Chemical Synthesis and Biological Activity of Bromohydrin Pyrophosphate, a Potent Stimulator of Human γδ T Cells*

Small phosphorylated metabolites from mycobacteria stimulate human γδ T lymphocytes. Although such phosphoantigens could prove useful in the composition of vaccines involving γδ T cell-mediated immunity, their very low abundance in natural sources limits such applications. Here, we describe the chemical production, purification, and bioactivity of a phosphorylated bromohydrin (BrHPP) analogue that mimics the biological properties of natural phosphoantigens. This compound can be obtained in gram amounts, is easy to detect, and is of high stability in aqueous solutions. Whereas unspecific binding of BrHPP to a wide panel of cell surface receptors is not detected even at micromolar concentrations, nanomolar concentrations specifically trigger effector responses of human γδ T lymphocytes. Thus, BrHPP is a novel molecule enabling potent immunostimulation of human γδ T lymphocytes.

Small phosphorylated metabolites from mycobacteria stimulate human ␥␦ T lymphocytes. Although such phosphoantigens could prove useful in the composition of vaccines involving ␥␦ T cell-mediated immunity, their very low abundance in natural sources limits such applications. Here, we describe the chemical production, purification, and bioactivity of a phosphorylated bromohydrin (BrHPP) analogue that mimics the biological properties of natural phosphoantigens. This compound can be obtained in gram amounts, is easy to detect, and is of high stability in aqueous solutions. Whereas unspecific binding of BrHPP to a wide panel of cell surface receptors is not detected even at micromolar concentrations, nanomolar concentrations specifically trigger effector responses of human ␥␦ T lymphocytes. Thus, BrHPP is a novel molecule enabling potent immunostimulation of human ␥␦ T lymphocytes.
Stimulating ligands for ␣␤ T lymphocytes are usually composed of single peptides complexed at the surface of major histocompatibility complex molecules. Some small non-peptidic structures, however, may also constitute specific agonist ligands for T cells, particularly ␥␦ T lymphocytes. In human blood, about 3% of T cells initiate their physiological function upon recognition of small phosphorylated non-peptide antigens (phosphoantigens). This cognate interaction involves on the one hand phosphoantigens in the absence of major histocompatibility complex-presenting molecules, and on the other hand, highly selective receptors (TCR) 1 of ␥␦ subtype. In nature, phosphoantigens that can activate human ␥␦ T cells at nanomolar concentrations are produced by Gram-positive and Gram-negative bacteria and also by some eukaryotic parasites and plants. Synthetic analogues of natural phosphoantigens are also known, but their stimulating concentrations for the reactive cells never go below the micromolar range. Mycobacterium tuberculosis, the agent of human tuberculosis, produces four distinct phosphoantigens. These molecules share a moiety that is responsible for the potent stimulation of ␥␦ cells seen in tuberculosis patients (1). The structure of this common core is 3-formyl-1-butyl-pyrophosphate, a recently described phosphoester (2). Its metabolic production might be related to the non-mevalonate (or so-called Rohmer's) pathway for isoprenoid precursor biosynthesis (3). 3-formyl-1-butyl-pyrophosphate is produced in very small amounts in slow-growing mycobacteria such as Mycobacterium tuberculosis and only accumulates to submicromolar concentrations in culture media from fast-growing mycobacterial species (4). Getting large amounts of highly bioactive phosphoantigens by purification routes from such natural sources is therefore hard to conceive.
Such molecules could prove therapeutically useful for immunotherapeutic approaches involving ␥␦ T cell-mediated immunity, such as elicitation of anti-infectious protection or antitumor immunity (5,6). To address the need for readily available highly bioactive phosphoantigens, we have developed a synthetic reagent called bromohydrin pyrophosphate (BrHPP), whose biological properties on human T cells are optimized compared with those of 3-formyl-1-butyl-pyrophosphate.

