PET micro/nanoplastic exposure alters the physicochemical fingerprint of epirubicin–albumin mixtures and modulates bacterial growth and stress responses in Escherichia coli and Staphylococcus aureus

Abstract

Polyethylene terephthalate micro/nanoplastics (PET MNPs) may alter the measured physicochemical behavior of drug–protein formulations. This in vitro study examined whether PET MNP exposure produced differences in epirubicin (EPI)–bovine serum albumin (BSA) formulations and in bacterial responses to their 0.2 µm filtrates. Formulations containing 1, 10, or 100 µM EPI and 2.5 mg mL⁻¹ BSA were evaluated with 0, 50, 250, or 750 µg mL⁻¹ PET MNPs using fluorescence and UV-vis spectroscopy, dynamic light scattering, zeta-potential analysis, Rayleigh scattering, reactive oxygen species measurements, and protein-stability assays. Empirical modeling of intrinsic BSA fluorescence yielded higher but imprecisely estimated K_resp point estimates for the 50 and 250 µg mL⁻¹ PET formulations than for the PET-free formulation, consistent with right-shifted fluorescence-response profiles. Broad bootstrap intervals indicated limited parameter identifiability. The non-monotonic response at 750 µg mL⁻¹ PET was not assigned to a numerical parameter. K_resp was interpreted as a descriptive response parameter rather than a thermodynamic binding constant. UV-vis-based estimation showed a graded reduction in apparent EPI-equivalent response with increasing PET exposure in formulations prepared without BSA, whereas the response remained comparatively stable in BSA-containing formulations. These estimates describe differences in detectable EPI optical response and do not directly quantify free EPI. Complementary measurements identified formulation-dependent differences in fluorescence, absorbance, apparent hydrodynamic populations, surface charge, scattering, and oxidative responses without evidence of extensive bulk protein denaturation. The filtrates produced species- and formulation-dependent effects on bacterial growth and stress-related endpoints in Escherichia coli and Staphylococcus aureus. Collectively, prior PET MNP exposure was associated with changes in the physicochemical fingerprint of EPI–BSA formulations and bacterial responses to their filtrates. However, free EPI was not independently quantified, and contributions from filter-passing PET species or leachable constituents could not be excluded. The findings therefore do not establish binding stoichiometry, thermodynamic affinity, competitive displacement, or redistribution of EPI among aqueous, albumin-associated, and particle-associated fractions.

PET micro/nanoplastic exposure was associated with formulation-dependent changes in empirical EPI–BSA optical and colloidal fingerprints and species-dependent bacterial responses to bulk particle-depleted filtrates.

Introduction

Epirubicin (EPI) is an anthracycline anticancer drug used in the treatment of solid tumors and hematological malignancies, including breast cancer, gastroesophageal cancers, and diffuse large B-cell lymphoma. Its antitumor activity is primarily associated with DNA intercalation, topoisomerase II inhibition, and reactive oxygen species (ROS) generation, which collectively disrupt nucleic acid synthesis and suppress cellular proliferation.^1,2 Despite its therapeutic effectiveness, EPI has several limitations, including rapid systemic clearance, dose-dependent toxicity, cardiotoxicity, multidrug resistance, and variable cellular bioavailability.^3,4 Because the pharmacological activity of EPI depends partly on its distribution between free and protein-bound forms, interactions with circulating proteins and coexisting particulate contaminants may influence its biological behavior.⁵

Serum albumin plays a central role in the transport, distribution, and pharmacokinetics of many therapeutic agents. Its high abundance, structural stability, long circulatory half-life, and multiple ligand-binding pockets make it an important determinant of drug biodistribution and free-drug availability.⁵ Bovine serum albumin (BSA) is frequently used as a model protein because of its structural and drug-binding similarities to human serum albumin. EPI has been reported to bind measurably to plasma proteins, including serum albumin, with approximately 51% albumin binding observed in vitro.¹ Therefore, factors that alter albumin conformation, ligand accessibility, or competitive binding may modify the soluble and protein-associated fractions of EPI and consequently affect its downstream biological activity.

Micro- and nanoplastics (MNPs), particularly polyethylene terephthalate (PET), are increasingly recognized as reactive particulate contaminants rather than biologically inert materials. PET has been detected in human whole blood, indicating that plastic-derived particles can reach the systemic circulation, while PET fibers, films, and fragments have also been identified in indoor and ambient air, supporting inhalation as an additional exposure pathway.^6–8 PET particles can adsorb proteins and pharmaceutical compounds, promote protein-corona formation, alter protein conformation and colloidal stability, and redistribute chemicals among dissolved, protein-bound, and particle-associated fractions.^9,10 Such interactions could potentially change both albumin-mediated drug transport and the amount of pharmacologically available EPI.

However, EPI–albumin binding and MNP–protein or MNP–drug interactions have largely been investigated as separate research areas. Consequently, it remains unclear whether prior PET MNP exposure produces measurable changes in the physicochemical behavior of EPI–albumin formulations and whether the resulting physicochemical changes are sufficient to alter biological responses. The EPI–BSA–PET system was therefore selected because EPI exhibits substantial albumin binding, BSA provides a well-established model of serum albumin-mediated transport, and PET MNPs provide a competing particulate interface capable of interacting with both proteins and aromatic pharmaceutical compounds.^9,10 Furthermore, the intrinsic fluorescence of BSA and the characteristic optical properties of EPI enable changes in their respective molecular microenvironments to be evaluated.

The present study addressed two experimental questions: (i) whether prior exposure to increasing PET MNP levels produces measurable differences in the spectroscopic, colloidal, oxidative, and protein-stability responses of EPI–BSA formulations; and (ii) whether the corresponding 0.2 µm filtrates produce species-dependent growth and stress responses in Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus). The experimental design also compared the UV-detectable EPI response in formulations prepared with and without BSA. However, the study was not designed to independently quantify free EPI or determine its distribution among aqueous, albumin-associated, and particle-associated fractions. E. coli and S. aureus were selected as complementary Gram-negative and Gram-positive models because their distinct cell-envelope structures may produce different responses to components present in the filtered formulations. The lipopolysaccharide-containing outer membrane of E. coli provides a selective permeability barrier, whereas the porous peptidoglycan- and teichoic-acid-rich wall of S. aureus presents different electrostatic and molecular-interaction characteristics.^11–17 Accordingly, formulations containing 2.5 mg mL⁻¹ BSA and EPI at 1, 10, and 100 µM were evaluated following prior exposure to three PET MNP levels. Following incubation, the samples were centrifuged and filtered, and the resulting bulk particle-depleted filtrates were used consistently for both physicochemical characterization (e.g., UV-visible spectroscopy, fluorescence spectroscopy, dynamic light scattering, Rayleigh scattering, reactive oxygen species (ROS) determination, total protein analysis, and protein-denaturation assessment) and bacterial exposure experiments. Bacterial growth, ROS generation, lipid peroxidation, superoxide dismutase activity, protein and FTIR-based biochemical fingerprints were subsequently assessed. The novelty of this study lies in evaluating associations between PET-exposure-related physicochemical differences and bacterial responses after depletion of bulk PET particles and larger particle-associated assemblies, while acknowledging that contributions from filter-passing PET species or leachable constituents cannot be excluded.

Materials and methods

Chemicals and reagents

BSA, EPI hydrochloride and all analytical-grade chemicals were used. EPI hydrochloride was selected as the water-soluble form of the drug, consistent with earlier analytical and binding studies.^1,18

Secondary PET MNPs were prepared using a previously reported top-down mechanical fragmentation procedure.⁷ Commercial light-blue PET drinking-water bottles were used as the source material. The caps, labels, and adhesive-containing sections were removed, after which the bottles were thoroughly washed with ultrapure water and dried. The bottle material was mechanically grated, and the resulting irregular fragments were passed through a stainless-steel sieve to isolate the selected particle fraction. The collected particles were washed again with ultrapure water, dried to constant mass, weighed, and stored in clean glass containers until use. The detailed preparation workflow was reported previously.^6,7 PET polymer identity, elemental composition, hydrodynamic behavior, surface area, and morphology were evaluated using FTIR, EDX, DLS, BET, and TEM, respectively. Detailed physicochemical characterization of this material is reported in Fig. 1 of our previous study.^6,7 TEM imaging showed predominantly small spherical and irregularly shaped particles, together with occasional larger particles of approximately 1.0 µm. DLS analysis showed a hydrodynamic particle-size distribution of approximately 250–490 nm, with a mean size of 395 ± 76 nm; the main number-based particle population was below 400 nm. The particles therefore constituted a heterogeneous MNP population rather than a monodisperse nanoparticle standard. The polymer identity was confirmed by characteristic FTIR bands at approximately 720, 1094, 1241, and 1710 cm⁻¹. EDX analysis of the pristine particles yielded elemental proportions of 62.9% C, 36.2% O, 0.3% N, 0.4% Na, and 0.1% Cl, with an O/C ratio of 0.58. The pristine PET MNPs had a zeta potential of −11.3 ± 1.3 mV, a BET surface area of 0.3257 m² g⁻¹, and a micropore area of 0.0027 m² g⁻¹. The same prepared and previously characterized PET MNP batch was used throughout the study.^6,7

PET MNPs were applied at final concentrations of 50, 250, and 750 µg mL⁻¹, designated MNP1, MNP2, and MNP3, respectively. These concentrations were selected as mechanistic, high-exposure in vitro levels intended to provide sufficient analytical resolution and continuity with the previously characterized PET MNP batch, rather than to reproduce typical environmental mass concentrations. The selected range was based on (i) previous validation of the same PET MNP batch under BSA-, A₅₄₉-, and S. aureus-based experimental conditions, in which these concentrations provided analytically measurable responses while remaining compatible with subsequent biological assessment;^6,7 (ii) the need to cover a sufficiently broad particle-loading range to resolve formulation-dependent changes in fluorescence, scattering, oxidative responses, and colloidal behavior; and (iii) comparability with concentration ranges used in mechanistic MNP hazard and biomolecular-interaction studies.^19–25 Reported mass concentrations of MNPs in freshwater and drinking-water environments are generally substantially lower than those employed in the present study,^19–22 although particle-number concentrations may vary widely depending on particle size, morphology, and analytical methodology.^23,24 Based on the approximate size characteristics of the PET material, the present mass concentrations correspond to estimated particle-number concentrations on the order of 10⁹ to 10¹⁰ particles per mL; however, these estimates are provided only for exposure-scale context and should not be interpreted as demonstrating environmental equivalence because the particles were heterogeneous in size and morphology. Following the rationale of Triebskorn et al.²⁴ and Prata et al.,²¹ although the PET MNP concentrations used here exceed typical ambient mass-based environmental levels, elevated laboratory doses can remain mechanistically informative for hazard identification by providing analytically resolvable exposure contrasts and allowing particle-load-dependent effects to be examined.^22,25 Accordingly, MNP1–MNP3 should be regarded as controlled high-exposure conditions for mechanistic investigation rather than environmentally representative exposure concentrations. The same concentration range and PET MNP batch were retained to enable direct comparison with our previously published PET MNP–protein interaction studies.^6,7

Phosphate buffer (pH 7.4) was used as the interaction medium unless otherwise stated. A BSA solution (5 mg mL⁻¹) was prepared in phosphate buffer. BSA was prepared at a stock concentration of 5 mg mL⁻¹ in 50 mM phosphate buffer at pH 7.4. For preparation of the experimental samples, the BSA stock was mixed 1 : 1 (v/v) with the corresponding EPI solution in the presence or absence of PET MNPs, resulting in a final BSA concentration of 2.5 mg mL⁻¹, equivalent to approximately 37.6 µM, in all assays. This concentration was selected to provide a sufficiently intense intrinsic protein-fluorescence signal while limiting excessive absorbance, inner-filter effects, self-absorption, and fluorescence reabsorption. The selected concentration is also comparable to those used in previous spectroscopic protein–ligand interaction studies, including studies employing approximately 40 µM BSA.²⁶

EPI stock solution was freshly prepared in ultrapure water and diluted with phosphate buffer to yield final concentrations of 1, 10, and 100 µM, referred to as EPI1, EPI2, and EPI3.

Experimental design for EPI–BSA interaction

BSA–EPI interactions were evaluated by fluorescence spectroscopy in the absence and presence of PET MNPs. A BSA solution of 5 mg mL⁻¹ was prepared in phosphate buffer (pH 7.4). EPI working solutions were prepared at 2, 20, and 200 µM. Equal volumes of the BSA and EPI solutions were subsequently mixed at a 1 : 1 (v/v) ratio, resulting in final concentrations of 2.5 mg mL⁻¹ BSA (approximately 37.6 µM, based on a BSA molecular mass of approximately 66.5 kDa) and 1, 10, and 100 µM EPI, respectively. The same BSA : EPI ratio and final concentrations were maintained in all PET-free and PET MNP-containing systems. For the PET-exposed groups, MNP1, MNP2, and MNP3 corresponded to final PET concentrations of 50, 250, and 750 µg mL⁻¹, respectively. Samples were incubated for 30 min under light-protected conditions before subsequent sample processing and physicochemical and bacterial measurement (SI Fig. 1). Following incubation, all BSA–EPI–PET MNP mixtures and corresponding controls were centrifuged and passed through a nominal 0.2 µm syringe filter to reduce bulk PET particles and larger particle-associated assemblies. The resulting 0.2 µm filtrates were used consistently for all physicochemical, oxidative, and bacterial analyses. Thus, the bacterial cultures were exposed to the filtrates obtained after prior incubation in the presence or absence of PET MNPs rather than to the original particle-containing mixtures.

To assess bulk PET MNP depletion during filtration and to reduce particle-related optical interference, PET MNP suspensions were passed through identical nominal 0.2 µm syringe filters and weighed before and after filtration. Filtration efficiency was calculated relative to the corresponding non-filtered controls. The measurements demonstrated reproducible depletion of bulk PET material during filtration, thereby reducing the contribution of larger suspended particles to subsequent spectroscopic measurements. However, this procedure did not establish complete removal of filter-passing PET-derived nanoscale species.

To evaluate possible filtration-induced loss of EPI, free EPI solutions were filtered through identical 0.2 µm syringe filters and analyzed before and after filtration using UV-vis spectroscopy. Filtration recovery was evaluated at 1, 10, and 100 µM EPI using three paired pre- and post-filtration measurements. Mean recoveries were 99.46 ± 1.18%, 94.35 ± 3.40%, and 99.71 ± 0.37% for 1, 10, and 100 µM EPI, respectively. Paired two-tailed t-tests revealed no significant differences between EPI concentrations measured before and after filtration at 1 µM (p = 0.506), 10 µM (p = 0.121), or 100 µM (p = 0.313), indicating that the filtration procedure did not cause statistically detectable EPI loss under the experimental conditions.

