Channeling Nicotinamide Phosphoribosyltransferase (NAMPT) to Address Life and Death

Abstract

Nicotinamide phosphoribosyltransferase (NAMPT) catalyzes the rate-limiting step in NAD⁺ biosynthesis via salvage of NAM formed from catabolism of NAD⁺ by proteins with NADase activity (e.g., PARPs, SIRTs, CD38). Depletion of NAD⁺ in aging, neurodegeneration, and metabolic disorders is addressed by NAD⁺ supplementation. Conversely, NAMPT inhibitors have been developed for cancer therapy: many discovered by phenotypic screening for cancer cell death have low nanomolar potency in cellular models. No NAMPT inhibitor is yet FDA-approved. The ability of inhibitors to act as NAMPT substrates may be associated with efficacy and toxicity. Some 3-pyridyl inhibitors become 4-pyridyl activators or “NAD⁺ boosters”. NAMPT positive allosteric modulators (N-PAMs) and boosters may increase enzyme activity by relieving substrate/product inhibition. Binding to a “rear channel” extending from the NAMPT active site is key for inhibitors, boosters, and N-PAMs. A deeper understanding may fulfill the potential of NAMPT ligands to regulate cellular life and death.

Significance

• A holistic perspective from orthosteric inhibitors to allosteric activators binding to the NAMPT “rear channel” as a basis for drug design for cancer, pain, diabetes, and neurodegeneration

• Caveats on metabolites/antimetabolites that may engage off-targets leading to toxicity or therapeutic benefit

• Notes on patient selection and the potential for combinations with modulators of related NAD⁺-dependent pathways

• A need for mechanistic understanding of NAMPT in cellular and extracellular compartments to underlie successful translation to the clinic

1. Introduction

Nicotinamide adenine dinucleotide (NAD⁺) is an essential enzyme cofactor. NAD⁺ and its 2-phosphate derivative (NADP⁺) are electron acceptors. Cellular processes including redox chemistry, glycolysis, metabolism, mitochondrial respiration, and bioenergetics are dependent on electron transfer from NADH and NADPH. We can conceptualize two compartments for NAD⁺. The first compartment is defined by electron transfer and the reversible interconversion of the oxidized and the reduced forms of the cofactor (e.g., NAD⁺ + e^– + H⁺ ⇌ NADH). The second is defined by the consumption and supply of NAD⁺ itself, dictated by (a) consumption by families of important enzymes with NADase activity that catabolize NAD⁺ to nicotinamide (NAM) and (b) production by the three mammalian pathways for cellular NAD⁺ biosynthesis (Figure 1A). Given the reliance of dehydrogenases and oxidoreductases on NAD⁺ cofactors, the first compartment must be protected against excessive fluctuations of NAD⁺ that may occur in the second compartment. Mitochondria represent the clearest example of a physical manifestation of the first compartment, with SLC25A51, the recently discovered mitochondrial NAD⁺ transporter, regulating mitochondrial NAD⁺.¹⁻³

Figure 1.

NAD⁺ biosynthesis, catabolism, and NAMPT enzyme mechanism. (A) The three biosynthetic pathways (kynurenine, Preiss–Handler, salvage) are shown with key enzymes and metabolites. Enzymes responsible for NAD⁺ catabolism are depicted in blue. See Figure S1 for full chemical structures of NAD⁺ biosynthesis intermediates and products. (B) Chemical reaction catalyzed by NAMPT converting NAM to NMN. (C) NAMPT inhibitor A-1326133 is a substrate for NAMPT and is converted to a PR adduct, which is an NMN analogue. (D) Structure of the archetypal NAMPT inhibitor FK866. (E) FK866 (magenta) and NAM (orange) bound to NAMPT showing the zones comprising the active site and rear channel (PDB 2GVJ, 8DSI).

NAD⁺ is a substrate for enzymes that possess NADase activity, which have in common the catalysis of ADP-ribosyltransferase (ART) reactions that use the anomeric carbon of NAD⁺ as an electrophile, releasing NAM. The action of these enzymes is diverse; however, all catabolize NAD⁺ to yield NAM. Several of the families of enzymes that catabolize NAD⁺ mediate cell signaling by regulation of protein post-translational modification (PTM), notably sirtuins (SIRTs) and poly(ADP-ribose)polymerases (PARPs).

The regulation of the cellular NAD⁺ pool is controlled by the demand for NAD⁺ counterbalanced by the NAD⁺ supply (Figure 1A). The demand for NAD⁺ is greatest in cells under high metabolic pressure, including many cancer cells that rely on glycolysis. This has stimulated efforts to develop anticancer drugs that restrict cellular NAD⁺ supply and ultimately restrict cellular ATP synthesis leading to cell death. Conversely, NAD⁺ supply is attenuated in cells under various types of pathological stress associated with aging and disease. The disruption and depletion of the NAD⁺ supply, linked to age-associated and metabolic disorders, has inspired the search for “NAD⁺-enhancing drugs” that replenish cellular NAD⁺ and may serve as therapeutics that enhance the lifespan and healthspan.⁴⁻⁷

The rate-limiting step in NAD⁺ biosynthesis, in mammals, is the salvage of NAM and conversion to nicotinamide mononucleotide (NMN), catalyzed by the enzyme nicotinamide phosphoribosyltransferase (NAMPT) (Figure 1A,B). In the liver and kidneys, de novo synthesis of NAD⁺ from tryptophan is important, with catabolism of NAD⁺ and secretion of NAM providing a source of NAM to other tissues, thus feeding the salvage pathway of NAD⁺ synthesis.⁸ To circumvent the rate-limiting step in salvage, both NMN and nicotinamide riboside (NR) have been explored as dietary supplements (Figure 1A). Recent work shows that NMN and NR are primarily metabolized to NAM in the liver, which when secreted into circulation provides the substrate for NAMPT-mediated NAD⁺ synthesis in peripheral tissues.⁸ Administration of NR and NMN was shown to have beneficial effects on the healthspan and/or lifespan in mice and humans.⁹⁻¹² For example, prediabetic postmenopausal women treated for 10 weeks with NMN showed improved muscle insulin sensitivity and insulin signaling.¹³ Recently, NMN was reported to improve “blood biological age” relative to placebo in middle-aged healthy adults.¹⁴ The status of NMN as a dietary supplement has been questioned by the FDA.¹⁵

In cancers in which NAMPT is upregulated, there is a clear justification for pursuing NAMPT inhibitors. NAMPT inhibitors have entered cancer clinical trials. Owing to the detrimental effects of NAMPT inhibition in noncancer cells, several of these clinical trials have used coadministration of nicotinic acid (NA), otherwise known as niacin. Niacin is a dietary component, a dietary supplement, and a prescription drug for certain forms of dyslipidemia. In the context of NAMPT inhibitor therapy, NA provides a source of NAD⁺, via the Preiss–Handler pathway in healthy cells, countering the effects of NAMPT inhibition in these cells but not cancer cells addicted to NAMPT for NAD⁺ biosynthesis (Figure 1A).

The potential of small molecule therapeutic agents that target NAMPT has been reviewed in the context of specific and broad disease states.¹⁶⁻²² NAMPT has also been discussed in reviews on therapeutic modulation of NAD⁺ in aging.^23,24 The focus of these therapeutic strategies has primarily been the regulation of cellular NAD⁺ levels, targeting intracellular NAMPT. Although not the focus of this perspective, and rarely discussed in the context of NAD⁺ salvage, we will need briefly to introduce extracellular NAMPT (eNAMPT).

The medicinal chemistry of NAMPT inhibitors was reviewed in 2013,²⁵ with an update provided in 2020, which further discussed the potential of these inhibitors in immunotherapy.²⁶ Our objective is to discuss the medicinal chemistry of small molecule NAMPT ligands in the context of their still untapped translational potential as human therapeutics. The goal of this current perspective is not a comprehensive, descriptive review of NAMPT inhibitors; rather the goal is to emphasize the complexity of the interactions with NAMPT that may inform drug design. The need for this review is driven by recent observations on (i) the pharmacology of inhibitors that are substrates for NAMPT and (ii) the subtle changes in inhibitor structure that lead to small molecules that increase the enzyme activity of NAMPT.

NAMPT inhibitors that are also substrates are bioactivated to analogues of NMN (Figure 1C) and NAD⁺ that may insert themselves into the cellular machinery perturbing cellular homeostasis. Thus, some NAMPT inhibitors function as classical antimetabolites. This is a serious and often unrecognized challenge in the drug design of NAMPT inhibitors. Furthermore, the regioisomeric switch from a 3-pyridyl inhibitor to a 4-pyridyl NAMPT activator has inspired the development of broader families of activators. The topological feature of NAMPT, which facilitates picomolar inhibition by 3-pyridyl (e.g., FK866 (Figure 1D)) and other inhibitors, is a rear channel accessing the NAM binding pocket (Figure 1E). This rear channel also mediates activation of NAMPT by small molecule NAMPT positive allosteric modulators (N-PAMs).

2. NAMPT Inhibition

2.1. Target Validation of NAMPT and eNAMPT

Target identification and validation are essential early steps in drug discovery and development. In contemporary studies, genetic manipulation of the target in cell cultures and animal models is commonly used for target validation.²⁷ In the context of NAMPT, genetic manipulation generally does not distinguish between the extracellular form (eNAMPT) and intracellular NAMPT that regulates NAD⁺ biosynthesis. In several disease states, including pulmonary hypertension, type 2 diabetes, inflammation, and cancer, eNAMPT has been identified as a target.^17,26,28,29 In the context of its activity as a cytokine, eNAMPT is referred to as pre-B-cell colony-enhancing factor (PBEF)^30,31 and is reported to be a ligand and agonist at Toll-like receptor 4.³² A similar function has been proposed for extracellular NAPRT.³³ As an adipokine, eNAMPT is referred to as visfatin.³⁴ In cancer, obesity, and inflammatory disease, eNAMPT has also been proposed as a disease biomarker.³⁵

Little is known about the interactions between eNAMPT and cellular NAMPT. Any crosstalk between the function of NAMPT as the rate-limiting enzyme in mammalian NAD⁺ biosynthesis and the actions of eNAMPT is ill-defined. Although not the focus of this perspective, eNAMPT must be considered in the context of genetic target validation. In at least one case, the role of NAMPT in cancer progression has been proposed as independent of its enzyme activity and dependent on eNAMPT.³⁶ We will not return to eNAMPT in this perspective.

2.2. The Archetypal “Highly Specific” NAMPT Inhibitor FK866

The archetypal NAMPT inhibitor is FK866 (APO866, daporinad; Figure 1D). In 2003, Fujisawa scientists identified FK866 from screening compounds that cause cell death in HepG2 cell cultures via apoptosis.³⁷ FK866 was identified to cause delayed depletion of cellular ATP and rapid depletion of NAD⁺, which could be rescued by the addition of NAM (10 mM). In THP-1 cells, apoptosis induced by FK866 was rescued by NA (0.1 mM) and FK866 did not inhibit NAD⁺ synthesis from NA. Thus, FK866 inhibits NAMPT and the salvage pathway and does not inhibit nicotinic acid phosphoribosyltransferase (NAPRT) and the Preiss–Handler pathway (see Figure 1A). In the measurement of NAMPT activity in HepG2 and K-562 cell lysates, it was shown that FK866 inhibited cellular NAD⁺ biosynthesis with high specificity at the level of the enzyme NAMPT.

It was concluded that FK866 acts as a noncompetitive NAMPT inhibitor binding at an allosteric site with K_I = 0.3–0.4 nM, and it was speculated that FK866 binds to NAMPT as an analogue of NAD⁺. This hypothesis was reasonable given that NAD⁺ itself is a NAMPT inhibitor. However, all recent work shows that FK866 is competitive with NAM/NMN and binds to the nucleobase pocket of the NAMPT active site competing for binding with substrate NAM (Figure 1E). In the context of the Preiss–Handler and salvage pathways, FK866 is specific for inhibition of NAMPT and the salvage pathway.

The pharmacokinetics of FK866 after iv infusion have been reported from clinical trials: C_SS = 14 nM (at MTD). After the termination of infusion, the drug rapidly clears.³⁸ Delivered iv to mice, T_1/2 ≈ 50 min, C_max = 14 μM (at 10 mg/kg) and 25 metabolites were identified including the N-oxide that was also measured in human trials.³⁹ Although the activity of the N-oxide metabolite has not been reported, rapid clearance likely translates to little or no contribution to the in vivo activity of FK866. In toxicodynamic studies in rats, FK866 (ip up to 60 mg/kg bd) in plasma reached C_max = 6.7 μM.⁴⁰ In other preclinical animal studies, FK866 has been dosed ip (e.g., 0.5–25 mg/kg qd).⁴¹⁻⁴³ Although pharmacokinetics data are rarely provided, the reliable effects of this dosing in regulating NAMPT activity and phenotype suggests adequate drug exposure. In general, the data support use of the prototypical NAMPT inhibitor, FK866, as a chemical probe both in vitro and in vivo. For in vivo studies it would always be preferable to measure plasma exposure to ensure submicromolar concentrations compatible with the low nanomolar potency of FK866. In this respect, FK866 is a pharmacological equivalent to genetic NAMPT silencing for use in NAMPT target validation.

