Design and Exploration of a Tricyclic Series of CNS-Penetrant CD38 Small Molecule Inhibitors

Abstract

Cluster of Differentiation 38 (CD38) plays a key role in NAD⁺ metabolism and calcium signaling, and recent reports highlight therapeutic opportunities for CD38 modulation in central nervous system (CNS) disorders. In this report, we describe the rational design, synthesis, and exploration of structure–activity relationships (SAR) for a novel and brain-penetrant series of tricyclic compounds that form an adduct with ADPR to inhibit CD38 in an uncompetitive manner. Rigidification of the bioactive conformation via covalent tethering of a 6-membered heteroaromatic core to an imidazole warhead served as the key design element that afforded improved biochemical potency against both mouse and human CD38. Incremental property-based optimization of core and linker motifs generated analogues with enhanced unbound brain-to-plasma ratios in mice. These efforts led to the discovery of brain-penetrant CD38 inhibitor 16, which was evaluated in vivo and demonstrated dose-dependent increases of NAD⁺ measured in mouse brain and lung tissue samples following oral doses ranging from 3 to 100 mg/kg.

Cluster of Differentiation 38 (CD38) is a multifunctional type II transmembrane glycoprotein highly expressed on the cell surface of immune cells but also expressed more broadly at lower levels intracellularly where it contributes to calcium signaling. , CD38 exerts significant influence on cellular levels of nicotinamide adenine dinucleotide (NAD⁺) via its NADase activity that catalyzes the breakdown of NAD⁺ and related metabolites. , Importantly, the cofactor NAD⁺ is essential to a variety of cellular processes such as cell metabolism, energy production, and DNA repair. Alterations in NAD⁺ homeostasis have been implicated in the pathogenesis of a range of diseases including diabetes, cancer, cardiovascular, inflammatory, and neurodegenerative disorders. , In addition to its role as an NADase, CD38 also demonstrates ADP-ribosyl cyclase activity that generates cyclic ADP-ribose to trigger calcium release from intracellular stores. Alternatively, the presence of CD38 in the acidic lysosomal compartment enables regulation of lysosomal calcium through NAADP production.

Due to its central role in NAD⁺ metabolism and calcium signaling, CD38 has emerged as an intriguing therapeutic target for a variety of indications. Accordingly, genetic deletion and pharmacological inhibition of CD38 have been shown to modulate levels of NAD⁺ or its metabolites in a variety of tissues and have beneficial effects in multiple disease models, − including brain-related injury models. , In response to the potential of this biology, there have been several reports within the scientific and patent literature that detail small molecule inhibitors of CD38, including 78c (1), MK-0159 (2), and pyrimidine 3 (Figure ). Brain-penetrant compounds have also started to emerge, exemplified by molecules CVN14 (4) and A-3190 (5). − With continued interest in the effects of CD38 modulation as a potential therapeutic strategy for CNS disorders, there is benefit to identifying additional and chemically differentiated CNS-penetrant tools.

1.

Examples of previously reported small molecule CD38 inhibitors.

Reported small molecule CD38 inhibitors typically feature a heteroaromatic warhead that forms an adduct with adenosine diphosphate ribose (ADPR) to generate an active, uncompetitive CD38 binder in situ. − This distinctive mode of product-assisted CD38 inhibition presents the advantage of amplifying compound potency while maintaining relatively low molecular weight (MW), which is a key optimization parameter for efficient drug delivery in vivo. Additionally, known CD38 inhibitors often contain a hydrogen bond donor (HBD), commonly in the form of an amide or vinylogous amide as exemplified in compounds 1–4, which provides a vector for a “tail” moiety. This tail extends into the solvent and may facilitate intramolecular stabilizing interactions with the adenine of the in-situ-formed ligand–ADPR complex within the CD38 active site. We had previously reported CD38 inhibitor 5 in which the conventional amide tail motif was replaced with a cyclopropyl sulfonyl pyrazole group, resulting in a molecule that lacks HBDs.