BSA-only, EPI-only, MNP-only, BSA + MNP, EPI + MNP, BSA + EPI, and BSA + EPI + MNP groups were analyzed under identical experimental and sample-processing conditions for all applicable physicochemical, spectroscopic, colloidal, oxidative, protein- and bacteria-related measurements. This comprehensive control design was used to distinguish the individual contributions of BSA, EPI, and PET MNPs from effects arising from their binary and ternary interactions. Accordingly, changes in fluorescence, UV-vis absorbance, absorbance-derived indices, Rayleigh scattering, hydrodynamic behavior, zeta potential, ROS generation, and protein-stability-related responses were interpreted by comparison with the corresponding single-component and binary controls rather than being attributed solely to formation of the ternary BSA–EPI–MNP system. Because PET MNP exposure itself altered several EPI-associated optical and physicochemical signals, including EPI fluorescence, these measurements were used as complementary indicators of changes in the molecular and colloidal microenvironment and were not interpreted as direct quantitative measures of free or unbound EPI. Consequently, changes in individual analytical signals were evaluated collectively across the complementary endpoints and controls rather than being assigned to a single molecular mechanism.

All experiments were performed using three independently prepared experimental replicates (n = 3), unless otherwise specified. When duplicate technical measurements were obtained from the same independent preparation, they were averaged before calculation of the group-level summary statistics. Data are presented as mean ± standard deviation (SD).

UV-vis spectroscopy

UV-vis spectra were recorded using a microplate UV-vis spectrophotometer over the 200–800 nm range.^6,7 Particular attention was given to the visible absorption region around 480 nm, corresponding to the characteristic anthracycline chromophore of EPI.¹⁸ Changes in absorbance intensity and spectral profile were subsequently evaluated comparatively to identify formulation-dependent alterations in the optical environment of the EPI–BSA system in the absence and presence of PET MNPs.

UV-vis-based estimation of apparent EPI-equivalent concentrations

Apparent EPI-equivalent concentrations were estimated from absorbance measurements at 480 nm using an external EPI calibration curve described by the equation (A₄₈₀) = 0.006256C + 0.041124, where C is the EPI concentration in µM. For each formulation, the absorbance of the corresponding EPI-free matrix control was subtracted before calculation. Values were expressed as matrix- and dilution-corrected apparent EPI-equivalent concentrations. Formulations prepared without BSA are described as BSA-free formulations. This term indicates only the absence of added albumin and does not represent an analytically isolated free-EPI fraction. Similarly, the calculated values were used for comparative evaluation of the detectable EPI optical response and not for quantification of free, BSA-bound, or PET-associated EPI.

Fluorescence spectroscopy and empirical EPI-dependent BSA fluorescence-response analysis

Following the established separation procedure, samples intended for fluorescence analysis were centrifuged and passed through a 0.2 µm syringe filter. Thus, fluorescence measurements of the MNP1–MNP3 groups represented the bulk particle-depleted filtrate or small-colloidal fraction remaining after prior interaction with PET MNPs rather than fluorescence originating directly from suspended PET particles. Intrinsic BSA fluorescence was recorded at an excitation wavelength of 280 nm over an emission range of 300–450 nm, and the fluorescence intensity at 335 nm was used for quantitative analysis. This spectral region predominantly reflects changes in the microenvironment of the intrinsic aromatic fluorophores of BSA and is widely used for evaluating drug–albumin interactions. EPI fluorescence was independently recorded at an excitation wavelength of 490 nm over an emission range of 520–700 nm. Epirubicin possesses intrinsic fluorescence that is sensitive to its molecular environment, and EPI–BSA interactions have previously been quantitatively investigated by fluorescence spectroscopy, including determination of EPI–BSA association parameters.²⁷

The EPI-sensitive fluorescence channel was used as complementary evidence of changes in the EPI microenvironment but was not used to calculate free EPI concentration because EPI fluorescence was itself substantially modified by PET MNP exposure. Consequently, EPI fluorescence intensity could not be assumed to be directly proportional to the unbound EPI concentration.

The concentration-dependent intrinsic BSA fluorescence data were analyzed using an empirical saturable response model. The model was applied only to summarize differences in the detectable fluorescence response among the tested formulations. It was not used to determine equilibrium binding affinity, binding stoichiometry, binding-site number, free EPI concentration, or the distribution of EPI among aqueous, albumin-associated, and particle-associated fractions.

Quantitative analysis of the EPI-dependent BSA fluorescence response was performed separately for the PET-free, MNP1, MNP2, and MNP3 conditions. The analysis included the condition-specific EPI-free baseline ([EPI] = 0) and the final EPI concentrations present in the measurement mixtures (1, 10, and 100 µM). BSA alone served as the EPI-free baseline for the PET-free system, whereas the corresponding BSA + MNP formulation served as the condition-specific baseline for each PET-exposed system.

The analysis was designed as an empirical description of the concentration-dependent BSA fluorescence response and not as a determination of equilibrium binding affinity, binding stoichiometry, binding-site structure, atomic-level binding mode, or free EPI concentration. The replicate-level fluorescence data were fitted by ordinary nonlinear least-squares regression to a saturable fluorescence-response model:F = F₀ − ΔF_max × [EPI]/(K_resp + [EPI])where F is the observed BSA fluorescence intensity at a given final EPI concentration, F₀ is the fitted fluorescence response in the absence of EPI, ΔF_max is the maximum modeled decrease in fluorescence, and K_resp is the apparent EPI concentration associated with half of the modeled maximum fluorescence decrease. The fitting constraints were F₀ > 0, K_resp > 0, and 0 ≤ ΔF_max ≤ F₀.

K _resp was interpreted exclusively as a descriptive fluorescence-response parameter within the tested in vitro formulation and not as an equilibrium dissociation constant, thermodynamic binding constant, or direct measure of EPI redistribution. No inverse association constant was calculated from K_resp. This conservative interpretation was adopted because the measured fluorescence response may include contributions from multiple, non-exclusive processes, including binding-associated quenching, inner-filter effects, ligand self-association, adsorption, scattering, aggregation, conformational or local environmental changes, and other formulation-dependent optical effects.²⁸

The independent experimental preparation was treated as the analytical unit. When duplicate technical readings were available from the same preparation, they were averaged before model fitting. Nonlinear regression was performed using the replicate-level observations rather than group means. Parameter uncertainty was evaluated using 95% bootstrap confidence intervals, which were reported together with the fitted parameter estimates. Residuals were calculated as the difference between the observed and fitted fluorescence responses. Model adequacy and parameter identifiability were evaluated using the observed-versus-fitted response profiles, residual-versus-concentration plots, root-mean-square error (RMSE), descriptive R² values, bootstrap-interval width, and evidence of boundary-limited or non-monotonic behavior.

A numerical response parameter was not reported when the concentration-dependent response was non-monotonic, when ΔF_max reached the imposed fitting boundary, when the bootstrap interval was unbounded or excessively broad, or when K_resp was otherwise poorly identifiable.

Because only three non-zero EPI concentrations were investigated and free EPI was not independently quantified, reciprocal binding linearizations such as the Benesi–Hildebrand and Scatchard approaches were not used for parameter estimation. In particular, no equilibrium association constant, dissociation constant, or numerical binding-site number was derived from the present dataset. The fitted model was therefore used only to summarize formulation-dependent differences in the empirical BSA fluorescence response.

Rayleigh scattering analysis

Rayleigh scattering (RLS) was assessed from fluorescence scans by monitoring the elastic scattering contribution at or near the excitation wavelength. This analysis was used as a comparative indicator of formulation-dependent light-scattering and colloidal behavior in the BSA–EPI–PET MNP systems. Samples were excited at 350 nm, and spectra were recorded over the range of 350–400 nm. RLS intensity was evaluated at Ex/Em = 350/350 nm.²⁹

Hydrodynamic size, polydispersity index, and zeta-potential measurements

The hydrodynamic diameter, polydispersity index (PDI), and zeta potential of BSA, EPI, PET MNPs, and their binary and ternary combinations were determined using a dynamic light-scattering instrument (Malvern Zeta Sizer) Samples were prepared independently using identical compositions, concentrations, incubation periods, and measurement conditions to those employed in the spectroscopic experiments. Before analysis, the samples were gently mixed to minimize sedimentation without inducing additional particle aggregation. Measurements were conducted at 25 °C. The reported values represent the mean ± standard deviation of independently prepared experimental setups rather than repeated measurements of a single aliquot.

DLS-derived hydrodynamic diameters, PDI values, and zeta potentials were used to characterize the ensemble-level colloidal behavior of the formulations. The hydrodynamic diameters were interpreted as apparent populations detected under the measurement conditions, while the PDI was used as an indicator of the breadth and heterogeneity of the size distribution. Zeta potential was interpreted as an ensemble measure of electrophoretic surface behavior. These measurements did not chemically identify the detected populations, demonstrate the formation of discrete BSA–EPI–PET complexes, or determine EPI binding stoichiometry or fractional distribution.

ROS assay

ROS generation in the BSA–EPI mixtures in the presence and absence of PET MNPs was determined using the DCFH₂-DA fluorescence assay. Samples were incubated with DCFH₂-DA for 30 min at 37 °C, after which fluorescence intensity was measured at an excitation wavelength of 485 nm and an emission wavelength of 528 nm.^6,30

Total protein determination

Total bulk particle-depleted filtrate protein in the interaction mixtures was determined using the Bradford method. Aliquots of each sample were mixed with Bradford reagent, incubated, and absorbance was measured at 595 nm. BSA standards were used for calibration.^6,7

Protein denaturation assay

The ability of the samples to protect BSA against heat-induced structural destabilization was assessed using a commonly applied in vitro protein-denaturation inhibition assay. The samples were exposed to 70 °C under the specified experimental conditions, and the resulting absorbance change was used as an indicator of thermally induced BSA unfolding and aggregation. Inhibition of heat-induced albumin denaturation is frequently employed as a preliminary in vitro indicator of anti-inflammatory potential because protein stabilization may reduce denaturation-associated inflammatory processes. Nevertheless, this accelerated thermal-stress model does not reproduce physiological denaturation and was used here only for comparative evaluation of BSA stabilization in the presence of EPI and PET MNPs.^30,31

Data preprocessing and normalization for physicochemical characteristics

Before normalization, the corresponding buffer, medium, reagent, or probe blank was subtracted from each measurement. For the cell-free BSA–EPI interaction experiments, the blank-corrected signals were expressed relative to the BSA-only control measured under the same assay conditions. Each replicate value was divided by the mean value of the corresponding BSA control, which was assigned a relative value of 1.00. This procedure was applied independently to the UV-vis absorbance values at 238, 277, 292, and 490 nm; fluorescence intensities obtained at Ex 280 nm and Ex 490 nm; ROS fluorescence; protein-denaturation absorbance; and Rayleigh-scattering intensity. The aromaticity and humification indices were first calculated from the blank-corrected absorbance values as Abs₂₅₃/Abs₂₀₃ and Abs₂₅₀/Abs₃₆₅, respectively. Each calculated index was then divided by the mean index of the BSA-only control. Therefore, values below or above 1.00 indicate a decrease or increase, respectively, relative to the BSA control. Normalization was performed separately for each analytical endpoint, and signals obtained using different excitation/emission configurations were not normalized against one another.

Bacterial strains, culture conditions, and bacterial assays

E. coli (ATCC 35218) and S. aureus (ATCC 25923) were used as Gram-negative and Gram-positive bacterial models, respectively. Bacteria were cultured in appropriate nutrient broth under aerobic conditions at 37 °C. Overnight bacterial cultures were adjusted to a standardized inoculum before treatment.^32,33 The standardized bacterial suspensions were exposed for 24 h at 37 °C to the medium control, BSA control, filtered BSA–EPI samples prepared without PET MNPs, or filtered BSA–EPI samples obtained after incubation in the presence of MNP1, MNP2, or MNP3. Before bacterial exposure, all BSA–EPI samples and controls were subjected to the same centrifugation and filtration procedure used for the physicochemical analyses. Thus, the bacterial cells were exposed to the bulk particle-depleted filtrates remaining after prior BSA–EPI–PET MNP incubation rather than to the original bulk particle-containing mixtures. This procedure reduced exposure to bulk PET particles and larger particle-associated assemblies but did not exclude possible contributions from filter-passing PET-derived species or leachable constituents. Following incubation, bacterial growth, oxidative-stress responses, biochemical parameters, and FTIR-based biochemical fingerprinting were assessed.^6,7,30,33 Bacterial growth was monitored by measuring optical density (OD) at 600 nm using a microplate reader. Protein content in the bacterial suspension was determined using the Bradford assay at 595 nm. Intracellular ROS in treated bacterial cells was measured using the DCFH₂-DA assay described above.^6,30 Lipid peroxidation was evaluated in the bacterial suspension using a thiobarbituric acid reactive substances (TBARS)-based spectrophotometric assay at 532 nm.³³ Following reaction with the appropriate assay reagents, absorbance was measured and the results were expressed relative to the corresponding control. Superoxide dismutase (SOD) activity was determined by reacting cell extracts with sodium carbonate buffer, nitroblue tetrazolium, Triton X-100, and hydroxylamine–HCl, followed by measuring absorbance at 560 nm against a blank.³³

Before normalization of the bacterial measurements, the corresponding medium, reagent, or probe blank was subtracted from each signal. An untreated bacterial control containing bacteria alone was included and incubated under the same experimental conditions as all treatment groups. However, for statistical presentation and normalization of the bacterial-response data, the BSA-only group was used as the reference condition because BSA constituted the common protein matrix of the EPI–BSA interaction system investigated in this study. For each bacterial species and analytical endpoint, the blank-corrected value of each replicate was therefore divided by the mean value of the corresponding BSA-only group, which was assigned a relative value of 1.00 (100%). Accordingly, normalized values below or above 1.00 indicate responses lower or higher, respectively, than those observed in bacteria exposed to BSA alone. The bacteria-only control was retained to verify the baseline response of untreated cells under the same incubation conditions but was not used as the normalization denominator. Normalization was performed independently for E. coli and S. aureus because the two species differed in baseline growth, metabolic activity, biochemical composition, and assay-specific optical responses. This procedure was applied separately to OD₆₀₀, Bradford protein content, ROS generation, lipid peroxidation, and SOD activity. After exposure, bacterial cells were harvested by centrifugation, washed, and dried before FTIR analysis. Spectra were recorded in the mid-infrared region, and characteristic bands associated with lipids, proteins, nucleic acids, and carbohydrates were evaluated.³⁰ All FTIR spectra were subjected to the same background subtraction and baseline-correction procedure and were subsequently vector-normalized over the complete analyzed spectral range before comparison of the selected biochemical regions. No additional normalization to the untreated bacterial control was applied.

