Rhodopsin pp 205-219 | Cite as

Detection of Rhodopsin Dimerization In Situ by PIE-FCCS, a Time-Resolved Fluorescence Spectroscopy

Protocol
Part of the Methods in Molecular Biology book series (MIMB, volume 1271)

Abstract

Rhodopsin self-associates in the plasma membrane. At low concentrations, the interactions are consistent with a monomer-dimer equilibrium (Comar et al., J Am Chem Soc 136(23):8342–8349, 2014). At high concentrations in native tissue, higher-order clusters have been observed (Fotiadis et al., Nature 421:127–128, 2003). The physiological role of rhodopsin dimerization is still being investigated, but it is clear that a quantitative assessment is essential to determining the function of rhodopsin clusters in vision. To quantify rhodopsin interactions, I will outline the theory and methodology of a specialized time-resolved fluorescence spectroscopy for measuring membrane protein-protein interactions called pulsed-interleaved excitation fluorescence cross-correlation spectroscopy (PIE-FCCS). The strength of this technique is its ability to quantify rhodopsin interactions in situ (i.e., a live cell plasma membrane). There are two reasons for restricting the scope to live cell membranes. First, the compositional heterogeneity of the plasma membrane creates a complex milieu with thousands of lipid, protein, and carbohydrate species. This makes it difficult to infer quaternary interactions from detergent solubilized samples or construct a model phospholipid bilayer that recapitulates all of the interactions present in native membranes. Second, organizational structure and dynamics is a key feature of the plasma membrane, and fixation techniques like formaldehyde cross-linking and vitrification will modulate the interactions.

PIE-FCCS is based on two-color fluorescence imaging with time-correlated single-photon counting (TCSPC) (Becker et al., Rev Sci Instrum 70:1835–1841, 1999). By time-tagging every detected photon, the data can be analyzed as a fluorescence intensity distribution, fluorescence lifetime histogram, or fluorescence (cross-)correlation spectra (FCS/FCCS) (Becker, Advanced time-correlated single-photon counting techniques, Springer, Berlin, 2005). These analysis tools can then be used to quantify protein concentration, mobility, clustering, and Förster resonance energy transfer (FRET). In this paper I will focus on PIE-FCCS, which interleaves two wavelength excitation events in time so that the effects of spectral cross-talk and FRET can be isolated. In this way it is possible to characterize monomer-dimer-oligomer equilibria with high accuracy (Müller et al., Biophys J 89:3508–3522, 2005). Currently, PIE-FCCS requires a customized equipment configuration that will be described below. There is an excellent protocol that outlines traditional FCCS on a commercially available instrument (Bacia and Schwille, Nat Protoc 2:2842–2856, 2007). The PIE-FCCS approach is a relatively recent advance in FCCS that has been used in live cell assays to quantify lipid-anchored protein clustering (Triffo et al., J Am Chem Soc 134:10833–10842, 2012), epidermal growth factor receptor dimerization (Endres et al., Cell 152:543–556, 2013), and recently the dimerization of opsin (Comar et al., J Am Chem Soc 136(23):8342–8349, 2014). This paper will outline the theory and instrumentation requirements for PIE-FCCS, as well as the data collection and analysis process.

Key words

Rhodopsin dimerization Membrane protein dynamics Fluorescence correlation spectroscopy Time-correlated single-photon counting Pulsed-interleaved excitation 

1 Introduction

Fluorescence microscopy is widely used in bio-imaging because of its high contrast, molecular specificity, and live cell/animal compatibility. Image contrast is generated by the Stokes shift, which splits the energy required to excite a fluorophore from the energy emitted as a photon when the fluorophore relaxes to the ground state. Using dichroic mirrors these energies are spatially separated so that the background light reaching the detectors is very low. In this way the signal-to-noise ratio can be quite high even for low numbers of molecules by using high-sensitivity cameras and point detectors now readily available. Molecular specificity can be achieved through several methods, but to simplify the discussion I will focus on fluorescent fusion proteins that are co-expressed with the protein of interest [9].

In fluorescence correlation spectroscopy and related methods, time-dependent fluctuations in fluorescence intensity are analyzed to extract microscopic parameters from the system [10]. A central requirement for these methods is that the amplitude of the fluctuations be significant compared to the average fluorescence intensity. This is achieved by observing relatively low fluorophore concentrations in small volumes. Experimentally, concentration is controlled by the expression level of the fluorescent fusion protein in the cell, and volume is limited via confocal detection.

