Genetics: BTN2A2 and BTN3A3

[Jonathan Edwards]

It would all fit together nicely. But then there always seems to be a way to fit things together nicely. I think Jackie Cliff is pleased the way data are coming out.

I am reminded of the film Amelie. If some set of T cells are causing trouble they are doing it in a very surreptitious way. Nobody can see where they are or what they are doing. Things turn up in research data but we cannot see how they got there or who they belong to. Maybe the writing on the wall is part of the story. Like the photos in the booths in Amelie.

Maybe one could image patterns of writing attached to TGF beta binding proteins on matrix and read what messages the T cells or macrophages have left there.

 

[V.R.T.]

Maybe one could image patterns of writing attached to TGF beta binding proteins on matrix and read what messages the T cells or macrophages have left there.

Are you talking about the kind of biopsies you describe here?

The first step would be to look at patterns of adsorbed signalling proteins on matrix fibres using biopsies and immunochemistry. Calibrating between individuals would probably be hard. I would look for changes in patterns. One the the best bits of evidence for our rheumatoid theory was that the immunoglobulin receptor CD16 is expressed on macrophages and adsorbed on to elastic fibres at exactly the site where rheumatoid nodules occur. I would not expect anything quite as clear-cut as that for ME/CFS but something equally striking might turn up around muscle or something.

Taking biopsies is not easy but if one knew exactly what was looking for a fairly small series might provide a very useful clue.

And if so do you expect any studies like this to happen in this new post-DecodeME landscape? You mentioned Wusts group as a possibility further down that thread.

 

[Jonathan Edwards]

Are you talking about the kind of biopsies you describe here?

No, in this instance I was talking about trying to find patterns by imaging techniques without taking tissue samples. It is a long shot but maybe technologies will come along.

 

[Jonathan Edwards]

I have tried to get it bit more information on BTN2A2 / its butyrophyllin protein.

It appears to be an MHC Class I protein. It is not a typical antigen presenting Class I protein as in HLA-A, -B and -C but it is a transmembrane protein of the same immunoglobulin supergene family. It also appears to be quite widely expressed on antigen presenting cells, as are both Class I and II. That includes expression on thymic epithelial cells, which interact specifically with T cells in triage by antigen recognition. It also includes breast epithelial cells and milk fat globule membrane, which expresses other MHC proteins like DR, presumably as part of some sort of immune signalling to fetus, along with maternal antibody transfer. (I remember in 1979 when my brother suggested I looked for HLA-DR [then called IA] in synovium because it was involved in immune response he [a breast tissue biologist] was chuffed because he had heard it was on milk fat globule membrane.)

So although this protein is pretty widely distributed, that is pretty standard for MHC molecules that function via interactions with T cells. Apparently some T cells express BTN2A2 but again, they do for other MHC.

The reviews talk of roles in moderating T cell functions like gamma interferon production. Some focus on regulatory T cells and some on gamma delta. My impression is that the interaction is fairly general. But I have not yet found suggestions that BTN2A2 does much else. I risk from a gene like this would certainly fit with the gist of the model Jo C, Jackie and I suggested - the FcRI aspect was only ever a way to try to make the model consistent with female predominance.

This analysis may be wrong but to me the gene looks like a very important clue. I was initially excited by the idea that DecodeME might have found an HLA link. That turned out to be a bit of a false start, although there still seems to be a suggestion of DQ. But in some ways BTN2A2 looks more relevant because it looks to be involved in a rather general 'innate' signalling mechanism to T cells (in the sense of not being tied to any particular peptide, which seems to me problematic).

 

[V.R.T.]

I have tried to get it bit more information on BTN2A2 / its butyrophyllin protein.

It appears to be an MHC Class I protein. It is not a typical antigen presenting Class I protein as in HLA-A, -B and -C but it is a transmembrane protein of the same immunoglobulin supergene family. It also appears to be quite widely expressed on antigen presenting cells, as are both Class I and II. That includes expression on thymic epithelial cells, which interact specifically with T cells in triage by antigen recognition. It also includes breast epithelial cells and milk fat globule membrane, which expresses other MHC proteins like DR, presumably as part of some sort of immune signalling to fetus, along with maternal antibody transfer. (I remember in 1979 when my brother suggested I looked for HLA-DR [then called IA] in synovium because it was involved in immune response he [a breast tissue biologist] was chuffed because he had heard it was on milk fat globule membrane.)

