Rotational Diffusion of the α_2a Adrenergic Receptor Revealed by FlAsH Labeling in Living Cells

Abstract

The fluorescein arsenical hairpin binder (FlAsH) shows much promise to determine the relative orientations of protein regions and structures even in living cells and in the plasma membrane. In this study, we characterized FlAsH's photophysical properties by steady-state anisotropy and time-resolved single photon counting for further applications with G-protein coupled receptors. We find that FlAsH has a relatively high initial anisotropy of 0.31 ± 0.01 and a three-component fluorescence lifetime with an average of 4.1 ± 0.1 ns. We characterized the FlAsH fluorophore orientation in the α_2A adrenergic receptor revealing rigid orientations of FlAsH in the membrane plane for rotational correlation times of ∼50 ns in living cells. To elucidate the fluorophore-membrane orientation and rotational correlation time, an anisotropy treatment similar to that of another researcher (Axelrod, D. 1979. Biophys. J. 26:557–573) was developed. The rotational correlation times were observed to increase by up to 16 ns after agonist addition. The rotational correlation time also allowed for a comparison to the theoretical relationship between translational and rotational diffusion (originally proposed by Saffman, P. G., and M. Delbrück. 1975. Proc. Natl. Acad. Sci. USA. 72:3111–3113) and revealed a discrepancy of a factor between 10 and 100.

Introduction

G-protein-coupled receptors (GPCRs) are membrane proteins with a common structural motif of seven transmembrane spanning α-helices (1). They represent not only the largest family of hormone and neurotransmitter receptors but also are a prominent pharmacological target (1). Upon binding of ligand hormones and transmitters, GPCRs display a plethora of activation signals mediated into nearly every type of cell (1). Ligand binding triggers a conformational change in the GPCR, placing it into an activated state, which then activates the common signal transducers, the G proteins. G proteins in turn activate proteins, which can be either ion channels or enzymes that produce second messengers (1).

Recent x-ray crystallography data of Rhodopsin, a light-sensitive GPCR activated by light absorption of its covalently coupled 11-cis retinal, which isomerizes after photoexcitation, shows that the largest conformational changes are shifting and twisting motions between transmembrane domain helices V and VI (2). Rhodopsin serves as a central guide for all GPCRs, because the chromophore ligand undergoes huge light-absorption chromatic shifts after activation, called the photocycle, which through electron microscopy, neutron diffraction, and electron spin resonance spectroscopy has been directly linked to the conformational changes (3,4). Thus, the photocycle of rhodopsin allowed the first direct studies of time-resolved changes in activation of GPCRs both in vitro and in vivo (3,4). Through the well-aligned dipole moment of the retinal chromophore, rhodopsin and bacteriorhodopsin—a bacterial proton pumping analog of rhodopsin—and their photocycle also allowed for measurements of rotational and reorientational motions in the membrane and of conformational changes between spectroscopic absorption states (3–8).

With the advent of the use of fluorescent and luminescent proteins and of fluorophores that can be covalently attached to GPCRs in living systems, the timing dynamics can be elucidated via photophysical techniques like Förster resonance energy transfer (FRET) and bioluminescence resonance energy transfer combined, at times, with optical microscopy (9,10). From these studies, it has been determined that the millisecond conformation switch in rhodopsin carries over to many GPCRs (9,11). However, some classes and specific examples, e.g., the parathyroid hormone receptor as a class II GPCR, have been shown to have activational switches, up to one-thousand times slower than Rhodopsin (4,9,11).

Attempts to reverify the membrane rotational and reorientational GPCR movements by optical fluorescence, however, have so far been distorted by artifacts due to the binding of the fluorophore. Monofunctionality allows for a spinning motion of the fluorophore whereas bifunctional attachment of fluorescent proteins to regions of the non-membrane-spanning random coils between the transmembrane domains are of a floppy nature and reflect rotational motions that cannot be distinguished from the fluorescent protein alone (3,6,12). FRET and bioluminescence resonance energy transfer measurements in regions between and very close to the transmembrane domains reveal primarily the displacement instead of orientational motions between the fluorophores (9,10). Several attempts have been made to directly attach small fluorophores other than fluorescent proteins to measure distinct activation timing differences (13). Among the fluorophores being used, one stands out to be very successful: fluorescein arsenical hairpin binder or FlAsH (14,15), originally described in 1998 by Griffin et al. (14). Four cysteines (two pairs with two-amino-acid separation) bridge and form a complex with arsenic groups of the FlAsH fluorophore, which increases its fluorescence quantum efficiency by a factor of 50,000 to 0.5 upon binding to the protein (15). The FlAsH in nonbound form is permeable to the plasma membrane and can be washed out from the cytosol (15). The FlAsH-labeling technique has been used to track gap junction structure and dynamics, to monitor AMPA receptors in synaptic vesicle trafficking, and to monitor the conformational changes in Myosin V upon binding to actin (16). For GPCRs, attachment of FlAsH between the transmembrane domains (TM) V and VI in replacement of the yellow fluorescent protein (YFP) revealed similar kinetics of the activation switch as compared to the YFP-labeled FRET construct. However, signaling, especially of the G_s-dependent activation of adenyl cyclase, was interrupted with CFP/YFP labeling, but was not affected by FlAsH labeling (13,17,18).

