Quaternary structure of the yeast pheromone receptor Ste2 in living cells

BBAMEM-82371; No. of pages: 9; 4C: 5, 6, 7 Biochimica et Biophysica Acta xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbamem

Quaternary structure of the yeast pheromone receptor Ste2 in living cells☆ Michael R. Stoneman a,⁎, Joel D. Paprocki a, Gabriel Biener a, Koki Yokoi a, Aishwarya Shevade b, Sergei Kuchin b, Valerică Raicu a,b,⁎ a b

Physics Department, University of Wisconsin-Milwaukee, Milwaukee, WI, USA Department of Biological Sciences, University of Wisconsin-Milwaukee, Milwaukee, WI, USA

a r t i c l e

i n f o

Article history: Received 14 October 2016 Received in revised form 5 December 2016 Accepted 8 December 2016 Available online xxxx Keywords: G protein coupled receptors Optical spectroscopy Membrane protein interactions Fluorescence Förster resonance energy transfer (FRET) Light microscopy

a b s t r a c t Transmembrane proteins known as G protein-coupled receptors (GPCRs) have been shown to form functional homo- or hetero-oligomeric complexes, although agreement has been slow to emerge on whether homo-oligomerization plays functional roles. Here we introduce a platform to determine the identity and abundance of differing quaternary structures formed by GPCRs in living cells following changes in environmental conditions, such as changes in concentrations. The method capitalizes on the intrinsic capability of FRET spectrometry to extract oligomer geometrical information from distributions of FRET efficiencies (or FRET spectrograms) determined from pixel-level imaging of cells, combined with the ability of the statistical ensemble approaches to FRET to probe the proportion of different quaternary structures (such as dimers, rhombus or parallelogram shaped tetramers, etc.) from averages over entire cells. Our approach revealed that the yeast pheromone receptor Ste2 forms predominantly tetramers at average expression levels of 2 to 25 molecules per pixel (2.8 · 10−6 to 3.5 · 10−5 molecules/nm2), and a mixture of tetramers and octamers at expression levels of 25–100 molecules per pixel (3.5 · 10−5 to 1.4 · 10−4 molecules/nm2). Ste2 is a class D GPCR found in the yeast Saccharomyces cerevisiae of the mating type a, and binds the pheromone α-factor secreted by cells of the mating type α. Such investigations may inform development of antifungal therapies targeting oligomers of pheromone receptors. The proposed FRET imaging platform may be used to determine the quaternary structure sub-states and stoichiometry of any GPCR and, indeed, any membrane protein in living cells. This article is part of a Special Issue entitled: Interactions between membrane receptors in cellular membranes edited by Kalina Hristova. © 2016 Elsevier B.V. All rights reserved.

1. Introduction G protein-coupled receptors (GPCRs) recognize and respond to a variety of stimuli ranging from light to molecular ligands such as odorants, hormones, and neurotransmitters [1–5]. While several experiments have indicated that GPCRs form functional homo- or hetero-oligomeric complexes in vivo as well as in vitro [6–13], there have been more than occasional suggestions that not all GPCRs are multimeric or that homooligomerization is not essential for function [14–16]. Whether this uncertainty stems from data over-interpretation or built-in structural and functional versatility of these receptors is yet to be clarified [12, 17]. In this work, we utilized a combination of fluorescent tag-based approaches introduced recently to investigate the oligomerization

☆ This article is part of a Special Issue entitled Interactions between membrane receptors in cellular membranes, edited by Kalina Hristova. ⁎ Corresponding authors at: University of Wisconsin-Milwaukee, Physics Department, P.O. Box 413, Milwaukee, WI 53201, USA. E-mail addresses: [email protected] (M.R. Stoneman), [email protected] (V. Raicu).

properties of the yeast pheromone receptor Ste2 [18], which is often used as a general model for fungal pheromone receptors [19]. Protein interactions in living cells may be probed via detection of light from fluorescent tags attached to the proteins of interest. If an “acceptor” (A) fluorescent molecule lies within less than 10 nm of an optically excited “donor” molecule (D), it can extract energy from the donor through non-radiative Förster Resonance Energy Transfer (FRET) [20] and re-emit it at longer wavelengths. Using FRET, it has been possible to determine intramolecular distances, probe association of molecules into oligomeric complexes, determine the spatial distribution of such complexes in living cells, and assess the effects of ligand binding [7, 20–25]. Previously, a strategy was proposed for determination, at every pixel in a fluorescence image [8], of the apparent FRET efficiency (Eapp), i.e., the average fraction of energy transferred within a population of donortagged and acceptor-tagged molecules, some of which may be interacting. The method relies on acquisition of full spectral information from sample voxels containing donors and acceptors, and their unmixing using known donor and acceptor spectra. The resulting

http://dx.doi.org/10.1016/j.bbamem.2016.12.008 0005-2736/© 2016 Elsevier B.V. All rights reserved.

