MazE: an Antidote from the Escherichia Coli mazEF System

19 Pages·2012·1.4 MB·English

JBC Papers in Press. Published on January 19, 2012 as Manuscript M111.318451
The latest version is at http://www.jbc.org/cgi/doi/10.1074/jbc.M111.318451
Energetics of TNF-α recognition

Recognition of Human Tumor Necrosis Factor TNF-α by a Therapeutic Antibody Fragment:
Energetics and Structural Features*
Jaka Marušiča,b, Črtomir Podlipnika, Simona Jevševarb, Drago Kuzmanb, Gorazd Vesnavera and
Jurij Laha
aUniversity of Ljubljana, Faculty of Chemistry and Chemical Technology, Aškerčeva 5, 1000 Ljubljana,
Slovenia
bSandoz Biopharmaceuticals, Mengeš, Lek Pharmaceuticals d.d., Kolodvorska 27, SI-1234 Mengeš,
Slovenia;

*Running title: Energetics of TNF-α recognition

To whom correspondence should be addressed: Jurij Lah, University of Ljubljana, Faculty of Chemistry
and Chemical Technology, Aškerčeva 5, 1000 Ljubljana, Slovenia. Tel.: [phone], Fax: +386 1
2419 425, E-mail: [email protected].

Keywords: tumor necrosis factor; adalimumab; protein-antibody recognition; protein unfolding;
D
ow
n
Background: Human TNF-α is a cytokine denaturation experiments has been supported lo
a
d
involved in many disease-related cellular by structural modeling. We demonstrate that e
d
processes. the observed high affinity binding of Fab to fro
m
Results: High affinity binding of therapeutic TNF-α is an enthalpy driven process due h
antibody (inhibitor) to native and molten globule- mainly to specific non-covalent interactions ttp://w
like TNF-α conformation is driven by specific taking place at the Fab-TNF-α binding w
w
non-covalent interactions. interface. It is coupled to entropically .jb
c
Conclusion: Binding-coupled conformational unfavorable conformational changes and .o
rg
changes are crucial for antibody-TNF-α accompanied by favorable solvation b/
y
recognition. contributions. Moreover, the three-state model g
u
e
Significance: Learning which forces drive analysis of TNF-α unfolding shows that at st o
unfolding of TNF-α and its recognition by physiological concentrations TNF-α may exist n M
monoclonal antibodies and how they affect TNF-α not only as a biologically active trimer but also arc
h
activity regulation. as an inactive monomer. It further suggests that 2
8
even small changes of TNF-α concentration , 2
0
1
SUMMARY could have a considerable effect on the TNF-α 9
Human tumor necrosis factor alpha (TNF-α) activity. We believe that this study sets the
exists in its functional state as a homotrimeric energetic basis for understanding of TNF-α
protein and is involved in inflammation inhibition by antibodies and its unfolding
processes and immune response of a human linked with the concentration dependent
organism. Overproduction of TNF-α results in activity regulation.
development of chronic autoimmune diseases
that can be successfully treated by inhibitors Human tumor necrosis factor alpha (TNF-α) is an
such as monoclonal antibodies. However, the important cytokine involved in many diverse and
nature of antibody-TNF-α recognition remains complex functions in human organism,
elusive due to insufficient understanding of its particularly in inflammation and cellular immune
molecular driving forces. Therefore, we studied response (1-3). Its wide role in biological
energetics of binding of a therapeutic antibody processes and observations of its ability to trigger
fragment (Fab) to the native and non-native regression of tumors makes the TNF-α one of the
forms of TNF-α by employing calorimetric and most intensively investigated proteins over past
spectroscopic methods. Global thermodynamic decades, although cytotoxicity prevents its
analysis of data obtained from the extensive clinical use (1,4). The crystal structure
corresponding binding and urea-induced determined in 1989 suggests that TNF-α consists
1
Copyright 2012 by The American Society for Biochemistry and Molecular Biology, Inc.
Energetics of TNF-α recognition

