structure of the silk fibers from Hymenoptera aculeata (bees, wasps, ants) R.D. Bruce Fraser a,1, David A.D. Parry a,b,⇑ a b
Institute of Fundamental Sciences, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand Riddet Institute, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand
a r t i c l e
i n f o
Article history: Received 31 July 2015 Received in revised form 9 October 2015 Accepted 24 October 2015 Available online 31 October 2015 Keywords: Honeybee silk Hornet silk Fiber diffraction Hymenoptera aculeata Coiled-coils Filaments
a b s t r a c t Silks from the Hymenoptera aculeata (bees, wasps, ants) contain ropes with four a-helical strands, rather than the more usual two strands found, for example, in a-keratin and myosin molecules. Extensive studies of the chemical structure of the silks have shown that each of the four chains in the molecule contains a central coiled-coil rod domain. However, little progress has been made in modeling the threedimensional structure. X-ray diffraction data on honeybee silk (Apis mellifera), recorded by Rudall and coworkers, has been re-examined in detail and possible structures developed for the various types of filament seen in the silk glands, and for the packing arrangement in the spun fibers. The original X-ray data were re-collected by scanning figures in the original publications, de-screening and averaging perpendicular to the direction of interest, thereby reducing the graininess of the original images. Sufficient numbers of equatorial and meridional reflections were collected to define the axial projection of the base of the unit cell in fibers drawn from the contents of the silk glands, and to suggest that the axial period is different from that suggested by Rudall and coworkers. Models for two types of filament of increasing diameter are developed based on the node–internode packing scheme observed in protein crystals containing four-strand a-helical ropes. The central domains of the four component chains in the molecule are enclosed by N- and C-terminal domains with widely different lengths and compositions. The fibers thus have a composite filament-matrix texture, and possible locations for the matrix are discussed. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction Crick (1953a, 1953b) showed that the X-ray diffraction patterns obtained from the k–m–e–f (keratin, myosin, epidermin, fibrinogen) group of a-fibrous proteins (Astbury and Woods, 1930, 1933) are consistent with a model in which the axes of the constituent a-helices are twisted around each other to form coiledcoil ropes. Subsequent studies on this class of structures in a diverse range of proteins, both fibrous and globular, have been summarized by Parry et al. (2008). The X-ray diffraction pattern predicted for a coiled-coil rope structure (Crick, 1953b) includes a meridional reflection at a spacing of 0.51 nm (rather than an off-meridional layer line of spacing 0.54 nm as expected of a simple a-helix), and a strong near-equatorial layer line (a feature not expected of an a-helix) with a spacing of |P|/ns, where |P| is the ⇑ Corresponding author at: Institute of Fundamental Sciences, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand. E-mail address: [email protected] (D.A.D. Parry). 1 Current address: 28 Satinay Drive, Noosa Parklands, Tewantin, Qld 4565, Australia. http://dx.doi.org/10.1016/j.jsb.2015.10.017 1047-8477/Ó 2015 Elsevier Inc. All rights reserved.
magnitude of the pitch length of the major helix and ns is the number of a-helical strands. Rudall (1962, 1965), Lucas and Rudall (1968), and Atkins (1967) studied the X-ray diffraction patterns of the silks of the Hymenoptera aculeata (bees, wasps and ants) and showed that they were typical of a-fibrous proteins, rather than the b-structures yielded by most other silks. The spacing of the near-equatorial layer line measured by Atkins (1967), however, was 3.5 nm, a value half that observed in other coiled-coil rope structures studied up until that time (Cohen and Holmes, 1963; Fraser et al., 1965; Elliott et al., 1968), typically about 7 nm. Sequence data are now available for the four constituent chains in honeybee silk (F1, F2, F3 and F4: Sutherland et al., 2006) as well as those from the four chains in a number of other aculeate silks (bumblebee, hornet, weaver ant, Australian bulldog ant and Indian jumping ant). These sequences, allied to structural and biochemical studies (Sutherland et al., 2006, 2007, 2011a,b, 2014; Campbell et al., 2014; Kameda et al., 2014a,b) afford us an opportunity to revisit the question of the molecular packing arrangement in aculeate silks. Each of the four chains has a central domain with a sequence characteristic of coiled-coil a-helical ropes. In contrast,
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
529
short sequences that lie N- and C-terminal to this region are predicted to contain very little regular secondary structure. A similar type of chain organization is found in the epidermal appendages of birds and reptiles and has been shown to lead to a filament-matrix composite (Filshie and Rogers, 1962; Fraser and Parry, 2014b). In this structure the filaments, consisting of an end-to-end aggregation of b-rich central domains, are embedded in a less-highly structured matrix composed of the remainder of the b-keratin sequences. Similarly, the two-stranded a-helical coiled-coil molecules in egg case silk from praying mantis (Bullough and Tulloch, 1990) are also characterized by an end-toend aggregation of the molecules. When fibers have a composite structure of this type it leads to a high tensile strength combined with flexibility. Consequently, the possibility that a similar type of structure also exists in aculeate silks has been examined here. In recent years proteins containing multi-stranded a-helical coiled-coil ropes have been crystallized and studied in detail by protein crystallography. These data indicate that two-strand ropes typically have a coiled-coil pitch length of about 14–15 nm, the three-strand ones about 17–18 nm, and the four-strand ones around 20 nm, though significant variations from these values have been found, especially on a local basis (Parry et al., 2008; Meier et al., 2009; Nicolet et al., 2010; Blocquel et al., 2014). In the present study we re-examine the original fiber diffraction data from honeybee silk and consider the quantitative fit between the main features of the observed high angle X-diffraction pattern and those predicted for filaments consisting of a single fourstrand rope, filaments consisting of 2 or 4 four-strand ropes, and also organized assemblies of filaments. Measurements of the low-angle diffraction pattern are used to derive a unit cell for the packing arrangement, enabling a detailed model to be developed for the molecular structure of honeybee silk.
then carried out using the formulae developed by Fraser and MacRae (1973, p. 29). The observed diffraction pattern is a sampling of a portion of a central section of the rotationally averaged intensity transform and, ideally, digitized scans would be used to map the data in reciprocal space using the techniques devised by Fraser et al. (1976), but insufficient information about the parameters is included in the original publications to be able to carry out this operation. The spacings reported in the present study are based on indirect calibrations using the reported period of the meridional reflections (Rudall, 1965) and the reported spacings of various features in the high-angle pattern (Atkins, 1967). As regards the calculated transforms the difference between film space and reciprocal space increases as R and Z increase but the errors over the ranges covered in the diagrams of the calculated intensity transform are minor. In these diagrams film space is matched to reciprocal space at the 0.51 nm meridional reflection and the result is adequate for visual comparison of the observed and calculated values. When better data become available conversion of the data from film space to reciprocal space can readily be accomplished using appropriate software, such as that available on the internet site http://fibernet.vanderbilt.edu/software.htm. Data for the scans of the meridian and the equator were obtained by selecting narrow strips parallel to the equator and to the meridian, and the optical density was averaged along lines perpendicular to the length of the strip. This procedure reduces the noise caused by the graininess in the film. Amino acid sequence comparisons were displayed using JalView (The Barton Group, University of Dundee, Scotland, UK). Calculations of volumes and molecular weights were carried out using the web facility http://www.basic.northwestern.edu/ biotools/proteincalc.html.
2. Methods
3. Results and discussion
X-ray diffraction patterns of fibers drawn from the silk glands of the honeybee (Apis mellifera) have been published by Rudall (1965) and by Lucas and Rudall (1968). The original publications were scanned at 600 dpi to obtain the data required for the present study and the results are shown in Fig. 1a and b respectively. It is worth noting that the preparation and handling of specimens of aculeate silks requires great patience and manual dexterity and the quality of the patterns is a tribute to the skill of the late Ken Rudall. In this work we use cylindrical polar coordinates (r, u, z) in real space and (R, w, Z) in reciprocal space, as illustrated in Fraser and MacRae (1973) chapter 1. The calculation of a coiled-coil diffraction pattern is a complex procedure (Crick, 1953a; Pardon, 1967) and the calculations in the present study were carried out using the simplified method described by Fraser et al. (1964) in which the coordinates of two turns of the distorted a-helix were generated using the formulae given by Crick (1953a). The intensity transform of the coiled-coil can then be calculated by setting this seven-residue unit as the repeating unit of a simple helix with a pitch equal to that of the coiled-coil. The intensity for a single coiled-coil is restricted to layer lines (m, n) that satisfy the condition:
3.1. Chain structure and packing
Z ¼ m=h þ n=P
ð1Þ
where Z is the reciprocal space coordinate parallel to the fiber axis, h is the unit height of the seven-residue repeating unit, n is the order of the Bessel function permitted on that layer line and P is the pitch of the major helix (Klug et al., 1958). The calculation of the intensity transform of the assemblies of coiled-coil ropes was
Sequence data are available for the four constituent chains in honeybee silk. These are termed F1, F2, F3 and F4 (Sutherland et al., 2006) and have 333, 309, 335 and 342 residues respectively (Sutherland et al., 2007). Each chain contains a 19-residue signal peptide at its N-terminus. It is believed, by analogy with observations on related silk proteins, that the signal peptides in the honeybee silk chains are removed through the action of a signal peptidase as the protein is secreted into the silk gland. However, direct experimental verification for such processing of the honeybee silk proteins is currently lacking. The sequence data, allied to other structural and biochemical studies undertaken by Sutherland and colleagues (Sutherland et al., 2007, 2011a,b, 2014; Campbell et al., 2014), provide an opportunity for us to understand the relationship between primary structure and the molecular packing arrangement in aculeate silks, as proposed here. The primary sequence data of honeybee silk can be used to predict the likely secondary structure of each of the four chains using various web-based methods (see, for example, Kelley and Sternberg, 2009; Källberg et al., 2012). These consistently indicate an a-helix content of 85 ± 5%. In almost all cases, the non-a-helical residues are found in short regions at either end of the chain. The sequence comparison of the four chains in honeybee silk and hornet silk (Fig. 2a) reveals a central domain with a heptad pattern (a–b–c–d–e–f–g)n in all four chains, where a and d are occupied predominantly by apolar residues (see also, for example, Sutherland et al., 2006, 2014). The heptad pattern, characteristic of a rod-like coiled-coil rope structure, is 217-residues long and has a length of approximately 32 nm, assuming a unit rise per
530
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Fig. 1. X-ray diffraction patterns obtained from drawn honeybee silk (Apis mellifera) by (a) Rudall (1965) and (b) Lucas and Rudall (1968). In (a) the position of the nearequatorial layer lines expected for a two-strand rope with a pitch of 14 nm is shown, as is the position of the layer line expected for a four-strand rope with the same pitch length. For a variety of reasons, but most specifically the fit of the model to the X-ray data, the four-strand structure was preferred over other possibilities (Atkins, 1967).
