Yoshiki Sasai

Yoshiki Sasai: stem cell Sensei

Gary cropped

Accelerated Frogs: Developmental Biology meets Particle Physics


Interview with SDB poster winner Niteace Whittington

euroevodevo logo

An evolving field- notes from the European Evo Devo meeting

Out with the old, in with the new: reassessing morpholino knockdowns in light of genome editing technology

Posted by on August 6th, 2014

This Spotlight article was written by Stefan Schulte-Merker and Didier Y. R. Stainier, and was first published in Development.


Morpholino oligomers have been used widely and for many years in the zebrafish community to transiently knock down the function of target genes. It has often been difficult, however, to reliably discriminate between specific and non-specific effects, and thus generally accepted guidelines to control for morpholino side effects do not exist. In light of recent methodologies to generate mutant lines in virtually any zebrafish gene, we discuss these different approaches with a specific focus on how the first description of a loss-of-function phenotype in zebrafish should be accomplished.


Initially, the genetic analysis of zebrafish development and physiology was dominated by mutants identified in small- and large-scale forward genetic screens (Chakrabarti et al., 1983; Driever et al., 1996; Haffter et al., 1996).Whereas forward genetics was instrumental in establishing zebrafish as an additional vertebrate model system, progress was hampered by the fact that there was no reliable technology to carry out reverse genetics in this model.TILLING (a reverse genetics approach based on high-throughput sequencing of ENU-mutagenized fish)was introduced only in 2002 (Wienholds et al., 2002) and requires considerable up-front investment in logistics and infrastructure. The introduction of morpholinos (MOs) in frogs (Heasman et al., 2000) and zebrafish (Nasevicius and Ekker, 2000) as an antisense reagent to transiently knock down gene function was therefore greeted with considerable excitement, as it appeared to fill a real void in the toolbox. Since its inception, countless studies using this technology have been published, including some using MOs to knock down maternally deposited transcripts to circumvent the generation of maternal-zygotic mutants, and others using caged MOs, which allow for inducible release of these antisense reagents (Shestopalov et al., 2012).

The MO antisense technology is based on nucleic acid bases that are linked tomorpholine rings and a non-charged phosphorodiamidate backbone. The rationale for this design was that MOs would not bind electrostatically to protein, hence causing less toxicity, while at the same time being resistant to nucleases (Summerton, 2007). MOs are injected into early zebrafish embryos using standard techniques. Commonly, they are ∼25-mers designed to be an exact antisense match against the region surrounding the first translated ATG (to block translation) or against a splice donor or acceptor site (to interfere with precursor mRNA splicing). It quickly became apparent that some MOs could work extremely well, and there are many MO phenotypes that efficiently mimic mutant phenotypes without any noticeable side effects. However, it has also become clear that MOs can lead to artifacts and that for many MOs the phenotypes caused by specific binding to the intended target RNA are difficult to separate from those caused by the non-specific binding to unintended targets (Eisen and Smith, 2008). In fact, a simple calculation suggests that binding to targets other than the intended precursor or mature mRNA is likely. A zebrafish embryo contains ∼500 ng of RNA, 2-5% of which is translatable (25 ng) (A. Giraldez, personal communication; see also Davidson, 1986). Assuming that at any given time there are more than 104 different mRNA species present in a cell (Davidson, 1986), and that those transcripts are equally represented among the 25 ng of mRNA, only 2.5 pg of a specific mRNA species is available for targeting. Injections typically deliver ∼1 ng of MO, often more. Assuming further that the target mRNA has an average length of 1.25 kb, whereas the MO is a 25-mer, this equates to a 2×104-fold molar excess of MO versus target mRNA. It is therefore most likely that this vast excess of MO will bind other RNA or other macromolecules. This situation would not be such a serious problem if there were reliable ways to distinguish specific from nonspecific effects. However, this is not the case, and one can at best only show that MOs affect the target sequence; non-specific effects cannot easily be identified, even when using mRNA rescues (see Del Giacco et al., 2010; Tao et al., 2011). The literature now contains several examples in which developmental delay, defects in organ asymmetry and pericardial edema (among many other ‘phenotypes’) are attributed to knocking down a specific gene, but in which subsequent generation of a mutation in that gene revealed a very different phenotype, and often no phenotype at all. Recent examples include mutations in sox18, nr2f1a and prox1a/b, all genes that had been reported to show morphant phenotypes within the lymphatic vasculature, whereas the mutant alleles do not (van Impel et al., 2014).