MATERIALS AND METHODS
Chemical Synthesis-All glassware and equipment were dried for several hours prior to use. Unless otherwise stated, the reagents and starting material were from Fluka. Trisodium (R,S)-3-(bromomethyl)-3-butanol-1-yl-diphosphate (BrHPP) was produced as white amorphous powder by the following procedure. Tosyl chloride (4.8 g, 25 mmol) and 4-(N,N-dimethylamino-) pyridine (3.4 g, 27.5 mmol; Aldrich) were mixed under magnetic stirring with 90 ml of anhydrous dichloromethane in a 250-ml three-necked flask cooled in an ice bath. A solution of 3-methyl-3-butene-1-ol (2.2 g, 25 mmol) in about 10 ml of anhydrous dichloromethane was then slowly introduced with a syringe through a septum in the flask, and the ice bath was then removed. The reaction was monitored by silica gel TLC (pentane/ethyl acetate, 85:15 (v/v)). After 2 h with constant stirring, the mixture was precipitated by dilution into 1 liter of hexane and filtered, and the filtrate was concentrated under reduced pressure. This filtration/suspension step was repeated using diethyl ether, and the resulting oil was purified by liquid chromatography on silica gel (pentane/ethyl acetate, 85:15 (v/v)), yielding a yellow oil of 3-methyl-3-butene-1-yl-tosylate (5.6 g, 23. 5  Disodium dihydrogen pyrophosphate (51.5 mmol, 11.1 g) dissolved in 100 ml of deionized water (adjusted to pH 9 with NH 4 OH) was passed over a cation exchange DOWEX 50WX8 (42 g, 200 meq of form H ϩ ) column and eluted with 150 ml of deionized water (pH 9). The collected solution was neutralized to pH 7.3 using tetra-n-butyl ammonium hydroxide and lyophilized. The resulting hygroscopic powder was solubilized with anhydrous acetonitrile and further dried by repeated evaporation under reduced pressure. The resulting Tris (tetra-n-butyl ammonium) hydrogenopyrophosphate (97.5% purity by HPAEC; see below) was stored (concentration, ϳ0.5 M) at Ϫ20°C in anhydrous conditions under molecular sieves. 100 ml of a solution containing 50 mmol of Tris (tetra-n-butyl ammonium) hydrogenopyrophosphate (0.5 M, 2.5 eq) in anhydrous acetonitrile under magnetic stirring in a 250-ml * This work was supported by institutional grants from INSERM, l'ARC, MENRT (action PRFMMIP 2000), and EU Cluster for TB vaccine. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed. Tel.: 335-6274-8364; Fax: 335-6274-8386; E-mail: fournie@purpan.inserm.fr. 1 The abbreviations used are: TCR, T cell receptor; BrHPP, bromohydrin pyrophosphate; ESI, electrospray ionization; MS, mass spectrometry; HPAEC, high pressure anion exchange chromatography; PBL, peripheral blood lymphocytes; TNF-␣, tumor necrosis factor-␣; 3fbPP, 3-formyl-1-butyl-pyrophosphate; IL, interleukin. three-necked flask cooled in an ice bath were slowly mixed with 20 mmol (4.8 g) of 3-methyl-3-butene-1-yl-tosylate introduced via a septum with a syringe. After 20 min, the ice bath was withdrawn, and the reaction was left under agitation at room temperature for 24 h. The reaction was analyzed by HPAEC (see below), evaporated, and diluted into 50 ml of a mixture composed of a solution (98 % volume) of ammonium hydrogenocarbonate (25 mM) and 2-propanol (2 volume %). The resulting mixture was passed over a cation exchange DOWEX 50WX8 (NH 4 ϩ , 750 meq) column formerly equilibrated with 200 ml of the solution (98 % volume) of ammonium hydrogenocarbonate (25 mM) and 2-propanol (2 volume %). The column was eluted with 250 ml of the same solution at a slow flow and collected in a flask kept in an ice bath. The collected liquid was lyophilized, and the resulting white powder was solubilized in 130 ml of ammonium hydrogenocarbonate (0.1 M) and completed by 320 ml of acetonitrile/2-propanol (v/v). After agitation, the white precipitate of inorganic pyro-and mono-phosphates was eliminated by