Polypeptide chain composition diversity of hexagonal-bilayer haemoglobins within a single family of annelids, the Alvinellidae

Authors


F. Zal, Equipe Ecophysiologie, UPMC–CNRS–INSU, Station Biologique, BP 74, 29682 Roscoff cedex, France. Tel.: + 33 2 98 29 23 09, Fax: + 33 2 98 29 23 24,E-mail: zal@sb-roscoff.fr

Abstract

Following previous analysis of the structure of Alvinella pompejana heaxagonal-bilayer haemoglobin (HBL Hb) [1], we report in this paper the structure of three other HBL Hbs belonging to Alvinella caudata, Paralvinella grasslei and Paralvinella palmiformis, members of the Alvinellidae, annelid family strictly endemic to deep-sea hydrothermal vents located on the ridge crests in the Pacific ocean. The multi-angle laser light scattering (MALLS) and fast protein liquid chromatography (FPLC) analysis revealed a broad range of molecular masses for the extracellular Hb molecules, 3517 ± 14 kDa (A. caudata), 3822 ± 28 kDa (P. grasslei) and 3750 ± 150 kDa (P. palmiformis). Native and derivative Hbs (reduced, carbamidomethylated and deglycosylated) were analysed by electrospray ionization mass spectroscopy (ESI-MS) and the data was processed by the maximum entropy deconvolution system (MaxEnt). The most important difference between alvinellid HBL Hbs was the variation in their composition, from two to four monomeric globin chains, and from one to four linker chains. Therefore, despite the fact that all these species belong to a single family, notable differences in the polypeptide chain composition of their HBL Hbs were observed, probably accounting for their different functional properties as previously reported by this group Toulmond, A., El Idrissi Slitine, F., De Frescheville, J. & Jouin, C. (1990) Biol. Bull.179, 366–373.

Abbreviations
HBL

hexagonal-bilayer

MALLS

multi-angle laser light scattering

Hb

haemoglobin

MaxEnt

maximum entropy

FPLC

fast protein liquid chromatography

ESI-MS

electrospray ionization mass spectroscopy

Extracellular hexagonal-bilayer haemoglobins (HBL Hbs) are present in the three annelid classes; polychaetes, oligochaetes (earthworms) and achaetes (leeches) as well as in vestimentiferans (deep-sea tube-worms, now considered to belong to the annelids at large). The HBL Hbs are giant biopolymers with molecular masses in the range 3000–4000 kDa, possessing a characteristic quaternary architecture, namely the hexagonal-bilayer structure. Each molecule consists of an assembly of around 200 polypeptide chains of two kinds: globin chains with heme group and molecular mass of 15–19 kDa and; linker chains with molecular mass between 22 and 27 kDa that possess few or no heme groups and play an important role in the assembly of the whole structure [3]. Recently, our understanding of the fine architecture of these complex biopolymers has progressed in an impressive way due to the emergence of new analytical tools, ESI-MS and its associated software MaxEnt (maximum entropy) [4,5], MALLS (multi-angle laser light scattering) [6,7] and three-dimensional reconstruction by cryo-electron microscopy reviewed in Lamy et al. [3]. Despite the fact that all HBL Hbs investigated to date possess similar architecture we found different native masses using the MALLS system (i.e. range of mass comprised between ≈ 3200 and 3900 kDa [8]). Recently, a number of HBL Hbs belonging to widely diverse annelid taxa have been examined using ESI-MS revealing an intriguing diversity in subunit composition. Hence, the main purpose of the present study was to focus on a single annelid family, to test whether this diversity reflects evolutionary divergences or an intrinsic plasticity of the HBL structure. Indeed, the central question is what does it take to build an HBL Hb? How many different globin/linker chains are necessary to achieve this very specific quaternary architecture? An in vitro study has already addressed these questions for Lumbricus terrestris HBL Hb, by attempting reassociation with different combinations. One linker (but not just any) may be sufficient and removing the monomeric globin does not impair proper reassociation of this peculiar structure [9–11].

To date, the alvinellid family comprises 11 species of tubiculous polychaete annelids divided in two genera, Alvinella and Paralvinella[12,13]. All these species are strictly endemic to the deep-sea hydrothermal vent ecosystem and colonize different environments. Thus, Alvinella caudata lives sympatricaly with Alvinella pompejana in whitish honeycomb-like structure tubes on the surface of black smoker chimneys and ‘snow ball’ diffusers on the East-Pacific Rise (EPR) [14]. Although it is unclear exactly where these worms reside in the settlement structure, temperature maxima of 40 °C–80 °C have been recorded in their vicinity [15]. Paralvinella grasslei and P. palmiformis are found in various habitats at two geographically distinct area (i.e. EPR and North EPR), on white chimneys, among vestimentiferan tubes, in diffuse venting areas, and bacterial mats [13–16]. However, P. grasslei seems to be mostly associated with the tube-worms Riftia pachyptila where the temperature varies between 8 and 30 °C [12].

