ARTICLE IN PRESS Journal of

Structural Biology Journal of Structural Biology xxx (2003) xxx–xxx www.elsevier.com/locate/yjsbi


Revisiting the structure of Alvinella pompejana hemoglobin at 20 A resolution by cryoelectron microscopy Ludovic Jouan,a Sergio Marco,a,b and Jean-Christophe Taveaua,* a

Laboratoire des Prot eines Complexes, J. E. 2320, Universit e de Tours. 2, bis Boulevard Tonnell e, F-37032 Tours Cedex, France b Institut Curie Section de Recherche, UMR 168, 11 rue Pierre et Marie Curie. 75231 Paris Cedex 05, France Received 17 January 2003, and in revised form 25 April 2003

Abstract
resolution from frozen-hydrated The hemoglobin of the polychaete worm Alvinella pompejana was reconstructed at 20 A samples observed by electron microscopy according to the random conical tilt series method. This three-dimensional reconstruction was mirror-inverted with respect to a previous volume published by de Haas et al. in 1996. In order to explain this handedness discrepancy, various 3D reconstructions using different reference volumes were carried out showing that the choice of the first volume was the keystone during the refinement process. The 3D reconstruction volume of A. pompejana Hb presented structural features characteristic of annelid Hbs with two hexagonal layers each comprising six hollow globular subassemblies and a complex of non-heme linker chains. Moreover, the eclipsed conformation of the two hexagonal layers and a HGS architecture similar to that described for Arenicola marina Hb led to the conclusion that A. pompejana Hb belonged to the architectural type II according to the definition of Jouan et al. (2001). A comparison between this cryo-EM volume and X-ray crystallography density maps of Lumbricus terrestris type-I Hb (Royer et al., 2000) showed that the triple stranded coiled coil structures of linker chains were different. Based on this observation, a model was proposed to explain the eclipsed conformation of the two hexagonal layers of type-II Hbs. Ó 2003 Published by Elsevier Science (USA). Keywords: Hexagonal bilayer hemoglobin; Annelids; Cryoelectron microscopy; Alvinella pompejana; 3D reconstruction

1. Introduction Alvinella pompejana worm is an annelida (polychaete) living in the deep-sea hydrothermal vent. The extracellular respiratory pigment of these annelids is a large multimeric protein composed of monomeric or trimeric globin chains and heme-deficient linker chains (Vinog-

* Corresponding author. Present address: IECB. University of Bordeaux I. 16, Avenue Pey-Berland, F-33607 Pessac Cedex, France. Fax: +33-5-40-00-34-84. E-mail address: [email protected] (J.-C. Taveau). 1 Abbreviations used: 3D, three-dimensional; Chls, chlorocruorins; ESI-MS, electrospray ionization mass spectrometry; EM, Electron microscopy; EMV, expected molecular volume; FSC, Fourier shell correlation; Hb, hemoglobin; HGS, hollow globular subassembly; MALLS, multiangle laser light scattering; TMV, tobacco mosaic virus; SD, standard deviation; SIRT, simultaneous iterative reconstruction technique.

radov, 1985). Like the other extracellular annelid hemoglobins (Hbs),1 the molecule is characterized by an acidic isoelectric point and low heme and iron contents (Vinogradov et al., 1982, 1991) appearing as hexagonal bilayers in the electron microscope (Antonini and Chiancone, 1977; Boekema and Van Heel, 1989; Lamy et al., 1996; Vinogradov et al., 1982). Molecular mass of this Hb was estimated to 3833 kDa by multiangle laser light scattering (MALLS) (Zal et al., 1997). The subunit composition comprises four types of monomeric globin chains termed a1 (16 633.4 Da), a2 (16 532.4 Da), a3 (16 419.6 Da), and a4 (16 348.9 Da) and one disulfidebonded trimer T (51 431.9 Da) composed of globin chains b (16 477.5 Da), c (16 916.1 Da), and d (18 048.8 Da) (Zal et al., 1997). Four linker chains termed L1–L4 were analyzed by electrospray ionization mass spectrometry (ESI-MS) leading to molecular mass of 22 887.1, 24 230.5, 26 233.6, and 25 974.4 Da, respectively.

1047-8477/03/$ - see front matter Ó 2003 Published by Elsevier Science (USA). doi:10.1016/S1047-8477(03)00115-1

ARTICLE IN PRESS 2

L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

Numerous 3D reconstructions of oligochaete (Mouche et al., 2001; Schatz et al., 1995; Taveau et al., 1999), achaete(de Haas et al., 1996b), polychaete (de Haas et al., 1996a; Jouan et al., 2001), and vestimentifera (de Haas et al., 1996c) Hbs were carried out by cryoelectron microscopy (cryo-EM) during the last ten years. From these data, a common architecture has emerged. It presents a D6 point-group symmetry composed of two hexagonal layers each made up by six hollow globular subassemblies (HGS) linked to a central piece. Each HGS possesses 12 globin chains disposed in four trimeric layers and symmetrically arranged around a local 3-fold axis. Moreover, six connections termed c1, c2, c3a, c3b, c4, and c5 described by Taveau et al. (1999) correspond to contact areas between neighboring HGS and to three linking units located at the HGS bottom. Finally, the central piece is linked to the 12 HGS through 12 connections termed c5. The structure of Lumbricus terrestris Hb solved by X-ray crystallography
resolution (Royer et al., 2000) has confirmed the at 5.5 A structural features observed by cryo-EM and the existence of 144 globin chains and 36 linker chains. Furthermore, the c5 connection observed in cryo-EM appears as a triple stranded coiled coil corresponding to the N-terminal parts of the three linker chains (described as c3a, c3b, and c4 in cryo-EM). More recently, two architectural types of annelid Hbs have been defined (Jouan et al., 2001). (i) Type I characterized by a rotation of
16° between the two hexagonal layers is found in Hbs of L. terrestris, Macrobdella decora, and Riftia pachyptila (de Haas et al., 1996b, 1996c, 1997) and in the chlorocruorins of Eudistylia vancouverii (de Haas et al., 1996d) and Sabella spallanzanii (Lanzavecchia et al., 1999). (ii) Type-II Hbs present an eclipsed disposition of the two hexagonal layers and have been observed in two polychaetes: A. pompejana (de Haas et al., 1996a) and A. marina (Jouan et al., 2001). The Hb of A. pompejana was first reconstructed in 1996 by de Haas et al. from frozen-hydrated specimen at
resolution. However, due to the low resolution, 27.7 A the 3D reconstruction volume showed a distorted organization of the inner masses of the HGS with respect to A. marina Hb and the linker chains were not discernable in the volume densities. In order to know whether the A. pompejana Hb has a specific architecture, a new 3D reconstruction is carried out at higher resolution. Data presented in this paper show that A. pompejana Hb unambiguously belongs to the architectural type II and that the structural differences described in 1996 are due to the computation of a mirror-inverted volume. Moreover, our new 3D reconstruction volume is in good agreement with the X-ray crystallography maps of L. terrestris Hb. A model based on the X-ray dodecamer architecture is proposed in order to explain the eclipsed conformation of the two hexagonal layers in type-II Hbs.