Statistical analysis

All experiments were performed using three independently prepared experimental replicates (n = 3). When duplicate technical measurements were obtained from the same independent preparation, the technical measurements were first averaged, and the resulting value was treated as one independent replicate. Unless otherwise stated, data are presented as mean ± standard deviation (SD). Inferential statistical analyses were performed using replicate-level data rather than group means.

For comparisons among BSA, EPI, PET micro/nanoplastic (MNP), and their binary and ternary formulations, one-way analysis of variance (ANOVA), followed by Tukey's multiple-comparison post hoc test, was applied within the predefined experimental comparison sets. The same comparison structure was used throughout the physicochemical and bacterial assays. Three predefined comparisons were used for graphical annotation: (a) comparison with the corresponding BSA reference; (b) comparison of each BSA + EPI + MNP formulation with the corresponding BSA + EPI group containing the same EPI concentration; and (c) comparison of each BSA + EPI + MNP formulation with the corresponding EPI + MNP group containing the same EPI and MNP concentrations. Tukey-adjusted p < 0.05 was considered statistically significant. The same comparison framework was applied independently to each analytical endpoint and, for the bacterial assays, separately to E. coli and S. aureus. Statistical annotations were therefore not used to imply comparisons between unrelated EPI concentrations, MNP levels, analytical endpoints, or bacterial species.

For normalized physicochemical measurements, statistical analyses were based on the corresponding independently prepared set values after blank correction and normalization to the BSA reference, as described above. For bacterial-response assays, normalization and statistical analysis were conducted separately for E. coli and S. aureus. The BSA-containing bacterial group served as the reference condition for the normalized bacterial-response data because BSA represented the common protein matrix of the interaction system. The bacteria-only control was retained as an untreated biological reference but was not used as the normalization denominator. This normalization procedure was applied independently to OD₆₀₀, Bradford response, ROS generation, lipid peroxidation (LPO), and superoxide dismutase (SOD) activity.

Pearson product-moment correlation analysis was additionally performed as an exploratory approach. First, relationships among the physicochemical, spectroscopic, oxidative, and colloidal responses of the BSA–EPI–PET MNP system were evaluated across 31 experimental formulation groups. This analysis included BSA intrinsic fluorescence, EPI fluorescence, absorbance at 238, 277, 292, and 490 nm, aromaticity and humification indices, Rayleigh light scattering (RLS), cell-free ROS generation, and the protein-denaturation response. Two-tailed Pearson correlation coefficients (r) and nominal p-values were calculated. Because observations from different EPI concentrations and PET MNP exposure levels were pooled, these correlations were interpreted as treatment-dependent patterns of co-variation rather than as evidence of causality or a specific molecular interaction mechanism. The corresponding physicochemical correlation matrix was presented in the SI. The existing physicochemical Pearson analysis likewise uses 31 observations and two-tailed correlations.

A second exploratory cross-domain Pearson analysis was performed to evaluate whether formulation-dependent physicochemical characteristics co-varied with bacterial responses. Only the 31 formulation-matched groups for which corresponding physicochemical and bacterial measurements were available were included; the bacteria-only control was excluded because it did not have a directly corresponding cell-free physicochemical formulation. Physicochemical variables included BSA and EPI fluorescence, absorbance at 238, 277, 292, and 490 nm, aromaticity and humification indices, RLS, cell-free ROS generation, protein-denaturation response, zeta potential, apparent hydrodynamic particle size, and polydispersity index (PDI). These variables were correlated separately with OD₆₀₀, Bradford response, ROS, SOD, and LPO for E. coli and S. aureus. The expanded analysis therefore directly incorporated the DLS- and zeta-potential-derived colloidal characteristics together with the spectroscopic and oxidative measurements.

For the physicochemical–bacterial correlation matrix, two-tailed Pearson r values and nominal p-values were calculated. Because of the relatively large number of pairwise correlations, Benjamini–Hochberg false-discovery-rate (FDR) adjustment was additionally applied as a sensitivity analysis, and adjusted q-values were reported alongside nominal p-values. Nominal associations were defined at p < 0.05, whereas correlations remaining significant after multiple-comparison adjustment were identified at q < 0.05. The FDR-adjusted results were used to distinguish the more robust associations from relationships that were significant only at the nominal level. Nevertheless, all correlation analyses were considered exploratory because the pooled experimental groups differed simultaneously in EPI concentration, PET MNP level, and formulation composition. Accordingly, correlation coefficients were not interpreted as demonstrating causal relationships, direct physicochemical control of bacterial responses, or a specific molecular mechanism.

Results and discussion

Spectroscopic responses of EPI–BSA formulations

Based on the intrinsic BSA fluorescence data recorded at Ex = 280 nm and Em = 335 nm, the fluorescence response was strongly dependent on EPI concentration and formulation composition (Fig. 1). Relative to BSA (1.000 ± 0.020), EPI1, EPI2, and EPI3 reduced the normalized BSA-associated fluorescence to 0.230 ± 0.005, 0.198 ± 0.011, and 0.115 ± 0.001, respectively, and all three EPI groups differed significantly from BSA (p < 0.05). The MNP-only groups also exhibited substantially lower fluorescence values of approximately 0.216–0.228 and differed significantly from BSA. In contrast, the BSA + MNP systems retained higher fluorescence, with normalized intensities of 0.803 ± 0.033, 0.805 ± 0.012, and 0.769 ± 0.010 for BSA + MNP1, BSA + MNP2, and BSA + MNP3, respectively; nevertheless, all three BSA + MNP formulations remained significantly different from the BSA reference (p < 0.05). BSA + EPI1, BSA + EPI2, and BSA + EPI3 showed progressively lower normalized fluorescence intensities of 0.950 ± 0.008, 0.846 ± 0.014, and 0.670 ± 0.005, respectively, and each differed significantly from BSA. Within the EPI1-containing systems, all three EPI1 + MNP binary groups also differed significantly from BSA. The corresponding BSA + EPI1 + MNP1, BSA + EPI1 + MNP2, and BSA + EPI1 + MNP3 ternary formulations differed significantly from BSA, from BSA + EPI1, and from their corresponding EPI1 + MNP groups (p < 0.05 for all predefined comparisons). A comparable pattern was observed at EPI2: BSA + EPI2 and all EPI2 + MNP binary groups differed from BSA, whereas each BSA + EPI2 + MNP ternary formulation differed from BSA, BSA + EPI2, and its corresponding EPI2 + MNP formulation. At EPI3, BSA + EPI3 and all three EPI3 + MNP binary groups differed significantly from BSA. The BSA + EPI3 + MNP1, BSA + EPI3 + MNP2, and BSA + EPI3 + MNP3 ternary systems also differed significantly from BSA and from their corresponding EPI3 + MNP groups; however, none differed significantly from BSA + EPI3 after Tukey adjustment. Thus, increasing EPI concentration progressively reduced the BSA-associated fluorescence response, while prior PET MNP exposure further modified the fluorescence behavior in a formulation-dependent manner. The absence of a significant ternary-versus-BSA + EPI difference at EPI3, despite clear differences from the matched EPI3 + MNP controls, indicates that the influence of PET exposure on the BSA-associated fluorescence response depended on EPI concentration rather than following a uniform MNP-dose-dependent pattern.

Fig. 1. Normalized intrinsic BSA fluorescence intensity in BSA–EPI–PET micro/nanoplastic (MNP) systems. Subpanels show (A) BSA and EPI groups, (B) BSA and MNP groups, (C) BSA + MNP groups, and the corresponding binary and ternary formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. BSA fluorescence was recorded at Ex = 280 nm and Em = 335 nm. Data are presented as mean ± SD from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations.

The EPI-associated fluorescence response showed a pronounced concentration dependence and was further modified by BSA and PET MNP exposure (Fig. 2). Relative to the BSA reference (1.000 ± 0.243), EPI1, EPI2, and EPI3 produced normalized fluorescence intensities of 5.340 ± 0.379, 100.374 ± 21.874, and 212.295 ± 12.276, respectively. After Tukey adjustment, EPI2 and EPI3 differed significantly from BSA (p < 0.05), whereas EPI1 did not. Among the MNP-only groups, MNP1 and MNP3 differed significantly from BSA, while MNP2 did not. None of the BSA + MNP groups showed a predefined significant difference from BSA. At the EPI1 level, BSA + EPI1 exhibited a normalized fluorescence intensity of 3.107 ± 0.622 and differed significantly from BSA. EPI1 + MNP1 and EPI1 + MNP2 also differed significantly from BSA, whereas EPI1 + MNP3 did not. The ternary BSA + EPI1 + MNP1 and BSA + EPI1 + MNP2 formulations differed significantly from BSA, from BSA + EPI1, and from their corresponding EPI1 + MNP groups. BSA + EPI1 + MNP3 differed significantly from BSA and from EPI1 + MNP3 but not from BSA + EPI1. These results demonstrate that the effect of PET exposure at the lowest EPI concentration was dependent on MNP level rather than being uniformly expressed across all three ternary formulations. At EPI2, BSA + EPI2 and all three EPI2 + MNP binary formulations differed significantly from BSA. BSA + EPI2 + MNP1 differed significantly from BSA, BSA + EPI2, and EPI2 + MNP1. BSA + EPI2 + MNP2 differed significantly from BSA and BSA + EPI2 but not from EPI2 + MNP2, whereas BSA + EPI2 + MNP3 differed significantly from BSA and EPI2 + MNP3 but not from BSA + EPI2. At EPI3, BSA + EPI3 and all EPI3 + MNP binary groups differed significantly from BSA. All three BSA + EPI3 + MNP ternary formulations differed significantly from BSA and from their matched EPI3 + MNP groups, but none differed significantly from BSA + EPI3. Overall, these findings confirm that the EPI fluorescence response was strongly concentration-dependent and that PET MNP exposure modified the detectable EPI microenvironment in a non-monotonic and formulation-specific manner. Because EPI fluorescence itself was strongly affected by PET exposure, these changes were interpreted as optical evidence of altered EPI microenvironment rather than as a direct quantitative measure of free EPI.

Fig. 2. Normalized EPI-associated fluorescence intensity in BSA–EPI–PET micro/nanoplastic (MNP) systems. Subpanels show (A) BSA and EPI groups, (B) BSA and MNP groups, (C) BSA + MNP groups, and the corresponding binary and ternary formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. EPI fluorescence was recorded at Ex = 490 nm over the emission range of 520–700 nm. Data are presented as mean ± SD from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations.

Taken together, the BSA- and EPI-centered fluorescence measurements provide complementary evidence that both EPI concentration and prior PET MNP exposure modified the optical environment of the BSA–EPI system. BSA fluorescence progressively decreased with increasing EPI concentration, whereas EPI-associated fluorescence increased strongly with EPI concentration and showed formulation-dependent modulation after PET exposure. In the BSA-containing systems, significant ternary-versus-BSA + EPI differences were observed mainly at MNP1 and MNP2 for the EPI1 and EPI2 formulations, whereas no such differences were detected at EPI3. In contrast, several ternary formulations differed from their matched EPI + MNP controls, including all three EPI3-containing systems. However, because fluorescence intensity may also be influenced by inner-filter effects, scattering, adsorption, and aggregation, these findings do not independently demonstrate competitive displacement or a defined conformational transition and should be interpreted together with the UV-vis, RLS, DLS, and protein-stability data. Taken together, the BSA- and EPI-centered fluorescence measurements showed that both EPI concentration and prior PET MNP exposure influenced the detectable optical responses of the formulations. These differences may arise from multiple, non-exclusive processes, including binding-associated quenching, local environmental changes, inner-filter effects, scattering, adsorption, and colloidal organization.^34,35 Because free EPI was not independently quantified, the fluorescence data were not used to assign redistribution among aqueous, albumin-associated, and PET-associated fractions or to demonstrate competitive displacement of EPI from BSA.

Empirical analysis of the EPI-dependent BSA fluorescence response following PET MNP exposure

The concentration-dependent intrinsic BSA fluorescence data were analyzed using an empirical saturable response model to summarize differences in fluorescence-response behavior among the PET-free and PET-exposed formulations (Table 1). The model was not used to determine equilibrium binding affinity, binding stoichiometry, binding-site number, or free EPI concentration. The analysis included the EPI-free baseline and final EPI concentrations of 1, 10, and 100 µM. The final BSA concentration was 2.5 mg mL⁻¹, corresponding to approximately 37.6 µM.

Table 1. Empirical EPI-dependent BSA fluorescence-response parameters following prior exposure to PET MNPs. Replicate-level BSA intrinsic fluorescence measurements obtained at 0, 1, 10, and 100 µM EPI were fitted separately for the PET-free, MNP1 (50 µg mL⁻¹), MNP2 (250 µg mL⁻¹), and MNP3 (750 µg mL⁻¹) conditions using a saturable fluorescence-response model. N/A, not applicable; no numerical fit was reported for MNP3.

Condition	F 0, estimate (95% bootstrap interval)	K resp (µM), estimate (95% bootstrap interval)	ΔFmax, estimate (95% bootstrap interval)	R 2	RMSE
BSA + EPI	0.988 (0.974–1.003)	14.89 (11.24–20.04)	0.364 (0.350–0.381)	0.987	0.0144
BSA + EPI + MNP1	0.798 (0.779–0.820)	76.83 (26.60–502.12)	0.266 (0.206–0.789)	0.944	0.0150
BSA + EPI + MNP2	0.790 (0.784–0.799)	275.76 (44.50–494.99)	0.550 (0.209–0.797)	0.942	0.0151
BSA + EPI + MNP3	Not reported	Not reported	Not reported	N/A	N/A

In the PET-free BSA–EPI formulation, increasing EPI concentration progressively decreased the intrinsic BSA fluorescence. Relative to the normalized BSA baseline of 1.000, the fluorescence responses were approximately 0.950, 0.846, and 0.670 at 1, 10, and 100 µM EPI, respectively. These values corresponded to fluorescence decreases of approximately 5.0%, 15.4%, and 33.0%, demonstrating a concentration-dependent change in the detectable BSA fluorescence response.

