1.

Introduction

Two-photon fluorescence lifetime imaging (FLIM) of endogenous fluorophores can non-invasively monitor metabolic activity on a cellular level.¹ Endogenous fluorophores related to metabolism include flavin adenine dinucleotide (FAD), reduced nicotinamide adenine dinucleotide (phosphate) (NAD(P)H), and flavin mononucleotide (FMN). FAD and FMN are fluorescent flavins, in which FAD is more prevalent in the cell.^2,3 FAD and FMN are primarily localized in the mitochondria and bound to enzymes as a cofactor, although some proteins that use FAD or FMN as a cofactor are localized to other parts of the cell.^4,5 Fluorescence spectra of NAD(P)H, FAD, and FMN have been previously published.^3,6,7 The fluorescence spectra of FAD and FMN show substantial overlap, so FMN may contribute to FAD lifetime in FLIM measurements.³ NAD(P)H and FAD each have two distinct lifetimes due to free- and protein-bound states. For NAD(P)H, the short lifetime (${\tau }_1$) corresponds to free, unbound NAD(P)H, and the long lifetime (${\tau }_2$) corresponds to protein-bound NAD(P)H.⁸ Conversely, for FAD, free FAD has a long lifetime, and protein-bound FAD has a short lifetime.⁹

Given that FLIM of endogenous fluorescence is highly sensitive to cellular changes in metabolism, it has been used in many biomedical applications, including detection and treatment of cancer, tissue metabolism, and pathologies in the skin and eye.^10–17 Fluorescence lifetimes are sensitive to a number of environmental factors, including pH, oxygen levels, and temperature.^10,11,18,19 Extracellular pH is particularly relevant, as its effects on the internal pH of the cell may result in a suboptimal pH for metabolic enzymes to function or disrupt the proton gradient within mitochondria.²⁰ Extracellular pH is typically maintained in interstitial fluid around pH 7.4.²¹ In some tissues, such as the GI system, the extracellular pH varies. In the intestinal tract, the pH may be as high as 8.0 or low as 4.0, and as low as 1.0 in the stomach.²² The extracellular pH is altered in common pathologies such as COPD, renal failure, and ischemia. The extracellular environment in tumors may also be more acidic (approximately pH 6.2 to 6.9) than healthy tissues.^23–25 Studies in tumors have demonstrated that pH changes inside and outside the tumor are a major factor in promotion of tumor growth and metastasis.^23,26

Several studies have investigated how pH affects the lifetime of various endogenous fluorophores. NAD(P)H mean lifetime in cells decreased with increased extracellular pH.^19,20 Other studies focused on intracellular pH, using nigericin to equalize intracellular and extracellular pH across the cell membrane. These studies found NAD(P)H and FAD mean lifetimes decreased with increased intracellular pH.^27,28 However, using nigericin/K+ to equalize pH across the membrane presents a confounding factor when used in conjunction with FLIM of metabolic cofactors. Nigericin decreases ATP concentrations, lowers the rate of protein synthesis, and affects the intracellular levels of other molecules, such as lactate.^29–31 It is unclear whether the effects of nigericin are due to the presence of the molecule itself, or due to the changes in ion concentration and pH that result. At least, one study suggests that nigericin affects metabolites independently of its effects on ion concentrations.²⁹ When nigericin/K+ equilibration is used with FLIM of NAD(P)H and FAD, molecules highly sensitive to metabolic changes, it may affect the results of such experiments. Since several studies have used nigericin to control intracellular pH, the effects of pH, independent of nigericin, on the fluorescence lifetimes of NAD(P)H and FAD remain ambiguous.

