Accelerating the identification of viable drug candidates across therapeutic areas demands autonomous systems for molecular generation that enforce chemical correctness by structural invariants rather than model capability. We present MolecureAI, an autonomous AI agent architecture for closed-loop drug discovery with cost-ordered cascade evaluation that addresses four capability gaps unresolved across the field: autonomous design strategy selection, persistent cross-campaign structural memory, in-loop ADMET safety gating, and literature-scale evidence integration into the generative loop. A cost-ordered cascade concentrates full structure-based evaluation on a small pre-filtered survivor set, achieving a 2.5 × reduction in candidates advanced relative to those generated. An in-loop metabolic gate enforces Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) safety by writing flagged reactive substructures to a per-target learned state, suppressing identified liabilities from all subsequent generation rounds. A persistent campaign memory influences generation trajectories by embedding prior structural success and failure patterns directly into the design context. We validate these architectural invariants across three structurally diverse benchmark targets (EGFR, KRAS G 12 C , and BRD4) through a five-condition component ablation and a controlled memory ablation, and confirm generator agnosticism across four language-model backends via a deterministic recalibration layer that overrides backend self-evaluation bias. Without explicit chemotype instruction, the pipeline computationally converged toward candidates bearing the 4-anilinoquinazoline scaffold characteristic of the approved inhibitors erlotinib and gefitinib, from structural feedback alone. Persistent campaign memory showed directional but non-significant trends toward lower inter-round score variance and higher late-stage advance rate (n ≈ 90 per condition; all p > 0.20). The component ablation reveals that optimizing for a single objective collapses performance on others; only full orchestration resolves this tension. Advancement rates above 79% across all backends confirm that the validated properties are generator-agnostic, enforced by architectural invariants rather than model capability. All reported outcomes are properties of the computational generation process; prospective biochemical validation of the advanced candidates remains the primary direction for future work.

Current systemic treatments for advanced hepatocellular carcinoma (HCC) such as antiangiogenic targeted therapies, immune checkpoint inhibitors (ICIs), and their combinations have shown benefit, but widespread drug resistance severely limits patient outcomes and underscores an urgent need for effective post-progression strategies. We herein present a case of a patient with advanced HCC who developed cutaneous, pulmonary, and lymphatic metastases following failure of multiple lines of tyrosine kinase inhibitors (TKIs) including donafenib and regorafenib, and lenvatinib plus anti-programmed death-1 (PD-1) monoclonal antibody (tislelizumab) alongside radiotherapy. Next-generation sequencing (NGS) of the primary liver surgical sample confirmed wild-type EGFR status without canonical driver mutations, whereas retrospective immunohistochemistry (IHC) and cyclic multiplexed tyramide signal amplification (CmTSA) of the cutaneous metastatic sample demonstrated EGFR protein overexpression. Subsequent spatial analysis demonstrated that EGFR-high tumor regions strongly correlated with hyper-proliferative Ki67+ tumor cells. Furthermore, this EGFR-high niche was spatially associated with an immunosuppressive microenvironment characterized by the infiltration of PD-L1+ myeloid cells, CD14+ monocytes, regulatory T cells (Tregs), and PD-L1+/IDO1+ tumor cells. Therefore, we formulated a spatial profiling-guided therapeutic hypothesis and added gefitinib to the prior regimen of lenvatinib plus tislelizumab concurrent with radiotherapy, achieving significant tumor regression despite subsequent intracranial progression. In conclusion, these findings suggest that spatial profiling offers valuable therapeutic insights, highlighting the potential benefit of an EGFR-targeted triplet strategy in refractory HCC with EGFR protein overexpression. While confounding effects from concurrent therapies cannot be excluded, this profiling-guided hypothesis warrants further prospective clinical and mechanistic validation.

