Thermodynamic analysis of monoclonal antibody binding to duplex DNA
Thermodynamic analysis of monoclonal antibody binding to duplex DNA
Jamshid
Tanha
and
Jeremy S.
Lee*
Department of Biochemistry, University of Saskatchewan, 107 Wiggins Road,
Saskatoon
, Saskatchewan S7N 5E5,
Canada
Received October 28, 1996;
Revised and Accepted February 4, 1997
ABSTRACT
A technique based on fluorescence polarization (anisotropy) was used to measure
the binding of antibodies to DNA under a variety of conditions. Fluorescein-labeled duplexes of 20 bp in length were employed as the standard because
they are stable even at low ionic strength yet sufficiently short so that both
arms of an IgG cannot bind to the same duplex. IgG Jel 274 binds duplexes in
preference to single-stranded DNA; in 80 mM NaCl
K
obs for (dG)20[middot](dC)20 is 4.1
*
10
7
M
-1
compared with 6.4
*
10
5
M
-1
for d(A
5
C
10
A
5
). There is little sequence specificity, but the interaction is very dependent
on ionic strength. From plots of log
K
obs against log[Na
+
] it was deduced that five or six ion pairs are involved in complex formation.
At low ionic strength,
K
obs is independent of temperature and complex formation is entropy driven with
[Delta]
H
o
obs
and
[Delta]
C
o
p,obs both zero. In contrast, in 80 mM NaCl
[Delta]
C
o
p,obs is -630 and -580 cal mol
-1
K
-1
for [d(TG)]10[middot][d(CA)]10 and (dG)20[middot](dC)20 respectively. IgG Jel 241 also binds more tightly to
duplexes than single-stranded DNA, but sequence preferences were apparent. The values for
K
obs to [d(AT)]20 and [d(GC)]20 are 2.7
*
10
8
and 1.3
*
10
8
M
-1
respectively compared with 5.7
*
10
6
M
-1
for both (dA)
20
[middot](dT)
20
and (dG)
20
[middot](dC)
20
. As with Jel 274, the binding of Jel 241 is very dependent on ionic strength
and four or five ionic bonds are involved in complex formation with all the
duplex DNAs which were tested.
[Delta]
C
o
p,obs for Jel 241 binding to [d(AT)]20 was negative (-87 cal mol
-1
K
-1
) in 80 mM NaCl but was zero at high ionic strength (130 mM NaCl). Therefore,
for duplex-specific DNA binding antibodies
[Delta]
C
o
p,obs
is dependent on [Na
+
] and a large negative value does not correlate with sequence-specific interactions.
INTRODUCTION
Interaction of proteins with DNA is important in many cellular processes, such
as replication and transcription. As a general classification, these proteins
can be characterized in terms of their sequence specificity as well as their
preferences for single-stranded or duplex DNA. An example of a well-studied single-strand binding protein is SSB from
Escherichia coli
(
1
,
2
). It only binds to single-stranded nucleic acids and has a preference for pyrimidines over purines.
Thermodynamic parameters have been measured and this preference is probably due
to the fact that purine oligonucleotides have considerable self-structure (
2
). In the case of sequence-specific duplex DNA binding proteins, such as phage [lambda]
cro
protein or the
trp
repressor, binding to the target sequence is many orders of magnitude greater
than to random duplex DNA. The sequence preference is based on specific
contacts between the protein and functional groups on the DNA. Thermodynamic
analysis has allowed this binding to be characterized in terms of a highly
complimentary or specific interface which is accompanied by a large negative [Delta]
C
o
p,obs
(
3
,
4
).
Monoclonal antibodies are another class of DNA binding protein which have
received considerable attention because of their involvement in autoimmune
diseases (
5
-
8
). They can be produced from autoimmune strains of mice in which they occur
spontaneously or, alternatively, from mice which have been immunized with
nuclease-resistant nucleic acids (
9
-
11
). Those produced by immunization include antibodies to Z DNA, triplex DNA and
poly(dG)[middot]poly(dC) (
12
-
14
). In general they are structure specific and show limited cross-reactivity with other nucleic acids. Antibodies of autoimmune origin, on
the other hand, tend to show limited sequence and structure specificity, which
is consistent with the idea that binding is dominated by ionic interactions
with the phosphodiester backbone. Exceptions to this rule include Hed 10 and
BV04-01, which are single-strand-specific, with a preference for poly(dT), and Jel 241 and 274,
which are duplex-specific, with minor sequence specificity (
15
-
17
).