of three identical 17.35 kDa peptide chains. Each circular dichroism (CD) spectroscopy. In addition,
of them is packed into an antiparallel β-sandwich the stabilization of TNF-α by Fab at denaturing
structure. Subunits are held together by conditions was investigated via urea-induced
noncovalent interactions and form a compact bell- unfolding of TNF-α-Fab complex, unbound TNF-
shaped trimer characterized by a 3-fold axis of α and unbound Fab by CD and fluorescence (FL)
symmetry (5,6). Biologically active TNF-α is a spectroscopy. Thermodynamic parameters
homotrimer (7-10) which may dissociate to obtained from global analysis of experimental data
monomers in the solution at physiological measured at various conditions (temperature,
concentrations (10,11). This regulatory feature has protein concentration, urea concentration) are
been suggested to be a part of a mechanism discussed in terms of structural alterations that
essential for maintenance of optimal biological accompany the observed binding and unfolding
activity of TNF-α (10,11). processes. In addition, we designed a structural
Overproduction of TNF-α in tissues model of the TNF-α-Fab complex which was, in
results in development of chronic autoimmune the absence of its 3D structure, used in molecular
diseases like rheumatoid arthritis, multiple interpretation of the obtained thermodynamic
sclerosis, psoriasis and Crohn's disease (12,13). parameters of binding.
Nowadays availability of certain
biopharmaceuticals to bind TNF-α and EXPERIMENTAL PROCEDURES
D
consequently block its binding to receptors enables Preparation of Protein Solutions - The o
w
successful treatment of these pathologies. Such preparation and purification of human TNF-α has n
lo
a
TNF-α inhibitors are in most cases monoclonal been described elsewhere (17). Monoclonal d
e
d
antibodies, e.g. infliximab, adalimumab, antibody adalimumab was purchased from Abbott fro
golimumab, however, some other approaches to Laboratories Ltd. Its digestion was performed by m
h
TNF-α inhibition are effective as well (14). papain immobilized on an agarose beads and Fab ttp
Approval of the first TNF-α inhibitors that fragments were isolated from digestion mixture by ://w
w
successfully treat the pathologies associated with protein A affinity chromatography. Immobilized w
.jb
elevated TNF-α level has induced over the past papain, protein A affinity column and Fab c
.o
decades an intensive search for more effective and fragment preparation protocols were obtained by rg
b/
safe inhibitors of TNF-α. Presently used Pierce, Thermo Fisher Scientific. SDS PAGE y
g
u
therapeutics are efficient in therapy, however, they electrophoresis under reducing and non-reducing e
s
still induce a number of undesired side effects. To conditions shows that Fab fragment is present in t o
n
M
solve this problem, an extensive research focusing solution in a form of polypeptide chains linked
a
on the design of new and better inhibitors is going with disulphide bond. Amino acid sequence of Fab rch
2
on in several pharmaceutical companies. fragment and SDS PAGE data are available in the 8, 2
Unfortunately, in spite of all these efforts, the Supplemental material (SI). 01
9
molecular basis of forces that drive recognition of Prior to calorimetric and spectroscopic
TNF-α by its inhibitors (15) remains poorly measurements, solutions of proteins were dialyzed
understood. Therefore, we decided to study extensively against phosphate buffer (0.02 M
energetics of recognition (stabilization) of native sodium phosphate, 0.14 M NaCl, pH 7.4). All of
and non-native forms of TNF-α by adalimumab the samples for the urea denaturation experiments
which is the first commercially available were prepared by mixing 10 M urea and protein
therapeutic human antibody. It has been shown stock solution to a final urea concentration
that mixing of solutions of adalimumab and TNF- between 0 and 8 M. The pH value of all solutions
α results in a heterogeneous mixture of TNF-α– was checked and adjusted to 7.4 by the addition of
adalimumab complexes (16). Since quantitative NaOH. The concentrations of protein solutions
thermodynamic analysis and structural modeling were determined spectrophotometrically from the
of binding events in such a complex interacting absorbance measured at 278 nm in 6 M GdnHCl
system is not possible, we simplified it by using a at 25 ºC using extinction coefficients obtained by
fragment of adalimumab (Fab) instead of the full- the method introduced by Gill and von Hippel (see
length antibody. SI) (18).
Binding of Fab to TNF-α at non- Isothermal Titration Calorimetry -
denaturing conditions was monitored by Experiments were performed between 5 and 54 ºC
isothermal titration calorimetry (ITC) and by in an iTC calorimeter from Microcal Inc.
200
2
Energetics of TNF-α recognition

