Structure–Activity Relationship of Adenosine 5′-diphosphoribose at the Transient Receptor Potential Melastatin 2 (TRPM2) Channel: Rational Design of Antagonists

Adenosine 5′-diphosphoribose (ADPR) activates TRPM2, a Ca2+, Na+, and K+ permeable cation channel. Activation is induced by ADPR binding to the cytosolic C-terminal NudT9-homology domain. To generate the first structure–activity relationship, systematically modified ADPR analogues were designed, synthesized, and evaluated as antagonists using patch-clamp experiments in HEK293 cells overexpressing human TRPM2. Compounds with a purine C8 substituent show antagonist activity, and an 8-phenyl substitution (8-Ph-ADPR, 5) is very effective. Modification of the terminal ribose results in a weak antagonist, whereas its removal abolishes activity. An antagonist based upon a hybrid structure, 8-phenyl-2′-deoxy-ADPR (86, IC50 = 3 μM), is more potent than 8-Ph-ADPR (5). Initial bioisosteric replacement of the pyrophosphate linkage abolishes activity, but replacement of the pyrophosphate and the terminal ribose by a sulfamate-based group leads to a weak antagonist, a lead to more drug-like analogues. 8-Ph-ADPR (5) inhibits Ca2+ signalling and chemotaxis in human neutrophils, illustrating the potential for pharmacological intervention at TRPM2.


■ INTRODUCTION
Transient receptor potential (TRP) channels are six-transmembrane polypeptide subunits that assemble as tetramers to form cation-permeable pores. 1 TRP subfamily melastatin, type 2 (TRPM2), is a Ca 2+ permeant channel which is also permeant to Na + , K + , and Cs + ions. 2 TRPM2 is unique among the known ion channels as it contains a C-terminal domain which is homologous to NUDT9 ADPR-hydrolase, and this has led to considerable interest. The NUDT9-homology (NUDT9H) domain of human TRPM2 extends from residue 1236 to the Cterminus. NUDT9 was identified in an EST database screen for proteins with homology to the C-terminus of TRPM2. It is an enzyme of the Nudix family of pyrophosphatases, with adenosine 5′-diphosphoribose (ADPR, 1, Figure 1) as sole substrate. A NUDT9 crystal structure illustrated that this is a two-domain enzyme with a C-terminal ADPRase and Nterminal domain which enhances affinity for ADPR. 3 Sequence alignment suggests that NUDT9H has the same features in TRPM2 channels and revealed that TRPM2 is gated by binding of ADPR, 2,4 controlling cation entry through the plasmamembrane channel. Interestingly, TRPM2 is not a very effective hydrolase. 5 The TRPM2 channel is gated by micromolar cytosolic concentrations of ADPR (EC 50 ∼ 100 μM). In contrast to other nicotinamide adenosine 5′-dinucleotide (NAD + ) metabolites, such as cyclic adenosine 5′-diphosphoribose (cADPR, 3, Figure 1) and nicotinic acid adenine dinucleotide phosphate (NAADP), free ADPR has only recently been considered a second messenger. 6 Cellular ADPR derives from NAD + glycohydrolases such as CD38 that predominantly produce ADPR but also hydrolyze cADPR to ADPR ( Figure 1) and other NAD + metabolizing processes, including activation of poly-ADPribose polymerase (PARP) and poly-ADPribose glycohydrolase (PARG) pathways. 7,8 TRPM2 is expressed in a variety of tissues, with highest transcript levels being detected in the brain, bone marrow, and cells of the immune system. 9 Because the free cytosolic Ca 2+ concentration ([Ca 2+ ] i ) sensitizes TRPM2 for activation by ADPR, 10 resulting in a positive feedback loop and massive Ca 2+ entry, it has been related to cell death from early on. 11−13 Its activation downstream of reactive oxygen species suggested that it might be involved in the pathogenesis of neurodegenerative diseases, 14 myocardial infarction, 15 and type I and type II diabetes. 16,17 More recent data point to an important physiological role for TRPM2 in cells of the innate immune system (reviewed in ref 18). In macrophages/monocytes, TRPM2 is required for the production of the pro-inflammatory chemokine CXCL2 in response to reactive oxygen species 19 and in the production of different cytokines in response to lipopolysaccharides. 20 In neutrophils, TRPM2 is involved in Ca 2+ signalling 21,22 and the chemotactic response to fMLP. 21 TRPM2 has also been shown to be involved in dendritic cell maturation and their chemotaxis to chemokines. 23 Because TRPM2 is upregulated in Tlymphocytes after activation and appears to play a role in cytokine secretion and proliferation, 24 it might also be important for the adaptive immune response.
