HPLC-MS was used to validate the simple colorimetric detection method for the determination of PFOS during dialysis process. The two methods have achieved the consistent results in the measurement of three PFOS dialysis solutions (see Additional File 1). Therefore, the colorimetric method is feasible for the determination of PFOS in the present work. The interaction of PFOS (L) with SA and DNA (M) can be summarized below:

N L + M → ML N Initiation c L0 c M0 0 Equilibrium c L = c L0 − N c M0 c M → 0 c M0 MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@703C@

Both c_L0 and c_M0 are the initial concentrations of PFOS and of SA or DNA pipeted into dialysis bags. The symbol c_L is the equilibrium concentration of PFOS, N the saturation binding number of PFOS, K_b the binding constant. Considering PFOS was diluted 4 times during the equilibrium dialysis, the effective fraction (f) and the molar ratio (γ) of PFOS bound to SA or DNA can be calculated by the relationships below 66:

f = 1 − 4 × c L c L 0 MathType@MTEF@5@5@+=feaagaart1ev2aqatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaemOzayMaeyypa0JaeGymaeJaeyOeI0scfa4aaSaaaeaacqaI0aaniiaacqWFxdaTcqWGJbWydaWgaaqaaiabdYeambqabaaabaGaem4yam2aaSbaaeaacqWGmbatcqaIWaamaeqaaaaaaaa@3A1A@

and

γ = f × c L 0 c M 0 MathType@MTEF@5@5@+=feaagaart1ev2aqatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaaccaGae83SdCMaeyypa0tcfa4aaSaaaeaacqWGMbGzcqWFxdaTcqWGJbWydaWgaaqaaiabdYeamjabicdaWaqabaaabaGaem4yam2aaSbaaeaacqWGnbqtcqaIWaamaeqaaaaaaaa@39D9@

It can be deduced that

γ = c L 0 − 4 c L c M 0 MathType@MTEF@5@5@+=feaagaart1ev2aqatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaaccaGae83SdCMaeyypa0tcfa4aaSaaaeaacqWGJbWydaWgaaqaaiabdYeamjabicdaWaqabaGaeyOeI0IaeGinaqJaem4yam2aaSbaaeaacqWGmbataeqaaaqaaiabdogaJnaaBaaabaGaemyta0KaeGimaadabeaaaaaaaa@3AE8@

By measuring a series of PFOS solutions containing known concentrations of SA or DNA, γ was calculated using Eq. 4. As is shown in Fig. 1A, the γ value increases with increasing PFOS concentration. The Langmuir isothermal model below was used to fit the experimental data.

1 γ = 1 N + 1 K N c L MathType@MTEF@5@5@+=feaagaart1ev2aqatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaqcfa4aaSaaaeaacqaIXaqmaeaaiiaacqWFZoWzaaGccqGH9aqpjuaGdaWcaaqaaiabigdaXaqaaiabd6eaobaakiab=TcaRKqbaoaalaaabaGaeGymaedabaGaem4saSKaemOta4Kaem4yam2aaSbaaeaacqWGmbataeqaaaaaaaa@3A69@

where K is the adsorption constant. The plots of γ^-1 versus c_L^-1 are shown in Fig. 1B. The good linear relationships of the plots indicate that the bindings of PFOS to peptide chains and DNA duplex obey the Langmuir isothermal model. The interactions of PFOS with SA and DNA are the chemical adsorption in monolayer. From the intercepts of the regression lines in Fig. 1B, N of PFOS molecules bound per molecule of SA and per bp of DNA was calculated to be 45 and 0.36, respectively.

