Tài liệu Báo cáo khoa học: Exposure of IgG to an acidic environment results in molecular modifications and in enhanced protective activity in sepsis doc

Exposure of IgG to an acidic environment results in
molecular modifications and in enhanced protective
activity in sepsis
Iglika K. Djoumerska-Alexieva
1,
*, Jordan D. Dimitrov
1,2,3,4,
*, Elisaveta N. Voynova
1
,
Sebastien Lacroix-Desmazes
2,3,4
, Srinivas V. Kaveri
2,3,4
and Tchavdar L. Vassilev
1
1 Department of Immunology, Stefan Angelov Institute of Microbiology, Bulgarian Academy of Sciences, Sofia, Bulgaria
2 Centre de Recherche des Cordeliers, Universite
´
Pierre et Marie Curie Paris 6, France
3 Universite
´
Paris Descartes, France
4 INSERM U 872, Eq. 16, Paris, France
Introduction
The ability of antibodies to interact with one single or
with multiple structurally unrelated antigens (monore-
activity versus polyreactivity) is believed to be an
inherent property of each individual immunoglobulin
molecule. However, it has been previously shown by
us, as well as by others, that the in vitro exposure of

monoclonal and of polyclonal IgG to various protein-
destabilizing factors may result in dramatic enhance-
ment of their binding polyreactivity. These treatments
include high-salt solutions, low-pH or high-pH buffers,
chaotropic agents, ferrous ions, reactive oxygen
species, and heme [1–7].
Keywords
antibodies; antibody polyreactivity; antigen–
antibody interaction; IgG; immunoglobulins
Correspondence
T. Vassilev, Stefan Angelov Institute of
Microbiology, Bulgarian Academy of
Sciences, Acad. G. Bonchev St., Block 26,
1113 Sofia, Bulgaria
Fax: +359 2 870 0109
Tel: +359 2 979 6348
E-mail:
*These authors contributed equally to this
work
(Received 22 February 2010, revised 22
April 2010, accepted 18 May 2010)
doi:10.1111/j.1742-4658.2010.07714.x
IgG molecules are exposed on a regular basis to acidic conditions during
immunoaffinity purification procedures, as well as during the production of
some therapeutic immunoglobulin preparations. This exposure is known to
induce in them an antigen-binding polyreactivity. The molecular mecha-
nisms and the possible biological significance of this phenomenon remain,
however, poorly understood. In addition to the previously reported ability
of these modified IgG antibodies to interact with a large panel of self-anti-
gens, enhanced binding to non-self-antigens (bacterial), an increased ability

to engage in F(ab¢)
2
⁄ F(ab¢)
2
(idiotype ⁄ anti-idiotype) interactions and an
increased functional antigen-binding affinity are reported here. The newly
acquired ‘induced polyreactivity’ of low-pH buffer-exposed IgG is related
to structural changes in the immunoglobulin molecules, and is at least
partly attributable to the enhanced role of the hydrophobic effect in their
interactions with antigen. Our results suggest that data from many previous
studies on monoclonal and polyclonal IgG antibodies purified by low-pH
buffer elution from protein A or protein G immunoaffinity columns should
be reconsidered, as the procedure itself may have dramatically affected
their antigen-binding behavior and biological activity. Low-pH buffer-trea-
ted pooled therapeutic immunoglobulins acquire novel beneficial properties,
as passive immunotherapy with the pH 4.0 buffer-exposed, but not with
the native therapeutic intravenous immunoglobulin preparation, improves
the survival of mice with bacterial lipopolysaccharide-induced septic shock.
Abbreviations
ANS, 8-anilinonaphthalene-1-sulfonate; CRP, C-reactive protein; IFN-c, interferon-c; IVIg, intravenous immunoglobulin;
LPS, lipopolysaccharide; RU, relative units.
FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS 3039
A comparative study of seven licensed commercially
available pooled therapeutic intravenous immunoglob-
ulins (IVIgs) revealed that those produced using a frac-
tionation step at low pH were significantly more
polyreactive when tested on a complex mix of self-anti-
gens [8]. The molecular mechanisms responsible for the
effect of low-pH buffer exposure on IgG molecule
have remained, however, poorly understood. IVIg

preparations with lower and with higher antigen-bind-
ing polyreactivity have quantitatively different effects
on cells in vitro. Low-pH buffer-exposed IVIg causes
significantly stronger suppression of Phaseolus vulgaris
agglutinin-induced proliferation of human peripheral
blood mononuclear cells than the native preparation
[8]. Polyreactive natural antibodies form part of the
innate immunity mechanism, and are known to play a
major role in preventing pathogen dissemination in the
preimmune host [9–11]. Natural polyreactive antibod-
ies are detected in the sera of all healthy individuals,
and their immunoreactivity increases dramatically in
the pure IgG fractions, purified from the same sera
using low-pH buffer elution from immunoaffinity col-
umns [12]. However, the physicochemical characteris-
tics and the biological activities of IgG antibodies
transiently exposed to low-pH conditions remain
unknown. We hypothesized that low-pH (£ 4) buffer
exposure could endow a commercially available IVIg
preparation with novel beneficial therapeutic proper-
ties. The present study shows that low-pH buffer treat-
ment of a commercial IVIg results in its enhanced
binding to bacterial antigens as well as to self-antigens,
owing to structural changes in the immunoglobulin
molecules. The modified preparation is shown to have
a protective effect in experimental sepsis.
Results
Exposure to a low-pH buffer increases the
binding polyreactivity of IgG to foreign antigens
Previous studies showed that low-pH buffer-exposed

