Holo-lMb, nevertheless, is difficult to characterize experimentally as it has been Keywords heme binding; micro-myoglobin; molecular dynamics simulation; protein stability; unfolding Corr
Trang 1of micro-myoglobin
Hong-Fang Ji1,2, Liang Shen1,2, Rita Grandori3and Norbert Mu¨ller2
1 Shandong Provincial Research Center for Bioinformatic Engineering and Technique, Center for Advanced Study,
Shandong University of Technology, Zibo, China
2 Institute of Organic Chemistry, Johannes Kepler University, Linz, Austria
3 Dipartimento di Biotecnologie e Bioscienze, Universita degli Studi di Milano-Bicocca, Milan, Italy
The monomeric heme protein myoglobin (Mb) is
found mainly in muscle tissue [1], where its principal
physiological functions are oxygen storage and the
facilitation of oxygen transport to the mitochondria
for oxidative phosphorylation [2,3] The capability of
Mb to bind oxygen depends on the presence of a heme
prosthetic group, in which an iron(II) cation is
che-lated by the four nitrogen atoms in the center of a
protoporphyrin ring The metal ion can form two
additional coordinative bonds on either side of the
heme plane, termed the fifth and sixth coordination
positions, which are essential for the three-dimensional
structure and oxygen-binding function of the protein
Mb folds into a globular, single-domain structure
with a high a-helix content It comprises eight
right-handed a-helices (A–H, from the N- to C-terminus),
which are linked to each other by short loop regions
(Fig 1) [4]
It has long been recognized that the central portion
of the globin fold (mid-helix B to mid-helix G) forms a compact subdomain containing almost all the protein– heme contact sites [5] Recent studies have indicated that a fragment corresponding to such a portion of the structure is capable of folding into a functional heme-binding unit forming a complex with the prosthetic group with characteristics similar to native Mb [6,7] The deletion product was subcloned and expressed as
a 77-amino-acid fragment spanning residues 29–105 of sperm whale Mb, and was named micro-myoglobin (lMb) [6] The earlier experimental studies revealed that, in the absence of heme, this fragment is mostly disordered, but acquires a native-like a-helix content
on interaction with the cofactor [6] (We note here that
‘native-like’ in the context of this paper refers to the native fold of holo-Mb.) Holo-lMb, nevertheless, is difficult to characterize experimentally as it has been
Keywords
heme binding; micro-myoglobin; molecular
dynamics simulation; protein stability;
unfolding
Correspondence
N Mu¨ller, Institute of Organic Chemistry,
Johannes Kepler University, 4040 Linz,
Austria
Fax: +43 732 2468 8747
Tel: +43 732 2468 8746
E-mail: norbert.mueller@jku.at
(Received 28 May 2007, revised 29 October
2007, accepted 5 November 2007)
doi:10.1111/j.1742-4658.2007.06176.x
Micro-myoglobin, the isolated heme-binding subdomain of myoglobin, is a valuable model system for the investigation of heme recognition and bind-ing by proteins, and provides an example of protein foldbind-ing induced by co-factor binding Theoretical studies by molecular dynamics simulations on apo- and holo-micro-myoglobin show that, by contrast with the case of the full-length wild-type protein and in agreement with earlier experimental evi-dence, the apo-protein is not stably folded in a native-like conformation With the cofactor bound, however, the protein fragment maintains its folded conformation over 1.5 ns in molecular dynamics simulations Fur-ther inspection of the model structures reveals that the role of heme in sta-bilizing the folded state is not only a result of its direct interactions with binding residues (His93, Arg45 and Lys96), but also derives from its shield-ing effect on a long-range electrostatic interaction between Arg45 and Asp60, which, in the molecular dynamics simulations, apparently triggers the unfolding process of apo-micro-myoglobin
Abbreviations
Mb, myoglobin; MD, molecular dynamics; PDB, Protein Data Bank; lMb, micro-myoglobin.
