Moss Tom. DNA-protein interactions: principles and protocols

Подождите немного. Документ загружается.

Calorimetry of Protein–DNA Complexes 513

constant and T is the absolute temperature. For the simple equilibrium describ-

ing the interaction of protein P with DNA, D:

P + D ⇔ PD then K

[PD]

(1)

[P][D]

Techniques such as equilibrium dialysis, ultracentrifugation, the gel-shift,

or radio-ligand binding assays yield direct values for [P], [D], and [PD] and,

therefore, values for K

and ∆G at any fixed temperature. More indirect, mostly

spectroscopic, methods allow the construction of a binding isotherm in terms

of the degree of saturation, Y:

Y =

∆p

[PD]

[P]

(2)

∆p

max

[D] + [PD] 1 + K

[P]

where ∆p

is the signal change obtained at some particular addition of P to a

constant amount of D and ∆p

max

is the maximal signal change at full saturation

(Y = 1). If the experiment is repeated at different temperatures, K

(T) and ∆G(T)

are obtained. ∆H(T) and ∆S(T) can be derived from the integrated form of the

van Hoff equation. Further, the s derivative of ∆G(T) with respect to tempera-

ture yields ∆C

. This noncalorimetric analysis of binding data (sometimes

referred to as a van Hoff analysis) can be flawed. First, the analysis assumes a

model, which may or may not be correct, to describe the interaction of the

components. Second, for technical reasons, experiments can be performed only

over a limited temperature range and the experimental errors propagate into

large errors in ∆H, ∆S, and especially ∆C

when the extrapolation from the

reference temperature to any other temperature is large.

1.2. The ITC Experiment

Modern mixing microcalorimeters allow precise measurements of the

enthalpy of binding, ∆H

app

, between 5°C and 80°C. The basic principle is that

one binding partner (protein, P) is titrated at a constant temperature into a

known amount of the other binding partner (DNA, D) placed in the sample cell

of the calorimeter. Heat is produced or released when binding occurs. The instru-

ment measures the electrical power (in units of J/s) required to maintain a small

temperature difference between the sample cell and the reference cell, which is

filled with buffer. The contents of the sample cell are stirred to allow rapid mixing

and effective heat transfer over the surface of the cell. If K

is high and the degree

of saturation is still low, electrical power peaks of similar magnitude appear in

the thermogram at each addition. Integration with respect to time yields the

apparent heat change, ∆q

i,app

, between additions i and i–1 as ∆q

i,app

= q

– q

–1.

As the fractional saturation increases, ∆q

i,app

gradually decreases and even-

tually all binding sites become saturated. Small nonspecific heat effects, ∆q

i,ns

514 Read and Jelesarov

registered after complete saturation may be caused by the heat of protein

dilution, an imperfect match between the buffer composition of the protein

solution and the DNA solution, or by other nonspecific effects. ∆q

i,app

is pro-

portional to the volume of the calorimetric cell, V

cell

, to the change in concen-

tration of the bound protein ∆[P]

i,bound =

[P]

i,bound

= [P]

i–1,bound

and to the

apparent molar enthalpy of association ∆H

app

. Thus,

∆q

i,app

= ∆q

+ ∆q

i,ns

(3)

= ∆[P]

i,bound

cell

∆H

app

∆H

app

at temperature T may therefore be calculated because V

cell

is known and

∆q

i,ns

may be obtained from a blank titration of protein into buffer.

For the binding of protein, P, with n identical and noninteracting sites to

DNA, D,

∆q

= ∆q

i,app

+ ∆q

i,ns

(4)

= nR V

cell

∆H

app

∆q

is the effective heat change caused by the formation of complex, PD, at

the i

step of the titration and R is the root of the quadratic equation

– Y

(

1 +

[P]

i,total

)

+ n[P]

i,total

[D]

total

= 0 (5)

[D]

total

n[D]

total

where the fractional saturation is defined by Y

= ∆[P]

i,bound

/ [D]

total

. [P]

i,total

the total concentration of protein, P, added until injection, i. A nonlinear

regression procedure based on Eq. 4 yields n, K

, and ∆H

app

at temperature T,

from a single titration experiment. More complicated systems with multiple

and interacting sites require a statistical mechanical treatment of the data

(11,12). ∆C

app

the apparent heat capacity change of association can be calcu-

lated from a plot of ∆H

app

versus temperature, because in general, ∆C

= ∆H / ∆T.

