The impact of mutations on the protein structure, in the context of generating a gp41 based vaccine antigen that resembles a fusion intermediate state, is dis-cussed.. The fusogenic stat
Trang 1HIV-1 gp41 and gp160 are hyperthermostable proteins
in a mesophilic environment
Characterization of gp41 mutants
Tino Krell1, Fre´de´ric Greco1, Olivier Engel1, Jean Dubayle1, Joseline Dubayle1, Audrey Kennel1,
Benoit Charloteaux2, Robert Brasseur2, Michel Chevalier1, Regis Sodoyer1and Raphae¨lle El Habib1
1 Aventis Pasteur, Marcy l’Etoile, France; 2 Centre de Biophysique Mole´culaire Nume´rique, Faculte´ Universitaire des Sciences Agronomiques de Gembloux, Belgium
HIV gp41(24–157) unfolds cooperatively over the pH range
of 1.0–4.0 with Tmvalues of > 100C At pH 2.8, protein
unfolding was 80% reversible and the DHvH/DHcalratio of
3.7 is indicative of gp41 being trimeric No evidence for a
monomer–trimer equilibrium in the concentration range of
0.3–36 lM was obtained by DSC and tryptophan
fluores-cence Glycosylation of gp41 was found to have only a
marginal impact on the thermal stability Reduction of the
disulfide bond or mutation of both cysteine residues had
only a marginal impact on protein stability There was no
cooperative unfolding event in the DSC thermogram of
gp160 in NaCl/Pi, pH 7.4, over a temperature range of
8–129C When the pH was lowered to 5.5–3.4, a single
unfolding event at around 120C was noted, and three
unfolding events at 93.3, 106.4 and 111.8C w ere observed
at pH 2.8 Differences between gp41 and gp160, and hyperthermostable proteins from thermophile organisms are discussed A series of gp41 mutants containing single, dou-ble, triple or quadruple point mutations were analysed by DSC and CD The impact of mutations on the protein structure, in the context of generating a gp41 based vaccine antigen that resembles a fusion intermediate state, is dis-cussed A gp41 mutant, in which three hydrophobic amino acids in the gp41 loop were replaced with charged residues, showed an increased solubility at neutral pH
Keywords: gp160; gp41; HIV; hyperthermostability; site-directed mutagenesis
HIV entry is mediated by the viral envelope proteins gp41
and gp120 Both proteins are derived from the gp160
precursor following proteolytic cleavage [1] After cleavage,
both proteins remain associated [2] and are trimeric in their
prefusogenic state [3] gp41 anchors the protein complex to
the viral membrane [4], whereas gp120 binds to the human
cell-surface receptor CD4 and other receptors on the target
cell This interaction leads to a dissociation of gp120 from
gp41 [5], which induces a conformational change in gp41 [2],
resulting in the fusion of viral and cellular membranes
Both envelope proteins are vaccine candidates against
HIV However, the results of the world’s first phase III
efficacy trial using gp120 were relatively modest [6] and
efforts are nowbeing made to explore the vaccine potential
of gp41 The sequence of gp41 contains four functional
regions, as follows: the N-terminal fusion peptide (which
is thought to insert into the host cell membrane) is followed
by an ectodomain, a transmembrane region and a
cyto-plasmic domain
The ectodomain of gp41 is thought to adopt at least two
different tertiary structures [7]: a prefusogenic state, and a
lower energy fusion-active (fusogenic) state The decrease in free energy during the transition into the latter state is thought to account for the energy necessary for the fusion of viral and cellular membranes There is nowevidence that the recombinantly produced gp41 ectodomain adopts sponta-neously this fusogenic state, and 3D structural information
on gp41 is entirely on this lower energy state [8–11] The fusogenic state of the gp41 ectodomain is a trimeric coiled-coil protein in which the three helices C pack in an antiparallel manner against the central trimer of parallel helices N
It has been shown that gp41 produced in Escherichia coli forms insoluble aggregates at neutral pH [12], and aggre-gation is proposed to occur at the loop region connecting helices N and C [13] This is consistent with the fact that recombinant constructs of gp41, which have this loop replaced with a small flexible linker (gp41 models), are soluble at neutral pH Therefore, a large proportion of biochemical and biophysical studies on gp41 have been carried out using either a gp41 model [14–16] or a stoichiometric mixture of peptides corresponding to the N- and C-terminal helices, which assemble to hetero-hexamers in a native-like manner [17,18] However, recom-binant gp41 was shown to be soluble at a pH of < 3.5, and several studies have been carried out at acidic pH [19,20] The NMR structure of gp41, at pH 3.0 [9], and its X-ray structure, using crystals grown at pH 4.25 [21], are identical
to structures of gp41 fragments determined at pH values of 4.6–8.0 [8,11,16,22] This demonstrates that the structure of gp41 is not altered by acidic pH
Correspondence to T Krell, Aventis Pasteur, 1541 avenue Marcel
Me´rieux, 69280 Marcy l’Etoile, France Fax: + 33 4 37 37 31 80,
Tel.: + 33 4 37 37 90 12, E-mail: tino.krell@aventis.com
Abbreviations: SIV, simian immunodeficiency virus; TCEP,
Tris(2-carboxyethyl)phosphine.
