Small Molecule Antagonists of NAADP-Induced Ca2+ Release in T-Lymphocytes Suggest Potential Therapeutic Agents for Autoimmune Disease

Nicotinic acid adenine dinucleotide phosphate (NAADP) is the most potent Ca2+-releasing second messenger known to date, but the precise NAADP/Ca2+ signalling mechanisms are still controversial. We report the synthesis of small-molecule inhibitors of NAADP-induced Ca2+ release based upon the nicotinic acid motif. Alkylation of nicotinic acid with a series of bromoacetamides generated a diverse compound library. However, many members were only weakly active or had poor physicochemical properties. Structural optimisation produced the best inhibitors that interact specifically with the NAADP/Ca2+ release mechanism, having no effect on Ca2+ mobilized by the other well-known second messengers d-myo-inositol 1,4,5-trisphosphate [Ins(1,4,5)P3] or cyclic adenosine 5′-diphospho-ribose (cADPR). Lead compound (2) was an efficient antagonist of NAADP-evoked Ca2+ release in vitro in intact T lymphocytes and ameliorated clinical disease in vivo in a rat experimental autoimmune encephalomyelitis (EAE) model of multiple sclerosis. Compound (3) (also known as BZ194) was synthesized as its bromide salt, confirmed by crystallography, and was more membrane permeant than 2. The corresponding zwitterion (3a), was also prepared and studied by crystallography, but 3 had more desirable physicochemical properties. 3 Is potent in vitro and in vivo and has found widespread use as a tool to modulate NAADP effects in autoimmunity and cardiovascular applications. Taken together, data suggest that the NAADP/Ca2+ signalling mechanism may serve as a potential target for T cell- or cardiomyocyte-related diseases such as multiple sclerosis or arrhythmia. Further modification of these lead compounds may potentially result in drug candidates of clinical use.


Results and Discussion
Synthesis. Small-molecule NAADP analogues were prepared by alkylation of nicotinic acid derivatives with a series of bromoacetamides (Fig. 2). Bromoacetamides (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) were produced in good to excellent yield by treatment of bromoacetyl bromide with selected amines at low temperature. Treatment of the acetamides with nicotinic acid (or nicotinic acid derivatives), generated the desired pyridinium salts 2, 3 and (18-32). Difficulties encountered during purification using column chromatography, because of the high polarity of both the carboxylic acid group and the positive charge on nitrogen, were overcome by precipitation using slow addition of ether to a methanol solution followed by several crystallizations from acetone/MeOH. Other analogues bearing an alkene and an acetamidine group were also prepared to ascertain the importance of the amide motif linked to the nitrogen atom of nicotinic acid. Compound (34) which has a two-carbon chain between the nitrogen and the amide bond, was conveniently prepared under the usual conditions using bromopropionyl chloride rather than bromoacetyl bromide as a starting material (Fig. 3A). Acetamidine analogue (35)  was synthesized from chloroacetronitrile that was first treated with NaOMe followed by n-octylamine hydrochloride to form the N-octyl substituted choloracetamidine 34,35 . This intermediate was reacted in situ with nicotinic acid to afford 35 in 22% yield (Fig. 3B). Alkene (37) was synthesized in two steps from commercially available n-octanol via a Mannich reaction 36 followed by a NaBH 4 reduction. The hydroxyl intermediate (36) thus obtained 37 was then converted to its iodo-derivative and reacted with nicotinic acid to give (37) in 55% yield (Fig. 3C). As pyridinium salts can be reduced to the 1,4-dihydropyridine form using Na 2 S 2 O 4 /NaHCO 3 38 , we designed compound (38) to act as a potential pro-drug of the pyridinium compound 26 (Fig. 3D). We hoped that removing the positively charged nitrogen in the pyridinium ring would generate a more cell-permeant analogue that in principle could be oxidized back to the active form within cells.
The amide may increase the rigidity of the analogue, and we therefore also synthesized compounds (41-43) based on the report of White et al. 39 (Fig. 3E), where such compounds have been used as inhibitors of the DNA repair enzyme poly(ADP-ribose) polymerase.
