Summary of ICMS Journal Club presentation on Friday, August 27, 2021.
Title: Increasing the rate of homologous recombination at Cas9 target sites
Speaker: Professor Dr. David B. Alexander, Nanotoxicology Project, Nagoya City University
On Friday, August 27, by Zoom, Prof. Alexander gave a presentation to the Journal Club of the Institute for Comprehensive Medical Science (ICMS) of Fujita Health University. This was a classical type of journal club meeting in which a paper published by others was discussed. After Prof. Alexander had given an introduction about Cas9/CRISPR and about the homology directed repair (HDR) versus non-homologous end-joining (NHEJ) routes of DNA repair, he discussed in detail the following paper:
There were 12 participants. The paper was quite complicated and for a proper understanding one should understand about (i) the goals and challenges of genetic engineering/therapy, (ii) the Cas9/CRISPR system, and (iii) the HDR and NHEJ routes of DNA repair. On top of that, we, including Prof. Alexander, commonly agreed that the paper could have been clearer and more complete in several of its descriptions. Prof. Alexander explained the contents of the paper very well, but without prior reading even I (who has some background in this field) probably would have had difficulties to follow the detailed story. Therefore, as an organizer, I was very glad that Ms. Sarantuya Enkhjargal, M.D., who is a graduate student at Fujita, expressed that she had read the paper beforehand and very much enjoyed the journal club event.
Prof. Alexander has a lot of experience with using viral vectors for gene manipulation and will use the construct described by Jayavaradhan et al. in his own research. This intimate knowledge showed in his comfortable answering of our questions. As always, many participants engaged actively in the discussion (for their questions see below). For me, the event was very gratifying because we learned more about one of the urgent research topics in medicine (how to optimize Cas9/CRISPR for use in gene therapy), and I enjoyed the puzzling together with other researchers about an article.
The contents of the presentation
For this blog post, I will not make a detailed description of the contents of the presentation and article but only a summary.
Background
Cas9/CRISPR. (As most of you know:) Many bacteria use a Cas/CRISPR system as an adaptive immune system for specifically recognizing and cutting genetic sequences of phages (viruses). CRISPR (clustered regularly interspaced short palindromic repeats) is a bacterial genomic region where DNA fragments derived from phages are stored between repeats. Cas9 (CRISPR-associated protein 9) acts as a single protein together with a guide RNA (gRNA) complex that consists of a CRISPR RNA (crRNA) and a trans-activating cRNA (tracrRNA) to specifically find and cleave DNA that matches with the unique (“spacer”) portion of the crRNA (Fig. 1). The finding that this bacterial system can also efficiently and precisely make doublestranded DNA (dsDNA) cuts in other organisms—upon recombinant expression and replacement of the spacer with a gene target sequence of choice—has revolutionized gene editing.
The HDR and NHEJ routes of DNA repair. The Cas9/CRISPR system only introduces double-strand (ds) DNA breaks and relies on the cell’s endogenous repair routs for fixing the break. The different methods by which a cell can do this can be broadly classified into two pathways, the non-homologous end-joining (NHEJ) pathway which is—more or less—the default system and the homology directed repair (HDR) pathway which is upregulated (but still not dominant) during the S and G2 cell cycle phases when there is more homologous DNA after DNA duplication. HDR integrates a copy of a homologous template at the site of DSB, which in many types of artificial gene manipulation—that in such cases provide a template for homologous recombination—is the desired effect (Fig. 2). NHEJ does not use such template and this error-prone type of repair tends to cause out-of-frame changes in the original DNA (Fig. 2). Because NHEJ is the dominant route of DNA repair, its inhibition leads to unrepaired DNA breaks and increased cell death.
53BP1 (tumor suppressor p53-binding protein 1 also known as p53-binding protein 1).
At the site of dsDNA breaks, the combination of histone modifications and accumulations of (repair) proteins form “DNA damage foci.” The types of proteins determine whether this is followed by NHEJ or HDR. The 53BP1 protein is an early and central element of the NHEJ cascade by directly binding to chromatin, by self-oligomerization, and by binding of the factors RIF1, PTIP (Fig. 3 left), and EXPAND. The RIF1protein keeps BCRA1, an early factor in the HDR cascade (Fig. 3 right), from accumulating at the foci and thereby RIF1 (as thus also does intact 53BP1) inhibits the HDR pathway.
Global inactivation of 53BP1, or of other factors of the NHEJ pathway, causes increased cell death because naturally occurring DNA damage is not efficiently repaired.