(Received 8 January 2004, revised 1 March 2004,
accepted 3 March 2004)
Trang 2Based on the evidence that the structure of gp41 is not
altered by acidic pH, and the functional importance of the
loop region [23] with its conserved cysteine residues [24],
we have carried out our analyses at acidic pH using
recombinant gp41(24–157) containing the cysteine
resi-dues
The potent HIV entry inhibitor 5-helix [25], a
recombin-ant gp41 trimer lacking a helix C, has been shown to bind
tightly to individual C peptides It has been proposed that
the mechanism of action of 5-helix is based on the binding to
a fusion intermediate of gp41, which is characterized by an
accessible and exposed helix C [26] These data can be
regarded as evidence that a protein is capable of blocking
fusion intermediates of gp41, which consequently leads to
the inhibition of virus entry Our vaccine approach is based
on the generation of gp41 mutants, which resemble the
fusion intermediate state It is generally accepted that this
intermediate state is trimeric and that helices N and C do
not interact Here we explore ways to stabilize the
interaction between helices N, to favour a trimeric state
and, on the other hand, to destabilize the interaction
between helices N and C in order to prevent helix contact
The study of recombinant gp41, either as separate
glycosylated or nonglycosylated protein, or as part of
gp160, by biophysical techniques, forms the first part of this
article In the second part, site-directed mutagenesis data of
gp41 are presented with the aim of assessing the influence of
amino acid replacements on protein stability and solubility
Materials and methods
Materials
Gp41(24–157) Recombinant ectodomain of gp41
corres-ponding to amino acids 537–669 of the envelope protein of
the LAI isolate (p03375) with a C-terminal extension of
amino acids GGGGSHHHHHH For details of protein
expression and purification see below
Gp41(34–170) Glycosylated recombinant ectodomain of
gp41 purchased from Tebu-bio (Le Perray en Yvelines
Cedex, France) corresponding to amino acids 546–682 of
the envelope protein of the HxB2 isolate (p04578) The
protein has been expressed in Pichia pastoris The gp41
domains of isolates BH10 and HxB2 share 99% sequence
identity
Gp160 Fusion of amino acids 30–500 of the envelope
protein of the isolate MN (comprising gp120) with amino
acids 501–744 (comprising a large part of gp41) of the LAI
isolate The amino acid sequence of the gp120–gp41
cleavage site KAKRRVVQREKR (502–513 in the LAI
sequence) has been altered to KAQNHVVQNEHQ This
change prevents cleavage of gp160 into gp120 and gp41
This protein has been expressed in vaccinia virus grown on
BHK21 cells and was purified by affinity chromatography
on immobilized antibodies Further details on the
construc-tion and biochemical characterizaconstruc-tion of this protein are
found in Kieny et al [27] This protein has been used for
several HIV vaccine trials [28,29]
TCEP/HCl [Tris(2-carboxyethyl)phosphine
hydrochlo-ride] was purchased from Pierce
Cloning procedures
A 0.6 kb DNA fragment, containing the sequence encoding the ectodomain of gp41 of HIV isolate LAI (amino acids 537–669), was obtained by PCR amplification using, as template, a plasmid containing the gp120 sequence of MN and the gp41 sequence of the LAI isolate Restriction sites for BspHI and XhoI (shown in italics) were, respectively, included in the forward and reverse primers, as follows: forward primer: 5¢-CTCTTTCATGACGCTGACGGTA CAGGCC-3¢; reverse primer: 5¢-CCGCTCGAGCTAATG GTGATGGTGATGGTGTGACCCTCCCCCTCCACT TGCCCATTTATCTAA-3¢ The start codon in the for-ward primer (in bold) is a naturally occurring methionine residue A stop codon (in bold) and the DNA sequence encoding the extension GGGGSHHHHHH (underlined) were added to the reverse primer Platinum HF polymerase (Gibco Invitrogen Corp.) was used, according to the manufacturer’s instructions, for PCR amplification The PCR amplified fragment was cloned directly into the vector pM1800 using the restriction sites NcoI and XhoI The expression vector pM1800 is a derivate of pET28c (Novagen), in which the F1 origin of replication has been deleted and replaced with the Cer fragment, allowing for multimer resolution
Site-directed mutagenesis: single and multiple point mutations
The various point mutations were created using the Quick-change site-directed mutagenesis kit (Stratagene), using the instructions provided by the manufacturer Mutations were carried out directly on the expression vector containing the sequence of native gp41, and multiple mutations were introduced in a sequential manner
NC and CN fusion mutants The NC and CN gp41 fusion constructs were generated using three consecutive PCRs In the first PCR (for the NC constructs) the following two partially overlapping DNA fragments were generated: a fragment corresponding to helix N containing a C-terminal extension encoding the first six amino acids of the helix C and a fragment corresponding
to helix C containing an N-terminal extension coding for the last six amino acids of helix N In the second reaction, a stoichiometric mix of both partially overlapping fragments (no primers added) was submitted to 10 PCR cycles The third reaction was carried out with the product of the second reaction in the presence of primers containing NcoI or XhoI restriction sites and which are complementary to the 5¢ end
of helix N and to the 3¢ end of helix C An equivalent approach was used to generate the CN fusion constructs Expression
