Physicochemical and Metabolic Optimization of NAMPT Inhibitors Enables Effective Antibody–Drug Conjugates for Liquid and Solid Tumors

Abstract

Antibody drug conjugates (ADCs) enable selective delivery of highly potent small molecules, yet most clinically validated payloads target solely cell-division pathways with several drawbacks such as high off-target toxicity and a lack of efficacy on nondividing tumor cells (e.g., tumor stem cells). Metabolic targets such as nicotinamide phosphoribosyltransferase (NAMPT) offer a complementary mode of action and are predestined as ADC payloads since systemic toxicities have prevented clinical use of NAMPT inhibitors (NAMPTi) as free drugs. Here we show that highly hydrophobic NAMPTi, especially cyanoguanidine-containing inhibitors, are chemically and metabolically suboptimal for ADC deployment due to limited efficacy likely due to lysosomal conversion to inactive guanylureas. Guided by structure-based design, and molecular dynamics simulations, we developed two next-generation NAMPT inhibitors featuring (i) a tertiary alcohol to balance hydrophilicity and (ii) an isoindoline-urea group to improve lysosomal stability in comparison to the cyanoguanidine. The optimized inhibitors retained high affinity to the NAMPT enzyme and showed strong cellular activity upon targeted delivery through ADCs. When conjugated to anti-CD30, anti-HER2, or anti-TROP2 antibodies, the resulting ADCs showed durable responses in hematologic and solid tumor models, including complete regressions in the metabolically stringent NCI-N87 gastric carcinoma xenograft after a single 2 mg/kg dose. These findings highlight physicochemical tuning and lysosomal stability as key design principles for NAMPT-based payloads and support NAMPT inhibition as a very promising mode of action (MoA) for next-generation ADC therapeutics.

Introduction

The ADC field is expanding rapidly and continues to reshape modern oncology. With the recent 2025 approval of Jiangsu Hengrui’s Avida (Trastuzumab Rezetecan) in China, the number of approved ADCs has risen to 19, and several hundred additional candidates are progressing through preclinical and clinical development. , The success of these bioconjugates comes from their ability to pair selective antigen targeting with highly potent payloads, resulting in improved efficacy and reduced systemic toxicity compared to conventional chemotherapy. However, most clinically validated ADCs rely on microtubule-disrupting payloads or TOP1i limiting their use to rapidly proliferating tumors since their activity depends on cell division. , In this work we sought to expand the current selection of available payloads to provide therapeutic tools to access low proliferating tumors by using a payload acting independent of cell division.

NAMPT is the rate-limiting enzyme of the nicotinamide (NAM) salvage pathway, the predominant route sustaining the intracellular reservoir of nicotine adenine dinucleotide (NAD⁺). Through its central role in NAD⁺ homeostasis, NAMPT supports energy metabolism and maintains the cellular redox environment. NAD⁺ also functions as a critical substrate for several enzyme classes, including the poly-ADP-ribosylpolymerases (PARP), which consume large amounts of NAD⁺ during DNA damage repair. Since cancer cells commonly exhibit metabolic rewiring, increased oxidative stress, and elevated activation of NAD⁺-consuming enzymes, their reliance on NAMPT is frequently heightened. , Indeed, NAMPT overexpression has been described in energy-demanding malignancies such as ovarian cancer. The important role of NAMPT in NAD⁺ homeostasis especially in cancer cells has driven the intensive development of small-molecule NAMPTi since the early 2000s. These compounds are synthetically easily accessible, capable to inhibit the enzyme at nanomolar concentrations and display strong antiproliferative activity in vitro and in vivo. − However, their systemic administration results in severe dose-limiting toxicities in humans, including hematologic, retinal, and cardiac effects, reflecting the essential role of NAMPT in all types of tissues and the lack of tumor-selective delivery. ,

To overcome these challenges, NAMPTi have been incorporated into ADCs to retain their potent metabolic mechanism while restricting exposure to antigen-positive tumor cells. − Yet, our studies show that previous NAMPTi generation, originally designed as free small molecules, are not intrinsically suited for ADC payload use. Their physicochemical and metabolic features impose significant limitations when deployed in bioconjugates. First, highly hydrophobic NAMPTi can reduce ADC solubility and promote aggregation, leading to suboptimal activity in vitro. Second, because many ADC formats rely on lysosomal processing for payload release, the payload must remain stable under acidic, enzyme-rich lysosomal conditions. In this context, cyanoguanidine containing NAMPTi may undergo hydration to inactive guanylurea species, potentially reducing intracellular exposure to the active inhibitor after release. Together, these findings underscore the need to redesign NAMPTi specifically for ADC delivery.

Results and Discussion

1. In Silico Lead Optimization of NAMPTi for ADC Payload Development

1.1. Balance of Hydrophilicity

Hydrophobic payloads like NAMPTi (LogP > 1) present significant challenges for ADC development, including aggregation, poor solubilization, precipitation-induced loss, and increased off-target toxicity driven by strong diffusion-driven bystander activity, resulting in a narrow therapeutic index compromising safety. On the other hand, such a pronounced bystander effect can also contribute to the potent activity of NAMPTi, typically in the nanomolar range. − These issues underscore the need for NAMPTi with improved but carefully balanced physicochemical properties to enhance formulation stability and therapeutic safety as ADC payloads.

To guide the design of optimized NAMPTi, we adopted the pharmacophore model proposed by Neumann et al. as a generalized description of NAMPTi binding within the enzyme’s funnel-shaped catalytic pocket (Figure A). This model defines five distinct pharmacophoric regions: (i) a solvent-exposed linker attachment point at the pocket entrance (linker anchor point); (ii) a central heteroaromatic “core ring” positioned at the tunnel opening; (iii) an aliphatic spacer; (iv) an H-bond polar capable group; and (v) a terminal heteroaromatic headgroup that mimics nicotinamide (NAM-mimicking group).

1.

Structure-Based Optimization of NAMPTi Guided by Docking and MD Simulations. (A) Pharmacophore representation of NAMPTi using reference compound NAMPTi 1A (4-piridinyl) or NAMPTi 1B (3-piridinyl), with NAMPTi 1B exhibiting a calculated LogP of 2.65. (B) NAMPTi 1B (ball-and-stick, green carbons) docked into the NAMPT catalytic pocket (PDB: 6E68), showing key noncovalent interactions and the distance (Å) between His191 (stick, orange carbons) and the C4 atom of the piperidine ring (purple line). (C) Structure of optimized NAMPTi 3, highlighting the newly introduced tertiary alcohol (red) with NAMPTi 3 exhibiting a calculated LogP of 1.92. (D) NAMPTi 3 (ball-and-stick, magenta carbons) docked into NAMPT (sticks, green carbons), showing the new H-bond between the His191 imidazole and the tertiary hydroxyl (yellow line, distance in Å). (E) Interaction diagram from a 20 ns MD simulation of NAMPTi 3 bound to NAMPT (PDB: 6E68), depicting the fraction of simulation time each interaction is maintained. The His191–OH hydrogen bond (circled in blue) persisted for over 50% of the trajectory.