(Northampton, MA). Before each experiment the naphthalene sulfonate (ANS), a hydrophobic probe
protein solutions were degassed for 10 min. A frequently used as an indicator of the molten-
solution of Fab was titrated into a solution of globule-like protein states (19). The ANS emission
TNF-α. The enthalpy of interaction (H ) was spectra were measured between 400 and 580 nm
T
with excitation at 380 nm.
obtained by integration of the raw signal, corrected
Molecular Modeling - Molecular
for the corresponding heat of dilution, and
modeling of the TNF-α-Fab complex was started
expressed per mole of added Fab fragment.
with building and refining a homology model of
CD Spectroscopy - CD spectroscopy
Fab using YASARA Structure (20,21). Protein-
measurements were performed with an AVIV
protein docking software Hex 6.0 (22-24) was
Model 62A DS spectropolarimeter (Aviv
used to obtain an ensemble of representative
Associates) at different temperatures lower than
complexes between TNF-α and Fab. 25 complexes
the one at which the proteins start to denature
with the lowest docking score were saved for
irreversibly. Changes in secondary structure at
further modeling and analysis. The complexes
increasing urea concentrations (0 – 8 M) were
were then refined using short 10 ps molecular
followed by measuring the ellipticity (Θ) of the
dynamic (YASARA Structure, Yamber2 FF) in
protein solutions at several wavelengths.
vacuum followed by energy minimization in water
Ellipticities of Fab solutions were measured at 234
(Convergence 0.05 kJ/mol per atom during 200
nm in a 1 mm cuvette and at protein concentration
D
of ~ 43 µM. Ellipticities of the TNF-α solutions steps) (25). Finally, FoldX (26) (as YASARA ow
plugin) was used for the final refinement and n
were measured at 224 nm in 1 mm and 10 mm lo
a
calculation of interaction energy between Fab and d
cuvettes at protein concentrations of ~ 17 µM and e
d
~ 1.8 µM respectively. Measurements of Fab- TNF-α. NACCESS (27) was used for the analysis fro
of protein surfaces. m
TNF-α complex solutions were performed at 222 h
and 234 nm in 1 mm and 10 mm cuvettes at the RESULTS ttp://w
complex concentrations of ~ 7 µM and ~ 0.3 µM w
Binding of Fab to TNF-α Monitored by w
respectively. The reversibility of the urea-induced .jb
ITC and CD Spectroscopy - The calorimetric c
transitions was checked by diluting the protein .o
biding isotherms accompanying association of Fab rg
solutions from the post-transition to the pre- b/
to TNF-α exhibit the characteristics of a 3:1 y
transition urea concentrations. By comparing the g
u
association process (Fig. 2). This observation is e
CD spectra measured for these diluted solutions confirmed by a very good fitting of the family of st o
with those obtained for the sample solutions n M
the ITC curves measured at different T with the
prepared directly from urea and buffer solutions, corresponding model function based on the arch
we estimated the extent of reversibility of the 2
observed transitions to be higher than 80%. binding model that assumes the existence of three 8, 2
Molar ellipticities ([Θ], 105 degrees cm2 dmol-1) equivalent independent Fab binding sites on the 019
TNF-α molecule (Fig. 2; see SI for details).
presented in the paper were obtained from raw
Therefore, the obtained model-based
data (ellipticities Θ) by subtracting the
thermodynamic parameters (Table 1) may be
corresponding Θ of the buffer solution and taking
considered as appropriate descriptors of the
into account the molar concentration of protein (c)
binding process. In line with the ITC data are the
and optical path length (l) through the relation [Θ]
results of the CD spectroscopic measurements
= Θ/(c·l).
which clearly indicate that the association of TNF-
Fluorescence Spectroscopy - Fluorescence
α with Fab is accompanied by significant
emission spectra were recorded using a Perkin
rearrangements of Fab and/or TNF-α structure
Elmer Life Sciences LS 50 luminescence
(Fig. 2).
spectrometer equipped with thermally controlled
To test whether protonation or deprotonation of
cell holder and a cuvette with a 1 cm path length.
the proteins is coupled to their binding we
All measurements were performed at 25.0 ± 0.1
performed the same ITC experiments also in TRIS
ºC. Concentration of TNF-α in solutions was ~ 2.2
buffer (0.02 M TRIS, 0.14 M NaCl, pH = 7.4, T =
µM while concentration of TNF-α-Fab complex
35 °C) which has about ten times higher ionization
was ~ 0.45 µM. Fluorescence spectra
enthalpy than the phosphate buffer (28). Since the
accompanying urea-induced denaturation were
titration curves observed in TRIS are very similar
measured in the presence of 15 µM 1-anilino-8-
to those observed in phosphate we concluded that
3
Energetics of TNF-α recognition