In vitro studies and structure activity relationship (SAR). Novel NAADP analogues were evaluated in rat effector T cells specific for myelin-basic protein using [ 3 H]deoxy-thymidine incorporation, as published previously 31 . Among different analogues 1-3, 21-23, 25 only 2 and 3 concentration-dependently antagonized antigen-dependent re-activation of rat effector T cells (Fig. 5 for the activity of 2, and published data for 3 31 ).
Although only a small library of NAADP analogues was synthesized, two inhibitors of antigen-dependent re-activation of rat effector T cells have been discovered that allow us to establish a very preliminary structure-activity relationship (SAR) for these small molecule antagonists (Fig. 4). The first conclusion relates to the C-3 position of the pyridinium ring: here, the carboxyclic acid moiety is crucial since the corresponding amide 31 or esters 32 are inactive. This might be expected, since NADP is inactive in Ca 2+ release.
An alkyl side chain attached to the amide linker was introduced to improve cell permeability compared to 1. Analogues with single alkyl chains; n-butyl 19, n-hexyl 27, n-heptyl 26, n-octyl 3 and n-decyl 25, were synthesized. It seems that apart from the nicotinic acid moiety, a side chain with greater than 4 but less than 10-carbon atoms is required for activity in intact T cells. Compounds with 10-or more carbon side-chains could not be studied quantitatively due to their poor solubility in biological buffers. Rings at the amide side-chain 21, 22, 23 are not tolerated. Hence an optimum balance between solubility and membrane permeability at this stage was achieved with eight carbons, spread over two chains 2 or one 3. The amide and its position relative to the nicotinamide also appear crucial for activity; increasing the length of the carbon tether to the nicotinamide 34, replacing the amide with an acetamidine 35 or alkene 37 all significantly decreased or abolished activity (data not shown).
Further in vitro studies with lead compound 2. Co-injection of 2 almost completely antagonized the effect of NAADP with an IC 50 of 1.7 ± 0.8 µM; highest inhibition (approx. 80%) was obtained at 1 mM 2 (Fig. 5A,B). Evidence for the specificity of the inhibitory effect of 2 was obtained by co-injection with  d-myo-Ins(1,4,5)P 3 or cADPR. In these experiments d-myo-Ins(1,4,5)P 3 induced a similar Ca 2+ signal consisting of a rapid (peak) and sustained (plateau) phase as compared to NAADP. However, no difference between the presence and absence of 2 was observed (Fig. 5C). Similarly, Ca 2+ signalling induced by microinjection of cADPR was not affected by 2 (Fig. 5D). Unexpectedly, the effect of cADPR was even enhanced in the presence of 2; however, the underlying mechanism is unclear.
To determine the role of the NAADP/Ca 2+ signalling system in T cell receptor/CD3 mediated activation, we studied the effect of 2 on stimulation of rat encephalitogenic CD4-positive 40 T cells (T MBP cells) via CD3 crosslinking. Here, 2 was shown to block both myelin-basic protein (MBP) and ConA-mediated proliferation (Fig. 6).
In vivo studies. Increasing concentrations of 2 up to 500 μM did not significantly reduce the number of T MBP cell blasts when incubated with these cells for 48 h, indicating that it does not have a large cytotoxic effect on resting cells, but specifically affects proliferating T cells (Fig. 7A). Therefore, 2 was studied in vivo in rat experimental autoimmune encephalomyelitis (EAE), a T cell-mediated animal model for MS 12 . Animals injected with  PBS (vehicle) or nicotinic acid (as a control compound with some structural similarities to 2) developed severe paralysis of tail and rear legs due to the inflammation. Clinical symptoms started on day 3.5 and reached their maxima on day 4 to 6.5. Between day 3.5 and day 6, no difference between PBS and nicotinic acid was observed. Animals treated with 2 showed a less rapid increase of symptoms, a decreased maximum score, a more rapid decline of clinical symptoms (p < 0.05 vs. either PBS or nicotinic acid, Mann-Whitney comparisons in EAE peak phase, day 4-8; Fig. 7B) and less decrease in body weight compared to the other two groups (p < 0.05 vs. either PBS or nicotinic acid, Mann-Whitney comparisons in EAE peak phase, day 3-5; Fig. 7C). Furthermore, the number of autoimmune T MBP cells invading the CNS was reduced by 50%, upon treatment with 2 (p < 0.05 vs. either PBS or nicotinic acid, ANOVA), while the T cells in parathymic lymph nodes (LNs) and spleen were found to be slightly increased (Fig. 7D). It is known from previous studies 41 that intravenously transferred encephalitogenic T cells first move to the spleen and para-thymic lymph nodes before reaching the CNS where they are re-activated by recognizing their target autoantigen (day 4 post transfer). This re-activation is essential for the recruitment of immune cells into the CNS and thus it is crucial for the induction of CNS inflammation and clinical disease.
