Small Molecule CD38 Inhibitors: Synthesis of 8-Amino-N1-inosine 5′-monophosphate, Analogues and Early Structure-Activity Relationship

Although a monoclonal antibody targeting the multifunctional ectoenzyme CD38 is an FDA-approved drug, few small molecule inhibitors exist for this enzyme that catalyzes inter alia the formation and metabolism of the N1-ribosylated, Ca2+-mobilizing, second messenger cyclic adenosine 5′-diphosphoribose (cADPR). N1-Inosine 5′-monophosphate (N1-IMP) is a fragment directly related to cADPR. 8-Substituted-N1-IMP derivatives, prepared by degradation of cyclic parent compounds, inhibit CD38-mediated cADPR hydrolysis more efficiently than related cyclic analogues, making them attractive for inhibitor development. We report a total synthesis of the N1-IMP scaffold from adenine and a small initial compound series that facilitated early delineation of structure-activity parameters, with analogues evaluated for inhibition of CD38-mediated hydrolysis of cADPR. The 5′-phosphate group proved essential for useful activity, but substitution of this group by a sulfonamide bioisostere was not fruitful. 8-NH2-N1-IMP is the most potent inhibitor (IC50 = 7.6 μM) and importantly HPLC studies showed this ligand to be cleaved at high CD38 concentrations, confirming its access to the CD38 catalytic machinery and demonstrating the potential of our fragment approach.

CD38 has relevance for a range of diseases, e.g., it is a marker of AIDS progression and a negative prognostic marker of chronic lymphocytic leukemia. A recent review categorized the enzyme as a druggable target, at least for human cancers [14]. It was also shown to be influential for social behaviour in mice [15] and plays a key role in age-related NAD + decline. NAD + metabolism is implicated in the aging process and in the pathogenesis of several diseases. CD38 inhibition can decrease NADase activity and boost cellular CD38 has relevance for a range of diseases, e.g., it is a marker of AIDS progression and a negative prognostic marker of chronic lymphocytic leukemia. A recent review categorized the enzyme as a druggable target, at least for human cancers [14]. It was also shown to be influential for social behaviour in mice [15] and plays a key role in age-related NAD + decline. NAD + metabolism is implicated in the aging process and in the pathogenesis of several diseases. CD38 inhibition can decrease NADase activity and boost cellular NAD + levels and such a therapy could be used to promote increases in longevity and health span in models of aging and age-related disease [14]. Therefore, there is significant interest to identify CD38 inhibitors and provide structural clues for design of potential drug candidates.
The greatest therapeutic success so far has been in multiple myeloma using CD38 inhibitors as an antibody-based therapy to target white blood cells in the bone marrow that cause the disease and where CD38 is found on the cell surface. Darzalex (daratumumab), an FDA-approved CD38 inhibitor for mono-and combination therapy of multiple myeloma, binds to CD38, blocks the growth of the cells and induces their death. Several other antibody therapies are currently being evaluated in clinical trials, for example Isatuximab, GBR 1342, TAK−079 and TAK−169 [16], but relatively few small molecule CD38 inhibitors have been reported to date and there is a need to identify lead structures.
Inhibitors of the CD38 NAD + glycohydrolase activity have mainly been investigated, the best being covalent mechanism-based agents that modify the active site. For example, nicotinamide ribose derivatives derived from NAD + exhibit Ki values in the nanomolar range [17,18]. Metabolically stable nicotinamide-based analogues can block endogenous CD38 activity [19]. A non-hydrolysable NAD + analogue is a weak micromolar competitive inhibitor [19]. Membrane permeable analogues are low mM inhibitors and could relax agonist-induced muscle contraction [20]. NAD + analogues with ribose, nucleobase, or pyrophosphate modifications have been explored [21]. Others explored non-nucleotide compounds and non-covalent compounds via screening methodologies [22]. Screening yielded a compound that after optimization afforded a non-covalent CD38 NADase inhibitor with an IC50 of 4.7 μM. Low micromolar concentrations of flavonoids inhibit CD38 [23]. A recent study reported the first small molecule allosteric modulator LX102 [24]. The structure of the CD38 catalytic domain and mechanism of cADPR breakdown have re- The greatest therapeutic success so far has been in multiple myeloma using CD38 inhibitors as an antibody-based therapy to target white blood cells in the bone marrow that cause the disease and where CD38 is found on the cell surface. Darzalex (daratumumab), an FDA-approved CD38 inhibitor for mono-and combination therapy of multiple myeloma, binds to CD38, blocks the growth of the cells and induces their death. Several other antibody therapies are currently being evaluated in clinical trials, for example Isatuximab, GBR 1342, TAK−079 and TAK−169 [16], but relatively few small molecule CD38 inhibitors have been reported to date and there is a need to identify lead structures.