The extreme character of their habitat has prompted several studies of their particular biology despite the difficulty in collecting these species [17]. Regarding their Hbs, two independent studies showed by gel electrophoresis that A. pompejana, A. caudata and P. grasslei HBL Hbs exhibit a very similar quaternary structure, typical of nonvent annelids, such as Arenicola marina and L. terrestris HBL Hbs [2–18]. However, Terwilliger et al. [18] also reported, the very high stability of A. pompejana Hb as no dissociation of the HBL Hb was evident below 50 °C, and they concluded that Alvinella HBL Hb might possess structural adaptations to the extreme temperatures of its environment [18]. Indeed, in a recent investigation, Zal et al. studying the same Hb molecule, identified an unusual number of linker chains that could improve the thermostability of A. pompejana HBL Hb [1]. In addition, Toulmond et al. found, by gel electrophoresis, similarities in the structures of A. caudata, A. pompejana and P. grasslei HBL Hbs [2]. Despite these similarities, these authors found different functional properties and concluded that the Hbs can be easily differentiated from each other based on these criteria.

The aims of the present work are: to refine existing data on the structure of A. caudata, P. grasslei and P. palmiformis HBL Hbs by providing detailed and complete information on their masses determined by MALLS and the masses and numbers of their polypeptide chains determined by ESI-MS; to compare the polypeptide chain composition of alvinellid Hbs together and with those of other annelids previously published and; to test the fitting of our data with models currently proposed for L. terrestris HBL Hb [6,11].

Experimental procedures

Animal collection and sample preparation

A. pompejana and A. caudata were collected at a depth of 2600 m at the 13°N site (12°46′N−103°56′W and 12°50′N−103°57′W) on the East-Pacific Rise during two expeditions HERO′92 (A. T. & F. Z.) and EPR-SPRING′95 (F. Z.) in April 1992 and 1995, respectively. Worms were collected with the manipulator arm of the American submersible ‘Alvin’ using a landing net-like tool, and maintained at deep-sea water temperature (2–4 °C) in a thermally insulated container during the trip to the surface (2–3 h). P. palmiformis specimens were collected from the North-East-Pacific Rise by the remotely operated vehicle ROPOS during the July 1995 dive series on the Endeavour Segment of the Juan de Fuca Ridge (47°56′N−129°05′W) (PM). P. grasslei were removed from the outside surface of Riftia tubes after their sampling from the bottom during HOT TIME′97 in December 1997 (F. Z. & J. J. C.). On board ship, animals that had not been damaged during collection were dissected dorsally on an ice-cooled tray, and the blood, contaminated with as little coelomic fluid as possible was withdrawn from the main vessel into glass micropipettes or syringes and pooled on melting ice. In all cases, blood samples were centrifuged at low speed for a few minutes, and the supernatants frozen in liquid nitrogen until use in the laboratory.

Samples were thawed and centrifuged for 10 min at 7000 g at 4 °C. Hb solutions were purified by gel filtration on a 1 × 30 cm Superose 6-C column (Pharmacia LKB Biotechnology, Inc.) using a low-pressure FPLC system (Pharmacia) as described previously [7]. The eluant was a saline buffer derived from Riftia's blood composition (bis-tris propane 50 mm, 400 mm NaCl, 3 mm KCl, 32 mm MgSO4 and 11 mm CaCl2, pH 7.5) and the elution was monitored simultaneously at 280 and 414 nm (Pharmacia LKB detector). For mass determination of P. palmiformis Hb, the column was calibrated with the following marker-proteins (Sigma, Pharmacia and worm Hbs): cross linked Hb (16 kDa, 32 kDa, 48 kDa and 64 kDa), aldolase (158 kDa), catalase (232 kDa), ferritin (440 kDa), R. pachyptila Hb V2 (433 kDa [7]), thyroglobulin (669 kDa), R. pachyptila V1 Hb (3503 kDa [7]), and A. pompejana HBL Hb (3833 kDa [1]).

Multi-angle laser light scattering

Native molecular mass of A. caudata and P. grasslei could be more precisely determined using a DAWN DSP system (Wyatt Technology Corp., Santa Barbara, CA, USA), generously provided by SOPARES SA (Gentilly, France), and fitted directly on-line with the FPLC system described above. The eluate was monitored with a UV detector at 414 nm typical for Hb. The Debye fit method was used for molecular mass determination [19]. In this method the variation rate of the refractive index as a function of concentration, dn/dc, was set to 0.190 mL·g−1, typical for proteins without glycan [19]. Furthermore, to avoid discrepancies due to aggregation or dissociation of the molecules (i.e. head and tail of elution peak), mass measurements used to compute mean values were restricted to the top of each peak, corresponding to an homogeneous fraction (T. Azoulay, Sopares, personal communication). Unfortunately, this equipment was not available to analyse P. palmiformis Hb samples.