During our continued investigations of CNS-active CD38 inhibitors, we separately identified compound 6 (Figure ). Although the core pyrimidine and imidazole warhead motifs of analogue 6 draw similarity to that of literature examples 3 and 4, it instead incorporates the cyclopropyl sulfonyl pyrazole “tail” functionality present in compound 5, which renders it devoid of HBDs. Compound 6 demonstrated modest potency against both mouse and human CD38 (mCD38 and hCD38) and has a CNS MPO score of 5.81, which is consistent with chemical space that often has favorable CNS druglike properties. , Indeed, compound 6 was found to be brain-penetrant with an unbound brain-to-plasma ratio (unb. b:p) equal to 0.55 in mice. Minimizing HBDs is a common approach for optimizing small molecules for blood–brain barrier (BBB) permeability, in addition to other strategies such as maintaining low MW, enhancing rigidity, modulating lipophilicity, increasing passive diffusion, and controlling active transport. − Given the therapeutic potential of CD38 modulation in the brain, the structure–activity relationship (SAR) studies detailed herein leverage these tactics to access novel, brain-penetrant CD38 inhibitors.

2.

Brain-penetrant CD38 inhibitor 6 with the cyclopropyl sulfonyl pyrazole motif highlighted in blue and an X-ray co-crystal structure (PDB 10OT) of 6 (blue sticks) bound to the mCD38 active site via the ADPR–ligand complex (white lines). The yellow arrow illustrates the proposed design strategy to reinforce the active planar conformation.

With compound 6 as a promising starting point, our aim was to enhance the CD38 inhibitory potency without sacrificing BBB permeability. To guide ligand optimization, an X-ray cocrystal structure of 6 bound to mCD38 was obtained, which confirmed formation of the ligand–ADPR complex in the active site. Structural analysis revealed that the imidazole moiety, which facilitates the in-situ ADPR adduct formation, adopts a nearly coplanar orientation with the pyrimidine core, with a dihedral angle of ∼2° between the two. This observation led to the idea that introducing a covalent tether between the pyrimidine and imidazole groups may improve the potency of CD38 inhibition by preorganization and rigidification of the bioactive conformation.

To interrogate this hypothesis, several cyclized ether analogues varying in ring size and warhead connection point were designed and synthesized (Figure ). Analogues 7 and 8, featuring two and three atom ether linkages to the carbon atom of the imidazole, respectively, resulted in modest CD38 IC₅₀ values comparable to uncyclized parent 6. However, improvements of greater than 10-fold in mCD38 and hCD38 inhibitory potencies were realized through rotation of the imidazole warhead to furnish isomeric analogue 9. Despite similar calculated pK _aH values, it is speculated that the N-alkyl imidazole of 9 is more nucleophilic than the N-aryl imidazole of 8 and thus can better facilitate the in-situ ligand–ADPR complex formation required to drive potent CD38 inhibition. An X-ray co-crystal structure of 9 and mCD38 was obtained and illustrated that the N-alkylated imidazole ring is indeed restricted to a planar conformation and successfully formed the active ligand–ADPR adduct. Growing the N-alkyl imidazole tether to four atoms, exemplified by 10, resulted in diminished activity against CD38. This could indicate a disruption of the desired coplanar alignment due to the flexibility of the larger fused-ring system or perhaps an intolerance of the binding pocket toward the added lipophilic bulk, suggesting that the three-atom linkage of N-alkyl analogue 9 is optimal.

3.

Cyclized analogues and their respective etheno-NAD hydrolase assay IC₅₀ values as the geometric mean. Corresponding pIC₅₀ ± SD and number of independent experiments are reported in the . All derivatives feature the cyclopropyl sulfonyl pyrazole motif (undrawn) present in parent molecule 6. An X-ray co-crystal structure (PDB 10LP) of compound 9 (green sticks) bound to the mCD38 active site via the ADPR–ligand complex (white lines) is shown.