For protein expression, E coli BL21(DE3) was transformed with the corresponding plasmids Typically, 1 L cultures were grown at 37C on LB (Luria–Bertani) broth supple-mented with kanamycin (25 lgÆmL)1) Protein expression was induced at an attenuance (D) of 0.6 at 600 nm, by the addition of isopropyl thio-b- -galactoside (Q-BIOgene,
Trang 3Illkirch, France) to a final concentration of 1 mM Bacteria
were harvested, 4 h after protein induction, by
centrifuga-tion For certain mutants, protein expression was optimized
by the replacement of LB medium with Terrific Broth
medium, induction with 0.02–0.1 mM isopropyl
thio-b-D-galactoside and a growth temperature of 30C
Protein purification
The bacterial pellet resulting from a 1 L culture was
resuspended at room temperature in 95 mL of 50 mM
Tris/HCl containing 100 lMCompleteTMEDTA-free
pro-tease inhibitor cocktail (Roche Molecular Biochemicals) and
100 lgÆmL)1lysozyme (Sigma-Aldrich), pH 8.0, and gently
agitated for 30 min The bacterial suspension was then
placed on ice and cell lysis was achieved by ultrasound
treatment (4· 2 min) using a Branson-Sonifer 450
After-wards, MgCl2and Benzonase (Merck) were added to final
concentrations of 1 mM and 1 UÆmL)1, respectively The
resulting solution was then centrifuged (20 000 g, 30 min,
4C) Aliquots of the resulting supernatant and pellet were
analysed by SDS/PAGE and recombinant proteins were
detected by Western blot analysis using a monoclonal
antibody raised against poly-histidine (Novagen)
Recom-binant protein was found to be almost exclusively present in
the pellet The pellet was resuspended in 100 mL of buffer A
(50 mMTris/HCl, 8Murea, 500 mMNaCl, 10 mM
imida-zole, pH 8.0) and agitated at 4C for 30 min After filtration
using a 0.45 lm cut-off filter, protein was loaded onto a 5 ml
Hi-Trap Chelating column (Amersham Pharmacia Biotech),
previously equilibrated in buffer A After washing with
50 mL of buffer A, protein elution was achieved using a
buffer comprising 50 mMTris/HCl, 8Murea, 500 mMNaCl
and 500 mM imidazole, pH 8.0 Protein refolding was
achieved by dialysis with 50 mMformate, pH 2.8 Protein
was then sterile filtered and stored at)45 C
MALDI-TOF MS
Mass spectrometric analyses were carried out on a Biflex III
MALDI-TOF mass spectrometer (Bruker Daltonics,
Wiss-embourg, France) Samples of native and mutant gp41
(1–3 mgÆmL)1) in 50 mM formate, pH 2.8, were diluted
with 30% acetonitrile (v/v) containing 0.07% trifluoroacetic
acid to a final concentration of 0.4 mgÆmL)1 (protein
solution) A saturated solution of sinapic acid
(Sigma-Aldrich) in 70% acetonitrile (v/v) containing 0.1%
trifluoro-acetic acid was prepared and subsequently diluted
four-fold with the same solvent (matrix solution) Droplets
(1 lL) of a 1 : 1 (v/v) mixture of protein and matrix solution
were deposited on the sample slide and allowed to dry at
room temperature Positive ion mass spectra were acquired
in the linear mode with pulsed ion extraction Mass
assignments were based on an external calibration of the
instrument
DSC
DSC experiments were performed on a MicroCal VP-DSC
apparatus (MicroCal, Northampton, MA, USA) Prior to
analysis, proteins were exhaustively dialyzed against the
buffer stated in the legend of each figure, and degassed The
dialysis buffer was used for baseline scans and was present
as a reference buffer for the protein scans The system was allowed to equilibrate at 5C for 15 min, and temperatures from 5 to 129C were scanned at a rate of 85 C/h Thermograms obtained were analysed using the MicroCal version of ORIGIN The standard deviation indicated for each parameter corresponds to the error of curve fitting Details concerning the calculation of thermodynamic parameters and instrumentation have been published pre-viously [30,31]
CD Far-UV CD measurements were made at 25C in a Jasco J-810 spectropolarimeter (Tokyo, Japan) using cuvettes with a pathlength of 0.1 mm Proteins were exhaustively dialyzed against 50 mMformate, pH 2.8 All proteins were analysed at a concentration of 60 lM and spectra were corrected using the spectra of the dialysis buffer
Fluorescence spectroscopy Measurements of the intrinsic tryptophan fluorescence were carried out at 25C using a Kontron SFM 25 spectrofluorimeter (Kontron, Zurich, Switzerland) Unless stated otherwise, proteins were present at a concentration
of 1 lM in 50 mM formate, pH 2.8 Emission spectra between 300 and 400 nm were collected after excitation at
295 nm Spectra were corrected using the spectra of the buffer
Results
Protein expression of native and mutant gp41(24–157) Histidine tagged gp41(24–157), and a substantial number of mutants, have been expressed in E coli After centrifugation
of the cell lysate, all proteins were present in the pellet, which was solubilized in buffer containing 8Murea Purification was also carried out in the presence of chaotropic agents and refolding was achieved by a simple dialysis into 50 mM
formate, pH 2.8 The DSC analysis of gp41(24–157) did not provide evidence of partially or wrongly folded protein (see below)
Typically, a yield of 60 mg of pure protein per litre of cell culture was observed for native gp41(24–157) This yield appeared to be lower for certain gp41 mutants Multiple point mutations to the loop region, e.g as in the triple mutant L91K/I92K/W103D (see below), did not change the protein yield
Analysis of gp41(24–157) by CD, DSC and MALDI-TOF MS The far-UV CD spectrum of gp41(24–157) in 50 mM
formate, pH 2.8, is shown in Fig 1 The spectrum is literally superimposable to the CD spectrum of glycosylated gp41(21–166) at pH 7.5, as reported by Weissenhorn et al [32] Both spectra showminima at around 208 and 222 nm,
