The intracellular B30.2 domain of butyrophilin 3A1 binds phosphoantigens to mediate activation of human Vγ9Vδ2 T cells

Abstract

In humans, Vγ9Vδ2 T cells detect tumor cells and microbial infections, including Mycobacterium tuberculosis, through recognition of small pyrophosphate containing organic molecules known as phosphoantigens (pAgs). Key to pAg-mediated activation of Vγ9Vδ2 T cells is the butyrophilin 3A1 (BTN3A1) protein that contains an intracellular B30.2 domain critical to pAg reactivity. Here, we have demonstrated through structural, biophysical, and functional approaches that the intracellular B30.2 domain of BTN3A1 directly binds pAg through a positively charged surface pocket. Charge reversal of pocket residues abrogates binding and Vγ9Vδ2 T cell activation. We have also identified a gain-of-function mutation within this pocket that, when introduced into the B30.2 domain of the nonstimulatory BTN3A3 isoform, transfers pAg binding ability and Vγ9Vδ2 T cell activation. These studies demonstrate that internal sensing of changes in pAg metabolite concentrations by BTN3A1 molecules is a critical step in Vγ9Vδ2 T cell detection of infection and tumorigenesis.

Figure 1

Phosphoantigen reactivity is mapped to the BTN3A1 intracellular B30.2 domain. A) Percentage of human Vγ9Vδ2 T cells expressing CD107a after co-culture with HEK sh284 cells mock-transfected (Mock) or re-expressing, after transient transfection, either wild-type (WT) BTN3A1 or BTN3A3 isoforms or chimeric BTN3A3 proteins assembled with the BTN3A1 B30.2 domain with (BTN3A3 B30.2 A1) or without the extra C-terminal tail of BTN3A3 (BTN3A3 B30.2 A1 short). Target cells were pre-treated for 2 hours with the α-CD277 mAb 20.1 (10 mg/mL) or the NBP zoledronate (100 mM). Data are representative of three independent experiments. B) Amino acid sequence alignment of the B30.2 domains of BTN3A1 (shown on top) and BTN3A3 on bottom. Positions identical to BTN3A1 are shown as dashes, differences are shown in capital letters. Dots indicate the region missing in BTN3A1. The swapped regions are indicated and labeled accordingly (Box 1, Box 2 and Box 3, abbreviated at b1, b2 and b3, respectively). Numbering corresponds to the full length sequence of BTN3A proteins. C) Percentage of human Vγ9Vδ2 T cells expressing CD107a after co-culture with NBP-treated mutant-BTN3A HEK293 transfectants. Full length BTN3A3, which does not mediate pAg activation of Vγ9Vδ2 T cells, had the corresponding BTN3A1 regions (b1, b2, b3 or a combination thereof) swapped into the intracellular B30.2 region. These mutants were used as targets for treatment with NBP (NBP) or vehicle (-) and the Vγ9Vδ2 response was measured as CD107a up-regulation.

Figure 2

Crystal structure of the BTN3A1 intracellular B30.2 domain at 1.4 Å resolution. A) Ribbon diagram of the BTN3A1 B30.2 domain dimer observed in the crystal lattice. Monomers of B30.2 are shown in orange and blue with the N and C termini of each shown as blue and red spheres respectively. The two β-sheets that compose the core of the B30.2 domain, sheet A and sheet B, are indicated. The putative position of the N-terminal linker regions in relation to the transmembrane domain are shown as dashed lines and the plasma membrane is modeled in pink. A monomer is oriented below the dimer to indicate the orientation shown as electrostatic surface, to the right. The electrostatic potential is shown rendered on the B30.2 domain surface. The surface is colored from red (-20 kT) to blue (20 kT) with the basic pocket indicated by a dashed yellow box. B) Close-up of the putative pAg binding pocket showing the six basic residues contributing to the highly positive charge of the pocket. These residues are shown as sticks under a transparent surface colored according to the electrostatic potential. C) Rotated view of (B) showing the results of molecular docking performed by Autodock Vina. A best fit docking position of the pAg, HDMAPP, is shown using a stick model; it is well-positioned in the binding pocket with a free energy (ΔG) calculated to be -5.6 kcal/mol, with the β-phosphate coordinated by the three arginines, R12, R418 and R469.

Figure 3

Binding of phosphoantigen to the BTN3A1 intracellular B30.2 domain. A) Raw ITC traces showing injection of pAg into solution containing either the BTN3A1 B30.2 domain (black) or buffer (gray). Binding measurements were performed using 100μM B30.2 in the cell and 2.0mM pAg in the syringe. The binding isotherms showing binding of the different pAgs to the BTN3A1 intracellular B30.2 domain are shown to the right, plotted on the same axis for comparison. Buffer control is shown as open circles, IPP as open triangles, EtPP as filled triangles, HDMAPP as open squares, and cHDMAPP as filled squares. The chemical structure of each pAg is shown under the binding isotherm traces. B) Binding measurements of the highly potent cHDMAPP to the extracellular domain of BTN3A1 (black) or buffer control (gray). Measurements were performed with 100μM BTN3A1 extracellular domain in the cell and 2mM cHDMAPP in the syringe. The binding isotherms are shown with buffer control (open circle) or cHDMAPP (filled square) to the right.

Figure 4

Charge reversal of putative pAg binding pocket residues prevents pAg binding and Vγ9Vδ2 activation. The residues of the B30.2 basic pocket were mutated as follows: H378D, K393D, R412E, R418E, and R469E. A) Shown are the binding isotherms of cHDMAPP with the B30.2 domain (filled squares), the B30.2 charge reversal mutant (filled triangle), and a buffer control (open circles). B) Vγ9Vδ2 T cell stimulation as assessed by CD107a up-regulation is shown in response to no stimulation (No stim.), addition of the α-CD277 agonist antibody 20.1, or in the presence of the bisphosphonate zoledronate (NBP). Data are mean of duplicates ± SD and are representative of three independent experiments. * P <0.05 (paired Student t-test).