Using this framework, we initiated a structure-based optimization campaign starting from the 4-pyridyl-cyanoguanidine NAMPTi 1A reported by Neumann et al. The original 4-pyridyl headgroup, responsible for NAM mimicry, was substituted by a 3-pyridyl moiety previously associated with stronger NAMPT engagement and improved potency (Figure A). The compound was synthesized following the route reported for its 4-pyridyl analogue, affording reference inhibitor NAMPTi 1B (full characterization in the Supporting Information). ,,

However, despite exhibiting potent nanomolar cytotoxicity of the free compound in vitro, its corresponding anti-CD30 ADC (DAR 10; linker–payload 2, a cathepsin B cleavable Val-Ala-PAB linker bearing a maleimide conjugation handle), generated via maleimide conjugation to reduced interchain cysteines and an additional engineered cysteine pair at position D265 (EU-Numbering), did not display a similar performance, since cell killing was limited to <70% even at micromolar ADC concentrations. These findings, described in detail in Section 4.1 (“In Vitro Evaluation of Free NAMPTi and NAMPTi-ADCs in Human Cancer Cell Lines”), motivated the design of a next-generation NAMPTi specifically tailored for ADC application with improved solubility and hydrophilicity while maintaining binding to the enzyme.

To this end, NAMPTi 1B was docked using Glide SP into two NAMPT cocrystal structures (PDB: 6E68, and 7PPE). The analysis aimed to identify substituents that could enhance physicochemical properties without compromising inhibitory potency. Analysis of the resulting binding conformations revealed that the imidazole nitrogen π of His191, located at the entrance of the narrow catalytic tunnel of NAMPT, is ideally positioned to engage with a hydrogen-bond donor (HBD) group. The interatomic distance between C4 carbon of the piperidine ring and His191 imidazole nitrogen π is approximately 4.8 Å (Figure B), suggesting that the introduction of an HBD at C4 on the piperidine could establish a favorable hydrogen bond that can compensate for potential destabilization of the complex due to the reduction of hydrophobic interactions. Topologically, this region represents a transition between the solvent-exposed pocket entrance and the more tight catalytic tunnel. In the absence of anchoring interactions (I.E. hydrogen bonding), conformational mobility of the inhibitor in this region may be entropically unfavorable and compromise binding stability. Introducing an HBD capable of engaging His191 was therefore expected to reduce local degrees of freedom and enhance conformational preorganization of the bound state, thereby improving overall complex stability. Based on geometric and electronic complementarity, the incorporation of a tertiary hydroxyl group at the piperidine C4 position was predicted to meet the requirements, while simultaneously lowering logP and improving hydrophilicity but maintaining a strong binding to the enzyme.

The resulting compound, NAMPTi 3, exhibited favorable docking interactions and a calculated logP of 1.92 (vs 2.65 for NAMPTi 1B), corresponding to a ∼ 5-fold increase in hydrophilicity (Figure C). The tertiary alcohol was predicted to modulate charge density between the core ring and linker-anchor point regions, thereby improving solubility and forming a stable hydrogen bond with His191 (Figure D). From a medicinal chemistry perspective, tertiary alcohols offer moderate polarity, high metabolic stability, and a nonbasic character that minimizes off-target reactivity compared with alternative HBDs such as amines. Their incorporation has been correlated with increased hydrophilicity, improved solubility, and sustained binding potency, properties that align well with the design goals of ADC payloads. Given the steric constraints at this position, a tertiary hydroxyl represented the optimal modification to balance polarity, stability, and productive target engagement.

To assess the persistence of the His191–OH hydrogen bond, 20 ns all-atom molecular dynamics (MD) simulations (OPLS4, SPC water, NPT ensemble, 300 K, 1 atm) were performed for two representative NAMPT–NAMPTi 3 complexes (PDB 6E68 and 7PPE; see Figure S5 and S6 in Supporting Information). Remarkably, the His191 hydrogen bond remained stable throughout 30–80% of the simulation time, supporting the proposed interaction (Figure E).

1.2. Design of a Lysosomal Stable NAMPT Inhibitor

NAMPTi 3 was designed to increase hydrophilicity (lower calculated log P than NAMPTi 1B) while maintaining key NAMPT binding interactions as supported by in silico modeling. For ADC evaluation, NAMPTi 3 was converted into linker-payload 5 using a cathepsin B cleavable Mal-Val-Ala-PAB linker and conjugated to an anti-CD30 antibody (see Section 2 for linker-payload synthesis and Sections 4.2-4.3 for in vivo study design). In vivo, this format was highly effective in a disseminated setting but showed attenuated activity in a solid tumor context, consistent with limited intratumoral exposure of the released payload. These observations indicate that, beyond solubility and target binding, ADC metabolism must be considered; in particular, payloads intended for solid malignancies should remain chemically intact following lysosomal release and reach sufficient intracellular concentration to enable controlled diffusion into surrounding tumor tissue.

This challenge is inherent to solid tumors, where poor vascularization and dense stroma restrict macromolecular penetration, placing greater reliance on the properties of the released small-molecule payload. For ADCs employing cathepsin-B cleavable linkers, effective activity therefore depends on both lysosomal stability and a fine-tuned balance of hydrophilicity that permits passive membrane diffusion sufficient for bystander activity without promoting off-target toxicity. Guided by these considerations, we pursued a second design cycle focused on improving lysosomal robustness and intratumoral distribution of the released NAMPT inhibitor.

Starting from NAMPTi 3, we noted that its cyanoguanidine moiety, although essential for hydrogen bonding with Asp219 and Arg311 in the NAMPT active site (Figure E), showed limited stability under lysosomal conditions (See the section below “3. Lysosomal Stability of Cyanoguanidine Containing NAMPTi”). To address this, we replaced the cyanoguanidine with a hydration-resistant urea group, a scaffold previously reported in NAMPTi.

Further optimization targeted the pyridine ring, which mimics the NAD⁺ moiety in NAMPT binding. Previous studies have shown that introduction of an isoindoline moiety into NAMPTi can enhance enzyme binding. This substitution was expected to strengthen the π–π stacking interactions already described between the inhibitor and aromatic residues in the NAMPT catalytic pocket and compensate for the potential loss of binding interaction between NAMPTi and NAMPT due to replacement of cyanoguanidine by a urea group. Structurally, replacing the pyridinyl group with an isoindolinyl moiety shifts the centroids of the aromatic systems, creating a more favorable offset relative to Tyr18 and Phe191. This geometry allows the negative π-cloud over the aromatic carbons of these residues to interact more effectively with the electron-deficient hydrogen edge of the isoindoline ring, stabilizing the interaction. Docking confirmed that the isoindoline analogue 4 exhibited a more pronounced offset than its pyridine counterpart, as reflected by a smaller angle between the aromatic rings centroids (142.5° vs 126.8°; Figure A, B). Since a 180° angle reflects coplanar, repulsive stacking, the smaller angle indicates less repulsion and stronger π–π stabilization.

2.

3D model of docking posed of NAMPTi 3 (A) and NAMPTi 4 (B) on NAMPT pocket. Zoom of pyridine (A) and isoindoline (B) moieties showing centroid distances and angles to each of π-π stacking participating amino acids in the enzyme (Phe193 and Tyr18), where the centroids are shown as cyan dots.