In a confocal detection geometry, the fluorescence intensity at a specific detector is denoted F(t). For this time-dependent variable, a temporal autocorrelation function can be calculated:
$$ C\left(\tau \right)=\frac{1}{T}\underset{0}{\overset{T}{{\displaystyle \int }}}F(t)\cdot F\left(t+\tau \right)\mathrm{d}t=\left\langle F(t)\cdot F\left(t+\tau \right)\right\rangle $$
At τ = 0, the correlation function is equal to the mean square value of the fluorescence intensity:
$$ C\left(\tau =0\right)=\left\langle F(t)\cdot F\left(t+0\right)\right\rangle $$
$$ C(0)=\left\langle F{(t)}^2\right\rangle $$
At long times, the correlation function is equal to the square of the mean value:
$$ \underset{\tau \to \infty }{ \lim }C\left(\tau \right)=\left\langle F(t)\right\rangle \left\langle F(t)\right\rangle ={\left\langle F(t)\right\rangle}^2 $$
The difference between these two values is the variance of the fluorescence intensity:
$$ {\sigma}^2=\left\langle F{(t)}^2\right\rangle -{\left\langle F(t)\right\rangle}^2 $$
Thus, the correlation function reflects the magnitude of the variance and how the microscopic variables that cause fluorescence fluctuations are averaged out over time.
In practice, the correlation function is defined using the fluctuation of fluorescence in channel i around the average value.
$$ \delta {F}_i(t)={F}_i(t)-\left\langle {F}_i(t)\right\rangle $$
The autocorrelation function or fluorescence correlation spectrum is defined as
$$ {G}_i\left(\tau \right)=\frac{\left\langle \delta {F}_i(t)\cdot \delta {F}_i\left(t+\tau \right)\right\rangle }{{\left\langle {F}_i(t)\right\rangle}^2} $$
In this formulation, Gi(0) is directly proportional to the variance of the fluorescence intensity fluctuations. Over a large range of time scales, fluorescence fluctuations are dominated by movement of molecules in and out of the confocal volume. This means that low numbers of molecules will exhibit large fluctuations (high variance), whereas large numbers of molecules will exhibit smaller fluctuations (low variance). The decay of the correlation function is the timescale over which these fluctuations are averaged out and is directly related to the diffusion coefficient of the molecules.
For a single species diffusing in three dimensions, the autocorrelation function can be fit to the following model function:
$$ {G}_i\left(\tau \right)=\frac{1}{N_i}\cdot \frac{1}{1+\frac{\tau }{\tau_{\mathrm{D},i}}}\cdot \frac{1}{\sqrt{1+\frac{\omega_0^2\cdot \tau }{z_0^2\cdot {\tau}_{\mathrm{D},i}}}} $$
Here, Ni is the average number of molecules in the confocal volume, τD is the average dwell time of the molecules in the confocal volume, ω0 is the lateral radius, and z0 is the axial radius of the confocal volume. In this form the effective confocal volume is
$$ {V}_{\mathrm{eff}}={\omega}_0^2\cdot {z}_0\cdot {\pi}^{3/2} $$
For the experiments described in Subheading 3, two corrections to the model function are necessary. First, membrane protein diffusion is restricted to two dimensions, so the third term in the model function is dropped. Second, fluorescent proteins display complex photophysics that can be fit with an expression for triplet relaxation, where T is the fraction of molecules in the triplet state and τT is the relaxation time for the triplet state:
$$ {G}_i\left(\tau \right)=\frac{1}{N_i}\cdot \frac{1-T+T{\mathrm{e}}^{-\tau /{\tau}_{\mathrm{Ti}}}}{1-T}\cdot \frac{1}{1+\raisebox{1ex}{$\tau $}\!\left/ \!\raisebox{-1ex}{${\tau}_{\mathrm{D},i}$}\right.} $$
Autocorrelation curves of membrane proteins can therefore be used to determine the local concentration, or surface density of the protein and the mobility, or effective diffusion constant of the proteins.
In the experiments described below, rhodopsin proteins are expressed as two populations—one fused to the green fluorescent protein, eGFP, and the other with a red fluorescent protein, mCherry. The fluorophores are excited by pulsed 488 and 561 nm laser light and fluorescence from each population is split by a dichroic mirror onto two detector channels. Channel A is filtered for “red” detection (λ ≈ 615 nm) and channel B is filtered for “green” detection (λ ≈ 520 nm). In a traditional fluorescence cross-correlation study, all of the photons recorded by the detector are used to calculate the fluorescence intensity function, FR(t) or FG(t) [6]. And the cross-correlation spectrum will be calculated according to
$$ {G}_{\mathrm{R}\mathrm{G}}\left(\tau \right)=\frac{\left\langle \delta {F}_{\mathrm{R}}(t)\cdot \delta {F}_{\mathrm{G}}\left(t+\tau \right)\right\rangle }{\left\langle {F}_{\mathrm{R}}(t)\left\rangle \cdot \right\langle {F}_{\mathrm{G}}(t)\right\rangle } $$
In PIE-FCCS, each of the photons is sorted into two time gates according to their arrival time relative to a synchronization pulse that clocks the arrival time of the 488 and 561 nm laser pulses, which are delayed ~50 ns with respect to each other (Fig. 1) [5]. This makes it possible to define two time gates, 0 and 1, that divide photons emitted after a 488 nm laser pulse from those emitted after a 561 nm laser pulse. This interleaved pulse scheme allows for the separation of several complicating events that arise from the broad excitation/emission spectra of fluorescent proteins. As seen in the schematic below, a 488 nm laser pulse will ideally excite GFP, which will emit in a photon detected in channel B. However, there is some probability that GFP will emit a photon that is detected by channel A and that a 488 pulse will directly excite mCherry. In addition, it is possible for a GFP in the excited state to resonantly transfer its energy to an mCherry protein, which will then emit a photon detected by Channel A. These processes can lead to excess photons in the red channel that complicate the number analysis and lead to false-positive cross-correlation (Fig. 2).
Fig. 1