IVIg acquired enhanced autoreactivity [8]. The first
aim of the study was to find out whether the same
broadening of IgG polyreactivity occurred when for-
eign antigens were used. The exposure of pooled
human IgG to a pH 4 buffer resulted in an increase
in its pre-existing binding to antigens present in an
Escherichia coli lysate, and in the appearance of some
new bands showing the acquisition of new antibody
reactivities (Fig. 1A). Interestingly, the same treatment
did not significantly change the reactivity to Bacillus
anthracis antigens. In contrast, the transient exposure
of IVIg to pH 2.8 buffer resulted in a significant
(P < 0.05) enhancement of immunoreactivity, and in
A
B
C
E
D
Fig. 1. Exposure of IgG to a pH 4 buffer results in increased antigen-binding polyreactivity. (A) Densitometric profiles of the reactivity of the
native (solid line) and low-pH buffer-exposed (dashed line) IVIg to Escherichia coli antigens. Migration distances (x-axis) expressed in pixels
were plotted against the intensity of binding (y-axis) expressed in relative units (RU) for each IVIg preparation. (B) Reactivity of native (lines 1
and 4), pH 4 buffer-exposed (lines 2 and 5) and pH 2.8 buffer-exposed (lines 3 and 6) IVIg with Bacillus anthracis antigens. The membranes
were incubated with two concentrations of IVIg: 100 lgÆmL
)1
(lines 1–3) and 50 lgÆmL
)1
(lines 4–6). (C) Increased binding of pH 4 buffer-
exposed IVIg to recombinant human IFN-c. (D) Increased binding of low-pH buffer-exposed IVIg to human factor H. In both panels, binding
of the native IVIg is indicated by a solid line, and that of the low-pH buffer-exposed IVIg is indicated by a dashed line. (E) Binding of two
commercially available IVIg preparations [Endobulin (solid line); Octagam (dashed line)] to human factor H. Data represent mean absorbance

values ± standard deviation of quadruplicate wells in one of three ELISA experiments.
Molecular modifications in low pH-exposed IgG I. K. Djoumerska-Alexieva et al.
3040 FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS
the appearance of a number of novel antigen-binding
specificities (Fig. 1B).
Enhanced binding of low-pH buffer-exposed IVIg
to recombinant human interferon-c (IFN-c) and
other self-antigens
Interactions of IVIg with cytokines and other mole-
cules of the immune system have been shown to play a
role in the immunomodulatory effect of the prepara-
tion [13]. To determine whether the IgG treatment
described affected binding to a typical proinflammato-
ry cytokine, the interactions of the native and the
low-pH buffer-exposed IVIg preparations with recom-
binant human IFN-c were compared by ELISA. The
pH 4 buffer-exposed IVIg showed significantly
(P < 0.05) stronger binding to this human proinflam-
matory cytokine (Fig. 1C). Next, the reactivity of
native and of low-pH buffer (pH 2.8)-exposed IVIg
towards a panel of structurally unrelated pure plasma
proteins or intracellular self-proteins was analyzed.
This transient exposure resulted in increases in their
binding to all tested target antigens [see Fig. 1D for
reactivity to factor H; data not shown for all other
antigens (see Experimental procedures)]. We also com-
pared the antigen-binding potentials of two commer-
cially available IVIg preparations that differ in the
absence or presence of exposure to low-pH conditions
during the production process (Endobulin versus Octa-

gam). As expected, the binding of the second to fac-
tor H was significantly (P < 0.05) higher (Fig. 1E).
The same was true for all other antigens in the panel
(not shown).
Low-pH buffer exposure increases the
anti-idiotypic reactivity of IgG
After a pH 4 buffer exposure, the studied IVIg prepa-
ration showed enhanced binding to IVIg F(ab¢)
2
frag-
ments (Fig. 2A) as well as to autologous pooled IgM
molecules (Fig. 2B). In contrast, no increase in reactiv-
ity of the modified IVIg to Fc-c (Fig. 2C) or Fc-l
fragments (Fig. 2D) was observed.
Antigen-binding kinetics of low-pH
buffer-exposed IVIg
In order to obtain quantitative information on the
effect of low-pH buffer exposure on the reactivity of
IgG, we used real-time kinetic measurements of the
interaction of IVIg with C-reactive protein (CRP). The
binding profiles obtained after single injections of
native and of low-pH buffer-exposed IVIg were com-
pared. As shown in Fig. 3A, transient (5 min) exposure
of IVIg to pH 2.8 buffer resulted in an increase in the
reactivity towards human CRP, as detected by this
nonequilibrium binding assay. The reactivity of the
pH 4 buffer-exposed preparation was also elevated,
but to a much lower extent (approximately nine-fold).
The native IVIg preparation showed no detectable
binding to CRP at the concentration and during the

period of observation used in the experiment.
Interaction analyses using increasing concentrations
of the native and of the low-pH buffer-treated IVIg
were also performed (Fig. 3B). The data obtained were
used to evaluate the kinetic constants of these inter-
actions. The bimolecular association rate constant of
Fig. 2. Low-pH buffer exposure of pooled
human IgG enhances its binding to F(ab¢)
2
immunoglobulin fragments. Dialysis of IVIg
against a pH 4 buffer results in increased
binding to F(ab¢)
2
fragments of IVIg (A) and
to pooled human IgM (B), but not to Fc-c
(C) or Fc-l fragments (D), as assessed by
ELISA. Data represent mean absorbance
values ± standard deviation of quadruplicate
wells (solid lines, native IVIg; dashed lines,
low-pH buffer-exposed IVIg).
I. K. Djoumerska-Alexieva et al. Molecular modifications in low pH-exposed IgG
FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS 3041
binding of the pH 2.8 buffer-exposed IVIg to CRP
was very low, at 33.5 ± 1.3 mol
)1
Æs
)1
. The estimated
dissociation rate constant had a value of (2.32 ·
10

)3
) ± (2.0 · 10
)5
s
)1
). The equilibrium dissociation
constant for CRP and low-pH buffer-exposed IVIg
was 69.1 lm. The absence of detectable binding or
negligible responses precluded the estimation of reli-
able values of kinetic parameters for the native and
the pH 4 buffer-exposed IVIg preparations.
Enhanced role of hydrophobicity in the binding
of a pH 4.0 buffer-treated monoclonal IgG
antibody to its target antigen
To evaluate the types of intermolecular interactions in
antigen binding, pH and salt concentration screening
assays were performed. This study was carried out
using the mouse monoclonal Z2 antibody, which
behaves in its native form as a typical monoreactive
antibody, as it interacts only with mouse IgG
2a
[14].
The interaction of the native Z2 antibody with its
immobilized target antigen was highly pH-dependent,
and characterized by a bell-shaped curve with a bind-
ing optimum at neutral pH (Fig. 4A). On the other
hand, the interaction of low-pH buffer-exposed Z2
antibody was much less dependent on the pH of the
buffer. From pH 4.5 upwards, the binding reached a
plateau and became almost independent of further

