RNAi Therapeutics: Qiagen High-Throughput RNAi User Meeting San Francisco (Part 2 of 2)

After the coffee break, John Hogenesch illustrated how RNAi has revolutionized our concept of mammalian genetics. Before RNAi, researchers essentially studied two expression states of a gene: when it was fully expressed (+/+), or when both alleles had either lost their function or weren’t expressed at all (-/-), and rarely the heterozygous state (+/-). RNAi, however, now allows the functional output of a gene to be measured in virtually all the states in between, depending on the RNAi dosage used.

Hogenesch’s systems biology team is interested in how the circadian rhythm is genetically wired that that it can be robust and yet adaptable, and consequently set up a beautiful system whereby the activity of master transcription factors of this genetic program could be measured in real-time over days based on the luminescence generated by binding of the transcription factors to a reporter gene. They then knocked down known regulators of this genetic circuit to varying degrees by titrating the siRNA amount and then assessed the correlation between the degrees of knock down and functional outputs. Pleasantly surprising not only Hogenesch, but also all the researchers worrying about how the degree of knockdown affects the phenotype they are seeing in RNAi experiments, in many cases the functional output appeared to almost directly correlate with the remaining expression level of a gene after knockdown.

However, in cases where there exist paralogues, that is highly related genes that have largely retained overlapping functions, the knockdown of one gene is often compensated almost entirely by the increased expression of the corresponding paralogue. You probably may want to avoid targeting these genes with an RNAi Therapeutics, or any other therapeutic for that matter. Finally, there were also cases where the knockdown gave non-linear responses, and this is mostly when enzymatic reactions were affected.

Nevertheless, as everybody in the audience having traveled from across the country for this one-day meeting knew, genetic circuits, long-term, are usually quite robust and can adjust to external fluctuations. It is therefore also of interest to RNAi Therapeutics (and then again, all classes of drugs, too), particularly for chronic applications, that out of the ~4400 gene knockouts in mice that have been generated so far, only 1000 are lethal in the homozygous state, and just 53 in their heterozygous state, and these already are probably biased numbers (scientists like to study genes with apparent phenotypes, lethality being a very obvious one). This has implications both for chronic drug treatment in general (drug effect may diminish) as well as when we worry about adverse consequences of off-targeting (off-targeting may be tolerated better than expected).

Next up was Loren Miraglia from the Genomics Institute of the Novartis Research Foundation (GNF, San Diego), a basic research institute with preclinical capabilities to feed into the Novartis drug development pipeline. The main part of the presentation focused on the technical aspects of establishing a lentiviral RNAi library for the investigation of difficult-to-transfect cells. The upshot was that these libraries once established can be very powerful, however it takes a considerable infrastructure to realize that goal.

There were two pieces of information that may give some quite interesting insights into Novartis’ RNAi Therapeutics plans. One was an overview of the RNAi libraries being used by GNF which essentially all somewhat disappointingly were focused on the “druggable genome”, disappointing because a lot of genes may be missed that could be addressed by RNAi Therapeutics. There was one notable exception, however, and that was a library devoted to genes involved in cancer. As the time approaches for Novartis to make their move on Alnylam, this confirms my belief that cancer will be an important part of the new license agreement. With screening efforts such as these, there should be more than enough targets to keep all the Alnylam licensees busy without having them compete too much with each other.

The second insight was Miraglia’s presentation of a new “gene X” that when mutated had been found to lead to an increase in LDL-R, the new star in the hypercholesterolemia field. When tested whether RNAi knockdown would also elevate LDL-R, this was indeed the case with a nice correlation between degree of target knockdown and LDL-R elevation and subsequent cholesterol lowering, just what you would like to see in an RNAi Therapeutic. So, like for Takeda, cancer may be well joined by metabolic disease as one of the main initial therapeutic fields for Novartis’ RNAi Therapeutics efforts and driven by current delivery capabilities.

