1/ Happy to share important work done with my co-author Andrew Khalil in the labs of Rudolf Jaenisch @WhiteheadInst @MITBiology @MIT and Dave Mooney @Harvard @wyssinstitute trying to assess and fix the major problem of transgene silencing in human ESC/iPSC based work

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 21/ in summary, this "iPSC-Xon" system overcomes critical limitations in current transgene systems in hPSC-based work, and will benefit from additional improvements. I'm on the job market and would love to bring and improve this technology to other researchers!

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 20/ strangely, leaky SALL1 is not due to unwanted splicing in the absence of drug, but rather the unspliced pre-mRNA somehow escapes the nucleus and is translated in-frame to make the full-length protein. We still don't know how this works.

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 19/ we thought larger transgenes (Cas9 is ~4kb) might be more prone to disrupting the Xon splicing switch, so we tried the ~4kb SALL1 TF, a critical driver of microglia homeostatic function. In contrast with EGFP, Xon-SALL1 is fully leaky and shows aberrant splicing patterns.

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 18/ we also show that we can get inducible gene editing with Xon hPSCs, but only if sgRNAs are introduced transiently. Cas9 has enough leakiness for maximal editing with constitutive lentiviral sgRNA introduced into hPSCs

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 17/ what about other genes? Here we show Xon works for inducible control of GRN expression, and we can overexpress GRN in iPSC-derived microglia, which already express boatloads of it. P2A peptides improve both GRN and SMAD, and importantly reduce N-terminal tag to a single aa.

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 16/ a couple important points here: in addition to resisting silencing and being highly tunable, Xon adds only 24aa N-terminal tag and needs only low nM drug doses, whereas degrons add ~500bp tags to BOTH termini to be effective, and need low uM drug doses.

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 15/ a note of caution is that while background EGFP is minor compared to even the lowest drug induced expression levels, this may need mitigation when using TFs or other strong cell state regulators

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 14/ Here's the commonly used H1 ESC background, differentiated into microglia with an embryoid body protocol (we used EB and monolayer)

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 13/ quantification shows how tunable this system really is! Below the 100nM dose, linear correlations between drug dose and EGFP is basically perfect both before and after differentiation

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 12/ again with AAVS1-targeted EGFP, what we saw was remarkable: first time we've ever seen robust inducibility after long differentiations, this time 4-7 weeks, across hPSC backgrounds and differentiation protocols!

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 11/ we then adapted the recently described "Xon" splicing switch, from this paper: https://t.co/eqMoGXurgK Could this approach work in hPSCs, resist silencing and allow for robust inducibility in differentiated microglia and macrophages?

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 10/ we also looked at leakiness, and leaky EGFP varied by system, with Tet-On being pretty tightly controlled (although silenced later on), and degrons showing variability. Note lack of degron tunability in panel C.

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 9/ AAVS1 targeting with identical selection methods, EGFP as the transgene, CAG promoter for constitutive expression, and ~4 week macrophage differentiation. Here, degrons vary in performance, but all show no tunability and only modest inducibility after differentiation.

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 8/ We next tested degrons, as this have been used successfully for transgene expression specifically in fully differentiated cells. However, their leakiness, tunability, and inducibility hasn't been thoroughly tested. To compare with Tet-On we used the same exact approach -->

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 7/ While these systems worked in hPSCs (albeit with slower kinetics), they still got silenced during differentiation

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 6/ we then tried to fix this by adding a Tet-Off-TET1 fusion protein, as DNA methylation has been proposed to underlie Tet-On silencing. Here's the idea, with existing systems (top) getting silenced before DOX is added, and new systems (bottom) we hoped would fix it

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 5/ First we showed that Tet-On-based transgenes are silenced early during microglia or macrophage differentiation, regardless of transgene delivery method, transgene components, hPSC background, or differentiation protocol

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 4/ We compared multiple existing transgenic systems, tried to fix the ones that don't work, and developed a new system that overcomes many existing problems, using #microglia and #macrophages as test cases #Immunotherapy #neuroinflammation LINK:

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 3/ this is a major problem for both basic biomedical research in disease-relevant cells/tissues, as well as CRISPR screens and other systematic gene/pathway dissections that must be done with careful control at the right stage of differentiation. #ESC #iPSC #genomics #synbio

@WhiteheadInst @MITBiology @MIT @Harvard @wyssinstitute 2/ we cannot currently alter genes and pathways in many differentiated hPSC-derived cell types without potential effects on the differentiation itself, due to transgene silencing, lack of transgene tunability, and toxicity or unwanted effects of electroporation and transduction