Morpholino review: how to knockdown
July 18, 2008 at 8:53 pm #9863
This is (IMHO 😀 ) a broad review of Morpholino techniques for a diverse range of organisms and strategies, including translation knockdown, splice modification and miRNA inhibition. Cracklin’ dry, but a good place to start if you are thinking about doing knockdowns.
Moulton JD, Yan YL. Using morpholinos to control gene expression. Curr Protoc Mol Biol. 2008 Jul;Chapter 26:Unit26.8.
July 29, 2008 at 1:30 pm #85256
With all due respect, I don’t think that an article about morpholino written by people that work for the company that makes morpholino would be a good place to start anything.
I’m actually surprised that this was allowed.
“specific, and lack non-antisense effects"? They must be joking!
I use morpholino a lot and one thing that can be said for sure is that they have tons of nonspecific effects, you need to be very careful with them and use multiple controls (I can elaborate if anyone is interested) to verify that the effect IS specific.
Its a great tool, the only established one of KD in various model systems, but be vary careful with it and be very aware that this paper was written by people who have a vested interest.
July 30, 2008 at 8:30 am #85287snowcapkParticipant
I’m interested in the different controls you used to check nonspecific effects. What’s the standard on this? We do other types of controls (control MASO injection to test for toxicity at a given concentration, in a given batch of embryos; design of reporter constructs to test the efficiency of the MASO) but don’t check for nonspecific effects.
When designing the MASO we pay attention to paralogs or other family members (e.g. will the brachyury MASO bind to the t-brain ATG?). However, we don’t check the effect of the MASO on those family members by QPCR because there are a lot of legitimate reg. interactions between these genes. No one would be surprised to see Erf go down after Ets1 MASO injection in our system. "One light goes out, they all go out," as it were.
July 30, 2008 at 12:39 pm #85289
As far as I am concerned the gold standard is rescue.
One way is to make mRNA using the Mmassage Mmachine kit and inject with your MO, if you are lucky it will rescue the phenotype and this means that the MO has a specific effect on the target gene.
However, more often than not this will not work, there are various reasons for this including that you are injecting a large amount of mRNA that is not translated in accordance with how the endogenous mRNA is produced and translated. So you are basically over-expressing which can itself cause abnormalities.
Another way is to make a construct that will express the mRNA that the MO is targeting under control of the promoter of the gene of interest. That will give you a more normal expression pattern and hopefully a rescue.
Note that these methods can only be used with a splicing MO, an ATG MO will bind to the injected/artificially produce RNA and block it as well.
There are ways to get around that but I have no experience with them.
Using two different MO (one ATG and one splicing) for the same target and getting the same phenotype is also considered a good control, because it is assumed that MO with different sequences would not causing the same nonspecific phenotype.
Another is using a mismatch MO that will not bind to target RNA but has a vary similar sequence. If this MO does not cause a similar phenotype that indicated that the effect of your MO is specific.
Lastly, there is a recent paper showing that injecting p53 MO together with your MO inhibits nonspecific MO induced cell death in the CNS, thus giving you a cleaner phenotype.
That’s all I can think of at the moment. Sometimes nothing works. I’ve seen many paper where it seems that the authors just raised the injected MO concentration until they see an effect and with no real controls. In my view this simply means that they probably caused general MO toxicity thus messing up the embryo. With many MO you will see non-specific effect (in zebrafish) that include curled tail, no trunk and small head and eyes (i.e. messed up CNS).
July 31, 2008 at 9:30 pm #85323
Good day chaka8,
I will respond first to your latter message.
I agree that an mRNA rescue of a Morpholino knockdown is an excellent proof-of-specificity experiment when it works. However, we should be careful to point out that many rescues based on Morpholino-RNA coinjections cannot in principle reproduce a wild-type phenotype. It is not just the amount of mRNA injected, but the timing of expression onset and the location of expression that can make rescuing a phenotype impossible. In many cases, the timing of the onset of a gene’s expression is critical to its function. While a Morpholino can prevent that expression, a coinjected rescue RNA is often present prior to the onset of transcription of the endogenous RNA. This inappropriately early expression can prevent rescue of the wild-type phenotype for many genes. Similarly, the location of endogenous gene expression and rescue RNA translation are often different, with the protein product of the rescue RNA distributed more homogeneously through the embryo. This, likewise, can preclude rescue of the phenotype after a Morpholino and its corresponding rescue RNA are coinjected. These are not simply problems of overexpression, but rather of mis-expression.
As you pointed out, expression in the cells of a rescue RNA under the control of its normal promoter in a stable construct can avoid the problems caused by expression at inappropriate times and places.