As in other binding experiments, the dissociation constant, K

(equal to 1/K

)

can only be reliably measured over a narrow concentration range. If the con-

centration of binding sites is much larger than K

, the binding isotherm is of

rectangular shape with a sharp step at saturation. In the case of the binding-site

concentration being much smaller than K

, the binding isotherm is very shal-

low and is of limited use for calculating the dissociation constant. In practice,

values of the ratio of the binding-site concentration to K

that range between

10 and 100 represent an optimal experimental window to obtain precise values

of the dissociation constant. Unfortunately, this window is not always acces-

sible to ITC. For very small K

values, the optimal concentrations are so low

that the released heat is below the sensitivity of the instrument, the specific

enthalpy of biomolecular interactions never being very large. For this reason,

there is an effective lower limit to the ITC measurement of K

to values of

Calorimetry of Protein–DNA Complexes 515

approx 10

–9

M (∆G about –50 kJ/mol). Alternatively, if K

is high, then very

high concentrations are required and the heat of binding may be obscured by

aggregation effects.

The two types of ITC measurement are normally obtained under different

conditions. The procedure to obtain n, K

, and ∆H

app

is best carried out as a

full titration at concentrations near the stoichiometric point of the interaction

(i.e., at a DNA concentration 10–100 times the estimated value for K

and with

the injection of up to a 2 M to 4 M excess of protein. However, the direct

measurement of ∆H

app

is best carried out in conditions where the system is

fully associated and the degree of saturation is low (i.e., at high DNA concen-

trations (>1000K

) and with the injection of low amounts of protein). The total

number of moles of added protein is then much less than the number of moles

of binding sites available and virtually all protein molecules are bound to DNA,

such that ∆[P]

i,bound

approximately equals ∆[P]

i,total

. Each injection of protein,

∆[P]

i,total

thus produces an approximately equal change in heat, ∆q

i,app

. If the

overall fractional saturation is still low after completion of a series of injec-

tions, then the average may be taken and Eq. 3 becomes

m m

∑∆qi

,app

∑

(

∆qi

,ns

)

(6)

i=1

∑

(

∆[P]

i,bound +

cell

∆H

app

)

i=1

where m is the number of injections. If low amounts of protein are injected,

aggregation effects and nonspecific binding are also minimized.

1.3. The DSC Experiment

In a DSC experiment, the heat capacity of a macromolecule is measured as a

function of temperature. Typically, two thermally insulated cells, a sample cell

containing the macromolecule of interest in buffer and a reference cell contain-

ing buffer, are electrically heated at a known rate. At a temperature-induced

transition, which is typically endothermic, the temperature of the sample cell

will lag behind that of the reference cell. An electrical feedback mechanism is

used to maintain the reference cell at the same temperature as the sample cell.

This amount of compensatory electrical power (in units of J/s or Watts) at

temperature T divided by the heating rate is the apparent difference in heat capac-

ity between the cell containing the sample and the reference cell, ∆C

p,app

(T) (in

units of J/K). Because the sample containing cell has a smaller volume fraction of

buffer as compared to that in the reference cell, the partial molar heat capacity of

the dissolved macromolecule C

p,f

(T) at temperature T (J/mol/K) is given by

516 Read and Jelesarov

p,φ

(T) =

p,buffer

(T)V

˚M∆C

p,app

(T)

(7)

buffer

where C

p,buffer

(T) and V

buffer

are the partial molar heat capacity and molar vol-

ume of buffer, respectively. V

˚ and M are the partial molar volume and the

molar mass of the macromolecule, respectively, and m is the mass of macro-

molecule in the sample cell.

The excess molar heat capacity function is the heat absorbed in the melting

transition above that of the intrinsic heat capacity of the macromolecule. Inte-

gration of the excess molar heat capacity function with respect to temperature

yields the enthalpy of the melting transition:

∆H = 兰具C

(T)典dT (8)

The intrinsic heat capacity of the initial (folded) and final (unfolded) states

of the macromolecule must be estimated within the melting transition. For

monomeric proteins, the heat capacity of the folded state is an approximate

linear function of temperature (13,14) and the unfolded state is a shallow para-

bolic function of temperature (13).

The relationship of the partial molar heat capacity to the excess molar heat

capacity functions are schematically shown in Fig. 1.