Figure 5

A single amino acid difference mediates pAg reactivity between the BTN3A1 (pAg-responsive) and BTN3A3 (pAg-inactive) B30.2 Domains. A) ITC binding isotherms comparing the binding of cHDMAPP to the wildtype BTN3A1 B30.2 domain or mutants containing the respective BTN3A3 residue are shown at left. Binding of cHDMAPP is shown to B30.2 (filled squares), H351R (filled diamonds), K426E (open triangle), or the buffer control (open circle). At right are the ITC binding isotherms comparing the binding of cHDMAPP to the wildtype BTN3A3 B30.2 domain or mutants in which the relevant BTN3A1 residues are introduced. A BTN3A3 B30.2 domain mutant lacking the additional 70 C-terminal residues (Δ485-555) was also assessed for pAg binding. Binding of cHDMAPP is shown to B30.2 (filled squares), R351H (filled diamonds), the C-terminal truncation Δ485-555 (filled triangles), E427K (open triangle), or the buffer control (open circle). B) FACS plots showing CD107a upregulation of Vγ9Vδ2 T cells in response to treatment of wildtype and sequence swapped B30.2 domain full-length BTN3A constructs with α-CD277 agonist antibody 20.1, or in the presence of the bisphosphonate zoledronate (NBP). C) Functionality of fluorescent wildtype and mutated BTN3A1 chimeras are shown via FACS plots showing CD107a upregulation on Vγ9Vδ2 T cells (polyclonal line) after coculture with CD277shRNA#284 transduced-HEK293FT cells transiently transfected for the expression of either wildtype (WT) or H351R B30.2 domain mCherry full-length BTN3A1 constructs. Cells were treated with α-CD277 agonist antibody (clone #20.1) or the aminobisphosphonate zoledronate (NBP) before coculture. Numbers adjacent to outlined areas indicate the percentage of CD107a+ T cells. D) Top: Representative confocal images of HEK293 cells expressing mCherry-WT or –H351R BTN3A1 molecules following overnight incubation with NBP, shown before (Pre-bleach), immediately after (T₀-bleach) and 120 seconds (T₁₂₀) after photobleaching of regions of interest (indicated rectangular areas). Scale bars, 3 μm. Bottom: FRAP analysis of HEK293 cells expressing either wildtype or H351R B30.2 domain mCherry full-length BTN3A1 molecules, without treatment (-) or following incubation with NBP or α-CD277 agonist mAb (20.1). Data are presented as the value for the percentage of immobile fraction. FRAP was collected every 5 s. Control: no treatment, (-). BTN3A1 WT: (-), n=48; NBP, n=34; α-CD277, n=20. BTN3A1 H351R: (-), n=19; NBP, n=15; α-CD277, n=19. Bars, mean values. Student's t-test: *** p < 0.005.

Figure 6

Structure of cHDMAPP in complex with B30.2 domain of BTN3A1. A) B30.2 is shown in ribbon format in orange, with pAg binding pocket residues shown as sticks. cHDMAPP is shown as sticks with spheres, and specific atoms are shown using standard coloring (carbon=yellow, red=oxygen, blue=nitrogen and orange=phosphate). Hydrogen-bonds ≤ 3.3Å are shown as yellow dashes. B) Omit map density of cHDMAPP is shown in teal scaled at 1σ. C) Positive density observed in the complex structure but missing in the apo structure as identified from the isomorphous difference map is shown in green scaled at 4σ.

Figure 7

BTN3A1 is necessary but not sufficient to stimulate Vγ9Vδ2 T cells. A) SPR analysis of injections of the BTN3A1 extracellular domain, 20.1 single chain antibody, and the BTN3A1 extracellular domain in complex with the 20.1 single chain antibody from 0.03125μM to 4μM (Black) flown over immobilized G115 Vγ9Vδ2 TCR. Control injections of buffer alone for each experiment are shown in gray. B) Multimerized G115 γδ TCRs do not bind recombinant, full-length BTN3A1. AmDex multimerized TCRs were made by conjugating C-terminally biotinylated TCRs at a 60:1 ratio with streptavidin conjugated AmDex polymers. Serial dilutions from 1ug/ml to 0.03125ug/ml of AmDex TCRs were injected over immobilized full length BTN3A1 and the murine T22 protein. Since non-specific interactions were observed between the TCR conjugates and the biotin coated reference flow cell (AmDex-G115 reference cell), specific binding was measured by subtracting non-specific interactions with T22 from the signal from BTN3A1 channel. This analysis was reversed for the G8 murine TCR. Co-injection of 1mM cHDMAPP with the G115 AmDex conjugate did not alter observed interactions. C) Rodent cells expressing full-length human BTN3A1 do not activate human Vγ9Vδ2 T cells following agonist 20.1 α-CD277 mAb or NBP pretreatment. Expression of CD107a on Vγ9Vδ2 T cells following co-culture with NXS2 murine neuroblastoma tumor cells expressing full-length human BTN3A1 or BTN3A2 after transient transfection. 36 hours after transfection, surface expression of BTN3A was confirmed by flow cytometry. Target cells were pretreated for 2 h with agonist 20.1 α-CD277 mAb (10 mg/ml) or the NBP Zoledronate (100 mM). Data are presented as the mean of specific values ± SD of the percentage of CD107a positive γδ T cells. Positive control: Human HEK293 cells. Similar results were obtained using the following rodent cell lines also transfected for the expression of human ICAM-1; AB1:mouse; MC38:mouse; M5T1:rat.