This new, atom-efficient design element incorporated into tricycle 9 resulted in improved CD38 inhibitory potency, maintained excellent CNS drug-like property space (CNS MPO score = 5.51), and decreased unbound plasma clearance (CL_p,u) in mice, relative to uncyclized compound 6 (Table ). However, despite enhanced molecular rigidity and only a minimal increase in molecular size, compound 9 exhibited a 10-fold lower unbound brain-to-plasma ratio in mice. It was hypothesized that the modest increase in polarity (as reflected by TPSA) and decrease in lipophilicity (as determined by cLogP) from 6 to 9 could contribute to the reduced CNS permeability.

1. CD38 Inhibitory Activities, Properties, and Mouse PK Parameters for 6, 9, and 11–16 .

Mse PK (0.5 mg/kg, (iv)
Cmpd	X	Y	Z	mCD38 IC50 (nM)	hCD38 IC50 (nM)	TPSA/cLogP	CLp,u (L/h/kg)	Unb. b:p,
6​	–	–	–	22.5	55.2	96/0.41	17.9	0.55
9​	O	N	N	1.82	1.28	105/0.12	<2.36	0.053
11​	O	CH	N	7.22	3.95	92/0.02	4.87	0.044
12​	O	N	CH	6.79	1.55	92/0.41	8.89	0.40
13​	NH	N	CH	5.51	3.94	95/0.04	20.7	0.00
14​	NMe	N	CH	23.2	57.8	86/0.67	n.d.	n.d.
15​	O	CH	N	3.27	1.70	92/0.56	24.7	0.096
16​	O	N	CF	2.71	1.94	92/0.55	12.9	0.23

^an.d. = not determined. Etheno-NAD hydrolase assay IC₅₀ values are reported as the geometric mean. Corresponding pIC₅₀ ± SD and number of independent experiments are reported in the .

^bPlasma and brain concentrations were measured 1 h post-dose in mice.

^cSee for f_u and C_1h measurements.

^dSee Figure for structure of compound 6.

^eCL_p,u calculated using AUC_last, because AUC_inf could not be determined.

^fNo exposure in brain tissue was detected.

^gPyridyl core of compound 15 is substituted with a fluorine atom ortho to the ether linkage.

In search of improved CNS penetration within this new series, several closely related derivatives were designed and synthesized with the goal of incrementally modulating polarity and lipophilicity. All analogues reported herein feature high CNS MPO scores (≥5.2, ) consistent with favorable CNS druglike property space. The relative uniformity of this metric across the series meant that CNS optimization was instead guided by direct in-vivo measurements of unbound brain-to-plasma ratio in mice. To start, SAR efforts were directed toward minor changes to the core heterocycle. Less-polar pyridyl analogues 11 and 12 maintained single-digit nM inhibitory potency against mouse and human CD38 and exhibited moderate CL_p,u in mice. However, only 12, the more lipophilic of the two, demonstrated an improved unbound brain-to-plasma ratio of 0.40. Benzazepine derivative 13 showed that a nitrogen atom in the linker is tolerated against both mouse and human CD38. Unfortunately, 13 proved to be a likely P-gp efflux substrate with diminished passive permeability and correspondingly achieved no exposure in mouse brain tissue. , To improve permeability, the HBD was removed by methylation of the nitrogen atom to give example 14. However, substitution at this position of the linker greatly attenuated CD38 inhibition potency, resulting in over a 10-fold decrease for the human enzyme. Therefore, 14 was not evaluated for plasma and brain exposure in vivo. Lastly, fluoropyridine analogues 15 and 16 were generated to increase lipophilicity incrementally. Although both compounds exhibited similar single-digit nM CD38 IC₅₀ values, compound 16 demonstrated lower CL_p,u and greater CNS penetration in vivo, compared to its isomer, 15, with an unbound brain-to-plasma ratio of 0.23 in mice.