Systems biologist Sumit Chanda (Burnham Insititute, La Jolla) praised RNAi screening as an important tool to learn more about the vast part of the genome that is essentially left unexplored. He bemoaned the fact that it is paradoxically genes of which so much is already known about, think of p53 and TNF-alpha, that get the bulk of the research funding.

Chanda’s laboratory used RNAi screening to discover host factors involved in HIV replication, which could therefore also be potential targets for drug intervention. As a virologist/molecular biologist myself, I also see targeting host factors as a very promising approach for treating viral infection. Unfortunately for Chanda’s group, another group from Harvard just published the results of a similar RNAi library screen, meaning that they were “scooped”. Nevertheless, it appeared that there was only limited overlap between the results and one could explain this by the slightly different assay conditions employed or the quality of each screen. In any case, the point was well made that one screen will never answer all the questions and should be complemented by additional ones. Moreover, the importance for multiple negative controls and multiple redundant siRNAs towards a single gene (to confirm the sequence-specificity of the hit) was emphasized.

Natasha Caplen from the National Cancer Institute (NCI; Rockville, MD) picked up the personalized medicine theme for cancer therapy. As noted before, while monoclonal antibodies have changed the way cancer is treated today, it is often only few patients that respond to them. Moreover, many of the drugs’ mechanism of action converge on the few same signaling pathways, meaning that the mechanisms of actions are still limited and prone to mutational escape. RNAi Therapeutics could then either play a role in exploiting new mechanisms of actions or for sensitizing towards existing drugs. Caplen’s group chose the latter approach and focused on devising new strategies for the use of the bacterial L-Asparaginase enzyme (L-ASP) that has been used to selectively starve acute lymphoblastic leukemia (ALL) cells. One problem associated with L-ASP, however, has been the quite variable response to this treatment.

Caplen’s group suspected that cancers that do not respond well to L-ASP treatment may overproduce the human asparagine synthetase (ASNS) to compensate for asparagine deprivation by L-ASP. To test their hypothesis they first screened a library of cancer cell lines for L-ASP responsiveness and, indeed, L-ASP negatively correlated with ASNS expression levels. To demonstrate a causal relationship, they then knocked down ASNS which greatly increased the sensitivity of the cells to L-ASP. The sensitization was so unbelievable, up to 500-fold, that Natasha Caplen made her co-worker repeat the experiment again and again to convince her of the veracity of the results.

ASNS expression levels could now be used to select patients for L-ASP treatment. Alternatively, ASNS knockdown may be a viable way to increase L-ASP responsiveness. It more and more appears therefore that cancer drug sensitization, and not just targeting novel oncogenes appears to be a very promising avenue for future cancer RNAi Therapeutics. Similar patient selection and drug sensitization strategies were subsequently discussed for the topoisomerase inhibitor camptothecin (Caplen) and the Wnt signaling pathway by Paul Kassner (Amgen).

Finally, I’d like to summarize the talk given by Xiao-Dong Yang from Intradigm (Palo Alto, CA) on the delivery of RNAi Therapeutics to cancer. Using siRNAs provided by Qiagen, it appears that Intradigm screens for siRNAs essentially based on potency. I am somewhat skeptical whether this is sufficient and consequently raises some questions as to their findings that were largely centered on targeting the VEGF pathway in mouse cancer models.

Intradigm’s RNAi Nanoplexes consist of a cationic polymer to which the siRNA is complexed, a hydrophilic steric polymer such as PEG, and optionally a cell targeting ligand such as an RGD peptide. The Polytran polymer-based system is particularly interesting as it is a polypeptide composed of branched histidine and lysine residues. While the positively charged lysine condenses the siRNA, histidine only becomes positively charged once internalized into the endosome as they suck up the incoming protons. This supposedly leads to an increase in osmotic pressure causing the rupture of the endosomes to release the siRNA into the cytoplasm.