1.3. In Silico Validation of Newly Designed NAMPTi

To assess whether the newly designed inhibitors retained comparable affinity for NAMPT, we evaluated NAMPTi 3 and 4 alongside the parental NAMPTi 1B in silico and benchmarked them against the reference inhibitors FK866 and the crystallized ligand in the PDB cocrystal 6E68 (lig-6E68). Molecular Mechanics generalized Born Surface Area (MM-GBSA) binding free energies derived from MD trajectories were used as a qualitative indicator of relative binding strength. While this approach does not provide absolute binding affinities, as methods such as free energy perturbation (FEP⁺) would, its purpose here was to guide lead optimization in conjunction with experimental validation that enables an integrative consideration of parameters such as solubility, chemical stability, and permeability. Tanimoto similarity analysis of molecular fingerprints showed that Lig-6E68 has the highest average similarity to the test compounds (NAMPTi 1B, 3, and 4). While Tanimoto similarity does not determine binding mode, this structural similarity supports the expectation that the binding mode of Lig-6E68 within the NAMPT pocket is more representative of the test compounds than that of the other two reference ligands (see Supporting Information). Accordingly, a receptor grid based on PDB 6E68 was used for docking all inhibitors with the Glide SP workflow. Docked complexes were visually inspected against the pharmacophore model proposed by Neumann et al. and subsequently subjected to all-atom MD simulations in Desmond (OPLS4, SPC water, NPT ensemble, 300 K, 1 atm). Each system was simulated in triplicate with randomized initial velocities and extended until RMSD stabilization (fluctuations ≤ 1 Å over 100 ns; see RMSD plots in the SI).

MM-GBSA binding energies were calculated from equilibrated 50 ns trajectory fragments. Statistical analysis indicated that both newly designed inhibitors maintained binding energies comparable to the reference compounds (Figure ). Hence, introduction of the tertiary alcohol in NAMPTi 3 and replacement of the cyanoguanidine-pyridyl unit with the urea-isoindoline scaffold in NAMPTi 4 helped compensate for potential losses in hydrophobic stabilization. Although median ΔG_binding values varied by up to ∼ 12 kcal mol^–1 between NAMPTi 4 (−73.57 kcal mol^–1) and FK866 (−61.49 kcal mol^–1), the overlap of confidence intervals and standard deviations of 4 to 5 kcal mol^–1 suggest that these differences are not chemically significant (Statistical descriptors can be found in Supporting Information Table S3). Accordingly, the MM-GBSA results are best interpreted as qualitative support for the preservation of NAMPT binding, consistent with experimental confirmation that the new inhibitors retain affinity comparable to the parental compound.

3.

MM-GBSA binding free energy calculations from 50 ns MD simulations of optimized NAMPTi and reference inhibitors. Truncated violin plots show the distribution of MM-GBSA binding energies for each NAMPTi; dotted lines indicate the first and third quartiles, and the dashed line denotes the median.

2. Lead NAMPTi (3 and 4) and Linker-Payload (5 and 14) Synthesis

The lead cyanoguanidine NAMPT inhibitor 3 was synthesized in seven steps (Scheme ). The sequence began with a Grignard reaction to introduce a tertiary alcohol onto the piperidine ring. Initial attempts at room temperature gave tertiary alcohol 8 in ≤ 5% yield, but optimization by lowering the reaction temperature to 8 °C and adding LaCl₃·LiCl to stabilize the enolizable 4-piperidone, as described by Knochel et al., improved the yield to 47%.

1. Synthesis of Payloads 3 and 4 and Linker-Payloads 5 and 14 .

^a Reagents and conditions: (a) Mg, LaCl₃–LiCl, THF (abs.), 8 °C, overnight; (b) RuCl₃*3 H₂O (cat), NaIO₄, H₂SO₄, H₂O, ACN, EtOAc, 0 °C, 30 min; (c) 10: Ammonium acetate (pH = 6), NaCNBH₄, MeOH (abs.), r.t, 4.5 h; 18: benzylamine, 10% Pd/C (70 × 4 mm) cartridge, max H₂ production, 1 mL/min reaction mix flow, MeOH, 30 °C, 1h then, 50 °C, 2h; (d) 11: Phenyl N-cyano-N’-3-pyridinylcarbamimidate; Et₃N; 1,4-dioxane; r.t., overnight; 19: 2,3-dihydro-1H-isoindole-2-carbonyl chloride, Et₃N, DCM (abs.), r.t, overnight; (e) 12: H₂, Pd/C, EtOH/EtOAc (1:1), r.t., overnight; 20: 20% TFA DCM solution, r.t, 20 min; (f) 13: Cbz-3-ABZ–OH, DCC/HOBt, DIPEA, DCM (abs.); r.t., overnight; 21: Boc-3-ABZ–OH, DCC/HOBt, DIPEA, DCM (abs.); r.t., overnight; (g) 3: H₂, Pd/C, EtOH/EtOAc (1:1), r.t., 4 h; 4: 20% TFA DCM solution, r.t, 20 min; (h) Mal-Val-Ala–PAB-PNP, HOAt, 2,6-lutidine, DIPEA, DMF (abs.), r.t., 4 days.

Then aldehyde 9 was obtained from olefin 8, functionalized with a tertiary alcohol, in a single one-pot reaction. In this step, periodate served a dual role: first oxidizing Ru­(III) to the active Ru­(VIII) species that catalyzes syn-dihydroxylation of olefin 8 and then promoting in situ oxidative cleavage of the resulting diol. This one-pot two-step reaction afforded aldehyde 9 in just 30 min at 0 °C, with a respectable yield of 45% and high purity.

Amine 10 was obtained from aldehyde 9 via reductive amination, which yielded amine 10 in 20% under the best conditions tested. Condensation of amine 10 with phenyl N-cyano-N′-3-pyridinylcarbamimidate afforded intermediate 11 in 40% yield. Pd-catalyzed hydrogenolysis of the benzyloxycarbonyl (Cbz) group followed by DCC/HOBt-mediated coupling of piperidine 12 with 3-(benzyloxy)­carbonylaminobenzoic acid furnished amide 13 in 37% yield. Final Pd-catalyzed hydrogenolysis removed the Cbz group to deliver inhibitor 3 in 80% yield and an overall yield for the synthesis of the free inhibitor of 0.33%.

The corresponding linker–payload 5, bearing a cathepsin B–cleavable Val-Ala-PAB linker, was then prepared by carbamate formation between nitrophenyl carbonate-activated Mal-Val-Ala-PAB linker and aniline inhibitor 3 in the presence of HOAt with a yield of 30%.

Next, the NAMPTi 4, which contains an isoindoline linked via a urea moiety as NAM-mimicking group, was synthesized in seven steps (Scheme ). While conceptually analogous to the route toward NAMPTi 3, this sequence enables the transformation of aldehyde 17 into amine 18 via a one-pot two-step process that combines reductive amination with benzylamine, followed by in-line hydrogenolytic debenzylation in a continuous flow setup, giving amine 18 in 58% overall yield. Subsequent condensation with commercially available isoindoline-2-carbonyl chloride under basic conditions afforded the asymmetric isoindoline–urea 19 in 44% yield.

