At the Keystone, CO, meeting, “MicroRNAs and siRNAs: Biological Functions and Mechanisms,” which was held late last month, a number of scientists focused their presentations on optimizing and expanding the applications of this technology for target validation. While the technology remains challenging to apply, researchers reported significant progress toward developing products based on a wider fundamental understanding of basic siRNA pathway mechanisms.
The scientists also noted that off-target events, or lack of specificity between siRNA and its target mRNA, have yielded new information about cancer therapeutic targets and how cancer cells develop drug resistance.
William S. Marshall, Ph.D., group vp for technology assessment and business development at Thermo Fisher Scientific (www.thermo.com), said that understanding core mechanisms of the RNAi pathway led to greatly enhanced awareness of best practices for use of RNAi technology, particularly in the control of off-target events.
In 2004 Fisher Scientific (now Thermo Fisher Scientific) acquired Dharmacon (www.dharmacon.com), which has been developing siRNA technology and its applications. Over the past five years, Dharmacon developed partnerships with a wide variety of companies, including Abbott Laboratories (www.abbott.com) for the development of siRNA-based cancer therapeutics, Lentigen (www.lentigen.com) for lentiviral reagents to deliver shRNA expression vectors into cells, and amaxa (www.amaxa.com) for the use of Dharmacon’s siRNA libraries with amaxa’s Nucleofector® nucleic acid delivery technology.
Predicting Gene Silencing Activity
Dharmacon was the first to develop and publish rational design principles and proprietary algorithms that could predict the gene-silencing activity of siRNA sequences in silico. Its research revealed dozens of parameters that influenced siRNA activity and allowed the development of collections of predesigned highly functional siRNA molecules targeting the entire genome.
Company scientists performed a study with detailed analyses of off-target gene silencing by conducting global expression profiling studies of human cells after transfection with siRNAs. They found that except for instances of near-perfect sequence homology, the level of overall complementarity between a siRNA and off-targeted mRNAs is not correlated with off-target silencing. The researchers concluded that many current bioinformatic protocols employed to minimize off-target effects may have little merit in eliminating all but the most obvious off targets.
Their work suggested that the root cause of most off targeting was because the siRNA was acting like a naturally expressed microRNA. The principle determinant of off-target silencing was due to the presence of “seed” regions in the siRNA nucleotide, positions 2-7 on the antisense strand, that are complementary to a “landing site” in the 3´ untranslated region of mRNAs that are regulated by microRNAs. The study thus identified new parameters for determining siRNA specificity that allow for further development of predictive algorithms to minimize off-target events and enhance siRNA design.
“In trying to find a valid cancer target, one would treat cancers cells with siRNAs targeting a variety of genes to identify those genes that when inhibited resulted in cancer cell death. Off targets may lead to the down regulation of a variety of genes that result in cell death, but the event isn’t correlated with down regulation of the targeted gene; these are false positives.” To enhance specificity of silencing he described two approaches developed by the company, including siRNA pooling and siRNA chemical modification.
Dr. Marshall said “an important part of what we have been doing is validating that the siRNA pooling concept leads to specificity enhancement. It is somewhat counter intuitive, but we make mixtures of four siRNAs that target the same gene— SMART-pool® siRNA reagents allow enhancement of specificity.”
Because each siRNA sequence contains a unique potential seed region that can lead to off targets, but all of them target the intended gene, by making a mixture one can maintain a higher concentration of on-target events while diluting the off-target potential of the individual sequences.
The Dharmacon team introduced chemical modifications to siRNA that could interfere with microRNA interactions but not affect siRNA activity. Dr. Marshall said that “by introducing chemical modifications into the siRNA in the seed region we can disrupt its microRNA-like behavior and reduce off-target activity.” Their research led to the development of On-Target plus™ siRNA that, according to the company, can reduce such events by almost 90%.
Abbott has applied siRNAs for cancer target identification and validation. Stephen Fesik, Ph.D., divisional vp, cancer research, global pharmaceutical R&D, described a number of these applications.
“In 2003, we thought siRNA was very specific but, as everyone now knows, siRNAs can produce off-target effects and care must be used in interpreting the results from screens using large siRNA libraries,” he said. However, according to Dr. Fesik, despite this lack of specificity, you can get useful information using siRNAs in large-scale, broad screens.