increases in pH.
The salt concentration dependence of the same inter-
action (within the range 0–4 m sodium chloride) was
also studied. The binding of the native Z2 antibody to
IgG
2a
was shown to be highly dependent on the salt
concentration in the buffer. In contrast, this interac-
tion was mostly independent of the salt concentration
in the case when the low-pH buffer-exposed Z2 was
left to bind with its target antigen (Fig. 4B). Both
observations could be explained by an increased role
for the hydrophobic effect in the interaction of the
modified monoclonal antibody with its immobilized
antigen.
Increase in the IgG hydrophobicity as evaluated
by 8-anilinonaphthalene-1-sulfonate (ANS)
fluorescence
In order to confirm the increased role of the hydro-
phobic effect upon low-pH buffer exposure of IgG,
fluorescence spectroscopy using a molecular probe for
protein hydrophobicity was applied. ANS changes its
A
B
Fig. 3. Real-time interaction analysis of the binding of IVIg to human CRP. (A) Comparison of interaction profiles of 5 lM native IVIg (black
line), pH 4 buffer-exposed IVIg (gray line) and pH 2.8 buffer-exposed IVIg (light gray line) with human CRP. (B) Profiles characterizing the
interactions of increasing concentrations (0.039–1.25 l
M) of native IVIg (left panel), pH 4 buffer-exposed IVIg (middle panel) and pH 2.8 buf-
fer-exposed IVIg (right panel) with the same human molecule. The sensorogram depicting the interaction of pH 2.8 buffer-exposed IVIg was
used for evaluation of the binding affinity by global analyses. All measurements were performed at 25 °C. The results obtained in one of

two independent experiments are shown.
Molecular modifications in low pH-exposed IgG I. K. Djoumerska-Alexieva et al.
3042 FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS
fluorescence properties with the polarity of the envi-
ronment. Thus, the transition from a polar to a non-
polar (hydrophobic) environment results in a dramatic
increase in its fluorescence signal. This property and
the ability of ANS to bind to proteins make it a widely
used molecular probe for the evaluation of hydropho-
bicity of proteins, as well as for exploring structural
alternations in protein molecules [15–18].
Incubation of native IVIg in the presence of ANS
resulted in a modest change in the fluorescence signal
(Fig. 4C). Similar spectral characteristics were mea-
sured in the case of IVIg exposed to a pH 4 buffer,
implying the absence of a significant increase in the
total hydrophobicity of the immunoglobulins. The lack
of correlation between these findings and the data
shown in Fig. 4A,B could well be explained by the dif-
ferent IgG preparations studied (pooled, polyclonal
versus monoclonal) and the different sensitivities of the
methods used. Dramatic changes in the fluorescence
characteristics of ANS were seen in the presence of
pH 2.8 buffer-exposed IVIg. Thus, a considerable
increase in the fluorescence intensity and a blue shift in
the fluorescence maxima were observed. These effects
were observed at different concentrations of ANS
(Fig. 4D). These findings further confirmed the results
from the pH and ionic strength dependencies of the
interactions, demonstrating the increased hydrophobic-

ity of low-pH buffer-exposed IgG molecules.
Fluorescence studies on low-pH buffer-exposed
immunoglobulins
Here, the physicochemical mechanisms responsible for
the increased IgG antigen recognition potential after
low-pH buffer exposure were studied. Aromatic amino
acids in proteins possess intrinsic fluorescence proper-
ties when excited at an appropriate wavelength. The
fluorescence characteristics (intensity and wavelength
of the fluorescence maxima) depend on the polarity of
the local protein environment. Thus, changes in the
positions of the fluorescent amino acids, caused by
structural modifications of the molecule, result in
changes in the microenvironment of the aromatic
amino acids that affect the fluorescence characteristics,
especially the position of the emission maxima [15,19].
For these reasons, fluorescence spectroscopy is widely
used for the analysis of structural changes and of the
stability of proteins.
We used tryptophan fluorescence in order to deter-
mine whether the exposure of immunoglobulin
Fig. 4. Low-pH buffer exposure of IgG antibodies results in an enhanced role of the hydrophobic effect in their antigen binding. (A) A
pH-scanning ELISA analysis of the interaction of the mouse monoclonal Z2 IgG antibody with its target antigen. (B) The same antigen–
antibody interaction in the presence of increasing concentrations of NaCl. In both experiments, binding intensities are represented in RU,
and each data point shows the mean value ± standard deviation of quadruplicate wells. Solid lines, native Z2; dotted lines, low-pH buffer-
exposed Z2. (C) Emission spectra of 32 l
M ANS in the absence or presence of 2 lM native, pH 4 buffer-exposed or pH 2.8 buffer-exposed
IVIg. (D) Comparison of the fluorescence characteristics of increasing concentrations of ANS (1–32 l
M) in the presence of 2 lM native IVIg
(solid lines) or pH 2.8 buffer-exposed IVIg (dashed lines). The emission spectra of ANS were recorded after excitation at 388 nm.

I. K. Djoumerska-Alexieva et al. Molecular modifications in low pH-exposed IgG
FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS 3043
molecules to an acidic milieu would result in structural
modifications. Indeed, an increase in the fluorescence
intensity of the pooled IgG preparation and a slight
red shift in the emission maxima following its exposure
to a pH 2.8 buffer were detected by fluorescence spec-
troscopy (Fig. 5). This effect is consistent with a
change in the positions of tryptophan(s) in the immu-
noglobulins. The red shift in the emission maxima is
indicative of the relocation of tryptophan(s) to a more
polar environment. In contrast, treatment of the same
preparation with a pH 4 buffer did not change the
tryptophan fluorescence of the same molecules.
Increase in the relative functional antigen-binding
affinity of a monoclonal antibody after its
exposure to a pH 4 buffer
The relative functional affinities of the native and the
low-pH buffer-exposed monoclonal Z2 antibody were
analyzed by thiocyanate elution ELISA [20]. In the
elution assay, 0–3.0 m potassium thiocyanate was used
to disrupt the binding of the Z2 antibody to its target,
mouse IgG
2a
. The functional affinity was defined by
the molar concentration of potassium thiocyanate
required for a 50% reduction in binding as detected in
ELISA at A
405 nm
. The modified mouse monoclonal

IgG antibody bound more strongly to the immobilized
antigen (Fig. 6).
Passive immunotherapy with low-pH
buffer-exposed, but not with native IVIg has
protective activity in mouse sepsis
The effect of treatment with low-pH buffer-exposed
IVIg on the survival of lipopolysaccharide (LPS)-
injected animals was studied. The decision to test the
modified immunoglobulin preparation in experimental
sepsis was based on its enhanced binding to IFN-c,on
the known role of polyreactive antibodies in infections
[10], and on the hypothesis of Antonio Coutinho and
Stratis Avrameas suggesting that polyreactive antibod-
ies may represent a buffering system that prevents
brisk changes in the levels of components of inflamma-
tion, coagulation, and other pathways [21]. The admini-
stration of a single dose of 500 mgÆkg
)1
of the
modified IVIg had significant (P < 0.05) protective
activity in this experimental sepsis model. The native
IVIg was not protective, regardless of the dose used
(Fig. 7A–D). Two sets of data strongly argue that the
therapeutic effect of the modified IVIg was not due to
better neutralization of the injected LPS: first, the
binding of the preparation to LPS in its two forms
was identical (tested by ELISA, data not shown); and
second, the pH 4.0 buffer-exposed IVIg significantly
(P < 0.05) decreased mortality, even if injected after
the administration of LPS (Fig. 7E).