In pancreatic β-cells, TRPM2 has been shown to be involved in modulation of insulin secretion. 25,26 In neurons, it is much less clear whether TRPM2 also performs a physiological function besides its potential role in the pathophysiology of neurodegeneration. However, there have been initial reports of a contribution of TRPM2 in synaptic transmission. 27,28 Research in these areas and identification of additional physiological processes involving ADPR/TRPM2 critically depend on the availability of specific small molecule tools. Over the years, some compounds that inhibit the TRPM2 channel have been described. The first TRPM2 inhibitor was flufenamic acid (FFA, IC 50 70 μM), 29 belonging to the fenamate group of nonsteroidal anti-inflammatory drugs. Other fenamates like mefenamic acid and niflumic acid also inhibit the channel although with lower potency (IC 50 124 and 149 μM, respectively). A fenamate with reduced nonspecific effects, 2-(3-methylphenyl)aminobenzoic acid (3-MFA, IC 50 76 μM), has recently been developed by Chen et al. 30 Other TRPM2 inhibitors include the PLA2 inhibitor N-(pamylcinnamoyl)anthranilic acid (ACA, IC 50 1.7 μM) 31 that shares the anthranilic acid motif with the fenamates, the antifungal imidazoles miconazole, 25 econazole, and clotrimazole (IC 50 < 3 μM for both), 32 2-APB (IC 50 ∼ 1 μM), 25 and copper ions (IC 50 2.6 μM). 33 All of these pharmacological compounds apparently act as channel blockers and inhibit a variety of other ion channels (see ref 30 for citations). ACA affects other signal transduction pathways, thereby severely limiting its usefulness.
In 2007, some of us developed 8-Br-ADPR (4, Figure 1) as the first specific NUDT9H-TRPM2 antagonist. 21 Both Ca 2+ signalling and chemotaxis of murine neutrophils were very sensitive to 4. 21 The introduction of an 8-bromo substituent turned the natural agonist ADPR into an antagonist, highlighting the sensitivity of ADPR to small structural modifications.
To better understand the ADPR/NUDT9H-TRPM2 system, it was necessary to develop further this early lead and begin to build a structure−activity relationship. Here, for the first time, we report a chemo-enzymatic approach involving chemical synthesis, coupled with use of Neurospora crassa NADase, to interrogate each major motif of the ADPR structure and

■ RESULTS AND DISCUSSION
To explore the SAR of ADPR at TRPM2, we undertook a systematic approach, splitting ADPR into four major motifs: adenine, the adenosine ribose, the pyrophosphate, and the terminal ribose ( Figure 1). We synthesized analogues with modifications in each of the four motifs to evaluate the resultant effect on antagonist activity and provide a comprehensive early SAR.
Adenine Modified Analogues: 8-Modified. The introduction of substituents at the 8-position of a nucleoside can directly affect the syn/anti orientation of the base residue. Following on from the discovery that 8-Br-ADPR (4) is an antagonist, 21 we wanted to further explore the effect of other substituents in this area. We prepared three analogues with hydrophobic substitutions by Suzuki coupling (compounds 5− 7, Scheme 1) and one more polar substitution in the form of 8-NH 2 -ADPR (12, Scheme 3).
8-Br-ADPR (4) was prepared by treatment of commercially available NAD + with NADase followed by bromination with Br 2 in NaOAc buffer (pH 4). The Suzuki reaction of 4 with phenylboronic acid was not straightforward. The conditions that had been successfully applied to 8-Br-cIDPR 34,35 did not result in any reaction with 4; possibly the less rigid phosphate conformation was able to bind with the palladium source, stalling the reaction. Re-examining the literature identified that Pd(OAc) 2 as the palladium source, and the water-soluble triphenylphosphine TPPTS had been applied to similar compounds with success. 36−38 These conditions were tested on a model system, the monophosphate 8-bromo-AMP 8, and the reaction with phenylboronic acid proceeded well. Unfortunately, this was not directly transferable to 4. As the reaction proceeded, a competing side reaction caused hydrolysis of the pyrophosphate, generating the 8-substituted AMP. Therefore, the reactions were carefully followed by HPLC, after which purification of the product by ion-exchange and reverse phase chromatography removed the monophosphate impurity and generated 8-phenyl-ADPR (5), 8thiophenyl-ADPR (6), and 8-(3-acetylphenyl)-ADPR (7). Disappointingly, the reaction was not general for all boronic acids; only a fraction of those tried resulted in product, and those that failed just resulted in conversion of 4 to 8. To confirm that the lack of reactivity observed with the boronic acids was due to 4, the reaction with 4-phenylalanine (4-Ph-ala) boronic acid was repeated under the same conditions using 8 as reactant (Scheme 2). Unlike the reaction with 4 that generated no product, conversion was complete in 5 min and 8-(4-Ph-ala)-AMP 9 was isolated by combined ion-exchange and reverse phase chromatography.