Schematic of PIE-FCCS experiments. (a) Confocal detection near a membrane. Drawing of rhodopsin diffusion in a membrane and the confocal detection volume. (b) Photon counting events. In PIE-FCCS each detected photon is time-stamped with its absolute arrival time and the delay time, δτ, of its arrival with respect to the synchronization pulse. In this way the data is filtered to ignore photons resulting from spectral cross-talk and FRET in the PIE-FCCS analysis. (c) PIE-FCCS for DNA standards. Example data for two DNA standards. On the left is data for a mixture of noninteracting single-strand DNA molecules labeled with TAMRA or FAM. On the right is data for a dual-labeled single strand of DNA, which serves as a positive control for cross-correlation. Panels a and b are adapted with permission from Comar, W.D., Schubert, S.M., Jastrzebska, B., et al., “Time-resolved fluorescence spectroscopy measures clustering and mobility of a G protein-coupled receptor opsin in live cell membranes,” J Am Chem Soc. Copyright (2014) American Chemical Society

Fig. 2

Schematic of complications in a two-color fluorescence experiment

In PIE-FCCS, these processes can be separated using time gates around the laser pulse arrival times [3, 4, 5]. Only photons arriving in channel A during time gate 1 and channel B during time gate 0 are used to calculate the auto- and cross-correlation functions:
$$ {G}_{\mathrm{A}1}\left(\tau \right)=\frac{\left\langle \delta {F}_{\mathrm{A}1}(t)\cdot \delta {F}_{\mathrm{A}1}\left(t+\tau \right)\right\rangle }{{\left\langle {F}_{\mathrm{A}1}(t)\right\rangle}^2} $$
$$ {G}_{\mathrm{B}0}\left(\tau \right)=\frac{\left\langle \delta {F}_{\mathrm{B}0}(t)\cdot \delta {F}_{\mathrm{B}0}\left(t+\tau \right)\right\rangle }{{\left\langle {F}_{\mathrm{B}0}(t)\right\rangle}^2} $$
$$ {G}_{\mathrm{A}\mathrm{B}}\left(\tau \right)=\frac{\left\langle \delta {F}_{\mathrm{A}1}(t)\cdot \delta {F}_{\mathrm{B}0}\left(t+\tau \right)\right\rangle }{\left\langle {F}_{\mathrm{A}1}(t)\left\rangle \cdot \right\langle {F}_{\mathrm{B}0}(t)\right\rangle } $$
Each of these is fit with the model function described earlier, except for GAB(τ), for which the triplet relaxation term is omitted.
The three correlation curves produce a large set of well-defined parameters related to the microscopic state of the system. These parameters are described in Table 1.
Table 1

Parameters obtained from the raw and fitted PIE-FCCS data

Parameter

Description

NR

This is equal to 1/GA1(0). For a monomeric population, NR is equal to the average number of red (mCherry)-labeled molecules. If dimers or larger oligomers are present, NR is equal to the number of diffusing species that contain a red-labeled molecule

NG

This is equal to 1/GB0(0). Similar to NR, NG is equal to the average number of green (eGFP)-labeled molecules for a monomeric population. If dimers or larger oligomers are present, NG is equal to the number of diffusing species that contain a green-labeled molecule

