Small Molecules to enhance HDR

I have tried all the chemicals including the ones you did not include here.
My results are all based on human cancer cell lines, not iPSCs or ESCs.
Therefore, please be advised that it could be different in your cells.

SCR7 never worked in any types of cells tested.
Nocodazole gave me 2~3 fold increase in HR in multiple cell types.
L755507 never worked.
RS-1 did not work in a couple of cell lines tested.
DNA-PK inhibitors worked the best. Between NU and KU, I got better efficiency using NU. (several cell lines have been tested)
You can also use a little more NU compound compared the amount you can find in literature. I use 3X without seeing any toxicity.

Please remember that you need to get enough NHEJ just by doing CRISPR itself.
For example, if you get 10% indels without doing HR, you will get less than 10% of HR at the end since you are trying to get HR by antagonizing NHEJ.

I typically get 80-90% genome modification efficiency (measured by TIDE or Mi- seq) in NHEJ, then I get about 50% HR efficiency by using NU compound.
I also strongly recommend electroporation method over lipofectamine.
I get 98% genome editing in HEK293 in most cases and >70% in other cells including primary cells using electroporation.

I forgot one thing.
IDT enhancer..
The enhancer has been developed in the same story as in this paper.
https://www.nature.com/articles/ncomms12463

It definitely helps when the efficiency is low.
Therefore, when you work with Cpf1, which is known to work a lot less efficiently compared to Cas9, it could be almost mandatory to use the enhancer to boost up the indel efficiency, specially for NHEJ.
In one of my experiments, I got 62% from 5% by adding the enhancer.
In your case, you are trying to do HR. I am not sure how much complication you would get by using two ssDNA, one of HR , the other as an enhancer. worth a try.
However, please remember that this could generate a significant off-target effects.
Although off-target studies from a strategy using ssDNA as a donor for HR have not been well documented, my experience is that it in fact generate significant insertion of your ssDNA into genome (we are still analyzing the data).

IDT's documentation says that the enhancer is unnecessary if you're using an ssDNA donor - the carrier effect of the ssDNA works whether it is homologous or not.

unfortunately I have to disagree with your claim that donor-mediated repair % (recent work by the Corn lab suggests that this process is actually distinct from HR) is limited by total editing % i.e. NHEJ. In my experience there is no clear correlation between the two, and knock-in efficiency can exceed indel %, paradoxical as it may seem. If you must choose between two guides, I think going for proximity to mutation of interest over editing % is advisable (provided the closer guide cuts at least a little).

I am wondering how IDT measures the carrier effect when they used homologous ssDNA donor.
I always include NHEJ only control when I do HR with a ssDNA donor. The total edit in most cases decreases a little bit with the addition of homologous ssDNA.
I may be able to get the answer directly from IDT.

I agree that ssODN-mediated repair mechanism could be different from the classical HDR, as it was reported that RAD51 knockdown does not have much effect on ssODN-mediated repair. This could be reason that I did not any improvement from using RS-1.
I also believe that RN observed no clear correlation between ssODN-mediated repair and NHEJ. It seems to me that RN claims that ssODN-mediated repair is independent from NHEJ.
Before the era of ZFN, TALEN and CRISPR, people in fact had used ssODN in several different forms to modify genome sequence without making dsDNA breaks. However, the efficiency of these methods remained quite low.
Without making dsDNA breaks, how much efficiency of ssODN-mediated repair can we expect? I tried this a couple of times as a control but do not include this control any more since it does not work.
This clearly means that ssODN-mediated repair is dependent on dsDNA break events. I guess this clear effect could be hard to be seen & analyzed if the total edit % is very low.
I have several different gene correction experiments in different cell lines showing clear correlation between the reduction of NHEJ and increase in ssODN-mediated repair while keeping total edit similar level.
I am also thinking that playing with very low % of gene correction may generate some inconsistency between experiments and conditions.
In addition, for the analysis of the sequence, having Mi-seq in the procedure may reduce unintentional misinterpretation of the results.

Therefore, I am wondering how much efficiency you are struggling with.

Yes detection of the point mutations inserted by HDR is difficult.