In several cases it has been possible to circumvent some of the non-specific phenotypes by suppressing p53 activity (Robu et al., 2007), which can reduce the ectopic cell death caused by nonspecific MO effects; however, this approach has its own caveats as it effectively generates a phenotype not on a wild-type, but on a p53-deficient, background. Applying such drastic corrective measures to allow a phenotypic analysis raises a number of questions that cannot be easily addressed.

Two surprisingly efficient alternatives for reverse genetics have been recently implemented in zebrafish (Chang et al. 2013; Huang et al., 2011; Sander et al., 2011; Hwang et al., 2013; Zu et al., 2013) and other organisms (Beumer et al., 2008; Tesson et al., 2011; Yang et al., 2013). TALE nucleases and the Crispr/Cas9 system are very efficient at generating mutations. As both techniques have been
reviewed extensively (Auer and Del Bene, 2014), we will restrict the discussion here to comparing the principles of the MO and TALEN/ Crispr approaches.

First, it should be noted that – like MOs – the implementation of these new technologies can be carried out in virtually any lab. Whereas TILLING really only makes sense for those willing to analyze large numbers of samples and genes, TALENs and Crispr do not require a substantial investment and, once established, can be used to generate targeting constructs within 1 (Crispr) to 2 (TALEN) weeks.

Second, as TALENs and Crisprs affect genomic DNA, rather than RNA transcripts, their molecular effect can be determined at the single embryo level (which is more difficult with MOs) to obtain a clear phenotype/genotype correlation. Of course, such an approach requires caution, as these nuclease-injected embryos are most likely to be mosaic for the resulting genomic lesions. Furthermore, TALEN and Crispr constructs can sometimes be efficient enough to generate loss-of-function situations in the actual injected embryos (Dahlem et al., 2012), and so, in a minority of cases, can be used almost like an MO; injection, scoring for phenotypes and confirming that the nuclease works efficiently can be performed within a few days.

Third, the published evidence, although currently limited, suggests that the side effects of these nucleases are often negligible (Hruscha et al., 2013), even though both TALE and Crispr-Cas nucleases can bind and cleave off-target loci (Reyon et al., 2012; Fu et al., 2014). When additional mutations are introduced, they can usually be segregated away from the mutation of interest by one or two outcrosses (as with mutations identified in ENU mutagenesis screens). This specificity is of course a tremendous advantage, and very different from MOs: an MO that binds non-specifically will most likely do so in every injected embryo. Lastly, it is relatively easy to generate multiple mutant alleles in one gene (e.g. by using TALEN pairs that affect different regions of the targeted gene), thus further reducing the chance of being misled by off-target mutations.

Hence, it seems fair to say that within the last year or so, the landscape of reverse genetics in zebrafish has changed, and it has changed for the better. Anyone can now, within a few weeks, generate reagents that can be used for reverse genetic experiments that appear to be of superior reliability and that are less burdened with side effects compared with MOs. Does that mean that we should do away with MOs altogether? Not necessarily: as we pointed out in the first paragraph, there are many MOs that are very useful and that appear to work specifically. We know they work specifically because we can compare them with a mutant phenotype. We would argue that this is the criterion that should be used in most cases: if one can show that an MO phenotype is an exact replicate of a mutant phenotype, then use of this MO is certainly acceptable and can save valuable time; for example, for injection into transgenic lines or for generating ‘double mutants’. However, the description of a phenotype that is provided for the first time and that is based solely on MOs without the ability to compare with a genetic mutant, should in the future be viewed very critically. In most cases, there are better alternatives in the form of nuclease based targeted approaches and there is no good reason not to use them.


