|Home | About | Journals | Submit | Contact Us | Français|
The allosteric respiratory protein hemocyanin occurs in gastropods as tubular di-, tri- and multimers of a 35 × 18 nm, ring-like decamer with a collar complex at one opening. The decamer comprises five subunit dimers. The subunit, a 400 kDa polypeptide, is a concatenation of eight paralogous functional units. Their exact topology within the quaternary structure has recently been solved by 3D electron microscopy, providing a molecular model of an entire didecamer (two conjoined decamers). Here we study keyhole limpet hemocyanin (KLH2) tridecamers to unravel the exact association mode of the third decamer. Moreover, we introduce and describe a more complex type of hemocyanin tridecamer discovered in fresh/brackish-water cerithioid snails (Leptoxis, Melanoides, Terebralia).
The "typical" KLH2 tridecamer is partially hollow, whereas the cerithioid tridecamer is almost completely filled with material; it was therefore termed "mega-hemocyanin". In both types, the staggering angle between adjoining decamers is 36°. The cerithioid tridecamer comprises two typical decamers based on the canonical 400 kDa subunit, flanking a central "mega-decamer" composed of ten unique ~550 kDa subunits. The additional ~150 kDa per subunit substantially enlarge the internal collar complex. Preliminary oxygen binding measurements indicate a moderate hemocyanin oxygen affinity in Leptoxis (p50 ~9 mmHg), and a very high affinity in Melanoides (~3 mmHg) and Terebralia (~2 mmHg). Species-specific and individual variation in the proportions of the two subunit types was also observed, leading to differences in the oligomeric states found in the hemolymph.
In cerithioid hemocyanin tridecamers ("mega-hemocyanin") the collar complex of the central decamer is substantially enlarged and modified. The preliminary O2 binding curves indicate that there are species-specific functional differences in the cerithioid mega-hemocyanins which might reflect different physiological tolerances of these gill-breathing animals. The observed differential expression of the two subunit types of mega-hemocyanin might allow individual respiratory acclimatization. We hypothesize that mega-hemocyanin is a key character supporting the adaptive radiation and invasive capacity of cerithioid snails.
The most urgent physiological need of animals is a constant supply of oxygen, usually provided by allosteric respiratory proteins such as the red, iron-based protein hemoglobin. In most molluscs, this crucial role is played by the blue, copper-containing protein hemocyanin. Its basic hemocyanin oligomer, the decamer, is composed of five identical subunit dimers forming a cylinder with a collar complex at one end [1,2] (Figure (Figure1).1). Gastropods express a didecamer assembled from two conjoined decamers, with the collar complexes facing outward at the ends. The subunit, a ~400 kDa polypeptide, is a concatenation of eight paralogous functional units termed FU-a to FU-h, each with a single copper active site for reversible oxygen binding . The exact topology of the functional units within the didecamer has only recently been established through 3D electron microscopy of keyhole limpet hemocyanin isoform 1 . Keyhole limpet hemocyanin (KLH) is an established immune response modifier and hapten carrier [4,5]. The two KLH isoforms (termed KLH1 and KLH2) are homo-oligomers, each assembled from a single subunit type of ~400 kDa. KLH1 forms didecamers and didecamer clusters, whereas KLH2 forms didecamers and multidecamers. The two subunit types differ substantially in primary structure and most probably furcated by a gene duplication event some 340 million years ago (for literature, see [2-5]).
The function of gastropod hemocyanins is characterized by moderate cooperativity, Bohr effect and oxygen affinity; high oxygen affinities as required under hypoxic conditions are not observed . In warm hypoxic ponds this might limit the diving time of lung breathers such as Lymnea . The lung breathing planorbid snails (e.g. Biomphalaria) evolved an intriguing respiratory alternative for prolonged diving, namely a multimeric hemoglobin with high oxygen affinity [7,8]. The convergent occurrence of hemoglobins in some bivalves that thrive under hypoxic conditions [see references in ] also suggests that molluscan hemocyanin has been unable to evolve into high affinity forms, needed especially by gill breathers for survival under hypoxic conditions. We have now found in the Cerithioidea a unique complex "mega-hemocyanin" tridecamer which is probably capable of adapting to a much broader range of oxygen affinities than typical hemocyanin. Cerithioid snails (Figure (Figure2)2) span a broad range of aquatic habitats, where they are often important components of the respective communities [9,10]. The present study provides the first structural and a preliminary functional analysis of this newly detected hemocyanin type.
