Definition of APC Presentation of Phosphoantigen (E)-4-Hydroxy-3-methyl-but-2-enyl Pyrophosphate to Vγ2Vδ2 TCR¹ Free

The γδ T cells appear to belong to nonclassical T cells with both innate and adaptive immune features (1, 2, 3, 4). Accumulating evidence suggests that human Vγ2Vδ2 (also called Vγ9Vδ2) T cells may play a role in mediating immunity against microbial pathogens (5, 6, 7, 8, 9, 10, 11, 12) and tumors (13, 14, 15). Vγ2Vδ2 T cells exist only in primates, and in humans they represent a major circulating γδ T cell subset that normally constitutes 65–90% of total peripheral blood γδ T cells. We have recently demonstrated that macaque Vγ2Vδ2 T cells can undergo major expansion and transendothelial migration in mycobacterial infections (5, 16) and during phospholigand + IL-2 treatment (17, 18). These findings are consistent with the remarkable expansion of human Vγ2Vδ2 T cells seen during various bacterial and protozoal infections (4), and they notably increase in patients with certain cancers (14, 19). Vγ2Vδ2 T cells appear to be activated by certain low m.w. foreign- and self-nonpeptidic phosphorylated metabolites of isoprenoid biosynthesis, e.g., (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP),⁴ isopentenyl pyrophosphate (IPP), and its isomer dimethylallyl pyrophosphate (20, 21, 22, 23, 24) commonly referred to as phosphoantigens. HMBPP is produced in the newly discovered 2-C-methyl-d-erythritol-4-phosphate pathway of isoprenoid biosynthesis of most eubacteria, apicomplexan protozoa, plant chloroplasts and algae but not in vertebrates and thus not in the human host (5). HMBPP from mycobacteria or other microbes is the most potent Vγ2Vδ2 T cell activator known, with an EC₅₀ of 0.1 nM (20, 25, 26).

Whereas the chemistry of phosphoantigens and their ability to activate Vγ2Vδ2 T cells have been well described, little is known about molecular mechanisms by which HMBPP interacts with γδ T cells (6, 27). Most studies done to date have been focused on prenyl pyrophosphates, particularly IPP, but rarely the naturally occurring microbial phosphoantigen HMBPP (6). Earlier experiments using Vγ2Vδ2 T cell activation as readouts demonstrated that IPP does not need to undergo cell entry or processing and that phosphoantigen activation of Vγ2Vδ2 T cells requires cell-cell contact (6, 28). Efforts using mutant APC cell lines deficient of MHC class I, MHC class II, CD1, and β₂-microglobulin or using blocking Abs demonstrate that none of these known Ag presenting molecules is required to present IPP for immune activation of Vγ2Vδ2 T cells (6). Based on these findings, two models have been proposed to explain IPP presentation and Vγ2Vδ2 T cell recognition (6). One is the simple cell surface contact model, in which IPP nonspecifically contacts on the cell surface of APC and then stimulates the activation of Vγ2Vδ2 T cells (6). The other model postulates that a membrane molecule on APC is required to present nonpeptide phosphoantigen to Vγ2Vδ2 TCR for immune recognition and activation (6). Despite decade-long studies, however, there has been no direct evidence indicating prenyl pyrophosphate Ags or HMBPP are indeed associated with an APC membrane molecule, and then presented and bound to Vγ2Vδ2 TCR. Furthermore, human or macaque Vγ2Vδ2 TCR transfectants have not been reported to respond to HMBPP (29, 30). Better experimental systems are needed to undertake in-depth studies of phosphoantigen presentation and interaction with γδ T cells.

The development of soluble, tetrameric Vγ2Vδ2 TCR may provide a useful approach to explore the mechanism by which phosphoantigen HMBPP is associated with APC, presented and bound to Vγ2Vδ2 TCR. This scenario is supported by a recent study in which murine γδ TCR multimers can be used to detect cell surface ligands (31). In this study, we took advantage of our decade-long TCR expertise and MHC class I tetramer application experience (32) to develop Vγ2Vδ2 TCR tetramer. We demonstrated that soluble, recombinant Vγ2Vδ2 TCR tetramer made it possible to visualize APC presentation of phosphoantigen HMBPP to Vγ2Vδ2 TCR. Our data provided evidence indicating that whereas exogenous HMBPP was associated firmly with the membrane of APC and then presented and bound to Vγ2Vδ2 TCR, endogenous phosphoantigen produced by intracellular Mycobacterium in the infected dendritic cells (DC) could also readily be transported and presented on the membrane to Vγ2Vδ2 TCR for immune recognition. Our data also showed that the putative APC membrane molecule presenting HMBPP appeared to be a protein or protein-associated component and exists in primate APC or T cells but not rodent APC.