Discussion
The brief exposure of polyclonal and at least some
monoclonal IgG preparations to a low-pH buffer
results in an alteration of the immunoglobulin struc-
ture, and in the acquisition of enhanced antigen recog-
nition behavior and new biological activities. Our data
show that the changes fall short of denaturation of the
immunoglobulin molecules. The main argument in
support of this claim that low-pH-modified IgG
molecules fully retain their F(ab¢)
2
-dependent and
Fig. 5. Increase in the fluorescence intensity of pooled IgG after
low-pH buffer exposure. Spectrofluorometric analyses of native
(solid line), pH 4 buffer-exposed (dotted line) and pH 2.8 buffer-
exposed (dashed line) IVIg. The ordinate represents fluorescence
intensity in RU.
Fig. 6. Increased functional affinity of a mouse monoclonal Z2 anti-
body after its exposure to a pH 4 buffer. Thiocyanate elution ELISA
was performed as described in Experimental procedures. The func-
tional affinity is represented by the molar concentration of potas-
sium thiocyanate required for a 50% reduction in binding as
detected at A
405 nm
. The results represent the average of at least
three independent measurements, with the standard deviation indi-
cated by error bars (gray bar, native Z2 antibody; black bar, pH 4
buffer-exposed Z2 antibody).
Molecular modifications in low pH-exposed IgG I. K. Djoumerska-Alexieva et al.
3044 FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS

Fc-dependent antibody functions is based on clinical
experience over many years with IVIg preparations
produced using a low-pH buffer fractionation step.
This study supports a previous suggestion [8] that com-
mercially available immunoglobulin preparations are
not equal, and shows that the differences between
them might be large enough to be clinically relevant.
The possibility that nonspecific aggregation of IgG
on surface-adsorbed antigens is responsible for the
observed enhanced immunoreactivity can be ruled out
by the results presented above. Spectroscopic data
(tryptophan fluorescence and ANS fluorescence) did
not imply the presence of aggregation of the IgG mole-
cules in solution after low-pH buffer treatment. In
contrast, we observed an increase in the fluorescence
signal after exposure of IVIg to pH 2.8 buffer. In the
case of IgG aggregation, a decrease in fluorescence
intensity would be expected. Nonspecific aggregation
on the surface of the adsorbed antigens is also ruled
out by the fact that the increased antigen recognition
is not observed in the case of all studied antigens. Our
real-time kinetic measurements imply that the complex
of low-pH buffer-treated IgG with CRP dissociates,
although at a slow rate.
Wymann et al. [22] have recently suggested that the
increased antigen-binding polyreactivity of pH 4 buf-
fer-exposed immunoglobulin preparations was mainly
caused by the dissociation of IgG dimers in them. Our
data strongly suggest, however, that the effect of this
exposure goes beyond IgG–IgG dimer dissociation,

and affects the IgG molecules themselves.
A possible explanation for the finding of new anti-
gen-binding specificities after exposure to low-pH con-
ditions is the induction of structural rearrangements in
the variable region of the antibody. Indeed, we
observed changes in the tryptophan fluorescence char-
acteristics of IVIg after its exposure to a low-pH (2.8)
buffer. The increase in the fluorescence intensity and
the red shift in the emission maximum are signs of a
change in the position of tryptophan(s) in the IgG
molecules – a mark of the existence of a structural
modification in the polypeptide chains of the immuno-
globulins. In addition, our results revealed that the
interactions of low-pH buffer-exposed IgG are less
dependent on changes in the ionic strength or the pH
of the medium. Such binding behavior is typical of
protein–protein bonds that rely on nonpolar types of
interaction (hydrophobic effect and van der Waals
contacts). Indeed, the increase in the hydrophobicity
may well be explained by exposure of previously bur-
ied hydrophobic amino acids to the solvent, as shown
previously for IgG treated with chaotropic agents [23].
By using the fluorescence molecular probe for hydro-
phobicity of proteins (ANS), we confirmed that the
exposure of IgG to a low-pH buffer results in molecu-
lar modifications characterized by a considerable
increase in their surface-exposed hydrophobicity. The
antigen-binding behavior of low-pH buffer-exposed
IgG preparations was enhanced for some, but not all,
antigens tested (e.g. IgG Fc fragments). We are cur-

rently investigating whether the proteins that are pref-
erentially recognized by the modified antibodies share
any common features.
It has been observed that the increased polyreacti-
vity of a monoclonal IgG after transient exposure to
urea correlates well with the elevated flexibility of the
antigen-binding site and the involvement of hydropho-
bic interactions as a driving force for the recognition
of the target antigen. All of these findings allow us to
hypothesize that elevated binding to various molecular
patterns of monoclonal and polyclonal IgG after their
transient exposure to low pH (pH 4 or less) may
well be caused by an augmentation of the structural
A
CD
E
B
Fig. 7. Treatment with pH 4 buffer-exposed IVIg reduces mortality
in bacterial LPS-induced septic shock. Survival curves of mice (15
per group) injected with 0.5 mg of LPS and treated with 4 mgÆkg
)1
(A), 20 mgÆkg
)1
(B), 100 mgÆkg
)1
(C) or 500 mgÆkg
)1
(D) native IVIg
(solid lines), or pH 4 buffer-exposed IVIg (dashed lines), or NaCl ⁄ P
i

