Polymeric labeled NPs were synthesized at the Joint Research Centre (JRC), Ispra, Italy, as illustrated in a previous paper (32).

Briefly, polypropylene and polyvinyl chloride nanoplastics (PP and PVC NPs) were prepared by the oil-in-water emulsion technique by dissolving 30 mg of plastic polymers (Sigma-Aldrich) in 3 g of toluene in a 50-ml flask, mixed at 100°C for 1 h to dissolve the pellets. QdotTM 655 ITKTM (ThermoFisher) organic quantum dots (200 μl) were inserted in the toluene solution. Afterwards, a boiling ultra-pure MilliQ water solution (27 ml) containing 7.5 mg of the dissolved surfactant sodium cholate (Sigma-Aldrich) was added to the hydrophobic phase and the two separating phases were homogenized using an Ultra-Turrax T25 (IKA) at 16,000 rpm for 2 min, ultrasonicated (Vibra-Cell Ultrasonic Liquid Processors, vCX 130) at 40% amplitude for 2.5 min, and cooled in an ice-water bath for 3 min. Five-micrometer polyether-sulfone membrane syringe filters filtered the reaction product to obtain the desired NPs size, while the organic phase was successfully evaporated using the rotavapor. The synthetic yield was calculated by weighing the freeze-dried powder. QDs NPs were dispersed in MilliQ and stored at 4°C under dark conditions (32).

Human monocytic THP-1 cells (ATCC TIB-202) were maintained in a sub-confluent state in complete culture medium, RPMI 1640 (Life Technologies, MI, Italy) supplemented with 10% (v/v) fetal bovine serum (Life Technologies, MI, Italy) and 1% (v/v) penicillin/streptomycin (Life Technologies, MI, Italy) under standard cell culture conditions (37°C, 5% CO₂, and 95% humidity). The cells were cultured in a 75-cm² Falcon flask for 2–3 days and then seeded onto silicon carbide (SiC) with 200-nm-thick membranes (Silson Ltd, Warwickshire, UK) in 24 multi-well plates. For THP-1 polarization, cells were seeded at 1 × 10⁵ cells/ml in the complete culture medium (CM) with phorbol-12-myristate 13-acetate (PMA) (Thermo Fisher Scientific, MA, USA) to a final concentration of 15 ng/ml. They were incubated at 37°C, 5% CO₂ for 48 h, to be activated into resting macrophages (RM), and washed with phosphate-buffered saline, Dulbecco’s formula (PBS) supplemented with Ca⁺⁺ and Mg⁺⁺ (Life Technologies), and the CM was replaced with a PMA-free CM to control the macrophages’ differentiation. RM was polarized into M1 macrophages by the addition of the following stimuli: IFN-γ (500 U/ml) and LPS (100 ng/ml) (PeproTech, London, UK) for 24 h. The different conditioned cells were incubated with CM containing the two different NPs labeled with CdSe-QDs at a final concentration of 20 μg/ml for 72 h. Control M1 cells were grown under the same conditions but were not exposed to NPs. The samples were then fixed at room temperature with a 4% aqueous paraformaldehyde solution for 20 min in the dark and washed with PBS and Milli‐Q water before the analyses.

Cells were seeded in a 96-well Falcon plate and treated with different concentrations of labeled and non-labeled PP and PVC NPs (0, 1, 5, 20, and 50 μg/ml), and Staurosporine 2 μM (Sigma Aldrich, Milan, Italy) as a positive control. After 24 h of incubation, they were activated and polarized into the M1 phenotype and then treated with the testing concentrations of NPs for 72 h. MTT reagent (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) (Sigma-Aldrich, MI, Italy) (10 μl) was added to each conditioned well in triplicate, incubated for 4 h at 37°C, 5% CO₂. The formazan crystals were visible through the light microscope; thus, they were dissolved in dimethyl sulfoxide (DMSO) (Euroclone, MI, Italy), and the corresponding absorbance was read using an ELISA Microplate Reader (Bio-Rad, Hercules, CA, USA) at 570 nm wavelength. The data were collected in triplicate for each condition, the negative control of the reagent used (MTT) was subtracted and all the values were normalized for the control cells (100%) using GraphPad software.