RNA were extracted from macaque PBMC using the TRIzol (Invitrogen)-based isolation method; cDNA were synthesized from the RNA using the First Strand cDNA Synthesis Kit (Clontech Laboratories) as previously described (5, 32, 33). Full-length cDNA of Vγ2- and Vδ2-bearng TCR were isolated by the PCR-based approach (5, 33) using respective oligonucleotide primers: Vγ2F 5′-GCGGTACCATGCTGTCACTGCTCCA-3′ and Vγ2R 5′-GCTCTAGAGGGACAATAACCGATGAG-3′; Vδ2F 5′-GCGGTACCATGCAGAGGATCTCCTC-3′ and Vδ2R 5′-GCTCTAGAAGTGTAGCTTCCTCATG-3′ (Operon). The TCR DNA fragments from PCR were digested and inserted into vector pGEM72f (Promega) for cloning and sequencing; the molecular TCR clones with correct sequences were chosen for subcloning for making constructs encoding soluble TCR. To facilitate production and purification of soluble recombinant Vγ2Vδ2 TCR heterodimer, Vγ2- or Vδ2-bearing cDNA was recombined with the DNA segment containing linker, Jun/Fos or BirA sequences following previously described procedures. First, the extracellular region of Vγ2- or Vδ2-bearing cDNA was recombined with Jun or Fos region encoding leucine zipper dimerization through PCR using a V region-specific 5′-primer and C region-specific 3′-primer with a 35-bp sequence corresponding to the N terminus of Jun or Fos. Secondly, the leucine zipper dimerization region of Jun and Fos was amplified by PCR using a 5′-primer containing corresponding sequences of the Cγ or Cα region and the Jun or Fos coding region and a 3′-primer for Jun or Fos 3′ (34, 35, 36). In this context, a fragment encoding a 7-aa linker (VDGGGGG) was introduced to connect Vγ2Cγ and Jun or Vδ2Cδ and Fos using the PCR-based approach as described above. Furthermore, a 45-bp DNA fragment encoding a 15-aa biotinylation sequence (GGGLNDIFEAQKIEWHE; Bir substrate peptide) was introduced to the downstream of the TCRδ for specific biotinylation of TCR by the enzyme BirA (37, 38) using the vector pPAC4 (Avidity). Finally, recombined DNA fragments containing the Vγ2-bearing Jun or Vδ2-bearing Fos-BSP were produced by the PCR-based recombination. The derived PCR products were digested with KpnI and ApaI (Invitrogen), ligated with the vector pMTV/V5-His A (Invitrogen) by T4 ligase, and then transformed into One Shot Top 10-competent cells (Invitrogen). The transformed colonies were identified; the correct sequences of fusion genes were confirmed by DNA sequencing again. As a control, DNA constructs encoding soluble Vγ1- and Vδ2-bearing TCR fused with the linker, Jun/Fos, and BirA were similarly developed.

Equal amounts of two plasmids bearing the paired TCR constructs (Vγ2Cγ-JB mammary tumor virus (MTV) and Vδ2Cδ-FB MTV, or Vγ1Cγ-JB MTV and Vδ2Cδ-FB MTV) were co-transfected into S2 cells (Life Technologies) together with pCoBlast helper vector (Invitrogen) using Cellfectin Reagent following the supplier’s instruction. Transfected cells were drug selected using 25 μg/ml blasticidin (Invitrogen) and cultured for 2 wk to screen drug-resistant cells. These drug-selected cells were sampled for extracting cellular DNA using DNeasy Blood & Tissue Kit (Qiagen) and assessed for the presence of transfected Vγ2- or Vγ1- and Vδ2-bearing TCR DNA through PCR amplification using primers 5′-GCGCCAGCCCGCCTGGAATGTG-3′; 5′-GCGCGAAGGAAGAAAAATAGTGGG-3′ and primers 5′-GCGCAAGTGCTCCATGAAAGGA-3′; 5′-GCGCTTGTCTTTCTGGTTCCAC-3′, respectively. The samples exhibiting a specific DNA fragment with ∼360 bp and ∼330 bp were considered to have Vγ2- or Vγ1-or Vδ2-bearing TCR construct. The Vγ and Vδ double-positive cell lines were then induced for expression of TCR proteins by adding 1 mM CuSO₄ to the culture for 72–96 h. The culture supernatants were concentrated and tested for secreted soluble recombinant Vγ2Vδ2 TCR heterodimer or Vγ1Vδ2 TCR heterodimer in standard dot-blot assays using anti-Vγ2, anti-Vδ2, and anti-Vγ1 mAbs, respectively (Fig. 1 d).

