Gel Shift Assays–EMSA

See all protein-nucleic acid interaction products

The EMSA technique is based on the observation that protein–DNA complexes migrate more slowly than free linear DNA fragments when subjected to non-denaturing polyacrylamide or agarose gel electrophoresis. Because the rate of DNA migration is shifted or retarded when bound to protein, the assay is also referred to as a gel shift or gel retardation assay.

The ability to resolve protein–DNA complexes depends largely upon the stability of the complex during each step of the procedure. During electrophoresis, the protein–DNA complexes are quickly resolved from free DNA, providing a "snapshot" of the equilibrium between bound and free DNA in the original sample. The gel matrix provides a "caging" effect that helps to stabilize the interaction complexes: even if the components of the interaction complex dissociate, their localized concentrations remain high, promoting prompt reassociation. Additionally, the relatively low ionic strength of the electrophoresis buffer helps to stabilize transient interactions, permitting even labile complexes to be resolved and analyzed by this method.

Protein–DNA complexes formed on linear DNA fragments result in the characteristic retarded mobility in the gel. However, if circular DNA is used (e.g., mini-circles of 200–400 bp), the protein–DNA complex may actually migrate faster than the free DNA, similar to what is observed when supercoiled DNA is compared to nicked or linear plasmid DNA during electrophoresis. Gel shift assays are also good for resolving altered or bent DNA conformations that result from the binding of certain protein factors. Gel shift assays need not be limited to protein–DNA interactions. Protein–RNA and protein–peptide interactions have also been studied using the same electrophoretic principle.

Overview of the gel shift assay method. The gel shift assay consists of three key steps: (1) binding reactions, (2) electrophoresis, (3) probe detection. The order of component addition for the binding reaction is often critical. Completed binding reactions are best electrophoresed immediately to preserve potentially labile complexes for detection. This idealized example shows complete elimination of the protein–probe complex with the addition of a specific competitor or protein-specific antibody. However, only a reduction in intensity is observed rather than the complete elimination of bands.

Typically, linear DNA fragments containing the binding sequence(s) of interest are used in EMSAs. If the target DNA is short (20–50 bp) and well defined, complementary oligonucleotides bearing the specific sequence can be economically synthesized and annealed to form a duplex. For most applications and sequences, standard desalting of the oligonucleotides yields sufficient purity for use in EMSA. However, for sequences that form strong secondary structure or have long repeats, gel- or HPLC-purification may be required to ensure that the majority of the product is of the correct length and sequence for the experiment.

For analysis of protein–DNA interactions involving the formation of multi-protein complexes with multiple protein binding sites, longer DNA fragments are typically used. These larger fragments (100–500 bp) are usually a restriction fragment or a PCR product obtained from a plasmid containing the cloned target sequence. In this case, the desired DNA fragments must be gel-purified and verified by restriction digestion or sequencing. Standard precipitation or desalting will not remove the enzymes used to generate the fragments, and the remaining plasmid or template DNA will often can cause nonspecific bands or compete for binding during the assay.

If large quantities of DNA are used in EMSA reactions, the DNA bands can be visualized by ethidium bromide staining or other fluorescent DNA stains. However, the carrier and nonspecific competitor DNA used in most EMSA binding reactions can also become stained and cause high background signal. This detection method also requires a lot of DNA, where it is usually preferable (and economical) to use low concentrations of DNA in binding reactions in order to limit nonspecific binding. Thus, the more common approach is to label the probe before performing the experiment to permit its specific detection after electrophoresis. Traditionally, DNA probes have been radiolabeled with ³²P by incorporating an [γ-³²P]dNTP during a 3' fill-in reaction using Klenow fragment or by 5' end labeling using [γ-³²P]ATP and T4 polynucleotide kinase. Following electrophoresis, the gel is exposed to X-ray film to document the results.

Many labs have moved to alternative EMSA detection systems due to the expense and regulatory concerns associated with radioactivity. Because dNTPs are available modified with haptens (e.g., biotin and digoxigenin) or fluorescent dyes, there are numerous nonradioactive methods for performing EMSA. Fluorescent probes can be detected in-gel with the aid of an appropriate imaging system, but this method has not been very popular to date because expensive instruments are required and their sensitivity does not yet compare to radioactive probes.