Inhibitors of the CD38 NAD + glycohydrolase activity have mainly been investigated, the best being covalent mechanism-based agents that modify the active site. For example, nicotinamide ribose derivatives derived from NAD + exhibit K i values in the nanomolar range [17,18]. Metabolically stable nicotinamide-based analogues can block endogenous CD38 activity [19]. A non-hydrolysable NAD + analogue is a weak micromolar competitive inhibitor [19]. Membrane permeable analogues are low mM inhibitors and could relax agonist-induced muscle contraction [20]. NAD + analogues with ribose, nucleobase, or pyrophosphate modifications have been explored [21]. Others explored non-nucleotide compounds and non-covalent compounds via screening methodologies [22]. Screening yielded a compound that after optimization afforded a non-covalent CD38 NADase inhibitor with an IC 50 of 4.7 µM. Low micromolar concentrations of flavonoids inhibit CD38 [23]. A recent study reported the first small molecule allosteric modulator LX102 [24]. The structure of the CD38 catalytic domain and mechanism of cADPR breakdown have recently been elucidated crystallographically using covalent inhibitors [25,26]. Glu−226 is the catalytic residue and mutation eliminates activity [27]. Glu−146 is critical to regulate the multi-functionality of CD38-mediated NAD + hydrolysis, the ADP-ribosyl cyclase and cADPR hydrolysis activities [27][28][29].
We previously designed the hydrolysis-resistant cADPR analogue, cyclic inosine 5′diphosphoribose (N1-cIDPR, 2, Figure 2) in which an oxo group at position 6 replaces the amino group [30,31]. cADPR hydrolysis by CD38 is inhibited with an IC50 of 276 μM and in T-cells N1-cIDPR induces Ca 2+ release almost indistinguishably to that induced by cADPR [30,32]. We also described the first total synthesis of the membrane permeant, hydrolytically stable, analogue analogue 8 bromo-cIDPR (3) via regio-and stereoselective N1-ribosylation of protected 8-bromoinosine [33]. A crystal structure of the ligand with wild-type CD38 showed N1-cIDPR to bind in the active site, close to catalytic Glu−226 with the two hydroxyl groups of the "northern" ribose forming hydrogen bonds [34]. This work facilitates, at least in principle, structure-based design of novel CD38 inhibitors using the N1-cIDPR template. In another approach to more drug-like inhibitors we deleted the pyrophosphate group of the macrocycle using a "click" approach without serious loss of activity (5, Figure 2) [35]. Structure and nomenclature of cADPR and previous analogues. NB the "northern" and "southern" riboses of the cyclic analogues are distinguished by adopting prime (′) and double prime (″) notation respectively for their sugar carbons.
Analogues based on the N1-cIDPR template replaced the "southern" N9-ribose with a butyl chain, illustrating the nonessential nature of the "southern" ribose for binding [36] and 8-amino-N9-butyl-cIDPR (6) compared to the best non-covalent CD38 inhibitors (IC50 = 3.3 μM). Crystallographic analysis of the complex with CD38 unexpectedly revealed an N1-hydrolyzed ligand in the active site and ligand cleavage at high protein concentrations was confirmed. We described X-ray crystal structures of CD38 in complex with two nonhydrolysable inhibitors, an 8-substituted N1-cIDPR analogue analogue [37] and cADP carbocyclic ribose (cADPcR, 7, Figure 2) [38] and the elucidation of a preliminary SAR for inhibitors [39].
More recently, we exploited the cIDPR template to generate CD38 inhibitors via total synthesis. In the first example of a sugar hybrid cIDPR analog, L-cIDPR (8), the natural "northern" N1-linked D-ribose of cADPR was replaced by an L-ribose [40] and other work has demonstrated the existence of conformers in these macrocycles [41]. Structure and nomenclature of cADPR and previous analogues. NB the "northern" and "southern" riboses of the cyclic analogues are distinguished by adopting prime ( ) and double prime ( ) notation respectively for their sugar carbons.
