Featured Article
Monday January 18, 2010
by Jeffrey M. Perkel
There's no denying that the small, non-coding regulatory molecules called microRNAs are hot. Implicated everywhere from floral development to oncogenesis, microRNAs account for 6500 references in PubMed, all but a few hundred of them published after 2004. miRBase, the repository of microRNA sequences, lists 10,883 sequences from some 115 organisms in its most recent release (v. 14, released September 2009)—up from 218 in 2002.
Needless to say, this quick pace presents something of a logistical problem for researchers. Each time miRBase updates—which happens several times per year—it not only swells with new RNAs, older sequences are also modified, and sometimes deleted, meaning existing data must constantly be reinterpreted. For scientists not yet steeped in the literature, the field can appear both alluring and treacherous.
For those undaunted by the challenge, vendors offer a healthy supply of microRNA-oriented reagents—microarrays, quantitative PCR assays, and so on. But an ever-growing segment of the research population is choosing another option: outsourcing.
"The service [side of our] business is growing far beyond our expectations," says Peer Staehler, chief scientific officer and vice president of marketing at febit, a German tool and service provider. "In 2009, we had almost twice as much revenue in services than we planned. We are overwhelmed by the response." Staehler says the company has taken on several hundred service jobs since 2008.
Chris Harbert, director of marketing for North American life sciences at Exiqon, a contract service provider based in both Boston and Denmark, cites two primary reasons researchers choose to outsource work to an organization like his. "First, everywhere scientists are being asked to do more with less," he says. Journal and grant referees are demanding ever more detailed analyses prior to acceptance or funding, for instance—demands that researchers must accommodate without additional personnel, space, or very often, funding.
Second, says Harbert, "it may not be the most efficient use of time for a scientist who is well versed in a particular pathway to spend the time to become an expert in processing arrays or in situ hybridization probes or qPCR."
That's where Exiqon comes in. "We look at it as a partnership," Harbert explains, "where the clients bring expertise in their particular biological system, and we bring expertise in applying methods to that biology, and together we get something greater than either could create alone."
It's a common theme. Across the globe, tool providers and service companies offer a diversity of microRNA-focused services for customers who just want the data, and have neither the time nor the expertise to collect it (or analyze it) themselves. Options run the gamut from RNA extraction and expression profiling, to sequencing and bioinformatics analysis.
In Exiqon's case, the menu includes RNA extraction, locked nucleic acid (LNA)-based microarrays (for expression profiling), qPCR (for validation), and ISH services, which generally return data within one to six weeks.
LC Sciences, based in Houston, also offers qPCR and miRNA microarray services—the latter based on the company's flexible microfluidic microarray platform, which can be updated within a day of a new miRBase release, according to Christoph Eicken, Head of Technical Services for Microarrays. In addition, the firm also offers next-generation sequencing using Illumina's Genome Analyzer. Costs range from $40 per sample for qPCR to about $2,500 per sample for sequencing, says Eicken, while turnaround time runs from one to eight weeks (for sequencing).
Eicken says LC Sciences tends to attract two types of customers. The first are researchers involved in either microRNA discovery or studies of non-model organisms (such as the tomato, cow, or chicken). "If you look at manufacturers of regular spotted arrays, they cover human, mouse, rat, and that's it," says Eicken. "We cover all species; there are 115 species and 12 viruses in miRBase, and you can put any of them on an array."
The second group: people just starting in the miRNA field. Such clients, he says, tend to be reluctant to use university core facilities for miRNA work, "because core labs give just the raw data." Given the pace of developments in the field, and the complexity of analyzing microarray or next-gen sequencing data overall, miRNA data analysis is no simple task. But with LC Sciences, he says, "You tell us how to group the samples, we do the statistical tests and give you the analyzed data as well as raw data files. That's a big advantage."
In a crowded marketplace such as this, each provider tries to position itself uniquely.