NAB

This is simply equal to 1/GAB(0) and is not to be confused with the total number of double-labeled species, which is calculated in Table 2

τD,i

Average time spent in the detection volume. This is a direct measure of the average mobility of the green-labeled \( \left(i=\mathrm{G}\right) \), red-labeled \( \left(i=\mathrm{R}\right) \), or double-labeled species \( \left(i=\mathrm{AB}\right) \)

cpsi

cpsi or counts per second, is the photon detection rate and is obtained by dividing the total number of photons collected in channel i by the total collection time

2 Materials

2.1 Microscope Instrumentation

  1. 1.
    Microscope: In the methods below we use a Nikon Eclipse Ti microscope system, but any research-grade inverted microscope is sufficient. Rather than purchase a commercial confocal system from Nikon or another supplier, we use custom-built laser excitation and confocal fluorescence detection systems that will be described later. Some useful microscope options include:
    1. (a)

      Focus drift correction system: This is especially important for membrane measurements, which are sensitive to drifts in the focal plane.

       
    2. (b)

      Motorized stage.

       
    3. (c)

      At least two 100 % output ports.

       
    4. (d)

      Two dichroic filter block turrets. In the Nikon Eclipse the upper turret can be used as the input port for laser excitation.

       
     
  2. 2.
    Laser source: For PIE-FCCS, two pulsed light sources are required. This can be achieved with electro-optic gating of continuous wave lasers [5] or with short pulse diode or mode-locked lasers [1, 7, 8, 11]. The latter will provide a simultaneous measure of fluorescent lifetime and are used in the instrument below. In our experience the supercontinuum source based on a mode-locked fiber laser is the best current choice because of the short pulse duration and wide spectral flexibility [1]. Below are several features to consider in a supercontinuum source:
    1. (a)

      Several companies now sell suitable laser systems including NKT photonics, Fianium Inc., and Toptica.

       
    2. (b)

      Base repetition rates for the laser sources is ~80 MHz, so an internal pulse picker is required to achieve a repetition rate of 10 MHz, which is ideal for the PIE-FCCS experiments.

       
    3. (c)

      Spectral selection can be achieved with an acousto-optic tunable filter often available as an integrated add-on to the system.

       
    4. (d)

      Spectral selection can also be achieved with a series of dichroic beamsplitters and filters. In either case, cleanup filters are needed to eliminate any spectral leakage away from the desired excitation bands. In the instrument described below, we use the following two filters: a 488 nm filter with a 1.9 nm full width half max (FWHM) bandwidth (LL01-488-12.5, Semrock) and a 561 nm filter with a 2.1 FWHM bandwidth (LL02-561-12.5, Semrock).

       
     
  3. 3.

    Pulse-delay generator: If one is using a supercontinuum source for both excitation beams, the pulses arrive coincident in time and a pulse-delay generator is needed. To achieve this we direct the 488 nm beam through a 3 m single mode fiber and the 561 nm pulse through an 18 m fiber (QPMJ-3AF3U-488-3.5/125-3AS-18-1-SP and QPMJ-3AF3U-488-3.5/125-3AS-3-1-SP, OZ Optics).

     
  4. 4.

    Beam combiner: To spatially overlap the beams, we use a 503 nm cut-off dichroic beamsplitter (LM01-503-25, Semrock).

     
  5. 5.

    Laser power adjustment: variable neutral density filters are available from several optomechanics supplies companies.

     
  6. 6.

    Microscope dichroic block: Two-color dichroic mirror and laser blocking filter (zt488/561rpc and zet488/561m, Chroma Technology).

     
  7. 7.

    Objective: 100× TIRF objective, NA 1.49, (Nikon).

     
  8. 8.

    Incubator stage: To maintain cells at 37 °C use a stage-top incubator (Chamlide IC, Quorum Technologies).

     
  9. 9.

    Confocal pinhole: 50 μm confocal pinhole (Thorlabs).

     
  10. 10.
    Detection dichroic beamsplitter and filters:
    1. (a)

      560 nm longpass dichroic filter (FF560-FDi01-25x36, Semrock).

       
    2. (b)

      612/69 nm filter (FF01-621/69-25, Semrock).

       
    3. (c)

      520/44 nm filter (FF01-520/44-25, Semrock).

       
     
  11. 11.

    Avalanche photodiodes: single-photon avalanche diode (SPAD) with a 50 μm active area, 30 ps timing resolution, and 25 dark counts per second (Micro Photon Devices).

     
  12. 12.

    Four-channel-routed time-correlated single-photon counting (TCSPC) device: PicoHarp 300 (PicoQuant).

     
  13. 13.