In addition, manipulation of extracellular pH often occurs in cell culture, as well as in other contexts, during FLIM experiments. For example, many cell media rely on ${\mathrm{CO}}_2$-dependent buffers to maintain media pH. If ${\mathrm{CO}}_2$ level is not maintained throughout the experiment, the buffer may no longer keep the extracellular pH consistent. Understanding the effects of extracellular pH on endogenous fluorescence lifetimes could improve the accuracy and reproducibility of in vitro and in vivo studies. Studies of extracellular pH effects on FAD lifetime are absent from the existing literature. Additionally, few studies have investigated pH effects on coenzyme lifetimes in more than one cell line, limiting our understanding of how extracellular pH may influence endogenous fluorescence lifetimes in different cell types.¹⁹

To further investigate the effects of extracellular pH on the fluorescence lifetimes of FAD, FMN, and NAD(P)H, human breast cancer (BT474) and HeLa cells were placed in pH adjusted media and imaged using two-photon FLIM. Additionally, 5-(and-6)-carboxy SNARF-1, a fluorescent intracellular pH indicator, was added to cells in a separate experiment to study the relationship between extracellular and intracellular pH changes. To investigate how these lifetime changes due to pH relate to the chemical properties of flavins, FAD (free and bound to cholesterol oxidase) and FMN solutions were prepared at varying pHs and imaged using FLIM.

2.

Methods

3.

Results

Representative images of the SNARF1 intensity ratio in calibration condition (nigericin treated) HeLa cells demonstrated a qualitative increase in the intensity ratio as the pH increased from pH 4 to pH 9 [Fig. 1(a)]. This was further supported by the quantitative calibration data, showing a sigmoidal relationship between the intensity ratio of SNARF-1 and the intracellular pH in both cell lines [Figs. 1(b) and 1(c)]. Using these calibration curves to calculate the intracellular pH in the experimental condition (no nigericin), a strong positive correlation between extracellular pH and intracellular pH emerged, present in both HeLa and BT474 cells [Figs. 1(d) and 1(e)]. The Pearson correlation coefficient ($R$) for this relationship was 0.8921 in BT474 cells and 0.9789 in HeLa cells.

Fig. 1

SNARF-1 measurement of intracellular pH. (a) Representative images of the intensity ratio of SNARF-1 fluorescent dye in BT474 cells. Intensity ratio is calculated on an image level and is equal to the intensity of the pH-dependent SNARF-1 intensity at $650/45\mathrm{nm}$ bandpass divided by the isoemissive SNARF-1 intensity at 575 nm low pass. (b) Calibration curve in BT474 (blue) and HeLa (red) cells, fit to a sigmoidal curve ($n=65$ images, $n=25$ images, respectively; each dot corresponds to one image). (c) Intracellular pH versus extracellular pH in BT474 (blue) and HeLa (red) cells, $\text{mean}\pm 95\%$ confidence intervals ($n=48$ images, $R=0.8785$ and $n=30$ images, $R=0.9789$, respectively). ***$P<0.001$. $R$-values correspond to Pearson correlation coefficients.

The SNARF-1 data demonstrated the existence of a strong positive correlation between the extracellular and intracellular pH in BT474 and HeLa cells. Therefore, altering environmental pH does affect the intracellular pH. Given the pH dependence of many enzymes and metabolic processes, this is especially relevant for studies of metabolism. The relationship between intracellular and extracellular pH characterized in this assay is consistent with other studies investigating the same relationship.^20,39,40

Representative images of NAD(P)H ${\tau }_m$ and FAD ${\tau }_m$ in HeLa cells qualitatively demonstrated an increase in FAD ${\tau }_m$ with increasing extracellular pH, and higher NAD(P)H ${\tau }_m$ at pH 7 compared to pH 4 and pH 9 [Fig. 2(a)]. The FAD ${\tau }_m$ trends were also demonstrated in the quantitative data, which showed a moderate linear correlation between extracellular pH and FAD ${\tau }_m$ in both BT474 and HeLa cells ($R=0.6520$ and $R=0.4914$ in BT474 and HeLa cells, respectively) [Fig. 2(b)]. Conversely, the overall linear correlation between NAD(P)H ${\tau }_m$ and extracellular pH was weakly positive in HeLa cells and moderately negative in BT474 cells [Fig. 2(c)]. However, with the pH 4 group excluded, NAD(P)H ${\tau }_m$ showed a moderately negative linear correlation with extracellular pH ($R=-0.7627$ and $R=-0.5655$ in BT474 and HeLa cells, respectively). In addition, NAD(P)H ${\tau }_m$ increased from pH 4 to pH 5 in both BT474 and HeLa cells.

