This protocol describes fluorescence imaging and analysis of the endogenous metabolic coenzymes, reduced nicotinamide adenine (phosphate) dinucleotide (NAD(P)H), and oxidized flavin adenine dinucleotide (FAD). Autofluorescence imaging of NAD(P)H and FAD provides a label-free, nondestructive method to assess cellular metabolism.
Cellular metabolism is the process by which cells generate energy, and many diseases, including cancer, are characterized by abnormal metabolism. Reduced nicotinamide adenine (phosphate) dinucleotide (NAD(P)H) and oxidized flavin adenine dinucleotide (FAD) are coenzymes of metabolic reactions. NAD(P)H and FAD exhibit autofluorescence and can be spectrally isolated by excitation and emission wavelengths. Both coenzymes, NAD(P)H and FAD, can exist in either a free or protein-bound configuration, each of which has a distinct fluorescence lifetime-the time for which the fluorophore remains in the excited state. Fluorescence lifetime imaging (FLIM) allows quantification of the fluorescence intensity and lifetimes of NAD(P)H and FAD for label-free analysis of cellular metabolism. Fluorescence intensity and lifetime microscopes can be optimized for imaging NAD(P)H and FAD by selecting the appropriate excitation and emission wavelengths. Metabolic perturbations by cyanide verify autofluorescence imaging protocols to detect metabolic changes within cells. This article will demonstrate the technique of autofluorescence imaging of NAD(P)H and FAD for measuring cellular metabolism.
Metabolism is the cellular process of producing energy. Cellular metabolism encompasses multiple pathways, including glycolysis, oxidative phosphorylation, and glutaminolysis. Healthy cells use these metabolic pathways to generate energy for proliferation and function, such as the production of cytokines by immune cells. Many diseases, including metabolic disorders, cancer, and neurodegeneration, are characterized by altered cellular metabolism1. For example, some cancer cell types have elevated rates of glycolysis, even in the presence of oxygen, to generate molecules for the synthesis of nucleic acids, proteins, and lipids2,3. This phenomenon, known as the Warburg effect, is a hallmark of many cancer types, including breast cancer, lung cancer, and glioblastomas4. Because of the alterations of cellular metabolism associated with cancer progression, cellular metabolism can be a surrogate biomarker for drug response5,6. Moreover, understanding drug efficacy at a cellular level is crucial as cell heterogeneity can lead to differing drug responses in individuals7,8.
Technologies that identify and quantify changes in cellular metabolism are essential for studies of cancer and drug response. Chemical and protein analyses are used to evaluate the metabolism of cells or tissues but lack single-cell resolution and spatial information. Metabolic plate reader-based assays can measure pH and oxygen consumption in the sample over time and the subsequent metabolic perturbation by chemicals. The pH can be used to calculate the extracellular acidification rate (ECAR), which provides an insight into the glycolytic activity of the cells9. Whole-body imaging methods, including 2-[fluorine-18] fluoro-D-glucose positron emission tomography (FDG PET) and magnetic resonance spectroscopy (MRS), are noninvasive imaging modalities used clinically to identify tumor recurrence and drug efficacy through metabolic measurements10,11,12,13,14.
FDG-PET images the tissue uptake of FDG, a radiolabeled glucose analog. Increased uptake of FDG-PET by tumors relative to surrounding tissue is due to the Warburg effect12,13. MRS images common nuclei of molecules used for metabolism, such as 13C and 31P, and can obtain dynamic information about how metabolism changes in response to stimuli, such as exercise or eating14. Although FDG-PET and MRS can be used clinically, these technologies lack the spatial resolution to resolve intratumoral heterogeneity. Likewise, oxygen consumption measurements are made on a bulk population of cells. Autofluorescence imaging overcomes the spatial resolution obstacle of these technologies and provides a noninvasive method of quantifying cellular metabolism.
Figure 1: NADH and FAD in common metabolic pathways. NADH and FAD are coenzymes used in glycolysis, the Krebs cycle, and the electron transport chain. Autofluorescence imaging of these molecules provides information about cellular metabolism. Please click here to view a larger version of this figure.
