Innate Immune Response and Off-Target Mis-splicing Are Common Morpholino-Induced Side Effects in Xenopus

摘要

class="head no_bottom_margin" id="sec1title">IntroductionPerturbing the function of a gene of interest in order to infer its biological role is a common approach in many biological disciplines including embryology and physiology. Since forward and reverse genetic approaches have not been readily applicable to many organisms, the injection of morpholino oligomers (MOs) has been widely adopted instead. This antisense technology is based on a nucleic acid analog with a backbone of morpholine rather than deoxyribose rings, and neutral phosphorodiamidate instead of negatively charged phosphate linkages. According to the manufacturer, MOs are more stable, efficient, and specific in knocking down genes than alternative knockdown (KD) technologies such as short interfering RNA, mainly because of their neutral features that prevent electrostatic interactions with endogenous proteins at physiological pH (). Depending on their design, MOs can block either translation or splicing when hybridizing almost perfectly over their full length of 23–25 bases to the translation start site or splice acceptor or donor site. In addition, they can disrupt the activity of non-coding RNA species, such as microRNAs (miRNAs) or Y-RNAs (, ).While both anecdotal evidence and a number of published studies have suggested that MOs can cause off-target effects, it was previously assumed that, if appropriate control experiments are performed, robust knockout (KO)-like phenotypes could be generated. However, the use of TALEN and type II CRISPR genome editing technologies has now brought this assumption into question. Such studies have found that morphant and mutant phenotypes can be significantly different even when the morphant phenotype can be rescued by the introduction of cognate RNA species (). While in some cases it is likely that the morphant phenotype is an off-target effect, in others it may be that the genetic mutation does not result in a complete loss of function either because it gives rise to a hypomorphic allele or because the gene product is maternally inherited (). Mutating a gene can also in some circumstances lead to genetic compensation that is not triggered by MO-mediated KD ().Nevertheless, bearing in mind that MOs can replicate corresponding null phenotypes at least at a morphological level, their use in vertebrate embryos may be legitimate and advantageous for several reasons. First, some maternal-effect genes are difficult to study due to an essential function in later life stages, and cumbersome germline-specific KO strategies are required to produce viable females with homozygous KO eggs (). In particular, rapid-turnover proteins translated from maternal transcripts can be efficiently depleted with the injection of a translation-blocking MO into the zygote. In contrast, splice-blocking MOs will only perturb zygotic protein synthesis. Second, the simultaneous KD of multiple genes can reveal functional redundancies (, , ), which can be informative for future KO strategies. However, this may require more MOs being injected into the embryo, thus increasing the likelihood of off-target effects. Third, MOs can be introduced into different mutant and transgenic backgrounds either to screen for genetic interactions or to help characterizing phenotypes, without the need for time-consuming intercrossing. Fourth, MO-injected embryos do not need genotyping, so that large numbers can be collected. This is of particular importance given the expanding use of molecular profiling to analyze loss-of-function effects at the chromatin level. However, this kind of differential analysis assumes that KD and KO animals with identical macro- or microscopic defects, such as in morphology or behavior, share similarly derailed genomic readouts. We examined this hypothesis in the western clawed frog Xenopus tropicalis by generating Brachyury null mutants using TALENs and comparing these with corresponding, previously validated morphants () at a transcriptome-wide level.Our results showed that, while depletion of Brachyury resulted in the same dramatic loss of posterior mesoderm regardless of the gene interference technology employed, only control and Brachyury-targeting MOs perturbed hundreds of splicing events and caused excessive immune response-related gene transcription. These MO side effects were caused, on the one hand, by the off-target binding of premature transcripts and, on the other hand, by a cell-intrinsic (innate) immune reaction. The latter strongly correlated with the guanine-cytosine (GC) content of the injected MO. Proper dose and design optimization of the injected MO can mitigate these inadvertent effects. However, some specific off-target effects could not be eliminated even when an elevated incubation temperature was used in an effort to increase hybridization stringency. This is further corroborated by the kinetic analysis of MO oligomers binding off-target RNA sequences far below the minimal concentration required to produce a t/t2 KO-like phenotype. We expect that our findings will be critical to keep unintended disruptions in tissue and organ development to a minimum.