Analogues based on the N1-cIDPR template replaced the "southern" N9-ribose with a butyl chain, illustrating the nonessential nature of the "southern" ribose for binding [36] and 8-amino-N9-butyl-cIDPR (6) compared to the best non-covalent CD38 inhibitors (IC 50 = 3.3 µM). Crystallographic analysis of the complex with CD38 unexpectedly revealed an N1-hydrolyzed ligand in the active site and ligand cleavage at high protein concentrations was confirmed. We described X-ray crystal structures of CD38 in complex with two non-hydrolysable inhibitors, an 8-substituted N1-cIDPR analogue analogue [37] and cADP carbocyclic ribose (cADPcR, 7, Figure 2) [38] and the elucidation of a preliminary SAR for inhibitors [39].
More recently, we exploited the cIDPR template to generate CD38 inhibitors via total synthesis. In the first example of a sugar hybrid cIDPR analog, L-cIDPR (8), the natural "northern" N1-linked D-ribose of cADPR was replaced by an L-ribose [40] and other work has demonstrated the existence of conformers in these macrocycles [41].
From a comparison of the N1-cIDPR and cADPcR complexes with wild-type CD38 it was clear that the "northern" ribose part of the cyclic dinucleotide (ribose and/or carbocycle) is more important in binding than the "southern" part [39]. The "northern" ribose monophosphate motif of N1-cIDPR and the carbocyclic ribose monophosphate of cADPcR overlap with the rest of the ligand accommodated more flexibly. This implied that  non-cyclic simple fragments of the macrocycle could maintain key interactions  with wild-type CD38 and might inhibit the enzyme. A small series of N1-hypoxanthine ribose 5 -monophosphate fragments (N1-IMPs, 9−11), derived from careful degradation of the parent cyclic compound [39,42] were indeed inhibitors of CD38-catalyzed cADPR hydrolysis (Scheme 1). Moreover, 8-amino N1-IMP (11) showed promise, being considerably better than its cyclic counterpart (7.6 µM cf. 8-NH 2 -N1-cIDPR (4) at 56 µM) and this was explored and rationalized in a preliminary fashion through docking experiments [39]. The reduced complexity and lower molecular weight of such fragments make them attractive as a starting point for further inhibitor design. 8-NH 2 -N1-IMP is among the best small non-covalent molecule inhibitors of CD38 activity reported so far; thus, its further development to design agents for pharmacological intervention is desirable and a straightforward synthesis is required.
From a comparison of the N1-cIDPR and cADPcR complexes with wild-type CD38 it was clear that the "northern" ribose part of the cyclic dinucleotide (ribose and/or carbocycle) is more important in binding than the "southern" part [39]. The "northern" ribose monophosphate motif of N1-cIDPR and the carbocyclic ribose monophosphate of cADPcR overlap with the rest of the ligand accommodated more flexibly. This implied that perhaps non-cyclic simple fragments of the macrocycle could maintain key interactions with wildtype CD38 and might inhibit the enzyme.
A small series of N1-hypoxanthine ribose 5′-monophosphate fragments (N1-IMPs, 9−11), derived from careful degradation of the parent cyclic compound [39,42] were indeed inhibitors of CD38-catalyzed cADPR hydrolysis (Scheme 1). Moreover, 8-amino N1-IMP (11) showed promise, being considerably better than its cyclic counterpart (7.6 μM cf. 8-NH2-N1-cIDPR (4) at 56 μM) and this was explored and rationalized in a preliminary fashion through docking experiments [39]. The reduced complexity and lower molecular weight of such fragments make them attractive as a starting point for further inhibitor design. 8-NH2-N1-IMP is among the best small non-covalent molecule inhibitors of CD38 activity reported so far; thus, its further development to design agents for pharmacological intervention is desirable and a straightforward synthesis is required.