The rigid nature of the FlAsH binding allows the determination of the transition dipole orientation as a measure of the protein orientation and thus can reveal reorientational and rotational motions. FlAsH has primarily been used for these types of experiments with cytosolic or non-membrane-bound proteins (19,20). Frequency domain fluorescence anisotropy uncovered calmodulin differences in the reorientational times from the non-Ca²⁺- to the Ca²⁺-bound states (20).

In this article, we report the monitoring of a GPCR, the α_2A adrenergic receptor (AR), with bound FlAsH in the plasma membrane of live cells with steady-state and time-resolved fluorescence anisotropy. First, we needed to characterize the FlAsH for its photophysical properties of the fluorescence lifetime <τ_F> and initial anisotropy, r₀. We, first of all, find that different attachment regions of GPCRs show, as expected, varying degrees of rigidity of the protein and that attachment regions on single α-helices or between transmembrane domains but nearly imbedded in the membrane are quite rigid and display fluorophore orientations within the membrane plane. For the cases of the FlAsH on rigid portions of the receptor, the steady-state anisotropies for each surface of the cell with respect to the excitation polarization can be used with the Perrin equation to globally solve for the rotational correlation time of the receptor and verify the rotational correlation time determined by time-resolved fluorescence anisotropy. Last, we demonstrate that the rotational correlation time of the receptor significantly increases after the addition of activating, agonist ligand. We compare our values of the two-dimensional rotational correlation time for the α_2A-AR in the membrane of living cells to measurements performed previously for Rhodopsin and to the theoretical determination of pure membrane mobility by Saffmann and Delbrück (8).

Materials and Methods

Sample preparation

Materials

The ligand norepinephrine, flavin mononucleotide (FMN), and phosphate-buffered saline were obtained from Sigma (Taufkirchen, Germany). PTH(1–34) was purchased from Bachem (Torrance, CA). FlAsH is commercially available from Invitrogen (Karlsruhe, Germany) as TC-FlAsH.

FlAsH-labeling

The labeling was performed as previously described by Hoffmann et al. (13). Transfected cells grown on coverslips as described above were twice washed with Hank's Balanced Salt Solution without Phenol Red supplemented with 1 g/L glucose (HBSS; Invitrogen). Immediately after washing, cells were incubated at 37°C for 1 h with 2 mL HBSS containing 500 nM FlAsH and 12.5 μM 1,2-ethane dithiol. Next, to reduce nonspecific labeling, cells were washed twice with HBSS, and incubated for 10 min with HBSS containing 250 μM 1,2-ethane dithiol. After this incubation, cells were washed again twice with HBSS before being used for fluorescence measurements (13,18).

Cell culture

HEK-293 cells were transfected using Effectene according to the manufacturers’ description (Quiagen, Hilden, Germany). Cells were maintained in Dulbeccos’ modified Eagle's medium supplemented with 10% fetal calf serum and 100,000 U/L penicillin and 100 mg/L streptomycin at 37°C in 7% CO₂. For fluorescence measurements, cells were seeded on round polylysine-coated coverslips that were placed in six-well plates and transfected 6 h later. Cells were kept in culture for an additional 48 h.

Molecular biology

The mouse α_2A-AR constructs used in this study were previously published (13,18). Three constructs contain the specific amino-acid binding motif FLNCCPGCCMEP in different positions within the third intracellular loop (IL): from A246 to R257 (I3-N), from S297 to R308 (I3-M), and from G350 to R361 (I3-C). A fourth construct binds the A247 region to the R361 region connecting a bridge between the N- and C-termini of IL3 (I3-D). All constructs contain an additional enhanced cyan fluorescent protein (CFP) fused to position R449 of the C-terminus of the receptor. The PTH-receptor construct (21) contains the specific motif fused to the very C-terminus of the human PTH-receptor following amino-acid position 515. Receptor cDNAs were cloned into pcDNA3 (Invitrogen) for transient expression in HEK-293 cells. A cytosolic reference construct was prepared by fusing the CFP C-terminally with the sequence AEAAAREACCPGCCARA. This sequence is derived from the original FlAsH-binding sequence by Griffin et al. (14) but substitutes the original CCRECC motif with CCPGCC for higher affinity (19). This construct was needed for control measurements because FlAsH is not fluorescent in an unbound state.

