Division of labor and cooperation between different butyrophilin proteins controls phosphoantigen-mediated activation of human γδ T cells

Abstract

Butyrophilin (BTN)-3A and BTN2A1 molecules control TCR-mediated activation of human Vγ9Vδ2 T-cells triggered by phosphoantigens (PAg) from microbes and tumors, but the molecular rules governing antigen sensing are unknown. Here we establish three mechanistic principles of PAg-action. Firstly, in humans, following PAg binding to the BTN3A1-B30.2 domain, Vγ9Vδ2 TCR triggering involves the V-domain of BTN3A2/BTN3A3. Moreover, PAg/B30.2 interaction, and the critical γδ-T-cell-activating V-domain, localize to different molecules. Secondly, this distinct topology as well as intracellular trafficking and conformation of BTN3A heteromers or ancestral-like BTN3A homomers are controlled by molecular interactions of the BTN3 juxtamembrane region. Finally, the ability of PAg not simply to bind BTN3A-B30.2, but to promote its subsequent interaction with the BTN2A1-B30.2 domain, is essential for T-cell activation. Defining these determinants of cooperation and division of labor in BTN proteins deepens understanding of PAg sensing and elucidates a mode of action potentially applicable to other BTN/BTNL family members.

Keywords: BTN2A1; BTN3A1; Butyrophilin; TCR; Vγ9Vδ2 T cells; juxtamembrane; phosphoantigens; γδ T cells.

Figure 1

Loss of function of BTN3A1-V domain deleted molecules can be compensated in complexes with BTN3A2 or BTN3A3 molecules. a 293T and BTN3 isoform-specific knock-out cell lines were cocultured with titrated concentration of HMBPP and 53/4 human Vγ9Vδ2 TCR reporter cells. The activation of reporter cells was measured by mouse IL-2 ELISA (n-3). b 293T and BTN3 isoform-specific knock-out cell lines were pulsed with zoledronate and cocultured with HMBPP expanded primary Vγ9Vδ2T cells. The T cell activation was measured by immuno flow cytometry with CD107a expression as readout detected by anti-CD107a-PE and anti-Vδ2-FITC (n-3). Surface-expressed BTN3A of the above-mentioned cells detected by mAb 103.2 followed by anti-mouse F(ab’)2-APC conjugate (right). c 293T, BTN3KO (3KO) cells and 3A-transductants of 3KO were cultured and tested as in a (n-3). Not shown are the results of 293T 3KO as they are consistently non-stimulatory . d Above-mentioned presenting cells were tested as in B. Surface-expressed 3A-molecules of the above-mentioned cells detected by mAb 103.2 followed by anti-mouse F(ab’)2-APC conjugate and their corresponding total mCherry expression were presented as histograms (right). e Histograms representing the total and surface-expressed FLAG protein of fix-permeabilized and live 3KO cells transduced with FLAG-tagged IgVdeleted-BTN3A1 (VΔ3A1) alone or cotransduced with other 3A-molecules detected by anti-FLAG and anti-mouse F(ab’)2-APC conjugate were analyzed by FACS. f 3KO cells transduced with 3A2 or 3A3 and the cells from e were cocultured with 53/4 Vγ9Vδ2 TCR reporter and titrated concentration of HMBPP. The activation of reporter cells was measured by mouse IL-2 ELISA (n-3). g 3KO cells expressing FLAG-IgVdeleted-BTN3A2 (VΔ3A2) alone or together with other BTN3As were analyzed as in e. h 293T wt and 3KO cells transduced with 3A1 and/or VΔ3A2 were analyzed as in G (n-3). i Schematic representations of different tagged constructs of 3A, 3A mutants, truncated 3A, and JM chimeras. The number of independent experiments was represented as n. Statistical significance in P-value is presented by asterisks (**** <0.0001; *** <0.001; ** <0.01; * <0.05; ns>0.05), and mean values with the SD are presented in graphs.