Reduced nicotinamide adenine (phosphate) dinucleotide (NAD(P)H) and oxidized flavin adenine dinucleotide (FAD) are coenzymes of metabolic reactions, including glycolysis, oxidative phosphorylation, and glutaminolysis (Figure 1). Both NAD(P)H and FAD are autofluorescent and provide endogenous contrast for fluorescence imaging1,15. NADPH has similar fluorescent properties to NADH. Because of this, NAD(P)H is often used to represent the combined signal of NADH and NADPH2,16.
Fluorescence lifetime imaging (FLIM) quantifies the fluorescence lifetime or the time for which a fluorophore is in the excited state. Fluorescence lifetimes are responsive to the microenvironment of the fluorophores and provide information about cellular metabolism17. NAD(P)H and FAD can exist within cells in either protein-bound or free conformations, each of which has a different lifetime. Free NAD(P)H has a shorter lifetime than protein-bound NAD(P)H; conversely, free FAD has a longer lifetime than bound FAD18,19. The lifetimes and lifetime component weights can be quantified from fluorescence lifetime decay data through Eq. (1)20:
I(t) = α1e–t/τ1 + α2e–t/τ2 + C (1)
Eq (1) represents the normalized fluorescence intensity as a function of time. The α1 and α2 in this equation represent the proportional components of short and long lifetimes (α1+ α2=1), respectively, τ1 and τ2 represent the short and long lifetimes, respectively, and C accounts for background light7,20. The amplitude-weighted lifetime, represented here as τm, is calculated using Eq. (2).
τm= α1τ1+ α2τ2 (2)
A mean lifetime can be computed by averaging "t" over the intensity decay of the fluorophore, which for a two-exponential decay is shown by Eq. (3)17,21.
τ*m= (α1τ12+ α2τ22)/ (α1τ1+ α2τ2) (3)
A fluorescence intensity image can be computed from the lifetime image by integrating the fluorescence lifetime decay. Autofluorescence imaging is a nondestructive and label-free method that can be used to characterize the metabolism of live cells at a subcellular resolution. The optical redox ratio provides an optical analog metric of the chemical redox state of the cell and is calculated as the ratio of NAD(P)H and FAD intensities. Although the formula for calculating the optical redox ratio is not standardized22,23,24,25, it is defined here as the intensity of FAD over the combined intensities of NAD(P)H and FAD. This definition is used because the summed intensity in the denominator normalizes the metric between 0 and 1, and the expected result of the cyanide inhibition is a decrease in the redox ratio. The fluorescence lifetimes of free NAD(P)H and FAD provide insight into changes in the metabolic solvent microenvironment, including pH, temperature, proximity to oxygen, and osmolarity17.
Changes in the fluorescence lifetime of the bound fractions of NAD(P)H and FAD can indicate metabolic pathway utilization and substrate-specific metabolism26. Component weights can be interpreted for changes in the free to the bound fraction of the coenzymes18,19. Altogether, these quantitative autofluorescence lifetime metrics allow the analysis of cellular metabolism, and autofluorescence imaging has been used for identifying neoplasms from normal tissues27,28, characterizing stem cells29,30, evaluating immune cell function31,32,33,34,35, gauging neurological activity36,37,38, and understanding drug efficacy in cancer types such as breast cancer and head and neck cancer21,39,40,41,42. High-resolution autofluorescence imaging can be combined with image segmentation for single-cell analysis and quantification of intrapopulation heterogeneity43,44,45,46,47.
NAD(P)H and FAD can be imaged on single-photon or multiphoton fluorescence microscopes configured for intensity or lifetime imaging. For single-photon microscopes, NAD(P)H and FAD are typically excited at wavelengths of 375-405 nm and 488 nm, respectively, due to common laser sources at these wavelengths48. In two-photon fluorescence excitation, NAD(P)H and FAD will excite at wavelengths of approximately 700 to 750 nm and 700 to 900 nm, respectively15,49. Once the fluorophores are excited, NAD(P)H and FAD emit photons at wavelengths between ~410 nm to ~490 nm and ~510 nm to ~640 nm, respectively15. The NAD(P)H and FAD maxima emission wavelengths are approximately 450 nm and 535 nm, respectively48.
