Facebook PixelHow can we make CRISPR therapies safer?
Brainstorming session

How can we make CRISPR therapies safer?

How can we make CRISPR therapies safer?

Image credit: Arek Socha/Pixabay

By Jamila on Oct 21, 2020

[1] Terns, Michael P., and Rebecca M. Terns. "CRISPR-based adaptive immune systems." Current opinion in microbiology 14.3 (2011): 321-327.

[2] Anderson, Keith R., et al. "CRISPR off-target analysis in genetically engineered rats and mice." Nature methods 15.7 (2018): 512-514.

Creative contributions

Using a synthetic material-based delivery system

[1] Tong, S., Moyo, B., Lee, C.M. et al. Engineered materials for in vivo delivery of genome-editing machinery. Nat Rev Mater 4, 726–737 (2019). https://doi.org/10.1038/s41578-019-0145-9

by Shubhankar Kulkarni on Oct 22, 2020

Shubhankar Kulkarni a month ago
I read a bit further and found some issues with the synthetic material-delivery systems in the same paper. They have also mentioned the use of viral-non-viral-hybrid delivery methods (a combination of viral and non-viral delivery approaches) - a synthetic material delivers the Cas9 nuclease and a viral vector delivers the gRNAs and the DNA donor template.

This reduces the off-target effect issue of the viral vehicles and the bulkiness of the synthetic materials. Also, although the positively charged surface of the synthetic material allows for the easy entry in the nuclear space and the release of the cargo, it leads to extravasation of the delivery vehicles into the interstitial space from highly permeable vessels. Hence, repeated injections are required to observe the desired amount of the on-target effect.

Although the viral-non-viral-hybrid delivery methods have not been used extensively, they seem to be advantageous than both the viral and synthetic material methods.
Jamila a month ago
Shubhankar Kulkarni The delivery method is an essential aspect as it will determine the CRISPR therapy's safety and effectiveness. In many studies, viruses have been used to deliver CRISPR-Cas to the cells. [1]

✔️Viruses are quite efficient – infecting the host genome is a natural ability.
✖️ There could be viral integration into the host genome.
✖️ Immunogenicity and toxicity associated with viruses
✖️ Adeno-associated viruses (AAV) have a limited capacity for gRNAs and template DNA. [1]

Previously, nonviral delivery methods were less efficient than viral delivery methods [2] – but I think the consensus is that they are improving rapidly. So, viruses may not be the best way to deliver CRISPR in the future.

I've never heard of this viral-non-viral-hybrid delivery method you mention – it sounds fascinating that researchers can use viruses to deliver the guide RNAs and synthetic methods to deliver Cas proteins. It seems like the perfect combination that might overcome significant issues, like viral integration, capacity limits, immunogenicity, improve efficiency, etc.

1. Wang, Dan, Feng Zhang, and Guangping Gao. "CRISPR-Based therapeutic genome editing: Strategies and in vivo delivery by AAV vectors." Cell 181.1 (2020): 136-150.
2. Nayerossadat, Nouri, Talebi Maedeh, and Palizban Abas Ali. "Viral and nonviral delivery systems for gene delivery." Advanced biomedical research 1 (2012).
Jamila a month ago
I also read a paper where researchers developed virus-like nanoparticles (VLN) to deliver CRISPR components. VLN has a core-shell - mesoporous silica nanoparticle (MSN)-based core and a lipid shell. Apparently, this structure allows VLN to remain stable in the blood. This method also showed no signs of inducing cytotoxicity in cells. So, VLN can be used to co-deliver CRISPR and drugs to cancer cells. [1]

1. Liu, Qi, et al. "Virus-like nanoparticle as a co-delivery system to enhance efficacy of CRISPR/Cas9-based cancer immunotherapy." Biomaterials 258 (2020): 120275.

Use Anti-CRISPRs as an antidote to CRISPR.

[1] Marino, Nicole D., et al. "Anti-CRISPR protein applications: natural brakes for CRISPR-Cas technologies." Nature Methods (2020): 1-9.

[2] Shin, Jiyung, et al. "Disabling Cas9 by an anti-CRISPR DNA mimic." Science advances 3.7 (2017): e1701620.

[3] Li, Chang, et al. "HDAd5/35++ adenovirus vector expressing anti-CRISPR peptides decreases CRISPR/Cas9 toxicity in human hematopoietic stem cells." Molecular Therapy-Methods & Clinical Development 9 (2018): 390-401.