Since the established cell lines were polyclonal, two rounds of limited dilution cloning were performed to select monoclonal population of cells that secreted soluble Vγ2Vδ2 or Vγ1Vδ2 TCR heterodimer. Thus, the Vγ⁺Vδ⁺ cell lines were diluted into a 10-cells/ml suspension, distributed 100 μl/well into a 96-well plate, and then cocultured with 1 × 10⁵/well gamma-irradiated S2 feeder cells. After culture for 1 wk at 27°C, 25 μg/ml blasticidin were added to the culture for another 2–3 wk until the drug-resistant monoclonal cells were fully grown. The established individual clones were identified whether they contain Vγ2 or Vγ1 and Vδ2 genes by PCR as described above, and these double-positive monoclonals were expanded and verified for TCR expression by dot-blot assays using conformation-dependent anti-Vγ1 mAb, anti-Vγ2 mAb or anti-Vδ2 mAb. The cloned cells secreting Vγ2⁺Vδ2⁺ TCR were adapted through gradual reduction of FBS in complete Schneider’s Drosophila medium (Life Technologies) from 10% FBS and 25 μg/ml blasticidin to Drosophila-SFM (Invitrogen) supplemented with 20 mM l-glutamine and 10 μg/ml blasticidin.

The secreted TCR heterodimers were purified by two-round purification procedures using Ni-NTA and avidin affinity columns. These adapted, cloned cells were expanded to 1-liter conical flasks (Wheaton) at 27°C with 90–120 rpm in a rotary shaker until they reached a density of 1 × 10⁷ cell/ml, and then induced for secretion of soluble TCR in the presence of 1 mM CuSO₄ for 3 days with 140 rpm rotation. After removal of the debris and pellets by centrifugation, supernatants containing recombinant TCR molecules were dialyzed to remove free Cu²⁺ by SnakeSkin Pleated Dialysis Tubing (Pierce) overnight at 4°C in PBS, and then concentrated 20-fold by ultrafiltration with a m.w. 30,000 cutoff Vivacell concentrator (Vivasciences). The concentrated samples were passed through a Ni-NTA agarose affinity column (Qiagen) under native conditions; the His-6-tagged recombinant proteins were collected by 250 mM imidazole elution.

The Ni-NTA-purified samples were balanced to 10 mM Tris, pH 8.0, by buffer exchange overnight at 4°C in Slide-A-Lyzer Dialysis Cassette (10K; Pierce), and then concentrated. The samples concentrated from the first-round purification by Ni-NTA were optimally biotinylated using d-biotin and BirA enzyme (Avidity) at 30°C according to the supplier’s protocol. Excess biotin was removed by PBS dialysis overnight in a Slide-A-Lyzer Dialysis Cassette (Pierce); the dialyzed biotinylated samples were then bound to the avidin agarose in an avidin affinity column (ImmunoPure Immobilized Monomeric Avidin; Pierce) for the second round of purification. After washings, the biotinylated TCR molecules were eluted by biotin-buffer exchange. The eluted samples were dialyzed to remove excess biotin by PBS dialysis, concentrated, and then determined for protein concentration via the bicinchoninic acid kit (Pierce). The yield of the biotinylated TCR heterodimer purified from the 1-liter supernatant is 5–10 μg.

