Probing protein aggregation at buried interfaces: distinguishing between adsorbed protein monomers, dimers, and a monomer–dimer mixture in situ

Protein adsorption on surfaces greatly impacts many applications such as biomedical materials, anti-biofouling coatings, bio-separation membranes, biosensors, antibody protein drugs etc. For example, protein drug adsorption on the widely used lubricant silicone oil surface may induce protein aggregation and thus affect the protein drug efficacy. It is therefore important to investigate the molecular behavior of proteins at the silicone oil/solution interface. Such an interfacial study is challenging because the targeted interface is buried. By using sum frequency generation vibrational spectroscopy (SFG) with Hamiltonian local mode approximation method analysis, we studied protein adsorption at the silicone oil/protein solution interface in situ in real time, using bovine serum albumin (BSA) as a model. The results showed that the interface was mainly covered by BSA dimers. The deduced BSA dimer orientation on the silicone oil surface from the SFG study can be explained by the surface distribution of certain amino acids. To confirm the BSA dimer adsorption, we treated adsorbed BSA dimer molecules with dithiothreitol (DTT) to dissociate these dimers. SFG studies on adsorbed BSA after the DTT treatment indicated that the silicone oil surface is covered by BSA dimers and BSA monomers in an approximate 6 : 4 ratio. That is to say, about 25% of the adsorbed BSA dimers were converted to monomers after the DTT treatment. Extensive research has been reported in the literature to determine adsorbed protein dimer formation using ex situ experiments, e.g., by washing off the adsorbed proteins from the surface then analyzing the washed-off proteins, which may induce substantial errors in the washing process. Dimerization is a crucial initial step for protein aggregation. This research developed a new methodology to investigate protein aggregation at a solid/liquid (or liquid/liquid) interface in situ in real time using BSA dimer as an example, which will greatly impact many research fields and applications involving interfacial biological molecules.

s2 S1 SFG sample geometry and SFG spectra reproducibility Fig. S1 Schematic of the SFG sample geometry used in this experiment.  The details of using the Hamiltonian method to generate SFG spectra can be found elsewhere and will not be repeated. 1 After the SFG spectra were generated as a function of protein orientation by using the Hamiltonian method, a new method (different from the published method 1 ) was used to match the calculated spectra and the fitted experimentally collected spectra. Here a linear least square method was used for evaluating the matching quality between the calculated spectra and the experimental data for each protein orientation. The matching process involves several steps: 1. Fit the experimentally collected SFG spectra and reconstruct the resonant SFG spectra from the fitting parameters.
2. For spectral feature comparison for a particular protein orientation: Normalize the calculated spectra and reconstructed experimental spectra according to the highest peak intensity (at 1645 cm -1 for this study).
3. Calculate the square of the difference between the normalized calculated and experimental spectra at each data point, then sum all the squares for all the data points: where Y is the normalized SFG intensity, and x is wavenumber.

4.
Repeat step 3 for all the orientations. Heat maps can be generated from SE1 as a function of protein orientation for SFG ssp and ppp spectra (These maps display the matching qualities of spectral features of ssp and ppp spectra). 8. Identify the orientation that has the highest final score (lowest square difference).

S3 Matching the fitted experimentally collected spectra of BSA after DTT treatment with calculated spectra based on the BSA monomer structure
Fig. S4 Matching qualities between the experimental data and calculated spectra based on BSA monomer structure for BSA after the DTT treatment: (a) Heat map showing the matching quality between the reconstructed resonant SFG spectra of BSA after the DTT treatment and calculated SFG spectra using the BSA monomer structure. The blue spectra in (b) and (c) show the calculated SFG yyz (b) and zzz (c) spectra of BSA after the DTT treatment with the best matching quality with the reconstructed resonant SFG spectra (shown in red). The matching qualities of the spectra are much worse than those shown in Figure 5 in the main text based on BSA dimer-monomer mixture. The SFG spectra shown in (d) are replotted fitted experimentally collected SFG spectra. The spectra shown in (e) are calculated SFG yyz (blue) and zzz (red) spectra with the best matching quality. The matching qualities of the spectral features can be seen from (b) and (c), while the matching quality for the ssp and ppp intensity ratio can be seen from (d) and (e). The ssp and ppp spectra can be converted to yyz and zzz spectra after considering the Fresnel coefficients.