We now report the synthesis of 8-NH2-N1-IMP (11) and a small focused SAR study to examine obvious points of substitution and clarify the importance of the "northern" ribose phosphate group motif for cIDPR-based inhibitors ( Figure 3). Analogues were evaluated for their inhibition of CD38-mediated hydrolysis of cADPR. We now report the synthesis of 8-NH 2 -N1-IMP (11) and a small focused SAR study to examine obvious points of substitution and clarify the importance of the "northern" ribose phosphate group motif for cIDPR-based inhibitors ( Figure 3). Analogues were evaluated for their inhibition of CD38-mediated hydrolysis of cADPR.
was clear that the "northern" ribose part of the cyclic dinucleotide (ribose and/or carbocycle) is more important in binding than the "southern" part [39]. The "northern" ribose monophosphate motif of N1-cIDPR and the carbocyclic ribose monophosphate of cADPcR overlap with the rest of the ligand accommodated more flexibly. This implied that perhaps non-cyclic simple fragments of the macrocycle could maintain key interactions with wildtype CD38 and might inhibit the enzyme.
A small series of N1-hypoxanthine ribose 5′-monophosphate fragments (N1-IMPs, 9−11), derived from careful degradation of the parent cyclic compound [39,42] were indeed inhibitors of CD38-catalyzed cADPR hydrolysis (Scheme 1). Moreover, 8-amino N1-IMP (11) showed promise, being considerably better than its cyclic counterpart (7.6 μM cf. 8-NH2-N1-cIDPR (4) at 56 μM) and this was explored and rationalized in a preliminary fashion through docking experiments [39]. The reduced complexity and lower molecular weight of such fragments make them attractive as a starting point for further inhibitor design. 8-NH2-N1-IMP is among the best small non-covalent molecule inhibitors of CD38 activity reported so far; thus, its further development to design agents for pharmacological intervention is desirable and a straightforward synthesis is required.
We now report the synthesis of 8-NH2-N1-IMP (11) and a small focused SAR study to examine obvious points of substitution and clarify the importance of the "northern" ribose phosphate group motif for cIDPR-based inhibitors ( Figure 3). Analogues were evaluated for their inhibition of CD38-mediated hydrolysis of cADPR.

Synthesis of Fragments
Initially, fragments 14-17 were designed based on the cIDPR structure with only the pyrophosphate deleted.

Synthesis of Fragments
Initially, fragments 14-17 were designed based on the cIDPR structure with only the pyrophosphate deleted.

Synthesis of Fragments
Initially, fragments 14-17 were designed based on the cIDPR structure with only the pyrophosphate deleted.

Scheme 4. Synthesis of N9-(4-hydroxybutyl)-N1-IMP analogues. Reagents and conditions
We previously synthesized 8-NH2-N1-IMP (11) by destruction of 8-N3-cIDPR (12, Scheme 2) through acid catalyzed hydrolysis at elevated temperature [39]. However, this route is inefficient and thus we sought an alternative via total synthesis. Starting from adenine 45, the tert-butyldimethylsilyloxymethyl group [43] was introduced in a two-step procedure. In contrast to other N9 protecting groups, such as benzyl, benzoyl or acetyl, this generates an organically soluble product that is considerably easier to handle. The N9-protected adenine 46 was then prepared for a regio-and stereoselective N1-glycosylation by introduction of a bromine group at C8 to give 47, followed by treatment with We previously synthesized 8-NH 2 -N1-IMP (11) by destruction of 8-N 3 -cIDPR (12, Scheme 2) through acid catalyzed hydrolysis at elevated temperature [39]. However, this route is inefficient and thus we sought an alternative via total synthesis. Starting from adenine 45, the tert-butyldimethylsilyloxymethyl group [43] was introduced in a two-step procedure. In contrast to other N9 protecting groups, such as benzyl, benzoyl or acetyl, this generates an organically soluble product that is considerably easier to handle. The N9-protected adenine 46 was then prepared for a regio-and stereoselective N1glycosylation by introduction of a bromine group at C8 to give 47, followed by treatment with sodium nitrite to effect conversion from adenine to hypoxanthine base (48). N1glycosylation was effected by treatment of 48 