2. Materials and methods 2.1. Hemoglobin Worms were collected from 2600 m depth at 13°N (12°460 N, 103°560 W and 12°500 N, 103°570 W) and 9°500 N sites at the East Pacific Rise during the HOPEÕ 99 campaign in April 1999. The polychaete annelids were picked using the manipulator of the Nautile submersible and maintained at 2.4 °C in a thermally insulated container during the trip to the surface (2–3 h). On board, the animals were dorsally opened in order to withdraw directly blood samples from the main vessel into glass micropipettes pooled in melting ice. Then the blood was centrifuged at low speed for a few minutes and the supernatant was frozen in liquid nitrogen. In the laboratory, samples were centrifuged at 10 000g for 10 min at 4 °C before being purified by gel-filtration on a Superose 6 column (Amersham Pharmacia Biotechnology, Uppsala, Sweden) as previously described (Zal et al., 1997). Finally, Hb was frozen and stored at liquid nitrogen temperature. 2.2. Electron microscopy After defrosting, to prevent dissociation, the A. pompejana Hb sample was dialyzed with a 0.025 lm Millipore filter against 100 ml of Tris–HCl buffer, pH 7.0, 10 mM CaCl2 , and 10 mM MgCl2 . A droplet of sample dilution (1.3 g/l) was applied onto 400-mesh copper grids coated with a holey carbon film. A thin liquid film was formed in the holes of carbon film by blotting the excess solution. Then, the grid was quickly plunged into liquid ethane (Adrian et al., 1984; Dubochet et al., 1988; Milligan et al., 1984). The grids were transferred into a CM12 or a CM120 Philips electron microscope by using the Gatan 626 cryotransfer system. First, images of frozen-hydrated specimens were recorded under low-dose conditions at accelerating voltage of 100 kV, by tilting the grid by an angle of 45° and without tilting at 1.8 lm defocus. Second, a new image set was collected at 0° tilt angle with a 0.8 lm defocus at accelerating voltage of 120 kV. All images were recorded on SO163 Kodak films which were developed for 12 min in full-strength Kodak D19 developer. Magnifications
layer line of tobacco were calculated from the 23 A mosaic virus (TMV) included in the sample (Henderson, 1992) and were in the range of 43 820–44 168 and 53 512–54 031 for the first and second image sets, respectively. 2.3. Image processing Selected micrographs of the two image sets were digitized using an Optronics P1000 drum densitometer, at 25 lm scanning step corresponding to pixel si-

ARTICLE IN PRESS L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx


for the first set and a zes in the range 5.70–5.75 A Leafscan 45 CCD-array microdensitometer with a
scanning step size corresponding to pixel size of 4.6 A for the second one. By using interactive selection programs, coordinates of single particles showing a background free from contamination artifacts and overlapping molecules were registered; then images were windowed, contrast inverted and normalized so that their noise statistics would match a reference distribution corresponding to a portion of micrograph free of particles and artifacts. All programs used for image processing are part of SPIDER (Frank et al., 1996), SIGMA (Taveau, 1996), and Xmipp (Marabini et al., 1996) softwares. 2.4. Three-dimensional reconstruction The 3D reconstruction process was carried out in two steps. The first image set comprising tilted/untiltedspecimen image pairs were used to compute a reference volume according to the random conical tilt series method (Radermacher et al., 1987a, 1987b) in order to determine the correct handedness. Then, this volume was refined from the second untilted-specimen image set to improve the final resolution. For the computation of the reference volume, the various key parameters involved in the determination of the handedness (tilt angle in the electron microscope, scanning process, etc.) were rigorously identical to those used for L. terrestris Hb (de Haas et al., 1997). The accuracy of our protocol was confirmed by the X-ray data of L. terrestris native Hb (Royer et al., 2000) which presented the same handedness as our cryoEM volumes (de Haas et al., 1997; Taveau et al., 1999). First, from the untilted/tilted-specimen micrograph pairs, the tilt angle and direction of tilt axis were calculated according to the method developed by Radermacher (1988) during the windowing step with WEB, the interactive tool of SPIDER (Frank et al., 1996). Then, the untilted-specimen images were subjected to a reference-free bidimensional alignment so as to sort the set in homogeneous classes by correspondence analysis (Lebart et al., 1982) followed by hierarchical ascendant classification (Lebart et al., 1982; Van Heel, 1984) using WardÕs merging criterion (Ward, 1963). After classification, the tilted-specimen image sets corresponding to the hexagonal (top) and rectangular (side) views were subjected to 3D reconstructions according to the simultaneous iterative reconstruction technique (SIRT) algorithm modified for electron microscopy images (Penczek et al., 1992) with an enforced D6 point-group symmetry. These primary volumes were then aligned in a common orientation and a ‘‘merged’’ volume was calculated and refined during six cycles of 3D projection alignment (Penczek et al., 1994) leading to a reference volume.