Typical gastropod hemocyanin didecamers tend to bind additional decamers at one or both ends, thereby forming tubular tridecamers and larger multidecamers; a well-studied example is KLH2 . However, the exact association mode of the additional decamers remained unknown and therefore was elucidated in the present context, in order to better understand the unique structure of the novel mega-hemocyanin tridecamer.
In the electron microscope, purified KLH2 shows the presence of di-, tri- and multidecamers (Figure (Figure3a),3a), which confirms earlier observations [11-15]. We enriched the KLH2 tridecamer fraction by gel filtration chromatography and produced, by electron microscopy of negatively stained molecules and single particle analysis, a 30-Å reconstruction (Figure 4a, a', b, b'). Although this resolution was low, it reproducibly allowed automated rigid-body fitting of three copies of the molecular model of the KLH1 decamer (Figure 4c, c', d, d'); this model was available from a previous study . The results clearly show that within the tridecamer, the additional decamer is rotated 36° with respect to the didecamer. This fits previous electron microscopical observations . Indeed, it can be demonstrated using several plaster 3D models of the KLH1 decamer, obtained by rapid prototyping of the published 9-Å volume , that a rotation angle of 36° joins them in the most stable manner. It should be noted that 36° is also the staggering angle of the two constituent decamers within the "nucleating" didecamer of KLH1 , although in this case the two decamers are assembled with their "open" ends facing each other, and their closed ends (with the collar complex) facing outwards. In contrast, in tri- and multidecamers the additional decamers are attached with their open end to a closed end (Figure (Figure5;5; see also ).
Molluscan hemocyanin decamers are assembled from five oblique subunit dimers, resulting in five major and five minor right-handed helical grooves encarving the surface of the cylinder wall. Due to the decamer↔decamer staggering angle of 36°, these grooves are steadily continued in the didecamer (see Figure Figure4)4) which electron microscopists observed several decades ago [17,18]. The recent 9-Å 3D structure of KLH1 ultimately confirmed this for the didecamer , and the present results show, for the first time, steady continuation of the helical grooves in the tridecamer. From spatial constraints it is clear that this association mode is also true for the additional decamers in the larger KLH2 multidecamers, as illustrated in Figure Figure55 for a pentadecamer. In Figure 6a-d, representations of the KLH2 tridecamer and smaller oligomers derived from electron microscopical images are shown, together with a 3D model of the basic decamer (Figure (Figure6e6e).
Leptoxis carinata (Pleuroceridae) is a small North American freshwater snail populating rocks in shallow, well-aerated water . During a routine electron microscopic survey of hemocyanin samples from this snail we detected exclusively tridecamers (Figure (Figure3b),3b), which is unprecedented for a molluscan hemocyanin. The typical hemocyanin tridecamer observed in KLH2 (see Figure Figure3a)3a) and many other molluscs (e.g. [11-13,16]) is a transient state between the comparatively stable didecamers and larger multidecamers; it is therefore generally present as a minor component. In previous careful dissociation/reassembly studies, conditions that would transform the bulk of material into tridecamers were not found [13-15]. Therefore, it was completely unexpected to find here an exclusive population of fully stable tridecamers.
Moreover, these tridecamers looked abnormal. In contrast to the partially hollow hemocyanin tridecamers of other molluscs (see Figure 6a, b), the centre of these tridecamers is almost completely filled with material (Figure 6a', b'). Partial dissociation of the Leptoxis tridecamers induced by prolonged storage revealed two unique sub-structures: a didecamer with a single mass extension (Figure 6c'), and a decamer with two extensions (Figure 6d'). Typical decamers were also detected. 3D electron microscopy of the Leptoxis tridecamer under cryo conditions yielded a ~13-Å reconstruction (Figure (Figure7).7). It revealed that this "mega-hemocyanin" is actually composed of two different units: the flanking typical decamer, with its collar complex shifted towards one cylinder opening (point-group C5 symmetry, see Figure Figure6e),6e), and the central "mega-decamer" with a much larger and more symmetrical collar complex (probably point-group D5 symmetry; Figure 6e'). The difference between a typical hemocyanin tridecamer and mega-hemocyanin is illustrated in Figure Figure88.