The electrostatic attraction plays an important role in the binding of sulfonic ligands to proteins. Several studies in our laboratory have revealed that electrostatic attraction between the sulfonic groups of azo compounds and the positively charged amino acid residues (AARs) of SA is the main contributor that fixes the position of these compounds in protein in acidic media and induces the subsequent combined actions of multiple non-covalent bonds, e.g., hydrogen bonds, hydrophobic interactions and van der Waals force 6768. Given the dissociation constants (K_R) of the side groups (R) of basic and acidic AARs of SA (10.53 for Lys, 6.00 for His, 12.48 for Arg, 3.65 for Asp, and 4.25 for Glu), only the side groups of Lys and Arg residues are protonated and positively charged in neutral solutions in our study, so selectrostatic attraction between the sulfonic acid group of PFOS and positively charged AARs of SA will be much weaker than that in acidic media. But concerning the high polarity of PFOS, the electrostatic attraction between the sulfonic acid group of PFOS and the positively charged -NH₃⁺ groups of Lys and Arg of SA can still play an important role in the PFOS-SA binding 6970. Moreover, the hydrophobic interaction will also occur between the long alkyl group of PFOS and the nonpolar side groups of AARs of SA such as Lys, Arg, His and so on. Similarly, PFOS may bind to DNA via hydrophobic interaction resulting from the long alkyl group and the homolateral bases in the groove of DNA.

The stability of non-covalent interaction is always affected by various environmental conditions such as pH, ionic strength and temperature 7172. As is shown in Fig. 2A, γ for the PFOS-SA binding decreases obviously with increasing pH. It implies that the acidic media are more favorable for PFOS binding to SA. As we know from the dissociation constants (K_R) of the side groups (R) of basic and acidic AARs of SA, more side groups (R) of AARs are protonated and positively charged in acidic media. Thus, the electrostatic attraction between the sulfonic acid head of PFOS and the positively charged AARs will be much stronger than that in neutral media 6768, which result in the increasing of binding number of PFOS to SA. Moreover, SA tends to unfold in acidic media, so steric hindrance decreases, leading to an increasing of possible binding sites as well as the binding number. As to DNA, increasing binding is obviously observed at pH 2. Similar to SA, DNA tends to unfold in acidic media, too. The increasing of possible binding sites resulting from the unfolding of DNA may be the reason that PFOS-DNA binding number increases. In addition, a peak of γ appears at pH 7.4 for the PFOS-DNA binding and then γ decreases at pH 8.4 and 9.4 afterwards. It implies the neutral media is comparatively more favorable for PFOS binding, indicating the potential risk of PFOS toxicity under the physiological condition of wildlife and human.

Sodium chloride was added in the PFOS-SA/DNA solutions to investigate the effect of electrolyte on the non-covalent interaction. Human blood normally contains approximately 0.15 mol/l electrolyte. Although γ of PFOS-SA binding (Fig. 2B) increases when the electrolyte is less than 0.15 mol/l, γ is only slightly different in 0.15 mmol/l electrolyte from that in 0.25 mol/l electrolyte. This implies that ions in human blood would not affect PFOS binding. From the curves in Fig. 2B, a peak of γ of the PFOS-DNA binding appears at 0.25 mol/l electrolyte, indicating that ions in body fluids will affect the PFOS-DNA binding more or less.

High temperature has two opposing effects on non-covalent interaction. On one hand, the peptide chain and DNA duplex will expand with heating and the three-dimensional conformation will be favorable for the insertion of organic molecules. On the other hand, such expansion will increase the distance between peptide chains and DNA duplex, leading to the redistribution of the effective binding sites on the peptide chain and DNA duplex and the subsequent desorption of small organic substances 73. The balance between these mechanisms decides the effect of temperature on the non-covalent interactions. As is shown in Fig. 2C, γ values for the PFOS-SA/DNA interactions decrease with increasing of temperature. Obviously the distance between the peptide chains or DNA duplex increases as a result of temperature increase and lead to the desorption of PFOS, which result from the subsequent decrease of binding sites for PFOS. Therefore, γ values of the PFOS-SA/DNA interactions decrease.