3

Second, this reference volume was used in a cycle of 3D projection alignment in order to determine Eulerian angles of the second image set (recorded at 54 000 magnification). A volume was then computed and refined during six cycles of 3D projection alignment. 2.5. Resolution For the resolution limit estimation of the 3D reconstruction volumes, two subsets were randomly drawn from the whole image set. The corresponding 3D volumes were then compared in reciprocal space on spherical shells with increasing radii using the Fourier shell correlation (FSC) criterion (Harauz and Van Heel, 1986; Saxton and Baumeister, 1982) according to the method described by Orlova et al. (1997). In this method, the number of symmetries for the noise curve estimation was taken into account, and the resolution was calculated from the value of the crossing point between the FSC curve and the random noise curve at a threshold of 3.0 SD. Resolution was also calculated by thresholding the experimental FSC curve at a value of 0.5 corresponding to signal-to-noise ratio of 1 (Penczek, 1998). 2.6. Computer graphics 3D reconstruction volume of A. pompejana Hb was analyzed by using SIGMA software (Taveau, 1996). Isosurface representations and slice visualization were used to observe the overall shape and inner densities of 3D reconstruction volumes. The extraction and segmentation of inner substructures were carried out by a recently devised method based on a tree-like representation of the volume densities, called ‘‘cladistic analysis’’ (Taveau et al., 1999). This method allowed the computation of a cladogram equivalent to a hierarchical classification of local maximum of densities depending of threshold values. Coupled with a tool of isosurface computation, different substructures were displayed by selecting one or several nodes in the cladogram without specifying any threshold value. The construction of the model of c5 connections was carried out by calculating the skeletons of the threestranded coiled coil observed in the X-ray crystallogra
resolution, the alpha helices phy density maps. At 5.5 A appear as rod-like structures and their skeletons can be computed by determining their medial axes. This method was based on the extraction of level sets following the medial axis of tubular shapes (Lazarus and Verroust, 1999). Such level sets were defined from isosurface vertices according to the geodesic distance to a tip of the tubular polyhedron. Then, centroids of these level sets were calculated and linked to form the skeletal axis. Finally, the various paths were fitted on a theoretical helical curve and rendered as rods.

ARTICLE IN PRESS 4

L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

3. Results 3.1. Electron microscopy and 3D reconstruction An electron micrograph field of A. pompejana Hb is shown in Fig. 1. In frozen hydrated state, molecules are randomly oriented. However, among these various orientations, it is possible to identify two specific views: (i) the so-called top view (white arrow), characterized by a hexagonal contour composed of six substructures and (ii) the side view, corresponding to a two-layered structure with a rectangular contour (black arrow). The latter views are observed occasionally in cryoelectron microscopy and most of the orientations, termed intermediate views, present oval shapes (black circles). The 3D reconstruction is carried out in two steps: (i) computation of a reference volume using image pairs of tilted- and untilted-specimen and (ii) refinement of a volume calculated from a second set of untilted-specimen images aligned from the reference volume. In the first step, 1532 image pairs of A. pompejana Hbs are windowed from a selection of eleven digitized micrographs collected by tilting the grid in the electron microscope by an angle of 45 ° and without tilting. After contrast inversion and normalization, images are subjected to four cycles of two-dimensional alignment before being sorted in 25 homogeneous classes from which one class of top views and two classes of side views are selected (containing 53, 74, and 78 images, respectively). Three primary volumes are then calculated according to the random conical tilt series method (Radermacher et al., 1987a, 1987b) with an enforced D6 point-group symmetry using the corresponding tilted-specimen im-

Fig. 1. Electron micrograph field of frozen-hydrated Hb of A. pompejana taken at 0° tilt angle. Three main different orientations of the molecule are visible: top (white arrows), side (black arrows), and intermediate views (black circles). The scale bar represents 50 nm.

ages of the three selected classes. Eulerian angles of the 205 (53 + 74 + 78) images are modified so as to orient the three volumes in a common orientation to compute a ‘‘merged’’ volume. Finally, twelve cycles of 3D projection alignment using the whole set of untilted-specimen images lead to a ‘‘reference’’ volume. From 40 selected micrographs of the second cryo-EM session, 3265 untilted-specimen images are windowed, contrast inverted, normalized, and submitted to a centration process. After a single cycle of 3D projection alignment carried out to determine the three Eulerian angles of each image, a 3D reconstruction is performed and followed by eight refinement cycles. Finally, the resolution of the A. pompejana Hb 3D volume is esti
according to the FSC method modified mated to 19.6 A
according to the FSC by Orlova et al. (1997) and 23 A method thresholded at 0.5 according to Penczek (1998) and a low-pass Fermi filtration is applied on the volume
limit. down to the 20 A 3.2. 3D exploration The threshold used for isosurface computation is based on a percentage of the expected molecular volume (EMV) of the Hb. This EMV is calculated from the 3.833 MDa molecular weight determined by MALLS (Zal et al., 1997) and the partial specific volume value of 0.73 cm3 g1 . The latter value is estimated from the partial specific volumes of the globin (0.738 cm3 g1 ) and linker (0.722 cm3 g1 ) chains assuming a weight ratio of 65/35 (Zal et al., 1997). The EMV value calculated with
3 and corresponds to a these assumptions is 4.15 106 A threshold value of 0.64 for the 3D reconstruction volume of A. pompejana Hb rescaled between 0 and 1. In Fig. 2, solid-body surfaces of A. pompejana Hb calculated at a threshold displaying 100% of the EMV, are rendered in the so-called top, side, and 45° orientations. In top orientation, the molecule is viewed along the 6-fold symmetry axis leading to a hexagonal structure composed of two layers each made up of six HGS (dashed circle) (Fig. 2(a)). In addition, two types of dyad axes termed P (white ellipse) and Q (black ellipse) pass between and through the vertices of the hexagonal layers
resolution, all the HGSs are (Royer et al., 2000). At 20 A linked to their neighbors and to the central piece by a set of six types of connections corresponding either to simple contact zones or to protein material termed linking units. When the volume is viewed along the P dyad axis (Fig. 2(b)), two neighboring HGSs of the same layer are in contact through c1 and c2 contact zones involving globin chains (i) 3B and 2B and (ii) 4B and 6A, respectively. Moreover, the contacts between the two hexagonal layers occur through six c4 connections located along the Q dyad axes. In the 45° orientation, removing one-third of the front surface (shaded zone of inset in Fig. 2(c)) allows the visualization of globin