From intermediate 19, incorporation of the aniline-containing fragment was achieved in three steps. Boc deprotection of the piperidine moiety furnished secondary amine 20 as TFA salt, which was used directly without purification. Coupling of secondary amine 20 with commercially available 3-(Boc-amino)­benzoic acid mediated by DCC/HOBt yielded amide 21 with 35%, followed by final Boc deprotection affording NAMPTi 4 in 92% yield and an overall yield for the synthesis of the free inhibitor of 3.2%.

The corresponding linker–payload 14 was then prepared following the same procedure described for linker–payload 5, with a yield of 26%.

3. Lysosomal Stability of Cyanoguanidine Containing NAMPTi

3.1. Guanylurea as Inactive Metabolite of Cyanoguanidine Containing NAMPTi

Most cleavable ADC linkers rely on the chemical or enzymatic conditions in the lysosome to trigger payload release, with those used in this study been selectively cleaved by lysosomal cathepsin B. Targeted delivery via ADCs improves safety compared to untargeted approaches using NAMPTi as small molecule drug formulations, since the uptake does not rely on passive diffusion through membranes but rather through target-dependent internalization followed by trafficking to the lysosome for drug release. This requires the payload itself to remain stable under acidic environments and in the presence of lysosomal enzymes. Cyanoguanidines, however, are known to undergo acid-catalyzed hydration to guanylurea derivatives, raising concerns about the lysosomal stability of cyanoguanidine-containing NAMPTi such as NAMPTi 1B or 3.

We hypothesized that conversion of cyanoguanidine-containing NAMPTi into an inactive guanylurea species such as 22 and 23 (Scheme ) could reduce the effective intracellular concentration of the inhibitor upon payload release from the ADC. In a solid tumor, the degradation or inactivation of the inhibitor would reduce its overall concentration within the tumor and the surrounding microenvironment, typically sustained by bystander activity. Consequently, cyanoguanidine hydration was expected to compromise efficacy particularly in solid tumor models, where payload diffusion through the tumor microenvironment (bystander effect) contributes to antitumor activity (see Section “ In Vivo Evaluation of NAMPTi-ADCs in a Solid L540 Mouse Xenograft Model”).

2. Acidic Hydration of Cyanoguanidine Containing NAMPTi.

To evaluate this possibility, we synthesized the guanylurea metabolites of NAMPTi 1B and 3 (compounds 22 and 23, respectively) as reference compounds for in vitro cytotoxicity assessment as well as to assess the instability of the cyanoguanidine group in the NAMPTi. Each NAMPTi was incubated in water/TFA/MeOH (pH < 2) at 37 °C for 24h (See Scheme ), resulting in an almost complete conversion of NAMPTi 1B to compound 22 but only in partial conversion of 23% of NAMPTi 3 to compound 23. We speculate that the lower reactivity of NAMPTi 3 may reflect intramolecular hydrogen bonding between its tertiary alcohol and the cyanoguanidine group hampering the hydration process.

In contrast, NAMPTi 4 lacking a cyanoguanidine moiety remained stable under the same acidic conditions for over 72 h with no detectable degradation or side products, demonstrating that the design concept of this new generation of NAMPTi effectively ensures stability.

In vitro assays showed that both guanylurea derivatives 22 and 23 were not cytotoxic relative to their parent NAMPTi 1B and 3 (Supporting Information Figure S26 A,B). These findings support our hypothesis and suggest that guanylurea formation represents a critical lysosomal inactivation mechanism limiting the efficacy of cyanoguanidine-containing NAMPTi based ADCs, particularly in solid tumors.

3.2. Lysosomal Degradation Study of Cyanoguanidine Containing NAMPTi

The potential lysosomal degradation of a cyanoguanidine-based NAMPTi into corresponding guanylurea derivatives was also directly investigated by assessing the metabolic stability in human liver lysosomal extracts. NAMPTi 1B was used as representative compound, and its degradation profile was quantified using multiple reaction monitoring (MRM) based LC–MS analysis (see Supporting Information for detailed experimental procedures).

Upon incubation at 37 °C, NAMPTi 1B showed progressive conversion to guanylurea 22. After 24 h, ∼ 10% of NAMPTi 1B was converted, while no conversion was observed in control samples lacking lysosomal enzymes (See Supporting Information). This demonstrates that not only the pH, but also lysosomal enzymes are contributing significantly to the conversion into the inactive guanylurea form. As shown in Figure , hydration increased over time, reaching ∼ 30% conversion after 72 h.

4.

MRM LC–MS analyses of (A) NAMPTi 1B and (B) guanylurea 22 over 72h in lysosomal medium with lysosome extracts. In both cases, the R² values of the trend lines indicate a good linear fit, showing that the degradation of NAMPTi 1B leads to the formation of guanylurea 22 through hydration in lysosomes.

In contrast, in case of NAMPTi 3, no formation of guanylurea 23 was detected. However, a slower or lower-level lysosomal degradation of NAMPTi 3 below the quantification limits of the current method cannot be ruled out, and further studies will be required to clarify its metabolic fate.

Collectively, these results demonstrate that cyanoguanidine-containing NAMPTi can undergo enzyme-driven lysosomal conversion, and, in the case of NAMPTi 1B, this involves formation of guanylurea 22, supporting the proposed metabolic inactivation pathway for this class. We note, however, that these in vitro lysosomal stability assays use incubation times that are expected to exceed the intracellular lysosomal residence time of released payload, which was not directly measured in this study.

Accordingly, these data are not intended to imply that hydration alone is sufficient to fully deactivate the payload during lysosomal transit. Rather, they establish hydration to the guanylurea as a measurable lysosomal chemical liability under lysosomal-relevant conditions that can contribute, together with factors such as release kinetics, intracellular trafficking and efflux, to reducing the amount of active NAMPTi available to reach the cytosol and engage its target.

4. In Vitro and In Vivo Evaluation of Free NAMPTi and NAMPTi-ADCs

4.1. In Vitro Evaluation of Free NAMPTi and NAMPTi-ADCs in Human Cancer Cell Lines

To assess the potency of the newly developed NAMPTi and their suitability as ADC payloads, cytotoxicity in selected human cancer cells cultured under standard conditions in serum-containing media was evaluated using either the BrdU-ELISA or CellTiter-Glo assay, depending on whether the cell line was adherent or nonadherent (see Supporting Information for experimental details). Representative results for two models that were directly used to guide the in vivo validation experiments are presented in Figure and Figure S23.

5.

Comparative cytotoxicity of NAMPTi and their corresponding anti-CD30 ADCs in L540 cells. All ADCs were DAR10 obtained by maleimide conjugation to reduced interchain cysteines (maximum DAR 8 for IgG1) plus two engineered cysteines introduced via the D265C mutation. (A) Dose–response curves of free NAMPTi 1B (blue), NAMPTi 3 (red), and NAMPTi 4 (green), assessed by CellTiter-Glo after 96 h. (B) Dose–response curves of anti-CD30 DAR 10 ADCs bearing linker–payloads 2 (blue), 5 (red), and 14 (green), assessed by CellTiter-Glo after 96 h. Maximal cell killing for the ADC bearing linker–payload 2 was ∼ 60–70% at the highest concentrations tested, whereas ADCs bearing linker–payloads 5 and 14 achieved >90% maximal cell killing.