2.3. Genetic and Pharmacological Target Validation

NAMPT has been indicated as a target in many cancers, for example, thyroid,⁴⁴ lymphoma,⁴⁵ colorectal,⁴⁶ gastric,⁴⁷ prostate,⁴⁸ endometrial,⁴⁹ breast,^50,51 and ovarian.⁵² A recent study combined both genetic validation and pharmacological validation, the latter using FK866. Chowdhry et al. studied >7000 tumors and 2600 matched normal tissue samples representing 19 cancer/tissue types.²¹ In addition to patient-derived tissues, 54 cancer cell lines were studied and compared using silencing of genes associated with the Preiss–Handler pathway (PH) or salvage pathway. This study set out to establish the dependence of NAD⁺ supply in tumors on PH (and associated NAPRT and NADSYN genes) versus the salvage pathway (and associated NAMPT and NMRK1 genes) (Figure 1A). The study concluded that cancer cell survival is either PH dependent or salvage dependent, whereas normal cells can use all biosynthetic pathways. Overexpression of PH genes was shown to drive “addiction” of PH-dependent cancer cells.

PH-dependent and salvage-dependent cell lines would be expected to be sensitive to silencing of NAPRT and NAMPT, respectively. In Figure 2, we replot data from Chowdhry et al. to illustrate graphically the segregation of cancer cell lines into salvage dependent and PH dependent. Salvage-dependent cells are vulnerable to both siNAMPT and FK866, whereas PH-dependent cells are invulnerable to both siNAMPT and FK866 (Figure 2A). Conversely, silencing of the essential PH gene, NAPRT, causes cell death only in cell lines that are insensitive to FK866 (Figure 2B). This finding identifies NAMPT as a target for inhibition in a specific subset of cancers that are salvage-pathway addicted. Further analysis emphasizes that the sensitivity of salvage-dependent cell lines is correlated with increased NAMPT copy number (Figure 2C).

Figure 2.

Cell killing by FK866 in NAMPT-addicted cancer cells. (A) Correlation of cell death from siNAMPT versus FK866 treatment. (B) Correlation of cell death from siNAPRT versus FK866 treatment. (C) Correlation of NAMPT amplification (copy number NAMPTc#) or expression (NAMPTex) versus cell death showing dependence of FK866 sensitivity on NAMPT copy number (c#). (D) Cell death from FK866 versus siNAMPT (black, vehicle; orange, 10 nM FK866; pink, 50 nM; red, 100 nM; R² = 0.9). Cell death measured by propidium iodide staining (Z score) data are from Chowdhry et al.²¹ and replotted. Each symbol represents a data point for an individual cell line.

The Chowdhry study²¹ supports the use of FK866 as a chemical probe selective for NAMPT, since a good correlation was observed across the 54 cell lines studied for cell death induced by FK866 treatment compared to that induced by NAMPT silencing (Figure 2D). Off-target, cytotoxic effects of FK866 are not seen in this study (Figure 2D).

In PH-dependent cells, supplementation with NAMN or NAD⁺ should be able to rescue the effects of NAPRT silencing. Similarly, supplementation of salvage-dependent cells with NR, NMN, or NAD⁺ should be able to rescue cells from NAMPT silencing. NR and NMN were observed only partially to rescue salvage-dependent cells from NAMPT silencing.²¹ NAD⁺ synthesis from NR requires phosphorylation of NR by the enzyme NMRK to yield NMN (see Figure 1A). In a salvage-dependent xenograft, NMRK1 silencing had no effect on NAD⁺ levels or tumor volume; however, in combination, NMRK1 and NAMPT silencing significantly enhanced cell death relative to NAMPT silencing alone.²¹ These experiments emphasize that cancer cells may survive and escape NAMPT inhibition either by using cellular NAD⁺ produced by the PH pathway or via NMRK-mediated NMN synthesis. Therefore, NAD⁺ biosynthesis via PH or NMRK may circumvent NAMPT inhibition and limit the therapeutic efficacy of NAMPT inhibitors.

2.4. NAMPT Inhibition and Synthetic Lethality

A 2012 report of the synthetic lethality of NAMPT inhibition came from a study using siRNA to silence NAMPT in triple-negative breast cancer (TNBC) cell lines.⁵³ Cells were screened for synthetic lethality in response to a focused siRNA library covering 44 genes associated with PARP activity and NAD⁺ metabolism. Screening was conducted in the presence of the PARP1/2 inhibitor olaparib. NAMPT was identified as a unique and significant, nonredundant modifier of response to olaparib. The predicted synthetic lethality in this study was corroborated in TNBC cell lines with the combination of olaparib and the NAMPT inhibitor FK866. Olaparib received FDA approval in 2014 and is now indicated for breast, pancreatic, ovarian, and peritoneal cancers. In most cancers, clinical use is limited to patients with BRCA mutations since the mechanism of action of PARP inhibitors is associated with blocking of DNA repair in cancer cells and BRCA mutations lead to deficiency in homologous recombination (HR) repair. In high-grade serous carcinoma, the most fatal ovarian cancer, another PARP inhibitor, niraparib, is now being used independent of BRCA status.⁵⁴

The potential advantages of a combination of PARP inhibitor with NAMPT inhibitor is synthetic lethality in the absence of BRCA mutations, which would significantly expand clinical use in TNBC and other cancers. Inhibition of NAMPT reduces the supply of intracellular NMN and NAD⁺, thereby restricting the ability of PARP to PARylate protein targets while the PARP inhibitor competes with NAD⁺ for binding to the PARP active site.

Following up on the report of synthetic lethality in TNBC, a recent study demonstrated synergistic interactions between olaparib and FK866 in multiple preclinical models of PARP-resistant and naïve ovarian cancer.⁵⁵ Several cell lines with acquired and de novo resistance were shown to have a 400-fold loss of potency toward olaparib relative to naïve ovarian cancer cell lines in clonogenic and proliferation assays. In all cell lines, only the combination of olaparib with FK866 gave significant antiproliferative activity that was fully or partially abrogated by the addition of NMN (0.5 mM). It was concluded that FK866 synergistically sensitizes cells to the DNA damaging effect of PARP inhibitors inducing apoptosis and cell death. Tumorigenesis was significantly attenuated in a mouse xenograft study with ip administration of olaparib (50 mg/kg) and FK866 (5 mg/kg), whereas monotherapy showed no efficacy. It is interesting that the combination induced no overt signs of toxicity or weight loss in mice. Two other NAMPT inhibitors, OT-82 and KPT-9274 (vide infra), showed efficacy in one cell line when substituting for FK866.

2.5. Pharmacological NAMPT Inhibition

The medicinal chemistry of NAMPT inhibitors was comprehensively reviewed in 2013,²⁵ with an update provided in 2020.²⁶ These reviewers concluded that, where cellular and enzymatic potency matched, inhibitors were “rather specific” for NAMPT.²⁵ FK866 and many other chemical families of NAMPT inhibitors have been reported with matching enzymatic and cellular potencies. In many cases, potency is single digit nanomolar. At submicromolar concentrations in cells, these NAMPT inhibitors will be highly selective for NAMPT inhibition in vitro and in vivo. Of further reassurance would be the measurement of plasma and/or tissue concentrations of NAMPT inhibitors to confirm target exposure at concentrations compatible with selectivity for NAMPT (i.e., submicromolar concentrations).

Although a cocrystal structure, per se, does not confirm target selectivity, cocrystal structures do confirm target binding. In the absence of such a structure, rigorous determination of binding affinity is required using an unambiguous biophysical method such as isothermal titration calorimetry (ITC). Over 70 structures of NAMPT are reported in the PDB database, and many of these are structures of the human enzyme complexed with bound ligands, including substrates and inhibitors. A compilation of reported NAMPT inhibitors with (Table S1) and without (Table S2) cocrystal structures is provided in the Supporting Information. The tabulated inhibitors are categorized by the warhead that binds to the nucleobase pocket, which in the majority of examples is an N-heterocycle. Where available, data are also tabulated in the Supporting Information for enzyme inhibition potency and potency for growth inhibition in cell lines.

In 2006, the cocrystal structure of NAMPT-bound FK866 was reported, demonstrating that FK866 binds competitively with NMN/NAM at the nucleobase pocket (Table S1, entry 4; Figure 1E).^56,57 The key structural features responsible for substrate and inhibitor binding are (1) the active site, (2) the nucleobase pocket, and (3) the “rear channel” (Figure 1E). The simple explanation for the specificity of NAMPT inhibitors for NAMPT over NAPRT is that there is no rear channel in NAPRT for inhibitors to bind (Figure 3A).⁵⁸

Figure 3.

NAMPT structural biology. (A) Comparison of the structures of NAPRT (PDB 4YUB) and NAMPT with FK866 (PDB 2GVJ) showing the homodimeric structures with active sites at the interface. The active sites are highlighted by showing active site residues in enhanced colors and docking NAM at the NAPRT active site. Both the depth of the rear channel and ready access to the active site are shown in the NAMPT structure in contrast to NAPRT. (B) H-bonding network (yellow dashes) at the doorsill between the nucleobase pocket and rear channel of NAMPT showing bound NAM (blue) and a network of three to four water molecules (in addition to three conserved water molecules, a fourth is often observed in crystal structures, which does not H-bond with NAM or NAMPT residues). (C) GMX1778 bound to NAMPT showing replacement of the water molecules by direct H-bonding of the ligand with residues in the doorsill (PDB 4O12). (D) Crystal structure of NAMPT active site and rear channel with four different cocrystallized inhibitors superposed to demonstrate the variety of warheads accommodated and the larger variety of cap groups accommodated in the rear channel (PDB 2GVJ, 6AZJ, 4LVA, 4KFN).

2.6. NAMPT Rear Channel: Inhibitors

The rear channel is a unique feature of NAMPT. At the doorsill between the nucleobase pocket and the rear channel is a H-bonding network with three conserved water molecules interacting with Ser-275, Asp-219, Ser-241, and Val-242 (Figures 1E and 3B). Binding of FK866 displaces two of the three water molecules, with the water interactions being replaced by H-bonding with the amide group of FK866. Among other inhibitors, the 4-amino-3-vinylpyridine NAMPT inhibitor, KPT-9274 (padnasertib; Table S1, entry 17; Scheme 1), binds similarly to FK866, also retaining one water molecule H-bonded to Asp-219, Ser-241, and the backbone carbonyl of Val-242 (PDB 5NSD).⁵⁹ In the structures of a 3-pyridyl-azetidine class and a cyanoguanidine class of NAMPT inhibitors, all three water molecules are displaced leaving direct H-bonds with Ser-275 and Glu-219 (Table S1, entry 11; PDB 6PEB; Figure 3C).⁶⁰

Scheme 1.

2.6.1. NAMPT Inhibitor Design

NAMPT catalyzes the ATP-dependent turnover of NAM to NMN. The three structural components relevant to inhibitor design are the active site, nucleobase pocket, and rear channel (Figure 1E). A large structural variety of NAMPT inhibitors show low nanomolar potency in enzyme assays; for example, the 3-pyridyl inhibitor compound 3 in the azetidine series (PDB 6PEB; Table S1, entry 11; Scheme 1) has biochemical IC₅₀ = 2.7 nM and for inhibition of A2780 cell growth IC₅₀ = 5 nM.⁶⁰ We note that other weaker NAMPT inhibitors, such as KPT-9274, maintain submicromolar potency (KPT-9274 biochemical IC₅₀ = 120 nM; from Caki-1 cell viability IC₅₀ = 600 nM).⁶¹ The 10-fold loss of potency, often assumed by medicinal chemists, going from biochemical to cell-based assays is rarely observed for NAMPT inhibitors.

Assessment of the available crystal structures suggests that optimal engagement of the nucleobase pocket while maintaining the Ser-275 H-bond allows diversity in inhibitor design. Structural variety is seen in NAMPT inhibitors, with the rear channel of NAMPT accommodating more structural diversity than the nucleobase pocket. Since NAPRT does not possess a rear channel, all potent NAMPT inhibitors are specific for NAMPT versus NAPRT.⁵⁸

Examples of the effective replacement of the 3-vinylpyridine warhead of FK866 are the 3-pyridyl-methylenethiourea and comparable urea analogues introduced by Zheng et al. in 2013,⁶² notably compound 50 (Table S2, entry 6; IC₅₀ = 7 nM for NAMPT; Scheme 1). The ureido linker perfectly replicates the H-bonding interactions of the vinylamides (such as FK866). Compound 5 (Table S1, entry 12; Scheme 1), a thiourea, inhibited NAMPT with IC₅₀ = 7 nM; however, the 4-pyridyl positional isomer, compound 6, was reported to have no activity.⁶² The urea series was able to accommodate a large number of changes in the warhead, including aniline and 2- and 3-fluoroaryls, while maintaining double-digit nanomolar potency. A number of modifications of the tail group led to potent NAMPT inhibitors with no cellular activity. The authors stated that a “sufficiently nucleophilic nitrogen atom” in the warhead was needed for potent cellular activity; however, we note that a nucleophilic nitrogen atom is not needed for NAMPT inhibition. The development candidate optimized in this work, compound 50 (Scheme 1), is a typical 3-pyridyl NAMPT inhibitor with cellular antiproliferative potency (IC₅₀ = 32 nM) and oral bioavailability in mice (F = 26%) (Table S2, entry 6). We will return to the Zheng et al. paper in section 3.1.