a crossover in sign at 202 nm, and a maximum at 193 nm, which are typical characteristics of a protein largely dominated by a-helix From the molecular ellipticity at
222 nm, an a-helix content of 75% has been calculated [33],
Trang 4which is consistent with the a-helix content of 80%
determined by Weissenhorn et al for glycosylated gp41(21–
166) at pH 7.5 [32]
Figure 2A (upper trace) shows the DSC thermogram of
gp41(24–157) after the renaturation process by dialysis into
50 mM formate, pH 2.8 (see the Materials and methods)
Two unfolding transitions at 110.4 and 119.5C are seen
(Table 1), demonstrating that this protein is
hyperther-mostable at lowpH The thermal unfolding was highly
cooperative with an DHvH/DHcalratio of 3.7 for the major
unfolding event, which is consistent with a cooperative
unfolding of a gp41 trimer Evidence that gp41 is trimeric
at pH 2.5–3 has previously been demonstrated using gel
filtration [13], analytical ultracentrifugation [19] and NMR
[9]
A major peak corresponding to the monomer, and two
smaller peaks corresponding to covalently linked dimeric
and trimeric forms of gp41(24–157), are seen in the
MALDI-TOF spectrum (Fig 2B) of the same sample It
is generally accepted that MALDI-TOF analysis results in
the disruption of all noncovalent interactions and that the
observed multimers correspond to covalently linked
multi-mers gp41(24–157) contains two cysteine residues which are
involved in an intrasubunit disulfide bond in the native
protein [34] The reaction of this protein sample with
Ellman’s reagent [35] showed that the two cysteine residues
are engaged in disulfide bonds (data not shown) This is
consistent with the monomer, observed by MS, having an
intrasubunit disulfide bond, whereas the dimer and trimer
peaks indicate the presence of intersubunit disulfide bonds
Fig 1 Far UV CD spectra of native gp41(24–157) (––) and of the
quadruple mutant W6OA/I124D/I131D/Q142N (- - - -) Proteins were
analyzed at a concentration of 60 l M in 50 m M formate, pH 2.8.
Fig 2 Analysis of native gp41(24–157) by DSC and MALDI-TOF
MS (A) DSC thermograms of gp41(24–157) before and after
reduc-tion with tris(2-carboxyethyl)phosphine (TCEP) Derived
thermo-dynamic parameters are shown in Table 1 (B) MALDI-TOF mass
spectra of gp41(24–157) before and after reduction with TCEP In
panels (A) and (B), the same samples were used for analysis The
sequence-derived mass of gp41(24–157) is 16801.5 l Spectra indicate
that the N-terminal methionine residue has been processed (C) Study
of the reversibility of the thermal unfolding of gp41(24–157) Shown
are segments of two consecutive DSC up-scans from 5 to 129 C.
Reversibility was defined as: % reversibility ¼ (DH cal2 /DH cal1 ) ·
100%, with DH cal2 being the change of enthalpy from the second
up-scan and DH cal1 the change of enthalpy from the first up-scan of the
same protein sample Reversibility data are given in Table 1 For
clarity reasons, DSC thermograms and mass spectra are moved
arbitrarily on the y-axis Proteins were analyzed in 50 m M formate,
pH 2.8.
Trang 5which have been described previously for glycosylated
gp41(21–166) [32] To verify the hypothesis of the presence
of intersubunit disulfide bonds, a gp41(24–157) sample was
analysed after reduction with TCEP, a reagent known to
reduce disulfides selectively over a pH range of 1.5–8.5 [36]
After reduction, only a single unfolding transition was
observed by DSC, which was characterized by a downshift in
Tmof 1.3C, with respect to the major peak before reduction
(Fig 2A, Table 1) This event was equally very cooperative
(DHvH/DHcal¼ 3.3), demonstrating that reduction does not
alter the trimeric state of gp41 The peaks corresponding
to covalent dimers and trimers in the mass spectrum of this
sample were significantly decreased as compared to the
sample analysed before reduction (Fig 2B) These minor
peaks can probably be attributed to the formation of
covalent multimers during ionization in the instrument, a
well-known phenomena of this technique, because multimer
peaks of similar size are observed in the spectrum of the
cysteine-free double mutant, C87S/C93S (data not shown)
This implies that the single transition seen in DSC after
reduction corresponds to the unfolding of noncovalently
associated trimers with reduced disulfide bonds, whereas
the major unfolding event before reduction represents the
unfolding of noncovalently associated trimers with
intra-subunit disulfide bonds The effect of disulfide reduction on
the thermal stability of gp41 trimers is thus relatively modest
(downshift in Tm of 1.3C) The absence of the second
unfolding transition after reduction demonstrates that this
event represents the unfolding of gp41 trimers characterized
by one or several intersubunit disulfide bonds
In summary, gp41(24–157) is, after renaturation, mainly
present in its native conformation, defined by noncovalently
associated trimers with intrasubunit disulfide bonds Based
on this result, all subsequent analyses were carried out using
protein taken after renaturation (nonreduced) The
param-eters of the major unfolding transition were used for data
analysis Furthermore, thermal unfolding of gp41(24–157)
is highly reversible, as calculated from two consecutive scans
of the protein (Fig 2C)
pH dependence of the thermal unfolding
of gp41(24–157)
gp41(24–157) has been shown to be hyperthermostable in
50 m formate, pH 2.8 To study the pH dependence of
protein unfolding, samples were analysed by DSC, CD and fluorescence spectroscopy over a pH range of 0.5–5.0 Protein was dialyzed into 50 mMformate and adjusted to the pH indicated by the addition of concentrated HCl or NaOH
