Báo cáo khoa học: Molecular mass of macromolecules and subunits and the quaternary structure of hemoglobin from the microcrustacean Daphnia magna ppt

Paul1 1 Institut fu¨r Zoophysiologie, Westfa¨lische Wilhelms-Universita¨t, Mu¨nster, Germany 2 Proteom Centrum Tu¨bingen, Eberhard-Karls-Universita¨t, Tu¨bingen, Germany 3 Institut fu¨r

quaternary structure of hemoglobin from the

microcrustacean Daphnia magna

Tobias Lamkemeyer1,2, Bettina Zeis1, Heinz Decker3, Elmar Jaenicke3, Dieter Waschbu¨sch3,

Wolfgang Gebauer4, Ju¨rgen Markl4, Ulrich Meissner4, Morgane Rousselot5, Franck Zal5,

Graeme J Nicholson6and Ru¨diger J Paul1

1 Institut fu¨r Zoophysiologie, Westfa¨lische Wilhelms-Universita¨t, Mu¨nster, Germany

2 Proteom Centrum Tu¨bingen, Eberhard-Karls-Universita¨t, Tu¨bingen, Germany

3 Institut fu¨r Molekulare Biophysik, Johannes Gutenberg Universita¨t, Mainz, Germany

4 Institut fu¨r Zoologie, Johannes Gutenberg Universita¨t, Mainz, Germany

5 Equipe Ecophysiologie: Adaptation et Evolution Mole´culaires, UPMC–CNRS UMR 7144, Station Biologique, BP 74, Roscoff, France

6 Institut fu¨r Organische Chemie, Eberhard-Karls-Universita¨t,Tu¨bingen, Germany

Keywords

glycosylation; hemoglobin; macromolecule;

molecular mass; quaternary structure

Correspondence

T Lamkemeyer, Interfakulta¨res Institut fu¨r

Zellbiologie, Proteom Centrum Tu¨bingen,

Eberhard-Karls-Universita¨t, Auf der

Morgenstelle 15, D-72076 Tu¨bingen,

Germany

Fax: +7071 295359

Tel: +7071 2970556

E-mail:

Tobias.Lamkemeyer@uni-tuebingen.de

(Received 23 March 2006, revised 17 May

2006, accepted 30 May 2006)

doi:10.1111/j.1742-4658.2006.05346.x

The molecular masses of macromolecules and subunits of the extracellular hemoglobin from the fresh-water crustacean Daphnia magna were deter-mined by analytical ultracentrifugation, multiangle laser light scattering and electrospray ionization mass spectrometry The hemoglobins from hypoxia-incubated, rich and normoxia-incubated, hemoglobin-poor Daphnia magna were analyzed separately The sedimentation coeffi-cient of the macromolecule was 17.4 ± 0.1 S, and its molecular mass was 583 kDa (hemoglobin-rich animals) determined by AUC and 590.4 ± 11.1 kDa (hemoglobin-rich animals) and 597.5 ± 49 kDa (hemo-globin-poor animals), respectively, determined by multiangle laser light scattering Measurements of the hemoglobin subunit mass of hemoglobin-rich animals by electrospray ionization mass spectrometry revealed a signi-ficant peak at 36.482 ± 0.0015 kDa, i.e 37.715 kDa including two heme groups The hemoglobin subunits are modified by O-linked glycosylation in the pre-A segments of domains 1 No evidence for phosphorylation of he-moglobin subunits was found The subunit migration behavior during SDS⁄ PAGE was shown to be influenced by the buffer system used (Tris versus phosphate) The subunit mass heterogeneity found using Tris buffer-ing can be explained by glycosylation of hemoglobin subunits Based on molecular mass information, Daphnia magna hemoglobin is demonstrated

to consist of 16 subunits The quaternary structure of the Daphnia magna hemoglobin macromolecule was assessed by three-dimensional reconstruc-tions via single-particle analysis based on negatively stained electron micro-scopic specimens It turned out to be much more complex than hitherto proposed: it displays D4 symmetry with a diameter of approximately

12 nm and a height of about 8 nm

Abbreviations

AUC, analytical ultracentrifugation; BN-PAGE, blue native polyacrylamide gel electrophoresis; ESI-MS, electrospray ionization mass

spectrometry; Hb, hemoglobin; MALLS, multiangle laser light scattering; MRA, multireference alignment; MSA, multivariate statistical analysis; RuBPs, ruthenium II tris(bathophenanthroline disulfonate).