Adenine Modified Analogues: Base-Modified. To probe the other interactions between the adenine base and receptor, we also prepared analogues modified at C6 [IDPR (13) and 6-O-Me-IDPR (24)], N7 [7-deaza-8-Br-ADPR (15) and 7-deaza-IDPR (14)], and C2 [2-F-ADPR (16)]. Such modifications are challenging to introduce, as they usually require specialized preparation of a customized base that must then be coupled to the adenosine ribose and elaborated in a stepwise manner. Therefore, in the case of analogues previously known to the group, or in the literature, the NAD + analogue was hydrolyzed enzymatically to generate the linear ADPR compound; NADase hydrolysis of NHD + gave 13, 7-deaza-8-Br-NAD +39 gave 15, and 2-F-NAD +40 gave 16 ( Figure 2). 7-Deaza-IDPR (14) was prepared by treatment of 7-deaza-NHD + with ADPRC because in this instance the lack of N7 for cyclization results in the linear analogue on treatment with cyclase rather than a cyclic product. 41 6-O-Methyl-IDPR (6-O-Me-IDPR, 24) was prepared from 6chloropurine (17) and tetraacetyl-D-ribose (18) that were coupled together at N9 using Vorbruggen conditions, 42 followed by simultaneous deprotection of the three hydroxyls and conversion of the 6-chloro to a 6-methoxy substituent using sodium methoxide. Phosphorylation of the 5′-OH to generate 6-O-Me-AMP (21) was followed by activation of the phosphate, subsequent coupling to β-NMN + , and hydrolysis of the 6-O-Me-NHD + (23) using NADase to generate the desired analogue 24 (Scheme 4).
Adenosine Ribose Modified Analogues. Three modifications were prepared at the adenosine ribose: individual deletion of either the 2′-or 3′-hydroxyl group or complete removal of the ribose by replacing it with a more flexible and hydrophobic butyl chain. An acyclic analogue (acyclic-ADPR, 30) was prepared by introduction of an n-butyl chain at N9, via alkylation of 17 with chlorobutylacetate (25). 43 Subsequent treatment with MeOH/NH 3 simultaneously removed the acetate protecting group and substituted the 6-chloro substituent with a 6-amino group. 44 Nucleoside analogue 27 was then phosphorylated under the standard POCl 3 /TEP conditions. 41 Activation of the phosphate, followed by coupling with β-NMN + and MnCl 2 in formamide, delivered the desired linear acyclic-NAD + 29. Treatment with NADase hydrolyzed the nicotinamide to afford 30 in very good yield (Scheme 5).
The synthesis of 3′-deoxy-ADPR (43) was more challenging as 3′-deoxy-AMP (41) is not commercially available. Using a published synthesis, 46 we prepared 3′-deoxyadenosine (37). However, in this instance, our usual phosphorylation conditions (POCl 3 , TEP) were not selective for the primary hydroxyl group, and a mixture of products was obtained. We therefore had to adopt a longer route, in which both hydroxyl groups were protected as the TBDMS ether, followed by selective cleavage of the primary ether and phosphitylation with a protected phosphoramidite, followed by oxidation under basic 3′-Deoxy-AMP (41) was activated and coupled to β-NMN + to generate 3′-deoxy-NAD + (42), and the nicotinamide cleaved with NADase to generate 43.
Pyrophosphate-Modified Analogues. The diphosphate linkage is in most cases essential for biological activity. We prepared an analogue with increased negative charge, length, and flexibility in the form of three phosphate groups (ATPR,     47). ATPR (47) was prepared by degradation of cyclic adenosine triphosphoribose (cATPR, 47 46, Scheme 8). The cyclic precursor 46 was synthesized in two steps from the coupling of ADP (44) and β-NMN + with EDC to afford the linear nicotinamide adenine trinucleotide (NAT, 45) that was cyclized by incubation with Aplysia cyclase as previously reported 47 (Scheme 8).
In contrast, we developed analogues in which the pyrophosphate was replaced entirely by a phosphate or pyrophosphate bioisostere. The limited bioavailability of pyrophosphate-bearing compounds, their instability in a physiological environment, low membrane permeability, and difficulties in synthesis and scale up have led to considerable effort toward the development of isosteres. These include phosphate-containing, sulfur-containing, and carboxylate linkages as well as more unique cyclic mimics. 48 However, no generally applicable bioisostere has so far been identified and with so many to choose from, our choices were based on ease of synthesis in the first instance. We therefore chose to utilize squarate, triazole, or sulfamate groups.
A small library of squarate compounds was prepared by coupling protected 5′-aminoadenosine 59 with a suitable squarate (Scheme 10). The analogue in which only the pyrophosphate is altered required 5′-aminoribofuranose (52) that was synthesized according to Scheme 9: One-step protection of the 1-, 2-, and 3-OH under acidic conditions generated 49. 49 The 5-OH was then converted to the tosylate (50) that was displaced using sodium azide to generate 51 and the product reduced to the desired amine 52.
