Editorial Article
Monday June 29, 2009
by Jeffrey M. Perkel
For all its information, DNA is biologically inert. The cell must work to propagate its genetic information, using RNA polymerases to transcribe genes into RNA, and ribosomes to make protein. Naturally, both of these processes are heavily regulated. But to a large extent, gene expression is controlled at the level of transcription, and the agents of that control are protein-DNA interactions.
Proteins called transcription factors latch on to DNA regulatory regions and tweak the expression of nearby genes by, for instance, loosening the genetic material's grip on the nucleosome spools around which it is wrapped, or by boosting the efficiency with which transcriptional machinery is recruited to gene promoters. Researchers intent on understanding these processes—that is, of analyzing protein-nucleic acid interactions—have a number of tools at their disposal, from low-tech old-school methods to cutting-edge genetic instrumentation. Which you choose depends on your throughput requirements, available reagents, existing knowledge, and most importantly, your question.
One of the simplest methods for mapping protein-DNA interactions, the electrophoretic mobility shift assay (EMSA), works exactly as its name implies: a small fragment of radiolabeled double-stranded DNA (for instance, a gene regulatory element) is incubated with a protein sample, such as a nuclear extract from stimulated cells. If that sample contains a protein capable of binding to the labeled probe, it will retard the DNA's migration through a gel (relative to a control reaction)—hence, the experiment's more common name, the "gel shift assay." (Some researchers add an additional electrophoretic shift, or supershift, by adding antibodies to candidate proteins to the reaction.)
Another old-school approach is the so-called Southwestern blot. Useful in cases where the size or identity of the protein binding to a DNA sequence is unclear, Southwestern blots combine the protein gel blotting aspect of a Western with the DNA-based probing of a Southern. Denatured proteins are resolved on an SDS-PAGE gel and transferred to a nitrocellulose filter. Then, the researcher renatures the proteins, probes the blot with radiolabeled DNA containing the protein-binding site, and washes away any unbound nucleic acid. By exposing the labeled filter to film, he can determine the size of the DNA-binding protein by comparison to a protein size ladder.
The problem with both gel shifts and Southwestern blots is their throughput and reliance on radiolabeled materials; many researchers are reluctant to use radioisotopes if they can avoid it. Colorimetric and chemiluminescent variants of these methods exist, but so do alternate approaches. One is the TransAM™ product line from Active Motif.
A cross between a gel shift and an ELISA, TransAM is a 96-well plate-based colorimetric (or sometimes chemiluminescent) assay in which each well contains a consensus sequence for a given transcription factor, such as NF-kappaB family members. The user adds a protein extract, incubates, washes, and then probes, first with antibodies to specific protein members of that family (for instance, p50, p52, p65, RelB, and c-Rel in the case of NF-kappaB), and then with an enzyme-coupled secondary antibody. If any of these proteins bind to the plate, the reaction generates a signal whose intensity is proportional to protein abundance.
Sensitive, quantitative, and reproducible, "It is an alternative procedure, a superior technology, to the gel shift assay," says Kyle Hondorp, product manager at Active Motif. "But, it runs along the same principle as a gel shift."
However, TransAM is only intended for cases in which you are interested in determining whether a given protein is present and/or activated; except in the specific case of NF-kappaB, it cannot be used to probe custom DNA sequences.
Sometimes researchers have a DNA site of interest, but don't know which proteins (if any) bind to it. One approach to addressing this problem is a genetic assay, such as Clontech Laboratories' Matchmaker Gold Yeast One-hybrid Library Screening System. A yeast one-hybrid assay relies on two DNA constructs. The "bait" vector is a reporter in which the DNA binding site of interest drives expression of a selectable marker (in this case, a gene encoding resistance to the antifungal agent, Aureobasidin A). The "prey" is a library of plasmids, each containing one potential DNA-binding protein fused to the Gal4 transactivation domain. By transfecting the prey library into a bait yeast strain and selecting for aureobasidin A resistance, researchers can identify those colonies containing a protein capable of binding the bait, and then sequence those clones to identify the prey.
According to Baz Smith, product manager at Clontech, Aureobasidin A selection is a new and highly effective addition to the Matchmaker system (it was introduced earlier this year). Previous versions, he explains, relied on nutritional selection. But nutritional markers don't kill negative cells; they simply slow their growth. As a result, "you can get a high background if you leave the plates in the incubator too long, or plate cells at high density," Smith explains. The traditional solution to this problem was to use multiple selectable markers; the new system, in contrast, has only one. "With Aureobasidin A, you get a really clean plate," Smith says. "With the negative control, you see zero colonies."