The complementary EPI-centered fluorescence measurements showed that the EPI signal was also influenced by formulation composition. BSA-free EPI formulations produced normalized fluorescence intensities of approximately 5.34, 100.37, and 212.30 at 1, 10, and 100 µM EPI, respectively, whereas the corresponding BSA-containing formulations produced values of approximately 3.11, 30.70, and 107.75. These differences represented reductions in the detectable EPI-centered fluorescence response of approximately 41.8%, 69.4%, and 49.2%, respectively. However, the EPI + MNP controls also exhibited PET-dependent differences in fluorescence, and fluorescence enhancement rather than attenuation occurred in several PET-containing formulations at the lowest EPI concentration following the addition of BSA. The EPI-centered fluorescence signal was therefore not treated as a direct quantitative measure of free EPI concentration or used to derive a free-versus-bound ligand relationship.

For the PET-free BSA–EPI formulation, nonlinear fitting yielded an F₀ estimate of 0.988 (95% bootstrap interval: 0.974–1.003), a K_resp estimate of 14.89 µM (11.24–20.04 µM), and a ΔF_max estimate of 0.364 (0.350–0.381). The model provided a close descriptive fit to the observed responses, with R² = 0.987 and RMSE = 0.0144.

The MNP1 formulation produced an F₀ estimate of 0.798 (0.779–0.820), a K_resp estimate of 76.83 µM (26.60–502.12 µM), and a ΔF_max estimate of 0.266 (0.206–0.789), with R² = 0.944 and RMSE = 0.0150. The higher K_resp point estimate relative to the PET-free formulation indicated a right shift in the empirical concentration–response profile. However, the wide bootstrap intervals for both K_resp and ΔF_max indicated considerable parameter uncertainty and limited identifiability.

For MNP2, the fitted F₀ was 0.790 (0.784–0.799), while K_resp was estimated as 275.76 µM (44.50–494.99 µM) and ΔF_max as 0.550 (0.209–0.797). The descriptive fit statistics were R² = 0.942 and RMSE = 0.0151. Although the K_resp point estimate was higher than those obtained for the PET-free and MNP1 formulations, it exceeded the highest experimentally tested EPI concentration of 100 µM. Consequently, the MNP2 K_resp and ΔF_max estimates depended substantially on model extrapolation and should not be interpreted as precise quantitative parameters.

At the highest PET exposure level, MNP3, the EPI-dependent BSA fluorescence response was non-monotonic. In particular, the fluorescence response at 10 µM EPI was approximately equal to or slightly higher than the corresponding EPI-free BSA + MNP3 response, whereas marked fluorescence attenuation occurred at 100 µM EPI. This behavior was incompatible with a reliably identifiable monotonic saturable response. Therefore, F₀, K_resp, ΔF_max, R², and RMSE were not reported for MNP3, and its observed responses were retained for qualitative interpretation only.

Observed-versus-fitted response profiles and residual diagnostics are presented in SI Fig. 2. These diagnostics supported the descriptive adequacy of the model for the PET-free, MNP1, and MNP2 formulations, whereas the non-monotonic MNP3 response was retained as observed data without assigning a numerical response parameter.

Overall, the empirical analysis identified formulation-dependent differences in the concentration-dependent BSA fluorescence response. MNP1 and MNP2 produced higher K_resp point estimates than the PET-free formulation, consistent with right-shifted empirical response profiles, whereas MNP3 exhibited non-monotonic behavior that could not be represented reliably by the same model. The relatively high descriptive R² values and low RMSE values for the fitted conditions indicate that the model reproduced the observed fluorescence measurements but do not establish precise parameter identification. This distinction is particularly important for MNP1 and MNP2 because their K_resp and ΔF_max estimates were accompanied by broad bootstrap intervals.

Because only three non-zero EPI concentrations were investigated and free EPI was not independently quantified, K_resp and ΔF_max were interpreted exclusively as descriptive fluorescence-response parameters. The observed changes support PET-associated optical perturbation of the tested formulations but do not establish altered thermodynamic affinity, competitive displacement, binding-site modification, binding stoichiometry, or redistribution of EPI among aqueous, BSA-associated, and PET-associated fractions.

UV-vis responses of EPI–BSA formulations following PET MNP exposure

The UV-vis absorbance profiles showed wavelength-dependent responses across the BSA–EPI–MNP systems (Fig. 3 and SI Fig. 3). The absorbance at 238 nm lies within the peptide-bond-associated far-UV region and can be sensitive to changes in protein structural state.³⁶ At 238 nm, most normalized absorbance values remained close to the BSA reference, indicating comparatively limited formulation-dependent variation in this spectral region (SI Fig. 3-I). EPI1, EPI2, and EPI3 did not differ significantly from BSA after Tukey adjustment. In contrast, MNP1, MNP2, and MNP3 each showed a significant difference from BSA (p < 0.05), whereas none of the BSA + MNP formulations differed significantly from the BSA reference. Within the EPI1-containing systems, the only predefined significant comparison involved BSA + EPI1 + MNP2, which differed from the corresponding EPI1 + MNP2 formulation. No predefined significant differences were detected in the EPI2-containing panel. At EPI3, EPI3 + MNP1 and EPI3 + MNP2 differed significantly from BSA, while BSA + EPI3 + MNP1 differed significantly from its corresponding EPI3 + MNP1 group. No other predefined comparisons reached significance. These findings indicate that the A₂₃₈ response was relatively stable overall, with statistically supported changes restricted to selected MNP-containing formulations rather than showing a systematic EPI- or MNP-concentration-dependent pattern.

Fig. 3. Normalized UV-vis absorbance responses of BSA–EPI–PET micro/nanoplastic (MNP) systems at (I) 277 nm, and (II) 490 nm. Within each main panel, subpanels show (A) BSA and EPI groups, (B) BSA and MNP groups, (C) BSA + MNP groups, and the corresponding binary and ternary formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. Data are presented as normalized mean ± SD relative to the BSA reference from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations.

The absorbance at 277 nm lies within the near-280 nm aromatic region of proteins, which is dominated by tryptophan and tyrosine residues, while EPI also contributes through its aromatic anthracycline chromophore.^18,37 A clearer formulation-dependent response was observed at 277 nm (Fig. 3-I). EPI1, EPI2, and EPI3 all showed significantly higher normalized absorbance than BSA (p < 0.05). Similarly, MNP1, MNP2, and MNP3 differed significantly from BSA. Among the BSA + MNP formulations, BSA + MNP2 and BSA + MNP3 differed significantly from BSA, whereas BSA + MNP1 did not. Within each EPI concentration series, all EPI + MNP binary formulations differed significantly from BSA. In contrast, the BSA + EPI + MNP ternary formulations did not differ significantly from BSA or from their corresponding BSA + EPI groups under the predefined comparison scheme, but each ternary formulation differed significantly from its matched EPI + MNP binary group. This pattern was observed consistently for EPI1, EPI2, and EPI3. Thus, the elevated A₂₇₇ response characteristic of the EPI + MNP binary systems was attenuated after inclusion of BSA, indicating that the protein-containing ternary environment altered the aromatic-region optical response associated with the corresponding EPI + MNP formulations.

The absorbance at 292 nm lies within the near-UV aromatic region and is particularly sensitive to the microenvironment of tryptophan residues; changes in this region can therefore provide complementary information on the tertiary structural state of proteins.^6,38 The A₂₉₂ response showed comparatively high variability and no statistically supported differences within any of the predefined comparison sets (SI Fig. 3-II). Although EPI1 and EPI2 showed numerical decreases relative to BSA and several EPI + MNP formulations exhibited formulation-dependent variation, none of them showed statistical differences. Accordingly, the observed A₂₉₂ differences should be regarded as descriptive trends rather than statistically established effects. Overall, this wavelength was less responsive to formulation composition than A₂₇₇ or A₄₉₀ under the present experimental conditions.

The absorbance at 490 nm predominantly reflects the anthracycline chromophore of EPI,^19,32 whereas the major intrinsic absorption bands of serum albumin occur in the UV region.¹⁸ The strongest concentration-dependent optical changes were observed at 490 nm (Fig. 3-II), consistent with the characteristic visible absorption of the EPI anthracycline chromophore. EPI3 differed significantly from BSA, whereas EPI1 and EPI2 did not reach significance in the EPI-only panel after Tukey adjustment. MNP-only and BSA + MNP groups showed no significant differences from BSA, and no predefined significant differences were detected within the EPI1-associated panel. At EPI2, BSA + EPI2, EPI2 + MNP1, and EPI2 + MNP2 differed significantly from BSA, whereas EPI2 + MNP3 did not. All three BSA + EPI2 + MNP ternary formulations differed significantly from BSA; however, none showed a significant predefined difference from BSA + EPI2 or from its matched EPI2 + MNP formulation. At EPI3, BSA + EPI3, all three EPI3 + MNP binary groups, and all three BSA + EPI3 + MNP ternary groups differed significantly from BSA. No significant b or c comparisons were detected in this panel. Thus, A₄₉₀ primarily reflected the strong EPI concentration-dependent optical contribution, while the absence of matched binary-versus-ternary significance indicates that this wavelength alone did not establish a specific PET-dependent redistribution between the BSA-containing and BSA-free EPI systems.

Overall, the wavelength-specific UV-vis results showed that the response of the BSA–EPI–PET MNP system depended strongly on the spectral region examined. A₂₃₈ and A₂₉₂ showed relatively limited statistically supported changes, whereas A₂₇₇ displayed a clear distinction between EPI + MNP binary formulations and the corresponding BSA-containing ternary systems. A₄₉₀ was dominated by the concentration-dependent EPI chromophore response. These findings support formulation-dependent modification of the optical environment but should be interpreted together with the fluorescence, RLS, DLS, and protein-stability measurements rather than as evidence of a single molecular mechanism.

Following the wavelength-specific UV-vis analysis, absorbance-derived spectral ratios were evaluated as complementary empirical descriptors of formulation-dependent optical changes in the BSA–EPI–PET MNP system (SI Fig. 4). The Abs₂₅₃/Abs₂₀₃ ratio has been used as an indicator of changes in aromatic-ring substitution and associated functional-group composition, whereas Abs₂₅₀/Abs₃₆₅ is commonly related to the apparent molecular size and aromatic character of absorbing organic matter.^39–42 Relative to BSA (1.000 ± 0.051), EPI1, EPI2, and EPI3 significantly increased aromaticity to 1.152 ± 0.004, 1.163 ± 0.009, and 1.137 ± 0.025, respectively (Tukey-adjusted p = 0.0009, 0.0006, and 0.0018, respectively). MNP1–MNP3 also showed significantly higher aromaticity than BSA (1.199–1.210; p = 0.0012, 0.0008, and 0.0008, respectively), whereas the corresponding BSA + MNP systems did not differ significantly from BSA. Binary EPI + MNP systems showed additional formulation-dependent changes that were generally attenuated after BSA incorporation. For example, aromaticity decreased from 1.174 ± 0.020 in EPI1 + MNP3 to 0.931 ± 0.149 in BSA + EPI1 + MNP3 (p = 0.007), and from 1.183 ± 0.026 in EPI3 + MNP2 to 1.005 ± 0.054 in BSA + EPI3 + MNP2 (p = 0.029). In contrast, the remaining matched ternary–binary comparisons and the ternary formulations relative to their corresponding BSA + EPI controls were not statistically significant. Overall, these findings indicate that the aromatic-region response was sensitive to EPI and PET MNP exposure, whereas BSA generally constrained the magnitude of these formulation-dependent optical changes.

The Abs₂₅₀/Abs₃₆₅ spectral ratio showed a less consistent response and substantially greater variability, particularly in BSA-containing formulations (SI Fig. 4-II). Relative to BSA (1.000 ± 0.436), EPI1 and EPI2 remained near unity, whereas EPI3 decreased numerically to 0.634 ± 0.045; however, none of the EPI-only groups differed significantly from BSA (p = 0.989, 0.998, and 0.420, respectively). Similarly, no significant differences were detected for MNP-only or BSA + MNP groups relative to BSA, or among the predefined BSA + EPI, EPI + MNP, and BSA + EPI + MNP comparisons (all Tukey-adjusted p > 0.05). Although the binary EPI3 + MNP series showed a numerical increase from 0.741 ± 0.030 to 0.981 ± 0.054 with increasing MNP level, this pattern should therefore be interpreted as a descriptive trend rather than a statistically established concentration-dependent effect.

Overall, Abs₂₅₃/Abs₂₀₃ aromatic-region ratio was more sensitive than Abs₂₅₀/Abs₃₆₅ spectral ratio to formulation composition. The statistically supported aromaticity changes, together with the comparatively stable Abs₂₅₀/Abs₃₆₅ spectral ratio response, suggest preferential modification of aromatic chromophore exposure or local molecular organization rather than a uniform shift toward greater structural condensation. Because both indices are derived from absorbance measurements, they may also be influenced by adsorption, aggregation, scattering, and overlapping EPI/BSA absorption and should therefore be interpreted together with the fluorescence, UV-vis, and colloidal data.

UV-vis-based apparent EPI-equivalent responses in formulations prepared with and without BSA across PET MNP exposure conditions

UV-vis analysis demonstrated different PET-associated response patterns in BSA-free and BSA-containing EPI formulations (Table 2). In the BSA-free system containing nominally 100 µM EPI, the apparent EPI-equivalent concentration decreased from 106.47 ± 1.24 µM in the PET-free formulation to 56.36 ± 0.05, 49.76 ± 0.75, and 43.17 ± 0.57 µM in the MNP1, MNP2, and MNP3 formulations, respectively. These values represented 52.9%, 46.7%, and 40.5% of the corresponding PET-free EPI response. A similar graded decrease was observed at the lower EPI levels. In contrast, the apparent EPI-equivalent concentration remained comparatively stable across the BSA-containing formulations. At nominally 100 µM EPI, the PET-free BSA–EPI formulation yielded 43.06 ± 2.12 µM, whereas the MNP1, MNP2, and MNP3 formulations yielded 43.67 ± 2.22, 41.13 ± 4.86, and 41.40 ± 1.58 µM, respectively. These responses corresponded to 101.4%, 95.5%, and 96.1% of the PET-free BSA–EPI response and did not exhibit a graded PET concentration-dependent pattern. Thus, PET MNP exposure was associated with a marked decrease in the detectable EPI response in BSA-free formulations, whereas this PET-associated difference was attenuated in the BSA-containing formulations. The observed contrast reflects formulation-dependent optical behavior and does not demonstrate redistribution or displacement of EPI among molecular fractions.

Table 2. Matrix-corrected apparent EPI-equivalent concentrations in BSA-free and BSA-containing formulations following exposure to increasing PET MNP levels. EPI-equivalent responses were estimated from absorbance at 480 nm using the corresponding matrix controls. Values are presented as mean ± SD from three independently prepared samples (n = 3). The percentage response relative to the nominal EPI concentration is shown in parentheses. Negative values following matrix-control subtraction indicate responses close to the analytical baseline and do not represent negative EPI concentrations.