residue of 0.1485 nm. Note also the work of Sutherland et al. (2006, 2007) who have earlier proposed a similar rod domain (210 residues long) in honeybee silk. The corresponding segments of the four chains are F1, residues 33–249(d); F2, residues 53–269(d); F3, residues 58–274(d) and F4, residues 65–281(d), where the letter enclosed in parentheses refers to the position in the heptad repeat of the first residue listed and the residue numbers refer to the chains lacking their signal peptides. This region is dominated by an unusually high content of alanine residues (Fig. 2b, green). Several significant differences in the heptad pattern are seen for the honeybee silk sequences studied here (Sutherland et al., 2006, 2007, 2011a,b). In central domains of honeybee silk chains the a and d positions are predominantly occupied by alanine residues rather than the larger leucine, valine and isoleucine residues that are characteristic of most other coiled-coil rope structures. The b, c, e, f and g positions also contain some alanine residues in roughly equivalent amounts but less than that observed in the core a and d positions. In addition, short a-helix-favoring and heptadcontaining segments in each chain lie immediately N- and C-terminal to the 217-residue rod domain characterized by its continuous heptad substructure. The question arises as to the spatial location of these particular segments of chain. One possibility is that if they are located in line with the 217-residue central coiled-coil domain they could extend its length in the N-terminal and C-terminal directions by respectively, 22 and 14 residues for F1, 35 and 21 residues for F2, 13 and 10 residues for F3, and 25 and 0 residues for F4. Specifically, they are as follows – F1, residues 11–32(b) and 250–263(b); F2, residues 18–52(f) and 270–290(d); F3, residues 45–57(e) and 275–284(d) and F4 residues 40–64(d). The total region of recognizable heptad substructure would then be about 253, 273, 240 and 240 residues respectively for F1, F2, F3 and F4. The maximum lengths of a coiled coil rope conformation in these cases would thus be about 38, 41, 36 and 36 nm, respectively, assuming that the axial rise per residue in a coiled-coil conformation is 0.1485 nm. These ranges are effectively limited by the presence of proline residues and/or the lack of apolar residues in heptad positions a and d. If it is assumed that the length of the coiled-coil rod domain is the same in all four chains then this value must lie in the range 32–36 nm. The apolar contents of the combined a and d positions for the entire heptad-containing stretches of sequences are about 70%, 78%, 66% and 83% respectively. These
figures compare favorably with the value of approximately 70–75% seen in a large variety of other a-fibrous proteins (Conway and Parry, 1990). An important proviso to the above discussion is that the methods used to predict the a-conformation in the terminal domains are all based on data from hydrated protein crystals and the predicted content of regular secondary structure will almost certainly be an over-estimate for a dehydrated fiber. Another proviso regarding the prediction of heptad patterns is that in a protein containing a high proportion of hydrophobic residues the occurrence of short heptad patterns will occur by chance and have no significance as regards coiled-coil ropes. The numbers of potential interchain ionic interactions have been calculated for a wide variety of two-strand coiled-coil ropes (Conway and Parry, 1990). In every case these have peaked for a parallel, in-register chain arrangement. In the case of the four honeybee silk chains the numbers of interchain ionic interactions for a four-chain coiled-coil have been calculated for all chain combinations (homo- and hetero-tetramers), both parallel and antiparallel, and with a range of relative axial staggers. The analysis is indicative that inter-chain ionic interactions that specify both the relative chain stagger and orientation in most coiled-coil structures are few in number in the aculeate silks. It is near certain, however, that the chains will lie in axial register (thereby maximizing heptad overlap) and will lie parallel to one another, as found in all of the other a-fibrous proteins studied to date. Sutherland et al. (2014) have also noted that the hydrophobic wax molecules in which the honeybee silk normally exists in vivo may also disrupt the hydrophobic interactions of the core a and d residues that largely specify the coiled coil structure and some evidence has been reported that isolated a-helices may be the predominant structure in honeybee silk (Kameda and Tamada, 2009). It is therefore not unlikely that the stabilities of a coiled-coil conformation and a bundle of individual a-helices are comparable, with the former being the structure adopted in the more aqueous environment of the silk gland and the latter in the final material when in the presence of lipids (Sutherland et al., 2014). If it is assumed that only one type of molecule exists then six different cyclic chain orders between parallel chains are possible. These are (in clockwise order) F1–F2–F3–F4, F1–F2–F4–F3, F1–F3–F2–F4, F1–F3–F4–F2, F1–F4–F2–F3 and F1–F4–F3–F2. If one
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
531
Fig. 2. Homologies in the amino acid sequences of the four constituent chains F1, F2, F3, F4 in honeybee silk and V1, V2, V3 and V4 in hornet silk. The signal peptides are not included. Over a length of 217 residues (about 32 nm) in honeybee silk all four chains have a segment, marked in mauve, in which there is an unbroken repeating heptad motif characteristic of coiled-coil a-helix ropes, and the sequences have been aligned to bring the central domains in each chain into register. These residues are regarded as delineating a central rod domain having with a four-strand rope structure. (a) Percentage identity, (b) the distribution of alanine (green), lysine (blue) and glutamine (red).
allows the possibility of antiparallel chains this increases to 48 different combinations. The simplest arrangement, as well as the most likely, is that in which the central domains in all of the four chains are in lateral register and are similarly directed. Campbell et al. (2014) have shown that the bee silks are stabilized predominantly by intermolecular covalent crosslinking,
particularly through the isopeptide crosslink e-(c-glutamyl)-lysine produced by the transglutaminase enzymes. It is therefore of interest to study the disposition of the lysine (blue) and glutamine residues (red) in F1, F2, F3 and F4 (see Fig. 2b). Overall, the glutamines are few in number and percentage in the terminal domains (7 and 1.8% respectively) but more common in the rod domain (42 and
532
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Fig. 2 (continued)
4.9%). Lysine, in contrast, is more common in the terminal domains (41 and 10.8%) than in the rod domain (53 and 6.2%). Specifically, in the N, rod and C-terminal domains, respectively, there are 0, 9 and 1 (F1), 2, 12 and 1 (F2), 1, 13 and 2 (F3) and 0, 8 and 0 glutamine residues (F4). The corresponding figures for the lysine residues are 1, 16 and 8 (F1), 2, 12 and 5 (F2), 7, 13 and 4 (F3) and 6, 12 and 8 (F4) respectively. These data suggest that the glutamine end of the crosslink is far more likely to be found in the rod domain than in a terminal domain but the figures are less clear-cut for
the lysine end of the crosslink. The isopeptide crosslinks are likely to be particularly important in stabilizing the filaments axially but are likely also to provide lateral stability. 3.2. Selection of model parameters and derivation of possible filament structures Atkins (1967) measured the spacing of the near-equatorial layer line in honeybee silk and obtained a value of 3.5 ± 0.1 nm
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
corresponding to a pitch length of 14.0 ± 0.4 nm for a four-strand rope. The spacing of the prominent meridional reflection that characterizes the a-pattern, which represents the axial length of 3.5 residues, was measured as 0.506 ± 0.005 nm. Thus, the number of residues in a pitch length lies between 3.5 13.6/0.511 = 93.2 and 3.5 14.4/0.501 = 100.6. For both the illustrations and the transform calculations it is convenient, though not essential, for the number of residues in the pitch length to be divisible by seven, the number of residues in the repeating unit of the coiled coil, and a choice was made of 98 residues which implies a pitch length of 98 2 0.506/7 = 14.2 nm. Rudall (1965) pointed out that the most compact structure that could be obtained for a pair of two-strand ropes occurred when the bulges on the second rope were aligned with the hollows on the first rope, and this node–antinode type of packing has been found in both the egg case of the praying mantis (Bullough and Tulloch, 1990) and in a-keratin (Fraser and Parry, 2014a). This condition is attained when the second rope is rotated by p/2 relative to the first or displaced axially by P/4, where P is the pitch of the helix, or some equivalent combination of rotation and displacement. In the case of the four-strand ropes the values have to be p/4 or P/8 or some equivalent combination. An excellent illustration of this type of packing is seen in the crystal structure of the measles virus phosphoprotein multimerization domain (Blocquel et al., 2014), shown in Figs. 3 and 4. The packing is tetragonal in projection and the 2 four-strand ropes in the unit cell (colored blue) are aligned axially so that they are packed in a node–antinode manner as shown in Fig. 4. At any particular level in the region of overlap there is a combination of rotation and axial displacement equivalent to a rotation of p/4 between the orientations of the two molecules in the unit cell. Several types of filament of differing diameter have been reported to be present in the contents of the honeybee silk glands (Rudall, 1965; Flower and Kenchington, 1967) and possible models for these filaments are illustrated diagrammatically in Fig. 5. The smallest, shown in Fig. 5a, has a central domain with a fourstrand rope structure (length 217 residues) and a matrix component consisting of the N- and C-terminal domains distributed over the surface. The second, shown in Fig. 5b, has two such ropes in lateral register with a rotational phase difference of p/4 between them so that node–antinode packing is achieved as in Fig. 4. In the four-rope model (c) two two-rope filaments are combined in such a manner that all four ropes are related by a progressive rotation of p/4. There is no reason to suppose that the axes of the coiled coils in either the two-rope model or the four-rope model are straight and in both cases are likely to have a slow twist. Again, the N- and C-terminal domains form a coating on the surface in both (b) and (c). The filaments, around 4–5 nm in diameter, visible in electron micrographs, would be compatible with the four-rope model (c), and Flower and Kenchington noted that each of these appeared twinned, comprising elements 2.0–2.5 nm in diameter. This is compatible with a pair of two-rope filaments with node–internode packing, similar to that observed in Fig. 5b, combining to form a four-rope filament as shown diagrammatically in Fig. 5c.
533
Fig. 3. Axial projection of the four-strand ropes in the crystal structure of the measles virus phosphoprotein multimerization domain (Blocquel et al., 2014).
3.3. Information derived from the X-ray diffraction patterns Common features of X-ray diffraction patterns obtained from fibrous proteins are a low-angle equatorial pattern, where the innermost reflections can be directly related to the packing of the filaments (often quasi-hexagonal), and a series of sharp meridional reflections associated with a very precise repetition of a structural unit in the filament. The observed pattern can be regarded as the transform of the filament with a superposed external interference function determined by fibril packing. This latter
Fig. 4. Longitudinal relationship between the two molecules in the unit cell of the measles virus phosphoprotein multimerization domain (Blocquel et al., 2014) shown in cartoon view. The 2 four-strand ropes are aligned axially so that they are packed in a node–antinode manner (Rudall, 1956; Bullough and Tulloch, 1990).
534
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538 Table 1 Spacings of the observed equatorial reflections compared with those calculated for a two dimensional lattice with a = 4.52 nm, b = 3.74 nm and c = 101.1° corresponding to an axial projection of the unit cell.
