Validation Study to Determine the Accuracy of Widespread Promoterless EGFP Reporter at Assessing CRISPR/Cas9-Mediated Homology Directed Repair

Round 1

Reviewer 1 Report

This manuscript describes and evaluates a visual reporter system that has been used in some other laboratories as an assay for homology directed repair.  These types of assays and their robustness are critical to the field of human gene editing because the gold standard for genetic engineering in humans is the repair of an inherited mutation or the total disablement of a malfunctioning gene.  Due to the complexity of the human genome, it is quite challenging to assess the effectiveness of such repair or replacement through the process of homology directed repair.  Scientists have traditionally developed biochemical or genetic readout systems that include a fluorescent tag or a selectable marker that can facilitate a visualization of the efficiency and efficacy of the gene editing outcomes.  One of these assays is based on the integration of a reporter gene (EGFR) that lacks a promoter.  In general, this type of assay is problematic at several levels, but in this case, the authors point out a central and  significant problem.  They report an inability to generate quality data in a reproducible fashion.  While the authors stop short of questioning the validity of the original assay system and subsequent papers that have used it, they do point out a number of errors that are present in the assay.  The paper is somewhat unique in the sense that it does not directly produce any new information but rather points out the limitations , robustness surrounding the previously described assay.  In fact, they point out that in their hands to control levels that are used as the standard against which homology-directed repair activity is judge, are flawed.  I have no specific experimental changes for the authors, but prior to publication, I would request they consider the following.  Since it is paramount to their argument that the expression levels existing prior to the action of CRISPR/Cas be presented , more detail should be given on the levels of expression and why they exist in the absence of gene editing activity.  It would also be informative and important to the scientific community if the authors provided a chart of papers that have purportedly used this assay to conclude specific things about homology directed repair.  A chart containing the authors, the name of the paper and the primary conclusions as well as the citation would strengthen the argument that this is not a one- off assay system, but rather one that has been percolating through the field for a number of years.  I personally do not think this assay has a lot of value, and I do believe the authors have struck upon the best way to inform the scientific community of its frailties.  I would also request that the authors spend a little bit more time strengthening in their conclusions.  A side-by-side comparison in either chart or narrative form comparing what others have found supporting this is a system I what they found in contrast, will also be helpful to the reader.  These kinds of papers are important in the field even if they point out the lack of robustness and reproducibility.  The authors should be commended for doing a detailed study of something that could have ramifications in the field if this assay/readout is relied upon to construct specific conclusions.

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors describe an issue they have encountered while cloning a deemed promoterless HDR template, which, despite their best efforts, turned out to contain transcription promoting elements. This sentence sums up most (if not all) of the MS.

I do apologize if I will come across very blunt, but this observation is not new, the ideas are not new, and the donor most probably is wrongly designed. Major concerns: (A) Everyone who is doing HDR experiments does test any donor for potential leaky reporter expression first thing. (B) If a design-flaw allows leaky expression, it will be fixed and corrected. (C) Linear donors are known to randomly integrate since decades. (D) 293T and similar cell culture cell lines are known to retain plasmids for several weeks to months.

I do not detect any novelty in the MS with my best intentions to find it. As such, I am having a hard time finding a justification to allow this MS to be carried on anywhere further.

It is a single observation based on a design error in the donor that potentially could alert researchers to check their donors in a “non-Cas9/non-sgRNA” setting and screen for unexpected expression. However, this is what everyone should do anyhow with any construct as good laboratory practice (called negative control).

 

I am adding my detailed concerns below in case the authors would like to re-submit their MS to another journal.

The introduction is fine. The authors draw the attention to several HDR testing approaches: (a) KI of reporter alleles, (b) SNP conversion of BFP to GFP and FACS evaluation and (c) 2A-tagging of endogenous genes, especially at the CTD to avoid bringing in promoter elements (lines 32 to 81). Concern and major suggestion: Of all promoterless donors reported in the literature, the authors could test those by simple transfection and see if any of them has the same issue as their donor? If so, this would bring previous publications into discreditation, if not it would point into the direction that the author’s donor was wrongly constructed.  

Line 82: “we established a promoterless eGFP reporter”. This is factually wrong since transcription emanates from the plasmid as repeatedly demonstrated. Concern: this points to a wrong/bad construction of the donor. I can tell from personal experience that none of my 100+ CTD donors ever gave rise to a leaky reporter expression.

Materials and Methods section is good. Concern: the details of the donor assembly (lines 128 to 137) could be strengthened if the authors would publish the entire clonechart to see if there is a hidden TSS/ATG upstream of the eGFP part? No TFBS database will ever pick up such a cryptic site.

The donor was linearized prior to electroporation into porcine cells (line 143). Concern: Linear DNA has a very (!) high propensity for random genomic integration, and as such the 2A could pick up several other promoters, explaining the high eGFP background level. The entire beauty of the CRISPR system (except you are aiming for HITI) is that the cut is in the gDNA and NOT in the donor?