Sample	EPI1 (1 µM)	EPI2 (10 µM)	EPI3 (100 µM)
EPI	0.71 ± 0.15 (71.1%)	9.32 ± 0.70 (93.2%)	106.47 ± 1.24 (106.5%)
EPI + MNP1	0.42 ± 0.01 (42.4%)	5.67 ± 0.20 (56.7%)	56.36 ± 0.05 (56.4%)
EPI + MNP2	0.19 ± 0.32 (19.2%)	3.62 ± 0.42 (36.2%)	49.76 ± 0.75 (49.8%)
EPI + MNP3	0.03 ± 0.59 (3.2%)	2.32 ± 0.38 (23.2%)	43.17 ± 0.57 (43.2%)
BSA + EPI	0.01 ± 0.01 (0.8%)	3.38 ± 0.31 (33.8%)	43.06 ± 2.12 (43.1%)
BSA + EPI + MNP1	−0.04 ± 0.19 (−4.0%)	3.58 ± 0.27 (35.8%)	43.67 ± 2.22 (43.7%)
BSA + EPI + MNP2	0.93 ± 0.02 (92.6%)	3.75 ± 0.71 (37.5%)	41.13 ± 4.86 (41.1%)
BSA + EPI + MNP3	0.35 ± 0.09 (35.2%)	3.43 ± 0.37 (34.3%)	41.40 ± 1.58 (41.4%)

The graded decrease observed in the BSA-free formulations may reflect multiple processes, including PET-associated adsorption or retention during sample processing and changes in the optical or colloidal environment of EPI. The comparatively stable response across the BSA-containing PET formulations suggests that the presence of BSA attenuated the PET-associated change in the detectable EPI signal. However, the lower absolute response of the BSA-containing formulations should not be interpreted as evidence of reduced EPI concentration or increased albumin binding because the EPI calibration was not matrix-matched to the BSA-containing samples. Accordingly, the UV-vis results do not establish competitive displacement, binding-mediated protection, or molecular redistribution of EPI.

Colloidal and thermal-stability responses of EPI–BSA formulations

DLS and zeta-potential measurements showed clear composition-dependent changes in the hydrodynamic and interfacial properties of the BSA–EPI–PET MNP systems (Table 3). BSA alone exhibited an apparent hydrodynamic diameter of 7.0 ± 13.2 nm and a zeta potential of −16.3 ± 6.0 mV. Addition of EPI increased the apparent diameter of the BSA-containing systems to 10.3–22.3 nm and shifted their mean zeta potentials to −9.0 to −1.5 mV, indicating modification of the BSA-associated hydrodynamic and surface-charge environment. Changes in particle size and zeta potential following albumin interaction have similarly been reported for PET micro/nanoplastics and other nanoplastic–albumin systems, supporting their use as complementary indicators of particle–protein interfacial reorganization.^6,7,43,44 PET MNPs alone showed increasing apparent diameters of 46.5, 56.4, and 126.0 nm for MNP1, MNP2, and MNP3, respectively, whereas their zeta potentials varied non-monotonically from −9.7 to −3.1 mV. Thus, particle loading influenced the hydrodynamic distribution more consistently than the surface-potential response.

Table 3. Zeta-potential and DLS results of the samples. The reported SD values were obtained from independently prepared samples analysed under identical experimental conditions.

Sample	Zeta mean (mV)	Zeta SD	Detectable/filter-passing hydrodynamic populations size mean (d. nm)	Detectable/filter-passing hydrodynamic populations size SD (d. nm)	PDI
BSA	−16.3	6.0	7.0	13.2	0.37
EPI1	−0.6	13.7	27.9	10.3	0.39
EPI2	−6.5	3.3	89.4	8.1	0.52
EPI3	−3.0	13.3	74.7	11.5	0.64
BSA + EPI1	−9.0	12.5	10.3	10.3	0.39
BSA + EPI1 + MNP1	−7.8	2.3	10.8	2.8	0.50
BSA + EPI1 + MNP2	−13.1	10.9	36.4	9.3	0.24
BSA + EPI1 + MNP3	−7.4	0.3	21.8	3.2	0.18
BSA + EPI2	−1.5	2.4	22.3	14.5	0.54
BSA + EPI2 + MNP1	−8.2	4.5	29.0	4.5	0.29
BSA + EPI2 + MNP2	−24.8	3.6	12.0	3.6	0.59
BSA + EPI2 + MNP3	−7.4	2.9	6.3	3.9	0.33
BSA + EPI3	−4.0	24.1	16.2	7.4	0.61
BSA + EPI3 + MNP1	−4.4	3.4	17.5	7.1	0.63
BSA + EPI3 + MNP2	−9.7	3.3	16.7	3.4	0.73
BSA + EPI3 + MNP3	−5.9	8.5	29.9	3.7	0.36
MNP1	−9.7	0.5	46.5	11.2	0.41
MNP2	−6.7	0.9	56.4	5.4	0.39
MNP3	−3.1	1.2	126.0	5.9	0.73
BSA + MNP1	−17.8	0.0	49.0	4.7	1.00
BSA + MNP2	−11.3	4.4	7.3	11.2	0.39
BSA + MNP3	−17.6	5.0	21.1	5.0	0.32
EPI1 + MNP1	−14.7	10.9	14.0	13.5	0.94
EPI2 + MNP1	−17.3	0.2	73.5	5.2	0.70
EPI3 + MNP1	−4.5	28.0	108.0	5.1	0.39
EPI1 + MNP2	−20.2	0.4	22.6	9.4	0.75
EPI2 + MNP2	−8.9	0.3	79.6	3.6	0.54
EPI3 + MNP2	−5.7	0.2	38.3	2.9	0.41
EPI1 + MNP3	−12.0	1.8	53.7	3.7	0.75
EPI2 + MNP3	−3.3	0.7	85.0	16.9	1.00
EPI3 + MNP3	−3.6	0.5	95.3	10.0	0.11

The binary EPI + MNP systems displayed highly formulation-dependent apparent sizes (14.0–108.0 nm) and PDI values (0.11–1.00), indicating substantial differences in the breadth and organization of the detectable hydrodynamic populations. Notably, addition of BSA reduced the apparent mean diameter in eight of the nine corresponding EPI + MNP-to-BSA + EPI + MNP comparisons. For example, the EPI2 + MNP1, EPI2 + MNP2, and EPI2 + MNP3 sizes decreased from 73.5, 79.6, and 85.0 nm to 29.0, 12.0, and 6.3 nm, respectively, after BSA addition. A comparable reduction was observed at EPI3, whereas EPI1 + MNP2 was the only condition showing an increase (22.6 to 36.4 nm). In contrast, zeta-potential changes showed no consistent direction across the ternary formulations, ranging from −4.4 to −24.8 mV. Together with the variable PDI values, these findings suggest that BSA inclusion was associated with differences in the apparent hydrodynamic populations detected across the formulations. Such heterogeneous behavior is consistent with protein-corona studies showing that albumin association can produce either modest changes in hydrodynamic diameter or extensive aggregation depending on particle size, surface properties, protein concentration, and corona state.^43–46 Because several formulations were moderately to highly polydisperse, the DLS values should be interpreted as apparent hydrodynamic populations rather than the dimensions of discrete BSA–EPI–PET formulations.⁴⁶

The differences in hydrodynamic diameter, PDI, and zeta potential indicate that PET exposure altered the ensemble-level colloidal behavior of the formulations. The relatively broad or heterogeneous PDI values observed in several conditions suggest the presence of overlapping hydrodynamic populations rather than a single uniform species. Because DLS and zeta-potential measurements do not provide chemical or molecular identification, these changes were not interpreted as evidence of discrete BSA–EPI–PET complex formation, altered binding stoichiometry, or redistribution of EPI among formulation fractions. Collectively, the DLS, PDI, zeta-potential, and fluorescence measurements support formulation-dependent colloidal and optical perturbations following PET MNP exposure. These responses may arise from multiple, non-exclusive processes, including adsorption, scattering, aggregation, changes in the local fluorophore environment, inner-filter effects, and reorganization of the colloidal populations. However, the combined measurements do not distinguish among these possibilities and cannot establish binding stoichiometry, thermodynamic affinity, competitive displacement, or redistribution of EPI among aqueous, BSA-associated, and PET-associated fractions.

The RLS results provided a complementary but comparatively stable response across the tested formulations (SI Fig. 5). Normalized RLS intensities remained close to the BSA reference, with MNP-only groups tightly distributed at approximately 1.030–1.035 and the BSA–EPI–MNP ternary formulations generally remaining within approximately 0.922–1.071. No predefined Tukey-adjusted comparisons reached statistical significance. Thus, the formulation-dependent reorganization of the apparent hydrodynamic populations detected by DLS was not accompanied by a statistically supported increase in bulk Rayleigh scattering. Together, the DLS, zeta-potential, and RLS data are therefore more consistent with heterogeneous changes in colloidal organization than with extensive large-scale aggregation. This interpretation is consistent with studies showing that albumin–nanoplastic association can generate distinct soft- and hard-corona states with markedly different hydrodynamic and aggregation behavior.^43,46

Oxidative and protein-stability responses of the EPI–BSA formulations following PET MNP exposure

The normalized cell-free ROS response varied markedly with EPI concentration and formulation composition (Fig. 4-I). Relative to BSA (1.000 ± 0.529), EPI1 reduced the ROS response to 0.223 ± 0.139, EPI2 produced a numerical increase to 1.395 ± 0.798, and EPI3 generated the strongest response at 5.175 ± 1.279. Among these groups, only EPI3 differed significantly from BSA after Tukey adjustment (p < 0.05). This pronounced increase indicates that the highest EPI level substantially enhanced the oxidative potential of the system, consistent with the well-established ROS-generating behavior of anthracyclines through redox-active quinone chemistry.^47,48 By contrast, the MNP-only groups showed consistently low ROS values of 0.323–0.331, indicating that the tested PET MNPs alone did not strongly promote ROS generation under the present cell-free conditions. The BSA + MNP systems produced higher mean ROS responses than the corresponding MNP-only groups, reaching 1.606 ± 0.425, 1.331 ± 0.382, and 1.317 ± 0.372 for BSA + MNP1, BSA + MNP2, and BSA + MNP3, respectively. Although these differences were not statistically significant relative to BSA, the numerical increase after combining BSA with MNPs may reflect changes in particle dispersion and interfacial organization following protein adsorption and the possibility of corona formation, processes known to modify the physicochemical behavior of nanoplastic–albumin systems.^43,45 The absence of a monotonic response across MNP1–MNP3 further indicates that nominal particle concentration alone did not determine ROS generation.

Fig. 4. Normalized reactive oxygen species (ROS) generation and heat-induced BSA denaturation/aggregation responses of BSA–EPI–PET micro/nanoplastic (MNP) systems. Main panel (I) shows normalized ROS generation and main panel (II) shows the normalized heat-induced BSA denaturation/aggregation response. Within each main panel, subpanels show (A) BSA and EPI groups, (B) BSA and MNP groups, (C) BSA + MNP groups, and the corresponding binary and ternary formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. Data are presented as mean ± SD from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations.

In the EPI1-containing systems, BSA + EPI1 showed a normalized ROS value of 0.863 ± 0.618. The binary EPI1 + MNP formulations remained lower, ranging from 0.408 to 0.505, whereas the corresponding ternary BSA + EPI1 + MNP systems ranged from 0.867 to 1.090. None of these groups differed significantly from BSA + EPI1. The lower ROS response of the binary EPI1 + MNP systems may reflect formulation-dependent differences in the optical, redox, or colloidal environment.² Although the BSA-containing ternary formulations showed numerically higher ROS responses than the corresponding binary EPI1 + MNP formulations, the predefined pairwise comparisons were not statistically significant. These differences were therefore interpreted as formulation-dependent descriptive variation without assigning a specific binding-, adsorption-, or corona-related mechanism.^27,43,45 For the EPI2 systems, BSA + EPI2 produced a normalized ROS value of 1.301 ± 0.864. The binary EPI2 + MNP1, EPI2 + MNP2, and EPI2 + MNP3 groups yielded 1.425 ± 0.116, 1.108 ± 0.055, and 1.044 ± 0.142, respectively. The binary EPI2 + MNP formulations showed a numerical decline in ROS response across MNP1–MNP3. However, pairwise Tukey-adjusted comparisons among the three PET levels were not statistically significant. This pattern was therefore considered descriptive and cannot be used to infer PET-load-dependent redistribution, adsorption, or changes in EPI accessibility.² The ternary BSA + EPI2 + MNP systems instead ranged from 1.372 ± 0.929 to 1.640 ± 1.116, but the comparatively large variation prevented statistical significance relative to BSA + EPI2. The highest absolute ROS responses occurred in the EPI3-containing systems. BSA + EPI3 reached 3.071 ± 2.007, whereas the binary EPI3 + MNP groups showed values of 4.816 ± 0.169, 4.226 ± 0.142, and 4.215 ± 0.035 for MNP1, MNP2, and MNP3, respectively. Numerically, EPI3 + MNP1 produced the highest response; however, no predefined comparison established a statistically significant difference among the tested PET levels. The corresponding ternary systems ranged from 3.376 to 3.482 and did not differ significantly from BSA + EPI3 because of their comparatively large standard deviations. Thus, the high oxidative response associated with EPI3 was retained in the binary EPI + MNP systems, whereas inclusion of BSA reduced the magnitude of the response relative to several corresponding binary formulations. The mean ROS responses of the BSA-containing ternary formulations were lower than those of several corresponding binary EPI3 + MNP formulations. However, because of the comparatively high variability and the absence of significant differences from BSA + EPI3, these results do not establish a BSA-mediated protective effect, altered EPI association, or a specific corona-related mechanism.^27,43,45

Overall, the cell-free ROS response was dominated by EPI concentration, with EPI3 producing the only statistically supported increase relative to BSA among the predefined comparisons. Although several EPI + MNP and BSA + EPI + MNP formulations showed numerical variation with MNP level, these differences did not reach significance under the predefined Tukey-adjusted comparison framework. PET-associated changes in the ROS response should therefore be interpreted as non-monotonic descriptive variation rather than as evidence of a statistically established particle-load-dependent effect. The numerical convergence of several BSA-containing formulations toward intermediate ROS values is compatible with the possibility that the protein environment modifies EPI-associated oxidative behavior, but the present data do not demonstrate a uniform BSA-mediated protective or enhancing effect. The comparatively high variability of several BSA-containing groups further supports cautious mechanistic interpretation.

