There are many ways of looking at RNA expression, from PCR to NGS. Yet microarrays’ ability to query thousands of different transcripts at a time leaves PCR in the dust, while their consistency and ease of interpretation keep them ahead of upstart rival RNA sequencing (RNA-seq)—especially when looking for differential expression patterns in well-characterized systems.
Hip, hip, array
A gene expression microarray typically has thousands of arrayed nucleic acid oligonucleotide probes. RNA samples that have been converted to labeled DNA are allowed to hybridize with the slide. After washing and staining, the array is scanned, with the intensity of signal (fluorescence, for example) at each spot indicating the quantity of each species of RNA present in the sample. Comparison samples are usually run on separate arrays (which may reside on the same slide, plate, or cartridge), although some platforms allow for so-called 2-color assays in which control and test samples compete with each other, resulting in red, green, or yellow spots indicating whether there was differential expression at a given location.
One often-heard comment is that microarrays can’t do true discovery—they only interrogate the known sequences that are being probed for. “You’re never going to get something new, something that’s not on the chip,” points out Chris Krebs, manager of the University of Michigan (UM) DNA Sequencing Core, which also handle’s UM’s microarray services. But “we get people who are studying diseases that have been around for a long time. There is a lot known about the cell biology and the genes that are involved, but it’s not clear as to how the sets of genes change.”
The field of players in the gene expression microarray market has certainly thinned.
The field of players in the gene expression microarray market has certainly thinned. There remain only two major manufacturers—Thermo Fisher Scientific, which acquired Affymetrix in 2016, and Agilent Technologies—and perhaps some smaller companies, says Krebs. “There are some core facilities that are phasing out or have dropped their microarray support of services.”
Biocompare’s Microarray Search Tool
Find, compare and review microarrays
from different suppliers Search
Yet for comparing the expression “of panels of known genes, microarrays are the industry standard,” notes Valentina Maran, global product manager, diagnostics and genomics division, at Agilent Technologies. She cites current uses of microarrays evaluating changes linked to specific diseases such as cancer and diabetes, validating the effect of changes due to gene editing, studying host/pathogen interactions, and analyzing the response to therapeutics, among others.
For the most part, says Erika Thompson, assistant director of the sequencing and microarray facility at the University of Texas MD Anderson Cancer Center, arrays are used for differential expression comparisons of pre- and post-treatment, and of different time points.
What’s on the array?
“There is so much real estate” on modern gene expression arrays, and the selection of markers is so superior to their predecessors, that—rather than being pathway- or disease-specific—they generally allow researchers to query the entire genome transcriptome in a single experiment, says Shantanu Kaushikkar, director, genotyping microarray platform, for Thermo Fisher Scientific. “By far I think most people are looking for signatures across the genome.”
Some microarrays like Thermo Fisher’s Applied Biosystems™ Clariom™ D assays query not only canonical mRNA but also splice variants, rare transcripts, as well as long non-coding RNA (lncRNA). Other arrays may specialize—for example, in the detection of splice variants, for which “a different probe design strategy is usually required … [they] consist of a minimum of one probe per exon, so that relative exon expression level can be used to determine the specific variant expressed by the sample,” notes Maran. Shorter RNA species such as microRNA (miRNA) can be detected on arrays designed specifically to do so, but cannot typically be assayed on arrays meant for the longer sequences.
To detect circular RNA (circRNA), Arraystar has developed a microarray that hybridizes to the unique sequences created by the formation of the circular junction, notes senior scientist Yanggu Shi.
While companies try to keep their arrays current and relevant with the latest content from annotation databases and literature, it can be difficult to keep any gene expression platform updated. Krebs recommends exercising due diligence “if you have a really specific question, with really specific SNPs, that you need answered”: make sure the array contains “the probe sets that are going to answer the question that you want.” If not, a customized array might be in order.
Why not NGS?
The cost of microarray analysis—about $300-350 per sample—is about the same as for RNA-seq, and both technologies theoretically have the capacity to look at tens or thousands of samples with roughly equal effectiveness. So why choose one over the other?
For Krebs, the choice often comes down to “whether it’s a discovery study—looking for something new that’s never been seen before—in which case you would use NGS. If you have a very well-established system and all you’re looking at is a perturbation in what’s known, then the microarray is the ideal format for that.”
Thompson points out that labs using microarrays are often continuing ongoing projects, with the need to compare to older data.
But there is also the question of bioinformatics burden: “Investigators don’t have to wait for a bioinformatician to do their analysis,” she says. “And that’s actually sometimes a very, very strong reason why an investigator will start new projects using microarrays.”
Microarrays are also better for quantification of low abundance transcripts including lncRNA, fusion events, and splice variants, says Shi. They can detect down to about one transcript per cell—making them several-fold more sensitive than RNA-seq at normal sequencing depths. Greater than 95% of circRNA cannot be quantified by sequencing—even if they can be detected—while they can be counted on microarrays.
Microarrays are very consistent and reliable, says Kaushikkar. “We use them in biobanks, we use them in precision medicine studies, we use them in applied” as well as clinical markets.
They are fully standardized, easy to implement, flexible in terms of content and throughput, and have a far lower informatics requirement than NGS. So what’s not to love?