**3. Results and Discussion**

The proposed methodology makes use of a high-resolution gel (Sephacryl S500) designed to resolve large macromolecules between 20 million and 400,000 daltons (g/mol). The upper protein size range is in the order of circa 750 kDa containing many thousands of amino acids, thus the gel matrix should be able to resolve plastic NPs larger from the protein/lipid pool in cells. Since SEC separates particles based on the Stokes radius rather than weight, the following discussion will be provided in size rather than mass. It is estimated that the protein size range is between 3 and 700 kDa where large 500 kDa globular proteins theoretically occupy a radius of 5.2 nm, which is well within the separation range of the gel matrix [14] and sets the theoretical lower size range of the column. Hence, the gel matrix offers the possibility to separate large plastic nanoparticles (10–1000 nm) from the protein pool in the sample without the need to extract/purify nanoparticles before running the column (Table 1). Large filamentous proteins (e.g., actinomyosin) should precipitate during centrifugation with appropriate speeds depending of the tissues. In the present study with molluscs, homogenisation with blender or Teflon pestle followed by 3000× *g* centrifugation appeared satisfactory. To validate these assumptions, the reported diameter of albumin is about 4 nm (mass of 66 kDa) [15] and should separate well from 50 or 100 nm NPs. A 40 cm × 1 cm column was filled with Sephacryl S-500 gel for separation of pure fluorescently labeled polystyrene nanoplastics (50 and 100 nm mean diameter), albumin, and NaCl (Figure 1). The total volume (Vt = 25 mL) was calculated with a conductivity (NaCl) peak, which permeates completely the gel beads, and the theoretical void volume (Vo) of the column was at Vo/Vt = 0.3 according to the supplier's information for Sephacryl S500 (Table 1). The 100 and 50 nm NPs were resolved at Ve/Vt = 0.4 and 0.45 respectively, albumin eluted at Ve/Vt of 0.75, and NaCl at Ve/Vt = 1 using the 0.2% Tween-20–14 mM NaCl elution buffer. The addition of the non-denaturing detergent (Tween-20) was necessary to maintain separation of albumin from nonspecific hydrophobic interactions of polystyrene NPs. Indeed, NPs adsorb proteins forming a corona [16] and could change the elution volume (Ve) for polystyrene NPs. This was observed by lowering the detergent concentration to 0.01%, which increased the Ve/Vo to 0.60 for the fluorescent NPs in the presence of 1 mg/mL albumin (results not shown). The addition of denaturing detergent such as SDS (0.1%) denatured albumin and eluted at the void volume Vo/Vt = 0.30 and overlapped upper size range of plastic nanoparticles (>100 nm). A significant and linear relationship between the size (log size in nm) and the Ve/Vt of the samples was obtained with r = 0.99 (Figure 1B). This relationship sets the upper and lower size range of NPs between 161–11 nm respectively.

We tested this chromatographic method with transparent polystyrene NPs (50 nm diameter) alone and in *Mya arenaria* soft tissues homogenates (3000× *g* supernatant) using NR staining for plastic particles (Figure 2). The solvatochromic NR dye is well-recognized as a fluorescent dye for plastics but also stains lipids [12,17]. Transparent polystyrene NPs eluted at Ve/Vt = 0.45 based on NR fluorescence (Figure 2A). The addition of soft tissues fractions revealed two peaks at 280 nm (Figure 2B): one at Ve/Vo corresponding to polystyrene NPs and the other at Ve/Vo = 0.9–0.95 corresponding to the protein/lipids pool. NR staining revealed also two major peaks overlapping with 280 nm absorbance corresponding to polystyrene NPs at Ve/Vo = 0.44 and the protein/lipid pool at Ve/Vo = 0.9. The elution profile of tissue fractions alone did not show any peaks at Ve/Vo < 0.7 in keeping with the resolution range of the gel matrix. The NR dye is a well-recognized fluorescent stain for many types of plastics such as polyethylene, polystyrene, nylon, polyethylene terephthalate, and polypropylene at the micro- and nanoscales, thus could be used with these plastic polymers [12,18]. The average length of phospholipids is about 2 nm [19] giving a Ve/Vo = 0.8–0.9 overlapping to the second NR peak, hence not suitable for measuring small plastic NPs (<10 nm). The presence of the detergent Tween-20 in the elution buffer also serves to solubilize membranes vesicles in the extracts removing larger lipid vesicles/structures in the sample.



<sup>1</sup> Chromatography column: 40 cm × 1 cm (31.4 mL total volume), sample buffer 0.70 M NaCl-0.05% Tween-20 and elution buffer 14 mM NaCl-0.2% Tween-20. Ve: elution volume, Vt: total volume (NaCl). The void volume Vo was Vo/Vt = 0.3 according to the suppliers' information for Sephacryl S500.