S4
Matching the fitted experimentally collected spectra of BSA after DTT treatment with calculated spectra based on the BSA dimer structure Fig. S5 Matching qualities between the experimental data and calculated spectra based on BSA dimer structure for BSA after the DTT treatment: (a) Heat map showing the matching quality between the reconstructed resonant SFG spectra of BSA after the DTT treatment and calculated SFG spectra using the BSA dimer structure. The blue spectra in (b) and (c) show the calculated SFG yyz (b) and zzz (c) spectra of BSA after the DTT treatment with the best matching quality with the reconstructed resonant SFG spectra (shown in red). The matching qualities of the spectra are much worse than those shown in Figure 5 in the main text based on BSA dimer-monomer mixture. The SFG spectra shown in (d) are replotted fitted experimentally collected SFG spectra. The spectra shown in (e) are calculated SFG yyz (blue) and zzz (red) spectra with the best matching quality. The matching qualities of the spectral features can be seen from (b) and (c), while the matching quality for the ssp and ppp intensity ratio can be seen from (d) and (e).
To quantify which BSA dimer-monomer mixture ratio can generate calculated SFG spectra with the best matching quality with the experimental data, we varied the dimer-monomer mixture ratio from 1:9 to 9:1, calculated SFG spectra as a function of orientation, and matched the calculated spectra with the experimental data.  Table S2. Fitting parameters of the SFG spectra of BSA on silicone oil shown in Figure 1( Table S3. Fitting parameters of the SFG spectra of BSA on silicone oil after the DTT treatment and the BSA-DTT mixture shown in Figure 4.

S6 PAGE experiments
We performed native PAGE experiments to determine BSA monomer and dimer amounts in the BSA solution before and after silicone oil surface contact, and before and after the DTT treatment. A BSA solution of concentration 1.0 mg/mL with a volume of 10 mL was prepared in a glass container. A silica window coated with silicone oil was placed in the above BSA solution for 30 minutes. The silicone oil surface was then removed from the BSA solution. 10 L of the BSA solutions before and after the silicone oil contact was used in the native PAGE experiment. A new BSA solution of concentration 1.0 mg/mL with a volume of 10 mL was prepared in a glass container. 77 mg of DTT was added to the BSA solution. A silica window coated with silicone oil was placed in the above DTT added BSA solution for 30 minutes.
The silicone oil surface was then removed from the DTT added BSA solution. 10 L of the DTT added BSA solutions before and after the silicone oil contact each was used in the native PAGE experiment.
The native PAGE experiment was performed using a 4-20% Mini-PROTEAN TGX polyacrylamide gel (Bio-Rad). 10 L of samples were mixed with 20 L Native sample buffer (Bio-Rad) and 10 L of this was loaded onto gels. Electrophoresis was performed at 30 V and 4℃ for 14 hrs in Tris/Glycine buffer (Bio-Rad). The gel was stained using InstantBlue Commassie Protein Stain (Novus Biologicals) and visualized via ChemiDoc Touch Imager (Bio-Rad). Protein bands were quantified using Image Lab (Bio-Rad).
We performed several PAGE experiments and qualitative reproducible results were obtained. Figure S4 shows the results from one example run. The four samples are BSA solutions before (1) and after (2) silicone oil exposure and BSA solutions with DTT added before (3) and after (4) silicone oil exposure. Table S4 presented the quantitative results obtained from the PAGE data shown in Figure S4. Figure S4 clearly shows that all the samples are dominated by the BSA monomers, with substantial amounts of BSA dimers. After the DTT treatment, the dimer amounts in the solution reduced noticeably. However, for the BSA solution before and after silicone oil surface contact, the dimer/monomer ratios are not very different.  Table 4 shows that the dimer/monomer ratios decreased after the DTT addition, indicating that DTT reduced some dimer molecules to monomers, as expected. We do not think that the native-PAGE experiments can provide quantitative correlations to SFG data. For quantitative correlations, after the silicone oil exposure, protein molecules (total intensity) should decrease, which was not observed. In the literature, similar methods were used to determine the possible monomer and dimer amounts adsorbed onto a surface. That is, a bulk method (e.g., PAGE) was used to measure the BSA dimer and monomer concentrations or amounts in a BSA solution. Then a surface was placed into contact with the solution.
After that, the bulk solution (after the surface contact) was analyzed again using the bulk method to determine the BSA monomer and dimer amounts. According to the difference before and after the surface contact, possible amounts of BSA monomer and dimer on the surface were deduced. Unfortunately the results obtained from this method may not be related to the amounts of BSA monomer and dimer on the (silicone oil) surface because BSA monomer and dimer can be re-equilibrated in the solution after the s12 surface contact. Also, the interfacial interactions between adsorbed BSA and the surface may lead to dimer dissociation or monomer dimerization on the surface, which cannot be probed using the PAGE method.