Because of their different excitation and emission wavelengths, the fluorescence of the two metabolic coenzymes can be spectrally isolated. An understanding of the spectral characteristics of NAD(P)H and FAD is necessary for the design and optimization of autofluorescence imaging protocols. Cyanide is an electron transport chain (ETC) complex IV inhibitor. The effects of cyanide on cellular metabolism and the autofluorescence intensities and lifetimes of NAD(P)H and FAD within cells are well characterized27,40. Therefore, a cyanide perturbation experiment is an effective means of validating NAD(P)H and FAD imaging protocols. A successful cyanide experiment provides confidence that the NAD(P)H and FAD imaging protocol can be used to assess the metabolism of unknown groups or perturbations.
1. Cell plating for imaging
2. Multiphoton FLIM imaging of NAD(P)H and FAD
3. Cyanide experiment preparation
4. FLIM image analysis
Figure 2: Measured IRF of urea crystal. (A) Intensity image obtained from the urea. A representative pixel was chosen to create the IRF decay curve (B) for subsequent analysis of fluorescence lifetime images of cells. Abbreviation: IRF = instrument response function. Please click here to view a larger version of this figure.
Figure 3: Identification and segmentation of individual cells. The NAD(P)H intensity image of MCF7 cells (A) obtained by integrating a fluorescence lifetime image. Cells were imaged using 750 nm excitation at 5 mW for 60 s. The x and y axes represent the pixel location of the image. (A) Individual cells were identified. The cells were masked (B) to eliminate any background noise from the data set. The nucleus was then identified (C) and projected onto the cell mask (D). The cells were then filtered (E) to remove masked areas that do not fit the size of typical cells. Scale bar = 50 µm. Please click here to view a larger version of this figure.
5. Alternative method: Fluorescence intensity imaging
The epithelial breast cancer cell line, MCF-7, was cultured in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. For fluorescence imaging, the cells were seeded at a density of 4 × 105 cells per 35 mm glass-bottom imaging dish 48 h before imaging. The cells were imaged before and after cyanide treatment using the protocols stated above. The goal of the cyanide experiment is to confirm spectral isolation of NAD(P)H and FAD fluorescence and validate the imaging system and analysis protocol for detecting metabolic changes in cells. Paired NAD(P)H and FAD fluorescence lifetime images were taken at five different locations before cyanide and five different locations after the addition of cyanide to the medium. Fluorescence lifetime parameters (optical redox ratio, NAD(P)H α1, NAD(P)H τ1, NAD(P)H τ2, NAD(P)H τm, FAD τ1, FAD τ2, FAD α1, FAD τ1, FAD τ2, and FAD τm) were calculated using the measured IRF from urea (Figure 2) and averaged across the pixels of the cytoplasm of each cell used for segmentation (Figure 3).
Multiphoton fluorescence lifetime imaging of NAD(P)H and FAD allows visualization of cell morphology and metabolism (Figure 4). The high resolution achieved with multiphoton microscopy allows for the identification of single cells. NAD(P)H and FAD are primarily located in the mitochondria and cytoplasm, while the nucleus, which lacks metabolic NAD(P)H and FAD, is dark in comparison23,53. The lifetime images provide visualization of the amplitude-weighted lifetimes of NAD(P)H and FAD throughout the cell and as a result of cyanide exposure (Figure 4).
Figure 4: Representative fluorescence lifetime images of MCF7 cells before and after cyanide treatment. (A) NAD(P)H amplitude-weighted fluorescence lifetime image and (B) FAD amplitude-weighted fluorescence lifetime image before cyanide treatment. (C) NAD(P)H amplitude-weighted fluorescence lifetime image and (D) FAD amplitude-weighted fluorescence lifetime image after cyanide treatment. The amplitude-weighted fluorescence lifetime (indicated by the color bar) measures the time for which a fluorophore, in these cases NAD(P)H and FAD, is in an excited state. NAD(P)H lifetime decreases with cyanide treatment, whereas the FAD lifetime increases after cyanide treatment. NADH signal was imaged using 750 nm excitation at 5 mW for 60 s, and the FAD signal was imaged using 890 nm excitation at 7 mW for 60 s. Images acquired with a 40x water-immersion objective, numerical aperture = 1.1. Scale bar = 50 µm. Please click here to view a larger version of this figure.