ARTICLE IN PRESS L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

5

Fig. 2. Surface representations at a threshold displaying 100% of the EMV of A. pompejana Hb viewed in (a) top, (b) side, and (c) 45° orientations. The shaded zone of inset in (c) corresponds to the removed part of isosurface presented in (c). The black hexagon, triangle, black and white ellipses indicate the 6-fold, 3-fold, Q and P dyad axes. The nomenclature of the globin chains (1A, 1B–6A, 6B), contact zones (c1, c2), and linking units (c3a, c3b, c4, c5) is detailed in Taveau et al. (1999). The isosurfaces are calculated with SIGMA and rendered with the ray tracer PoVRay (http:// www.povray.org).

chains 1A and 1B. On both sides of mass 1B, two different elongated substructures designated as c3a and c3b linking units are visible. The central piece is connected to the 12 HGSs by 12 rod-like structures called c5 connections. 3.3. Hollow globular subassemblies The HGS of the A. pompejana Hb extracted from the whole molecule is represented with a wireframe mesh in order to reveal six high densities (numbered from 1 to 6) organized around a local 3-fold axis (white triangle in Fig. 3). Each mass is itself subdivided in two parts leading to 12 masses per HGS corresponding to 12 globin chains termed 1A, 1B, 2A, 2B, 3A, 3B, 4A, 4B, 5A, 5B, 6A, and 6B, according to the nomenclature used by Taveau et al. (1999). These globin chains are arranged in four parallel planes perpendicular to the local 3-fold axis of HGS. When looking from the top of HGS, sections 1–4 in Fig. 3 comprise respectively centers of globin chain triplets (2A, 3A, 5A), (2B, 3B, 5B), (1A, 4A, 6A), and (1B, 4B, 6B). In sections 1 and 2, the globin chains (2A, 2B), (3A, 3B), and (5A, 5B) project radially along the 3-fold axis, whereas globin chains of sections 3 and 4 project on a circle. In sections 5 and 6 carried out in the HGS bottom four linking units are found in each HGS as it was described in L. terrestris Hb (Taveau et al., 1999). The contour levels of section 5

show medium density zones corresponding to the linking units c3a, c3b, and c4. Unlike the slice 5 in A. marina Hb volume, the density area of c4 appears split in two perpendicular zones. This different density pattern is due to a strong contact with the neighboring c4 of the HGS belonging to the other hexagonal layer. However, in 3D the elongated shapes of c4 linking units of A. pompejana and of A. marina HGSs are similar. In the last section, the c5 connection is visible near the local 3-fold axis of the HGS. This rod-like structure (Fig. 3(b)) corresponds to the ends of the c3a, c3b, and c4 linking units. The comparison of the inner densities between A. pompejana and A. marina HGSs shows that the organization and orientation of the main densities are identical. By extension, since the HGSs of L. terrestris and A. marina are also similar (Jouan et al., 2001), the HGS presents a common architecture in hexagonal bilayered Hbs regardless of the hexagonal layer rotation. 3.4. Central piece In annelid Hbs, two types of central pieces are observed corresponding either to an ellipsoid or to a toroid. These two shapes are not related to the architectural types of Hbs. For example, A. marina and A. pompejana Hbs, two type-II Hbs, are respectively composed of an ellipsoid and a toroid. Top and side view orientations of the central piece extracted from the

ARTICLE IN PRESS 6

L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

Fig. 3. HGS architecture of A. pompejana Hb. (a,b) A wireframe mesh of a HGS embedding six high density masses corresponding to duets of globin chains is viewed (a) along and (b) at 90° of the local 3-fold axis of the HGS (white triangle). Six slices numbered from 1 to 6 are extracted perpendicularly to the 3-fold axis as it is shown in (b). The white lines in (b) indicate the level at which the various slices are carried out. The same six slices are also calculated in the polychaete A. marina Hb as an element of comparison.

cladogram of A. pompejana Hb are shown in Figs. 4(a) and (b). In top view (Fig. 4(a)), three isosurfaces computed at three different thresholds are superposed and viewed along the 6-fold symmetry axis of the Hb (black hexagon). At higher threshold than 100% of the EMV, the toroid of A. pompejana Hb is composed of 18 masses located in the center of the toroid and in the median plane of the molecule for six of them (termed tb) and in the prolongation of the 12 c5 linking units for the others (termed ta). The second representation of A. pompejana toroid (Fig. 4(b)) is a solid-body surface viewed along a

Q dyad axis (black ellipse) which underlines the thickness of this structure and the straightness of the spokes linking the HGS of the upper and lower hexagonal layers. These two types of representations are used to display the central pieces of L. terrestris (Figs. 4(c) and (d)) and A. marina (Figs. 4(e) and (f)) Hbs. On the one hand, at each vertex of the central piece, two spokes are visible corresponding to the c5 connections linking HGSs of the upper and lower hexagonal layers. On the other hand, the central ring of each central piece is composed of six

ARTICLE IN PRESS L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

7

nally, the two central masses eb of A. marina ellipsoid have no equivalent in the toroids. In conclusion, the curved and straight shapes of c5 connection respectively found in L. terrestris, A. pompejana and A. marina Hb seem to be directly related to type-I and type-II Hbs. Moreover, this result confirms the fact that there is no correlation between the rotation of the two hexagonal layers and the shape of the central piece. Indeed, L. terrestris Hb (a type-I Hb with 16° rotated hexagonal layers) has a toroid-like A. pompejana Hb (type-II Hb).