HEK293T cells show eGFP expression >10 days post transfection (line 165). Concern/comment: 293 are well known to engulf any plasmid DNA and keep expression levels up for several months if challenged. This is another hint into the direction that the chosen donor has cryptic initiation signals.

Results in line 198 claims “Successful establishment of the promoterless EGFP reporter system”. As above, this is factually wrong.

Results part in line 232/234 claims “This unexpected EGFP expression without specific DSB in the target sequence would confuse the assessment of CRISPR/Cas9-mediated HDR efficiency.” CONCERNs: With due respect, this statement can only invoke the conclusion that the donor is badly designed and does not work as anticipated and you want to sell that as a story? Removal of fragments of the homology arms influences non-specific eGFP expression but does not abrogate the effect. In my understanding the only remaining question is: does the 2A-eGFP have an ATG that will allow expression. Presumably yes, and this makes the entire construct design faulty. Omission of the ATG will only allow expression upon correct fusion to the previous exon.

In the subsequent part the authors test their faulty donor in various cell lines and with different HA deletions and conclude it is still expressed. This is not really unusual or unexpected when the donor is wrongly designed?

Random insertion (line 278) is predicted to be the predominant outcome due to linearization of the donor. The authors could try and see if other promoters drive expression (random insertion into a correct frame after splicing) or if it is inherent to their donor design. Apart from removing the ATG and go on with the real experiments?

Minor comments 

Line 78:          “combing” must be a typo

Line 140         “linearized electroporation buffer” must be a typo.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Overview

Accurate assessment of HDR rates in mammalian cells is of a critical importance for development of methods to increase the efficiency of HDR-mediated knock-in. Several reporter systems are commonly used including Traffic Light Reporter and its modifications, promotorless fluorescent protein KI systems or systems which are based on the correction of non-functional fluorescent protein. For reliable measurement of HDR efficiency which could be quite low, it is of a great importance to reduce non-specific expression of reporter constructs. In the work by Hu et al, the usage of widespread promotorless EGFP reporter was critically evaluated. Here authors transfected several porcine, human and hamster cell lines with different forms of promotorless plasmid and found that EGFP reporter was quite efficiently expressed in the absence of promotor and sgRNA-Cas9. The observed phenomenon is largely due to random integration of the reporter plasmids into genome.

Major issues

As stated in the Results section 3.1 the promotorless reporter plasmid contained EGFP CDS, i.e., it contained an ATG codon. In my opinion the design of the donor plasmid with P2A-EGFP cassette should not include ATG before EGFP CDS to minimize non-specific EGFP expression/translation. Authors should delete the ATG codon in the EGFP CDS while preserving the reading frame with GAPDH and compare such construct with those used in the study.

Lines 209-210: The evaluation of sgRNA efficiency refers to Fig1B and Suppl fig1B. First, Suppl Fig1B show pDNA electrophoresis, not the results of sgRNA evaluation. Second, there are no “multiple” peaks near Cas9 cutting site in any shown Sanger sequencing file. To evaluate and compare the efficiencies of sgRNAs authors should preform T7EI or Surveyor assay. Low efficiency of knock-in into the target locus could be attributable to poor sgRNA performance.

The discussion section should include the discussion of possible ways to overcome this non-specific expression of EGFP.

Specific commentaries

Fig1A – The targeting scheme is too small; it should be enlarged. The same notion is applied for Fig4C.

Line 89 – Typing error: Specie -> Species

Line 140 – “Linearized electroporation buffer” -> I guess the reporter was linearized, not the buffer.

Lines 235-240: For Fig2A it is better to show the schemes of linear restrictions fragments that were used as donors instead of referring the readers to Supp Fig1B showing the circular pDNA map.

Lines 284-285: The notion is similar to that for text in lines 235-240, for better understanding it is worth showing the schemes of BbsI and NsiI-linearized fragments. Moreover, BbsI site is not even shown on the map in Suppl fig1B.

Section 3.4 starts with the text refereeing to Fig4a, and then continues with the text for Fig.5, but Fig4c is not discussed.

In fact, authors have not shown correct knock-in of the reporter into GAPDH locus. When knock-in efficiency is low, the PCR-based detection of the correct integration may be not sensitive enough. In order to increase the sensitivity of the system authors could apply the antibiotic selection to enrich for cells with integrated Neo cassette and to reduce the number of cells with transient expression of reporter genes. That might help to detect cells with correct knock-in by genomic PCR.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors provided text to my suggestions, however, did not provide any answer or made the MS better. I am afraid, the MS still is a plain description of a failed HDR donor assembly in no context to CRISPR HDR as claimed in the title.

I appreciate it is interesting that few/some promoter-less reporters do allow transcription/translation or do catch a neighbouring promoter by random insertion. However, the vast majority of "promoter less" HDR donors actually is not and will never be expressed unless properly recombined. The conclusion derived from some outlying examples is not substantial, and even with best intentions I cannot warrant accepting this MS as a solid scientific story.

I am afraid but I have to stick to my previous decision and recommend rejection of the MS.