The DSC scans of gp41(24–157) at different pH values are show n in Fig 3A A plot of Tmand DHvH/DHcal, as a function of pH, is shown in Fig 3B Cooperative unfolding events at temperatures of > 100C are seen in the pH range
of 1.0–4.0 The variation of Tmas a function of pH is of a slight valley shape, with the minimum at pH 2.25 (102C) The DHvH/DHcal ratio at pH 1.0–2.5 was 1 and increased to 3–4 at pH 2.8–3.5 This sudden increase was accompanied by an increase in Tm(Fig 3B) To evaluate whether the increase in DHvH/DHcal from 1 to 3 represents an association of monomers to trimers, or whether this increase corresponds to an increase in coop-erativity of the unfolding of gp41 trimers, 14 samples of gp41 at different pH values (1.0–4.0) were analysed by fluorescence spectroscopy A gp41 monomer contains eight tryptophan residues Three are involved in a buried tryptophan cluster [11], characterized by the tight packing
of W60 of helix N between W117 and W120 located at helix C of a neighbouring monomer Trimer dissociation into monomers disrupts this cluster, leading to an exposure
of the three buried tryptophan residues to the solvent, giving rise to a shift in the maximum of the emission spectrum, as shown for the gp41 model [14] The maximum of the fluorescence emission spectrum of gp41 at pH 2.8 was
349 nm (data not shown) This maximum of fluorescence emission for the analysis of gp41 at pH between 1.0 and 4.0 (data not shown) was unchanged (349 ± 1 nm), suggesting
no disruption of the tryptophan cluster and no trimer dissociation at this pH range
Samples of gp41 (all at 60 lM) at this pH range have also been analysed by CD Over the range of pH 1.0–4.0, spectra are closely superimposable and no shift in the minima or maxima is observed (data not shown) Differences in molecular ellipticity are in the range of the error associated with determination of the protein concentration
Monomer–trimer equilibrium
A monomer–trimer equilibrium has been described for the cysteine double mutant of simian immunodeficiency virus
Table 1 Thermodynamic parameters derived from the DSC analysis of gp41(24–157), gp41(34–170) and gp160 (Figs 2 and 5) ND, not determined.
Sample
T m
(C)
DH cal
(kcalÆmol)1)
DH vH
(kcalÆmol)1) DH vH /DH cal
Reversibility (%) gp41(24–157) nonreduceda 110.4 61 ± 0.4 228 ± 2 3.7 80
119.5 21 ± 0.5 165 ± 5 7.8 85 gp41(24–157) reduced (36 l M )b 109.1 94 ± 0.4 308 ± 2 3.3 85 gp41(24–157) reduced (1 l M )b 109.1 91 ± 1.0 303 ± 4 3.3 ND gp41(34–170), glycosylated c 112.6 38 ± 0.5 201 ± 4 5.3 25
106.4 32 ± 1.5 218 ± 6 6.8 100 111.8 41 ± 1.5 191 ± 7 4.6 0
a Non-reduced protein corresponds to gp41(24–157) after the dialysis step into 50 m M formate at pH 2.8 (Materials and methods).
b
Reduction was carried out by overnight dialysis of the protein into 50 m M sodium formate, 150 l M Tris(2-carboxyethyl)phosphine (TCEP)
at 4 C c See the Materials and methods for a description of the protein.
Trang 6(SIV) gp41 and the data obtained have been extrapolated to
HIV gp41 [19,37,38] Serial dilutions of gp41(24–157), at
concentrations between 36 lM and 0.3 lM, have been
prepared in 50 mMformate, pH 2.8, incubated for 5 h to
achieve equilibration, and then analysed by DSC (down to
1 l ) and intrinsic tryptophan fluorescence spectroscopy
The thermodynamic parameters of unfolding of the mono-mer are expected to be different from those of the trimono-mer Trimer dissociation is bound to alter the ratio of DHvH/
DHcal, indicative of the content of the cooperatively unfolding unit All DSC scans of serial dilutions of gp41(24–157) were very similar, providing no evidence for trimer dissociation Figure 4 shows a superimposition of the scans obtained with 36 and 1 lM of protein Derived thermodynamic parameters for these thermograms are given in Table 1
The same samples were analysed by fluorescence spectroscopy, and the disruption of the tryptophan cluster (see above) as a result of trimer dissociation is expected to change the maximum of the fluorescence emission spectrum However, the emission spectra of all samples in the concentration range between 36 and 0.3 lM were very similar (data not shown) and the maxima of all tryptophan fluorescence emission spectra were 349 ± 1 nm, indicating
no trimer dissociation
Analysis of glycosylated gp41 and gp160
To evaluate the influence of glycosylation on the thermal stability, experiments were carried out with glycosylated gp41 Furthermore, experiments were also carried out with glycosylated gp160 (see the Materials and methods for a detailed description of both proteins) to study the thermo-dynamics of protein unfolding of the entire surface protein Interestingly, the DSC scan of gp160 at 0.3 mgÆmL)1in NaCl/Pi, pH 7.4, showed no cooperative unfolding event (data not shown) A rescan of this sample showed no significant difference from the first scan However, a cooperative unfolding event at 121.2, 120.6 and 118.3C was seen when the pH was lowered to 5.5, 4.0 and 3.4, respectively (Fig 5) The DSC scan of gp160 at pH 2.8 (using the same conditions as for gp41, see above) showed three unfolding transitions at 93.3, 106.4 and 111.8C
Fig 3 Study of the pH dependence of the thermal denaturation of
gp41(24–157) (A) DSC thermograms of gp41(24–157) in the pH range
of 0.5–5.0 Protein was analyzed in 50 m M formate and adjusted to the
pH indicated by the addition of concentrated HCl or NaOH For
clarity reasons, traces have been moved arbitrarily on the y-axis (B)
Plot of T m (ÆÆÆÆ) and DH vH /DH cal (––) as a function of pH Values have
been taken from the data presented in (A).