ESI-MS

Electrospray data were acquired on either a Quattro II or a Quattro LCZ mass spectrometer (Micromass UK Ltd, Altrincham, UK) and accumulated over 5–10 min whilst scanning over the m/z range 600–2500 in 10 s per scan. Sample preparation and analysis conditions were extensively explained in previous publications as well as the deglycosylation assay [20,21]. Molecular masses are based on the atomic masses of the elements [22]. The raw ESI-MS spectra were processed using a MaxEnt based approach employing the memsys5 program (MaxEnt Solutions Ltd, Cambridge, UK) incorporated as part of the Micromass masslynx software package.

Results

Hb molecular mass determined by FPLC and MALLS

Samples of alvinellid Hbs generally showed two major symmetrical peaks by gel filtration, but only the first peaks were considered for MALLS and FPLC mass measurements (Fig. 1A,B and Fig. 2). Several minor peaks were observed after the second peak for P. palmiformis (Fig. 2). These fractions, showing an absorption only at 280 nm, were not collected and probably correspond to dissociation products of the Hb (e.g. linker chains without heme groups) or other proteins dissolved in the blood. However, despite the fact that the second major peak eluted from P. palmiformis blood absorbed at both wavelengths (i.e. 280 nm and 414 nm), it was not collected in this study. Molecular masses determined by MALLS analysis during the whole elution of the peaks were 3517 ± 14 kDa (n = 841) for A. caudata (Fig. 1A) and 3822 ± 28 kDa (n = 728) for P. grasslei (Fig. 1B). The molecular mass estimated for P. palmiformis Hbs by FPLC was 3750 ± 150 kDa for the first peak and 58 ± 10 kDa for the second (n = 5) (Fig. 2).

Figure 1.

Gel filtration separation and MALLS mass measurement of Alvinella caudata (A) and P. grasslei (B) HBL Hbs. Optical density at 414 nm (lines) and molecular mass (Da; dots) vs. elution volume (mL). The slope of mass estimates along the elution profile indicates the mass polydispersity.

Figure 2.

Elution profiles of Paralvinella palmiformis haemoglobins on Superose 6-C gel eluted with saline buffer (Materials and methods). Only the position of the markers surrounding the major peaks are shown on the chromatogram. (A) A. pompejana Hb HBL (3833 kDa); (B) R. pachyptila V1 Hb (3503 kDa); (C and D). cross linked hemoglobin (64 and 32 kDa).

Polypeptide chain composition determined by ESI-MS

The subunits and polypeptide chains are named according to the nomenclature used for A. pompejana HBL Hb [1]. Figure 3A shows the MaxEnt deconvoluted spectrum obtained from native A. caudata HBL Hb analysed in denaturing solvent. It reveals three monomeric chains (a1-a3), three linker chains (L1-L3) and a trimer (T). Their masses and relative intensities are summarized in Table 1. The MaxEnt deconvoluted spectrum of the Hb after reduction with dithiothreitol (Fig. 3C), shows the disappearance of T concomitant with the emergence of three new components (b, c and d). The sum of the masses of these new components match well with the mass of the trimer, clearly indicating that T is a covalent trimer of b + c + d.

Figure 3.

MaxEnt-processed ESI spectra of Alvinella caudata HBL Hb. (A) native (B) carbamidomethylated (C) reduced with 10 mm dithiothreitol for 15 min (D) reduced with 10 mm dithiothreitol for 30 min and carbamidomethylated. The insets to (A) and (B) show the trimer region on the same intensity scale. The trimer was not observed in the spectra shown in (C) and (D). In (A), the peaks at 23370.2, 26706.7 and 51480.9 Da are 345 adducts of L1, L3 and T, respectively. Their origin is unknown.

Table 1. Summary of ESI-MS results of Alvinella caudata HBL Hb Subunits and polypeptide chains. ND, not determined.
Chain/subunitMean native
mass (Da) a
Rel.
int.b
Cam
mass (Da) c
Free
Cysd
Reduced
mass (Da) e
Red/Cam
mass (Da) f
Corrected
mass (Da) g
Total
Cys h
  • a

     Mean of four determinations on native Hb ± estimated error.

  • b

     Relative intensity in each group for the native chains a and L, and after reduction for the three trimer components (b, c and d).

  • c

     Mean of four determinations on carbamidomethylated Hb ± estimated error.

  • d

     Number of free Cys in native chain or subunit derived by dividing (Cam mass – native mass) by 57 and rounding to nearest integer.

  • e

     Mean of four determinations on reduced Hb ± estimated error.

  • f

     Mean of eight determinations on reduced/carbamidomethylated Hb ± estimated error.

  • g

     Red/Cam mass corrected for carbamidomethylation (57.052 Da/Cys). Values are reduced masses ± estimated error.

  • h

     Total Cys residues derived by dividing (Red/Cam mass - Native or reduced mass) by 58 and rounding to nearest integer.