Fig. 2

FLIM of BT474 and HeLa cells vary with extracellular pH. (a) Representative images of NAD(P)H ${\tau }_m$ and FAD ${\tau }_m$ in HeLa cells at pH 4.0, pH 7.0, and pH 9.0 in pseudocolor. (b) Box-and-whisker plot of FAD ${\tau }_m$ versus extracellular pH. Box is centered at median and reaches to first and third quartiles. Whiskers reach to 1.5 × interquartile range with dots as outliers beyond this range. (c) Box-and-whisker plot of NAD(P)H ${\tau }_m$ versus extracellular pH. (d) Box-and-whisker plot of FAD ${\alpha }_1$ versus extracellular pH. (e) Box-and-whisker plot of NAD(P)H ${\alpha }_1$ versus extracellular pH. $\text{Blue}=\mathrm{BT}474$, $\text{red}=\mathrm{HeLa}$. (f) Heatmap of Pearson correlations, measuring the linear correlation with pH for all metabolic parameters in both cell lines, for the extracellular pH range 4 to 9; all $R$-values are significant at $p<0.05$. $n=4831$ cells for BT474, $n=3870$ cells for HeLa. $R$ values correspond to Pearson correlation coefficients. ***$P<0.001$.

We then investigated which parameters contributed to the mean lifetime trends. In these results, ${\alpha }_1$ and ${\tau }_1$ correspond to free NAD(P)H and protein-bound FAD. Increases in FAD ${\tau }_m$ with increasing extracellular pH were driven primarily by a decrease in FAD ${\alpha }_1$ with increasing extracellular pH [Fig. 2(d)]. The trend in NAD(P)H ${\tau }_m$ appears to have been driven by different components between the two cell types. In BT474 cells, there was a moderate linear correlation in NAD(P)H ${\alpha }_1$ with increasing extracellular pH [$R=0.6475$, Fig. 2(e)]. These results indicate an increase in the amount of free FAD and NAD(P)H [due to increased FAD ${\alpha }_2$ and NAD(P)H ${\alpha }_1$] when extracellular pH increases in BT474 cells. The linear correlation between NAD(P)H ${\tau }_m$ and extracellular pH was weaker in HeLa cells [$R=0.1821$, Fig. 2(e)]. In contrast, NAD(P)H ${\tau }_m$ in HeLa cells was more strongly driven by decreased NAD(P)H ${\tau }_1$ with increasing extracellular pH [$R=0.5960$, Fig. 2(f)]. FAD ${\tau }_1$, FAD ${\tau }_2$, and NAD(P)H ${\tau }_2$ had moderate to weak linear correlations with extracellular pH in both cell lines [Fig. 2(f)]. The redox ratio in HeLa cells demonstrated a strong linear correlation with extracellular pH ($R=0.8059$), but the correlation was much weaker in BT474 cells [$R=0.2091$, Fig. 2(f)].

These data suggested that extracellular pH affects the lifetimes of both FAD and NAD(P)H in the cells. FAD ${\tau }_m$ linearly decreases between extracellular pH 4 to 9. NAD(P)H ${\tau }_m$ linearly decreases between extracellular pH 5 to 9, with pH 4 breaking this trend. Interestingly, the trends and the strength of correlations between fluorescence lifetime variables and extracellular pH were not the same in both cell lines, with differences particularly evident in the NAD(P)H lifetime variables and the optical redox ratio [Fig. 2(f)]. This suggests that the extracellular pH may not affect the metabolism of every cell line in the same way or to the same degree.