Invertebrate hemoglobins show very high structural

diversity in comparison with the uniform tetrameric

structure in vertebrates They range from the 17 kDa

single-chain globins found in bacteria, algae, protozoa

and plants to the large multisubunit, multidomain

hemoglobins found in nematodes, molluscs and

crusta-ceans up to the giant annelid and vestimentiferan

hemoglobins of about 3600 kDa, which are composed

of globin and nonglobin subunits [1]

Specifically, the structure and function of the

hemo-globin (Hb) of the microcrustacean Daphnia magna

have been addressed by many studies over the last

dec-ades Daphnia magna Hb is freely dissolved in the

extracellular fluid It is a multisubunit Hb composed

of didomain globin chains The synthesis of this Hb is

regulated by ambient oxygen concentration,

tempera-ture and juvenoid hormones [2–4] Environmental

hypoxia, for example, may cause an increase of Hb

concentration in D magna by a factor of 16,

corres-ponding to Hb concentrations between 55 and

888 lmol hemeÆL)1 [5,6] Concomitant changes of

oxy-gen affinity between 1.02 and 0.15 kPa (P50) have been

reported [7], which are related to Hb multiplicity

(iso-hemoglobins composed of different Hb subunits)

[1,8,9] The biological advantages of increased Hb

concentration and oxygen affinity under hypoxia are

manifold [1,9] Accordingly, Daphnia is currently the

focus of investigations aimed at the structure and

evo-lution of its globin genes as well as of its physiologic

adaptations [1]

Hb synthesis takes place in the fat cells and

epipo-dite epithelial cells of D magna [10], which are the

only known sites of Hb synthesis in crustaceans [1]

The seven known Hb subunits (DmHb) [3,8,11] are

encoded by at least six Hb genes (dmhb) [12] At least

four of them are organized in a cluster in the order

dmhb4, dmhb3, dmhb1 and dmhb2 [11] On the basis of

nucleotide and derived amino acid sequences, the

molecular masses of Hb subunits are 36.228 kDa

(dmhb1), 36.177 kDa (dmhb2), 36.217 kDa (dmhb3)

(sequence data from [11]) and 35.921 kDa (dmhb4)

(sequence data from [12]) However, the molecular

masses of the Hb subunits experimentally determined

by gel electrophoresis were 36.2 (DmHbA–DmHbD),

37.9 (DmHbF, DmHbG) and 40.6 kDa (DmHbE) [8],

raising the question of the reason for this difference

For the molecular mass of the native D magna Hb

complex, reported values vary between 494 and

670 kDa, depending on the methods used (gel

filtra-tion, ultracentrifugafiltra-tion, gel electrophoresis [13–15])

Suggesting very low molecular masses of about 31–

33 kDa for the Hb subunits and 494 kDa for the

native multimer, Ilan et al [14] concluded that 16

polypeptide chains, each carrying two heme-binding domains, form one D magna Hb macromolecule From electron micrographs, two models of the three-dimensional structure have been suggested [14]: (a) a cyclic structure composed of all 16 subunits; and (b) a dihedral structure, in which the subunits are grouped

in two layers stacked in an eclipsed orientation Accordingly, the present data on the molecular mass

of the native Hb complex as well as the Hb subunits are inconsistent In addition, there are only two hypo-thetical models concerning the structure of the multi-subunit assembly, which has stimulated further studies

on the structure of D magna HB Three-dimensional reconstruction from transmission electron microscopy promises to be a satisfactory way to describe the qua-ternary structure of high molecular mass invertebrate respiratory proteins [16–18]

To elucidate the important structural characteristics

of the extracellular Hb of D magna, the molecular mass of the native Hb complex as well as those of denatured Hb subunits were determined by analytical ultracentrifugation and multiangle laser-light scattering (MALLS) or by MS and gel electrophoresis, respect-ively, in normoxically (pale) and hypoxically (red) raised D magna In addition, possible post-transla-tional modifications were investigated To investigate the quaternary structure of D magna Hb, transmission electron microscopy and three-dimensional reconstruc-tion of the macromolecule were carried out

Results The native Hb complex from the hemolymph of red

D magna purified by chromatofocusing and diluted in

10 mm ammonium acetate buffer was subjected to ultracentrifugation to determine the sedimentation velocity (Fig 1A) Use of the van Holde–Weischet analysis on the data (Fig 1B) showed that approxi-mately 75% of the native protein sedimented with a sedimentation coefficient of 17.4 ± 0.1 S (Fig 1C) The remaining proteins sedimented at 16.3–17.1 S Subsequent sedimentation equilibrium experiments (Fig 2) resulted in a molecular mass of 583 kDa for the native Hb complex from the hemolymph of red

D magna Owing to their small quantity, the molecu-lar mass of slower-sedimenting proteins could not be analyzed with this method

