#### Loading of Jurkat cells with fura 2-AM

To remove any growth media, the Jurkat cells were centrifuged (5 min, 1000 r.p.m.) and re-suspended in basal salt solution (BSS, composition (mM): NaCl 125.0, KCl 5.0, MgCl_{2} 1.0, CaCl_{2} 1.5, HEPES 25.0, D-glucose 5.0 and 1 mg ml^{−1} bovine serum albumin, pH 7.3). In a loading volume of BSS (3 ml), the washed Jurkat cells were incubated with fura 2-AM (17 μM) and apyrase (2 u ml^{−1}) at 37°C with shaking for 45 min. Apyrase was included to prevent tonic desensitization of receptors by ATP or ADP (Dainty *et al.*, 1997). Cells were then centrifuged (5 min, 1000 r.p.m.), the supernatant discarded and the pellet re-suspended in BSS to a final concentration of 1 ± 10^{6} cells ml^{−1}.

#### Preparation of human washed platelets

Suspensions of human washed platelets were prepared as previously described by Humphries *et al.*, (1994). Venous blood was obtained from healthy male and female volunteers and anti-coagulated with 1/10 volume, 3.2% trisodium citrate. The blood was centrifuged (15 min, 240 *g*) to obtain platelet rich plasma (PRP) to which prostacyclin (PGI_{2}; 300 ng ml^{−1}) was added to stabilize platelets during the washing procedure. Red cell-free PRP was obtained by centrifugation (10 min, 125 *g*), and following further centrifugation (15 min, 640 *g*), the supernatant was discarded and the pellet re-suspended in 20 ml calcium-free Tyrode solution (CFT, composition (mM): NaCl 137.0, NaHCO_{3} 11.9, NaH_{2}PO_{4} 0.38, KCl 2.68, MgCl_{2} 1.05, D-glucose 5.55, gassed with 95% O_{2}/5% CO_{2} and maintained at 37°C) containing 300 ng ml^{−1} PGI_{2} to give washed platelets.

#### Drugs

Adenosine 5′-diphosphate (sodium salt; ADP), adenosine 5′-triphosphate (disodium salt; ATP), adenosine 5′-O-(3-thiotri-phosphate) (tetralithium salt; ATPyS), adenosine 3′-phosphate-5′-phosphate (monosodium salt; A3P5P), prostacyclin, apyrase (grade V) and Fura 2-AM were obtained from the Sigma Chemical Co. (Poole, U.K.). 2-pro-pylthio-D-*β,γ*-difluoromethylene ATP (trisodium salt AR-C66096) was synthesized in the Medicinal Chemistry Department, Astra Charnwood; 2-methylthio-adenosine 5′-di-phosphate (trisodium salt; 2-MeSADP), 2-methylthio-adenosine 5′-tri-phosphate (tetrasodium salt; 2-MeSATP) and 2-chloro-adenosine 5′-tri-phosphate (tetrasodium salt; 2-ClATP) were supplied by Research Biochemicals Inc. (St. Albans, U.K.). Fura 2-AM was dissolved in dimethylsulphoxide (DMSO); all other drugs were dissolved and diluted in distilled water.

Heat inactivated foetal calf serum, RPMI 1640, L-glutamine (200 mM) and penicillin/streptomycin (5000 iu ml^{−1}, 5000 ug ml^{−1}) were obtained from Gibco BRL (Paisley, U.K.).

#### Data analysis

*Logistic curve fitting* Individual agonist E/[A] curve data were fitted to the following form of the Hill equation:

- (1)

in which α, [A]_{50} and n_{H} are the asymptote, location and slope parameters, respectively. [A]_{50} values were assumed to be log-normally distributed and quoted as p[A]_{50} (-log[A]_{50}) values.

*Antagonist affinity estimation* [A]_{50} data obtained in antagonist experiments were fitted to the following linear form of the Schild equation (Trist & Leff, 1985):

- (2)

where [A]^{c}_{50} is the estimated control [A]_{50} value, [B] is the concentration of the antagonist, *K*_{B} is the antagonist equilibrium constant and n is equivalent to the Schild plot slope parameter (unity for simple competition). If n was not significantly different from unity, it was constrained to unity for pK_{B} estimation.

*Operational model-fitting* The affinity and efficacy of the partial agonist, AR-C66096 (in bovine P2Y_{1}-receptor trans-fected Jurkat cells), was estimated by use of the operational model of agonism (Black & Leff, 1983; Black *et al.*, 1985):

- (3)

in which E and [A] are the pharmacological effect and agonist concentration respectively; E_{m} is the maximum possible effect; *K*_{A} is the agonist dissociation constant (estimated as the negative logarithm, i.e. pK_{A}); T is the efficacy of the agonist (estimated as a logarithm) and n determines the steepness of the occupancy-effect relation.

The analysis was performed with the comparative method, in which the partial agonist (AR-C66096) E/[A] curve data were fitted to the operational model (Equation 3) simultaneously to fitting the full agonist (ADP) E/[A] curve data to a Hill equation of the form:

- (4)

where E_{m} and n are as defined above and [A]_{50} is the location parameter of the full agonist curve. This allows *K*_{A} and τ for the partial agonist to be estimated as well as E_{m} and n from each pair of curves (see Leff *et al.*, 1990 for details).

*Theoretical analysis of the effect of ATP contamination by ADP* Increases in [Ca^{2+}]_{i} induced by commercially obtained‘ATP’ may actually be attributed to the agonist effects of contaminating ADP. To determine whether the partial agonist effects of ATP observed in the present study were the result of receptor activation by ADP and concomitant receptor blockade by ATP, we fitted the ADP and ATP E/[A] curve data from human washed platelets simultaneously to the following equation:

- (5)

in which ADP has been assumed to be a full agonist and ATP to be a competitive antagonist; X represents the total concentration of agonist (A) and antagonist (B), that is, ([A] + [B]) in the sample; q represents the fraction of the total sample that is antagonist (q = [B]/([A] + [B]) and therefore (1-q) represents the fraction that is agonist; E_{m} is the maximum possible effect; n is the slope index of the occupancy-effect function; [A]_{50} is the midpoint location parameter and *K*_{B} is the antagonist equilibrium dissociation constant.

q was measured for the ATP sample by high performance liquid chromatography (h.p.l.c.) whereas q was assumed to be zero for the ADP sample. The analysis was carried out by fitting data from individual donors which allowed the estimation of *K*_{B} (as well as E_{m}, n and [A]_{50}) for each donor.

All of the data fitting procedures and simulations were carried out with Microsoft Excel. Results are expressed and plotted as mean values ± s.e. Statistical differences were assessed by the use of Student's *t* test and considered significant at the level *P* < 0.05.