In the typical gastropod hemocyanin collar complex, five FU-g pairs form five hidden "arcs" protruding into the cylinder lumen. Moreover, five FU-h pairs build five exposed "slabs"; the latter are arranged as a flat ring (Figure 8a, a'). In spite of the fully anti-parallel subunit arrangement in the wall, the whole collar complex is shifted towards one end of the cylinder (see Figure Figure1).1). This has been explained on the basis of the subunits forming an asymmetric homo-dimer , and it is amazing that in mega-hemocyanin this apparently has turned into a more symmetric homo-dimer (Figure 8b, b'). However, such a mutation is not unique, but has a parallel in cephalopod hemocyanins. The latter always occur as single decamers, with FU-h missing [20,21]. In Nautilus hemocyanin, the wall architecture of this decamer and the asymmetric position of FU-g pairs forming the collar are indistinguishable from the situation in KLH1 [1,2,22]. However, in Sepia hemocyanin an additional FU, resulting from gene duplication of FU-d, also contributes to the collar complex by substantially enlarging it and forcing it into a more symmetrical structure .
Mutations of the collar complexes in cerithioid and sepioid hemocyanins clearly result from convergent evolution, but they demonstrate that such rearrangements of the collar are possible without destabilizing the hemocyanin particle in the intermediate phases. This is due to inherent stability of the isolated wall which has been demonstrated in cases in which the collar complex is lacking .
In addition to the typical ~400 kDa subunit, SDS-PAGE of Leptoxis carinata hemocyanin shows a previously unknown, significantly larger polypeptide (Figure (Figure9a).9a). A comparable pattern was obtained from Leptoxis praerosa hemocyanin (not shown) and Terebralia palustris hemocyanin (Figure (Figure9b).9b). Terebralia palustris (Potamididae) is a mangrove snail from Kenya. Biophysical molecular mass determination of molluscan hemocyanin subunits is generally difficult due to the enormous size of the subunits and the lack of marker proteins in this range, and therefore a matter of debate (see [1,6]). In the past we applied a variety of methods [12,13] and ultimately found, as soon as "true" molecular masses predicted from primary structures became available, that at least in our hands SDS-PAGE yielded the most accurate results (e.g. [3,21]). Molluscan hemocyanins are glycosylated, but their glycan content is generally not high enough to cause irregular migration on SDS gels. For example, in case of Haliotis tuberculata hemocyanin, ~400 kDa was determined by SDS-PAGE, and 392 kDa was later predicted from the amino acid sequence; from their total yield, the glycans might well account for the balance of ca. 8 kDa . From these previous studies we had available various well-characterized molluscan hemocyanin subunits and subunit fragments that could be used in SDS-PAGE as molecular mass markers of appropriate size. Using KLH1 (400 kDa, see Figure Figure9a)9a) as the largest marker, we determined ~540 kDa for the larger mega-hemocyanin subunit; however, critically, the larger subunit was determined by graphical extrapolation (Figure 10a).