Confocal microscopy and analysis for fluorescence and anisotropy

The fluorescence images, steady-state, and time-resolved fluorescence anisotropy and anisotropy images were acquired on a modified TCS SP5 commercial confocal laser-scanning microscope (Leica, Mannheim, Germany) operating with LAS AF software (Leica). The normal FlAsH, YFP, and CFP detection for non-FLIM and nonanisotropy imaging was acquired with various internal lasers (405 nm for CFP and 514 nm for FlAsH and YFP) and with internal spectral emission settings of 475–495 nm for CFP and 520–590 nm for YFP and FlAsH. All measurements on the Leica TCS SP5 confocal microscope were performed at 512 × 512 pixels and a line frequency of 400 Hz. The confocal pinhole was always set to 1.0 Airy units.

The Leica TCS SP5 was custom built by the manufacturer to include an external laser input port with a 30/70 beam splitter and an external emission port or multifunction port after the confocal pinhole (22,23). The excitation port was used to input the polarized light to excite the FlAsH. A custom adaption system was inserted into our confocal microscope, and polarization selective, pulsed laser beams could be directed into the microscope (23).

A custom emission path allowed both the simultaneous detection of the parallel and perpendicular emission with respect to the excitation laser while preserving the time-resolved detection capabilities with time-correlated single photon counting and with imaging for fluorescence lifetime imaging microscopy (FLIM) and polarization variations for anisotropy imaging (see the Supporting Material). Control measurements of the instrumental response of 762 ps, g-factor of 0.73, and control fluorescence lifetimes of standard fluorophores were recorded (23).

To establish these methods, we compared FMN and YFP to FlAsH bound to the peptide-CFP construct all in solution. Under our measurement conditions, we were able to observe fluorescence lifetimes, anisotropies, and rotational correlation times in good agreement to measurements previously reported in the literature (24,25) for FMN and YFP (Table 1).

Table 1.

Control fluorescence lifetimes and anisotropies of FMN, YFP, and FPC

Fluorophore	〈τF〉 (ns)∗	r0†	〈τR〉 (ns)†
FMN	4.5 ± 0.1	0.2 ± 0.1‡	0.2 ± 0.1
YFP	3.4 ± 0.1	0.34 ± 0.02§	13.5 ± 1.4
FPC	4.5 ± 0.1	0.24 ± 0.02	11.0 ± 0.3
FPC (10% glycerol)	4.3 ± 0.1	0.28 ± 0.02	14.1 ± 0.5
FPC (30% glycerol)	4.0 ± 0.1	0.31 ± 0.02	20.9 ± 0.9

^∗Mean value in each case determined by more than two independent measurements.

^†Value determined directly from more than two anisotropy decay fittings.

^‡Reduced initial anisotropy due to quenching by very high fluorophore concentration.

^§Borst et al. (24) reports 0.38.

Measurements were controlled by the SymPhoTime software (PicoQuant, Berlin, Germany). The software allowed selection of regions of interest (ROI) and extraction of the data for postprocessing with IGOR (WaveMetrics, Portland, OR). Fitting was performed with FluoFit (PicoQuant) (23).

For more-detailed information on the Materials and Methods, including the theoretical anisotropy and orientation treatment, corrections for such items as high numerical aperture objectives and fitting, and for more-detailed information on the results, please refer to the Supporting Material.

Results

Experimental controls and labeling

FlAsH labeling should produce a rigid binding fluorophore to the protein in order for fluorescence to occur (14). However, in our case, we also wanted to test whether specific labeling regions of GPCRs would also be rigidly aligned around the membrane plane in living cells. To establish the system, we needed to verify that we could first observe the fluorescence of FlAsH with our system and then establish convenient and verifiable ways with our system for measuring fluorescence anisotropy, both time-resolved and steady state.

The proteins of interest for FlAsH labeling are the FlAsH binding motif CCPGCC fused to the CFP, referred to as FlAsH-peptide-CFP (FPC), as a simple control which can be expressed in the cytosol of cells or can be prepared for use in vitro; the parathyroid hormone receptor (PTHR); and the α_2A-AR. The PTHR plays an important role in the regulation of calcium levels as well as in osteolysis. It can be activated by binding the parathyroid hormone (1–34) (PTH) and activates the G-proteins G_q and G_s (11,26). Here, the FlAsH binding motif was attached to the end of the 150 amino-acid-long C-terminus of the PTHR and will be referred to as PTHR-F (21). PTHR-F should serve as a nonrigid membrane FlAsH binding GPCR motif, as the C-terminal tail is thought to be randomly coiled and in the cytosol of the cell (21).