4. Discussion 4.1. Handedness of the A. pompejana Hb The present 3D reconstruction volume of A. pompejana Hb and that published by de Haas et al. in 1996 are enantiomers. Although the two hexagonal layers are superposed in both 3D reconstruction volumes, the organization of the various subunits is inverted. For example in Fig. 5(a), the elongated mass 6AB located in the front HGS of the upper layer (double arrow) is located on the left of the central subunit 5AB whereas mass 4AB (only the 4A is visible in this orientation) is located on the right side. In the de Haas et al. 3D reconstruction volume (Fig. 5(b)), the mass 6AB is now at

Fig. 4. Structures of central pieces of (a,b) A. pompejana, (c,d) L. terrestris, and (e,f) A. marina Hbs. Surface representations are observed in top (a,c,e) and side (b,d,f) orientations. (a,c,e) In top view, the central pieces are rendered with three superposed isosurfaces corresponding to main nodes in the cladistic exploration. (b,d,f) In side view, the Hb is viewed along a Q dyad axis and only one isosurface calculated at 100% of the EMV is visualized for sake of clarity. The internal masses are respectively termed ta and tb in the toroids (a,c), and ea and eb in the ellipsoid (e). The black and dotted arrows indicate the paths of c5 connections linking the central piece to the upper and lower hexagonal layers, respectively. The hexagons and ellipses symbolize the 6-fold and Q 2-fold axes.

masses tb in toroids of A. pompejana (Fig. 4(a)) and L. terrestris Hbs (Fig. 4(c)) and ea in A. marina ellipsoid (Fig. 4(e)). However, some discrepancies are found between these two toroids and this ellipsoid. First of all, the shape of the spoke is straight in the two polychaete Hbs whereas c5 connections of L. terrestris central piece appear curved (Figs. 4(c) and (d)). Second, the inner densities tb of A. pompejana toroid are located between two neighboring vertices like in L. terrestris Hb. Conversely, main inner densities ea of A. marina toroid are only located at the basis of c5 connections. The toroids are not analogous because in side views (Figs. 4(b) and (d)), A. pompejana toroid is flat with respect to that of L. terrestris suggesting a different inner organization. Fi-

Fig. 5. Handedness of 3D reconstruction volumes of A. pompejana Hb calculated from various reference volumes. (a) Volume presented in this paper and computed from a reference obtained by the conical tilt series method. (b) Volume published in 1996 (de Haas et al., 1996a). (c) Slice extracted perpendicularly to the local 3-fold axis of an HGS of the 1996 A. pompejana Hb volume. This slice must be compared to the slice number 3 of Fig. 3. (d) Volume calculated with a reference volume of M. decora Hb (de Haas et al., 1996b) from the image set collected in the present work. This new volume of A. pompejana Hb presents the same handedness as (b) and consequently is mirror-inverted with respect to (a). The double arrows indicate the locations of the mass 6AB. For sake of convenience, the mass 5B is encircled in black.

ARTICLE IN PRESS 8

L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

the right side of subunit 5AB. The mirror-inversion is not only limited to isosurfaces. For example, a section carried out perpendicularly to the 3-fold axis in the HGS of 1996 reconstruction volume (Fig. 5(b)) is inverted with respect to the same slice in the new volume (section number 3 of Fig. 3). A question remains to choose the good enantiomer between these two 3D reconstruction volumes. While the refinement process consisting of several cycles of 3D projection alignment and 3D reconstruction computation with the SIRT method are identical in both projects, the two reference volumes are calculated differently. In 1996, de Haas et al. used as reference the volume of M. decora Hb, a ‘‘structurally related’’ Hb which was the best resolved volume at that time. This choice was reasonable due to the great similarities between the previous series of Hb reconstructions (de Haas et al., 1996a, 1996b, 1996c, 1996d). In this paper, the reference volume is calculated from tilted/ untilted image pairs of A. pompejana Hb. This last procedure has the advantage of determining the correct handedness. In order to verify whether the choice of the reference volume is responsible of the mirror inversion, we recalculated a 3D reconstruction volume of A. pompejana Hb from the 3101 untilted image set collected for this project by using the M. decora Hb as reference. The result presented in Fig. 5(d) shows that the side view of this new volume is similar to that of 1996 (Fig. 5(b)). The M. decora Hb belongs to the type I with two 16° rotated hexagonal layers and has not the same architecture as A. pompejana Hb: a type-II Hb. The hexagonal layer rotation seems to be responsible of the computation of the enantiomeric volume because we have built a model from L. terrestris Hb volume by extracting one hexagonal layer which is then duplicated, exactly superposed, and assembled to form two hexagonal layers. This symmetrical volume is then used to calculate a new 3D reconstruction volume of A. pompejana Hb which converged to the good solution (data not shown). This experiment confirms that the 16° rotated disposition of the two hexagonal layers leads to a mirror-inverted volume. This is a good example of the influence of the reference volume on the final result converging, in this case, to a local stable minimum. 4.2. Architecture of A. pompejana Hb The exploration of A. pompejana Hb shows that the main structural features already described in L. terrestris and in A. marina Hbs are present in this architecture. The A. pompejana HGS comprises 12 globin chains leading to a total number of 144 globin chains in the whole Hb. Moreover, the c3a, c3b, and c4 connections correspond to three linking units in one HGS giving 36 linker chains per Hb. The twelve c5 connections and the central piece correspond likely to the N-terminal parts of the linker chains as described in L. terrestris Hb from

X-ray maps (Royer et al., 2000). As shown in the next section, a ‘‘modified’’ triple stranded coiled coil structure similar to that determined in L. terrestris maps can be aligned in the A. pompejana c5 connection. Thus, the model of 144 globin and of 36 linker chains observed in X-ray maps is also valid for A. pompejana Hb. However, this model does not fit with that published by Zal et al. (1997) from results of MALLS and ESI-MS experiments. The authors proposed alternative models with 144 globin and 60 linker chains or with 120 globin and 72 linker chains. This result was also found for the Hb of Paralvinella grasslei whereas the model of 144 globin and 36 linker chains is compatible with experimental molecular weights determined for Alvinella caudata and Paralvinella palmiformis Hbs (Zal et al., 2000). In the case of A. pompejana Hb, the total mass of linker chains represents 1301 kDa (35% of the 3833 kDa) yielding 51 copies of linker chains L3. From a structural point of view, this number of 51 (36 + 15) corresponds approxi
resolution, mately to a fourth linker per HGS. At 20 A there is no clue indicating the presence of this additional structure in the A. pompejana volume. Therefore, the molecular weight of A. pompejana Hb can be estimated from the weighted means of ESI-MS masses of subunits (Zal et al., 1997) for the stoichiometry of 144 globin and 36 linker chains and yields 3.427 MDa. This value corresponds to 42 680 voxels (threshold value of 0.664) in the 3D reconstruction volume versus 47 414 voxels for the 3.833 MDa molecular weight. The surfaces calculated for a threshold value of 0.664 are slightly eroded with respect to surface representations of Fig. 2. However, the differences between the surfaces are not visually detectable and the change has no consequence on the Hb architecture. The discrepancy between cryo-EM and ESI-MS data could reflect the variability of subunit composition. Zal et al. (2000) have shown that Alvinellidae Hbs contain from four to one linker chains and the composition variability is also dependent of the environment because two animals (A. marina and R. pachyptila) collected at different geographical locations have Hbs of different subunit compositions. This high variability could not be detected in this 3D reconstruction because the sample is a mixture of Hbs collected from several animals and therefore the 3D volume is an averaged structure containing features of various Hbs weighted by their frequencies. Detecting such features should only be possible if samples are collected from a single animal. According to the model of 144 (12  12) globin and 36 (3  12) linker chains, there is no reason for the presence of four different types of linker chains in A. pompejana Hb. Among Alvinellidae, P. palmiformis Hb has only one type of linker chains (Zal et al., 2000) and reassembly studies of Hbs of L. terrestris lacking one or two linker chains (Kuchumov et al., 1999; Lamy et al., 2000) have shown that the linker chains are inter-