with DBU, followed by TMSOTf and 1,2,3,5tetra-O-acetyl-β-D-ribofuranose to afford 49. The three acetyl protecting groups were removed using methanolic ammonia and exchanged for a 2 ,3 -O-isopropylidene group by treatment with 2,2-dimethoxypropane and acetone under acidic conditions to afford 50. The protected precursor 50 was phosphitylated at the free 5 -OH using di-tert-butyl protected phosphoramidite, followed by oxidation to the corresponding phosphate using hydrogen peroxide and triethylamine to afford 51. A convenient global deprotection of the tert-butyldimethylsilyloxymethyl, isopropylidene ketal and two tert-butyl phosphate esters using 50% aqueous TFA generated 10, the N1-IMP scaffold with an 8-bromo substituent. This could be conveniently manipulated to generate 8-NH 2 -N1-IMP 11 by sequential treatment with TMS-N 3 and dithiothreitol (Scheme 5). Attempts to convert the 8-Br substituent to the 8-N 3 analogue at an earlier stage in the synthesis were unsuccessful as the N7-tert-butyldimethylsilyloxymethyl protecting group was cleaved concurrently. sodium nitrite to effect conversion from adenine to hypoxanthine base (48). N1-glycosylation was effected by treatment of 48 with DBU, followed by TMSOTf and 1,2,3,5-tetra-Oacetyl-β-D-ribofuranose to afford 49. The three acetyl protecting groups were removed using methanolic ammonia and exchanged for a 2′,3′-O-isopropylidene group by treatment with 2,2-dimethoxypropane and acetone under acidic conditions to afford 50. The protected precursor 50 was phosphitylated at the free 5′-OH using di-tert-butyl protected phosphoramidite, followed by oxidation to the corresponding phosphate using hydrogen peroxide and triethylamine to afford 51. A convenient global deprotection of the tert-butyldimethylsilyloxymethyl, isopropylidene ketal and two tert-butyl phosphate esters using 50% aqueous TFA generated 10, the N1-IMP scaffold with an 8-bromo substituent. This could be conveniently manipulated to generate 8-NH2-N1-IMP 11 by sequential treatment with TMS-N3 and dithiothreitol (Scheme 5). Attempts to convert the 8-Br substituent to the 8-N3 analogue at an earlier stage in the synthesis were unsuccessful as the N7-tertbutyldimethylsilyloxymethyl protecting group was cleaved concurrently. We next sought to explore the SAR of these more accessible, less negatively charged molecules. Following on from our earlier interesting results with inhibition of CD38 from compounds with an L-configuration "northern" ribose [40], we synthesized fragments with an L-ribose (Scheme 6). Briefly, N1-glycosylation of 48 was affected by treatment with DBU, TMSOTf and 1,2,3,5-tetra-O-acetyl-β-L-ribofuranose to afford 52. Treatment with methanolic ammonia removed the three acetyl groups (53), followed by introduction of a 2′,3′-O-isopropylidene ketal to afford 54. The protected precursor was phosphitylated at the free 5′-OH and oxidized using the methods described above to afford 55. Global deprotection using 50% aqueous TFA generated 26, the L-N1-IMP scaffold with an 8-bromo substituent. The 8-bromo substituent was reduced to generate L-N1-IMP 27 using hydrogenation with palladium on carbon. We next sought to explore the SAR of these more accessible, less negatively charged molecules. Following on from our earlier interesting results with inhibition of CD38 from compounds with an L-configuration "northern" ribose [40], we synthesized fragments with an L-ribose (Scheme 6). Briefly, N1-glycosylation of 48 was affected by treatment with DBU, TMSOTf and 1,2,3,5-tetra-O-acetyl-β-L-ribofuranose to afford 52. Treatment with methanolic ammonia removed the three acetyl groups (53), followed by introduction of a 2 ,3 -O-isopropylidene ketal to afford 54. The protected precursor was phosphitylated at the free 5 -OH and oxidized using the methods described above to afford 55. Global deprotection using 50% aqueous TFA generated 26, the L-N1-IMP scaffold with an 8bromo substituent. The 8-bromo substituent was reduced to generate L-N1-IMP 27 using hydrogenation with palladium on carbon.
We next explored the potential for N1-phosphate replacement with a bioisostere (Scheme 7). Phosphate bioisosteres present binding partners without negative charge, which is more attractive for drug design [44].