Since GFP-transduced encephalitogenic T cells were used for the transfer EAE experiment (T MBP-GFP cells 40 ), the numbers and localization of these cells could be determined on day 4 post transfer. For the animals treated with 2, the number of CNS-infiltrating T MBP-GFP cells decreased by about 50% (Fig. 7D). In addition, the number of T cells was reduced significantly. Our data thus strongly suggest that inhibition of NAADP/Ca 2+ signalling will block the re-activation of effector T cells and render beneficial effects to EAE animals. Further in vitro studies with different batches of 2 ( Supplementary Fig. S4) suggested possible less than optimal membrane permeability of 2, as some batch-dependent biological variability was noted in proliferation blockade efficacy, despite all batches of material showing equal inhibitory effects when co-injected with NAADP. Although 2 is uncharged overall as the zwitterion, the polar structure possibly limits its membrane permeability. We sought to optimize the structure of 2 and to improve cell permeability and this was achieved through modification of 2 to produce 3 (BZ194) (Fig. 8), representing structurally the rearrangement of the di-substituted-amide of 2 to the mono-substituted 3; the hydrophobic tails in the former have the same number of carbon atoms as in 3, but in 2 there are two tails (2 × 4 carbon atoms) and in 3 only one (1 × 8 carbon atoms). 3 Thus has the same molecular formula as 2, but is mono N-substituted, and is less water-soluble. It has much better cell permeability than 2 and shows highly reproducible inhibitory effects in intact T cells 31 . Like 2, 3 is an inhibitor of NAADP-induced Ca 2+ release, and similarly did not interact with other mechanisms such as Ins(1,4,5)P 3 and cADPR and has been studied extensively both in vitro 31 and in vivo 42 . In EAE, 3 not only ameliorates autoimmune disease when given before the onset of disease but also after, suggesting that a compound from this class, perhaps after further optimization, may find therapeutic use in human autoimmune disease. To date, 3 as the best optimized version of 2 so far, has emerged as the most successful agent for biological applications from this work.
We used the recently reported SwissADME program 43 to compare the drug-likeness of our three most promising compounds, 1-3. In agreement with our initial studies, 1 was predicted to be more water soluble, but less membrane permeant compared to 2 and 3. Unlike 1, both 2 and 3 were predicted to cross the blood-brain barrier. All three were predicted to have good gastrointestinal absorption. Interestingly, the long alkyl chains of both 2 and 3 were highlighted as less desirable due to their increased flexibility. Predicted LogP values were 1 = −1.24, 2 = 1.52 and 3 = 1.65. The evaluation also flagged up the quaternary nitrogen, traditionally somewhat unattractive for medicinal chemistry design but, we believe at least anecdotally, to be essential to mimic the quaternary nitrogen of NAADP. Further optimization of these lead compounds may address this and future work will investigate wider SAR issues. 1 H-NMR analysis of these pyridinium compounds showed a characteristic singlet around 5.5-5.6 ppm, representing the CH 2 directly connected to nicotinic acid. It was also found that these pyridinum salts all contained a bromide counter ion as verified by microanalysis ( Supplementary Information, Table 1). This suggested that the carboxylic acid group of nicotinic acid was protonated, rather than ionized, and that the counter ion ensures neutrality of the molecule (Fig. 8). This was investigated crystallographically for 3. Figure 7. (A) Toxicity assay of 2 on non-proliferating myelin-basic protein (MBP) specific, CD4 + rat T cell blasts; (B and C) Protective effect of 2 in transfer experimental autoimmune encephalomyelitis (EAE); Animals (6 per group, body weight approx. 150 g) were injected i.p. twice per day with either PBS (vehicle control), nicotinic acid (50 μmol/100 g body weight), or 2 (50 μmol/100g body weight). Clinical scores indicate the degree of paralysis of tail and legs -as previously reported 42 . Data are presented as mean ± SD from one representative experiment (n = 6) of two independently conducted in vivo studies; (D) Effect of 2 on the localization of encephalitogenic T cells on day 4 post transfer in transfer experimental autoimmune encephalomyelitis (EAE); Animals (4 per group, body weight approx. 150 g) were injected i.p. twice per day with either PBS (vehicle control), nicotinic acid (50 μmol/100 g body weight), or 2 (50 μmol/100 g body weight). Animals were sacrificed on day 4 and the number of T MBP-GFP cells was determined for the organs displayed (n = 4 per group). Data are corrected for organ mass and presented as mean ± SEM from two independent experiments. Non parametric Kruskal-Wallis test shows statistical relevant differences (p < 0.05) in spinal cord migrating T MBP-GFP cells following treatment with 2, as compared to vehicle-treated animals. Crystallography. X-Ray crystallography of a suitable crystal of 3 ( Fig. 8) showed an asymmetric unit formed by linking O2 of the protonated species with O3 of a zwitterionic counterpart via hydrogen bonding (Fig. 9). Only one bromide counterion was observed for this unit, suggesting that 3 could exist as a zwitterion.
