Believe the hype

Is RNAi the real deal? Graeme O'Neill asked Australian scientist and global authority John Mattick.

Since 2000, RNA interference (RNAi) has been sluicing a path through biology's Augean stables, vastly simplifying the task of assigning functions to the myriad anonymous gene sequences now held in international databases.

RNAi has ushered in the era of high-throughput assays for screening protein function in cell-culture systems and living species.

The first antiviral and cancer therapies to exploit RNAi's matchless precision in knocking down expression of target genes are moving into clinical trials, harbingers of a drugless therapeutic revolution.

But the revolution has barely begun, according to Prof John Mattick, director of the University of Queensland Centre for Molecular Bioscience, and a pioneer in the new science of 'R-nomics', the study of RNA structure and function, and RNA regulatory networks.

Consider a striking statistic, says Mattick. Living cells seethe with RNA molecules, yet fewer than half are messenger RNAs specifying proteins.

Once dismissed as cellular detritus, because they tended to migrate off the edges of electrophoresis gels, the function of this menagerie of non-mRNA molecules has been enigmatic.

There is increasing evidence that they are the functional elements of the intricate 'digital' regulatory system that Mattick believes is responsible for modulating and integrating gene function across genomes.

Because these small RNAs do not code for proteins, they are not amenable to normal protein-analysis techniques.

Molecular geneticists have regarded them as an intractable problem -- without a functional assay system, how do you determine the function of small RNAs that don't code for proteins?

A new role overnight

Mattick says that, almost overnight, RNAi technology has found an unexpected and important new role. It's not limited to shooting messengers -- all RNA species are fair game. Select your anonymous regulatory RNA, make an RNAi construct to knock it down, then check for phenotypic effects.

"For the first time, we have a tool to assess the function of regulatory RNAs in vivo in cells and animals," he said.

Recent evidence has indicated a key role for both small and large regulatory RNAs in embryogenesis and cell differentiation. Many so-called microRNAs (miRNAs) -- only 22 bases long -- have been shown to control a wide range of developmental and physiological pathways in plants and animals, including hematopoietic differentiation, adipocyte differentiation and insulin secretion in mammals, and have been shown to be perturbed in cancer and other diseases.

At the other end of the scale the non-coding RNA called Air is more than 100,000 bases long and is also involved in regulating developmental processes, in this case the differential expression of IGF-2 (Insulin-like Growth Factor-2) in different types of cells, via methylation-induced imprinting.

Moreover, says Mattick, the sheer abundance of regulatory RNAs suggests points to the astonishing probability that most of the 98 per cent of the human genome, once dismissed as non-coding genetic junk, is actually transcribed into functional RNAs. While some may code for peptides and as yet unknown proteins, most appear to have no protein-coding capacity and are presumably regulatory in function.

The tip of the iceberg

The perennial game of 'guess the number of genes' has come almost full circle. Less than a decade ago, geneticists still estimated there were around 100,000 genes in the human genome. Mattick now suspects that estimate, revised down to just 30,000-odd genes by the Human Genome Project, was actually close to the mark. Genes coding for functional RNAs may outnumber protein-coding genes by a factor of two or three.

He said the discovery of these small regulatory RNAs calls into question the traditional concept of a gene, especially since many regulatory RNAs come from the introns of genes, both protein-coding and non-coding, and many RNA transcripts overlap each other.

The list grows apace -- indeed, says Mattick, recent evidence suggests mRNAs from active genes are probably in the minority in cells. The current roll-call of regulatory RNAs is almost certainly just the tip of the iceberg.

It includes more than 1000 micro-RNAs cleaved from the cell's own homespun hairpin RNAs, small RNAs of non-hairpin origin, endogenous short interfering RNAs (siRNAs), small double-stranded RNAs (dsRNAs) and small nucleolar RNAs (snoRNAs), that appear to work exclusively in the nucleus.

Meanwhile, the RNAi genomics Big Bang has spawned a universe of possibilities in basic and applied research. Gene-knockdown by short interfering RNAi (siRNA) and DNA-directed (ddRNAi) technology is now routine in bioscience and medical research laboratories across Australia, and around the world.

Determining gene function could hardly be simpler. Sift your cells for messenger RNAs, convert them to cDNAs, and plug into an off-the-shelf cassette like CSIRO's Hellsgate vector, which automatically generate transgenes for 'hairpin' dsRNA-knockdown molecules.

Transform your plant or animals cells with the vector and wait for their resident 'elves' -- RNA-induced silencing complexes (RISCs) - to cleave the hairpin molecule into small (20- to 23-meric) targeting templates. The RISCS are now armed for their high-precision search-and-destroy missions against the mRNAs of the originating genes.

CSIRO recently applied to the Office of the Gene Technology Regulator to field-trial hundreds of RNAi-knockdown rice lines on land near Wagga. Plant Industry researchers created the lines as part of the division's contribution to the international 'Green Machine' project to investigate gene function in rice.

Currently RNAi works beautifully in invertebrates, less so in plants and animals. But its relative inefficiency in plants and animals is not necessarily a problem. "In most cases, just knocking down the gene produces positive functional effects," Mattick says.

"But a negative result is harder to interpret. Sometimes, a fraction of normal gene activity is still enough to give a wild-type phenotype.

"Most researchers experiment with more than one RNAi construct. It's not easy to predict which ones will work.

"But the really big advantage of RNAi is high throughput. It allows you to run hundreds or thousands of experiments at a time, or to easily make constructs to knock down your favourite protein-coding or non-coding sequences.

"When DNA sequencing was first introduced, the doomsayers said it wasn't perfect because replication errors occurred -- although far fewer than with protein sequencing. But if gene-sequencing errors occur at a rate of only one in 10,000 to 100,000, you don't need perfection.

"If RNAi can be made more efficient, great. If it can't, it's still a great research tool."