MOLECULAR MECHANISMS OF PROTEIN BINDING BY PERFLUOROALKYL SUBSTANCES (PFASs)

Comparative investigation of the intermolecular chemical interaction (binding) between bovine serum albumin (BSA) and four polyfluoroalkyl substances (PFOA, PFNA, PFHxS, PFOS) by F NMR spectroscopy with synchronous observation of the F signals from both ends (head and tail) of the polyfluoroalkyl molecules was performed at three temperatures of 298K, 304K and 310K. Chemical shifts of F NMR peaks in solutions of PFAS with BSA were used for evaluation of the dissociation constants, Kd, for both known mechanisms of PFAS binding with BSA: by hydrophobic interaction of the PFAS molecule carbon chain tail in the hydrophobic pockets of BSA (Ω mechanism) and by hydrogen bond and electrostatic interaction of the PFAS molecule head group with charged regions on the BSA surface (α mechanism). It was established that highest affinity of all four PFAS:BSA complexes is by the Ω mechanism of binding with Kd at 310K reaching values as low as 3.9×10, 6.5×10, 7.7×10 and 1.9×10 M for PFOS, PFOA, PFNA, and PFHxS, in comparison with 5.7×10, 5.6×10, 6.6×10 and 5.4×10 M values for α mechanism of binding. Evaluation of the thermodynamic parameters (enthalpy ΔH, entropy ΔS, and Gibb’s free energy ΔG) showed that binding of PFOA, PFNA, PFHxS, and PFOS with BSA by both α and Ω mechanisms is accompanied by negative ΔH and ΔS and positive ΔG which are characteristic for binding of two large hydrophobic molecules with each other by weak hydrogen bond and van der Waals’ forces. Kd for binding of the branched isopropyl isomers of PFHxS and PFOS with BSA were measured at 310K as 8.8×10 and 7.6×10 M, correspondingly, which indicated less affinity of isomers with the surface of BSA in comparison with the linear structure of PFHxS and PFOS molecules possibly due to “bulky” structure of the branched isomer head.

iv ACKNOWLEDGMENTS First, I would like to thank The Almighty God for giving me strength and wisdom to finish this degree, even at the most trying of times.
Secondly, my wife Abby for being my rock, helping me whenever I asked even with your own 3D printing project going on. Words cannot express how happy you make me and how proud I am of you. Thank you for your endless love and being the     were detected in the blood samples of over 99% of the individuals examined. 5 They have also been found in cord serum of infants and in breast milk of nursing mothers. 6,7 Their persistency is demonstrated by their long half-lives in humans, estimated to be 3.8 years for PFOA, 2.5 years for PFNA, 8.5 years for PFHxS, and 5.4 years for PFOS. 1,8 From epidemiological studies, the critical effects of PFASs are an increase in serum total cholesterol in adults, 9 a decrease in antibody response for vaccinations, 10 pregnancy-induced hypertension and preeclampsia, 11 and cancer. 12,13 The mechanisms by which PFASs interact and transport throughout the human body are not well understood and are still being researched. 10,14 These PFASs are detected primarily in the blood and the liver of humans, highlighting their proteinophilic nature. 15 Human serum albumin (HSA), being a ligand binding protein, plays an important role in the accumulation pattern of PFASs in the blood and the liver tissue. 16,17 The ability of a molecule to bind to HSA influences its lifetime and excretion from the body. 18 HSA at 0.6 mM is the most abundant protein in humans, transporting different natural and exogenous ligands including fatty acids, pharmaceuticals and small organic anions throughout the human body. 18 Studies have estimated that over 90% of the total PFASs in the body will be bound to HSA. 19 With an aliphatic tail and anionic head group, PFASs are analogous to fatty acids and bind primarily with HSA due to its abundancy. 20 HSA contains seven distinct fatty acid binding sites that are asymmetrically distributed around the protein. 18 Competition for binding sites between molecules can significantly affect the equilibrium between HSA-bound and HSA-unbound forms of the PFAS molecule. 16,18 Dissociation constants, Kd, for PFAS-HSA binding reportedly range from 10 -2 M to 10 -6 M. 17,19,21 The range of Kd over four orders of magnitude can be explained in part by the variety of experimental techniques amenable to different mechanisms of PFAS-HSA binding. 22 HSA binds PFASs by two thermodynamic interactions: hydrophobic forces via the carbon tail in the hydrophobic pockets of HSA (specific binding) or by hydrogen bonds and electrostatic forces with the anionic head group in charged regions on the HSA surface (non-specific binding). Literature is conflicted on the mechanisms of PFAS-HSA binding and the number of binding sites on HSA. [18][19][20][21] Fluorescence spectroscopy experiments have shown that PFASs interact with HSA specifically in hydrophobic cavities. These specific binding sites are sterically hindered and have a particular geometry that binds a limited number of PFAS molecules per protein. Studies using equilibrium dialysis report PFASs binding specifically and non-specifically to HSA. Once the specific sites with the higher binding affinity have been filled the PFASs will continue to be adsorbed nonspecifically throughout the charged regions on the surface of HSA with greater orientational freedom. Many PFASs are bound by non-specific adsorption because there is more available surface area than there are hydrophobic pockets. 24 The adsorption phenomena can also help explain the number of PFAS binding sites on HSA ranging from 1 to 50 as described in literature. 20,25 Fatty acids have Kd values in the same range as PFASs. 18 (1) and (2), and is inversely proportional to the association constant, Ka.