ARTICLE IN PRESS L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

changeable. In A. pompejana Hb, according to the studies of ESI-MS (Zal et al., 1997), L4 is a minor linker chain compared to the three others. Thus, a reasonable hypothesis is that the L1, L2, and L3 linker chains are the major components in the HGS and are equally distributed. 4.3. Central piece: model of the c5 connection The main structural differences among the hexagonal bilayered Hbs are located in the central piece. Moreover,
resolution, the toroid of L. terrestris Hb is at 5.5 A composed of 12 coiled coil structures corresponding to the N-terminal ends of 36 linker chains. Therefore, understanding the two different types of shapes and the curvature of c5 connections is a step towards the comprehension of Hb assembly. Unlike the toroidal or ellipsoidal shapes of the central piece which are found in type-I and type-II Hbs, the bend of c5 connection is only observed in type-I Hbs. For example, various 3D reconstruction volumes obtained at resolution close to
present curved shapes like L. terrestris Hb (type-I 20 A Hb) and straight shapes for polychaete Hbs (type-II Hbs) like A. marina (Jouan et al., 2001), A. pompejana (this paper), and Tylorrhynchus heterochaetus (Jouan et al., in preparation). A model of c5 connection has been carried out to understand the origin of this curvature. The 3D alignment of the X-ray density maps within the L. terrestris cryo-EM data of Taveau et al. (1999) shows that the c5 connection corresponds to the triple stranded coiled coil described by Royer et al. (Fig. 6(a)). Moreover, as shown in Fig. 6(c) where the globin chains have been removed, this structure is composed of two coiled
long, respectively. The offset coil spokes of 45 and 20 A between the two coiled coils at the basis of the HGS (black arrow) could compensate the 16° rotation of the two hexagonal layers. This assumption is confirmed by the following clues. The superposition of A. pompejana Hb isosurface and Xray crystallographic dodecamer in Fig. 6(b) shows that (i) the coiled coil structure does not match the c5 isosurface and (ii) the neighboring c5 connection of the lower HGS pointed out by a 3D arrow in Fig. 6(b) cannot interact due to a too large distance. In order to improve the fitting between X-ray and cryoEM data, a model was built in two steps. First, skeletons of the triple stranded coiled coil structure were extracted and represented by rods as shown in Fig. 6(d). Second, a new model was built by removing the offset between the two parts in order to build one single triple stranded helix. The superposition of this model with A. pompejana Hb isosurface (Fig. 6(e)) gives a better result and the distance between the two c5 connections are now comparable to that presented in Fig. 6(a) for L. terrestris Hb. One may ask if this ‘‘modification’’ is related to the 16° rotation of hexagonal layers. To answer this question, models of central piece with the four different

9

combinations of curved/straight c5 shapes and 0 and 16°-rotated hexagonal layers are carried out. The genuine configurations corresponding to curved c5 and 16° rotation (L. terrestris Hb) and straight c5 and 0° rotation (A. pompejana Hb) are represented in Figs. 7(a) and (b), respectively. In Fig. 7(a), the interdigitation of the spokes of the upper (dark gray) and lower (light gray) hexagonal layers is in good agreement with the X-ray density maps. Due to the absence of layer rotation, the c5 connections in the model of A. pompejana Hb (Fig. 7(b)) appear straight and parallel when viewed in top orientation. The c5 contacts between lower and upper HGSs are similar to those observed in L. terrestris Hb. The ‘‘crossed’’ configurations (curved c5 and 0° rotation and straight c5 and 16° rotation) are unlikely due to large interpenetrations of the different spokes (Figs. 7(c) and (d)) suggesting that the bend of c5 connections is related to the hexagonal layer rotation. Although the central pieces of A. pompejana and L. terrestris Hbs are toroids, they do not present the same inner masses tb. According to the 3D reconstruction volume of L. terrestris Hb carried out by Mouche et al.
resolution, the mass tb (Fig. 4(c)) ap(2001), at 14.9 A pears as a complex organization of rod-like structures which correspond to the overlapping of N-terminal ends of linker chains belonging to neighboring HGSs. However, one may notice that the thinner appearance of A. pompejana toroid at high threshold value (Figs. 4(a) and (b)) compared to that of L. terrestris (Figs. 4(c) and (d)) means that the intricate contacts are different in A. pompejana toroid. Two hypotheses can be summarized. First,
3D reconthe presence of sinusoidal pillars in 14.9 A struction volume of L. terrestris Hb could correspond as mentioned by the authors to a ‘‘feature specific for one of the four linker’’ or a ligand. This extra-material is perhaps not present in our A. pompejana volume. Second, due to the absence of offset in the coiled coil structure (Fig. 7(a)), the contacts appear weaker in the model of A. pompejana toroid. Consequently, the overlapping of the neighboring spokes of the triple stranded coiled coil structures is less important than in L. terrestris Hb explaining why masses tb are thinner in A. pompejana volume. Thus, A. pompejana and L. terrestris toroids could correspond to different structural organizations although their shapes at
resolution appear identical. Further experiments on 20 A other annelid Hbs must be carried out to understand the implications of these differences in Hb assembly.