1.4. Preliminary Characterization of the Sox-5–DNA Interaction

It must be emphasized that the in vitro interaction of the protein with DNA

must be characterized prior to the calorimetry experiments. Preliminary

knowledge of the number of (independent) binding sites, the stoichiometry of

the interaction, and the magnitude of the association constant are all required

for the optimal design of the ITC experiments (see Subheading 1.2.). For DSC,

the experiments must be performed with a fully associated complex (at low

temperature) and therefore at a concentration well above the estimated K

Furthermore, the protein–DNA complex and its components need to be soluble

at this concentration, even at raised temperatures, and all temperature-induced

conformational transitions must be fully reversible. Knowledge of the mode of

interaction is also required to obtain van Hoff enthalpies from the DSC data.

Site-selection and DNAse I footprinting assays (Chapter 3) defined the

Sox-5 target sequence as 5’-AACAAT-3' within a DNAse I protected region of

approx 14 nucleotides on both DNA strands (15,16). This determined the

sequence and size of the DNA duplexes that were to be used. A circular permu-

tation assay showed that the binding of Sox-5 to these duplexes was able to

introduce a DNA bend of similar magnitude to that obtained with the complete

Calorimetry of Protein–DNA Complexes 517

Fig. 1. Schematic representation of (A) the partial molar heat capacity function as

would be observed for the melting of a single domain monomeric protein. Over the

transition region the intrinsic heat capacity function is an interpolation of the folded

and unfolded states, weighted in proportion to their relative contributions (dashed line).

(B) replots the C

/T function above that of the intrinsic heat capacity of the system.

This is known as the molar excess heat capacity function.

518 Read and Jelesarov

Sox-5 protein (6,16). The dissociation constant and the stoichiometry of the

Sox-5–DNA interaction were determined from circular dichroism (CD) and

gel-shift assays (Chapters 2 and 34):

1. The binding of Sox-5 to a 12-bp DNA duplex (10 µM) was followed by CD,

monitoring the ellipticity of the positive DNA peak at 280 nm upon successive

additions of the protein. The CD data were fitted to a 1:1 binding model with an

estimated K

of lower than 100 nM (6). This same estimate was obtained at tem-

peratures between 10°C and 37°C, suggesting no significant variation of binding

constant with temperature.

2. A gel-shift assay using a radioactively labeled 12-bp DNA duplex (at 1 nM) was

titrated in molar excess with Sox-5 (6). This indicated that the primary binding

site is characterized by a K

of about 35 nM at 4°C. No secondary protein binding

was observed with <1 µM Sox-5.

These assays thus indicate a bimolecular interaction of Sox-5 with the DNA,

with a single primary binding site characterized by a K

in the low nanomolar

range. Furthermore any secondary binding of Sox-5 to DNA, if present, must

have a K

of >10 µM (i.e., at least two orders greater than the primary binding

site). Thus, for the DSC measurements, typically at a concentration of 100–

500 µM complex, the preformed Sox-5–DNA complex is fully associated and

the effect of secondary binding is negligible.

Given the estimated K

value, the ITC measurement of a full binding iso-

therm under optimal conditions requires a DNA concentration of between

approx 0.35 and 3.5 µM (see Subheading 1.2.). Preliminary titration experi-

ments however showed only a small heat effect of association (a result of the

temperature dependency of ∆H

app

which passes through zero at 17°C) and the

presence of a secondary binding event, with a large exothermic effect, after

saturation of the primary binding site (Fig. 2). It was concluded that measure-

ment of the entire binding isotherm was not possible by ITC.

However, conditions could be chosen that were optimal for the direct deter-

mination of ∆H

app

by ITC (i.e., total association at partial saturation (see Sub-

heading 1.2.). At the chosen DNA concentration of 60 µM, total association of

injected Sox-5 in an ITC experiment is achieved (the ratio of the DNA concen-

tration to the estimated K

being approx 1700). The fractional molar saturation

was always kept lower than 0.5 to avoid any secondary binding effects. Because

high concentrations were to be used, the total heat effect was greater and more

easily detectable, which meant that ∆H

app

values could be accurately obtained

at temperatures close to 17°C.

Additional experiments included low-speed analytical ultracentrifugation at

different temperatures to demonstrate that the free Sox-5 is monomeric. For

the free DNA, UV melting was performed as an initial check on its melting

temperature and for two-state melting of the duplexes. UV melting of the

Calorimetry of Protein–DNA Complexes 519

Sox-5–DNA complex showed a single cooperative transition with a melting

temperature above that of the free DNA.