Four amines were coupled to diethylsquarate 53 in the presence of DIPEA and EtOH; 52 gave 54, cyclopentylamine gave 55, butylamine gave 56, and hexylamine gave 57. The squarate compounds 54−57 were then coupled to protected 5′aminoadenosine 59 that was synthesized in two steps from 2′,3′-O-isopropylideneadenosine (58) via the 5′-azido intermediate to generate 60−63. The isopropylidene protecting group was then cleaved under standard acidic conditions, except in the case of the ribosyl compound 60 for which heating to 60°C for 16 h was required to remove the 1-Omethyl protecting group, to furnish ADPR analogues 64−67 (Scheme 10). An analogue was prepared in which the pyrophosphate was replaced by a 1,2,3-triazole. Suitable precursors were prepared by generating an azide at the 5′ of adenosine and introducing a propargyl ether on the terminal ribofuranose for copper(I)catalyzed Huisgen 1,3-dipolar cycloaddition ("click" cyclization). 2′,3′-O-Isopropylidene adenosine (58) was treated with diphenylphosphoryl azide followed by sodium azide to generate 5′-azidoadenosine 68 that was then brominated in the 8position. The propargyl counterpart was prepared by deprotonating 49 with sodium hydride, followed by addition of propargyl chloride to prepare the protected 5-O-propargyl ether 70. The two precursors 69 and 70 underwent a 1,3dipolar cycloaddition promoted by Cu(I), generated in situ from copper sulfate and sodium ascorbate, to afford 71. Triazole 71 underwent Suzuki coupling with phenyl boronic acid at the preinstalled 8-bromo substituent to afford 72. This was deprotected under acidic conditions to generate 8phenyladenosine-1,4-triazole ribose (8-Ph-ATrR, 73, Scheme 11). Despite multiple attempts, it was not possible to introduce the propargyl ether on the 5′-position of adenosine as the nucleoside was degraded during deprotonation with sodium hydride.
The third pyrophosphate replacement explored was a sulfonamide. The sulfonamide was introduced to the 5′hydroxyl group of a protected adenosine by treatment of 58 with sulfamoyl chloride. However, introduction of a terminal ribose proved difficult, both in the preparation of a suitable amide or carboxylic acid and in the condensation reaction. Condensation of 74 with salicylic acid afforded an opportunity to introduce a terminal ribose surrogate and generate a known phosphate bioisostere. The protected sulfonamide analogue 75 was then deprotected under acidic conditions to afford salicyladenosine monosulfamide (Sal-AMS, 76, Scheme 12).
Terminal Ribose-Modified Analogues. A cyclopentyl group was introduced in place of the terminal ribose to assess its importance in binding to TRPM2. This has none of the polar hydroxyl groups but a similar overall shape. Analogues modified at the terminal ribose have the advantage that they do not require coupling using expensive β-NMN + . Cyclopentanol (77) was phosphorylated and deprotected to give the corresponding free-phosphate 79 that was coupled to AMPmorpholidate to afford cyclopentyl-ADP (80, Scheme 13).
Combined Modifications. After initial positive results, we designed analogues combining substitutions that appeared to improve antagonist properties. Cyclopentyl-8-Ph-ADP (84, Scheme 14) was prepared by introduction of an 8-bromo substituent to 58, followed by Suzuki cross coupling with phenyl boronic acid to introduce an 8-phenyl substituent in 82. The free 5′-OH was phosphorylated to afford 8-Ph-AMP (83) that was activated and coupled to 79 to afford 84 (Scheme 14).

■ PHARMACOLOGY
To evaluate ADPR analogues for antagonist activity, HEK293 clones with stable expression of human TRPM2 were established. Clones derived from HEK293 cells transfected with an expression vector for human TRPM2 were selected according to their increase in [Ca 2+ ] i in response to hydrogen peroxide ( Figure 3A). Further, expression of TRPM2 at the protein level was confirmed in responsive clones by Western blot analysis ( Figure 3B). The specificity of the antibody was checked by preincubation with a blocking peptide ( Figure 3B, right). For different clones, maximum [Ca 2+ ] i response to hydrogen peroxide increased with magnitude of TRPM2 expression ( Figure 3).
Whole-cell patch-clamp experiments were performed to evaluate ADPR analogues using clone #24 showing intermediate expression of TRPM2. To avoid excessive inward currents, the sodium chloride within the bath solution for the patch clamp experiments was replaced by N-methyl-D-glucamine (NMDG) hydrochloride that does not pass through the TRPM2 channel. Under these conditions, activation of TRPM2, after break-in with 100 μM ADPR (1) in the pipet solution, resulted in a small but measurable inward current of Ca 2+ and a more pronounced outward current mainly carried by K + ions ( Figure 4A, gray line) similar to previous observations. 50 In contrast, no such currents were observed when a control cell line stably expressing EGFP (EGFP#8, Figure 4A, dashed gray line) was used instead of TRPM2#24 or the pipet was filled with a solution devoid of 1 ( Figure 4A, dashed black line). As a read-out for TRPM2 activation the maximum net outward current at +15 mV obtained from repetitive voltage ramps between −85 and +20 mV was chosen.