Analogues 12 and 16 were the most brain-penetrant of this new set of tricyclic CD38 inhibitors and proved to have similar CL_p,u in mice. However, oral administration distinguished the two, as a single 30 mg/kg dose of compound 16 yielded more persistent unbound exposures in both plasma and brain tissue compared to the same dose of 12 (). Based on its potent mCD38 inhibition, moderate CL_p,u, and favorable oral bioavailability and tissue distribution, compound 16 was selected for further evaluation in an in-vivo PK/PD mouse model. Oral dosing of 16 in mice (QDx4) resulted in dose-dependent increases in NAD⁺ concentrations as measured in brain and lung tissue samples (Figure ). Notably, compound 16 exhibited minimal to no inhibitory activity across a panel of off-target NAD⁺-consuming enzymes, consisting of sirtuins (SIRTs) and poly­(ADP-ribose) polymerases (PARPs) (). This suggests that the observed increases in NAD⁺ are likely a result of targeted CD38 inhibition, supporting the profile of 16 as a selective, CNS-active CD38 inhibitor.

4.

NAD⁺ concentrations measured in brain (left) and lung (right) tissues following daily oral administration of compound 16 with escalating doses in mice (QDx4, N = 8 per group). All samples were collected at experiment termination, 3 h post-dose on Day 4. Individual NAD⁺ sample measurements are indicated by black spheres and correspond to a unique mouse. (*) p < 0.05 and (***) p < 0.001 compared to vehicle-treated control via one-way ANOVA followed by Dunnett’s multiple comparison test. Unbound drug concentrations in brain tissue (left) and plasma (right) are overlaid and denoted by blue squares. Plasma concentrations from Day 1 (3 h post-dose) can be found in the .

Additional SAR efforts were conducted to further modulate lipophilicity via substitution of the linker motif (Table ). For example, methyl enantiomers 17 and 18 retained similar mCD38 IC₅₀ values relative to parent compound 9, though hCD38 activity was reduced by a full log unit for both. This suggests that the hCD38 binding site does not tolerate much additional lipophilic bulk in this region of the ligand. Unbound brain-to-plasma ratio for compounds 17 and 18 remained low, despite an increased cLogP. In contrast, introduction of geminal fluorine atoms to the central ether linkage yielded a small set of matched comparators, all of which exhibited higher ratios of unbound brain-to-plasma in mice relative to their des-fluoro parent molecules. These smaller fluorine atom substituents were tolerated by hCD38 with no observed loss in IC₅₀, apart from pyridyl analog 20, which lost potency against both the mouse and human enzyme. An additional consequence of the geminal difluoro-substitution was that the imidazole motif of each analogue was calculated to be at least a log unit of pK _aH less basic due to inductive effects of the highly electron-withdrawing fluorine atoms. Compound 21 is particularly noteworthy because it demonstrated a 9-fold improvement in mCD38 IC₅₀ relative to des-fluoro 12 and was measured to have almost equal unbound distribution between brain and plasma in vivo, making it both the most biochemically potent and brain-penetrant example within this newly reported tricyclic series of CD38 inhibitors. However, the increased lipophilicity imparted by the geminal fluorine atoms was accompanied by a higher CL_p,u, relative to 12 and 16, suggesting a less favorable systemic clearance profile. Therefore, 21 was not prioritized for additional in-vivo evaluation at the time of these studies.

2. CD38 Inhibitory Activities, Properties, and Mouse PK Parameters for 17–22 .

							Mse PK (0.5 mg/kg, iv)
Cmpd	Y	Z	R1, R2	mCD38 IC50 (nM)	hCD38 IC50 (nM)	TPSA/cLogP	CLp,u (L/h/kg)	Unb. b:p ,
17	N	N	H, Me	3.09	11.5	105/0.53	7.00	0.099
18	N	N	Me, H	1.94	19.6	105/0.53	<4.45	0.10
19	N	N	F, F	1.31	1.87	105/0.98	19.0	0.58
20	CH	N	F, F	8.20	50.2	92/0.87	4.75	0.20
21	N	CH	F, F	0.727	1.37	92/1.25	<26.1	0.94
22	N	CF	F, F	0.852	1.50	92/1.40	61.6	0.51

^aEtheno-NAD hydrolase assay IC₅₀ values are reported as the geometric mean. Corresponding pIC₅₀ ± SD and number of independent experiments are reported in the .