Dr. Fesik presented results from studies using siRNA-based screens for identifying novel cancer targets using apoptosis assays and synthetic lethal screens as well as for elucidating the mechanisms of drug resistance in cancer cells. In addition, he described a strategy for using inducible shRNAs (small hairpin RNAs) for cancer target validation in vivo.
Dr. Fesik discussed the use of an siRNA library to identify new cancer targets by conducting a cell death assay screen against 3,700 possible targets. Many false positive hits were obtained in this screen. However, by testing several siRNAs against the same protein, the off-target hits could be eliminated and novel cancer targets were identified.
The Abbott group also identified the cancer target survivin, which was specific for cancer cells that have a ras mutation. This was accomplished by identifying siRNAs in a large-scale screen that killed cancer cells containing a ras mutation but which had fewer effects on isogenic cell lines containing wild type ras.
Another use of siRNA that was presented involved the identification of proteins that cause drug resistance. This application of siRNA was demonstrated using a Bcl-2 family inhibitor of antiapoptotic Bcl-2 proteins, including Bcl-2, Bcl-XL, and Bcl-w, effectively inducing cell death in many small cell lung carcinoma (SCLC), lymphoma, and leukemia cell lines. (Many cancers evade programmed cell death by over-expressing the antiapoptotic proteins of this family.)
Other similarly derived tumor cell lines proved resistant to the drug, as were cell lines from other solid tumors. To study the mechanism of these cancer cells’ resistance, the investigators screened an siRNA library containing siRNAs against about 4,000 potential drug targets in a resistant SCLC-derived cell line in the presence of a low dose of ABT-737. The top three “hits” resulted from off-target gene silencing by three siRNAs that prevented transcription of mRNA for Mcl-1, a member of the Bcl-2 family that is not inhibited by ABT-737. The off-target silencing resulted from complementarity between the siRNA’s seed region and the 3’ untranslated region of the Mcl-1 mRNA.
Reduction of Mcl-1 with siRNAs or two small molecules, Bay43-9006 and Seliciclib, was sufficient to restore cancer cell sensitivity to ABT-737 in the resistant SCLC cell line and in other solid tumor derived cell lines. According to Dr. Fesik, siRNAs can be employed to discover mechanisms of cancer cell resistance, which may give insight into how to treat resistant tumors. Abbott scientists have also relied on RNAi-based screening to explore the development of anticancer agents that inhibit multiple cellular kinases. “A major challenge in identifying such molecules is finding out which kinases to inhibit in each cancer to maximize efficacy,” said Dr. Fesik.
To identify Akt-cooperating kinases, the scientists screened a library of kinase-directed siRNAs for their ability to enhance cell killing in the presence of a potent Akt inhibitor, A-443654. SiRNAs targeted at a casein kinase, or the inositol polyphosphate multikinase, enhanced cell killing in the presence of the drug, suggesting that small molecules targeting these kinases in combination with Akt inhibitors may increase their therapeutic index.
Dr. Fesik also talked about an inducible siRNA system in vivo to evaluate the potential therapeutic effect of hypoxia-inducible factor 1a (HIF-1a) inhibition on established tumors in SCID mice. In these cells, expression of shRNA that silences HIF-1a can be controlled by doxycycline, the shRNA being induced in the presence of the drug. The investigators produced xenograft tumors in the mice using these cells and evaluated the effect of controlling HIF-1a in two tumor types.
Results showed that while HIF-1a inhibition by siRNA caused temporary tumor growth cessation or regression, the tumors adapted to the protein loss and continued to grow. HIF-1a inhibition in early-stage tumors was more effective than inhibition in established tumors, and different tumor types responded differently to the inhibition.
Most importantly, the inducible shRNA suppression system allowed the investigators to determine that the tumors adapted to HIF-1a loss through a mechanism independent of its expression.
“The efficiency of in vitro and in vivo knockdown of various HIF-1 targets including PGK1, LDH, and VEGF at the mRNA or protein level that we observed with target inhibition by shRNA is comparable to that achieved with targeted gene ablation,” noted Dr. Fesik.