THP-1 cells (10⁴ cells/ml) were seeded in a 96-well plate and treated with different concentrations of CdSe-QDs PP and PVC NPs (0, 5, 10, 20, 50, and 100 μg/ml). After 24 h of incubation, monocytes were activated and polarized into the M1 phenotype and then treated with the testing concentrations of NPs for 24 and 72 h. At this point, cells were washed twice with PBS, and 100 μl of PFA, 4% in PBS, was added to each well and left for 20 min in the dark. Then, sterile water was gently rinsed into the wells. Isopropanol (100 μl) was added to each well and left to sit for 2–5 min. Oil Red O (Acros Organics, Geel, Belgium) working solution (100 μl) was added to the wells, and the multi-well was slowly rotated to favor the spread of the solution along the cells. After 5 min, the wells were washed with sterile water until the water ran clear. Isopropanol (200 μl) was added to each well, whose 150 μl was transferred to a new well plate for quantification. The corresponding absorbance was read using an ELISA Microplate Reader (Bio-Rad, Hercules, CA, USA) at 450 nm wavelength. The t-test calculated the significance of the data using GraphPad software.

For each experimental condition, two to three cells among the samples grown on the same or different SiC windows were selected by performing light fluorescence microscopy, to be afterwards analyzed by X-ray microscopy at Elettra Sincrotrone Trieste. SiC windows were mounted on the sample holder for fluorescence microscopy (BioTek Cytation 5 Cell Imaging Multimode Reader, Agilent, CA, USA), operating both in bright field mode and with Texas Red ^® filter cube (excitation wavelength 560 ± 55 nm and emission wavelength 645 ± 75 nm). Images were acquired at 10× and 20× magnification objectives for each experimental condition and were carried out using FiJi software (35).

STXM and LEXRF were performed at the TwinMic beamline (33, 36) at Elettra Sincrotrone Trieste (Trieste, Italy). The TwinMic microscope was operated in the scanning transmission mode (STXM), and the beam was focused on the sample through a zone plate (600 μm diameter and 50 nm outermost zone width), delivering a micrometric or sub-micrometric probe size. Samples were raster-scanned perpendicularly to the incoming monochromatic beam. At the same time, a fast readout charge-coupled device (CCD) camera collected the transmitted X-rays (37), and an eight-silicon drift detector-based XRF system acquired the emitted fluorescence photons (36, 38). The obtained absorption and phase contrast images outlined the morphological sample features at sub-micrometer length scales, while the simultaneous detection of LEXRF correlated the elemental distribution to the morphology. Elemental distribution has been obtained with PyMCA software (39) by deconvolving and fitting the XRF spectra. For high-resolution images operating in the STXM mode, a photon energy of 1.2 keV was used, with a spot size of approximately 200 nm and a dwell time of 60 ms. Parameters were then changed to acquire XRF maps of a smaller area, using a photon energy of 1.7 keV to excite and obtain optimal emission conditions for the element of primary interest, Se, with a spot size of 490 nm and a dwell time of 10 s per pixel for XRF mapping, and a CCD dwell time of 60 ms for STXM imaging. Each map lasted approximately 8−12 h, depending on the dimensions of the scanned area. We mapped approximately two to three cells per SiC window.