Fig 4 Study of gp41(24–157), at different concentrations, by DSC Superimposition of DSC scans of gp41(24–157) at 36 l M (ÆÆÆÆ) and at
1 l M (––) Derived thermodynamic parameters are listed in Table 1 Proteins were analyzed in 50 m M formate, pH 2.8 Proteins were incubated for 5 h prior to analysis.
Trang 7(Fig 5, Table 1) Most interestingly, the thermodynamic
parameters of glycosylated gp41 (Fig 5A, Table 1), which
shares 99% sequence identity with the corresponding
sequence of gp160, were very similar to the corresponding parameters of the third transition seen for gp160
The Tmvalues of the major unfolding events of glycosyl-ated gp41 (112.6C) and nonglycosylated gp41 (110.4 C) were very similar (Table 1, Figs 2 and 5) Differences in DH values of the major events can rather be attributed to the pre-sence of two minor transitions in the scan of the glycosylated gp41 (which might be attributed to misfolded protein) than to differences between the proteins The unfolding of glycosylated gp41 was less reversible than that of the nonglycosylated form (Table 1, Figs 2C and 5B) However,
in interpreting these data it has to be taken into account that, although the sequences of both proteins are almost identical, the molecular boundaries of glycosylated gp41(34–170) and nonglycosylated gp41(24–157) differ slightly
Analysis of gp41(24–157) with a single point mutation Our vaccine approach is based on the generation of gp41 mutants that resemble a fusion intermediate of the protein
It is generally accepted that this fusion intermediate state is trimeric and that helices N and C do not interact Ways have been explored to destabilize the interaction between helices
N and C by site-directed mutagenesis Alternatively, muta-tions were carried out to stabilize the interaction between the three helices N in order to favour the trimeric state of gp41
An initial series of mutations involved amino acids in positions a and d, which are critical for helix interaction Replacement of T58 with I resulted in an increase in Tm
of 2.2C (Fig 6A, Table 2) This mutation results in the generation of an additional hydrophobic cluster between helices N I124 is located at position a in helix C and interacts, at the same time, with two hydrophobic amino acids located at two neighbouring helices N Replacement
of I124 with S and D significantly destabilizes the protein (Fig 6, Table 2) Interestingly, replacement of I124 with D rendered the unfolding irreversible
Analysis of gp41(24–157) with several point mutations Amino acids Q64 and Q66 are on positions c and e, respectively Their substitution with A resulted in a decrease
in Tmof 5 C (Fig 6B, Table 2), and no changes in the
CD spectrum were noted
There is nowevidence that the loop region of gp41 interacts with gp120 [39] This loop contains a highly conserved di-cysteine motif [40], which forms an intrasub-unit disulfide bond It has recently been suggested that the loop region of the mutant lacking the disulfide bond may be less stable and more dynamic than that of the wild type [40] Here, we show that the replacement of both cysteines with serine has a modest impact on the thermal stability (Fig 6B, Table 2) of the protein, showing a downshift of 2.3C in Tm compared to the wild type These data are consistent with the DSC analysis of native gp41 after reduction with TCEP, which resulted in a decrease in Tmof 1.3C (Table 1) The exposed surface area of the C-terminal half of helix C
is highly charged, whereas helix N is mainly neutral Several
of these charged residues of helix C have been mutated to alanine Both double mutants E143A/K144A and K144A/ E148A showed a decrease in Tm of 4 C (Fig 6B, Table 2)
Fig 5 DSC analysis of gp160 and gp41(34–170) See the Materials and
methods for further information (A) Both proteins are glycosylated.
Proteins were analyzed at pH 2.8, 3.4 and 4.0 in 50 m M formate and
adjusted to the pH indicated by the addition of concentrated NaOH.
Protein, at pH 5.5, was analysed in 50 m M Mes DSC scans of gp160
and gp41(34–170) were performed at different pH values gp41(34–
170) shares 99% sequence identity with the corresponding sequence
in gp160 (B) Shown are segments of two consecutive up-scans of
gp41(34–170) from 5 to 129 C, at pH 2.8 Scans have been moved
arbitrarily on the y-axis Derived data are given in Table 1.