    Camera for cell identification and localization: Evolve 512 EMCCD (Photometrics).

     
  14. 14.

    Laser power meter with μW sensitivity.

     

2.2 Data Processing

  1. 1.

    Computer should have a modern processor and 16–32 GB memory for data analysis.

     
  2. 2.
    MatLab scripts (available upon request from the author):
    1. (a)

      Convert *.pt3 file (PicoQuant) to MatLab data format file (*.mat).

       
    2. (b)

      Implement time gate and sort *.mat single-photon data to intensity file.

       
    3. (c)

      Calculate red and green autocorrelation spectra and the cross-correlation spectrum for each data set.

       
     
  3. 3.

    Additional files for averaging and displaying data.

     

2.3 Control Samples

Several organic dyes and dye-labeled systems are useful for instrument alignment and as a control to verify proper alignment. These include:
  1. 1.

    A free dye of known diffusion coefficient. These are used to calibrate the confocal detection volume and should be chosen to match the absorbance/emission of the dyes of interest. Two examples are fluorescein and Alexa Fluor 568.

     
  2. 2.

    Positive control for dimerization. It is valuable to have a molecular system that is labeled with a red and a green dye to serve as a positive control for cross-correlation. The most common example of this is a single strand of DNA labeled with a red dye on the 5′ end and green dye on the 3′ end. The requirements for this are that the DNA strand should contain enough bases to minimized FRET and not form any secondary structure. A commercial version of this is available (e.g., Cat. #5-0000-604, IBA GmbH). We have also used the following custom synthesized sequence, which showed superior cross-correlation compared to the IBA standards in our hands: 5′-/56-TAMN/CCC TAG AGT GAG TCG TAT GAT AGT GAC AGC TGG ATC GTT AC/36-FAM/-3′.

     
  3. 3.

    In cellulo controls: In recent live cell work, it has been shown that live cell controls are necessary because of the complications arising from the photophysics of fluorescent proteins and the statistics of dimerization [8]. Plasmids for these constructs are available upon request from the author.

     
  4. 4.
    For the alignment procedure below, prepare the following solutions:
    1. (a)

      Mixture of 2 μM fluorescein and 2 μM Alexa Fluor 568.

       
    2. (b)

      100 nM solution of the dual-labeled DNA sample.

       
     

2.4 Cell Culture Requirements

  1. 1.

    Cell line: COS-7 (kidney fibroblast), African green monkey (Cercopithecus aethiops), available from ATCC (CRL-1651).

     
  2. 2.

    Media: 1× Dulbecco’s Modified Eagle’s Medium (DMEM) + GlutaMAX (Life Technologies) supplemented with 10 % fetal bovine serum (FBS) (Life Technologies) and 1 % penicillin/streptomycin (BioReagent, Sigma-Aldrich). Opti-MEM I media without phenol red (Life Technologies).

     
  3. 3.

    Plasmids: pEGFP-N3 and pmCherry-N1 original vectors (Clontech) and constructs encoding mouse opsin gene (AAH31766) as a fusion with eGFP or mCherry, respectively.

     
  4. 4.

    Lipofectamine 2000 transfection reagent (Life Technologies).

     
  5. 5.

    Plastic ware: 35 × 10 mm uncoated glass bottom dishes with #1 coverslips (MatTek).

     

3 Methods

3.1 Instrument Setup

  1. 1.

    Configure supercontinuum source to emit light with center wavelengths 488 and 561 nm for eGFP and mCherry excitation. Position bandpass filters listed in Subheading 2 to clean up the excitation spectra.

     
  2. 2.

    Set laser repetition rate to around 10 MHz with internal pulse picker.

     
  3. 3.

    Spatially separate beams with dichroic beamsplitter.

     
  4. 4.

    Align the 488 nm beam to the 3 m fiber and the 561 nm beam to the 18 m fiber using commercial fiber aligners.

     
  5. 5.

    Decouple the fibers with commercial fiber aligners near the microscope using a high-quality apochromatic lens.

     
  6. 6.

    After exiting the fibers, overlap both beams using mirrors and a laser mixing filter (see Subheading 2).

     
  7. 7.

    Align overlapped beam to the excitation pathway. For the Nikon Eclipse Ti, this is done by directing the beams into the rear of the microscope to the top dichroic filter turret, which houses the two-line laser dichroic filter block described in Subheading 2. The lasers are aligned to the optical axis using the two preceding pointing mirrors.

     
  8. 8.

    Position the confocal pinhole at the image plane of one of the side ports of the microscope.

     
  9. 9.

    Position a lens after the pinhole to collimate the light.