Trang 2found to be prone to aggregation and precipitation at
high concentration By contrast, a longer Mb-derived
fragment, spanning residues 32–139, and named
mini-myoglobin (mini-Mb), is capable of independent
fold-ing into a native-like conformation, even in the
absence of heme [8] Other deletion products and
circu-lar permutations have also shown that the structural
determinants for heme binding can be segregated from
those for protein stability and solubility [9] Thus,
lMb represents a valuable and, so far, unique model
system for the investigation of heme recognition by
proteins and its possible role as a structural scaffold
for the minimal heme-binding subdomain
Molecular dynamics (MD) simulations can aid in
the understanding of the physical basis of the
struc-tural and functional features of biological
macromole-cules by providing details concerning intra- and
intermolecular motions as a function of time
Although results of MD simulations do not always
and easily translate to assessment of conformational
stability in a thermodynamic sense, they can be used
to address specific questions about the properties of a
model system, particularly for comparison of
mole-cular variants under controlled conditions Although
finding the thermodynamically most stable form of a
multiparticle system (and proving it) through MD
alone is generally impossible, instability revealed in
MD simulations is more straightforward to interpret
[10] In this study, we performed MD simulations on
‘heme-excised’ apo-lMb and holo-lMb with the
fol-lowing aims: (a) to compare the stability of native-like
structural models for apo-lMb and holo-lMb; (b) to
investigate the mechanisms by which heme affects
protein conformation and dynamics; and (c) to probe the transitions of each individual helix during the ini-tial step of thermal unfolding
Results and Discussion
Apo-lMb Snapshots of the MD simulation trajectory of apo-lMb at 0.0, 0.25, 0.5, 0.75, 1.0, 1.25 and 1.5 ns have been extracted and visualized in Fig 2 by ribbon
Fig 1 Ribbon representation of the crystallographic structure of
sperm whale myoglobin The eight helices of the myoglobin fold
are labeled A–H from the N- to C-terminus (PDB entry 1A6N).
Fig 2 Snapshots of apo-lMb (residues 29–105) and holo-lMb simulations extracted from the MD trajectories at 400 K Red, a-helix; green, random coil.
Trang 3drawings It can be seen that apo-lMb quickly moves
away from the initial native-like conformation, i.e it
unfolds easily over the timescale of the simulation
Therefore, we can conclude that the native-like fold is
not even a metastable conformation of this protein To
compare the unfolding of the individual helices of
apo-lMb, the average rmsd of each helix after 1.5 ns of
simulation was calculated over the backbone atoms
relative to the initial energy-minimized conformation
The values are 1.87 A˚ for helix B, 1.95 A˚ for helix C,
1.69 A˚ for helix D, 5.23 A˚ for helix E, 1.92 A˚ for helix
F and 2.07 A˚ for helix G A close look at Fig 2
reveals that helix G is the first to unfold, which is understandable because it spans only five residues despite its relatively low rmsd (2.07 A˚) However, it is interesting to find that, although helix E is the longest and is positioned close to the core of the sequence, it
is the second to unfold, immediately following helix G,
as indicated in Fig 2 This stimulated us to further inspect the unfolding process of helix E Previous investigations on native apo-Mb have indicated that helix E is the first to unfold (or the last to fold) [11,12] However, according to the present study in apo-lMb, helix G unfolds first This difference may
Fig 3 Variation of hydrogen-bond distances
in helix E during the 1.5 ns simulation of
apo-lMb at 400 K The distances are
mea-sured between CO(i) and NH(i + 4), where
i ranges from residue 58 to residue 73.