5. Conclusion The results presented in this paper revealed unambiguously that A. pompejana Hb belongs to the architectural type II of annelid Hbs. The HGS architecture and the locations of the various globin and linker chains are in good agreement with previous structural results of

ARTICLE IN PRESS 10

L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

Fig. 6. Models of the c5 connection. (a,b) 3D alignments of the X-ray crystallography dodecamer and cryo-EM wireframe isosurfaces of (a) L. terrestris and (b) A. pompejana Hbs. The location of the neighboring c5 connection belonging to a HGS of the other hexagonal layer is pointed out by the 3D arrow. (c) Segmentation and visualization of the triple stranded coiled coil within the wireframe surface of L. terrestris Hb. The offset between the two spokes is pointed out by the black arrow. (d) Model of the corresponding coiled coil structure. (e) Stereoscopic view of the model built from a ‘‘modified’’ triple stranded coiled coil structure. The contacts between two c5 connections are now close to each other allowing the same inner interactions described by Royer et al., for L. terrestris Hb.

3D reconstructions of annelid Hbs. Therefore, the extreme living environment of A. pompejana Hb has no specific consequence on the structure of this Hb. Moreover, the comparison of various annelid Hbs be-

longing to type II has allowed the construction of a ‘‘modified’’ model of triple stranded coil coiled structure simulating the curvature of c5 connection. This new conformation could result of primary sequence differ-

ARTICLE IN PRESS L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

11

Fig. 7. Models of central piece with different combinations of c5 connections and hexagonal layer rotation: (a) curved c5 and 16° rotation (L. terrestris Hb) (b) straight c5 and 0 ° rotation (A. pompejana Hb), (c) straight c5 and 16° rotation, (d) curved c5 and 0° rotation. The two last models (c,d) are impossible due to large interpenetrations of the different spokes. The light gray and dark gray rods are respectively connected to the lower and upper hexagonal layers.

ences or by additional constraints between c4 connections of the hexagonal layers when they are rotated by 16° in type-I Hbs.

Acknowledgments We are grateful to Dr. William E. Royer for providing phased amplitudes of the Lumbricus terrestris hemoglobin, to Professor Lallier for providing Alvinella pompejana samples and to Dr. Jean-Louis Rigaud for letting us use Philips CM120 electron microscope.

References Adrian, M., Dubochet, J., Lepault, J., McDowall, A.W., 1984. Cryoelectron microscopy of viruses. Nature 308, 32–36. Antonini, M., Chiancone, M., 1977. Assembly of multisubunit respiratory proteins. Ann. Rev. Biophys. Bioeng. 6, 239–271. Boekema, E.J., Van Heel, M., 1989. Molecular shape of Lumbricus terrestris erythrocruorin studied by electron microscopy and image analysis. Biochim. Biophys. Acta 957, 370–379. de Haas, F., Zal, F., You, V., Lallier, F., Toulmond, A., Lamy, J.N., 1996a. Three-dimensional reconstruction by cryoelectron microscopy of the giant hemoglobin of the polychaete worm Alvinella pompejana. J. Mol. Biol. 264, 111–120, doi: 10.1006/jmbi.1996.0627. de Haas, F., Boisset, N., Taveau, J.-C., Lambert, O., Vinogradov, S.N, Lamy, J.N., 1996b. Three-dimensional reconstruction of Macrobdella decora (leech) hemoglobin by cryoelectron microscopy. Biophys. J. 70, 1973–1984.

de Haas, F., Zal, F., Lallier, F., Toulmond, A., Lamy, J.N., 1996c. Three-dimensional reconstruction of the hexagonal bilayer hemoglobin of the hydrothermal vent tube worm Riftia pachyptila by cryoelectron microscopy. Proteins 26, 241–256. de Haas, F., Boisset, N., Taveau, J.-C., Lambert, O., Vinogradov, S.N., Lamy, J.N., 1996d. Three-dimensional reconstruction of the chlorocruorin of the polychaete hemoglobin Eudistylia vancouverii. J. Mol. Biol. 255, 140–163, doi: 10.1006/jmbi.1996.0012. de Haas, F., Kushumov, A., Taveau, J.-C., Boisset, N., Vinogradov, S.N., Lamy, J.N., 1997. Three-dimensional reconstruction of native and reassembled Lumbricus terrestris extracellular hemoglobin. Localization of the monomeric globin chains. Biochemistry 36, 7330–7338. Dubochet, J., Adrian, M., Chang, J.-J., Homo, J.-C., Lepault, J., McDowall, A.W., Schultz, P., 1988. Cryoelectron microscopy of vitrified specimens. Q. Rev. Biophys. 21, 129–228. Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y., Ladjadj, M., Leith, A., 1996. Spider and web: processing and visualization of images in 3D electron microscopy and related fields. J. Struct. Biol. 116, 190–199, doi: 10.1006/jsbi.1996.0030. Harauz, G., Van Heel, M., 1986. Exact filters for general geometry three-dimensional reconstruction. Optik 73, 146–156. Henderson, R., 1992. Image contrast in high-resolution electron microscopy of biological macromolecules. TMV in ice. Ultramicroscopy 46, 1–10. Jouan, L., Taveau, J.-C., Marco, S., Lallier, F., Lamy, J.N., 2001. Occurence of two architectural types of hexagonal bilayer hemoglobin in annelids: comparison of 3D reconstruction volumes of Arenicola marina and Lumbricus terrestris hemoglobins. J. Mol. Biol. 305, 757–771, doi: 10.1006/jmbi.2000.4344. Kuchumov, A.R., Taveau, J.C., Lamy, J.N., Wall, J.S., Weber, R.E., Vinogradov, S.N., 1999. The role of linkers in the reassembly of the 3.6 MDa hexagonal bilayer hemoglobin from Lumbricus terrestris. J. Mol. Biol. 289, 1361–1374.