The first model, L540 (CD30⁺ human Hodgkin lymphoma) representative for hematological malignancies, was selected for its well-established sensitivity to NAMPT inhibition, enabling clear comparison between the new inhibitors NAMPTi 3 and 4, the parental compound NAMPTi 1B, and the corresponding ADCs. As shown in Figure A (blue curve), NAMPTi 1B exhibited potency in the nanomolar-range, consistent with its profile as a highly active NAMPT inhibitor, suitable for ADC payload development. The corresponding ADC Anti CD30-NAMPTi 2-DAR10 consisting of an anti-CD30 mAb and linker–payload 2 with a DAR of 10 retained strong cytotoxicity in the low-nanomolar range (Figure B, blue curve). However, despite the high potency of the naked compound, the ADC reached only a maximum of ∼ 60% reduction in viable cells, suggesting incomplete cell killing even at the highest concentration.

In contrast, the optimized more hydrophilic inhibitor NAMPTi 3 (Log P = 1.92 vs 2.65 for NAMPTi 1B) displayed markedly reduced potency as a free compound, with IC₅0 values in the low-micromolar range (Figure A, red curve). This apparent loss of activity might account for a reduced membrane permeability rather than weaker NAMPT binding. Supporting this interpretation, the corresponding ADC Anti CD30-NAMPTi 5-DAR10 carrying linker–payload 5 with a DAR of 10 restored low-nanomolar potency and achieved complete cytotoxicity with complete cell killing in L540 cells (Figure B, red curve). Consistent with ADC metabolism where payload release internalization occurs via endolysosomal trafficking rather than passive diffusion, these results confirm that NAMPTi 3 maintains strong intracellular activity if delivered by the ADC.

Building on these findings, NAMPTi 4 and its corresponding linker–payload 14 were designed to balance hydrophilicity and lysosomal stability, properties essential for achieving effective bystander activity in solid tumors. Unlike under in vivo conditions, in vitro assays maintain a nearly constant extracellular ADC concentration and a uniform cell population, which can mask the impact of payload stability or diffusion because any metabolic instability of the payload is compensated by continuous ADC influx and direct accessibility of the ADC to the cells. Consequently, differences among the ADC constructs are expected to be less pronounced in vitro. Despite its slightly higher hydrophilicity (Log P = 2.47) compared with NAMPTi 1B, NAMPTi 4 demonstrated roughly 1 order of magnitude greater potency as a free compound (Figure A, green curve), confirming that the structural modification that improved the physicochemical properties did not compromise NAMPT engagement in line with the in silico predictions. In agreement with this, ADCs conjugated with linker–payload 14 maintained strong cytotoxicity and nearly complete cytotoxicity with only 10% viable cells at highest concentrations what might resemble an increased concentration of the lysosomal released payload due to its increased metabolic stability compared to NAMPTi 1B (Figure B, green curve).

To extend these studies toward solid malignancies, we evaluated the HER2⁺/TROP2⁺ human gastric carcinoma cell line NCI-N87 in the same in vitro assays. NCI-N87 has been reported to exhibit high nicotinate phosphoribosyl transferase (NAPRT) expression, enabling NAD⁺ biosynthesis via the Preiss–Handler pathway and thereby reducing dependence on NAMPT. , In agreement with literature reports NAPRT was detectable by Western blot in NCI-N87 cell lysates, whereas NAPRT was not detected in L540 lysates (Supporting Information, Figure S24). Consistent with this NAPRT-positive phenotype, neither the free NAMPT inhibitors nor their corresponding anti-HER2 ADCs showed measurable cytotoxicity in vitro (Figure S23).

Despite this lack of in vitro activity, NCI-N87 was selected for in vivo assessment of NAMPTi-loaded ADCs because tumor metabolism differs markedly between nutrient-rich cell culture conditions and the physiological tumor microenvironment. In vitro, the supplementation with vitamin rich FBS can provide enough niacin to support NAPRT-mediated NAD⁺ production, thereby minimizing the contribution of the NAMPT salvage pathway. In contrast, solid tumors in vivo commonly experience nutrient limitation and hypoxia, conditions that restrict niacin availability and increase dependence on the NAMPT salvage pathway to maintain intracellular NAD⁺. Under these constraints, tumor cell lines that appear insensitive in vitro may become susceptible to NAMPT inhibition in vivo. In line with this, the NAMPTi-loaded ADCs show strong antitumor activity in vivo efficacy results in the NCI-N87 xenograft model as shown in Section 4.4, despite the limited in vitro sensitivity under standard culture conditions. To prove this rationale, follow-up studies using serum-free culture conditions with controlled niacin supplementation/depletion would provide a direct test of whether niacin availability can restore in vitro sensitivity in NCI-N87.

4.2. Efficacy of NAMPTi-ADCs in a Disseminated L540 Mouse Xenograft Model

To translate our in vitro observations into a physiologically more relevant context, we conducted in vivo xenograft studies in NXG mice. The disseminated Hodgkin lymphoma (CD30⁺) model L540 was selected in the first place because it has shown high sensitivity against NAMPTi and serves as a representative model to evaluate the NAMPTi-ADC efficacy in hematological cancers. This disseminated model was established by intravenous injection (i.v.) of L540 cells, which allows for almost full tumor exposure to the ADC (see Supporting Information for full experimental description). ,,

In this study, we focused exclusively on linker–payload 5, incorporating NAMPTi 3. NAMPTi 4 was deliberately engineered to support bystander activity in solid tumors, balancing its cellular permeability, as reflected by its enhanced potency in vitro relative to NAMPTi 3. While advantageous for treating spatially heterogeneous solid malignancies, this property raises concerns in disseminated disease, where unrestricted release of the more permeable NAMPTi 4, following lysosomal cleavage of linker–payload 14, could result in widespread exposure and unacceptable off-target toxicity, consistent with the dose-limiting toxicities reported for systemically delivered NAMPT inhibitors. , Accordingly, evaluation of NAMPTi 4 and linker–payload 14 was restricted to solid tumor models, where the tumor mass itself can act as a diffusion barrier and locally confine the released payload (see section 4.4. “Efficacy of NAMPTi-ADCs in NCI-N87 mouse xenograft model”).

The Kaplan–Meier analysis demonstrated a significant survival benefit in animals treated with the brentuximab-based anti-CD30-NAMPTi-5-DAR10 ADC carrying linker-payload 5, compared to all control groups, including those treated with the unconjugated anti-CD30 mAb and the isotype control ADC (Figure ). Anti-CD30-ADC-treated mice survived beyond 80 days, whereas control group animals succumbed by day 30 (PBS, isotype control) or day 39 (naked mAb).

6.