The Jayavaradhan et al. study
Rationale
Cells use NHEJ more abundantly than HDR to repair dsDNA breaks, which results in many cells acquiring undesired DNA mutations at the Cas9/gRNA target site when using a Cas9/CRISPR + homologous template system for gene therapy. A possible way to improve the HDR-to-NHEJ ratio is by inhibiting NHEJ. This, for example, can be done by a dominant negative form of 53BP1 that still binds to the chromatin near DSB breaks but lacks the ability to recruit other NHEJ factors. However, if such NHEJ inhibition is done globally, this results in increased cell death. Therefore, Jayavaradhan and co-workers coupled a dominant negative form of 53BP1 to Cas9 in the hope that it would inhibit NHEJ (and therefore increase HDR) only at the Cas9/gRNA target site.
The (Cas9-)53BP1 Constructs
For making dominant negative 53BP1 mutants, of 53BP1 the authors:
– Removed the N- and C-termini which recruit NHEJ factors such as RIF1, PTIP, and EXPAND.
– Retained all or some of the domains: oligomerization domain (OD), glycine-arginine rich (GAR) motif, Tudor domain (this domain mediates interactions with histone H4 dimethylated at K20), and ubiquitination-dependent recruitment motif (UDR; this domain binds ubiquinated histone H2AK15ub).
They tried several variants, which they named (DN for dominant negative): DN1, DN1S, DN2, DN3, DN4 (Fig. 4A). Based on the results, DN1S—which retains all the motifs OD, GAR, Tudor domain, and UDR—was used in most experiments.
Based on results with the above recombinant 53BP1 proteins, the authors made four types of fusion constructs with Cas9: Cas9-DN1, Cas9-DN1S, Cas9-DN2, and Cas9-DN2L (Fig. 4B). The Cas9-DN1S variant was chosen for usage in most experiments. For localization experiments, Cas9 was replaced by a catalytically dead variant (dCas9) so that the molecule complex would remain at the Cas9/gRNA target site.
The Results
(for simplicity of this blog post, the below results are only for experiments using DN1S or Cas9-DN1S)
Results for DN1S
DNA damage foci:
In cell lines, recombinantly expressed DN1S colocalized with endogenous 53BP1 at DNA damage foci (Fig. 5c and -d). This binding was competitive, as concluded from the fact that high expression levels of DN1S prohibited detectable accumulation of endogenous 53BP1 at the foci (Fig. 6). However, see below a question of the audience about this matter.
Expression of DN1S in cell lines resulted, at DNA damage foci, in a decrease in RIF1 (which belongs to the NHEJ pathway) (Fig. 5e and -f) and an increase in BRCA1 (which belongs to the HDR pathway) (Fig. 5g and -h). This is consistent with DN1S inhibiting NHEJ and promoting HDR.
Recombination:
In cells with a GFP-reporter system for NHEJ recombination upstream of the GFP gene, expression of DN1S reduced the number of GFP-positive cells (see the article). Thus, DN1S interferes with NHEJ recombination.
Cells transfected with free DN1S and Cas9 showed higher HDR rates than cells transfected with Cas9 alone (see the article). Thus, DN1S promotes HDR.
Cell viability:
Expression of DN1S in Jurkat T cells and K562 myeloid cells significantly increased their mortality (see the article), probably because the deficiency in the NHEJ pathway led to cells not being able to sufficiently repair DNA damage. Expression of DN1S also reduced the viability of HeLa cells (Fig. 7).
Results for Cas9-DN1S(-gRNA)
The Cas9/gRNA target site and DNA damage foci:
For localization experiments, a catalytically dead variant of Cas9 was used (dCas9). The dCas9-DN1S or dCas9-DN1S/gRNA complexes were not detected at DNA damage foci. The dCas9-DN1S/gRNA was only detected at the dCas9/gRNA target site, which could clearly be seen if the CENPB gene was targeted because this gene has many repeats at the centromere (see the article).
Recombination:
At the target loci, the use of Cas9-DN1S/gRNA compared to the use of Cas9/gRNA increased HDR and reduced NHEJ (see the article and Fig. 8). Quantitatively, the authors summarize this as follows: “Notably, the degree of HDR varied at different loci and in different cell types, but overall, the Cas9-DN1S fusion nucleases increased HDR by 2–3 fold in all the cell lines and gene loci tested, with a 3–4 fold reduction in NHEJ.”
Their medically most relevant result was that, compared to Cas9/gRNA, the use of Cas9-DN1S/gRNA in immortalized B lymphocytes derived from a leukocyte adhesion deficiency (LAD) patient helped to more efficiently insert an intact CD18 gene (which LAD patients are lacking) at the safe AAVS1 locus. In this model, when using Cas9-DN1S/gRNA instead of Cas9/gRNA, the frequency of cells expressing CD18 increased from 23% to 45% while NHEJ repair decreased from 30% to 7% (Fig. 8). Immunofluorescence intensity experiments showed that many of the cells expressing CD18 had undergone HDR at both AAVS1 loci (see the article), and the authors calculated that at the allelic level nearly 70% of alleles were repaired by HDR.