The α_2A-AR can be activated by several agonists including adrenaline and noradrenaline. They, in turn, can activate G-proteins G_i and G_o. Physiological effects mediated by the α_2A-AR include the regulation of blood pressure and the inhibition of neurotransmitter release. Apart from the above-mentioned full agonists, there are so-called partial agonists—leading to a partial activation of the receptor, with antagonists occupying the binding site without affecting receptor activity and inverse agonists impairing the activity of the receptor. Today, it is not known whether the α_2A-AR undergoes partial steps of one conformational change or adopts distinct conformations in response to the various ligands. Some of the most recent results seem to favor a model in which several distinct conformations exist and challenge the definition of an active-versus-inactive receptor (18,27,28). The activation of the α_2A-AR seems to have a strong influence on TM V and VI while juxtamembrane regions of the intracellular loops (ILs) mediate the interaction with G-proteins (18,27). Here, the FlAsH binding motif was introduced at different sites in IL3 connecting TM V and VI as shown in Fig. 1 B, without affecting the functionality of the receptor (18). IL3 is thought to be of high importance concerning the functionality of the receptor because it is the region of highest structural deviation from other adrenoreceptors (18). We refer to the α_2A-AR-FlAsH constructs by their N-terminal, middle, and C-terminal IL3 tetracysteine positions as I3-N, I3-M, and I3-C, respectively (18). Another construct, called I3-D, has two cysteines both on the N-terminal and C-terminal sides of IL3 that upon FlAsH binding form a bridge (see Materials and Methods and (13)).

Figure 1.

Experiment diagram of fluorophore (A), receptor labeling sites (B), cell measuring regions of interest (C), and fluorophore orientation in cellular coordinates. (A) The FlAsH fluorophore and its two arsenic groups and four binding sites (14), (B) α_2A-AR labeling regions I3-N and I3-C are located below TM V and TM VI, respectively. I3-M is located in the middle of IL 3. I3-D artificially creates a junction of TM V and TM VI near the membrane (17,18). A CFP is located at the C-terminus. All constructs retain normal functional activity (17,18). (C) Assignment of the cell measurement regions on a spherical diagram (see text for description) (31).

These three proteins and, in total, six FlAsH binding constructs, have been previously described by some of the authors for their studies with FlAsH as a small molecule alternative to fluorescence proteins in the determination of protein conformational changes observed by FRET microscopy (Fig. 1 A). In the previous studies, the constructs were already well tested and described for both fluorescence and functional properties (13,18,21,27,28). The six FlAsH-labeled constructs were first imaged with the standard internal laser, as used in one of these previous studies, and could be subsequently imaged with high confocal sectioning and resolution with our custom-aligned laser (Fig. 2 A).

Figure 2.

Initial experimental measurements with FlAsH labeling. (A) Confocal transmission, fluorescence, and overlay images of cells and specific FlAsH membrane binding to the I3-D. (Scale bar: 10 μm.) (B) Fluorescence lifetime, instrument response function, three-exponential fit (top), and residuals (bottom) of the FPC. The average fluorescence lifetime 〈τ_F〉 = 4.1 ± 0.1 ns. (C) Anisotropy images of the I3-D (left) and the PTHR-F (right) in the equatorial region (B and C) of the cell as assigned in Fig. 1C. The region C can be clearly observed in the I3-D (left) with the higher anisotropy parallel to the excitation polarization (solid arrow). Anisotropy scaling: 0–0.5. (Scale bar: 10 μm.) (D) Comparison of the average time-resolved anisotropy decay fitting of the I3-D, the PTHR-F, the FPC-30% glycerol, −10% glycerol, and −0% glycerol.

Before this report, few articles have dealt with both the fluorescence lifetime and anisotropy (initial anisotropy, r₀) of the FlAsH fluorophore (15,19,20,29). We tested the FlAsH-peptide construct in solution, in dried film, and in viscous media to reveal that the FlAsH always displayed three significant fluorescence emission decay components (Table 1 and Fig. 2 B) with a long component of nearly 5 ns, $\left({\tau }_F^{\left(1\right)}\approx 4.8\text{ns}\right),$ a middle component of just over 2 ns, $\left({\tau }_F^{\left(2\right)}\approx 2.3\text{ns}\right)$a fast component of 0.3 ns, $\left({\tau }_F^{\left(3\right)}\approx 0.3\text{ns}\right)$ and with mean fluorescence lifetimes $<{\tau }_{\mathit{F}}>\text{of}4\text{.}1\pm \text{0}\text{.}1\text{ns}\text{.}$ The fluorescence lifetime values are in agreement with those performed by frequency domain measurement of <τ_F> of 4.1 ± 0.1 ns, but reveal three components instead of two. However, the three-component model has a significantly better goodness of fit (20,29,30).