Trang 8The DSC scan of a triple mutant, in which three glutamine residues (at positions 40, 41 and 51, all located
on the N-terminal half of the N-terminal helix) were replaced with alanine, showed a major unfolding event
at 119.1C (Fig 7A, Table 2), which is significantly higher than that found for the native protein (110.4C) This unfolding transition was preceded by a broad transition centred at 97 C
Another triple mutant was aimed at combining mutations which stabilize the interaction between the three N-terminal helices (Q51I/T58I) with a mutation that destabilizes the interaction between helices N and C (I124D) The stabilizing effect of the T58I exchange has been described above, and the Q51 is located in a similar position in the heptad repeat
as T58, indicating that its replacement with Ile has a similar effect as the T58 replacement I124 is located on the interface between helices N and C, and its replacement with glutamate dramatically altered the thermodynamics of protein unfolding (see above) In contrast to I124D, mutant Q51I/T58I/I124D unfolds in a single transition, character-ized by a moderate downshift in Tmby 4 C, as compared
to the native protein (Fig 7A, Table 2)
All four amino acids replaced in the quadruple mutant W6OA/I124D/I131D/Q142N are located in the interface between helices N and C W60 is located on helix N and is part of the tryptophan cluster comprising W117 and W120
of helix C The other three amino acids mutated (I124, I131 and Q142) form part of helix C These four mutations have dramatically changed the thermodynamics of protein unfolding Protein unfolding starts at 60 C (Fig 7B) and the peak can be deconvoluted into three transitions centred on 72, 76 and 81C At 95 C, the protein starts
to aggregate The CD spectrum of this quadruple mutant was significantly different from that of the native protein (Fig 1) The percentage of a-helix calculated for that mutant was 56% [33], significantly lower than for the native protein (75%)
In another mutant, three solvent-accessible hydrophobic amino acids (L91, I92, W103) in the loop region were replaced with charged residues (K, K, D, respectively) in order to render this protein soluble at neutral pH This mutant was soluble to 80 lgÆmL)1 in 10 mM Na2HPO4/ NaH2PO4, 0.05% Tween-20, pH 7.5, a considerable improvement on the native protein which is retained qualitatively on 0.22 lm filters after dialysis into this buffer The DSC scan of this triple mutant, at pH 2.8, shows a single event at 108C, indicative of only small alterations to protein stability as a result of these three, nonconservative replacements (Fig 7A, Table 2) When the DSC analysis
is repeated at pH 7.5, a single unfolding event, with a
Tmof 110.8C, is seen (Fig 7A)
Analysis of gp41 constructs of directly fused helices
We created mutants in which helices N and C (or vice versa) were joined directly in order to generate con-straints that might prevent the adoption of this fusogenic state and favour an alternative arrangement This alternative arrangement might be characterized by the exposure of epitopes which are hidden in the helix bundle arrangement Details on these mutants are found
in Table 3
Fig 6 DSC analysis of gp41(24–157) mutants containing single (A) or
double (B) point mutations Thermodynamic parameters derived are
listed in Table 2 Proteins were analysed in 50 m M formate, pH 2.8,
and traces were moved arbitrarily on the y-axis.
Trang 9The CD spectra of these five mutants showthe
characteristics of a highly a-helical protein (data not
shown) All five constructs are very thermostable and their
unfolding is at least partially reversible (Fig 7C, Table 3)
The size of the cooperatively unfolding unit was between
1.7 and 2.8, indicating some sort of monomer association
Interestingly, the two NC mutants (fusion of helix N to
helix C), which are closer to the native form, are less thermostable than the CN constructs (fusion of helix C to helix N) CN3 is characterized by an upshift in Tm of
2 C, accompanied by an increase in the enthalpy change
as compared to the wild-type protein (Table 3) Mutants CN1, CN2 and CN3 differ only in tw o or, respectively, three additional amino acids on the C-terminal extension
Table 2 Thermodynamic parameters for the major unfolding event for the DSC analysis of gp41(24–157) mutants containing one to four point mutations (Figs 6 and 7) The second column shows the location of point mutations in the structure in the six-helix bundle of gp41; letters N, C, and
L correspond to helix N, helix C and loop, respectively; lower case letters indicate the position of the mutation in the heptad repeat ND, not determined.
Sample
Position of the mutation
in the structure
T m
(C)
DH cal
(kcalÆmol)1)
DH vH
(kcalÆmol)1) DH vH /DH cal
Reversibility (%)
96.0 36 ± 3.8 159 ± 15 4.4 0 105.3 33 ± 1.6 127 ± 7 3.8 0 Q64E/Q66E N-c/N-e 105.5 69 ± 2.0 172 ± 4 2.5 66 C87S/C93S L/L 108.1 90 ± 0.5 300 ± 2 3.3 33 E143A/K144A C-f/C-g 106.6 56 ± 1.2 263 ± 8 4.7 ND K144A/E148A C-g/C-d 105.6 61 ± 1.3 216 ± 5 3.5 ND Q40A/Q41A/Q51A N-g/N-a/N-d 119.1 55 ± 0.4 164 ± 2 3.0 55 Q51I/T58I/I124D N-d/N-d/C-a 106.2 77 ± 0.8 243 ± 3 3.1 52 L91K/I92K/W103D L/L/L 108.0 91 ± 0.4 293 ± 2 3.2 59 W6OA/I124D/I131D/Q142N N-f/C-a/C-a/C-e 71.7 13 ± 2.5 125 ± 8 9.8 ND
a Parameters of the three transitions.
Fig 7 Analysis of gp41(24–157) mutants by DSC (A) DSC scans of gp41(24–157) mutants containing three point mutations The upper three scans were carried out with protein in in 50 m M formate, pH 2.8, the lower trace with protein in 10 m M Na 2 HPO 4 /NaH 2 PO 4 , 0.05% Tw een-20, pH 7.5 (B) DSC scan of a mutant containing four point mutations; note the difference in scale of the y-axis Derived thermodynamic parameters are shown
in Table 2 (C) DSC scans of recombinant constructs of gp41(24–157) in which helices N and C have been joined directly A definition of these molecules and the thermodynamic parameters derived are given in Table 3.