KPT-9274 is a 3-vinylpyridine NAMPT inhibitor (Scheme 1) currently in clinical trials for cancer (NCT04914845). KPT-9274 is an interesting compound with an extensive literature. KPT-9274 has been described as both a highly specific NAMPT inhibitor and a dual and specific NAMPT and PAK4 inhibitor.^61,63 It is also noted that in several papers the focus is largely or entirely on either PAK4 inhibition^64,65 or NAMPT inhibition.^63,66 As noted above (and in Table S1), KPT-9274 is a moderately potent 3-vinylpyridine NAMPT inhibitor, whereas interactions with PAK4 are more complex. An early report of the compound in 2015 reports anticancer activity in cell lines, primary human acute myeloid leukemia (AML) cultures, and an AML xenograft model (at 150 mg/kg po).⁶⁷ In this paper, KPT-9274 is described exclusively as a PAK4 allosteric modulator. Another early report introduces the development rationale for such agents as potential breast cancer therapeutics.⁶⁸ KPT-9274 reduces levels of PAK4 in cells, with the mechanism supported by a 70-fold right shift in cytotoxic activity when PAK4 is knocked out by CRISPR-Cas9.⁶¹

It is important to emphasize that KPT-9274 is not a PAK4 kinase inhibitor but selectively inhibits PAK4 downstream signaling. A study comparing KPT-9274 with an authentic PAK4 kinase inhibitor, PF-3758309, is highlighted.⁶⁹ KPT-9274 was studied among other 3-vinylpyridines, all described as PAK4 modulators.⁶⁴ This study focused on the discussion of KPT-9274 as an anticancer agent, with the authors stating that NAMPT activity would be examined in the future, based on the contemporaneous report of PAK4/NAMPT dual activity.⁶¹ It was speculated that PAK4/NAMPT dual targeting was beneficial, because NAMPT was reported to activate Cdc42 upstream of PAK4.⁶⁴ More fascinating was a paper on targeting PAK4 with KPT-9274 in multiple myeloma, in which it was stated that NAMPT formed a complex with the PAK binding domain, independent of Cdc42, which was disrupted by KPT-9274.⁷⁰ This study proposed inhibition of ERK and mTOR signaling as a NAMPT-dependent contributor to KPT-9274 activity. More recently, PAK4 overexpression in cancer cells was shown to be a driver of mTOR signaling and implicated in the KPT-9274 mechanism of action.⁷¹

Impressive data continue to be published on KPT-9274. The compound was reported to increase sensitivity to DNA-damaging agents in Waldenstrom macroglobulinemia, a form of non-Hodgkin lymphoma.⁷² The combination with checkpoint inhibitors has been studied in a renal cell carcinoma model.⁷³ The combination of KPT-9274 with KRAS(G12C) inhibitors, sotorasib and adagrasib, was studied in pancreatic ductal adenocarcinoma (PDAC) models with the conclusion that this strategy should be tested in PDAC patients who do not respond to KRAS inhibitors.⁶⁵ This paper did not mention NAMPT as a contributor. The compound was also reported to potentiate gemcitabine activity in PDAC.⁷⁴ HDAC inhibitors were shown to confer synthetic lethality to KPT-9274 with the combination having minimal toxicity toward normal cells.⁷⁵ Finally, KPT-9274 was explored in autosomal dominant polycystic kidney disease (ADPKD), showing no toxicity in the ADPKD animal model used.⁷⁶ The first-in-human study, dosing patients with solid tumors, alone or in combination with NA, was first reported in 2017.⁷⁷

2.6.2. NAMPT Inhibitor Discovery and Development

Inhibitors described in this section were discovered serendipitously. The archetypal NAMPT inhibitor, FK866, was discovered in a screen for cancer cell death. Perhaps surprisingly, NAMPT inhibitors continue to be discovered through phenotypic screening and target deconvolution. CB30865 (Table S2, entry 3; Scheme 1) was developed as a thymidylate synthase inhibitor and cytotoxic agent. The poor SAR for thymidylate synthase led to a chemoproteomics study that identified NAMPT as a target.^78,79 CB30865 contains a characteristic 3-pyridyl group with the appropriately positioned amide to interact with Ser-275. To facilitate affinity pulldown, a derivative with water-soluble and clickable propargylamine was synthesized (MPI-0479883). Given the high potency of NAMPT inhibition subsequently measured for MPI-0479883 (IC₅₀ = 0.23 nM; HCT116 EC₅₀ = 0.68 nM) and the presence of a classical NAMPT pharmacophore, the veracity of NAMPT target identification is without doubt. CB30865 lowered cellular NAD⁺, and downstream PARylation was studied to measure the cellular effect of starving PARPs of NAD⁺. Subsequent optimization was reported, although the most potent NAMPT inhibitors were not significantly better than the original probe, MPI-0479883. The downstream inhibition of PARylation was reported to correlate with cytotoxicity. In 2010–2011, preclinical data on MPI-0486348 (renamed MPC-9528) with oral bioavailability (F = 47%) was reported at a conference.⁸⁰⁻⁸³

A cytotoxicity screen led to identification of a hit, OT-1901, that was subsequently optimized (although no SAR or other details are published) to yield OT-82 (Scheme 2). NAMPT was identified as a target of OT-82 by affinity chromatography, although in this work LC-MS/MS proteomics was not conducted; instead, a 55 kDa band was identified by staining and confirmed to be NAMPT.⁸⁴ That NAMPT is a target for OT-82 is not immediately obvious. In the absence of a crystal structure, we can speculate that either the 3-propylpyrazole or 4-pyridyl groups might bind in the nucleobase pocket (Table S2, entry 26). Inhibition of recombinant NAMPT by OT-82 with IC₅₀ = 41 nM was reported. We will revisit OT-82 in consideration of clinical trial outcomes for NAMPT inhibitors in section 2.8.

Scheme 2.

Chidamide/tucidinostat (Table S2, entry 25; Scheme 2) is an HDAC inhibitor that received approval in 2015 in China for relapsed or refractory peripheral T-cell lymphoma and further approval for similar clinical use in Japan in 2021. In 2020, it was reported that NAMPT is a new target for this drug.⁸⁵ Chidamide is an accidental hybrid drug that contains the 3-vinylpyridine warhead archetypal of NAMPT inhibitors and a benzamide HDAC warhead. It is a modest class I HDAC inhibitor (0.1 < IC₅₀ < 0.3 μM) and an extremely weak NAMPT inhibitor (IC₅₀ = 2.1 μM). Given the rapid metabolic deactivation of pyridyl NAMPT inhibitors by N-oxidation, it is not confirmed that this activity contributes to clinical efficacy. At 30 μM in HCT116 cells, a 50% reduction in NAD⁺ was reported.⁸⁵

Hybrid HDAC inhibitors are very popular in academic drug discovery,⁸⁶ in part, because of the ease of construction with benzamide or hydroxamate metal-chelating warheads that confer inhibition of metalloproteases, including HDACs. Both 3-pyridylthiourea (Table S2, compound 35, entry 17)⁸⁷ and imidazolyl (Table S2, compound 7f, entry 21; Scheme 2)⁸⁸ dual HDAC/NAMPT inhibitors have been reported with potency for NAMPT inhibition 100-fold better than that of chidamide. Conversely, design of hybrid NAMPT inhibitors is facilitated by the ready incorporation of a 3-vinylpyridine warhead as reported in a dual NQ01/NAMPT inhibitor (Table S2, entry 28).⁸⁹

CHS-828/GMX1778 (Table S1, entry 40; Scheme 2) was first reported in 1997 as an antiproliferative agent in three cancer cell lines and a xenograft study.⁹⁰ GMX1778 is a close analogue of a compound identified in an in vivo tumoricidal screen of antihypertensive compounds by Leo Pharmaceutical Products. GMX1778 contains a 4-pyridyl warhead conjugated with a relatively bulky cyanoguanidine group. Interestingly, in the first report, 3-pyridyl congeners showed no antiproliferative activity. GMX1778 entered clinical trials and was reported to be selective for cancer cells over normal cells, with a good efficacy/toxicity profile in xenografts delivered po, and a novel mechanism of action.⁹¹ In 2008, GMX1778 was shown to lower the concentration of cellular NAD⁺.⁹² Subsequently, Watson et al. demonstrated GMX1778 to inhibit NAMPT (IC₅₀ < 25 nM).⁹³

No crystal structure was reported for GMX1778 or later potent cytotoxic compounds that were either optimized based on the same pharmacophore or replaced the cyanoguanidine group with a squaric acid derivative. Inhibition of NAMPT activity using cell lysates and a radiometric assay indicated potent NAMPT inhibition.⁹⁴ Cyanoguanidine inhibitors were pursued by a different research team leading to an inhibitor with IC₅₀ = 4.4 nM for recombinant NAMPT and informative cocrystal structures of GMX1778 and compound 15 (Table S1, entry 41; Scheme 2). A similar binding mode to 3-pyridyl inhibitors is seen with the replacement of all three waters of the H-bond network with direct H-bonding of the ligand to Asp-219 and Ser-275 (Figure 3B).⁹⁵

2.6.3. Inhibitors That Are Also Substrates for NAMPT

In the absence of a crystal structure for CHS-828/GMX1778 and unable to determine binding by ITC, a GMX1778 fluorescence probe was used to prove NAMPT binding, estimating K_d = 120 nM.⁹³ Mutation of G217R in the NAMPT rear channel was used to support the binding mode of GMX1778. Most interestingly, the presence of a phosphoribosylated (PR) adduct was indicated after incubation with NAMPT. The data were generally supportive of GMX1778 being a substrate for NAMPT leading to a PR adduct that was a more potent NAMPT inhibitor.

A detailed characterization of NAMPT inhibitors that act as substrates was provided by Zheng et al. PR adducts were identified for GMX1778, compound 50, and other inhibitors with nitrogenous warheads.⁶² Adduct formation was confirmed by crystal structures of PR adducts bound to NAMPT (Table S1, entries 15, 19, 32).⁹⁶⁻⁹⁸ Not all inhibitors that formed PR adducts were active in cell cultures, which led to questioning of the hypothesis that the formation of PR adducts was important for high cellular potency. To test this hypothesis, a pair of PR adduct forming imidazopyridine inhibitors were compared to their carbocyclic, indolizine analogues (Scheme 3). All inhibitors showed good MDCK cell permeability and NAMPT potency (IC₅₀ < 80 nM). The indolizines had no activity in cell cultures, whereas both imidazopyridines were active. This would confirm the hypothesis that PR-adduct formation for imidazopyridine inhibitors is essential for cell-based activity.

Scheme 3.

In a more detailed analysis, one imidazopyridine (GNE-643; Table S1, entry 20; Scheme 3) was relatively less potent as an antiproliferative agent. This observation was explained by the PR–GNE-643 adduct being a weaker NAMPT inhibitor than its parent GNE-643, whereas the PR–GNE-617 adduct (Table S1, entry 19; Figure 4A) was equipotent with its precursor GNE-617 (Table S1, entry 18). This explanation was supported by time dependence of biochemical NAMPT inhibition, the amount of adduct formation, and the analysis of adduct binding interactions in crystal structures. The conclusion was that PR–GNE-643 was unable to form ideal interactions with the active site and rear channel.⁹⁹

Figure 4.

NAMPT substrate and substrate/inhibitors. (A) The enzymic conversion of NAM via NMN to NAD⁺. (B) PR-adduct formation for N-heterocyclic NAMPT inhibitors exemplified by as GNE-617. (C) Cartoon representation of the structural elements of the NAMPT homodimer. (D) The ATPase reaction phosphorylates His-247, leading to PRPP binding and an active site primed for phosphoribosyl transfer. (E) NAM binding leads to product formation. (F) Inhibitor cocrystal structures are of the apoenzyme. (G) Inhibitor binding to the primed active site can lead to phosphoribosylation to give a PR adduct. (H) Several crystal structures of PR adducts bound to the apoenzyme have been reported. (I–N) Clamps extracted from crystal structures, with substrate/inhibitor truncated to the N-heterocycle for clarity: (I) NAM (PDB 8DSD); (J) NMN (PDB 2H3D); (K) PR–GNE-617 (PDB 4L4L); (L) adenine in alternate clamp (PDB 8DSH); PR–SAR154782 (PDB 5LX5) in π-clamp (M) and alternate clamp configurations (N).