The first specific ADPR antagonist 8-Br-ADPR (4) 21 significantly inhibited TRPM2 activation when applied at a 9fold excess over ADPR (1). Thus, alongside the positive control (100 μM of 1) in most experiments, a combination of 100 μM 1 and 900 μM 4 was included as an inhibitor control ( Figure  5). Most ADPR analogues were tested for antagonism to 100  , EGFP#8, and wild-type HEK293 cells was determined by Western blot analysis. Specificity of TRPM2 detection was tested by preincubation of the primary antibody with a corresponding blocking peptide (right part). Specific bands of full length TRPM2 (black arrow) and degradation products (gray arrows) are marked. None of these bands were seen in wild-type HEK293 cells, in EGFP#8 control cell line, or in the presence of blocking peptide. contrast in activity may be due to the alteration of the purine base conformation from anti to syn upon substitution with large groups in the 8-position. 41 In contrast to an 8-bromo or 8amino substituent, the 8-aryl group introduces a new conformational parameter, as not only will such groups tend to a predominantly syn configuration around the glycosidic bond, but they also introduce the possibility of torsional conformational effects of, as yet, unknown significance.
Adenosine Ribose Modifications. The adenosine ribose was replaced by an alkane chain, resulting in the weak antagonist acyclic-ADPR 30 that showed nearly complete inhibition at 900 μM, whereas at 100 μM, no inhibition of the ADPR induced current was observed, indicating a steep dose response relationship with an IC 50 value in the upper micromolar range ( Figure 6). It was not possible to measure the antagonist activity of 2′-deoxy-or 3′-deoxy-ADPR (33,34). In contrast, 2′-deoxy-AMP (31) was active as a weak antagonist, whereas AMP 88 showed no antagonist activity in    Figure 7A). There are conflicting reports on the antagonistic effect of 88; while it was previously described to inhibit TRPM2 activation completely with an IC 50 value of 70 μM in whole cell patch clamp experiments, 51 other studies have not shown any inhibition of ADPR-mediated currents in cell free patches from Xenopus oocytes expressing TRPM2, even at 60-fold excess of 88 over 1. 52 Interestingly, the combination of this 2′-deoxy-modification and the 8-phenyl substitution at C8 in 8-Ph-2′-deoxy-ADPR 86 also completely inhibited activation of TRPM2 when applied at a concentration of 900 μM. Full concentration−response relationships for 4 and 86 are displayed in Figure 10. The adenosine ribose, particularly the C2′ position, seems to be important for binding of analogues to NUDT9H; possibly the absence of the hydroxyl group results in lower steric hindrance or lower repulsion between ligand and channel ultimately resulting in stronger binding.
Pyrophosphate Modifications. Introduction of an additional phosphate in ATPR (47) generated an analogue that appears to have some antagonist effect at 900 μM, although this was not statistically significant ( Figure 7a). It remains unclear whether this is a result of increased length or the increased charge, the latter unattractive from a drug discovery point of view. Squarate analogues 64−67 and triazole analogue 73 could only be tested at 100 μM because the usual method of solubilizing the compounds in HEPES buffer was not successful, so DMSO was used. However, only as much as 0.1% (v/v) DMSO was tolerated in the pipet solution, allowing a maximum concentration of 100 μM to be reached. However, at this concentration, none of these compounds showed antagonist activity ( Figure 7B). The acyl sulfamate analogue 76 was soluble in aqueous media and showed noticeable antagonist activity at 900 μM. Although this was not statistically significant, this analogue (in particular, because it combines both a neutral pyrophosphate bioisostere and simple substitution of the terminal ribose) represents a tractable lead toward future, more drug-like ADPR antagonists. Interestingly, Sal-AMS has previously been prepared as an antibiotic lead for the inhibition of siderophore biosynthesis in tuberculosis and plague. 53,54 However, in ADPR, the pyrophosphate moiety is more than simply a "spacer" between the two riboses; to date, no analogues with a pyrophosphate substitution have shown a significant antagonist effect. These results indicate that the negative charge at the pyrophosphate is important for interaction with the NUDT9H domain. This finding is of great significance for all applications of such an antagonist in cell and animal models because the pyrophosphate moiety potentially may be attacked by pyrophosphatase enzyme activity. Thus, analogues containing the pyrophosphate moiety may suffer from a short half-life. Metabolically stable phosphonate-type analogues or alternative pyrophosphate replacement groups might be a future solution for this problem. 55 Terminal Ribose Modifications. A number of analogues were prepared or are commercially available, including completely truncated "AMP-like" compounds. The latter are attractive analogues due to the presence of only one phosphate group; they are therefore likely to be more membrane permeant and are more easily synthetically accessible as they do not require the difficult coupling to expensive reagent β-NMN. In contrast to AMP (88) that was ineffective, 2′-deoxy-AMP (31) and 8-(4-phenylalanine)-AMP (9) suppressed activation of TRPM2 by 1 when applied at 900 μM ( Figure  8). However, when equimolar to 1 (100 μM each), the antagonist effect was lost, indicating a steep concentration− response relationship with an IC 50 value considerably higher than 100 μM. Interestingly, 8-Cl-AMP (87) showed no antagonist effect at 900 μM ( Figure 8). The observed variation among AMP analogues suggests that 8-substitution plays a role in introducing antagonist properties, but these are much reduced in magnitude compared to the full-length analogues.