**Figure 1.** *Cont*.

**Figure 1.** Gel chromatographic analysis of fluorescently labeled plastic NPs. Elution profile of polystyrene NPs (**A**) using 100 nm, 50 nm polystyrene NPs (fluorescence), albumin (absorbance 280 nm, MW = 60 kDa), and NaCl (conductivity: mS); calibration of the size exclusion chromatography column (**B**).

This chromatographic procedure was used to screen for plastic NPs in two real-life case studies, one involving wild *Mya arenaria* clams collected at a harbor/marina and the second with caged *Elliptio complanata* mussels at downstream sites of a major urban area, both generally associated to plastics pollution [20]. The first case consisted of feral *Mya arenaria* clams collected at a reference site under no direct source of pollution and a harbour site supporting intense commercial and touristic boating activities (Figure 3A,B). Clam tissues collected at the reference site showed one major UV absorbance (280 nm) peak at Ve/Vt = 0.8 corresponding to the protein pool, one NR peak at Ve/Vt = 0.8–0.9 corresponding to hydrophobic proteins and lipids, and a conductivity peak at Ve/Vt = 1 corresponding to salts/NaCl (Figure 3A). Clams collected at the contaminated harbor site showed two major UV and NR peaks at Ve/Vo = 0.4 and 0.8 (Figure 3B) corresponding to large-size compounds consistent with UV-absorbing plastics (polystyrene and polyethylene terephthalate) followed by the protein/lipid pool peak. Large hydrophilic carbohydratebased polymers (sugars, glycogen), if present, do not absorb at 280 nm and should not interact with the NR dye. The second case concerned exposure of caged mussels exposed to pristine lake (reference) and downstream of a highly populated city of circa 3 million inhabitants in the Saint-Lawrence river (Figure 3C,D). Sediments in the Saint-Lawrence river were recently reported contaminated by microplastics [20] and prompted investigation on plastic NPs in this study. The major form of plastic was polyethylene (UV absorbance negative and NR positive) and reached densities of 1.4 × 105 microbeads·m−2. *Elliptio complanata* digestive gland homogenates caged at the reference lake revealed one major band at Ve/Vt = 0.75 and one small band at the void volume (Figure 3C). One conductivity peak was observed at the Ve/Vt = 1 corresponding to dissolved salts with no evidence of conducting large-size molecules (i.e., absence of peaks at Ve/Vt < 0.8). The NR fluorescence

peak generally followed the UV peak with a major band at Ve/Vt = 0.8 of the protein/lipid pool. In mussels caged downstream of a large city, the UV signal was distributed over the elution profile (Figure 3D) with a maxima at Ve/Vt = 0.75. The same conductivity peak at Ve/Vo = 1 was obtained, suggesting no presence of conductive large-molecular-size compounds (e.g., elemental nanoparticles). In the case of NR staining, we observed two major peaks at Ve/Vt = 0.6–0.65 and Ve/Vt = 0.35–0.5, corresponding to compounds of sizes ranging from 11 to 120 nm. This suggests that exposure of mussels to urban activities leads to a complex distribution pattern of NR-stained materials outside the protein and lipids pools in tissues. This is in keeping with the reported levels of microplastics in sediments in the Saint-Lawrence river [20].

**Figure 2.** NP addition in mussel tissue homogenates. Pure transparent 50 nm polystyrene NPs at 10 μg/mL (**A**) and spiked mussel tissue homogenates (**B**), centrifuged at 3000× *g* for 10 min, the supernatant mixed with one volume of 1.5 M NaCl containing 0.1% Tween-20 and injected into the column. The eluted samples (1 mL) were collected and were analyzed for absorbance at 280 nm, conductivity and neutral red (NR) staining.

**Figure 3.** Chromatographic profile marine and freshwater bivalves exposed to anthropogenic pollution. Representative chromatographic profiles of *Mya arenaria* clam tissue homogenates from a reference site (**A**), a harbor supporting intense boat traffic (**B**), *Elliptio complanata* mussel digestive glands caged at reference lake (**C**), and downstream of a large-population city (**D**) for 3 months during the summer of 2017. The homogenates were centrifuged at 3000× *g* for 10 min, the supernatant mixed with one volume of 1.5 M NaCl-0.1% Tween-20 and directly injected to the size exclusion chromatography column. The data represented the measurements at each 1 mL fraction and normalized to 1 for viewing.