To characterize soluble TCR heterodimer, the purified samples were electrophoresed in 12% SDS-PAGE gel (Bio-Rad) under reducing conditions (Fig. 1,b) or in 10% PAGE gel (Bio-Rad) under nonreducing conditions (Fig. 1,c), and then stained with Imperial protein stain (Pierce). To determine whether soluble γδ TCR heterodimers mimic some degree of conformation for native TCR expressed on the γδ T cell surface, dot-blot assay was performed using Vγ1, Vγ2, or Vδ2 mAbs that react only with cell surface Vγ- or Vδ-bearing TCR (Fig. 1,d). Supernatant from untransfected S2 cells served as control. As another control, the purified recombinant TCR samples were intentionally denatured with SDS at 100°C for 10 min and loaded on the dot-blot membrane for the dot-blot assay (Fig. 1 d). In the dot-blot assay, methanol-treated membrane (0.45 μm pore size; Millipore) was loaded with 10–20 μl of concentrated supernatants or purified samples, blocked for nonspecific binding by using SuperPBS blocking buffer (Pierce), incubated for 1 h at room temperature with anti-human Vγ2 mAb (clone 7A5; Endogen), anti-human Vγ1 mAb (clone 23D; Endogen), or anti-human Vδ2 mAb (clone 15D; Endogen). The membrane was then washed four times with PBST (PBS + 0.05% Tween 20), incubated with a 1/5000 dilution of goat anti-mouse HRP-conjugated Ab (Bio-Rad) for 1 h, washed, and developed by Supersignal West Pico Chemiluminescence substrate (Pierce) and then exposed to BioMax MR film (Kodak). The dot-blot assay determined whether recombinant TCR heterodimers were recognized by respective anti-Vγ or anti-Vδ mAbs that reacted only with native conformation of cell surface Vγ- or Vδ-bearing TCR (39, 40, 41, 42).

To assemble Vγ2Vδ2 TCR or Vγ1Vδ2 TCR tetramer, biotinylated TCRs were incubated with one-fourth of its molar amount of FITC (BD Pharmingen)- or Qdot 655-conjugated streptavidin (Invitrogen), respectively, for 30 min at 4°C. The assembled TCR tetramers were separated from free fluorescents by a gel filtration column (Superose 6 PC 3.2/3.0; Amersham). All TCR tetramers were freshly prepared for staining experiments using purified Vγ2Vδ2 or Vγ1Vδ2 TCR heterodimers frozen down at −70°C, and stored short term in the dark at 4°C for subsequent staining.

APC cell lines used in the experiments included three human monocyte/macrophage cell lines: SC (ATCC CRL-9855), THP-1 (ATCC TIB-202), and U937 (ATCC CRL-1593.2); three mouse monocyte/macrophage cell lines: J774 (ECACC 85011428), RAW 264.7 (ATCC TIB-71), and C8-B4 (ATCC CRL-2540); rat monocyte cell line 158.2 (ATCC CRL-8466); and pig macrophages 3D4/31 (CRL-2844). These cell lines were grown in modified DMEM or RPMI 1640 completed medium at 37°C in a 5% CO₂ atmosphere, as recommended by the Ameerican Type Culture Collection. Macaque T cells, B cells, monocytes, and DC were prepared as follows. PBMCs isolated from rhesus monkeys were incubated with biotin-conjugated anti-CD3 mAb and, subsequently, anti-biotin beads (Miltenyi Biotec). The cell suspensions were then used to isolate CD3⁺ T cells using a MACS separation column following the supplier’s protocols. The remaining CD3⁻ cells were thereafter used to enrich monocytes by plastic adhering in a 2-h incubation. The adherent monocytes were collected by extensive washing using chilled (4°C) PBS. Nonadherent cells were used to isolate B cells using a CD20⁺ magnetic beads column. To generate monocyte-derived DCs, the isolated monkey monocytes were stimulated in culture with 0.08 ng/ml GM-CSF (R&D Systems) and 1 ng/ml IL-4 (Sigma-Aldrich) for 5–6 days. The cultured cells were assessed for the differentiation of DCs as defined by morphology analyses and CD1a expression.

To ensure consistent BCG infection of DCs, mid-log phase BCGs were prepared as previous described (43). The monocytes and DCs were infected with mid-log phase BCG at 10 multiplicities of infection for 6 h. Free BCG was removed by washings. The BCG-infected cells were cultured for another 12–36 h, and then used for surface staining by the FITC-labeled TCR tetramers and CD14 APC (eBiosciences) and analyzed by flow cytometry and confocal microscopy. The CFU analyses of lysate from cells incubated with BCG for 6 h showed that a single DC ingested an average of 6–8 bacteria of BCG, and the viability exceeded 85%. As controls for FITC-labeled TCR tetramer staining, DCs not infected with BCG, or DCs infected with SIV for 16–48 h were also prepared and analyzed.