In previous studies on the typical molluscan hemocyanin subunits, limited proteolytic cleavage (by using low amounts of proteases) proved to be a strong tool in analyzing the functional unit concatenation. Under the appropriate conditions, certain proteases cleave one or several of the exposed, elongated linkers between functional units, whereas the globular core of the functional units remains intact. In SDS-PAGE this yields well-defined subunit fragments containing one, two or several functional units (e.g. [12,13]). Using this approach on Terebralia hemocyanin, we obtained a very complex cleavage pattern when exposing the subunit mixture to 5% trypsin (not shown). However, if 5% trypsin was applied to the intact oligomers, after subsequent denaturation a clear pattern was revealed in SDS-PAGE: The 400 kDa subunit was still intact, but the larger subunit yielded two well defined cleavage products (Figure (Figure9c),9c), with 370 ± 10 and 180 ± 10 kDa, respectively (Figure 10a, b); the process was still incomplete after 1 hour but complete after 3 hours. Together, the two fragments represent ~550 kDa, which is taken here for the apparent molecular mass of the intact subunit, as we still lack the amino acid sequence. The stability of the typical hemocyanin decamers of Terebralia to this protease treatment conforms with our previous observations on keyhole limpet and Haliotis hemocyanin. In contrast, the ~550 kDa subunit apparently has a tryptic cleavage site that is freely accessible for the enzyme even in the native quaternary structure.
In all Leptoxis individuals studied thus far (n = 51), the ratio of the 400 kDa to 550 kDa subunit is ca. 2:1 (see Figure Figure9a),9a), suggesting that this is indeed the correct ratio within the mega-tridecamer. Interestingly, in the studied Terebralia individuals (n = 13), the ratio of the 400 kDa subunit was higher than in Leptoxis, and was further increased in samples enriched in didecamers via gel filtration chromatography (see Figure Figure9b).9b). The electron microscope showed typical didecamers in addition to mega-tridecamers (Figure (Figure3c).3c). In Melanoides tuberculata (Thiaridae), variable proportions of the two subunits are present, but in all studied individuals (n = 22) with a clear excess of the 550 kDa subunit (Figure (Figure9d).9d). Electron microscopy revealed a variable mixture of mega-hemocyanin oligomers (Figure (Figure3d).3d). Some individuals even lack the 400 kDa subunit and possess mega-decamers exclusively in their hemolymph. The two subunit types are immunologically distinct (Figure 9e, f), which usually corresponds to major differences (>30%) in primary structure. Reassembly experiments with isolated Melanoides subunits confirmed that the 550 kDa subunit exclusively forms mega-decamers (Figure (Figure9g),9g), whereas the 400 kDa subunit produces typical decamers which further assemble into typical di-, tri- and multidecamers (Figure (Figure9h9h).
The presence of two or more different subunit types within the same hemocyanin particle is usual in arthropods [24,25], but rarely observed in molluscs. In the latter, hemocyanin heterogeneity is obvious at the level of the functional units [3,12,13,20]. In keyhole limpet hemocyanin and Haliotis tuberculata hemocyanin, it has been demonstrated by many reassembly experiments that the two 400 kDa subunit isoforms are unable to co-polymerize into hetero-oligomers: reassembly attempts from subunit mixtures always resulted in distinct homo-oligomers which fits the observation that the native hemocyanin is devoid of hetero-oligomers [14,26]. The only previously published example of a mollusc with hetero-oligomeric hemocyanin is Concholepas, but this animal has a typical hemocyanin decamer made from two types of 400 kDa subunit . A situation such as we have now detected in cerithioid snails has not previously been reported.
In contrast to other gastropod hemocyanins , oxygen binding curves of whole hemocyanin from Leptoxis praerosa, Melanoides and Terebralia all indicate low cooperativity (Hill coefficient of 1.0-1.2 at half-saturation). Oxygen affinity (p50) is moderate in Leptoxis (mega-tridecamers: 9 ± 1 mmHg at pH 7.6; 1 mmHg ~133 Pa), but very high in Melanoides (mega-decamers: 2.6 ± 0.5 mmHg at pH 7.6) and even higher in Terebralia (mostly mega-tridecamers and some typical didecamers: 1.8 ± 0.5 mmHg at pH 7.6). These data are preliminary, but seem to reflect the different physiological tolerances of these gill-breathing species: Leptoxis is stenotrophic and requires swift-flowing, well-oxygenated waters; in contrast, Melanoides is found in quiet waters and can tolerate stagnant, eutrophic conditions, and Terebralia thrives in warm mangrove mud. In these cases, the observed difference in p50 indicates an evolutionary adaptation, and could mean the difference between survival or death.