N1-(β-D-Ribofuranosyl)-N9-tert-butyldimethylsilyloxymethyl−8-bromohypoxanth-ine (50) was treated with triethylamine followed by addition of sulfamoyl chloride. In addition to the desired introduction of a 5-O-sulfonamide group, the 8-bromo substituent was also substituted by an 8-chloro substituent in the reaction mixture, confirmed by MS (ES + 588.13 and 590.13, 3:1) to afford 56. Deprotection of the N9-protecting group with aqueous TFA gave the 8-chloro sulfonamide analogue 28. Removal of the 8-chloro substituent by treatment with palladium on carbon under an atmosphere of hydrogen gave the N9-protected parent analogue 57, which was then deprotected using aqueous TFA to give sulfonamide analogue 29. We next explored the potential for N1-phosphate replacement with a bioisostere (Scheme 7). Phosphate bioisosteres present binding partners without negative charge, which is more attractive for drug design [44].

N1-(β-D-Ribofuranosyl)-N9-tert-butyldimethylsilyloxymethyl−8-bromohypoxanth-
ine (50) was treated with triethylamine followed by addition of sulfamoyl chloride. In addition to the desired introduction of a 5-O-sulfonamide group, the 8-bromo substituent was also substituted by an 8-chloro substituent in the reaction mixture, confirmed by MS (ES + 588.13 and 590.13, 3:1) to afford 56. Deprotection of the N9-protecting group with aqueous TFA gave the 8-chloro sulfonamide analogue 28. Removal of the 8-chloro substituent by treatment with palladium on carbon under an atmosphere of hydrogen gave the N9-protected parent analogue 57, which was then deprotected using aqueous TFA to give sulfonamide analogue 29.  We next explored the potential for N1-phosphate replacement with a bioisostere (Scheme 7). Phosphate bioisosteres present binding partners without negative charge, which is more attractive for drug design [44].

N1-(β-D-Ribofuranosyl)-N9-tert-butyldimethylsilyloxymethyl−8-bromohypoxanth-
ine (50) was treated with triethylamine followed by addition of sulfamoyl chloride. In addition to the desired introduction of a 5-O-sulfonamide group, the 8-bromo substituent was also substituted by an 8-chloro substituent in the reaction mixture, confirmed by MS (ES + 588.13 and 590.13, 3:1) to afford 56. Deprotection of the N9-protecting group with aqueous TFA gave the 8-chloro sulfonamide analogue 28. Removal of the 8-chloro substituent by treatment with palladium on carbon under an atmosphere of hydrogen gave the N9-protected parent analogue 57, which was then deprotected using aqueous TFA to give sulfonamide analogue 29.
The two analogues with an L-ribose as the N1-ribose configuration (26,27) showed contrasting activity. While the activity of 8-Br analogue 26 was similar to the 8-Br N1-IMP analogue 10, the 8-H analogue 27 showed a 30-fold reduction in activity compared to its D-ribose counterpart 9 (IC 50 = 460 and 14 µM, respectively). In other studies, where the L-ribose was constrained as part of a cyclic analog, the L-ribose substitution highlighted differences in binding activity that were attributed to likely different binding modes [40]. For fragments such as 26 and 27, however, there would be free rotation around the N1ribosyl bond and a smaller overall ligand to fit into the binding pocket.
CD38 is predominantly an ectoenzyme, but to a small degree its catalytic site can also face the intracellular environment, e.g., Type III CD38 has its C-terminal facing intracellularly, CD38 is present in the nucleus and mitochondrial membrane and a soluble form of CD38 is likely present in the cytoplasm [14]. Thus, approaches to neutralize the mono-phosphate charges of inhibitors could be useful for wider targeting of the enzyme. Attempted phosphate replacement with a sulfonamide bioisostere (Scheme 7), however, did not generate CD38 inhibitors, as neither of the two analogues (28,29) showed any activity. For further development there are obviously many further phosphate bioisosteres that could be explored, as well as perhaps more importantly alternative methods to mask phosphate negative charges, such as acetoxymethyl-esters and the Protide approach, using groups that may be cleaved intracellularly [44].