ARTICLE IN PRESS 12

L. Jouan et al. / Journal of Structural Biology xxx (2003) xxx–xxx

Lamy, J., Kuchumov, A., Taveau, J.C., Vinogradov, S.N., Lamy, J.N., 2000. Reassembly of Lumbricus terrestris hemoglobin: a study by matrix-assisted laser desorption/ionization mass spectrometry and 3D reconstruction from frozen-hydrated specimens. J. Mol. Biol. 298, 633–647. Lamy, J.N., Green, B.N., Toulmond, A., Wall, J.S., Weber, R.E., Vinogradov, S.N., 1996. Giant hexagonal bilayer hemoglobins. Chem. Rev. 96, 3113–3124. Lanzavecchia, S., Wade, R.H., Ghiretti Magaldi, A., Tognon, G., Bellon, P.L., 1999. A two-exposure technique for ice-embedded samples successfully reconstructs the chlorocruorin pigment of Sabella spallanzanii at 2.1 nm resolution. J. Struct. Biol. 127, 53–63, doi: 10.1006/jsbi.1999.4142. Lazarus, F., Verroust, A., 1999. Level set diagrams of polyhedral objects. ACM Solid ModelingÕ99, June 1999, Ann-Arbor, MI, USA. Lebart, L., Morineau, A., Fenelon, J.-P., 1982. Traitement des donnees statistiques methodes et programmes. Dunod. Marabini, R., Masegosa, I.M., San Martın, M.C., Marco, S., Fern andez, J.J., de la Fraga, L.G., Vaquerizo, C., Carazo, J.M., 1996. Xmipp: an image processing package for electron microscopy. J. Struct. Biol. 116 (1), 237–241. Milligan, R.A., Brisson, A., Unwin, P.N.T., 1984. Molecular structure determination of crystalline specimens in frozen aqueous solutions. Ultramicroscopy 13, 1–10. Mouche, F., Boisset, N., Penczek, P.A., 2001. Lumbricus terrestris hemoglobin—the architecture of linker chains and structural variation of the central toroid. J. Struct. Biol. 133, 176–192. Orlova, E.V., Dube, P., Harris, J.R., Beckman, E., Zemlin, E., Markl, J., Van Heel, M., 1997. Structure of keyhole limpet hemocyanin
resolution by electron cryomicroscopy and type 1 (KLH1) at 15 A angular reconstruction. J. Mol. Biol. 271, 417–437, doi: 10.1006/ jmbi.1997.1182. Penczek, P., Radermacher, M., Franck, J., 1992. Three-dimensional reconstruction of single particles embedded in ice. Ultramicroscopy 40, 33–53. Penczek, P.A., Grassuci, R.A., Franck, J., 1994. The ribosome at improved resolution/new technique for merging and orientation refinement in 3D cryoelectron microscopy of biological particles. Ultramicroscopy 53, 251–270. Penczek, P., 1998. Appendix: measures of resolution using Fourier shell correlation. J. Mol. Biol. 280, 115–116. Radermacher, M., 1988. Three-dimensional reconstruction of single particles from random and nonrandom tilt series. J. Electron Microsc. Tech. 9, 359–394. Radermacher, M., Wagenknecht, T., Verschoor, A., Frank, J., 1987a. Three-dimensional reconstruction from a single-exposure, random

conical tilt series applied to the 50S ribosomal subunit of Escherichia coli. J. Microsc. 146, 113–136. Radermacher, M., Wagenknecht, T., Verschoor, A., Frank, J., 1987b. Three-dimensional structure of the large ribosomal subunit from Escherichia coli. EMBO J. 6, 1107–1114. Royer, W.E., Strand, K., van Heel, M., Hendrickson, W.A., 2000. Structural hierarchy in erythrocruorin, the giant respiratory assemblage of annelids. PNAS 97, 7107–7111. Saxton, W.O., Baumeister, W., 1982. The correlation averaging of a regularly arranged bacterial cell envelope protein. J. Microsc. 127, 127–128. Schatz, M., Orlova, E.V., Dube, P., J€ager, J., van Heel, M., 1995.
resolution Structure of Lumbricus terrestris hemoglobin at 30 A determined using angular reconstitution. J. Struct. Biol. 114, 28–40. Taveau, J.-C., 1996. Presentation of the SIGMA software: software of imagery and graphics for molecular architecture. J. Struct. Biol. 116, 223–229, doi: 10.1006/jsbi.1996.0034. Taveau, J.-C., Boisset, N., Vinogradov, S.N., Lamy, J.N., 1999. Threedimensional reconstruction of Lumbricus terrestris hemoglobin at
resolution: intramolecular localization of the globin and linker 22 A chains. J. Mol. Biol. 289, 1343–1359, doi: 10.1006/jmbi.1999.2824. Van Heel, M., 1984. Multivariate statistical classification of noisy images (randomly oriented biological macromolecules). Ultramicroscopy 21, 164–184. Vinogradov, S.N., 1985. The structure of invertebrate extracellular hemoglobins (erythrocruorins and chlorocruorins). Comp. Biochem. Physiol. B 82 (1), 1–15. Vinogradov, S.N., Kapp, O.H., Ohtsuki, L., 1982. The extracellular hemoglobins and chlorocruorins of annelids. In: Harris, J. (Ed.), Electron Microscopy of Proteins, 3. Academic Press, New York, pp. 135–163. Vinogradov, S.N., Sharma, P.K., Walz, D.A., 1991. Iron and heme contents of extracellular hemoglobins and chlorocruorins of annelids. Comp. Biochem. Physiol. B 98 (2-3), 187–194. Ward Jr., J.H., 1963. Hierarchical grouping to optimize an objective function. J. Am. Stat. Assoc. 58, 236–244. Zal, F., Green, N., Lallier, F.H., Toulmond, A., 1997. Investigation by electrospray ionization mass spectrometry of the extracellular hemoglobin from the polychaete annelid Alvinella pompejana: an unusual hexagonal bilayer hemoglobin. Biochemistry 36, 11777– 11786. Zal, F., Green, B.N., Martineu, P., Lallier, F.H., Toulmond, A., Vinogradov, S.N., Childress, J.J., 2000. Polypeptide chain composition diversity of hexagonal-bilayer haemoglobins within a single family of annelids, the alvinellidae. Eur. J. Biochem. 267 (16), 5227–5236.