Planorbid freshwater snails such as Biomphalaria glabrata are lung breathers, adapted to prolonged diving in warm ponds; they switched from "blue to red blood" by almost exclusively expressing a multimeric hemoglobin with p50 ~6 mmHg [7,8]. It has been suggested that hemocyanin is probably unable to evolve into high affinity oxygen-binding forms needed under conditions of hypoxia . However, mega-hemocyanin seems to have this capacity and possibly has the added advantage of allowing respiratory acclimatization through differential expression of the 400 kDa and 550 kDa subunit types. It also seems to us that the functional plasticity of mega-hemocyanin, notably its ability to function properly under hypoxic as well as normoxic conditions, exceeds that of typical hemocyanin and also of multimeric hemoglobin.
Our previous 9-Å structure of the KLH1 didecamer essentially solved the overall architecture, intricate subunit pathway and near-atomic structure of a gastropod hemocyanin decamer and didecamer . The present 3D model of the KLH2 tridecamer (see Figure Figure4)4) and its extrapolation to larger KLH2 multidecamers (see Figure Figure5)5) shows that within all types of KLH molecule, the constituent decamers are rotated 36° with respect to their adjoining decamers, thereby yielding a steady continuation of the right-handed helical grooves characteristic of the cylinder outer wall. As deduced from multiple sequence alignments and biochemical analyses [21,29,30], the different association forms of KLH can serve as structural models for gastropod (and bivalve) typical hemocyanin in general.
The newly detected mega-hemocyanin, apparently a synapomorphy of the Cerithioidea, can be interpreted on the basis of its present 13-Å cryoEM structure and our detailed knowledge on KLH. We provide strong evidence that the mega-tridecamer is assembled from two peripheral typical decamers (each with ten copies of the 400 kDa subunit), and a central mega-decamer made up from ten 550 kDa subunits. The wall of the mega-decamer is apparently constructed according to the typical scheme, which means that the six wall functional units FU-a to FU-f that exist in the 400 kDa subunit are also present in the 550 kDa subunit.
In contrast, the collar complex of the mega-decamer remains unresolved (shown in red in Figures Figures77 and and8).8). It is formed by ten copies of the remaining segment of the 550 kDa subunit, encompassing ca. 250 kDa, respectively ca. 2000 amino acids. This might correspond to four or five functional units, instead of only two (FU-g and FU-h) in the collar complex of typical hemocyanin. In total, this newly defined collar complex has a molecular mass of 2.5 Megadalton. Whether or not it shows a true D5 symmetry and how the 250 kDa segment is folded requires its full-length sequence and a higher resolution 3D reconstruction.
In view of the wealth of comparative data available on molluscan hemocyanins (for literature, see ), the discovery of a substantially modified type such as mega-hemocyanin, and moreover in some of the most abundant snails, was completely unexpected. The 400 kDa subunit has existed for ~740 million years  and is wide-spread among the Mollusca [30,32]. The 550 kDa subunit with its isolated occurrence, probably only in the superfamily Cerithioidea  which appeared in the Paleozoic, is a more recent evolutionary specialization. The unique 550 kDa hemocyanin subunit is one of the largest polypeptides ever reported, second only to the myoelastic titin family .
Cerithioideans are important members of many freshwater communities spanning a broad range of habitats , and might be highly suitable for providing fundamental insights into the mechanisms that generate biodiversity, pattern in historical biogeography, and the underlying processes of speciation and radiation [34,35]. We hypothesize that mega-hemocyanin may have been a key character in their success. From our preliminary O2-binding data it appears that Terebralia and Melanoides hemocyanin have unusually high oxygen affinities. If this holds true in future more detailed analyses, mega-hemocyanin might allow Terebralia to populate hypoxic mangrove muds, and presumably has contributed to the invasive capacity of Melanoides which thrives in disturbed habitats and has been introduced worldwide through the aquarium trade.