Fig. 5. Cross-sections of models for the central domains in three types of filament that could be formed in an aqueous environment, for example in the silk glands. The small white circles represent a-helices. (a) A single four-strand rope, (b) a pair of four-strand ropes in which the two central domains have a phase shift of p/4 resulting in node/internode packing as in Fig. 4. The likelihood is that the filament will have a slow twist about its axis, (c) 4 four-strand ropes with a cyclic phase shift of p/4, again the likelihood is that the filament will have a slow twist. In all cases the N- and C-terminal domains form a surface coating.
function has maxima and minima close to the origin but tails off to unity with increasing R. In most cases, the observed intensity away from the meridian and equator contains no maxima associated with the external interference function and is determined solely by filaments. This indicates that when viewed down the fiber axis the projections of the filaments are well ordered but the axial shift and/or rotation between adjacent filaments is subject to cumulative variation which leads to a rapid increase in both band-width at half-height and rapid attenuation of the intensity (Fraser and MacRae, 1973, pp. 31–39). The absence of significant evidence for external interference remote from the equator and the meridian in honeybee silk is indicative of a texture with precisely ordered filaments and cumulative disorder in the rotation and/or z-shift between neighboring filaments. The fact that the equatorial pattern in honeybee silk has well developed equatorial reflections (Fig. 1) suggests that the cumulative disorder in projection is low whilst the cumulative disorder in rotation and/or z-shift is much greater. The terminal domains are presumed to have much less secondary structure but may still have a well-defined conformation that enables the lateral order to be largely maintained even in the presence cumulative disorder in the rotation and/or z-displacement of the filament. The spacings of the observed equatorial reflections, as determined in this work from a scan of the X-ray pattern in Lucas and Rudall (1968), are summarized in Table 1. They are consistent with an axial projection of the filament packing with a two-dimensional lattice with a = 4.52 nm, b = 3.74 nm and c = 101.1° (Table 1) and an area of 4.52 3.74 sin (101.1°) = 16.59 nm2. The volume of a cylinder with this base and a height equal to the mean axial rise per residue in the coiled coil (0.1485 nm) is 2.46 nm3. The mean residue weight in the central domain is 96.0 and assuming that the density is 1.45 g cm3 the volume per chain is 1.66 96.0/(1.45 1000) = 0.110 nm3. Thus the number of four-strand ropes that could be accommodated in a 0.1485 nm length of the unit cell is therefore estimated to be 2.46/ (4 0.110) = 5.59. The simplest interpretation of this would be that the asymmetric unit consists of a four-rope filament together with a matrix occupying the remaining volume of 2.46 16 0.110 = 0.70 nm3 as shown in Fig. 6. The spacings of six meridional reflections extracted from a digitized scan of the Lucas and Rudall X-ray pattern (1968) are given in Table 2. The implied possibilities for the value for the period are illustrated in Fig. 7a. These may be compared with those with derived from the seven meridional spacings measured by Kameda et al. (2014b) for hornet silk Fig. 7b. In each case, the plotted values refer to the root-mean-square deviations between the
h
k
Calc
Obs
0 0 1 1 1 1 1 2 2 2 2 2
1 2 2 1 0 1 2 2 1 0 1 2
3.74 1.87 1.86 3.20 4.52 2.64 1.62 1.60 2.12 2.26 1.79 1.32
3.74 1.84 1.84 – 4.52 2.64 1.61 1.61 2.07a 2.21a 1.72 1.34
Notes Overlaps [1, 2] Overlaps [0, 2] Not observed
Overlaps [2, 2] Overlaps [1, 2]
a Not resolved in the pattern obtained by Lucas and Rudall (1968) due to overexposure. The observed values refer to the pattern obtained by Rudall (1962) which was obtained with a smaller exposure.
Fig. 6. The projection of the base of the unit cell, determined from an analysis of the low-angle equatorial X-ray diffraction patterns of honeybee silk. The simplest model for the contents of the cell is a single four-rope filament and a layer of matrix formed from the terminal domains.
indices corresponding to the spacings of the observed meridional reflections and those of the nearest integer values. The trace in Fig. 7a is overlaid by noise acquired during the extraction of the data from the published photographs due to the short specimen-film distance and graininess in the film. With only six meridional reflections the results of statistical tests need to be treated as indicative rather than precise measures of goodness of fit. Two of the possible values in honeybee (33.1 and 43.0 nm) have counterparts in hornet, which is perhaps not surprising given the close similarities of their amino acid compositions of their central domains (Fig. 2). The first of these (33.1 nm.) is very close to the value used by Kameda et al., 2014b to index the meridional reflections in hornet silk so is a natural choice to check whether a repeat parallel to the fiber axis of this length is compatible with our suggestion of a cell containing 4 four-strand ropes. The combined molecular weights of the 4 four-strand ropes in a filament similar to that in Fig. 5c is M = 494,260 g mol1 and the
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
535
Table 2 Spacings of the observed meridional reflections compared with those calculated for an axial period of 33.1 nm. Index: Calculated (nm): Observed (nm): Observed index:
11 3.00 3.11 10.6
13 2.55 2.55 13.0
16 2.07 2.06 16.1
20 1.66 1.63 20.3
27 1.23 1.23 26.9
33 1.00 1.00 33.1
Meridional scan.
associated volume V, measured in nm3, is 0.00166M/q, where q is the mean density of the protein measured in g cm3. As the area of the cell base A is 16.59 nm2, the predicted value of c is V/A = 0.00166 494,260/(16.59 q). The value of q is not known and various algorithms have been used in attempts to predict its value (Quillin and Matthews, 2000; Fischer et al., 2004). The former authors suggest 1.43 ± 0.03 and 1.47 ± 0.03 g cm3 as possible ranges. Using these values the estimated value of c is 33.64– 34.58 nm. The former value is very close to one of the possible values (33.1 nm) in honeybee silk, and to the value of c derived for hornet silk by Kameda et al. (34.4 nm in Fig. 7b). Given the close similarity between the compositions of the central domains in the two materials this suggests that the minimum in Fig. 7a at 33.1 nm, is the appropriate choice for the period in honeybee silk. It will be seen from Table 2, however, that there is an unacceptable discrepancy between the spacing calculated for the 11th order of the axial repeat, and the observed spacing, which corresponds to an index of 10.6. A possible explanation for this is that the rod domains are packed head-to-head and tail-to-tail leading to a doubling of the period. Unfortunately, the number of reflections that can be extracted from the published photograph is insufficient to speculate further and this anomaly can only be resolved when data collected with modern high-resolution X-ray equipment become available. The number of residues in each chain in the central domain is estimated to be 217 and with a mean axial rise per residue of 0.1485 nm this will account for 32.2 nm of the period. The remaining distance of 0.9 nm, equivalent to about six residues in an ahelical conformation, must include residues from one or both of the terminal domains immediately proximal to the central domain. The filaments are thus envisaged as parcels of 4 four-strand ropes packed head-to tail with a small number of residues at the ends of the central domains bonding the parcels together. The above model, however, is only one of several possibilities. Its main merit lies in its simplicity, with four-rope filaments aggregating into sheets parallel to the b-axis (Fig. 6) and these sheets, in turn, being separated from one another by a layer of matrix formed from the distal parts of the terminal domains. When high
Fig. 7. Possible axial periods for (a) honeybee silk and (b) hornet silk derived by plotting the root-mean-square (RMS) deviation of the calculated indices from integral values. The trace in (a) is overlaid by noise acquired during the extraction of the data from the published photographs due to imperfect descreening and is exacerbated by the short specimen-film distance, which is inadequate for the precise measurements of low-angle reflections, due to graininess in the film.
resolution digital data become available they can be used to compare the intensity transform calculated for both this and alternative models, and then used as a basis for further refinement. An alternative choice for the period in honeybee silk is 30.9 nm (Fig. 7), which has the lowest RMS deviation. The problem, however, is that this would require a density of 1.56 g cm3 which is well outside the established range. 3.4. The high-angle intensity transform The calculated high-angle intensity transform for the three types of filaments shown in Fig. 5 are shown in a graphical form (Fig. 8), in which values are plotted above and below the layer line to enable an easier visual comparison with the observed pattern. It should be emphasized that this is not a simulated pattern, which would require many extra assumptions about the lengths of the rod-like portions, the orientation density function and the distribution functions for positional, axial, and rotational disorder. All of the predicted maxima appear in the observed patterns, as do subsidiary maxima in the profile of the (2, 0) meridional reflection in the multi-rope models. This arises from interference between the scattering from the component ropes. The spacings
536
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Fig. 8. Comparisons between the observed intensities in film space with the calculated distribution of intensity in reciprocal space for (a) a single four-strand rope, (b) a pair of four-stand ropes with node–internode packing, (c) 4 four-strand ropes packed tetragonally with node–internode packing, and (d) 4 four-strand ropes packed hexagonally. The latter may be appropriate to specimens from which water has been completely removed. The Z positions of the layer lines in the calculated intensity transform corresponding to the (m, n) values are shown as lines, and R is the radial coordinate as defined in Section 2. All of the predicted maxima appear in the observed patterns. However, subsidiary maxima in the meridional streak on layer line (2, 0) for multi-rope models are present (as noted by Atkins, 1967), but are not adequately accounted for. It is suggested here, however, that they arise either from an external interference function associated with the scattering from filaments in different cells, or with a slow twist in the filament that results in the axes of the four-strand ropes being inclined to the fiber axis. The spacings of the maxima of the observed peaks on the equator and the nearequatorial layer line are close to the values of the maxima in the predicted intensity transforms. The shapes of these reflections are appropriate to an infinite four-strand coiled coil rope of poly-L-alanine and will be smoothed in honeybee silk because of the limited length, variations in the local radius and pitch length in the coiled coils, and the incorporation of residues other than alanine.
Fig. 9. An extract from the X-ray pattern obtained from honeybee silk by Lucas and Rudall (1968), showing the satellites of the meridional reflection on layer line (2, 0). R is the radial coordinate as defined in Section 2. One possible explanation for the satellites is that they are produced by an external interference function associated with the a-axis of the projected unit cell (Fig. 6).