Cas9-DN1S with or without gRNA had no effects on DNA repair, except for at the Cas9/gRNA target site in the case of presence of gRNA (see the article). Thus, the authors succeeded in having DN1S improve the HDR rate of Cas9-induced DNA breaks without this Cas9-fused DN1S interfering with repair elsewhere in the genome.
Cell viability:
In contrast to free DN1S, Cas9-DN1S did not reduce cell viability (e.g., Fig. 7). Thus, the authors appear to have succeeded in retaining the good points of DN1S while removing its toxicity.
Major conclusion by the authors
“Our CRISPR-Cas9-DN1S system is clinically relevant to improve the efficiencies of precise gene correction/insertion, significantly reducing error-prone NHEJ events at the nuclease cleavage site, while avoiding the unwanted effects of global NHEJ inhibition.”
Remarks by Prof. Alexander
Prof. Alexander finds the major conclusions of the article trustworthy, which is why he will start doing experiments with the Cas9-DN1S construct himself.
However, he has some concerns about whether the study results will translate well in vivo because, as he stated:
(i) The study used rapidly dividing cells, and these cells spend much less time in G0 and G1 (at which HDR efficiency is poor).
(ii) Tissue culture cell lines typically do not have a functional p53 pathway leading to cell death. If HDR is very inefficient in G1, a CRISPR-Cas9 double strand cut occupied by a dominant negative form of 53BP1 may persist for a prolonged period of time, and this persistance may result in p53 (or other pathway) mediated apoptosis.
(iii) The amount of CRISPR-Cas9-gRNA and donor template that can be delivered to target cells in vivo is limited.
Questions by the audience
(These are the questions and answers that I remember. They are modified for clarity)
Question 1.
How did the authors decide on the somewhat peculiar quantifications such as, for example, “% of cells with >3 BRCA1+ foci”? Doesn’t that somehow interfere with proper statistic analysis, which in some cases calculates surprisingly low p values even though results are very similar (e.g., see comparison between the irradiated samples in their Fig. 1h) (shown in this blog post Figs. 5 and 11).
Answer: The authors seem not to explain this well.
Question 2.
In Fig. 2c, it shows that DN1S expression significantly increases the percentage of cells with >3 53BP1 foci. This conflicts with a main message of the paper, which is that DN1S competes for forming foci with endogenous 53BP1 (see their Supplementary Fig. 1c in our Fig. 6).
Answer: This is puzzling indeed and the authors seem not to explain this well.
Question 3.
In Supplementary Fig. 5a, it looks as if the dCas9-DN1S fusion without guide RNA does not localize in the nucleus. The respective figure legend also says “absence of dCas9-DN1S …… in nuclei.” In the main text, the authors seem not decided on whether the absence of detectable nuclear staining for dCas9-DN1S is related to this molecule not entering the nucleus or only to not associating with foci. Can you say something more about that?
Answer: The paper, indeed, seems not decisive on this.
Question 4.
There are many proteins involved in the NHEJ and HDR recombination pathways. Why did the authors specifically choose 53BP1?
Answer: The 53BP1 molecule functions at the very beginning of the NHEJ pathway and is a scaffold for several other NHEJ proteins. Moreover, other studies had already shown that interference with 53BP1 function could increase HDR.
Question 5.
The paper finding seems very important. However, it is from a few years back already, and I wonder about the developments since then. Have there been newer studies using this technique?
Answer: Although the data are very promising indeed, there are several complications for their translation into in vivo gene therapy. That may explain why the impact on the field has not been dramatic yet. However, this line of research is being continued, and it is of note that the authors have made their Cas9-DN1S construct available for the research community at addgene for a very small cost and the request for signing an Material Transfer Agreement (MTA).
Question 6.
In recombinant fusions of two proteins it is frequently seen that one or both of the proteins lose (part of) their functionality. Isn’t that a problem in the Cas9-DN1S fusion proteins?
Answer: This is precisely why it may work. Without fusion to Cas9, DN1S is toxic for the cells by reducing the global level of NHEJ through competition with endogenous 53BP1. After fusion to Cas9, DN1S appears to have lost this competitive advantage, except for at the site where it is directed by Cas9/gRNA.
Question 7.
What do you think of the use adenovirus associated virus (AAV) in gene therapy?
Answer: The advantage is that for humans AAVs are not immunogenic. The disadvantage is that the length of recombinant DNA that can be delivered per single AAV is only small, so that for delivery of multiple genes you would need multiple AAVs and hope that they enter the same cell.
Question 8.
If in the future you will have some results with the Cas9-DN1S system, can you let our Journal Club know?
Answer: Yes, I will.
