Thread by @GallowayLabMIT

So you want to engineer your hiPSCs, but targeting DNA payloads requires multiple slow, inefficient steps for each construct. What if we could accomplish multi-site integration seamlessly? Come hear about STRAIGHT-IN Dual now out at Nature Biomedical Engineering! 🧵

Link at end!

STRAIGHT-IN Dual enables simultaneous, allele-specific, single-copy integration of two DNA payloads with 100% efficiency within one week. Engineering the landing pad (LP) with a promoter trap supports 100% efficiency through selection (e.g. burn-the-hay strategy). (2/n)

STRAIGHT-IN Dual employs a single recombinase, Bxb1, to integrate two donor plasmids carrying payloads using the attP/attB-GT and -GA sites. Positive selection enables the enrichment of cells carrying the payloads from ~1% to ~100%. (3/n)

One special feature of STRAIGHT-IN Dual is that you can specifically target one (or two alleles based on the donor plasmids that you deliver). The orthogonality of the recombination sites can be seen by selection or by FPs. GT donors go to the GT allele and GA goes to GA. Beautiful! (4/n)

For quantitative biology, this platform is going to be exceptionally powerful. Integrating the same fluorescent reporter into both landing pads resulted in an ~2-fold increase in fluorescence signal, suggesting transgene expression levels are similar from both CLYBL alleles. (5/n)

Accessing this level of quantitative control over the expression of two cargoes opens up so many previously inaccessible. So how fast and efficient can we get to large numbers of uniformly integrated lines? (6/n)

Using eeBxb1 from @davidrliu ’s lab and p53DD modRNA, we generated ~10-fold higher rates of integration with 10% of cells with the payload (pre-selection), providing more rapid expansion to large cell numbers of uniformly integrated hiPSCs. How fast can we go? (7/n)

The entire integration procedure can done in less than a week with ~100% of the cells carrying and homogeneously expressing the payloads for a single allele or for both alleles. 🚀 (8/n)

The optimized rapid integration protocol also allows dual and simultaneous integration and enrichment of cells carrying and expressing both payloads within one week, offering unprecedented combinatorial screenings. But there’s more! (9/n)

🚨Flanking sequences in landing pads can impact expression!!! 🚨
You can see how tdTomato expression increases when we deliver Cre to excise the sequences flanking the GT allele(compare minus to 3 modes of Cre delivery).
Excision of the auxiliary elements also enables recycling these markers for downstream applications! doi.org/10.1038/s41596…. (10/n)

We knew that optimizing excision for STRAIGHT-IN Dual would be important if we wanted to achieve a complete workflow for markerless and virtually scarless integration of the payload. (11/n)

With some optimization, we could perform the entire procedure (integration and excision) with a single payload in the same 24-well WITHOUT having to harvest and replate the cells. We maintained ~100% efficiency in both integration and excision steps using Flp-2A-BleoR modRNA for excision. (12/n)

With this speed, efficiency, and reduced handling times, STRAIGHT-IN Dual should easily scale to characterization of libraries of parts and circuits for characterization and functional screening via automation. (13/n)

With the power of two alleles, there’s an enormous advantage in controlling and observing allele-specific expression profiles and precisely narrowing down some thorny challenges in cell engineering (14/n)

With two alleles, we turned to examining sources of transgene silencing. Transgene silencing impedes robust circuit performance (10.1016/j.cels.2022.11.005). There’s a number of potential sources of silencing, but how much does each affect silencing? (15/n)

First, we looked at promoters. Putting CAG and EF1a at the same site with different FPs, we find that the hEF1a promoter silences over time. At the same allele, the CAG promoter displays strong and robust expression, indicating the silencing is specific to EF1a and is not just the effect of the locus. (16/n)

Putting EF1a at a separate allele or changing the FP didn’t help either. The problem is clearly EF1a. Given the ubiquity of EF1a and relatively small set of promoters for hiPSCs, our observations stress the need to identify novel promoter sequences to expand the hiPSC toolbox....(see further down for a few more stable promoters!) (17/n)

Since we identified that EF1a silenced, we swapped in CAG for the promoter trap on the donor plasmid. We reasoned that more robust expression from CAG would increase the number of colonies we selected for; and indeed, it did! (18/n)

So we did a much deeper dive to examine promoter stability in hIPSCs and in differentiated cells (see extended figures!). (19/n)

We found that there are very few promoters that are stable in hIPSCs over passaging (CAG, UBC, B-actin, PGK (sorta)) (20/n).

