Gene Expression Platform Selection Guide

Understanding which genes are being expressed in a cell, or in a population of cells, is the basis for much of modern biomedical science. Comparing RNA levels across time, between species, or between healthy and diseased tissue, for example, can provide a window into the molecular basis of the phenotypic differences. It may allow clinicians to determine a disease etiology, and researchers the pathways and targets worthy of further study.

Many platforms can be employed to query the transcriptome—or a part of it—each with its own set of advantages and disadvantages. Here we look at the most common of these, and glance at other noteworthy platforms as well.

Whole transcriptome or targeted?

Gene expression platforms can be categorized in several different ways. Among these is whether the entire transcriptome (or a significant portion of it) or a more targeted selection is to be interrogated.

Techniques such as RNA sequencing (RNA-seq) and microarrays, while fundamentally different technologies, are both used to simultaneously examine transcripts from a very large number of genes—a potentially infinite number in the case of RNA-seq, and up to tens of thousands in the case of microarrays. These are often used “for fishing” for differential gene expression, for example to globally compare a new cancer cell line to normal, explains Yanping Zhang, Scientific Director at the University of Florida’s ICBR Gene Expression and Genotyping Core.

In contrast, quantitative reverse transcription PCR (herein referred to as qPCR) as well as digital PCR (dPCR) are normally used for “just a few genes,” she says.

Using digital PCR and qPCR may be the choice to evaluate a large number of samples across a smaller panel of genes—to confirm the results of broader gene expression screen, for example—points out Angelica Olcott, Senior Applications Manager at Bio-Rad Laboratories.

RNA-seq

RNA-seq relies on reverse transcription of RNA into cDNA, followed by next-generation sequencing (NGS) of that cDNA. It boasts a virtually unlimited dynamic range and, because it is based on read counts it produces absolute (rather than relative) expression values. NGS requires no prior knowledge of the genome to perform, although the resulting sequences are typically mapped back to a reference sequence during analysis. But it is costly and generates large amounts of data that requires a non-trivial level of bioinformatics expertise to interpret.

“RNA-seq has become very popular in cancer for its ability to comprehensively analyze transcriptome changes or to profile genome-wide gene expression levels in a single experiment,” Olcott notes.

“Currently, RNA-seq is used to perform bulk gene expression of all cells collectively in a tissue sample, or at the single-cell level to evaluate heterogeneity of expression levels based on cell type,” she adds.

Microarrays

Gene expression microarrays contain an array of oligonucleotide probes bound to the surface of a chip, to which complementary cDNA fragments hybridize, triggering the probe to fluoresce. Spots are counted, and the data analyzed to reveal the relative abundance of each RNA target, typically in the form of a heat map. “Microarray analysis is very cost-effective for profiling numerous genes, using an already-defined large panel of gene targets,” Olcott points out.

Yet for most studies, RNA-seq has largely supplanted the microarray. Zhang’s core, for example, still uses microarrays for genotyping but “we just retired the microarray for gene expression.”

There are areas for which microarrays are still the preferred, if not the only, method to query expression of large numbers of RNA species. Because they are much less abundant than mRNA, expression of long non-coding RNA (lncRNA), including circular RNA (circRNA) and many other non-coding species, may be masked by RNA-seq, being either severely miscounted, undercounted, or not found at all, explains Yanggu Shi, Senior Scientist at Arraystar.

PCR

Whereas RNA-seq and microarrays examine the entire transcriptome in one go—typically in one or very few samples—qPCR and dPCR look at only a comparative few gene products, but are capable of doing so for a large number of samples. Both can be used for multiplexed detection of multiple targets in a single well. Both PCR techniques provide sensitive and specific detection and quantitation.

Of the two, qPCR “offers greater sample throughput, including the ability to run 384-well plates, which has encouraged its use with automation,” Olcott notes. “The method is very cost-effective and rapid, and its technology is familiar and accessible to many researchers … and is now routine in many laboratories.” Its wide dynamic range allows it to handle both dilute and concentrated samples under the same conditions, and it can discriminate among splice variants.

dPCR offers absolute target quantitation, without the need for standard curves. It is better at detecting and discriminating rare alleles, and is also called for when small changes (less than twofold) need to be distinguished. It is more tolerant of PCR inhibitors found in many complex biological samples, less sensitive to secondary structure, and requires less sample, than qPCR. But it is more expensive and has a slower time-to-result.

Spatial transcriptomics

A host of other platforms are vying for gene expression customers as well, each offering its own unique spin. Some, for example, rely on a bead-based capture process. Others avoid PCR bias by using a non-enzymatic process. Some will query a very large number of genes, while others offer higher throughput for more targeted expression. Some look at bulk RNA while others focus on individual cells. Some boast lower cost, or ease-of-use, or faster turnaround, or more robust data analysis, or proprietary databases.

Among the most exciting innovations are those based around spatial biology, which allow RNA to be viewed as it is expressed in the cell or tissue, in its two- or even three-dimensional context. “It’s a really hot topic, so everybody’s talking about it,” Zhang observes. “But at least in our facilities, there are not so many people using it [yet].”