Methods for studying protein-DNA interactions

BIOC/MCB 568 -- Fall 2010
John W. Little--University of Arizona

BIOC/MCB568 Home Page


There are several methods in common use for analyzing specific protein-DNA interactions. It is important to understand these methods, and to be aware of the capabilities and limitations of each method. 

DNase I Footprinting

Gel mobility shift assay

Other methods

Nitrocellulose filter binding assay

Genetic analysis

X-ray crystallography

click for larger version
click for larger version

Description of method
Description of method



Table comparing the various methods

DNase I footprinting:



Consider first the left side of the figure, with no protein.

A restriction fragment containing a specific binding site is labeled at one end, usually with 32P.
A restriction fragment is used because all the molecules end at exactly the same place. Its source is almost always from recombinant DNA.
Molecules are treated lightly with DNase I, which makes single-strand breaks (nicks) in the DNA. A small amount of enzyme is used so that there is an average of <1 nick/strand.
With this mild digestion, many molecules are not cut at all, and most are cut only once. Different molecules are cut in different places, so that one gets a family of labeled fragments ending at many positions throughout the DNA.
The reaction is stopped, the DNA is denatured, and the mixture is run on a denaturing polyacrylamide gel.
These gels are capable of resolving molecules differing in length by a single nucleotide. They keep the DNA denatured because they contain 8 M urea and are run hot.
The distribution of radioactivity is visualized by autoradiography or by the use of a Phosphorimager.
Although many different fragments are present in the mixture, only the labeled ones show up in this analysis; fragments derived from the bottom strand are not seen at all, for example.
The result is a ladder of bands, representing the various sites at which DNase I cuts.
DNase I has a small degree of specificity, so that it does not cut uniformly at all sites.

Now consider what happens if a specific DNA-binding protein is present:

The procedure is the same, except that the protein is incubated with the DNA for a period of time to allow binding.
Binding is not instantaneous. Particularly if you want to measure equilibrium constants, you need to give it enough time to reach equilibrium.
All the subsequent steps are the same.
Where the protein is bound, the DNase I can usually not attack the DNA.
Result: Where the protein was bound, there is a gap in the ladder -- the footprint.
This technique provides a visually pleasing demonstration of where the protein binds.

Back to Diagrams of Methods

An example of a footprinting gel is in Little et al (1999) [abstract, available anywhere; HTML file, available from U of A computers]. Look at Fig. 2. This shows the binding of two phage lambda proteins to each of four different templates. Consider first the wild-type (upper left panel). Incubations for the various lanes contained no protein (lane in center indicated by "-"), or differing amounts of one of two proteins, CI or Cro; increasing amounts are indicated by wedges. The template contains three binding sites (indicated by oR1, oR2 and oR3) for these proteins. In the various lanes, differing affinities for the three sites can be seen as differences in the amounts of protein needed to give protection. For instance, Cro binds tightly to oR3 and weakly to oR1 and oR2; CI binds tightly (and cooperatively) to oR1 and oR2, weakly to oR3.

Now consider the other three panels. In these cases, the templates were changed (by site-directed mutagenesis) so that the two outside sites (oR1 and oR3) are the same. At least for Cro, now the protein binds about equally well to the two flanking sites. Comparisons among panels shows that Cro's affinity for these sites differs on different templates.

Back to Diagrams of Methods

Refinements to footprinting methodology 

1. Other agents that break or damage DNA can be used in addition to DNase I.

Dimethyl sulfate (DMS) methylates G residues, creating an adduct that can be broken chemically. Typically, complexes are formed, then treated with DMS, then the reaction is stopped. The position of the breaks is then determined by primer extension. This approach has several advantages. First, it can be carried out under conditions (e.g., absence of magnesium) in which DNase I is inactive. Second, it can be done on covalently closed, supercoiled DNA, which often binds proteins differently than linear DNA.

Potassium permanganate (KMnO4) is specific for thymine residues in single-stranded regions of DNA. Again, it is usually used in conjunction with primer extension. This can be used, for example, in transcription complexes to determine the location of RNA polymerase and the structure of the transcription bubble.

2. When used properly, this methodology can be used to measure dissociation constants. One carries out a series of binding reactions over a range of protein concentrations, and determines the fraction of the binding site that is occupied at each protein concentration. The dissociation constant is equal to the concentration of protein when 50% of the binding site is occupied, as we now discuss. To do this, the DNA concentration in the assay must be far lower than the dissociation constant for the interaction. This constraint arises directly from the definition of the dissociation constant:

Kd = [R] [O] / [R:O]

 where [R] and [O] are the concentrations of the free DNA-binding protein and binding site, respectively, and [R:O] is the concentration of the complex. At 50% occupancy, [O] = [RO], and Kd = [R]. However, we only know [R] if the total amount of DNA is so small that it does not significantly deplete the pool of free R; in other words, [R:O]<<[R].

An example of this approach is in Liu and Little (1998),[ abstract available anywhere, HTML file from U of A computers]. Look at Fig. 4, panel e for the curve given by a single site. The remaining data here are used to evaluate cooperative binding of the protein to three adjacent sites.

Back to Diagrams of Methods

 Gel Mobility Assay 

This assay has several different names: Gel mobility assay, gel shift assay, EMSA (electrophoretic mobility shift assay), gel retardation assay. All these terms describe the same technique.



Prepare a labeled restriction fragment containing one or more binding sites for a specific DNA-binding protein.
The fragment can be labeled uniformly throughout, at one end, or at both ends. A restriction fragment is uniform in size and forms a single band in the gel.
Set up a series of incubations containing your fragment and either no protein or increasing amounts of the protein. Incubate to allow binding to occur.
A series of samples with graded amounts of protein will often give more information, but is not necessary.
Apply your samples to a polyacrylamide gel. Run the gel.
Agarose gels can also be used for large fragments. The gel buffer should contain a low concentration of salt to stabilize protein-DNA interactions.
Visualize the distribution of radioactivity (see Footprinting).