Trang 4arise from the fact that helix G is truncated in lMb,
which allows it to unfold faster than helix E, which is
complete in the fragment It should be noted, however,
that helix F has been found to be disordered in native
apo-Mb under neutral and equilibrium conditions [13]
A close look at Fig 2 reveals that helix E of
apo-lMb begins to unfold from its N-terminus at 0.5 ns,
and approximately one-half of the helix is disordered
at 1.0 ns Within the first 1.5 ns, helix E disappears
completely To describe helix E unfolding, 14
hydro-gen-bond distances [CO(i)–NH(i + 4)] that
character-ize the a-helix between residues 58 and 77 were
tracked throughout the MD simulation (Fig 3)
The first hydrogen bond that breaks is that between
CO(60) and NH(64), at the N-terminus of the helix
Starting at 0.3 ns, the CO(60)–NH(64) distance begins
to increase, reaching 6.0 A˚ at 0.4 ns, clearly indicating
that this hydrogen bond no longer exists
Subse-quently, significant fluctuations in the distances of
other hydrogen bonds are observed The CO(61)–
NH(65) and CO(62)–NH(66) hydrogen-bond distances
start to increase at 0.5 and 0.7 ns, respectively, which
is accompanied by a further increase in the distance
between CO(60) and NH(64) Subsequent break-up of
all the other hydrogen bonds within helix E ultimately
results in complete unfolding of this helix
Further inspection of the unfolding snapshots of
model structures of apo-lMb reveals that, during the
unfolding simulation, the distance between the
carbox-ylate oxygen of Asp60 (helix E) and the imino nitrogen
of Arg45 (helix C) decreases rapidly from 8.0 to 3.0 A˚
This can be interpreted as the formation of a salt
bridge between these two residues (Figs 4 and 5) The
occurrence of a salt bridge at approximately 0.5 ns
coincides with the beginning of the unfolding of helix E described above (Fig 2) Therefore, the forma-tion of this salt bridge between Arg45 and Asp60 seems to play an important role in triggering apo-lMb unfolding
Previous experimental studies have shown that apo-lMb has a disordered conformation in aqueous solu-tion at ambient temperature in the absence of heme [6], whereas, under the same conditions, the full-length apo-protein can fold into a native-like confor-mation and pre-organize the heme pocket before making contact with the cofactor [13] For compari-son, MD simulations were also performed on ‘heme-excised’ apo-Mb, under the same conditions and initial assumptions as for apo-lMb The average backbone rmsd for apo-Mb compared with the initial energy-minimized conformation is 3.51 A˚, much lower than that for apo-lMb: 4.99 A˚ (Fig 6) These data corroborate the effectiveness of MD simulations in
Fig 4 Variation of distances between the amino nitrogen of Arg45
and the carboxylate oxygen of Asp60 during the 1.5 ns simulation
of apo-lMb (residues 29–105).
Fig 6 Evolution of backbone rmsd during the 1.5 ns simulation of apo-lMb, ‘heme-excised’ apo-Mb and holo-lMb at 400 K, relative
to the respective initial structures.
Fig 5 Three-dimensional model of the salt bridge between Arg45 and Asp60 in apo-lMb based on the structure at approximately 0.5 ns of the simulation.
Trang 5discriminating between stable and unstable protein
conformations Thus, in agreement with the
experi-mental evidence [6], the results presented here show
that a native-like Mb conformation does not
repre-sent a stable state for the truncated lMb fragment in
the absence of heme
A comparison of the dynamics of helix E, in
par-ticular between apo-Mb and apo-lMb, also shows
considerable differences The backbone rmsd between
the starting and final structures of the fragment
corre-sponding to helix E is 3.18 A˚ for apo-Mb and 5.23 A˚
for apo-lMb Inspection of the Mb structure
indi-cates that, in apo-Mb, helix E is stabilized by several
salt bridges formed between helix E side-chains and
neighboring residues from helices A and B (Fig 7),
such as Glu4-Lys79, Glu18-Lys77 and Asp27-Lys56,
as also pointed out in a previous study [14] The
sta-bilizing effect of these salt bridges in apo-Mb
pre-vents fluctuations at the N-terminus of helix E, and
apparently also retards the formation of the salt
bridge between Asp60 and Arg45, which is an
impor-tant factor triggering apo-Mb unfolding By contrast,
these salt bridges are absent in apo-lMb because of
the absence of the helices A and B, providing a
possi-ble explanation for the distinct behavior of helix E in
apo-lMb and apo-Mb
Holo-lMb
In order to investigate the role of heme in protein
dynamics and stability, MD simulations were
per-formed on holo-lMb under the same conditions as
employed for apo-lMb In Fig 2, the extracted
snap-shots of holo-lMb simulation are shown juxtaposi-tioned to the corresponding positions of apo-lMb It
is evident that the holo-protein displays much reduced dynamics By contrast with the almost unfolded struc-ture of apo-lMb at the end of the simulation, most helices of holo-lMb are still folded at the end of the simulation period This remarkably different behavior
of the protein in the presence and absence of heme is also reflected by the respective rmsd values from a comparison of the backbone conformations before and after the MD simulation runs, which are 3.28 A˚ for holo-lMb and 4.99 A˚ for apo-lMb (Fig 6)
These findings provide some clues for the interpreta-tion of the experimental result that apo-lMb is almost unfolded [6], whereas the same fragment folds into a predominantly a-helical secondary structure on interac-tion with heme, thus forming a complex with the pros-thetic group with characteristics similar to the fold of native holo-Mb According to the crystal structure of holo-Mb [15], the heme ligand allows for coordination interaction between the iron ion and an imidazole ring nitrogen of His93, and its propionate groups are engaged in salt bridges with the basic side-chains of Arg45 and Lys96 (Fig 8) These interactions conceiv-ably also contribute to the conformational stability of holo-lMb Consistently, they are maintained over the 1.5 ns of the simulation (Fig 2)
A comparison of the hydrogen-bond distances in Figs 3 and 9 indicates that helix E also displays reduced dynamics in holo-lMb relative to its counter-part in apo-lMb As mentioned above, in apo-lMb, Arg45 forms a salt bridge with Asp60 in the first 0.5 ns of MD simulation, triggering the unfolding of helix E and the subsequent global unfolding of the structure In holo-lMb, the shielding effect of the heme and the engagement of Arg45 in the salt bridge
Fig 7 Three-dimensional model of salt bridges Glu4-Lys79,
Glu18-Lys77 and Asp27-Lys56 in apo-Mb (PDB entry 1A6N).