of the maxima of the observed peaks on the equator and the nearequatorial layer line are close to the values of the maxima in the
predicted intensity transforms (Fig. 8). The intensities calculated for the equatorial and near-equatorial layer lines exhibit pronounced maxima and minima, which will be smoothed out, to a large extent, for several reasons. For example, the calculations were carried out for poly-L-alanine coiled coil, assuming a constant radius, pitch and axial rise per residue, whereas studies of coiled coils in crystalline proteins show that substantial variations occur in these parameters (see, for example, Meier et al., 2009; Nicolet et al., 2010). In addition, the finite length of the coiled coils and the presence of non-alanine residues will produce further blurring. The unusual profile of the (2, 0) meridional reflection, noted by Atkins (1967), is better resolved in the X-ray pattern obtained by Lucas and Rudall (1968), and is reproduced in Fig. 9. The specimen has been tilted so that the center of the (2, 0) layer line intercepts the sphere of reflection and it will be seen that subsidiary maxima, characteristic of a lateral periodicity, are evident. The spacing of
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
the satellite maxima on the (2, 0) layer line is difficult to judge but appears to be around 4.5 nm which suggest that it may be related to the inter-sheet distance in the unit cell (Fig. 6). Another possibility is that the filaments have a slow twist that results in the axes of the four-strand ropes being inclined to the fiber axis. 3.5. Polymorphism It is to be expected that the dimensions of the unit cell and possibly also the packing arrangement of the molecules will vary with water content. As far as we have been able to determine, the X-ray data obtained by Kameda et al. (2014b), Rudall (1962) and Lucas and Rudall (1968) were from specimens with very low water content, whereas the data obtained by Rudall (1965) and Atkins (1967) were from hydrated specimens. The axial period and the dimensions of the axial projection of the base of the unit cell derived in the present study from the X-ray data of Lucas and Rudall (1968) are probably appropriate to the dehydrated material. 4. Summary The aim of the present study was to re-examine the original X-ray diffraction data from honeybee silk obtained by Rudall, Atkins and colleagues to see if a model for the molecular packing could be devised. Atkins (1967) had previously shown that much of the high-angle X-ray diffraction data could be explained by molecules having a four-strand coiled-coil conformation. Calculations presented here confirm this conclusion and, furthermore, show a close relationship between the entire high-angle X-ray pattern and the intensity transforms of the models that are described in this work. Evaluation of the low-angle X-ray diffraction data has now enabled us to determine the likely dimensions of the axially projected base of the unit cell and to construct a model for the filaments, each containing 4 four-strand ropes packed in a node–antinode manner, thereby maximizing packing density. In addition, a value for the axial period has been derived from the measured meridional spacings, which is consistent with a near common axial period in both honeybee and hornet silks. The latter, in turn, is compatible with a packing arrangement in which the central domains are arranged in a head-to-tail manner to form continuous filaments in a manner closely akin to that found with the coiled-coil proteins in the egg cases of praying mantises (Bullough and Tulloch, 1990), and with the b-crystallites in the b-keratins (Fraser and Parry, 2014b). The data as a whole suggest that, in common with a number of other natural fibers, the aculeate silks are composites with a filament-matrix texture. In the proposed model the four-rope filaments are tightly packed in one dimension i.e. in sheets. In turn, these sheets are separated from one another by a matrix composed of the bulk of the terminal domains. There is, therefore, a close parallel between the molecular structures of the two silks and the keratins of birds and reptiles. Both materials display a common central domain, though this is characterized by a high content of a-structure for the silks but b-structure for the avian and reptilian keratins. In both structures the central domains aggregate end-toend to give the fiber their high tensile strength. In honeybee silk the contribution of the isopeptide crosslinks to the tensile attributes of the fiber is also likely to be very significant. The terminal domains, with their wide range of amino acid compositions, form a matrix between the sheets of the filaments, thereby adding pliability to the structure. The present study highlights the need for further highresolution studies of the low-angle X-ray diffraction pattern, and electron microscope studies of longitudinal and transverse sections, using modern equipment, to resolve the uncertainties about the exact nature of the filaments and their packing.
537
Acknowledgments The authors wish to express their sincere thanks to Emeritus Professor E.D.T. (Ted) Atkins for unearthing and making available to us many of the X-ray diffraction patterns of honeybee silk and other members of the aculeate family taken by him some 50 years earlier. These have provided an invaluable context to the studies that we have undertaken. His interest and encouragement were also very much appreciated, as is that of Dr Tara Sutherland. Thanks are also due to Professor Martin Hazelton for his assistance in identifying possible periods in honeybee silk and assessing their significance. RDBF is indebted to Massey University for the award of an Honorary Research Fellowship.
References Astbury, W.T., Woods, H.J., 1930. The X-ray interpretation of the structure and elastic properties of hair keratin. Nature 126, 913–914. Astbury, W.T., Woods, H.J., 1933. X-ray studies on the structure of hair, wool and related fibres II: the molecular structure and elastic properties of hair keratin. Phil. Trans. R. Soc. Lond. A 232, 333–394. Atkins, E.D.T., 1967. A four-strand coiled-coil model for some insect fibrous proteins. J. Mol. Biol. 24, 139–141. Blocquel, D., Habchi, J., Durand, E., Sevajol, M., Ferron, F., Erales, J., Papageorgiou, N., Longhi, S., 2014. Coiled-coil deformations in crystal structures: the measles virus phosphoprotein multimerization domain as an illustrative example. Acta Crystallogr. D Biol. Crystallogr. D70, 1589–1603. Bullough, P.A., Tulloch, P.A., 1990. High-resolution spot-scan electron microscopy of microcrystals of an a-helical coiled-coil protein. J. Mol. Biol. 215, 161–173. Campbell, P.M., Trueman, H.E., Zhang, Q., Kojima, K., Kameda, T., Sutherland, T.D., 2014. Cross-linking in the silks of bees, ants and hornets. Insect Biochem. Mol. Biol. 48, 40–50. Cohen, C., Holmes, K.C., 1963. X-ray diffraction evidence for a-helical coiled-coils in native muscle. J. Mol. Biol. 6, 423–432. Conway, J.F., Parry, D.A.D., 1990. Structural features in the heptad substructure and longer range repeats of two-stranded a-fibrous proteins. Int. J. Biol. Macromol. 12, 328–334. Crick, F.H.C., 1953a. The Fourier transform of a coiled-coil. Acta Crystallogr. 6, 685– 689. Crick, F.H.C., 1953b. The packing of a-helices: simple coiled-coils. Acta Crystallogr. 6, 689–697. Elliott, A., Lowy, J., Parry, D.A.D., Vibert, P.J., 1968. Puzzle of the coiled coils in the aprotein paramyosin. Nature 218, 656–659. Filshie, B.K., Rogers, G.E., 1962. An electron microscope study of the fine structure of feather keratin. J. Cell Biol. 13, 1–12. Fischer, H., Polikarpov, I., Craievich, A.F., 2004. Average protein density is a molecular-weight-dependent function. Prot. Sci. 13, 2825–2828. Flower, N.E., Kenchington, W., 1967. Studies on insect fibrous proteins: the larval silk of Apis, Bombus and Vespa (Hymenoptera: Aculeata). J. Microsc. Soc. 86, 297–310. Fraser, R.D.B., MacRae, T.P., 1973. Conformation in Fibrous Proteins and Related Synthetic Polypeptides. Academic Press, New York and London. Fraser, R.D.B., Parry, D.A.D., 2014a. Keratin intermediate filaments: differences in the sequences of the Type I and Type II chains explain the origin of the stability of an enzyme-resistant four-chain fragment. J. Struct. Biol. 185, 317–326. Fraser, R.D.B., Parry, D.A.D., 2014b. Amino acid sequence homologies in the hard keratins of birds and reptiles, and their implications for molecular structure and physical properties. J. Struct. Biol. 188, 213–224. Fraser, R.D.B., MacRae, T.P., Miller, A., 1964. Fourier transform of the coiled-coil model of a-keratin. Acta Crystallogr. 17, 813–816. Fraser, R.D.B., MacRae, T.P., Miller, A., 1965. X-ray diffraction patterns of a-fibrous proteins. J. Mol. Biol. 14, 432–442. Fraser, R.D.B., MacRae, T.P., Miller, A., Rowlands, R.J., 1976. Digital processing of fibre diffraction patterns. J. Appl. Crystallogr. 9, 81–91. Källberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H., Xu, J., 2012. Templatebased protein structure modeling using the RaptorX web server. Nat. Protoc. 7, 1511–1522. Kameda, T., Tamada, Y., 2009. Variable-temperature 13C solid-state NMR study of the molecular structure of honeybee wax and silk. Int J. Biol. Macromol. 44, 64– 69. Kameda, T., Walker, A.A., Sutherland, T.D., 2014a. Evolution and application of coiled coil silk from insects. In: Asakura, T., Miller, T. (Eds.), Biotechnology of Silk. Springer, pp. 87–106. Kameda, T., Nemoto, T., Ogawa, T., Tosaka, M., Kurata, H., Schaper, A.K., 2014b. Evidence of a-helical coiled coils and b-sheets in hornet silk. J. Struct. Biol. 185, 303–308. Kelley, L.A., Sternberg, M.J.E., 2009. Protein structure prediction on the web: a case study using the Phyre server. Nat. Protoc. 4, 363–371. Klug, A., Crick, F.H.C., Wyckoff, H.W., 1958. Diffraction by helical structures. Acta Crystallogr. 11, 199–213.
538
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Lucas, F., Rudall, K.M., 1968. Extracellular fibrous proteins: the silks. In: Florkin, M., Stotz, E.H. (Eds.), Comprehensive Biochemistry, vol. 26B. Elsevier, Amsterdam, pp. 475–558. Meier, M., Padilla, G.P., Herrmann, H., Wedig, T., Hergt, M., Patel, T.R., Stetefeld, J., Aebi, U., Burkhard, P., 2009. Vimentin coil 1A – a molecular switch involved in the initiation of filament elongation. J. Mol. Biol. 390, 245–261. Nicolet, S., Herrmann, H., Aebi, U., Strelkov, S.V., 2010. Atomic structure of vimentin coil 2. J. Struct. Biol. 170, 369–376. Pardon, J.F., 1967. A modified theory for Fourier transform for a coiled-coil. Acta Crystallogr. 23, 937–939. Parry, D.A.D., Fraser, R.D.B., Squire, J.M., 2008. Fifty years of coiled-coils and ahelical bundles: a close relationship between sequence and structure. J. Struct. Biol. 163, 258–269. Quillin, M.L., Matthews, B.W., 2000. Accurate calculation of the density of proteins. Acta Crystallogr. D56, 791–794. Rudall, K.M., 1956. Protein ribbons and sheets. Lectures on the Scientific Basis of Medicine, vol. 5. Athlone Press, London, Ch 12, pp. 217–230. Rudall, K.M., 1962. Silk and other cocoon proteins. In: Florkin, M., Mason, H.S. (Eds.), Comparative Biochemistry, vol. 4. Academic Press, New York, pp. 397–433.
Rudall, K.M., 1965. Skeletal structure in insects. In: Aspects of Insect Biochemistry. Academic Press, London/New York, pp. 83–92. Sutherland, T.D., Campbell, P.M., Weisman, S., Trueman, H.E., Sriskantha, A., Wanjura, W.J., Haritos, V.S., 2006. A highly divergent gene cluster in honey bees encodes a novel silk family. Gen. Res. 16, 1414–1421. Sutherland, T.D., Weisman, S., Trueman, H.E., Sriskantha, A., Trueman, W.H., Haritos, V.S., 2007. Conservation of essential features in coiled coil silks. Mol. Biol. Evol. 24, 2424–2432. Sutherland, T.D., Church, J.S., Hu, X., Huson, M.G., Kaplan, D.L., Weisman, S., 2011a. Single honeybee silk protein mimics properties of multi-protein silk. PLoS ONE 6, e16489. http://dx.doi.org/10.1371/journal.pone.0016489. Sutherland, T.D., Weisman, S., Walker, A.A., Mudie, S.T., 2011b. The coiled coil silk of bees, ants and hornets. Biopolymers 97, 446–454. Sutherland, T.D., Trueman, H.E., Walker, A.A., Weisman, S., Campbell, P.M., Dong, Z., Huson, M.G., Woodhead, A.L., Church, J.S., 2014. Convergently-evolved structural anomalies in the coiled coil domains of insect silk proteins. J. Struct. Biol. 186, 402–411.