So although this protein is pretty widely distributed, that is pretty standard for MHC molecules that function via interactions with T cells. Apparently some T cells express BTN2A2 but again, they do for other MHC.

The reviews talk of roles in moderating T cell functions like gamma interferon production. Some focus on regulatory T cells and some on gamma delta. My impression is that the interaction is fairly general. But I have not yet found suggestions that BTN2A2 does much else. I risk from a gene like this would certainly fit with the gist of the model Jo C, Jackie and I suggested - the FcRI aspect was only ever a way to try to make the model consistent with female predominance.

This analysis may be wrong but to me the gene looks like a very important clue. I was initially excited by the idea that DecodeME might have found an HLA link. That turned out to be a bit of a false start, although there still seems to be a suggestion of DQ. But in some ways BTN2A2 looks more relevant because it looks to be involved in a rather general 'innate' signalling mechanism to T cells (in the sense of not being tied to any particular peptide, which seems to me problematic).

So you're saying this gene could potentially be more useful/relevant than the kind of HLA link that turned out to be a mirage? That's encouraging. I was a little concerned your pre DecodeME enthusiasm might have been heavily weighted by that HLA skyscraper.

BTN2A2 does seem like it could fit quite well (in terms of my surface level understanding anyway)

On that note- I've been looking into DQ a little and it seems really relevant to a lot of the stuff we've been talking about lately (T cells, graft vs host, autoantibody response).

 

[V.R.T.]

Are there any plans to entice experts in these genes to study them in ME/CFS as with the UCL Ca10 people?

 

[Jonathan Edwards]

I was a little concerned your pre DecodeME enthusiasm might have been heavily weighted by that HLA skyscraper.

In fact it was an MHC skyscraper significantly due to the MHC but non-HLA BTN2A2 I think.
An HLA link would have seemed easier to interpret maybe but BTNs may be more interesting.

I don't know who in the UK would have expertise in BTNs. Jackie Cliff is aware of the finding and continues to look at T cells. I am hoping to do a bit more networking in February.

 

[Kitty]

I don't know who in the UK would have expertise in BTNs.

Google sayeth this:

Ongoing UK Research Context
UK institutions such as the Division of Immunology at the University of Cambridge have been instrumental in characterizing the basic biology of BTN2A2. While many BTN2A2-specific clinical trials are in early or preclinical stages, they represent a broader shift in UK cancer research—led by organizations like the Institute of Cancer Research (ICR) and The Royal Marsden—toward finding next-generation checkpoints for patients who do not respond to existing immunotherapies.

 

[SNT Gatchaman]

The Phylogenomic Approach Suggests That Butyrophilins Have Ligands Beyond Gamma–Delta Receptors

Spoiler: Authors

Marenco, Ludovic; Olive, Daniel; Pontarotti, Pierre

Since γδ T cells are present in all jawed vertebrates, we wondered whether butyrophilins, proteins that play a key role in the activation of these cells, were also present in these organisms. Our analyses revealed the presence of genes encoding butyrophilins across all jawed vertebrates, including in squamates, a reptilian clade that is nonetheless reported in the literature to have lost γδ T cells. The conservation of butyrophilins in this group, despite the absence of their only known cellular partner, suggests that they may fulfill an alternative function, possibly through interaction with another ligand. Given their strong conservation across jawed vertebrates, it is reasonable to hypothesize that this alternative ligand may also be present in humans.

Web | DOI | PDF | International Journal of Molecular Sciences | Open Access

Butyrophilins are primarily known for their ability to interact with γδ T lymphocytes, leading to their activation.

Despite this, butyrophilins remain largely understudied, and their specific interactions have been characterized only in a handful of mammals, most notably in humans, mice, and alpacas. This paucity of data motivated our interest in butyrophilins, and more generally in their evolution relative to the γδ TCR. Our initial hypothesis was that butyrophilins and γδ TCRs might have co-evolved.