The intrinsic fluorescence analysis of tryptophan residue (Trp) of SA is often used in the investigation of a ligand-protein binding. The quenching of protein fluorescence depends on the degree of exposure of Trp residue to the polar, aqueous solvent and its proximity to specific quenching groups such as protonated carboxyl, protonated imidazole and deprotonated ε-amino groups 74. Two models have been proposed for the quenching of protein fluorescence: static quenching and dynamic quenching 75. In static quenching, the quencher binds to Phe, Tyr, especially Trp residues of SA and consequently the compound in ground state that has no fluorescence is formed, resulting in the decrease in the fluorescence intensity of SA. On the contrary, in dynamic quenching the decrease in the fluorescence intensity of SA is only because of the molecular collision between the quencher and SA. Higher temperatures usually result in faster diffusion, collision and hence larger amounts of dynamic quenching. In order to confirm the quenching mechanism, the fluorescence quenching data presented here was assumed to be dynamic quenching. The classical relationship often employed to describe the dynamic quenching process is the Stern-Volmer equation:

F 0 F = 1 + K q τ 0 [ Q ] MathType@MTEF@5@5@+=feaagaart1ev2aqatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaqcfa4aaSaaaeaacqWGgbGrdaWgaaqaaiabicdaWaqabaaabaGaemOrayeaaGGaaOGae8xpa0JaeGymaeJaey4kaSIaem4saS0aaSbaaSqaaiabdghaXbqabaGccqWFepaDdaWgaaWcbaGae8hmaadabeaakiab=TfaBnXvP5wqSXMqHnxAJn0BKvguHDwzZbqegCuzVj3zPfgaiqaacaGFrbGae8xxa0faaa@464E@

Where F₀ is the fluorescence intensity of SA without PFOS, F is the fluorescence intensity of SA in the presence of PFOS, τ₀ is the lifetime of the fluorophore of SA in the absence of quencher, usually 10^-8 s for biomolecule, and Q is the concentration of PFOS (mol/l), the quencher, K_q is the bimolecular quenching rate constant and 2.0 × 10¹⁰ l/(mol·s) is the highest value of K_q for dynamic quenching. In this study if K_{q, PFOS-SA} (K_q for PFOS-SA binding) is calculated to be much larger than 2.0 × 10¹⁰ l/(mol·s), static quenching occurs for PFOS-SA binding; if K_{q, PFOS-SA} is calculated to be less than 2.0 × 10¹⁰ l/(mol·s), dynamic quenching occurs. By regression of plots F₀/F vs Q (Fig. 3B), the relationship y = 0.0436x + 0.962 (R = 0.9903) was arrived. The K_qτ₀ value could be equivalent to 0.0436 (PFOS in umol/l in Fig. 3B). K_{q, PFOS-SA} is calculated to be 4.36 × 10¹² l/(mol·s), which is much greater than the highest dynamic quenching constant, 2.0 × 10¹⁰l/(mol·s). So the interaction between PFOS and SA resulted in static quenching. Moreover, the fluorescence spectra of SA were obtained in the presence of PFOS (Fig. 3A). Not only a gradual decrease in the fluorescence intensity of SA but also a blue shift (10 nm) in the emission wavelength were observed, indicating that PFOS bound to the Trp residue, W214 of SA and the microenvironment around Trp, Phe and Tyr residues was changed consequently. This phenomenon confirmed the occurrence of static quenching between PFOS and SA again.

SA contains a hydrophobic cavity and all hydrophobic AARs were present in the cavity 76, including Trp residue, which is favorable for hydrophobic group to enter. PFOS enters the cavity and interacts with the hydrophobic AARs via hydrophobic interaction between the alkyl group of PFOS and the aromatic group of AARs, causing the quenching of SA fluorescence. In addition, concerning the high polarity of the sulfonic acid group of PFOS, the electrostatic attraction could occur between the sulfonic acid group of PFOS and -NH₃⁺ of Lys and Arg residues of SA.

The combinations of covalent and non-covalent interactions among a protein or DNA result in the specific conformation and the corresponding function of biomolecules. When organic compounds such as a pollutants, drugs or toxicants interact with protein or DNA, the internal non-covalent bonds of the peptide chain or DNA duplex are often disrupted, possibly changing the original conformation or even their special function. CD spectrometry is often used to evaluate the secondary structure of DNA 77 or a protein 68, such as the fractions of β-pleated sheet, α-helix and β-turn.