(pH 7.4) alone (gray lines). The protective activity of the modified
IVIg is retained, even when its administration has been postponed
for 1 h (E). *P < 0.05, Mann–Whitney test.
I. K. Djoumerska-Alexieva et al. Molecular modifications in low pH-exposed IgG
FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS 3045
plasticity of their paratopes. Urea and low-pH buffer
exposure are both known to induce some degree of
melting of the protein conformation. The increased poly-
reactivity observed could well be explained by limited
melting of immunoglobulin molecules by either of
these agents.
Our previous data have indicated that treatment of
some IgG antibodies with different redox-active agents
is also able to enhance antigen-binding polyreactivity
by modulating the properties of the antigen-combining
sites of some, but not all, studied IgG antibodies. The
binding behavior of some, generated in response to
repeated immunizations and expected to have rigid,
high-affinity binding sites, is not modified by exposure
to these conditions [5]. The precise mechanisms that
make an individual IgG molecule resistant to polyreac-
tivity-inducing treatment remain to be determined.
Redox-active agents (heme, iron ions, reactive oxygen
species) are released in vivo in inflammation sites as
well as after trauma, hemorrhages, etc. Exposure to
low pH is part of the production process for some
commercial IVIg preparations, and is used on a daily
basis for IgG immunoaffinity purification.
The dramatically increased antigen-binding polyreac-
tivity of polyclonal and of some monoclonal IgG anti-

bodies that have been briefly exposed to a pH 4.0 or
pH 2.8 buffer suggests that conclusions from many
previous studies on IgG, purified by low-pH buffer
elution from Protein A, Protein G or other immuno-
affinity columns, have to be carefully re-examined. We
[24,25] and others [26] have previously used a similar
approach to immunopurify anti-self-antigen-binding
IgGs from IVIg by recirculating the pooled immuno-
globulin preparation through immunoaffinity columns,
containing pure immobilized self-antigens, and then
eluting the bound fractions by washing the columns
with a pH 2.8 buffer. The latter were found to be
enriched in natural autoantibodies with the expected
specificities. These antibodies were further tested in
various in vitro assays, and shown to engage in biologi-
cally relevant interactions. Spalter et al. claimed that
all individuals, regardless of their ABO histo-blood
antigen groups, possessed both anti-A and anti-B anti-
gen IgG in their sera. Their main argument was based
on the ability of pure IgG, isolated from the sera of
donors with an A, B, AB or a O blood group, to bind
to the A as well as to the B antigens. The pure IgG
fraction was obtained, however, by pH 3 buffer elution
from a Protein G Sepharose affinity column. We pro-
pose an alternative explanation for the same observa-
tions. Even a brief exposure to a low-pH milieu (pH 4
or lower) modifies some of the circulating IgG
molecules, resulting in their enhanced antigen-binding
polyreactivity and possibly in the acquisition of the
ability to bind to a polysaccharide self-antigen (A or

B). The low-pH buffer elution of the same fraction
may also enhance its capacity to engage in
F(ab¢)
2
⁄ F(ab¢)
2
(idiotype ⁄ anti-idiotype) interactions
with other immunoglobulin molecules (see Fig. 2).
IVIg preparations are used in patients with primary
and secondary immunodeficiencies, as well as in an
increasing number of autoimmune and inflammatory
diseases. To the best of our knowledge, no compara-
tive clinical studies on the immunomodulatory effects
of IVIg preparations produced by different protein
fractionation technologies have been performed so far.
Data from this and from an earlier study [8] strongly
suggest that licensed therapeutic IVIgs exposed to pro-
duction steps at low pH do acquire new, clinically rele-
vant, properties. Their use could be beneficial in the
early stages of sepsis, which are characterized by
uncontrolled production of proinflammatory mediators
(‘cytokine storm’). Previous studies have shown that
binding to bacterial LPS is not affected in IgG mole-
cules modified by protein-destabilizing agents [5],
strongly suggesting that the prevention of LPS-induced
sepsis death is due to the ability of this preparation
to attenuate the hyperreactivity of body defense
mechanisms.
In addition to sepsis, there is an increasing number
of emerging infectious diseases in which the severe gen-

eralized inflammatory reaction of the infected host is a
major factor in the poor outcome. Recent additions to
the list are H5N1 influenza (avian flu), dengue,
Marburg and Lassa hemorrhagic fevers, and West Nile
virus infection [27–31]. One could speculate that pas-
sive immunotherapy with ‘modified’ IVIg preparations
would be beneficial in patients with these diseases. An
important argument in favor of low-pH buffer-exposed
IVIg is that it has already been in clinical use for a
long time, whereas the ferrous ion and heme-exposed
IVIg preparations are at an early preclinical evaluation
stage.
Experimental procedures
Monoclonal antibody immunoglobulin
preparations, and immunoglobulin fragments
The Z2 hybridoma producing a mouse monoclonal IgG
2b
antibody against mouse IgG
2a
was kindly provided by
E. Rajnavolgyi (Department of Immunology, Lorand Eotvos
University, Budapest, Hungary). The commercial intrave-
nous immunoglobulins Endobulin S ⁄ D (Baxter, Deerfield,
IL, USA), and Octagam (Octapharma, Lachen, Switzerland)
were used in the experiments. F(ab¢)
2
and Fc fragments of
Molecular modifications in low pH-exposed IgG I. K. Djoumerska-Alexieva et al.
3046 FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS
IVIg, as well as an experimental pooled human IgM

preparation, were prepared as described previously [32,33].
Pure human Fc-l fragments, obtained from a patient with
l-heavy chain disease, were a gift from L. Mouthon (Cochin
Hospital, Paris).
The Z2 antibody and IVIg samples were diluted in 0.1 m
sodium acetate buffer (pH 4.0 or 2.8) and incubated for
5 min. The pH was then brought to 7.0, and the samples
were dialyzed against NaCl ⁄ P
i
(pH 7.2) [8]. The pH 4.0
buffer was chosen because several commercial IVIg prepa-
rations are produced using a fractionation step with this
pH value. Buffers of pH 2.8 are widely used for the isola-
tion of pure IgG by affinity chromatography.
Immunoblot analysis
The total lysate from a nonpathogenic strain of B. anthra-
cis was kindly provided by S. Mesnage (Centre de Recher-
che des Cordeliers, Paris, France). A lysate of E. coli was
prepared as described elsewhere [5]. Both bacterial antigen
extracts were subjected to 10% SDS ⁄ PAGE and trans-
ferred to nitrocellulose membranes (Scheicher & Schuell,
Dassel, Germany) with a Mini Transfer Blot system (Bio-
Rad, Richmond, CA, USA) in a buffer containing 48 mm
Tris, 110 mm glycine, and 20% (v ⁄ v) methanol. Then, they
were incubated for 1 h at room temperature in NaCl ⁄ Tris
containing 0.3% Tween-20. Membranes were further cut
into strips or fixed in a miniblot system, and incubated
for 1 h at room temperature with the native, the pH 4.0
buffer-exposed or the pH 2.8 buffer-exposed IVIg prepara-
tions (at 0.1 mgÆmL