Contents lists available at ScienceDirect
Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi
The molecular structure of the silk fibers from Hymenoptera aculeata (bees, wasps, ants) R.D. Bruce Fraser a,1, David A.D. Parry a,b,⇑ a b
Institute of Fundamental Sciences, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand Riddet Institute, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand
a r t i c l e
i n f o
Article history: Received 31 July 2015 Received in revised form 9 October 2015 Accepted 24 October 2015 Available online 31 October 2015 Keywords: Honeybee silk Hornet silk Fiber diffraction Hymenoptera aculeata Coiled-coils Filaments
a b s t r a c t Silks from the Hymenoptera aculeata (bees, wasps, ants) contain ropes with four a-helical strands, rather than the more usual two strands found, for example, in a-keratin and myosin molecules. Extensive studies of the chemical structure of the silks have shown that each of the four chains in the molecule contains a central coiled-coil rod domain. However, little progress has been made in modeling the threedimensional structure. X-ray diffraction data on honeybee silk (Apis mellifera), recorded by Rudall and coworkers, has been re-examined in detail and possible structures developed for the various types of filament seen in the silk glands, and for the packing arrangement in the spun fibers. The original X-ray data were re-collected by scanning figures in the original publications, de-screening and averaging perpendicular to the direction of interest, thereby reducing the graininess of the original images. Sufficient numbers of equatorial and meridional reflections were collected to define the axial projection of the base of the unit cell in fibers drawn from the contents of the silk glands, and to suggest that the axial period is different from that suggested by Rudall and coworkers. Models for two types of filament of increasing diameter are developed based on the node–internode packing scheme observed in protein crystals containing four-strand a-helical ropes. The central domains of the four component chains in the molecule are enclosed by N- and C-terminal domains with widely different lengths and compositions. The fibers thus have a composite filament-matrix texture, and possible locations for the matrix are discussed. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction Crick (1953a, 1953b) showed that the X-ray diffraction patterns obtained from the k–m–e–f (keratin, myosin, epidermin, fibrinogen) group of a-fibrous proteins (Astbury and Woods, 1930, 1933) are consistent with a model in which the axes of the constituent a-helices are twisted around each other to form coiledcoil ropes. Subsequent studies on this class of structures in a diverse range of proteins, both fibrous and globular, have been summarized by Parry et al. (2008). The X-ray diffraction pattern predicted for a coiled-coil rope structure (Crick, 1953b) includes a meridional reflection at a spacing of 0.51 nm (rather than an off-meridional layer line of spacing 0.54 nm as expected of a simple a-helix), and a strong near-equatorial layer line (a feature not expected of an a-helix) with a spacing of |P|/ns, where |P| is the ⇑ Corresponding author at: Institute of Fundamental Sciences, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealand. E-mail address: [email protected] (D.A.D. Parry). 1 Current address: 28 Satinay Drive, Noosa Parklands, Tewantin, Qld 4565, Australia. http://dx.doi.org/10.1016/j.jsb.2015.10.017 1047-8477/Ó 2015 Elsevier Inc. All rights reserved.
magnitude of the pitch length of the major helix and ns is the number of a-helical strands. Rudall (1962, 1965), Lucas and Rudall (1968), and Atkins (1967) studied the X-ray diffraction patterns of the silks of the Hymenoptera aculeata (bees, wasps and ants) and showed that they were typical of a-fibrous proteins, rather than the b-structures yielded by most other silks. The spacing of the near-equatorial layer line measured by Atkins (1967), however, was 3.5 nm, a value half that observed in other coiled-coil rope structures studied up until that time (Cohen and Holmes, 1963; Fraser et al., 1965; Elliott et al., 1968), typically about 7 nm. Sequence data are now available for the four constituent chains in honeybee silk (F1, F2, F3 and F4: Sutherland et al., 2006) as well as those from the four chains in a number of other aculeate silks (bumblebee, hornet, weaver ant, Australian bulldog ant and Indian jumping ant). These sequences, allied to structural and biochemical studies (Sutherland et al., 2006, 2007, 2011a,b, 2014; Campbell et al., 2014; Kameda et al., 2014a,b) afford us an opportunity to revisit the question of the molecular packing arrangement in aculeate silks. Each of the four chains has a central domain with a sequence characteristic of coiled-coil a-helical ropes. In contrast,
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
529
short sequences that lie N- and C-terminal to this region are predicted to contain very little regular secondary structure. A similar type of chain organization is found in the epidermal appendages of birds and reptiles and has been shown to lead to a filament-matrix composite (Filshie and Rogers, 1962; Fraser and Parry, 2014b). In this structure the filaments, consisting of an end-to-end aggregation of b-rich central domains, are embedded in a less-highly structured matrix composed of the remainder of the b-keratin sequences. Similarly, the two-stranded a-helical coiled-coil molecules in egg case silk from praying mantis (Bullough and Tulloch, 1990) are also characterized by an end-toend aggregation of the molecules. When fibers have a composite structure of this type it leads to a high tensile strength combined with flexibility. Consequently, the possibility that a similar type of structure also exists in aculeate silks has been examined here. In recent years proteins containing multi-stranded a-helical coiled-coil ropes have been crystallized and studied in detail by protein crystallography. These data indicate that two-strand ropes typically have a coiled-coil pitch length of about 14–15 nm, the three-strand ones about 17–18 nm, and the four-strand ones around 20 nm, though significant variations from these values have been found, especially on a local basis (Parry et al., 2008; Meier et al., 2009; Nicolet et al., 2010; Blocquel et al., 2014). In the present study we re-examine the original fiber diffraction data from honeybee silk and consider the quantitative fit between the main features of the observed high angle X-diffraction pattern and those predicted for filaments consisting of a single fourstrand rope, filaments consisting of 2 or 4 four-strand ropes, and also organized assemblies of filaments. Measurements of the low-angle diffraction pattern are used to derive a unit cell for the packing arrangement, enabling a detailed model to be developed for the molecular structure of honeybee silk.
then carried out using the formulae developed by Fraser and MacRae (1973, p. 29). The observed diffraction pattern is a sampling of a portion of a central section of the rotationally averaged intensity transform and, ideally, digitized scans would be used to map the data in reciprocal space using the techniques devised by Fraser et al. (1976), but insufficient information about the parameters is included in the original publications to be able to carry out this operation. The spacings reported in the present study are based on indirect calibrations using the reported period of the meridional reflections (Rudall, 1965) and the reported spacings of various features in the high-angle pattern (Atkins, 1967). As regards the calculated transforms the difference between film space and reciprocal space increases as R and Z increase but the errors over the ranges covered in the diagrams of the calculated intensity transform are minor. In these diagrams film space is matched to reciprocal space at the 0.51 nm meridional reflection and the result is adequate for visual comparison of the observed and calculated values. When better data become available conversion of the data from film space to reciprocal space can readily be accomplished using appropriate software, such as that available on the internet site http://fibernet.vanderbilt.edu/software.htm. Data for the scans of the meridian and the equator were obtained by selecting narrow strips parallel to the equator and to the meridian, and the optical density was averaged along lines perpendicular to the length of the strip. This procedure reduces the noise caused by the graininess in the film. Amino acid sequence comparisons were displayed using JalView (The Barton Group, University of Dundee, Scotland, UK). Calculations of volumes and molecular weights were carried out using the web facility http://www.basic.northwestern.edu/ biotools/proteincalc.html.
2. Methods
3. Results and discussion
X-ray diffraction patterns of fibers drawn from the silk glands of the honeybee (Apis mellifera) have been published by Rudall (1965) and by Lucas and Rudall (1968). The original publications were scanned at 600 dpi to obtain the data required for the present study and the results are shown in Fig. 1a and b respectively. It is worth noting that the preparation and handling of specimens of aculeate silks requires great patience and manual dexterity and the quality of the patterns is a tribute to the skill of the late Ken Rudall. In this work we use cylindrical polar coordinates (r, u, z) in real space and (R, w, Z) in reciprocal space, as illustrated in Fraser and MacRae (1973) chapter 1. The calculation of a coiled-coil diffraction pattern is a complex procedure (Crick, 1953a; Pardon, 1967) and the calculations in the present study were carried out using the simplified method described by Fraser et al. (1964) in which the coordinates of two turns of the distorted a-helix were generated using the formulae given by Crick (1953a). The intensity transform of the coiled-coil can then be calculated by setting this seven-residue unit as the repeating unit of a simple helix with a pitch equal to that of the coiled-coil. The intensity for a single coiled-coil is restricted to layer lines (m, n) that satisfy the condition:
3.1. Chain structure and packing
Z ¼ m=h þ n=P
ð1Þ
where Z is the reciprocal space coordinate parallel to the fiber axis, h is the unit height of the seven-residue repeating unit, n is the order of the Bessel function permitted on that layer line and P is the pitch of the major helix (Klug et al., 1958). The calculation of the intensity transform of the assemblies of coiled-coil ropes was
Sequence data are available for the four constituent chains in honeybee silk. These are termed F1, F2, F3 and F4 (Sutherland et al., 2006) and have 333, 309, 335 and 342 residues respectively (Sutherland et al., 2007). Each chain contains a 19-residue signal peptide at its N-terminus. It is believed, by analogy with observations on related silk proteins, that the signal peptides in the honeybee silk chains are removed through the action of a signal peptidase as the protein is secreted into the silk gland. However, direct experimental verification for such processing of the honeybee silk proteins is currently lacking. The sequence data, allied to other structural and biochemical studies undertaken by Sutherland and colleagues (Sutherland et al., 2007, 2011a,b, 2014; Campbell et al., 2014), provide an opportunity for us to understand the relationship between primary structure and the molecular packing arrangement in aculeate silks, as proposed here. The primary sequence data of honeybee silk can be used to predict the likely secondary structure of each of the four chains using various web-based methods (see, for example, Kelley and Sternberg, 2009; Källberg et al., 2012). These consistently indicate an a-helix content of 85 ± 5%. In almost all cases, the non-a-helical residues are found in short regions at either end of the chain. The sequence comparison of the four chains in honeybee silk and hornet silk (Fig. 2a) reveals a central domain with a heptad pattern (a–b–c–d–e–f–g)n in all four chains, where a and d are occupied predominantly by apolar residues (see also, for example, Sutherland et al., 2006, 2014). The heptad pattern, characteristic of a rod-like coiled-coil rope structure, is 217-residues long and has a length of approximately 32 nm, assuming a unit rise per
530
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Fig. 1. X-ray diffraction patterns obtained from drawn honeybee silk (Apis mellifera) by (a) Rudall (1965) and (b) Lucas and Rudall (1968). In (a) the position of the nearequatorial layer lines expected for a two-strand rope with a pitch of 14 nm is shown, as is the position of the layer line expected for a four-strand rope with the same pitch length. For a variety of reasons, but most specifically the fit of the model to the X-ray data, the four-strand structure was preferred over other possibilities (Atkins, 1967).