no protonation or deprotonation is involved in 3TD(denatured). Global fitting of the model
association of Fab with TNF-α. resulted in its good agreement with the family of
Urea induced unfolding monitored by CD TNF-α CD denaturation curves measured at
and FL spectroscopy - Because of the different T and different TNF-α concentration
irreversibility of the thermal denaturation of Fab, (Fig. 3). The corresponding thermodynamic
TNF-α and the TNF-α-Fab complex we attempted parameters are presented in Table 1.
to investigate the thermodynamics of their TNF-α-Fab complex. Analysis of ITC and CD
unfolding transitions via the reversible urea binding experiments and molecular modeling of
denaturation (Figs. 3 and 5). The quantitative TNF-α-Fab formation suggest that the TNF-α-Fab
thermodynamic analysis of such denaturation complex is a heterohexamer consisting of three
processes has been discussed extensively Fab molecules bound on a TNF-α trimer (T F ).
3 3
elsewhere (29-31) and therefore the details of its Denaturation curves of the TNF-α-Fab complex
application to our systems are discussed in SI. were measured at various temperatures and two
Fab. SDS PAGE electrophoresis (see SI) shows concentrations of the complex. CD spectra in 0 M
that Fab molecule is composed of two polypeptide urea show local maximum at about 234 nm in the
chains linked by a disulphide bond. Therefore, Fab case of Fab and TNF-α-Fab complex but not in the
(F) urea-induced unfolding can be treated as a case of the unbound TNF-α (Fig. 4). Since molar
monomolecular transition. A very good global fit ellipticities of TNF-α at 234 nm in 0 M and 8 M
D
is obtained for the two-state transition model urea are practically the same it may be assumed o
w
expressed as: F(native)  FD(denatured) with the that denaturation curves of TNF-α-Fab complex nlo
a
family of CD denaturation curves measured at measured at 234 nm reflect mostly changes in the d
e
d
various temperatures (Fig. 3). This suggests that secondary structure of Fab. On the other hand, at fro
the model is appropriate and gives reliable 222 nm both proteins contribute to the measured m
h
thermodynamic parameters of Fab unfolding signal. The complexity of denaturation curves ttp
(Table 1). measured at 222 nm suggests that denaturation of ://w
w
TNF-α. Brief examination of CD denaturation TNF-α-Fab complex should be considered at least w
.jb
curves (Fig. 3b,c) suggests that between 0 and 8 M as a three-state process. In contrast to TNF-α, c
.o
urea TNF-α may populate three conformational solutions of the TNF-α-Fab complex exhibit no rg
b/
states. At 0 - 2.5 M urea TNF-α predominantly induced ANS FL between 0 and 8 M urea y g
u
exists as a native trimer. The intermediate state of suggesting that in the intermediate state Fab is e
s
TNF-α is discernible mainly from CD denaturation strongly bound to the binding sites of ANS (Fig. t on
M
curves measured from 25 ˚C to 45 ˚C at 2.5 - 6 M 5). Moreover, concentration dependence of urea a
urea and high TNF-α concentration. Observation unfolding curves suggests that the overall process rch
2
of the corresponding ANS FL spectra suggest that of denaturation of the complex is not a 8, 2
TNF-α in the intermediate state binds a known monomolecular process. Based on these findings 01
9
molten globule-specific fluorescence probe ANS and known TNF-α unfolding features we assumed
(Fig. 3c-inset). CD spectra measured in 8 M urea that TNF-α-Fab complex (T F ) unfolding may be
3 3
(Fig. 4) indicate that in its urea-denatured state described as: T F (native)  I (intermediate) 
3 3 TF
TNF-α still contains a considerable amount of 3TD + 3FD. Global fitting of the model resulted in
secondary structure. Careful comparison of CD good agreement of the corresponding model
denaturation curves measured at high and low function with the family of TNF-α-Fab CD
TNF-α concentration (Fig. 3b,c) suggests that the denaturation curves measured at different T and
transition from the native to the intermediate state TNF-α-Fab complex concentrations. The
is concentration independent (monomolecular) corresponding thermodynamic parameters are
while the transition from the intermediate state to presented in Table 1. It should be mentioned that
the denatured state is concentration dependent in contrast to the model analysis of Fab and TNF-α
(non-monomolecular). Based on these qualitative unfolding curves that resulted in reasonably
findings, previously reported results on TNF-α reliable and uniform thermodynamic descriptors,
folding (32) and known TNF-α structural features TNF-α-Fab complex unfolding curves could be
(5,8,9) we assumed that TNF-α (T ) unfolding equally well described by various sets of
3
thermodynamic parameters. Since all these
may be described by the following three-state
parameters are state functions the problem of their
transition model: T (native)  I (intermediate) 
3 3 high correlation was solved in the following way.
4
Energetics of TNF-α recognition