There are many cases where rescues have been reported using translation blocking Morpholino oligos targeting the 5′-UTR of an mRNA. A construct is made with the coding sequence of the RNA of interest fused downstream of the 5′-UTR of a different mRNA, then the fusion is expressed and coinjected with the Morpholino. Since the 5′-UTR sequence is different, the protein encoded by the fusion mRNA can complement the knocked-down protein without the Morpholino binding to the fusion mRNA. Other approaches for rescuing a translation-blocked Morphant include using the mRNA from a different species (which often has sufficient 5′-UTR sequence divergence but can complement the missing protein’s activity) or using a Morpholino that targets the start of the coding sequence and introducing wobble-base mutations at five or more positions in the start of the rescue mRNA coding sequence (however, while there have been some reports of success with this latter method, other labs have fought with it without success and I do not recommend that strategy).
Injection of rescue RNA transcribed from antisense cDNA can rescue a splice Morphant as long as the timing and location of expression are not critical.
The two-nonoverlapping-oligo strategy is commonly used to confirm specificity of a Morpholino knockdown, especially in the zebrafish community. Some investigators report two translation blockers, some use two splice modifying Morpholinos, and others use the combination of a translation blocking Morpholino with a splice modifying Morpholino. In all of these cases, when two nonoverlapping sequences targeting the same gene produce the same phenotype, this supports the hypothesis that the observed phenotype is due to the knockdown of the targeted gene and not an interaction with an off-target RNA.
The mismatch experiment typically uses a five-mispair Morpholino oligo. First a targeted oligo is used at a range of concentrations and the minimum concentration needed to produce an experimentally useful frequency of phenotype is determined. Then the five-mispair oligo is used at that same concentration. If no phenotype is observed, this suggests that the Morpholino is specific for its target, at least within a range of sequenced affinity defined by the mispairs. However, this experiment fails more often than the two-nonoverlapping-oligo strategy fails, usually requiring purchase of another oligo to try another mispair experiment; therefore I do not recommend the five-mispair experiment for Morpholino specificity controls.
For a recent excellent review of Morpholino controls, see:
Eisen JS, Smith JC. Controlling morpholino experiments: don’t stop making antisense. Development. 2008 May;135(10):1735-43. Epub 2008 Apr 9.
The "general MO toxicity" you describe at the end of your second message matches the p53-mediated apoptosis phenotype described in the recent paper you mention. That paper is:
Robu ME, Larson JD, Nasevicius A, Beiraghi S, Brenner C, Farber SA, Ekker SC. p53 activation by knockdown technologies. PLoS Genet. 2007 May 25;3(5):e78. Epub 2007 Apr 10.
The authors showed that the "general MO toxicity" can be caused by other antisense types and is a consequence of the knockdown, not of some chemical toxicity specific to the oligo type used. The same phenotype was observed when a gene was knocked down with a Morpholino or with a griPNA (a peptide nucleic acid with a phosphate linkage between every other base) and the "general MO toxicity" phenotype could be rescued by coinjection with a p53-targeted Morpholino. Therefore, what you are calling "general MO toxicity" is in fact a general knockdown toxicity, a cascade effect that is the organism’s response to success of the knockdown instead of the oligo chemistry employed.
As for your first message, many folks rely on customer support from Gene Tools to assist with oligo design, experimental design and troubleshooting. We have a broad information base about Morpholino experiments as we discuss experiments with investigators working with many different organisms, delivery systems and oligo targeting strategies.
You write in regards to the phrase "specific, and lack non-antisense effects" in my abstract that "They must be joking!" No, to the contrary, I am very confident that the specificity of Morpholinos is far better than most antisense types including all RNase-H competent oligos (DNA, RNA, phosphorothioate RNA), high-affinity RNase-H independent oligos (LNA, PNA) and siRNA, for which widespread off-target gene modulation is well documented. The only non-antisense effect associated with Morpholinos that I know of is the knockdown cascade effect triggering p53-mediated apoptosis, an effect shared by any gene knockdown. Even this is not the sort of sequence-independent non-antisense effect associated with older antisense types like phosphorothioates. Morpholinos do not produce the toxic degradation products typical of phosphorothioate oligos and, because because they are not ionic and form no salt bridges, unlike phosphorothioates the Morpholinos do not interact strongly with proteins and unlike phosphorothioates do not trigger physiological-level responses independent of binding to RNA.
Finally, I wholeheartedly agree that the use of careful specificity controls is very important. Relative to other antisense types, the specificity of Morpholinos is very good, but minor non-specific effects are occasionally observed and catastrophic non-specific effects do occur with some oligo sequences. Non-specific effects were carefully studied in a particular case in sea urchin:
Coffman JA, Dickey-Sims C, Haug JS, McCarthy JJ, Robertson AJ. Evaluation of developmental phenotypes produced by morpholino antisense targeting of a sea urchin Runx gene. BMC Biol. 2004 May 07;2(1):6.
Note that interactions of Morpholino with off-target RNA are favored by lower temperature, and this study was done in embryos raised at unusually low temperatures. It is sometimes possible to overcome off-target interactions by raising the incubation temperature of cells or embryos, increasing the stringency of the oligo hybridization (control cells or embryos must, of course, be incubated in similar conditions).