A subsequent paper from an AbbVie research team contradicted the PR-adduct hypothesis, although a similar approach was used in comparing pairs of adduct-forming azaisoindoline NAMPT inhibitors with their carbocyclic isoindoline analogues. Inhibitors tested included the A-1326133/A-1293201 pair for which crystal structures were reported (Table S1, entries 27, 36; Scheme 3).¹⁰⁰ Of 32 pairs, the majority of isoindolines inhibited PC3 cell growth with submicromolar potency, although, with most pairs, the cellular potency was weaker with the isoindoline. The conclusion is that formation of a PR adduct is clearly not necessary and sufficient for cellular NAMPT inhibition. However, inhibitors that can form PR adducts are usually more potent in cells.

Further study, using cellular washout studies, showed that the azaisoindoline A-1326133 was more potent in reducing both cellular NAD⁺ and ATP and caused delayed recovery of cellular NAD⁺ levels after washout. Again, not all PR-adduct-forming NAMPT inhibitors showed evidence of cellular retention. It is noted that GMX1778 formed a PR adduct that was not identified in cells and FK866 did not form a PR adduct, consistent with other studies. In conclusion, PR-adduct formation may increase cellular potency; however, this will depend on the inhibitor characteristics and is also likely to be cell and context dependent.

Although of academic interest, biologically active metabolites cause complexity in preclinical drug development and clinical trials because of species-dependent bioactivation mechanisms and the need to track and quantify active metabolites. Carbocyclic NAMPT inhibitors that cannot form PR adducts are expected to have simpler DMPK properties. The AbbVie team clearly viewed the simplified ADMET of the isoindoline series to be persuasive, and the in vivo efficacy of the isoindoline matched that of the azaisoindoline, A-1326133, albeit at twice the oral dose. We should also add that phase 1 metabolism is a factor that influences mitigation of DMPK liabilities in the design of NAMPT inhibitors as in all drug classes, elegantly exemplified in the creation of the spirocyclic piperidine class of inhibitors by the Genentech group (see Table S1, entry 21; Table S2, entry 14), which bypass CYP metabolism.¹⁰¹

A further study from the Genentech group illustrates differences between different chemical classes of NAMPT inhibitors.⁹⁷ Cell lines were treated with the potent imidazopyridine NAMPT inhibitor GNE-618 (Table S1, entry 32; Scheme 3) to evolve NAMPT escape mutants. These mutations (G217R, D93-del, G217A, G217V, S165F/Y) were incorporated in recombinant NAMPT. In cell-based and cell-free assays, NAMPT mutants did not show cross-inhibition with other NAMPT inhibitors. Pyrazolopyridine and imidazopyridine inhibitors were compared to FK866 and GMX1778. FK866 and GMX1778 suffered little loss of potency with all NAMPT mutants, except G217R. This study emphasizes that different NAMPT inhibitor chemotypes may display different phenotypes.

2.6.4. Inhibitor Summary: Part 1

We have described target identification and validation for NAMPT in cancer. There is justification for NAMPT inhibitors in specific cancers that are dependent on NAMPT for NAD⁺ biosynthesis. Synthetic lethality in combination with other modalities, such as HDAC and PARP inhibitors, may represent new approaches to cancer therapy. We have encountered a great structural variety of NAMPT inhibitors with single/double digit nanomolar potency in cells and illustrated their common interactions with the active site and rear channel of NAMPT. Despite the common binding mode, differences are observed with respect to NAMPT mutants and most markedly for inhibitors that are also NAMPT substrates.

The NAMPT active site primes the anomeric carbon of PRPP for reaction with NAM pyridyl nitrogen to yield NMN (Figure 4B–D). Presumably, a NAMPT inhibitor that aligns a warhead nitrogen to act as a nucleophile will generate a PR adduct (as shown in Figure 4A,G,H). There is no algorithm to predict which inhibitors will act as NAMPT substrates, forming PR adducts. It is possible that all inhibitors with N-heterocycle warheads act as substrates, with some reversibly reacting with PRPP to form transition state analogues. This would be compatible with the very high inhibitory potency. The dozens of cocrystal structures of inhibitors and PR adducts show binding to the dimeric NAMPT apoenzyme (Figure 4F,H). Since the interactions with the primed active site (Figure 4D,G) control PR adduct formation, the apoenzyme structures may not reveal the binding pose required for phosphoribosylation and adduct formation.

An interesting feature of the active site is the Phe-193/Tyr-18/Arg-311 π-clamp (Figure 4C,I–N). The nucleobase pocket must accommodate binding of nicotinamide (NAM, NMN) and adenine rings (ATP, ADP), because NAMPT catalysis is driven by an ATPase reaction and binding of ATP. The orientation of adenine in the pocket is rotated approximately 100° with respect to the orientation of NAM. NAM and NMN are bound in a Phe-193/Tyr-18 π–π clamp (with an Arg-311 edge interaction): see Figure 4, parts I and J, respectively. Adenine binds in an orthogonal Phe-193/Arg-311 π-cation clamp (with Tyr-18 edge interactions Figure 4L). Analysis of the structures of potent NAMPT inhibitors demonstrates that all bind with the “warhead” inserted in the Phe-193/Tyr-18 π–π clamp. The PR adduct PR–GNE-617 (Figure 4A) binds identically to NAM and NMN (Figure 4K). As expected, the 5-amino-3-pyridylyurea inhibitor (SAR154782; Table S1, entry 14; Scheme 3) binds in the π-clamp. However, crystal structures show that the PR adduct (PR–SAR154782, Table S1, entry 15) can bind to the π–π clamp (Figure 4M) or the alternate π–cation clamp (Figure 4N). A predictive framework for PR-adduct formation is lacking.

2.7. NAMPT Inhibitors That Are NAMPT Substrates and Rat Poison

The PR adduct that is formed by phosphoribosylation of a NAMPT inhibitor is an analogue of NMN (Figure 4A,B). Theoretically, NMN analogues that bind in the active site of NMNAT may act as substrates for NMNAT to be converted to analogues of NAD⁺. Again, theoretically, these NAD⁺ analogues can be substrates for PARPs and other enzymes using NAD⁺ as substrate. This concept has been exploited in a clever use of chemical tags.¹⁰² The designed chemical tags are incorporated in the NAD⁺ biosynthetic pathway to provide NAD⁺ analogues that PARPs use in the polyadenosine-diphosphate ribosylation PTM termed “PARylation”. This approach allows identification of protein PARylation using the designed chemical tags.¹⁰² Incorporation in the NAD⁺ biosynthetic pathway has been exploited accidentally by the rat poison Vacor.

In 2018, Vacor (Table S2, entry 20) was shown to be a NAMPT inhibitor and substrate (Figure 5A), providing a rationale for the neurotoxicity of Vacor compatible with earlier theories.¹⁰³ Human poisoning by Vacor, an erstwhile rat poison, leads to type 1 diabetes and neurotoxicity, and it can be lethal. Vacor can act as a diabetogenic agent in rodents, destroying pancreatic β-cells, with one proposed mechanism being depletion of cellular NAD⁺.¹⁰⁴ The diabetogenic effect was claimed to be mediated primarily via inhibition of NAD-dependent mitochondrial respiration in pancreatic islet cells, with a similar mechanism likely contributing to neurotoxicity.¹⁰⁵ Vacor intoxication can be partially treated by administration of a high concentration of NAM.

Figure 5.

Vacor and other pyridyl compounds hijack enzymes binding nicotinamides. (A) Vacor is converted to analogues of NMN, NR, and NAD⁺ (VMN, VR, and VAD⁺, respectively), which are biologically active. VMN is claimed to dominate the biological effects of Vacor. (B) VMN is argued to activate SARM1 to catabolize NAD⁺ to yield cADPR, ADPR, and NAM, leading to axonal degeneration in neurons. (C) Enzymes with NADase or ART activity, such as SARM1 and CD38, are proposed to be hijacked by N-heterocyclic drugs (including NAMPT inhibitors) leading to ADPR adducts that mediate their biological activity. (D). The CD38 inhibitor, 78c, is also a substrate for CD38, which forms an ADPR adduct.

Vacor contains a typical 3-pyridylureido NAMPT inhibitor warhead. Vacor cytotoxicity in cancer cell lines was observed to correlate with cellular NAD⁺ levels, with sensitivity varying between cell lines.¹⁰³ Cells were protected by a very high concentration of NAM (1 mM). With recombinant NAMPT and/or NMNAT2 and in cell cultures, the PR–Vacor adduct, the NMN analogue (VMN), and the NAD⁺ analogue formed from VMN (VAD⁺) were identified (Figure 5A).¹⁰³ Vacor was reported to inhibit NAMPT (IC₅₀ = 400 nM, K_I ≈ 2 μM) and act as a substrate (K_M < 2 μM). The adducts formed by sequential bioactivation by NMNAT are also NMNAT inhibitors (Figure 5A). VMN inhibited NMNAT2 with K_I ≈ 200 nM, compared with K_M = 14 μM for the endogenous substrate NMN. VAD⁺ was also shown to inhibit NAD-dependent dehydrogenases with IC₅₀ ≈ 50–100 μM. It is reasonable to anticipate that other NAMPT inhibitors may be incorporated into cellular biosynthesis to form NMN and NAD⁺ analogues that are biologically active.

In simile with observations on other NAMPT inhibitors, Vacor attenuated tumor growth in xenografts.¹⁰³ In an SH-SY5Y neuroblastoma xenograft study, the concentrations of Vacor and VAD⁺ in tumor tissues were reported to be 278 and 2.8 μM, respectively. The antitumor effects were interpreted to support a mechanism of Vacor-induced cancer cell cytotoxicity dependent on conversion of Vacor to VMN and VAD⁺, catalyzed by NAMPT and NMNAT2 (Figure 5A).

2.7.1. NAMPT Inhibitors and the Role of SARM1

SARM1 is a multidomain protein that has ADP-ribosyltransferase (ART) and NADase activity in the TIR domains that bind NAD⁺ as a substrate. Importantly, activation of SARM1 in neurons, in response to cellular stress, causes axonal degeneration and is implicated in neurodegenerative disorders and in the neurotoxicity of Vacor and speculatively other NAMPT inhibitors. Selective axonal degeneration is also a mechanism for relieving peripheral pain and neuropathy. A brief discussion of SARM1 is warranted, because of potential implications for off-target effects that may mediate toxicity or provide other therapeutic opportunities for NAMPT inhibitors.

Cryo-EM structures have provided precise information on the structure and function of the human SARM1 (hSARM1) homo-octamer.^106,107 Self-assembly of the TIF domains, required for NADase catalytic activity, is prevented in the enzymically repressed conformation. The repressed conformation is stabilized by NAD⁺ binding to the ARM allosteric sites.^106−108 NAD⁺ is also a substrate for SARM1 (K_M = 28 μM in constitutively active hSARM1^106,109). At “normal” cellular NAD⁺ concentrations, NAD⁺ binding to the allosteric site inhibits SARM1 activation, maintaining SARM1 in a repressed state (NAD⁺ IC₅₀ ≈ 500 μM).^106,107 Binding of VMN or NMN to the allosteric site derepresses SARM1. Interpretation of Vacor data has significantly contributed to the hypothesis that an increase in the ratio of NMN/NAD⁺ in neurons is necessary and sufficient to activate SARM1 leading to NAD⁺ catabolism and axonal degeneration (Figure 5B).^110,111 Vacor is commonly used as a pharmacological tool, in tandem with SARM1 silencing, in the study of axonal neurodegeneration. This research has laid the foundation for the development of new ARM-targeted pharmacological agents that inhibit SARM1 activation and axonal degeneration via negative allosteric modulation.¹¹²

2.7.2. Alternative Therapeutic Uses of NAMPT Inhibitor/Substrates

The initial work on Vacor suggested that Vacor-like chemicals could selectively induce necrosis of NMNAT2-expressing tumors, claiming superior antineoplastic activity compared to other NAMPT inhibitors.¹⁰³ These Vacor-like agents would need to be substrates for NAMPT, because NAMPT is required for the formation of VMN analogues. Subsequently, the focus has shifted from cancer to disease states, tissues, and cells associated with NMNAT2 and SARM1 expression. Work on 3-acetylpyridine was presented in the context of developing “SARM1 agonists” (SARM1 positive allosteric modulators) as selective neurolytic agents to destroy nerve axons in severe pain conditions, such as visceral cancer pain, trigeminal neuralgia, and back pain.¹¹¹ Again, such a SARM1 agonist would need to mimic NMN or be a Vacor-like substrate for NAMPT.