Non-small cell lung cancer (NSCLC) remains a major cause of cancer-related mortality, while acquired resistance to EGFR-targeted therapy and an immunosuppressive tumor microenvironment continue to limit therapeutic efficacy. Here, we developed a tumor cell membrane-coated, acid-degradable hollow calcium-manganese nanoplatform loaded with gefitinib (CM@HCC-Mn/Gef) for combined redox modulation and antitumor immune activation. Under acidic and redox-active conditions, CM@HCC-Mn/Gef released gefitinib, Ca[2+], and Mn[2+], leading to intracellular Ca[2+] accumulation, mitochondrial depolarization, ATP depletion, reactive oxygen species generation, glutathione consumption, GPX4 downregulation, and lipid peroxidation. These coordinated effects induced ferroptosis-associated oxidative damage and immunogenic cell death (ICD), characterized by cell-surface calreticulin exposure and HMGB1 redistribution, and subsequently promoted dendritic cell maturation and T-cell activation. In gefitinib-resistant NSCLC cells, CM@HCC-Mn/Gef produced substantially greater cytotoxicity than free gefitinib, indicating that multimechanistic stress amplification enhanced the therapeutic response of resistant tumor cells. In immunocompetent syngeneic LLC tumor models, the nanoplatform inhibited primary tumor growth, increased tumor-infiltrating T cells, alleviated immunosuppressive features of the tumor microenvironment, and delayed contralateral secondary tumor progression. Combination with anti-PD-1 further enhanced tumor suppression and systemic immune activation. Collectively, CM@HCC-Mn/Gef integrates targeted drug delivery, calcium-associated mitochondrial stress, redox disruption, ferroptosis-associated damage, and immune activation, providing a biomimetic nanotherapeutic strategy for NSCLC chemoimmunotherapy. STATEMENT OF SIGNIFICANCE: Conventional nanocarriers often provide drug delivery without coordinating multiple tumor-specific stresses and immune activation. This study introduces a tumor cell membrane-coated, acid-degradable hollow calcium-manganese nanoplatform that co-releases gefitinib, Ca[2+], and Mn[2+] in the tumor microenvironment. The material couples calcium-associated mitochondrial dysfunction with manganese-driven redox catalysis, thereby enhancing oxidative damage, lipid peroxidation, and immunogenic cell death-associated responses. It also improves tumor accumulation, promotes dendritic cell maturation and T-cell activation, and enhances anti-PD-1 therapy in immunocompetent lung cancer models. This work establishes a direct link between biomaterial composition, stimulus-responsive degradation, catalytic function, and antitumor immunity, offering a versatile design strategy for biomaterials-enabled chemoimmunotherapy.

Aldehyde oxidase 1 (AOX1) is a cytosolic drug-metabolizing enzyme. Idelalisib is a phosphoinositide-3-kinase-delta inhibitor that undergoes extensive biotransformation, mainly via 8-oxidation, which is catalyzed predominantly by human AOX1 and, to a lesser extent, cytochromes P450. In the present study, the primary objective was to test the hypothesis that idelalisib 8-oxidation is an enzyme-selective catalytic marker of AOX1 in HepG2 human hepatocellular carcinoma cells that express AOX1 but has little or no CYP3A functionality, and this approach is effective in identifying AOX1 inactivators. Enzyme kinetic analysis indicated comparable values of kcat, Km, and ratio of kcat/Km (catalytic efficiency) for idelalisib 8-oxidation catalyzed by HepG2 cells and human liver cytosol, whereas Vmax was greater for liver cytosol-catalyzed idelalisib 8-oxidation. Human recombinant AOX1, CYP3A4, and CYP3A5 were major catalysts of idelalisib 8-oxidation. However, experiments with a panel of enzyme-selective chemical inhibitors indicated that AOX1, but not cytochrome P450 or xanthine oxidase, was responsible for idelalisib 8-oxidation in HepG2 cells and human liver cytosol. The HepG2 cell-catalyzed idelalisib 8-oxidation assay correctly identified time-dependent AOX1 inactivators (erlotinib, O-desmethylerlotinib, O-didesmethylerlotinib, and hydralazine) and noninactivators (gefitinib, O-desmethylgefitinib, O-desmorpholinopropylgefitinib, and raloxifene). The data from the experiment to address our secondary objective indicated that erlotinib dosing increased the accumulation of idelalisib and 8-oxo-idelalisib in mouse plasma, liver, and kidney. In conclusion, (1) idelalisib 8-oxidation is an AOX1-selective catalytic marker in HepG2 cells that can be used to study AOX1 catalytic function; and (2) the in vivo mouse pharmacokinetic approach is not effective in identifying erlotinib as an inhibitor of idelalisib 8-oxidation. SIGNIFICANCE STATEMENT: Idelalisib 8-oxidation as an aldehyde oxidase 1 (AOX1)-selective catalytic marker in HepG2 cells provides a convenient experimental platform to study AOX1 catalytic function, eg, time-dependent inactivation. The experimental tools for AOX1 research can now include a dual AOX1-CYP3A substrate to be used as an AOX1-selective catalytic marker in a cellular system (eg, HepG2 cells) that expresses AOX1 but lacks CYP3A functionality.