MALLS analyses of pale and red D magna Hb obtained by purification of crude animal extracts via gel filtration or chromatofocusing provided molecular mass determinations during the elution of peaks (Fig 3) On average, the molecular mass was 597.5 ± 49 kDa (n¼ 3 samples: 562.3, 576.4 and

653.9 kDa; for each sample, the molecular mass is

averaged over 130 measured points) for pale D magna

Hb and 590.4 ± 11.1 kDa (n¼ 5 samples: 576.2,

582.2, 593.2, 596.0 and 604.2 kDa; for each sample,

the molecular mass is averaged over 130 measured

points) for red D magna Hb A second peak in case of

pale D magna (Fig 3A) was heterogeneously

com-posed (polydispersity), as indicated by the inclination

of individual measuring points The molecular mass of

these proteins (Hb dissociation products or other

pro-teins) was 385 ± 9 kDa

Using a multiphasic buffer system according to

Laemmli [19], gel electrophoresis (SDS⁄ PAGE)

resul-ted in the separation of D magna Hb subunits into

three different bands (Fig 4A) with molecular masses

of 40.0, 38.1 and 35.1 kDa (pale D magna) and 40.0,

36.9 and 35.1 kDa (red D magna) As such

separa-tions may be due to a specific buffer system instead of

reflecting actual mass differences [20], the protocol of

Weber and Osborn [21] was employed: in this case,

only one band for both Hb from the hemolymph of

pale D magna and Hb from red D magna appeared

(Fig 4B), corresponding to a molecular mass of

approximately 39.2 kDa Bands with molecular masses

above 66 kDa may originate from undissociated Hb

molecules or impurities in the sample

For an exact determination of the molecular mass of

D magna Hb subunits, electrospray ionization mass

spectrometry (ESI-MS) analyses were performed For

red D magna Hb purified by gel filtration, an ion

ser-ies in the mass⁄ charge ratio (m ⁄ z) range of 1500–3000

occurred under denaturing conditions (Fig 5A) De-convolution of the spectrum resulted in a single signifi-cant peak which corresponded to a molecular mass of 36.482 ± 0.0015 kDa (Fig 5B) The acidic conditions used for ESI-MS analysis led to the dissociation of the heme group (616.5 Da) from the polypeptide chains Thus, the final mass for the didomain subunit is 37.715 kDa, including two heme groups

To test for post-translational modifications of red

D magna Hb subunits, a staining technique specific for glycosylated proteins was used (Fig 6A), followed

by staining with ruthenium II tris(bathophenanthroline disulfonate) (RuBPs) (Fig 6B) All Hb subunit spots and the glycosylated proteins of the CandyCane mar-ker (Fig 6C) were specifically labeled after staining for glycoproteins The remaining marker proteins became visible only after silver staining for total protein Obvi-ously, D magna Hb subunits are glycosylated

These staining results are corroborated by enzymatic deglycosylation of Hb subunits After two-dimensional gel electrophoresis, the Hb subunits A–D appear as a train of spots, whereas the subunits E–G also differ in molecular mass [3,8] (Fig 7A) To test whether this separation is influenced by subunit glycosylation, Hb (purified by gel filtration) was incubated with different sets of enzymes removing only N-linked sugars (Fig 7B), only O-linked carbohydrates (Fig 7C) or both types of glycans (Fig 7D), respectively The untreated sample (Fig 7A, control) represents the typ-ical spot pattern of red D magna Hb consisting of the subunits A–D and F The Hb pattern after incubation

radius (cm) 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0

a 0.0

0.1 0.2 0.3 0.4 0.5

0.6

A

time-0.5 (min-0.5) 0.00 0.01 0.02 0.03 0.04

12 14 16 18 20 22

24

B

sedimentation coefficient (S) 16.0 16.5 17.0 17.5 18.0

% 0

20 40 60 80

100

C

time

Fig 1 Sedimentation velocity analysis of

the hemoglobin (Hb) of red (Hb-rich)

Daph-nia magna A preparation of Hb purified by

chromatofocusing was centrifuged at

116 480 g at 20
C; the protein

concen-tration in the cell was 1.0 mgÆmL)1in

10 m M ammonium acetate buffer, pH 6.7.

(A) The absorbance at 280 nm along the

length of the cell was recorded every 4 min.