In vivo efficacy of Anti CD30-NAMPTi 5-DAR10 ADCs in a disseminated L540 model in NXG mice (n = 10 per group). Survival was monitored following treatment with vehicle (PBS, black), isotype control (red, dose regime: 12.5 mg/kg, once weekly for 4 weeks, as indicated by black arrows), or anti-CD30 monoclonal antibody (salmon). Treatment regimens included a multiple-dose schedule shown in green (multiple dose (m.d.); 6.25 mg/kg i.v., once weekly for 4 weeks, as indicated by black arrows) and a single-dose schedule shown in blue (single dose (s.d.); 12.5 mg/kg i.v.). Anti-CD30 linker-payload 5 NAMPTi ADCs significantly prolonged survival in the Hodgkin lymphoma model, demonstrating potent therapeutic efficacy. DAR = 10.

In summary, NAMPT inhibitor ADCs showed strong in vivo antitumor efficacy in the CD30⁺ disseminated L540 mouse xenograft model, leading to a significant improvement in survival despite the intrinsic activity of the unconjugated mAb.

4.3. Efficacy of NAMPTi-ADCs in a Solid L540 Mouse Xenograft Model

To further evaluate the applicability of NAMPTi 3-based ADCs in the more demanding context of solid tumors, L540 cells were implanted subcutaneously in CB17 SCID mice as xenografts coinjected with Matrigel (see Supporting Information for details). The resulting solid L540 model allowed assessment of ADC efficacy under conditions of limited tissue perfusion and restricted macromolecular diffusion, both clearly limiting the bioavailability of the ADC.

Despite the pronounced intrinsic antitumor activity of the unconjugated anti-CD30 mAb, the L540 cell line was further evaluated as a subcutaneous xenograft to examine whether the activity of linker-payload 5 based ADCs observed in the disseminated setting could be retained under the physical constraints of a solid tumor. Using the same tumor model enabled assessment of the impact of tissue architecture, limited macromolecular diffusion, and reduced payload exposure on ADC performance without altering the underlying NAMPT sensitivity of the target cells.

As shown in Figure (salmon line), and consistent with the disseminated model (Figure ), the unconjugated anti-CD30 mAb showed measurable antitumor activity. The magnitude of this mAb-driven effect was unexpected, as immunocompromised mice were used and the Fc-engineered antibody carrying the LALA (L234A/L235A) and D265C substitutions was designed to minimize Fcγ-receptor–mediated effector functions. Although these mutations significantly reduce FcγR binding, residual low-affinity interactions have been reported. Such residual binding could promote limited Fc-dependent receptor clustering in vivo, effectively cross-linking mAb-CD30 complexes at the cell surface.

7.

In vivo efficacy of Anti CD30-NAMPTi 5-DAR10 ADCs in a subcutaneous L540 model in SCID mice (n = 10 per group). Kaplan-Meyer survival curves following treatment with vehicle (PBS, black), isotype control (red; 12.5 mg/kg, once weekly × 4, as indicated by black arrows), anti-CD30 mAb (salmon), and anti-CD30 NAMPTi 5 ADCs: single dose 12.5 mg/kg (green), multiple-dose 6.25 mg/kg once weekly × 4, as indicated by black arrows (light blue), and multiple-dose 12.5 mg/kg once weekly × 4, as indicated by black arrows (blue). Since the unconjugated mAb produced a marked antitumor response, only the high-dose multiple-administration ADC group (blue) achieved a statistically significant further reduction in tumor volume. The strong intrinsic activity of the mAb control partially masked the contribution of the ADC. See also tumor volume plot in Supporting Information section.

This mechanism aligns with previous findings for chimeric anti-CD30 antibodies such as SGN-30, where receptor cross-linking enhanced CD30-mediated signaling, leading to cell-cycle arrest and apoptosis via disruption of TNF-receptor–associated factor (TRAF) pathways. In solid tumors, where cell density is higher and antibody diffusion more restricted, this effect would likely be amplified by perivascular accumulation and localized receptor coalescence. Together, these factors help to explain the differentiated antitumoral activity observed for the unconjugated anti-CD30 groups in vivo, while underscoring the importance of evaluating NAMPTi-loaded ADCs in systems where intrinsic antibody effect is minimized.

Despite the strong baseline activity of the unconjugated antibody, multiple dose ADC treatment (DAR10, linker–payload 5) still produced significant tumor suppression in the solid L540 model, albeit with reduced potency compared to the disseminated setting (Figure ). This attenuation is consistent with the limited penetration of macromolecules into solid tumor tissue, leading to suboptimal local concentrations of ADC relative to the administered dose. For this reason, ADCs targeting solid malignancies often rely on payloads with sufficient passive diffusion capacity to elicit a bystander killing effect within the tumor stroma as well as optimal lysosomal stability. In contrast, NAMPTi 3, the active payload released from Anti CD30-NAMPTi 5-DAR10, is more hydrophilic than the parental compound NAMPTi 1B and likely exhibits restricted passive membrane diffusion. This limitation can lead to subtherapeutic intratumoral payload levels, possibly compounded by lysosomal stability issues associated with the cyanoguanidine group.

Together, these findings motivated the design of a next-generation NAMPT inhibitor (NAMPTi 4) with improved resistance to lysosomal degradation and a more balanced permeability profile to enhance intratumoral distribution. Accordingly, ADCs incorporating the modified linker–payload 14 were advanced for evaluation in subsequent solid tumor studies.

4.4. Efficacy of NAMPTi-ADCs in NCI-N87 Mouse Xenograft Model

Following the promising results obtained with linker–payload 5 ADCs in the L540 models, we assessed efficacy in a more stringent solid tumor setting. For this purpose, we employed the NCI-N87 gastric cancer xenograft, a HER2⁺/TROP2⁺ model. Unlike the L540 model, free anti-HER2 and anti-TROP2 monoclonal antibodies show no intrinsic antitumor activity in this model, providing a clean background to evaluate ADC-specific effects (see Figure ).

8.

In vivo efficacy of Anti HER2- and Anti TROP2-NAMPTi 14-DAR10 ADCs in a subcutaneous xenograft NCI-N87 model in nu-NMRI mice (n = 10 per group). Survival was monitored following treatment with vehicle (PBS, black dashed line), isotype control (gray dashed line), or free monoclonal antibodies (anti-HER2 and anti-TROP2, green dashed lines circles and triangles, respectively) or ADCs directed against HER2 (red lines) and TROP2 (blue lines). Linker-payload 14 NAMPTi ADCs shows remarkable tumor inhibition in gastric cancer models independent of target (I.E. HER2 or TROP2). TI ≥ 50 (in mice). DAR = 10.

The NCI-N87 xenograft presents two major challenges for ADC development: (i) restricted macromolecular penetration and heterogeneous antigen distribution, both intrinsic to solid malignancies, and (ii) intrinsic lower sensitivity toward NAMPTi, as demonstrated in the in vitro studies section. ,,, For these reasons, this model was selected to evaluate our optimized linker–payload 14 based on NAMPTi 4, designed to improve lysosomal stability and balance hydrophilicity for enhanced yet controlled bystander activity. Linker–payload 5 was not tested further in this context, given its suboptimal performance in the L540 subcutaneous solid model and suitability primarily for liquid malignancies.

