Developing Analysis Protocols for Monitoring Intracellular Oxygenation using Fluorescence Lifetime Imaging of Myoglobin-mCherry

Abstract

Oxygen (O₂) is a critical metabolite for cellular function as it fuels aerobic cellular metabolism; further, it is a known regulator of gene expression. Monitoring oxygenation within cells and organelles can provide valuable insights into how O₂, or lack thereof, both influences and responds to cell processes. In recent years, fluorescence lifetime imaging microscopy (FLIM) has been used to track several probe-concentration-independent intracellular phenomena, such as pH, viscosity, and, in conjunction with Förster resonance energy transfer (FRET), protein-protein interactions. Here, we describe methods for synthesizing and expressing the novel FLIM-FRET intracellular O₂ probe Myoglobin-mCherry (Myo-mCherry) in cultured cell lines, as well as acquiring FLIM images using a laser scanning confocal microscope configured for two-photon excitation and a time correlated single photon counting (TCSPC) module. Finally, we provide step-by-step protocols for FLIM analysis of Myo-mCherry using the commercial software SPCImage and conversion of fluorescence lifetime values in each pixel to apparent intracellular oxygen partial pressures (pO₂).

1. Introduction

Imaging physically measurable quantities (“quantitative imaging”) within subcellular environments can reveal crucial information about biological processes [1, 2]. Fluorescence lifetime imaging (FLIM) has emerged as a reliable method for measuring physiological processes in cells and subcellular organelles with high signal-to-noise ratios (SNRs) and submicron (<< 1 μm) planar resolution when combined with confocal or multiphoton microscopy (yielding Z sectioning). Improvements in data acquisition using time-correlated single photon counting (TCSPC), have played a significant role in making FLIM more reliable and of more general use [3, 4, 5, 6]. Cell biologists have used FLIM to measure intracellular viscosity, bio-membrane heterogeneity, molecular species concentrations, redox state of cells, pH, and, in conjunction with Förster Resonance Energy Transfer (FRET), protein-protein interactions, among other parameters [7, 8, 9]. We recently published a novel, genetically encoded sensor, myoglobin-mCherry (Myo-mCherry), to monitor intracellular oxygen partial pressure (pO₂) that expands the list of measurables accessible through FLIM [10, 11]. In an effort to expand the accessibility of FLIM measurements by other researchers, we herein present a protocol for utilizing FLIM to image Myo-mCherry in transfected cells. The protocol details the use of a laser scanning microscope (LSM) configured for two-photon microscopy, but one-photon excitation is also suitable and it should yield very similar results in cultured cells, although it is not advisable for tissue slices and thicker samples. We would also like to emphasize that we make the plasmid encoding for the sensor available to the community and requests for material should be addressed to the corresponding author.

The fluorescence ‘lifetime’ for a homogenous population of $N$ excited molecules in a singlet state (S₁, Fig. 1a) is the time required for this excited state population to be reduced by a factor of $e$ as a result of fluorescence (toward S₀, Fig. 1a) [12]. Since the lifetime is a ‘state function’, its value depends on the type and conformation of the fluorophores, as well as how they interact with their chemical and physical environment [1, 5]. The fluorescence lifetime is not related to a fluorophore’s concentration across a wide range [12]: this makes it an invaluable tool for probing biological environments, which often have unknown concentrations and/or require low fluorophore concentrations (on the order of 10s −100s of nM) to preserve system integrity. For fluorophore populations with more heterogeneous / complex fluorescence decays, which are often found in biological environments, the molecular mean fluorescence lifetime, $\tau$, corresponds to the weighted arithmetic mean of the individual lifetimes for each component in the decay (Eq. 1) [4].

Fig. 1.