Replacing only the terminal ribose, while maintaining the complete pyrophosphate, generated the antagonist cyclopentyl-ADP (80) that showed partial inhibition at 900 μM (Figure 9). This suggested that the hydroxyl groups on the terminal ribose may not be critical for antagonist activity. In contrast, the parent ADP (44) was inactive, suggesting that the fivemembered ring does play a critical role in filling the binding site. Combining the terminal cyclopentyl substituent with the 8phenyl substituent afforded cyclopentyl-8-phenyl-ADPR (84), which showed significant antagonist activity when applied in an  Figure 9). Replacing the terminal ribose with a cyclopentyl group is attractive because, as well as removing any complication of both anomeric center stereochemistry and equilibrium with a ringopened ribose form, it negates the possibility of intramolecular attack on the pyrophosphate by a ribose hydroxyl group, which may lead to instability. Thus, analogue 84 was analyzed in more detail (see Figure 10).
The five most potent inhibitors are all substituted with an aromatic ring at the 8-position of adenine. Comparison of the three 8-aromatic analogues 5−7 suggest that appended groups (e.g., 3-acetylphenyl-) or alternative ring sizes (e.g., thiophene) do not improve antagonist effects, although this has not been fully exploited due to the difficulties encountered with Suzuki chemistry to prepare these analogues. The small improvement in activity observed for cyclopentyl-8-phenyl-ADP (84) suggests that the terminal ribose is not critical.
Importantly, we also studied the biological activity of one of the best antagonist compounds with a low IC 50 value, in neutrophil Ca 2+ signalling and chemotaxis; a system where we have already established the physiologically important role of TRPM2. 21 8-Phenyl-ADPR (5) was accessed via 4, as opposed to 8-phenyl-2′-deoxy-ADPR (86), which was prepared by degradation of its valuable cyclic counterpart. Thus, we assessed the effect of 5 on Ca 2+ signalling and chemotaxis in human neutrophils stimulated with the chemotactic peptides A5 or fMLP (Figures 11 and 12).
To investigate the effect of NUDT9H-TRPM2 antagonism on Ca 2+ signalling, freshly isolated primary human neutrophils from healthy donors and loaded with Fura2 were stimulated either by 1 μM fMLP (high affinity ligand for human FPR1) or 10 μM of the FPR2 specific A5 peptide. 56,57 When the   neutrophils were preincubated with 100 μM of 5 before application of the chemoattractant, the increase in [Ca 2+ ] i upon stimulation was significantly reduced for both ligands (31.8% for fMLP and 31.4% for A5 peptide, Figure 11), indicating that like in mouse neutrophils, 21 TRPM2 is activated by 1 and contributes to the Ca 2+ signal by mediating Ca 2+ entry downstream of chemoattractant receptors.
To analyze the effect of 8-phenyl-ADPR (5) on chemotaxis, we analyzed the migratory pattern of freshly isolated neutrophils from healthy donors. In the absence of a chemoattractant, the cells showed spontaneous migration that was largely nonvectorial as the center of gravity of all the cells changed only slightly ( Figure 12A). This basal migration was nearly abolished when extracellular Ca 2+ was removed by EGTA ( Figure 12B), demonstrating the requirement for millimolar Ca 2+ in the extracellular space. Upon a fMLP gradient, the cells moved toward the higher attractant concentration ( Figure 12C). However, if extracellular Ca 2+ was removed while the fMLP gradient was applied, directed migration was reduced but was still higher as compared to controls (compare parts D to C and A of Figure 12), indicating that fMLP induced chemotaxis partially depends on a high extracellular Ca 2+ concentration. Importantly, when the neutrophils were preincubated with 100 μM of 5 and then placed within the fMLP gradient, the decrease in directed migration was similar to migration in the experiment with EGTA (compare parts E to D of Figure 12). Taken together, these data demonstrate the importance of ADPR mediated TRPM2 activation in directing the chemotactic response of neutrophils.
8-Phenyl-ADPR (5) partially reduced Ca 2+ signalling by the FPR1 ligand fMLP and the FPR2 specific agonist A5 peptide in human neutrophils and also partially inhibited the directional migration in a gradient of fMLP. These results demonstrate that the novel ADPR analogue 5 not only inhibits activation of TRPM2 directly on target in the "cell free" whole-cell patch clamp system but also affects TRPM2-dependent processes in a cellular environment, making it a useful tool for the investigation of ADPR/TRPM2 dependent processes in other cell systems and physiological processes.
