Ross D. Jones, PhD

For the first time, researchers at UBC have demonstrated how to reliably produce #TCells from stem cells in a lab setting.

See the work from Dr. @jonesr18.bsky.social, @kevinsalim.bsky.social, and the Zandstra & Levings labs: https://bit.ly/4qJcXhi

Abstract: https://bit.ly/45QKIFo

Thanks! Let us know if you have any questions!

UBC research associate Dr. Ross Jones in the lab where they are working to develop cell-based therapies from stem cells.

Canadian breakthrough in stem cell engineering paves way for next-generation living drugs www.med.ubc.ca/news/stem-ce... @jonesr18.bsky.social @ubcmedicine.bsky.social

More news → CanadaHealthwatch.ca 🍁

A pug chomping a tomato, or at least attempting to!

If you made it this far, thanks for reading! Your reward is a photo of my Pug (Elliot) being not-too-sure about a tomato! He did finish it not long after this 😋

Thanks to funders: @engagewcs.bsky.social Wellcome Leap, @genomebc.bsky.social, CIHR, NSERC, and @healthresearchbc.bsky.social

@sbmeubc.bsky.social @ubcpress.bsky.social @bcchresearch.bsky.social

The iPSC-T cell platform in our lab continues to grow, with more cell types, more improvements to CD8/CD4 production, and CAR/TCR integration ongoing! It's an exciting time to be in this field, with a lot to learn, even more to do! Thanks to all who contributed to this work!

Overall, we think this is a big step for expanding the potential of allogeneic, stem cell-derived cell therapies, while also teaching us more about T cell development along the way. We can now learn a lot without needing mouse models, and we hope to keep learning more while we scale up for therapy!

Schematic for CITE-seq + scTCRseq experiment. iPSCs were induced to an approximately equal ratio of CD4 and CD8 T cells, then expanded and sequenced.

Beyond this, we also looked in detail at cell phenotypes via CITE-seq and measured TCR repertoire composition via scTCRseq. The results were mostly as expected with CD4-relevant gene expression and diverse V(D)J-generated TCR sequences (with some twists of course!) See the paper for more details!

Schematic of polarization study, cells were maintained in non-polarizing conditions (Th0), or polarized to Th1, Th2, or Th17.

Graphs comparing cytokine expression (IFNg, IL-4, and IL-17A/F) in iPSC-, thymic-, and blood-derived CD4 T cells.

Kevin also put the cells through polarization assays and confirmed that they could change phenotypes mostly as expected. However, like with thymocyte-derived CD4 T cells, these changes weren't as clean as with primary blood T cells, again likely reflecting the extreme naivety of our cells.

Graphs showing induction of co-stimulatory markers CD71, 4-1BB, OX40, and CD40L on iPSC-, thymic- and blood-derived CD4 T cells.

When stimulated, our iPSC-CD4 T cells up-regulated co-stimulatory markers similarly to thymus-derived CD4 T cells (but less so than primary blood CD4 T cells). This likely reflects the cells being at a similar stage to mature thymocytes that have not yet reached a fully-mature naive state.

Graphs showing expansion of iPSC-CD4+ T cells over time and retention of CD4+ CD8- phenotype.

Kevin did a lot of functional testing, finding that the cells could expand following additional TCR stimulation while largely retaining the CD4+ CD8- phenotype (this is working in both IL-2- and IL-7/15-based expansion medias)

Now that we're through the overly-detailed mechanism bit, we can zoom back out and look at the cells we made. Are they legit, or do they just happen to express CD4 and not CD8? Fortunately, I had a lot of help by my co-author Kevin (Levings Lab) and others in the Zandstra Lab!

Notably, these findings of Notch's effect on GATA-3 and ThPOK were also recently seen in a pre-print describing a new mouse model of conditional Notch over-expression

www.biorxiv.org/content/10.1...

Graphs showing effect of different TCR/Notch stimulation conditions on GATA-3, ThPOK, and RUNX3 levels over time. Inset T, D, S indicate statistical significance for Time, DLL4/DAPT presence, or TCR Stim level (x indicates 2D/3D effects via ANOVA)

Looking at key lineage-specifying transcription factors, we can see another important effect. Notch reduces activation of GATA-3 after TCR stimulation, and completely blocks activation of the CD4 master factor ThPOK. Meanwhile, the CD8 master factor RUNX3 is boosted by CD8-optimal TCR stimulation.

This connects nicely with the kinetic signaling model where CD4 lineage induction requires stable downstream TCR signaling dynamics. Akt is an interesting counter-case that we speculate may connect to IL-7 signaling - see the paper for more details/speculation.

Phosphorylation of key TCR signaling pathways members over time following different TCR and Notch stimulation inputs that differentially favor CD4 or CD8 T cells. Inset T, D, S indicate statistical significance for Time, DLL4 presence, or TCR Stim level (x indicates 2D effects via ANOVA)

Diagram showing important TCR signaling proteins and the CD4 vs CD8 lineage induction circuit, which is essentially a bistable switch between transcription factors ThPOK (CD4) and RUNX3 (CD8), with GATA-3 also playing an important role for the CD4 lineage.

So how do our conditions tie into the underlying mechanism of lineage choice? One answer comes from looking at TCR signaling kinetics, where we see that CD4-optimal input conditions (light blue) lead to much more stable NFkB and Erk signaling than CD8-optimal (dark red), and less Akt activation.

