Pheophorbide A, a chlorophyll degradation product, exhibits various pharmacological activities and is a good candidate for photodynamic therapy of various tumors as second-generation photosensitizer. In this study, the analytical method was developed and validated to quantify pheophorbide A in biological samples and to assess its absorption, distribution, and excretion in mice. A sensitive, simple, and reliable liquid chromatography-high resolution mass spectrometry method was developed using protein precipitation method as sample preparation procedure, and showed linearity over the concentration range of 1- 500 ng/mL (r = 0.9966), with an intra- and inter-day accuracy and precision at four quality control levels of 93.3-110.0% and 3.5-13.5%, respectively. After an intravenous (2.5 mg/kg) and oral (5 mg/kg) administration of pheophorbide A to male ICR mice, its fecal recoveries for 48 h were 3.2% and 16.3% of the administered dose, respectively, but it was not excreted in the urine. Pheophorbide A showed high volume of distribution at steady state, low oral bioavailability (13.2%), low total recovery, and the delayed time to maximum plasma concentration, indicating the extensive tissue distribution and systemic clearance via the metabolism.
Antibody-drug conjugates (ADCs) have emerged as a powerful class of targeted therapeutics, integrating the selectivity of monoclonal antibodies with the cytotoxic potency of small-molecule payloads. Accurate quantification of payloads within ADCs is essential for evaluating conjugation efficiency, stability, and drug-to-antibody ratio (DAR), yet remains challenging due to the structural heterogeneity and complexity of conjugation chemistry. In this study, an LC-MS/MS based analytical workflow was developed and validated using IgG-FITC as an ADC-like model to establish a practical approach for payload detection and quantification. The workflow consisted of six systematic steps: characterization of free payload fragments, enzymatic digestion of the ADC, verification of overlapping m/z ions between free and conjugated forms, confirmation of their absence in native IgG peptides, construction of an MRM calibration curve, and quantitative evaluation of digested samples. This stepwise approach provides a robust, reproducible, and transferable framework for developing LC-MS/MS methods applicable to payload quantification and characterization in ADC systems, while ensuring methodological reliability and safety during early-phase analytical development.
Mistletoe (Dendrophthoe pentandra (L.) Miq.) is a parasitic plant commonly found on teak trees and has been traditionally used to treat various ailments. This species contains bioactive compounds with significant pharmacological activities, including potential anti-cancer properties. In this study, mistletoe leaf extracts were obtained through maceration using n-hexane as the solvent. Toxicity testing was conducted on Artemia salina larvae using the Brine Shrimp Lethality Test (BSLT), and bioactive compounds were identified through gas chromatography-mass spectrometry (GC-MS) analysis. In silico anticancer activity was assessed using molecular docking, targeting sirtuin 1 (SIRT1) receptors. The toxicity test revealed an LC50 value of 14.38 ppm, indicating strong bioactivity and high toxicity potential. GC-MS analysis identified three major compounds with potential antioxidant and anti-cancer activities: phytol, octadecanoic acid, and n-hexadecanoic acid. Molecular docking results showed that phytol exhibited the highest binding affinity to the SIRT1 receptor at -7.3 kcal/mol, followed by octadecanoic acid and n-hexadecanoic acid, both at -6.4 kcal/mol. These three compounds are predicted to inhibit SIRT1 activity, a key protein involved in the of cancer cells, their as natural anti-cancer agents.
Isotopic analysis of ultra-trace uranium particles for nuclear safeguards requires the highest level of accuracy and reliability. Previously, uranium isotope ratios in NUSIMEP-7 (the 7th Nuclear Signatures Interlaboratory Measurement Evaluation Programme; an interlaboratory comparison/proficiency-test exercise for uranium microparticles) particles were determined using thermal ionization mass spectrometry (TIMS) with an optimized static detection method, yielding satisfactory results. However, the data integration procedure can also significantly influence the final analytical results, especially for ultra-trace level samples analyzed by the continuous heating method. In this study, raw data from the original NUSIMEP-7 particle analysis were retrospectively re-processed using a recently optimized data integration protocol: 'Method I' (Summed Intensity Ratio), with an 'over 25%' signal integration range. The results showed improvements in accuracy for all isotope ratios. Most notably, the z-scores and zeta-scores, which are critical for interlaboratory comparison programs, were significantly improved, moving from 'acceptable' or 'warning' levels to 'excellent'. For the key n( U)/n( U) ratio, the z-score improved from 1.9 to-0.66, and the zeta-score improved from 1.3 to-0.12. This study demonstrates that applying optimized data processing protocols, even retrospectively, can significantly enhance the fidelity and reliability of ultra-trace isotopic analyses.
The recent emergence of large language models (LLMs) has transformed the process of machine learning (ML) model development, markedly reducing the need for advanced coding expertise and enabling domain scientists to directly construct computational workflows through natural language prompts. In this study, we demonstrate the application of Google Gemini Pro 2.5 for developing classification models of erectile dysfunction (ED) drug analogues using tandem mass spectrometric (MS/MS) data. The dataset consisted of 149 compounds, including sildenafil, vardenafil, tadalafil analogues, and structurally unrelated compounds, represented as binary barcode spectra (m/z 50-800) derived from fragment ion intensities. Through stepwise prompting, the LLM generated executable Python code for data preprocessing, model construction, hyperparameter optimization, and ensemble learning using random forest, artificial neural networks (ANN), and support vector machines (SVM). The resulting models achieved high classification performance comparable to that of a manually programmed ANN reported in our previous work, while requiring markedly less programming effort. Beyond reproducing classification accuracy, this study highlights the efficiency, accessibility, and reproducibility of LLM-assisted ML workflows, underscoring their potential to democratize computational methods in mass spectrometry and analytical chemistry.