Acquired resistance to epidermal growth factor receptor (EGFR)-targeted therapy remains a major obstacle in the treatment of EGFR-mutant non-small cell lung cancer (NSCLC). This study investigated whether repurposing mebendazole (MBZ), a microtubule-disrupting agent, could restore sensitivity to gefitinib in gefitinib-resistant HCC827 GR5 cells. Gefitinib monotherapy showed minimal cytotoxicity, reducing cell viability only marginally (92.7 ± 1.5%), whereas MBZ significantly decreased viability to 79.7 ± 0.6% at 0.1 µM and 29.7 ± 0.6% at 0.5 µM (p < 0.001). Combination treatment (gefitinib 1 µM + MBZ 0.1 µM) further reduced viability to 37.7 ± 0.6% (p < 0.001). Mechanistically, alkaline comet assay revealed significantly increased single-strand DNA damage in the combination group (tail intensity 22.33 ± 0.56%, p < 0.001), while neutral comet assay confirmed elevated double-strand DNA breaks. MBZ also induced dose-dependent γ-H2AX activation (p < 0.001), supporting replication stress-associated DNA damage signaling. In vivo, combination therapy significantly suppressed tumor growth, prolonging tumor doubling time from 6.2 days in controls to 25 days and increasing growth delay to 18.8 days (p < 0.01), without substantial systemic toxicity. Collectively, these findings identify replication stress vulnerability as a therapeutically exploitable axis and support MBZ repurposing as an adjuvant strategy to overcome gefitinib resistance.

Triptolide, a bioactive diterpenoid epoxide isolated from Tripterygium wilfordii Hook. F. (Thunder God Vine), is a potent multi-target natural compound with significant anticancer potential. This systematic review with narrative synthesis evaluates preclinical evidence published between 2015 and 2026 on its molecular mechanisms, cancer-specific responses, and translational prospects, introducing a confidence-based framework to grade the strength of reported mechanisms. Across 51 studies, triptolide showed broad-spectrum anticancer activity in breast, lung, colorectal, pancreatic, hepatocellular, ovarian, prostate, glioma, and osteosarcoma models, acting mainly through mitochondrial apoptosis (92% of studies), reactive oxygen species generation, cell-cycle arrest, and modulation of NF-κB, PI3K/Akt/mTOR, STAT3, and Wnt/β-catenin signaling, with chemosensitization to cisplatin, gemcitabine, pemetrexed, and gefitinib reported in 31% of studies. Clinical translation, however, remains constrained by a therapeutic-index paradox: the same multi-target actions driving efficacy also underlie systemic toxicity, compounded by poor aqueous solubility, low bioavailability, and a narrow therapeutic window. The prodrug Minnelide has reached phase I/II trials, but substantial toxicity persists and no phase III data yet exist. Future progress will likely depend on improved delivery and toxicity-mitigation strategies rather than the discovery of additional molecular targets.

Targeting the epidermal growth factor receptor (EGFR) remains a proven strategy for the treatment of lung adenocarcinoma; however, acquired drug resistance and suboptimal long-term efficacy of the currently available EGFR inhibitors spurred the pursuit of new EGFR-targeted molecules. Hence, a series of pyridine-cored derivatives was rationally designed by strategically integrating pharmacophoric features of Gefitinib. Subsequently, the molecules were synthesised via promising synthetic approaches (Pd-catalyzed CN coupling, nucleophilic substitution reaction), and the characterized compounds (3-5 and 7a-h) were screened at the cellular level using the lung cancer cells (A549) via MTT assay. Consequently, the evaluated compounds were identified as promising leads at the cellular level. Specifically, molecule 5 exhibited potent antiproliferative activity against the A549 cells, with an IC50 value of 8.90 ± 1.6 μM, which was comparable to that of Gefitinib (IC50 = 9.22 ± 1.153 μM). Importantly, molecule 5 exhibited reduced cytotoxicity over the non-cancerous mouse fibroblast cells (L929). Moreover, mechanistic studies of molecule 5 demonstrated enhanced clonogenic survival, modulated intracellular ROS levels, substantially enhanced apoptotic cell populations, and modulation of EGFR protein expression as determined by ELISA. Notably, the combination of molecule 5 with gefitinib revealed the apparent IC50 value of 7.13 ± 0.53 μM under the specific experimental condition. On the other hand, molecular docking studies predicted the favourable binding modes with the EGFR binding sites, with predicted favourable binding affinities ranging from -7.4 to -10.3 kcal/mol. In addition, MD simulation and DFT calculations provide supportive insights into the predicted stability of the ligand-receptor complex and the electronic properties of the synthesised molecules. Collectively, molecule 5 represents as promising pyridine-based scaffold for further investigation as an anti-NSCLC lead, through additional structural modification, and further studies are required to establish its precise molecular mechanism.

