Bovine Serum Albumin Bends Over Backward to Interact with Aged Plastics: A Model for Understanding Protein Attachment to Plastic Debris

Plastic pollution, a major environmental crisis, has a variety of consequences for various organisms within aquatic systems. Beyond the direct toxicity, plastic pollution has the potential to absorb biological toxins and invasive microbial species. To better understand the capability of environmental plastic debris to adsorb these species, we investigated the binding of the model protein bovine serum albumin (BSA) to polyethylene (PE) films at various stages of photodegradation. Circular dichroism and fluorescence studies revealed that BSA undergoes structural rearrangement to accommodate changes to the polymer’s surface characteristics (i.e., crystallinity and oxidation state) that occur as the result of photodegradation. To understand how protein structure may inform docking of whole organisms, we studied biofilm formation of bacteriaShewanella oneidensison the photodegraded PE. Interestingly, biofilms preferentially formed on the photodegraded PE that correlated with the state of weathering that induced the most significant structural rearrangement of BSA. Taken together, our work suggests that there are optimal physical and chemical properties of photodegraded polymers that predict which plastic debris will carry biochemical or microbial hitchhikers.

. W-fluorescence fit parameters Table S2 -S4.Summarizing changes observed in BSA upon polymer property changes with aging Additional Methods:

Polymer Preparation and Characterization
Low density polyethylene 30 μm films were purchased from Goodfellow Cambridge Limited (Huntingdon, UK).Samples were prepared into rectangular pieces and soaked in hexane, methanol, and doubly distilled water for 24 h each to ensure the removal of processing additives and unpolymerized monomers or oligomers.Films were mounted on cardstock and irradiated in a Rayonet merry-go-round photoreactor (Southern New England Ultraviolet Corp, Bamford, CT) using 16 Hg vapor lamps for exposure with 254 nm light.
Attenuated Total Reflectance-Fourier Infrared Spectroscopy (ATR-FTIR).Irradiated films were scanned and analyzed using a Nicolet iS510 FTIR spectrometer (Thermo Scientific, Waltham, MA) with a diamond crystal ATR scanning 400-4000 cm -1 with a resolution of 4 cm

Fractional Helicity
The fractional helicity quantifies the structural change in terms of a percentage and provides additional insight into whether a protein is structurally rearranging or unfolding.To better tease out trends in helical structure, fractional helicity was determined using the 230-240 nm slope method of analysis outlined by Wei and colleagues. 1 The resultant slopes were then averaged and converted to fractional helicity (Figure 4 in Main Text).The slope method permits extraction of secondary structural changes in cases where noise prohibits secondary structural elucidation below 230 nm.Essentially, the steeper the curve between 230 -240 nm, the higher the slope and helical content.A decrease in slope corresponds to a decrease in helical content and represents a loss of secondary structure.The slope method is able to elucidate the loss of alpha helical content but does not distinguish between 3.613 and 310 helical substructures.

W-fluorescence Center of Mass (COM) Calculations
Center of mass (COM) data was extracted from individual spectra for each time point in the the fluorescence study and fit following the methodology outlined by Tigner et al. 2 Briefly, the fluorescence signal at each wavelength was weighted by its location and normalized by the sum of all wavelengths reporting fluorescence using the following equation where xnm is the specific wavelength, mnm is the photon count or fluorescence intensity at that wavelength and Xcom is the resultant location of the center of the fluorescence emission signal W-fluorescence fitting W-fluorescence data were fit with a 1-or 2-layer (or 2-or 3-state) cooperative Langmuir sorption model to extract binding time for protein exposed to variably aged polymer.For the spectroscopic constants So.S1, and S2, the bulk population of a system can be fit to a 1-layer (2state) system, Or 3-state system, Where k1 and k2 are the rates associated with transitions and n1 and n2 is the hill coefficient characterizing the sorption cooperativity and () is the detected spectroscopic signal.Fit data reported in Table S1.