Hapten-modified DNA probes can be visualized via secondary detection reagents such as streptavidin or anti-DIG antibodies in systems with enzymatic substrates similar to those used for western blotting. The only additional step required for this method is to transfer the separated protein and DNA samples onto an appropriate membrane support. When used with an appropriate substrate, the biotin-streptavidin system is very robust and can be optimized to achieve detection limits equal to those obtained with radioactive probes. It is important to note that performing such blotting assays with nucleotides requires the use of positively charged membranes in order to capture and immobilize the small nucleic acid fragments. Although this type of membrane presents certain technical challenges related to blocking and nonspecific binding, these can be overcome with optimized kits. For example,  the Thermo Scientific Pierce RNA 3' End Biotinylation Kit is optimized for labeling the 3-prime terminal end of RNA probes to facilitate their use as probes or targets in EMSA and other methods for studying protein–RNA interactions.

3’ end labeling of RNA. T4 RNA ligase catalyzes the ligation of the 3’-OH from the RNA with the biotinylated cytidine bisphosphate from the Pierce RNA 3' End Biotinylation Kit. This biotin label allows for detection with avadin proteins. Use of this kit results in the 3’ end labeling of RNA probes.

A nonspecific competitor is any irrelevant, unlabeled nucleic acid used as a blocking/quenching agent in the binding reaction to minimize the binding of nonspecific proteins to the labeled target DNA. The most common nonspecific competitors used in DNA gel shift assays are sonicated salmon sperm DNA and poly(dI•dC). These repetitive fragments, or polymers, provide an excess of nonspecific sites to adsorb proteins in crude lysates that will bind to any general DNA sequence. The order of addition of this reagent to the binding reaction is important in that, in order to maximize its effectiveness, the competitor DNA must be added to the reaction along with the extract before adding the labeled DNA target.

The specific competitor is an important control used to verify the specificity of a band resulting from protein binding to the labeled probe. The specific competitor typically has the identical sequence as the labeled probe or contains a known consensus binding sequence for the target protein. Generally, a 200-fold molar excess of unlabeled probe is sufficient to out-compete any specific interactions with the same labeled probe, eliminating or reducing any positive shift results. Conversely, in an optimized experimental system, the addition of excess mutant or unrelated sequence containing a low-affinity binding site will not compete with specific interactions, and the shifted band will be preserved. As with the nonspecific competitor, the unlabeled specific competitor must be added to the reaction before the labeled probe but after the nonspecific competitor has been incubated with the protein.

Nonspecific interactions can be affected by order of addition. Depending on the sequence of the probe and the presence of additional DNA binding proteins, the order of addition can influence nonspecific binding in gel shift assays. When probe for either Oct-1 or AP1 was added last, nonspecific bands were prevented. However, when HeLa extract was added last (after the probe), nonspecific bands occurred despite the inclusion of nonspecific competitor DNA (poly( dI•dC)) in each reaction. Nonspecific binding persisted with the addition of specific competitor when the protein extract was added to the reaction last. Nonspecific bands are indicated at the position of the arrow.

Factors that affect the strength and specificity of the protein–DNA interactions under study include the ionic strength and pH of the binding buffer, the presence of nonionic detergents, glycerol or carrier proteins (e.g., BSA), the presence/absence of divalent cations (e.g., Mg2+ or Zn2+), the concentration and type of competitor DNA present, and the temperature and time of the binding reaction. If a particular ion, pH or other molecule is critical to complex formation in the binding reaction, it may need to be included in the electrophoresis buffer to stabilize the interaction until the sample has entered the gel matrix.

Because there are many unique requirements for different nucleic acid binding proteins, there is no universal set of reaction conditions for EMSA assays. However, EMSA is a commonly used procedure with popular interaction targets, and it is often possible to find information about the binding reaction requirements from the research articles published on a protein of interest. If no information on prior work is available, conditions must be determined empirically. In this case, it is best to start with general assumptions based on the type of protein being studied or the consensus sequences for putative binding proteins within the probe being tested. For example, nuclear hormone receptors often require ligand to be activated. Also, because they are zinc-finger proteins, nuclear hormone proteins require zinc ions to be functional. Thus, the protein ligand and zinc ions are likely required in the binding reaction. Furthermore, strong chelators such as EDTA and EGTA may need to be avoided when setting up binding reactions for a novel nuclear hormone receptor.

Depending on the requirements for proteins binding to the probe, another important consideration is how the sample will be loaded in the gel. Usually, glycerol or Ficoll* (GE Healthcare) are added to binding reactions to increase the density of the sample so that it sinks into the well when loaded. If the loading agent is not added as part of the binding reaction, it can be added immediately before loading the samples. Bromophenol blue is typically added to stain so they can be seen during loading. Bromophenol blue also functions as a tracking dye, permitting the overall progress of electrophoresis to be monitored. It should be noted that bromophenol blue and other dyes can bind to proteins and inhibit the intended binding reaction. Therefore, it is important to test binding reactions and samples with and without the tracking dye if problems occur. In such cases, tracking dye can be loaded into lanes with no binding reactions or samples without protein.