     
  10. 10.

    Split off the green light with a longpass dichroic mirror.

     
  11. 11.

    Position bandpass filters for green and red beams.

     
  12. 12.

    Position lenses to focus light onto each of the SPAD detectors. In our current setup, we include one adjustable mirror between the lens and the detector to align the beam to the active area of the detector. In other setups, the detector itself is on a movable stage, eliminating the need for an additional mirror.

     

3.2 Instrument Alignment

  1. 1.

    Set laser powers to ~10 μW by measuring with laser power meter before light enters the microscope path.

     
  2. 2.

    Position fluorescein/Alexa Fluor 568 sample on the stage and set the focal plane ~5 μM above the glass slide.

     
  3. 3.

    Configure the microscope settings so that the laser is illuminating the sample and the light path is directed to the camera.

     
  4. 4.

    Confirm that the lasers are illuminating the sample within the camera field of view.

     
  5. 5.

    Adjust laser trajectory so that the illumination spot is a symmetric Gaussian shape.

     
  6. 6.

    Confirm that the 488 and 561 lasers are spatially overlapped to within the resolution of the camera. Adjust relative positions using mirrors not common to both beams.

     
  7. 7.

    Minimize the size of each laser illumination region in the camera using the fiber output couplers.

     
  8. 8.

    Note the pixel position of laser illumination. If this is a repeated alignment, confirm that the spot is near an earlier position. Some drift is tolerable, but a large displacement will make realignment of the SPAD positions difficult.

     
  9. 9.

    Direct microscope light path to the confocal detection port.

     
  10. 10.

    Turn on TCSPC instrument and SPAD detectors.

     
  11. 11.

    With confocal pinhole removed, align the SPADs in axial and lateral dimensions to maximize photon counts. (If no signal is visible, seeNote1.)

     
  12. 12.

    Replace pinhole and adjust axial and lateral positions of pinhole to maximize the signal on the SPADs.

     
  13. 13.

    Remove free dye solution from the sample stage and place 100 nM DNA sample. Set focus to ~5 μm above the glass-water interface.

     
  14. 14.

    Collect TCSPC data with PicoQuant software in full time-tagged mode (*pt3 files). For the dilute solutions of the FCCs control, the ideal data collection time is 30–60 s repeated five to ten times. With proper alignment, this typically gives PIE-FCCS data with low noise.

     

3.3 Cell Culture and Preparation

  1. 1.

    Cells are cultured in 100 × 20 mm petri dishes with 1× DMEM + GlutaMAX supplemented with 10 % FBS and 1 % penicillin/streptomycin. Cells are routinely split upon reaching 80–90 % confluency.

     
  2. 2.

    Three days prior to imaging, split cells and seed into 35 × 10 mm glass bottom dishes.

     
  3. 3.

    One day prior to imaging, co-transfect cells with control plasmids or plasmids encoding fusion opsin-mCherry and opsin-EGFP. Since low concentrations are ideal for FCCS, we find that the best DNA transfection volume is about ten times lower than for typical transfection protocols.

     
  4. 4.

    At least 30 min prior to imaging, exchange media to one without phenol red.

     

3.4 Cell PIE-FCCS Data Collection

  1. 1.

    Make sure TCSPC instrument and SPAD detectors are on and aligned according to Subheading 3.2.

     
  2. 2.

    If using stage-top incubator, position it and an objective warmer and allow them to temperature stabilize for about 30 min.

     
  3. 3.

    Set laser powers to between 0.4 and 2.0 μW. Note power and keep consistent throughout experiments. The specific power depends on fluorophore, protein mobility and clustering, and instrumentation. Choose powers that are high enough for good signal to noise and low enough to avoid significant photobleaching over the ~2 min illumination of the cell.

     
  4. 4.

    Load cultured transfected cells in phenol-red-free media into the stage-top incubator.

     
  5. 5.

    Set focal plane at glass-water interface.

     
  6. 6.
    Using wide-field fluorescence imaging with camera, raster scan the stage until a cell is found with relatively low level of brightness (seeNote2 and Fig. 3a).
    Fig. 3

    Example data for live cells expressing opsin. (a) An epi-fluorescence image of an opsin-eGFP-expressing Cos-7 cell is merged with an image of fluorescence excited by the laser used for PIE-FCCS (scale bar = 5 μm). The arrow points to the laser illumination area shown in green, which has a radius of ~220 nm. (b) Representative PIE-FCCS data for two live cells expressing opsin-eGFP and opsin-mCherry. In each plot, colored dots are the measured data points, whereas the solid black lines indicate the fitted functions. Red dots are the FCS data for the mCherry fusion protein, GA1(τ); green dots are the FCS data for the eGFP fusion, GB0(τ); and blues dots are the FCCS data, GAB(τ). In each plot, a horizontal dashed line marks the zero value for comparison with the cross-correlation amplitude, GAB(τ). Figure and caption is adapted with permission from Comar, W.D., Schubert, S.M., Jastrzebska, B., et al., “Time-resolved fluorescence spectroscopy measures clustering and mobility of a G protein-coupled receptor opsin in live cell membranes,” J Am Chem Soc. Copyright (2014) American Chemical Society

     
  7. 7.

