Empirical equation for the correction of fluorescence quenching of proteinaceous substance by Suwannee river natural organic matter

Fluorescence quenching of proteinaceous substances by natural organic matter is a well-known phenomenon, but there are no knownmethods for correcting it. The main objective of this research was to develop an empirical equation to correct the fluorescence quenching of different concentrations of bovine serum albumin (BSA – 0.15, 0.25, 0.5, 0.75, 1, 1.25 μmol/L (μM)) by Suwannee river natural organic matter (SWNOM 0,2,4,6,8,10 mg-C/L) using the fluorescence titration method. The excitation emission matrix (EEM) data were analyzed by parallel factor analysis with inner filter effect removal. With increasing SWNOM concentration, BSA peak intensity quenching was in the range 29–85%, with a linear relationship for increment of either BSA or SWNOM concentration. A higher ratio of SWNOM to BSA resulted in greater BSA peak intensity quenching. The unquenched BSA peak (BSA (RU)) is given by the empirical equation. BSA (RU) 1⁄4 {2:052 peak C (RU)þ 2:522} quenched peak T (RU) The calculated unquenched BSA peak intensities using the empirical equation agreed well with the actual unquenched peak values (R1⁄4 0.98, mean absolute error1⁄4 0.33 RU). The equation is expected to help in rapid estimation of the quenching effect of SWNOM on BSA.


INTRODUCTION
Fluorescence quenching (quenching) is one of the commonest phenomena in fluorescence analysis, whereby any component (quencher) present in a system or related physico-chemical process reduces the intensity of fluorescence of a sample (fluorophore) (Lakowicz 1983;Eftink 1991). This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/). Quenching mechanisms include excited-state reactions, molecular rearrangement, energy transfer, ground-state complex formation, and collisional quenching. Collisional quenching can be further divided into static and dynamic quenching. Although there are numerous studies on quenching mechanisms (Yuan et al. 1998;London & Ladokhin 2002;Soares et al. 2007;Keppler et al. 2014;Joye et al. 2015;Wang et al. 2015;Heo et al. 2016;Ali & Al-Lohedan 2018), an empirical method for quenching correction to enable calculation of the actual concentration of proteinaceous fluorophore substances quenched by humic substances in a sample matrix is still lacking. In earlier studies, the post-chlorination fluorescence for trihalomethane formation potential was found to be quenched by humic substances, leading to reduced accuracy in predicting trihalomethane formation in water samples (Saipetch & Yoshimura 2019). In another experiment, multi-spectroscopic studies revealed static quenching between humic and protein-like substances arising from rearrangement of the secondary polypeptide carbonyl hydrogen bonding network of BSA (Saipetch et al. submitted).
In water sources, the amount of each component in dissolved organic matter (DOM) varies widely (Carstea et al. 2016;Singh et al. 2017). The nature of the various DOMs affects the extent of fluorescence quenching, but no previous study has been designed to investigate the effect of DOM concentration on the level of quenching (Wang et al. 2015;Lin et al. 2017). Saipetch et al. (submitted) found that the protein fluorescence of bovine serum albumin (BSA) is quenched by interaction with Suwanee River natural organic matter (SWNOM), indicating the possibility that the concentration of each of BSA and SWNOM can influence the quenching uniquely. No study has yet demonstrated the correction accuracy of quenching by BSA and SWNOM using empirical methods, however.
The relationships between BSA and SWNOM concentrations on quenching were investigated in this study, leading to the proposal and verification of a method for correcting quenching.

MATERIALS AND METHODS
Fluorescence titration was carried out between varying concentrations of SWNOM (0, 2, 4, 6, 8, 10 mg-C/L), and BSA (0.15, 0.25, 0.5, 0.75, 1, 1.25 μmol/L (μM)) was performed. The concentrations of SWNOM and BSA were equivalent to NOM concentrations found in surface water samples. The excitation emission matrix (EEM) data were analyzed using parallel factor analysis with inner filter effect removal (drEEMtoolbox in MATLAB -Murphy et al. 2013). The effect of the BSA and/or SWNOM concentration on BSA's quenched fluorescence intensity was investigated using nonlinear regression, on the basis of the hypothesis that the fluorescence intensity of BSA (unquenched BSA fluorescence, peak T location (Figure 1   intensity (Figure 2(a2)). The quenching effect of increases in SWNOM concentration can also be seen in the BSA peak intensity (Figure 2(a2)). There was a linear relationship between SWNOM concentration and the fluorescence intensity of peaks A (R 2 ¼ 0.9987) and C (R 2 ¼ 0.9991) (Figure 2(b1)), and peak T's fluorescence intensity (R 2 ¼ 0.9886) with BSA concentration (Figure 2(b2)).
The fluorescence intensities of BSA at peak T, in the absence of SWNOM, and of SWNOM at peak C were used to represent the concentrations of BSA and SWNOM due to their linear relationship (Figure 2(b1) and 2(b2)). As the SWNOM concentration increased, the quenching of BSA's peak T was between 29 and 85% (Figure 2(c)), and the relationship was linear for every concentration increment of either BSA or SWNOM. Hence, the hypothesis that BSA's fluorescence intensity in the absence of SWNOM is a function of the intensities of quenched fluorescence of BSA in the presence of SWNOM and the fluorescence intensity of SWNOM alone can be said to be true.