Result: DNA molecules to which proteins bind move more slowly in the gel and are retarded relative to the sample with no protein.
At intermediate protein concentrations of protein, only a portion of the DNA molecules are in complexes, so that several bands are seen. In the example shown in the diagram, three different bands can appear, depending on the concentration of protein.


An example of gel mobility assay (includes a supershift assay (see below) in Fig. 5) is given in Dennisova et al. (2000). [abstract , HTML file], Figs. 4 and 5 .

 Back to Diagrams of Methods

Refinements to the gel mobility assay:

 Several different refinements to the basic methodology provide additional types of information.

First, this assay can be carried out in the presence of complex mixtures of proteins, either in crude extracts or with partially purified components.

Second, a so-called supershift assay can be used. When one uses a complex mixture of proteins, it's not clear which one is binding to the DNA. When an antibody is available that interacts with a protein of interest (call it the antigen), one can ask whether a particular shifted band contains the antigen by having a second incubation that includes the antibody. If the complex shifts further up in the gel (the "supershift"), this is evidence that the antigen was present in the initial complex; the reason it shifts further up is that now the complex also contains the antibody. An example is cited above.

Third, one can do competition experiments, to ask whether the addition of unlabeled DNA can compete with the labeled DNA for the protein. The unlabeled DNA is often in the form of synthetic oligonucleotides. This assay is done when there is a limited amount of protein, so that it can be saturated. Competition experiments can be done for several reasons:

a. If you want to ask whether a particular protein is in the complex, you can add a known high-affinity site for that protein. If the protein is in the complex, it should be competed away; if not, then addition of the competitor has no effect.

b. If you want to determine which of the bases in the DNA are important, you can add competitor DNA with changes in particular bases. If the base is important, then the mutant template will have less effect than the same amount of the wild-type template. Eventually the mutant template should compete; the relative amount that one has to add can give a good measure of relative affinities for mutant and wild-type template.

Back to Diagrams of Methods


Nitrocellulose filter-binding assay

This assay is no longer used widely, but it is rapid and simple, and can give a lot of information. It would be used for relatively detailed analysis of a particular protein-DNA interaction, for example.



Incubate labeled DNA with protein
Allow enough time to allow the system to reach equilibrium
Filter the mixture through a filter disk made of nitrocellulose
Proteins bind to nitrocellulose, but DNA does not. Any DNA that is retained on the filter is there because it is interacting with the protein.
Dry the filters and count

Back to Diagrams of methods

Genetic analysis

Isolation of mutants in the DNA binding site help to identify which residues in the binding site are important. Examples of such sites are up-promoter and down-promoter mutants, which affect binding of RNA polymerase (or subsequent steps in initiation), and operator-constitutive mutants in repressor binding sites.

Mutants can be identified in a DNA-binding protein that affect its ability to bind to DNA; however, it is difficult to be certain how these mutants affect the interaction, except when high-resolution structural information is available.

X-ray crystallography

X-ray crystallography of protein-DNA complexes provides the most detailed look at these interactions. This is the source of most of the PDB files used in our work with RasMol (here's a link to the class page on RasMol ). However, it is important to realize that structural data by themselves are not enough. Important contacts are also identified by the combination of genetic evidence that particular bases in the binding site are crucial and biochemical measurements of the changes in affinity. Such evidence allows us to look in detail at the important interactions in the structure.

Back to Diagrams of methods


Comparison of the various methods: 

Each of these methods has its advantages and disadvantages. These are listed here, though the comparison is not exhaustive.

One important use of many of these methods is to determine equilibrium constants (defined, for example, in the section on footprinting above). For the method to give an accurate value, the components must be at equilibrium under the conditions of the assay. For the gel mobility assay, this is rather problematic. The components may be at equilibrium when they are loaded onto the gel, but once they enter the gel one would expect them to re-equilibrate. What one sees depends on the conditions. Weak interactions often lead to dissociation during the gel run, giving a smeary shifted band or a smear extending above the free DNA. In other cases, the species move as well-behaved entities, without interconversion. In such cases, workers often assume that the values derived are equilibrium values, but usually don't validate this assumption. I usually take such numbers with a grain of salt. In any case, it's not well understood why complexes don't dissociate; one reason may be that the gels are run under low ionic-strength conditions, which favor strong DNA-protein interactions.



Disadvantages and limitations

1. It is an equilibrium method
2. Other agents besides DNase I can be used
3. It localizes the binding site to within a few bp.
1. Tedious, technically demanding

2. DNase I can only be used under conditions that support its activity (presence of Mg++ and Ca++ ions).

Gel mobility assay

1. Technically easy
2. Gives some information about the mass of the protein bound
3. Can reveal multiple complexes
4. Can be used to measure DNA bending
1. Not really an equilibrium method, but can give info about relative affinities
2. Does not localize binding site

Filter-binding assay

1. Quick and easy
2. Can measure on- and off-rates
1. Does not localize binding site

Genetic analysis

1. Helps identify important bases in the binding site
2. With site-directed mutagenesis, can test models based on x-ray structures.
Analysis of protein is hard to interpret except in conjunction with x-ray crystallographic data

X-ray crystallography

1. Can give structural information at atomic resolution
2. Can give extremely explicit testable models for specificity of interaction.
1. Not for everyone! Requires a lab dedicated to this approach.

2. Not all DNA-binding proteins can be solved; many are "floppy". One solution is to use fragments or domains of protein.

Back to Diagrams of methods

BIOC/MCB 568 -- University of Arizona
Last modified August 18, 2010
© 2010. All rights reserved.