DCs, 1.0 × 10⁵ per tube, or 1.0 × 10⁶ per tube of other cells were incubated with HMBPP (80 ng/ml) in 10% FCS-RPMI 1640 at 4°C at different time points (30–120 min). Cells were washed two or three times with PBS and then stained for 15–60 min at 37°C with FITC-labeled TCR tetramers either alone or in combination with CD14 APC (eBiosciences), CD20 PE-Cy7 (BD Biosciences), or CD4 allophycocyanin and CD8 PE-Cy7 (BD Biosciences), respectively. The stained cells were washed and fixed with 2% formalin. For flow cytometry analyses, the stained cells were gated based on forward and side scatters and pulse width, and at least 40,000 gated events were analyzed using Summit Data Acquisition and Analysis Software (DakoCytomation) as previously described (17).

To assess whether the TCR tetramer preincubated with HMBPP could directly stain APCs, Vγ2Vδ2 TCR tetramer was incubated with HMBPP (80 ng/ml) for 30 min, and free HMBPP was removed by PBS dialysis. After dialysis, the buffers containing the HMBPP-treated TCR tetramer were collected, concentrated, quantified, and used to stain HMBPP-pulsed APC or APC not pulsed with HMBPP.

APC cells, T cells, B cells, monocytes, or DCs (1.0 × 10⁵ each) were added to an eight-well coverglass (NUNC) treated overnight with polylysine. The cells were washed with PBS, fixed with 2% paraformaldehyde, and blocked with avidin/biotin blocking kit (Vector Laboratories) and, subsequently, 2% BSA-PBS. The cells were incubated with or without HMBPP at different concentrations (20–80 ng/ml) at 4°C for 60 min. After washings for removing free HBMPP, cells were stained with the Vγ2Vδ2 TCR tetramer or Vγ1Vδ2 TCR tetramer at 4°C from 60 min and then washed three times. Samples were examined under Carl Zeiss upright LSM 510 META confocal microscope (Zeiss) using the 488 nm laser line as we previously described (18). The Vγ1Vδ2 TCR tetramer and medium alone served as controls. To assess further whether the putative HMBPP-associated molecule was distributed on the cell surface of APCs, the HMBPP-pulsed APCs were stained with Qdot 655-labeled Vγ2Vδ2 TCR or Vγ1Vδ2 TCR tetramer and assessed for membrane distribution in a series of 1-μm-thick section views on Z-stack images using the standard Z-stack technique as we described (18).

To assess whether the putative phosphoantigen-associated molecule on the surface of APC is protease sensitive, APC cell line, THP-1 cells were treated with or without 0.05% trypsin-EDTA solution (Invitrogen) for 10–30 min at room temperature. After extensive washings to remove trypsin, cells were incubated with HMBPP, 80 ng/ml, at 4°C for 60 min and stained with the FITC-labeled TCR tetramer as described above. To exclude the possibility that trypsin treatment caused cell death resulting in negative TCR binding, one portion of the trypsin-treated cells was stained with trypan blue to test the cell viability, and the remainder were continuously cultured for 24 h and then tested for their ability to present HMBPP to Vγ2Vδ2 TCR tetramer. The buffer containing Vγ2Vδ2 TCR tetramer for staining the trypsin-treated cells was reused to stain newly HMBPP-pulsed THP-1 cells to ensure the absence of potentially residual trypsin left in the buffer, which might cause degradation of Vγ2Vδ2 TCR tetramer.