residue of 0.1485 nm. Note also the work of Sutherland et al. (2006, 2007) who have earlier proposed a similar rod domain (210 residues long) in honeybee silk. The corresponding segments of the four chains are F1, residues 33–249(d); F2, residues 53–269(d); F3, residues 58–274(d) and F4, residues 65–281(d), where the letter enclosed in parentheses refers to the position in the heptad repeat of the first residue listed and the residue numbers refer to the chains lacking their signal peptides. This region is dominated by an unusually high content of alanine residues (Fig. 2b, green). Several significant differences in the heptad pattern are seen for the honeybee silk sequences studied here (Sutherland et al., 2006, 2007, 2011a,b). In central domains of honeybee silk chains the a and d positions are predominantly occupied by alanine residues rather than the larger leucine, valine and isoleucine residues that are characteristic of most other coiled-coil rope structures. The b, c, e, f and g positions also contain some alanine residues in roughly equivalent amounts but less than that observed in the core a and d positions. In addition, short a-helix-favoring and heptadcontaining segments in each chain lie immediately N- and C-terminal to the 217-residue rod domain characterized by its continuous heptad substructure. The question arises as to the spatial location of these particular segments of chain. One possibility is that if they are located in line with the 217-residue central coiled-coil domain they could extend its length in the N-terminal and C-terminal directions by respectively, 22 and 14 residues for F1, 35 and 21 residues for F2, 13 and 10 residues for F3, and 25 and 0 residues for F4. Specifically, they are as follows – F1, residues 11–32(b) and 250–263(b); F2, residues 18–52(f) and 270–290(d); F3, residues 45–57(e) and 275–284(d) and F4 residues 40–64(d). The total region of recognizable heptad substructure would then be about 253, 273, 240 and 240 residues respectively for F1, F2, F3 and F4. The maximum lengths of a coiled coil rope conformation in these cases would thus be about 38, 41, 36 and 36 nm, respectively, assuming that the axial rise per residue in a coiled-coil conformation is 0.1485 nm. These ranges are effectively limited by the presence of proline residues and/or the lack of apolar residues in heptad positions a and d. If it is assumed that the length of the coiled-coil rod domain is the same in all four chains then this value must lie in the range 32–36 nm. The apolar contents of the combined a and d positions for the entire heptad-containing stretches of sequences are about 70%, 78%, 66% and 83% respectively. These
figures compare favorably with the value of approximately 70–75% seen in a large variety of other a-fibrous proteins (Conway and Parry, 1990). An important proviso to the above discussion is that the methods used to predict the a-conformation in the terminal domains are all based on data from hydrated protein crystals and the predicted content of regular secondary structure will almost certainly be an over-estimate for a dehydrated fiber. Another proviso regarding the prediction of heptad patterns is that in a protein containing a high proportion of hydrophobic residues the occurrence of short heptad patterns will occur by chance and have no significance as regards coiled-coil ropes. The numbers of potential interchain ionic interactions have been calculated for a wide variety of two-strand coiled-coil ropes (Conway and Parry, 1990). In every case these have peaked for a parallel, in-register chain arrangement. In the case of the four honeybee silk chains the numbers of interchain ionic interactions for a four-chain coiled-coil have been calculated for all chain combinations (homo- and hetero-tetramers), both parallel and antiparallel, and with a range of relative axial staggers. The analysis is indicative that inter-chain ionic interactions that specify both the relative chain stagger and orientation in most coiled-coil structures are few in number in the aculeate silks. It is near certain, however, that the chains will lie in axial register (thereby maximizing heptad overlap) and will lie parallel to one another, as found in all of the other a-fibrous proteins studied to date. Sutherland et al. (2014) have also noted that the hydrophobic wax molecules in which the honeybee silk normally exists in vivo may also disrupt the hydrophobic interactions of the core a and d residues that largely specify the coiled coil structure and some evidence has been reported that isolated a-helices may be the predominant structure in honeybee silk (Kameda and Tamada, 2009). It is therefore not unlikely that the stabilities of a coiled-coil conformation and a bundle of individual a-helices are comparable, with the former being the structure adopted in the more aqueous environment of the silk gland and the latter in the final material when in the presence of lipids (Sutherland et al., 2014). If it is assumed that only one type of molecule exists then six different cyclic chain orders between parallel chains are possible. These are (in clockwise order) F1–F2–F3–F4, F1–F2–F4–F3, F1–F3–F2–F4, F1–F3–F4–F2, F1–F4–F2–F3 and F1–F4–F3–F2. If one
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
531
Fig. 2. Homologies in the amino acid sequences of the four constituent chains F1, F2, F3, F4 in honeybee silk and V1, V2, V3 and V4 in hornet silk. The signal peptides are not included. Over a length of 217 residues (about 32 nm) in honeybee silk all four chains have a segment, marked in mauve, in which there is an unbroken repeating heptad motif characteristic of coiled-coil a-helix ropes, and the sequences have been aligned to bring the central domains in each chain into register. These residues are regarded as delineating a central rod domain having with a four-strand rope structure. (a) Percentage identity, (b) the distribution of alanine (green), lysine (blue) and glutamine (red).
allows the possibility of antiparallel chains this increases to 48 different combinations. The simplest arrangement, as well as the most likely, is that in which the central domains in all of the four chains are in lateral register and are similarly directed. Campbell et al. (2014) have shown that the bee silks are stabilized predominantly by intermolecular covalent crosslinking,
particularly through the isopeptide crosslink e-(c-glutamyl)-lysine produced by the transglutaminase enzymes. It is therefore of interest to study the disposition of the lysine (blue) and glutamine residues (red) in F1, F2, F3 and F4 (see Fig. 2b). Overall, the glutamines are few in number and percentage in the terminal domains (7 and 1.8% respectively) but more common in the rod domain (42 and
532
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Fig. 2 (continued)
4.9%). Lysine, in contrast, is more common in the terminal domains (41 and 10.8%) than in the rod domain (53 and 6.2%). Specifically, in the N, rod and C-terminal domains, respectively, there are 0, 9 and 1 (F1), 2, 12 and 1 (F2), 1, 13 and 2 (F3) and 0, 8 and 0 glutamine residues (F4). The corresponding figures for the lysine residues are 1, 16 and 8 (F1), 2, 12 and 5 (F2), 7, 13 and 4 (F3) and 6, 12 and 8 (F4) respectively. These data suggest that the glutamine end of the crosslink is far more likely to be found in the rod domain than in a terminal domain but the figures are less clear-cut for
the lysine end of the crosslink. The isopeptide crosslinks are likely to be particularly important in stabilizing the filaments axially but are likely also to provide lateral stability. 3.2. Selection of model parameters and derivation of possible filament structures Atkins (1967) measured the spacing of the near-equatorial layer line in honeybee silk and obtained a value of 3.5 ± 0.1 nm
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
corresponding to a pitch length of 14.0 ± 0.4 nm for a four-strand rope. The spacing of the prominent meridional reflection that characterizes the a-pattern, which represents the axial length of 3.5 residues, was measured as 0.506 ± 0.005 nm. Thus, the number of residues in a pitch length lies between 3.5 13.6/0.511 = 93.2 and 3.5 14.4/0.501 = 100.6. For both the illustrations and the transform calculations it is convenient, though not essential, for the number of residues in the pitch length to be divisible by seven, the number of residues in the repeating unit of the coiled coil, and a choice was made of 98 residues which implies a pitch length of 98 2 0.506/7 = 14.2 nm. Rudall (1965) pointed out that the most compact structure that could be obtained for a pair of two-strand ropes occurred when the bulges on the second rope were aligned with the hollows on the first rope, and this node–antinode type of packing has been found in both the egg case of the praying mantis (Bullough and Tulloch, 1990) and in a-keratin (Fraser and Parry, 2014a). This condition is attained when the second rope is rotated by p/2 relative to the first or displaced axially by P/4, where P is the pitch of the helix, or some equivalent combination of rotation and displacement. In the case of the four-strand ropes the values have to be p/4 or P/8 or some equivalent combination. An excellent illustration of this type of packing is seen in the crystal structure of the measles virus phosphoprotein multimerization domain (Blocquel et al., 2014), shown in Figs. 3 and 4. The packing is tetragonal in projection and the 2 four-strand ropes in the unit cell (colored blue) are aligned axially so that they are packed in a node–antinode manner as shown in Fig. 4. At any particular level in the region of overlap there is a combination of rotation and axial displacement equivalent to a rotation of p/4 between the orientations of the two molecules in the unit cell. Several types of filament of differing diameter have been reported to be present in the contents of the honeybee silk glands (Rudall, 1965; Flower and Kenchington, 1967) and possible models for these filaments are illustrated diagrammatically in Fig. 5. The smallest, shown in Fig. 5a, has a central domain with a fourstrand rope structure (length 217 residues) and a matrix component consisting of the N- and C-terminal domains distributed over the surface. The second, shown in Fig. 5b, has two such ropes in lateral register with a rotational phase difference of p/4 between them so that node–antinode packing is achieved as in Fig. 4. In the four-rope model (c) two two-rope filaments are combined in such a manner that all four ropes are related by a progressive rotation of p/4. There is no reason to suppose that the axes of the coiled coils in either the two-rope model or the four-rope model are straight and in both cases are likely to have a slow twist. Again, the N- and C-terminal domains form a coating on the surface in both (b) and (c). The filaments, around 4–5 nm in diameter, visible in electron micrographs, would be compatible with the four-rope model (c), and Flower and Kenchington noted that each of these appeared twinned, comprising elements 2.0–2.5 nm in diameter. This is compatible with a pair of two-rope filaments with node–internode packing, similar to that observed in Fig. 5b, combining to form a four-rope filament as shown diagrammatically in Fig. 5c.
533
Fig. 3. Axial projection of the four-strand ropes in the crystal structure of the measles virus phosphoprotein multimerization domain (Blocquel et al., 2014).
3.3. Information derived from the X-ray diffraction patterns Common features of X-ray diffraction patterns obtained from fibrous proteins are a low-angle equatorial pattern, where the innermost reflections can be directly related to the packing of the filaments (often quasi-hexagonal), and a series of sharp meridional reflections associated with a very precise repetition of a structural unit in the filament. The observed pattern can be regarded as the transform of the filament with a superposed external interference function determined by fibril packing. This latter
Fig. 4. Longitudinal relationship between the two molecules in the unit cell of the measles virus phosphoprotein multimerization domain (Blocquel et al., 2014) shown in cartoon view. The 2 four-strand ropes are aligned axially so that they are packed in a node–antinode manner (Rudall, 1956; Bullough and Tulloch, 1990).
534
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538 Table 1 Spacings of the observed equatorial reflections compared with those calculated for a two dimensional lattice with a = 4.52 nm, b = 3.74 nm and c = 101.1° corresponding to an axial projection of the unit cell.