(a) Jablonski diagram illustrating the physical mechanism of fluorescence. A photon (or multiple lower energy photons) excites the fluorophore to a higher energy singlet state (S_{0, 0} → S_{1, 3}), where it undergoes non-radiative relaxation beyond the scope of this chapter (S_{1, 3} → S_{1, 0}). The fluorophore decays to its ground state with emission of a fluorescent photon (S_{1, 0} → S_{0, 2}). (b) Jablonski diagram illustrating the physical mechanism of FRET. After excitation and relaxation of the donor, a “virtual photon” would excite the acceptor (S_{1, 0 (D)} → S_{0, 0 (A)}), leading to acceptor fluorescence. (c) Factors that cause fluorescence decay spectra to deviate from their model. To account for the nonzero width of the excitation pulse (left) and the nonzero temporal response of the detector (middle), the excitation pulse width and the photodetector response function are convolved, generating the IRF for the imaging system (right). ((c) reproduced from Becker 2017 [4] with permission from Becker & Hickl Gmbh).

$$\tau ={\sum }_{i=1}^N{\alpha }_i{\tau }_i/{\sum }^{\text{}}{\alpha }_i$$ (1)

If two different fluorophores are in close proximity to one another (~ 2-6 nm [13]) and the emission spectrum of one overlaps with the absorption spectrum of the other, energy from the first molecule, the donor, can transfer to the second molecule, the acceptor, through a nonradiative dipole-dipole interaction called FRET [14, 15] (Fig. 1b). This energy transfer results in a reduction in both donor intensity and lifetime, as well as an increase in acceptor intensity. This reduction in donor lifetime in one way that enables FLIM to discern where donor-acceptor pairs are in close proximity within a sample based on the lifetimes of the interacting fluorophores [1, 4, 12].

In practice, however, fluorescence decay data do not appear as a sum of exponentials model due to the nonzero width of the excitation beam and the detector’s imperfect temporal response to photon detection (Fig. 1c) [4]. To account for these factors, the instrument response function (IRF) for the imaging system is measured, especially in multi-photon microscopy where detected second harmonic generation (SHG) can confound fluorescence data [4]. After determining the appropriate IRF, a sum of exponentials model would be iteratively reconvolved with the IRF and the result would be fit to the acquired fluorescence decay data.

As discussed at length in prior work [10], Myo-mCherry is a biocompatible FLIM-FRET probe that is capable of tracking pO₂ within cells. The probe combines myoglobin, an oxygen (O₂) binding protein that experiences spectral shifts between its oxygenated (oxyMyo) and deoxygenated (deoxyMyo) states [16], with the fluorescent protein mCherry. In Fig. 2a, an artistic ribbon depiction of the probe is presented. The absorption spectrum of myoglobin in each biochemical state has different overlap with the emission spectrum of mCherry. The latter acts as a donor for the dark (i.e.: virtually nonfluorescent for the excitation wavelength) acceptor myoglobin [16]. The absorption spectrum of deoxyMyo has more overlap with mCherry than when myoglobin is in its oxygenated state. This overlap yields large FRET in deoxyMyo-mCherry and as a result, shorter lifetimes are observed in the deoxygenated state of the probe. Upon oxygenation, oxyMyo-mCherry is formed and the decrease in spectral overlap yields longer lifetimes (see Fig. 2c and full explanation in Ref. 11). Given these two different populations (deoxy and oxyMyo-mCherry), and assuming a single discrete orientation factor in the FRET, a biexponential model can be used to fit this FLIM data:

$$F\left(t\right)=a_1{e}^{\raisebox{1ex}{$-t$}\!\left/ \!\raisebox{-1ex}{${\tau }_1$}\right.}+a_2{e}^{\raisebox{1ex}{$-t$}\!\left/ \!\raisebox{-1ex}{${\tau }_2$}\right.}$$ (2)

where $a_1$ and $a_2$ are pre-exponential weights used to represent the fractions of fluorophore populations with lifetimes ${\tau }_1$ and ${\tau }_2$, assuming the natural lifetime is constant [17]. Orientation is likely distributed rather than unique and mCherry itself has been reported to have two lifetimes [18] such that the ideal model must be multiexponential; however, a biexponential model appears to capture the main FRET trends in the Myo-mCherry system and changes in the energy transfer are easily monitored using the average lifetime as an indicator [16].

Fig. 2.