The overall parameters of TNF-α-Fab complex recognition by Fab is the very large negative
unfolding were calculated independently from Ho suggesting that a large number of strong
BIND
parameters describing Fab (Fig. 3a) and TNF-α
intermolecular contacts are formed upon binding
(Fig. 3b,c) unfolding and Fab binding to TNF-α
of the pre-structured Fab and TNF-α. This is in
(Fig. 2) and further used in the fitting procedure as
accordance with our structural model that predicts
fixed parameters. The validity of this approach
the formation of about 20 H-bonds (per binding
was justified experimentally by measuring CD
site) and several van der Waals contacts in the
spectra of Fab, TNF-α and their complex 8 M
TNF-α-Fab binding interface (Fig. 1). Moreover,
urea. Namely, it can be seen in Fig. 4 that CD
the molecular recognition is to a large extent
spectrum (physical property) corresponding to the
driven by favorable solvation entropy
denatured TNF-α-Fab complex is equal to the
contributions (Co < 0 andCo < 0)
corresponding sum of spectra (physical properties) P,BIND P,CONF
of the individual denatured proteins. arising mainly from desolvation of a relatively
large non-polar surface area (ΔA  ΔA 
N,BIND N,CONF
DISCUSSION -1400 Å2; see SI). The calculated significantly
Driving forces of TNF-α recognition by Fab - positive Go is in accordance with
Very good agreement of the proposed model with CONF
experimental observation that the pre-structured
the experimental ITC data measured at various
Fab and TNF-α conformations are not present in D
temperatures suggests that each TNF-α binds three o
the unbound state at given experimental conditions w
Fab molecules. The thermodynamic parameters of n
and thus supports the validity of the presented loa
binding of Fab to TNF-α (Table 1) obtained by d
e
model analysis of ITC data show that the binding parsing of thermodynamic parameters. GCoONF > d fro
of Fab to TNF-α is an enthalpy driven process 0 may be associated to a large loss of m
h
accompanied by an unfavorable entropy configurational entropy that overcompensates the ttp
contribution (Fig.6). The measured large negative corresponding small favorable enthalpy and large ://w
w
heat capacity change suggests that hydrophobic favorable solvation entropy contribution. An w
.jb
effect is an important driving force of the observed estimate of configurational entropy loss upon c
.o
binding event. Standard thermodynamic folding [(No. of residues)·4.3 cal K-1 (mol residue)- rg
b/
parameters of binding, , can be dissected into 1] (33) suggests that the observed binding-coupled y g
u
contributions coming from conformational conformational transition involves about ten e
s
changes, , of Fab and/or TNF-α residues most likely located on the hypervariable t on
M
(experimentally observed by CD spectroscopy; loops in the Fab binding region. Structural a
rc
Fig.2) and from the rigid-body association, rearrangements of this region are expected in the h
2
, of the pre-structured Fab and TNF-α: light of inhibition of TNF-α activity which 8, 20
requires tight fitting of Fab into the narrow groove 19
, (1) formed in the contact of two TNF-α subunits

(location of TNF-α receptor epitop).

High binding affinity and specificity of inhibitors
whereG, H, S and C represent Gibbs free
P are of particular importance in the development of
energy, enthalpy, entropy and heat capacity,
efficient therapeutics. In this light our approach of
respectively. Fo values were obtained from dissecting thermodynamics of inhibitor binding to
model analysis of ITC experiments (see Table 1 TNF-α may be used as a general methodology for
uncoupling the specificity and affinity factors.
and SI) while the Fo contributions were
BIND Namely, binding specificity is predominantly
estimated using empirical parameterization that is
determined by the ability of inhibitor to form
mainly based on changes in polar and non-polar
energetically favorable interactions with its target.
solvent-accessible surface areas (SASA). The
Since highly specific binding results in highly
SASA values were calculated from the structural
negative Ho , this parameter that can be
model of the TNF-α-Fab complex (see Fig. 1 and BIND
SI). Finally, the contributions of structural changes estimated on the basis of structural data (see Fig. 6
and SI), may be used as a criterion of the binding
were estimated as Fo Fo Fo . Fig.
CONF BIND specificity. Often, Ho predominantly
6 shows that the main driving force of TNF-α BIND
5
Energetics of TNF-α recognition

