CRISPR often damages human embryo chromosomes

An artistic depiction of human embryo CRISPR starting with an image of a real embryo.
An artistic depiction of a CRISPR’d human embryo starting with an image of a real embryo. Knoepfler.

Three new studies reported in preprints all show severe DNA damage to human embryos from CRISPR a surprisingly high percentage of the time including in some cases loss of heterozygosity (LOH). Anywhere from about a quarter to half of CRISPR’d embryos exhibited major genome damage.

As readers of The Niche know, I never was a fan of the idea of doing CRISPR on human embryos for reproductive use, although I have been supportive of limited, thoughtful in vitro gene-editing studies on human embryos.

The chromosomal chaos reported in the new work including in particular the LOH reported in the preprints collectively throw a big monkey wrench into both strictly in vitro research and any even hypothetical future reproductive uses. For instance, you can imagine if you are trying to study OCT4 by knocking it out but you’re actually losing an entire chromosome, then your results are not going to be OCT4-specific.

These issues raise deep safety concerns too if anyone were to try clinical human embryo CRISPR again.

Imagine telling a parent, “It turns out that we corrected your just arrived newborn’s specific disease-associated mutation as we’d hoped, but unfortunately she’s also missing one entire chromosome.”

“Oops, sorry,” does seem to cut it in such a case.

CRISPR causes big genome damage including LOH

What’s do these new reports tell us more specifically?

First of all, even now much-improved gene editing technology inserted into human embryos does not work very precisely. It’s messy in the sense that even when it makes a desired change it often also causes other large-scale changes.

While their studies were quite different in some ways, the groups all found this kind of big chromosomal problem with embryo gene editing. We shouldn’t shrug it off. A newsy Nature piece by Heidi Ledford CRISPR gene editing in human embryos wreaks chromosomal mayhem highlights the troubles.

3 preprints point to challenges moving forward

Dieter Egli
Professor Dieter Egli.

Here are the 3 new preprints I mentioned above including from the Kathy Niakan, Dieter Egli, and Shoukhrat Mitalipov groups:

Beyond finding chromosomal damage in CRISPR’d human embryos some of the time, another challenge is that the kinds of alterations they defined sometimes would be easy to miss by other researchers without going the extra mile to delineate them.

Note that some past reports including this one by Niakan have suggested large-scale chromosomal changes could be occurring, but things weren’t so clear in part because the undesired changes make their own detection more difficult.

The groups also reported some differences from each other in the types of aberrant events, with Mitalipov’s team uniquely reporting gene conversion.

Egli’s take on where this leaves things

I asked Egli for comment on what this could mean for the idea of future clinical, reproductive use (and will it eventually be used this way) and also for his sense of where the field goes from here:

“Whether it will ever be used? I think we need to first answer a basic question, what the outcomes of a double strand break are to know what it could be useful for.

Clearly this is the time to do basic research, since even basic and simple questions are not resolved. It’s a fundamental difference whether a gene is corrected or a chromosome removed.

Differences between studies remain unresolved. Ma et al. 2017 and Liang (preprint) reported efficient correction; by contrast, Alanis-Lobato and Zuccaro preprints report chromosome loss/removal. This question needs to be answered conclusively (and I think we do this in our study).

The discussion is ahead of the science. We are talking about clinical use when basic questions are not answered, and potentially the wrong conclusions have been made. The discussion about clinical use of Cas9 needs to be based on sound published literature (preprints are not published). This is a contribution the journals can make. Thereafter, the discussion about your question will be possible.

There are also other basic and relevant questions that should be addressed as well. e.g. what is the suitability of Cas9 to study the function of genes in preimplantation development? How do the different outcomes of a DSB affect phenotypes?”

In some further emails, Dieter also clarified their main goal and provided additional perspectives on the new findings, including potential utility of chromosome loss if it could be controlled:
“It was our goal to determine the outcomes of a DSB in human embryos. The correction of a mutation was not our primary goal, as the mutation is recessive. While the public discussion right now is focused on the gene correction function of Cas9, what has not until now been considered is that in some contexts, the targeted removal of a chromosome could actually be useful, such as in the case of a trisomy.”

Still early days for embryo CRISPR

So looking to the future, near and far, where does this leave us?

It seems even clearer now that taking this tech down a human reproductive path is beyond unwise and would be reckless even if done somewhere in a technically lawful manner in the future.

