Live-cell FLIM-FRET using a commercially available system.

Abstract

Förster resonance energy transfer (FRET)-based sensors have been powerful tools in cell biologists’ toolkit for decades. Informed by fundamental understanding of fluorescent proteins, protein-protein interactions, and the structural biology of reporter components, researchers have been able to employ creative design approaches to build sensors that are uniquely capable of probing a wide range of phenomena in living cells including visualization of localized calcium signaling, sub-cellular activity gradients, and tension generation to name but a few. While FRET sensors have significantly impacted many fields, one must also be cognizant of the limitations to conventional, intensity-based FRET measurements stemming from variation in probe concentration, sensitivity to photo-bleaching, and bleed-through between the FRET fluorophores. Fluorescence lifetime imaging microscopy (FLIM) largely overcomes the limitations of intensity based FRET measurements. In general terms, FLIM measures the time, which for the reporters described in this chapter is nanoseconds (ns), between photon absorption and emission by a fluorophore. When FLIM is applied to FRET sensors (FLIM-FRET), measurement of the donor fluorophore lifetime provides valuable information such as FRET efficiency and the percentage of reporters engaged in FRET. This chapter introduces fundamental principles of FLIM-FRET towards informing the practical application of the technique and, using two established FRET reporters as proofs of concept, outlines how to use a commercially available FLIM system.

I. Introduction

For years the application of fluorescence lifetime imaging microscopy (FLIM) techniques has largely been the purview of specialists, the objective of this methods chapter is to make FLIM-FRET accessible to generalists – specifically cell biologists who are already familiar with live-cell fluorescence microscopy approaches. Many FLIM systems in use today are custom-built and housed in labs with expertise in developing and maintaining advanced microscope systems. The microscope described in this chapter is a commercially available “turn-key” system that is stable enough to be housed in a microscopy core facility. For most, the core facility will be the best economic model for incorporating FLIM-FRET into their research program as the hardware requirements include a laser scanning confocal microscope, high-frequency pulsed lasers, and time-correlated single photon counting (TCSPC) detectors. The protocol will also include standard operating procedures for acquiring and analyzing FLIM-FRET data using well-established software from Nikon (NIS-Elements) and Becker & Hickl.

FLIM measures how long a fluorophore is in its excited state to determine its fluorescent lifetime (Elangovan, Day, & Periasamy, 2002). There are two major modalities of acquiring FLIM data referred to generally as “frequency domain” FLIM and “time domain” FLIM. This chapter will deal exclusively with time domain FLIM using time-correlated single photon counting (TCSPC). TCSPC uses a pulsed laser with one or more single-photon detectors. Photons are sorted based on arrival time to the detector in relation to the excitation signal. A decay curve is constructed by plotting the time of photon arrivals following an excitation pulse versus intensity, which is the number of photons detected at a given arrival time (Figure 1). The lifetime of a fluorophore (τ) is defined as the arrival time where the population of photons has decayed to 1/e of the original population or intensity. Using TCSPC with a scanning confocal head, high spatial resolution can be attained, but depending on the lasers, detectors and reporter brightness, long acquisition times (minutes) may be required to attain a sufficient number of photons for robust data analysis (Padilla-Parra, Auduge, Coppey-Moisan, & Tramier, 2008).

Figure 1.

Schematic of how a decay curve is generated in the time correlated single photon counting (TCSPC) FLIM technique. Printed with permission from Wolfgang Becker.

This methods chapter will begin by outlining basic principles of FRET and FLIM before shifting focus to practical considerations such as hardware requirements (with associated cost estimates) and appropriate donor selection for FLIM-FRET sensors. A step-by-step protocol for acquiring FLIM images on a Nikon scanning confocal system using NIS-Elements in sync with SPCM64 (Becker & Hickl) will be detailed using the real world example of a previously characterized FRET biosensor. Step-by-step protocols are presented for conducting FLIM analyses in SPCImage (Becker & Hickl) to determine: 1) the donor lifetimes in the presence and absence of FRET, 2) the percentage of reporter engaged in FRET in your sample, and 3) FRET efficiency. The principles of the Phasor Plot – an approach for displaying and analyzing FLIM images – will be introduced and a protocol for Phasor Plot analysis in SPCImage is presented. A second proof-of-concept example of FLIM-FRET using the system described here is presented followed by the introduction of a FLIM-FRET acquisition and analysis platform that will soon be released by Nikon to interface with their NIS-Elements software. In summary, this chapter describes how to visualize and analyze FRET sensors using a commercially available FLIM system that has proven robust enough to become a workhorse microscope in a core facility.

II. Principles of FLIM-FRET

FRET is a non-radiative energy transfer from one fluorophore in the excited state (the donor) to another molecule in the ground state (the acceptor) which is typically located less than 10 nm away. In order for FRET to occur three conditions must be fulfilled:

1) The donor emission spectrum must overlap with the absorption spectrum of the acceptor

2) Both molecules must be within angstroms to nanometers distance to each other

3) Both molecules must have a matching dipolar orientation

Since the energy transfer efficiency changes as the inverse sixth power (r⁻⁶) of the distance r between the donor and acceptor, selection of the proper FRET pairs allows the researcher to monitor distance changes in the angstrom and nanometer range, which makes the approach relevant for probing macromolecular structural dynamics at the scale of proteins and nucleic acids within living samples. A common measurement technique is intensity-based FRET, which can be performed with confocal or widefield microscopy approaches. Intensity-based FRET can be measured either through the donor (acceptor-photobleaching) or the acceptor (sensitized emission). Unfortunately, both techniques generate experimental challenges due to spectral cross-talk, photobleaching sensitivity, and variable reporter concentration. A better way to determine FRET efficiencies has been the use of measuring the fluorescence lifetime of the donor. The lifetime is the half-time a molecule spends in the first excited state and is on the nanosecond time scale (10⁻⁹ seconds) for common FRET donors.