However, the fact that butyrophilins have been highly conserved across jawed vertebrates, despite their absence of association with γδ T lymphocytes in squamates, contradicts the notion of their co-evolution. The loss of γδ T lymphocytes in squamates appears to have occurred after their divergence from the tuatara (sphenodon punctatus), approximately 240 million years ago. This timeline is supported by fossil evidence placing the oldest squamate ancestor at roughly 168 million years ago. The persistence of genes encoding butyrophilins for such an extended period, despite the absence of γδ T lymphocytes, strongly suggests that these molecules have assumed an alternative role, likely through interactions with a distinct ligand.

further corroborated by recent work highlighting a novel function for butyrophilins. In particular, engagement of BTN2A1 expressed on M2-polarized macrophages and tumor-associated macrophages (TAMs) via a specific antibody promotes their reprogramming into M1-polarized macrophages through the SYK and MAPK signaling pathways [23]. This finding, which illustrates an alternate role for butyrophilins, supports the hypothesis that ligands other than the γδ TCR may interact with these molecules, although such ligands have yet to be identified. It is very likely that this ability to bind an alternative partner has enabled the conservation of butyrophilins in squamates despite the loss of γδ T lymphocytes.

[23] is BTN2A1 targeting reprograms M2-like macrophages and TAMs via SYK and MAPK signaling (2024)

 

[jnmaciuch]

That’s interesting! The other known non-lymphocyte-related function of butyrophilins is in facilitating globule production for milk secretion—wouldn’t be relevant in non-mammals obviously, though may have implications for butyrophilins in other roles related to lipid membrane regulation. This would explain BTN importance in breast milk, though maybe its presence also helps facilitate antibody transfer from mother to baby. It seems like butyrophilins have quite a wide range of physiological functions

Butyrophilin controls milk fat globule secretion - PMC

The molecular mechanism underlying milk fat globule secretion in mammary epithelial cells ostensibly involves the formation of complexes between plasma membrane butyrophilin and cytosolic xanthine oxidoreductase. These complexes bind adipophilin in ...

pmc.ncbi.nlm.nih.gov

 

[Jonathan Edwards]

[23] is BTN2A1 targeting reprograms M2-like macrophages and TAMs via SYK and MAPK signaling (2024)

I hate the M1/M2 nomenclature because it oversimplifies the functional reality but this might fit with a role for FcR1 and gamma interferon. Still, depending on which way function shifts with ME/CFS linked BTN variants I guess it would fit with lots of stories of broad shifts in behaviour of B, T and monocyte derived cells. It would be interesting to know whether it affects SYK signalling in neurons too.

 

[V.R.T.]

may have implications for butyrophilins in other roles related to lipid membrane regulation

Could this potentially be a link to @DMissa's recent preprint?

 

[jnmaciuch]

Could this potentially be a link to @DMissa's recent preprint?

Maybe! I don't really have a way to connect them since there are so few findings on that specific aspect of BTN

 

[jnmaciuch]

Unfortunately I don’t have time for this myself, but this locus might be a good candidate to explore with AlphaGenome. Since it falls in the midst of several histone genes, there’s a possibility that it could be strongly influencing transcription of many genes and the eQTL data just picks up on the BNTs since they are the closest genes with strong expression. Checking expression differences farther out from the locus could confirm or deny this

 

[V.R.T.]

Is anyone in the PRIME network well positioned to look into this locus?

 

[forestglip]

Unfortunately I don’t have time for this myself, but this locus might be a good candidate to explore with AlphaGenome.

Since I previously ran AlphaGenome with all the significant variants, all the brain-expression protein-coding scores can be found in the spreadsheet in this post. And scores for expression in all tissues and with all types of genes can be found in the much larger file at this GitHub repo.

Filtering to just the chr6 locus in protein-coding brain expression, there's still a lot of different genes predicted to be affected by a lot of different variants.