To study 3 in more detail, we prepared the pure zwitterionic form (3a (Fig. 8) by removal of the bromine counter ion. Traditional methods, including treatment with Ag(I)O to precipitate AgBr 32 , were not possible as the insolubility of the resulting zwitterion resulted in an inseparable mixture. However, treatment of 3 in methanol with DOWEX ® Monosphere ® 550 A (OH) anion exchange resin, followed by hot filtration to remove the resin afforded small white crystals of 3a. Zwitterion 3a is however disappointingly insoluble in H 2 O, soluble in hot MeOH and only sparingly soluble in DMSO. In contrast to 3, that has a melting point of 195-197 °C, 3a decomposes above 180 °C. Crystallography confirmed the integrity of the zwitterion, with the unit cell containing one molecule of the zwitterion hydrogen-bonded to one molecule of water (Fig. 10A). Hydrogen bonding between the hydrogen of the water molecule and the carboxylic acid, and the oxygen of the water and the amide N-H (Fig. 10B) generates a gross structure that is dominated by H-bonded sheets (Fig. 10C).
On balance, given the more difficult solubility and apparently better membrane permeability characteristics of 3a, 3 seems to be the preferred compound for biological applications, with the optimal balance of solubility and membrane permeability. We have therefore not studied 3a biologically.
The best compound in terms of potency and membrane permeability discovered in this series so far is 3. The crystal structure study revealed that this compound presents with a bromide equivalent, crystallizes as a mixture of the zwitterionic form and free acid form that is balanced with half an equivalent of a bromide counter ion per unit cell. In vivo the expectation would be that it is active in the zwitterionic form, similar to that of the natural alkaloid trigonelline. 3 Has found multiple applications, being used specifically in a model of T-cell mediated autoimmunity to inhibit Ca 2+ mobilization in intact T-cells and to attenuate downstream signalling events with relevance to autoimmune therapy; 31 it blocked antigen-dependent increases of Ca 2+ signals in T cells 44 and induced a transient state of non-responsiveness in post activated effector T-cells, showing the ability to ameliorate significantly in vivo, both prophylactically and therapeutically, the clinical symptoms of EAE 42 . In T-cells 3 was also recently shown by intravital imaging specifically to reduce long-lasting Ca 2+ signalling at spinal cord leptomeninges, confirming an activation checkpoint as a potential therapeutic target 45 . Finally, in a cardiovascular setting 3 blocked isoproterenol-induced diastolic Ca 2+ transients in myocytes and almost completely prevented isoproterenol-induced cardiac arrhythmias in vivo, indicating a pivotal role for NAADP/Ca 2+ signalling in excitation-contraction coupling and uncovering a new target for antiarrhythmic therapy 46 .