  • i

     Composed of chains b + c + d. Note that the sum of corrected masses for b + c + d-10H (for three intra- and two inter-chain disulfide bonds) is 51132.4 Da for the native trimer.

Monomers
 a116163.2 ± 1.00.916220.0 ± 1.01ND16336.8 ± 1.016165.6 ± 1.0 3
 a216203.2 ± 1.01.016260.5 ± 1.01ND16376.2 ± 1.016205.0 ± 1.0 3
 a316531.8 ± 1.00.816589.2 ± 1.01ND16705.2 ± 1.016534.0 ± 1.0 3
Trimer Ti51133.9 ± 4.0 51189.9 ± 4.01    1
 b 0.7  16147.5 ± 1.016319.2 ± 1.016148.0 ± 1.0 3
 c 0.6  16928.6 ± 1.017101.0 ± 1.016929.8 ± 1.0 3
 d 1.0  18063.4 ± 1.018350.0 ± 1.018064.7 ± 1.0 5
Linkers
 L123025.2 ± 2.01.025025.1 ± 2.00ND23606.4 ± 2.023035.9 ± 2.010
 L224368.2 ± 2.00.424368.5 ± 2.00ND24950.5 ± 3.024380.0 ± 3.010
 L326362.2 ± 2.00.926362.7 ± 2.00ND26942.5 ± 2.026372.0 ± 2.010

Spectra obtained from denatured native P. grasslei Hb (not shown), revealed two monomeric globin chains (a1 and a2), four linker chains (L1a,b,c and L4) and a trimer (T). As with A. caudata Hb, reduction of P. grasslei with dithiothreitol produced three new components (b, c and d) concomitant with the disappearance of the trimer (T). The masses and relative intensities of the polypeptide chains comprising P. grasslei Hb are summarized in Table 2.

Table 2. Summary of ESI-MS results of Paralvinella grasslei HBL Hb Subunits and polypeptide chains. ND, not determined.
Chain/subunitMean native
mass (Da) a
Rel.
int.b
Cam
mass (Da) c
Free
Cysd
Reduced
mass (Da) e
Red/Cam
mass (Da) f
Corrected
mass (Da) g
Total
Cys h
  • a

     Mean of four determinations on native Hb ± estimated error.

  • b

     Relative intensity in each group for the native chains a and L, and after reduction for the three trimer components (b, c and d).

  • c

     Mean of two determinations on carbamidomethylated Hb ± estimated error.

  • d

     Number of free Cys in native chain or subunit derived by dividing (Cam mass – native mass) by 57 and rounding to nearest integer.

  • e

     Mean of three determinations on reduced Hb ± estimated error.

  • f

     Mean of 5 determinations on reduced/carbamidomethylated Hb ± estimated error.

  • g

     Red/Cam mass corrected for carbamidomethylation (57.052 Da/Cys). Values are reduced masses ± estimated error.

  • h

     Total Cys in chain derived by dividing (Red/Cam mass - Reduced mass) by 57 and rounding to nearest integer.

  • i

     Composed of chains b + c + d. Note that the sum of corrected masses for b + c + d-10H (for 3 intra- and 2 inter-chain disulfide bonds) is 51056.7 Da for the native trimer.

Monomers
 a116685.1 ± 1.01.016742.2 ± 1.0116685.5 ± 1.016859.7 ± 1.016688.5 ± 1.0 3
 a216814.1 ± 1.01.016871.0 ± 1.0116815.0 ± 1.016986.6 ± 1.016815.4 ± 1.0 3
Trimer Ti51059.2 ± 4.0 51115.2 ± 4.01    1
 b 0.6  16245.8 ± 1.016419.0 ± 1.016247.8 ± 1.0 3
 c 0.4  16754.3 ± 1.016925.7 ± 1.016754.5 ± 1.0 3
 d 1.0  18063.4 ± 1.018349.8 ± 1.018064.5 ± 1.0 5
Linkers
 L1a24303.7 ± 2.01.024303.5 ± 2.0024305.2 ± 2.024885.1 ± 2.024314.6 ± 2.010
 L1b24334.8 ± 2.00.724336.4 ± 2.0024343.0 ± 2.024918.4 ± 2.024347.9 ± 3.010
 L1c24349.8 ± 2.00.824350.2 ± 2.0024357.6 ± 2.024930.5 ± 2.024360.0 ± 3.010
 L226565.7 ± 2.01.026565.7 ± 2.0026575.2 ± 2.027147.7 ± 2.026577.2 ± 2.010

Similarly, data obtained for P. palmiformis Hb showed that it is composed of two major (a1a2) and two minor (a3a4) monomeric globin chains, only one linker chain (L1) and a trimer (T). As with the previous Hbs, reduction of P. palmiformis Hb with dithiothreitol showed that the trimer is composed of three globin chains b, c and d. The masses and relatives intensities of the chains comprising P. palmiformis Hb are summarized in Table 3.