    Using the stage controller, position cell over the pixel area where the laser illumination has been aligned (see Subheading 3.2, step 8, and Note3).

     
  8. 8.

    Change microscope setting to the multi-line laser dichroic and illuminate cell with both lasers. View laser-induced fluorescence on camera and fine-tune the focus to minimize the spot size (see Fig. 3a, laser illumination spot).

     
  9. 9.

    Change microscope light path to the confocal detection box.

     
  10. 10.

    With lasers on, collect several 10–20 s TCSPC data sets. We find that ~5 repeats are sufficient for robust fitting and data analysis. If using the PicoHarp TCSPC system, collect data as *.pt3 files using the correlator option. This allows you to monitor the FCS data in real time but saves the raw photon data for PIE correction later.

     
  11. 11.

    After collecting TCSPC data, record and save fluorescence images of both eGFP and mCherry. This can be done before and after FCS data collection to ensure that no significant photobleaching or photodamage has occurred within the cell. If there is damage, adjust the laser power.

     

3.5 Data Analysis

  1. 1.

    After collecting TCSPC data, it needs to be parsed before calculating the FCS and FCCS curves. This can be done via custom MatLab routines available from the author.

     
  2. 2.

    Choose the time delays that define the 0 and 1 time gates. For the analysis shown below, two intensity traces are constructed. The first includes photons arriving in the red channel during time gate 1 (~1 ns before and ~40 ns after the arrival of the 561 nm laser pulse). The second includes photons arriving during time gate 0 (~1 ns before and ~40 ns after the arrival time of the 488 nm laser pulse).

     
  3. 3.

    Once the intensity traces are constructed, calculate the auto- and cross-correlation functions as defined in the introduction (Fig. 3b).

     
  4. 4.

    The resulting curves are then fit using a nonlinear least squares minimization routine to the functions described in the Introduction. In practice, several FCS curves from the same position in the same cell can be averaged together for a better fit (seeNote4, and Fig. 3b).

     
  5. 5.
    The fit parameters can then be used to calculate observables like those shown in Table 2.
    Table 2

    Observables that can be calculated from the parameters in Table 1

    Equation

    Description

    \( {N}_{\mathrm{X}}=\frac{G_{\mathrm{A}\mathrm{B}}(0)}{G_{\mathrm{A}1}(0)\cdot {G}_{\mathrm{B}0}(0)} \)

    NX is the number of diffusing species that have at least one eGFP and one mCherry probe

    \( {f}_{\mathrm{c}}=\frac{N_{\mathrm{X}}}{N_{\mathrm{R}\;\mathrm{or}\;\mathrm{G}}} \)

    fc is the fraction correlated. It reports on the how many of the red-labeled species co-diffuse with a green-labeled molecule. It varies from 0 to 1

    \( {\eta}_i=\frac{N_i}{{\mathrm{cps}}_i} \)

    ηi is called the molecular brightness and has units of counts per molecule per second (cpms). For a monomeric population it can reveal the number of photons each fluorophore emits per unit time. For an oligomeric population, it will be proportional to the oligomer size

    \( {D}_{\mathrm{eff},i}=\frac{\omega_{0,i}^2}{4\cdot {\tau}_{\mathrm{D},i}} \)

    The effective diffusion coefficient, Deff,i, is calculated by using the fit mobility, τD,i, and the calibrated radius of the detection volume, ω0,i. For model systems, this is equal to the true diffusion coefficient, but for live cell membranes, the mobility is not purely Brownian and is thus referred to as the effective diffusion coefficient

    \( {C}_{\mathrm{X}}=\frac{N_{\mathrm{X}}}{\pi \cdot {\omega}_{0,\mathrm{G}}^2} \)

    CX is the area concentration of dual-labeled complexes

    \( {C}_i=\frac{N_i-{N}_{\mathrm{X}}}{\pi \cdot {\omega}_{0,i}^2} \)

    Ci is the area concentration of single-labeled species \( \left(i=\mathrm{R}\;\mathrm{or}\;\mathrm{G}\right) \)

     

4 Notes

  1. 1.
    If no light is detected by SPADs during early alignment, follow this procedure:
    1. (a)

      Turn off and unplug SPADs to avoid irreversible photodamage.