In a FRET experiment a fluorescent molecule (the donor) in the excited state can return to the ground state either by emitting a photon (fluorescence) or in a non-radiative way by transferring energy to the acceptor, as discussed above. This transfer is associated with a probability over time - meaning the longer the donor is in the excited state the higher the likelihood it will engage in FRET. Since there are two competing kinetic pathways (fluorescence and non-radiative) under FRET conditions, the lifetime of the donor becomes shorter than its lifetime under non-FRET conditions. The lifetime (τ) can be described and mathematically fitted using an exponential function or a sum thereof of the fluorescence intensity as a function of time:

$$I\left(t\right)={\sum }_i{a}_i{e}^{\raisebox{1ex}{$-t$}\!\left/ \!\raisebox{-1ex}{${\tau }_i$}\right.}$$

where ∑_i a_i = 1

An ideal FRET donor should exhibit a single-exponential decay (further details below). When engaged in FRET, the single-exponential decay curve now exhibits a double-exponential behavior (Figure 2). Since some fraction a₁ of donors will not have acceptor in their vicinity, these donors would fluoresce with the normal baseline lifetime τ₁. However, a fraction of donor a₂ will be undergoing FRET, which introduces a second lifetime component τ₂ that is shorter (or faster) than the native donor. The result is an average fluorescence lifetime (τ_FRET) where:

$${\tau }_{\mathit{\text{FRET}}}=a_1{\tau }_1+a_2{\tau }_2$$

Figure 2.

Schematic showing how a single exponential decay donor behaves in the absence of FRET and when engaged in FRET with an acceptor.

For the FRET-FLIM approaches outlined in this chapter you will need two measurements: 1) cells with only the donor expressed to determine the lifetime (τ) of the donor in your model system, and 2) cells expressing the FRET reporter that includes both the donor and the acceptor in the same sample system. Once the baseline lifetime of the donor under non-FRET conditions in your cell line is known (τ_D) and the lifetime of the donor under possible FRET conditions (in the presence of an acceptor) has been measured (τ_FRET), the FRET efficiency (E) can also be calculated by:

$$E=\left(1-\frac{{\tau }_{\mathit{\text{FRET}}}}{{\tau }_D}\right)\times 100\%$$

Unlike intensity-based FRET experiments the acceptor can be “dark” or non-fluorescent (Ganesan, Ameer-Beg, Ng, Vojnovic, & Wouters, 2006; Murakoshi, Lee, & Yasuda, 2008; H. Murakoshi, A. C. E. Shibata, Y. Nakahata, & J. Nabekura, 2015) because only the lifetime of the donor needs to be measured. In addition, you can measure each experiment independent of excitation intensity and integration time as long as your data has sufficient signal-to-noise to reveal its decay characteristics. Consequently, when FLIM is applied to FRET sensors (FLIM-FRET), measurement of only the donor fluorophore lifetime, which negates bleed-through concerns, provides valuable information such as FRET efficiency and the percentage of reporter engaged in FRET in a manner that is independent of local probe concentration and less sensitive to photo-bleaching than conventional FRET measurements.

III. Equipment and Materials.

A FLIM system can be set up on an existing scanning confocal microscope system such as the Nikon A1 confocal system described here, which is depicted in Figure 3. In the method outlined in this chapter the system was set up to visualize the FRET donor mTurquoise2 (CFP variant) and FRET acceptor mVenus (YFP variant) although other FRET pairs can be used with the appropriate pulsed lasers, filters, and dichroic mirrors. A parts list and estimated costs for a system with 4 pulsed lasers (405 nm, 445 nm, 488 nm, and 561 nm), accompanying filters and dichroic mirrors, and two TCSPC detectors is provided in Table I. A system need not require this number of laser lines (plus appropriate filters and dichroic mirrors) and detectors and could function with as little as a single pulsed laser for your FRET donor of choice and a single detector to count the emitted photons. It is also important to note that the addition of FLIM capabilities does not negatively impact the other imaging modalities on the microscope to which it is added.

Figure 3.

Layout of the FLIM system outlined in this methods chapter. CW: Continuous wavelength. PMT: Photomultiplier tube. Note 1: the CW lasers are contained in the box to the right most edge of the photograph. Note 2: The “Spectral detector” is part of the FLIM system in the local user facility but is not used in the FLIM imaging protocol.

Table I.

Parts list and cost estimates for the components of TCSPC FLIM system.

	Product Number	Item	Quantity	*Price per unit	Total Price
Detection	SPC-152	TCSPC module card	2	$16,000	$32,000
	DCC-100	PCIe Control modules	3	$2,500	$7,500
	HPM-100-40	TCSPC detector units	2	$10,000	$20,000
Lasers & Delivery	BDL-xxx-SMN/Y	Visible pulsed lasers	4	$17,000	$68,000
	O-LAS-FM	Fiber socket	4	$700	$2,800
	O-BDL-COMBINER-4-FC	Laser combiner hardware	1	$13,000	$13,000
Connectivity & Emmision Side	CSC-N-A1	Scan cable for A1 connection	1	$100	$100
	C10-SMAM-1-SMAM	Coax cable	1	$15	$15
	F100-005-FC-FC	Fiber cable	1	$200	$200
	SPC-BOB-104	Break out box	1	$400	$400
	A-FC-OD-MOUNT	Fiber adapter	1	$400	$400
	O-CUBE	4 Sets of filter cubes	4	$1,900	$7,600
	O-FLP-XXX 25MM	Laser-blocking LP filters	4	$550	$2,200
	O-UNI-BSU-HPM-2-XF	Beamsplitter holer	1	$1,300	$1,300
	W-THO-ANGLE-S	Short right angle plate	1	$125	$125
	O-SHUT-CMOUNT	Shutter	1	$3,500	$3,500
Software	SPC-Image	Analaysis software	1	$1,600	$1,600
				Total:	$160,740

^*Estimated at 2019 prices.

Live-cell imaging

1. Scanning confocal microscopy equipment

Images and FLIM data were acquired on a Nikon Ti-E microscope body that was fitted with an A1 confocal scan head. The scan head has two laser input ports and three emission output ports. Four continuous-wave lasers (Nikon LUN-V; 405, 488, 561, and 640 nm) were combined and fiber-coupled into one input port and four pulsed lasers (Becker & Hickl; BDL-405-SMN, BDL-445-SMN, BDL-488-SMN, and BDS-561-SMY; 405, 445, 488, and 561 nm respectively) were combined and fiber-coupled into the second input port. The first output port of the scan head was connected to the Nikon A1-DU4G that contains 5 PMT detectors, the second port connected to the Nikon 32-channel spectral detector A1-DUs, and the third port connected to a pair of Becker & Hickl HPM-100-40 detectors.