The autoimmune antibodies Jel 241 and 274 are of particular interest because
duplex binding antibodies of this type have been implicated in the pathogenesis
of the disease systemic lupus erythematosus (
18
,
19
). They appear to have a propensity for binding to the glomeruli of the kidney,
causing an inflammatory response which may eventually lead to nephritis (
20
). The origins of the antibodies are obscure, since most duplex DNAs are not
immunogenic (
7
). It has been proposed that they arise from stimulation by some, as yet,
unidentified antigen and by chance they cross-react with DNA (
21
,
22
). Alternatively, their presence may represent some defect in the immune system,
such as inappropriate self-tolerance (
23
,
24
).
Previously, binding of Jel 241 and 274 was studied by a competitive solid phase
radioimmunoassay (SPRIA), which allows measurement of relative binding
constants (
17
). By this technique it was demonstrated that Jel 274 binds well to most duplex
nucleic acids, whereas Jel 241 prefers duplexes with an alternating
pyrimidine/purine sequence. In both cases binding to single-stranded DNA was at least 100-fold lower (
17
). However, this technique cannot be used to measure thermodynamic parameters.
In this report binding studies for both Jel 241 and 274 have been extended to
include a thermodynamic analysis of their interaction with duplex DNA. This was
made possible by the use of fluorescence polarimetry (
25
-
27
).
Briefly, a fluorescein-labeled oligonucleotide is titrated with increasing concentrations of the
antibody.
The free oligonucleotide tumbles rapidly and thus the fluorescein has a low
polarization. The complex with bound antibody has a much higher molecular
weight, tumbles more slowly and, thus, has a higher polarization. Binding
parameters can then be calculated from a plot of polarization against antibody
concentration (a modified Klotz plot) (
28
). The technique is rapid, versatile and applicable to any nucleic acid which
can be labeled with fluorescein.
MATERIALS AND METHODS
Oligonucleotides
5'-Fluorescein-labeled oligonucleotides were purchased from the Calgary
Regional DNA Synthesis Facility. They were gel purified before use, except for
(dG)
20
, which tends to form insoluble aggregates. All the oligonucleotides had the
fluorescein label attached to their 5' purine with Pharmacia Fluoreprimetm. The oligonucleotide concentrations were estimated using the
published extinction coefficients at 260 nm (
29
-
32
). Control experiments showed that the presence of fluorescein did not interfere
with these determinations. (dA)
20
[middot](dT)
20,
[d(TG)]
10
[middot][d(CA)]
10,
[d(AT)]
20
, [d(GC)]
20
and [d(AT)]
10
were prepared in phosphate-buffered saline, pH 7.4 (2.7 mM KCl, 137 mM NaCl, 10 mM Na
2
HPO
4
, 1.4 mM KH
2
PO
4
), whereas (dG)
20
[middot](dC)
20
was prepared in 10 mM Tris-HCl plus 0.1 mM EDTA, pH 8.0. Briefly, to prepare (dA)
20
[middot](dT)
20
,
[d(TG)]
10
[middot][d(CA)]
10
and(dG)
20
[middot](dC)
20
, equimolar amounts of the complementary strands were mixed, heated at 95oC for 10 min and slowly cooled down (>30 min) to room temperature. To
prepare [d(AT)]
20
, [d(GC)]
20
and [d(GC)]
10
, the oligonucleotides were heated at 95oC for 10 min and quenched in an ice bath. Under these conditions the
alternating sequence DNAs will preferentially form intramolecular duplexes with
hairpin structure. Duplex formation was monitored by an ethidium bromide
fluorescence assay (
13
) and
T
m
measurements showed a single helix to coil transition. The triplex (dT)
20
[middot](dA)
20
[middot](dT)
20
was formed by incubating the strands at a 2:1 ratio in 10 mM Tris-HCl, pH 8.0 plus 2 mM MgCl
2
for 2 h at 20oC. Triplex formation was monitored by the ethidium bromide fluorescence
assay (
13
). Oligonucleotide duplexes (0.4 [mu]g 20mer or 1.6 [mu]g 10mer) were analyzed on non-denaturing 20% acrylamide gels and stained with ethidium (
33
).
Antibodies
Jel 274 is a re-clone of Jel 229 which has retained the original specificity as judged by
SPRIA (
34
). IgG Jel 241 and 274 were purified by gel exclusion and ion exchange
chromatography as described previously (
35
). Analysis of the antibodies on SDS-polyacrylamide gels showed the absence of contaminating protein. Fab 274
was prepared by papain digestion and purified as described previously (
35
). Concentrations were expressed in terms of binding sites (i.e. per Fab) and
were calculated assuming 1.5 A
280
= 1 mg/ml and a molecular mass of 150 000.