Auer, T. O. and Del Bene, F. (2014). CRISPR/Cas9 and TALEN-mediated knock-in approaches in zebrafish. Methods pii: S1046-2023(14)00129-7 (in press).

Beumer, K. J., Trautman, J. K., Bozas, A., Liu, J.-L., Rutter, J., Gall, J. G. and Carroll, D. (2008). Efficient gene targeting in Drosophila by direct embryo injection with zinc-finger nucleases. Proc. Natl. Acad. Sci. USA 105, 19821-19826.

Chakrabarti, S., Streisinger, G., Singer, F. and Walker, C. (1983). Frequency of gamma-ray induced specific locus and recessive lethal mutations in mature germ cells of the zebrafish, Brachydanio rerio. Genetics 103, 109-123.

Chang, N., Sun, C., Gao, L., Zhu, D., Xu, X., Zhu, X., Xiong, J.-W. and Xi, J. J. (2013). Genome editing with RNA-guided Cas9 nuclease in zebrafish embryos. Cell Res. 23, 465-472.

Dahlem, T. J., Hoshijima, K., Jurynec, M. J., Gunther, D., Starker, C. G., Locke, A. S., Weis, A. M., Voytas, D. F. and Grunwald, D. J. (2012).  Simple methods for generating and detecting locus-specific mutations induced with TALENs in the zebrafish genome. PLoS Genet. 8, e1002861. Davidson, E. (1986). Gene Activity in Early Development. Waltham: Academic Press.

Del Giacco, L., Pistocchi, A. and Ghilardi, A. (2010). prox1b Activity is essential in zebrafish lymphangiogenesis. PLoS ONE 5, e13170.

Driever, W., Solnica-Krezel, L., Schier, A. F., Neuhauss, S. C., Malicki, J., Stemple, D. L., Stainier, D. Y., Zwartkruis, F., Abdelilah, S., Rangini, Z. et al. (1996). A genetic screen for mutations affecting embryogenesis in zebrafish. Development 123, 37-46.

Eisen, J. S. and Smith, J. C. (2008). Controlling morpholino experiments: don’t stop making antisense. Development 135, 1735-1743.

Fu, Y., Sander, J. D., Reyon, D., Cascio, V. M. and Joung, J. K. (2014). Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nat. Biotechnol. 32, 279-284.

Haffter, P., Granato, M., Brand, M., Mullins, M. C., Hammerschmidt, M., Kane, D. A., Odenthal, J., van Eeden, F. J., Jiang, Y. J., Heisenberg, C. P. et al. (1996).

The identification of genes with unique and essential functions in the development of the zebrafish, Danio rerio. Development 123, 1-36.

Heasman, J., Kofron, M. and Wylie, C. (2000). Beta-catenin signaling activity dissected in the early Xenopus embryo: a novel antisense approach. Dev. Biol. 222, 124-134.

Hruscha, A., Krawitz, P., Rechenberg, A., Heinrich, V., Hecht, J., Haass, C. and Schmid, B. (2013). Efficient CRISPR/Cas9 genome editing with low off-target effects in zebrafish. Development 140, 4982-4987.

Huang, P., Xiao, A., Zhou, M., Zhu, Z., Lin, S. and Zhang, B. (2011). Heritable gene targeting in zebrafish using customized TALENs. Nat. Biotechnol. 29, 699-700.

Hwang, W. Y., Fu, Y., Reyon, D., Maeder, M. L., Tsai, S. Q., Sander, J. D., Peterson, R. T., Yeh, J.-R. J. and Joung, J. K. (2013). Efficient genome editing in zebrafish using a CRISPR-Cas system. Nat. Biotechnol. 31, 227-229.

Nasevicius, A. and Ekker, S. C. (2000). Effective targeted gene ‘knockdown’ in zebrafish. Nat. Genet. 26, 216-220.

Reyon, D., Tsai, S. Q., Khayter, C., Foden, J. A., Sander, J. D. and Joung, J. K. (2012). FLASH assembly of TALENs for high-throughput genome editing. Nat. Biotechnol. 30, 460-465.