Fig. 5. Cross-sections of models for the central domains in three types of filament that could be formed in an aqueous environment, for example in the silk glands. The small white circles represent a-helices. (a) A single four-strand rope, (b) a pair of four-strand ropes in which the two central domains have a phase shift of p/4 resulting in node/internode packing as in Fig. 4. The likelihood is that the filament will have a slow twist about its axis, (c) 4 four-strand ropes with a cyclic phase shift of p/4, again the likelihood is that the filament will have a slow twist. In all cases the N- and C-terminal domains form a surface coating.
function has maxima and minima close to the origin but tails off to unity with increasing R. In most cases, the observed intensity away from the meridian and equator contains no maxima associated with the external interference function and is determined solely by filaments. This indicates that when viewed down the fiber axis the projections of the filaments are well ordered but the axial shift and/or rotation between adjacent filaments is subject to cumulative variation which leads to a rapid increase in both band-width at half-height and rapid attenuation of the intensity (Fraser and MacRae, 1973, pp. 31–39). The absence of significant evidence for external interference remote from the equator and the meridian in honeybee silk is indicative of a texture with precisely ordered filaments and cumulative disorder in the rotation and/or z-shift between neighboring filaments. The fact that the equatorial pattern in honeybee silk has well developed equatorial reflections (Fig. 1) suggests that the cumulative disorder in projection is low whilst the cumulative disorder in rotation and/or z-shift is much greater. The terminal domains are presumed to have much less secondary structure but may still have a well-defined conformation that enables the lateral order to be largely maintained even in the presence cumulative disorder in the rotation and/or z-displacement of the filament. The spacings of the observed equatorial reflections, as determined in this work from a scan of the X-ray pattern in Lucas and Rudall (1968), are summarized in Table 1. They are consistent with an axial projection of the filament packing with a two-dimensional lattice with a = 4.52 nm, b = 3.74 nm and c = 101.1° (Table 1) and an area of 4.52 3.74 sin (101.1°) = 16.59 nm2. The volume of a cylinder with this base and a height equal to the mean axial rise per residue in the coiled coil (0.1485 nm) is 2.46 nm3. The mean residue weight in the central domain is 96.0 and assuming that the density is 1.45 g cm3 the volume per chain is 1.66 96.0/(1.45 1000) = 0.110 nm3. Thus the number of four-strand ropes that could be accommodated in a 0.1485 nm length of the unit cell is therefore estimated to be 2.46/ (4 0.110) = 5.59. The simplest interpretation of this would be that the asymmetric unit consists of a four-rope filament together with a matrix occupying the remaining volume of 2.46 16 0.110 = 0.70 nm3 as shown in Fig. 6. The spacings of six meridional reflections extracted from a digitized scan of the Lucas and Rudall X-ray pattern (1968) are given in Table 2. The implied possibilities for the value for the period are illustrated in Fig. 7a. These may be compared with those with derived from the seven meridional spacings measured by Kameda et al. (2014b) for hornet silk Fig. 7b. In each case, the plotted values refer to the root-mean-square deviations between the
h
k
Calc
Obs
0 0 1 1 1 1 1 2 2 2 2 2
1 2 2 1 0 1 2 2 1 0 1 2
3.74 1.87 1.86 3.20 4.52 2.64 1.62 1.60 2.12 2.26 1.79 1.32
3.74 1.84 1.84 – 4.52 2.64 1.61 1.61 2.07a 2.21a 1.72 1.34
Notes Overlaps [1, 2] Overlaps [0, 2] Not observed
Overlaps [2, 2] Overlaps [1, 2]
a Not resolved in the pattern obtained by Lucas and Rudall (1968) due to overexposure. The observed values refer to the pattern obtained by Rudall (1962) which was obtained with a smaller exposure.
Fig. 6. The projection of the base of the unit cell, determined from an analysis of the low-angle equatorial X-ray diffraction patterns of honeybee silk. The simplest model for the contents of the cell is a single four-rope filament and a layer of matrix formed from the terminal domains.
indices corresponding to the spacings of the observed meridional reflections and those of the nearest integer values. The trace in Fig. 7a is overlaid by noise acquired during the extraction of the data from the published photographs due to the short specimen-film distance and graininess in the film. With only six meridional reflections the results of statistical tests need to be treated as indicative rather than precise measures of goodness of fit. Two of the possible values in honeybee (33.1 and 43.0 nm) have counterparts in hornet, which is perhaps not surprising given the close similarities of their amino acid compositions of their central domains (Fig. 2). The first of these (33.1 nm.) is very close to the value used by Kameda et al., 2014b to index the meridional reflections in hornet silk so is a natural choice to check whether a repeat parallel to the fiber axis of this length is compatible with our suggestion of a cell containing 4 four-strand ropes. The combined molecular weights of the 4 four-strand ropes in a filament similar to that in Fig. 5c is M = 494,260 g mol1 and the
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
535
Table 2 Spacings of the observed meridional reflections compared with those calculated for an axial period of 33.1 nm. Index: Calculated (nm): Observed (nm): Observed index:
11 3.00 3.11 10.6
13 2.55 2.55 13.0
16 2.07 2.06 16.1
20 1.66 1.63 20.3
27 1.23 1.23 26.9
33 1.00 1.00 33.1
Meridional scan.
associated volume V, measured in nm3, is 0.00166M/q, where q is the mean density of the protein measured in g cm3. As the area of the cell base A is 16.59 nm2, the predicted value of c is V/A = 0.00166 494,260/(16.59 q). The value of q is not known and various algorithms have been used in attempts to predict its value (Quillin and Matthews, 2000; Fischer et al., 2004). The former authors suggest 1.43 ± 0.03 and 1.47 ± 0.03 g cm3 as possible ranges. Using these values the estimated value of c is 33.64– 34.58 nm. The former value is very close to one of the possible values (33.1 nm) in honeybee silk, and to the value of c derived for hornet silk by Kameda et al. (34.4 nm in Fig. 7b). Given the close similarity between the compositions of the central domains in the two materials this suggests that the minimum in Fig. 7a at 33.1 nm, is the appropriate choice for the period in honeybee silk. It will be seen from Table 2, however, that there is an unacceptable discrepancy between the spacing calculated for the 11th order of the axial repeat, and the observed spacing, which corresponds to an index of 10.6. A possible explanation for this is that the rod domains are packed head-to-head and tail-to-tail leading to a doubling of the period. Unfortunately, the number of reflections that can be extracted from the published photograph is insufficient to speculate further and this anomaly can only be resolved when data collected with modern high-resolution X-ray equipment become available. The number of residues in each chain in the central domain is estimated to be 217 and with a mean axial rise per residue of 0.1485 nm this will account for 32.2 nm of the period. The remaining distance of 0.9 nm, equivalent to about six residues in an ahelical conformation, must include residues from one or both of the terminal domains immediately proximal to the central domain. The filaments are thus envisaged as parcels of 4 four-strand ropes packed head-to tail with a small number of residues at the ends of the central domains bonding the parcels together. The above model, however, is only one of several possibilities. Its main merit lies in its simplicity, with four-rope filaments aggregating into sheets parallel to the b-axis (Fig. 6) and these sheets, in turn, being separated from one another by a layer of matrix formed from the distal parts of the terminal domains. When high
Fig. 7. Possible axial periods for (a) honeybee silk and (b) hornet silk derived by plotting the root-mean-square (RMS) deviation of the calculated indices from integral values. The trace in (a) is overlaid by noise acquired during the extraction of the data from the published photographs due to imperfect descreening and is exacerbated by the short specimen-film distance, which is inadequate for the precise measurements of low-angle reflections, due to graininess in the film.
resolution digital data become available they can be used to compare the intensity transform calculated for both this and alternative models, and then used as a basis for further refinement. An alternative choice for the period in honeybee silk is 30.9 nm (Fig. 7), which has the lowest RMS deviation. The problem, however, is that this would require a density of 1.56 g cm3 which is well outside the established range. 3.4. The high-angle intensity transform The calculated high-angle intensity transform for the three types of filaments shown in Fig. 5 are shown in a graphical form (Fig. 8), in which values are plotted above and below the layer line to enable an easier visual comparison with the observed pattern. It should be emphasized that this is not a simulated pattern, which would require many extra assumptions about the lengths of the rod-like portions, the orientation density function and the distribution functions for positional, axial, and rotational disorder. All of the predicted maxima appear in the observed patterns, as do subsidiary maxima in the profile of the (2, 0) meridional reflection in the multi-rope models. This arises from interference between the scattering from the component ropes. The spacings
536
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Fig. 8. Comparisons between the observed intensities in film space with the calculated distribution of intensity in reciprocal space for (a) a single four-strand rope, (b) a pair of four-stand ropes with node–internode packing, (c) 4 four-strand ropes packed tetragonally with node–internode packing, and (d) 4 four-strand ropes packed hexagonally. The latter may be appropriate to specimens from which water has been completely removed. The Z positions of the layer lines in the calculated intensity transform corresponding to the (m, n) values are shown as lines, and R is the radial coordinate as defined in Section 2. All of the predicted maxima appear in the observed patterns. However, subsidiary maxima in the meridional streak on layer line (2, 0) for multi-rope models are present (as noted by Atkins, 1967), but are not adequately accounted for. It is suggested here, however, that they arise either from an external interference function associated with the scattering from filaments in different cells, or with a slow twist in the filament that results in the axes of the four-strand ropes being inclined to the fiber axis. The spacings of the maxima of the observed peaks on the equator and the nearequatorial layer line are close to the values of the maxima in the predicted intensity transforms. The shapes of these reflections are appropriate to an infinite four-strand coiled coil rope of poly-L-alanine and will be smoothed in honeybee silk because of the limited length, variations in the local radius and pitch length in the coiled coils, and the incorporation of residues other than alanine.
Fig. 9. An extract from the X-ray pattern obtained from honeybee silk by Lucas and Rudall (1968), showing the satellites of the meridional reflection on layer line (2, 0). R is the radial coordinate as defined in Section 2. One possible explanation for the satellites is that they are produced by an external interference function associated with the a-axis of the projected unit cell (Fig. 6).