(B) The sedimentation coefficient was

deter-mined with the van Holde–Weischet

extra-polation plot (C) The distribution plot shows

that approximately 75% of the protein

sedi-mented at 17.4 ± 0.1 S, whereas only a

small fraction sedimented more slowly

(16.3–17.1 S).

with N-glycosidase F (Fig 7B) is obviously identical to

that of the untreated sample Incubation with enzymes

specific for the removal of O-linked glycans led to the

occurrence of additional spots (marked with white

cir-cles in Fig 7C) with a higher electrophoretic mobility

Consequentially, these spots also emerged, when the

Hb sample was treated with the whole set of enzymes

removing N-linked and O-linked glycans (Fig 7D)

Potential N-linked glycosylation sites (NXT⁄ S) were

not found in the amino acid sequences of D magna

Hb [22] Prediction of O-linked glycosylation was

per-formed using the recently released NetOGlyc 3.1 server

[23], and the results indicated the pre A-segments of

domain 1 as sites for glycosylation events in all Hb sequences The numbers of possible glycosylation sites exceeding the threshold (G-score¼ 0.5) were 11 for DHb 1 and DHb2 and 12 for DHb3 and DHb4, respectively

For identification of the carbohydrates bound to

D magna Hb subunits, saccharides were released by methanolysis and separated by GC followed by MS Chromatograms of the carbohydrates of D magna Hb,

of the corresponding blank and of a mannose standard are shown in Fig 8 The dominant carbohydrate in

D magna Hb is mannose, which was unambiguously identified by the analysis of a mannose standard (gray-shaded curve in Fig 8) A lower, but still highly signi-ficant, increase (compared to the blank; Fig 8, inset)

in the intensity of galactose and glucose was also observed (the dashed line in Fig 8 represents the

0.0

0.2

0.4

0.6

0.8

1.0

III

II I

A

-0.05

0.00

0.05

er -0.05

0.00

0.05

position, rmp2 - rmc2 (cm 2 ) 0.0 0.5 1.0 1.5 2.0 2.5

-0.05

0.00

I

II

B

Fig 2 Sedimentation equilibrium analysis of the hemoglobin (Hb)

of red Daphnia magna A preparation of Hb purified by

chromatofo-cusing was centrifuged at 4660 (I), 5900 (II) and 7280 (III) g at

4
C, until equilibrium was achieved (A); the protein concentration

was 1.0 mgÆmL)1in 10 m M ammonium acetate buffer, pH 6.7 The

results at all speeds were globally fitted, resulting in a molecular

mass of 583 kDa The residuals (B: I–III) reflect the deviation of

individual measuring points from the calculated fit, indicating the

quality of the fit and the calculated molecular mass mp, measuring

point; mc, meniscus.

Fig 3 MALLS analysis of the hemoglobin (Hb) of (A) pale (Hb-poor) and (B) red (Hb-rich) Daphnia magna purified by gel filtration or chromatofocusing, during the elution on a gel filtration column (Su-perose 6-C) The solid curve represents the refractive index signal profile versus the elution volume, and the distribution of the molecular weight values is represented by crosses The latter infor-mation yielded molecular masses of 597.5 ± 49 kDa (n ¼ 3 sam-ples from different animal groups) for pale and 590.4 ± 11.1 kDa (n ¼ 5 samples from different animal groups) for red D magna Hb The samples from pale D magna additionally showed a second protein component of a lower molecular mass (384.5 kDa).

maximum total ion current of the blank: glucose,

about 2500)

The results of all experiments to demonstrate

phos-phorylation of Hb subunits, however, were negative

Neither specific staining of phosphorylated proteins in

gels (Pro Q Diamond stain) nor detection by western

blotting using antibodies specific for phosphoserine

and phosphotyrosine residues or different MS methods

showed any indication of phosphorylation of D magna

Hb subunits (data not shown)

To assess the organization of the native Hb

com-plex, electron microscopy and three-dimensional

recon-structions were employed After staining with 2%

uranyl acetate, the highly concentrated Hb

macromole-cules in the hemolymph of red D magna (Fig 9C,D)

frequently showed specific clover-leaf structures

(marked by arrows) These characteristic clover-leaf

structures were scarcely found in the hemolymph of

pale D magna (Fig 9A,B) In addition to Hb

mole-cules, another type of protein was detected in the

hemolymph of pale and red D magna (probably

fer-ritin, marked by asterisks in Fig 9A,C,D)

Because of its higher hemolymph concentration and

higher degree of structural detail, red D magna Hb

was considered to be more promising for single-parti-cle image processing As purification by gel filtration did not completely remove the ferritin, which would have interfered with image analysis, Hb was purified

by chromatofocusing, resulting in a small quantity of ferritin in comparison to the large number of Hb mole-cules in the samples In addition, structures clearly representing ferritin molecules were omitted from sub-sequent image analysis The macromolecules within a sample were spread across the holes in a perforated carbon support film, causing the molecules to have many different axial orientations relative to the elec-tron beam As the three-dimensional organization of subunits in a native Hb complex of D magna is as yet unknown, different symmetry features had to be tested and verified Therefore, the simplest symmetry with minimum assumptions (C2; two-fold symmetry, bot-tom and top of the molecule different) was chosen for

an initial estimation From the established three-dimensional model, it could be expected that the bot-tom and the top of the molecule would be identical (D symmetry) Manual and automated (anchor set) Euler searches were then systematically performed for two-fold to eight-two-fold D symmetries (D2–D8) For the Hb

of red D magna, the class averages calculated from electron micrographs seemed to suggest a tetrameric (D4) symmetry (Fig 9E) Moreover, the best corres-pondence between class averages and reprojections was found for this symmetry, so that D4 symmetry was finally selected and used for three-dimensional recon-structions The preliminary three-dimensional model of red D magna Hb has a sub-30 A˚ resolution (Fig 9F)