Cyanide inhibits complex IV of the electron transport chain, which inhibits oxidative phosphorylation2,15. After cyanide exposure and prior to cell death, NAD(P)H accumulates within the mitochondria, and FAD decreases1,15. Due to these well-defined changes in NAD(P)H and FAD intensity due to cyanide inhibition of metabolism, the perturbation is a standard test to verify autofluorescence imaging and analysis protocols2,54. As expected, the optical redox ratio (FAD/(FAD+NAD(P)H)) of MCF-7 cells decreased after cyanide treatment (Figure 5, p = 0.044, Welch's t-test). The optical redox ratio is not a standardized formula, yet all intensity formulas have been shown to be equivalent20. FAD/(NAD(P)H+FAD) was chosen to define the optical redox ratio because the combined sum of NAD(P)H and FAD in the denominator provides a normalized value between 0 and 120,55. The opposite effect-an increase in optical redox ratio-is expected for cyanide perturbations with optical redox ratios calculated with NAD(P)H in the numerator.
Figure 5: Optical redox ratio of MCF7 cells decrease with cyanide treatment. The boxplots show the median and the first and third quartiles calculated from the cell data. The mean is represented by the black dot symbol, and the median is represented by the black line inside the box. The gray data points overlaid on each boxplot represent the averaged value of all the cytoplasm pixels within each cell. The control group consisted of 91 cells from five different images, and the cyanide group consisted of 95 cells from five different images. p < 0.01, Welch's t-test. Please click here to view a larger version of this figure.
The amplitude-weighted NAD(P)H lifetime (τm) of MCF7 cells decreased with cyanide exposure (Figure 6A, p < 2.2 × 10-16, Welch's t-test). Both the short and long lifetimes decreased for NAD(P)H (Figure 6B,C) but increased for NAD(P)H α1 (Figure 6D). These changes in NAD(P)H fluorescence lifetimes, decrease in τm, increase in α1, and decrease in τ2 matched published values of cyanide perturbations40,56. The decrease in NAD(P)H amplitude-weighted fluorescence lifetime with cyanide exposure indicates increased quenching within the microenvironment of NAD(P)H. An increase in α1 indicates more free NAD(P)H, as expected from the increase of NAD(P)H due to the effects of cyanide on cellular metabolism21.
Figure 6: NAD(P)H fluorescence lifetime of MCF7 cells before and after cyanide treatment. The boxplots in panels A–D show the median and the first and third quartiles calculated from the cell-level data. The mean is represented by the black dot symbol, and the median is represented by the black line inside the box. The gray data points overlaid on each boxplot represent the averaged value of all the cytoplasm pixels within a cell. The control group consisted of 91 cells from five different images, and the cyanide group consisted of 95 cells from five different images. (A–D) exhibit the changes in NAD(P)H amplitude-weighted lifetime (τm), NAD(P)H short lifetime (τ1), NAD(P)H long lifetime (τ2), and NAD(P)H-proportional component of the short lifetime (α1) due to cyanide treatment. p-values were calculated using Welch's t-test. Please click here to view a larger version of this figure.
The amplitude-weighted FAD lifetime (τm) of MCF7 cells increased after cyanide exposure (Figure 7A, p = 3.688 × 10-12, Welch's t-test). Both the short and long lifetimes increased for FAD (Figure 7B,C) but decreased for FAD α1 (Figure 7D). Changes in FAD fluorescence lifetimes, an increase in τm and τ2, and a decrease in α1 are consistent with published FAD fluorescence lifetime data of cyanide perturbation57. The change in lifetime values and α1 suggest metabolic changes within the cells, including an increased amount of free FAD21.