The protein-denaturation assay was used to evaluate the resistance of BSA-containing formulations to heat-induced structural destabilization and aggregation (Fig. 4-II). Thermal treatment of BSA is known to induce progressive unfolding followed by intermolecular association and aggregation, making heat-stressed BSA a useful comparative model of protein structural stability.^49,50 Although albumin-denaturation assays are also widely employed as preliminary in vitro protein-stabilization or anti-inflammatory screening models, the present results were interpreted only in terms of relative thermal stability and not as evidence of physiological anti-inflammatory activity.^51,52 The protein-denaturation assay showed only limited variation among the tested formulations (Fig. 4-II). Normalized values generally remained close to the BSA reference, within approximately 0.98–1.08. EPI-only, MNP-only, BSA + MNP, EPI + MNP, BSA + EPI, and BSA + EPI + MNP formulations showed no predefined Tukey-adjusted significant differences within any of the six panels. These results indicate that the pronounced optical and oxidative responses observed in selected formulations were not accompanied by statistically detectable changes in the bulk heat-induced BSA denaturation/aggregation response. The data are therefore more consistent with formulation-dependent changes in local molecular environment and colloidal organization than with extensive global destabilization of BSA under the tested conditions.

Taken together, these findings indicate that EPI and prior PET MNP exposure produced only limited changes in bulk BSA thermal stability under the applied conditions. Importantly, the pronounced oxidative response observed at EPI3 was not accompanied by a comparable increase in the heat-induced denaturation/aggregation signal. Thus, the fluorescence, UV-vis, ROS, DLS, and RLS responses are more consistent with localized changes in molecular environment and formulation-dependent colloidal organization than with extensive global unfolding or large-scale protein aggregation. This interpretation is consistent with molecular-level studies showing that PET nanoparticle association with serum albumin can occur without major disruption of the overall protein secondary structure.⁵³

Exploratory relationships among spectroscopic and biochemical responses in the BSA–EPI–PET MNP system

Pearson correlation analysis was used as a descriptive exploratory approach to identify patterns of co-variation among the spectroscopic, oxidative, and colloidal parameters of the BSA–EPI–PET MNP system across 31 experimental groups (SI Fig. 6). Because the analysis pooled observations obtained at different EPI and PET MNP concentrations, the correlations may partly reflect shared treatment- and concentration-dependent responses. In addition, the reported p-values were not adjusted for multiple comparisons. Therefore, the observed relationships were interpreted as exploratory associations rather than evidence of causality or definitive proof of a specific interaction mechanism.

BSA intrinsic fluorescence exhibited a very strong inverse correlation with absorbance at 277 nm (r = −0.972, p < 0.001) and a strong inverse correlation with the aromaticity index (r = −0.763, p < 0.001). In contrast, BSA fluorescence was positively correlated with absorbance at 292 nm (r = 0.603, p < 0.001). A₂₇₇ was also strongly positively associated with aromaticity (r = 0.806, p < 0.001) and negatively associated with A₂₉₂ (r = −0.621, p < 0.001). These relationships indicate that changes in BSA fluorescence occurred concurrently with treatment-dependent alterations in the UV spectral region associated with aromatic protein residues. However, because some of these indices are derived from related absorbance regions, the correlations should not be interpreted independently as direct evidence of protein unfolding or a particular conformational transition.

A distinct cluster of positive correlations was observed among EPI fluorescence, absorbance at 490 nm, and ROS generation. EPI fluorescence was strongly correlated with A₄₉₀ (r = 0.912, p < 0.001) and ROS (r = 0.952, p < 0.001), while A₄₉₀ was also strongly correlated with ROS (r = 0.880, p < 0.001). These findings demonstrate that increasing EPI-associated optical responses occurred concurrently with increasing oxidative response across the experimental groups. Nevertheless, BSA fluorescence was not significantly correlated with EPI fluorescence (r = −0.344, p = 0.058), A₄₉₀ (r = −0.293, p = 0.110), or ROS (r = −0.139, p = 0.457). Therefore, the pooled correlation analysis does not support a direct statistical relationship between BSA fluorescence quenching and ROS generation and cannot establish that changes in BSA fluorescence were responsible for the observed oxidative effects.

Absorbance at 238 nm showed inverse correlations with EPI fluorescence (r = −0.685, p < 0.001) and ROS generation (r = −0.653, p < 0.001). This pattern suggests that the treatment-dependent increase in EPI-associated fluorescence and oxidative response was accompanied by a reduction in the A₂₃₈ response. Because absorbance in this region may be influenced by peptide-bond transitions, compound absorption, turbidity, and particle-associated optical effects, this association should be considered a system-level spectral response rather than direct evidence of a specific change in the protein backbone.

The Abs₂₅₀/Abs₃₆₅ spectral ratio displayed strong inverse correlations with EPI fluorescence (r = −0.742, p < 0.001), A₄₉₀ (r = −0.782, p < 0.001), and ROS (r = −0.800, p < 0.001), but was positively correlated with Rayleigh light scattering (RLS; r = 0.748, p < 0.001). RLS was also negatively correlated with EPI fluorescence (r = −0.487, p = 0.005), A₄₉₀ (r = −0.551, p = 0.001), and ROS (r = −0.539, p = 0.002). These relationships indicate that conditions associated with stronger EPI-related optical and oxidative responses generally exhibited lower Abs₂₅₀/Abs₃₆₅ spectral ratio (humification) and scattering responses, whereas increased RLS occurred together with increased humification. The correlations may reflect formulation-dependent differences in aggregation, molecular packing, particle–protein association, or the distribution of optically active species. However, RLS alone cannot determine the size, composition, or formation mechanism of the underlying colloidal structures.

Protein denaturation did not show strong correlations with the major fluorescence, absorbance, or ROS parameters. Its only statistically significant association was a moderate inverse correlation with humification (r = −0.421, p = 0.018). This limited correspondence indicates that the protein-denaturation response did not closely parallel the principal optical or oxidative changes across the pooled experimental groups. Consequently, the correlation analysis does not support assigning the observed fluorescence, ROS, or scattering responses to a single generalized protein-denaturation mechanism.

Overall, the exploratory correlation analysis identified coordinated but distinct patterns of co-variation within the BSA–EPI–PET MNP system. The strongest relationships comprised an EPI fluorescence–A₄₉₀–ROS cluster, inverse associations of BSA fluorescence with A₂₇₇ and aromaticity, and an opposing pattern between the humification/RLS responses and EPI-associated optical and oxidative signals. When considered together with the individual group comparisons, these relationships are consistent with formulation-dependent variation in the physicochemical behavior of the BSA–EPI–PET MNP system. However, because the analysis pooled groups differing simultaneously in EPI concentration, PET MNP level, and formulation composition, the correlations cannot distinguish direct molecular interactions from shared concentration effects, overlapping optical responses, or formulation-dependent changes in dispersion and colloidal organization. Accordingly, they should not be interpreted as resolving the molecular mechanism of the observed responses.

Bacterial responses to bulk-particle-depleted filtrates of PET-exposed EPI–BSA formulations

To complement the biochemical bacterial-response assays, concentration-dependent growth–inhibition profiles were evaluated for E. coli and S. aureus across the tested EPI concentrations (SI Fig. 7). These profiles were used as supportive microbiological evidence rather than as formal broth-microdilution MIC determinations. A clear species-dependent difference was observed. In E. coli, ≥90% growth inhibition was not achieved in any treatment group even at the highest EPI concentration of 100 µM. At this concentration, inhibition reached 74.7% for EPI alone and 69.7% for EPI + BSA, while the corresponding values in the presence of MNP1, MNP2, and MNP3 were 49.2%, 54.8%, and 64.3%, respectively. In contrast, S. aureus showed substantially greater inhibition at 100 µM EPI, reaching 93.5% with EPI alone, 92.2% with EPI + BSA + MNP1, and 90.8% with EPI + BSA + MNP3. EPI + BSA and EPI + BSA + MNP2 produced 89.3% and 74.1% inhibition, respectively. Thus, while the tested concentration range did not permit assignment of a conventional MIC for E. coli, the S. aureus profiles indicate that the highest EPI concentration approached or exceeded the 90% inhibition level in several formulations. The non-monotonic differences among the BSA–EPI–MNP formulations further suggest that prior PET MNP exposure modulated the bacterial response to the BSA–EPI system in a formulation- and species-dependent manner rather than uniformly enhancing or suppressing antimicrobial activity.

To place these concentration-dependent inhibition profiles in the context of the main bacterial-response experiments, bacterial growth was subsequently evaluated by OD₆₀₀ under the complete BSA–EPI–MNP exposure design (Fig. 5). The OD₆₀₀ measurements provide a comparative assessment of bacterial growth among the selected BSA, EPI, MNP, binary, and ternary formulations. Lower normalized OD₆₀₀ values therefore indicate reduced bacterial growth and correspondingly greater growth inhibition.

Fig. 5. Normalized bacterial growth responses of (I) Escherichia coli and (II) Staphylococcus aureus exposed to BSA, EPI, PET micro/nanoplastics (MNPs), and their binary and ternary formulations, based on optical-density measurements at 600 nm (OD₆₀₀). For each bacterial species, subpanels show (A) BSA and EPI groups, (B) MNP-only groups, (C) BSA + MNP groups, and the corresponding formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. Data are presented as normalized mean OD₆₀₀ ± SD from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations. Statistical analyses were performed separately for E. coli and S. aureus.

For E. coli, the BSA-normalized OD₆₀₀ value was 1.000 ± 0.135 (Fig. 5-I). Among the EPI-only groups, EPI1 and EPI2 slightly reduced normalized OD₆₀₀ to 0.791 ± 0.213 and 0.817 ± 0.037, respectively, whereas EPI3 produced the strongest decrease, reaching 0.400 ± 0.012 and differing significantly from BSA. The MNP-only groups remained close to the BSA reference, with values of 0.941 ± 0.018, 0.983 ± 0.015, and 1.014 ± 0.019 for MNP1, MNP2, and MNP3, respectively. Similarly, the BSA + MNP groups ranged from 1.060 ± 0.097 to 1.102 ± 0.157, indicating that MNP exposure alone did not substantially alter bacterial growth. In the EPI1-associated groups, BSA + EPI1 yielded a normalized OD₆₀₀ of 0.936 ± 0.112. The binary EPI1 + MNP formulations remained narrowly distributed between 0.963 ± 0.010 and 0.984 ± 0.019, while the ternary BSA + EPI1 + MNP groups ranged from 0.878 ± 0.138 to 1.013 ± 0.252. None of these formulations differed significantly from the corresponding BSA + EPI1 or binary EPI1 + MNP groups. For the EPI2-associated systems, BSA + EPI2 reduced normalized OD₆₀₀ to 0.813 ± 0.127. The binary EPI2 + MNP1, EPI2 + MNP2, and EPI2 + MNP3 groups yielded 0.873 ± 0.022, 0.891 ± 0.013, and 0.934 ± 0.007, respectively, whereas the corresponding ternary groups showed values of 0.790 ± 0.191, 0.859 ± 0.061, and 0.879 ± 0.046. None of these differences reached statistical significance relative to their corresponding BSA + EPI2 or EPI2 + MNP formulations. In contrast, the EPI3 panel showed clear formulation-dependent differences. BSA + EPI3 (0.503 ± 0.046), EPI3 + MNP1 (1.359 ± 0.094), EPI3 + MNP2 (0.628 ± 0.153), and EPI3 + MNP3 (0.561 ± 0.070) each differed significantly from BSA. Among the ternary formulations, BSA + EPI3 + MNP1 differed significantly from BSA and from the corresponding EPI3 + MNP1 group, whereas BSA + EPI3 + MNP3 differed significantly from BSA only. BSA + EPI3 + MNP2 did not show a significant predefined comparison. Importantly, none of the EPI3 ternary formulations differed significantly from BSA + EPI3 after Tukey adjustment. Thus, the strongest statistically supported reduction in E. coli growth was associated with EPI3, whereas the effect of prior PET MNP exposure within the BSA-containing systems remained formulation-specific.

For S. aureus, the BSA-normalized OD₆₀₀ value was 1.000 ± 0.173 (Fig. 5-II). EPI1, EPI2, and EPI3 yielded normalized values of 1.502 ± 0.129, 1.069 ± 0.323, and 0.607 ± 0.260, respectively. Despite these numerical differences, none of the EPI-only groups differed significantly from BSA after Tukey adjustment. Likewise, no significant differences were detected in the MNP-only or BSA + MNP panels. In the EPI1-associated systems, EPI1 + MNP1 and EPI1 + MNP2 differed significantly from BSA, whereas the corresponding BSA + EPI1 + MNP1 and BSA + EPI1 + MNP2 ternary formulations differed significantly from their matched EPI1 + MNP binary groups. No significant difference from BSA + EPI1 was detected for the ternary formulations. At EPI2, EPI2 + MNP1 differed significantly from BSA, while BSA + EPI2 + MNP3 also differed significantly from BSA; no matched b or c comparisons reached significance in this panel. The clearest ternary response occurred at EPI3. BSA + EPI3 differed significantly from BSA. In addition, BSA + EPI3 + MNP1, BSA + EPI3 + MNP2, and BSA + EPI3 + MNP3 each differed significantly from BSA and from their corresponding EPI3 + MNP groups, but not from BSA + EPI3. Thus, the largest formulation-dependent contrasts in S. aureus occurred at the higher EPI exposure, with BSA-containing ternary formulations producing substantially lower growth responses than the corresponding EPI3 + MNP binary systems.

Overall, both bacterial species showed their greatest growth reduction under EPI3, whereas effects associated with prior PET MNP exposure were species- and formulation-dependent. Differences between the two organisms may partly reflect their contrasting cell-envelope architectures. The Gram-negative envelope of E. coli contains an outer membrane that provides an additional permeability barrier,⁵⁴ whereas S. aureus lacks an outer membrane and instead possesses a thick peptidoglycan-rich wall containing negatively charged wall teichoic acids.^55–57 These structural differences may contribute to differential interactions between the bacterial surface and components present in the bulk particle-depleted filtrates, including EPI-, BSA-, and other formulation-associated species. The absence of a consistent monotonic MNP-dose response nevertheless indicates that the bacterial growth effects cannot be explained by nominal PET MNP concentration alone.