For determination of the initial anisotropy, r₀, the FPC was measured by both steady-state anisotropy and time-resolved anisotropy in solutions of increasing viscosity and in dried film. The time-resolved anisotropy decays showed for the FPC in solution a rotational correlation time of 11 ± 1 ns, a typical value for CFP alone in solution (24) with an initial anisotropy of 0.29 ± 0.01 (Fig. 2 D). Upon increasing the viscosity by addition of 10% and 30% glycerol into the solution, the rotational correlation time decreased linearly with the viscosity, demonstrating that the changes were just due to the Stokes-Einstein-Debye relationship and revealed an initial anisotropy of 0.30 ± 0.01 (Table 1 and Fig. 2 D). Upon acquisition of steady-state anisotropies corrected to the Perrin equation (23) with the average lifetime of <τ_F> = 4.1 ns, an average, initial, fundamental anisotropy of r₀ = 0.308 ± 0.018 could be determined. Although a higher value of r₀ = 0.38 ± 0.03 could be determined from the dried film (data not shown), we were uncertain whether this represented the correct value in biological media, due to the extremely different environment which also caused the fluorescence lifetime to be deviant from all other data. To our knowledge, only four other reports exist on this subject (15,19,20,29), and the data were acquired by steady-state or frequency domain techniques to reveal r₀ = 0.30 or values slightly higher but within experimental error of our acquired value (Fig. 3).

Figure 3.

Fluorescence image, anisotropy image, anisotropy distribution, fluorescence lifetime, and anisotropy decay example for every FlAsH construct under investigation: I3-N, I3-M, I3-C, I3-D, PTHR-F, and FPC. The anisotropy images are of the ROI-B and -C equatorial regions of the cell-sphere assignment (Fig. 1C) and indicate that the I3-C and I3-D are quite rigidly aligned with the plasma membrane of the cell and with the excitation polarization (solid arrow). The I3-N displays slight alignment. The fluorescence lifetimes did not significantly differ between all measurements. The anisotropy decay times were also the longest for I3-C and I3-D. (Anisotropy scaling: 0–0.5. Scale bar: 10 μm.)

Steady-state anisotropy and determination of fluorophore rigidity in the membrane of cells

The FlAsH, which becomes only fluorescent upon binding, might also display random orientations with respect to the membrane plane due to rapid motions of the protein, especially in the random coil regions of a GPCR between the transmembrane α-helices and in the N- and C-terminal regions. To initially test the rigid nature of the fluorophore and protein in the membrane, we performed florescence anisotropy imaging with confocal microscopy in the center sections of the GPCR-FlAsH labeled cells (Fig. 1 C). If, for example, the fluorophore is aligned with the membrane plane, then it should respond with higher anisotropy for membrane regions that are aligned with the excitation polarization and lower anisotropy for membrane regions orthogonal to the excitation polarization as previously observed for rigid lipids aligned with the membrane plane (31) (Fig. 1 C).

The resulting steady-state anisotropies for each of the FlAsH binding constructs are summarized in Fig. 2 C and Fig. 3. The steady-state anisotropy images with polarization of the excitation indicated with an arrow and with scaling and calculation of the anisotropy to both the three- and two-dimensional models (see the Supporting Material) revealed that two constructs, the I3-C and I3-D, display rigid fluorophore and protein orientation with distinctly and statistically higher anisotropy for membrane regions parallel to the excitation polarization. On the other hand, the PTHR-F and the I3-M would display random orientation of the fluorophore, because the labeling sites of these proteins were relatively far away from the transmembrane α-helices (Fig. 2 C and Fig. 3). However, in contrast to its I3-C counterpart, the I3-N with labeling site near the TM V region visually displayed seemingly random orientation (Fig. 3).

Anisotropy model for membrane plane orientation—transition dipoles aligned within a sphere

For cases where the fluorophore (and the fluorophore's transition dipole) rigidly remain in the membrane plane, the two-dimensional anisotropy case above becomes more complex upon measuring over the entire surface of a living cell (Fig. 1 C). Such a treatment has been previously described for oriented lipids on the surface of erythrocyte ghost cells (31). In this case, we approximate the cell to a sphere to define three distinct regions for anisotropy measurements with an incident light beam from the top with a specific polarization direction (Fig. 1 C). The first region, region A (ROI A), is the top of the cell or sphere with the membrane plane approximately orthogonal to the incident beam of light, which is essentially approximated to the two-dimensional anisotropy case (Fig. 1 C). The next region, region B (ROI B), has the spherical section (or membrane) parallel to the incident beam and parallel to the polarization of the incident beam (Fig. 1 C). The last region, region C (ROI C), has the spherical section (the membrane) parallel to the incident beam but perpendicular to the polarization of the incident beam (Fig. 1 C). A development of the theoretical treatment of both the steady-state and time-resolved anisotropies of each region can be found in the Supporting Material.

Application of the model to steady-state anisotropy data

The newly defined regions of interest A, B, and C were applied to the steady-state anisotropy images for the receptors (Fig. 2 C and Fig. 3). The anisotropies of the regions of interest were extracted pixelwise and displayed a Gaussian distribution from which the mean value could be determined (Fig. 3). The data from these measurements is summarized in Table 2 for the determined anisotropies with standard error for every construct. The PTHR-F and the I3-N displayed no difference in the anisotropy between regions B and C beyond experimental error, and all anisotropies between regions A, B, and C were the same within experimental error for the PTHR-F (Table 2). The measured difference between B and C for I3-M of the α_2A receptor was just barely significant, but the anisotropy differences between B and C were the largest for I3-C and I3-D (Table 2).

