Keep calm & CRISPR on: perspectives on report of human Cas9 immunity

Keep calm & CRISPR on

The news that CRISPR-Cas9 gene editing in its current form may not work in a substantial fraction of people due to many of us having immunity to Cas9 came as a shock to many, but if you think about it, maybe it’s not so surprising. I don’t see it as the end of the world.

Motley Fool CRISPR stocks
Motley Fool Headline on CRISPR stocks

(preprint) from a group led by Matthew Porteus started this lively discussion a few days ago. The preprint is entitled, “Identification of Pre-Existing Adaptive Immunity to Cas9 Proteins in Humans.” Some people are freaking out about this finding for a variety of reasons including for some investors their financial concerns (see the headline snapshot from the investing site Motley Fool above, for instance).Keep calm & CRISPR on

Everyone needs to take a deep breath.

If you think back to the fact that CRISPR-Cas9 as a toolbox for “gene editing” was developed from a natural bacterial defense system against viral infections, it’s logical that the bacterial Cas9 protein (the most commonly used nuclease by which CRISPR gene editing works via DNA repair) might be sometimes viewed as foreign by our immune systems as we grow up.

The new preprint reports that the two main sources of Cas9 protein so far from the bacteria Staph aureus (S. aureus) and Strep pyogenes (S. pyogenes), are likely going to be recognized as a sign of infection in some people. In fact, probably in many people. A lot of us are likely to be immune sensitive to these bacterial Cas9 proteins (reportedly, 79% to S. aureus Cas9 and 65% to S. pyogenes Cas9 in the limited number of donors examined in the preprint) because we’ll have antibodies against these Cas9 proteins.

CRISPR-Cas9 immunity
Adaptive Immunity to Cas9, challenge for CRISPR gene editing, Figure 1, Charlesworth, et al.

You can see in a pic from Figure 1 from the preprint that Western blots demonstrate endogenous human antibodies against Cas9 proteins often recognize clear protein bands. These human antibodies that make Western blot bands show up for Cas9 are in all likelihood present at a pretty decent concentration because so many of us have had and responded to S. aureus and S. pyogenes infections during our lifetimes and those bacteria express Cas9. Every time we are infected with a pathogen we stand a good chance of developing antibodies against unique proteins expressed by the pathogen and apparently Cas9 is fairly immunogenic.

Since many of us likely have Cas9 antibodies in our bloodstreams, if we were to receive CRISPR-Cas9 gene therapy our bodies may effectively inactive the CRISPR-Cas9 system via those antibodies. In addition, Porteus’ report (Charlesworth, et al.) found reactive T-cells as well in humans. Thus, not only might our antibodies inactivate CRISPR-Cas9, but also the resulting immune response could pose risks to patients getting CRISPR-Cas9 gene therapy. These data are so new though that it’s hard to know what will actually happen in patients who have Cas9 antibodies after getting CRISPR-Cas9 introduced into their systems. So, we know these antibodies are present in some people, but we don’t know the functional significance in a gene editing therapeutic setting. Another caveat for this paper is that the group of human donors was relatively small so in a wider, more genetically diverse population it is possible not such a high percentage of people are reactive to Cas9…or it could unfortunately go the other way and even more people may have antibodies. Functionally significant levels of Cas9 antibodies could be present yet undetectable so that’s another potential headache.

So what happens next?

A number of discussions are ongoing now about workarounds to this immunity hurdle, including some discuss in the piece from Antonio Regalado:

“New CRISPR systems are out there, just waiting to be discovered in bacteria that the human body has never seen—like those living in hydrothermal vents, say. Extracting cells from our bodies, treating them with CRISPR, and then putting them back might also work.”

These are good ideas.

Those of us with antibodies to S. aureus and/or S. pyogenes Cas9 may react with Cas proteins from other bacterial species. Still, it seems reasonably likely that other nucleases can be found that work well for gene editing but against which humans generally do not have antibodies. Importantly, transient use of Cas9 in patients’ cells in vitro followed by read ministration in vivo back into the patient could prove effective in getting good gene editing and at the same time avoiding immune responses. However, such an approach may not always be practical for gene editing of cell types other than those of the immune system. For example, you can’t easily remove brain precursor cells to gene edit in vitro without damaging the brain.

Other longer-standing issues are still out there too as some have pointed out such as human immune reactions to the viruses such as AAV sometimes used to deliver CRISPR-Cas9 systems. I don’t see any of these things as insurmountable though across the board. I believe some CRISPR-Cas9-based gene therapy will be proven safe and effective down the road.

Even more challenges and uncertainty await those who would use CRISPR in human embryos for proposed heritable prevention of genetic diseases as we can see exemplified by the complications and limbo status of the Mitalipov lab human embryo CRISPR paper from last year, the main conclusions of which were challenged by the Egli, et al. preprint (this is not even including societal and bioethical issues with heritable human gene editing). Note that I don’t see such ethical issues with CRISPR use for gene therapy.

The bottom line from this past week’s new finding on Cas9 antibodies and reactive T cells is that using CRISPR-Cas9 for applications such as gene therapy in humans is still almost certainly going to be workable, but in many cases it’ll be more complicated than hoped and other enzymes besides Cas9 might often be needed. The road ahead, as has been found to be true for so many transformative biomedical translational pipelines including for stem cells, is going to be tougher than imagined at first with more steps involved to maximize chances of success and lower risks for patients and the field. As a technology like CRISPR matures, the field needs to mature in our expectations and realize there will be big challenges along the way without panicking. Challenges are just the norm for science. In fact, usually the more exciting something is, the more hurdles we’ll run into along the way.