Empirical equation for the quantitative impact of SWNOM concentration on BSA fluorescence intensity
Higher ratios of quencher (SWNOM) to proteinaceous fluorophore (BSA) resulted in greater quenching of the intensity of BSA's peak T. The effects of BSA and SWNOM concentration on the quenched and unquenched intensities of BSA's peak T are shown in Figure 3. There were linear relationships between BSA peak intensity and the concentrations of either BSA or a BSA-SWNOM mixture (Figure 3(b)). At the same initial concentration of BSA, an increase in SWNOM concentration lowered the fluorescence intensity of peak T linearly (Figure 3). At fixed SWNOM concentration, however, an increase in BSA concentration reduced the quenching of BSA's peak T. The relationship between the quenched intensity of peak T, and the concentration of either BSA or the BSA-SWNOM mixture, can be explained by the power function Equation (1)  (1)

Correction of fluorescence quenching by SWNOM
In order to develop the correction method, the term α was defined to represent the ratio between BSA's unquenched and quenched peak Ts, and assumed to be a function of quenched peak T intensity and peak A intensity (Equation (2)). The plots derived using Equation (2) are shown in Figure 4(a). which shows clearly that α is a power function of quenched peak T. The values of the constants in  Equation (2) were derived by selecting the empirical values of the slope (a) and constant (b) from Figure 4(a) - Table 1. It is to be noted that peak C is the direct representative of SWNOM concentration, as the intensity of SWNOM's peak C is not quenched by BSA addition.
where a and b are constants.
Values of both the empirical constants 'a' and 'b' changed with an increase in peak C's relative intensity (Table 1). Coefficient 'a' was a linear function of peak C, whereas 'b' was a third-order of a polynomial function. Application of a polynomial function of peak C for 'b' was considered unnecessary as it produced no improvement in R 2 between the modelled and actual fluorescence intensities of unquenched peak T compared to the use of an average value for 'a'. The calculated intensities of unquenched peak T also agreed well with its actual intensities (R 2 ¼ 0.98 and mean absolute error 0.33 RU (Figure 4(b))).
At lower intensities of unquenched peak T fluorescence (1.4-5.7 RU), the variation in modelled unquenched peak T was lower than that at unquenched intensities of 7.7-10.7 RU. The fluorescence intensities in the lower range are comparable to the protein fluorescence intensity of clean water (Tama River, Japan), while intensities higher than that can be considered to arise from increases in organic quencher concentrations from wastewater.
Use of Equation (3) is expected to improve the quality of estimates of the unquenched intensity of peak T in surface waters. The proposed empirical correction equation (Equation (3)) is also expected Figure 4 | (a) Relationship between unquenched peak T/quenched peak T vs. quenched peak at different peak C levels, (b) modelled vs. actual unquenched peak T fluorescence intensity. to be applicable to various types of protein-related compound other than BSA, as BSA is the standard used to represent substances like algae cell proteins (Hong et al. 2009;Wei et al. 2011), cyanobacterium (Microcystis aeruginosa) (Pivokonsky et al. 2015) and membrane foulants (Guan et al. 2018).

CONCLUSIONS
The impact of natural organic matter (SWNOM) concentrations on the fluorescence quenching of a proteinaceous substance (BSA) was studied using fluorescence titration. Increasing the concentration of SWNOM increased its quenching effect on BSA fluorescence intensity, and the magnitude of the quenching was proportional to the SWNOM/BSA ratio in solution. At fixed BSA concentrations, a linear increase in SWNOM concentration lowered the intensity of BSA's peak T fluorescence, while, at fixed SWNOM concentration, an increase in BSA concentration reduced the fluorescence quenching level in line with a power function. When used to correct for fluorescence quenching, Equation (3) fitted well with the actual unquenched and modelled intensities of peak T (Figure 4).