The chemical shift of a ligand NMR signal in the presence of a protein is commonly used to monitor the formation of a protein-ligand complex. 1D NMR spectra of small-molecules (MW ≤ 500 Da) typically have sharp peaks due to a shorter dipole-dipole relaxation. Binding of a ligand to a high molecular weight molecule such as a protein induces peak broadening and a corresponding chemical shift in the NMR signal because the bound ligand experiences the slow relaxation time of the protein compared to the free state of the ligand. 31 Kd values were determined based on the resonance chemical shift, ∆δ, of the PFAS bound to the BSA relative to its unbound state in solution. 19,20,26 Chemical shift perturbation (CSP) theory can determine the binding affinity of the ligand, the binding site(s) and the structure of the complex. 28 The observed chemical shift is the population-weighted average of free and bound ligands, which allows the determination of Kd from measurement of the peak positions. 19,20,22,28 The chemical shift of the PFAS resonance peak is sensitive to structural differences of its bound and unbound states, meaning that a genuine binding interaction of PFAS with BSA will produce a perturbation. 22 Kd values were determined graphically based on equation (3): where ∆δ = δ obs − δ free is the net chemical shift of the monitored resonance of the  Dissociation of PFAS with BSA is accompanied by a change of the Gibb's free energy, ΔG, that can be evaluated using equation (4): where R is the ideal gas constant and T is absolute temperature. For small temperature ranges the change in the enthalpy ΔH and entropy ΔS of a thermodynamic system are essentially constant and equation (5) can be used for ΔG without the need to take into account the temperature dependencies of ΔH and ΔS: on the van't Hoff equation (6): The homologues. The ECF of short-chain sulfonates gives significantly better percentage of the linear compounds. 33,38,39 This is confirmed in Figure 2; the 1D 19 F NMR spectra for PFHxS and PFOS exhibited roughly 5% and 30% branched isomers based on peak integration of the linear and branched peaks. Understanding how isomers differ in binding to HSA has not been studied in detail due to challenges connected with the coalescing and splitting of isomer peaks. 38

Dissociation Constants (Kd) and Binding Mechanisms
19 F NMR spectroscopy observation is a powerful tool to study protein-ligand interactions because each fluorine atom gives an individual signal in the spectrum that carries information on the local chemical environment. (22,40) An advantage of 19 F NMR spectroscopy is that it can measure Kd in the μM range which is not well covered by traditional biochemical binding assays. 22 The following criteria are required to identify PFAS binding mechanisms: (1) the molecular recognition event is sufficiently defined to provide a well-structured binding complex; (2) there are a number of independently varying 19   Langmuir sequence and that the favorite binding site is located in the protein hydrophobic core. 42 This two-step binding mechanism is supported by the data found in this study shown in Figure 5 and in Figure S1 (see supplementary material in Appendix A). An animated illustration of this two-step sequence is shown in Figure 6.

Sulfonates (PFHxS and PFOS)
PFOS 19 F NMR peaks had almost two orders of magnitude lower signal intensity in comparison with the other three PFASs with or without BSA due to the significant content of isomers (~30%). 28 PFHxS exhibited less isomers content (~5%).
