Knocked down by a hairpin

This feature appeared in the May/June 2009 issue of Australian Life Scientist. To subscribe to the magazine, go here.

In 2003, Ross Dickins finished his PhD and headed to Cold Spring Harbour in New York to seek out a postdoc with cancer geneticist Scott Lowe. Dickins was certainly not disappointed with his choice and describes this time as “awesome”. Lowe was renowned for some landmark work on the genetics of chemotherapy, and particularly what determines chemosensitivity and chemoresistance.

“I was really attracted by Scott’s work,” Dickins says. “A lot of cancer research in mouse models focuses on recapitulating the human disease according to various pathological and genetic criteria. Scott was more interested in the genetics of the therapy response – using a fairly standard DNA-damaging chemotherapy agent and then looking at how the genetics of the mouse influenced outcome.

“Mice with essentially identical disease would respond completely differently to a drug based on a single genetic change in the tumour. The responses he got were pretty much black and white, as is often seen in the clinic. It was very cool work.”

By the time Dickins landed on his doorstep in New York, things there had evolved even further and RNAi was becoming a very big thing. Lowe wanted to use these tools in his mice to mimic what a therapy might do in a patient.

Fortunately, just down the road and already collaborating with Lowe was RNAi guru Greg Hannon. Dickins found himself “positioned between these guys, possibly inadvertently, but it ended up working really well”.

The RNA silencing phenomenon for gene suppression was discovered in 1998 in worms, and then found to be active in mammalian systems over the subsequent few years. In mammalian cells, it was initially used only as an experimental tool whereby cultured cells were transfected with siRNAs to transiently silence gene expression.

According to Dickins, the idea emerged around 2001 of driving short hairpin RNA (shRNA) production in cells. “This is basically a linear RNA sequence that can fold up into a double-stranded region in a kind of stem-loop structure; it is then metabolised by the cell to siRNA, resulting in the knockdown of a particular gene.”

What Lowe and Hannon wanted to do about the time Dickins came along was design a new style a loss-of-function genetics using RNAi whereby gene output could by reversibly switched off and then back on again, where they wanted and when they wanted.

In other words, they were aiming for reversible and inducible control of gene expression.

“Such a reversible system would be unique, because in existing mouse models you could inducibly delete things or switch them on, but it is unidirectional,” Dickins says. “Instead of altering the gene itself, our concept used an shRNA to block the gene product being made. This could be then controlled from a distance by switching the shRNA on and off.” This is exactly what they did.

With a lot of intelligent guesswork and some luck, Dickins and colleagues just happened to pull a couple of vectors out of their toolbox that worked a treat once assembled. In fact, they worked better than anything tried before.

It turned out that the process they had designed was mimicking a natural cellular process, which is why it worked so well experimentally. “Although we did not realise it at the time, we were simply making short hairpins that mimicked naturally occurring microRNAs in the cell,” he says.

“Then we used Pol II promoters to drive these shRNAs and it worked beautifully. These promoters yield abundant and stable transcripts and, most importantly, can be made inducible.”

This was achieved by virtue of a specific transcription expression system that has been used experimentally for about 20 years – the tetracycline-regulated ‘Tet on/off’ system. ---PB---

Fame and fortune

In 2005, Dickins reported the group’s first reversible gene knockdown in cultured cells by shRNA expression in Nature Genetics. The chronological extension of the cell work, expression in mice models, was published in 2007, also in Nature Genetics.

“We simply added the tetracycline analogue doxycycline to the drinking water of transgenic animals to turn specific gene expression on and off,” he says.

Refinements since then also allow the gene silencing effect to be restricted to a desired cell type or tissue, furthering the applications of this technology in disease research.

There are a couple of different ways this technology can be used, Dickins says. One approach is gene-by-gene testing and the other involves screening a multitude of genes at once.

Dickins’ group at WEHI takes both approaches in their work. They make transgenic animals inducibly expressing specific hairpin RNAs to turn off a specific candidate gene, such as one suspected to affect tumour survival. After inducing tumour formation, they just feed the mice doxycycline and see what happens.

These sorts of experiments take at least a year to complete. In other screening-type work, Dickins’ group might test 1000s of different shRNAs in cell culture and assay a specific readout, such as cell death. Genes he identifies in this way can then be explored further as potentially druggable cancer targets.

The wider impact of this hairpin RNA approach is growing, with many people using it across the board and not just in cancer research.

“It is really good for anyone who wants to do some quick and easy loss-of-function genetics, particularly if they already have a candidate gene or protein,” Dickins says.

“It is not all that expensive and there is no need to make a knockout mouse, which is time-consuming and expensive, and something not everyone is set up to do.

“Others do the same sort of things with siRNAs, but there you are more restricted experimentally – the work can only be done in cell culture and expression is transient. With shRNAs, you can reversibly switch them on and off, and you can stably express a hairpin in a cell to get ongoing gene knockdown and thus loss of function.”

This allows the techniques to also be used for making disease models, he says. “Let’s say that in your disease of interest you identify a candidate expression product that is druggable, such as a cancer cell survival factor. You simply make an shRNA to target that gene and use our system to get loss of function in an animal model just by switching that hairpin on.

“The model is then useful for both studying the disease and the genetic alterations that affect therapeutic responses, as well as for testing new drugs.” ---PB---

Targeted therapies

For Dickins’ group, the ultimate hope for this technology is to accelerate cancer drug discovery by creating such mouse models and finding better therapeutic targets. Traditional cancer treatments induce significant side effects because they act basically by damaging cells, and not just tumour cells.

The increasing development and use of so-called targeted therapies aim to make cancer treatment more patient- or cancer-specific, as a patient’s response to therapy can differ with the genetics of the tumour itself.

“Using inducible RNAi to inhibit specific genes or proteins in established mouse tumours, we can model likely effects of different targeted therapies in a high-throughput manner,” he says. This approach would reduce and time and cost of conventional drug development.

The group now works mainly on mouse models of blood-borne cancers including leukaemias and lymphomas.

“Of course, it is not all our doing, but this shRNA technology is a revolution in loss-of-function genetics. We hope that ultimately it can help match the genetic profile of a tumour with an appropriate therapeutic agent. So, in an ideal future, a cancer patient might initially have a biopsy to genotype their tumour, and then a therapeutic can be selected that will work best to treat that tumour.”

As for his personal research goal, Dickins would really like to find new ways to successfully treat one of the countless mice in which he has induced cancer over the years. “Just curing one mouse of cancer would make me really happy.”