Robu, M. E., Larson, J. D., Nasevicius, A., Beiraghi, S., Brenner, C., Farber, S. A. and Ekker, S. C. (2007). p53 activation by knockdown technologies. PLoS Genet. 3, e78.

Sander, J. D., Cade, L., Khayter, C., Reyon, D., Peterson, R. T., Joung, J. K. and Yeh, J.-R. J. (2011). Targeted gene disruption in somatic zebrafish cells using engineered TALENs. Nat. Biotechnol. 29, 697-698.

Shestopalov, I. A., Pitt, C. L. and Chen, J. K. (2012). Spatiotemporal resolution of the Ntla transcriptome in axial mesoderm development. Nat. Chem. Biol. 8, 270-276.

Summerton, J. E. (2007). Morpholino, siRNA, and S-DNA compared: impact of structure and mechanism of action on off-target effects and sequence specificity. Curr. Top. Med. Chem. 7, 651-660.

Tao, S., Witte, M., Bryson-Richardson, R. J., Currie, P. D., Hogan, B. M. and Schulte-Merker, S. (2011). Zebrafish prox1b mutants develop a lymphatic vasculature, and prox1b does not specifically mark lymphatic endothelial cells. PLoS ONE 6, e28934.

Tesson, L., Usal, C., Ménoret, S., Leung, E., Niles, B. J., Remy, S., Santiago, Y., Vincent, A. I., Meng, X., Zhang, L. et al. (2011). Knockout rats generated by embryo microinjection of TALENs. Nat. Biotechnol. 29, 695-696.

van Impel, A., Zhao, Z., Hermkens, D. M. A., Roukens, M. G., Fischer, J. C., Peterson-Maduro, J., Duckers, H., Ober, E. A., Ingham, P. W. and Schulte- Merker, S. (2014). Divergence of zebrafish and mouse lymphatic cell fate specification pathways. Development 141, 1228-1238.

Wienholds, E., Schulte-Merker, S., Walderich, B. and Plasterk, R. H. A. (2002). Target-selected inactivation of the zebrafish rag1 gene. Science 297, 99-102.

Yang, H., Wang, H., Shivalila, C. S., Cheng, A. W., Shi, L. and Jaenisch, R. (2013). One-step generation of mice carrying reporter and conditional alleles by CRISPR/Cas-mediated genome engineering. Cell 154, 1370-1379.

Zu, Y., Tong, X.,Wang, Z., Liu, D., Pan, R., Li, Z., Hu, Y., Luo, Z., Huang, P., Wu, Q. et al. (2013). TALEN-mediated precise genome modification by homologous recombination in zebrafish. Nat. Methods 10, 329-331
GD Star Rating

Tags , , ,
Category Discussion | 6 Comments »

  1. We’d be really interested to hear reaction and feedback from the community on this opinion piece; please leave any thoughts you’d like to share here in the Comments section.

    Rate comment: Thumb up 0 Thumb down 0

  2. Richard Maguire says:
    First a disclaimer: I don’t particularly like morpholinos. Frankly, I don’t know anyone who does…

    However, I think the approach outlined in the last paragraph sounds reasonable on the face of it, but should be toned down a little. The main reason being that of where to draw the line. For instance, if an MO phenotype generally recapitulates the mutant but someone describes a phenomenon in a particular cell population only in either the MO or induced mutant, would this be bounced in review? I can see the lack of corroborating mutant data being used as a rather blunt tool during peer review to bounce papers, along the lines of “This study is good preliminary data, but limited in scope”.

    It’s not as if mutants and knock-ins don’t have their own foibles, eg strain specific effects. Will reviewers soon be insisting in F4 outcross data repeatable in 2 or more lines only?

    Yes, everyone starting a project now will probably be working on raising the TALEN/CRISPR mutants in tandem with the MO. However there’s a lot of good science with rescue data out there from small labs which don’t have the manpower or space for mutants which would be consigned to the scrap heap (or “low impact” journals).