(a) The Myo-mCherry construct, where mCherry (PDB: 2H5Q) was coupled to the C-terminus of myoglobin (PDB: 1MBO) by using a 2 amino acid linker. (b) Theoretical energy transfer efficiency for the Myo-mCherry system as a function of distance between the two proteins. The dashed line at 4 nm highlights the average distance between the two proteins as inferred from the available structures of the individual proteins. (c) Absorption spectra of oxyMyo and deoxyMyo overlapped with the emission spectrum of mCherry, highlighting the spectral overlap region (shaded area) between the species. R₀ is the Förster radius calculated based on the actual spectral overlap integral (not shown here). (This figure was reproduced from Penjweini 2018 [10] with permission from Journal of Biomedical Optics)

In prior work [10], cells transfected with Myo-mCherry yielded shorter average lifetimes at lower external pO₂ than at higher external pO₂ across A549 cells and its mitochondrial DNA-depleted (mtDNA) counterparts, A549 ${\rho }_0$ cells. Additionally, transfected A549 cells capable of aerobic respiration yielded shorter average lifetimes at all external O₂ concentrations (excluding near-anoxia) than A549 ${\rho }_0$ cells with inhibited aerobic respiration [10]. To image Myo-mCherry using time-domain FLIM, we used a Becker & Hickl (B&H) time correlated single photon counting (TCSPC) module. A high-repetition near-infrared (NIR) pulsed laser (≈80 MHz, 780 nm) excites a sample with a focused beam swept across the sample using a laser scanning microscope configured for two-photon microscopy [1]. The arrival times of the fluorescent photons are then recorded with respect to the laser pulses and the 2D spatial coordinate of the scanner [1]. As a large number of excitation cycles is repeated over time, photons are accumulated and a spatiotemporal histogram of photon counts is built, creating a spatial map of fluorescence decay curves with the highest time resolution and best lifetime accuracy of any FLIM technique [1, 19, 20, 21]. To analyze the acquired data, we used the proprietary FLIM analysis software, SPCImage, from B&H configured for batch analysis followed by user corrections and refinement to optimize fitting parameters.

Here below we outline details on how to setup and configure FLIM measurements, record data and subsequently analyze them using an appropriate analysis pipeline suitable to extract meaningful information from recordings of the genetically encoded oxygen probe Myo-mCherry expressed in cells.

2. Materials

The plasmid containing the Myo-mCherry (or mito-Myo-mCherry, nucleus-Myo-mCherry) gene is available through our laboratory to anyone interested in using it. The expression of the gene is constitutive and driven by a CMV promoter The plasmid encodes for antibiotic resistance (Kanamycin/Neomycin). This feature can be used for amplification and propagation in E. coli, or for transfected cells selection to produce a stable cell line (topics not covered here) (see Notes 1 and 2).

2.1. Transfection of Living Cells with Myo-mCherry

1. Living cells to be transfected.

2. Cell culture media consisting of modified eagle’s medium (DMEM) with 10% non-heat inactivated fetal bovine serum (FBS, Invitrogen, Grand Island, New York) and 1% Penicillin/Streptomycin.

3. Phosphate-buffered saline (PBS)

4. Low or serum-free media such as Opti-MEM.

5. Lipofectamine® or FuGENE® transfection reagent.

6. Plasmid containing Myo-mCherry DNA.

7. 1.5 mL Eppendorf tubes as needed.

8. Welled microscope slides for cell plating and suitable for confocal microscopy (i.e., Ibidi 4-well μ-Slide, with a single-well volume of 700 μL). Make sure that they are compatible with the stage mounted O₂/CO₂ incubator.

2.2. Cell Treatment with Rotenone and Antimycin A

1. Transfected cells cultured in cell media.

2. DMSO for preparing stock solutions of drugs.

3. Antimycin A and rotenone, a mitochondrial complex III and complex I inhibitor, respectively [22, 23].

2.3. Imaging Setup

1. A laser-scanning confocal microscope configured for two-photon microscopy and equipped with an objective that can transmit IR signal, typically with a numerical aperture (NA) greater than one. For subcellular resolution, an NA above 1.2 (for example, a Leica SP5 or Zeiss LSM 510 with oil objective, NA = 1.4) is best to obtain pixel resolution smaller than subcellular organelles.

2. A femtosecond, high-repetition (40–80 MHz) laser tunable to 780 nm (for example, a mode-locked Mai Tai®, Spectra-Physics laser operated at40 MHz or 80 MHz with a peak wavelength of 780 nm) (see Notes 3 and 4).

