Thread by @connor_rosen

More on immune evasion in glioblastoma! Online now in @NatureMedicine, intracranial tumors drive sequestration of T cells in the bone marrow through downregulation of S1PR1 receptor surface expression. nature.com/articles/s4159…

GBM patients, including treatment-naive, and mice with a variety of intracranial tumors (regardless of tumor origin) show extremely low levels of circulating T cells and reduced spleen size, with accompanying increase in bone marrow T cells (especially naive T cells).

This seems to be due to lower S1PR1 on T cells, driven by ligand-independent internalization (RNA levels are identical between control and tumor-bearing mice and there is no increase of S1P in the tumor-bearing mice).

KO or antagonism of S1PR1 results in increased T cell sequestration in bone marrow during adoptive transfer, and a knock-in mouse that can't internalize S1PR1 do not experience bone marrow sequestration of T cells with intracranial tumors.

They then show that rescuing ability of T cells to avoid sequestration in the bone marrow (expressing non-internalizable S1PR1) improves outcome in combination with immunotherapy - 4-1BB agonism or 4-1BB agonist + anti-PD-1 combo (but not anti-PD-1 monotherapy, interestingly).

Importantly, they show that the S1P1-KI does not have any tumor benefit alone - sequestration is a redundant immune evasion strategy, and there are other suppression mechanisms in the tumor that block T cell activity even if the cells manage to get in.

Very interesting results on a novel immune evasion mechanism, and anything getting at the root of why GBM is so challenging and its unique immune evasion mechanisms is a very welcome finding! Also raises many new interesting questions...

How do intracranial tumors induce S1PR1 internalization, if it's not ligand-dependent? What about T cell entry into the brain itself - this paper shows bone marrow and periphery only, but is there a S1PR1-dependent defect in brain entry (consistent with FTY720 efficacy in MS)?

The kinetics of T cell entry to the bone marrow (Fig S3C) show it's 2 weeks before there's any increase in BM T cells, so it's probably an effect of persistent activity - but is it tumor intrinsic, or would any persistent brain inflammation drive this as a feedback response?

There's no S1P1-KO into Control (no tumor) adoptive transfer, so it's unclear if S1PR1 downregulation is sufficient to drive T cells into the bone marrow or if the tumors have other signals acting on the T cells that force them into the bone marrow instead of e.g. SLOs.

The supplement shows G-CSF rescues T cells from sequestration and synergizes with 4-1BB. What's the regulatory mechanism going on - is it acting indirectly on e.g. microglia to do something to the T cells? What cells are making / responding? Is G-CSF downregulated in GBM?

Recombinant G-CSF can be given to people, so that may be one way to tackle this axis, since enforcing S1PR1 surface expression is right now pharmacologically challenging. Would a Grk inhibitor to prevent internalization be safe?

Vaccination is big in GBM, and this paper might suggest one reason to favor it - naive T cells are more prone to sequestration, so there's no T cells around to prime. Overcoming that with a vaccine may drive memory formation, and those cells then can get into the tumor site.

And since I like NK cells in GBM (

https://twitter.com/connor_rosen/status/1019366609973104641

), it's interesting to note that NK cells don't show any sequestration in the tumor-bearing mice, which is consistent with a lower level of S1PR1 expression.

Overall this paper shows some very exciting and interesting results from GBM patients and mouse models, and opens up a lot of areas for understanding a new immune evasion mechanism and some very interesting brain immunology. #Glioblastoma #Immunotherapy #Neuroimmunology

Also, @NatureMedicine, there are some confusing figure references, the text points to Fig 6F for PD-1 results but that's actually shown in Fig S6B, and the G-CSF results are shown in Fig S6c-e but referenced in the text as S6a-c.