contributes to the overall Ho of binding. In such significantly protects TNF-α native and
intermediate form against urea denaturation.
cases the measured Ho itself could be used as a Moreover, the population of states in the presence
valuable specificity criterion. Therefore, Ho and in the absence of urea (Fig. 7) is significantly
affected by changes in TNF-α concentration. In
and/or Ho can complement the affinity data the absence of urea the changes of populations of
BIND
that in the process of development of new, TNF-α native, intermediate and denatured forms
improved versions of inhibitors usually (Fig. 8c) may be important for regulation of TNF-
characterize the therapeutic candidates (34). α activity. Namely, at physiological concentrations
Thermodynamics stability of TNF-α and (38) (picomolar range) TNF-α exists not only as a
regulation of its activity – Inspection of native (biologically active cytotoxic) trimer but a
thermodynamic parameters in Table 1 suggests considerable fraction of monomeric (inactive non-
that the folding of Fab (FD  F), TNF-α (3TD  cytotoxic) TNF-α may be present as well (11). Our
T ) and TNF-α-Fab complex (3TD + 3FD  T F ) model of TNF-α denaturation predicts that in the
3 3 3
absence of urea decreasing TNF-α concentration
in the corresponding standard states at
for an order of magnitude (e.g. from 100 pM to 10
physiological temperatures is an enthalpy driven
pM) may result in about 100 times lower
process which is a general feature of globular
concentration of its active form and in the
proteins (35,36). To estimate the degree of
D
unfolding of TNF-α and Fab in the denatured state complete absence of the intermediate form. It ow
follows that at such conditions even small changes n
the observed overall heat capacities of unfolding lo
a
of TNF-α concentration could have a considerable d
were compared to the corresponding averages over e
d
a large set of proteins presented by Robertson & effect on the TNF-α activity. The predictions of fro
our model are in very good agreement with the m
Murphy (37). The expected average Co values h
p corresponding data measured by ELISA (11). ttp
were calculated from the linear relation between Taken together, we believe that the ://w
w
the number of residues in the protein cooperatively measured energetics of the protein-protein binding w
unfolding unit and the heat capacity change. In and protein unfolding together with the .jbc
.o
case of TNF-α the experimentally obtained Cpois corresponding molecular interpretation will brg/
improve our understanding of forces that drive y
only 15 % of the expected average value, and in g
u
recognition of TNF-α by monoclonal antibody e
the case of Fab only 14 %, which is in agreement fragments and understanding of molecular st o
with CD spectra showing significant amount of n M
mechanism of TNF-α activity regulation in
residual protein structure even in 8 M urea (Fig. general. Moreover, since many new monoclonal arch
4). 2
We used thermodynamics to determine the antibodies and their fragment-based therapeutics 8, 2
are in different stages of preclinical and clinical 01
mechanism and driving forces of TNF-α unfolding 9
development we believe that thermodynamic
and binding to Fab. Namely, it enables prediction
approach presented in this work can be
of TNF-α behavior at different conditions
successfully used in more rapid design,
(temperature, urea concentration, TNF-α
development, stability assessment and production
concentration, Fab concentration). In this context
of more efficient and safer drugs.
speciation diagrams (Fig. 7) indicate that Fab

REFERENCES
1. Barbara, A. J., Van Ostade, X., Lopez, A. (1996) Immunol. Cell Biol. 74, 434-443
2. Vilček, J., Lee, T. H. (1991) J. Biol. Chem. 266, 7313-7316
3. Old, L. J. (1985) Science 230, 630-632
4. Terlikowski, S. J. (2001) Rocz. Akad. Med. Bialymst. 46, 5-18
5. Eck, M. J., Sprang, S. R. (1989) J. Biol. Chem. 264, 17595-17605
6. Jones, E. Y., Stuart, D. I., Walker, N. P. C. (1989) Nature 388, 255-258
7. Lewit-Bentley, A., Fourme, R., Kahm, R., Prange, T., Vachette, P., Tevernier, J., Hauquier, G.,
Fiers, W. (1988) J. Mol. Biol. 199, 389-392
6
Energetics of TNF-α recognition