FTIR micro-spectroscopy measurements were performed at the Chemical and Life Science branch of the SISSI-Bio beamline, Elettra Sincrotrone Trieste (Trieste, Italy) (40), using a Hyperion 3000 Vis-IR microscope (15× condenser/objective) and an MCT detector coupled with a Vertex 70v interferometer (Bruker Optics GmbH, Ettlingen, Germany). FTIR measurements were gathered in transmission mode in the MIR region (4,000–800 cm⁻¹ at 4 cm⁻¹ spectral resolution). For each condition, 40 to 300 single spectra (averaging 512 scans per spectrum and 1 min per spectrum) were acquired in both cytoplasmatic and nuclear regions. Background spectra were collected for each measure in a cell-free area with the same parameters. In addition, cellular imaging was performed using a 64 × 64 pixel focal plane array (FPA) detector, collecting two to three maps per condition averaging 256 scans (20 min per image) with 4 cm⁻¹ spectral resolution and a pixel size of 2.6 microns (15× condenser/objective). Data analyses were conducted employing OPUS software (Bruker Optics, Billerica MA, US) for water vapor correction and then Quasar (http:\\quasar.codes) for spectra processing: baseline correction, cut, normalization when required, calculation of the band integrals and ratios, and principal component analysis (PCA). To screen the pixels belonging to empty areas that could affect the results, an intensity filter on the peak at 1,650 cm⁻¹ was applied. Then, statistical analysis was carried out with OriginPro 2023 (Originlabs). For univariate analysis, the following band integrals were calculated: 2,800–3,000 cm⁻¹ for lipids, peak height at 2,925 cm⁻¹ for the CH₂ asymmetric stretching, and at 2,960 cm⁻¹ for the CH₃ asymmetric stretching, 1,710–1,765 cm⁻¹ for carbonyl group (CO), 1,716 cm⁻¹ for free fatty acids (FFAs), 1,740 cm⁻¹ for triacylglycerols (TAGs), and 1,700–1,480 cm⁻¹ for proteins. Multivariate analysis was accomplished using Quasar. The second-order derivative of absorbance spectra was calculated by the Savitzky–Golay algorithm using 21 points of smoothing, a third-degree fitting polynomial function. Then, the data were normalized using vector normalization, and PCA was performed using 10 components. The results are shown as a scatterplot of PC1 vs. PC2 vs. PC3 and line plots of the respective loadings.

Mean and standard deviations were calculated for continuous variables, whereas percentages were reported for categorical variables. Sample data were analyzed by Student’s t-test. Analysis of different groups of data was performed using two-way analysis of variance (ANOVA). Results were expressed as mean ± standard deviations. p-values <0.05 were considered statistically significant. All statistical analyses were performed using GraphPad Prism software 10 (GraphPad Software Inc., La Jolla, CA, USA).

Light fluorescence microscopy analysis, able to detect the presence of NPs inside the cells, was preliminarily carried out to select the best cell samples for synchrotron-based studies. The light fluorescence images in Figures 3A, F , respectively, show the specific bright red emitting property of CdSe-QDs-PVC and -PP drops (1 mg/ml) deposited on 200-nm-thick SiC windows, acquired with the Texas Red filter cube (excitation 560 ± 55 nm and emission 645 ± 75 nm). Thus, this particular sized CdSe-QDs was confirmed to be an optimal labeling strategy to track NPs in cell samples (32). Despite a slight artifactual autofluorescence (red) of macrophages incubated with CdSe-QDs-PVC or -PP, visible in Figures 3B, G , respectively, the QDs-labeled NPs can be identified by the red spots at the highest brightness as depicted in Figures 3C–E, H–J . The light fluorescence images were extraordinarily supportive and used as a preliminary guide in the selection of two to three cells per sample to be analyzed at higher spatial resolution by the synchrotron-based technique, specifically LEXRF, to confirm the presence of NPs by following the emission L line of Se, one of the two components linked to NPs.

To evaluate the NPs’ impacts on cellular morphology and to simultaneously track the NPs across the macrophages by following the Se L line of the label of NPs at sub-cellular spatial resolution, STXM and low-energy-X-ray fluorescence (LEXRF) were performed. The monocyte-polarized macrophage topographic information before and after the 72-h incubation of cells with NPs has been provided by STXM absorption and phase contrast images visible in Figures 4 , 5 , respectively, for CdSe-QDs-PP and -PVC NPs. In Figures 4A , 5A , the morphology of unexposed cells is shown, with a more absorbent nuclear region. Instead, when cells were exposed to NPs, especially PP, the formation of bright vesicles was often observed in the absorption and phase contrast images, like those reported in Figures 4C, D . Based on already published studies (41), the bright vesicles evidenced by STXM suggest LDs that frequently localize perinuclearly and at the plasma membrane. Contrary to fluorescence images (Figure 3), the presence of NPs aggregates at the cell surface cannot be directly inferred from absorption and phase contrast images.

More importantly, LEXRF analyses performed at 1.7 keV proved a suitable tool for tracing a component of NPs, selenium. This component is shown in the third column of Figures 4 , 5 . The chemical distribution map of low-Z elements (C, O, Na, and Mg) was acquired and is displayed in the same figure. In this way, the cellular integrity and the specific localization of the NPs in the cells can be monitored.