Table 2.

Receptor steady-state anisotropies and orientations

Sample	rSA	rSB	rSC	(rSB + rSC)/2	ΔBC∗
α2A I3-M	0.234 ± 0.005	0.265 ± 0.005	0.281 ± 0.001	0.273	0.016
α2A I3-D	0.238 ± 0.005	0.248 ± 0.006	0.303 ± 0.002	0.276	0.055
α2A I3-N	0.244 ± 0.005	0.263 ± 0.005	0.273 ± 0.006	0.268	0.010
α2A I3-C	0.248 ± 0.005	0.263 ± 0.004	0.292 ± 0.013	0.278	0.029
PTHR-F	0.241 ± 0.005	0.247 ± 0.009	0.245 ± 0.001	0.247	0.001
	Θ [°]	Θ′ [°]	Ψ [°]	β [°]	r0
α2A I3-M	11.58 ± 0.30	29.82 ± 0.06	31.77 ± 0.01	34.7	0.293
α2A I3-D	11.45 ± 0.28	28.88 ± 0.10	31.50 ± 0.02	34.2	0.297
α2A I3-N	12.47 ± 0.55	30.44 ± 0.27	30.52 ± 0.06	34.7	0.301
α2A I3-C	12.02 ± 1.61	29.58 ± 0.94	30.23 ± 0.27	33.1	0.303
PTHR-F	15.23 ± 0.70	31.76 ± 0.02	30.58 ± 0.01	32.4	0.307

^∗Δ_BC = r_S^C - r_S^B.

The data could then be applied to the theoretical model of the transformation between the coordinates of the membrane plane and the fluorophore's absorption and emission axis (23) (Table 2) to determine the average angle β between the absorption and emission dipole to again solve for the initial anisotropy, r₀. For I3-D, the absorption dipole moment of FlAsH (Θ) is only at a roughly 11° angle from the membrane plane, but the emission dipole moment (Θ′) extends roughly 29° further from the membrane plane. This angular difference, the consideration that the absorption and emission axes also have a shift of nearly 32° in the membrane plane (Ψ), and the total angular difference between the absorption and emission axes (β) of ∼34° reveal an initial anisotropy of 0.297. The average initial anisotropy for all five constructs was determined using this method to be 0.300.

Time-resolved measurements in each of the cellular regions

Encouraged from the well-aligned axis of the FlAsH in the membrane, we were interested in determining both the fluorescence lifetime and the rotational correlation times of the FlAsH-labeled constructs. Acquisition of the same samples with the same microscope required only the application of a time readout filter built in and synchronized to our confocal microscope. First, polarization-independent fluorescence lifetimes were acquired to reveal nearly the same three-component fit in comparison to the FPC (Fig. 3 and Table 3). The anisotropy decay for all of the constructs was very long, displaying no major artifacts due to a spin or wobble of the FlAsH fluorophore (Fig. 3). However, upon fitting, a minor but significant component of 0.01–0.10 ns always appeared. The majority of the anisotropy decay always occurred in a second longer component of >15 ns, depending on the construct (Fig. 3 and Table 3). The measurements were recorded for each of the regions and fit for regions A and C (Table 3). However, the rotational correlation time was not affected within experimental error for the regions measured. High rigidity of the FlAsH fluorophore and protein construct to the membrane resulted in accordingly longer rotational correlation times of the anisotropy decay of the FlAsH (Table 3). The I3-C and I3-D displayed an average rotational correlation time of nearly 50 ns. The I3-M and I3-N displayed an average rotational correlation time of nearly 36 ns. The mostly unrigid PTHR-F still decayed with a rotational correlation time of nearly 21 ns. The initial anisotropy for the time-resolved data also did not vary too much from the previously determined values (Table 2).

Table 3.

Receptor rotational correlation times and fluorescence lifetimes

Sample	rA(t)	n	rC(t)	n	〈r(t)〉	n	τRPERRIN	n	〈τF〉 (ns)†
	〈τR〉 (ns)∗		〈τR〉 (ns)∗		〈τR〉 (ns)∗		〈τR〉 (ns)∗
α2A I3-M	36.6 ± 1.3	6	32.2 ± 0.4	2	35.5 ± 1.2	8	30.5 ± 4.1	4	4.3 ± 0.2
α2A I3-D	47.7 ± 1.2	7	48.8 ± 2.1	5	48.1 ± 1.1	12	32.7 ± 2.9	4	4.1 ± 0.5
α2A I3-N	40.2 ± 1.4	4	32.0 ± 1.4	4	36.1 ± 1.8	8	25.7 ± 3.2	6	4.0 ± 0.3
α2A I3-C	51.3 ± 1.6	4	48.6 ± 2.6	2	50.4 ± 1.3	6	34.0 ± 13.9	4	4.1 ± 0.2
PTHR-F	20.3 ± 1.4	5	21.9 ± 1.3	8	21.3 ± 1.0	13	15.6 ± 1.4	2	4.3 ± 0.2

^∗Mean value determined by n independent evaluations.