Once you have identified a particular DNA-binding protein of interest, the tool of choice for analyzing its interaction with DNA in vivo (as opposed to in vitro, as with gel shift assays) is Chromatin ImmunoPrecipitation (ChIP).
ChIP is really just the first of a two-part process, a DNA enrichment step that precedes detection. The enrichment step works like this: Normally transient protein-DNA interactions in a cell are locked in place using a cross-linker. The genomic DNA (now with associated proteins attached) is then collected, sheared into small fragments, and subjected to immunoprecipitation using an antibody against a particular protein of interest (such as a specific transcription factor, modified histone, or RNA polymerase). Finally the immunoprecipitated protein-DNA complexes are dissociated by reversing the cross-link.
With enriched DNA in hand, researchers can analyze what they’ve pulled down in one of three ways. The simplest, and lowest throughput approach involves PCR. Using primers flanking specific regions of interest, researchers probe the immunoprecipitated DNA to determine whether any of a small number of sites is enriched in one sample (such as drug-treated cells) relative to another, suggesting, for instance, that treatment induces enhanced protein binding to those regions.
For more comprehensive analysis, the DNA can instead be applied to a microarray comprising many or all potential binding sites in the genome—a strategy called ChIP-on-chip. Affymetrix, Agilent Technologies, Illumina, and Roche NimbleGen all offer off-the-shelf DNA microarrays that can be used for this purpose, the primary advantage of which is that a very large number of sites can be interrogated simultaneously (up to 244,000 in the case of Agilent's arrays, with a million probe array on the way, and 2.1 million in the case of Roche NimbleGen).
Custom arrays can simplify higher-throughput, but require more focused, follow-up experiments. For instance, using Agilent's eArray tool, researchers can build DNA chips containing eight arrays each with about 15,000 probes per array. The company can synthesize those arrays using either user-provided probe sequences, Agilent's existing library of probes, or by asking the company to design a probe set for a specific sequence (such as a model organism or particular genomic region).
Finally, the ChIP-enriched DNA can be sequenced, an approach called ChIP-Seq. In this case, the tally of sequence reads per genome location acts as a map of protein binding under a given set of conditions, and its biggest advantage, says Chris Streck, product manager at Illumina (which offers ChIP-Seq kits and software for its Genome Analyzer sequencer system), is that it sidesteps the fact that microarrays cannot detect interactions in regions for which they have no probes.
ChIP-Seq "provides a truly global view of the precise binding location on a genome-wide scale," Streck says. And, he adds, it is among the most widely adopted applications of the Genome Analyzer to date.
Though initially developed for protein-DNA interactions, ChIP is evolving as a growing number of researchers devise ways to expand the method's reach. One example is Active Motif's new Re-ChIP-IT assay, which uses two back-to-back ChIP pull-downs to identify DNA fragments to which multiple factors are bound. Other investigators are applying ChIP to protein-RNA interactions, for instance, to study interactions of splicing factors and mRNAs and RNAi machinery with siRNAs and miRNAs.
"This is very cutting edge," says Sallie Cassel, director of antibodies and immunoassays at Millipore (which offers a wide range of antibodies and kits for ChIP), "looking at interactions of RNA with protein and trying to understand the transcripts that are bound."
Millipore will be launching a kit later this summer in support of what it calls RNA immunoprecipitation, or RIP, Cassel says. Similarly, Illumina's users have developed an RNA-based ChIP-Seq variant called cross-linked immunoprecipitation (CLIP-Seq), says Streck.
By all accounts, ChIP and its variants pose only one significant problem: the antibodies used to perform the assay. Obviously, you cannot do ChIP without an antibody, but not just any antibody will do, says Robert Brazas, epigenetics product manager at Roche NimbleGen.
"The antibody is critical," Brazas says. "If you don't get a good immunoprecipitation, if you get a lot of non-specific binding, you will get a lot of background noise and it will be difficult to identify the DNA-binding sites." Your best bet is to use antibodies that have been validated for ChIP. Alternatively, says Alicia Burt, director of microarray applications at Agilent Technologies, use an antibody that recognizes the native (as opposed to denatured) protein.
In other words, if you don't have a ChIP-validated antibody, look for one that works in immunoprecipitation or immunohistochemistry. When in doubt, adds Brazas, you can always pre-test using ChIP-PCR.