)1
). After extensive washing, they were
incubated with goat anti-human IgG (Fc-specific), conju-
gated to alkaline phosphatase (Southern Biotech, Birming-
ham, AL, USA), and finally developed using the Nitro
Blue tetrazolium and bromo-chloro-indolyl-phosphate
substrates (both from Sigma-Aldrich, Taufkirchen,
Germany). The quantitation of bound antibodies in immu-
noblots was performed by densitometry in reflective mode,
using a UMAX 1220p scanner linked to a PC. The data
were analyzed using imagetool v2.0 for Windows
(UTHSCSA, San Antonio, TX, USA). Migration distance
(x-axis) was expressed in pixels, and intensity of binding
level (y-axis) was expressed in relative units (RU).
ELISA
Ninety-six-well polystyrene plates (Brand GMBH, Wert-
heim, Germany, or Nunc, Denmark) were coated with
5 lgÆmL
)1
recombinant human IFN-c (a gift from I. Iva-
nov, Institute of Molecular Biology, Bulgarian Academy of
Sciences, Sofia, Bulgaria), with 2 lgÆmL
)1
human fac-
tor VIII, 10 lgÆmL
)1
human factor IX (LFB, France),
10 lgÆmL
)1
human CRP, 10 lgÆmL

)1
human C3 (both
from Calbiochem), 10 lgÆmL
)1
human factor H,
10 lgÆmL
)1
human factor B (both from Complement
Technology, TX, USA), 20 lgÆmL
)1
porcine thyroglobulin,
20 lgÆmL
)1
rabbit tubulin, and 10 lgÆmL
)1
bovine myelin
basic protein (all three from Sigma-Aldrich), for 2 h at
room temperature. Plates were blocked with 0.25–
0.4% (v ⁄ v) Tween-20 in NaCl ⁄ P
i
for 2 h. After washing
with NaCl ⁄ P
i
containing 0.05% Tween-20, the plates were
incubated overnight at 4 °C (in the case of IFN-c) or for
2 h at 25 °C (in the case of other proteins) with increasing
concentrations of the immunoglobulin preparations under
study. The plates were then extensively washed, and goat
anti-(human IgG) (c-chain-specific) coupled to alkaline
phosphatase was added and incubated for 1 h at room tem-

perature. Immunoreactivities were revealed by adding p-ni-
trophenyl phosphate (Sigma-Aldrich) diluted in appropriate
buffers.
The pH and salt concentration dependence of the binding
of the native or low-pH buffer-exposed mouse monoclonal
Z2 antibody to its cognate antigen were analyzed by
ELISA. Polystyrene plates were coated with 20 lgÆmL
)1
of
a mouse IgG
2a
monoclonal antibody (clone IP2-11-1). The
free binding sites were blocked with NaCl ⁄ P
i
containing
0.5% (v ⁄ v) Tween-20 for 2 h at room temperature. After
washing, the plates were incubated overnight at 4 °C with
dilutions of the native or the low-pH buffer-exposed Z2.
For the pH-scanning analysis, the Z2 antibody was diluted
to 15 lgÆmL
)1
in buffers with different pH values, as fol-
lows: in citrate ⁄ phosphate-buffered saline [0.05 m sodium
citrate, 0.05 m Na
2
HPO
4
, 0.14 m NaCl, 0.05% (v ⁄ v)
Tween-20] with a pH in the range 3–7.5, or in carbonate-
buffered saline [0.05 m NaHCO

3
, 0.05 m Na
2
CO
3
, 0.14 m
NaCl, 0.05% (v ⁄ v) Tween-20] with a pH in the range
8.5–12. In the ELISA with increasing salt concentrations,
the Z2 antibody was diluted to 15 lgÆmL
)1
in buffers with
0–4 m sodium chloride. After a 2 h incubation step under
the described conditions, the plates were washed and
further incubated with an alkaline phosphatase-conjugated
goat anti-mouse IgG
2b
(PharMingen, San Diego, CA,
USA) for 1 h at room temperature. The following steps of
the assay were performed as described above. The results
were represented in RU. The binding at pH 7.0 or in the
presence of 0 m NaCl buffers, respectively, was referred to
as 1 RU.
The abilities of both IVIg variants to engage in idio-
type ⁄ anti-idiotype interactions were compared by ELISA.
Polystyrene plates were coated with F(ab¢)
2
or Fc IVIg
fragments, with pooled IgM, or with pure Fc-l fragments
(all at 10 lgÆmL
)1

in coating buffer). The blocking and
washing steps were performed as described above. The
plates were incubated with increasing concentrations of the
IVIg preparations under study, and after extensive washing,
goat anti-human IgG (Fc-specific; Sigma-Aldrich) coupled
to alkaline phosphatase was added for an additional 1 h at
room temperature. In the case when Fc-c fragments were
used as coating antigen, goat anti-human IgG [F(ab)
2
-spe-
cific] (PharMingen, San Diego, CA, USA) was used to
I. K. Djoumerska-Alexieva et al. Molecular modifications in low pH-exposed IgG
FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS 3047
reveal antibody binding. The final steps were performed as
described above.
Real-time kinetic measurements
The kinetic constants of the interactions between IVIg and
human CRP were determined by surface plasmon resonance
(BIAcore 2000; GE Biacore, Uppsala, Sweden). CRP was
immobilized on research-grade CM5 chips, using an amino-
coupling kit (Biacore) as described by the manufacturer. In
brief, CRP was diluted in 5 mm maleic acid (pH 5) to a
final concentration of 100 lgÆmL
)1
, and coated on the pre-
activated sensor surface. Experiments were performed using
HBS-EP (0.01 m Hepes, pH 7.4, containing 0.15 m NaCl,
3mm EDTA, and 0.005% Tween-20) as running and sam-
ple dilution buffer. IVIg (native or low-pH buffer-exposed
Endobulin) was injected at concentrations in the range

