This is the second part of the three-part series in which Tobias Wolfram and I are looking at the scientific merits of the first three notable RNAi Therapeutics candidates that have entered the clinic for solid cancer indications. Solid cancer is arguably the one therapeutic area where RNAi Therapeutics currently have highest value because of a) the ability to target many of the important, but otherwise undruggable cancer-related genes; b) systemic delivery technologies that ought to be able to reach solid cancer tissues; and c) a patient population that is essentially indifferent to the exact mode of administration (e.g. oral vs intravenous infusion).
In Part I of our series we looked at Calando’s CALAA-01. We came out slightly disappointed with that program because it seemed stuck in the quite early, non-biological/physico-chemical characterization stages and did not provide too much bona fide evidence of anti-cancer activity in pre-clinical animal models. Neither was Calando’s RONDEL delivery technology convincingly shown to facilitate in vivo gene silencing in the first place. We are therefore relieved, and I personally also somewhat pleasantly surprised at the same time, to be able to report a considerably better validated cancer RNAi Therapeutics candidate in Silence Therapeutics’ Atu-027. While the precise theory behind Atu-027 remains somewhat uncertain, we are encouraged by the scientific rigor with which the target had been validated as a component of the well known PI3-kinase pathway and the careful demonstration that lipoplex-siRNA delivery can be used for knocking down genes in the vascular endothelia from mice to monkeys.
The Target: PKN3, a Downstream Component of the Cancer-related PI3K pathway
The precursor of Silence Therapeutics, Atugen AG, was a child of the millennium genomics era involved in gene discovery and target validation. The technological workhorses underlying this work were a combination of gene knockdown approaches such as ribozymes, antisense (‘GeneBlocs’), expressed RNAi that were combined with RNA expression analysis and proteomics.
For some reason that may be related to the prior training of Atugen’s scientific staff, Atugen specialized in studying the genetic components of the PI3-kinase pathway. This pathway has attracted wide attention because it plays important roles in various biological processes and has consequently been implicated in numerous diseases, particularly in cancer where a number of genetic abnormalities that affect a range of phenotypes ranging from cell growth to metastasis have been linked to it. While clearly an attractive target for cancer therapy, and there has been some success in drug development in that regard, its many roles in molecular biology means that targeting the pathway upstream, for example at the PI3 kinase level, is likely to be associated with a poor therapeutic index. Atugen therefore reasoned that discovering targets downstream of PI3K could result in equally potent, but also more specific therapeutics for cancer.
Applying their genomic tool kit that placed a high priority on specificity controls, they came up with a handful of candidate target genes that they then nicely validated in subsequent in vitro and in vivo tumor-related assays. From this, PKN3 eventually emerged as a gene whose expression and phosphorylation was not only PI3 kinase-dependent, but when knocked down also abolished the ability of PI3 kinase-activated cancer cells to grow under culture conditions thought to reflect the natural tumor environment. When such cancer cells were modified with a DNA-directed RNAi construct against PKN3 and then introduced into mice in an orthotopic prostate tumor model, the primary tumor formation was, somewhat unexpectedly, not much affected whereas metastasis formation was significantly inhibited. Thus, the authors speculated that it is PI3 kinase-dependent functions like cell motility and/or cell-cell/cell-matrix contacts that are disturbed following PKN3 knockdown, and that the substrate-dependent growth defect observed in vitro may reflect deficiencies in tissue invasion that are shared with the facilitate metastasis in vivo. Consistent with a role for PKN3 in real-life cancers, tissue immunohistochemistry from prostate cancer patients revealed PKN3 to be upregulated in the cancer, but not surrounding normal tissue.
The theory of why PKN3 RNAi might have utility as a cancer treatment had to be modified when it was found that the delivery system did not target the cancer cells in which the initial target validation experiments had been carried out, but instead endothelial cells of the blood vasculature. The late change in the theoretical framework for Atu-027 that occurred in the otherwise very impressive 2009 landmark Atu-027 mouse-to-monkey validation study (Aleku et al., 2009), is probably the major weakness in Atu-027’s rationale. The question arises why take such a target risk at all when there are many other well-validated, but undruggable cancer targets already out there to choose from? Is PKN3 a suitable cancer target after all, and, even if it were, are they knocking it down in the right cell types? It is here that the relative lack of knowledge on the biology of PKN3 (all the functional studies on this gene were conducted by Silence Therapeutic) turns out to be a disadvantage and may complicate things like the development of biomarker tests to monitor early clinical efficacy.