Gefitinib improves outcomes in patients with EGFR-mutant non-small cell lung cancer; however, treatment-related gastrointestinal and hepatic toxicity can limit its long-term use. Here, we investigate whether probiotic strains can improve gefitinib efficacy while reducing treatment-associated toxicity. Among the tested strains, Lacticaseibacillus rhamnosus GG (LGG) shows the strongest enhancement of gefitinib efficacy while reducing intestinal and hepatic injury without altering systemic gefitinib exposure. Metabolomic analyses identify propionyl-L-carnitine as an LGG-associated metabolite detected in both LGG culture and LGG-treated mice. Propionyl-L-carnitine treatment reproduces the antitumor effects of LGG and is associated with altered mitochondrial metabolism, decreased NF-κB/IL-6 signaling, and increased apoptosis. A gut-liver-tumor organ-on-a-chip model further supports these effects. LGG treatment also improves short-chain fatty acid profiles, intestinal barrier markers, and bile acid-related metabolism. Together, these findings suggest that LGG may represent a microbiota-based strategy to enhance gefitinib response while limiting treatment-associated toxicity.

Breast cancer is a leading cause of cancer-associated mortality and morbidity worldwide, and there is a continued need for structurally diverse anticancer agents with improved properties. The present study reports a series of diphenylpropan-1-one (DPP) derivatives (4a-4i), rationally designed around a chalcone framework and synthesized via a three-step reaction sequence. The physicochemical and spectral characterization of the synthesized derivatives was performed using spectral techniques. Their anticancer potential was evaluated through a preliminary in vitro cytotoxicity screening against the MCF-7 mammary cancer cell line, followed by an in vivo efficacy assessment using a DMBA-induced mammary cancer animal model. To explore a possible molecular target, in silico molecular docking was performed against the epidermal growth factor receptor (EGFR) tyrosine kinase domain as an exploratory computational analysis. Amongst the synthesized derivatives, compound 4h demonstrated the most potent cytotoxic effects (IC50 = 1.7 µg mL[-1]), comparable to the standard drug gefitinib (IC50 = 1.8 µg mL[-1]). The in vivo studies revealed a significant and dose-dependent decline in cellular proliferation and mammary gland tumor progression, along with histopathological abnormalities upon treatment with compound 4h. Interestingly, the quantitative gene expression analysis demonstrated a significant reduction in the DMBA-induced overexpression of BRCA1, BRCA2, and TP53 in the 4h-treated groups in a dose-dependent manner. Molecular docking studies showed favorable binding of the DPP derivatives within the EGFR active site, with compound 4h exhibiting strong affinity (binding energy = -9.2 kcal mol[-1]) and favorable predicted interactions in the ATP-binding region of EGFR. Computational ADMET profiling provided additional preliminary predictions of the pharmacokinetic and toxicity properties of compound 4h. The identified DPP derivatives demonstrated encouraging preliminary anticancer potential and may serve as promising lead molecules for future investigations.