Importantly, the expression is different in transient expression compared to integrated. So we couldn’t use a transfection to reliable predict stability of integrated promoters. This is a VERY important point for folks looking to characterize their circuits! Transfection =/ integration! (21/n)

Integrating a pool and binning the defined set, we could pull out the promoters based on their expression, suggesting we can use this platform to identify new sequences and hopefully more stable promoters. (22/n)

But promoters aren’t the only thing that can lead to silencing. Suppose you have a circuit and you want to stably integrate it. Should you separate transcriptional units? Put them together? Which strategy is better? With two sites with similar expression, we could ask this question (23/n)

To examine the effect of separate vs co-local (trans vs cis) designs, turned to the doxycycline-inducible Tet-On 3G system, which use constitutive expression of the transactivator rtTA to drive expression from the Tet-3G promoter and thus has two essential transcriptional units (24/n)

For more on syntax, stay tuned for our other work on syntax that developed on long with and because of STRAIGHT-IN Dual. (25/n)

To study the effect of syntax, we generated one hiPSC line in which the rtTA and the inducible cassettes are in separate alleles (Trans) and 4 all-in-one hiPSC lines (Cis) that differed in syntax (Convergent, Divergent, Downstream Tandem and Upstream Tandem). (26/n)

Remarkably, we observed major differences between these 5 hiPSC lines! The Divergent and Downstream Tandem syntaxes robustly induce mScarlet while the Convergent is weaker. The Upstream Tandem and Trans designs showed very poor induction. (27/n)

These trends (and particularly the decrease in mTagBFP2 in the Downstream Tandem!) align with our models of silencing based on transcription-induced supercoiling (see here: )... [More on this soon] (28/n)

Focusing on the Downstream Tandem, we explored the induction in cardiac organoids. While doxycycline can fine-tune the expression, duration of induction is more potent in setting the level of expression, demonstrating two distinct handles to control these 3D in vitro models. (29/n)

With the insights from silencing, we turned to examine what functional impact syntax might have in forward programming. (30/n)

We compared the two Tet-On Tandem orientations for their ability to drive cell commitment into neurons by overexpressing Ngn2, a pioneer TF for neurons. We could only achieve forward programming into neurons using the Downstream Tandem syntax. (31/n)

Beyond neurons, we wanted to make motor neurons. Adding Isl1 and Lhx3 to Ngn2 should direct neurons to motor neurons. Surprisingly, how we designed these encoding of these TFs matters. (32/n)

For activation of the definitive motor neuron gene HB9 (shown in yellow), splitting Ngn2 from Isl1 and Lhx3 at different alleles resulted in much more efficient programming! Even adding a second copy of NIL (Ngn2-Isl1-Lhx3) wasn’t as good for making motor neurons as splitting across alleles which surprised us. (33/n)

Having identified a robust forward programming design, we wanted to full leverage the Dual allele system by building orthogonal inducible systems into either allele for dual-independent inducible control of two cargoes. We established an orthogonal inducible system based on the grazoprevir-inducible synZiFTR-NS3 complex from the @MoKhalilLab doi: 10.1126/science.ade0156. Overexpressing Ngn2 using using grazoprevir drove rapid differentiation from hiPSCs into neurons. (34/n)

By combining both inducible systems using STRAIGHT-IN Dual, we can sequentially and independently overexpress two genes of interest using doxycycline and grazoprevir. (35/n)

So can we program two different fates from the identical cells? (36/n)

Can the addition of small molecule determine the direction of differentiation? Indeed, yes! We can add GZV and make endothelial cells by expressing ETV2 or we can add Dox and make neurons by inducing Ngn2. And this is all from the exact identical cells! Very exciting to think about how this could be used for patterning cell fate! (37/n)

Two independent induction systems in hiPSCs opens endless opportunities for SynBio in stem cell biology. One inducer can drive rapid differentiation while the second can control functional biosensors, regulators for specialized cell types, or transgenes to induce disease phenotypes. (38/n)

We’re excited to see how the STRAIGHT-IN platform can be applied in research areas such as stem cell biology, cancer biology, developmental biology, and synthetic biology. If you are interested to know more, let us know! (39/n)

This work was led by Albert Blanch Asensio @A_BlanchAsensio from the Davis and Mummery labs @mummerylab, and who joined us at MIT as a visiting student and then as a postdoc. He worked with @DeonPloessl to get it optimized along with @nbwang22 . Albert set an extraordinary pace on this work, inspiring everyone to hustle and reach for the stars! (40/n)

Building the collaboration with the Davis and @MummeryLab labs was so rewarding. It was a pleasure to work on building STRAIGHT-IN Dual for more of the prior work, see here:
For more, on the development, see below:
v1.0: doi.org/10.1016/j.crme…
v2.0: doi.org/10.1016/j.scr.…
Protocols: doi.org/10.1038/s41596…

(41/n)

Special thanks to our editor @rita_strack at @natBME and the whole team at Nature Publishing who helped move this work forward and wrapped up! Thanks also to our funders including @NIH , @NSF , @AFOSR , @PershingSqFdn and others! In a challenging year, this funding has made a huge difference in our ability to finish the work! (42/n)

It’s been exciting to watch the development and see how this tool is accelerating so many of our projects and ideas! Stay tuned for more as this tool was essential for really pushing the limits of what we could understand about how to design gene circuits that are integrated into the genome!

Link:

(fin)nature.com/articles/s4155…

@A_BlanchAsensio @MummeryLab @DeonPloessl @nbwang22 @rita_strack @natBME @NIH @NSF @AFOSR @PershingSqFdn @threadreaderapp unroll