Fig 8 Three-dimensional model of the salt bridges made by Arg45 and Lys96 with the two propionate groups of heme in holo-lMb.
Trang 6with the heme propionate prevent the build-up of
attractive electrostatic interactions between Arg45 and
Asp60 that are seen to lead to apo-lMb unfolding
Therefore, in addition to direct interaction with
bind-ing residues, heme seems to stabilize the lMb folded
state also by counteracting long-range electrostatic
interactions that would otherwise act as destabilizing
forces for native-like conformations
Conclusions
The comparison of apo- and holo-lMb by MD
simu-lations reveals a dramatic effect of the heme cofactor,
which is seen to stabilize native Mb-like folded
confor-mations of the protein fragment, in agreement with the
available experimental evidence The simulation results
suggest that, in the absence of heme, the unfolding process is triggered by the formation of a non-native salt bridge between Arg45 and Asp60 The presence of heme at the active site counteracts this unfolding mechanism, in addition to providing stabilizing inter-actions with the folded chain per se Although the starting conformation used here for apo-lMb simula-tion has not been observed experimentally, it can be taken as a model of an unstable intermediate in holo-lMb unfolding on loss of the cofactor The MD simu-lation results provide an explanation of why such a conformation is not a stable state for apo-lMb Electrostatic interactions, in general, have attracted considerable interest in protein folding studies, as it has been shown that surface charges can play an important role in protein stability [16] Nevertheless,
Fig 9 Variation of hydrogen-bond distances
in helix E during the 1.5 ns simulation of holo-lMb at 400 K The distances are between CO(i) and NH(i + 4), where
i ranges from residue 58 to residue 73.
Trang 7the contribution of engineered salt bridges to protein
stability is frequently smaller than predicted on the
basis of theoretical models [17] This observation has
been interpreted to be the result of the survival of
partially folded structures stabilized by electrostatic
interactions in the denatured state [18] Indeed, the
available tools for modeling protein conformations in
the denatured state are still insufficient However, we
have good reasons to assume that, in the presence of
denaturants or, even more so, at high temperatures,
electrostatic interactions can contribute to maintain
residual structure in the denatured state The
perva-siveness of structured clusters under denaturing
condi-tions has also been found experimentally for different
proteins [19–21] The results reported here strengthen
the notion of the structural complexity of proteins in
the denatured state and provide an example of the role
of non-native ion pairs in protein unfolding
Experimental procedures
The MD simulations on holo- and apo-lMb were
per-formed in parallel The starting structure of holo-lMb was
obtained from the highest resolution X-ray crystallographic
structure of sperm whale Mb [Protein Data Bank (PDB)
entry 1A6N] [15] [which matches well the 12 structure
bun-dle obtained earlier by NMR (PDB entry 1MYF)] [22], by
excision of the corresponding fragment (residues 29–105)
whilst retaining the bound heme The structure of apo-lMb
was obtained by additional excision of the heme ligand
Likewise, the starting geometry for the simulations of
apo-Mb were obtained by deleting the heme from the above
crystal structure All the simulated structures were
immersed in two layers (20 A˚ for the inner and 15 A˚ for
the outer) of explicit water molecules The inner layer was
dynamic, whereas the outer layer was static, and served as
a solvent boundary to prevent the escape of the inner layer