We presumed that the development of high-affinity-binding soluble Vγ2Vδ2 TCR tetramer would make it possible to define APC presentation of phosphoantigen to Vγ2Vδ2 TCR. As an initial effort to characterize Vγ2Vδ2 TCR interaction with phosphoantigen, soluble Vγ2Vδ2 TCRs were expressed, purified, and assessed for the desired conformation mimicking the membrane Vγ2Vδ2 TCR expressed on Vγ2Vδ2 T cells. Jun and Fos leucine zipper domains fused to the extracellular C terminus of Vγ2- and Vδ2-bearing TCR allowed the dimerization of recombinant Vγ2Vδ2 TCR (Fig. 1,a). The His-tag and biotinylation sequences of the secreted TCR chains made it possible to perform two-round purification of Vγ2Vδ2 TCR by passing through a Ni-NTA agarose affinity column and an avidin affinity column (Fig. 1). The two-round purification procedures gave rise to considerably pure Vγ2Vδ2 TCR (Fig. 1, b and c). To determine whether soluble Vγ2Vδ2 TCR resembled native TCR molecules expressed on the Vγ2Vδ2 T cell surface, soluble Vγ2Vδ2 TCR heterodimer was loaded on the dot-blot membrane and assessed for the ability to be recognized by anti-Vγ2 and anti-Vδ2 mAbs that bind only to native cell surface Vγ2Vδ2 TCR. Soluble Vγ2Vδ2 TCR, but not denatured Vγ2Vδ2 TCR, was indeed recognized by anti-Vγ2 and anti-Vδ2 mAbs, respectively (Fig. 1 d), suggesting that soluble Vγ2Vδ2 TCR heterodimer resembled native surface TCR to some extent. The biotinylated Vγ2Vδ2 TCR heterodimer was then incubated with FITC- or quantum dot-conjugated streptavidin, and assembled into fluorescence-labeled Vγ2Vδ2 TCR tetramer to define APC presentation of phosphoantigen to Vγ2Vδ2 TCR.

We proposed that the soluble Vγ2Vδ2 TCR tetramer would be better than the heterodimer to confer a high affinity binding to APC membrane-associated phosphoantigen. Because human and macaque Vγ2Vδ2 T cells share the capability to undergo activation in response to phosphoantigen stimulation (5, 17, 30), macaque Vγ2Vδ2 TCR tetramer was assessed for the ability to bind to membrane-associated HMBPP presented by three different human APC cell lines (SC, THP-1, and U937) using flow cytometry-based analyses. Exogenous HMBPP was associated with the APC membrane in such an appreciable affinity that the membrane-associated HMBPP was readily recognized and stained by the Vγ2Vδ2 TCR tetramer, but not Vγ1Vδ2 TCR tetramer (Fig. 2,a). In fact, the FITC-labeled Vγ2Vδ2 TCR tetramer was able to hold and stain membrane-associated HMBPP for ≥60 min before fixation (data not shown). Importantly, there was no direct binding of Vγ2Vδ2 TCR tetramer to APC that were not incubated with HMBPP (Figs. 2,b and 3, f and g). In addition, preincubation of HMBPP with Vγ2Vδ2 TCR tetramer did not confer on the tetramer the ability to bind directly to APC membrane. When the Vγ2Vδ2 TCR tetramer was incubated with HMBPP and then dialyzed or spun in a column to remove free HMBPP in the buffer, the treated tetramer was not able to bind to the APC cell surface (Fig. 2,b, upper right). However, the treated Vγ2Vδ2 TCR tetramer maintained the ability to bind to and stain the HMBPP-pulsed APC cell lines (Fig. 2 b, lower right). These results demonstrated that while recognition of HMBPP by Vγ2Vδ2 TCR was APC membrane dependent, the membrane-associated HMBPP stably bound to the Vγ2Vδ2 TCR tetramer.

We then visualized the Vγ2Vδ2 TCR tetramer binding to APC membrane-associated HMBPP using confocal microscopy and three-color flow cytometry. The confocal microscopy showed that the Vγ2Vδ2 TCR tetramer-stained cell surface of HMBPP-pulsed macrophages, DCs, and lymphocytes, with some clusters formed on the membrane (Fig. 3, a and b). Interestingly, extensive flow cytometry analyses demonstrated that the FITC-labeled Vγ2Vδ2 TCR tetramer stained not only macrophages/monocytes and DCs, but also primary macaque B cells and T cells when these cells were pulsed with HMBPP (Fig. 3, f and g, and data not shown). Importantly, both confocal microscopy and flow cytometry showed that the Vγ2Vδ2 TCR tetramer did not stain or bind to HMBPP-pulsed APC cell lines from mouse, rat, and pig (Fig. 3, d and g). These results therefore provided further evidence that the Vγ2Vδ2 TCR tetramer bound to membrane-associated HMBPP on APCs from humans and nonhuman primates, suggesting a membrane molecule on various cells from primates, but not APCs from mouse, rat, or pig, presents HMBPP for recognition by Vγ2Vδ2 TCR.