The molar ellipticity CD curves of PFOS-SA/DNA solutions are given in Fig. 4. As is shown in Fig. 4B, the fractions of β-pleated sheet, α-helix, and β-turn of SA changed obviously in the presence of PFOS. With the addition of PFOS, the β-pleated sheet fraction of SA decreases from 20.2% to 11.3% whereas that of α-helix and β-turn increases from 22.8 to 26.1% and 26.5 to 29.8%, respectively. Obviously the disappearance of β-pleated sheet results in the increase of α-helix and β-turn content. Although in neutral solutions the non-covalent bonds between PFOS and SA are much weaker compared with that in acidic solutions 6768, the addition of PFOS causes obvious change of secondary structure of SA. From the data above, the addition of PFOS transforms some β-pleated sheet into α-helix and β-turn.

The CD spectra of DNA in the 230–300 nm scope show a signal characteristic of a B-form helix with a negative band at 245 nm and a positive band at 275 nm. The enhancement of CD spectra at both 245 and 275 nm in the presence of PFOS and the shape change of the positive and negative band (Fig 4C) indicated that PFOS binds in the groove of DNA. On account of the hydrophobic alkyl group of PFOS, PFOS may bind in the groove of DNA paralleling with the phosphate backbone via van der Waals force and hydrophobic interaction. Thus, the electron cloud of bps becomes expanding owing to the reversed pulling of PFOS so as to the enhancement of CD of DNA at 275 nm.

Relating thermodynamic parameters to structural and biochemical data allows a better understanding of the mechanism of the biomolecule-ligand reaction. The ITC measurements provide information on thermodynamic quantities such as enthalpy and heat capacity changes during the molecular interaction based on the heat produced by reactions In recent years, ITC measurements have been widely applied to study e.g. protein-ligand interaction 78, DNA-ligand interaction 77.

The ITC experiments were conducted by injecting PFOS solutions into the ITC cell containing SA or DNA in pH 7.40 media at 37°C. In each experiment, an exothermic heat pulse was detected following each injection. Its magnitude progressively decreased until a plateau is reached indicating saturation of binding. The heat involved at each injection was corrected for the heat of dilution, which was determined separately by injecting the PFOS solutions into the B-R buffer and then divided by the number of moles injected. Values for the equilibrium binding constant (K_b), enthalpy change (ΔH), and entropy change (ΔS) of the reaction were obtained and calculated by the Gibbs free energy (ΔG) equation:

ΔG = -RTln K_D = ΔH - TΔS

As is shown in Fig. 5, all ΔH are much less than 60 kcal/mol, so the PFOS-SA/DNA interactions are non-covalent 79, involving the electrostatic attraction, hydrophobic interaction, and van der Waals force. Moreover, the heat released from both PFOS-SA (Fig. 5A) and PFOS-DNA (Fig. 5B) interactions approaches zero when c_L0/c_M0 > 10 and 0.3, respectively, indicating the plateau is being reached. All the reactions are driven by entropy change and thus spontaneous because ΔH is much less than that -TΔS (Fig. 5B-2).

SA is composed of three homogenous and helical domains (Fig. 6A). Every domain consists of two sub-domains, which face each other like two notches. All the hydrophobic amino acid residues embedded in the sub-domains formed three hydrophobic cavities 2. During protein-ligand interaction, polar bonds are likely to be formed outside of the hydrophobic cavities because polar residues of SA predominate here. The reactions are usually exothermic due to the high energy of polar bond. On the contrary, hydrophobic interactions are likely to occur inside of the hydrophobic cavity. The reactions are usually endothermic with the folding of protein and decreasing of entropy change.