Trang 10of helix C, whereas the fragment corresponding to helix N
is unchanged
Discussion
Proteins are the result of evolution and their features reflect
an optimal adaptation to a multitude of environmental,
mechanistic and other factors Some organisms have
evolved in a way which enables them to live in a
hyperthermophilic environment characterized by
tempera-tures of > 80C Elevated growth temperature has thus
been a driving force for the evolution of their proteins
towards hyperthermostability Early studies aimed at
understanding thermostability were based on sequence
comparisons, but the increasing availability of 3D structures
has made it possible to identify structural determinants of
protein hyperthermostability by comparing 3D structures of
hyperthermophilic organisms with their mesophilic
coun-terparts [41,42] At least 15 physical and chemical factors
giving rise to thermostability have been identified [43]
However, a major role in achieving thermostability has been
attributed to an increase in salt bridges and H-bonds, better
hydrophobic internal packing, enhanced secondary
struc-ture propensity, helix dipole stabilization and burying of
a hydrophobic accessible surface area [41]
gp41 is a special case in terms of thermal stability The
protein has evolved to be hyperthermostable, but, in
contrast to the studies described above, elevated growth
temperatures can be excluded as a factor driving protein
evolution This hence raises the question of whether the
characteristics of thermostable proteins, which have evolved
as an adaptation to high temperature, are different from a
protein that has evolved to be hyperthermostable as a
consequence of the adaptation to a factor other than
elevated temperature
The availability of an increasing number of complete
genome sequences has allowed comparison of the amino
acid composition of entire genomes of mesophiles and
hyperthermophiles [44,45] It emerged that proteins of
hyperthermophilic organisms have a higher percentage of
charged residues (K,R,D,E) and a lower number of polar
noncharged residues (Q,N,S,T) than mesophiles The
difference between charged and polar noncharged residues
(termed CvP bias) was found to be the only criterion that
distinguishes mesophiles (CvP: )1 to 5%) from
hyper-thermophiles (CvP: 10–15%) on a global basis [45] The
authors conclude that the evolutionary adaptation of
proteins to elevated temperatures (in hyperthermophile
species) is dominated by the replacement of polar
noncharged residues with charged ones This implies that the formation of ion bonds is the dominating mechanism leading to hyperthermostability as a result of adaptation to elevated temperatures This statistical approach is suppor-ted by experimental data Lowering the pH results in the perturbation of electrostatic interaction, including ion pairs, which is complete at pH 2.0 [46] For a large number of proteins from hyperthermophiles, monitoring thermostabil-ity as function of pH shows a dramatic decrease as the pH approaches 2.0 [46–49], consistent with protein destabiliza-tion by perturbadestabiliza-tion of ionic interacdestabiliza-tion
This was not the case for gp41 The native protein was shown to be hyperthermostable over the pH range of 1–4 (Fig 3) and a similar thermostability has been observed at
pH 7.5 for the soluble triple mutant L91K/I92K/W103D (Fig 7), indicating that protonation has little effect on the thermal stability of the protein Interestingly, the CvP bias for gp41 was, at )13%, the opposite from proteins in hyperthermophile organisms (CvP¼ 10–15%) A dense internal hydrophobic packing is thus more likely to determine thermostability, which is supported by the fact that gp41 contains 29% aliphatic amino acids, above the average found in mesophilic and thermophilic species [45]
It can be concluded that the evolution of gp41 towards hyperthermostability was not dominated by an increase in ionic or electrostatic interaction, shown to be the major feature of proteins from thermophilic species It can thus be hypothesized that structural features giving rise to thermo-stability are different for proteins which have evolved in adaptation to elevated temperatures and hyperthermostable proteins found in mesophile organisms
It appears that the formation of trimeric coiled-coil structures is a common principle of viral membrane fusion [34] It has been demonstrated that the influenza virus fusion protein, HA2 [50,51], and the paramyxovirus fusion protein [52], are equally thermostable It can further be hypothes-ized that this trimeric coiled-coil fold might be associated with thermostability
DSC analysis of the double cysteine to alanine mutant of SIV gp41(27–149) (in 50 mMformate, pH 3.0) has recently been reported [38] The protein was equally hyperthermo-stable with a Tm of 110.7C, and the cooperatively unfolding unit was shown to contain a trimer (DHvH/
DHcal¼ 2.91) Here we report very similar data (Tm¼ 110.7 C, DHvH/DHcal¼ 3.53) for the analysis of HIV gp41(24–157) using the same buffer conditions Based
on the observations of Wingfield et al [20], that SIV gp41 can be expressed and refolded with substantially higher yields than HIV gp41, the authors [37,38] have used the SIV
Table 3 Parameters derived from the DSC analysis of the helix–fusion constructs of gp41(24–157) (see Fig 7C) Molecules are defined in the second column of the table All molecules contain the C-terminal tag, GGGGSHHHHHH.
Sample
Position of the mutation
in the structure
T m
(C)
DH cal
(kcalÆmol)1)
DH vH
(kcalÆmol)1) DH vH /DH cal
Reversibility (%) NC1 A30-D78_N113-K154 91.0 103 ± 3.1 176 ± 4 1.7 65
NC2 A30-K77_W117-K154 86.5 84 ± 1.0 234 ± 3 2.8 71
CN1 W117-K154_A30-K77 101.9 59 ± 1.3 140 ± 5 2.4 37
CN2 W117-A156_A30-K77 106.8 81 ± 0.5 206 ± 2 2.5 39
CN3 W117-W159_A30-K77 112.5 140 ± 0.6 241 ± 1 1.7 18