3. Dual-band dichroic mirror that transmits the excitation beam into the objective and reflects emitted fluorescent photons consistent with the mCherry emission spectrum to the photodetector (380–740 nm, with a maximum at 610 nm [24]). This dual-band dichroic is an essential component for two-photon fluorescence microscopy (for example, an HFT KP 700/488-nm).

4. Long-pass dichroic mirror that can reflect emitted fluorescent photons downstream into the nondescanned detector (e.g., 735 nm, Thorlabs, DMLP735B).

5. Two short pass filters that remove residual signals representing scattered light from the laser (e.g. 750 nm, Thorlabs, DMSP750B; 700 nm, Edmund Optics Inc., 12.5 mm diameter, OD 2).

6. Band pass filter to transmit the mCherry emission to the detector (e.g. 641/75 nm, Semrock, BrightLine® single-band) (See Note 5).

7. A fast (low transit time spread) photomultiplier module (for example, B&H, HPM-100 or Hamamatsu, H7422P-40).

8. Photon counting card equipped with a histogramming TCSPC module that is synchronized with the laser pulses and the pixel and line clocks from the microscope (for example, a B&H SPC-150 or SPC-830).

2.4. IRF Measurements (Multiphoton Microscopy)

1. Same imaging setup described in Subheading 2.3 except:

   Short-pass filter in front of the detector that transmits SHG signal $({\lambda }_{SHG}\approx \frac{1}{2}{\lambda }_{ex.})$ to the photomultiplier (e.g. below 425 nm (for two-photon excitation at 780 nm), Edmund Optics, 25 mm diameter, OD 2).

2. A strong SHG generation source (for example, high purity urea crystals or sucrose [25]) (See Note 6).

2.5. Controlled Experimental Environment with a Range of Stable Oxygen Concentrations for Imaging and Calibration

1. A stage-mounted incubator to provide temperature control and an environment suitable for homeostasis during imaging (i.e. a TC-MI-20×46, Bioscience tools, San Diego, CA).

2. An incubator temperature controller, such as a TC-1-100-I (Bioscience Tools).

3. A gas mixing system to deliver mixtures of N₂, O₂ and 5% CO₂ inside the incubator (i.e. CO₂-O₂-MI, Bioscience Tools) and capable of delivering O₂ concentrations from normoxia to hypoxia.

4. Tanks of high purity CO₂ and N₂ to supply the gas mixing system.

5. An oxygen monitoring system, (i.e., an OxyLite Pro 2-channel bare-fiber O₂ sensor (NX-BF/O/E, Optronix Ltd., Oxford, United Kingdom) for measuring extracellular pO₂ and temperature at and above (~100 μm) the cell layer.

3. Methods

Methods are partially adapted from Refs. 10, 11. The plasmid provided by our laboratory is generally delivered in a final volume of 50 μL, with a concentration between 0.5 and 1 μg/μl. No further dilution is necessary. For transfection in cell lines (as described below), we generally advise to use plasmid solutions with a concentration between 0.5 and 1.0 μg/μL, to minimize dilution of the transfection reagents used in the following step (see Note 7).

3.1. Transfection of Living Cells with Myo-mCherry

This transfection protocol is suitable for an Ibidi 4 well μ-Slide, with a single well volume of 700 μL. The protocol can be scaled appropriately for a range of well numbers and volumes so long as the final concentration of DNA per well during the 24-48 h incubation period is 0.57 ng/μL.

1. Seed cells such at the end of the 24-48 h incubation period confluency will be in the range of 70-80%. High confluency will result in increased O₂ consumption, and therefore alter the local O₂ environment, and vice versa for low confluency. As such confluency should be kept consistent across samples. 70-80% provides an adequate pool of cells to choose from and image individually or in groups.

2. Remove old cell culture medium. Add 400 μl of fresh culture media.

3. Add 150 μL of OptiMEM and 400 ng of the Myo-mCherry plasmid to a 1.5 mL tube.

4. Add 150 μL of OptiMEM and 2 μL of Lipofectamine® 2000 to a different 1.5 mL tube. Lipofectamine® 3000 can be used in cell lines, such as suspension and primary cells, that are difficult to transfect [26].