Ligand Hydrolysis by CD38
Previous studies demonstrated that cADPR analogues inhibiting CD38-mediated hydrolysis could be turned over by the high concentrations of CD38 catalytic domain (shCD38) used during crystallography [36]. Indeed, 8-NH 2 -N9-butyl-cIDPR (6, Figure 2) was captured as the hydrolyzed product in the crystal structure with shCD38. Demonstration of hydrolysis by shCD38 using HPLC suggests that the fragment is indeed binding to the cADPR pocket, probably in an orientation that places the N1-ribosyl bond within reach of the catalytic residue. Incubation of 11, 8-NH 2 -N1-IMP (1 mM final concentration) with 4 mg/mL shCD38 was monitored using RP-HPLC. The peak corresponding to 8-NH 2 -N1-IMP (R T = 10.5 min.) reduced in intensity over time, alongside the appearance of a new peak (R T = 2.0 min.) that was characteristic of an 8-amino substituted hypoxanthine analogue (See Supplementary Information, Figure S1). No change in the original peak was observed in a parallel control experiment containing no shCD38 (data not shown). 8-NH 2 -N1-IMP (11) was hydrolyzed more rapidly than cIDPR (2), but more slowly than 8-NH 2 -N9-butyl-cIDPR (6) (Figure 4). tration) with 4 mg/mL shCD38 was monitored using RP-HPLC. The peak corresponding to 8-NH2-N1-IMP (RT = 10.5 min.) reduced in intensity over time, alongside the appearance of a new peak (RT = 2.0 min.) that was characteristic of an 8-amino substituted hypoxanthine analogue (See Supplementary Information, Figure S1). No change in the original peak was observed in a parallel control experiment containing no shCD38 (data not shown). 8-NH2-N1-IMP (11) was hydrolyzed more rapidly than cIDPR (2), but more slowly than 8-NH2-N9-butyl-cIDPR (6) (Figure 4). Both 8-NH2-substituted analogues are more potent inhibitors of CD38-mediated hydrolysis than cIDPR (2) (IC50 values of 7.6 and 3.3 μM compared to 276 μM). Perhaps most surprising is that the small fragment, 8-NH2-N1-IMP (11) is hydrolyzed at the N1-ribosyl bond, suggesting that it not only binds in the active site of CD38 but also that it orientates itself to bind with the N1-ribosyl bond accessible to the catalytic residue. This would seem more likely for the larger cyclic ligands and adds further weight to the argument that the "northern" ribose makes key interactions in the CD38 binding site [36,40].

General
All reagents and solvents were of commercial quality and were used without further purification, unless described otherwise. Unless otherwise stated, all reactions were carried out under an inert atmosphere of argon. 1 H, 13 C, and 31 P-NMR spectra were collected on a Varian Mercury 400 MHz or Bruker Avance III 500 MHz spectrometer. All 1 H and 13 C NMR assignments are based on gCOSY, gHMBC, gHSQC, and DEPT−135 experiments. Abbreviations for splitting patterns are as follows: br, broad; s, singlet; d, doublet; t, triplet; m, multiplet. Coupling constants are given in hertz (Hz). High resolution timeof-flight mass spectra were obtained on a Bruker Daltonics micrOTOF mass spectrometer Both 8-NH 2 -substituted analogues are more potent inhibitors of CD38-mediated hydrolysis than cIDPR (2) (IC 50 values of 7.6 and 3.3 µM compared to 276 µM). Perhaps most surprising is that the small fragment, 8-NH 2 -N1-IMP (11) is hydrolyzed at the N1ribosyl bond, suggesting that it not only binds in the active site of CD38 but also that it orientates itself to bind with the N1-ribosyl bond accessible to the catalytic residue. This would seem more likely for the larger cyclic ligands and adds further weight to the argument that the "northern" ribose makes key interactions in the CD38 binding site [36,40].