On the other hand, the fact that PKN3 emerged after having considered a range of targets, suggests that based on practical experience Silence had highest confidence in the effectiveness of PKN3 as a cancer target. This argument is supported by what Tobias and I felt to be very well controlled target validation experiments, including the use of multiple positive and negative control oligonucleotides to confirm that the observed phenotypes were on-target. In that regard, I would score the relevant dataset generated by Silence in the top 5% of the RNAi literature. Maybe not surprising because of Silence's roots in gene discovery.
Being the only one working on a gene, however, has a few benefits, too. As such, Silence has made inroads in obtaining broad patent protection for the use of PKN3 as a target in PI3 kinase-related diseases. Given the large interest in the PI3 kinase pathway, this could be a valuable asset to the company and somewhat further justifies what otherwise would look like a disproportionate investment for a platform company into a single pathway!
The RNAi Trigger: A 23 base-pair Atu-siRNA
Although Atugen’s background was more in ribozymes and antisense, it decided to make drugs based on synthetic siRNAs after its transformation into a drug development company. This was around the time that RNAi had just been discovered in humans by Tuschl. In the case of Atu-027 this meant that while antisense, expressed RNAi, and small molecules acting on the PI3 kinase pathway were used for target validation, the active ingredient would be a 23 base-pair blunt-end siRNA in which alternating 2’-O-methyl nucleotides would face unmodified residues, a pattern that also coincides with the eventual scope of the issued Atu-siRNA patents in the US and Europe.
Since siRNAs may have anti-tumor activity independent of RNAi, the result of non-specific innate immunostimulation by some siRNAs, it is notable that such extensive 2’-O-methylation was shown only later in the field to result in siRNAs with no or very little innate immunostimulatory potential. This not only lends credibility to their pre-clinical efficacy data, the choice of 2’-O-methylation at such an early stage can therefore be more generally regarded as a very lucky one considering that the scope of the Atu-siRNA patents would eventually only encompass 2’-O-methylations. Speaking of RNAi trigger IP, a strong Tuschl II would actually benefit the blunt-end Atu-027, while a strong 21-23 base-pair/nucleotide covering Tuschl I could cause problems.
Lipoplex-siRNA Delivery to Endothelial Cells of Blood Vasculature
The delivery system for Atu-027 was developed in parallel to identifying the target siRNA. It is a pegylated lipoplex-siRNA system where the siRNA sits on the outside of positively charged, peylated liposomes ('Atuplex'), instead of the more commonly used SNALP-like liposomes where the siRNA is encapsulated inside. Because of this fundamental difference, the two systems vary considerably in their biological properties. Whereas SNALPs deliver their cargo mainly to the liver, but can be further stabilized to reach tissues beyond that, lipoplex-siRNAs have a preference for the endothelia lining the inside of blood vessels pretty much throughout the body.
In careful labeling studies, Silence Therapeutics scientists were able to show convincingly that they can achieve bona fide RNAi knockdown in the 50-70% range when administering around 1mg/kg- from mice to monkeys. Moreover, the invariable pharmacokinetics following repeat administration support that there is little or no cytokine activation. It is curious that most of their in vivo studies involve repeat administration. This is in contrast to what for example Tekmira and Alnylam are practicing which carefully study both single and repeat administration both of which together should inform much better on eventual dosing schedules, causes for toxicities etc. Similarly, only one cationic lipid, AtuFECT01, is used throughout the years, and a relatively simple opportunity to improve the therapeutic index has probably been missed. There is also a lack of a detailed description of the precise formulation process and early-stage physico-chemical characterization, although part of this may be for competitive reasons.
Apparently limiting themselves to delivery systems developed in-house, the performance of the lipoplex-siRNA technology means that in order for Silence to pursue their long-standing interest in cancer, they have to address the disease where it interfaces with blood vessel endothelia. Importantly, the concept of starving cancer cells by inhibiting neoangiogenesis is well accepted and one only has to look at the commercial success of Roche/Genentech’s anti-VEGF monoclonal antibody Avastin. The more pressing task therefore is to find the right target genes.