In top view (Fig 9F, center), the molecule has the overall shape of a square Each side of the square has

a length of approximately 12 nm A single mass in the center of the molecule was observed However, this apparent mass may be an artefact of the negative staining procedure (see Discussion) In side views (Fig 9F, left and right), the molecule appears as a compressed sphere with a height of about 8 nm With-out additional information such as X-ray data, the molecule’s handedness cannot be defined, meaning that

it cannot be determined whether the real structure or its mirror image was reconstructed

Discussion Molecular mass and sedimentation coefficient

of Hb macromolecules Crustacean hemoglobins are large polymers with molecular masses between 240 and 800 kDa, and sedi-mentation coefficients between 11 and 19 S Reported

B

14.2

66 45 36 29 24 20.1

phosphate buffer

Marker (kDa) pale red

A

66

45

36

29

24

20.1

Tris buffer

Marker

(kDa) pale red

Fig 4 The influence of the buffer system used for gel

electrophor-esis on the band pattern of pale and red Daphnia magna

hemoglob-in (Hb) The hemolymph of pale (2.0 lg of Hb) and red (2.0 lg of

Hb) D magna was subjected to electrophoresis according to

Laemmli [19], using either Tris buffer (A) or sodium phosphate

buffer (according to Weber and Osborn [21]) (B) In the first case,

Hb was separated into three bands (pale D magna, 40.0, 38.1 and

35.1 kDa; red D magna, 40.0, 36.9 and 35.1 kDa), whereas in the

second case, only a single band at 39.2 kDa was detected.

values for the molecular mass of D magna Hb vary

considerably (494–670 kDa)

Molecular mass determinations by gel filtration and

gel electrophoresis yielded values between 500 kDa

[24] and 670 kDa [15] However, interactions between

protein molecules and matrix, whether beads (gel fil-tration) or polyacrylamide⁄ agarose (gel electrophor-esis), may cause false estimations of molecular mass Actually, it is not the molecular mass, but the effect-ive molecular radius, that determines the mobility of

m/z

2147.0 1921.1

1825.1

A

1659.3

2281.2 2433.2 2606.9

0

100

50

B

0

molecular mass (Da)

36482

100

50

Fig 5 Mass spectra of red Daphnia magna hemoglobin (Hb) purified by chromatofocus-ing (A) Charge state ESI-MS spectrum (B) MaxEnt-deconvoluted spectrum Prior to

MS, the Hb samples had been desalted and denatured by adding acetonitrile ⁄ water con-taining 0.2% formic acid In (A), seven of the 10 peaks representing the subunit monomer are labeled In (B), the molecular mass (Da) of the Hb subunit (without heme)

is given above the peak.

180 82

42

A

B

C D F

glycoprotein stain

A

B

RuBPs stain

A

B

C

D F

C

97 66

29

18

Fig 6 Staining for glycosylation of red Daphnia magna hemoglobin (Hb) Hb puri-fied by gel filtration (100 lg) was subjected

to two-dimensional gel electrophoresis and

a staining procedure specific for

glycosylat-ed proteins (A) Hb spots became visible after staining for glycosylated proteins Sub-sequently, the gel was stained with ruthen-ium II tris(bathophenanthroline disulfonate) (RuBPs) for total protein analysis (B) To prove the specificity of the staining tech-nique, the CandyCane molecular mass mar-ker (C) consisting of glycosylated and nonglycosylated proteins in alternating order was stained for glycoproteins (left lane, bold italicized digits: after staining with Pro Q Emerald 488) and total protein (right lane: after silver staining).

a macromolecule during gel filtration [25] The

accu-racy of SDS⁄ PAGE is also limited to about 10–

40%, because of, for example, unusual amino acid

composition, glycosylation or phosphorylation

[21,26,27] Covering the charge of native proteins by

Coomassie Brilliant Blue during gel electrophoresis

(blue native PAGE; BN-PAGE) resulted in a

molecular mass of about 600 kDa in the case of

native D magna Hb (data not shown)