ADCs carrying linker–payload 14 induced complete tumor remission for over 50 days after a single 2 mg/kg dose, both as anti-HER2 and anti-TROP2 ADC, demonstrating efficacy largely independent of the targeting antibody (Figure ). In both cases, the naked antibodies did not show any antitumor effect, attributing the efficacy of the ADCs clearly to the payload. The minimum effective dose, defined as lowest dosing leading to active tumor remission as compared to treatment start for at least 1 week, was 1 mg/kg, whereas the highest tested dose (50 mg/kg, see SI) was well tolerated, corresponding to a therapeutic index of at least 50. These findings establish linker–payload 14 as a potent and versatile linker-payload for ADCs targeting solid tumors.

We attribute this robust efficacy to the combination of high lysosomal stability, strong NAMPT binding affinity, and fine-tuned hydrophilicity, which together promote sustained intracellular activity and an effective but controlled bystander killing effect. This balanced pharmacological profile enables broad tumor cell engagement even under conditions of restricted antigen accessibility and limited tissue diffusion.

Conclusions

Although several NAMPTi-ADCs have demonstrated encouraging activity, this study shows that significant room for optimization remains and how it can be used to produce best-in-class NAMPTi ADCs. Earlier designs, while effective, raised concerns regarding selectivity, stability, and toxicity. For example, the bioconjugates reported by Karpov et al. exhibit pronounced aggregation and only moderate selectivity, whereas the glucuronide-carbamate linker systems employed by Neumann et al. mitigate payload hydrophobicity by the hydrophilic linker but potentially introduce issues associated with premature extracellular β-glucuronidase cleavage and uncontrolled tumor-microenvironment release of NAMPTi payloads, all factors that may increase systemic toxicities. ,, Similarly, although well tolerated, the NAMPTi-ADCs described by Böhnke et al. required comparatively high doses (10 mg/kg) and multiple administrations to maintain activity in solid tumors. This likely reflects limited intratumoral dispersion of the released NAMPT inhibitor, leaving poorly accessible tumor regions insufficiently exposed and at risk of relapse. In addition, the need for repeated high-dose ADC treatment raises concerns about systemic accumulation, which may constrain tolerability in a clinical setting.

Guided by these limitations, we sought to refine the physicochemical and metabolic properties of NAMPTi and developed two payloads distinctively tailored for the demands of hematologic (NAMPTi 3) and solid (NAMPTi 4) malignancies. Initial in silico and cellular analyses revealed that highly hydrophobic NAMPTi although potent as free agents, they display suboptimal ADC performance, leading to incomplete target-cell killing. We further demonstrated that cyanoguanidine-containing NAMPTi, widely used in first-generation series, exhibit lysosomal instability by undergoing hydration to inactive guanylureas. This intrinsic metabolic instability compromises intracellular exposure of active payloads and constrains ADC efficacy. These limitations establish lysosomal stability and tuned hydrophilicity as essential parameters for NAMPTi-based ADC payload design.

Applying these principles, we developed two next-generation inhibitors incorporating (i) a tertiary alcohol to increase hydrophilicity and improve solubility, and (ii) in NAMPTi 4, an isoindoline-urea headgroup to eliminate the cyanoguanidine-associated lysosomal liability. Computational docking and MD simulations suggested preservation of key binding interactions with NAMPT, while cellular in vitro assays showed that both inhibitors retain potent intracellular activity as free compounds in L540 cells. When integrated into anti-CD30, anti-HER2, and anti-TROP2 ADCs, the corresponding linker-payloads produced consistent and durable antitumor effects in vivo, including complete tumor regressions in the NAMPTi low-sensitivity NCI-N87 model. Notably, the superior performance of linker-payload 14, including robust efficacy at a single 2 mg/kg dose, highlights the value of balancing hydrophilicity and metabolic stability to achieve sustained antitumor activity in solid-tumor settings.

These results position NAMPT inhibition as a viable and tunable MOA and open compelling opportunities for future ADC development.

Finally, consistent with the view that NAMPT inhibition lowers intracellular NAD⁺ and thereby compromises PARP-dependent DNA repair, combination approaches, such as pairing NAMPTi-ADCs with PARP inhibitors or developing dual-payload formats may produce synergistic activity, allow lower dosing, and help overcome resistance to PARP inhibitors.

Taken together, these findings show that the therapeutic value of NAMPTi critically depends on matching their physicochemical profile to the metabolic conditions imposed by ADC trafficking and tumor physiology. By integrating insights from tumor metabolism, lysosomal processing, and small-molecule design, we established a blueprint for optimizing NAMPTi.

Materials and Methods

Detailed methods are provided in the Supporting Information.

5. In Silico Materials and Methods

All in silico studies were performed using Schrödinger Suite 2024–4 (Maestro, Glide, Prime, Desmond) with the OPLS4 force field. Human NAMPT cocrystal structures (PDB IDs: 6E68 and 7PPE) were prepared using the Protein Preparation Wizard (pH 7.4 ± 2.0; restrained minimization; removal of waters beyond 5 Å of ligand). Receptor grids were centered on the cocrystallized ligands with no positional or H-bond constraints. Ligands were prepared with LigPrep/Epik (pH 7.4 ± 2.0; up to 32 states per compound). Docking was performed with Glide SP, and top poses were visually inspected against the pharmacophore model described by Neumann et al. Docked complexes were used for Desmond MD simulations (SPC water; NPT, 300 K, 1 atm; standard relaxation protocol; triplicate simulations with different random seeds as applicable). MM-GBSA binding energies were calculated from equilibrated trajectory segments using Prime as described in the Supporting Information.

6. Biological Techniques: Materials and Methods

6.1. ADC Production

Maleimide-containing payloads were conjugated to fully reduced cysteine residues on Thiomabs to generate high-DAR conjugates. A 5 mg/mL antibody solution in PBS was adjusted to pH 7.4 and 1 mM EDTA. Reduction was performed with 40 mol equiv of freshly prepared TCEP (2 h, 37 °C), followed by two rounds of dialysis at 4 °C against PBS containing 1 mM EDTA. Conjugation was carried out by adding 20 equiv of the respective maleimide linker–NAMPTi (10 μg/μL in DMSO) to the reduced antibody (2 h, room temperature, gentle agitation). For hydrophobic payloads, DMSO was adjusted to a final concentration of 10% (v/v). Residual thiols were capped with 12 equiv of N-ethylmaleimide (1 h, room temperature), and unreacted maleimide was quenched with 40 equiv of N-acetyl-l-cysteine (15 min, room temperature). Crude ADCs were purified by gel filtration (ÄKTA Start FPLC; Superdex 200), dialyzed into PBS, concentrated (Amicon Ultra filters), quantified by A280 (Nanodrop One C; extinction coefficients calculated from sequence using ProtParam), and sterile-filtered (0.22 μm).