Chemotherapy-induced peripheral neuropathy (CIPN) is a significant clinical issue affecting up to 80% of cancer patients, and a role has been proposed for NMNAT2 and SARM1.^113,114 In neuronal cultures, axonal degeneration induced by the chemotherapeutic agents bortezomib (a pyrazine) and vincristine was prevented by SARM1 silencing or NR supplementation.⁹³ These approaches were also effective in mouse models of taxol-induced and diabetic neuropathy.^114−116 Genetic deletion or silencing of SARM1 is protective in disease models beyond peripheral neuropathy and is protective against loss or silencing of NMNAT2.^117−119 Genetic links with both SARM1 and NMNAT2 have also been reported in pediatric neuropathies and ALS (motor-neuron disease).^120−122 NMNAT2 is also a “nominated target” for upregulation to treat Alzheimer’s disease.¹²³

A4276 is a recently reported 3-pyridyl NAMPT inhibitor with a classical binding pose that appears unremarkable with relatively weak NAMPT inhibition: IC₅₀ = 492 nM (Table S1, entry 13).¹²⁴ It was compared to KPT-9274 and FK866 in a wide range of cancer cell lines. In HGC27 xenografts, it was reported to be superior to KPT-9274 and to reduce tumor NAD⁺ in contrast to KPT-9274. In early seminal papers,^125,126 it was speculated that, given [NAM] ≫ K_M(NAM), inhibitors would have a hard time competing for binding to NAMPT. Thus, weak inhibitors, such as A4276, would struggle to compete with NAM for binding to NAMPT. The second point of interest with A4276 is that it is effective in reducing axonal degeneration in retinal axons and is neuroprotective in counteracting taxol-induced CIPN in mice. In the same study, KPT-9274 and LSN3154569 (Table S2, entry 19) did not protect axons and FK866 had minimal effect. Protective effects in CIPN models are associated with NAMPT activators and were reported for activator NAT5r, discussed below.

2.7.3. Potential Pyridine Promiscuity and NAMPT Inhibitors

N-Heterocyclic compounds able to bind to NAM binding sites may act as substrates not only for NAMPT but also for other enzymes for which NAM is a substrate or product. The ART/NADase activity of SARM1 has been reported to convert “any free base”, such as 3-acetylpyridine, to its ADP-ribosyl derivative, and this enzymic reaction was reported to be dominant over NAD⁺ catabolism to ADPR or cADPR (Figure 5B).¹¹⁰ Recently 4-pyridyl derivatives (NB-1, NB-7) have been reported to act as substrates for SARM1 forming ADPR adducts (Figure 5C).¹²⁷ This work was focused on the neuroprotective actions of SARM1 in preclinical models of nerve injury and disease.

An interesting side observation was made on the “highly potent and specific” CD38 inhibitor (compound 78c), which was converted to an ADPR adduct at the active site of CD38 (Figure 5D). CD38 is a target for cancer therapy and has ART/NADase enzyme activity. The observations on compound 78c are important for two reasons: (1) compound 78c is reported as a neuroprotective agent that elevates cellular NAD⁺ and increases the healthspan and the lifespan in mice;¹²⁸ and (2) the weakly nucleophilic thiazole moiety of compound 78c, which occupies the NAM nucleobase pocket, is capable of mimicking NAM to form an ADPR adduct (Figure 5D). This would imply that, in addition to being substrates for NAMPT, the many NAMPT inhibitors that contain a weak N-heterocyclic base may potentially be substrates for enzymes with ART and NADase activity, including SARM1 and CD38, forming bioactive adducts, including PR adducts and ADPR adducts.

2.7.4. Inhibitor Summary: Part 2

Vacor is a submicromolar NAMPT inhibitor and substrate that is converted by NAMPT to a PR adduct, VMN. This adduct mimics NMN and is further converted to an NAD⁺ analogue. Both NMN and NAD⁺ analogues have been shown to bind to proteins, competing with endogenous NMN and NAD⁺, to subvert their biological activity. Observations on Vacor illustrate the complexity that can be introduced when NAMPT inhibitors are also NAMPT substrates. Similar pyridyl and N-heterocyclic compounds are reported as substrates for SARM1 and other enzymes with NADase activity, such as CD38, raising the possibility of NAMPT inhibitors being converted not only to NMN and NAD⁺ analogues but also to ADPR adducts.

Vacor and its metabolites are biologically active. Despite the multiple potential binding partners and cellular mechanisms, recent interest has focused on positive allosteric activation of SARM1 by VMN. In addition to neurotoxicity via SARM1, Vacor is toxic to pancreatic β-cells, and although these cells share phenotypic traits with neurons, the mechanism is relatively understudied. Research on Vacor illustrates expanded therapeutic potential for NAMPT inhibitor/substrates, for example in the treatment of pain. These observations should increase the prioritization of metabolite identification for NAMPT ligands, since the bioactivated metabolites may mimic NAM, NR, NMN, and NAD⁺ conferring biological activity beyond that associated with NAMPT inhibition.

2.8. NAMPT Inhibitors: Clinical Translation for Cancer

The preclinical anticancer efficacy of NAMPT inhibitors in various mouse xenograft studies has been reported for over 15 years. The lack of successful translation to the clinic is likely due to an unconvincing balance of safety and efficacy in early-stage clinical trials (for tabulation of clinical trials, see refs (19 and 129)). Failed phase 1 clinical trials often do not appear in peer-reviewed reports. Clinical data for the promising NAMPT inhibitor MPI-0486348 are unavailable, and the clinical status of KPT-9274 is unclear.^77,130 Informative reports have appeared for FK866,^38,131 GMX1775,^91,132−134 and GNE-618.¹³⁵

There are only two explanations for the lack of progress in clinical trials. First, when adverse effects and on-target, dose-limiting toxicity (DLT) are observed at drug exposure levels that are below the effective therapeutic dose, efficacy will not be achieved (i.e., DLT prevents testing of a therapeutic dose). Second, the pharmacological inhibition of NAMPT may not present an effective anti-cancer mechanism.

Dose-dependent thrombocytopenia was observed as a grade 3–4 DLT for FK866 after iv infusion,³⁸ and in refractory or relapsed cutaneous T-cell lymphoma both lymphocytopenia and thrombocytopenia were observed, with a lack of efficacy.¹³¹ Thrombocytopenia was also observed consistently for GMX1775.^132−134 These observations have not been replicated in the standard, preclinical rat model for thrombocytopenia, although the rat model was able to replicate the lower grade lymphopenia seen in clinical trials.⁴⁰ Only cytotoxicity toward colony forming unit human megakaryocytes (CFU-MK) provided a model of the thrombocytopenia observed for FK866 and GMX1778.⁴⁰ Observations on other NAMPT inhibitor chemotypes, GNE-617 and GNE-643, and the rescue of CFU-MK cells by NA indicated thrombocytopenia as a reversible, on-target effect of NAMPT inhibitors.

Cardiotoxicity and retinal toxicity, although not reported in clinical trials, were observed in rats for the same four NAMPT inhibitors studied in thrombocytopenia models. GNE-617 (30 mg/kg, po qd) caused mortality after 4 days due to cardiotoxicity.¹³⁶ On-target toxicity was supported by similar observations being made on NAMPT inhibitors with different chemistries and the ability of NA supplementation to reverse or protect against cytotoxicity through compensatory elevation of NAD⁺ via the PH pathway (although we note that NA was unable to protect fully against in vivo cardiotoxicity). Several studies support the use of NA in clinical trials to support NAD⁺ synthesis in noncancer cells and to extend the therapeutic window (right shift the cytotoxic response in healthy cells) sufficiently to achieve a therapeutic dose. The use of a biomarker to titrate NA and NAMPT inhibitor dosing in the clinic can be enabled by the use of FDG-PET imaging.¹³⁷

In preclinical safety pharmacology, FK866 induced retinal atrophy in rats but retinal changes were not seen in monkeys. The clinical trial of FK866 included baseline ophthalmic evaluation followed by repeated in-trial evaluation and ERG measurements, revealing no signs of retinopathy or visual acuity loss.³⁸ A study of FK866, GNE-617, GNE-643, GNE-875, GNE-618, and GMX1775 in human and rat retinal cells, accompanied by in vivo rat studies, concluded that retinal toxicity is (i) the result of on-target actions of NAMPT inhibitors, (ii) cannot be mitigated by NA cotreatment, and (iii) is likely translatable to humans. For example, retinal degeneration was observed after a 4–7 day treatment with GNE-617 at relatively high drug exposure levels: 2 and 6 μM in retina and plasma, respectively.¹³⁵ All inhibitors in this study contained N-heterocyclic warheads, although not all have shown evidence of PR-adduct formation.

Comparison was made of six carbocyclic NAMPT inhibitors (including A-1293201, Table S1, entry 36) with one N-heterocyclic inhibitor (A-1326133, Table S1, entry 27) in a zebrafish retinopathy study (Scheme 3).¹³⁸ A reasonable correlation was observed between potency for vision loss and antiproliferative potency in a human PC3 cancer cell line. The most potent antiproliferative agent, A-1326133, was not the most toxic; however, for this compound, retinal toxicity was reported in rats and dogs. A-1326133 is a substrate/inhibitor forming a PR adduct and was observed to accumulate in zebrafish eye tissues, which was speculated to be a result of PR-adduct formation in retinal cells associated with high NAMPT levels. No obvious safety advantage was identified with carbocyclic NAMPT inhibitors.

In mice, toxicity was observed for GMX1778 (125 mg/kg po qd); however, it was reported that, by morphological analysis, OT-82 (100 mg/kg po qd) caused neither retinal toxicity nor cardiotoxicity after 5 days treatment in mice.⁸⁴ Both compounds caused similar decreases in body weight and spleen weight. The reported plasma concentration of OT-82 (10 mg/kg po) was 1.2 μM, suggesting that OT-82 may not cause retinal toxicity or cardiotoxicity at the same exposure level as GNE-617, GMX1778, and other NAMPT inhibitors. The reasons for these positive observations are not yet fully understood.

It is possible that the failure of NAMPT inhibitors in trials has resulted from the ability of cells to replace the salvage pathway by the PH pathway and/or NMRK-mediated NAD⁺ synthesis. Certainly, the work by Chowdhry et al.,²¹ discussed above, indicates that genetic identification of cancers as addicted to the salvage pathway will assist in selecting appropriate phase 1 clinical trial cancer patients. NA, although itself not devoid of side effects, can minimize toxicity in cancer therapy. There is also scope to use NAMPT inhibitors in combination with other pathway modulators, such as NMRK inhibitors, that would block conversion of NR to NMN. The recent enthusiasm for the dual NAMPT/PAK4 inhibitor KPT-9274 is a case in point. The mechanistic differentiation of this and newer NAMPT inhibitors, such as OT-82, requires further study.

It is intriguing that the retinal toxicity shared by Vacor and some NAMPT inhibitors may be associated with NAMPT-mediated conversion to metabolites that mimic NMN. There is no evidence that FK866 is a substrate for NAMPT; however, many N-heterocyclic NAMPT inhibitors are substrates for NAMPT and may be substrates for enzymes with ART or NADase activity. These observations require detailed understanding of active metabolites and support the use of potent, carbocyclic inhibitors as NAMPT chemical probes, since they cannot form PR adducts.

Preclinical and clinical studies on NAMPT inhibitors suggest similarities between NAMPT inhibitors and FDA-approved PARP inhibitors that also have on-target toxicity. Hematologic toxicities, including anemia, neutropenia, and thrombocytopenia, are common adverse events for all PARP inhibitors, although significant differences in incidence are observed.¹³⁹ Nevertheless, PARP inhibitors are clinically effective in breast and ovarian and other gynecological cancers and are in clinical trials in prostate and other cancers. PARP inhibitors have had a significant, positive effect in the pharmacotherapy of ovarian cancer. All patients with high-grade ovarian cancer will be treated with a PARP inhibitor. It has been suggested that switching between different PARP inhibitors can be used to manage toxicity.⁵⁴ Used alone, NAMPT inhibitors must ablate cellular NAD⁺ to deplete cellular ATP in cancer cells to cause death by mechanisms including autophagy and apoptosis. The synergistic actions of combinations of NAMPT inhibitors with PARP inhibitors have been demonstrated in preclinical studies.⁵⁵ Used in combination with PARP inhibitors, depletion of NAD⁺ by NAMPT inhibition may provide an improved balance of efficacy and safety. Beyond NAD⁺-dependent enzymes, combinations with proapoptotic anticancer agents, such as BH3 mimetics, may benefit from metabolic sensitization induced by NAMPT inhibitors.¹⁴⁰

2.8.1. Inhibitor Summary: Part 3

Many NAMPT inhibitors were discovered in phenotypic screens for cancer cell death, including screens for apoptosis. These and the majority of designed NAMPT inhibitors contain an N-heterocyclic warhead that mimics NAM. The original rationale for selective killing of cancer cells by NAMPT inhibitors is based upon the dependence of cancer cell bioenergetics on glycolysis and the ability of NAMPT inhibitors to cut off cellular NAD⁺, leading to loss of cellular ATP and bioenergetic collapse. The ability of NAMPT inhibitors, notably Vacor, to hijack NAMPT and NAD⁺ biosynthetic pathways provides additional mechanisms of cell death, including necrosis in NMNAT2-expressing tumors, notably brain cancer. Since such hijacking may contribute to efficacy and/or toxicity, more understanding of the influence of chemotype is needed. Identification of cancers addicted to NAMPT and the salvage pathway for NAD⁺ supply may lead to more focused clinical trials resulting in a more favorable efficacy/safety ratio, especially when combined with translational biomarker measurements. Combination with inhibitors of other NAD⁺-dependent enzymes may be beneficial and requires further study. Ameliorating the therapeutic index through an ADC (antibody-conjugated NAMPT inhibitor) has also been considered.¹⁴¹

3. NAMPT Activation

3.1. From Pyridines That Inhibit to Those That Activate NAMPT

A typical example of a 3-pyridylmethyleneurea NAMPT inhibitor is compound 50, reported in work that culminated in a successful mouse xenograft study in A2780 ovarian cancer cells (Figure 6A).⁶² A thioureido analogue, compound 5, showed similar potency. Replacing the 3-pyridyl with 4-pyridyl led to loss of NAMPT inhibition (compound 6). Nevertheless, a 3-pyridyl warhead is not required for NAMPT inhibition, since GMX1778 is a potent inhibitor with a 4-pyridyl warhead (Scheme 2). The pyrrolopyridine inhibitor (A-1326133) and imidazopyridine inhibitor (GNE-617), like GMX1778, form PR adducts, which we have discussed above (Figures 1C and 4A). As we have seen, there is a diversity of NAMPT inhibitor structures and no predictive structural pattern for substrate versus nonsubstrate inhibitors. The picture is further complicated by the discovery that several 4-pyridyl analogues of 3-pyridyl NAMPT inhibitors actually increase NAMPT enzyme activity.