centrifugation (2100 ϫ g, 10°C, 8 min). This procedure was repeated three times, the supernatant was collected and dried, and the resulting oil was diluted in 120 ml of water. Remainders of unreacted tosylates were extracted three times by chloroform/methanol (7:3 (v/v)) in a separatory funnel, and the water phase was finally lyophilized. The resulting white powder was again washed twice by acetonitrile/chloroform/methanol (50:35:15 (v/v)) and dried under gentle N 2 flow. 11.25 mmol of pure 3-methyl-3-butene-1-yl-pyrophosphate triammonium salt were obtained by this procedure (75% yield) and were then dissolved in 200 ml of water for oxidation. For 6 mmol of 3-methyl-3-butene-1-ylpyrophosphate, an aqueous solution of Br 2 (0.1 M) kept at 4°C was added dropwise until appearance of a persistent yellowish color, yielding after evaporation 5.8 mmol (2.3 g) of an acidic solution (pH 2.1) of BrHPP, which was immediately neutralized by passing over DOWEX 50WX8 -200 (NH 4 ϩ , 48 meq). The ammonium salt of BrHPP obtained after lyophilization was dissolved in water and separated from bromides by passing through Dionex OnGuard-Ag (2 meq/unit) cartridges and an on-line column of (100 meq, 21 g) DOWEX 50WX8 -200 (Na ϩ ) eluted by milli-Q water. Colorless stock solutions of BrHPP (Na ϩ ) were filtered over Acrodisc 25 membranes of 0.2 M and kept as aliquots at Ϫ20°C.
HPLC-Final purification of BrHPP was achieved by HPLC (Spectra system P1000 XR device) on an analytic Symmetry 5 C18 column (Waters) eluted at 1 ml/min and 20°C with the ternary gradient indicated below. Upstream of detectors, a split of eluent distributes 190 l/min in the online MS detector (see below), and the remaining 810 l/min was sent to the Waters 996 photodiode array detector. Single wavelength detection at ϭ 226 nm was of 7 milliabsorbance units for 6 g of BrHPP injected in 25 l (Rheodyne injector). The gradient program was as follows: solvent A, acetonitrile; solvent B, 50 mM ammonium acetate; solvent C, water; 0 -7 min, 5% B in C; 7.1-11 min, 100% C; 12-15 min, 100% A; 15-17 min, 100% C.
Mass Spectrometry and NMR-Mass spectrometry was performed with an LCQ ion trap mass spectrometer (Finnigan MAT, San Jose, CA). LC-ESI-ion trap MS was performed with a standard Finnigan ESI source in negative ion mode at a voltage of 4.3 kV. The heated transfer capillary was kept at a temperature of 200°C, the sheath gas (N 2 ) flow was 80 units, and the auxiliary gas (N 2 ) flow was 15 units. The maximum ion collection time was set at 800 ms, and two microscans were summed per scan. The scan MS range was m/z 75-500. MS n was performed with a commercial nanospray ESI source (The Protein Analysis Co., Odense, Denmark) using glass capillaries (The Protein Analysis Co.), which were positioned directly at a distance of about 1 mm from the entrance hole of the heated transfer capillary with the help of a stereomicroscope. The capillaries were palladium and gold-coated for electrical contact. The glass capillaries were filled with 5 l of analyte solution. Voltage of the nanospray needle was set to 700 V; no nebulizer gas was used in this spray mode. The heated transfer capillary was kept at a temperature of 150°C. Samples were analyzed in the negative ion polarity mode. Collision energy was adjusted manually (range, 15-25%). NMR was done as previously published (2).
Cell Culture-Isolation of peripheral blood lymphocytes (PBL):heparinized peripheral blood was taken from healthy donors, and PBL were separated on Ficoll-Paque PLUS (Amersham Pharmacia Biotech), washed three times, and then cultured at 10 6 cells/ml in culture medium RPMI 1640 with Glutamax-I (Life Technologies, Inc.), supplemented with 10% AB human serum, 25 mM Hepes, 100 units/ml penicillin G, 100 g/ml streptomycin, and 1 mM sodium pyruvate.