2. Time-correlate single photon counting (TCSPC) Detectors

Two Becker & Hickl HPM-100-40 detectors were coupled to a changeable filter cube that consisted of an appropriate dichroic mirror and two band pass emission filters; for CFP/YFP 488/50BP, 520LP, 550/49BP (Semrock, FF01-488/50, FF520-Di02, FF01-550/49 respectively.) .

3. Final comments on approaches to live-cell imaging

Once cell lines expressing the FRET reporters have been made, sufficient time should be spent at the microscope familiarizing oneself with the cells prior to doing FLIM-FRET experiments. If it is difficult to readily assess the extent of chromosome alignment by phase contract or DIC microscopy, as is the case for Drosophila S2 cells, then “scanning” for cells will require the use of epifluorescence and; therefore, limiting photobleaching of the reporters is critical. If CFP-YFP variants are used as your FRET pairing (as is the case in this protocol) then co-transfecting the cells with an RFP-tagged construct such as mCherry-α-tubulin will aid in this process as you can scan on the RFP channel to avoid photobleaching of the donor or acceptor. If you will be scanning cells with epifluorescence then aim for a level of proficiency in which the necessary cellular states (e.g. mitotic versus interphase) can be identified and centered in the field of view within ~3–5 seconds.

IV. Proof of concept 1 – CyclinB1-Cdk1 activity reporter in Drosophila S2 cells

The first proof of concept will make use of a previously characterized reporter for CyclinB1-Cdk1 kinase activity (Gavet & Pines, 2010) in which the donor (mCerulean) has been replaced with mTurquoise2 (Goedhart et al., 2012), which exhibits a mono-exponential decay, and the acceptor (YPet) has been replaced with mVenus. The FRET pairs flank the autophosphorylation site of human CyclinB1, which is a substrate of active CylinB1-Cdk1, and a phospho-binding domain comprised of the Polo Box Domain (PBD) of Polo-like kinase 1 (Plk1). Consequently, the sensor undergoes a conformational change when it is phosphorylated by Cdk1 that results in an increase in intramolecular FRET (Figure 4). Prior work demonstrated that this reporter specifically reports on CyclinB1-Cdk1 activity and exhibits increased FRET in mitosis and decreased FRET in interphase. In the case of the CyclinB1-Cdk1 reporter used in this proof of principle, a non-phosphorylatable control was also used to confirm that any measured changes in donor lifetime and FRET efficiency were indeed due to its phosphorylation rather than non-specific changes in its conformation between interphase and mitosis. It is also important that the lifetime of the donor in the absence of an acceptor (no FRET condition) be measured in the cell line of interest to accurately analyze FLIM-FRET data under possible FRET conditions. The Cdk1 activity reporter, control non-phosphorylatable reporter, and mTurquoise2 (donor alone) were cloned into the Drosophila cell expression vector pMT-V5-HisB (Thermo Fisher) in which their expression was under the control of a promoter sequence that drives constitutive, but moderate levels of expression. The chapter will not detail the transfection protocol or general culturing of Drosophila S2 cells as we have detailed this elsewhere (Ye & Maresca, 2016).

Figure 4.

CyclinB1-Cdk1 FLIM-FRET reporter design. FRET increases when the sensor is phosphorylated during mitosis due to a conformation change triggered by the Polo Box Domain (PBD) of Polo-like kinase 1 (Plk1) binding to a phosphorylated recognition sequence from human CyclinB1.

A. Measuring donor lifetime

Ideally, the donors used for FLIM should exhibit single exponential decay (Martin et al., 2018). Unfortunately, this property excludes many commonly used fluorophores. Since analysis of donors that exhibit multi-exponential decay is technically challenging and not as clean or reliable as the analysis of mono-exponential decay data we strongly encourage researchers to invest time up front in engineering their FRET sensors to contain an appropriate mono-exponential donor. There are now multiple fluorophore options that are well-suited for use with FLIM-FRET since newly innovated fluorophores are increasingly becoming available (Goedhart et al., 2010). These include the modification of multi-exponential decay fluorophores to exhibit single exponential decay (Goedhart, et al., 2010) as well as modifications of single exponential decay fluorophores to increase their fluorescence intensity and/or increase fluorescence lifetime (Goedhart, et al., 2012; Kremers, Goedhart, van Munster, & Gadella, 2006; Markwardt et al., 2011). Table II shows fluorophores that exhibit single exponential decay, and thus are best-suited for use with FLIM-FRET. The reporters highlighted in this chapter as proof of principles all involve intramolecular FRET using mTurquoise2 as the donor and mVenus as the acceptor and the protocols/hardware are outlined as such. If you are not using a CFP/YFP pairing then pulsed lasers, filters, and dichroic mirrors should be changed accordingly to match your FRET pair of choice.

Table II.

Properties of fluorescent protein donors that exhibit mono-exponential decay. Appropriate FRET acceptors can be identified at https://www.fpbase.org/fret/.

Mono-exponential Donors	Source (s)	Lifetime (ns)	Peak Excitation λ	Peak Emission λ	Pulsed Laser	Color
mScarlet-I	(Padilla-Parra, Audugé, Coppey- Moisan, & Tramier, 2008)	3.1	569 nm	593 nm	561 nm	Red
dsRed	(Heikal, Hess, Baird, Tsien, & Webb, 2000; Piatkevich & Verkhusha, 2011)	3.7	558 nm	583 nm	561 nm	Red
LSSmOrange	(Demeautis et al., 2017; Shcherbakova et al., 2012)	2.75	437 nm	572 nm	445 nm	Orange
mOrange2	(Shaner et al., 2010)	2.7	549 nm	565 nm	532 nm	Orange
Citrine	(Heikal et al., 2000)	3.6 at pH=9	516 nm	528 nm	515 nm	Yellow
EYFP	(Kremers, Goedhart, van Munster, & Gadelk, 2006)	2.8	513 nm	527 nm	515 nm	Yellow
mVenus	(Kremers et al., 2006)	2.7	515 nm	527 nm	515 nm	Yellow
mNeonGreen	(Shaner et al., 2013)	3.1	506 nm	517 nm	488 nm	Green
mTFPl	(Ai, Henderson, Remington, & Campbell, 2006)	3.2	462 nm	492 nm	445 nm	Cyan
mCereulean3	(Markwardt et al., 2011)	4.3	433 nm	475 nm	445 nm	Cyan
mTurquoise	(Goedhart et al., 2010)	3.7	434 nm	474 nm	445 nm	Cyan
mTurquoise2	(Goedhart et al., 2012)	3.8	434 nm	474 nm	445 nm	Cyan
Aquamarine	(Betolngar et al., 2015; Erard et al., 2013)	4.1	430 nm	474 nm	445 nm	Cyan