Figure 6.

From NAMPT inhibitors to activators. (A) Structures of pyridyl and other N-heterocyclic NAMPT inhibitors and activators showing the evolution from 3-pyridyl inhibitors to non-pyridyl activators. (B) Phenolic activators include validated HTS hits and optimized compounds such as NAT5r with crystallographic evidence for binding to the rear channel (Table S3). (C) Optimized N-PAMs from two different chemical series shown by crystal structures to bind to the rear channel (Table S3).

SBI-797812 is the 4-pyridyl isomer of the NAMPT inhibitor compound 50, reported by Zheng et al.⁶² (Figure 6A). This compound was reported to increase the enzyme activity of NAMPT, in contrast to the compound 5.¹⁴² Although Zheng et al. studied 4-pyridyl compounds such as compound 6, they did not report the closely related structure of SBI-797812 (Figure 6A).

The original discovery of SBI-797812 derived from an HTS campaign using a thermal shift assay followed by measurement of enzyme activity that identified the 4-pyridyl hit SBI-0136892.¹⁴² Subsequently, a series of four papers reported medicinal chemistry optimization.^143−146 As we discuss below, assessment of “NAMPT activators” is more complicated than inhibitors and is defined not only by potency (EC₅₀) but also by maximal efficacy/activity (E_max), relative to control. As we shall see, in contrast to inhibitors, the concentrations of assay components also influence activator response.

Optimization of the 4-pyridyl series led to both activators (compounds 21, 22) and weak inhibitors (compound 8), further showing that a 4-pyridyl group is not necessary and sufficient to switch an inhibitor to an activator (Figure 6A). Of the compounds subsequently published, some showed significant improvement in efficacy and potency (e.g., compound 34). The primary focus of optimization was to improve ADMET properties, leading to DS68702229. Subsequently, compound 20L was reported in attempts to reduce the mutagenicity observed for DS68702229.^143−146 In place of the 4-pyridyl in earlier activators, later compounds tested alternative heterocycles, including the pyrazole that is common both to the activator compound 34 and NAMPT inhibitor OT-82. Thus, this series of activators is very similar to traditional NAMPT inhibitors, containing an N-heterocycle, ureido linker, and cap group.

The pyridyl group (or alternative N-heterocycle) in NAMPT inhibitors invariably occupies the nucleobase pocket, blocking substrate binding. This presents a conundrum for both the binding mode and mechanism of action leading to NAMPT activation by 4-pyridyl and other N-heterocycle activators. No cocrystal structures of NAMPT with SBI-797812 (or other SBI-like activator) have been published, and although the mechanism of SBI-797812 (referred to as an NAD⁺ booster) was examined, fundamental questions remain unanswered. The most pertinent question is the binding mode that allows the highly potent 3-pyridyl inhibitor, compound 50, to switch phenotype to a less potent, 4-pyridyl NAMPT activator, SBI-797812 (Figure 6A). The report of a benzyl alcohol analogue of SBI-797812 that is designed to occupy the nucleobase pocket and to penetrate further into the active site adds to the conundrum.¹⁴⁷ In the published binding pose, compound 10 (Figure 6A) would inhibit binding of both substrates, NAM and PRPP, yet it functions as a NAMPT activator. We need to delve deeper into the enzyme mechanism and a role for the rear channel.

3.2. NAMPT Enzyme Mechanism

A detailed discussion of NAMPT enzymology was not necessary to understand the mechanism of action of NAMPT inhibitors. For example, we did not need to reference the ATPase reaction catalyzed by NAMPT. To understand enzyme activation, we will need to probe the enzyme mechanism more deeply. Regarding progress in NAMPT activator medicinal chemistry, it would be beneficial to understand why a ligand switches from inhibitor to activator.

Two papers in 2006, studying NAMPT structural biology, defined Asp-219 as important for NAM binding and indicated His-247 as a site of autophosphorylation by ATP, essential for enzyme catalysis.^56,57 Subsequently, the detailed enzyme mechanism was elucidated by Schramm, Burgos, and co-workers in three papers focused on kinetic analysis,¹⁴⁸ structural biology,¹²⁵ and kinetic isotope effects.¹²⁶ NAMPT catalyzes two reactions: (i) the formation of NMN from reaction of NAM with α-d-5-phosphoribosyl-1-pyrophosphate (PRPP) (Figures 1B, 4B, and 7A) and (ii) an ATPase reaction phosphorylating His-247 at the same active site (Figure 5C,D).

Figure 7.

(A) Superposition of crystal structures showing NAMPT active site with His-BeF₃^– mimicking phospho-H247, substrates (NAM, PRPP), or products (NMN, PPi) (PDB 3DKL, 3DHF). (B) The high energy phosphoenzyme intermediate breaks down via capture of PO₃^– by water, PPi, or ATP. (C) After autophosphorylation by ATP to give the phosphoenzyme (E*), the NAMPT enzyme mechanism partitions toward either (green) productive NAM binding and turnover or (red) nonproductive NAM binding. (D) Superposition of NAMPT crystal structures shows the nicotinamide ring H-bonding network with three water molecules in the doorsill (NP-A1 shown in gold; PDB 3DKL, 3DHF, 8DSC, 8DSD, 8DSE). (E) Crystal structure of quercitrin occupying the rear channel in the presence of an ADP analogue bound (PDB 8DSH). (F) The dependence of enzyme activity on [NAM] leads to a bell-shaped profile (the curve depicted is a fit of data at cellular concentrations of ATP and PRPP). The simulation shown, of a 6-fold right shift, leads to increased enzyme activity. (G) NAM, NP-A1, and three waters bound to NAMPT (PDB 8DSC). (H) Superposition of N-PAMs and NAMPT activators in the rear channel (PDB 8F7L, 8DTJ, 8DSC, 8DSD).

The first step in the NAMPT enzyme mechanism has ATP binding to the active site with adenine occupying the nucleobase pocket. The ATPase reaction, leading to autophosphorylation of His-247, is required to drive the conversion of NAM to NMN. Autophosphorylation changes the characteristics of the NAMPT active site from one in which polar residues stabilize substrates (as shown in cocrystal structures with bound substrates and their analogues) to one in which the phosphohistidine coordinates two Mg²⁺ ions with PRPP to stabilize the transition state and prime the enzyme for phosphoribosyl transfer (Figures 4D and 7A). His-247 phosphorylation (i) increases affinity for PRPP (10-fold lowering of K_M), which in turn (ii) increases affinity for NAM (170-fold lowering of K_M).¹⁴⁸ These K_M-driven effects increase catalytic efficiency (k_cat/K_M). A 20-fold increase in k_cat results from stabilization of a transition state that is poised for phosphoribosyl (PR) group transfer from the ribose anomeric carbon of PRPP to the pyridyl-N of NAM (Figure 7A).¹⁴⁸ Since PRPP is poised for PR transfer to a pyridyl-N, it is easy to understand the facile transfer that occurs for Vacor, GNE-643, and other N-heterocyclic substrate/inhibitors, leading to PR adducts (Figures 1C and 4A).

The active site, consisting of the nucleobase pocket and polar catalytic site, has been mapped using cocrystal structures with substrate analogues in which phospho-H247 is mimicked by His-BeF₃^– (Figure 7A).¹²⁵ The fate of the high energy H247-PO₃^– after formation of NMN is trapping of metaphosphate (PO₃^–) either by water or by the NAMPT reaction product PPi to yield Pi or PPPi, respectively (Figure 7B). ATP has also been shown to trap metaphosphate formed in the NAMPT active site to yield Ap4.¹⁴⁹ The NAMPT activator, SBI-797812, was reported to promote formation of Ap4 and PPPi and to stabilize and “shield” phospho-H247 from water. The authors indicated that further details of this shielding mechanism “remain to be determined”.¹⁴²

The nominal stoichiometry of NAMPT is misleading and has sometimes been misrepresented as a 1:1 ratio of ATP consumption to NMN formation. Under most experimental conditions, the ATPase reaction of NAMPT is uncoupled from NAM salvage (Figure 7C); i.e., more ATP is consumed by NAMPT than is required for the conversion of NAM to NMN.¹⁴⁸ In the presence of ATP and absence of substrates (PRPP and NAM), ATP is consumed by NAMPT.¹⁴⁸ In the presence of substrates (NAM + PRPP) or products (NMN + PPi), the rate of ATP consumption is decreased. Conversely, ATP consumption is increased by binding of NAM or PPi in the absence of PRPP or NMN, respectively. In this context, ATP consumption is increased and uncoupled from turnover of NAM to NMN. This dictates that binding of NAM in the absence of PRPP is a nonproductive pathway (Figure 7C, red). In contrast, productive binding, leading to NAM turnover to NMN, results from binding of PRPP that lowers K_M(NAM) and increases affinity for NAM binding (Figure 7C, green).

3.3. NAMPT Rear Channel: Channeling for Allosteric Activation

The rear channel did not feature in the detailed enzyme mechanism delineated by Schramm and co-workers. The rear channel is a key feature for NAMPT inhibitors as discussed above and evidenced by multiple cocrystal structures. Small molecule N-PAMs were discovered by HTS using a coupled enzyme assay (NAMPT/NMNAT/NAD cycling) that identified hits that increased enzyme activity.¹⁵⁰ Cocrystal structures were obtained for early leads, NP-A1R, NP-A1S, and ZN-2-43S, and for ZN-2-29S obtained from further optimization (Figure 6C; Table S3).^150,151 These structures demonstrated that N-PAMs bind to the rear channel of NAMPT in the absence or presence of NAM (Figure 7D). The structure of an N-PAM from a different chemical series, NP-A3, was also obtained, again showing occupancy of the NAMPT rear channel (Figure 6C; Table S3).²⁰⁰ N-PAMs do not bind in the nucleobase pocket and do not disturb the water mediated H-bonding network at the doorsill to the rear channel (Figure 7D). As we describe above, NAMPT inhibitors bind to the nucleobase pocket replacing waters in this H-bonding network (Figure 3B). The rear channel is an allosteric site for N-PAMs, which is remote from the active site. The rear channel is also a key feature for small molecules that increase NAMPT activity.

From the coupled-enzyme HTS assay used to discover N-PAMs, several biogenic or bioactive phenols were identified as NAMPT activators, including quercitrin. Crystal structures were obtained for quercitrin bound in the rear channel, in the presence of either NAM or an ADP mimic (Table S3). In this case, one water molecule at the doorsill was replaced by a quercitrin phenol (Figure 7E).¹⁵⁰ A crystal structure of a synthetic phenolic NAMPT activator, NAT (Figure 6B), has also been reported, also showing binding to the rear channel (Table S3).¹⁵² The structures of quercitrin and the NP-A1 series N-PAMs displayed H-bonding to K189; however, the structure of neither NAT nor NP-A3 displayed such an H-bond. Further, the position of K189 was not influenced by N-PAM binding.

Optimization of the early lead, NP-A1S, led to syntheses of over 70 analogues.¹⁵¹ An excellent correlation was observed between binding affinity and potency for enzyme activation. Together with crystal structure data, these observations show that binding to the rear channel is responsible for enzyme activation. An explanation for NAMPT activation in response to N-PAMs must incorporate the rear channel as an allosteric site.