Figure 7: FAD fluorescence lifetime before and after cyanide treatment. The boxplots in A–D show the median, the first and third quartiles, and mean calculated from the cell-level data. The mean is represented by the black dot symbol, and the median is represented by the black line inside the box. The gray data points overlaid on each boxplot represent the averaged value of all the cytoplasm pixels within a cell. The control group consisted of 91 cells from five different images, and the cyanide group consisted of 95 cells from five different images. (A–D) exhibit the changes in FAD amplitude-weighted lifetime (τm), FAD short lifetime (τ1), FAD long lifetime (τ2), and FAD-proportional component of the short lifetime (α1) from cyanide treatment. p-values were calculated using Welch's t-test. Please click here to view a larger version of this figure.
Autofluorescence intensity and lifetime imaging have been widely used to assess metabolism in cells21,55. FLIM is high resolution and therefore resolves single cells, which is important for cancer studies because cellular heterogeneity contributes to tumor aggression and drug resistance7,39,41,44,45,46,58. Likewise, autofluorescence imaging of cellular metabolism is useful for imaging immune cells as immune cell function is linked to cellular metabolism, and immune cell populations are often heterogeneous20,31. Standard biochemical assays typically evaluate immune cells at the population level or require intracellular labeling following cell permeabilization34,59,60. Autofluorescence imaging is also well-suited for high-resolution in vivo and dynamic measurements of metabolism due to the non-destructive nature of light and lack of chemical or genetically encoded labels36,37,38,41,48,55. Critical steps for imaging NAD(P)H and FAD fluorescence intensity and lifetime include the selection of appropriate wavelengths for excitation and emission, verification that the cells do not contain synthetic or additional endogenous fluorophores that will contribute overlapping fluorescence, and the use of nondamaging laser powers.
Autofluorescence imaging of NAD(P)H and FAD fills a unique niche as a label-free, high-resolution, quantitative metabolic imaging technology. Other imaging methods, such as FDG-PET and MRS, image tissue metabolism but lack cellular level resolution and thus cannot evaluate cellular heterogeneity. Other biochemical techniques, such as oxygen consumption assays, measurements of metabolites in the media, or protein analysis, require expensive single-use reagents, obtain measurements from pooled cells, and require cell-destructive protocols, preventing time-course studies or in vivo analysis61,62.
While autofluorescence imaging of NAD(P)H and FAD provides high-resolution images in a label-free and nondestructive manner, some limitations of autofluorescence imaging must be considered when designing and interpreting experiments. FLIM requires specialized and expensive equipment that is not widely available. The FLIM excitation requires a pulsed-excitation source with picosecond or femtosecond pulses at a repetition rate between 40 and 100 MHz with output power > 50 mW63. Additionally, image acquisition is relatively slow, with a trade-off between the number of pixels or image resolution and image acquisition time. The parameters recommended in this protocol of 256 x 256 pixels and 60 s integration time provide an image with reasonable resolution within about 1 mi. The user can choose to image smaller areas with fewer pixels or perform line scans to improve the imaging speed.
Alternatively, higher-pixel images can be acquired with increased image integration times. The data analysis and interpretation of autofluorescence lifetime images can be challenging as FLIM provides specific biophysical information about the metabolic coenzymes and their molecular environment rather than specific metabolic pathways. Optical microscopy is also limited by the depth of light penetration in tissue due to light scattering and absorption. In vivo studies can be done, at depths of ~0.5 mm with multiphoton imaging systems, of surface tissues or through window chambers2,20,21,64,65.
While NAD(P)H and FAD are the primary endogenous fluorophores in isolated cells, additional molecules, including collagen and elastin, can contribute autofluorescence signals in tissues. The high-resolution images of multiphoton microscopy allow visualization of cellular and noncellular compartments for segmentation of NAD(P)H and FAD pixels from the extracellular proteins40. Some cells contain endogenous molecules with overlapping fluorescence, such as lipofuscin, retinol, tryptophan, and melanin66. Therefore, autofluorescence images of NAD(P)H and FAD may contain background contributions from other endogenous molecules.