Table 3. Summary of ESI-MS results of Paralvinella palmiformis HBL Hb Subunits and polypeptide chains. ND, not determined.
Chain/subunitMean native
mass (Da) a
Rel.
int.b
Cam
mass (Da) c
Free
Cys d
Reduced
mass (Da) e
Red/Cam
mass (Da) f
Corrected
mass (Da) g
Total
Cys h
  • a

     Mean of four determinations on native Hb ± estimated error.

  • b

     Relative intensity in each group for the native chains a and L, and after reduction for the three trimer components (b, c and d).

  • c

     Mean of three determinations on carbamidomethylated Hb ± estimated error.

  • d

     Number of free Cys in native chain or subunit derived by dividing (Cam mass - Native mass) by 57 and rounding to nearest integer.

  • e

     Mean of 5 determinations on reduced Hb ± estimated error.

  • f

     Mean of four determinations on reduced/carbamidomethylated Hb ± estimated error.

  • g

     Red/Cam mass corrected for carbamidomethylated (57.052 Da/Cys). Values are reduced masses ± estimated error.

  • h

     Total Cys in chain derived by dividing (Red/Cam mass - Reduced mass) by 57 and rounding to nearest integer.

  • i

     Composed of chains b + c + d. Note that the sum of corrected masses for b + c + d-10H (for 3 intra- and 2 inter-chain disulfide bonds) is 50981.6 Da for the native trimer.

Monomers
 a116616.8 ± 1.01.016674.2 ± 1.0116616.6 ± 1.016791.7 ± 1.016620.5 ± 1.03
 a216660.5 ± 1.00.916716.4 ± 1.0116660.8 ± 1.016832.8 ± 1.016661.6 ± 1.03
 a316745.9 ± 1.00.316801.6 ± 1.01NDNDNDND
 a416790.1 ± 1.00.216847.2 ± 1.01NDNDNDND
Trimer T i50983.2 ± 4.0 ND    1
 b0.7   16245.6 ± 1.016418.8 ± 1.016247.6 ± 1.03
 c0.5   16723.4 ± 1.016895.1 ± 1.016723.9 ± 1.03
 d1.0   18019.0 ± 1.018305.5 ± 1.018020.2 ± 1.05
Linkers
 L126598.8 ± 2.01.026598.1 ± 2.0026600.8 ± 2.027177.1 ± 2.026606.6 ± 2.010

Cysteine residue and carbohydrate content

The numbers of free Cys and disulfide bonds in each chain were estimated by comparing the native or reduced masses with those obtained after carbamidomethylation. Figure 3B shows the MaxEnt deconvoluted spectrum obtained after carbamidomethylation of the native A. caudata Hb and implies that the monomeric chains (a1-a3) and the trimer (T) possess one free Cys. This is inferred from the addition of one, and only one, carbamidomethyl group to the native components (+57.052 Da per Cys). Moreover, the masses of L1L3 remained the same after carbamidomethylation, indicating the absence of free Cys on these chains.

Figure 3D presents the spectrum obtained after carbamidomethylation of reduced A caudata Hb and shows that the monomeric chains a1 to a3 each contain 3 Cys, two accounting for an intrachain disulfide bond and one free Cys. The chains b, c and d, constituting the trimer (T), contain 3, 3 and 5 Cys, respectively. The data in Fig. 3D also imply that there are 10 Cys in each of the three linker chains. These results are summarized in Table 1. Similar results were obtained for the Cys content in P. grasslei and P. palmiformis HBL Hbs and are summarized in Tables 2 and 3, respectively. The interpretation of these data in terms of intra- and interdisulfide bridges is discussed later (Fig. 4)

Figure 4.

(A-C) Schematic representations of the three possible arrangements of the trimer T of A. caudata, P. grasslei and P. palmiformis HBL Hb based on data from Fig. 3 and Tables 1–3 involving a single disulfide bond between chains b, c and d and having one free Cys residues. Cys residue are either free ( SH) or involved in intra or interchain disulfide bridges (S-S-). (D) Similar representation for the trimer of A. pompejana[1]. Note: the locations of the Cys residues on the chains are arbitrary.

Deglycosylation experiments performed on HBL Hbs of A. caudata, P. grasslei and P. palmiformis revealed no glycosylated polypeptide chain in any of these molecules (data not shown).