Artemis Pharmaceuticals (www.artemis-pharmaceuticals.de), a wholly owned subsidiary of Exelixis (www.exelixis.com), is developing genetically engineered mouse models for the in vivo functional analysis of selected genes. In November, the company signed a research agreement with Merck & Co. (www/merck.com) that, it says, represents a “significant initiative” supporting a large-scale high-throughput approach to gene function analysis in vivo using shRNA knockdown in genetically engineered mice as models for human biology.
To make its shRNA knockdown mice, company scientists transfect F1 embryonic stem cells (ES cells) with inducible shRNA expression vectors, introduce the cells into tetraploid blastocysts, then implant these into pseudo-pregnant surrogates. Artemis’ ArteMice™RNAi mice reportedly can be produced in four months. The company says that constitutive RNAi knockdown of over 80% can be induced in almost all body tissues, and that endogenous gene knockdown reproduces the phenotype seen with conventional gene knockout technologies. This is the first time that such a significant and widespread knockdown has been demonstrated using RNAi in vivo.
According to Paul Rounding, Ph.D., managing director of business development at Artemis, “All of the offspring mice come from the ES cells and immediately carry the target gene of interest. We then can induce shRNA-mediated, reversible in vivo gene knockdown, organism-wide in all tissues and cells, by treating the mice with Doxycycline.
He also noted that the knockdowns specifically reflect known drug activities.
Dr. Rounding credits Jost Seibler, Ph.D., Artemis’ research group leader for RNAi, with the pioneering work for Artemis’ knockout systems technology development.
“These mice allow in vivo drug target validation in a way that was not previously possible,” explains Dr. Rounding. “The ability to inducibly cause gene knockdown means you can knock down the gene once the disease has developed and validate the role of the gene therapeutically. Previously, this was only possible through gene knockdown prior to disease development, thereby mimicking prophylaxis, not treatment.”
Dr. Rounding further pointed out that a second important application of inducible shRNA in vivo models is that expression of disease-causing genes can be suppressed during the course of the disease by removing the inducer. These models can then be used in a compound testing modality and can supply a reversible positive control.
Single Nucleotide Mutant Targets
Neil Aronin, M.D., professor of medicine and cell biology and chief of endocrinology at University of Massachusetts Medical School, noted that while the potential for using siRNA to treat neurodegenerative diseases is exciting, Huntington Disease (HD) protein mRNA poses a particular silencing challenge. HD is caused by a mutant protein that, as it accumulates in specific CNS neuronal cells, eventually destroys them. “All patients with the disease have a mutated allele with an expansion on the CAG glutamine coding region that extends beyond about 35 repeats,” explained Dr. Aronin. “The resulting HD phenotype is dependent at least in part on the extra CAG repeats and the resulting aberrant protein is at the root of the disease.”
Further, he said, the aberrant allele differs from the wild-type, normal allele by only a single nucleotide. Accumulation of the protein causes, among other cellular abnormalities, mitochondrial dysfunction, impairment of transcription, abnormal NMDA receptor function, and abnormal intracellular vesicular trafficking.
“We reasoned,” continued Dr. Aronin, “that if we could eliminate the mutant protein in susceptible neurons by reducing its translation through specific siRNA silencing while allowing continued translation of the wild-type mRNA, we could reduce the number of problems associated with the disease.”
To accomplish this, an RNAi specifically targeting the mutant allele and containing the CAG repeat and the single mutation was required.
By studying the normal genetic polymorphisms present in HD genes, the investigators were able to produce RNAi that could distinguish between the abnormal allele coding for HD protein and the normal allele.
At the meeting, Dr. Aronin described a recent in vivo proof of principle in which he and his colleagues could, using siRNA against huntingtin mRNA, delay the onset of disease phenotype in viral transgenic mouse models of HD. They further demonstrated in vitro in human cells preferential knockdown of huntingtin alleles with particular heterozygosity at sites of single nucleotide polymorphism.
The next step is to use this technology in vivo to knock down one huntingtin allele product while preserving the other huntingtin allele mRNA. In theory, this approach could be used for other CAG triplet repeat diseases and other disease genes in which the site of mutation is not amenable to RNAi.