  1. The proposed mechanism doesn’t make sense to me- I don’t understand how the Cas9 used in genome editing approaches would come into contact with serum anti-Cas9 antibodies. The in vivo strategies for genome editing currently being applied involve viral gene delivery, where, for example, an injected adeno-associated virus would deliver DNA encoding Cas9 to the targeted cell, and the Cas9 would then be expressed intracellularly to do its job- the Cas9 protein would not exist extracellularly. It’s not like people are injecting Cas9 protein into the blood where it would encounter the antibodies described in this study. Even if anti-Cas9 antibodies are present in the serum as shown by this study, why would they think that there would be an immune response, especially an immune response sufficient to disrupt the genome editing, if there is no Cas9 in the bloodstream? Why would they think that intracellular Cas9 would come into contact with extracellular antibodies to trigger an immune response? Did I miss something here?

    • @Shinsakan,

      I think the main answer is “antigen presentation”. Even non-immune cells regularly present peptide fragments on the cell surface of the proteins they express. This would include Cas9 in targeted cells.

      Some researchers now use pre-assembled CRISPR-Cas9 RNA-protein machinery (gene-editing ribonuclear protein) including gRNAs and Cas9 protein, rather than using virus or plasmid transfection, but I’m not sure if this is going to be the case for gene therapy where the focus seems more on virus. Assuming viral delivery, this passage from the NatureNews piece that I linked to in my post may be somewhat off-base: “Antibodies against Cas9 can bind to the enzyme in the bloodstream, before it has had a chance to act.”

      Other likely important factors. Since Cas9 is being overexpressed in the target cells, its levels are going to be far higher than most endogenous proteins too so there may be bucket loads of Cas9 peptides on the cell surface for the immune system to find.

      Also keep in mind that with (and due to) viral transduction alone and protein overexpression that some of those targeted cells are likely to die early after transduction and spill potentially a lot of Cas9 into the bloodstream.

      Finally, with many delivery approaches of CRISPR-Cas9 into most types of organs, I’d imagine that some virus coding for gene editing machinery including Cas9 will by chance directly end up in the bloodstream (e.g. via perturbation of capillaries with needle injection) and/or infect some immune/blood cells too.


      • Paul,

        Thanks for your response!

        As far as I know, the envisioned clinical applications are all based on gene therapy approaches, especially viral gene therapy. Correct me if I’m wrong, but pre-assembled CRISPR-Cas9 machinery doesn’t really seem realistic for clinical use because of the challenge of cellular uptake. So I think that the idea of antibodies binding to Cas9 in the bloodstream before it can act seems bogus.

        As for cells dying after transduction and releasing Cas9, I don’t think that AAV is actually that toxic to cells, and on/off systems can be used for controlled expression such that cells might not end up containing ridiculous amounts of Cas9. I think that the concerns with an immune response against Cas9 are (1) would it result in a severe systemic immune response that could threaten the life of the patient and (2) even if the immune response was minor, would it still interfere with the genome editing? With regard to (1) even if some trandsuced cells died (assuming that they died after they had time to express and build up a lot of Cas9) and released their Cas9 into the blood, I wonder whether the amount released would actually be enough to trigger a full-on life-threatening immune response. With regard to (2), I don’t see how release of Cas9 from dead cells into the blood would interfere with the genome editing in the viable transduced cells, because again their Cas9 is still intracellular.

        As for virus coding for gene editing machinery infecting immune/blood cells, (1) people are working on making AAVs that target only specific tissues, and (2) tissue-specific promoters for the genes coding for the gene editing machinery can also be used.

        I totally forgot about antigen presentation, though- thanks for mentioning this! As you indicated, this seems like the most legitimate potential problem. I think that this might be a potential problem for all gene therapies, and not just those where the transgene will code for Cas9. One solution might be to borrow viruses’ own immunoescape strategies. See, for example:

        Thanks for the discussion!

        • @Shinsakan,
          You raise some good points and questions.
          Conditional expression systems and selective transduction approaches may indeed help, but they tend to lower efficiency even as they increase specificity.

          I think you’re also right that many including some in the media somewhat missed the boat on this story in terms of digging deeper into actual mechanisms whereby Cas9 immunity would be problematic.

          Antigen presentation is still what seems most key in my mind, but I might be missing things too as I’m not an immunologist or clinical trial person.

          There’s probably going to be a Catch-22 situation of a sort for this kind of in vivo CRISPR-Cas9 gene therapy. On the one hand to have the highest possible chance of meaningful efficacy, you need the highest % transduction of the target cells and robust expression/levels of the gene editing machinery, but on the other hand as you approach those goals, you are also getting more and more likely to trigger an immune response at the same time. Maybe there’s a sweet spot of effective genetic modification without functionally problematic immune response triggering.
          For some patients who don’t express significant amounts of the antibodies, this may not be such a tricky issue.

          Alternatives to Cas9 could also allow resolve much of the problem.

  2. Supp. Fig1 is supposed to show the specificity of the Abs from each patient but they show only two donors and not even both Cas9 homologues. Thus, what authors are showing in Fig1 is just unspecific Ab binding on a western blot!! The serum was diluted only 1:10. So much Ab the binding can be unspecific!
    Indeed, any diagnostic kit that would tell you if you have antibodies against (let’s say) hepatitis C or any other previous infection is ELISA. And the serum dilutions would be way higher, not using a whole mg of Abs …
    Soon to be rejected paper (unless improved).

Leave a Reply