During analytical ultracentrifugation, protein

molecules are freely dissolved and no interactions

with a matrix take place Provided that the partial specific volume of a protein analyzed is exactly known, the error of measurement is below 3% [20] Using the partial specific volume of D magna Hb (0.749 mLÆg)1; measured by Ilan et al [14]), sedimen-tation equilibrium experiments on red D magna Hb (Figs 1 and 2) revealed a molecular mass of 583 kDa The value used for the partial specific volume is sim-ilar to those of other invertebrate hemoglobins such

as those of Caenestheria inopinata (0.747 mLÆg)1 [28]), Lepidurus apus lubbocki (0.745 mLÆg)1 [29]), Lumbri-cus terrestris (0.740 mLÆg)1 [30]), Planorbis corneus (0.745 mLÆg)1 [31]), and Triops longicaudatus (0.743 mLÆg)1 [32]) The measured sedimentation coef-ficients were comparable (between 17.4 and 17.8 S) in all studies on D magna Hb [13,14] (this study) How-ever, the molecular mass suggested by Sugano and Hoshi [13] for D magna Hb (670 kDa) was not deter-mined by sedimentation equilibrium experiments, but was deduced from the measured sedimentation coeffi-cient (17.8 S) As they found identical sedimentation coefficients for D magna and Moina Hb, they conclu-ded that the molecular masses of D magna and Moina Hb (molecular mass: 660–670 kDa, determined

by ultracentrifugation [33]) were identical Previous sedimentation equilibrium experiments on D magna

Hb [14] gave a molecular mass of 505 ± 35 kDa in a first experiment and one of 483 ± 27 kDa in a sec-ond experiment, resulting in an overall molecular mass of 494 ± 33 kDa For the subunits of D magna

Hb, the reported molecular mass (about 31 kDa; also determined by ultracentrifugation) was distinctly lower than that from the ESI-MS data of our study (37.715 kDa) or the value deduced from amino acid sequence (36.2 kDa) Accordingly, molecular masses seemed to be generally underestimated in that previ-ous ultracentrifugation study, which may be due to the buffer system used (0.1 m sodium phosphate,

pH 6.8)

To additionally verify the results from the ultra-centrifugation experiments, the molecular mass of the

D magna Hb complex was determined by MALLS (Fig 3) Zhu et al [34] have shown that the deter-mination of molecular mass by MALLS could have

an error of measurement below 2% The determined values (590.4 ± 11.1 kDa for red D magna and 597.5 ± 49 kDa for pale D magna) agree well with the data from ultracentrifugation (583 kDa for red

D magna) The second peak in case of pale

D magna Hb could be a dissociation product of Hb,

as partial dissociation of D magna Hb at pH 9–10 yielded a 353 kDa fragment determined by gel filtra-tion [35]

control

A1A2

B1B2

C1C2

D1 D2

F1 F2

*

A

de-N

A1A2

B1B2

C1C2

D1 D2

F1 F2

*

B

de-O

*

C

de-N/O

*

D

Fig 7 Mobility shift assays after enzymatic deglycosylation of

Daphnia magna hemoglobin (Hb) To determine the type of

glycosy-lation (N- or O-linked), Hb was deglycosylated using different sets

of enzymes, and this was followed by two-dimensional gel

electro-phoresis (A) Control: addition of water instead of enzymes to the

reaction mixture (B) Hb incubated with N-glycosidase F (cleavage

of N-linked sugars) (C) Release of O-linked sugars (D) Release of

N- and O-linked sugars New spots occurring due to the removal of

O-linked or O- and N-linked glycans are marked with white circles.

One spot, which is usually not found in Hb subunit patterns, is

labeled with an asterisk.

Molecular mass of Hb subunits

The reported molecular mass of Daphnia Hb didomain

subunits varies between 31 and 40 kDa [14,24] Hb

proteins of twice this molecular mass have also been

reported for crustaceans (Lepidurus [36]; Daphnia pulex

[37,38]; Daphnia magna [15]), presumably resulting

from subunit dimerization For Triops longicaudatus

and Cyzicus Hb subunits, molecular masses between

15 and 21 kDa have been found [32,39]