^†Mean value in each case determined by more than two independent measurements.

The fluorescence lifetime data was then applied to the Perrin equation adaptation for each of the regions to determine the rotational correlation time and to verify the time-resolved analysis (23) (Table 3). Upon application of the initial anisotropy of 0.30 and fluorescence lifetimes, generally good agreement occurred between the fit and the calculated rotational correlation times.

Rotational correlation time increases upon ligand activation

Although much is already known in terms of many details of the signaling cascade upon activation of both the α_2A-AR and the PTHR, little is known about what occurs to GPCRs and their physical status in the membrane with the exception of a few cases (32,33). To further test this, we continued our experimentation to add activating doses of 1 mM noradrenaline or 1 μM parathyroid hormone peptide (PTH(1–34)) and continued our readout of many of the same cells, previously studied in this investigation. The data registration was continued, but a time marker was placed in the electronic recording file to indentify measurements after ligand addition (23).

The data was again analyzed by the previously described models to reveal that the fluorescence lifetime and the initial anisotropy were unaffected by the addition of ligand. However, the rotational correlation time measured both by the time-resolved method and by the steady-state method with the Perrin equation, starting seconds up to minutes, after ligand addition became significantly longer for all constructs (Fig. 4). The average increase of the rotational time for all constructs was roughly 10 ns. However, both the I3-C and the PTHR-F were observed to reduce their rotational correlation times again to almost the basal, preactivation level ∼2–3 min after ligand addition (Fig. 4).

Figure 4.

Average rotational correlation time, τ_R, before and after ligand addition for the I3-N, I3-M, I3-C, and I3-D with 1 mM norepinephrine and for the PTHR-F with 1 μM PTH(1–34). Measurements were accumulated in 2-min intervals before and after ligand addition. The measurements before ligand addition demonstrate that little or no change was observed before ligand addition. After ligand addition, all receptors significantly increased the rotational correlation time, but the I3-C and I3-D showed the most significant and constantly increasing rotational times.

Discussion

Within this study, we report artifact free rotational correlation times of the α_2A adrenergic receptor that increase upon ligand stimulation in living cells. The labeling of the receptor with the rigid binding fluorophore FlAsH along with iterative labeling sites that maintained the α_2A receptor activity while rigidly keeping the fluorophore's dipole within the membrane plane was a key first step to the undertaking of this study (13,18). To assess the motion of the receptors that are in the membrane plane, a new evaluation scheme considering the geometrical positions of the fluorophore, protein, and cellular membrane was derived to approach a global solution to the orientation of the fluorophore and verifying rotational motions in the membrane plane. This assessment shares similarity with the treatment of the well-aligned membrane lipid probe, DiI, in the study on erythrocyte ghost cells performed by Axelrod (31).

The use of proper control experiments allowed us to predetermine both the fluorescence lifetimes and initial anisotropies of FlAsH in different media. The FlAsH has a three-component fluorescence decay with an average lifetime of ∼<τ_F> = 4.1 ns and an initial anisotropy of r₀ = 0.31 (Fig. 2, B and D, and Table 1). These control values are in agreement with others (19,20,29,30). We do note that our data result in a significant 0.3-ns component that is unlike the other reports in the fluorescence lifetime and a 0.1-ns component is the anisotropy decay time. One attribution of this might be due to the higher hydrophobicity of the fluorophore, which might be more susceptible to influences of the arsenide groups near the fluorophore.

The measurements performed on the α_2A-AR and PTHR constructs in living cells further reinforced the values obtained in the control experiments (Fig. 2 D, Fig. 3, and Table 1). By determination that the I3-D were rigidly membrane-aligned by steady-state anisotropy (Fig. 2 C, Fig. 3, and Table 2), we were able to roughly determine the orientation of the FlAsH dipole moment in the membrane plane and could obtain an initial anisotropy of r₀ = 0.30. The rest of the α_2A-AR and PTHR constructs agreed to this value within their experimental errors. Further application of time-resolution with analysis of the regions at the top (region A) and side (region C) of the cell and their corresponding anisotropies, r^A(t) and r^C(t), allowed for the direct measurement of the rotational correlation time, beyond the further verification of initial anisotropy and fluorescence lifetime (Fig. 2 D, Fig. 3, and Tables 2 and 3). The rotational correlation times of the I3-C and I3-D were determined to be ∼50 ns in the plasma membrane of living HEK 293 cells. The other α_2A-AR and PTHR constructs, which were less or not rigidly aligned to the membrane plane, had slightly lower values between 20 and 35 ns.