5–0.039 lm at a flow rate of 10 lLÆmin
)1
. The association
and dissociation phases of the interaction were monitored
for 5 min. The regeneration of the chip surface was
performed using a 5 m solution of guanidine-HCl (Sigma-
Aldrich). The binding to the surface of the uncoupled
control flow cell was always subtracted from the binding to
the protein-coated flow cells. biaevaluation software (ver-
sion 4.1; Biacore) was used for the calculation of the kinetic
rate constants. Calculations were performed by global
analysis of the experimental data using the kinetic models
included in the software, fitting the data with lowest value
of v
2
.
Fluorescence spectroscopy
Intrinsic emission spectra measurements of the native and
of the low-pH buffer-exposed IVIg were performed on a
Hitachi F-2500 spectrofluorometer. Samples of the IVIg
were exposed at 4 °C to pH 4 or pH 2.8 acetate buffers.
After incubation, the samples were dialyzed against
NaCl ⁄ P
i
. All analyses were carried out at 25 °C, using 1 cm
quartz cuvette. The samples were diluted in NaCl ⁄ P
i
to a
final concentration of 2 lm. A wavelength of 295 nm,
which excites tryptophans, was used for the fluorescence

spectra measurements. The fluorescence emission spectra
were recorded between 300 and 450 nm. The excitation and
emission slits were both 10 nm, and the scan speed was
1500 nmÆmin
)1
.
ANS fluorescence
ANS was obtained from Sigma-Aldrich. IVIg prepara-
tions (at 2 lm, native as well as low-pH buffer-exposed)
were mixed with increasing concentrations of ANS
(1–32 lm). After excitation at 388 nm, the fluorescence
emission spectra of ANS were recorded between 425 and
600 nm in a 1 cm quartz curette. The excitation and
emission slits were set to 10 nm, and the scan speed was
1500 nmÆmin
)1
.
Thiocyanate elution ELISA
The thiocyanate elution ELISA was performed as described
previously [20]. Briefly, ELISA plates were coated with a
mouse IgG
2a
monoclonal antibody and, after washing, were
further incubated overnight at 4 °C with 15 lgÆmL
)1
of the
native or low-pH buffer-exposed mouse monoclonal Z2
antibody in the presence of increasing concentrations of
potassium thiocyanate (ranging from 0 to 2.0 m). After
incubation and washing, the antibody binding was mea-

sured using a goat anti-mouse IgG
2b
as described above.
Experimental septic shock
Outbred ICR mice were purchased from the Breeding Farm
of the Bulgarian Academy of Sciences. The experimental
protocols were approved by the Animal Care Commission
of the Institute of Microbiology, in accordance with
National and European Regulations. The number of ani-
mals used was kept at the minimum that still ensured statis-
tical significance of survival differences between the
experimental groups. Septic shock was induced in 16–18-
week-old animals by the intraperitoneal administration of
400 lg of bacterial LPS (from E. coli B 055:B5, Sigma-
Aldrich, #L2880). Minutes later, groups of mice (15 per
group) were injected intravenously with increasing doses of
the native IVIg or of the low-pH buffer-exposed IVIg prep-
aration, or with NaCl ⁄ P
i
alone. Survival was observed for
5 days. In a separate experiment, the native and the modi-
fied IVIg (500 mgÆkg
)1
, 10 animals per group) were infused
1 h after the administration of LPS. Any effect of a treat-
ment started at this time-point should be the result of its
influence on the pathophysiological mechanisms of the sep-
sis syndrome [34].
Statistical analysis
Statistical analyses were performed using graphpad prism,

version 4.00 (GraphPad Software, San Diego, CA, USA).
Statistical analyses of the ELISA data and the areas under
the curve of the immunoblot densitometric profiles were
performed using the paired Student t-test. For survival
analyses, differences between groups were analyzed by the
Mann-Whitney U-test. In all cases, P-values < 0.05 were
considered to indicate statistical significance.
Acknowledgements
This study was supported by the Bulgarian National
Science Fund (grants VU-L-314 ⁄ 07 and TK-X-
1710 ⁄ 07), by the NATO Science for Peace Program
Molecular modifications in low pH-exposed IgG I. K. Djoumerska-Alexieva et al.
3048 FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS
(SfP 982158), and by INSERM and CNRS, France.
J. D. Dimitrov is supported by a fellowship from the
Fondation pour la Recherche Me
´
dicale en France. The
authors are grateful to A. Pashov (University of
Arizona, Little Rock, AR, USA) for reading the
manuscript.
References
1 Bouvet JP, Stahl D, Rose S, Quan CP, Kazatchkine
MD & Kaveri SK (2001) Induction of natural autoanti-
body activity following treatment of human immuno-
globulin with dissociation agents. J Autoimmun 16,
163–172.
2 Bussone G, Dib H, Dimitrov J, Camoin L, Broussard
C, Tamas N, Guillevin L, Kaveri S & Mouthon L
(2009) Identification of target antigens of self-reactive

IgG in intravenous immunoglobulin preparations.
Proteomics 9, 2253–2262.
3 Dimitrov J, Roumenina L, Doltchinkova V, Mihaylova
N, Lacroix-Desmazes S, Kaveri S & Vassilev T (2007)
Antibodies use heme as a cofactor to extend their path-
ogen elimination activity and to acquire new effector
functions. J Biol Chem 282, 26696–26706.
4 Dimitrov J, Roumenina L, Doltchinkova V & Vassilev
T (2007) Iron ions and haeme modulate the binding
properties of complement subcomponent C1q and of
immunoglobulins. Scand J Immunol 65, 230–239.
5 Dimitrov JD, Ivanovska ND, Lacroix-Desmazes S,
Doltchinkova VR, Kaveri SV & Vassilev TL (2006)
Ferrous ions and reactive oxygen species increase
antigen-binding and anti-inflammatory activities of
immunoglobulin G. J Biol Chem 281, 439–446.
6 McIntyre J, Wagenknecht D & Faulk W (2005)
Autoantibodies unmasked by redox reactions.
J Autoimmun 24, 311–317.
7 McMahon MJ & O’Kennedy R (2000) Polyreactivity as
an acquired artefact, rather than a physiologic property
of antibodies: evidence that monoreactive antibodies
may gain the ability to bind to multiple antigens after
exposure to low pH. J Immunol Methods 241, 1–10.
8 Djoumerska I, Tchorbanov A, Pashov A & Vassilev T
(2005) The autoreactivity of therapeutic intravenous
immunoglobulin (IVIG) preparations depends on the
fractionation methods used. Scand J Immunol 61,
357–363.
9 Notkins A (2004) Polyreactivity of antibody molecules.