Interferon-␥ released by activated T cells was measured by a sandwich enzyme-linked immunosorbent assay. 5 ϫ 10 4 ␥␦ T cells/well were incubated with stimulus plus 25 units of IL2/well in 100 l of culture medium during 24 h at 37°C. Then, 50 l of supernatant were harvested for enzyme-linked immunosorbent assay using mouse monoclonal antibodies (BIOSOURCE, Camarillo, CA).
Microphysiometry-The cell acidification rate was monitored using a cytosensor microphysiometer (Molecular Devices, Crawley, UK), which measures pH of extracellular fluid using a silicon-based method (7). The raw data from sensor output give mV ϭ f(t), which may be converted to pH ϭ f(t). The system allows cells (8 ϫ 10 5 ) disposed in sensor chambers to be irrigated (flow on period, 90 s) by low buffered RPMI medium (Molecular Devices) containing phosphoantigen or not; then, during a flow off period (30 s), the sensor data are used to calculate a slope, giving ⌬pH/⌬t and referred as to the acidification rate. In each experiment, 32 ϫ 10 5 ␥␦ T cells were resuspended in 30 l of low buffered medium plus 10 l of melted low temperature-melting agarose at 37°C. 10 l of the mixture were rapidly spotted on a cytosensor cell capsule. After 10 min the cell capsule was assembled and loaded in the sensor chamber of a microphysiometer. The experiments were run at 37°C, and the low buffered medium (pH 7.4) was perfused at 100 l/min.
Phenotype Analysis by Flow Cytometry-5 ϫ 10 5 cells were washed in phosphate-buffered saline containing 5% fetal calf serum and incubated for 30 min at 4°C with anti-CD3-PE and anti-␦2-fluorescein isothiocyanate monoclonal antibodies (Beckman Coulter) or isotypic controls. Samples were then washed in phosphate-buffered saline, 5% fetal calf serum and immediately acquired by an EPICS XL flow cytometer (Beckman Coulter).

Molecular Overlay of 3-Formyl-1-butyl-pyrophosphate and
BrHPP-Former structure-activity relationship studies of natural and synthetic phosphoantigens have shown that among monoesters of pyrophosphate (8,9), several organic esters with chemical reactivity (e.g. of Sn-2 type) (10) presented ␥␦ cellstimulating bioactivities higher than that of single chain alkyl, such as ethyl pyrophosphate. In addition to these parameters, a topological fit of the alkyl chain clearly contributes to optimize recognition by the ␥␦ TCR. To select a synthetic phosphoantigen matching 3-formyl-1-butyl-pyrophosphate (3fbPP) as much as possible, we overlaid this latter compound and several synthetic compounds. The bromohydrin phosphate BrHPP was selected for its good superimposition to 3fbPP (arbitrarily shown as the same enantiomers in Fig. 1) and two other ␥␦ cell-stimulating phosphoantigens, isopentenyl pyrophosphate and ethyl pyrophosphate (8,9). Both 3fbPP and BrHPP molecules distribute hydrogen bonds from the carbonyl and hydroxyl groups similarly in their surrounding volume while aligning for the remaining part of the molecule. Shorter homologues (C4) of BrHPP have been produced and were described elsewhere, but they do not present bioactivity for human T lymphocytes (10). In addition, this compound may be conveniently derived from oxidation of isopentenyl pyrophosphate (see below).