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

2

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

pixel-level Eapp values in an image are used to generate an Eapp distribution (i.e., a histogram), rather than an average over a region of interest. Such distributions are then interpreted using models of molecular complexes (or oligomers) with certain quaternary structures (i.e., sizes and geometries) [8,9]. Each quaternary structure entails multiple configurations of donors and acceptors, each of which generates a specific peak in the Eapp histograms. Taken collectively, the peaks represent a unique FRET fingerprint corresponding to a certain oligomeric structure, while the Eapp histograms constitute veritable FRET spectrograms. The underlying method therefore is called FRET spectrometry [26]. Using this FRET spectrometry approach, it has been shown that the α-factor receptor Ste2 forms complexes as large as tetramers [8]. Ste2 is a class D GPCR found in the yeast Saccharomyces cerevisiae [27–30] of the mating type a, and binds the pheromone α-factor secreted by cells of the mating type α. Previous work has not been able to demonstrate whether the Ste2 tetramers were stable structures or formed by reversible association of dimers, and whether structures larger than tetramers formed at higher expression level of the receptors. The novel FRET imaging platform introduced herein expands on the previous methods to allow determination of the proportion of different quaternary structures of the α-factor receptor in living cells for two different ranges of concentrations. We used the FRET spectrometry method to determine the dominant quaternary structure of Ste2 from FRET spectrograms. The method for determination of Eapp histograms relied on two photon excitation of the donor at a wavelength (800 nm) at which the acceptor is virtually unexcited, concomitant with spectrallyresolved detection of fluorescence which is spectrally unmixed (i.e., deconvoluted) into donor and acceptor components for each pixel in an image. A second excitation at a wavelength (960 nm) which directly excited both the donors and the acceptors allowed determination of average values for donor and acceptor concentrations as well as Eapp for each cell. This in turn allowed us to probe the partition of receptor populations among different quaternary structures, using the classical statistical ensemble approach [26,31,32] supplemented with geometrical information from FRET spectrometry. Our results suggested that Ste2 assumes tetrameric and octameric forms, with tetramers dominating at low concentrations. 2. Materials and methods 2.1. Sample preparation Yeast cells (Saccharomyces cerevisiae) were engineered to express the sterile 2 α-factor receptor protein fused to one of two different types of fluorescent tags, i.e. either GFP2 [33] or YFP [34], at position 304 in the Ste2 amino acid sequence. All but eight amino acids of the Ste2 cytoplasmic tail were removed for increased FRET efficiency [35]. This removed tail contains a sequence of amino acids that is required for receptor internalization and desensitization [18,36,37]. Therefore, the resulting Ste2Δtail-GFP2 and Ste2Δtail-YFP are endocytosis defective, a property that we used to our advantage in this work. The tagged Ste2 Δ tail proteins do remain biologically active (see Fig. S7 in Supporting materials). Both GFP2 and YFP are variants of the green fluorescent protein (GFP) and contain the following amino acid substitutions: F64L, A206K (GFP2), and S65G, S72A, T203Y, A206K (YFP). This choice of fluorescent tags offers the advantage of a high spectral overlap between the donor emission and acceptor excitation spectra, as well as a low direct excitation of the acceptor, because GFP2 has a blue-shifted excitation maximum (398 nm), compared to the commonly used eGFP. The A206K mutation was incorporated into both fluorescent tags to eliminate their natural propensity to dimerize at high concentrations [38]. The DNA constructs, Ste2Δtail-GFP2 and Ste2Δtail-YFP, were expressed (either singly or together) from high-copy (2 μ) plasmids in the yeast strain KBY58 (MATa leu2-3,112 ura3-52 his3-Δ1 trp1 sst1-Δ5 ste2Δ) lacking a functional copy of the chromosomal STE2 gene [39].