Cocrystal structures showed no significant effect of N-PAM binding on active site residues or NAM binding to the nucleobase pocket. Therefore, an explanation was provided associated with regulating NAM binding via the rear channel.¹⁵⁰ Remember that NAM binding in the absence of PRPP leads to nonproductive consumption of ATP. Productive binding, leading to NAM turnover to NMN, results from binding of PRPP lowering K_M(NAM) to increase catalytic efficiency. This is compatible with the NAM dependence of enzyme activity, which shows a bell-shaped curve with nanomolar K_M(NAM) and micromolar K_I(NAM) (Figure 7F). A small molecule that binds reversibly in the rear channel to inhibit low affinity NAM binding will cause a right shift in K_I(NAM), resulting in increased NAMPT enzyme activity (Figure 7F).

The proposed mechanism of NAMPT activation requires an activator to have affinity above the K_M(NAM) and below the K_I(NAM) (Figure 7C,F). Such an activator will inhibit nonproductive NAM binding without perturbing productive NAM binding. The corollary is that rear channel ligands with double-digit nanomolar affinity will inhibit both high affinity and low affinity NAM binding. Therefore, as we increase the binding affinity of ligands to the rear channel, both productive and nonproductive pathways will be inhibited. Across the 70 NP-A1 series N-PAMs, increased affinity led to increased potency for activation but reduced maximal activation, with some high affinity ligands showing minimal or no activation.¹⁵⁰ From a lead optimization perspective, it is frustrating that increasing affinity/potency does not translate to increased activity. This “Goldilocks effect” means that the highest activation will be achieved with N-PAMs that have potency in the triple-digit nanomolar range: not too hot and not too cold.

It is conceivable that SBI-797812 (and perhaps SBI-type activators) share a similar mechanism with N-PAMs in inhibiting nonproductive NAM binding. SBI-797812 displaces an FK866-like fluorescent probe from the NAMPT rear channel with affinity comparable to the potency for enzyme activation.¹⁵⁰ Nonproductive binding occurs when the phosphoenzyme breaks down without turnover of NAM to NMN. SBI-797812 was proposed to stabilize the phosphoenzyme by an unknown mechanism.¹⁴² Compound 10 (Figure 6A) is also proposed to stabilize the phosphoenzyme. If these activators bind in a similar pose to NAMPT inhibitors but with lower affinity, the prediction is that there will be a transition from enzyme inhibition to activation for those ligands that stabilize the phosphoenzyme.

In contrast to N-PAMs, SBI-797812 induced a profound left shift in K_M(ATP) and K_I(ATP).^151,153 Although SBI-like inhibitors and compound 10 (Figure 6A) can be docked in the allosteric site in simile with N-PAMs, the very different ATP dependence of SBI-797812 suggests a different mechanism. The ATP dependence is compatible with SBI-797812 competing with ATP for binding to the nucleobase pocket. This shift results in SBI-797812 being very effective at low ATP concentrations but losing activity at [ATP] > 3 mM. Physiological concentrations of cellular ATP vary from 2 to 7 mM depending on cell type.¹⁵⁴ Plasma [ATP] is over 1000-fold lower than K_M(ATP); therefore, eNAMPT has been concluded not to be catalytically active.¹⁵⁵ It is interesting to speculate that at the very low extracellular concentrations of ATP experienced by eNAMPT that SBI-797812 would “turn on” NAMPT enzyme activity.

Feedback inhibition is a feature of enzymes involved in NAD⁺ biosynthesis and catabolism. We glimpsed aspects of this feedback inhibition in our brief discussion of SARM1, CD38, NMNAT, and NAMPT above (Figure 5). NAMPT is inhibited by NAM, ATP, NADH, and NAD⁺. The published K_M for ATP is 7.4 mM; however, inhibition by ATP begins to be observed at concentrations of ATP > 4 mM.¹⁴⁸ NAMPT inhibition by NAD⁺ reported by Burgos and Schramm was IC₅₀(NAD⁺) ≈ 50 μM and K_I(NAD⁺) = 2 μM. In the presence of [NAD⁺] = 250 μM, from enzyme kinetics, the apparent K_M(NAM) was ≈900 nM versus K_M(NAM) = 5 nM extrapolated to zero NAD⁺.¹⁴⁸ In other work, 500 μM NAD⁺ was reported to increase K_M(NAM) from 1 to 25 μM.^156,157 These data indicate that normal cellular concentrations of NAD⁺ decrease NAMPT catalytic efficiency by increasing K_M(NAM). Therefore, relieving NAMPT inhibition by NAD⁺ is a mechanism that leads to increased NAMPT enzyme activity. SBI-797812 and N-PAMs have been reported to attenuate NAMPT inhibition by NAD⁺, which may contribute to increased activity in a cellular context.¹⁵³

It is likely that inhibition of nonproductive NAM binding via the rear channel, in combination with relief of NAMPT inhibition by NAD⁺, contributes significantly to the increase in NAMPT enzyme activity reported for N-PAMs and for SBI-like activators. It is also likely that other mechanisms contribute to k_cat enhancement. Normal mode analysis of NAMPT cocrystal structures hinted at the ability of rear channel binding to modulate remote amino acid residues across both subunits of the NAMPT dimer. Further evidence is needed to support a mechanism whereby activator binding to the rear channel of one active site would increase k_cat for the second active site of the dimer.

3.4. Small Molecule NAMPT Activators

Quercitrin at 30 μM increased NAMPT activity 300% over vehicle.¹⁵³ Several other biogenic or bioactive phenols were observed to be activators of NAMPT in biochemical assays, including onenetin, genistein, and dienestrol. The aglycone, quercetin, was not observed to increase NAMPT activity. The observation that several biogenic and/or bioactive phenols are biochemical NAMPT activators may indicate that this activity contributes to their observed biological actions. Interestingly, quercetin and analogues are inhibitors of NAD⁺-dependent SIRTs.¹⁵⁸ Since phenols are classical antioxidants and bioactive phenols, such as genistein, bind to multiple cellular targets, there are likely many contributors to biological activity. The dietary supplement EH301 (available commercially as Basis) has been studied in human trials. EH301 is a combination of the NAD⁺ supplement, NR, with the biogenic phenol pterostilbene.^159,160

An HTS campaign led to reporting of a series of synthetic phenols as NAMPT activators highlighted by NAT and the optimized nitrile derivative, NAT5r (Figure 6B).¹⁶¹ NAMPT binding affinity was measured by ITC and potency was reported for NAMPT activation, although E_max was not reported. These phenols were tested in a CIPN model at 3–30 mg/kg ip qd using a paclitaxel insult.¹⁶¹ Neuroprotective activity was quantified by sciatic nerve demyelination and by sensitivity to mechanical nociception with von Frey hairs: NAT5r was significantly more effective and potent than NAT in paw withdrawal and equivalent in myelin protection. This CIPN model has previously been used to test P7C3-A20⁴³ and, as we discussed above, the NAMPT activator A4276. One can speculate that the relatively low affinity inhibitor A4276 is able to act as a NAMPT activator under some cellular conditions.

Of the biogenic phenolic NAMPT activators, genistein has been shown to relieve neuropathic pain after sciatic nerve injury^162,163 and ononetin and quercitrin have been reported to have antinociceptive activity.^164,165 Phenolic compounds are antioxidants and contribute to observed neuroprotective activity by engagement of targets other than NAMPT.¹⁶⁶ Nevertheless, the observations on NAT and NAT5r are promising and suggest that the neuroprotective and other biological activities of biogenic and bioactive phenols may have a contribution from NAMPT activation.

The biochemical evidence shows that SBI-797812 relieves inhibition by NAD⁺ and acts as a NAMPT activator, especially under conditions of low ATP concentration. SBI-797812 was tested in a model of coronavirus infection by treating 17CL-1 cells with murine hepatitis virus (MHV).¹⁶⁷ It is proposed that MHV infection depletes cellular NAD⁺, thus limiting PARP activity. Induction and activation of PARP activity is an essential component of the innate immune response. Treatment with 10 μM NAM, NR, or SBI-797812 led to reductions in viral titer of 8.3-fold, 6.4-fold, and 4.4-fold, respectively, leading to the suggestion that these agents be tested in COVID-19 patients.

The 4-pyridyl NAMPT activator, DS68702229 (in the SBI class), was tested in mice fed high-fat diets. This is an obesogenic model of type 2 diabetes (T2D) and metabolic syndrome, which has been used to test NAD⁺ modulators. DS68702229 treatment led to a small reduction in body weight of approximately 6% relative to vehicle, over 21 days of treatment (30 mg/kg po qd).¹⁴³ Significant effects on other T2D-related endpoints were not reported.

We will briefly refer to the carbazole P7C3-A20.^142,153 Three labs have independently failed to detect any binding or biochemical activation of recombinant human NAMPT by this compound.^{142,147,151,153} However, in multiple reports, the pharmacological effects of P7C3-A20, including neuroprotection, have been reported as consistent with those predicted for a NAMPT activator. A recent paper reported that treatment of db/db mice (a genetic model of T2D) with P7C3-A20 for 4 weeks “rescued diabetes”.¹⁶⁸ NAD⁺ was measured in gastrocnemius muscle from P7C3-A20 treated db/db mice, reporting approximately 102% of that measured in vehicle control mice, which is problematic. In a separate experiment comparing NAMPT^+/– vs WT mice administered a single dose of P7C3-A20 or vehicle, it was reported that an antiglycemic effect was observed only in the treated WT mice (which would suggest a dependence on NAMPT).

3.5. Target Validation for NAMPT Activation

NAMPT mediates an adaptive response to genotoxic, inflammatory, and oxidative cellular stress; however, a reported significant decline in levels of NAMPT and NAD⁺ with age leads to diminution of this response.^169−171 The role of NAMPT in physiology and pathophysiology is complex. Enhancing NAMPT activity to treat diseases of aging, such as Alzheimer’s disease and related dementia (ADRD), and equally diseases of accelerated aging, such as type 2 diabetes (T2D), is compelling. NAMPT overexpression and silencing can be used to support the pursuit of activators of NAMPT enzyme activity as therapeutics.

A study on diabetic nephropathy in STZ-treated mice reported downregulation of NAMPT in proximal tubules (PT) and albuminuria that was alleviated by PT-targeted NAMPT overexpression.¹⁷² PT-targeted NAMPT conditional knockout caused a fibrotic phenotype associated with SIRT6 downregulation, which was further recapitulated by SIRT6 knockout. In earlier work by this group, kidney NMN concentration was reported to be doubled by SIRT1 targeted overexpression in STZ-treated mice and NMN was proposed to be a key upregulator of SIRT expression.¹⁷³ SIRT activity is dependent on NAD⁺ and NMN.

Female NAMPT^+/– mice exhibit glucose tolerance and impaired insulin secretion: these phenotypes are ameliorated by the administration of NMN.^169,174,175 Oxidative stress and diabetic cardiomyopathy, induced by HFD, were exacerbated in NAMPT^+/– mice. Cardiac-specific overexpression of NAMPT led to upregulation/enhancement of antioxidant and anti-inflammatory mechanisms and alleviation of the cardiomyopathic phenotype (fibrosis, apoptosis, diastolic dysfunction) in HFD-treated mice.¹⁷⁶ In another HFD mouse study, targeted overexpression of NAMPT in skeletal muscle tissue did not protect against insulin resistance but did lead to exercise-related positive changes in endurance and mitochondrial gene expression.¹⁷⁷ NAMPT protein expression was increased well over 10-fold in quadriceps and gastrocnemius. Enzyme activity in quadriceps was increased 6-fold, leading to 2-fold, 1.7-fold, and 1.4-fold increases in NMN, NAD⁺, and NAM concentrations, respectively. The large disconnect between the upregulation of NAMPT and the small changes in NAD⁺ and metabolites questions the use of genetic upregulation to validate NAMPT activators.

3.6. Effect of NAMPT Activation on Cellular NAD⁺ Concentrations

NAMPT overexpression in mouse models of T2D delivers positive phenotypes; however, a common observation in these studies is that the high levels of NAMPT expression and NAMPT activity do not lead to an equivalent elevation of cellular NAD⁺ concentration. This phenomenon, a 10–20-fold increase in NAMPT expression in cells or tissues with only a 1.4–1.6-fold increase in NAD⁺, has been reported in multiple studies and explained by a combination of NAD⁺ feedback inhibition and enhanced NAD⁺ degradation.^{157,178−180} Hara et al. used stable isotope LC–MS/MS to measure the rate of NAD⁺ biosynthesis and degradation in a variety of cell lines and primary cell cultures, correlating data with NAMPT activity.¹⁵⁷ NAMPT was overexpressed in HeLa cells: a 6-fold increase in NAMPT activity translating to a 2-fold increase in the rate of NAD⁺ biosynthesis and a 1.6-fold increase in NAD⁺. Addition of NA to these cells further increased NAD⁺, via endogenous NAPRT activity. The availability of PRPP, a substrate for NAMPT and NAPRT, was not a limiting factor for NAMPT activity. In addition, increased expression of enzymes with NADase activity (CD38, PARPs, SIRT1) was not observed.