       
    2. (b)

      Replace sample with a silver mirror or aluminum-coated glass coverslip.

       
    3. (c)

      Oil between the objective and mirror is necessary, so re-use of the same mirror is suggested.

       
    4. (d)

      Remove laser blocking filter (not dichroic mirror) from the two-line laser dichroic block.

       
    5. (e)

      Direct reflected laser light to confocal detection box.

       
    6. (f)

      Remove pinhole.

       
    7. (g)

      Adjust focus so that the reflected laser beam comes to a focus at the confocal pinhole position.

       
    8. (h)

      Follow beam and adjust mirrors to direct it to the active area of the SPADs.

       
    9. (i)

      Replace pinhole and adjust until some light passes through.

       
    10. (j)

      Replace laser blocking filter in the dichroic block and remove the mirror from the sample stage.

       
    11. (k)

      Align sample as above, Subheading 3.2, steps 212.

       
     
  2. 2.

    To gauge the appropriate level of fluorescence, take FCS data as described in Subheading 3.4, step 10, checking to see if the autocorrelation amplitude is between 0.005 and 0.5. Assuming constant imaging conditions, the camera intensity will be directly proportional to fluorophore population and can be used to choose the cells within the appropriate range of expression.

     
  3. 3.

    When positioning the cell relative to the laser focus, be sure to avoid bright features indicative of internal organelles. Also avoid being too close to the edge of the cell. The best position is one where the plasma membrane shows even intensity around the periphery of the cell.

     
  4. 4.

    During the averaging process, some FCS curves will display large amplitudes, long decay times, or periodic oscillations. These normally result from bright features like internal organelles or the edge of the cell crossing the confocal detection volume. These do not reflect 2D membrane diffusion and should be excluded from analysis.

     

References

  1. 1.
    Comar WD, Schubert SM, Jastrzebska B et al (2014) Time-resolved fluorescence spectroscopy measures clustering and mobility of a G protein-coupled receptor opsin in live cell membranes. J Am Chem Soc 136(23):8342–8349CrossRefPubMedGoogle Scholar
  2. 2.
    Fotiadis D, Liang Y, Filipek S et al (2003) Atomic-force microscopy: rhodopsin dimers in native disc membranes. Nature 421:127–128CrossRefPubMedGoogle Scholar
  3. 3.
    Becker W, Hickl H, Zander C et al (1999) Time-resolved detection and identification of single analyte molecules in microcapillaries by time-correlated single-photon counting (TCSPC). Rev Sci Instrum 70:1835–1841CrossRefGoogle Scholar
  4. 4.
    Becker W (ed) (2005) Advanced time-correlated single photon counting techniques, vol 81. Springer, BerlinGoogle Scholar
  5. 5.
    Müller BK, Zaychikov E, Bräuchle C et al (2005) Pulsed interleaved excitation. Biophys J 89:3508–3522CrossRefPubMedCentralPubMedGoogle Scholar
  6. 6.
    Bacia K, Schwille P (2007) Practical guidelines for dual-color fluorescence cross-correlation spectroscopy. Nat Protoc 2:2842–2856CrossRefPubMedGoogle Scholar
  7. 7.
    Triffo SB, Huang HH, Smith AW et al (2012) Monitoring lipid anchor organization in cell membranes by PIE-FCCS. J Am Chem Soc 134:10833–10842CrossRefPubMedCentralPubMedGoogle Scholar
  8. 8.
    Endres NF, Das R, Smith AW et al (2013) Conformational coupling across the plasma membrane in activation of the EGF receptor. Cell 152:543–556CrossRefPubMedCentralPubMedGoogle Scholar
  9. 9.
    Giepmans BNG, Adams SR, Ellisman MH et al (2006) The fluorescent toolbox for assessing protein location and function. Science 312:217–224CrossRefPubMedGoogle Scholar
  10. 10.
    Digman MA, Gratton E (2011) Lessons in fluctuation correlation spectroscopy. Annu Rev Phys Chem 62:645–668CrossRefPubMedCentralPubMedGoogle Scholar
  11. 11.
    Olofsson L, Margeat E (2013) Pulsed interleaved excitation fluorescence spectroscopy with a supercontinuum source. Opt Express 21:3370–3378CrossRefPubMedGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2015

Authors and Affiliations

  1. 1.Department of ChemistryUniversity of AkronAkronUSA

Personalised recommendations