If I look at just the top variant:

AlphaGenome Scores

variant_id: chr6:26239176:A>G
biosample_name: brain
Assay title: polyA plus RNA-seq

Gene	Quantile Score
H2AC8	0.9824973
H1-3	0.9799247
H2BC8	0.96761274
H3C7	0.96289337
H2AC7	0.9050901
H4C1	-0.90298235
H1-2	0.8940993

If I look at all variants at the chr6p22.2 locus (p<1x10^-7), and filter to abs(quantile score) > 0.99:

AlphaGenome Scores

biosample_name: brain
Assay title: polyA plus RNA-seq

Variant	Gene	Quantile Score
chr6:26328470:A>AT	BTN1A1	-0.999922
chr6:26328470:A>AT	BTN3A1	0.99987906
chr6:26328470:A>AT	BTN2A2	-0.99987334
chr6:26328470:A>AT	BTN2A1	-0.9997847
chr6:26298408:GC>G	BTN3A3	-0.9997745
chr6:26298408:GC>G	BTN2A2	0.99965847
chr6:26298408:GC>G	ZNF322	-0.99961674
chr6:26328470:A>AT	ABT1	0.99951726
chr6:26298408:GC>G	ABT1	-0.9994945
chr6:26298408:GC>G	BTN2A1	-0.9994827
chr6:26298408:GC>G	BTN3A1	0.9994327
chr6:26328470:A>AT	HMGN4	-0.9993333
chr6:26328470:A>AT	BTN3A3	0.9990358
chr6:26205065:G>A	H4C5	-0.998573
chr6:26298408:GC>G	HMGN4	0.9984708
chr6:26259579:G>A	H3C3	-0.9973411
chr6:26328470:A>AT	H1-6	0.9965742
chr6:26214245:T>C	H2BC8	-0.99641275
chr6:26298310:G>A	H4C5	-0.995975
chr6:26225804:C>G	H3C6	0.9956871
chr6:26325805:C>T	H4C4	0.9946945
chr6:26291363:T>C	H2AC7	-0.994183
chr6:26289082:T>C	H4C4	0.99404764
chr6:26328470:A>AT	H1-3	-0.99376744
chr6:26259579:G>A	H4C5	0.99332225
chr6:26338206:T>G	H4C4	0.99250877
chr6:26298408:GC>G	H4C5	0.9923346
chr6:26328470:A>AT	H4C3	0.9914013
chr6:26322245:G>A	H3C7	-0.9914013
chr6:26264972:G>A	H4C2	0.9912015
chr6:26325954:C>T	H3C8	0.9909971
chr6:26205065:G>A	H2BC8	-0.99035496
chr6:26322245:G>A	H3C3	-0.99013096

Not totally sure how to interpret. I think it's an issue of too many variants.

 

[jnmaciuch]

Not totally sure how to interpret. I think it's an issue of too many variants.

Thanks. If I'm interpreting correctly, your spreadsheet is for results before quantile score filtering, right? If I filter down to chr6:26239176:A>G I'm still only getting genes right near the locus--I'm wondering if that's a result of the training data being limited to cis-eQTLs near the locus for that region

 

[forestglip]

If I'm interpreting correctly, your spreadsheet is for results before quantile score filtering, right?

Yeah, every score returned by AlphaGenome.

If I filter down to chr6:26239176:A>G I'm still only getting genes right near the locus--I'm wondering if that's a result of the training data being limited to cis-eQTLs near the locus for that region

The predictions are limited to the same length as the strand of DNA sent for inference, with the limit for that being 1MB. The interval it scored genes on can be seen in the scored_interval column, which for this variant is chr6:25714888-26763464. When I filter for scores for just that variant, I see 50 different genes, which seems to be pretty close to how many genes I see in that interval on UCSC Genome Browser.

 

[jnmaciuch]

The predictions are limited to the same length as the strand of DNA sent for inference, with the limit for that being 1MB. The interval it scored genes on can be seen in the scored_interval column, which for this variant is chr6:25714888-26763464. When I filter for scores for just that variant, I see 50 different genes, which seems to be pretty close to how many genes I see in that interval on UCSC Genome Browser.

Ah I see so it is limited to looking at local effects. That's a shame since if it is the histone genes that are relevant you'd expect much more widespread effects, but that may just be beyond the ability of eQTL experiments to detect anyways