To determine the molecular mass of protein

sub-units, SDS⁄ PAGE is often used because of its

advanta-geous properties (rapid and sample-saving) However,

a frequently neglected problem arises from the buffer

system used A common protocol makes use of a Tris

buffer system [19] Weber and Osborn [21] introduced

another buffer system (sodium phosphate buffer) For

Hb, the number of separated subunits is lower with

the Weber and Osborn method than with a Laemmli

SDS⁄ PAGE protocol For the subunits of Daphnia

pulexHb, Dangott and Terwilliger [37] determined five

bands using the Laemmli protocol, but only two

prom-inent bands employing the Weber and Osborn

proto-col In this study on D magna Hb subunits, three

bands were found using the Laemmli protocol and

only a single band with the Weber and Osborn

proto-col (Fig 4) Gielens et al [40] have reported that

hemocyanin and some other proteins bind only 0.7 g

sodium dodecyl sulfate (SDS) per g protein in

Tris⁄ HCl and Tris ⁄ glycine buffers, whereas 1.4 g SDS

per g protein is bound in phosphate buffers Rochu

and Fine [20] concluded that Tris ions may be

attrac-ted by amino acid residues, which are negatively

charged at the pH values used in SDS⁄ PAGE, and

accordingly, proteins may not be fully saturated with

the detergent, with the consequence that the lower

density of negative charges may not lead to a strictly mass-dependent electrophoretic mobility Moreover, Tris ions may decrease the number of SDS monomers

in solution by promoting the formation of micelles Consequently, during Laemmli SDS⁄ PAGE the separ-ation of polypeptides is influenced not only by the molecular mass but also by the degree of coverage of surface charges, which depends on the amino acid composition of proteins Phosphate buffers, however,

do not specifically interfere with the binding of SDS

to polypeptide chains, permitting their saturation with SDS [20] Accordingly, a Weber and Osborn SDS⁄ PAGE protocol may better reflect molecular mas-ses The deviating electrophoretic mobility of Hb sub-units in Tris-buffered gels despite similar molecular masses may also be caused by differences in their pri-mary structure Actually, it has been reported for a variant of serum prealbumin that a single-point muta-tion, which results in methionine instead of threonine, leads to a different migration behavior and an appar-ently lower mass of the variant prealbumin form dur-ing SDS⁄ PAGE [41] In addition, glycosylated proteins are known to show unusual migration behavior during gel electrophoresis [27] The observed difference in migration behavior of D magna Hb subunits in buffer systems according to Laemmli [19] compared to Weber and Osborn gels [21] may therefore indicate a post-translational modification of Hb subunits (see below)

To determine the molecular mass of D magna Hb subunits exactly, ESI-MS experiments were performed After deconvolution of the raw spectrum, a single peak was high above background level The molecular mass

of this major component was calculated to be 37.715 kDa, including two heme groups The difference between the detected mass and the molecular masses cal-culated from amino acid sequences may be caused by

Fig 8 GC mass analysis of carbohydrate moieties of Daphnia magna hemoglobin (Hb) Saccharides were released by methanolysis and separated by GC Compar-ison with a mannose standard (gray-shaded curve) identifies this saccharide as the dom-inant glycan in D magna Hb A significant increase of galactose and glucose in com-parison to the background level [dashed line ¼ maximum total ion current of the blank (inset)] was also observed One peak marked with an asterisk could not be identi-fied.

glycosylation (see below) Considering that the

relation-ship between Hb genes and Hb subunits is not yet

com-pletely determined, this post-translational modification

may also be the reason for the discrepancy between the

result of one dominating component in ESI-MS

meas-urements and the expected number of six gene products (i.e the maximum number assuming expression of all known Hb genes [12]) or at least four different subunit types that can be detected as main components by two-dimensional electrophoresis in Hb from red animals

*

A

C

B

D

E

F

*

*

Fig 9 Electron micrographs (A–D) and three-dimensional reconstruction (E–F) of Daphnia magna hemoglobin (Hb), which was purified by gel filtration and negatively stained with 2% uranyl acetate for electron microscopy or was purified by chromatofocusing and negatively stained with 5% ammonium molybdate containing 0.1% trehalose on holey carbon grids for three-dimensional reconstructions (A) Hb of pale ani-mals (B) Magnification of a section in (A) (C) Hb of red aniani-mals (D) Magnification of a section in (C) (Hb molecules showing the clover-leaf structure are indicated by arrows Ferritin molecules are indicated by asterisks Bar, 50 nm.) (E) After a multireference alignment, matchable images were treated by multivariate statistical analysis Three examples of the resulting class averages are shown in (E) which correspond approximately to the views of the three-dimensional reconstruction (F), showing the characteristic clover-leaf structure of red D magna Hb: presumed top (F, center) and two side views (F, left and right) of the three-dimensional reconstruction of D magna Hb at sub-30 A ˚ resolu-tion [Note different scaling of molecules in (E) and (F).] In the center of the Hb molecule, a single mass was observed (F, center) (White areas represent molecular masses.)