If you take the analogy of going down a series of stepping stones along a path, there are many stones from where we are now to a future of any clinical heritable use of CRISPR or other gene-editing tech in human embryos, without even considering ethical and social justice issues.

On the practical side of things, I think accurate germ or pluripotent stem cell (gamete precursor) gene-editing is far more feasible than working on embryos.

He Jiankui’s “CRISPR babies”

You’ve also got to wonder even more now about the CRISPR gene-edited children that He Jiankui made last year and their long-term health.

It seems to me given the state of the technology at the point he introduced it into human embryos that there is a high probability the children have some kind of major genomic changes potentially both at the CCR5 locus and elsewhere.

I hope someone is privately giving the families involved the information they need such as via WGS and providing genetic counseling.

Maybe that 3-year jail sentence wasn’t so severe?

Field needs to tone things down

Are the folks who before seemed relatively upbeat about future reproductive human CRISPR now going to scale back their exuberance?

Will Mitalipov continue CRISPR’ing huge numbers of human embryos created by IVF for research that is at least implied to have clinical intent in the future?

What about the researcher Hui Yang who last year in the media even in the post-He Jiankui era seemed to want to make CRISPR babies in China in a lawful manner?

The guy Rebrikov in Russia?


Also, for those who liked the analogy of CRISPR gene-editing being “genome surgery” and some folks literally called themselves “genome surgeons”, at least in human embryos the tool is nowhere near that precise. More caution is needed with the language we use to describe it or we risk misleading others including the public.

At a basic science level the field needs to learn a vast amount about early human development. In particular,  human embryo DNA repair is unique compared to somatic cells and not well understood. Niakan makes a compelling case for this kind of research in the video above. CRISPR can aid in deciphering some key elements of our development, but it’s a tough road given the lack of consistent precision. This also means relatively few researchers are going to be able to do this kind of basic research.

For those who look for silver linings, these new preprints are sort of good news in the one sense that we now better understand the realities of the tough human embryo gene-editing challenges and we’re in the process of getting to know human embryos much better.

This in turn will help us understand our own biology better and provide new insights into diseases.

3 thoughts on “CRISPR often damages human embryo chromosomes”

  1. As you know, Paul, I oppose CRISPR’ing babies at this point largely on the grounds of the babies’ safety (as well as, of course, things like breaking the law). It does seem to me, however, that these studies are not as strongly negative for the long term possibilities as you expect. If one were to take, say, 10 embryos from parents who were both homozygous for an autosomal recessive condition and find 5 with chromosomal mayhem, you’re still left with 5 without it. (Even if the technology doesn’t get better, though, for example, base editing the embryos – or gametes or gametic precursors used to make embryos. Am I wrong in that?

    1. I still see it as a much rougher road than that.

      You won’t be able to be really sure which embryos have LOH just based on screening by PGD on a few cells from early embryos and there may be other changes that are even more difficult to detect/rule out by PGD. This will likely vary depending on the particular gene/mutation being targeted too. These are tough assays to do on single cells.

      Further, for both LOH and such smaller mutations, you can’t be sure of their absence in the embryo as a whole without screening all the cells, leaving you with no embryo to proceed with to pregnancy.

      In addition, you’re assuming the 5 embryos in that hypothetical scenario both do not have deleterious mutations and also have the desired gene edit.

      Accurate gamete precursor gene-editing is far more feasible since you have more time and cells to work with for QC. I should have mentioned that in the post, which I just added now.

  2. I’ve had time to consider perspective over the last few months. This is the longest I’ve been away from a lab since my undergraduate research days- more than 40 years ago. I’m still interested in solving the problem right in front of me- like predicting that a single neural organoid in a sealed cryovial with a few mls of medium would thrive during a month on the space station. I love a moment when other scientists say that something can’t work and then it does! I have decades of both failures and successes to draw on.
    I guess I think like a stem cell.
    Of course a double stranded break is going to mess up the genome of an embryo. We know that structural changes in the genome occur all the time, and that there are very few tools to detect them. I hope someone uses BioNano Genomics analysis to remap the sequence of human cells, gene targeted or not. We published on using it for iPSCs a long time ago- 2016! Too soon for the technique to routinely be adopted. If interested,

Leave a Reply