From Fig. 5A, the binding of PFOS to SA corresponds to two-step sequence model: on the surface and inside the cavity. First, the heat released from PFOS-SA interactions was negative when c_L0/c_M0 < 10, indicating the occurrence of exothermic reaction and the formation of polar bonds. Because the polar bonds are more likely to be formed outside of the hydrophobic cavity of SA, it is suggested that PFOS binds to the surface of SA via electrostatic attraction between the sulfonic acid group of PFOS and -NH₃⁺ of Lys and Arg residues of SA (Fig. 6A). After the binding sites of SA surface had been occupied, PFOS molecules entered the cavity of SA easily (Fig. 6A). When c_L0/c_M0 > 10, the heat released from the PFOS-SA interaction was positive, indicating the occurrence of endothermic reaction and hydrophobic interaction. Because the hydrophobic interaction is more likely to occur inside of the hydrophobic cavity of SA, it is suggested that PFOS enters hydrophobic cavity of SA and binds to the nonpolar AARs there. For example, PFOS interacts with the aromatic side group of Trp residue of SA. Owing to the small cavity where inside binding occurred and the consequent effect on SA conformation, SA folds and its conformation changes.

As is shown in Fig. 5B, ΔH is very small. Most of the heat released from the PFOS-DNA interaction was positive, indicating the occurrence of endothermic reaction and hydrophobic interaction. PFOS probably interacts hydrophobically with the homolateral bases of DNA via the alkyl group, paralleling with the phosphate backbones of DNA (Fig. 6B). The conformation of DNA changed owing to the hydrophobic interaction that pulls the contiguous bps closer. This is confirmed from high ΔS values (Fig. 5B-2).

The relationship between structural transformation of protein and its functioning is of great significance in organisms. During interaction process, a small organic compound may bind to the peptide chain, regulating its three-dimensional structure and even changing its corresponding function 8081. Although non-covalent binding is often weak and non-specific, a combination of many non-covalent bonds may alter the conformation and function of the protein 82. Serum albumin is the most abundant protein of blood plasma. It is responsible for the maintenance of both the oncotic pressure and pH of blood. Moreover, it is the major plasma carrier protein in blood which can bind a large number of ligands, e.g., amino acids, vitamins, fatty acids, drugs and so on.

The effects of PFOS on SA function were determined as shown in Fig. 7. With the addition of PFOS, the binding ratio of VB₂ to SA decreases obviously comparing to that in the absence of the pollutants and it indicates the inhibition of SA carriage capacity by the pollutants. At the normal physiological condition, 1.2 mmol/l PFOS reduces the binding ratio of VB₂ to SA by more than 30%. It's likely that the competition of binding sites in SA by pollutants and VB₂ and the subsequent conformation change of SA are unfavorable for the binding of VB₂ to SA. Therefore, non-covalent binding of the organic compound severely affects the physiological function of protein by altering its conformation and overlapping its active sites.

A Model F-4500 fluorospectrophotometer (Hitachi High-Technologies Cooperation, Tokyo, Japan) was used for the concentration measurement of vitamin B2 and fluorescence measurement of protein solutions in the presence of PFOS. A Model Lambda-25 spectrometer (Perkin-Elmer, Shelton, CT 06484, USA) was used to determine the concentration of PFOS during the process of equilibrium dialysis. The spectrometer was computer-controlled using UV WinLab software (Version 2.85.04). The Isothermal Titration Calorimetery (ITC) experiments were carried out on a Model VP-ITC system (MicroCal Inc., USA) with VPViewer 2000 (Version 1.04.0018). A Model J-715 CD spectropolarimeter (Jasco Instruments, Tokyo, Japan) with secondary structure estimation-standard analysis measurement software (715/No. B014460524, Jasco) was used to determine the conformation of protein and DNA. Model RC 30-5K semi-permeable membranes (Molecular Weight Cut Off 5 KDa, Shanghai Green Bird STD) were used for equilibrium dialysis. A Model DK-8D electrothermic multiporous constant temperature water-bath (Shanghai Yiheng Technol., Shanghai, China) was used in the temperature experiment. Solution pHs were measured with a Model pHS-25 acidity meter (Shanghai Precise Sci. Instrum., Shanghai, China). An Accela U-HPLC-system connected to a TSQ Quantum Access mass spectrometer (Thermo Fisher Scientific, America) was used to validate the CPC-ECR colorimetric method used for PFOS determination in this study.