[42], respectively The subunits contributing to the Hb

of red D magna are mainly A, B, D and F, and

accord-ingly, more than one main subunit type would have been

expected in the ESI-MS spectrum Although three of

these subunits (A, B and D), comprising almost 70% of

the total subunits present in Hb from red Daphnia [42],

seem to be of similar size in both gel electrophoresis

sys-tems used, their expected size differences should have

been resolved by MS Again, glycosylation may

contrib-ute to the observed discrepancy Different glycan

struc-tures seem to generate a complex mixture of masses

leading to a variety of charge states, which are difficult

to resolve by ESI-MS Actually, under the experimental

conditions used for ESI-MS, only a single subunit type

could be detected reliably It is most likely that this

phe-nomenon may also be affected by electrospray

ioniza-tion suppression, which is typically observed in ESI-MS

of extracts from biological samples [43] However, the

value measured was used to calculate the number of

sub-units in the native aggregate

Based on the molecular mass of macromolecule and

subunits, the number of subunits per macromolecule

can be calculated For red D magna, the molecular

mass of the native Hb complex (ultracentrifugation,

583 kDa; MALLS, 590.4 kDa, red, and 597.5 kDa,

pale) divided by the subunit mass (ESI-MS:

37.715 kDa) results in 16 subunits per Hb

macromol-ecule, independent of the method applied for the

deter-mination of the Hb complex mass Because of an

underestimation of both macromolecule and subunit

mass (see above), Ilan et al [14] came to the same

con-clusion of 16 subunits per D magna Hb macromolecule

Hb glycosylation

The mean molecular mass of D magna Hb subunits

calculated from the nucleotide and the derived amino

acid sequences (Hb genes dmhb1–dmhb4 [11,12]) is

36.207 ± 0.027 kDa Accordingly, the experimentally

determined value for the predominant peak found in

ESI-MS is 275 Da higher (red D magna Hb subunits)

than the calculated value Actually, the subunits of red

D magna Hb were found to be glycosylated (Fig 6)

using the Pro-Q Emerald 488 stain Although Pro-Q

Emerald 300 dye is capable of detecting proteins with a

higher sensitivity (300–1 ng) and a broader dynamic

range (500–1000-fold), the Pro-Q Emerald 488 dye is the

most sensitive dye for detection of glycoproteins in gels,

when a laser-based gel scanner such as the FLA-2000 is

used for imaging [44] For proteins with a high

carbohy-drate content (9–42%), the detection sensitivity is

repor-ted to be between 5 and 9 ng with a linear dynamic

range of 128–255-fold Even glycoproteins with lower

carbohydrate content (3–7%) were successfully detected (sensitivity 19 ng, linear dynamic range 64-fold) [44] Protein isoforms that have the same amino acid sequence, but different glycosylation profiles, often appear as trains of spots on two-dimensional separa-tions, which can differ in pI and⁄ or molecular mass [45] Hence, in a next step, enzymatic deglycosylation was performed using different sets of N- and O-glyco-sidases to remove only one type of glycan or com-pletely remove all common glycans (Fig 7) Additional spots occurred when the Hb samples were treated with enzymes releasing specifically O-linked glycans or with all enzymes, respectively After treatment with N-gly-cosidase F, which cleaves only N-linked glycans, addi-tional spots were not found These results indicate that the carbohydrates of Hb are O-linked

This is in accordance with amino acid sequence analyses for determination of the glycosylation type Whereas no N-linked glycosylation sites are present in

D magna Hb [22], analyses using the NetOGlyc 3.1 server [23] revealed 11–12 potential O-linked glycosyla-tion sites in Hb subunits exclusively in the pre-A seg-ments This server is reported to correctly predict 76%

of glycosylated residues and 93% of nonglycosylated residues It is intended for extracellular proteins and can predict sites for completely new proteins without losing its performance [23]

In order to identify the carbohydrates bound to Hb subunits, saccharides were released by methanolysis and analyzed by GC followed by MS (Fig 8) Man-nose was identified as the dominant sugar, whereas galactose and glucose were found in smaller quantities Presently, directly O-linked mannose cannot be removed enzymatically However, the occurrence of additional spots in two-dimensional gel electrophoresis (Fig 7) indicates (a) that some mannose is indirectly bound to the protein, allowing a cleavage, and⁄ or (b) that the mobility shifts originate from the removal of galactose and glucose The fact that the spots A–D and F were still present after 3 h of incubation with deglycosylating enzymes can be explained by incom-plete deglycosylation or the presence of mannose directly bound to the proteins

Although glycosylation is a common post-transla-tional modification of the respiratory pigments of invertebrates [46], Hb subunits of Daphnia pulex were reported to be not glycosylated [38] However, the ana-lysis of distribution of glycosylation among taxonomic groups showed that even closely related species may not necessarily share close similarities in their glycan diversity [47]

In conclusion, all experimental results concerning the molecular mass of Hb subunits can be explained