Figure 2

The JM region regulates BTN3A-protein and function. a 293T 3KO cells transduced with FLAG-VΔ3A1 alone and or cotransduced with N-terminus HA-tagged 3A-JM chimeras were analyzed in FACS for the total and surface expression of HA-3A molecules (Left) and FLAG-VΔ3A1 (right). The measurements were presented as histograms. b Live (left) and fix-permeabilized (right) 3KO cells transduced with FLAG VΔ3A1, cotransduced with HA-3A1 or HA-3A1_A3JM chimera were stained with mouse anti-FLAG and rabbit anti-HA followed by anti-mouse-Alexa Fluor 647 (red) and anti-rabbit Alexa Fluor 568 (blue), respectively. c 3KO cells transduced with FLAG-VΔ3A1, HA-3A1, HA-3A1_A3JM, FLAG-VΔ3A1 + HA-3A1, and FLAG-VΔ3A1 + HA-3A1_A3JM were labeled as 1 – 5, were subjected to anti-FLAG immunoprecipitation (IP) and samples were blotted against human vinculin (input, top), FLAG (middle) and HA (bottom) for their input (left) and immunoprecipitated proteins (right) (n-2).dSchematic presentation of FLAG-VΔ3A1-CFP, FLAG-3A1-CFP, 3A1-YFP and 3A1_Y3JM-YFP constructs (left), scheme describing the FRET with 440 LED laser, D is the donor (CFP), A is the acceptor (YFP) and A will emit a signal when exited by D if it is close proximity showing FRET. e Schematic presentation of probable ectodomain dimers and cytoplasmic B30.2 dimers based on the literature. Different cytoplasmic dimers expected were marked as A, B, C & D. f Ratiometric FRET analysis of 3KO transduced with 3A1-YFP and FLAG-3A1-CFP (upper left) or FLAG-VΔ3A1-CFP (lower left); 3KO transduced with 3A1_A3JM-YFP and FLAG-3A1-CFP (upper middle) or FLAG-VΔ3A1-CFP (lower middle); FRET ratio (FR) calculated chart (right).

Figure 3

Homomeric 3A3_JM and Heteromeric 3A_JM promote optimal stimulation via inter-BTN3 PAg signaling. a 293T and 3KO transductants of 3A1, 3A3, 3A3_R381H, or 3A_JM chimeric constructs were cultured and tested as in A (n-3). b Surface-expressed 3A-proteins of the above-mentioned cells detected by mAb 103.2 followed by anti-mouse F(ab’)2-APC conjugate (left) and their corresponding total mCherry expression (right) were presented as histograms. c The cellular distribution of BTN3A-mC fusion constructs is presented as images captured by confocal microscopy. d mCherry fusion constructs of 3A or 3A-JM chimera transduced 3KO cells were subjected to FRAP and the percentage of the immobile fraction of BTN3A-mC was measured. The number of cells (n) subjected to FRAP for 3KO_3A1mC (n-15), and other cell types (n-10) for each condition. e293T, 3KO transduced with mCherry fusion constructs of 3A3_R381H, 3A3_K136A_R381H, and cotransduced with eGFP reporter constructs of 3A1_H381R or 3A3 were analyzed by FACs for their total mCherry, total GFP, and surface-expressed BTN3As detected by mAb 103.2 and anti-mouse F(ab’)2-APC conjugate, the measurements were presented as histograms (bottom right). fThe above-mentioned cells were tested as in a (n-3). The predicted intermolecular signaling within the BTN3A proteins viz 3A3_R381H, 3A3-K136A-R381H, and 3A3/3A1_H381R and the observed stimulation strength was presented as a scheme in g III, IV and V, respectively. g Schematic presentation of predicted intermolecular signaling within the BTN3A proteins correlated to the observed outcomes in terms of 53/4 human Vγ9Vδ2 TCR reporter activation strength with antigen-presenting cells (3KO) expressing VΔ3A2 and 3A1 (I), VΔ3A1 and 3A2 (II) including the 3A-constructs mentioned in f. Statistical significance in P-value is presented by asterisks (**** <0.0001; *** <0.001; ** <0.01; * <0.05; ns>0.05), and mean values with the SD were presented in graphs.