The next central question was whether endogenous phosphoantigen HMBPP could be presented to Vγ2Vδ2 TCR through its association with the putative APC membrane molecule. Because Mycobacterium BCG could infect monocytes/macrophages and produce phosphoantigen, we sought to determine whether Vγ2Vδ2 TCR tetramer was able to bind to phosphoantigen presented through an intracellular pathway by a putative membrane molecule on BCG-infected DCs. Thus, DCs were infected with BCG for 0, 18, 24, or 36 h and then assessed for a stable formation of membrane-associated endogenous phosphoantigen that could bind to Vγ2Vδ2 TCR tetramer. DCs infected for 18 h with BCG were recognized and surface stained by the Vγ2Vδ2 TCR tetramer (Fig. 4). These results demonstrated that endogenous BCG phosphoantigen was able to be presented and associated stably with the putative membrane component on DCs and thereafter recognized by the Vγ2Vδ2 TCR tetramer.

Given the possibility that a putative membrane molecule on APC could present both exogenous and endogenous phosphoantigen for recognition by Vγ2Vδ2 TCR, we sought to investigate the biochemical nature of the HMBPP-associated membrane molecule. To this end, we attempted to treat APC with a proteolytic enzyme that might digest or denature membrane proteins but keep cells alive (31), and then assessed the protease-treated APC for a loss of the ability to present HMBPP to the Vγ2Vδ2 TCR tetramer. The capability of human THP-1 macrophages to present HMBPP to the Vγ2Vδ2 TCR tetramer was diminished after these cells were treated with trypsin for different lengths of time (Fig. 5). The diminished capability of trypsin-treated APC to present HMBPP and bind to Vγ2Vδ2 TCR was not due to the cell death or membrane disruption after trypsin treatment, because these trypsin-treated cells remained alive (resistant to trypan blue staining) and regained their ability in a subsequent 24-h culture to present HMBPP and bind to the Vγ2Vδ2 TCR tetramer (data not shown). Moreover, the diminished ability of APC-HMBPP complex to bind to Vγ2Vδ2 TCR was not attributed to the subsequent digestion of Vγ2Vδ2 TCR tetramer by the residual trypsin potentially existing on the membrane of trypsin-treated cells, because the TCR tetramer recovered from the buffer after staining trypsin-treated macrophages was still able to stain HMBPP-pulsed macrophages compared with the original Vγ2Vδ2 TCR tetramer (85 ± 3.4 vs 91 ± 4.2 binding percentages). Thus, these results suggest that a putative protein or protein-associated membrane component on the APC cell surface may present HMBPP to Vγ2Vδ2 TCR for immune recognition.

The soluble Vγ2Vδ2 TCR tetramer has made it possible to visualize APC presentation of phosphoantigen to Vγ2Vδ2 TCR for immune recognition. The Vγ2Vδ2 TCR tetramer produced using the insect cell-based expression system appears to have a unique capability to bind to APC-associated phosphoantigen compared with the soluble Vγ2Vδ2 TCR dimers reported in previous studies (44, 45). The Vγ2Vδ2 TCR heterodimers previously produced by prokaryotic cells or Chinese hamster ovary cells have not been shown to be able to bind to APC-associated phosphoantigens (44, 45). The soluble Vγ2Vδ2 TCR produced in our expression system may be more representative of native TCR structures on the cell surface of Vγ2Vδ2 T cells and therefore might more readily bind to APC-associated phosphoantigen than those TCRs reported in earlier studies. It is also likely that the stabilization of Vγ2Vδ2 TCR heterodimer using the Jun/Fos dimerization strategy and tetramerization of such stably formed TCR heterodimers confer the unique high-affinity binding to APC-associated phosphoantigen. In fact, the paired Jun/Fos dimerization structures have been shown to form unique heterodimeric coiled coils termed leucine zipper and thus to facilitate dimerization of two chimeric polypeptide chains (36, 41, 46, 47). Consistently, tetrameric MHC and αβ TCR molecules have proved to be much more likely than dimeric molecules to have high-affinity binding to their respective ligands on cell surface (41, 47).