References for the table: (Ai, Henderson, Remington, & Campbell, 2006; Betolngar et al., 2015; Demeautis, et al., 2017; Erard et al., 2013; Goedhart, et al., 2010; Goedhart, et al., 2012; Heikal, Hess, Baird, Tsien, & Webb, 2000; Kremers, et al., 2006; Markwardt, et al., 2011; Padilla-Parra, et al., 2008; Shaner et al., 2013; Shaner et al., 2008; Shcherbakova, et al., 2012)

1. Establishing donor fluorophore lifetime in the absence an acceptor/FRET

1. Procedure

   1. Seed Drosophila S2 cells expressing the donor (mTurquoise2) on a concanavalin A coated glass bottom 35 mm dish (Cellvis, Part No. D35-20-1.5-N) at ~70% confluency. After the cells adhere (~30 minutes), bring the volume up to 2 mls using fresh Schneider’s media supplemented with 10% heat inactivated FBS (Life Technologies). Allow the cells to flatten for an additional 30 minutes.
   2. Once mounted on the microscope, quickly scan by eye under widefield fluorescence settings on the GFP channel (or the RFP channel if co-transfected with an RFP-tagged construct) and center cells of interest in the field of view.
   3. Setup the confocal imaging parameters in the A1Plus Compact GUI. In our system the settings were set to image size 512×512 with a pixel dwell time set to 2.2 μs/pixel and a pinhole size set to 1.0 airy units (AU), which is based on the excitation wavelength and is automatically defined in the GUI. Note: these settings matched the FLIM acquisition parameters allowing a direct comparison of the fields between the confocal image and the FLIM image.
   4. In the A1Plus Compact GUI, click on “Scan” to obtain the best focal plane of your cell of interest. If you expect to present confocal images alongside your FLIM images then define parameters such as laser power and gain, avoiding pixel saturation, to acquire a quality confocal image of your sample. Save the image to the appropriate folder.
   5. In the A1plus Compact GUI click on “AUX”. This changes the output from the scan head so that emitted photons will be sent to the TCSPC detectors.
   6. In the A1plus Settings GUI in the “Filter and Dye” tab click on “AUX”. The “AUX Settings” GUI will appear to allow you to define the system settings such that the input into the scan head will be the pulsed laser and the proper dichroic mirror will be positioned in the light path to allow the appropriate emitted photons to reach the TCSPC detectors. On our system, under “Filter and Dye” 400–457/514 was selected in the dropdown next to “1^st DM” and the laser was set to “Port 2” in the dropdown under “Detector External”., All boxes under “Acquisition” in the AUX Settings GUI were unchecked.
   7. Click on “START” at the top of the AUX SETTINGS interface. You will hear the scan head turn on although no laser light will be hitting the sample yet. At this point, be sure that the appropriate pulsed laser (ours was 445 nm set to 50 mHz) is turned on.
   8. Open SPCM64 and load settings from a template file supplied by the system installer or, if a return user, by loading a file from a prior experiment. To load a file, select the “MAIN” tab at the top of the window and click on “LOAD” in the dropdown. In the “SPC-150 Load SPC-file” pop-up window select “file name” to open a file from a destination folder. After selecting the file name be sure to complete the action by clicking on “LOAD (F10)”. Once completed, the appropriate hardware settings will be loaded and appear in the “M1/M2 DCC-100” pop-up window.
   9. Configuration settings will vary depending on the hardware you have and its installment. In the hardware configuration for our system, Module 1 controls the TCSPC detectors and a shutter that allows light into the detectors, Module 2 controls the 405 and 445 nm pulsed lasers, and Module 3 controls the 488 and 561 nm pulsed lasers. Thus the configuration was as follows:

      Module 1:	Connector 1	Detector A	(turn on)
      	Connector 2	Shutter	(turn on)
      	Connector 3	Detector B	(turn on)
      Module 2:	Connector 1	405 laser	(turn on if using)
      	Connector 2	Not Connected
      	Connector 3	445 laser	(turn on if using)
      Module 3:	Connector 1	488 laser	(turn on if using)
      	Connector 2	Not Connected
      	Connector 3	561 laser	(turn on if using)