of the maxima of the observed peaks on the equator and the nearequatorial layer line are close to the values of the maxima in the
predicted intensity transforms (Fig. 8). The intensities calculated for the equatorial and near-equatorial layer lines exhibit pronounced maxima and minima, which will be smoothed out, to a large extent, for several reasons. For example, the calculations were carried out for poly-L-alanine coiled coil, assuming a constant radius, pitch and axial rise per residue, whereas studies of coiled coils in crystalline proteins show that substantial variations occur in these parameters (see, for example, Meier et al., 2009; Nicolet et al., 2010). In addition, the finite length of the coiled coils and the presence of non-alanine residues will produce further blurring. The unusual profile of the (2, 0) meridional reflection, noted by Atkins (1967), is better resolved in the X-ray pattern obtained by Lucas and Rudall (1968), and is reproduced in Fig. 9. The specimen has been tilted so that the center of the (2, 0) layer line intercepts the sphere of reflection and it will be seen that subsidiary maxima, characteristic of a lateral periodicity, are evident. The spacing of
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
the satellite maxima on the (2, 0) layer line is difficult to judge but appears to be around 4.5 nm which suggest that it may be related to the inter-sheet distance in the unit cell (Fig. 6). Another possibility is that the filaments have a slow twist that results in the axes of the four-strand ropes being inclined to the fiber axis. 3.5. Polymorphism It is to be expected that the dimensions of the unit cell and possibly also the packing arrangement of the molecules will vary with water content. As far as we have been able to determine, the X-ray data obtained by Kameda et al. (2014b), Rudall (1962) and Lucas and Rudall (1968) were from specimens with very low water content, whereas the data obtained by Rudall (1965) and Atkins (1967) were from hydrated specimens. The axial period and the dimensions of the axial projection of the base of the unit cell derived in the present study from the X-ray data of Lucas and Rudall (1968) are probably appropriate to the dehydrated material. 4. Summary The aim of the present study was to re-examine the original X-ray diffraction data from honeybee silk obtained by Rudall, Atkins and colleagues to see if a model for the molecular packing could be devised. Atkins (1967) had previously shown that much of the high-angle X-ray diffraction data could be explained by molecules having a four-strand coiled-coil conformation. Calculations presented here confirm this conclusion and, furthermore, show a close relationship between the entire high-angle X-ray pattern and the intensity transforms of the models that are described in this work. Evaluation of the low-angle X-ray diffraction data has now enabled us to determine the likely dimensions of the axially projected base of the unit cell and to construct a model for the filaments, each containing 4 four-strand ropes packed in a node–antinode manner, thereby maximizing packing density. In addition, a value for the axial period has been derived from the measured meridional spacings, which is consistent with a near common axial period in both honeybee and hornet silks. The latter, in turn, is compatible with a packing arrangement in which the central domains are arranged in a head-to-tail manner to form continuous filaments in a manner closely akin to that found with the coiled-coil proteins in the egg cases of praying mantises (Bullough and Tulloch, 1990), and with the b-crystallites in the b-keratins (Fraser and Parry, 2014b). The data as a whole suggest that, in common with a number of other natural fibers, the aculeate silks are composites with a filament-matrix texture. In the proposed model the four-rope filaments are tightly packed in one dimension i.e. in sheets. In turn, these sheets are separated from one another by a matrix composed of the bulk of the terminal domains. There is, therefore, a close parallel between the molecular structures of the two silks and the keratins of birds and reptiles. Both materials display a common central domain, though this is characterized by a high content of a-structure for the silks but b-structure for the avian and reptilian keratins. In both structures the central domains aggregate end-toend to give the fiber their high tensile strength. In honeybee silk the contribution of the isopeptide crosslinks to the tensile attributes of the fiber is also likely to be very significant. The terminal domains, with their wide range of amino acid compositions, form a matrix between the sheets of the filaments, thereby adding pliability to the structure. The present study highlights the need for further highresolution studies of the low-angle X-ray diffraction pattern, and electron microscope studies of longitudinal and transverse sections, using modern equipment, to resolve the uncertainties about the exact nature of the filaments and their packing.
537
Acknowledgments The authors wish to express their sincere thanks to Emeritus Professor E.D.T. (Ted) Atkins for unearthing and making available to us many of the X-ray diffraction patterns of honeybee silk and other members of the aculeate family taken by him some 50 years earlier. These have provided an invaluable context to the studies that we have undertaken. His interest and encouragement were also very much appreciated, as is that of Dr Tara Sutherland. Thanks are also due to Professor Martin Hazelton for his assistance in identifying possible periods in honeybee silk and assessing their significance. RDBF is indebted to Massey University for the award of an Honorary Research Fellowship.
References Astbury, W.T., Woods, H.J., 1930. The X-ray interpretation of the structure and elastic properties of hair keratin. Nature 126, 913–914. Astbury, W.T., Woods, H.J., 1933. X-ray studies on the structure of hair, wool and related fibres II: the molecular structure and elastic properties of hair keratin. Phil. Trans. R. Soc. Lond. A 232, 333–394. Atkins, E.D.T., 1967. A four-strand coiled-coil model for some insect fibrous proteins. J. Mol. Biol. 24, 139–141. Blocquel, D., Habchi, J., Durand, E., Sevajol, M., Ferron, F., Erales, J., Papageorgiou, N., Longhi, S., 2014. Coiled-coil deformations in crystal structures: the measles virus phosphoprotein multimerization domain as an illustrative example. Acta Crystallogr. D Biol. Crystallogr. D70, 1589–1603. Bullough, P.A., Tulloch, P.A., 1990. High-resolution spot-scan electron microscopy of microcrystals of an a-helical coiled-coil protein. J. Mol. Biol. 215, 161–173. Campbell, P.M., Trueman, H.E., Zhang, Q., Kojima, K., Kameda, T., Sutherland, T.D., 2014. Cross-linking in the silks of bees, ants and hornets. Insect Biochem. Mol. Biol. 48, 40–50. Cohen, C., Holmes, K.C., 1963. X-ray diffraction evidence for a-helical coiled-coils in native muscle. J. Mol. Biol. 6, 423–432. Conway, J.F., Parry, D.A.D., 1990. Structural features in the heptad substructure and longer range repeats of two-stranded a-fibrous proteins. Int. J. Biol. Macromol. 12, 328–334. Crick, F.H.C., 1953a. The Fourier transform of a coiled-coil. Acta Crystallogr. 6, 685– 689. Crick, F.H.C., 1953b. The packing of a-helices: simple coiled-coils. Acta Crystallogr. 6, 689–697. Elliott, A., Lowy, J., Parry, D.A.D., Vibert, P.J., 1968. Puzzle of the coiled coils in the aprotein paramyosin. Nature 218, 656–659. Filshie, B.K., Rogers, G.E., 1962. An electron microscope study of the fine structure of feather keratin. J. Cell Biol. 13, 1–12. Fischer, H., Polikarpov, I., Craievich, A.F., 2004. Average protein density is a molecular-weight-dependent function. Prot. Sci. 13, 2825–2828. Flower, N.E., Kenchington, W., 1967. Studies on insect fibrous proteins: the larval silk of Apis, Bombus and Vespa (Hymenoptera: Aculeata). J. Microsc. Soc. 86, 297–310. Fraser, R.D.B., MacRae, T.P., 1973. Conformation in Fibrous Proteins and Related Synthetic Polypeptides. Academic Press, New York and London. Fraser, R.D.B., Parry, D.A.D., 2014a. Keratin intermediate filaments: differences in the sequences of the Type I and Type II chains explain the origin of the stability of an enzyme-resistant four-chain fragment. J. Struct. Biol. 185, 317–326. Fraser, R.D.B., Parry, D.A.D., 2014b. Amino acid sequence homologies in the hard keratins of birds and reptiles, and their implications for molecular structure and physical properties. J. Struct. Biol. 188, 213–224. Fraser, R.D.B., MacRae, T.P., Miller, A., 1964. Fourier transform of the coiled-coil model of a-keratin. Acta Crystallogr. 17, 813–816. Fraser, R.D.B., MacRae, T.P., Miller, A., 1965. X-ray diffraction patterns of a-fibrous proteins. J. Mol. Biol. 14, 432–442. Fraser, R.D.B., MacRae, T.P., Miller, A., Rowlands, R.J., 1976. Digital processing of fibre diffraction patterns. J. Appl. Crystallogr. 9, 81–91. Källberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H., Xu, J., 2012. Templatebased protein structure modeling using the RaptorX web server. Nat. Protoc. 7, 1511–1522. Kameda, T., Tamada, Y., 2009. Variable-temperature 13C solid-state NMR study of the molecular structure of honeybee wax and silk. Int J. Biol. Macromol. 44, 64– 69. Kameda, T., Walker, A.A., Sutherland, T.D., 2014a. Evolution and application of coiled coil silk from insects. In: Asakura, T., Miller, T. (Eds.), Biotechnology of Silk. Springer, pp. 87–106. Kameda, T., Nemoto, T., Ogawa, T., Tosaka, M., Kurata, H., Schaper, A.K., 2014b. Evidence of a-helical coiled coils and b-sheets in hornet silk. J. Struct. Biol. 185, 303–308. Kelley, L.A., Sternberg, M.J.E., 2009. Protein structure prediction on the web: a case study using the Phyre server. Nat. Protoc. 4, 363–371. Klug, A., Crick, F.H.C., Wyckoff, H.W., 1958. Diffraction by helical structures. Acta Crystallogr. 11, 199–213.
538
R.D.B. Fraser, D.A.D. Parry / Journal of Structural Biology 192 (2015) 528–538
Lucas, F., Rudall, K.M., 1968. Extracellular fibrous proteins: the silks. In: Florkin, M., Stotz, E.H. (Eds.), Comprehensive Biochemistry, vol. 26B. Elsevier, Amsterdam, pp. 475–558. Meier, M., Padilla, G.P., Herrmann, H., Wedig, T., Hergt, M., Patel, T.R., Stetefeld, J., Aebi, U., Burkhard, P., 2009. Vimentin coil 1A – a molecular switch involved in the initiation of filament elongation. J. Mol. Biol. 390, 245–261. Nicolet, S., Herrmann, H., Aebi, U., Strelkov, S.V., 2010. Atomic structure of vimentin coil 2. J. Struct. Biol. 170, 369–376. Pardon, J.F., 1967. A modified theory for Fourier transform for a coiled-coil. Acta Crystallogr. 23, 937–939. Parry, D.A.D., Fraser, R.D.B., Squire, J.M., 2008. Fifty years of coiled-coils and ahelical bundles: a close relationship between sequence and structure. J. Struct. Biol. 163, 258–269. Quillin, M.L., Matthews, B.W., 2000. Accurate calculation of the density of proteins. Acta Crystallogr. D56, 791–794. Rudall, K.M., 1956. Protein ribbons and sheets. Lectures on the Scientific Basis of Medicine, vol. 5. Athlone Press, London, Ch 12, pp. 217–230. Rudall, K.M., 1962. Silk and other cocoon proteins. In: Florkin, M., Mason, H.S. (Eds.), Comparative Biochemistry, vol. 4. Academic Press, New York, pp. 397–433.
Rudall, K.M., 1965. Skeletal structure in insects. In: Aspects of Insect Biochemistry. Academic Press, London/New York, pp. 83–92. Sutherland, T.D., Campbell, P.M., Weisman, S., Trueman, H.E., Sriskantha, A., Wanjura, W.J., Haritos, V.S., 2006. A highly divergent gene cluster in honey bees encodes a novel silk family. Gen. Res. 16, 1414–1421. Sutherland, T.D., Weisman, S., Trueman, H.E., Sriskantha, A., Trueman, W.H., Haritos, V.S., 2007. Conservation of essential features in coiled coil silks. Mol. Biol. Evol. 24, 2424–2432. Sutherland, T.D., Church, J.S., Hu, X., Huson, M.G., Kaplan, D.L., Weisman, S., 2011a. Single honeybee silk protein mimics properties of multi-protein silk. PLoS ONE 6, e16489. http://dx.doi.org/10.1371/journal.pone.0016489. Sutherland, T.D., Weisman, S., Walker, A.A., Mudie, S.T., 2011b. The coiled coil silk of bees, ants and hornets. Biopolymers 97, 446–454. Sutherland, T.D., Trueman, H.E., Walker, A.A., Weisman, S., Campbell, P.M., Dong, Z., Huson, M.G., Woodhead, A.L., Church, J.S., 2014. Convergently-evolved structural anomalies in the coiled coil domains of insect silk proteins. J. Struct. Biol. 186, 402–411.