A series of quinazoline-based PAK4-selective inhibitors were designed via structure-based modification of leads Czh-226 and compound 12i using a heterocyclic tail strategy. Kinase and cell viability assays identified two close analogues, 11j and 11k, differing only by the position of one nitrogen atom. Both showed low nanomolar PAK4 inhibition, high selectivity over PAK1, and low-micromolar antiproliferative activity against three TNBC cell lines, with 11j consistently more potent. Molecular docking and dynamics simulations indicated that 11j formed more stable interactions with the PAK4 catalytic pocket. Mechanistically, 11j inhibited PAK4 and downstream signaling pathways, reduced MAPK activation, up-regulated p53, induced G0/G1 arrest, promoted apoptosis, and raised ROS levels, thereby suppressing TNBC cell proliferation, migration, and invasion. Notably, 11j down-regulated c-Src phosphorylation and sensitized TNBC cells to Gefitinib, an EGFR-TKI. Co-treatment with a non-toxic concentration of 11j markedly lowered the IC50 of Gefitinib across three TNBC lines, and Gefitinib reciprocally enhanced 11j's anti-TNBC effects. In an MDA-MB-231 orthotopic xenograft model, 11j alone at 25 and 50 mg/kg/d significantly and dose-dependently inhibited tumor growth (TGI = 33.70% and 53.17%, respectively), while the combination of 11j (15 mg/kg/d) with Gefitinib (10 mg/kg/d) produced enhanced efficacy (TGI = 60.39%) and reduced lung metastasis without overt toxicity. Collectively, these results validate 11j as a promising lead for TNBC therapy, especially in combination with EGFR-targeted agents.

The quinazoline/quinazolinone ring is known as a unique scaffold, and its derivatives possess a broad biological activity profile, including antibacterial, antifungal, anticonvulsant, anti-inflammatory, anti-HIV, and analgesic activity, primarily focusing on anticancer activity. In this study, the synthesis of 2-[[4-oxo-3-(substituted phenyl)-3,4-dihydro-(substituted quinazolin-2-yl)]thio]-N'-(aryl/heteroaryl methylene)acetohydrazide (4a-4x) derivatives and their potential anticancer activities were investigated on the lung cancer A549 cell line, the breast cancer MCF-7 cell line, and healthy fibroblast L929 cell line. Compounds 4c, 4i, 4m, and 4u were identified as the most cytotoxic and selective molecules on the A549 cell line (IC50: 44.75-78.13 µM), while 4i, 4l, 4m, and 4u were identified as the most cytotoxic and selective molecules on the MCF-7 cell line (IC50: 14.43-29.39 µM). The mechanisms of action of their anticancer activities were examined and studied. It was determined that these compounds induced strong apoptosis and significantly activated caspase-3 activation in both cell types and that they interrupted the cell cycle in the pre-G (sub G0) phase. Compounds 4m and 4u exhibited EGFR inhibition (IC50: 4.80 µM, IC50: 5.40 µM, respectively) at a level similar to the standard drug gefitinib (IC50: 1.86 ± 0.43 µM). Based on the results of the biological activity assays, molecular docking, and molecular dynamics simulation studies, the 4-quinazolinone-acetyl hydrazone scaffold can be considered a promising structural framework with potential anticancer activity. More specifically, the findings of this study indicate that the acetyl moiety may function as an important pharmacophoric group, while the trisubstituted quinazolinone core may represent a favorable structural feature for caspase-3 activation. However, the same structural framework appears to be less favorable for EGFR inhibition, possibly due to steric constraints within the EGFR binding pockets.

Acquired resistance to epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors remains the principal barrier to durable responses in non-small cell lung cancer (NSCLC), yet how the molecular interaction networks underlying this resistance reorganize over time is poorly understood. We reasoned that a systems-level signature of resistance should be visible in the way protein interaction networks remodel under sustained drug exposure, and therefore applied time-resolved network analysis to isogenic gefitinib-sensitive PC9 and gefitinib-resistant PC9R cells across 24 h of gefitinib treatment, computing network entropy and centrality measures for temporal protein interaction networks and interrogating co-expression-augmented networks for candidate resistance-associated bottleneck proteins. Network entropy rose in both phenotypes, indicating that entropic remodeling is a general response to EGFR pathway perturbation rather than a signature of resistance. In contrast, eigenvector entropy was higher in resistant cells at the earliest post-treatment time point, and resistant cells preserved giant-component connectivity and small-world topology early before fragmenting later. Temporal centrality analysis nominated BIRC3 as a resistant-cell-associated high-centrality node at 24 h, and co-expression analysis identified delta-catenin (CTNND1) as a high-betweenness bottleneck in the PC9R network, topologically bridging EGFR, VAV3, HIF3A, and NOTCH2. These findings nominate early post-treatment eigenvector entropy and a CTNND1-centered, EGFR-enriched subnetwork as candidate resistance-associated features that require validation in independent datasets and functional models.