   10. To begin acquiring an image toggle to Module 1 in the lower left hand corner of M2 DCC-100 pop-up window to control the two detectors (Connectors 1 and 3 – with the +12V, +5V, and −5V boxes highlighted and “Gain /HV set to 84.00) and the shutter (Connector 2 – only “b0” highlighted). Click on “Enable outputs” to provide power to the detectors and shutter in module 1. After doing this step you will see baseline signal in the detectors in the lower left of the SPCM64 screen above columns labeled “CFD” and “TAC”.
   11. Since the 445 nm laser is being used, toggle to Module 2 in the pop-up window. The interface will now allow you to power the appropriate laser and set its power. To do so highlight +5v under Connector 3 (445 nm pulsed laser) and adjust laser to desired power. For our experiments, we found that up to 50% laser power worked well. Click on “Enable Outputs” to provide power to the laser. After doing this step you will see laser light through the objective and increased signal will be evident in the lower left of the SPCM64 screen above all four columns labeled “SYNC”, “CFD”, “TAC”, and “ADC”.
   12. Click on the “Start!” tab at the top of the window to acquire data. Images of the FRET donor and acceptor will appear in two separate windows. In our system, the “W1-FLIM1” window on the left shows the mVenus image while the “W2-FLIM2” window on the right shows the mTurquoise2 image – both of which will appear brighter over time as more photons are collected. Aim to acquire 30–50 million photons per scan, which will be tallied (as well as the elapsed time) in a blue highlighted window in SPCImage next to “Collecting data”.
   13. Once enough photons have been collected click on the “Stop!” tab at the top of the window to stop acquiring photons.
   14. In the M1/M2 DCC-100 pop-up window click on “Disable all modules” to turn off the detectors, shutter, and pulsed laser.
   15. Save the data by selecting the MAIN tab and clicking on “SAVE” in the drop down. In the “SPC-150 Save SPC file” pop-up click on “Select File (F9)” and select or create a file folder in which to save the FLIM images and to name the file. After selecting the destination folder and creating the file name, which should have a similar naming convention as the confocal image file saved in step 4, be sure to complete the saving action by clicking on “Save (F10)”.
   16. Return to NIS-Elements and click on “STOP” in the AUX SETTINGS GUI and then click on “Close” at the bottom of the AUX SETTINGS window. This will close the window and allow you to search for new cells to image.
   17. Repeat steps 2–16 for as long as your sample is viable or the duration of your microscope time – whichever comes first.
2. Donor lifetime measurements using SPCImage in donor alone conditions

   1. Open your FLIM data in the SPCImage software by selecting the “File” tab at the top of the SPCImage window and clicking on “Import”. In the “Open” pop-up window select your file of interest and click on “Open”.
   2. A pop-up window entitled “Options for YOUR FILE NAME.sdt” will appear. Under “Select data” in the lower right hand corner of the pop-up, identify the “Module Number” to open the data from the detector used to collect photons from the donor. Since our data was collected on detector 2, we selected Module Number “2” for both column options. After clicking “OK” in the pop-up window, the FLIM data will appear in the left window in SPCImage (Figure 5).
   3. In the “Multiexpontial Decay” interface (bottom of screen, right of the decay curve) select “1” for “Components”. None of the values in the Multiexpontial Decay interface should be fixed except “Scatter”, which should be fixed at “0” by checking the “Fix” box. (Figure 5 – Inset 1).
   4. Select the “Options” tab at the top of the SPCImage window and click on “Color” in the drop-down. In the pop-up window entitled “Color”, click on “Continuous” and set the appropriate lifetime “Range” to encompass the expected lifetime ranges – for the mTurquoise2 donor this was set from 600 (ps) to 3800 (ps). The lifetimes will be displayed in a color scale depending on whether the “Direction” is set to “R-G-B” or “B-G-R”. For example, if “R-G-B” is selected then the longest lifetimes (3800 ps) will be displayed as blue and the shortest lifetimes (600 ps) will be displayed as red, which would be inversed if “B-G-R” is selected. Since longer average lifetimes indicate lower FRET and shorter average lifetimes indicate higher FRET, we chose to display as R-G-B. Under “Coding of” at the bottom of the pop-up select “tm (ps)” in the dropdown for the “Value”. Click on “OK”.
   5. Blue crosshairs will automatically center upon the brightest pixel when you import an image. Move the cursor to the dimmest area of the cell you wish to analyze. In the decay curve window below the image, adjust the binning such that the peak intensity of the decay function (bottom of screen, below FLIM image window) is at least 100 A.U. (y-axis) (Figure 5 – Inset 2).
   6. At this point, if you plan on analyzing multiple images using the same analysis parameters then it is helpful to “Store the fit conditions” by clicking on the camera icon in the tab to the left of the FLIM image window (Figure 5 – Inset 3).
   7. Select the “Calculate” tab at the top of the SPCImage window and click on “Decay Matrix” in the drop-down, which will generate a color-coded FLIM lifetime image in the right window in SPCImage.
   8. The blue crosshairs can be positioned in either image window to assess local lifetimes. When the crosshairs are moved the lifetime value displayed above the decay curve as “tm = X ps” (Figure 5 – Inset 4) will change. The values in the “Multiexponential Decay” interface will also change. In the case of the mono-exponential fit being done in this section, the t1[ps] value will match the lifetime value above the decay curve and the a1[%] will read “100”.
   9. The pixels encompassing a region of interest (ROI) can be grouped together to measure the lifetime parameters for an entire cell or region of a cell. The “τ_m” value (shown above the decay curve) is generated by binning all the pixels in the ROI. To define an ROI, click on the “Define mask (move red crosshair)” icon in the tab to the left of the FLIM image (Figure 5 – Inset 5). Red crosshairs will appear in the FLIM image windows. Left click on the red crosshair to position it where you want to draw the ROI and then alternate between left clicking and moving the cross hair around your cell of interest in the left image window to draw the ROI.
   10. Once the ROI is drawn, “Activate binning of all pixels inside mask” by clicking on the red lock icon in the left tab (Figure 5 – Inset 6). Once the mask is locked, the blue crosshairs will disappear and the ROI will become bold. At this point, all the lifetime value parameters will be for the area within the ROI.
   11. When you are ready to generate a new ROI, “Deactivate binning of all pixels inside mask” by clicking on the red unlocked lock icon in the left tab (Figure 5 – Inset 7). The ROI will no longer be in bold and the blue crosshairs will reappear. Click on “Undefine the mask” in the left tab (Figure 5 – Inset 8) and the ROI will disappear.
   12. Repeat steps 9 – 11 on all the cells or regions of cells you would like to analyze in a given image. Import a new file once you are done analyzing an image. If you are sequentially analyzing the same experimental conditions then “Load fit conditions” (Figure 5 – Inset 9) by clicking on the yellow arrow in the left tab – confirm parameters in the “Multiexponential Decay” interface are appropriate and repeat steps 7 – 12 as needed.
   13. We analyzed and averaged data from 50 cells expressing mTurquoise2 to attain the lifetime (τ_D) of 3.57 ns +/− 0.07 ns (mean +/− S.D.) in Drosophila S2 cells. The τ_D value of 3573.2 ps (note SPCM64 software deals in ps rather than ns) was used for all future analyses (described below) to fit a double-exponential decay function under possible FRET conditions in which the acceptor mVenus was present.
   14. Protocol note: FLIM analysis can be done on a smaller region of the field of view by dragging in the white crosshairs from the corners of the donor image in the left window. To “Zoom In” on a highlighted region, which will be outlined by white lines, click on the “+” magnifying glass icon in the left tab (Figure 5 – Inset 10). Zooming in will speed up the calculation of the decay matrix. You can “Zoom Out” to the original field of view by clicking on the “-” magnifying glass icon in the left tab (Figure 5 – Inset 11).