Leptoxis carinata and Leptoxis praerosa (Gastropoda, Pleuroceridae) samples were collected in shallow water on rocks at a single location in the Potomac River, Maryland, USA, and at a single location in Cypress Creek, Alabama, USA, respectively. Both species were collected in spring 2009, L. carinata additionally in November 2008 and summer 2009. Melanoides tuberculata samples were taken throughout 2009 from two independently established freshwater aquaria in the Institute of Zoology, Mainz, Germany. Terebralia palustris was collected in summer 2009 in a mangrove mud near Gazi, Kenya. The keyhole limpet Megathura crenulata, source of KLH, was kindly provided by the biosyn company, Fellbach, Germany. Hemolymph from the following numbers of snails has been studied individually: L. carinata: 51 (SDS-PAGE), 12 (EM); T. palustris: 13 (SDS-PAGE), 8 (EM); 3 (O2 binding); M. tuberculata: 22 (SDS-PAGE), 10 (EM), 3 (O2 binding). L. praerosa: 5 (SDS-PAGE), 3 (EM), 1 (O2 binding). In addition, we studied pooled hemocyanin from 4 L. praerosa individuals in all three categories.
Generally, the living animals were transported to the laboratory in Mainz for bleeding, and the hemolymph further processed immediately after bleeding. Some individuals were bled immediately after appearing in our laboratory, others were kept in aquaria and bled days or weeks later. Hemolymph was withdrawn by puncturing the foot muscle by a syringe. The following procedures were as described previously [12,13]: Blood cells were removed by centrifugation and the hemocyanin was purified from the supernatant by pelleting it in a Beckman airfuge. The initial hemocyanin concentration in the animals was 10-25 mg/ml. Purified hemocyanin was routinely stored at 4°C in "stabilizing buffer" (50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 5 mM CaCl2, 5 mM MgCl2). Pure KLH2 was obtained from an animal that lacked KLH1. Different hemocyanin oligomers were separated by gel filtration chromatography using BioGel A15m. The individual Melanoides hemocyanin subunits were separated by cutting the respective protein bands from the native PAGE gel, followed by elution. Alternatively, the 550 kDa subunit was obtained from animals that lacked the 400 kDa component. Hemocyanin oligomers were dissociated into native subunits by dialysis overnight against 0.1 M glycine/NaOH buffer of pH 9.6. Reassembly experiments were performed by overnight dialysis of subunits versus "stabilizing buffer", containing 100 mM CaCl2 and 100 mM MgCl2.
SDS polyacrylamide gel electrophoresis (SDS-PAGE) was routinely performed with 7.5% acrylamide in the system of Laemmli  as described [12,13]. For limited tryptic cleavage, native Terebralia hemocyanin (see Figure Figure3c)3c) dissolved in stabilizing buffer at a final protein concentration of 16 mg/ml was treated for 1, 2, and 3 hours at 37°C with 5% trypsin (Sigma, Taufkirchen, Germany), and then immediately SDS-denatured to inactivate the protease. The resulting fragments were subsequently analysed by SDS-PAGE. The molecular mass of mega-hemocyanin subunits and fragments was graphically determined according to their migration distances in SDS-PAGE, using several molluscan hemocyanins and a commercial protein standard mixture as markers (for details, see Figure Figure10);10); 5%, 6% and 7.5% acrylamide concentrations were applied in this case. Crossed and tandem-crossed immunoelectrophoresis with the system of Weeke  was performed as described [12,13], using rabbit anti-Melanoides hemocyanin antibodies for the second dimension.
Negative staining electron microscopy (using 2% uranyl acetate) was done as described . Rapid freezing/vitrification sample grid preparation for cryo-electron microscopy under oxygen-rich conditions (25% oxygen/75% nitrogen) was done as published . Negatively stained specimens were studied with a FEI Tecnai 12 transmission electron microscope at an accelerating voltage of 120 kV. Images were taken with a 1392 × 1040 SIS Megaview camera. Cryo-electron microscopy was done with a FEI Polara transmission electron microscope, operating at an accelerating voltage of 300 kV and 59,000 × instrumental magnification. The negatives (Kodak SO 163 film) were developed for 12 minutes in full-strength Kodak D19 developer.