Fig. 2. ITC titration experiment of Sox-5 into a 12-bp DNA duplex. (A) Fifteen

injections of 540 µM Sox-5 into 10 µM of 12-bp DNA at 9°C. The injection rate was

5 µL in 8 s, with a time interval of 5 min between injections. (B) a plot of ∆H

app

against

ratio of Sox-5 to DNA showing the primary endothermic and secondary exothermic

reactions that occur on either side of the stoichiometric point.

520 Read and Jelesarov

2. Materials

2.1. Calorimeters

Isothermal titration calorimetry was performed on an OMEGA or MCS-ITC

titration calorimeter (MicroCal Inc., Northampton, MA). Technical details on

the construction of mixing microcalorimeters, their performance and sensitiv-

ity, and on the theory of data analysis have been described elsewhere (17–20).

Differential scanning calorimetric experiments were usually carried out on

a Nano-DSC calorimeter (Calorimetric Science Corp., Utah). The instrument’s

performance and data acquisition are detailed in ref. 21. In brief, the instru-

ment operates over the temperature range –20°C to 130°C at any set heating/

cooling rate. The sample and reference calorimetric cells are of approx 900 µL

volume. Control of the calorimeter and scan acquisition is achieved via the

supplied DSC_Acquisition program running on a PC computer connected to

the calorimeter. In experiments in which the amount of sample was limiting, a

new version of the Nano-DSC calorimeter having cells of approx 300 µL vol-

ume was used.

2.2. Reagents and Solutions

2.2.1. Calorimetry

1. For calorimetric experiments, it is important to ensure that the protein and DNA

samples are homogeneous, because even a few percent of contaminating species

might cause significant heat effects. Homogeneity of the samples is most reliably

verified by mass spectrometry.

2. Concentrations must be accurately determined because this affects the DSC and

ITC data (see Subheading 3.4.). The preparation of equimolar mixtures (of

complementary oligonucleotides or Sox-5-DNA complex) can then be made

precisely, without further purification. Dilutions are made by weight on a preci-

sion balance.

3. The DSC and ITC experiments with the Sox-5–DNA complex used a working

buffer of 100 mM KCl, 10 mM potassium phosphate, and 1 mM EDTA (pH 6.0)

(see Note 1). All buffer solutions were prepared from the highest-quality reagents

and ultrapure water. For ITC experiments, the solutions were filtered through

a 0.45-µm membrane and thoroughly degassed prior to use.

2.2.2. Preparation of Sox-5 HMG Box–DNA Complexes

and Components

1. DNA oligonucleotides were synthesized using phosphoramidite chemistry and

purified on a Mono Q HR16/10 column fitted to a Pharmacia FPLC system, elut-

ing with a linear 0.1 M to 1.0 M NaCl gradient in 10 mM Tris-HCl, 1 mM EDTA,

and 20% (v/v) acetonitrile (pH 7.0). Fractions containing oligonucleotide were

precipitated with 3 vol of ethanol at –20°C overnight, then centrifuged down and

Calorimetry of Protein–DNA Complexes 521

redissolved in water. Oligonucleotides were extensively dialyzed using

Spectrapor tubing (molecular-weight cutoff 500 Daltons) against three changes

of 1 L working buffer at 4°C.

2. DNA duplexes were prepared by mixing equimolar amounts of the complemen-

tary oligonucleotides in working buffer and annealed by heating to 95°C in a

water-bath followed by slow cooling to 4°C over a period of approx 4 h. DNA

duplexes were then extensively dialyzed using 3000-Dalton molecular-weight

cutoff tubing against three changes of 1 L working buffer at 4°C.

3. The HMG box of mouse Sox-5 (amino acids 182 to 260, [15]) was expressed as

a fusion protein in pGEX-2T, using E. coli BL21 (DE3) plysS cells. After affinity

purification with glutathione-agarose and thrombin cleavage while still

attached to the column (22), reverse-phase high-performance liquid chromatog-

raphy (HPLC) was used to purify the protein. Protein was redissolved in water

and refolded by extensive dialysis against three changes of 1 L working buffer

at 4°C.