ArrayStar, based in Rockville, MD, distinguishes itself with a collection of "novel" arrays that extend beyond miRNAs and into the broader area of non-coding RNAs, says Director of Business Development Jenny Zhu. "This can be very convenient," she says. The company offers arrays to profile the expression of both long non-coding RNAs and piwi RNAs (piRNA), and to assess DNA methylation up- and downstream of miRNA genes. Turnaround time ranges from two to three weeks for standard arrays, to six weeks for methylation services, Zhu says.
For febit, the distinguishing characteristic is data analysis. Like LC Sciences, febit's service menu includes flexible customizable microarrays, qPCR, and next-gen sequencing (using Life Technologies' SOLiD sequencers). But what sets the company apart, Staehler says, is its bioinformatics service, which he calls "the most important part of our service now."
"We have a very advanced biostatistics solution," Staehler says. "So we provide not just data, but also really meaningful biomarker results. That is something our competitors don't do." According to Staehler, the company "identified biomarkers in 13 major diseases in 2009 alone. That is obviously value added over a simple microRNA analysis."
In some cases, Staehler says, final reports can run to 200 pages.
Indeed, data analysis can be so challenging, some companies specialize in it. One such company is InteRNA Genomics, located in the Netherlands.
"In most cases, people have all this small RNA sequencing data, but they cannot analyze it," says Edwin Cuppen, professor of genome biology at Utrecht University in the Netherlands, who also is a company advisor. "It requires quite a bit of dedicated know-how to do that properly. That is where InteRNA Genomics steps in, as a dedicated bioinformatics provider."
The difficulty, Cuppen explains, is that because microRNAs are very short (about 20-25 bases long), they can be difficult to unambiguously map back to the genome. Plus, given their length, determining whether potentially novel sequences actually meet the criteria for true miRNAs is non-trivial.
"Figuring out if something is forming a stable hairpin requires thermodynamic modeling," he says. "If you have to do that for millions of reads, it requires a large computer cluster to do that." InteRNA Genomics has a 250-CPU cluster (as well as manual curators) dedicated to this problem, and can complete sequence analyses in under two weeks, Cuppen says. The cost ranges from 2000 to 6000 euros for from four to 20 samples.
Jens Schwamborn, junior group leader at the Westfälische Wilhelms-Universität Münster (University of Muenster), in Germany, has taken advantage of that service—albeit via a non-commercial arrangement. While working at the Austrian Academy of Sciences, he collaborated with InteRNA Genomics' Co-Founder, Eugene Berezikov, to characterize microRNA sequences associated with a complex containing the proteins Argonaute and TRIM32.
"What they did was making sense of these [sequence] data," he says—cataloging the RNAs, determining which were enriched in certain samples, running quality control checks to determine whether a sequence actually was an miRNA, and so on.
But the sequences themselves came from a Munich-based contract provider called Vertis Biotechnology.
"We had no experience with that [next-gen sequencing]," Schwamborn explains. "We could have done this ourselves, [but] it would have taken time, and you need these deep sequencing machines, and we didn't have one. [Outsourcing] was a much faster and cheaper approach."
Schwamborn didn't select Vertis at random, of course; a colleague recommended them. "I would go to a company where I know somebody has done work with them and it worked," he says.
There are other variables you can consider, as well. Eicken advises making sure the company's reagents reflect the most current data (that is, the most recent miRBase release). On a more practical note, he says, "Regarding sample prep: Whatever method you pick, whether it's column-based or TRIzol, stick with it throughout the experiment, because the miRNA profile might vary depending on which kit you are using."
Staehler suggests looking for a local provider, if possible, whereas Cuppen encourages planning ahead. "Working with a partner who knows the ins and outs of the sequencing and the analysis, who can talk about follow-up studies, I think that would be my advice: To think beyond the first step of data generation," Cuppen says.
Finally, consider such variables as cost, turnaround time, follow-up support, and especially, publication record. Says Eicken, whose LC Sciences lists some 130 papers on its web site, "That's the best proof."