[4] Lee, Jooyoung, et al. "Tissue-restricted genome editing in vivo specified by microRNA-repressible anti-CRISPR proteins." RNA 25.11 (2019): 1421-1431.

by Jamila on Oct 27, 2020

Shubhankar Kulkarni a month ago
That is a good idea! I could think of a few questions after reading the suggestion:
1. Did the anti-CRISPRs decrease the ON-TARGET efficiency of CRISPR?
2. Adding both CRISPR and anti-CRISPR in a living system is like the movie "Aliens vs predators" - two space species fighting a war on Earth. Will our body suffer any collateral damage?
3. If these therapies become mainstream, can pathogens use either of these systems (CRISPR and anti-CRISPR) to launch an infection?

I know that the field is in its infancy and we may not know the answers but thought that these questions are important. Do we know the answers to any of these?
Antonio Carusillo 25 days ago
Shubhankar Kulkarni

1- from this more recent paper regarding Anti-CRISPR.Protein ( https://advances.sciencemag.org/content/6/6/eaay0187 ) they show that the fine tuning of such ACr ( either over-expressed or fused directly to the Cas9) has an effect of both the ON target and Off-target. However, the effect on the ON-target depends on the target site ( in same instances the ACr did not have any effect in others the effect was more evident ). While the Off-targets within a certain concentration of the ACr used are dramatically reduced compared to the ON-Target. So you may give up on a bit of ON-target editing in order to achieve a preciser Nuclease. Nevertheless we have to take in consideration that the " target-dependency " issue is always a present factor. So, regardless the strategy you should always evaluate 3 to 5 different guiding RNAs for the same target and seee which one ( in the absence of any "tricks" ) gives you the bast ON-target/ OFF-target ratio and from there you can start using all the sort of strategies that I may explain in upcoming contribution. I just need to find the time to write it ( and to make it short enough ).
2- I do not expect Any collateral damage ( neither toxicity was reported in the paper). The ACr acts on CRISPR system by preventing it to interact with the DNA. The CRISPR system acts on the DNA inducing a DSB ( the real threat, but this is somenthing I will explain in the contribution or contributions ). So no Alien vs Predator scenario. As the " preys " are different 😄
3- Anti-CRISPR are indeed found in bacteriophage ( https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6086933/ ) as system to fight back the CRISPR system ( is an endless Cold War down there ). However, this system was evolved cause bacteria were indeed expressing CRISPR. So from an evolutionary point of view, the ACr was a form of " adaptation". Human therapy - so far - does not include integrating a CRISPR constantly expressed by our cells. So, pathogenes won´t take ( theoretically ) any advantage from evolving anti-CRISPR proteins.
Jamila 21 days ago
Shubhankar Kulkarni Thanks for the questions.

1. In Shin’s study, they found that anti-CRISPRs seemed to reduce the off-target effects of CRISPR editing but didn’t heavily impact the on-target efficiency – this effect was seen in Shin’s study, at least! (https://advances.sciencemag.org/content/advances/3/7/e1701620.full.pdf)

2. Yes, that is very true. More research definitely needs to be conducted on the safety of adding CRISPR-Cas and anti-CRISPRs into the body together. The in vivo studies are pretty much in their infancy. However, in one study by Lee, the group added CRISPR and anti-CRISPRs simultaneously into adult mice – the researchers reported no toxicity at all. (https://pubmed.ncbi.nlm.nih.gov/31439808/) In Lee’s study, the mice were euthanized 7 days after receiving the vector containing CRISPR-Cas and Acr.

The safety of anti-CRISPRs and CRISPR-Cas combination needs to be evaluated for extended periods because, in humans, these CRISPR treatments will be long term, not just a 7-day treatment. In conclusion, more in vivo studies using CRISPR-Cas and anti-CRISPRs need to be carried out to validate the safety further.

3. I’m not sure if pathogens can use CRISPR/anti-CRISPRs to launch an infection, but I would think they wouldn’t be able to do that. The clinical trials using CRISPR in humans have shown to be safe with no immune responses. (https://www.nature.com/articles/d41586-020-00339-3) However, this might be different when using anti-CRISPRs too. I can’t say for sure, so I’d suggest determining the immune response when adding CRISPR and anti-CRISPR components in vivo.