To date, to the best of our knowledge, 8-Br-ADPR (4) has been the only specific antagonist available for activation of TRPM2 by ADPR (1). 21 Therefore, for the first time, our systematic study of ADPR antagonism allows us to build up an early SAR profile for the inhibition of TRPM2 by ADPR analogues (Figure 13 and Table1) and provides more potent agents for the study of TRPM2.

■ CONCLUSION
In this study, we systematically derivatized the new second messenger ADPR and monitored the antagonist activity of the derivatives at TRPM2 during activation by ADPR. The replacement of bromine by the bulky, flat, and torsionally mobile, hydrophobic phenyl substitutent at C8 of adenine together with a 2′-deoxy motif at the adenosine ribose converted the poor, but specific, antagonist 8-Br-ADPR (4) into the highly active specific NUDT9H-TRPM2 antagonist 8-Ph-2′-deoxy-ADPR (86). We have established that the amino/ imino groups at C6 and the N7 nitrogen within the adenine base are important for antagonist activity. Likewise, the charges of the pyrophosphate bridge are critical for retaining antagonist activity because the bridge cannot be replaced by another noncharged linker with similar dimensions. However, substitution of the pyrophosphate by a sulfamate in Sal-AMS did lead to some antagonist activity and this compound is a promising lead for further development of neutral and more drug-like antagonists. Finally, we have also established that the terminal ribose is important but can be partially replaced by a cyclopentyl moiety. This SAR provides a key starting point for further development of potent modulators of TRPM2 and development of chemical biology tools to further probe this new signalling pathway.      Table 1. Summary of Analogues and Antagonist Activity at TRPM2 f a See ref 21. b No significant inhibition at 100 μM; compounds were not evaluated further. c Antagonistic effect observed but not statistically significant; see relevant figure. d Could only be tested at 100 μM due to solubility difficulties. e Partial inhibition was observed at 900 μM; not evaluated further. f n.d. = not determined. n.s. = no significant antagonist effect observed. The resulting clear brown solution was stirred for 2 h at 60°C, after which it was cooled to room temperature and aq satd NaHCO 3 (400 mL) was added. The aqueous phase was extracted with DCM (3 × 300 mL), dried (Na 2 SO 4 ), filtered, and evaporated under reduced pressure, giving a brown oil. The crude was purified by column chromatography on silica gel (DCM−acetone, 9:1 v/v) to afford the desired product as a white foam (4.9 g, 91%). 1  . 6-O-Methylinosine 20 (80 mg, 0.264 mmol) was dissolved in triethylphosphate (1 mL) by heating with a heatgun. The resulting colorless solution was cooled to 0°C, and water (2 μL) was added followed by POCl 3 (0.1 mL, 1.056 mmol). It was stirred at 0°C until disappearance of starting material and formation of a single peak as shown by HPLC. After 1 h, the reaction mixture was quenched by addition of ice/water (15 mL) and stirred for 15 min at 0°C, after which it was warmed up to rt. Triethylphosphate was extracted with EtOAc (6 × 6 mL), and the aqueous phase was neutralized with 2 M NaOH. It was then applied to a reverse phase gradifrac column eluted with a gradient of 5−65% MeCN in 0.05 M TEAB. The appropriate fractions were collected and lyophilized overnight to afford the desired monophosphate as its triethylammonium salt. 1  Synthesis of Acyclic-ADPR 30. 9-(4-Hydroxybutyl)adenine-5′monophosphate 28. 9-(4-Hydroxybutyl)adenine 27 (80 mg, 0.386 mmol) was dissolved in trimethylphosphate (1.3 mL) by heating with a heatgun. Phosphorus oxychloride (144 μL, 1.545 mmol) and water (2 μL) were added at 0°C, and the resulting solution was stirred at rt for 3 h. Ice/water (15 mL) was then added, and the mixture was stirred for further 15 min, after which it was extracted with EtOAc (×6). The aqueous layer was neutralized with 5 N NaOH and applied to a reverse phase column and the product eluted with a gradient of 0.05 M TEAB against MeCN. The appropriate fractions were combined and evaporated. The residue obtained was coevaporated with MeOH to remove excess TEA salt, leaving the desired monophosphate as its triethylammonium salt (92 mg, 72%). 1  Nicotinamide-9-(4-hydroxybutyl)adenine-5′-dinucleotide 29. 9-(4-Acetoxybutyl)adenine-5′-monophosphate·1TEA 28 (92 mg, 0.277 mmol) was dissolved in dry DMSO (1 mL) and coevaporated with dry DMF (5 × 3 mL). The residue was dissolved in DMSO (400 μL) to which was added morpholine (106 μL, 1.233 mmol), dipyridyldisulfide (130 mg, 0.592 mmol), and triphenylphosphine (155 mg, 0.592 mmol), at which point the solution became bright yellow. It was stirred for 2 h at rt, after which HPLC analysis showed completion of the reaction. Precipitation of the product occurred by dropwise addition of a solution of NaI in acetone (0.1 M, 20 mL). The resulting precipitate was filtered, washed with acetone, and dried ( 31 P: δ = 6.7 ppm). It was then reacted with β-NMN + (84 mg, 0.250 mmol) and MgSO 4 (53 mg, 0.454 mmol) in a 0.2 M solution of MnCl 2 in formamide (1.5 mL) at rt overnigh,t after which HPLC analysis showed completion of the reaction (R t = 2.9 min). Precipitation occurred by dropwise addition of MeCN. The precipitate was filtered, dissolved in Milli-Q, and applied to a reverse phase column eluted with a 5−65% gradient of MeCN in 0.05 M TEAB. Further treatment with Chelex 100 to remove any paramagnetic particles afforded the desired dinucleotide as its sodium salt.