The exact rationale for the tight regulation of cellular NAD⁺ (400–800 μM), even with 10-fold increases in NAMPT expression levels, has not been resolved. It is reasonable to hypothesize that NAMPT inhibition by NAD⁺ and its metabolites blunts the effect of NAMPT overexpression, enforcing a ceiling in cellular NAD⁺ concentration. Furthermore, measurements of enzyme activity from cell lysates will not be reflective of compartmentalized cellular feedback networks and, therefore, will be artificially high. Against this background, upregulation of NAMPT and even supplementation with NAD⁺ and its precursors (NAM, NMN, NR) will be self-limiting. The observed relief of NAD⁺ feedback inhibition by N-PAMs and SBI-like NAMPT activators contributes significantly to the ability of these compounds to elevate cellular NAD⁺.

In vivo target engagement has been reported for SBI-797812 and DS68702229 in mice. At 4 h after administration (20 mg/kg ip), SBI-797812 increased NAD⁺ in liver (from 2.0 to 2.8 nmol/mg), with no significant effect in heart and muscle tissues.¹⁴² At 1 h (100 mg/kg po), DS68702229 increased NAD⁺ in liver (from 0.32 to 0.45 nmol/mg) and increased NAD⁺ in muscle tissues to similar levels.¹⁴³ In elegant work from Dutta et al., a more complete metabolomic study was made of the effects of SBI-797812 in cell cultures and NAD⁺ flux in mice and humans.¹⁸¹ These stable isotope tracer studies were interpreted to indicate that NAD⁺ biosynthesis is differentially modulated in different cell compartments. This is compatible with other literature showing that NAD⁺ levels vary in different subcellular compartments, in particular to protect NAD⁺ stores in vital compartments, such as mitochondria.^182−185 Compartmentalization can be understood with reference to transporters, such as SLC12A8 and SLC25A51, although there is controversy about the role of the Na⁺/K⁺ Cl^– transporter SLC12A8 as a cell membrane NMN transporter.^186,187 The recently discovered role of SLC25A51 as an NAD⁺ transporter has been definitively established.¹⁻³ Moreover, the silencing of SLC25A51 has been shown to increase cytoplasmic and nuclear NAD⁺ levels leading to increased PARylation and PARP1-mediated DNA repair in cell cultures.¹⁸⁸ The authors suggested that tumors overexpressing SLC25A51 would be more sensitive to PARP inhibition, and the same can be speculated for sensitivity to NAMPT inhibitors.

4. Summary

4.1. Channeling NAMPT

After H247 autophosphorylation by ATP and PRPP binding, the affinity of NAM for NAMPT is increased, leading to high affinity binding and efficient turnover of NAM to NMN (Figure 8A). In the absence of PRPP, NAM binds to the nucleobase pocket with low affinity. This low affinity NAM binding results in nonproductive degradation of phospho-H247 and futile consumption of ATP (Figure 8B). Low affinity NAM binding is an autoinhibitory mechanism. Since these binding events are reversible, a ligand that competes with the low affinity, futile binding of NAM will relieve autoinhibition (Figure 8B).

Figure 8.

Mechanisms of orthosteric inhibition and allosteric activation. Schematic representation of small molecule regulation of NAMPT activity via binding to the rear channel. (A) High affinity NAM binding occurs after ATP autophosphorylation and PRPP binding leading to productive turnover to NMN. (B) Low affinity NAM binding leads to nonproductive ATP consumption and breakdown of the phosphoenzyme, which is inhibited by N-PAMs. (C) NAMPT inhibitors occupy the rear channel and nucleobase pocket preventing productive NAM binding. (D) NAMPT inhibitors that act as substrates are converted to PR adducts (NMN analogues) that may be further transformed to NAD⁺ analogues.

Cancer therapy was the original target for NAMPT ligands that lower cellular NAD⁺ by binding in the rear channel with a “warhead” occupying the nucleobase binding pocket (Figure 8C). NAMPT inhibitor ligands compete with NAM for binding to the active site that incorporates the nucleobase binding pocket. Consequently, many NAMPT inhibitors are also substrates that are converted to NMN analogues and sometimes further to NAD⁺ analogues (Figure 8D).

The relief of NAM (and NAD⁺) autoinhibition leading to increased productive turnover of NAM by NAMPT provides a possible explanation for the observation that potent inhibitors with a 3-pyridyl warhead become less potent activators when substituted with a 4-pyridyl warhead (SBI-like activators). We propose that both inhibitors and SBI-like activators bind to the orthosteric site (nucleobase pocket) and rear channel. N-PAMs bind only to the rear channel, and therefore are classed as positive allosteric modulators. Relief of NAD⁺ inhibition and the different dependence of SBI-like activators on ATP require further investigation.

4.2. Channeling NAMPT for Human Health

The majority of small molecule ligands designed to target NAMPT therapeutically incorporate a “warhead” that binds to the nucleobase pocket in the enzyme active site, leading to low nanomolar inhibitory potency that generally translates to equivalent potency in cells. In reviewing the literature, a surprising number of NAMPT ligands have been identified from phenotypic assays. Target pulldown experiments have frequently been used to identify NAMPT as a protein target, and many pulldown studies are supported by biophysical assays and/or cocrystal structures.

The rodenticide Vacor was identified as a NAMPT inhibitor, incorporating a classical warhead. Vacor is converted to an analogue of NMN that is enzymatically transformed to analogues of NR and NAD⁺. These analogues mimic their endogenous counterparts, disrupting cellular response, including disruption of the activity of dehydrogenases and other NAD⁺-dependent enzymes.

NAMPT inhibitors that are also substrates for NAMPT are often more potent in cells than nonsubstrate congeners. However, it is not necessary for NAMPT inhibitors to be substrates to have efficacy in cells or animal models. Theoretically, NAMPT inhibitor/substrates could target NAMPT-overexpressing cancer cells to cause selective cell death via formation of NMN and NAD⁺ analogues. Cell death would result from mechanisms in addition to NAD⁺ depletion; however, this has not been fully explored. The retinotoxicity observed for Vacor and several NAMPT inhibitors may result from a common substrate/inhibitor mechanism or may simply be an on-target adverse effect caused only by NAMPT inhibition. For orthosteric NAMPT ligands that occupy the nucleobase pocket, carbocyclic inhibitors are recommended as chemical probes, because adduct formation is not possible. Metabolite identification is exceptionally important for N-heterocyclic NAMPT ligands.

Enzymes that contain a nucleobase pocket that binds NAM include NAMPT, PARPs, SIRTs, CD38, and SARM1. For many of these enzymes, N-heterocycles that bind in the nucleobase pocket have been shown to be adducted. Inhibitors that target one of these enzymes at the nucleobase/orthosteric site may cross-react with other enzymes, either directly or indirectly via adduct formation. Although this promiscuity may sometimes lead to beneficial phenotypes, the alternative targeting of allosteric sites on these enzymes will lead to more selective modulators that do not have potential for adduct formation. The allosteric approach has recently been applied to SARM1 inhibition.¹¹²

Validated allosteric NAMPT inhibitors have yet to be reported. Inhibitor specificity for NAMPT over NAPRT results from the lack of a rear channel in NAPRT. The rear channel represents an allosteric site, and N-PAMs have recently been reported that bind to this allosteric site.¹⁵³ Biogenic phenols may also increase NAMPT activity by weak interactions at this site, and design of synthetic phenolic NAMPT activators provides another interesting strategy. Although the binding mode of SBI-like activators has not been fully defined, their N-heterocyclic warheads are likely to bind to the nucleobase pocket. Under certain cellular conditions, orthosteric activators may act as NAMPT inhibitors and vice versa. Observations on N-PAMs and SBI-like activators illustrate a key component of increasing NAMPT activity, which is the attenuation of feedback inhibition by NAM and NAD⁺.

NAMPT ligands that modulate NAMPT activity and regulate the levels of cellular NMN and NAD⁺ have not yet achieved clinical success (if we exclude the accidental, weak NAMPT inhibitor chidamide). NAMPT inhibitors have primarily been viewed from the perspective of causing a catastrophic drop in NAD⁺ and subsequent drop in cellular ATP leading to cancer cell death. NAMPT inhibitor strategies will be aided by identification of salvage-addicted tumors, versus those that circumvent NAMPT inhibition via the Preiss–Handler pathway or NMRK. The combination of NAMPT inhibitors with those of other neoplastic targets, such as PARP and CD38, also remains to be fully explored. Understanding retinotoxicity can also assist in NAMPT inhibitor design.

Allosteric regulation of NAMPT activity provides an entirely new therapeutic avenue that is yet to be explored fully in preclinical models. Given the central role of NAD⁺ and emerging roles for NMN in human physiology and pathophysiology, there is broad scope for therapeutic applications. The dependence of NAMPT activity and NAMPT activation on concentrations of NAM, PRPP, ATP, NADH, and NAD⁺ is likely to lead to tissue, cell, and cell-compartment specific effects. The NAMPT rear channel is fundamental to small molecule therapeutic approaches and appears to have a role in autoinhibition. Enzyme activation results from allosteric binding to the channel, begging the question of endogenous allosteric activators. Finally, our brief introduction of eNAMPT reflects the two solitudes of the intracellular and extracellular roles of NAMPT and the need for a deeper understanding of their interplay in health and disease.

Glossary

Abbreviations Used

AD

Alzheimer’s disease

ART

ADP-ribosyltransferase

CD38

cluster of differentiation 38

CNS

central nervous system

HTS

high-throughput screen

NA

nicotinic acid

NAD⁺

nicotinamide adenine dinucleotide

NADase

NAD⁺ catabolic enzyme

NAM

nicotinamide

NAMPT

nicotinamide phosphoribosyltransferase

NAPRT

nicotinic acid phosphoribosyltransferase

NMN

nicotinamide mononucleotide

NMNAT

nicotinamide mononucleotide adenylyltransferase

N-PAM

NAMPT positive allosteric modulator

NR

nicotinamide riboside

PARP

poly ADP ribose polymerase

SARM1

sterile α and TIR motif containing 1

SIRT

sirtuin

Biographies

Ganga Reddy Velma received his B.Sc. and M.Sc. degrees in chemistry from Osmania University and his Ph.D. in medicinal chemistry from CSIR—Indian Institute of Chemical Technology, Hyderabad, India. He is skilled in lead optimization, multistep organic synthesis, and purification of high-quality target molecules. He coordinates medicinal chemistry research across broad therapeutic targets in the Thatcher lab.

Isabella S. Krider received a bachelor’s degree in chemistry from Bridgewater College in 2019. She is currently studying for a Ph.D. in chemistry and biochemistry at the University of Arizona. She is working on the design and synthesis of NAMPT small molecule ligands.

Erick T. M. Alves completed his undergraduate studies in pharmacy and biochemistry at the University of Sao Paulo, Brazil. His undergraduate years were marked by his involvement in synthetic biology and drug discovery research. A three-month internship at the University of North Carolina, Chapel Hill, further refined his expertise, focusing on the intersection of AI and drug discovery. He is currently a graduate student in the drug discovery and development track of the pharmaceutical sciences program at the University of Arizona studying the mechanism and design of NAMPT activators. With a strong academic foundation and practical experience, he is dedicated to advancing pharmaceutical innovation at the nexus of science and technology.

Jenna M. Courey is a graduate student in the Department of Chemistry & Biochemistry at the University of Arizona. She earned a B.S. in biochemistry from Iowa State University. Her research interests center around the study of enzyme mechanisms for targeted drug discovery

Megan S. Laham received her bachelor’s degree in chemistry from Saint Anselm College in 2020. She is currently studying for her Ph.D. in biological chemistry at the University of Arizona. Her work is focused on the design and elucidation of the mechanism of action of nonlipogenic ABCA1 inducers and differential nuclear hormone receptor modulation.

Gregory R. J. Thatcher is professor and R. Ken and Donna Coit Endowed Chair of Drug Discovery in the R. Ken Coit College of Pharmacy and professor in the Department of Chemistry & Biochemistry in the Colleges of Science and Medicine at the University of Arizona in Tucson. He moved to Arizona during COVID from his position as the Hans W. Valteich Chair in Medicinal Chemistry and Founding Director of UICentre, the campuswide drug discovery center at the University of Illinois Chicago. He began independent faculty research in 1988 as a physical organic chemist at Queen’s University, Canada. An early interest in Alzheimer’s disease, which persists, led to a drug entering clinical trials in 2003. He has subsequently created novel therapeutics that are in clinical trials for breast cancer.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jmedchem.3c02112

• Chemical structures of NAD⁺ biosynthetic intermediates; tables providing structures, pdb codes, detailed information, and references for NAMPT inhibitors and activators (PDF)

Author Contributions

^# These authors contributed equally.

This work is supported by NIH Grant RF1AG067771. M.S.L. and J.M.C. are supported by NIH T32 GM008804.

The authors declare the following competing financial interest(s): G.R.J.T. is an inventor on patents assigned to the University of Illinois.