^bPlasma and brain concentrations were measured 1 h post-dose in mice.

^cSee for f_u and C_1h measurements.

^dCL_p,u calculated using AUC_last, because AUC_inf could not be determined.

A few different approaches were employed to construct the previously unreported tricycle motifs in the series of CD38 inhibitors described herein, as illustrated in Schemes , , and by representative syntheses of compounds 9, 12, and 21, respectively. First, synthesis of pyrimidine analogue 9 began with a Mitsunobu reaction between corresponding dichlorophenol 9a and a TBS-protected glycol. The resulting intermediate was then subjected to a Suzuki coupling with N-trityl imidazole boronic acid 23 to afford cyclization precursor 9b. This intermediate was deprotected under acidic conditions, tosylated, and then cyclized via an intramolecular S_N2 displacement facilitated by sodium hydride to give penultimate 9c. This sequence was also used analogously to obtain intermediates that precede pyrimidines 17 and 18. Alternatively, the core of pyridyl compound 12 was achieved by an efficient palladium-catalyzed CH annulation. , First, cyclization precursor 12b was accessed via a Mitsunobu reaction between phenol 12a and 2-(1H-imidazol-1-yl)­ethanol. Intermediate 12b was then heated in the presence of Pd­(OAc)₂, tri­(furan-2-yl)­phosphine, and a carbonate base in DMF to deliver chloro-tricycle 12c. This annulation chemistry was applied similarly to furnish the tricyclic scaffolds of pyridines 10, 11, and 13–16. The difluoro ether framework present in compound 21 was accessed by first subjecting phenol 21a to an addition reaction with difluorovinyl tosylate 25 facilitated by KOH. From there, a Suzuki coupling with boronic acid 23 afforded cyclization precursor 21b. Trityl group removal using in-situ-generated methanolic HCl and subsequent base-mediated intramolecular S_N2 displacement revealed chloro-tricycle 21c. The tricyclic cores for difluoro ether analogues 19, 20, and 22 were accessed in a similar manner. All penultimate chloro-heterocycle intermediates were subjected to a final Suzuki coupling with N-cyclopropylsulfonyl pyrazole boronate 24 to produce the CD38 inhibitors discussed above.

1. Synthesis of Compound 9, Representative of Compounds 17 and 18 .

^a (i) Ph₃P, DIAD, 2-((tert-butyldimethylsilyl)­oxy)­ethanol, THF/toluene, 0 °C. (ii) PdCl₂(dppf)•CH₂Cl₂, K₂CO₃, 23, dioxane/H₂O, 110 °C. (iii) 2 M HCl/MeOH, 70 °C. (iv) TsCl, pyridine, 5 °C. (v) 60% NaH, THF, rt. (vi) PdCl₂(dppf)•CH₂Cl₂, K₂CO₃, 24, dioxane/H₂O, 110 °C.

2. Synthesis of Compound 12, Representative of Compounds 10, 11, and 13–16 .

^a (i) Ph₃P, DIAD, 2-(1H-imidazol-1-yl)­ethanol, THF/toluene, 0 °C. (ii) Pd­(OAc)₂, tri­(furan-2-yl)­phosphine, K₂CO₃, DMF, 110 °C. (iii) XPhos PdG3, Cs₂CO₃, 24, THF/H₂O, 100 °C.