Likewise, while NAD(P)H and FAD can be multiplexed with exogenous fluorophores that emit at wavelengths above 600 nm, exogenous labels, such as DAPI or genetically encoded proteins such as GFP, spectrally overlap with autofluorescence imaging. The cyanide perturbation experiment described here helps verify that the excitation and emission wavelengths isolate NAD(P)H and FAD sufficiently to capture metabolic changes in cells. Other cell types can be applied to this protocol; however, the user should optimize imaging parameters, including laser power and image integration time for each cell type, and experiment to prevent photobleaching. Photobleaching can be minimized by monitoring the photon count rate or average fluorescence intensity during imaging for the duration of the lifetime scan. An increase or decrease in fluorescence intensity indicates that the laser power is too high. If laser powers induce photobleaching, the power can be reduced, and the total image acquisition increased to achieve sufficient photon collection for lifetime analysis.
Although the fluorescence lifetimes are independent of imaging parameters, including laser power and detector gain, the fluorescence intensity is dependent on these parameters20. Therefore, when conducting autofluorescence imaging to quantify the optical redox ratio, it is critical to use consistent imaging parameters. The protocol recommends that 5-6 images of different FOVs be acquired from each experimental group. This sample size provides sufficient data at both the image and cell levels to resolve the expected differences in fluorescence lifetime parameters of confluent MCF7 cells due to the cyanide perturbation. The optimal number of images acquired per group depends on the experimental design and effect size. When switching between NAD(P)H and FAD images, the focal plane should remain the same at each location for consistency. Although lifetime imaging typically does not have a saturating number of photons, intensity images acquired with cameras or photomultiplier tubes can be saturated. Pixel saturation should be avoided to ensure that the entire range of physiological intensities is captured in the images.
When analyzing fluorescence lifetime images, either a measured IRF or a simulated IRF can be used. The simulated IRF is estimated from the upshoot of the fluorescent lifetime curve; however, the real IRF obtained from the system might be broader depending on the system and thus result in more accurate lifetime values. While the IRF typically only changes with hardware or software alterations, a daily IRF measurement is a good practice to ensure the fluorescence lifetime system is working as expected.
Overall, autofluorescence imaging of NAD(P)H and FAD provides a label-free, nondestructive method to image and analyze cellular metabolism at the single-cell level. This method provides a step-by-step approach to validate the imaging of NAD(P)H and FAD using cyanide to induce a well-characterized metabolic perturbation in cells.
The authors have nothing to disclose.
Funding sources include the Cancer Prevention and Research Institute of Texas (CPRIT RP200668) and Texas A&M University. Figure 1 was created with BioRender.com.
2-deoxy-d-glucose (2-DG) | Sigma | AC111980000; AC111980010; AC111980050; AC111980250 | |
Antibiotic Antimicrobial (pen-strep) | Gibco | 15240096 | |
Cell Samples | American Type Culture Collection | N/A | MCF-7 cancer line |
CellProfiler | Broad Institute | N/A | Image analysis software |
Conical Tube | VWR | 89039-664 | 15 mL conical tube |
DMEM | ThermoFisher | 11965092 | Culture media |
FAD dichroic mirror | Semrock | FF495-Di03-25×36 | 495 nm |
FAD emission filter | Semrock | FF01-550/88-25 | 550/88 nm |
FAD excitation filter | Semrock | FF01-458/64-25 | 458/64 nm |
FBS | ThermoFisher | 16000036 | |
Fluorescence Lifetime Microscope | 3i | N/A | |
Glass bottom dish | MatTek Corp | P35G-1.0-14-C | |
Multiphoton Laser | Coherent | N/A | 2P Coherent Laser, Tunable 680 nm-1080 nm |
NAD(P)H dichroic mirror | Semrock | FF409-Di03-25×36 | 409 nm |
NAD(P)H emission filter | Semrock | FF02-447/60-25 | 447/60 nm |
NAD(P)H excitation filter | Semrock | FF01-357/44-25 | 357/44 nm |
PBS | ThermoFisher | 70011044 | |
Potassium Cyanide | Sigma-Aldrich | 380970 | |
SlideBooks 6 | 3i | N/A | Image acquisition software |
SPCImage | Becker & Hickl GmbH | N/A | Fluorescence lifetime analysis software |
Stage Top Incubator | okoLab | N/A | |
Trypsin | Biosciences | 786-262 | |
Urea | Sigma-Aldrich | U5128 | |
YG beads | Polysciences | 19096-2 | Yg microspheres (20.0 µm) |