In my case, my percentage of a successful HDR was possibly ~.1% I had used ddPCR for this: (Miyaoka et al, Isolation of single-base genome-edited human iPS cells without antibiotic selection)

This is the only protocol I know that can give accurate (once optimized) results when it comes to efficiency of HDR repairs.

However, even this system can misinterpret frameshifts for HDR so I cannot be sure of my results entirely

Here is an update about the methods I have tried to increase HDR. However, all the efforts were dropped midway because I was able to isolate clones which appear to have the inserted mutations

Nocodazole: in HEK293T cells and iPSC

Results: Appeared lessened via ddPCR with the addition of Nocodazole before transfection at the recommended dosage and treatment time (citation in the previous post)

Confounding factors: The ddPCR gave me a percentage of editing however I do not know if frameshifts - NEHJ repairs are involved in this percentage

                               Scenarios 1: If Nocodazole reduced frameshifts then less positive signal in the ddPCR.  Therefore ddPCR shows nocodazole to be worse, however, in reality, it may increase HDR 

                               Scenario 2: Nocodazole may stress out the iPSCs -HEK  and this could case editing to be reduced.  I notice they look different at the end of treatment

transient Gentamycin selection with plasmids: Hopefully to select for transfected cells only and aid in mutant enrichment. (Steyer et al 2018) G418 is hard to work with: kill curves were inconsistent. iPS seeding and density affected the kill curves ie clumps lived. Many problems also plasmids and RNP in iPS transfection is difficult to optimize

Other ways to try and get rid of your nontransfected cells may be better. For my cells and my problem, this was difficult. In the background was always doing the sib selection anyway.

MDM2: (Ihry et al 2018, p53 inhibits CRISPR–Cas9 engineering in human pluripotent stem cells) Newer publication. Results are unknown

      Lipofectamine RNP complex transfection with fluorescent tracr RNA (IDT) with or without the p53 inhibitor (MDM2) as described in the citation.

To add more variables I attempted to sort out the fluorescent (transfected cells) however it appeared all the cells were fluorescent to some degree so that may not have helped. FACs sorting out nonfluorescent cells may be more useful in plasmid experiments where fewer cells are transfected.

In my experience enrichment of HDR edited iPSCs is difficult and slow using the sib selection however for your experiment It may make sense to really try and optimize CRISPR and transfection efficiencies and then do typical cloning.

But then it is hard to diagnose which guide works best in vitro and cloning is hard because you could invest a lot of effort to pick clones and have none be edited correctly. From the transfection, my initial T7 assay in HEK cells gave some indication of my best guide. This was confirmed by ddPCR- editing efficiency readout.

So that leaves us with the original problem what is the best way to test systems and then choose the best guide and then how do you isolate clones. Hek cell testing looked like a good way to test which guide was the most efficient however they are not the same cells and results in HEK may not be easily valid in iPS. For example, my most efficient guide in HEK cells was able to generate a correctly edited mutant. However, in iPS I only got one mutation integrated with this guide. Eventually, I switched guides to my second most efficient and got correct editing. I only knew this after months of mutant enrichment via sib selection and cloning. Unfortunately, results come at the end of months of enrichment in culture for me. This may not be the case for you because my mutations affected the maintenance of pluripotency, therefore, the mutant grew slower in culture (harder to enrich).

Anyway, there may be other ways that I need to look into more about how to do this, especially for mutations that can change the way iPSCs grow.

I heard something about: sleeping beauty - piggyBac transposons for the introduction of a selectable marker and HDR. Then theoretically the selectable marker can be removed once the clone is isolated leaving the scarless HDR. But I need to read through the literature to verify.

If anyone knows of any papers or has any suggestion about the following problems they would be very appreciated

Transposons for Scarless HDR

Better detection methods (efficiency of HDR insertion of base pairs in the first transfection)

Drugs for HDR in iPS

Another method that can work as a quantitative measure of initial knock-in efficiency is RFLP-based DNA capillary electrophoresis. I used it as part of my sib selection protocol (https://blog.benchling.com/2017/03/28/gblocks-crispr-human-ipscs/).