Analytical characterization included SEC–HPLC for aggregate quantification (Agilent 1260 Infinity HPLC; Tosoh TSKgel UP-SW3000; isocratic mobile phase described in SI), SDS–PAGE (reducing and nonreducing; Bio-Rad TGX Stain-Free gels), DAR determination by UPLC–MS after Rapid PNGase F deglycosylation (Waters ACQUITY UPLC I-Class Plus with BioAccord RDa; Waters BioResolve column; step gradient with water/acetonitrile +0.1% TFA), and endotoxin measurement by EndoZyme II rFC assay (Hyglos).

6.2. Cell Viability and Cytotoxicity Assays

Cell proliferation of adherent cell lines was assessed using BrdU-ELISA (Roche; 4 h BrdU incubation), and cell viability of nonadherent cell lines was assessed using CellTiter-Glo 2.0 (Promega). Luminescence was measured on a CLARIOstar plate reader. Data were processed in GraphPad Prism 10 by blank subtraction, log­(x) transformation, normalization to untreated controls (100% viability), and sigmoidal dose–response fitting to determine EC₅0 values. Cell line-specific culture conditions are provided in the Supporting Information.

6.3. Western Blot Analysis of NAPRT Expression (L540 vs NCI-N87)

Equal amounts of protein (16 μg total cell lysate per lane) were separated by SDS-PAGE and transferred to PVDF membranes. Membranes were incubated with anti-NAPRT1 (ThermoFisher #PA5-100073; 1 μg/mL) and anti-GAPDH (Abcam #ab181602; 0.2 μg/mL), followed by HRP-conjugated secondary antibody (Cell Signaling #7074; 0.4 μg/mL) and ECL detection. Under these conditions, a NAPRT band (∼55 kDa) was detected in NCI-N87 lysates, whereas no NAPRT signal was detected in L540 lysates.

6.4. In Vitro Lysosomal Catabolism Study of NAMPT Inhibitors

Lysosomal catabolism was investigated using human liver lysosomal extract (Sekisui Xenotech; diluted 1:10 in Catabolism Buffer) over 72 h at 37 °C with sampling at 0, 2, 4, 24, 48, and 72 h. Lysosomal extracts diluted in Catabolism Buffer were preincubated for 10 min at 37 °C before addition of test compound. Reactions were terminated by addition of cold internal standard in acetonitrile (1:20, v/v) and stored at – 80 °C. Catabolism Buffer was pH 5 and contained 2 mM DTT. Control incubations in buffer alone were run in parallel.

Quantification used MRM LC–MS/MS in positive ESI mode (Agilent 1200 LC; QTRAP 5500). Samples were processed by protein precipitation with acetonitrile/internal standard followed by phospholipid removal (Phree 96-well plates). Separation used a Waters ACQUITY BEH Amide column (2.1 × 100 mm, 1.7 μm) at 60 °C with mobile phases of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B), applying a short linear gradient as described in the Supporting Information.

6.5. In Vivo ADC Efficacy Studies in Mice

All animal experiments were performed according to the German Animal Welfare Act and approved by the local regulatory authorities and ethics committee (G64-21, G-149/22). Mice were housed under SPF conditions with ad libitum access to food and water. Animals were monitored daily; body weight was recorded at least twice per week; humane end points were applied per approved protocols.

Disseminated L540 Model (NXG Mice)

Female NXG mice were injected intravenously (tail vein) with 5 × 10⁶ L540 cells suspended in 200 μL RPMI medium without phenol red. Three days after inoculation, animals were randomized and treated by intravenous injection (10 mL/kg) with PBS, ADCs, or antibody controls according to the dosing regimens specified in the figure legends.

L540 Solid Tumor Model (CB17-SCID Mice)

Female CB-17 SCID mice were injected subcutaneously in the flank with 5 × 10⁶ L540 cells suspended in 240 μL RPMI medium without phenol red supplemented with 16.7% Matrigel. Once tumors reached ∼ 150 mm³, mice were randomized and treated with a single intravenous injection (10 mL/kg) of PBS, ADCs or antibody controls as indicated in the figure legends. Tumor volume was measured by calipers and calculated as (W² × L)/2 (L > W).

NCI-N87 Solid Tumor Model (NMRI-Nude Mice)

Female NMRI-nude mice were injected subcutaneously in the flank with 5 × 10⁶ NCI-N87 cells suspended in 240 μL RPMI medium without phenol red supplemented with 16.7% Matrigel. Once tumors reached ∼ 150 mm³, mice were randomized and treated with a single intravenous injection (10 mL/kg) of PBS, ADCs or antibody controls as indicated in the figure legends. Tumor volume was measured by calipers and calculated as (W² × L)/2 (L > W).

6.6. Statistical Analysis

All statistical analyses were performed using GraphPad Prism 10.6.0. Statistical significance thresholds were P ≤ 0.05, 0.01, 0.001, and 0.0001. Tumor growth curves were analyzed using multiple t tests with Holm–Šidák correction. Survival analyses used Kaplan–Meier plots with both log-rank (Mantel–Cox) and Gehan–Breslow–Wilcoxon tests. MM-GBSA distributions were visualized as truncated violin plots, with normality assessed by the Kolmogorov–Smirnov test.

7. Materials and Technical Equipment: Chemistry (Summary)

Preparative RP-HPLC purification was performed using either an ACCQ Prep HP150 (Teledyne ISCO) or an Agilent 1260 Infinity II preparative system, using Phenomenex Luna C18(2) columns (250 × 21.2 mm or 250 × 30 mm; 10 μm) with a matching precolumn; mobile phases and gradients are specified in the Supporting Information. ¹H and ¹³C NMR spectra were acquired (500 MHz for ¹H; 126 MHz for ¹³C) using CDCl₃, DMSO-d ₆, or MeOD, with chemical shifts referenced to residual solvent signals; data were processed using MestReNova (v14.1.2). ESI-MS was performed on an Expression CMS L single-quadrupole instrument (Advion), and HRMS was performed on a Q Exactive Plus (Thermo Scientific) by direct infusion in positive ion mode. Exact synthetic procedures and full characterization data for all intermediates and final compounds are provided in the Supporting Information (including HPLC method tables and NMR spectra).

7.1. Safety

No unexpected or unusually high safety hazards were encountered in the procedures described.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomedchemau.6c00047.

• Additional experimental details, materials and methods, including the full computational workflow (docking and MD simulations), biological procedures (ADC preparation/analytics, cytotoxicity assays, NAPRT Western blotting, lysosomal catabolism LC–MS/MS), in vivo study protocols and statistics, as well as complete chemistry procedures (HPLC methods, step-by-step syntheses/purifications, ¹H and ¹³C NMR spectra for all payloads and linker-payloads) (PDF)

The authors declare the following competing financial interest(s): PR declares no conflict of interest. AMP is a full-time employee of Heidelberg Pharma AG; PR, HG, AH, MV, SJN, VP, DG, AP and TH are full-time employees of Heidelberg Pharma Research GmbH. HG, AH, MV, SJN, VP, DG, AP, AMP, and TH are shareholders of Heidelberg Pharma AG (stocks and/or stock options). Heidelberg Pharma Research GmbH is a German biotech company operating in the field of antibody-drug conjugates and is a subsidiary of Heidelberg Pharma AG.