For E. coli, the BSA-normalized Bradford response was 1.000 ± 0.178 (Fig. 6-I). EPI1, EPI2, and EPI3 remained statistically comparable to BSA. In contrast, MNP1, MNP2, and MNP3 showed significantly lower Bradford responses of 0.363 ± 0.012, 0.399 ± 0.003, and 0.454 ± 0.030, respectively (p < 0.05). The BSA + MNP1, BSA + MNP2, and BSA + MNP3 groups also differed significantly from BSA. Within the EPI1-associated systems, all three EPI1 + MNP binary groups differed significantly from BSA. BSA + EPI1 + MNP1 and BSA + EPI1 + MNP2 differed significantly from their corresponding EPI1 + MNP controls, whereas BSA + EPI1 + MNP3 differed from both BSA + EPI1 and EPI1 + MNP3. At EPI2, all three EPI2 + MNP binary formulations differed significantly from BSA, and each corresponding BSA + EPI2 + MNP ternary formulation differed significantly from its matched EPI2 + MNP group. None of the EPI2 ternary groups differed significantly from BSA + EPI2. A similar pattern was observed at EPI3. All EPI3 + MNP binary groups differed significantly from BSA. BSA + EPI3 + MNP1 differed significantly from BSA + EPI3 and from EPI3 + MNP1, whereas BSA + EPI3 + MNP2 and BSA + EPI3 + MNP3 differed significantly from their corresponding EPI3 + MNP groups. Overall, the E. coli Bradford response was markedly reduced in MNP-only and EPI + MNP binary systems, whereas incorporation of BSA generally increased the signal relative to the matched binary EPI + MNP formulations.

Fig. 6. Normalized Bradford responses of (I) Escherichia coli and (II) Staphylococcus aureus exposed to BSA, EPI, PET micro/nanoplastics (MNPs), and their binary and ternary formulations. For each bacterial species, subpanels show (A) BSA and EPI groups, (B) MNP-only groups, (C) BSA + MNP groups, and the corresponding formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. Data are presented as normalized mean Bradford response ± SD from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations. Statistical analyses were performed separately for E. coli and S. aureus.

For S. aureus, the BSA-normalized Bradford response was 1.000 ± 0.283 (Fig. 6-II). EPI1 increased the response to 1.748 ± 0.137 and differed significantly from BSA, whereas EPI2 and EPI3 did not. Among the MNP-only formulations, MNP2 and MNP3 differed significantly from BSA, whereas MNP1 did not. All three BSA + MNP formulations differed significantly from BSA. Within the EPI1-associated systems, EPI1 + MNP2 differed significantly from BSA, and BSA + EPI1 + MNP2 differed significantly from its matched EPI1 + MNP2 formulation. No other predefined comparisons reached significance in this panel. At EPI2, EPI2 + MNP1 and EPI2 + MNP3 differed significantly from BSA. The corresponding BSA + EPI2 + MNP1 and BSA + EPI2 + MNP3 formulations differed significantly from their matched binary EPI2 + MNP groups, whereas no significant predefined comparison was detected for the MNP2-containing ternary formulation. At EPI3, EPI3 + MNP1, EPI3 + MNP2, and EPI3 + MNP3 all differed significantly from BSA. The corresponding BSA + EPI3 + MNP1, BSA + EPI3 + MNP2, and BSA + EPI3 + MNP3 formulations also differed significantly from BSA; however, no significant b or c comparison was detected within this panel. Thus, the S. aureus Bradford response showed strong formulation dependence, but statistically supported binary-to-ternary shifts were restricted to selected EPI1- and EPI2-containing formulations rather than occurring uniformly across all EPI concentrations.

Overall, the Bradford response of S. aureus depended strongly on EPI concentration and formulation composition, with several binary EPI + MNP-associated decreases being partially reversed upon inclusion of BSA. For S. aureus, statistically supported binary-to-ternary differences were restricted to selected EPI1- and EPI2-containing formulations, indicating a more heterogeneous BSA-associated modulation than that observed for E. coli.

Bacterial ROS responses were species- and formulation-dependent (Fig. 7). Fluorescence-based ROS measurements provide an indicator of oxidative stress, which can contribute to antimicrobial-associated cellular injury and bacterial death.^58,59 The bacterial ROS response of E. coli was formulation-dependent (Fig. 7-I). No significant predefined comparisons were detected among the EPI-only, MNP-only, or BSA + MNP groups, despite numerical differences among their normalized ROS values. Similarly, none of the binary BSA + EPI or EPI + MNP groups in the EPI1 panel differed significantly from BSA. Within the EPI1-associated systems, BSA + EPI1 + MNP1 showed the clearest response, reaching 1.527 ± 0.182 and differing significantly from both BSA + EPI1 and the corresponding EPI1 + MNP1 formulation (p < 0.05), although its comparison with BSA did not reach the Tukey-adjusted significance threshold. At EPI2, BSA + EPI2 + MNP2 and BSA + EPI2 + MNP3 were significantly different from BSA + EPI2, whereas neither differed significantly from BSA or from its matched EPI2 + MNP formulation. At EPI3, BSA + EPI3 differed significantly from BSA. BSA + EPI3 + MNP1 differed significantly from the corresponding EPI3 + MNP1 formulation but not from BSA or BSA + EPI3. In contrast, BSA + EPI3 + MNP2 and BSA + EPI3 + MNP3 differed significantly from BSA + EPI3 but not from BSA or their corresponding binary EPI3 + MNP groups. Thus, prior PET MNP exposure modified the E. coli ROS response in a formulation-specific manner, with statistically supported effects reflecting different comparison levels at EPI1, EPI2, and EPI3 rather than a uniform increase or decrease in oxidative stress.

Fig. 7. Normalized reactive oxygen species (ROS) responses of (I) Escherichia coli and (II) Staphylococcus aureus exposed to BSA, EPI, PET micro/nanoplastics (MNPs), and their binary and ternary formulations. For each bacterial species, subpanels show (A) BSA and EPI groups, (B) MNP-only groups, (C) BSA + MNP groups, and the corresponding formulations containing (D) EPI1, (E) EPI2, and (F) EPI3. Data are presented as normalized mean ROS response ± SD from three independently prepared samples (n = 3). Letters indicate predefined comparisons reaching Tukey-adjusted p < 0.05 following one-way ANOVA and Tukey's multiple-comparison post hoc test: a, versus BSA; b, versus the corresponding BSA + EPI group at the same EPI concentration; and c, versus the corresponding EPI + MNP group at the same EPI and MNP concentrations. Statistical analyses were performed separately for E. coli and S. aureus.

The S. aureus ROS response differed from that of E. coli and was generally lower than the BSA reference across many formulations (Fig. 7-II). No significant differences were detected among the EPI-only or MNP-only groups. Among the BSA + MNP formulations, only BSA + MNP3 differed significantly from BSA. The clearest statistically supported response occurred in the EPI1-associated panel. BSA + EPI1 and EPI1 + MNP1, EPI1 + MNP2, and EPI1 + MNP3 each differed significantly from BSA. BSA + EPI1 + MNP1 and BSA + EPI1 + MNP2 also differed significantly from BSA, whereas BSA + EPI1 + MNP3 did not. Importantly, none of the EPI1 ternary formulations differed significantly from BSA + EPI1 or from their corresponding EPI1 + MNP groups after Tukey adjustment. No predefined significant differences were detected in the EPI2- or EPI3-associated panels. Overall, S. aureus showed a comparatively restricted pattern of statistically supported ROS modulation, concentrated mainly in the BSA + MNP3 and EPI1-associated systems. The lack of significant matched binary-versus-ternary differences indicates that the observed ROS response did not provide evidence of a consistent PET-dependent modification of EPI–BSA-associated oxidative stress in this species.

The species-specific patterns may partly reflect differences in bacterial envelope organization. The outer membrane of E. coli provides an additional barrier between extracellular formulations and the cytoplasmic membrane,⁵⁴ whereas the surface of S. aureus is dominated by a thick peptidoglycan layer and wall teichoic acids.^55–57 The ROS results should therefore be interpreted together with growth, SOD, and lipid-peroxidation responses rather than as an isolated indicator of downstream cellular damage.

The TBARS assay was used as an established indicator of lipid-peroxidation products and oxidative damage to lipids.⁶⁰ LPO showed a response pattern distinct from that of ROS (SI Fig. 8). In E. coli, normalized LPO values generally remained close to the BSA reference, and none of the predefined comparisons reached statistical significance. The EPI3 ternary formulations, for example, yielded values of 0.895 ± 0.030, 0.936 ± 0.192, and 0.983 ± 0.174 for MNP1–MNP3, respectively. In S. aureus, a more consistent numerical separation was observed between binary EPI + MNP and BSA-containing ternary formulations. At EPI3, the binary EPI3 + MNP groups showed LPO values of 1.414–1.512, whereas the corresponding ternary systems remained lower at 0.959–1.065. Similar numerical patterns occurred at EPI1 and EPI2. Although these differences did not reach statistical significance because of the relatively large variability in the BSA-containing systems, the trend was notable when considered together with the exploratory physicochemical–bacterial correlation analysis.

Specifically, S. aureus LPO correlated positively with A₂₇₇ (r = 0.742, p < 0.001), aromaticity (r = 0.595, p < 0.001), and apparent hydrodynamic particle size (r = 0.548, p = 0.001), while correlating inversely with BSA intrinsic fluorescence (r = −0.715, p < 0.001). These associations remained among the most robust relationships after false-discovery-rate adjustment. The combined results therefore indicate that S. aureus lipid-peroxidation behavior co-varied with formulation states characterized by stronger aromatic-region responses, lower BSA fluorescence, and larger apparent hydrodynamic populations. Because these correlations were derived from pooled formulation groups, however, they should be interpreted as exploratory co-variation rather than evidence that these physicochemical changes directly caused membrane lipid oxidation.

Comparing the two species, E. coli exhibited only minor and mostly nonsignificant LPO fluctuations, whereas S. aureus showed a clearer tendency toward elevated LPO in the binary EPI + MNP systems. Differences in cell-envelope architecture may contribute to this species-dependent behavior.^54–56 Nevertheless, membrane lipid oxidation was not a dominant or uniformly significant outcome under the tested conditions and should be interpreted together with ROS, SOD, and bacterial growth.

SOD activity showed a strongly formulation-specific antioxidant response (SI Fig. 9). SOD enzymes contribute to bacterial antioxidant defense by removing superoxide radicals, although their expression and activity depend strongly on physiological and environmental conditions.⁶¹ In E. coli, the clearest effects occurred in the EPI3 systems: EPI3 + MNP3 increased SOD to 3.152 ± 0.078, whereas BSA + EPI3 + MNP1 reached 2.233 ± 0.127; the other EPI3 ternary formulations remained close to the BSA reference. In S. aureus, the most pronounced responses were selectively associated with MNP2, with BSA + EPI2 + MNP2 and BSA + EPI3 + MNP2 increasing SOD to 4.964 ± 0.443 and 3.671 ± 0.737, respectively, and differing significantly from their corresponding reference and binary groups (p < 0.05).

Taken together, SOD responses were formulation-specific rather than monotonically related to nominal PET MNP concentration. In E. coli, the clearest modulation occurred within the EPI3 formulations, whereas in S. aureus the strongest responses were selectively associated with MNP2. Thus, the antioxidant response depended on the combined EPI–BSA–MNP formulation rather than simply increasing with MNP load. These responses indicate differential engagement of bacterial antioxidant defenses under the tested exposure conditions, although the variability observed in several groups warrants cautious mechanistic interpretation.^61,62

Overall, ROS, LPO, and SOD revealed complementary aspects of bacterial oxidative stress. ROS reflected changes in the oxidative environment, SOD indicated selective activation of antioxidant defense, and LPO provided the clearest exploratory association with the physicochemical state of the BSA–EPI–PET system, particularly in S. aureus. The overall pattern therefore suggests species- and formulation-dependent oxidative adaptation rather than a uniform progression from ROS generation to membrane lipid damage.

The effects of BSA–EPI formulations and prior PET MNP exposure on the biochemical fingerprints of E. coli and S. aureus were subsequently evaluated using selected FTIR regions associated with phosphodiester/cell-envelope components and the amide I and amide II protein bands (Fig. 8 and 9). The bacterial mid-infrared spectrum contains characteristic contributions from proteins, nucleic acids, phospholipids, and polysaccharides within defined spectral regions.^63,64 Because the spectra are presented as transmittance, higher transmittance corresponds to lower relative infrared absorption. Accordingly, the observed spectral changes were interpreted as relative differences in bacterial biochemical fingerprints rather than as direct evidence of membrane rupture, protein denaturation, or cell lysis.

Fig. 8. Selected FTIR transmittance regions of Escherichia coli following exposure to BSA–EPI and BSA–EPI–PET micro/nanoplastic (MNP) formulations. The three rows correspond to increasing EPI concentrations: EPI1 (1 µM; upper row), EPI2 (10 µM; middle row), and EPI3 (100 µM; lower row). For each EPI concentration, spectra are presented for three representative FTIR regions: the phosphodiester/cell-wall-associated region (∼1082 cm⁻¹; left column), amide II/protein region (∼1540 cm⁻¹; middle column), and amide I/protein region (∼1650 cm⁻¹; right column). The black solid line represents untreated E. coli (control), while the blue dashed line represents bacteria exposed to the corresponding BSA–EPI formulation without PET MNPs. The red, green, and orange solid lines represent the corresponding BSA–EPI formulations containing PET MNP1, MNP2, and MNP3, respectively. Spectral differences among the control, BSA–EPI, and MNP-containing formulations illustrate formulation- and concentration-dependent changes in FTIR transmittance within regions associated with bacterial cell-wall/phosphodiester components and protein secondary-structure-related amide vibrations.

Fig. 9. Selected FTIR transmittance regions of Staphylococcus aureus following exposure to BSA–EPI and BSA–EPI–PET micro/nanoplastic (MNP) formulations. The three rows correspond to increasing EPI concentrations: EPI1 (1 µM; upper row), EPI2 (10 µM; middle row), and EPI3 (100 µM; lower row). For each EPI concentration, spectra are shown for three representative FTIR regions: the peptidoglycan/cell-wall-associated region (∼1082 cm⁻¹; left column), the amide II/protein region (∼1540 cm⁻¹; middle column), and the amide I/protein region (∼1650 cm⁻¹; right column). The black solid line represents untreated S. aureus (control), whereas the blue dashed line represents bacteria exposed to the corresponding BSA–EPI formulation without PET MNPs. The red, green, and orange solid lines represent the corresponding BSA–EPI formulations containing PET MNP1, MNP2, and MNP3, respectively. Differences among the control, BSA–EPI, and MNP-containing spectra indicate formulation- and concentration-dependent variations in FTIR transmittance within regions associated with bacterial cell-wall/peptidoglycan components and protein-related amide vibrations.