Chemical Synthesis of BrHPP-The procedure for BrHPP synthesis was modified from Ref. 11 and is summarized in Fig.  2. A fully detailed description of this synthesis, which yields the racemate of BrHPP, due to asymmetry of the C3 position, is given under "Materials and Methods." Thus, in the absence of further enantiomer resolution of the synthetic mixture, the produced BrHPP compound corresponds to the racemic structure represented in Fig. 2. The produced BrHPP was essentially devoid of unreacted reagents when using the separation scheme based on differential solvent reprecipitation, as described under "Materials and Methods." 4 -20% of the recovered material still corresponded to other products (e.g. phosphate), warranting a final step of HPLC separation (see below). At this step, BrHPP was obtained as a triammonium salt, which was found to interfere with several cell culture assays (data not shown). It was then converted to BrHPP (Na ϩ form) by cation exchange. This latter form is stable in aqueous solutions and can be stored at Ϫ20°C for 4 months without detectable structural degradation. Chromatographic and Structural Assignment of BrHPP-Because of the halohydrin structure of BrHPP, its chromatographic analysis could not be undertaken by HPAEC as described for natural phosphoantigens (12), because hydroxide eluents of HPAEC rapidly catalyzed an epoxide rearrangement by HBr elimination (data not shown). Therefore, a chromatographic procedure for BrHPP analysis was based upon use of near-neutral pH eluents. The stock solutions of BrHPP were analyzed by ion pair reverse-phase C18 HPLC and monitored by UV-visible diode array and MS detection following a procedure described previously, with minor modifications (see "Materials and Methods" and Ref. 1). As shown in Fig. 3A, several UV-absorbing and phosphorylated contaminants eluted close to the major peak of BrHPP (sample injected: 0.75 mM, R t ϭ 5.3 min.). Ion-trap MS in negative mode of the collected fractions gave a unique set of signals at m/z 341 and 343, as expected for the pseudomolecular anion of BrHPP. Its composition, C 5 H 12 P 2 O 5 Br, was supported by the relative abundance of natural bromine isotopes evidenced by zoom scan (Fig. 3B). V␦2-encoded antigen receptors borne by phosphoantigen-reactive human ␥␦ T lymphocytes (1, 8, 9, 14 -16).
BrHPP Activates Proliferation and Cytokine Release by TCR ␥␦ ϩ T Lymphocytes-T lymphocytes respond to antigenic activation by proliferating, secreting cytokines, and/or mediating toxicity for target cells. Accordingly, when they are stimulated by the mycobacterial phosphoantigen 3fbPP, ␥␦ T cells expressing the V␥9/V␦2-encoded TCR expand specifically in culture and secrete TNF-␣ (2). Thus we tested whether increasing concentrations of BrHPP would lead to the selective outgrowth of the V␥9/V␦2 T lymphocytes in in vitro cultures of PBL. The result of a representative experiment (of 10 independent ones) is shown in Fig. 4A. BrHPP proved to be as potent a stimulus as whole mycobacterial extract for the in vitro expansion of the V␥9/V␦2 T cells. This property was abrogated by dephosphorylation of BrHPP using alkaline phosphatase in the culture, demonstrating that BrHPP owes its bioactivity on ␥␦ T cells to its pyrophosphate moiety, a characteristic hallmark of all natural and synthetic phosphoantigens (1,17). In addition to proliferative responses, human ␥␦ T cells reactive to phosphoantigens frequently secrete cytokines in response to antigen stimulation. When ␥␦ T cells were exposed to BrHPP, the production of high levels of interferon-␥ (Fig. 4C) and TNF-␣ (Fig. 4D) in the culture medium correlated strongly with the concentrations of BrHPP used. These observations indicate that BrHPP presents the same T cell-stimulating property as natural phosphoantigens.