Both 0.100 mmol/l SA (Sigma, A7906) and 3.7 mmol/l bps of DNA (Shanghai Chemical Reagents Company, Shanghai, China) were prepared in deionized water as a stock solution and stored at 4°C. The precise concentration of protein and DNA were determined by the UV method. PFOS stock solution (10 mmol/l) was prepared by diluting 1 ml of PFOS water solution (40% in water, Fluka, USA) in 100 ml of deionized water and stored at 4°C. Vitamin B₂ stock solution (0.1276 mmol/l) was prepared by dissolving 12 mg of Vitamin B₂ (VB₂) (purity 99%, Shanghai Chemical Reagents Company, Shanghai, China) in 250 ml of deionized water and stored at 4°C. The 1.25 mol/l NaCl and Britton-Robinson (B-R) buffer (pH 7.40) containing 0.040 mol/l phosphoric acid, acetic acid and boric acid was used for the experiments.

PFOS reacts with cetylpyridinium chloride (CPC) by ion-pair binding to form the PFOS-CPC complex 83. The anionic ligand, eriochrome cyanine R (ECR) was used to replace CPC from the PFOS-CPC complex to form a CPC-ECR complex. The color change was used for determination of PFOS only in this work. The replacement procedures are followed. 2.0 ml pH 3.80 acetate buffer, 0.30 ml 1.00 mM CPC and 0.50 ml 1.00 mM ECR were mixed in 5-ml calibrated flasks. After mixing for 15 min, various amounts of PFOS were added and the solutions diluted to 5 ml. After 10 min, the absorbance of each solution was measured at 626 nm against a reagent blank without PFOS.

A schematic diagram of equilibrium dialysis experiment was shown in Fig. 8. For equilibrium dialysis assay of PFOS, 12.5 ml mother liquor containing 2.5 ml of B-R buffer, 0.15 mol/l NaCl, a known volume of PFOS and deionized water was pipeted into dialysis bags (1). 37.5 ml of the dialysate solution containing 0.15 mol/l NaCl, 7.5 ml of B-R buffer and deionized water was added to the dialysis bag (3). The temperature of the water bath (4) was kept constant at 37°C by adjusting the thermostat magnetic stirrer (5). 1 ml of the dialysis solution (3) was collected into 5-ml calibrated flask every 2 h from the sampling tube (6). According the above detection method, the PFOS concentrations (c_L) of the dialysis solutions were determined. Variation of c_L with the dialysis time was shown in Fig. 9B. The dialysis of PFOS remains at equilibrium between 8 and 12 h. After dialysis for 10 hours, 1 ml of the dialysis solution (3) was collected in 5-ml calibrated flasks from every sampling tube (6). According to the detection method above, the absorbance of the dialysis solution was measured at 626 nm. The working standard curve of PFOS was established as shown in Fig. 9A. By five repetitive determinations of 0.200 mmol/l PFOS, the result was 0.199 ± 0.006 mmol/l PFOS. The recovery rates of the detection method are between 96.5 and 102% and the relative standard deviation 3%. The detection method was used for the determination of PFOS in the dialysis equilibrium solution, where no complicated interference substance co-existed.