8. Arakawa, T., Yphantis, D. A. (1987) J. Biol. Chem. 262, 7484-7485
9. Wingfield, P., Pain, R. H., Craig, S. (1987) FEBS Lett. 211, 179-184
10. Smith, R. A., Baglioni, C. (1987) J. Biol. Chem. 262, 6951-6954
11. Corti, A., Fasina, G., Marcucci, F., Barbanti, E., Cassani, G. (1992) Biochem. J. 284, 905-910
12. Bradley, J. R. (2008) J. Pathol. 214, 149-160
13. Tracey, K. J., (2002) Nature 420, 853-859
14. Tracey, D., Klareskog, L., Sasso, E., Salfeld, J. G., Tak, P. P. (2008) Pharmacol. Ther. 117, 244-
279
15. Rigby, W. F. C. (2007) Nat. Rev. Rhevmatol 3, 227-233
16. Santora, L. C., Kajmakcalan, Z., Sakorafas, P., Krull, I. S., Grant, K. (2001) Anal. Biochem. 299,
119-129
17. Kenig, M., Gaberc-Porekar, V., Fonda, I., Menart, V. (2008) J. Chromatogr. B. 867, 119-125
18. Gill, S. C., von Hippel, P. H. (1989) Anal. Biochem. 182, 319-326
19. Matulis, D., Lovrien, R. E. (1998) Biophys. J. 74, 422-428
20. Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W. and Lipman, D. J.
(1997) Nucleic. Acids. Res. 25, 3389-3402.
21. King, R. D., and Sternberg, M. J. (1996) Protein Sci. 5, 2298–2310 D
o
22. Ritchie, D.W., and Venkatraman, V. (2010) Bioinformatics 26, 2398-2405 w
n
lo
23. Macindoe, G., Mavridis, L., Venkatraman, V., Devignes, M., and Ritchie, D.W. (2010) Nucleic a
d
e
d
Acid Res. 38, W445-W449. fro
24. Ritchie, D.W., Kozakov D., Vajda S. (2008) Bioinformatics 24, 1865-1873 m
h
25. Krieger, E., Darden, T., Nabuurs, S.B., Finkelstein, A., and Vriend, G. (2004) Proteins 57, 678- ttp
683. ://w
w
w
26. Schymkowitz, J., Borg, J., Stricher, F., Nys, R., Rousseau, F., and Serrano, L. (2005) Nucleic Acids .jb
c
Res. 33, 382-388. .o
rg
27. Hubbard, S. J., and Thornton, J. M. (1993) NACCES, University College, London b/
y
28. Bradshaw, J. M., and Waksman, G. (1998) Biochemistry 37, 15400-15407 gu
e
s
29. Simic, M., De Jonge, N., Loris, R., Vesnaver, G., and Lah, J. (2009) J. Biol. Chem. 284, 20002- t o
n
20010 M
a
30. Drobnak, I., Vesnaver, G., and Lah, J. (2010) J. Phys. Chem. B 114, 8713-8722 rch
2
8
31. Drobnak, I., Korenčič, A., Loris, R., Marinovsky, I., Glaser, G., Jamnik., A., Vesnaver, G., and , 2
0
Lah, J. (2009) J. Mol. Biol. 392, 63-74 19
32. Holdan, R., and Pain, R. H. (1994) FEBS Lett. 343, 256-260
33. Lee, K. H., Xie, D., Freire, E., and Amzel, L. M. (1994) Proteins 20, 68-84
34. Ladbury, J. E., Klebe, G., and Freire, E. (2010) Nature Reviews 9, 23-27
35. Mayers, J. K., Pace, C. N., and Scholtz, M. J. (1995) Protein Sci. 4, 2138-2148
36. Murphy, K. P., and Freire, E. (1992) Adv. Protein Chem. 43, 313-361
37. Robertson, A. D., and Murphy, K. P. (1997) Chem. Rev. 97, 1251-1267
38. Warzocha, K., Salles, G., Bienvenu, J., Bastion, Y., Dumontet, C., Renard, N., Neidhardt- Berard,
EM., and Coiffier, B. (1997) J. Clin Oncol. 15, 499-508

FOOTNOTES
This work was supported by the Ministry of Higher Education, Science and Technology and by the
Agency for Research of Republic of Slovenia through the Grants No. P1-0201 and P1-0340.