BrHPP Causes Early Activation of Specific ␥␦ T Cells-Intracellular transduction of activating signals delivered by encounter of the natural phosphoantigen 3fbPP involves extracellular acid release by reactive ␥␦ T cells (2). A similar pathway of signal transduction could be expected for recognition of a phosphoantigen mimicking 3fbPP, such as BrHPP. We tested this using Cytosensor microphysiometry, to detect whether ␥␦ T cell activation with BrHPP led to early acid release, as already reported for ␣␤ T cells (18). Fig. 5A shows that a transient (10 min) pulse of ␥␦ cells with BrHPP does result in acid release (7). As illustrated in Fig. 5B, the doseresponse relationship in this experiment matched the dose responses for cytokine release (Fig. 4, C and D). Whereas this early metabolic response was not sustained at low BrHPP concentrations (5 and 25 nM), the initial burst was followed by sustained signaling for at least 1 h when cells were triggered by high BrHPP concentrations (50 -100 nM, Fig. 5A). In this latter case, the metabolic burst presented a particularly rapid onset, as evidenced by comparing the raw data from an unstimulated chamber (Fig. 5C, lower trace) and from a chamber perfused with BrHPP (Fig. 5C, upper trace). When taking into account the delay for BrHPP diffusion onto the cells (7 s in these assays), the mV collapse recorded in the perfused chamber indicated that extracellular acidification by ␥␦ T cells had occurred about 10 s after exposure to BrHPP. Thus, under saturating concentrations, phosphoantigen recognition by the ␥␦ TCR leads to very rapid activating events. DISCUSSION This paper details the structure, chemical synthesis, and biological property of bromohydrin pyrophosphate, a novel molecule activating human ␥␦ T lymphocytes. Based on the molecular overlay of this compound with the natural ␥␦ T cell ligand found in mycobacteria, it was hoped that this synthetic compound could mimic the biological properties of the naturally occurring 3fbPP, i.e. activation of a specific ␥␦ T cell subset in human blood (see Ref. 19 for a review). We describe a convenient and straightforward mode of synthesis for producing BrHPP. This simple method is based on pyrophosphorylation of the tosylated C5 precursor, followed by stoichiometric oxidation of the pyrophosphoester product in aqueous bromine. Because the compound carries a chiral C3, the resulting product is a racemic mixture used without further resolution of enantiomers. This straightforward and inexpensive synthesis is followed by a purification scheme involving solvent precipitation, LC, and HPLC to eliminate residual inorganic phosphate and bromide. The final stock solutions of BrHPP (Na ϩ ) salts are very stable and can be stored for several months without degradation. Little information about structural changes of the organic moiety of phosphorylated metabolites is usually drawn from HPLC-MS in negative mode, thereby limiting its use as an analytical tool. The mass spectral data of BrHPP presented here demonstrate a highly sensitive detection of BrHPP in aqueous phases, and its bromine content enables unambiguous detection for pharmacological follow-up studies. For the reasons listed above, BrHPP appears to be a promising lead candidate for therapeutic explorations among synthetic phosphoantigens.
In vitro cultures of bulk human lymphocytes carried out in the presence of 100 nM BrHPP and IL2 lead to the systematic expansion of T lymphocytes, which express the phosphoantigen-reactive ␥9␦2 TCR, and no other cell subset. This has been described previously for total lymphocyte populations stimulated by crude extracts from M. tuberculosis and is known to rely upon the presence of several stimulating phosphoantigens. Here, BrHPP also acts as a phosphoantigen agonist, because BrHPP dephosphorylation abolishes this bioactivity. Exposure to BrHPP also elicits TNF-␣ and interferon-␥ release, indicating that the full range of ␥␦ T cell effector response is activated by this ligand. Microphysiometric analysis showed that this activation results from exposure to a typical activating agonist (20). Its early signal transduction involves a BrHPP dose-dependent extracellular acid release, as was shown for ␣␤ T cells stimulated with peptide-major histocompatibility complex tetramers (21). Whereas high bioactive doses of BrHPP triggered a strong acidification burst followed by a sustained intracellular signaling, suboptimal BrHPP concentrations (5 nM) led to barely detectable signaling. When rapidly exposed to saturating BrHPP concentrations, ␥␦ cells respond by extracellular acidification within about 10 s, indicating that little (if any) intermediate processing of the stimulating BrHPP occurs prior to triggering the T cell reaction.
In summary, although synthetic BrHPP presents the same biological properties as natural phosphoantigens, the possibility of synthesizing gram amounts from simple procedures bypasses the production drawbacks of the natural counterparts. This makes BrHPP an attractive candidate for investigations of selective ␥␦ T cell-based immunomodulation approaches. Future studies will evaluate the potential of this novel immunostimulating molecule in subunit vaccines where ␥␦ T cell contribution in vivo is expected to be beneficial, such as antituberculous immunity and protection against acute leukemia.
Acknowledgments-We thank Sanofi-Synthélabo (Labège) for kind gifts of rIL2 and G. Cassar (IFR Claude de Préval) for advice with cytometry assays. We acknowledge A. T. N. Joly for critical reading of this manuscript.  TABLE I-continued   TABLE I-continued