Journal of Medicinal Chemistry
9-(4-Hydroxybutyl)adenine-5′-diphosphoribose (Acyclic-ADPR, 30). Nicotinamide-9-(4-acetoxybutyl)adenine-5′-dinucleotide 29 (10 μmol) was treated with NADase under general procedure B to leave the desired acyclic-ADPR (8.1 μmol, 81%). 1  989 mmol) in dry pyridine (1 mL), externally cooled with ice, was added p-toluenesulfonyl chloride (0.7 g, 3.668 mmol) and a catalytic amount of DMAP. The reaction mixture was stirred at rt under nitrogen for 5 h. Water (0.3 mL) was added and stirring continued for 30 min. This mixture was extracted with chloroform (3 × 10 mL) and the combined organic phases washed sequentially with CuSO 4 (10% w/v, aq satd), NaHCO 3 (aq satd) and water and then dried over anhydrous sodium sulfate. The solvent was evaporated, and the residue was purified on an Isco chromatographic system (petrol−EtOAc, 7:3 v/v) to yield the desired compound as a white solid (0.92 g, 81%). 1  Ratiometric Ca 2+ Imaging of Human Neutrophils. Neutrophils were incubated with 4 μM Fura2/AM for 30 min at 37°C in the dark, washed twice, and resuspended in Ca 2+ measurement buffer (see above) at a concentration of 1 × 10 6 cells/mL. For each measurement, 5 × 10 4 cells were transferred to a small chamber consisting of a rubber O-ring fixed with silicon grease on a glass coverslip coated subsequently with 25 ng of BSA and 250 ng of fibronectin. The cells were incubated for 15 min at ambient temperature in the presence of 10 mM L-Ascorbic acid (pH 7.4) and, if applicable, varying concentrations of 8-phenyl-ADPR. The loaded coverslip was mounted on the stage of a PerkinElmer/Imrovision imaging system built around a Leica DM IRE 2 fluorescence microscope. Approximately 70 s after the beginning of the measurement, the cells were stimulated by addition of fMLP (final concentration 1 μM) or A5 peptide (final concentration 10 μM). Ratiometric Ca 2+ imaging was performed as described previously. 62 Chemotaxis Measurement of Human Neutrophils. The migration of neutrophils was observed microscopically in microfluidic devices (μ-Slide Chemotaxis, Ibidi, Martinsried, Germany). First the μ-Slides were coated with 50 μg/mL fibronectin for 30 min at ambient temperature before washing three times and drying. Isolated neutrophils were resuspended to a concentration of 3 × 10 6 cells/ mL in Ca 2+ measurement buffer supplemented with 10% (v/v) plasma obtained from the same donor. If applicable, 8-phenyl-ADPR or EGTA was added and the whole slide was loaded according to the manufacturer's instructions and incubated at room temperature for 15 min. Adding 18 μL of fMLP (125 nM) to the upper reservoir resulted in a chemotactical gradient from 0 → 50 nM fMLP across the observation chamber. Plasma supplemented buffer without chemoattractant was used in control experiments. The slide was mounted on the stage of the imaging system, and the main chamber observed at 10 times magnification in bright-field mode. After a 5 min resting period, greyscale images with a resolution of 672 × 510 pixels were recorded every 30 s for 1 h using Openlab Software 4.0.4. Cell migration was tracked manually with a 5 × 5 pixel maximum intensity centering correction using the manual-tracking plugin for ImageJ (1.45e). Migrational parameters were calculated from the movement paths using the Chemotaxis and Migration Tool software (v2.0, Ibidi GmbH).

* S Supporting Information
General experimental information, experimental procedures for known compounds, and 1 H NMR spectra for compounds 5, 6, 7, 84, and 86. This material is available free of charge via the Internet at http://pubs.acs.org.

Author Contributions
The manuscript was written by contributions from all authors and all authors have approved the final version. C.M and T.K. contributed equally.