7
Energetics of TNF-α recognition

FIGURE LEGENDS
Figure 1. Structural model of TNF-α-Fab complex. (a) TNF-α subunits are presented in green, blue
and yellow. Three Fab fragments (in gray color) are bound on TNF-α surface at the contact of the TNF-α
subunits. (b) Closer view of the binding site in different orientation.
Figure 2. Binding of Fab to TNF-α monitored by ITC and CD spectroscopy. (a) Enthalpies of Fab
binding to TNF-α at various Fab/TNF-α molar ratios (r) and temperatures (T) monitored by ITC. Lines
represent the best global fit of the model (see SI; schematically presented as an inset) from which the
corresponding thermodynamic parameters were extracted (Table 1 and Figure 6). (b) Changes in
secondary structure upon Fab binding to TNF-α at various Fab/TNF-α molar ratios (r) monitored by CD
spectroscopy at 222 nm and 25 ºC. Δ[Θ] was calculated by subtracting molar ellipticities of unbound
TNF-α and Fab at 222 nm from the measured ellipticities of the TNF-α-Fab complex normalized to TNF-
α concentration of 1 mol L-1 (Δ[Θ] = [Θ] - [Θ] - r[Θ] ). Blue line represents the model function
exp TNF-α Fab
calculated from best fit binding parameters obtained from analysis of ITC measurements.
Figure 3. Modeling urea induced denaturation transitions of Fab and TNF-α. Molar ellipticities
measured as a function of urea concentration for Fab (a) and TNF-α at high (17 µM) (b) and low (1.8
µM) (c) concentration and various temperatures. Lines represent the best global fit of the denaturation
models (see SI; schematically presented as insets in (a) and (b)) to the experimental denaturation curves.
D
o
Inset (c): ANS fluorescence of TNF-α at 2.2 µM protein concentration and 15 µM ANS concentration w
n
measured at 458nm and 25 ºC. lo
a
d
e
d
Figure 4. CD spectra of TNF-α, Fab and TNF-α-Fab complex. (a) Far-UV CD spectra of TNF-α and fro
m
Fab in 0 M and 8 M urea at 25 ºC. (b) Far-UV CD spectra of TNF-α-Fab complex in 0 M and 8 M urea at h
25 ºC in comparison to the corresponding sum of spectra of unbound TNF-α and Fab. ttp://w
w
w
Figure 5. Modeling urea induced denaturation transitions of TNF-α-Fab complex. Molar ellipticities .jb
c
of TNF-α-Fab complex measured as a function of urea concentration at 222 nm at high (7 µM) (a) and .o
rg
low (0.3 µM) (b) concentration of the complex. Molar ellipticities of TNF-α-Fab complex measured as a b/
y
function of urea concentration at 234 nm at high (7 µM) (c) and low (0.3 µM) (d) concentration of the g
u
e
complex. Lines represent the best global fit of the denaturation model (see SI). Insets: (a) schematic s
t o
presentation of the TNF-α-Fab complex denaturation; (d) ANS fluorescence of TNF-α-Fab complex n
M
recorded at 0.45 µM complex and 15 µM ANS concentration measured at 458 nm and 25 ºC. a
rc
h
2
8
Figure 6. Dissection of binding parameters. Standard thermodynamic quantities of Fab binding to TNF- , 2
0
α (Exp.) at 37 °C obtained by global analysis of ITC data (Table 1) were dissected into the contributions 19
of rigid-body binding (Bind.) and conformational change (Conf.) (see eq. 1 and the corresponding
discussion). Bind. parameters were calculated from the structural model (Fig. 1) while Conf. parameters
were estimated as Conf. = Exp. –Bind.
Figure 7. Population of different TNF-α and TNF-α-Fab complex forms at various conditions at 37
ºC. Population of TNF-α forms (a) and population of TNF-α-Fab complex forms (b) at different urea and
protein concentrations. Population of different forms of TNF-α in the absence of urea as a function of
TNF-α concentration (c). Black diamonds represent the values measured by ELISA (Corti et.al.; ref. 11).

8
Energetics of TNF-α recognition

TABLE
Table 1. Thermodynamic parameters at T = 37 °C obtained from global fitting of the model functions
o
(see SI) to the ITC binding data (Fig. 2) and urea denaturation data (Figs. 3 and 5).
process Parametera / unit
Go / Ho / Co / m/
(To) (To) P
kcal mol-1 kcal mol-1 kcal mol-1 K-1 kcal mol-1 M-1
Fab binding to TNF-α -13.1 (±0.07)b -17.8 (±0.1) -0.78 (±0.01) /
Urea unfolding of
Fab F FD 23.8 (±0.7) 38.8 (±1.5) 0.8 (±0.1) 3.6 (±0.1)
T I 5.5 (±0.4) 8.2 (±0.2) 1.5 (±0.3) 1.1 (±0.1)
3 3
TNF-α
I 3T 21.8 (±0.4) 6.5 (±1.5) -0.5 (±0.2) 2.1 (±0.05)
3
T F I 12.1 (±0.5) 18.5 (±1.5) 0.3 (±0.1) 1.9 (±0.1)
3 3 TF
TNF-α-Fab complex
I 3TD3FD 126 (±2) 166 (±5) 5.6 (±0.4) 12.8 (±0.1) Do
TF w
a ± values represent standard deviations estimated from diagonal elements of the corresponding variance- nlo
a
d
covariance matrixes. e
d
b The corresponding value of the binding constant calculated as K expGo /RT = 1.7 fro
(T ) (T ) m
o o h
(±0.3)·109 M-1 ttp
://w
w
w
.jb
c
.o
rg
b/
y
g
u
e
s
t o
n
M
a
rc
h
2
8
, 2
0
1
9
9
Energetics of TNF-α recognition

FIGURES
Figure 1:
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
://w
w
w
.jb
c
Figure 2: .o
rg
b/
y
g
u
e
s
t o
n
M
a
rc
h
2
8
, 2
0
1
9

10