Differences in Neonicotinoid and Metabolite Sorption to Activated Carbon Are Driven by Alterations to the Insecticidal Pharmacophore

Abstract

Widespread application of neonicotinoids has led to their proliferation in waters. Despite low neonicotinoid hydrophobicity, our prior studies implicated granular activated carbon (GAC) in neonicotinoid removal. Based on known receptor binding characteristics, we hypothesized that the insecticidal pharmacophore influences neonicotinoid sorption. Our objectives were to illuminate drivers of neonicotinoid sorption for parent neonicotinoids (imidacloprid, clothianidin, thiamethoxam, and thiacloprid) and pharmacophore-altered metabolites (desnitro-imidacloprid and imidacloprid urea) to GAC, powdered activated carbon, and carbon nanotubes (CNTs). Neonicotinoid sorption to GAC was extensive and largely irreversible, with significantly greater sorption of imidacloprid than desnitro-imidacloprid. Imidacloprid and imidacloprid urea (electronegative pharmacophores) sorbed most extensively to nonfunctionalized CNTs, whereas desnitro-imidacloprid (positive pharmacophore) sorbed most to COOH-CNTs, indicating the importance of charge interactions and/or hydrogen bonding between the pharmacophore and carbon surface. Water chemistry parameters (temperature, alkalinity, ionic strength, and humic acid) inhibited overall neonicotinoid sorption, suggesting that pharmacophore-driven sorption in real waters may be diminished. Analysis of a full-scale drinking water treatment plant GAC filter influent, effluent, and spent GAC attributes neonicotinoid/metabolite removal to GAC under real-world conditions for the first time. Our results demonstrate that the neonicotinoid pharmacophore not only confers insecticide selectivity but also impacts sorption behavior, leading to less effective removal of metabolites by GAC filters in water treatment.

Combining Experimental Sorption Parameters with QSAR to Predict Neonicotinoid and Transformation Product Sorption to Carbon Nanotubes and Granular Activated Carbon

“… 47 Further information regarding sorbents is provided in the Supporting Information . 70 Sorption data for imidacloprid, clothianidin, thiacloprid, thiamethoxam, imidacloprid urea, and desnitro-imidacloprid were analyzed from our previously published data. 47 Additional batch sorption isotherms for acetamiprid, dinotefuran, and thiacloprid amide were generated for this study and quantified following our previously reported methods (see the Supporting Information for details, Figures S1–S3, and Tables S1 and S2).…”

“… 70 Sorption data for imidacloprid, clothianidin, thiacloprid, thiamethoxam, imidacloprid urea, and desnitro-imidacloprid were analyzed from our previously published data. 47 Additional batch sorption isotherms for acetamiprid, dinotefuran, and thiacloprid amide were generated for this study and quantified following our previously reported methods (see the Supporting Information for details, Figures S1–S3, and Tables S1 and S2). 47 All experiments were conducted at pH 7 in buffered solutions, as we had conducted previously, 47 where neonicotinoids are in their neutral/nonionic state ( Table S3 ) and hydrolysis will not be relevant.…”

“… 32 , 72 Additionally, GAC is a complex, heterogenous surface that we previously identified as undergoing numerous interactions that contribute to neonicotinoid sorption (e.g., electrostatic interactions with the pharmacophore, π–π stacking, hydrogen bonding, and nonspecific interactions). 47 Therefore, multiple parameters are required to discern the subtle structural nuances with significant impacts on the sorption propensity of neonicotinoids/transformation products. Initial models for log K ow were constructed around a set of 76 pesticides/pharmaceuticals 53 to compare two modeling approaches, which are presented in detail in the Supporting Information : (A) a traditional approach (akin to the pp-LFER with Abraham-inspired solute/system parameters) that served as a status quo benchmark for predicting partitioning and (B) a tailored approach that is composed of QSAR descriptors most correlated to the specific data set and partition system.…”

“…We recently reported on the extensive sorption of parent neonicotinoids and pharmacophore-altered imidacloprid transformation products to granular activated carbon (GAC, 60–150 mg g –1 ), powdered activated carbon (PAC, 50–90 mg g –1 ), and CNTs (30–140 mg g –1 ). 47 Although neonicotinoids and their transformation products sorbed extensively to black carbon, sorption to GAC for the pharmacophore-altered transformation products desnitro-imidacloprid and imidacloprid urea was significantly lower than was observed for the parent, imidacloprid. 47 Experiments with functionalized and nonfunctionalized carbon nanotubes (nF-CNTs) indicate that the pharmacophore is a significant driver in neonicotinoid sorption, 47 similar to previous observations in soils.…”

“… 32 Utility of pp-LFERs/LSERs is further limited with more complex, heterogenous, phases (e.g., biochar, CNTs, and GAC). 47 Researchers have begun using machine learning and neural networks to find quantitative structure–activity relationships (QSARs) to describe carbon surfaces and adjust pp-LFERs/LSERs to better predict sorption; 61 − 69 however, these models may continue to lead to inaccurate predictions for contaminants like neonicotinoids where sorption is driven by physiochemical interactions with the pharmacophore.…”