Why CRISPR is already safe but it may still need improvements

[1] Hirakawa MP, Krishnakumar R, Timlin JA, Carney JP, Butler KS. Gene editing and CRISPR in the clinic: current and future perspectives. Biosci Rep. 2020;40(4):BSR20200127. doi:10.1042/BSR20200127

[2] Zhang, J., Li, X., Neises, A. et al. Different Effects of sgRNA Length on CRISPR-mediated Gene Knockout Efficiency. Sci Rep 6, 28566 (2016). https://doi.org/10.1038/srep28566

[3] Havlicek S, Shen Y, Alpagu Y, et al. Re-engineered RNA-Guided FokI-Nucleases for Improved Genome Editing in Human Cells. Mol Ther. 2017;25(2):342-355. doi:10.1016/j.ymthe.2016.11.007

by Antonio Carusillo on Oct 31, 2020

Jamila 20 days ago
Antonio Carusillo Thanks for the insight.

1. I completely agree with you. It must be safe to some extent because we have clinical trials using this technology right now. With ex vivo editing, researchers can easily check whether there are any abnormalities before infusing the cells back into the patient. However, in vivo editing hasn’t got this pre-check – so it seems a lot riskier! (https://www.researchgate.net/publication/282392356_Advances_in_therapeutic_CRISPRCas9_genome_editing) Of course, the researchers can monitor the patients during the trial with blood tests and other health checks, but the pre-check gives extra satisfaction.

2. The way I see it, CRISPR can be made safer with various strategies like the ones you mention. We can optimize the guide RNA and Cas to reduce the risk of off-target effects:
- The size and GC content of the guide RNA can alter the off-target effects.
- Some Cas variants may have fewer off-target effects. For example, Cas9 derived from Staphylococcus aureus (SaCas9) might have less off-target effects compared to Cas9 derived from Streptococcus pyogenes (SpCas9). Furthermore, SaCas9 is smaller than SpCas9, which is better for delivery purposes.

3/4. Yeah, that is a problem. We could either use:
- “DSB free” methods
- regularly check the patient’s cells after in vivo editing

Maybe the P53 issue was limited to those two specific research papers – it could be to do with the methods used in their research paper. So, it might not be a problem.

Perhaps the DSB free CRISPR methods would be safer and better alternatives to traditional CRISPR gene editing. Base editing and prime editing are both outstanding DSB free methods. There is also RNA editing, epigenetic editing, and transcriptional regulation – all of which are also DSB free methods! (https://www.nature.com/articles/s41467-018-04252-2) It seems there is a vast array of CRISPR editing methods that could be used in the future – but of course, these methods are still in their infancy compared to CRISPR gene-editing.

Antonio Carusillo 20 days ago
Jamila Ahmed 1. Yes exactly what I meant. They took the chance to do it directly in vivo without any pre check beside previous experiments in animal model or in vitro. Meaning that for now CRISPR is considered safe enough to be used in a trial

2. yes diffeent Cas9 species may have different or lower off targets, you may also have different activities and off-target. You may also forced evolution to develop brand new Cas9 variants like the so called XCas9 (https://blog.addgene.org/xcas9-engineering-a-crispr-variant-with-pam-flexibility)

3. Yes RNA editing or Epigenome Editing are DSB free methods but they are also transient. Theoretically Epigenetic modifictions based on DNA methylation should be fixed or pretty stable, but no studies has be done so far showing 3 or 4 years monitoring DNA methylation status. RNA editing per se will be also transieent. So for such applications, you have to consider multiple administrations although you may find out that one or two administrations are enough.

Bioinformatics and sequencing to increase the safety of CRISPR

[1] https://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi

[2] https://genome.ucsc.edu/cgi-bin/hgPcr

by Juran K. on Oct 31, 2020

Jamila 18 days ago
For CRISPR editing, the guide RNA and target sequence must match, and the right PAM should be located near the target sequence. Bioinformatics and sequencing are quite useful tools that can make CRISPR safer. Specifically, these tools can help design guide RNAs with the lowest risk of off-target effects and detect off-target effects after CRISPR editing. [1]

Here's a list of some bioinformatics tools used in CRISPR:
- Cas-OFFinder
- CCTop
- CRISPRtarget
- E-CRISP [1]

Very interesting ethical questions. All the drugs on the market do have side-effects – but these drugs mainly target proteins, not the DNA itself. DNA-targeting drugs are relatively novel, so I guess there is an added worry that you might make considerable changes to the DNA, which you might not be able to reverse later down the line. So far, the CRISPR therapies have been well-tolerated and "safe" in the clinical trials.

I guess the overall aim is to make the CRISPR therapies as safe as possible by minimizing the risks of severe side effects – but yes, there will always be side effects, just like any drug.