    What about the growing number of anecdotes telling of induced zebrafish mutants which have no phenotype whatsoever? None, nada. Is this due to compensation from the other zebrafish allele(s), or does it mean induced mutations aren’t generally sufficient to disrupt gene expression? Are Zebrafish a good model anymore? Will this bring another fish model to the fore? What about Xenopus tropicalis? They’re genome is undiplicated and diploid. What if an advantage MO’s have is that they target both alleles simultaneously?

    Furthermore, what about tissue specific knockdown with MOs using targeted injections - a technique which is a mainstay of the Xenopus system. Cre-lox systems are not without their own downsides, so this approach could consign a major feature of a model system to the history books - unfairly, in my view.

    Rate comment: Thumb up 4 Thumb down 0

  3. I would like to add a few considerations to this discussion.

    The authors of the paper argue that: “However, the description of a phenotype that is provided for the first time and that is based solely on MOs without the ability to compare with a genetic mutant, should in the future be viewed very critically.”

    But should actually the description of morpholino phenotypes be viewed more critically than they had to be viewed before the advent of the new genomic technologies? I think that such a position is unjustified. The morpholino technology is as good or bad as it used to be one or two or more years ago, so we have to be exactly as critical as before, but it makes no logical sense to be more critical.

    Should reviewers and the community rather trust a loss-of-function phenotype in the actual TALEN/CRISPR-injected embryos or a morpholino experiment in the actual morpholino-injected embryos? I think until we have a quantitative understanding of the side-effects of these technologies, there is no big difference. One would want to see at least two different morpholinos or TALENS showing the same phenotype, not seen with a control, or a rescue experiment.

    I would also like to point out one important use of morpholinos (at least for my lab), namely to test the specificity of newly generated antibodies. We routinely generate neuron-type-specific antibodies to study neuroanatomy and then use morpholinos for specificity tests. As long as we see the loss of immunostaining with two different morpholinos, we don’t worry too much about the side effects. I certainly would be rather nervous if a reviewer asked me to do a proper knock-out for such an experiment.

    It is also important to remind the reader that morpholinos are widely used outside fish, in particular by the sea urchin community. Since it is not possible to rear a sea urchin through the entire life cycle, one cannot obtain mutant lines. The only way is to use the actual injected embryos. If one cannot obtain a super-efficinet nuclease, still the best way for LOF analyses is to use morpholinos.

    Rate comment: Thumb up 4 Thumb down 0

  4. Peter Walentek says:
    The use of new techniques always comes with new potential insights - so is the case with zinc-finger nucleases, TALENs and CRISPR - which surely have revolutionized the way we plan and conduct studies these days.
    BUT: we always have to remember to take into account that these new techniques are still in development, and compared with the number of studies using MOs the data set is still rather small.

    From my point of view there are a couple of considerations about gene-targeting approaches that should be taken into account:

    - 1st: the big advantage of genome editing tools like TALENs / CRISPR is that you can target every genome; but at the same time (and as mentioned in the post above) not in all systems we are able to generate a line in a reasonable time and back-cross it as suggested in the text. In some species we just lack the knowledge/technique to raise embryos to adulthood for example. In other cases it would take a very long time to establish a line and then back-cross and wait for generation 2 or 3 for the actual analysis, e.g. Xenopus is such a case. In the latter case it would require an international initiative and big investments into resource centers to do that sort of work - in my opinion- , as it would be not very efficient (timewise and economically) for smaller/mid-size labs.

    - 2nd: TALEN/CRISPR targeting is not “knock-out”, at least most of the times. You induce lesions in the gene of interest, and the site of disruption is in part dictated by the presence of specific nucleotide combinations, which are required for the design of e.g. a working guide RNA. So, in many cases we actually don`t know what we are generating by targeting that gene. It might be a situation comparable to a full knock-out, but we might also change protein function, e.g. by generating hypomorphic allels, dominant-negative acting versions of the proteins, or even more difficult to interpret variants. In conclusion one would need to have at least two independent lines, targeting different sites of the gene, generating different types of mutations, and which would still give the exactly same phenotype. I would bet that in many cases one would end up with the same result as the CRISPR vs MO comparison - in many cases you would have the same phenotype, and in many cases you would have some degree of differences, and in a few cases you would have very different effects. Then, we would have to come to a standard approach similar to biological replicates - at least 3 independent lines, two of which should have the same effect to make a convincing case. And here it already starts to get a bit complicated on the practical side of things (requiring rather large facilities to house many lines, which will need back crossing and so on…) - one lab would have to invest a lot of time, money and space before starting to analyze the phenotypes - a huge gamble for most labs.