Figure 5.

Lifetime analysis using SPCImage software (Becker & Hickl). This example highlights analysis of the donor mTurquoise2 in the absence of acceptor using a mono-exponential fit, which allowed us to measure the lifetime of the donor in our cell line.

2. Measuring donor lifetime and FRET efficiency in the presence of an acceptor

1. Procedure

   1. Repeat the procedure from the “Establishing donor fluorophore lifetime in the absence an acceptor/FRET” on cells expressing the FRET reporter in which both the donor (mTurquoise2) and acceptor (mVenus) are present.
2. Donor lifetime measurements via double exponential fitting of FLIM-FRET data using SPCImage

   1. Repeat steps 1 and 2 from Donor lifetime measurements using SPCImage in donor alone conditions
   2. In the “Multiexpontial Decay” interface (bottom of screen, right of the decay curve) select “2” for “Components”. Fix the long (no FRET) donor lifetime for t2[ps] by entering the lifetime measured for the donor alone in your sample of interest and checking on the “Fix” box. In our case the t2 [ps] was fixed at “3573.2” (measured above). Fix the “Scatter” at “0”. No other parameters should be fixed.
   3. Repeat steps 4 – 7 from Donor lifetime measurements using SPCImage in donor alone conditions
   4. The blue crosshairs can be positioned in either image window to assess local lifetimes. When the crosshairs are moved the lifetime value displayed above the decay curve as “tm = X ps” will change. This value is the lifetime calculated from the double exponential fit values shown in the “Multiexponential Decay” interface. In the case of the 2-exponential fit being done in this section, the interface will display the percentage of the reporter engaged in FRET as a1[%], the short lifetime of the donor (due to FRET) as t1[ps], and the percentage of reporter not engaged in FRET as a2[%]. The long lifetime t2[ps] will read the value at which it was fixed (for us this was 3573.2).
   5. Repeat steps 9 – 12 from Donor lifetime measurements using SPCImage in donor alone conditions as necessary to acquire a sufficient amount of data for your needs. FLIM-FRET lifetime measurements for the cylinB1-Cdk1 reporters in this proof of principle are reported in Table III.

3. Measuring FRET efficiency from FLIM-FRET data using SPCImage

Table III.

Lifetime measurements of cyclin B1-Cdk1 reporters in proof of concept 1.

Condition	Average Lifetime (ns)	% Short Lifetime	Short Lifetime (ns)	% Long Lifetime
Nonphospho-FRET Interphase	3.44	32%	2.04	68%
Nonphospho-FRET Mitosis	3.43	33%	2.05	67%
Phospho-FRET Interphase	3.19*	34%	1.89	66%
Phospho-FRET Mitosis	2.53*#	46%**	1.48*#	54%*#

^*indicates p < 0.05 relative to the control, non-phosphorylatable FRET reporter in interphase.

^#indicates p < 0.05 relative to the phosphorylatable FRET reporter in interphase.

Two-tailed P-values from a Student’s t-test are reported.

FRET efficiency can be calculated given the fluorescent lifetime of the donor fluorophore under both FRET and non-FRET conditions. The equation:

$$E=1-\frac{{\tau }_{\mathit{\text{FRET}}}}{{\tau }_D}\times 100\%$$

calculates FRET efficiency (E), where τ_FRET is the lifetime of the donor under FRET conditions, and τ_D is the lifetime of the donor under non-FRET conditions (measured above). FRET efficiency can be calculated manually by recording the average lifetime of your ROI (displayed as “tm = X ps” above the decay curve) and plugging this value into the equation above for τ_FRET and using the τ_D that you have measured. FRET efficiencies calculated from the τ_FRET and τ_D values measured using the protocols described above for the cylinB1-Cdk1 reporters in this proof of principle are reported in Table IV. SPCImage also has a useful function to display the decay matrix as a function of FRET efficiency using the long lifetime that you enter into the software in the double-exponential fitting interface. The following protocol discusses how to use SPCImage to convert lifetime data to FRET efficiency for visual representations.

Table IV.

FRET efficiency measurements of soluble cyclin B1-Cdk1 reporters in proof of concept 1.

Condition	n	Mean E	SD	SEM	p
Nonphospho-FRET Interphase	23	3.88	4.13	0.88	1
Nonphospho-FRET Mitosis	23	4.13	4.73	1.01	0.849
Phospho-FRET Interphase	23	10.56	9.51	2.03	0.002
Phospho-FRET Mitosis	20	29.26	12.33	2.83	<0.001

P-values calculated using the PlotsOfDifferences web app (Postma & Goedhart, 2019). N-values reported in the table apply to table III.

1. Repeat steps 1 and 2 from Donor lifetime measurements using SPCImage in donor alone conditions

2. In the “Multiexpontial Decay” interface (bottom of screen, right of the decay curve) select “2” for “Components”. Fix the long (no FRET) donor lifetime for t2[ps] by entering the lifetime measured for the donor alone in your sample of interest and checking on the “Fix” box. In our case the t2 [ps] was fixed at “3573.2” (measured above). Fix the “Scatter” at “0”. No other parameters should be fixed.

3. Select the “Options” tab at the top of the SPCImage window and click on “Color” in the drop-down. In the pop-up window entitled “Color”, click on “Continuous” and set the appropriate lifetime “Range” to encompass the expected FRET efficiency range – for the CyclinB1-Cdk1 reporter this was set from 0 (%) to 55 (%). The FRET efficiencies will be displayed in a color scale depending on whether the “Direction” is set to “R-G-B” or “B-G-R”. For example, if “R-G-B” is selected then the highest FRET efficiency (55%) will be displayed as blue and the lowest FRET efficiency (0 %) will be displayed as red, which would be inversed if “B-G-R” is selected. We chose to display the FRET efficiencies as B-G-R as this is the convention for displaying FRET data. Under “Coding of” at the bottom of the pop-up select “1 – τ_m/τ₂” in the dropdown for the “Value”. Click on “OK”.