1. Alkhnbashi, Omer S., et al. "CRISPR-Cas bioinformatics." Methods 172 (2020): 3-11.
Antonio Carusillo 21 days ago
Juran K. Bioinformatics is at the forefront of predicting and analysing CRISPR mediated off-targets. A possible tool is, for example, you can use CHOPCHOP (https://chopchop.cbu.uib.no/) to design a guide RNA with the highest efficiency and lowest off-target activity. However, these prediction tools have the problem to be not always " extendable " to all the conditions like cell type, nuclease delivery system, experimental time point ( days, weeks, month ). So they can return an "average" but there is no way for you, at the moment, to just pick one guide RNA and be sure that this one will be 100% effective and 100% precise. In the lab where I work, we usually select 3 to 5 gRNAs based on CHOPCHOP prediction and we try them all to see which one - in our conditions - perform the best.
At this point, bioinformatics comes handy for in cellula prediction like you would do using CIRCLE-Seq (https://www.nature.com/articles/nmeth.4278). This kind of analysis gives you the " worst-case scenario". By performing such an assay in vitro, you won´t take into account elements like chromatin status. Chromatin accessibility may affect the way CRISPR cuts at the on and off-target. Such assay assumes that CRISPR can equally access to both on-target and off-targets and so you can detect also off-targets that in vivo won´t be cut.
Now with these 2 first steps, you have selected the guide RNA with the best performance and is time to go in vivo ( human cells, mouse model etc ). After the editing, you can perform further bioinformatics pipelines that will tell you in your final model how many off-targets did you get (https://science.sciencemag.org/content/364/6437/286). Thereby the best recommendation is to apply this 3-layers system where you check: in silico, in vitro and in vivo for the precision of CRISPR. This is what is done in most of the labs willing to devise therapeutics based on gene editing :)
Shubhankar Kulkarni 21 days ago
Juran K.This is thoughtful progress. I like the idea of utilizing the available technology than creating a new one. Although we can design the CRISP Repeats after carefully screening the genome, the genome is only about 1.2% of the DNA that our cells contain. There is a high probability that the remaining 98.2% "junk" may have the target sequences and those can be modified. Although such changes may usually not have a drastic effect, since these regions do not translate into proteins, the modification using CRISPR can create new start and stop codons in the "junk" part of our DNA leading to their translation into alien proteins. This will be also, probably, considered as an "off-target" effect of CRISPR.

Perhaps drug-induced CRISPR systems could make CRISPR safer.

[1] Zhang, Jingfang, et al. "Drug inducible CRISPR/Cas systems." Computational and Structural Biotechnology Journal 17 (2019): 1171-1177.

[2] Davis, Kevin M., et al. "Small molecule–triggered Cas9 protein with improved genome-editing specificity." Nature chemical biology 11.5 (2015): 316-318.

by Jamila on Nov 13, 2020

Why gene editing precision and safety is not only determined by off-target effects.

[1] Iyama, T. and D.M. Wilson, 3rd, DNA repair mechanisms in dividing and non-dividing cells. DNA Repair (Amst), 2013. 12(8): p. 620-36.

[2] Takata, M., et al., Homologous recombination and non-homologous end-joining pathways of DNA double-strand break repair have overlapping roles in the maintenance of chromosomal integrity in vertebrate cells. EMBO J, 1998. 17(18): p. 5497-508.

[3] Delacote, F. and B.S. Lopez, Importance of the cell cycle phase for the choice of the appropriate DSB repair pathway, for genome stability maintenance: the trans-S double-strand break repair model. Cell Cycle, 2008. 7(1): p. 33-8.

[4] Anguela, X.M., et al., Robust ZFN-mediated genome editing in adult hemophilic mice. Blood, 2013. 122(19): p. 3283-7.

[5] Li, H., et al., In vivo genome editing restores haemostasis in a mouse model of haemophilia. Nature, 2011. 475(7355): p. 217-21.

[6] Sharma, R., et al., In vivo genome editing of the albumin locus as a platform for protein replacement therapy. Blood, 2015. 126(15): p. 1777-84.

[7] McVey, M. and S.E. Lee, MMEJ repair of double-strand breaks (director's cut): deleted sequences and alternative endings. Trends Genet, 2008. 24(11): p. 529-38.

[8] Shen, M.W., et al., Predictable and precise template-free CRISPR editing of pathogenic variants. Nature, 2018. 563(7733): p. 646-651.

by Manel Lladó Santaeularia on Nov 18, 2020

Juran K. 4 days ago
Hey! Thank you for a nice introduction of a really interesting part in CRISPR-Cas9 therapy that we missed.