water molecules All of the simulated structures were
calcu-lated for a 1.5 ns MD simulation in a neutral environment
The initial structures were first energy minimized by 1000
conjugate gradient steps, and then heated from 2 to 300 K
over 35 ps, with temperature increments of 50 K per 5 ps,
and kept at 300 K for 20 ps using the constant pressure
and temperature algorithm The velocity Verlet integrator
was used with an integration step of 2 fs As it has been
reported that a few nanoseconds are sufficient to assess the
relative stabilities of the initial structures [23], the
produc-tion MD phase (i.e the unfolding) was carried out for
1.5 ns, which was sufficient to reach the quasi-equilibrium
states of both holo- and apo-lMb, using 2 fs steps and
9.5 A˚ cut-off, as demonstrated by plateaus in the MD
tra-jectories To accelerate the unfolding processes of both
pro-teins, the simulations were performed at 400 K During the
simulations, the non-bonded interaction cut-off was set to
12 A˚, which is sufficiently large to include long-range elec-trostatic interactions Main-chain hydrogen bonds within a-helices were assigned when the distance between the CO group of residue i and the NH group of residue i + 4 was shorter than 2.5 A˚ A salt bridge was identified when the distance between the nitrogen of the base and the oxygen
of the carboxylate was shorter than 5.0 A˚ [24] Structures were saved every 0.5 ps for a total of 3000 snapshots for each trajectory All of the constant volume constant tem-perature MD simulations were performed by the Discover module of the insightii program package (Accelys Inc., San Diego, CA), which has been widely employed in MD simulation studies [25–28] The consistent valence force field [29–32] was used for all of the simulations The calculations were carried out on an SGI Origin 3800 server (Silicon Graphics Inc., Sunnyvale, CA)
Acknowledgements
We thank Elena Papaleo, Luca De Gioia and Stephan Schwarzinger for helpful discussions and critical read-ing of the manuscript This research was supported in part by the European Union in the REGINS-INBIO program and the Austrian Science Funds (project P15380) Hardware and software support was provided
by the Supercomputing Department at Johannes Kep-ler University, Linz, Austria
References
1 Cole RP (1982) Myoglobin function in exercising skeletal muscle Science 216, 523–525
2 Wittenberg BA & Wittenberg JB (1989) Transport of oxygen in muscle Annu Rev Physiol 51, 857–878
3 Wittenberg BA, Wittenberg JB & Caldwell PB (1975) Role of myoglobin in the oxygen supply to red skeletal muscle J Biol Chem 250, 9038–9043
4 Antonini E & Brunori M (1971) Hemoglobin and Myoglobin in their Reactions with Ligands North Holland Publishers Co., Amsterdam
5 Go M (1981) Correlation of DNA exonic regions with protein structural units of hemoglobin Nature 291, 90–92
6 Grandori R, Schwarzinger S & Mu¨ller N (2000) Cloning, overexpression and characterization of micro-myoglobin, a minimal heme-binding fragment Eur J Biochem 267, 1168–1172
7 Schwarzinger S, Ahrer W & Mu¨ller N (2000) Proteolytic preparation of a heme binding fragment from sperm whale myoglobin: micro-myoglobin Monatsh Chem 131, 409–416
8 DeSanctis G, Falcioni G, Giardina B, Ascoli F & Brunori
M (1988) Mini-myoglobin The structural significance of heme–ligand interactions J Mol Biol 200, 725–733
Trang 89 Ribeiro EA Jr & Ramos CH (2005) Circular
permuta-tion and delepermuta-tion studies of myoglobin indicate that
the correct position of its N–terminus is required for
native stability and solubility but not for native-like
heme binding and folding Biochemistry 44, 4699–
4709
10 Karplus M & McCammon JA (2002) Molecular
dynam-ics simulations of biomolecules Nat Struct Biol 9, 646–