Discussion

Molecular structures of alvinellid HBL Hbs

Native molecular masses. In order to determine the molecular masses of the native Hbs of A. caudata and P. grasslei Hbs we used a MALLS instrument, except for the molecular mass of P. palmiformis Hb which was obtained only by gel filtration. This method has the advantage of determining the molecular mass of molecules in physiological buffer quite similar to real conditions inside the organism [7]. The correct molecular mass of the native molecule is crucial information for the construction of the quaternary structure model, since this data allows us to check the validity of a theoretical assembly. However, despite the fact that all HBL Hbs investigated to date possess the same global architecture (i.e. hexagonal-bilayer structure) notable differences of native molecular mass values were found, even by MALLS (i.e. molecular mass comprised between 3261 ± 80 kDa for the chlorocruorin of Eudistylia vancouverii and 3833 ± 14 kDa for the HBL Hb of A. pompejana[1,8]) strongly indicating differences between these molecules. Similar findings were found inside the alvinellid family and seem to suggest once again different structures, even for these closely related species. The molecular mass found for P. grasslei Hb by MALLS was 3822 ± 28 kDa, a value close to that found using the same method for A. pompejana, 3833 ± 14 kDa [1], but distant from the molecular mass of A. caudata Hb, 3517 ± 14 kDa. The molecular mass of P. palmiformis HBL Hb as determined by gel filtration, 3750 ± 150 kDa is similar to the mass determined for other polychaetes Hbs using the same method (i.e. 3600 kDa), as for example A. pompejana[18], A. marina[24] or Nephtys incisa[23].

Structural analysis of the hexagonal-bilayer Hbs. SDS/PAGE technique is not resolutive enough to show mass differences of a few daltons between polypeptide chains as observed for this kind of assembly and SDS/PAGE was recently abandoned for a more accurate technology, ESI-MS. Therefore, structure divergences were also established for HBL Hbs presently investigated by this method as for A. pompejana[1], Riftia pachyptila[21], A. marina[24], L. terrestris[25], Eudystilia vancouverii[26] and Macrobdella decora[27]. However, these species belong to different classes or families, and we were wondering if similar differences could exist within a single family. Within the alvinellids, the four HBL Hbs studied show several differences in the composition of their polypeptide chains which were not observed by gel electrophoresis in previous study [2] and could correspond to a polymorphism.

Linker chains. One of the major differences between the four alvinellid HBL Hbs concerns the number of different linker chains, four for A. pompejana and P. grasslei[1], three for A. caudata and only one for P. palmiformis. Hence, the existence of different types of linker chains does not appear necessary to achieve a HBL arrangement, since even a single sort of linker chain is sufficient. This observation is in total agreement with the recent finding published for L. terrestris concerning the reassembly of its HBL Hb [10–11]. These authors showed that L. terrestris Hb, partially dissociated, can be reassembled with each of the four different linker chains and the present data show that it can be also the case inside a natural assembly as for P. palmiformis HBL Hb. Consequently, the additional linker chains observed for other species could be the result of evolutionary processes (i.e. duplication events) of an ancestor gene coding for this first primitive linker chain. This primitive linker chain should have contained 10 Cys making five intrachain disulfide bridges, a common characteristic of all alvinellid linker chains, in contrast to the linker chains of Riftia pachyptila V1 Hb which possess six disulfide bridges [21]. Furthermore, inside the same genus (i.e. Alvinella or Paralvinella) variation can be observed. Thus, it appears that gene coding for linker chains evolved at the species level.

Trimer complex. Data derived from ESI-MS analysis establish that the HBL Hbs of all alvinellid species are composed of a trimer complex clearly placing alvinellid HBL Hbs in the polychaete/oligochaete subgroup as opposed to the achaete/vestimentiferan subgroup which possesses dimers instead of trimers [3]. Amongst these trimers, only the one of A. pompejana is different with the presence of two free Cys on the globin chain b and d and the absence of an intradisulfide bridge on chain b (Fig. 4D). The presence of this additional free Cys could increase the potential to bind sulfide, following a mechanism demonstrated in Riftia pachyptila[28], as suggested by sulfide binding experiments with A. pompejana HBL Hb (F. Zal, unpublished observation). Sulfide binding could be an adaptive answer to the specific environment of this species, since this worm lives in close contact with the hydrothermal vent fluid where high concentrations of sulfide are discharged.

Another interesting observation is that several globin chains associated inside the trimeric subunits of these worms retain exactly the same molecular masses, and it seems highly improbable that these characteristics are due to a coincidence (e.g. molecular mass of chain b identical in P. palmiformis and P. grasslei, molecular mass of chain d identical in A. caudata and P. grasslei), although only the primary sequences would permit determination if it is really the same globin chains. Using this criteria, we can reassemble together P. palmiformis and P. grasslei as well as A. caudata and P. grasslei. Similar association can be made between chain e making a dimer in R. pachyptila V1 Hbs and chain b from A. caudata trimer, since these two chains are characterized by the same molecular mass [21]. Only A. pompejana trimer does not share characteristics with the three other ones coming from A. caudata, P. palmiformis and P. grasslei HBL Hbs (Fig. 4A–D).