Second derivative Analysis
While the 330-350 nm emission peak is the most commonly used spectral characteristic of Wfluorescence, there is a secondary emission at 380 nm as well.By taking the second derivative of fluorescence data in this region and comparing over time, it is possible to track if/when a population increases or decreases fluorophores were quenched during interactions with BSA This emission corresponds to how confined the tryptophan is, with a higher signal intensity reporting a more confined fluorophore.
Second derivative analysis clearly demonstrates that fluorescence subpopulations are well established for BSA and may change but do not evolve over the course of experiment making overall intensity max an appropriate technique for further analysis.

Additional Fluorescence Analysis Discussion
Fractional fluorescence intensity was calculated for all BSA and BSA-PE species and showed a decrease in percent intensity in all BSA exposed to PE (Figure 6A).All BSA-PE samples show evidence of binding kinetics and can be fit using a langmuir binding model.Clear 2-state transitions show up again in the 12 h PE samples.The 3-state system reappears in 0 h and 48 h samples, with local minima occurring around 21 minutes.Conversely, it is not clear whether 24 h reflects a standard 2-or 3-state system.The kinetics extracted from 2-or 3-states models is not as critical here as the necessity of more than one model to describe the system.Because of the difference in degradation landscape on the surface of the PE samples, it is not surprising that different models were required to describe degraded PE samples.The variation in data behavior strongly supports a requisite dynamic response of the protein to initiate polymer surface binding.

Figure S1 .
Figure S1.PDB 3V03 3 of BSA model created using MacPyMol, Domains I, II and III (yellow, orange and pink ribbon) highlighted to show locations of Trp-134 (blue) and Trp-212 (magenta) with respect to the interior and exterior of protein.

Figure S2 .
Figure S2.Diagram of cuvette set-up for fluorescence binding studies.The light path and the positioning of PE sample (blue) and glass insert (green) are shown.Created with BioRender.com

Figure S4 .
Figure S4.SEM images of polyethylene thin films where (a) is pristine, (b) 24 h UV aged, and (c) 48 h UV aged.Qualitative assessment supports the increase of surface roughness with greater weathering.

Figure S5 .
Figure S5.Brightfield images of biofilms of S. oneidensis stained with crystal violet grown on pristine and UV weathered polyethylene.Biofilms appear as purple within the images.

Figure S6 .
Figure S6.Averaged secondary structural CD spectra of free BSA (A), BSA exposed to 0 h PE (B), BSA exposed to 12 h PE (C), BSA exposed to 24 h PE (D) and BSA exposed to 48 h PE (E).All samples were prepared to 7 μM protein in either 10 mM Phosphate buffer or 100 mM HEPES (BSA), pH 7. Data was collected every 4 minutes over the course of 60 minutes (blue → green → red).Arrows denote CD spectral shift over the probe time.

Figure S7 .
Figure S7.Averaged W-fluorescence spectra of free BSA (A), BSA exposed to 0h PE (B), BSA exposed to 12h PE (C), BSA exposed to 24h PE (D), BSA exposed to 48h PE (E) All samples were prepared to 7 μM protein in 10 mM Phosphate buffer, pH 7. Data was collected every 5 minutes over the course of 60 minutes (blue → green→ red).Arrows denote fluorescence spectral shift over the probe time.

Figure S8 .
Figure S8.Second derivative analysis of averaged W-fluorescence of free BSA (A), BSA-0 h PE (B), BSA-12 h PE (C), BSA-24 h PE (D), BSA-48 h PE (E), over the course of one hour.Thermal mapping is reported from negative to zero value (blue →green→ yellow).
-1and scan averaging of 64 scans per spectra.Backgrounds were taken prior to each new scan and the film was sampled in at least 3 different locations to ensure uniformity of the photochemical transformation over the film's surface area.All spectra were acquired using OMNIC software in triplicate (Thermo Scientific, Waltham, MA)

Table S1 .
Rate and hill coefficient fit to either 1-or 2-layer (2 state or 3 state) population corresponding to Eqns S2 and S3 for W-fluorescence maximum decrease corresponding to data reported in SI FigureS7Aand D.