5. Combine the lipofectamine and plasmid solutions. Mix gently.

6. Incubate the transfection solution for 10-20 min at room temperature.

7. Add dropwise to a single well on the multi-well slide.

8. Allow the cells to incubate at 37°C and 5% CO₂ for 24-48 h. After 24-48 h, wash the cells twice with warm PBS and then cover with 500 μL (2-3mm) of fresh media.

9. When imaging, use phenol red free media to reduce background signal. Imaging media can be either DMEM or Opti-MEM formulated without phenol red, or Bright-MEM optimized to give low background for imaging. In alternative, DPBS or HBSS can also be used as low background imaging solutions, care must be taken in this case if longer imaging sessions are planned since these buffer/salt solutions do not contain any nutrients.

3.2. Fluorescence Two-Photon Image Acquisition

1. Turn on and (if necessary) tune the laser apparatus such that the excitation beam provides consistent mode-locked pulses (typically <200 fs width) at a high repetition rate (40 – 80 MHz) (see Note 8).

2. Turn on and configure the microscope with mirrors and filters (or their two-photon compatible functional equivalents). It is possible that some mirrors are built into your microscope system and others need to be installed manually depending on the configurations available. Consult your system documentation for further details.

3. Turn on the photon-counting card module and (if necessary) sync it with the laser pulses, either with a ns photodiode such as Thorlabs det 10A observing a small pickoff (for example, a cover slip at 45° deflecting the excitation beam) or from an attenuated manufacturer’s “sync” electrical output from the laser. Additionally, sync the module with the time and pixel clocks on your microscope. See your TCSPC module documentation for specific instructions (see Note 9).

4. Mount your sample on the microscope stage and eliminate as much room light as possible prior to acquiring images (see Note 10).

5. Ensure that the microscope and the TCSPC software have the same image dimensions (for example, 2ⁿ x 2ⁿ pixels²). The dimensions you can obtain depend on your TCSPC system (see Note 11).

6. Find your sample of interest and record the spatial TCSPC histogram (often for 30 – 100 seconds, depending on the intensity of your sample). When imaging single cells, zoom in to fill as much of the field as possible (see Notes 12 and 13).

7. Ascertain that the photon count rate in bright areas is still below 4% of the laser repetition, unless you intend to employ a “pileup correction” algorithm.

3.3. IRF Measurement (Multiphoton Microscopy)

1. Place the short-pass filter that transmits SHG wavelengths in front of the detector.

2. Mount the optically thin SHG sample onto the microscope stage.

3. Collect a spatial (or spectral) TCSPC histogram using the same parameters used during image acquisition in Subheading 3.2 (e.g., laser power, gain, offset, number of time channels, optics configuration. Attenuate emission as needed without changing power). See your TCSPC module documentation for further details (See Note 14).

3.4. Controlled Experimental Environment with a Range of Stable Oxygen Concentrations for Imaging

1. Mount a small, enclosed incubator onto the stage of the microscope. Use the temperature controller to preheat the incubator to the target temperature of 37 °C for live cell imaging. Connect the incubator to the gas mixing system, and be sure that the gas tanks are open, tubing is unobstructed, and gas is being delivered inside the chamber.

2. Once the incubator is at temperature, place the slide containing the cells inside. Remove the lid from the sample slide to allow equilibration with the atmosphere inside the chamber. Check the readout on the gas mixing system to verify 5% CO₂ and desired O₂ concentration. In this environment cells remain stable and can be imaged at the desired oxygen concentration for several hours.

3. We recommend obtaining FLIM images from at least 30 cells at each imposed oxygen concentration to obtain a sample size large enough for meaningful statistical analysis [27].

3.5. Myo-mCherry Calibration and Cell Treatment with Rotenone and Antimycin A

The lifetime of Myo-mCherry is used to determine the concentration of intracellular O₂. However, cells, through oxidative phosphorylation within mitochondria, consume O₂. This means that cells act as an O₂ sink, and therefore the concentration of O₂ within cells as measured by Myo-mCherry is not equivalent to the imposed O₂ concentration within the atmosphere, or even the media. Treatment with both rotenone and antimycin A nearly eliminates this sink by inhibiting cellular respiration. This allows for intra- and extra-cellular O₂ concentration to be treated as equivalent, and subsequent determination of cellular pO₂ as described below.