Importantly, a paper came out during our revisions that also used PMA to make CD4 T cells. They showed some success w/ CD4 T cell-derived iPSCs, but predominantly got CD8 T cells when using non-T cell-derived iPSCs (like us). Nevertheless important work, esp making Tregs!

doi.org/10.1093/stmc...

Further, while PMA+Iono achieved the highest purity of CD4 T cells, cell yields were much lower than with the other reagents. It had to be dosed way down compared to normal assays with mature T cells to avoid toxicity.

Flow cytometry plots comparing optimal conditions for inducing CD4 T cells with different TCR-stimulating reagents. Notably, PMA + Ionomycin-induced cells express less of the Naive T cell markers CD27, CD45RA, and CD62L.

Nicely, the CD4 T cells induced with anti-CD2/3/28 and PHA reagents look really good and express expected markers of mature thymocytes / naive T cells. Not so much for PMA+Iono-induced cells, which seemed to be stuck in a partly-differentiated state.

A graph comparing the percent of cells going to the CD4 vs CD8 T cell lineage, plotting all combinations of different TCR and Notch stimulating inputs, highlighting the tunability of the fate switch based on these inputs.

Combining different doses of Notch and TCR-stimulating reagents, we now had nearly complete and tunable control over lineage induction between CD4 and CD8 T cells, hence the title of the paper 😁

Graphs showing dose-dependent effects of anti-CD2/3/28 on CD4 or CD8 T cell induction when cultured on different concentrations of the Notch ligand DLL4. Intermediate TCR stimulation inputs yield the most CD4 T cells.

Graphs showing dose-dependent effects of PHA-based TCR stimulation on CD4 or CD8 T cell induction when cultured on different concentrations of the Notch ligand DLL4. Intermediate TCR stimulation inputs yield the most CD4 T cells.

Graphs showing dose-dependent effects of PMA + Ionomycin-based TCR stimulation on CD4 or CD8 T cell induction when cultured on different concentrations of the Notch ligand DLL4. In this case, higher-range TCR stimulation inputs yield the most CD4 T cells, though note the doses are still VERY small compared to normal usage of these reagents. Higher doses kill the cells.

Dotplot graph showing dose-dependent TCR stimulation by anti-CD2/3/28, anti-CD3/28, anti-CD3 (coated vs soluble), all in the absence of Notch. There is an optimum for CD4 T cell induction for each (though that is more spread out for soluble anti-CD3). The color intensity is yield and the circle size corresponds to population frequency.

Later, we repeated this with different doses (notches) of the strong Notch ligand DLL4, and saw the same effect. Likewise with the TCR-stimulating lectin PHA, PMA+ionomycin (will come back to this one) and other anti-CD3-based reagents.

Graphs showing dose-dependent effect of CD4 vs CD8 T cell generation by stimulation of DP T cells with anti-CD2/3/28 complexes. The optimal percent and yield of CD4 T cells are at ~0.1% and ~0.3% reagent, respectively.

But it turns out that for nearly all the TCR-stimulating reagents we've tested, there is a Goldilocks zone not just for avoiding cell death, but also for making CD4 T cells. In early experiments where we had not yet removed/blocked Notch, it looked like this, with just a small bump of CD4 T cells.

When we make T cells from iPSCs, we don't usually let the cells randomly go through positive selection on their own. Instead, we typically stimulate the TCR with antibodies or other proteins to replicate positive selection, greatly increasing the fraction of surviving cells.

Thus, there is a kinetic difference in TCR signaling that quantitatively leads to the two different lineages. This is called the "kinetic signaling" model. Updated models account for other complexities, but this basic phenomenon remains and is critical.

But how does one receptor binding to two different proteins lead to different outcomes? The details are an long debate, but the simple version is that after TCR stimulation, the CD8 protein goes down, interrupting signaling from TCRs that recognize Class I proteins only.

Within the Goldilocks zone, cells become CD4 or CD8 T cells based on whether their TCR was stimulated by MHC Class II or Class I proteins, respectively. The immune system uses these proteins to surveil the environment (Class II) or the inner working of every cell (Class I).

The process of going from DP to CD4 or CD8 T cell is called positive selection, and is driven by TCR signaling. Too much or too little signaling, and the cell dies. Just right, and it lives. This Goldilocks zone is part of an intricate mechanism to prevent T cells from attacking healthy cells.

But Notch isn't the only story. Some papers already didn't include Notch ligands during the DP to CD8/CD4 T cell step, but still lacked CD4 T cells, so what gives?

TCR signaling!

(Note that cytokines also matter a lot, but weren't a focus since we could get CD4+ T cells without changing ours)

Graphs showing dose-dependent effects of the Notch Inhibitor DAPT on DP to CD4 or CD8 T cell induction. As more DAPT is added, the resulting cell fates switch from CD8-dominant to CD4-dominant.

While iPSC-derived CD8 T cells were seemingly "easy" to make, CD4 T cells were elusive. We now know that this was in part because of continued, supraphysiological Notch stimulation when DP T cells are choosing the CD4 or CD8 fate. Blocking Notch restores CD4 T cells in a dose-dependent manner.

A cartoon of Notch ligand DLL4 plus the adhesion protein VCAM working together to induce T cell lineage genes in developing cells. Shukla 2017: doi.org/10.1038/nmeth.4258 Primbetova 2024: doi.org/10.1089/genbio.2024.0010

Cue iPSC-T cells, protocols for which started popping up around 2009. To make these cells, iPSC-derived blood stem cells are grown in the presence of a lot of strong Notch ligands (initially, attached to co-cultured "feeder" cells, more recently attached to the plate or other surfaces like beads)