Josh P. Roberts has an M.A. in the history and philosophy of science, and he also went through the Ph.D. program in molecular, cellular, developmental biology, and genetics at the University of Minnesota, with dissertation research in ocular immunology.
MicroRNAs (miRNAs), virtually unknown little more than a decade ago, affect virtually every aspect of our biology, from cancer to development. Although each of the more than 1,000 known miRNAs can potentially regulate more than 100 distinct mRNAs, few of these interactions have been validated. But that, of course, is key: miRNAs are useless in the absence of a target.
Scientists can detect those interactions computationally and experimentally. Here’s how.
In silico
The most common method of identifying miRNA targets relies on computer algorithms such as TargetScan, MiRanda and PicTar. These predict binding of the miRNA’s seed region—the most evolutionarily conserved segment of the miRNA, nucleotides 2 to 8, which is most frequently complementary to target sites in the mRNA 3’ untranslated region (3’-UTR)—principally by matching that region to 3’-UTRs found in databases such as Ensembl or that of the University of California, Santa Cruz. Each algorithm has its own distinct set of rules, such as where and how many mismatches are allowed; other sequences, such as introns, coding sequences, or the 5’-UTR, to be examined; expression data; and secondary structure.
Yet because such algorithms are based upon canonical rules, only a fraction of which are known, many of which are generic and probably none of which are absolute, they are only right about half the time, says Carol Kreader, R&D principle scientist at Sigma-Aldrich.
That being said, the highest scoring matches, meaning those with extensive base-pairing located at the best strategic locations, “have a reasonably good success rate in follow-ups,” says Gregory Goodall, head of the gene regulation laboratory at the Centre for Cancer Biology and professor of medicine at the University of Adelaide. “So the false positive rate for those is not so bad.”
Expression analysis
To find a few strongly responding targets with the hope that they will have especially interesting biological functions, Goodall recommends boosting or inhibiting the activity of the miRNA and then examining the resulting mRNA expression either by microarray analysis or RNA-seq. “Using arrays is quick to analyze and get your results; sequencing requires more extensive bioinformatics analysis, but it goes a little deeper.”
miRNA activity can be boosted by overexpressing the miRNA or mimics, either transiently or by retroviral transduction. There are various caveats associated with overexpression, including the generation of off-target effects both by overwhelming a balanced system and by affecting nonendogenous targets—such as those not normally even expressed in the same cell type or stage as the miRNA—or missing endogenous targets for the same reason.
Specific miRNA activity similarly can be knocked down by such means as antisense technologies and chemically engineered oligonucleotide mimics, techniques which are not mutually exclusive.
“In animals, miRNA binds to the transcript and prevents it from being translated. But in plants, the miRNA binds to the transcript and degrades it, completely breaks it up,” explains Chris Hebel, vice president of business development at LC Sciences. His company offers a service called “degradome sequencing,” which counts the number of fragments found in the RNA sample. Those that are over-represented are likely to be targets of miRNA that have been actively degraded.
Some researchers look for changes in protein expression—typically by mass spectrometry or gel-based approaches like SILAC (Stable Isotope Labeling by Amino acids in cell Culture) and 2D-DIGE (2D-Difference Gel Electrophoresis), respectively—to query for targets that may not have resulted in a detectable drop in mRNA levels by sequestration or degradation. Yet most responses—other than those in oocytes and early embryogenesis—show a stronger effect on mRNA than on protein expression, Goodall points out.
Expression analysis cannot distinguish between direct miRNA::mRNA interactions and secondary effects of such interactions. To do this, many researchers look for help from bioinformatics tools to help predict whether an interaction is likely to be a direct one.
Direct interactions
Another way to identify mRNA::miRNA interactions is to look for physical interactions experimentally, typically by co-immunoprecipitation (co-IP) using an antibody to Argonaut (AGO) or another protein component of the RNA-induced silencing complex (RISC).
A more recent take is to precede the co-IP by ultraviolet irradiation to crosslink the complex, and follow it with deep sequencing to identify the RNAs found—an approach called high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP), or alternatively, crosslinking immunoprecipitation (CLIP-seq). The crosslinking can be made up to 1,000-fold more efficient by pre-incubating the cells with a photoactivatable nucleoside such as 4-thiouridine (in a procedure termed photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation (PAR-CLIP)), which has the additional advantage of allowing the crosslinking site to be mapped.
Clontech’s MirTrap system accomplishes identifies direct interactions by making use of a FLAG-tagged dominant negative component of RISC, which expresses and assembles into a nonfunctioning complex. miRNA and its target “are fed into the RISC and will be stuck there,” explains Clontech technical support scientist Suvarna Sathe. “The complex can then be captured on beads, and the RNA isolated and sequenced.”
Sigma-Aldrich’s MISSION® Target ID cDNA library enables researchers to test for interaction via a functional assay. After transfection of the library and selection, cells are transfected with known miRNA. Cells containing cDNA with which the miRNA interacts will survive a second round of selection, allowing the targets to be sequenced. “An advantage is for labs that are used to transfecting and selecting. It’s a lot easier to do than other types of technology,” says Kreader.
Validation
Of course, any mRNA targets that are identified must be validated, a process very similar to that used to find the targets in the first place. “It’s pretty standard,” says Goodall. “Inhibit and/or overexpress the miRNA of interest and check for response at the mRNA and possibly protein level for the target of interest.”
To “nail it as a direct interaction,” Goodall continues, researchers can insert a luciferase reporter gene in the 3’-UTR and see how that construct responds to the miRNA. “Ideally—not everybody does it—one would mutate the sites, as well, to finally confirm that it’s direct.”
Many luciferase reporter vectors are on the market for the purpose of querying a putative target mRNA 3’-UTR, and off-the-shelf, pre-cloned vectors also are available.
Thus far, miRNA target identification and validation is for “labs that really want to put a bit of effort in to look at targets they’re interested in,” says Goodall, and the process is not yet routine. Perhaps as the tools and reagents for studying miRNA::mRNA interactions hit critical mass, it will become more so.