Direct sample analysis of homogenate fractions offers the advantage of measuring other endpoints (enzyme activities, proteins) or specific metabolites in addition to the presence of nanoplastic materials. In the attempt to increase the sensitivity of this methodology and as a means to remove potential interfering large protein filaments, a salting-out step with saturated NaCl and acetonitrile (NaCl/ACN) was used to extract and concentrate NPs. Mussel homogenates were spiked with increasing amounts of fluorescent polystyrene NPs, fractionated with NaCl/ACN methodology, and centrifuged at 3000× *g* for 5 min. The ACN upper phase was mixed with 0.1 volume of 1.4 M NaCl (for total volume calibration of the column) and resolved by SEC with the same elution buffer. Because polystyrene absorbs at 280 nm, the absorbance readings of the elution profile were included (Figure 4A). The data revealed that UV absorbance increased linearly but at Ve/Vt = 0.75–0.8 with the added NPs. In an unspiked sample, low absorbance at 280 nm was measured at Ve/Vt = 0.95–1, suggesting the presence of nonprotein UV-absorbing compounds—perhaps nonpolar metabolites such as tyrosine or phenylalanine amino acids/peptides. Lipids would elute near this volume but would not absorb at 280 nm. The fluorescence peak of the fluorescent NPs eluted at Ve/Vt = 0.75 and peak height/area increased linearly with added NP in the homogenates (Figure 4B). The elution of polystyrene at these fractions (Ve/Vt = 0.75–0.8) suggests that polystyrene NPs shrunk in size with ACN. Polystyrene is completely soluble in ACN or acetone and was shown to shrink in the presence of organic solvents. Polystyrene shrinks in apolar environments from increased electrostatic interactions between the polystyrene polymer layers [21,22]. Hence, the salting-out step displaces polystyrene nanoparticles from water to the ACN phase and leads to osmotic shrinkage. This phenomenon is readily observed with soft polystyrene polymers but seemingly less in hard plastic polymers.

**Figure 4.** *Cont*.

**Figure 4.** Calibration of the NaCl/acetonitrile fractionation method. Mussel homogenates were spiked with increasing quantities of fluorescently-labeled polystyrene NPs, fractionated using the salting-out step, and the acetonitrile upper phase injected in the SEC column. The elution buffer consisted of 0.2% tween-20 and 14 mM NaCl at pH 7.4. Absorbance at 280 nm (**A**) and fluorescence (**B**) were taken at each 1 mL volume of the eluate. The flow rate was 0.75 mL/min.

This approach was also tested on clam and mussel homogenates at sites under anthropogenic pollution/activity, which are likely sources of plastic pollution [23,24]. The eluted fractions of the Sephacryl S500 column were stained with 10 μM NR dye to detect plastic (apolar) material (Figure 5). In mussels caged for 3 months at downstream and rain runoff sites, NR staining was increased at Ve/Vt = 0.75–0.85 compared to reference Lake values (Figure 5A). In control mussels, the NR peak was Ve/Vt approximately 1 co-eluted with salts, suggesting the presence of other nonpolar low molecular weights compounds soluble in ACN. Moreover, the NR positive and staining peak was also detected at Ve/Vt = 0.3–0.4, suggesting the presence of larger plastic nanomaterials in the 100-nm size range. In *Mya arenaria* clam homogenates collected at a polluted site, the same pattern was observed, i.e., the appearance of NR staining at Ve/Vt = 0.75–0.80 at the polluted site and NR staining at Ve/Vt approximately 1 at the reference site (Figure 5B). Interestingly, the presence of large plastic compounds near the void volume of the column (at 160 nm) was not observed in this case.

The present methodology proposed a convenient and cost-effective means to isolate plastic NPs in tissues based on NR staining. Other, more specific, fluorescent stains could be used, such as molecular rotor probes (9-(2,2-dicyanovinyl)julolidine) that are more specific to plastic materials [12,13]. These approaches could be used as a screening tool to separate and detect plastic NPs but this methodology should be confirmed by more specific methods such as single particle plasma–mass spectrometry or pyrolysis gas chromatography–mass spectrometry [25,26]. The isolated fractions could be further analyzed for metals (e.g., arsenic, copper) or organic compounds to detect *Trojan horse* interactions i.e., xenobiotics sorbed to plastic NPs. They could also be measured using FT–infrared spectroscopy

directly or after further ultrafiltration [27]. Nevertheless, the SEC methodology using NR or molecular rotor probes represents a convenient tool for the screening of biological tissues for NPs. However, the proposed methodology is convenient and easy to implement, and more accessible to low-budget laboratories interested in animals exposed to plastic NPs.

**Figure 5.** Chromatographic profile of freshwater mussels *Elliptio complanata* caged for 3 months at downstream a large population city (Montréal, Québec, Canada) (**A**) and in wild *Mya arenaria* clams at a polluted harbor, Saint Catherine (S-C) Bay (**B**). Digestive gland homogenates were prepared were extracted using the salting out step and injected to the SEC column using 0.2% Tween-20 and 14 mM NaCl elution buffer. The data represents NR fluorescence at each 1 mL fraction.

**Funding:** The project was funded under the Saint-Lawrence Action Plan of Environment and Climate Change Canada.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Supplementary or raw data are available at ECCC upon demand.

**Acknowledgments:** This work was funded by the Aquatic Contaminants Research Division of Environment and Climate Change Canada.

**Conflicts of Interest:** The author declares no conflict of interests.

#### **References**