    - 3rd: MOs are very useful tools as they can be injected in various amounts, and therefore be used in lower amounts to address the functions of genese, which would be lethal during early development when removed altogether by full knock-out (in addition to specific targeted injections, e.g. in Xenopus, which are a great tool!).

    The above thoughts come to my mind immediately whan thinking about gene targeting strategies and their development over the past few years. In addition some of the standard thoughts were mentioned in the posts or the text already (e.g. off-target effects of CRISPRs). Also, the field was moving forward is such speed over the past years, that we cannot be sure that the CRISPR technology which is now regarded as the way to go, will be the gene targeting technology we will all be using in 2, 3, 4 years… Just think about the half-life of zinc finger nucleases and how many people work with them today… basically the same thing happened to TALENs, which work great, but CRISPR seems to be a bit cheaper and a bit faster, and all of a sudden everybody is using CRISPR and less TALENs… what will be the next hype? do we have to follow it? when our next paper will be rejected for having only CRISPR data and not a combination of CRISPR, MO and TALEN results? …in short: we might easily push it all a bit far, and therefore massively slow down the speed of our research (and the research of people who’s papers we are reviewing), which is a dangerous thing in times of short funding cycles.

    Rate comment: Thumb up 3 Thumb down 0

  5. Thanks for the comments. Just to say that this Opinion piece does not reflect journal policy. We do agree with many of the sentiments expressed by Stefan and Didier in terms of the problems with morpholino-based experiments and the potential of CRISPR technology to provide cleaner genetic data. However, we also recognise that there are many experiments that (at least for the moment) can be tackled using MO approaches but not using other techniques, and that properly controlled MO experiments can provide valuable insights.
    I do think that CRISPR technology is here to stay - its success in a wide range of model systems in such a short time has been immensely impressive. But that’s not to say that it’s perfect, or that new technologies won’t improve on or complement the CRISPR system.

    Rate comment: Thumb up 1 Thumb down 0

  6. Stefan Schulte-Merker says:
    Dear Katherine,

    I think the issue of ‘properly controled’ is where the problem with morpholinos lies. When Judith Eisen and Jim Smith published their editorial guidelines in Development a few years ago, these recommendations made a lot of sense at the time: use mRNA rescue, and use a combination of ATG/UTR and splice donor/splice acceptor MOs. By now we have learned that mRNA rescue can be misleading, and that there is no reliable way to test ATG or UTR morpholinos either: we have, in my lab, cloned MO-binding sites in front of a GFP cassette and tested whether MOs co-injected with the corresponding mRNA will block the appearance of fluorescence. This has worked in every single case attempted (!), which basically means that it is not a useful control. Hence, in the absence of an antibody against most targeted gene products, this only leaves splice donor or splice acceptor MOs, whose efficacy can be controlled for by RT-PCR. In our hands, these ‘work’ about 60% of the time, which means one has to order 3 or 4 MOs to be sure to get a pair of MOs that shows an effect on the molecular level. In our experience, it is more time- and cost-efficient to generate TALENs. But even if this was not the case: my main point here is that while it is possible to control for the intended efficacy of MOs (i.e. a effect on the intended target mRNA as tested by RT-PCR), it is impossible to control for unspecific effects. As long as the ‘proper controls’ are unable to discriminate between specific effect und unspecific side-effect, morpholinos will continue to be difficult reagents.

    Rate comment: Thumb up 2 Thumb down 3

Leave a comment


Copyright 2010 - 2012 The Company of Biologists Ltd

Company of Biologists