Trends Immunol 25, 174–179.
10 Ochsenbein AF, Fehr T, Lutz C, Suter M, Brombacher
F, Hengartner H & Zinkernagel RM (1999) Control of
early viral and bacterial distribution and disease by
natural antibodies. Science 286, 2156–2159.
11 Zhou Z, Zhang Y, Hu Y, Wahl L, Cisar J & Notkins A
(2007) The broad antibacterial activity of the natural
antibody repertoire is due to polyreactive antibodies.
Cell Host Microbe 1, 51–61.
12 Mouthon L, Nobrega A, Nicolas N, Kaveri SV,
Barreau C, Coutinho A & Kazatchkine MD (1995)
Invariance and restriction toward a limited set of self-
antigens characterize neonatal IgM antibody repertoires
and prevail in autoreactive repertoires of healthy adults.
Proc Natl Acad Sci USA 92, 3839–3843.
13 Kazatchkine MD & Kaveri SV (2001) Immunomodula-
tion of autoimmune and inflammatory diseases with
intravenous immune globulin. N Engl J Med 345,
747–755.
14 Fazekas G, Rajnavolgyi E, Kurucz I, Sinta’r E, Kiss K,
Laszlo G & Gergely J (1990) Isolation and characteriza-
tion of IgG2a-reactive autoantibodies from influenza
virus-infected BALB ⁄ c mice. Eur J Immunol 20,
2719–2729.
15 Lakowicz JR (2006) Principles of Fluorescence Spectro-
scopy, 3rd edn. New York, Springer.
16 Easterbrook-Smith SB & Dwek RA (1980) The use of
ANS fluorescence as a probe for immunoglobulin
flexibility. FEBS Lett 121, 253–256.
17 Matulis D, Baumann C, Bloomfield V & Lovrien R

(1999) 1-Anilino-8-naphthalene sulfonate as a protein
conformational tightening agent. Biopolymers 49,
451–458.
18 Sudhakar K & Fay P (1996) Exposed hydrophobic sites
in factor VIII and isolated subunits. J Biol Chem 271,
23015–23021.
19 Harding SE & Chowdhry BZ (2001) Protein–Ligand
Interactions: Structure and Spectroscopy. Oxford
University Press, Oxford.
20 Pullen G, Fitzgerald M & Hosking C (1986) Antibody
avidity determination by ELISA using thiocyanate
elution. J Immunol Methods 86, 83–87.
21 Coutinho A & Avrameas S (1992) Speculations on
immunosomatics: potential diagnostic and therapeutic
value of imune homeostasis concepts. Scand J Immunol
36, 527–532.
22 Wymann S, Ghielmetti M, Schaub A, Baumann M,
Stadler B, Bolli R & Miescher S (2008) Monomerization
of dimeric IgG of intravenous immunoglobulin (IVIg)
increases the antibody reactivity against intracellular
antigens. Mol Immunol 45, 2621–2628.
23 Dimitrov J, Lacroix-Desmazes S, Kaveri S & Vassilev T
(2007) Transition towards antigen-binding promiscuity
of a monospecific antibody. Mol Immunol 44, 1854–1863.
24 Vassilev T, Gelin C, Kaveri SV, Zilber MT, Boumsell L
& Kazatchkine MD (1993) Antibodies to the CD5 mol-
ecule in normal human immunoglobulins for therapeu-
tic use (intravenous immunoglobulins, IVIg). Clin Exp
Immunol 92, 369–372.
25 Vassilev TL, Kazatchkine MD, Van Huen JPD,

Mekrache M, Bonnin E, Mani JC, Lecroubier C,
I. K. Djoumerska-Alexieva et al. Molecular modifications in low pH-exposed IgG
FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS 3049
Korinth D, Baruch D, Schriever F et al. (1999) Inhibi-
tion of cell adhesion by antibodies to Arg-Gly-Asp
(RGD) in normal immunoglobulin for therapeutic use
(intravenous immunoglobulin, IVIg). Blood 93, 3624–
3631.
26 Spalter SH, Kaveri SV, Bonnin E, Mani JC, Cartron JP
& Kazatchkine MD (1999) Normal human serum con-
tains natural antibodies reactive with autologous ABO
blood group antigens. Blood 93, 4418–4424.
27 Atrasheuskaya A, Petzelbauer P, Fredeking T &
Ignatyev G (2003) Anti-TNF antibody treatment
reduces mortality in experimental dengue virus
infection. FEMS Immunol Med Microbiol 35, 33–42.
28 Cheung CY, Poon M, Lau AS, Luk W, Lau YL,
Shortridge KF, Gordon S, Guan Y & Peiris JS (2002)
Induction of proinflammatory cytokines in human
macrophages by influenza A (H5N1) viruses: a
mechanism for the unusual severity of human disease?
Lancet 360, 1831–1837.
29 Ignatyev G, Steinkasserer A, Streltsova M, Atrasheus-
kaya A, Agafonov A & Lubitz W (2000) Experimental
study on the possibility of treatment of some
hemorrhagic fevers. J Biotechnol 83, 67–76.
30 Rimmelzwaan G, van Riel D, Baars M, Bestebroer T,
van Amerongen G, Fouchier R, Osterhaus A & Kuiken
T (2006) Influenza A virus (H5N1) infection in cats
causes systemic disease with potential novel routes of

virus spread within and between hosts. Am J Pathol
168, 176–183.
31 Venter M, Burt F, Blumberg L, Fickl H, Paweska J &
Swanepoel R (2009) Cytokine induction after labora-
tory-acquired West Nile virus infection. N Engl J Med
360, 1260–1262.
32 Debre M, Bonnet MC, Fridman WH, Carosella E,
Philippe N, Reinert P, Vilmer E, Caplan C, Teillaud JL
& Griscelli C (1993) Infusion of Fcg fragments for the
treatment of children with acute immune thrombocyto-
penic purpura. Lancet 342, 945–949.
33 Hurez V, Kazatchkine MD, Vassilev T, Ramanathan S,
Basuyaux B, Pashov A, De-Kozak Y, Bellon B &
Kaveri SV (1997) Pooled normal human polyspecific
IgM contains neutralizing antiidiotypes to IgG
autoantibdies of autoimmune patients and protects
from autoimmune disease. Blood 90, 4004–4013.
34 Cohen J (2002) The immunopathogenesis of sepsis.
Nature 420, 885–891.
Molecular modifications in low pH-exposed IgG I. K. Djoumerska-Alexieva et al.
3050 FEBS Journal 277 (2010) 3039–3050 ª 2010 The Authors Journal compilation ª 2010 FEBS