Pre-clinical safety and efficacy
The lipoplex-siRNA formulations were generally well tolerated at dosages predicted to be therapeutic. No significant differences in body weights were observed with various siRNAs. Immune responses were also not detected. This is particularly important in studies like these as it further suggests that the observed anticancer activity was not due to an off-target immune-related mechanism. Further supporting such a lack of immune responses was, as noted above, the fact that the pharmacokinetics did not change with repeat administration. Immune reactions would have likely facilitated antibody formation against the PEG component and caused rapid clearance of the lipoplex-siRNA upon repeat administration. I originally had been skeptical of Silence’s claims that there were no immune issues since they never seemed to have paid much attention to this issue, but they may have just gotten lucky in their early choice of extensive 2’-O-methylation as part of the Atu-siRNA architecture.
There were, however, some toxicities noted that mostly involved the liver and included 2-3 fold minor elevations in liver enzymes with an PTEN siRNA (this could be an siRNA-specific response different from Atu-027) and an apparent increase in DNA replication in livers of mice treated with control lipids alone which may be indicative of liver cell damage (Santel et al.,2006). Liver safety and immune reactions may therefore be the two expected dose-limiting adverse events for Atu-027.
In terms of efficacy, Atu-027 had ED50s in mice and monkeys of 1 and 0.3mg/kg using liver and lung endothelia as surrogate tissues. Since the samples include non-endothelial contaminants that could express some PKN3, this may actually somewhat underestimate the true extent of the knockdown in the target endothelial cells. This knockdown efficacy translated into a 60% reduction of the primary tumor in an orthotopic mouse prostate cancer model, and an even greater 80% reduction in the lymph node metastases. It should be noted that while the effect on the primary tumor is to be welcome, the results somewhat differ from those in the initial characterization of PKN3 as a target 5 years earlier where an effect was only observed for the metastases (Leenders et al., 2004). The anti-tumor response was confirmed in a range of other tumor models. Taken together, the safety and efficacy profile of Atu-027 are in support of initiating phase I studies.
Target and Delivery: A match made in heaven, or marriage of necessity?
As alluded to in the target validation segment, we are somewhat concerned about target risk since, as a result of the emerging properties of the lipoplex-siRNA system and in vivo PKN3 knockdown phenotypes, the rationale for PKN3 as an anticancer target has undergone multiple transformations, first validated as playing an important role in the cancer cells itself, but then thought to act directly on blood vessel and eventually lymph vessel formation.
Cynics could say that once they got stuck with the lipoplex-siRNA system, they would have done anything to force the model to fit their data. Realizing this dilemma, they did, however, undertake bridging studies and were for example able to show that PKN3 knockdown in isolated (non-cancer) vascular and lymphatic endothelial cells disturb well known endothelial phenotypes such as ability to form tubes in tissue culture.
Despite these twists and turns, from a 30,000 foot level, PKN3, due to its well-validated role in the PI3 kinase pathway, does look like a credible anticancer target in endothelial cells. Are there potentially targets that are better suited for the lipoplex siRNA system? Most likely yes, as it ought to be a general rule in RNAi Therapeutics that ‘target follows delivery’. Thus armed with knowledge about the ‘Atuplex’ delivery system, the next target-delivery combination by Silence has a better chance of looking like a match made in heaven.
In the end, we came to the conclusion that, theoretical concerns notwithstanding, by targeting a skillfully validated PI3K pathway gene with a delivery system that facilitated bona fide gene knockdown from mice to monkeys and the clear anticancer phenotypes observed in a number of rodent cancer models, Atu-027 is a promising clinical candidate for which the pre-clinical proof was simply in the pudding. Given the pioneering and innovative nature of this particular approach, it should also be able to draw the attention of the likes of Novartis and Pfizer that should be able and willing to take such risks as they move into RNAi Therapeutics.
Atu-027 may also be viewed as the start of a franchise that focuses on attacking cancer from its endothelial side with all the unique advantages of RNAi, for example the almost unlimited choice of targets. The merger with Intradigm, which also has developed expertise in neoangiogenesis and cancer, should add depth to this by, most importantly, expanding the pool of potential targets and pathways so that potentially less risky target-delivery matches can be found. There should also be ample scope for improving the Atuplex delivery system as it is questionable whether the first cationic lipid that they published and held onto, Atu-FECT01, is really the non plus ultra of lipid chemistry.
Meanwhile, Atu-027 has started clinical development half a year ago with a multi-dose phase I study in patients with advanced solid tumors. First results are eagerly awaited.