4. The Sox-5–DNA complex was prepared by the mixing together of equal volumes

of the components (at equimolar concentration) in working buffer at 4°C. Sox-5

was added in 10% aliquots, at 5-min intervals, to the DNA. The complex was

then extensively dialyzed using 3000-Dalton molecular-weight cutoff tubing

against three changes of 1 L working buffer at 4°C. The accuracy of the concen-

trations of protein and DNA used to form the 1:1 complex may be checked by

trial additions of Sox-5 to DNA at various protein:DNA ratios followed by elec-

trophoresis on a nondenaturing polyacrylamide gel.

2.2.3. DNA Concentrations

1. 100 mM Tris-HCl (pH 8.0).

2. Snake venom phosphodiesterase I (PDE1, from Crotalus durissus terrificus, Sigma).

3. Methods

3.1. Isothermal Titration Calorimetry

Before a series of experiments, the calorimeter was calibrated either by

applying electrically generated heat pulses or by standardized chemical reac-

tions (e.g., the protonation of tris[hydroxymethyl] aminomethane or the bind-

ing of Ba

to 18-crown-6 ether). It is recommended to equilibrate the jacket

with a circulating water-bath for about 10 h. at a temperature lower than the

temperature of the experiment by 3–5°C. Although such equilibration is not

strictly necessary when operating at above room temperature, it substantially

improves the baseline stability. The stirring speed during both the equilibra-

tion and the experiment was 350 rpm (see Note 2).

1. Sample and reference cells are first rinsed with dialysis buffer. The reference cell

is then filled with buffer and the sample cell filled with the DNA solution. The

system is heated to the working temperature and equilibrated until the differen-

522 Read and Jelesarov

tial power signal levels off. The injection syringe containing Sox-5 is inserted

into the sample cell, stirring is initiated, and the baseline is established over a

period of 30–60 min. Typically, a baseline drift (differential power signal drift)

of less than 20–30 nJ/min and an rms noise of less than 15–20 nJ/s (or nW) indi-

cates complete thermal equilibration of the system under stirring (see Note 3).

2. The experiment is started with a small injection of 1–2 µL. The reason for this is

that during the long equilibration period, diffusion through the injection ports

occurs, thus causing a change in protein concentration near the syringe needle

tip. The actual injection schedule is then executed. To measure the enthalpy of

Sox-5 binding to DNA, six to eight injections each of 8–12 µL and of 12–15 s

duration were performed, with a 5-min interval between injections (see Note 4).

Typical thermograms are depicted in Fig. 3A, traces a and c.

3. After completion of the experiment, the cells are thoroughly cleaned. Cleaning of

the calorimetric cells and filling syringes follows a standard laboratory protocol

(see Note 5).

4. The cells may then be filled with buffer and step 2 repeated. Injections of protein

into buffer will yield the heat associated with protein dilution and other non-

specific effects. Typical control thermograms are depicted in Fig. 3A, traces b

and d. Because the heats obtained in steps 2 and 4 are directly compared in the

data analysis, it is crucial that (1) the blank titration is performed at exactly the

same temperature as the main experiment, (2) the same protein solution and

dialysis buffer are used, and (3) an identical injection scheme is executed. If

any of the listed requirements is not fulfilled, the results of the experiment could

be misleading.

5. The titrations are performed at different temperatures to collect data on the tem-

perature dependence of the binding enthalpy.

3.1.1. ITC Data Analysis

If a complete binding isotherm has been recorded over an optimal concen-

tration range, the data can be subjected to a nonlinear least-squares analysis to

obtain a full set of parameters (∆H

app

, K

, and n) according to Eq. 4. In the

case of Sox-5 binding to DNA, the experiments were designed to measure only

the enthalpy and heat capacity changes. Thus integration of the differential

power peaks (see Note 6) collected as in step 2 of Subheading 3.1. yields

∆q

i,app

and the nonspecific heats ∆q

i,ns

are obtained by integration of the peaks

collected as in step 4 of Subheading 3.1. Under conditions of total association

∆[P]

i,bound

equals ∆[P]

i,total

and is easily calculated from the known concentra-

tion of protein in the injection syringe, the volume of each injection and the

Fig. 3. (opposite page) Representative records of a Sox-5 titration into DNA duplex

(A) and enthalpies of binding of Sox-5 to a 12-bp DNA duplex as measured by ITC

(B). (A) Trace a: six injections of 550 µM Sox-5 into 58 µM of 16-bp DNA at 9°C (i.e.,

up to a fractional saturation of 0.45); trace b: control titration of Sox-5 into buffer at