Monomeric globin chains. We also observed a disparity in the number of the monomeric globin chains constituting these HBL Hbs, 4 for P. palmiformis, 4 for A. pompejana[1], 3 for A. caudata and only 2 for P. grasslei. In addition, as for the globin chain forming the trimers, one of the monomeric globin chains, a2, constituting A. pompejana HBL Hb has similar molecular mass of chain a3 of A. caudata HBL Hb. Consequently, as for the linker chains, the number of monomeric globin chains constituting these molecules is variable and fixed only at the species level. Remarkably, we can observe that P. palmiformis HBL Hb that possesses the highest diversity of monomeric globin chains is also the molecule that contains the lowest diversity of linker chains.

Cysteine content. Almost all globin chains of Alvinella and Paralvinella HBL Hbs hold free Cys. Indeed, all the monomeric globin chains constituting these molecules possess one free Cys in addition to an intradisulfide bond. Some years ago, several authors proposed to classify annelid globin chains into four categories according to their Cys content, all four groups holding an intrachain disulfide bond, and three of these groups containing a supplementary Cys, at different positions, participating in an interchain disulfide bond [29–31]. Nevertheless, we showed recently that it was not possible to place inside this classification the globin chains holding free Cys as it seems to be the case for the majority of globin chains involved in HBL Hb of annelids and vestimentiferans settling in sulfidic environments [1,21–24]. Moreover, we also showed that two of the globin chains in the trimer complex, globin chains b and d, of A. pompejana HBL Hb had an unusual number and association of the Cys [1] (Fig. 4D).

Model of quaternary structure of alvinellid HBL Hbs. Recently, the three-dimensional reconstruction of L. terrestris HBL Hb was obtained by cryo-electron microscopy with a very high resolution (i.e. 22 Å) [10]. This study allowed to localize precisely the globin and linker chains within the molecule. In view of this new finding it was interesting to test, for the three alvinellid molecules investigated in this study and that of A. pompejana previously investigated [1], the quaternary structure model proposed for L. terrestris with 36 monomeric globin chains (M), 36 trimeric complexes (T) and 42 linker chains (L). Using the weighted mean masses for M and L we can calculate the native mass for each molecule (Table 4). These calculations showed a very good agreement between the calculated and experimental masses for A. caudata and P. palmiformis with only 0.88 and 2.98% of difference, respectively. However, the differences found for A. pompejana and P. grasslei HBL Hbs is around 7%. These variations seem too high to be acceptable due to the accuracy of the techniques used in this study (i.e. ± 3% of the real mass for the MALLS instrument and around ± 0.01% for ESI-MS). Alternative models previously proposed for A. pompejana with 144 globin and 60 linker chains or 120 globin and 72 linker chains fit better with the experimental mass of this molecule with only 4.2 and 0.5% difference, respectively [1].

Table 4. Models for the quaternary structure of alvinellid HBL Hbs using the model proposed by Taveau et al.[10]. Two of the four alvinellid HBL Hbs do not fit this model (i.e. A. pompejana and P. grasslei) if we accept a difference of ± 3% between the native and calculated masses (given by MALLS accuracy). Interestingly, one of the model previous proposed for A. pompejana with 120 globin and 72 linker chains give a difference of only 0.5% [1].
AssemblyA. caudataA. pompejanaP. grassleiP. palmiformis
  • a  Mean masses in Da of the monomeric chains (cf. Tables 1–3 and for A. pompejana[1]). b An heme molecule possesses a mass of 616.5 Da and 144 corresponds to 36 monomeric globin chains and 36 trimeric complexes (i.e. 36 × 3 = 72). c Mean masses in Da of the linker chains (cf. Tables 1–3 and for A. pompejana[1]). d Theoretical mass in Da calculated using a model of 36 monomeric globin chains, 36 trimeric complexes and 42 linker chains [10].

  • e

     Experimental masses obtained by MALLS (A. caudata, A. pompejana and P. grasslei) and FPLC (P. palmiformis).

  • f

     Difference in percentage between the theorical masses of the models and the experimental masses of alvinellid HBL Hbs.

Monomersa16 29916 48316 75016 664
Monomers x 36586 764593 409603 000599 904
Trimer51 13451 43251 05950 983
Trimer x 361 840 8241 851 5521 838 1241 835 388
Heme x 144b88 77688 77688 77688 776
Linkerc24 56424 83124 88926 598
Linker x 421 031 6881 042 9021 045 3381 117 116
Total massd3 548 0523 576 6393 575 2383 641 184
Native masse3 517 0003 833 0003 822 0003 750 000
Difference (%)f0.887.166.902.98

However, due to this discrepancy and although this model does not fit with the three-dimensional reconstruction of HBL Hbs obtained by cryo-electron microscopy for L. terrestris[9–11,32], for A. pompejana and for R. pachyptila HBL Hbs [33–34], the model proposed for L. terrestris HBL Hb by Zhu and collaborators [6] was also tested. This model consists of the association of 48 M, 48 T and 24 L making an HBL Hb with 216 polypeptide chains, each one-twelfth consisting of an hexadecamer M4T4 instead of the dodecamer M3T3 constituting the ‘bracelet model’ as proposed by Vinogradov for L. terrestris HBL Hb and reviewed by Lamy et al. [3]. The calculations using Riggs's model are presented in Table 5. Nevertheless, the differences between calculated and experimental masses are too high to be acceptable for HBL Hbs coming from alvinellid species and consequently this model does not seem to fit as well.