The Polytran system consists of fairly uniform 100nm particles that are slightly positively charged. Like many RNAi nanoparticle systems, of the Nanoplexes that make it into a tissue following administration, many are either found in the liver, the spleen, or tumor tissue. Many of them are also taken up by phagocytic cells such as macrophages which may pose challenges with dose prediction and requires that siRNAs are carefully characterized as to their potential to activate the innate immune system. Unfortunately, this aspect of their work was little discussed although unmodified siRNAs, prone to elicit innate immune responses and could therefore cause non-specific anti-tumorigenic effects, were used. With this in mind and the fact that only one control siRNA sequence was used, quite potent mouse antitumor effects were found at low mg/kg dosages with apparently knockdown efficiencies of up to over 90%.

While these approaches are valid for the development of an RNAi delivery system, the presentation reminded me of the fact that for companies like Intradigm it would really make sense to collaborate on the siRNA chemistry side with one of the established RNAi operations rather than committing insufficiently characterized siRNAs into the clinic.

On the way home, I was once again amazed by the fact that RNAi has emerged as the only gene knockdown technology that can be applied in high-throughput to explore the depths of our genomes. This both speaks to the potency and specificity of RNAi as well as the ability to translate our understanding of how RNAi silencing works on a molecular level into bioinformatic tools to separate the wheat from the chaff, as Sumit Chanda put it.

Thanks a lot to Qiagen for organizing a day full of excellent scientific talks in a pleasant Mission Bay setting, and I’d recommend everybody running qPCRs Qiagen’s robust and reliable SYBR Green QuantiTect reagents.

RNAi Therapeutics: Qiagen High-Throughput RNAi User Meeting San Francisco (Part 1 of 2)

An army of RNAi enthusiasts descended today onto the UCSF Mission Bay Campus in San Francisco, just across the street from Sirna Therapeutics/Merck, to discuss the latest in the use of RNAi for high-throughput screening. Illustrating the ever increasing popularity of RNAi, 175 scientists registered for this year’s one-day meeting, some from as far as Japan, compared to 75 when it was held for the first time two years ago in Boston. It was therefore fitting that a number of presenters noted as an aside that the discovery of RNAi and microRNAs not only transformed molecular biology, but also a number of lab personnel from bench-weary scientists to those that live and breathe RNAi every day. In the next two posts, I will try and highlight a few points from the meeting seen through my RNAi Therapeutics filter.

Carl Novina (Harvard and scientific advisor to mdRNA/Nastech) gave the keynote address with an introduction to microRNAs and their molecular mechanism of action. In a paper earlier this year, his lab had developed a biochemical model system to study microRNA repression that led them to conclude that microRNAs repress translation through the inhibition of the joining of the large 60S ribosomal subunit to the mRNA-bound 40S subunit of the protein translation apparatus. In today’s presentation, he expanded on that work by describing an RNAi-based screen for the discovery of genes involved in microRNA function. For this, his lab created luciferase cell lines in which one of the reporters contained a target site for an endogenously expressed microRNA and was therefore repressed. Knocking down a gene contributing to microRNA-mediated repression should therefore lead to an increase in the signal intensity from that reporter gene.

Encouragingly and attesting to the high specificity of RNAi, the screen uncovered many of the known microRNA pathway components such as Drosha, Dicer, and Argonautes 1 and 2. In addition, the screen making use of an siRNA library targeting the “druggable genome”, that is the ~7000 genes in our genome encoding for kinases, cell surface receptors etc. that are amenable to small molecule and/or monoclonal antibody inhibition, uncovered a number of proteasomal components. This raises the intriguing question whether an old hypothesis, namely that microRNAs may induce co-translational protein degradation may indeed be part of the complicated reality of the molecular microRNA repression process.

As an aside, Carl Novina noted that one problem with working on RNAi-related aspects in Boston is the considerable demand for talent there by the many RNAi companies located there. A nice problem to have for the many hungry post-docs at a time when public funding for basic research has stalled to pay for apparently more important projects overseas.