In the FAD solutions [Fig. 3(a)], the magnitude of changes in FAD lifetime was small from pH 5 to 9, with a range of 2414 to 2534 ps. The FAD lifetime decreased with increasing solution pH [$R=-0.8375$, Fig. 3(a)], and the FAD lifetime in solution was significantly different from the pH 7 control at pH 5 and pH 9.2. FMN lifetime in solution [Fig. 3(b)] also exhibited only small changes, with a range of 3591 to 3704 ps over pH 5 to 9. Significant changes in FMN lifetime, compared to the control group at pH 7, occurred only in the pH 5 and pH 9 groups [Fig. 3(b)]. Cholesterol oxidase solutions containing FAD were fit to two components to account for both bound and free FAD. The proportion of free, unbound FAD in solution was $33.26\%\pm 5.19\%$. Both FAD ${\tau }_1$ (bound FAD) and FAD ${\tau }_2$ (free FAD) were significantly different from the control pH 7 only at pH 5 [Figs. 3(c) to 3(d)].

Fig. 3

Fluorescence lifetimes of FAD and FMN solutions. (a) Fluorescence lifetimes of FAD solutions versus solution pH, $\text{mean}\pm 95\%$ confidence intervals (linear fit is included with $R=-0.8375$, $n=36$ images). (b) Fluorescence lifetimes of FMN solutions versus solution pH, $\text{mean}\pm 95\%$ confidence intervals ($n=56$ images). (c) Short fluorescence lifetime of FAD in solution with cholesterol oxidase versus solution pH, $\text{mean}\pm 95\%$ confidence intervals ($n=37$ images). (d) Long fluorescence lifetime of FAD in solution with cholesterol oxidase versus solution pH, $\text{mean}\pm 95\%$ confidence intervals ($n=37$ images). FAD and FMN solutions in (a) and (b) are fit to one component, FAD-cholesterol oxidase solutions in (c) and (d) are fit to two components. Data presented as $\text{mean}\pm 95\%$ confidence intervals, $p$ values compare group to control at pH 7.0. **** $P<0.0001$, *** $P<0.001$, * $P<0.05$.

FAD ${\tau }_2$ trends in the cholesterol oxidase solutions differed from the FAD lifetime in solutions of FAD alone, but this difference may be attributed to the presence of cholesterol oxidase and its pH dependence in addition to pH-influenced changes in FAD itself. Cholesterol oxidase activity has been shown to vary with changes in pH.⁴¹ These changes in FAD and FMN lifetimes in solutions over pH 5 to 9 (Fig. 3) do not account for the observed changes in vitro in HeLa and BT474 cells (Fig. 2). This indicates that changes in flavin lifetimes observed in vitro are likely due to metabolic changes within the cell in response to pH rather than local changes in pH alone.

4.

Conclusions

The results of this study demonstrate that alterations in extracellular pH result in changes to NAD(P)H and FAD fluorescence lifetimes. The strength of the correlations between endogenous fluorescence lifetime variables and extracellular pH were not conserved across cell lines, with differences being particularly evident in the NAD(P)H lifetime variables and the optical redox ratio. Studies of free FAD, bound FAD, and FMN solutions indicated that changes in cellular FAD lifetimes with pH are likely due to metabolic and/or enzymatic changes within the cell, rather than microenvironmental changes in pH around the fluorophore alone. This is consistent with changes in enzyme function that are known to occur with changes in cytosolic and mitochondrial pH.²⁰ These results suggest that the relationship between extracellular pH and endogenous fluorescence lifetimes differ between cell lines, which could provide a method of cell phenotyping. This relationship between extracellular pH and endogenous fluorescence lifetime could also be applicable to quality control of cell conditions during FLIM imaging and to study metabolic changes in cells due to extracellular pH.