Conclusion
In summary, we present for the first time a synthetic route to stable, membrane-permeant NAADP fragment analogues, the best of which are effective inhibitors of NAADP-induced Ca 2+ release in T-lymphocytes. This also provides the wider synthetic context in which to view compound 3 BZ194, data for which have already been published 31 . These can be used as molecular tools to investigate and modulate NAADP/Ca 2+ signalling. Lead compound 2 showed positive and specific effects on Ca 2+ signalling in vitro and in vivo in an EAE model of multiple sclerosis; it showed beneficial effects towards autoimmune disease, suggesting that further modification of such compounds may lead to potential therapeutic agents. Another compound from this series 3 has also been investigated in vitro and in vivo, demonstrating the NAADP/Ca 2+ signalling pathway as a potential novel target in autoimmunity 31,42,44,45 and for ventricular cardiomyocyte arrhythmias 46 . Thus, suitable inhibitors of NAADP/ Ca 2+ signalling might potentially be further developed for clinical use. The fact that these active compounds were developed through evaluation of only a relatively small library of compounds indicate that more optimized lead compounds, perhaps with better potency and more fine-tuned physicochemical properties, can likely be discovered in future work with implications for therapeutic intervention.

Experimental
General. NAADP was supplied by Sigma. cADPR and Ins(1,4,5)P 3 were obtained from Biolog (Bremen, Germany). All Lewis and DA rats were bred in the animal facility of the Max Planck Institute of Neurobiology (Martinsried, Germany) and all experiments were conducted according to the guidelines of The Committee on Animals of the Max Planck Institute of Neurobiology and were approved by the Regierung von Oberbayern. All reagents and solvents were of commercial quality and were used directly unless otherwise described. 1

Microinjection into Jurkat T cells.
Microinjections were carried out as described. Briefly, an Eppendorf system was used (transjector type 5246, micromanipulator type 5171, Eppendorf-Netheler-Hinz, Hamburg, Germany) with Femtotips II as pipettes. NAADP was diluted to its final concentration in intracellular buffer (20 mM HEPES, 110 mM KCl, 2 mM MgCl 2 , 5 mM KH 2 PO 4 , 10 mM NaCl, pH 7.2) and filtered (0.2 μm) before use. Injections were made using the semi-automatic mode of the system with the following instrumental settings: injection pressure 60 hPa, compensatory pressure 30 hPa, injection time 0.3-0.5 s and velocity of the pipette 700 μm/s. Under such conditions the injection volume was 1-1.5% of the cell volume Ca 2+ measurements in intact T cell suspensions. Intact Jurkat T-lymphocytes were loaded with fura2/ AM. Ratiometric determination of [Ca 2+ ] i was carried out in cell suspension in a Hitachi F2000 fluorimeter at room temperature at excitation wavelengths of 340 and 380 nm (alternating) and an emission wavelength of 495 nm. Each experiment was calibrated by addition of Triton X100 (10% v/v final concentration) to obtain the maximal ratio and subsequent addition of 4 mM EGTA/40 mM Tris-base to obtain the minimal ratio.
In vitro proliferation assay (rat encephalitogenic T cells). Encephalitogenic T cells specific for the myelin protein myelin basic protein were established as described previously. The cells were retrovirally engineered to express green fluorescent protein (T MBP-GFP cells) 40 . Resting T MBP-GFP cells were plated in 96-wells (5 × 10 4 /well) and stimulated by the addition of thymocytes (1.5 × 10 6 / well) as antigen-presenting cells and MBP (5 µg/mL) as specific antigen, or ConA (2 µg/mL).

EAE induction and in vivo toxicity assay (Lewis rat).
Nicotinic acid or 2 was injected i.p. to recipient healthy Lewis rats and to EAE-induced rats twice a day for 6 days at the following concentrations: 100 µM (15 µmol substance /150 g body weight) and 500 µM (75 µmol substance /150 g body weight). To induce EAE, transfer of encephalitogenic T cells were performed as follows: 5 × 10 4 activated T MBP-GFP cells were transferred i.v. into recipient Lewis rats on day 0. Clinical EAE was graded in five scores: 0.5, loss of tail tonus; 1, tail paralysis; 2, gait disturbance; 3, hind limb paralysis; 4, tetraparesis; 5, death 31  (I > 2σ(I)) and wR 2 was 0.0820 (all data). The asymmetric unit in the structure was seen to comprise one cation, and one bromide anion (Fig. 9A). The cation presented as a dimer of two carboxylates, linked by a single proton (H2), located between O2 and O3A, which was refined without restraints [O2-H2 1.18 (3) unique (R int = 0.0352, R sigma = 0.0277) which were used in all calculations. The final R 1 was 0.0413 (I > 2σ(I)) and wR 2 was 0.1148 (all data). The asymmetric unit comprises one zwitterion and a molecule of water (Fig. 10A). Fig. 10B shows the hydrogen-bonded interactions and Fig. 10C a portion of the gross structure.