The results from the current studies do not support the simple cell surface contact model that intends to explain the interaction between phosphoantigen and Vγ2Vδ2 TCR (6). This simple model speculates that a nonspecific phosphoantigen (HMBPP) contact on membrane is sufficient enough for HMBPP interaction with Vγ2Vδ2 TCR and subsequent activation of Vγ2Vδ2 T cells (6). If such a simple cell surface contact model were correct, we would expect to see that a potential binding complex formed by theVγ2Vδ2 TCR tetramer and membrane-touched HMBPP would be readily washed out in the staining procedures. This will lead to the absence of the tetramer staining of HMBPP-pulsed APC as seen by flow cytometry or confocal microscopy. On the contrary, we find that the Vγ2Vδ2 TCR tetramer is able to tightly stain or bind to HMBPP-pulsed APC. The fact that Vγ2Vδ2 TCR tetramer can stably stain or bind to HMBPP-pulsed APC for 60 min (see Results) strongly suggests that HMBPP is associated with or bound to an accessory membrane molecule, rather than simply contacts on membrane (6), for immune recognition by Vγ2Vδ2 TCR. Furthermore, the simple HMBPP-cell surface contact model (6) could not explain why only APCs from human and monkeys, but not those from mouse, rat, or pig, were able to present HMBPP to Vγ2Vδ2 TCR tetramer.

The results from the current study strongly support the hypothesis that a putative membrane molecule is required as an accessory component to present phosphoantigen to Vγ2Vδ2 TCR for immune recognition. Our studies provide evidence indicating that exogenous HMBPP must associate itself with APC or a membrane molecule before it can be recognized by and bound to Vγ2Vδ2 TCR. The data are in agreement with earlier reports that indicate that IPP does not need to be processed before Ag presentation and TCR recognition (6). The hypothesis for a membrane-presenting molecule is also supported by the infection experiment showing that the endogenous phosphoantigen (HMBPP) produced in BCG-infected DCs is associated with the APC membrane molecule and stained by the Vγ2Vδ2 TCR tetramer. DCs infected with BCG for 18 h can readily be stained by the Vγ2Vδ2 TCR tetramer. Because BCG is slowly replicated Mycobacterium, a very limited amount of BCG phosphoantigen may be produced and released into cell culture medium within 18 h. It is likely that endogenous phosphoantigen produced by BCG-infected DCs may encounter the putative presenting molecule and then be presented on cell surface for recognition by Vγ2Vδ2 TCR. More importantly, the hypothesis for a membrane-presenting molecule is strongly supported by the finding that the capability of APCs to present HMBPP to Vγ2Vδ2 TCR tetramer was diminished after protease treatment of these APCs. The data also suggest that a putative protein or protein-associated membrane component on the APC cell surface presents HMBPP to Vγ2Vδ2 TCR (31).

The putative phosphoantigen-presenting molecule appears to be expressed only in cell sources of humans and nonhuman primates but not of mice, rats, or pigs. Our data implicate that rodents and pigs may not express the putative phosphoantigen-presenting molecule or have not evolved to have the homolog (s) that is conserved enough for presentation of HMBPP to Vγ2Vδ2 TCR. Our data also help to explain why murine γδ T cells do not recognize mycobacterial phosphoantigen HMBPP, although it may also be due to the lack of murine γδ T cell repertoires that confer specific recognition of the HMBPP complex. Alternatively, this putative molecule is well conserved in humans and nonhuman primates. Monocytes or mooneye/macrophage cell lines from humans and macaques possess the putative membrane molecule that presents HMBPP to Vγ2Vδ2 TCR. In addition, monocyte-derived DCs, B cells, and T cells including CD8⁺ T cells are also able to present the membrane-associated HMBPP to Vγ2Vδ2 TCR. These findings help to explain why cell-cell contacts among Vγ2Vδ2 T cells themselves are sufficient enough for IPP-mediated activation of Vγ2Vδ2 T cells (6, 28).

The studies of the putative phosphoantigen-presenting molecule have been hindered by the absence of useful tools. The earlier experiments using mutant APC cell lines and Ab blocking assays for studying IPP activation of Vγ2Vδ2 T cells indicate that MHC class I, β2M, CD1, or MHC class II molecules do not appear to be the putative molecule that is able to present prenyl pyrophosphate Ags to Vγ2Vδ2 T cells (6). Now, the development of the novel Vγ2Vδ2 TCR tetramer raises the possibility that the putative phosphoantigen-presenting molecule can be further studied and eventually defined. The Vγ2Vδ2 TCR tetramer will also provide a useful tool for elucidating potential pathways for uptake and presentation of microbial phosphoantigen in infections.

We thank Dr. Hassan Jomaa for providing HMBPP and Dr. Karen Hagen, Jewell Graves, and Brenda Paige for technical advice on flow cytometry analysis.

The authors have no financial conflict of interest.