One recurrent theme of the conference was the role of small RNAs in the new era of personalized medicine, particularly for cancer, which not coincidentally also was the loud message heard from last week’s ASCO mega meeting. The case was made by Lynne Bemis (University of Colorado) as she compared the costs of today’s monoclonal antibody cancer therapeutics that run at about $30k per annum with clinical responses in only a subset of patients. To avoid patients wasting precious time on ineffective therapies and also to reduce healthcare costs, it is therefore important that these treatments are accompanied by diagnostics that can predict the response of a patient to the therapy, by the way an area first pioneered by the local Genentech.

The striking conclusion from her research was that microRNAs were by far emerged the best biomarkers for this purpose, far better than single protein or mRNA diagnostics. When compared to proteins this is likely due to the fact that microRNAs function as master regulators, often with direct and critical functions in cancer itself, while a single protein may only reflect a much smaller aspect of a much larger picture. When compared to mRNAs, it is the relatively high stability in most tissue preservation methods and their compatibility with formalin fixation that adds to the competitive advantage of microRNA diagnostics.

As an example, EGFR tyrosine kinase inhibitors (TKI) are now widely used for cancer therapy, yet only 10% respond to treatment. Lynne’s team therefore looked at whether EGFR protein levels, EGFR mutations, or genomic amplification levels of the EGFR gene would be good predictors of drug response. Unfortunately, all of them either failed to do so (EGFR levels) or were not practical (EGFR mutations). Running out of options, they then turned to microRNAs. First they made use of microRNA target prediction software to discover two microRNAs that might target the EGF receptor mRNA. They then hypothesized that in cancers where EGFR had been aberrantly activated one of these microRNAs was lost. As an interesting technical aspect, unlike most other microRNA diagnostics currently being developed by Asuragen, Rosetta Genomics, Exiqon and others, they did not make use of qRT-PCR or microRNA microarray for the detection of microRNA expression levels, but the even simpler pPCR from DNA isolated from tumor tissue microdissected from formalin-fixed samples.

Amazingly, from a sample of 60 tumors of the lung, the miR-128b gene dosage was able to nicely segregate those patients that responded to TKI treatment versus those that did not. None of the other methods tested came anywhere close. As, if not even more spectacularly, it turns out that miR-128b is located on a region of the short arm of chromosome 3 that had been previously implicated and was frequently deleted in lung cancer (around 90% of late-stage lung cancers), but of which the identity of the underlying gene had long been a mystery.

In another elegant study, her group employed mRNA microarray analysis to look for genes either over- or under-expressed in a panel of cancer cell lines. Once identified, microRNA target prediction was used to come up with potential clusters of genes being targeted by the same microRNA. Demonstrating yet another promising method of discovering microRNA diagnostic candidates, 5 out of the top 7 overexpressed genes shared a common microRNA signature, and sure enough, this microRNA alone was more predictive of drug responsiveness than any of the proteins alone.

Eric Lader from the host Qiagen rounded up the session on microRNAs with an overview of products for the quantitation and inhibition of microRNAs sold by Qiagen. Qiagen’s qRT-PCR quantitation method allows for the one-step conversion of all microRNAs into cDNA so that many individual microRNAs can be measured from this one cDNA sample thereby reducing the use of precious clinical material. As to the microRNA inhibition technology, a number of nucleotide modifications were tested in the antagomirs and it was interesting that while a number of them were able to inhibit microRNA function, depending on the modification, they probably did so at various steps of microRNA biogenesis and function and included both microRNA degradation and competitive binding. LNAs and other chemistries with high melting temperature generally worked best while RNase H-dependent antisense did not appear to be an effective microRNA inhibition technology.

In my next post, I will summarize the rest of the meeting including more RNAi screens, cancer, and systems biology. Stay tuned…