To understand the isomer impact on PFAS interaction with BSA, an additional resonance peak at -71.868 ppm, labeled (I), was monitored throughout the CSP analysis. This resonance peak corresponds to the most abundant isopropyl branched isomer found in both PFHxS and PFOS based on spectra peak integration. Figure 9 shows the structure of the isopropyl isomer determined by previous studies. 33 Figure 11. Figure 10 show that in case of PFHxS, the values of the chemical shift for the isomer peak, I, are not significantly different from the α and Ω peaks for the linear PFHxS molecules. Figure   11 and Figure S2 show the return of the chemical shift for the α, Ω and I peaks with   Figure 13. Figure 12 shows that at the same concentration of PFOS the isomer peak, I, has the larger chemical shift, Δδ, in comparison with α and Ω peaks, corresponding to the linear PFOS molecules. PFOS peaks contained a significant chemical shift up to 100:1 PFOS:BSA, as seen in Figure   13 and Figure S2. PFOS was the only compound to have neither α, Ω and I peaks return to their original shifts as seen in Figure 13; this could be due to the large percent of isomers present causing there to be less free linear PFOS in solution with BSA. (28) Further work on isomer effects on PFAS binding is necessary to better understand their impact on PFAS-BSA binding.     (Figures S3 and S4). ΔG is calculated using equation (5) for the α and Ω binding mechanisms of the PFASs with BSA. It was found that both ΔH and ΔS were negative and ΔG was positive for all four PFASs studied, shown in  Simultaneous evaluation of both ΔH and ΔS from only a van't Hoff plot using equation (6) can give erroneous results due to the enthalpy-entropy compensation effect. Enthalpy-entropy compensation (EEC) is a well-known phenomenon manifested in many chemical and biochemical systems. Linear plots of ΔH versus ΔS are often treated as authentic representations of a thermodynamic relationship, or an EEC effect. (45) Errors can arise by the long extrapolation of the linear plot by an unknown law to get the y-intercept for ΔS, especially in biological systems with weak intermolecular interactions.
To check the PFAS-BSA binding thermodynamic parameters for the EEC effect as a result of errors in extrapolation of ΔS from the van't Hoff plot of equation (6) an additional method to calculate ΔS is used. ∆H can still be obtained from the van't Hoff plot of equation (6) as the temperature range is narrow. The ΔG is calculated from equation (4) with subsequent calculation of ΔS using equation (5).
Additional measurements of ΔG and ΔH through independent isothermal titration calorimetry (ITC) may provide more accurate evaluation of the thermodynamic parameters of the molecular interactions between PFAS and BSA.
Comparison of the ΔS calculated using equation (4) and equation (5) with ΔS received from van't Hoff plot of equation (6) seen in Table 1 show that they are almost identical for the α and Ω mechanisms of binding for PFOA, PFNA, PFHxS and PFOS with BSA. A good match of ΔS calculated by two different methods in Table 1 indicates that the thermodynamic parameters of the PFAS binding with BSA defined in this work were not compromised by the EEC effect.
The thermodynamic parameters found in this study characterizing the binding of PFOA, PFNA, PFHxS and PFOS with BSA are in agreement with published data received by different techniques. 13 (23) These values also suggest that PFASs have a stronger binding affinity than previously thought by other studies and they may have the ability to displace natural fatty acids from the hydrophobic pockets of BSA. Further investigation into this two-step binding sequence and its effect on PTE, as well the ability of PFASs to displace natural fatty acids in physiological conditions should be performed.
The Kd for the α mechanism of binding within an order of magnitude of the Ω mechanism of binding suggests that both carboxylic and sulfonate head groups have a strong binding affinity. This is not seen in their hydrocarbon analogs that primarily rely on their hydrophobic tail to bind with BSA. The ability of the PFAS head group to reversibly bind with BSA should be looked at more closely, to better comprehend their physiological effects and how they are transported throughout physiological systems. Additionally, determining a single Kd for a PFAS based on both α and Ω mechanism of binding, such as equation (9) should be investigated to better predict their binding affinity and PTE, such as the relationship seen below. Additional studies on the ability of BSA to stabilize PFAS micelle formation and aggregation should be performed. Limited work has been published in this area and many PFASs still lack reported physiochemical property data. 19 F NMR is a powerful tool to get high throughput results for a large quantity of PFASs.
Over the last decade, 19 F NMR equipment and software has advanced greatly allowing experiments to be performed at lower concentrations, with larger more complicated proteins. 19 F NMR should become a standard procedure when analyzing the ability of PFASs to bind and how their properties are affected in various systems.
Future work will be dedicated to multiple peak analysis programs and peak decoupling programs, such as PeakFit and Originlab, which are useful in analyzing a large number