1. Prepare 1 mM stock solutions for both rotenone and antimycin A by dissolving them in DMSO.

2. Dose the desired sample such that concentration of each agent is 100nM.

3. Re-dose samples every 4 hours during imaging, or as needed to maintain respiratory inhibition.

4. Use the same controlled atmosphere imaging set up as before. Image the treated cells at various stable O₂ concentrations from 20% down to 0.5% in steps of 3-4% with a constant CO₂ concentration of 5%. Begin with the highest O₂ and end with the lowest to avoid undue stress and shock to the cells.

5. Allow 30-45 min between step-downs in O₂ concentration. This gives time for atmospheric gasses to diffuse into the media and reach a new equilibrium. Larger step-downs require longer equilibration periods.

6. We again recommend obtaining at least 30 FLIM images at each imposed O₂ concentration.

3.6. Measurements of the Oxygen Partial Pressure (pO₂) at Each Imposed Oxygen Concentration

This measurement can be performed during the imaging session, or as a separate experiment. Be aware that if performed during imaging, care must be taken not to record O₂ concentration through the OxyLite Pro while imaging: the fiber probe uses light to measure O₂ concentration, and emits light at 650 nm which could interfere with FLIM measurements by increasing background noise.

1. Setup the same controlled atmosphere chamber and same dish used for imaging.

2. Plate cells to the same confluency used for FLIM, and transfect them with the Myo-mCherry probe.

3. After 24-48 h, proceed to prepare the sample as if it was going to be used for FLIM measurements.

4. Prepare a fiber optic polymer phosphorescence quenching probe such as OxyLite Pro (NX-BF/O/E, Optronix Ltd., Oxford, United Kingdom). This bare-fiber O₂ sensor, or similar O₂ monitoring probe, can be used to measure pO₂ in the media at the bottom of the Petri dish close to the cell layer. Obtain measurements for untreated cells and samples treated with rotenone and antimycin A. Obtain an average pO₂ for each condition by collecting six point measurements in different areas of each sample. Repeating using five samples each of treated and untreated cells to record sample to sample variability and statistics is strongly advised.

5. Observe the temperature and pO₂ readings on the OxyLite Pro display. Record pO₂ for imposed O₂ concentrations of 20% down to 0.5%. Use the same step-downs used during imaging.

6. Use the partial pressure measured at each imposed O₂ concentration to plot the lifetime values of Myo-mCherry versus the measured pO₂ and perform a hyperbolic curve fit of the data (see 3.9). pO₂ values vary between cell lines due to different O₂ consumption rates [28]; as such, this procedure has to be repeated for each different cell line used, and ideally should be done in parallel with imaging to achieve greater accuracy.

3.7. FLIM Analysis

1. The software SPCImage (B&H) is used to quantify lifetime values and calculate decay coefficients, as shown in Fig. 3. If there is a large enough laser interpulse time so that the entire Myo-mCherry fluorescence decay curve is recorded (e.g. 12.5 ns), then it is possible to use the experimental decay function with the “standard decay” model without incurring “wrap around” issues.

2. The SHG transient measured from urea crystals as described in Subheading 3.3 was used as the IRF data. To load the IRF data into SPCImage, open the raw FLIM file containing the IRF data and select an appropriate region in the image using the blue crosshairs. Observe the wave form in the curve window and increase binning as needed to obtain a smooth curve (typically >10,000 peak photons). Then use the sliders in the same curve window to select the start and end of the portion of the TCSPC histogram to be used as the IRF. Store the IRF for use in image analysis by clicking the “curve to IRF” button on the left-hand tool bar.

3. Open a FLIM file containing the desired data to be analyzed. Open the IRF tab from the toolbar and select “Paste from clipboard.” The IRF appears as a green wave form in the curve window, as shown in Fig. 3d.