General
All reagents and solvents were of commercial quality and were used without further purification, unless described otherwise. Unless otherwise stated, all reactions were carried out under an inert atmosphere of argon. 1 H, 13 C, and 31 P-NMR spectra were collected on a Varian Mercury 400 MHz or Bruker Avance III 500 MHz spectrometer. All 1 H and 13 C NMR assignments are based on gCOSY, gHMBC, gHSQC, and DEPT−135 experiments. Abbreviations for splitting patterns are as follows: br, broad; s, singlet; d, doublet; t, triplet; m, multiplet. Coupling constants are given in hertz (Hz). High resolution time-of-flight mass spectra were obtained on a Bruker Daltonics micrOTOF mass spectrometer using electrospray ionization (ESI). The purity of new tested compounds was determined to be ≥95% by analytical HPLC (see Supplementary Materials). Analytical HPLC analyses were carried out on a Waters 2695 Alliance module equipped with a Waters 2996 photodiode array detector (210-350 nm). The chromatographic system consisted of a Hichrom Guard column for HPLC and a Phenomenex Synergi 4 µm MAX-RP 80A column (150 mm × 4.60 mm), with elution at 1 mL/min with either (a) isocratic ion-pair buffer: 0.17% (m/v) cetrimide and 45% (v/v) phosphate buffer (pH 6.4) in MeOH or (b) a gradient of 0.05M Triethylammonium bicarbonate (TEAB):MeCN (95:5 → 35:65 v/v). TEAB was prepared by bubbling CO 2 (g) through a 0.05M solution of triethylamine in MilliQ water to pH ≤ 8. "MilliQ" water refers to purified water from a MilliQ ® Reference Water Purification system, resistivity of 18.2 MΩ.cm (at 25 • C). Semi-preparative HPLC was performed on a Waters 2525 pump with manual FlexInject using a Phenomenex Gemini column (5u, C18, 110 Å, 250 × 10.00 mm), eluted at 5 mL min −1 with a gradient of 0.05M TEAB:MeCN (95:5 → 35:65 v/v). Synthetic phosphates were assayed and quantified by the Ames phosphate test [45]. Non-phosphate final compounds were quantified by quantitative 1 H-NMR. Note the use of 0.5% pyridine in chromatography systems with acid sensitive functional groups (e.g. phosphates protected with tert-butyl ethers) or free monophosphates. This system was used (rather than triethylamine) as it was less basic and therefore prevented decomposition of the analogues. Pyridine was evaporated from the TLC plate using a heat gun before visualization under UV light.

HPLC Studies
HPLC studies were carried out as previously described [36]. Briefly, the solution containing shCD38 was adjusted to the desired concentration (4 mg/mL) using Tris-HCl buffer (20 mM, pH 8) and 50 µL was added to the inhibitor (0.05 µmole in MilliQ (2 mL)-1 mM final concentration) in an Eppendorf tube at room temperature (T = 0). At a given time point, a sample of 5 µL was removed and diluted with 95 µL MilliQ water. 10 µL Of this sample was injected directly into the analytical HPLC system (see General Experimental), eluting at 1 mL/min with an isocratic ion-pair buffer: 0.17% (m/v) cetrimide and 45% (v/v) phosphate buffer (pH 6.4) in MeOH.

Conclusions
Five fragment scaffolds were prepared, each with multiple 8-substitutions. N1-ribosylinosine derivatives 14-17 and N9-Hydroxybutyl-N1-inosine derivatives 18-21 are nonphosphorylated analogues that retain the key "northern" ribose motif. These analogues illustrate the importance of the 5 -phosphate group on the "northern" ribose for CD38 inhibitory activity. Introduction of the 5 -phosphate group to in N9-hydroxybutyl-N1-IMP analogues 22-25 shows some improvement in activity; however, the unconstrained N9butyl chain appeared to be detrimental, compared to its effect in cyclic analogues [36]. The promising fragment 8-NH 2 -N1-IMP (11) was prepared via total synthesis for the first time, which affords a route to generate this analogue in more significant amounts (compared alternatively to the previously reported degradation of the cyclic parent analog) and to access other related analogues for SAR studies. To illustrate the utility of this new synthetic route, L-ribose (26)(27) and sulphonamide (28)(29) analogues were prepared. In summary, this work illustrates the potential for design of much simpler and mono-phosphorylated CD38 inhibitors, through a key structural motif derived from its macrocyclic pyrophosphate ligand that could be worthy of future optimization and development. Importantly, their continuing, albeit weak, substrate activity implies that such compounds bind closely mimicking the relevant part of the natural ligand, which should aid structure-based design strategies. CD38 while generally an ectoenzyme, does also exist inside cells, so the reduction of the inhibitor class to a simple monophosphate derivative as here makes available wellestablished prodrug strategies that should improve inhibitor access.
Supplementary Materials: The following are available online, Figure S1: Hydrolysis of 8-NH 2 -N1-IMP by high concentrations of shCD38; 1 H, 13 C and 31 P spectral data and HPLC profiles for novel compounds.