Table 5. Models for the quaternary structure of alvinellid HBL Hbs using the model proposed by Zhu et al.[6]. None of the four alvinellid HBL Hbs fits this model at the threshold of ± 3% (given by MALLS accuracy) between the native and calculated masses.
AssemblyA. caudataA. pompejanaP. grassleiP. palmiformis
  • a  Mean masses in Da of the monomeric chains (cf. Tables 1–3 and for A. pompejana[1]). b An heme molecule possesses a mass of 616.5 Da and 192 corresponds to 48 monomeric globin chains and 48 trimeric complexes (i.e. 48 × 3 = 72). c Mean masses in Da of the linker chains (cf. Tables 1–3 and for A. pompejana[1]). d Theoretical mass in Da calculated using a model of 48 monomeric globin chains, 48 trimeric complexes and 24 linker chains [6].

  • e

     Experimental masses obtained by MALLS (A. caudata, A. pompejana and P. grasslei) and FPLC (P. palmiformis).

  • f

     Difference in percentage between the theorical masses of the models and the experimental masses of alvinellid HBL Hbs.

Monomersa16 29916 48316 75016 664
Monomers x 48782 352791 184804 000799 872
Trimer51 13451 43251 05950 983
Trimer x 482 454 4322 468 7362 450 8322 447 184
Heme x 192b118 368118 368118 368118 368
Linkerc24 56424 83124 88926 598
Linker x 24589 536595 944597 336638 352
Total massd3 944 6883 974 2323 970 5364 003 776
Native masse3 517 0003 833 0003 822 0003 750 000
Difference (%)f10.843.553.746.33

These data together raise two main questions, why similar molecules, belonging to closely related species, are so different in their assembly? and how, despite these differences, can they hold a similar architecture? Although, the results presented in this article do not allow us to solve these questions directly, we can propose some hypotheses. Indeed, we show that Hb composition shows a high degree of plasticity at the genus level (i.e. Alvinella and Paralvinella). Consequently, the HBL Hb structure seems only fixed at the species level, and can even differ between different populations of the same species that could reflect a polymorphism inside the HBL Hb structure. Indeed, two of us observed different polypeptide chain compositions of Hbs from the same species (i.e. A. marina and Riftia pachyptila) on animals collected at two distinct places (F. Zal and B. N. Green, unpublished observations). Thus, it seems that either the environment colonized by these organisms or the evolution of each population can induce differences in the structure of their Hbs. Similar observations were done on the haemocyanin composition of the blue crab Callinectes sapidus[35]. Indeed, these authors showed that the subunit composition of its haemocyanin was modified as a function of the variation of the salinity of the environment. Concerning the second question we raised, it appears that HBL Hb architecture is not dependent of a strict presence of one panel of globin and linker. The critical point to form this structure seems to be as earlier mentioned the presence of linker chains [3]. In addition, these data show that with only one sort of linker chain this architecture can be realized in agreement with recent finding obtained with L. terrestris HBL Hb reassembled [11].

In conclusion, the results presented in this paper revealed a high level of plasticity in the composition of the HBL Hb molecules, and consequently the HBL Hb of each species may possess its own assembly. These different structures are probably linked to the different functional properties previously highlighted for these HBL Hbs [2]. Furthermore, P. palmiformis Hb is the first molecule investigated which is able to form a native HBL structure with only one kind of linker chain. In addition, the molecular mass of some of the globin chains of these HBL Hbs are characterized by similar molecular masses, and it seems highly unlikely that it could be a coincidence.

Acknowledgements

We are very grateful to the members of the ‘Alvin’ groups, the captains and crews of the R/V ‘Vickers’, and R/V “Atlantis II', and to the chief scientists of the HERO’92 and EPR-SPRING’95 expeditions, Horst Felbeck, Daniel Desbruyères and Lauren Mullineaux, who allowed us to collect these samples. We also thank T. Azoulay of the Sopares SA (Gentilly, France) for his help and technical support in MALLS analysis. Finally, we acknowledge the two anonymous referees for constructive criticism that allowed us to improve this article. This work was supported by research grants from CNRS (UPR 9042), INSU, UPMC, IFREMER (URM no. 7) (F. Z., F. L. and A. T.), the US National Science Foundation (NSF OCE 9632861) (J. J. C.), the US National Institute of Health (NIH DK 38674) (S. V. N.), the Conseil Régional de Bretagne who enabled the first author to work in the University of California at Santa Barbara as a postdoctoral fellow and to perform this work.

Ancillary