Polarization measurements
A Panvera bioluminescent polarimeter (PanVera Corporation) was used for all
experiments. The standard buffer was 10 mM potassium phosphate, pH 7.2,
supplemented with NaCl to give the required ionic strength. Except where noted,
the oligonucleotide concentration was 1 nM. For experiments with triplex (dT)
20
[middot](dA)
20
[middot](dT)
20
10 mM Tris-HCl, pH 8, with 2 mM MgCl
2
was used. Briefly, antibodies were serially diluted in 12 * 75 mm borosilicate tubes (Fisher Scientific) containing a constant
amount of fluorescein-labeled oligonucleotide in a total volume of 1 ml. Complex formation was
rapid and reached equilibrium in <1 min (data not shown). Polarization values (in millipolarization units, mP)
for each tube (and associated fluorescence intensity) were determined with the
polarimeter in single blank mode. In this mode the polarimeter automatically
subtracts the background fluorescence values for each sample and reports the
corrected fluorescence intensity and polarization values. The maximum
fluorescence intensity quenching ranged from 25 to 70%, depending on the
combination of antigen and antibody used. The mP values were corrected for this
drop in intensity as described (
36
). The results were plotted as polarization versus antibody concentration.
K
obs
, the association constant, was determined by fitting the data to the single
site binding isotherm
mP
= (
mP
max
K
obs
[
AB
] +
mP
min
)/(1 +
K
obs
[
AB
])
1
with Deltaplottm.
mP
max
and
mP
min
are the maximum and minimum millipolarization values and [
AB
] is the free antibody concentration.
mP
min
was fixed and
mP
max
and
K
obs
were allowed to vary independently.
It should be noted that identical values for
K
obs
can be calculated using anisotropy as the variable, rather than polarization.
Control experiments showed that neither Jel 274 nor Jel 241 showed binding to
free fluorescein. Also, binding to [d(TG)]
10
[middot][d(CA)]
10
was not dependent on which strand carried the fluorescein label (data not
shown). In some cases
K
obs
could also be determined from the fluoresence quenching data as described
previously (
15
). For measurements at 5, 16, 30 and 38oC, the samples were incubated in a water bath at preset temperatures for ~5 min, wiped dry and then read immediately. Control tubes showed that
the variation in temperature was +-1oC. Reported
K
obs
values are based on the average of at least three determinations and
experimental errors were 5-10%.
Thermodynamic parameters
K
obs
, [Delta]
G
o
obs
(the observed standard free energy change), [Delta]
H
o
obs
(the observed standard enthalpy change), [Delta]
S
o
obs
(the observed standard entropy change) and [Delta]
C
o
p,obs
(the observed standard heat capacity change) are related by the following
functions
[Delta]
G
o
obs
= -
RT
ln
K
obs
2
[Delta]
G
o
obs
= [Delta]
H
o
obs
-
T
[Delta]
S
o
obs
3
ln
K
obs
= (-[Delta]
H
o
obs
/
R
)(1/
T
) + ([Delta]
S
o
obs
/
R
) (van't Hoff plot)
4
[Delta]
C
o
p,obs
= ([part][Delta]
H
o
obs
/[part]
T
)
p
5
Where
R
is the gas constant and
T
is the absolute temperature.
Thus, by measuring
K
obs
as a function of temperature all the parameters can be calculated. [Delta]
H
o
obs
is obtained from -[Delta]
H
o
obs
/
R
, the slope of the plot of ln
K
obs
against 1/
T
(van't Hoff plot). In cases where van't Hoff plots were non-linear (i.e. [Delta]
H
o
obs
varied with temperature), the best fit was obtained by the second order
function
ln
K
obs
=
a
(1/
T
)
2
+
b
(1/
T
) +
c
6
The derivative of the above function with respect to 1/
T
is equal to the slope of the van't Hoff plot
-[Delta]
H
o
obs
/
R
= 2
a
(1/
T
) +
b
7
Therefore, [Delta]
H
o
obs
can be obtained at any temperature from equation
7
. As a result, if [Delta]
H
o
obs
varies with temperature, [Delta]
C
o
p,obs
is not zero and can be obtained from equation
5
.
RESULTS
ACKNOWLEDGEMENT
This research was supported by the Medical Research Council of Canada.
REFERENCES
1 Bujalowski,W., Overman,L.B. and Lohman,T.M. (1988) J. Biol. Chem., 263, 4629-4640.MEDLINE Abstract
2 Ferrari,M.E. and Lohman,T.M. (1994) Biochemistry, 33, 12896-12910.MEDLINE Abstract
3 Takeda,Y., Ross,P.D. and Mudd,C. (1992) Proc. Natl. Acad. Sci. USA, 89, 8180-8184.MEDLINE Abstract
4 Ladbury,J.E., Wright,J.G., Sturtevant,J.M. and Sigler,P.B. (1994) J. Mol. Biol., 238, 669-681.MEDLINE Abstract