Enough about the technology. Your statement "I’m actually surprised that this was allowed" was offensive. I doubt you read the paper, or you would know that I tackled many of these issues and pointed out many pitfalls that may occur in experimental design and execution. There is no question about my industry affiliation, it is clearly stated in the publication. Are you suggesting that only academics should be allowed to publish in the area of their expertise? The only way my organization can survive in the long run is if we tell the truth about our products and help people to use them effectively to do good experiments and publish well.
July 31, 2008 at 10:26 pm #85324
Thank you for your detailed and illuminating response. As I said in my first post, MO is an excellent tool, in many cases the only tool available.
As a researched working on ZF CNS development, I see many MOs non-specific effects that make research very difficult indeed. For this reason I feel that claiming that MO is highly specific is somewhat inaccurate.
I use Gene-Tools customer service myself and am happy with the service; my only complaint is the cost of the MO… 😕
However, I do not think that accepting review articles (rather than experimental articles) for a peer-reviews scientific journal about a certain tool written by the people that are selling this tool is a very good idea.
If a review on the topic is needed it is easy to turn to an independent researcher that has experience with the tool for this purpose. I’m not saying that the paper is inaccurate in any way; just that I don’t this is a healthy practice due to conflict of interests.
August 1, 2008 at 7:36 pm #85331snowcapkParticipant
Thank you both for your insightful replies! I’m familiar with the rescue experiment you both mentioned (rescue through coinjection with a construct containing the gene’s coding region in its native regulatory context, e.g. a BAC) and it is clearly a powerful tool. As you know these constructs tend to incorporate in concatomers. Our lab models gene regulatory networks and hysteresis is often a factor…expression levels from a concatomer are not sufficiently similar to expression from only two native loci in this context. Furthermore, incorporation may occur in just a subset of cells so one must build in a method for determining which cells have actually been rescued. This is quite time-consuming whether the construct is a BAC (requiring homologous recombination to insert a reporter) or even just a small fragment, because there must then be additional checks to determine whether both the native gene and the reporter are being expressed equivalently.
Jon, the two non-overlapping MO control is an excellent idea but unfortunately it’s beyond our budget. I didn’t know it was possible to make two non-overlapping translation blockers – all of our morpholinos are centered (more or less) over the ATG. How do you position these so that both can be effective? Also thank you for the paper suggestions! An aside: Jim Coffman was my session chair at my first conference. Small world! 😉
Chaka8, FWIW I’ve never noticed cross-reactivity of morpholinos despite working with around a dozen of them…though I apparently haven’t used the right controls. *nervous laugh* I assume that is what you mean by non-specific effects. Were you able to identify the other targets of the MOs? I am curious whether this is a more common problem in vertebrates due to gene duplication because I have seen no evidence of it in sea urchin. If you weren’t able to demonstrate that your MOs were cross-reacting, how can you be sure that the effects you observed were really "non-specific?" This is quite a claim to be making but if you have support I’d really like to know about it. "Pics or it didn’t happen." 😆
August 4, 2008 at 5:30 pm #85367
For blocking cap-dependent translation, a Morpholino can target anywhere between the 5′ cap and the start codon and can extend downstream into the coding sequence as long as the start codon is covered. To see why this works, consider the steps at the beginning of translation. A group of proteins and the small ribosomal subunit bind at the 5′ cap and then other initiation factors bind, forming the initiation complex. The initiation complex scans through the UTR to the start codon. At the start codon the large subunit binds, the initiation factors dissociate and translation proceeds through the coding region.
If we can get in the way of the initiation complex by binding a Morpholino to the UTR we can prevent the initiation complex from reaching the start codon, but once the large subunit binds and forms an entire ribosome then a Morpholino oligo cannot stop its progression; the ribosome just displaces the downstream oligo from the mRNA and reads through. This is why the targetable region for translation blocking extends from the 5′ cap to the start + 25 bases.
There are two reasons why we prefer to target at the start. First, the quality of sequencing deposited in public databases is often poor in the UTR, especially for older sequences. More than once we’ve found cloning vector sequence reported in UTR. Second, though rare in vertebrate genomes, internal ribosome entry sites (IRES) do exist and can allow a ribosome to "short-circuit" a Morpholino-blocked target. So, when selecting oligo sequences, I like to start at the start codon and analyze the possible oligos in that region, then if necessary I’ll work upstream into the UTR until I find a good target.
For results of targeting some actual translation-blocking oligos, see figure 1 on the page:
If the sequence allows it, I’ll choose a pair of nonoverlapping translation blockers so that one oligo covers the AUG and the second is targeted just upstream of the first oligo. These oligos can then be used individually to confirm specificity and then in separate experiments they can be coinjected to take advantage of dose synergy, allowing much less oligo to be used per injection.
- You must be logged in to reply to this topic.