3. Synthesis of Compound 21, Representative of Compounds 19, 20, and 22 .

^a (i) KOH, 25, MeCN/H₂O, 70 °C. (ii) XPhos PdG3, Cs₂CO₃, 23, dioxane/H₂O, 100 °C. (iii) AcCl, MeOH, 50 °C. (iv) Cs₂CO₃, DMF, 60 °C. (v) XPhos PdG3, Cs₂CO₃, 24, THF/H₂O, 100 °C.

In summary, these studies highlighted the rational design and SAR exploration of a novel series of tricyclic CD38 inhibitors. The strategic covalent reinforcement of the bioactive imidazole conformation via a three-atom ether linkage afforded a novel series of compounds with potent mouse and human enzyme inhibition. Additionally, this design element enabled analogues that maintained low molecular weight and excellent cell permeability, while increasing molecular rigidity without introducing additional HBDs, all of which are parameters that favor BBB permeability. Compounds reported herein were all determined to have high CNS MPO scores, indicating desirable chemical space for potential CNS-active molecules. Thus, several CNS-penetrant CD38 inhibitors with modest unbound plasma clearances in mice were identified after some incremental structural modifications to adjust compound polarity and lipophilicity. This work resulted in the discovery of potent and brain-penetrant CD38 inhibitor 16, which was dosed orally in mice and correspondingly demonstrated dose-dependent increases in NAD⁺ levels, as measured in brain and lung tissue samples. These findings provide a foundation for further exploration of CD38 modulators in the context of CNS disorders and support the growing body of literature describing targeted therapeutics that leverage key physicochemical parameters for improved brain delivery and efficacy.

Glossary

ABBREVIATIONS

ADPR

adenosine diphosphate ribose

BBB

blood–brain barrier

CD38

Cluster of Differentiation 38

cLogP

calculated LogP

CL_pu

unbound plasma clearance

CNS

central nervous system

DIAD

diisopropyl azodicarboxylate

HBD

hydrogen bond donor

hCD38

human CD38

mCD38

mouse CD38

MPO

multiparameter optimization

NAD⁺

nicotinamide adenine dinucleotide

n.d.

not determined

PARP

poly­(ADP-ribose) polymerase

SAR

structure–activity relationship

SIRT

sirtuin

TPSA

total polar surface area

unb. b:p

unbound brain-to-plasma ratio

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.6c00301.

• Experimental methods, synthetic procedures, compound characterization data, supplementary figures and tables, and X-ray data collection and refinement statistics for PDB IDs 10OT and 10LP (PDF)

• NMR spectra for key compounds 6, 9, 12, 16, and 21 (PDF)

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The design and study conduct for this research were provided by AbbVie. AbbVie and Calico Life Sciences LLC participated in the interpretation of data, review, and approval of the publication. Financial support for this research was provided by AbbVie and Calico Life Sciences LLC.

No unexpected or unusually high safety hazards were encountered. All methods in the current study were performed in accordance with the ARRIVE guidelines (https://arriveguidelines.org/) and all other relevant guidelines and regulations. AbbVie is committed to ensuring the humane care and use of laboratory animals in the company’s research and development programs. Our programs exceed regulatory agency standards, and we are committed to the internationally accepted principles of 3Rs (refinement, reduction, replacement). All animal studies were performed in accordance with AbbVie’s Institutional Animal Care and Use Committee (IACUC) guidelines and the protocol was reviewed and approved by AbbVie’s IACUC (Ethics Approval ABBVLC00126). Animal studies were conducted in an AAALAC accredited program where veterinary care and oversight were provided to ensure appropriate animal care.

The authors declare the following competing financial interest(s): T.R.H., Y.J., A.O., J.T.R., R.S., B.D.B., C.B., P.J., M.M.S., A.B., L.B., R.J., N.P.B., A.K., A.M.S., and J.B.S. are employees of AbbVie. K.K., R.N.S., K.A.L., J.W.B., I.Z., and A.C. were employees of AbbVie at the time of the study. M.B. is an employee of Calico Life Sciences LLC.