76 cryoEM micrographs of Leptoxis carinata hemocyanin (LcH) were digitised using a PRIMESCAN rotating drum scanner with a sampling size of 1-Å at the specimen level. The contrast transfer function (CTF) for each image was determined using the program CTFFIND3 . 850 particles were manually picked with the module BOXER from the EMAN software package 1.7 , and extracted into 736 × 736 pixel boxes. The cut-out particles were corrected for the phase reversals of the CTF in IMAGIC , and, prior to band pass filtering and normalization, 2 × 2 averaged to give a final sampling of 2-Å/pixel.
Image processing was performed mainly with the IMAGIC software package. In case of Leptoxis carinata mega-hemocyanin, initial centring, classification and subsequent refinement of the class averages was done as described . The initial map was calculated from class averages by angular reconstitution with imposed D5 symmetry and then refined using projection matching with an initial angular increment of 5° until there was no significant change in Euler angle assignments between successive refinement rounds. Subsequent refinement rounds were performed with decreasing angular steps (4°, 3°, 2° and 1°) until no further improvement could be detected. The final map was obtained from 627 superior particles, representing exclusively side-views of the tridecamer. The resolution of the final reconstruction was 13-Å according to the 1/2-bit criterion of the Fourier shell correlation .
In case of the KLH2 tridecamer, 97 negative staining EM micrographs were digitised with a sampling size of 2-Å and particles were manually picked within 368 × 368 boxes. A total of 110 single particles representing side views were manually collected. The final 30-Å (FSC1/2-bit) density map was calculated from 29 class-averages by angular reconstitution with imposed C5 symmetry.
Automated rigid-body fitting of the molecular model of the KLH1 decamer  into the 30-Å cryoEM structure of the KLH2 tridecamer and the LcH tridecamer, respectively, was done in UCSF CHIMERA-1.3 , using the tool "fit in map". This was straightforward even in the 30-Å structure, because the right-handed helical wall features and the major elements of the collar complexes were clearly defined. The tool "color zone/split map" of the CHIMERA software was applied for segmentation of the maps; this procedure is based on the fitting. The molecular graphics images were also produced using CHIMERA.
Continuous oxygen binding equilibrium curves for determination of oxygen affinity (p50) and cooperativity (Hill coefficient) of hemocyanin were recorded by the fluorimetric polarographic method . The change in fluorescence as indicator of the oxygen saturation level was measured with a standard fluorimeter (Hitachi F4500, Binninger Analytic, Germany) at an emission wavelength of 338 nm (excitation 280 nm) while the oxygen concentration was determined simultaneously with an oxygen electrode (Microelectrodes Inc., Bedford, USA) equipped with a locally made amplifier. This electrode allowed determination of oxygen partial pressure to a precision of ca. 1 mmHg. All experiments were performed at 20 ± 0.05°C at a protein concentration of 0.2 mg/ml in stabilizing buffer.
CTF: contrast transfer function; FSC: Fourier shell correlation; FU: functional unit; KLH: keyhole limpet hemocyanin (from Megathura crenulata); LcH: Leptoxis carinata hemocyanin; MtH: Melanoides tuberculata hemocyanin; PAGE: polyacrylamide gel electrophoresis; TpH: Terebralia palustris hemocyanin
The authors declare that they have no competing interests.
MGH, EES and BL provided Leptoxis and information on cerithioid snails. NH collected Terebralia. BL initiated the first electron microscopical images leading to the detection of mega-hemocyanin. BL, WG, CG, FD, NH and JM designed experiments and interpreted data. WG, CG, FD and NH performed the experiments. JM coordinated the experiments, produced the graphics and wrote the manuscript. All authors commented and approved the manuscript.
We thank Dr Jeffrey T. Garner (Alabama Fish & Game Division) for the samples of Leptoxis praerosa, Dr J. Robin Harris for critical reading of the manuscript, Christoph Kühne for the Terebralia images, Prof. Dr Werner Kühlbrandt and Deryck Mills (MPI Biophysics, Frankfurt, Germany) for kindly providing access to the FEI Polara microscope and the Deutsche Forschungsgemeinschaft (DFG-GK 1043 grant to JM) for financial support.