For E. coli, formulation-dependent differences were most apparent around 1082 cm⁻¹ and, to a lesser extent, in the amide II region around 1540 cm⁻¹ (Fig. 8). The approximately 1080–1085 cm⁻¹ region contains major contributions from phosphate-containing groups, including phosphodiester vibrations associated with nucleic-acid and cell-envelope components, whereas the amide II and amide I bands occur predominantly around 1540 and 1650 cm⁻¹, respectively.^64,65 Most BSA–EPI and BSA–EPI–MNP formulations exhibited higher transmittance than the bacterial control near 1082 cm⁻¹, indicating lower relative absorption within this spectral region. Differences among MNP levels were also evident, particularly within the EPI3 formulations. In contrast, the amide I profiles remained comparatively closely clustered, indicating smaller treatment-related changes in this protein-associated region. These observations are consistent with formulation-dependent differences in the relative biochemical composition or organization of the bacterial samples. However, FTIR spectral intensity can also be influenced by sample preparation, biomass deposition, hydration, sample thickness, and normalization procedures; consequently, the spectral differences were not interpreted as direct evidence of a specific damage mechanism.^63,64

For S. aureus, the clearest treatment-related differences were similarly observed around 1082 cm⁻¹ and in the amide II region (Fig. 9). In Gram-positive S. aureus, the 1200–800 cm⁻¹ fingerprint region includes substantial contributions from cell-wall polysaccharides and wall teichoic-acid-associated phosphate groups.⁶⁶ The BSA–EPI formulations generally showed higher transmittance than the control near 1082 cm⁻¹, corresponding to lower relative absorption within this cell-wall-associated spectral region. Addition of PET MNPs produced further formulation-dependent differences, although no uniform concentration-dependent pattern was evident across all EPI groups. Changes in the amide II region were modest, whereas the amide I spectra remained closely clustered. Thus, the data indicate changes in the relative contribution or organization of cell-wall- and protein-associated spectral components but do not independently demonstrate peptidoglycan degradation or protein unfolding.

Comparison of the two bacterial species indicates that the BSA–EPI–PET MNP formulations produced species- and formulation-dependent FTIR responses. Changes around 1082 cm⁻¹ were evident in both organisms, whereas the protein-associated regions, particularly amide I, showed comparatively limited separation among treatments. These differences are compatible with the distinct envelope architectures of Gram-negative and Gram-positive bacteria,^64–66 but FTIR measurements alone cannot determine whether the observed changes result from direct molecular interactions, altered biomass, cell-envelope perturbation, aggregation, or other treatment-related effects. The FTIR findings were therefore interpreted together with bacterial growth, oxidative-stress, and biochemical responses rather than as independent proof of an antimicrobial mechanism or synergistic cellular damage.

Exploratory cross-domain correlations between physicochemical measurements and bacterial responses

Pearson correlation analysis was used to explore co-variation between the physicochemical measurements of the BSA–EPI–PET MNP formulations and the bacterial responses of E. coli and S. aureus (SI Fig. 10). The analysis included 31 formulation-matched experimental groups and incorporated BSA and EPI fluorescence, UV-vis absorbance at 238, 277, 292, and 490 nm, aromaticity and humification indices, Rayleigh light scattering (RLS), cell-free ROS generation, protein-denaturation response, zeta potential, apparent hydrodynamic population size, and PDI. These variables were correlated with bacterial growth (OD₆₀₀), Bradford-detectable protein response, ROS, SOD, and lipid peroxidation (LPO) for each bacterial species. Because the analysis pooled formulations differing simultaneously in EPI concentration, PET MNP exposure level, and composition, the resulting correlations were interpreted only as exploratory patterns of treatment-dependent co-variation. They were not used to infer causal relationships, direct physicochemical control of bacterial responses, or a molecular mechanism.

For E. coli, the clearest correlations involved bacterial growth and the Bradford response. OD₆₀₀ was inversely correlated with A₄₉₀ (r = −0.638, p < 0.001), EPI fluorescence (r = −0.574, p < 0.001), and cell-free ROS generation (r = −0.515, p = 0.003). Thus, formulations with higher EPI-associated optical or oxidative responses tended to show lower E. coli OD₆₀₀ values across the pooled dataset. The Bradford response was inversely correlated with A₂₇₇ (r = −0.639, p < 0.001) and aromaticity (r = −0.499, p = 0.004), and positively correlated with BSA intrinsic fluorescence (r = 0.591, p < 0.001). Apparent hydrodynamic population size was also inversely correlated with the E. coli Bradford response (r = −0.484, p = 0.006). Zeta potential showed a weaker inverse correlation with OD₆₀₀ (r = −0.406, p = 0.023), whereas no strong correlations were observed between zeta potential or PDI and the oxidative-stress endpoints. Collectively, these results identify co-variation between selected optical, redox, colloidal, and bacterial-response measurements without establishing a direct functional relationship among them.

A different correlation profile was observed for S. aureus, with the strongest associations involving LPO. LPO was positively correlated with A₂₇₇ (r = 0.742, p < 0.001) and aromaticity (r = 0.595, p < 0.001), and inversely correlated with BSA intrinsic fluorescence (r = −0.715, p < 0.001). Apparent hydrodynamic population size was also positively correlated with S. aureus LPO (r = 0.548, p = 0.001), and this association remained significant after false-discovery-rate adjustment. Accordingly, formulations with higher aromatic-region responses or larger apparent hydrodynamic populations tended to show higher LPO responses across the pooled formulation set. These correlations do not demonstrate that either spectral or hydrodynamic changes directly produced the LPO response. RLS was positively correlated with the Bradford response (r = 0.551, p = 0.001), indicating additional co-variation between these two measurements. PDI showed a weaker inverse correlation with the Bradford response (r = −0.411, p = 0.022), but this relationship did not remain significant after multiple-comparison adjustment.

As observed for E. coli, S. aureus growth also showed inverse correlations with several EPI-associated physicochemical measurements. OD₆₀₀ correlated negatively with A₄₉₀ (r = −0.500, p = 0.004), cell-free ROS (r = −0.493, p = 0.005), and EPI fluorescence (r = −0.423, p = 0.018). A modest positive correlation was observed between A₄₉₀ and SOD activity (r = 0.385, p = 0.032); however, this association did not remain significant after multiple-comparison adjustment.

Overall, the exploratory analysis identified different cross-domain correlation profiles for the two bacterial species. In E. coli, several of the larger correlations linked growth or Bradford responses with EPI-associated optical/redox measurements and BSA-associated spectroscopic responses, whereas in S. aureus, several of the larger correlations involved LPO and aromatic-region or hydrodynamic measurements. Because a relatively large number of correlations were examined, nominal p-values were interpreted cautiously. Benjamini–Hochberg false-discovery-rate adjustment showed that 12 of the 20 nominally significant physicochemical–bacterial correlations remained significant at q < 0.05. Among the FDR-robust associations were A₂₇₇–LPO and BSA fluorescence–LPO in S. aureus, A₄₉₀–OD₆₀₀ and A₂₇₇–Bradford in E. coli, and apparent hydrodynamic population size–LPO in S. aureus. These FDR-robust correlations strengthen the evidence for reproducible co-variation within the present dataset but do not convert the associations into mechanistic evidence. Because EPI concentration, PET MNP exposure level, and formulation composition varied concurrently across the pooled groups, shared treatment dependence cannot be separated from direct cross-domain coupling by this analysis. The correlation results should therefore be regarded as hypothesis-generating associations that identify relationships for future mechanistic testing rather than as evidence of causal physicochemical–bacterial pathways.

Limitation statement

Several limitations should be considered when interpreting the present findings. First, although the PET MNPs were previously characterized in terms of particle size, morphology, surface charge, chemical identity, surface area, and elemental composition, a PDI value was not available from the original DLS characterization of the pristine PET material. Targeted chemical analyses of potential bottle-derived pigments, additives, or other leachable constituents were also not performed. The heterogeneous PDI values observed in several experimental formulations further indicate that the DLS-derived diameters should be interpreted as apparent hydrodynamic populations rather than as the dimensions of discrete BSA–EPI–PET complexes. Moreover, nominal 0.2 µm filtration reduced bulk PET particles and larger particle-associated assemblies but did not demonstrate that the resulting filtrates were particle-free. Sub-200 nm hydrodynamic populations were detectable by DLS in post-filtration samples, although their chemical composition could not be identified by this method. Because polymer-specific quantification of filter-passing PET species and targeted analysis of PET-derived leachables were not performed, possible contributions from PET fragments, PET-containing nanoscale colloids, and leachable constituents to the post-filtration physicochemical and bacterial responses cannot be excluded.

Second, the fluorescence-response analysis included only three EPI concentrations and did not include independent quantification of free EPI. The fitted parameters may therefore be influenced by multiple, non-exclusive processes, including binding-associated quenching, inner-filter effects, adsorption, scattering, aggregation, and changes in the local fluorophore environment. Accordingly, the derived values should be regarded as descriptive apparent fluorescence-response parameters rather than thermodynamic binding constants. This limitation is particularly relevant at the highest PET concentration, where the fluorescence response was non-monotonic and could not be reliably described by the same saturable response model. Although complementary fluorescence, UV-vis, DLS, zeta-potential, Rayleigh-scattering, oxidative, and protein-stability measurements were included, none of these methods independently resolved the molecular distribution of EPI among free, albumin-associated, and PET-associated fractions. Consequently, the combined measurements support formulation-dependent physicochemical differences but do not establish competitive displacement, binding stoichiometry, thermodynamic redistribution, or a unique molecular mechanism. The UV-vis-based apparent EPI-equivalent concentrations provide a comparative measure of the detectable EPI response but do not represent direct quantification of free EPI. Neither the UV-vis analysis nor the complementary fluorescence and colloidal measurements physically separated aqueous, BSA-associated, and PET-associated EPI fractions. In addition, matrix-dependent changes in EPI absorptivity, adsorption, scattering, and colloidal organization may influence the calculated values. Consequently, these measurements cannot establish EPI redistribution, competitive displacement, or a specific molecular mechanism. Fraction-specific separation followed by chromatographic quantification would be required to resolve the distribution of EPI.

Third, the concentration-dependent bacterial inhibition profiles were not standardized MIC or MBC determinations. CFU-based killing assays and live/dead viability measurements were not performed. The bacterial results should therefore be interpreted as comparative growth- and stress-response outcomes under the tested conditions rather than as evidence of bactericidal activity, altered antimicrobial efficacy, or changes in clinically relevant susceptibility. In addition, because the bacteria were exposed to the 0.2 µm filtrates rather than to the original particle-containing formulations, the observed responses cannot be attributed exclusively to altered soluble EPI–BSA species or interpreted as direct effects of the bulk PET MNPs.

Fourth, the final BSA concentration of 2.5 mg mL⁻¹ was lower than the physiological concentration of human serum albumin in plasma, and BSA does not fully reproduce all structural and ligand-binding characteristics of human serum albumin. The PET MNP concentrations were also selected as controlled high-exposure in vitro conditions intended to provide analytically resolvable contrasts and should not be considered representative of typical environmental or systemic exposure levels. Accordingly, the present system should be regarded as a controlled in vitro physicochemical and microbiological model rather than as a quantitative reproduction of whole-plasma conditions. The findings should not be interpreted as direct evidence of altered clinical pharmacokinetics, free-drug availability, therapeutic efficacy, or patient-level risk.

DLS, PDI, zeta-potential, and fluorescence measurements provide complementary information on ensemble-level colloidal and optical behavior but do not directly resolve molecular binding or fraction-specific EPI distribution. In particular, DLS cannot chemically identify the detected hydrodynamic populations, and fluorescence responses may be influenced by binding-associated quenching, inner-filter effects, adsorption, scattering, aggregation, and changes in the local fluorophore environment. Consequently, these measurements support formulation-dependent physicochemical perturbations but do not establish binding stoichiometry, thermodynamic redistribution, or a unique molecular mechanism. Direct separation and chromatographic quantification of free, BSA-associated, and PET-associated EPI fractions, together with orthogonal binding measurements, would be required to address these questions.

Finally, the experiments included three independently prepared replicates per condition. Although replicate-level data were used for statistical testing, the limited number of independent replicates restricts the precision of parameter estimates and the assessment of response variability. In addition, the exploratory correlation analyses pooled formulations that differed simultaneously in EPI concentration, PET MNP exposure level, and composition. These correlations therefore represent treatment-dependent patterns of co-variation and cannot establish causal relationships between individual physicochemical characteristics and bacterial responses.

Conclusion

This study showed that prior PET MNP exposure was associated with formulation-dependent differences in the fluorescence, UV-vis, colloidal, redox, and protein-stability responses measured in 0.2 µm filtrates of EPI–BSA formulations. Empirical modeling identified differences in the concentration-dependent BSA fluorescence response among the PET-free, MNP1, and MNP2 formulations, whereas the non-monotonic MNP3 response could not be reliably represented by the same model and was not assigned a numerical response parameter. UV-vis-based estimation further showed a graded reduction in the apparent EPI-equivalent concentration with increasing PET exposure in formulations prepared without BSA, whereas the corresponding response remained comparatively stable across PET levels in BSA-containing formulations. These results indicate formulation-dependent changes in the detectable EPI optical response but do not demonstrate molecular redistribution of EPI.

The 0.2 µm filtrates also produced species- and formulation-dependent differences in bacterial growth and stress-related endpoints. Collectively, the findings identify associations among prior PET MNP exposure, the measured physicochemical fingerprint of the EPI–BSA formulations, and the bacterial responses to their filtrates within a controlled in vitro model. However, because only three EPI concentrations were investigated and free EPI was not independently quantified, the present measurements do not establish thermodynamic binding affinity, binding stoichiometry, or competitive displacement and do not resolve the distribution of EPI among aqueous, albumin-associated, and PET-associated fractions. Future studies combining fraction-specific separation and chromatographic EPI quantification with orthogonal binding measurements will be required to determine whether PET exposure alters the molecular distribution or binding behavior of EPI.

Author contributions

A. B. and H. S. contributed to conceptualization, performed data curation, formal analysis, and wrote the original draft. S. K. contributed to investigation. All authors reviewed and approved the final manuscript.

Conflicts of interest

The authors declare no competing interests.

Data availability

The data supporting the findings of this study are available within the article and are available from the corresponding author upon reasonable request.

Supplementary information (SI): Fig. S1–S10, including: a schematic overview of the experimental workflow; observed-versus-fitted fluorescence-response plots and residual diagnostics for the empirical BSA fluorescence model; additional UV-vis absorbance data at 238 and 292 nm; absorbance-derived aromaticity and humification indices; Rayleigh light-scattering data; exploratory Pearson correlation heatmaps for physicochemical variables; concentration-dependent bacterial growth-inhibition profiles; normalized lipid peroxidation and superoxide dismutase responses in Escherichia coli and Staphylococcus aureus; and exploratory correlation analyses linking formulation physicochemical characteristics with bacterial-response endpoints. See DOI: https://doi.org/10.1039/d6ra07666a.