Figure 4

JM regions modulate the conformation of BTN3A dimers. a Amino acids encoding juxtamembrane (JM) region of BTN3A1, BTN3A2, BTN3A3, and alpaca BTN3 (Vp) were aligned, and KKK and ETE residues of BTN3A1 and BTN3A3 were marked in red and blue, respectively. b Total and surface-expressed FLAG protein of permeabilized and live 3KO cells transduced with FLAG VΔ3A1 alone or cotransduced with 3A3 or 3A3_KKK mutant detected by anti-FLAG and anti-mouse F(ab’)2-APC conjugate were shown as histograms. c 3KO cells transduced with 3A1mC, 3A3_R381H-mC, or 3A3_R381H_KKK-mC mutant were cocultured with 53/4 Vγ9Vδ2 TCR reporter cells and titrated concentration of HMBPP. The activation of reporter cells was measured by mouse IL-2 ELISA (n-3). d Models of the BTN3-JM coiled-coil dimers. Models of the predicted JM coiled-coil dimers Q273–L312 were generated using CCBuilder2 (see Methods). Dimer interface residues at positions 283–285 are shown as ball and stick. I) BTN3A3 coiled-coil homodimer, II) BTN3A2 coiled-coil homodimer, II) Alpaca BTN3 (VpBTN3) coiled-coil homodimer, IV) BTN3A1 coiled-coil homodimer, V) BTN3A1-BTN3A2 coiled-coil heterodimer, VI) BTN3A1-BTN3A3 coiled-coil heterodimer, VII) BTN3A3-KKK (replacing ETE with KKK at positions 283–285) coiled-coil homodimer. Polar interactions are highlighted (red dashed lines). Each monomer within the homodimer has been labeled A or B. Statistical significance in P-value is presented by asterisks (**** <0.0001; *** <0.001; ** <0.01; * <0.05; ns>0.05), and mean values with the SD were presented in graphs.

Figure 5

4-M-HMBPP bound BTN3A1 did not interact with the BTN2A1-B30.2 domain. ITC titrations show that 4-M-HMBPP binds to BTN3A1 but does not support the binding of BTN3A1 to BTN2A1. a Titration of 960 μM 4-M-HMBPP into the buffer. bTitration of 960 μM 4-M-HMBPP into 60 μM BTN3A1 BFI. c Titration of 600 μM BTN2A1 ID271 into 60 μM BTN3A1 BFI. d Titration of 300 μM BTN2A1 ID271 into a mixture of 60 μM BTN3A1 BFI and 120 μM HMBPP. e Titration of 300 μM BTN2A1 ID271 into a mixture of 60 μM BTN3A1 BFI and 120 μM 4-M-HMBPP. Results are representative of n-3 independent experiments.

Figure 6

PAg induced Vγ9Vδ2 T cell activation by BTN3A-BTN2A1 composite ligand. In a resting state of the target cell, the heteromeric BTN3A (BTN3A1-BTN3A2/BTN3A3) interacts with BTN2A1 via their V-domains, and the BTN2A1-V domain interacts with germ-line encoded HV4 and CRR2 regions of Vγ9 chain of Vγ9Vδ2 TCR. Such interaction may act like a tonic TCR signal for maintaining homeostasis or even could be involved in the thymic selection of T cells. However, in case of stress in the target cell, the accumulated PAg binds to the B30.2 domain of BTN3A1, which further interacts with the B30.2 domains of BTN2A1. Consequently, the heteromeric JM region in the BTN3A complex permits the formation of appropriate topology where the V-domain of partnering BTN3A (BTN3A2/BTN3A3) distal to the PAg-B30.2 domain of BTN3A1, either on its own or in combination with unknown hypothetical ligand could be activating the TCR in which molecular interaction triggering remains elusive.