#  Yuanfan Sun, Yingyin Wu & Colin Hemez 

 



    ![Sun_Wu_Hemez](/sites/g/files/omnuum4021/files/styles/hwp_5_4__480x385/public/2025-04/Sun_Wu_Hemez2.png?itok=NiTJ0zqT) 

 



 

####  calendar\_today Date and Time 

 **April 14, 2025** 

 01:00PM - 02:00PM EDT 

####  pin\_drop Location 

 **https://harvard.zoom.us/j/94394339529**  



 

 [ https://harvard.zoom.us/j/94394339529 arrow\_circle\_right ](https://harvard.zoom.us/j/94394339529) 

 



 

 ![Sun Wu Hemez](/sites/g/files/omnuum4021/files/2025-04/Sun_Wu_Hemez.png)

 

**Yuanfan Sun** & **Yingyin Wu** ([Rui Zhang lab](https://lifesciences.sysu.edu.cn/en/teacher/1079), Sun Yat-Sen University)

[Type III CRISPR-mediated flexible RNA excision with engineered guide RNAs](https://www.sciencedirect.com/science/article/pii/S1097276525000553)

Molecular Cell, 2025

**Colin Hemez** ([Liu Lab](https://www.liugroup.us/), Broad Institute)

[Systematic optimization of prime editing for the efficient functional correction of CFTR F508del in human airway epithelial cells](https://www.nature.com/articles/s41551-024-01233-3)

Nature Biomedical Engineering, 2025

**Recording:** [https://harvard.zoom.us/rec/share/5Sjs-tX\_EznuvQCwhAG5rKqMofnFGmL2eECu2d-C2yoolWuywwvm2qZw8Ymbys1K.WF6qXw8Aohpe\_s9C](https://harvard.zoom.us/rec/share/5Sjs-tX_EznuvQCwhAG5rKqMofnFGmL2eECu2d-C2yoolWuywwvm2qZw8Ymbys1K.WF6qXw8Aohpe_s9C)

**Follow-up Q&A:**  
  
*Dandan Mao (Guest) 1:52 PM*

*Great work! Is it possible to fix allels in multiple sites?*

Yes! It is possible to use the same pegRNA to fix multiple adjacent alleles. CFTR F508del is an interesting case of this, because another CF-causing gene variant is I507del—a deletion right next door to the one for which we optimized a PE strategy. We are currently looking into whether or not our F508del-optimized strategy is also effective at correcting I507del.

We know a little bit less about how effective it would be to codeliver multiple pegRNAs to correct multiple alleles in multiplex. However, this sort of thing has been done with base editors, and editing is often observed at both targeted sites in more or less equal measure. So my guess is that multiplex prime editing strategies are feasible.

*Clare Lewis (Guest) 1:53 PM*

*What was the were the improvements for the PE6c? How do you design the dsgRNA for accessibilty?*

PE6c is a prime editor that uses an engineered and evolved Tf1 reverse transcriptase. It is particularly good at reverse transcribing long RTTs in a prime editing context (i.e. 30 bp or longer) as well as AT-rich sequences, but it is also good across the board. For more information, see Doman, Pandey, Liu, et al., Cell, 2023.

You can design dsgRNAs the same way as you would design sgRNAs, selecting protospacer sequences adjacent to the pegRNA target protospacer. The dsgRNAs we used have 6 nucleotides truncated from the 5' end of their protospacer. We found no clear pattern in where the dsgRNA is targeted relative to the pegRNA and the dsgRNAs effectiveness at improving editing efficiency (i.e. on the same strand or on the opposite strand, 5' of the pegRNA or 3' of it—all can lead to substantial boosts in editing). We recommend trying a few different dsgRNAs within ~100 bp of the pegRNA target site.

*Antonino Napoleone (Guest) 1:55 PM*

*Thanks for the talk. Two points: 1. What does the delivery landscape look like in this case? What are the best and most effective delivery options for your application and what are the main challenges?*

There are lots of promising delivery options available for prime editors, including AAVs, LNPs, and VLPs. A major challenge for any delivery strategy is to find a delivery modality that efficiently transduces relevant tissue types—LNPs are very good at targeting hepatocytes, and some AAV serotypes have very potent delivery to the central nervous system, for example. There is no clear winner yet on which delivery technologies will be most effective at targeting the lung and the airway, a major target for any genetic therapy for cystic fibrosis. But progress has been made in lung-targeting LNPs (see SORT LNPs) and AAVs (see AAV6.2FF).

2\. Looking at the conditions that need to be tested for PRIME editing, would AI or machine learning help fill this gap and what do you think about that usage for therapeutic applications?

There is definitely room for ML-inspired design to improve prime editing. PRIDICT and DeepPrime are two ML-based models used to generate efficient pegRNAs. There is likely also room to use protein design methods to engineer more efficient prime editors.

*Clare Lewis (Guest) 1:57 PM*

*Have you found optimization in the HEK cells does not translate into the other cell line?*

We have found that certain enhancements, particularly those that we hypothesize modulate target site accessibility (i.e. ngRNAs and dsgRNAs), do not always correlate well between cell lines. Enhancements intrinsic to the prime edit (protospacer, PBS length, RTT length, prime editor protein variant) do often correlate between cell lines well. We recommend optimizing ngRNAs and dsgRNAs in therapeutically or functionally relevant cell types, when possible.



 

 



 

 

 Share on:- [     Facebook ](#)
- [     Twitter ](#)
- [     Linkedin ](#)
 


 Save: [ Add to calendar calendar\_today ](https://gess.hms.harvard.edu/node/1887671/event-feed.ics)  Copy link link