Synthesis of bromoacetylamides (protocol A).
To a stirred solution of 2-bromoacetyl bromide Procedure for alkylation reaction (protocol B). Nicotinic acid (1.62 mmol) and 2-bromoacetylamides (1.62 mmol) were dissolved in dry DMF (4 mL) and the reaction solution was heated at 60-70 °C for 16 h. DMF was evaporated in vacuo and the resulting residue was dissolved in small amount of MeOH. The desired compounds were precipitated by dropwise addition of ether.
2) 5-Bromo-nicotinic acid methyl ester (400 mg, 1.87 mmol) and phenyl boronic acid (452 mg, 3.7 mmol) were suspended in toluene (4 mL). To the suspension was added MeOH (1 mL) and an aqueous solution of Na 2 CO 3 (2 M, 2 mL). The resulting mixture was bubbled with argon for 40 min and Pd(PPh 3 ) 4 (35 mg, 0.03 mmol) was added under argon atmosphere. The reaction mixture was heated at reflux for 4.5 h, then filtered through Celite, diluted with DCM (150 mL), washed with H 2 O (50 mL) and dried over MgSO 4 . The solvent was removed in vacuo and the residue was purified by column chromatography on silica gel (DCM: EtOAc 10:1 v/v) giving 5-phenyl nicotinic acid methyl ester (192 mg, 48%). 1    3-Carbamoyl-1-(2-methyleneoctyl)pyridinium iodide (37). 1) To a solution of 2-hexylacrolein 37,38 (10 mmol) in EtOH (15 mL) was added NaBH 4 (11 mmol). The reaction was stirred at rt for 2 h and was quenched by addition of ice. It was extracted with hexane (2 × 20 mL) and DCM (2 × 20 mL) and the organic layers were combined, dried and evaporated. Purification by column chromatography (DCM-hexane 10:1 → 1:0 v/v) gave 36 as a colourless liquid (47%). NMR data agreed with that reported 37 . To a solution of 36 (300 mg, 2.11 mmol), PPh 3 (820 mg, 3.13 mmol) and imidazole (217 mg, 3.19 mmol) in dry THF (22 mL) was added I 2 (795 mg, 3.13 mmol) in one portion at 0 °C under argon. The resulting mixture was stirred at rt for 2 h and the reaction was quenched by addition of Na 2 S 2 O 3 (10% aq.). The mixture obtained was diluted with ether (100 mL). The organic layer was separated, washed with water and dried over MgSO 4 . The solvent was removed under reduced pressure and the resulting residue was purified by column chromatography (DCM) giving crude iodo-derivative in 86% yield, which was used directly in the next step.

3-Carboxy-1-(3-(octylamino)-3-oxopropyl)pyridinium bromide (34
2) Nicotinic acid (74 mg, 0.60 mmol) was added to a solution of the iodo-compound (165 mg, 0.65 mmol) in DMF (4 mL) and the resulting mixture was stirred at 65 °C for 16 h. DMF was evaporated under reduced pressure and the resulting residue was dissolved in small amount of MeOH. Addition of ether to the solution gave the title compound as a yellow oil (125 mg, 55% based on nicotinic acid). 1

1-(2-(Heptylamino)-2-oxoethyl)-1,4-dihydropyridine-3-carboxylic acid (38). A suspension of 26
(200 mg, 0.56 mmol) and NaHCO 3 (236 mg, 2.81 mmol) in MilliQ water was bubbled under argon in a sonicator for 40 min. Na 2 S 2 O 4 (292 mg, 1.68 mmol) was added under argon and to the suspension MeOH (15 mL) was added. The resulting clear solution was stirred at rt under argon for 1H during which a white precipitate was produced. CHCl 3 (15 mL) was added and the mixture was stirred for a further hour. The organic layer was separated and the aqueous phase was extracted with CHCl 3 (2 × 15 mL). The organic layers were combined, dried over MgSO 4 and the solvent was removed in vacuo, giving the title compound as a yellow solid (120 mg, 77%). The compound was used in the biological assay without further purification. 1