The cells reported in Figures 4B–D , 5B–D are the sameshown as the fluorescence image insets reported in Figure 3 , whose red spot (NPs) localization is in good agreement with the selenium (QDs) distribution across the cell. Moreover, a possible internalization of NPs at submicron spatial resolution can be observed. Although the STXM coupled with LEXRF is not a 3D microscopy, the peculiar distribution of selenium in nanometric spots across the cells is suggestive of a cellular cytoplasmatic accumulation, with no evidence of interaction with the LDs.

The same measurement conditions were used as a negative control to acquire unexposed cell images which revealed no Se-related peak. These results do confirm the specificity of the Se signal for NPs presence. The hotspot appearing in Figure 4A , Se XRF map, corresponds to the Al contamination on the sample, as the Se L-line falls on the shoulder of the Al K-line.

To examine the cellular modifications at the macromolecular level induced by NPs exposure, FTIR microscopy and imaging was used.

In detail, FTIR microscopy characterized macromolecular changes in the cell samples exposed to NPs. Univariate analysis was accomplished to evaluate the spectral profiles of the different populations of cells and the values of specific band ratios. In Figure 6A , the normalized average spectra of the three groups of treated (PP_M1 and PVC_M1) and non-treated cells (Ctrl_M1) are shown (solid lines), compared with the spectra acquired on the labeled polymeric NPcs drops (dashed lines), prepared separately.

Specific signals from the NPs and the cells are highlighted in Figure 6 as vertical dashed line and listed in Table 1 for reference. Based on that, in our analyses, it was not possible to identify the NPs characteristic bands in the treated cells’ average spectra by looking at the solid orange line spectra for PP_M1, and the solid blue line spectra for PVC_M1 concerning the Ctrl_M1 in black, since the NPs size goes below the resolution limit for FTIR microscopy and their concentration within the samples is low. Thus, one must disentangle the FTIR vibrational changes of the polymeric material from those due to the cellular characteristic peaks of lipids (3,050–2,800 cm⁻¹), proteins (1,700–1,480 cm⁻¹), and nucleic acids (1,300–900 cm⁻¹).

In Figure 6B , a zoom in the 3,050–2,800 cm⁻¹ range, characteristic of the stretching C-H vibrations of aliphatic chains, is reported with the spectra of the NPs (dashed lines) overlapping those of the cells (solid lines). The spectra of the cells are characterized by a different absorbance: PP_M1 cells indicate a higher intensity band in this spectral region, but no CH₃ signals from the plastic can be directly seen ( Table 1 ). PVC_M1 cells, instead, reveal signals with a lower intensity than the Ctrl_M1 and the PP_M1 cells.

Figure 6C presents the 1,800–1,700 cm⁻¹ spectral range, where it is possible to identify the signal from C=O groups, either from the polar heads of lipids or from eventual oxidation by-products: (I) 1,737 cm⁻¹ comes from C=O of esters of TAGs (47), while (II) the 1,716 cm⁻¹ signal refers to the C=O of the FFAs and/or aldehydes, and ketones (54). From Figure 6C , it is possible to see that both NPs have no distinctive signals in this spectral region. In PP_M1 spectra, there is an increase of the ~1,737 cm⁻¹ component of the C=O band concerning the control cells, whereas in PVC_M1, there is an increase of the ~1,716 cm⁻¹ band. Any imbalance between the carbonyl signal at ~1,737 cm⁻¹ and those from the aliphatic chains (C-H_x stretching) at 3,050–2,800 cm⁻¹ could be caused by the presence of oxidative processes induced by the NPs exposure (55).