Because the anisotropy values for the cell regions B and C were significantly different for the I3-N, I3-C, and I3-D (Fig. 3 and Table 2), a steady-state anisotropy analysis similar to Axelrod (30) allowed us to determine the freedom of motion between these regions. The fact that the PTHR has the same freedom of motion in all regions (A, B, and C) served as verification. The difference Δ_BC = r_S^B – r_S^C, we propose, could then be used to determine the rigidity of a fluorophore with respect to the membrane plane. In this way, the relatively nonrigid PTHR had Δ_BC = 0.001, in stark contrast to the I3-N, I3-C, and I3-D with Δ_BC = 0.010–0.055.

Activation of the receptors with ligand could be further measured in intervals of 2 min and 4 min both before and after addition that avoided key times of internalization (27,34,35). Within these intervals, significant increases in the rotational correlation time could be observed for all of the receptor constructs, measured both by analysis of the time-resolved anisotropy and by measuring the steady-state anisotropy with subsequent calculation via the Perrin equation (see Eq. 9, and see also Fig. 4 and Table 3). The increase of rotational correlation time of between 8 and 16 ns in the α_2A-AR (meaning a retardation in the motion) could be attributed to the formation of oligomers with itself or other GPCRs or membrane receptors (28,36), to the binding of either its G_αG_βγ G-protein, which from FRET measurements has been known to form within milliseconds after α_2A-AR activation and remains stably attached for seconds and up to minutes after activation (37) or of binding to the β-arrestin 2 factor with the timing dynamics described for the G protein (38). To our knowledge, this is one of the first measurements of the pure rotational diffusion of a membrane protein by fluorescence, unhindered by artifacts due to spinning or wobbling of the fluorophore. Furthermore, this is the only measurement to our knowledge that measures the pure rotational diffusion of a GPCR other than Rhodopsin in native membranes (5). The rotational correlation time τ_R = 50 ns for the I3-C and I3-D in the membrane of HEK cells represents a two-dimensional rotation and thus a rotational diffusion constant D_R = (4τ_R)⁻¹ or specifically of D_R = 5 × 10⁶ s⁻¹. Such a rotational diffusion constant is roughly two orders-of-magnitude higher than the traditionally reported value from Cone (5) of D_R ≈ 2.5 × 10⁴ s⁻¹ to 10⁵ s⁻¹ of Rhodopsin in disk membranes of native rod pigment.

Saffman and Delbrück (8) created a simple model to relate translational D_T and rotational D_R motion and calculated a theoretical relationship for Rhodopsin which should apply, in general, to all similar structured GPCRs,

$${\left(\frac{D_T}{D_R}\right)}_{theor}=16\times {10}^{-14}{\text{cm}}^2,$$

whereby they reasoned that the GPCR should be treated like a cylinder with radius a and height h in a membrane of viscosity η with a theoretical two-dimensional rotational diffusion constant:

$$D_R=\frac{kT}{4\pi \eta a^{2}h}\text{.}$$ (9)

Experimentally, the relationship

$${\left(\frac{D_T}{D_R}\right)}_{\exp }$$

for Rhodopsin was very close, because Poo and Cone (39) measured D_T = 2–5 × 10⁻⁹ cm²/s, for

$${\left(\frac{D_T}{D_R}\right)}_{\exp ,Rhodopsin}\approx 2-20\times {10}^{-14}{\text{cm}}^2\text{.}$$

However, in the case of the α_2A-AR, PTHR, and similar GPCRs, we and others have found the translational diffusion in HEK 293 and similar cell types to be D_T = 10⁻⁹ cm²/s to 10⁻⁸ cm²/s (Fig. S6 and (32,40)). Thus, we estimate

$${\left(\frac{D_T}{D_R}\right)}_{\exp ,{\alpha }_{2A}-AR}\approx 2-20\times {10}^{-16}{\text{cm}}^2,$$

suggesting from the first theoretical calculation a factor of 10–100 times’ difference, which has been observed for lipid and protein motion in the membrane by one of us and by others (41–44). Among the suggestions about why the discrepancy exists is that the Saffman and Delbrück model (8) does not compensate or consider for long- and short-range viscosity effects, for protein binding to the cytoskeleton or for oligomerization of the membrane proteins (42–44). Given the rotational diffusion constant for the α_2A-AR D_R = 5 × 10⁻⁶ s⁻¹ and the size of the typical GPCR from Rhodopsin monomers (2) (2a = 3.8 nm and h = 8.4 nm), we calculate for our measurement from the equation above, the membrane viscosity to be η = 0.2 P or only a factor-of-two greater than water. This is in huge contrast to the factor of 100 for the Rhodopsin case from Saffman and Delbrück (8) and gives rise to our assertion that the concept proposed by others might be correct in that the extreme local viscosity might be much lower in some regions of the plasma membrane allowing for relatively high rotational diffusion and that these effects are attenuated on larger size scales, affecting the translational diffusion (42–44).