Non-denaturing TBE-polyacrylamide gels or TAE-agarose gels are used to resolve protein–DNA complexes from free DNA. Traditionally, large format gels were used to resolve EMSA reactions, but mini-gels generally work very well for reactions with probes up to about 300 bp, depending on the protein complexes associated with them. Large probes with small binding proteins may require longer gels to adequately resolve any positive shift. The gel percentage required depends on the size of the target DNA as well as the size, number and charge of the protein(s) that bind. It is important that the protein–DNA complex enters the gel and does not remain trapped at the bottom of the loading well. Polyacrylamide gels in the range of 4–8% are typically used, although it is not uncommon for higher percentage gels to be used with smaller probes and complexes. Agarose gels (0.7–1.2%) can be used to resolve very large complexes, as is the case with E. coli RNA polymerase (~460 kDa).

Before a binding reaction is performed, gels are pre-run at a constant voltage until the current no longer varies with time (45–90 minutes) in a cold room or using other methods to keep the gel cool. The primary reasons for pre-running gels is to remove all traces of ammonium persulfate (used to polymerize polyacrylamide gels), to distribute/equilibrate any special stabilizing factors or ions that were added to the electrophoresis buffer, and to ensure a constant gel temperature. After loading samples onto the gel, it is important to minimize the electrophoretic dead time required for the free DNA to enter the gel matrix, especially when analyzing labile complexes. Load the samples in the gel gently and quickly and re-apply the voltage. Separation of the samples will take 30–90 minutes depending on the size of the probe and complexes. The use of a tracking dye such as bromophenol blue is useful to monitor migration of the free probe. Bromophenol blue will migrate to 10–20 millimeters behind a double-stranded DNA probe of approximately 30 bp. It is usually desirable to run the free probe to near, but not off, the bottom of the gel.  To obtain the best result, it is important to choose the correct gel percentage, buffer system, gel format and thickness. We offer a wide variety of Invitrogen Novex precast gels. The following figure was designed to aid in gel selection and to help optimize experiments. The table shows migration patterns of nucleic acids run on varying Novex gel formats. 

Gel migration chart. This chart shows patterns of nucleic acid separation in different types and percentage Novex polyacrylamide gels.

While there are characteristic shifts caused by specific protein(s) binding to target DNA, a relative change in mobility does not identify the bound protein in a shifted complex. Identification of the protein bound to the probe is frequently accomplished by including an antibody that is specific for the putative DNA-binding in the binding reaction. If the protein of interest binds to the target DNA, the antibody will bind to that protein–DNA complex, further decreasing its mobility relative to unbound DNA in what is called a "supershift". In some cases, the antibody may disrupt the protein–DNA interaction, resulting in loss of the characteristic shift but not causing a supershift. The disappearance of a shifted band and absence of a supershift is a negative result, but it can support the identification of the protein of interest when used with the proper controls. Supershift reactions need not be limited to antibodies, but could include other secondary or indirectly bound proteins as well.

Performing supershift experiments creates additional considerations for setting up binding reactions. Antibodies and binding partners may require a specific order of addition. For example, some antibodies may block binding of a protein to DNA if added before the complex is formed. Antibodies and other proteins being used to create a supershift may not function in a binding buffer that is optimal for a protein–DNA shift complex. Many nuclear proteins function well for shift complexes under reducing conditions. Therefore, concentrations of DTT at 1 mM and higher are often added to the EMSA binding buffer. Yet, many antibodies become reduced and loose functionality in reducing conditions. When possible, choose antibodies known to function in gel shift assays. Additionally, antibodies that are effective for western blotting but not for ELISA (i.e., detect only denatured proteins) are not likely to work for supershift assays.

An alternative identification process would be to perform a combination "shift–western blot." This involves transferring the resolved protein–DNA complexes to stacked nitrocellulose and anion-exchange membranes. Protein captured on the nitrocellulose membrane can be probed with a specific antibody (western blot) while autoradiography or chemiluminescent techniques can detect the DNA captured on the anion-exchange membrane.

1. Hendrickson W (1985) BioTechniques 3:346–354.
2. Revzin A (1989) BioTechniques 7:346–354.
3. Fried M, Crothers DM (1981) Nucleic Acids Res 9:6505–6525.
4. Garner MM, Revzin A (1981) Nucleic Acids Res 9:3047–3059.
5. Kneale GG, editor. Protocols. Totowa (NJ): Humana Press Inc.
6. Fried MG, Crothers DM (1984) J Mol Biol 172:263–282.
7. Fried MG (1989) Electrophoresis 10:366–376.