As a strong believer in computers and the giant data sets being the future experimental systems, I support the solution that you proposed as a nice alternative to manage CRISPR off-targets.

As I read, two questions emerged:

- what percentage of "beneficial" gene editing outcome do you think is "acceptable"? Except the tissue specificity which you already mentioned, do you think we need an "ethical layer of decision-making" (e.g. 50% of good gene editing outcome is enough for ovarian cancer stage III, but it must be 80% for diabetes type 2, because there are other safer alternatives)?

- is this method already doable while treating states/diseases that include the extraction of the cells from blood, plasma or any other body liquid, modification of cells and reintroducing them back to a body? We could do these kind of in silico safety checks after the modification and before the reintroducement to the body.
Manel Lladó Santaeularia 4 days ago
Juran K. Fist of all, thanks for your comment. You raise some very interesting points.

-For the percentage of "beneficial" gene editing, that is a crucial point but also very difficult to establish. Different applications of gene editing could have different tresholds. For example, if you are trying to edit tumor cells, you want to be very sure that the INDELs you are generating are really having an effect on the cells. This would be crucial especially in this case in order to make sure you are eliminating as many cells as possible, considering how easily cancer cells can adapt to mutations. In a scenario where you want to generate a knock-out of a gene for an inherited disease, a high percentage of "beneficial" gene editing would clearly be preferable, but the treshold would probably be set at the minimum gene editing efficiency that allows you to improve the disease phenotype in the patients. Your point on safer alternatives is interesting, although one of the main advantages of gene editing is that it allows us to target diseases that previously we had no alternative how to treat. A paradigmatic example would be allele-specific gene editing to knock-out mutant alleles causing autosomal dominant diseases. Although miRNAs and other kinds of transcriptional regualtion have been tested, allele-specific gene editing seems to be the option with the most potential in , and thus setting this kind of treshold could indeed be quite important.

-This method would be useful also for ex vivo gene editing, where as you mentioned before, cells are extracted and modified after reimplantation. However, ex vivo gene editing protocols already contemplate in-depth characterization of the different clones generated, as well as selection of particular clone/s with the most desirable INDELs before amplification and reimplantation. Interestingly, in silico prediction could allow to choose a gene editing strategy with a higher turnout of cells edited in the desired way, thus probably reducing costs and improving the pipeline for production of edited cells. However, as far as I know this kind of machine learning would have to be adapted to each particular cell type, so I don't know if it would be readily available to use in, for example, hematopoietic stem cells. On the other hand, in vivo gene editing would benefit way more from this approach, since editing outcomes can only be analyzed after the editing has already occured in the tissue, meaning the potential for unintended gene editing outcomes is significantly higher. In that situation, having a high percentage of a particular, desirable outcome could be especially relevant in order to bring those approaches to the clinic.

Juran K. 3 days ago
Manel Lladó Santaeularia Thank you for answering the questions.

I understand and agree with you concerning the tissue/disease-specific percentage of the outcome. The ethnicity layer is, therefore, as I understood, not needed due to the very specific usage of CRISPR in no-alternative treatments.

Okay, I didn/t know the ex vivo characterization is already being actively done. Also, the small area where you proposed this in silico method could be used to reduce the costs and time of the gene-editing experiments seems fair concerning the fact that finances orchestrate most of the fundamental research experiments nowadays.
With the fast development of CRISPR technology, especially after the Nobel, I think this kind of data will be available soon, giving us the opportunity to reduce the experiment off-targets.

Temporary and tissue-specific expression of Cas9 could be crucial to improve its safety.

[1] Zhang, H.X., Y. Zhang, and H. Yin, Genome Editing with mRNA Encoding ZFN, TALEN, and Cas9. Mol Ther, 2019. 27(4): p. 735-746.

[2] Zuris, J.A., et al., Cationic lipid-mediated delivery of proteins enables efficient protein-based genome editing in vitro and in vivo. Nat Biotechnol, 2015. 33(1): p. 73-80.

[3] Merienne, N., et al., The Self-Inactivating KamiCas9 System for the Editing of CNS Disease Genes. Cell Rep, 2017. 20(12): p. 2980-2991.

[4] Suzuki, K., et al., In vivo genome editing via CRISPR/Cas9 mediated homology-independent targeted integration. Nature, 2016. 540(7631): p. 144-149.

by Manel Lladó Santaeularia on Nov 20, 2020

Your creative contribution

0 / 200

Added via the text editor