Jeffrey Perkel has been a scientific writer and editor since 2000. He holds a PhD in Cell and Molecular Biology from the University of Pennsylvania, and did postdoctoral work at the University of Pennsylvania and at Harvard Medical School.
RNA-seq. As a method, it sounds straightforward enough, but appearances can be deceiving, And much rides on the details.
With RNA sequencing, there is no “one” method because there is no single desired outcome, says William Cresko, professor of biology at the University of Oregon, whose lab created the online resource RNA-seqlopedia to serve as a kind of handbook on RNA-seq.
Some researchers, for instance, use RNA-seq as a next-generation sequencing (next-gen) alternative to DNA microarrays, for identifying differences in gene-expression levels between samples or conditions. Others use it to build a reference transcriptome, as an alternative to genome sequencing or at least to help annotate a newly sequenced reference. Some work with total RNA, others prefer to enrich specific subpopulations, such as mRNA or miRNA. And though many work with fresh frozen tissue or abundant cultured cells, which yield high-quality nucleic acids, others are forced to harvest RNA from archived blocks of FFPE biopsies, which typically yield small amounts of highly degraded material.
“The devil is in the details of the exact application, and there’s a lot of branch points along the way,” Cresko says. “And it’s important to be aware of those, as people go on.”
Purification
One such branch point involves the process of RNA purification and cDNA synthesis. Here, says Curtis Knox, global strategic marketing manager for NGS at Promega, quality is key. “Next-gen sequencing really is the epitome of ‘garbage in, garbage out,’” he says. Promega, for instance, sells both manual and automated kits for RNA purification from a variety of sample types and at a variety of throughputs, using both bead- and spin column-based methods.
Thermo Fisher Scientific also sells a variety of tools for purification of both total RNA and RNA subpopulations, says senior product manager Min Li. These include the automation-ready MagMax™ mirVana™ TotalRNA Isolation Kit and PureLink(r) spin column-based kits, Dynabeads® mRNA Direct Kits, and the mirVana miRNA Isolation Kit. Total RNA preparations can subsequently be depleted of ribosomal RNA using Thermo’s RiboMinus™ or KAPA Biosystem’s RiboErase technology (available from Roche Diagnostics), or enriched for mRNA using dedicated isolation tools, such as Thermo's Dynabeads mRNA DIRECT Kit.
Theoretically, both mRNA enrichment and rRNA depletion achieve the same goal: eliminating the ribosomal RNA sequences that constitute the bulk of total RNA and thus exposing rarer transcripts and isoforms, says Cresko.
But mRNA enrichment specifically captures only polyadenylated transcripts, whereas rRNA depletion retains everything but rRNA, including, for instance, microRNAs and long noncoding RNAs.
“RiboMinus approaches in general allow you to analyze not only protein-coding but also functional noncoding RNAs,” Cresko adds.
Target enrichment
Another approach to squeezing the most out of RNA-seq runs is sequence enrichment, for instance using Thermo’s PCR-based AmpliSeq or Roche Diagnostics’ hybridization-based SeqCap technologies, both of which provide off-the-shelf and custom panels to enrich transcripts of interest. “The transcriptome is very complex,” explains Casey Matthews, marketing manager at Roche Diagnostics. By selecting only the transcripts they want, researchers can “take a deeper dive” into select regions, focusing sequencing bandwidth and increasing sensitivity to detect rare events, Matthews says.
According to Cresko, read length and read depth represent yet another logical “branch point.” For instance, researchers interested in comparing gene expression across samples tend to require large numbers of short reads and multiple biological replicates. But those interested in building a reference transcriptome or studying splice variants may be better served with a long-read technology, such as Pacific Biosciences’, applied across multiple tissue types, to fully capture transcript structure.
Single cell analysis
Another option is downsizing RNA-seq to the single-cell level—an option that allows researchers to explore cell-to-cell variation.
According to Roger Lasken, director of single-cell genomics at the J. Craig Venter Institute, single-cell RNA-seq poses fresh challenges, especially regarding the isolation and amplification of full-length cellular mRNA. “It’s not easy to know how accurate you are,” he explains, “because cells vary.” Thus, it is difficult to differentiate biological variation from amplification bias and sample degradation.
Researchers have developed methods for single-cell RNA-seq, of course, including SMART-seq2. Recently, Lasken, with Professor Fred Gage of the Salk Institute in La Jolla, Calif., developed a method for sequencing RNA from individual nuclei rather than whole cells—a strategy that enables analysis of certain complex samples, particularly brain tissue. As it turns out, though the nucleus contains less RNA than the cytoplasm, relative transcript abundance appears relatively constant in both compartments. “You can isolate nuclei without altering the transcriptome, whereas proteolytic dissociation of whole cells from tissue is known to alter gene expression,” according to Lasken.
Given the variety of options in RNA-seq protocol design, Li suggests researchers match their purification and library-preparation methods carefully to their sample and application. For instance, when working with small cell numbers, select a product specifically designed to handle that use case.
But don’t neglect your controls, Li adds. Ensure RNA is of good quality and length (for instance, using Agilent’s Bioanalyzer system),and use spike-in controls, such as the ERCC standard, to monitor RNA recovery during purification and library preparation.
Single-cell RNA-seq requires even more controls. “Every new project, I would definitely encourage new investigators to put a lot of time into the details of RNA preparation and preserving RNA,” Lasken says. He typically incorporates a no-cell control (to control for the possibility of RNA contamination), as well as positive controls and whole-tissue controls, which should produce similar expression patterns to the aggregate transcriptome of multiple individual cells. And he tests amplified cDNA populations by qPCR for housekeeping genes to ensure RNA quality prior to sequencing.
Proper controls
Indeed, controls are key no matter what type of RNA-seq one does. Alexey Pronin, an associate scientist at the University of Miami School of Medicine, has used RNA-seq to study GPCR expression in the mouse cornea, finding—quite unexpectedly—that certain olfactory receptors also are expressed in the eye. Upon detecting those transcripts, he attempted to validate them using qPCR and RNA-in situ hybridization with Advanced Cell Diagnostics’ RNAscope technology. Though several were confirmed, about half the genes that appeared to be transcribed by RNA-seq could not be validated using these alternative approaches, Pronin says. And, some genes he expected to find by RNA-seq (such as housekeeping genes) could not be found. “My general advice would be to do confirmation by other means,” Pronin says. “It is the first key step. Otherwise, you cannot trust what you see.”
Pronin allows that he didn’t check every step to know precisely why his experiment produced some unreliable results, but he remains skeptical of the method in its current state. Others are more upbeat. Cresko’s advice for RNA-seq newbies: Read the literature, or talk to your colleagues. “Get input from others who have done this. Talk with them, because that devil in the details becomes most clear when having those conversations with people in the proverbial trenches who have actually done it.”