As previously described in Figures 6A, B , PVC_M1 cells are characterized by a lower intensity of the lipidic bands (solid blue line at 3,050–2,800 cm⁻¹) concerning the Ctrl_M1 (solid black line), confirmed in Figure 6D where the amides-to-lipids ratio is shown. The values of the ratio for PVC_M1 are significantly higher than that of Ctrl_M1. The present evidence can be better explained by observing Figure 7 : from both optical image and FTIR maps, it is possible to notice that PVC_M1 cells look rounder and slightly thicker in the nuclear area. Instead, the PP_M1 condition reveals a slight yet significant ( Supplementary Table 1 ) decrease in the amide/lipid ratio regarding Ctrl_M1. From Figure 7 , Ctrl_M1 are more adherent and spread onto the SiC windows (our substrate); thus, the decrease of the amide/lipids ratio can be associated with an enhanced lipid synthesis (56). In Figure 6E , the boxplot of the carbonyl/lipids ratio demonstrates that PP_M1 are characterized by a slightly higher value of the CO/lipids ratio (check Supplementary Table 1 for statistical significance of the data) that might be linked to oxidative stress conditions (55, 57, 58). PVC_M1, instead, revealed a lower ratio than that of Ctrl_M1, explained by the lower content of triacylglycerols (TAGs, band height at ~1,737 cm⁻¹) for FFAs (band height at ~1,716 cm⁻¹). Indeed, in Figure 6F , the boxplot of the FFAs/TAGs ratio for the three conditions has the opposite trend of that in Figure 6E : the lower value of the mean of PP_M1 (check Supplementary Table 1 for statistical significance of the data) could be due to a higher formation of TAGs, as confirmed by the Oil Red O staining, whereas PVC_M1 contain more FFAs (check Supplementary Table 1 for statistical significance of the data). The commented results can be more readily seen in Figure 7 , where the chemical maps of FFAs and TAGs are displayed for several groups of cells for the three conditions. In Ctrl_M1 and PVC_M1, the two types of fatty acids were comparably distributed and only a few differences can be seen; in contrast, in PP_M1, hotspots of FFAs tend to cluster at the cytoplasmic level, whereas the highest concentration of TAGs tends to accumulate in the cellular outer edge (white arrows in panels in Figure 7 ), as previously observed in the absorption and phase contrast images ( Figure 4D ). These considerations are very evident in Figure 7 , where the related RGB composite is reported, which combines the three signals mentioned above using the same range of the intensities (see caption) of the three maps of FFAs, TAGs, and amides, normalized from 0 to 1, to favor a direct comparison.

A PCA was performed on the second derivative of the absorbance spectra to evaluate subtler differences induced by the treatments, not only due to variations in the content of specific chemical compounds but also due to conformational or structural modifications. Figure 8A presents the 3D scatterplot of the results of the PCA, with the scores plotted along PC1, PC2, and PC3. The Ctrl_M1 are at the center of the graph for all the axes, while PP_M1 leans towards negative values for PC1 and PVC_M1 leans toward positive ones, respectively. The data are separated mainly on PC1, as seen in Figure 8B ; the graph was obtained by projecting all the points on PC1, allowing for a comparison of the mean values. The outcome revealed that the three distributions are significantly different along PC1, representing the 46.5% of the total variance of the dataset. Thus, the PCA underlines a different response of macrophages according to the composition of NPs. Figure 8C illustrates the loading vectors: PC1 in red, PC2 in yellow, and PC3 in blue are presented. The PC1 loading vector contains positive signals from lipids at 2,920 and 2,850 cm⁻¹ in the CH stretching region and at 1,737 cm⁻¹ of the C=O, meaning that they are higher for PP_M1, average for Ctrl_M1, and lower for PVC_M1. PC1 showed negative protein-related signals at 1,647 cm⁻¹ for amide I and at 1,545 cm⁻¹ for amide II, meaning that their trend is the opposite of those of the lipids previously described. These protein signals can be assigned to unordered structures (59). Even if of minor amplitude, the three groups also separate along PC2 and PC3, representing 12.7% and 9.3% of the total variance of the dataset, respectively. PC2 mainly separates PP_M1 from the other two groups, and the loading vector has stronger signals in the protein region at ~1,666 cm⁻¹, ~1,568 cm⁻¹, and the carbonyl signal at ~1,743 cm⁻¹.

PC3 instead have the PVC_M1 in the negative hemispace and the PP_M1 in the positive hemispace. The PC3 loading vector contains positive signals from lipids at 2,927 and 2,857 cm⁻¹ in the CH stretching region, blue-shifted concerning those of PC1, possibly attributed to less ordered lipids and more fluid membranes (46). The signals in the C=O region are negative instead and are located at ~1,737 cm⁻¹ and ~1,716 cm⁻¹ of the C=O. PC3 contains protein signals, as well, at 1,632 cm⁻¹ for amide I and at 1,490 cm⁻¹ for amide II, negative, which can be assigned to β-sheet structures (50, 59).