Abstract
The recent article by Damle et al. addresses the important question of potential off-target effects of PS antisense oligonucleotides (ASOs). I commend the authors for addressing this complex and important topic and particularly for the thorough dose-response studies and analyses and the efforts to enhance quality control. Their study is a step in the right direction. That said, I think that understanding the risk of hybridization-based off-target effects of PS gapmer ASOs designed to reduce target RNAs via RNase H1 cleavage is actually quite complex and experimentally challenging, and to fully understand that risk and the factors that might influence it will require a great deal more work and quite a few more types of controls than evident in the Damle et.al article.
Confounding Issues and Potential Solutions
More precisely define the objectives of the study
Based on the focus of the article, I assume that the authors want to better understand the potential of PS gapmer ASOs to cause hybridization-based off-target reduction and the factors that alter the risk of hybridization-based off-target effects. Though one could argue that irrespective of molecular mechanism, off-target effects of an ASO reduce the selectivity of effect and potentially reduce the therapeutic index, understanding the molecular mechanisms of each type of off-target effect is important because potential solutions are likely to vary as a function of molecular mechanism.
Differentiation of the direct hybridization-based off-target effects from secondary effects and compensatory changes in the transcriptome
As is the case for all pharmacological agents, PS ASOs can result in direct pharmacological effects, effects that are secondary to alteration of the pharmacological target, or effects that represent compensatory cellular responses attempting to maintain homeostasis in response to the pharmacological effects of the ASO. Since, by definition, direct reduction of an off-target must precede secondary or compensatory changes in the transcriptome, this confounding issue can be resolved to a large extent by evaluated the timing of transcriptional changes by time-course studies.
Differentiation of transcriptomic changes caused by true hybridization-based off-target reduction from transcriptional changes due to cytotoxic effects or innate immune activation
PS ASOs can induce changes in the transcriptome that are secondary to cytotoxic or immune stimulatory effects. For example, apoptosis causes a rapid reduction in many transcripts 1 and innate immune activation causes many changes in the transcriptome that can vary as innate immunity is actively terminated. 2 The molecular mechanisms by which some PS ASO cause cytotoxicity and innate immune activation are known.3,4 Both mechanisms are secondary to binding to proteins, and both cytotoxic and innate immune effects are influenced by PS content, the lipophilicity of the 2′modifications of the ASO, and ASO sequences. Consequently, either can result in transcriptomic effects that may be mistaken for hybridization-based off-target effects.
To evaluate the potential contributions to the observed changes in the transcriptome, investigators can add known cytotoxic PS ASOs and innate immune stimulatory PS ASOs of the same chemical class and assess the effects of those PS ASOs on the transcriptome of the cell line being studied. The transcriptional effects caused by known cytotoxic and immune stimulatory PS ASOs can then be used to “weed out” transcriptomic effects caused by cytotoxicity or innate immune effects, and if any dose used in the study proved to result in transcriptomic changes significantly influence cytotoxicity or innate immune activation, the results of those doses can be eliminated from consideration.
The activity of RNase H1 is influenced by RNA sequence and RNA modifications
As the event actually being evaluated is, in fact, the cleavage of RNA by RNase H1, it is essential to understand the enzymology and interactome of the enzyme. Fortunately, a great deal is known about the structure, enzymological properties, and interactome of RNase H1.5–9 Since the number of cleavage sites and rates of site-specific cleavage by RNase H1 are affected by the sequence of the hetero-duplex formed by the PS gapmer-ASO and the RNA, it quite difficult to predict the likelihood of cleavage or the rate of cleavage at any site in a target or off-target RNA. Further, the most common RNA modification, 2′methoxy, prevents H1 cleavage at the nucleotide so modified and reduces or ablates cleavages at nearby sites; assessing the impact of a mismatch or bulge is challenging unless one understands the sites of cleavage in the target sequence and the nucleotides that are modified in some way.
Given the above challenges, I would encourage caution in drawing conclusions about how different mismatches, bulges, etc. may contribute to off-target cleavage. Certainly, I would not limit thinking to the RNA-ASO structure without considering the characteristics of RNase H1.
Additional Suggestions
I would group that suggestions under the concept of taking advantage of all that is known and the many tools created.
Fortunately, site-directed mutagenesis of RNase H1 has already created a group of well-characterized mutant RNase H1s, overexpression vectors, an understanding of the interactome and the effects of certain members of the RNase H1 interactome on RNase H1 activity, and viable RNase H1 knockout mice, and it is known that RNase H1 levels are rate- and effect-limiting with regard to PS gapmer ASO-induced-reduction of RNAs. 10 Unfortunately, it has been quite difficult to establish viable RNase H1 knock-out cell lines.5,11
To confirm that a transcript that is reduced is secondary to hybridization-based off-target effects, one could use one of the inducible systems to overexpress RNase H1. If the transcript is, in fact, reduced by hybridization-based off-target effects, the IC50 should be reduced and the maximum reduction increased. Alternatively, one could overexpress a catalytically disabled RNase H1, which should act as a dominant negative and reduce effects on the putative off-target. One could also reduce either NAT10 or DDX21 as both are known to enhance RNase H1 cleavage. Reduction of either should lower the effects of the ASO if the effects of the putative hybridization-based off-target are, in fact, the cause of the transcriptomic change. 8
Conclusions
I fully agree with the authors that advancing our understanding of the risks of hybridization based off-target reduction and characterizing the factors that alter the potential for these off-target effects is an important exercise. Though the Damle et al. article is a step in the right direction, much more work is needed to achieve this goal. Fortunately, the knowledge and tools exist to thoroughly answer this important question.
Footnotes
Author Disclosure Statement
S.T.C. is the founder, chairman, and CEO of n-Lorem Foundation, a nonprofit foundation devoted to meeting the needs of nano-rare patients. S.T.C. holds founder stock in Ionis Pharmaceuticals.
Funding Information
No funding was received for this article.