652
11 Onufriev A, Case DA & Bashford D (2003) Structural
details, pathways, and energetics of unfolding
apomyo-globin J Mol Biol 325, 555–567
12 Nishimura C, Dyson HJ & Wright PE (2002) The
apomyoglobin folding pathway revisited: structural
heterogeneity in the kinetic burst phase intermediate
J Mol Biol 322, 483–489
13 Eliezer D, Yao J, Dyson HJ & Wright PE (1998)
Structural and dynamic characterization of partially
folded states of apomyoglobin and implications for
protein folding Nat Struct Biol 5, 148–155
14 Ramos CH, Kay MS & Baldwin RL (1999) Putative
interhelix ion pairs involved in the stability of
myoglo-bin Biochemistry 38, 9783–9790
15 Vojtechovsky´ J, Chu K, Berendzen J, Sweet RM &
Sch-lichting I (1999) Crystal structures of myoglobin–ligand
complexes at near-atomic resolution Biophys J 77,
2153–2174
16 Sanchez-Ruiz JM & Makhatadze GI (2001) To charge
or not to charge? Trends Biotechnol 19, 132–135
17 Nakamura H (1996) Roles of electrostatic interaction in
proteins Q Rev Biophys 29, 1–90
18 Pace CN, Trevino S, Prabhakaran E & Scholtz JM
(2004) Protein structure, stability and solubility in water
and other solvents Philos Trans R Soc Lond B Biol Sci
359, 1225–1234
19 Pace CN, Alston RW & Shaw KL (2000) Charge–
charge interactions influence the denatured state
ensem-ble and contribute to protein stability Protein Sci 9,
1395–1398
20 Millet IS, Townsley LE, Chiti F, Doniach S & Plaxco
KW (2002) Equilibrium collapse and the kinetic
‘fold-ability’ of proteins Biochemistry 41, 321–325
21 Denisov VP, Jonsson BH & Halle B (1999) Hydration
of denatured and molten globule proteins Nat Struct
Biol 6, 253–260
22 Osapay K, Theriault Y, Wright PE & Case DA (1994) Solution structure of carbon monoxymyoglobin deter-mined from nuclear magnetic resonance distance and chemical shift constraints J Mol Biol 244, 183–197
23 Fan H & Mark AE (2003) Relative stability of protein structures determined by X-ray crystallography or NMR spectroscopy: a molecular dynamics simulation study Proteins Struct Funct Bioinform 52, 111–120
24 Kumar S & Nussinov R (2002) Relationship between ion pair geometries and electrostatic strengths in pro-teins Biophys J 83, 1595–1612
25 Iyer KL & Qasbal PK (1999) Molecular dynamics simulation of a-lactalbumin and calcium binding c-type lysozyme Protein Eng Des Sel 12, 129–139
26 Fritze O, Filipek S, Kuksa V, Palczewski K, Hofmann
KP & Ernst OP (2003) Role of the conserved NPxxY(x)5,6F motif in the rhodopsin ground state and during activation Proc Natl Acad Sci USA 100, 2290– 2295
27 Lins RD, Briggs JM, Straatsma TP, Carlson HA, Greenwald J, Choe S & McCammon JA (1999) Molecu-lar dynamics studies of the HIV-1 integrase catalytic domain Biophys J 76, 2999–3011
28 Ji HF, Zhang HY & Shen L (2005) The role of electro-static interaction in triggering the unraveling of stable helix 1 in normal prion protein A molecular dynamics simulation investigation J Biomol Struct Dyn 22, 563– 570
29 Dauber-Osguthorpe P, Roberts VA, Osguthorpe DJ, Wolff J, Genest M & Hagler AT (1988) Structure and energetics of ligand binding to proteins: E coli dihy-drofolate reductase–trimethoprim, a drug–receptor sys-tem Proteins 4, 31–47
30 Hwang MJ, Ni X, Waldman M, Ewig CS & Hagler AT (1998) Derivation of class II force fields VI Carbohy-drate compounds and anomeric effects Biopolymers 45, 435–468
31 Peng Z, Ewig CS, Hwang MJ, Waldman M & Hagler
AT (1997) Derivation of Class II force fields 4 van der Waals parameters of alkali metal cations and halide anions J Phys Chem A 101, 7243–7252
32 Maple JR, Hwang MJ, Jalkanen KJ, Stockfisch TP & Hagler AT (1998) Derivation of Class II force fields 5 Quantum force field for amides, peptides, and related compounds J Comput Chem 19, 430–458