4. Select a bright pixel in the region of interest using the blue cursor. If there is any unwanted region of the image, select the ROI tool on the left side menu and click around the perimeter of the region to be included for analysis. Alternatively, use the ROI tool to remove any unwanted areas or use the white cursors on the edge of the image frame to exclude undesired sections.

5. In the presence of the FRET acceptor Myoglobin, the fluorescence decay of mCherry is better described by a multiexponential function with more than two exponential components [29]. However, the 4-parameter function in Eq. (1) (two exponentials) adequately represents most data and can be used to glean FRET trends in terms of average lifetime. In the menu on the right side, increase the multiexponential decay components from 1 to 2.

6. For dimmer samples, increase the bin number to values greater than 2 to avoid using decays with a peak count lower than 1,000. For our samples, the typical bin value per image is 5 (which corresponds to an area of 11x11 (pixels)²). If such a loss of spatial lifetime resolution is prohibitive, one must collect more frames or increase photon rates (see Note 15).

7. Increase the threshold at least by a factor of N²+1, where N is the bin number, to exclude dark pixels. Avoid a high threshold value that results in missing cellular features.

8. The shift of the IRF is determined by fitting the decay of the pixel with the highest intensity in each image, while leaving the shift parameter free. Once the shift is determined from the brightest pixel, it can be fixed for the subsequent pixel-by-pixel fitting of the lifetime decay across the whole image. A known standard sample with a single lifetime can be fit with shift allowed to vary. Shift is then fixed for subsequent unknown samples according to the known standard (see Note 16).

9. Only fix the offset at zero if detectors with low background and negligible afterpulsing are used in conjunction with maximum rejection of ambient and stray light. (for example, an HPM-100 hybrid photo detector from B&H, or Leica HyD detectors) [4].

10. Unfix scatter correction to account for any photon generation as a result of second harmonic components or other zero-delay scattering [4] (see Note 17).

11. Select Options from the toolbar, and open the Color panel. Adjust the width of the lifetime histogram to an interval of 500 to 1500 ps. This covers the range of Myo-mCherry mean lifetimes. If left unspecified, SPCImage will autogenerate the histogram width.

12. Press F2, or select “decay matrix” in the calculate tab to start the pixel-by-pixel fitting procedure.

13. Once the fit is completed, the average lifetime is calculated for each pixel via amplitude weighting, and a lifetime distribution histogram is generated for the image or selected ROI.

14. If you have more than one image, go to the Calculate tab and select Batch Processing to analyze the rest of the images in the set using the same fit parameters.

15. Once the set of images has been batch processed, open another image in the set and observe that the same parameters are used. Readjust fit parameters as needed: to achieve a peak photon count close to 1,000 binning might have to be changed; to exclude dark pixels from analysis thresholding might have to be adjusted. Recalculate the fit to improve accuracy of the lifetime values for the current image of the set. Repeat for all images to verify the results.

Fig. 3.

Lifetime analysis in SPCImage for mitochondrial targeted Myo-mCherry in a primary mouse fibroblast. (a) Intensity image of sample. (b) Amplitude-weighted lifetime image of sample. (c) Lifetime histogram with mean lifetime value (μ) and its error range (μ−/μ+sig). (d) Lifetime decay curve fitting options. The time gates (T1 & T2), the lifetime binning parameter (Bin), and the photon threshold are held constant across the image. X and Y give coordinates for the crosshair tool overlaying the images. The top plot shows the data (blue), the model iteratively re-convolved with the IRF (red), and the IRF (green). The bottom plot shows the fit residuals. (e) Lifetime decay curve model results for the selected pixel. The parameters a1, t1, a2, and t2 correspond to $a_1, {\tau }_1, a_2,$ and ${\tau }_2$. The shift parameter corrects for the difference in timing between the detector response at the IRF collection wavelength and response in the emission collection band, usually a fraction of a channel. ‘Scatter’ represents the amount of scattered light (with same timing as IRF, and no decay tail) detected during measurement. ‘Offset’ corrects for the upward shift of the decay curve due to background signal.