4. Blue crosshairs will automatically center upon the brightest pixel when you import an image. Move the cursor to the dimmest area of the cell you wish to analyze. In the decay curve window below the image, adjust the binning such that the peak intensity of the decay function (bottom of screen, below FLIM image window) is at least 100 A.U. (y-axis) (Figure 5 – Inset 2).

5. At this point, if you plan on analyzing multiple images using the same analysis parameters then it is helpful to “Store the fit conditions” by clicking on the camera icon in the tab to the left of the FLIM image window (Figure 5 – Inset 3).

6. Select the “Calculate” tab at the top of the SPCImage window and click on “Decay Matrix” in the drop-down, which will generate a color-coded image in the right window in SPCImage displaying FRET efficiencies with the lowest FRET in blue and the highest FRET in red.

7. The FRET efficiencies can be analyzed with ROIs by repeating steps 9 – 12 from Donor lifetime measurements using SPCImage in donor alone conditions; however, the value reported above the decay curve will now display the FRET efficiency of the binned pixels in the ROI rather than lifetime.

B. Phasor plots

The Phasor plot originates from measuring lifetime in the frequency domain and was developed by Enrico Gratton (Digman, Caiolfa, Zamai, & Gratton, 2008). In such, the lifetime data is illustrated in a polar system in which the data is presented in a semicircle. The vector pointer (also called “phasor”) from the left hand coordinate origin is drawn with the phase as its angle and the amplitude as its length (Figure 6). The lifetime is plotted based on these two numbers independent of the pixel location within the image. The advantages are:

1. The phasors of pixels with single-exponential decay profiles are located along the semicircle

2. Phasors of multiple decay components are linear combinations of the phasors of the components and will end up inside the semicircle

3. Pixels with similar phase and amplitude values form visible clusters. Therefore, pixels with similar lifetime signatures can be identified in the phasor plot.

4. Lifetime clusters in the dataset can be back-annotated and identified in the image display using a selection tool (Figure 7).

Taking the time-domain lifetime data you get from TCSPC measurements we can obtain the phase and amplitude values by its first Fourier components.

Figure 6.

Layout and principles of a phasor plot

Figure 7.

FLIM-FRET data analysis and visual display using the phasor plot. (A) Phasor plot with both populations displayed in the FLIM image. (B) Single-exponential population (no FRET) in the phasor plot selected and displayed in the FLIM image. Note that the interphase cell is all that is displayed in the FLIM image. (C) Double-exponential population (FRET) in the phasor plot selected and displayed in the FLIM image. Note that the mitotic cell is all that is displayed in the FLIM image.

Phasor plot analysis to determine spatial distribution of fluorescent lifetimes in SPCImage

• 1.Repeat steps 1–3 from Donor lifetime measurements via double exponential fitting of FLIM-FRET data using SPCImage
• 2.Select the “Tools” tab at the top of the SPCImage window and click on “Phasor Plot” in the drop-down.
• 3.In the “Phasor Plot” GUI that opens set the “Repetition Time (Laser):” to 20.0 ns and check on the box next to “Combine with FLIM analysis.” Click on the “Recalculate” button, which will be blinking in red and read “Please Recalculate”. A pixel distribution will now appear within the phasor plot. Pixels on the semicircle are exhibiting mono-exponential decay and their position on the semicircle will indicate their lifetime, which is labeled in “ns” along the semicircle. Pixels inside the semicircle indicate that they are best fit with higher multi-exponential decay functions. In the case of FRET, these pixels should fit a double-exponential decay function.
• 2.The crosshairs can be dragged and resized with the mouse to highlight clusters of pixels that are apparent in the phasor plot (Figure 7B, C). To selectively view pixels in the ROI from the phasor plot in the FLIM image, click on the “Select cluster” box. Pixels exhibiting the lifetime within the selected circle will be displayed in the FLIM image while pixels not exhibiting the lifetime within the selected circle will be displayed in the grayscale image on the left. The image displays will be update when you move and/or resize the ROI in the phasor plot.
• 3.Another useful feature of the phasor plot analysis interface is that when you move the blue crosshair in the donor/FLIM image windows, a corresponding blue crosshair will be positioned in the phasor plot so that you see if the FLIM ROI falls into a cluster in the phasor plot.

V. Proof of concept 2 – Camui CaMKII sensor in human HEK293T cells

Ca²⁺-calmodulin dependent protein kinase II (CaMKII) is a Ser/Thr kinase that plays crucial roles in memory formation, cardiac signaling, and fertilization (Chang, Minahan, Merriman, & Jones, 2009; Eisner, Caldwell, Kistamas, & Trafford, 2017; Herring & Nicoll, 2016). As suggested by its name, CaMKII is turned on by a Ca²⁺ elevation, leading to Ca²⁺ loaded calmodulin (Ca²⁺-CaM), which ultimately leads to the relief of autoinhibition in CaMKII (Stratton, Chao, Schulman, & Kuriyan, 2013). Camui is a FRET biosensor for CaMKII activity (Takao et al., 2005). When CaMKII binds Ca²⁺-CaM, it undergoes a significant conformational change, which is reported in Camui by placing a FRET pair at the N- and C-termini (Figure 8).

Figure 8.

Camui FLIM-FRET sensor for CaMKII activation. (A) The Camui sensor is comprised of CaMKII, the domains are highlighted in the figure. In this version of Camui, there is mVenus at the N terminus and mTurquoise at the C-terminus of CaMKIIα. In the presence of Ca2+, Ca2+-bound calmodulin (CaM) binds to CaMKII and turns activity ON. This also results in a large conformational change where the kinase domain moves away from the hub domain, causing a decrease in FRET, or higher lifetime measurement. (B) Representative images of HEK293 cells expressing Camui are shown beneath the cartoon, corresponding to low lifetime (left image) and high lifetime (right image). FRET efficiencies and lifetimes are listed in the table below and numbers correspond to those ROIs highlighted in the images.