To validate the above detection method, a high performance liquid chromatography (HPLC) system coupled to a triple quadrupole mass spectrometer (MS) was used with the method being modified 31 (Additional File 2). After equilibrium dialysis of 100.0, 150.0, 200.1 mg/l PFOS in SA presence at 37°C for 10 hours, 1.00 ml of the dialysis solution was collected and determined by the above detection method. Meanwhile, another 1.00 ml of the dialysis solution was also collected and determined by HPLC-MS after being diluted for 1000 times. Then a total of 10 μl was injected onto a Hypersil Gold C18 column (150 × 2.1 mm, 5 μm, Thermo) with a 5 mM ammonium acetate (pH 3.5)/methanol mobile phase starting at 10% methanol. At a flow rate of 250 μL/min, the gradient increased to 95% methanol at 13 min before reverting to original conditions at 16 min. Column temperature was maintained at 25°C. For quantitative determination, the HPLC system was interfaced to a TSQ Quantum Access (Thermo) mass spectrometer employing electron spray ionization in the negative ion mode and highly selective reaction monitoring. The fragment ions for PFOS m/z 499 (C₈F₁₇SO₃^-) were monitored for quantification.

In order to determine the interactions of PFOS with SA or DNA, B-R buffer, 0.15 mol/l NaCl, a known volume of SA or DNA solution (c_M0), deionized water and a known volume of PFOS solution (c_L0) were added in dialysis bags (1). The temperature (2) of the water bath (4) was kept constant at 37°C. Using the same method, 1 ml of the dialysis solution (3) was collected in 5-ml calibrated flasks from every sampling tube (6). According to the detection method above, the absorbance of the dialysis solution was measured at 626 nm and c_L of PFOS was calculated by the working curve. γ of PFOS to SA or DNA was calculated and data were plotted.

The B-R buffer (2.00 ml) and 0.050 ml of 0.100 mmol/l SA were mixed with 0, 2.5, 5, 7.5, 10, 12.5, 15, 17.5 μmol/l of PFOS in 10-ml calibrated flasks. The solutions were diluted to 10 ml with deionized water and their fluorescence intensities were measured at the excitation wavelength (280 nm) and the emission wavelengths (300–450 nm).

The B-R Buffer (1 ml), 0.04 ml SA (0.100 mmol/l) were mixed with 0, 16, 20, 24, 40 μmol/l PFOS in flasks. The solutions were diluted to 10.0 ml with deionized water. Each sample was allowed to equilibrate for 15 min before measurement. CD spectra were taken on a spectropolarimeter with a 0.1 cm light path cell at 25°C. The mean residue ellipticity (θ) of SA was measured between 190 and 240 nm. From the θ curves, the relative contents of secondary structure forms of SA, α-helix, β-pleated sheet, β-turn and random coil, were calculated in all solutions. The 100 μmol/l DNA containing 0, 20, 30 and 40 μmol/l PFOS were measured between 190 and 240 nm using the same method.

ITC experiments were carried out as follows: the PFOS solution (2.50 mmol/l in pH 7.40 B-R buffer) was injected about 27 times in 10-μl increments at 270-S intervals into the isothermal cell containing SA (0.010 mmol/l in pH 7.40 B-R buffer) or DNA (0.500 mmol/l in pH 7.40 B-R buffer). The cell temperature was kept at 37°C. Heats of dilution of PFOS, obtained separately by injecting PFOS into the buffer, were used to correct the raw data. The corrected heats were divided by the number of moles injected and analyzed using the Origin software (version 7.0) supplied by the manufacturer. The titration curve was fitted by a nonlinear least-squares method and ΔH and ΔS were determined.

The fluorescence intensity of Vitamin B₂ (VB₂) solution was measured at the excitation wavelength (440 nm) and the emission wavelength (525 nm) against deionized water using fluorospectrophotometer. The determination of the VB₂-SA interaction was investigated according to the same equilibrium dialysis method as described above.

For the equilibrium dialysis assay of PFOS-VB₂-SA, 12.5 ml solution containing 2.5 ml of B-R buffer, 0.15 mmol/l NaCl, 0.02 mmol/l SA (c_M0), 0.041 mmol/l VB₂, a known volume of PFOS solution (c_L0) and deionized water were pipeted into the dialysis bag (Fig. 8). After 10 hours at 37°C, 2.5 ml of the dialysis solution was collected from the sampling tube and the concentration of VB₂ was determined. The binding ratio of VB₂ to SA was calculated to investigate the effect of PFOS on the transport function of SA.