3.8. Obtaining the Intracellular pO₂ from Lifetime Data at Each Imposed Oxygen Concentration

The τ([pO₂]) for respiring cells is compared to the data for the same cell type treated with rotenone and antimycin A, which inhibits intracellular O₂ consumption [30]. If O₂ freely diffuses through cell membranes, and oxidative phosphorylation does not occur in treated cells, the τ([pO₂]) cell response in these conditions can determine the relationship between pO₂ and lifetime value, thus providing a calibration curve [31]. By using the calibration curve obtained with this method, it is then possible to back calculate the pO₂ in untreated cells based on the lifetime of Myo-mCherry measured.

Myoglobin has a hyperbolic oxygen dissociation curve [32, 33]. Use MATLAB’s 4Curve Fitting Toolbox (The MathWorks Inc., Natick, Massachusetts) or any other fitting software to fit a hyperbolic curve to the data:

$$\tau \left(\left[p{O}_2\right]\right)=\left({\tau }_{max}-{\tau }_{min}\right)\frac{\left[p{O}_2\right]}{a+\left[p{O}_2\right]}+{\tau }_{min}$$ (3)

where ${\tau }_{max}\mathrm{ }\mathrm{a}\mathrm{n}\mathrm{d}{ \tau }_{min}$ are the recorded lifetime values at the highest and lowest imposed partial oxygen partial pressure, [pO₂] respectively, and a is a fitting parameter related to oxygen affinity [34] (Fig. 4a).

Fig. 4.

(a) Average Myo-mCherry lifetimes for A549 cells and A549 cells treated with rotenone and antimycin A versus imposed [pO₂]. Since rotenone and antimycin A inhibit cellular respiration, there is more oxyMyo-mCherry in the cell, FRET decreases, and the average lifetime values increase relative to the respiring cell line. (b) Intracellular [pO₂] for A549 cells was plotted against imposed pO₂. Values were calculated using a, ${\tau }_{max}$ and ${\tau }_{min}$ values obtained from the calibration curve.

1. Fit τ([pO₂]) for the non-respiring cell to the model described in Eq. (3) to obtain a, which serves as a conversion factor between lifetime and external [pO₂] (See Note 18).

2. Fit $\tau \left(\left[{\text{pO}}_2\right]\right)$ for the respiring cell to the same model to confirm the hyperbolic behavior of Myo-mCherry in the cell line of interest.

3. Given a lifetime value for a respiring cell, find the external [pO₂] that produces the same lifetime value on the non-respiring cell calibration curve. This [pO₂] corresponds to the intracellular [pO₂] in the respiring cell (Fig. 4b).

3.9. Statistical Analysis

Using statistics software packages (SPSS, R, etc.), Mann-Whitney U tests can be used to determine differences between groups for sets of cells with at least 30 samples across external [pO₂]. Mann-Whitney U tests are alternatives to t-tests, where the lifetimes for the two sets of cells is not assumed to follow a normal distribution [35].

Fig. 5.

(a) A C2C12 cell shows homogeneous distribution of the probe. (b) A differentiated C2C12 myo-tube shows heterogeneous distribution within the cytoplasm and organelles.

Fig. 6.

(a) Oversampling of the Airy disk in the intensity image (left) and binning of lifetime data (right). Since the pixel size is smaller than the physical image resolution, lifetime data can be binned with no loss in spatial resolution. (b) Overlapping binning of pixels for lifetime calculation. Since adjacent pixels in the intensity image overlap in the binned lifetime data, sampling artifacts are minimized. (This figure was reproduced from Becker 2017 [4] with permission)

Fig. 7.

The effects of (a) IRF, (b) shift parameter, (c) bin number, and (d) thresholding on average Myo-mCherry lifetimes. Error bars represent ±μ sig values given by SPCImage. Shift, threshold, and bin values taken from a single A549 sample. Bin and threshold should be done together to avoid these biasing trends.

Fig. 8.

Effect of intermolecular FRET on Myo-mCherry lifetime. (a) Lifetime distribution of Myo-mCherry in non-respiring SCO2 KO cell at 20% oxygen concentration. (b) Shorter lifetime of Myo-mCherry due to intermolecular FRET. (c) Lifetime histogram distribution of Myo-mCherry (solid black line with a peak at 1.33 ns) and a shift in the average lifetime (solid white line with a peak at 0.97 ns) due to intermolecular FRET.