A. Expression of Camui in HEK293T cells

1. Seed 150,000 cells in a round dish (35 mm glass bottom dish with 20 mm micro-well #1 cover glass, D35-20-1-N) with 1 mL of Dulbecco’s Modified Eagle’s Medium – high glucose (Sigma Aldrich) + 10% of Fetal Bovine Serum (Sigma Aldrich).

2. Once the cells are 60–70 % confluent (typically after 2–3 days), transfect with 600–700 ng of Camui (mTurquoise2/mVenus) using Lipofectamine (Thermo Fischer) according to the manufacturers protocol.

3. Subject Camui expressing cells to FLIM imaging as described in the procedure from “Establishing donor fluorophore lifetime in the absence an acceptor/FRET” ~48 hrs post-transfection.

B. Establishing timing of Ca²⁺ release in HEK293T cells by addition of ionomycin

Ca²⁺ entry into the cell leads to CaMKII activation. We use ionomycin to induce Ca²⁺ release in HEK293T cells. In order to determine the kinetics of Ca²⁺ release after ionomycin addition, we used Fluo-4 AM, a cell permeant Ca²⁺ indicator. Robust Ca²⁺ release occurred roughly 40 sec after addition of 2.5 μM ionomycin.

C. Measuring CaMKII activity using Camui

As described above, Camui is a biosensor for CaMKII activation where there is a FRET decrease coincident with CaMKII activation. Prior to Ca²⁺ release, CaMKII should be in the OFF conformation and this is read out as a short lifetime from Camui. After ionomycin addition and subsequent Ca²⁺ release, CaMKII should be in the ON conformation and this is read out as a long lifetime from Camui. To provide a direct comparison to the experiment described above, we cloned mTurquoise2 into Camui in place of Cerulean. Furthermore, this version of Camui is better suited to FLIM analysis since mTurquoise2 exhibits a mono-exponential decay in its lifetime while Cerulean does not.

Data acquisition and analysis were performed as described above. We acquired 60 million photons before ionomycin addition, then ionomycin was added and after 40 seconds we began another acquisition for 60 million photons. The only change compared to the proof of concept 1 was in establishing the color display to a 600 ps – 4000 ps range.

VI. Utilizing the Nikon A1-FLIM platform

Nikon and Becker & Hickl have collaborated to integrate the Becker & Hickl hardware into a seamless module that is now part of Nikon’s software, NIS-Elements. The A1-FLIM module combines advanced lifetime imaging technology with the software platform used to drive conventional scanning confocal systems, one that many Nikon microscope users are fully comfortably. The result is the ability to use typical confocal system workflows to collect, visualize, and analyze lifetime data. The module allows for multiplexed FLIM acquisition, providing Z-stacks, time-lapses, multi-points, or any combination therein. All experiments are set up like normal confocal imaging on a Nikon system using NIS-Elements. Thus, anyone familiar with a Nikon A1 confocal system can begin to carry out lifetime experiments without specialized training.

To begin an experiment, one must first consider confocal imaging parameters including pinhole, image speed and size and laser power. Once this is set, there are two settings necessary for lifetime imaging; the capture mode and the time channels. Capture mode can be either a set number of frames or can be dynamic based on the number of photons collected in a given pixel. Generally, better contrast can be achieved with more data; however, the cost of doing so is time. Once these parameters are decided, one simply needs to press the “capture” button in order to acquire lifetime images.

VII. Conclusions

The hardware and protocols outlined here are meant to serve as an entry point for researchers to begin thinking about how FLIM-FRET could benefit their research program and, if it is deemed a worthwhile pursuit, how to assemble/purchase a system in their lab or imaging facility. While we described Becker & Hickl TCSPC components connected to a Nikon scanning confocal, other commercially available FLIM systems are available from PicoQuant Photonics and as upgrade packages from major microscope companies such as Leica, Olympus, and Zeiss and we hope this chapter will inform comparison shopping between the available options. It is also noteworthy that, in addition to FLIM-based applications, these systems provide Fluorescence Correlation Spectroscopy (FCS) capabilities.

In conclusion, FLIM-FRET offers significant improvements over conventional intensity-based FRET approaches. Furthermore, creative FLIM-FRET applications that would be impossible to accomplish with intensity-based FRET are and will continue to arise. For example, the development of large Stoke’s shift (LSS) and “dark” fluorescent proteins (H. Murakoshi, A. C. Shibata, Y. Nakahata, & J. Nabekura, 2015; Shcherbakova, Hink, Joosen, Gadella, & Verkhusha, 2012) have facilitated the application of dual FLIM-FRET applications in which multiple FRET-based sensors can be subjected to FLIM simultaneously (Demeautis et al., 2017; Ringer et al., 2017). In this method, a single wavelength (~440 nm) can be used to excite two separate mono-exponential donors such as the CFP variants mTurquoise2 or mTFP1 and LSSmOrange. The mTurquoise2 is paired with a “dark” acceptor such as the GFP variant ShadowG, which has a 120-fold lower quantum yield than EGFP and; therefore, emits very few photons when energy is transferred to it from a donor, while the LSSmOrange is paired with mKate2 as an acceptor. Since there is no overlap in the emission spectra of mTurquoise2/mTFP1 and LSSmOrange the lifetimes of each donor can be measured simultaneously with two TCSPC detectors following excitation with the 445 nm pulsed laser. In the system described in this chapter this would simply require changing the dichroic mirror and bandpass (BP) filters in the TCSPC module and a dichroic mirror in the confocal scanning microscope. When acquiring dual FLIM data in a modified setup, one window in the SPCM64 interface will display the mTurquoise2 signal and the other window will display the LSSmOrange signal. Thus, analysis of each donor can be conducted in SPCImage as described above to measure lifetimes and FRET efficiencies of two FRET sensors in the same cell simultaneously. The fast pace at which TCSPC and scanning confocal hardware is improving combined with the continued development of new fluorescent proteins and dye-based technologies means that significant advancements in FLIM-FRET in the vein of dual FLIM imaging will continue for years to come. We hope this chapter will enable cell biologists to take advantage of these emerging technologies by incorporating FLIM-FRET into their research programs.