These haploid cells of mating type a carry a mutation (ho) that prevents them from switching their mating type from type a to type α. In this way, any homo-oligomerization of Ste2 detected in the experiments performed in the absence of the ligand α-factor is not affected by endogenous ligand from cells with switched mating type. Yeast cells carrying one or two of the above plasmids were grown at 30 °C on synthetic complete medium lacking uracil and/or tryptophan, to provide plasmid selection. Cells gently scraped from the solid medium were suspended in 800 μL of 100 mM KCl buffer; 200 μL of cell solution was then pipetted onto 35-mm gridded glass-bottom dishes (Grid 50; Ibidi, Martinsried, Germany). Prior to adding the cell suspension, the dishes were coated with concanavalin A (Sigma Aldrich, St. Louis, MO) to prevent the cells from being moved when additional solutions were introduced to the medium. To this end, 100 μL of a 0.5 mg/mL solution of concanavalin A (in deionized water) was placed onto a gridded dish, and the dish was then covered for 30 min to allow the deposition to occur. The remaining solution was then removed and the dishes were allowed to air-dry for a minimum of 24 h. The yeast cell suspension was pipetted onto the dish prepared as above, and after allowing the cells to fully adhere to the dish for 10 min, the cells were imaged using a twophoton optical micro-spectroscope we have developed (see next). 2.2. Two-photon fluorescence micro-spectroscopy The two-photon optical micro-spectroscope was comprised of a Nikon Eclipse Ti™ (Nikon Instruments Inc., Melville, NY) inverted microscope stand and a modified OptiMiS scanning/detection head from Aurora Spectral Technologies (Grafton, WI). The scanning/detection head was modified to incorporate a line-scan protocol and a module for automatically changing the excitation wavelength while maintaining the excitation power. The line-scan protocol leads to signals two orders of magnitude higher than those achievable with a point-scanbased system for the same line dwell time, which reduces the sample photobleaching and thereby improves the accuracy of the method [40]. A mode-locked laser (MaiTai™; Spectra Physics, Santa Clara, CA), which generates 100 fs pulses with center wavelengths tunable between 690 nm and 1040 nm and a full-width half maximum of ~7 nm, was used for fluorescence excitation. The excitation beam was focused to a line in the plane of the sample using an infinity-corrected, plan apochromat, oil immersion objective (100×, NA = 1.45; Nikon Instruments Inc.). The OptiMiS detection head employed a nondescanned detection scheme, in which the emitted fluorescence was projected through a transmission grating onto a cooled electron-multiplying CCD (EMCCD) camera (iXon X3 897, Andor Technologies, Belfast, UK). The spectral components of the fluorescence emission are separated when passing through the transmission grating, and strike the CCD camera in the form of a line perpendicular to the excitation line. The different wavelengths of light comprising the emitted fluorescence are separated as a function of pixel position on the CCD array. Therefore, the micro-spectroscopic data set captured by the OptiMiS detection system contains three-dimensional information; two of the dimensions are spatial dimensions (440 × 300 pixels), and the third dimension is wavelength, i.e. each pixel in the 2D image is sampled at 200 different wavelength channels. The spectral bandwidth of the wavelength channels ranges from 415 nm to 615 nm with a spectral resolution of ~ 1 nm. The optical scanning head (for laser beam scanning) and EMCCD camera used for image acquisition were controlled by the same computer using in-house custom software written in C++. Using the OptiMiS detection system, spectral information was obtained from each sample voxel on a time scale much shorter than that which would correspond to molecular diffusion of the fluorescently labelled receptors [8]. Furthermore, by exciting the donor with a wavelength which leaves the acceptor virtually unexcited (800 nm in this work), any signal detected from the acceptor could be safely attributed to FRET. This sample scan, which we term “the FRET scan,” allowed determination of both the FRET efficiency and donor concentration at pixel

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

level [26]. It nevertheless, did not allow us to determine acceptor concentration. 2.3. Light delivery device for successive excitations at two wavelengths In order to determine the acceptor concentration, we built a light delivery system that allows one to tune the excitation wavelength of the laser from 800 nm to 960 nm and automatically adjust the laser power immediately after the FRET scan, and then perform a “concentration scan.” To control the power, an achromatic half-wave plate was mounted in a motorized precision rotation stage (AHWP05M-980 and PRM1Z8E; Thorlabs, Newton, NJ) and combined with a polarizing beamsplitter (CM1-PBS252; Thorlabs, Newton, NJ). Only the portion of the beam transmitted through the beam-splitting cube was directed into the OptiMiS scanning head. Therefore, by simply rotating the motorized stage housing the wave plate, the amount of laser power entering the OptiMiS scanning head was changed. In order to set this power to a specified value via the OptiMiS control software program, a small portion of the laser beam was picked off and directed to a high-speed photodiode (DET100A; Thorlabs, Newton, NJ). The voltage across the photodiode was measured with a National Instruments (Austin, TX) PCI-E 6229 DAQ card in real time; this voltage was converted to the amount of laser power transmitted through the microscope objective using a voltage-to-power calibration. During the calibration procedure, the voltage read out from the photodiode and the power measured with an Integrating Sphere Photodiode Power Sensor (Thorlabs, Newton, NJ) placed at the microscope objective front lens were obtained simultaneously. Three different power levels were measured in this manner for each laser excitation wavelength ranging from 700 to 1040 nm (in 20-nm increments). The total laser power illuminating the sample for all measurements presented herein was 300 mW for both the 800 nm and 960 nm excitation scans. Because the laser beam profile at the location of the sample was not a diffraction limited spot, but was shaped into a line, the total power of the beam was distributed across multiple voxels in the sample at any given time. Therefore, each voxel in the sample was illuminated with an average power of 0.2 mW for a duration of 35 ms. The total acquisition time for a full set of micro-spectroscopic images, i.e., two sets of images each containing 440 × 300 pixels and 200 wavelengths per pixel was ~ 60 s, including 10 s for each scan at the two laser wavelengths, and ~ 40 s needed to change the laser excitation wavelength. Since the molecular diffusion could potentially change the distribution of receptors and their oligomers during the relatively long time elapsed between FRET and concentration scans, the acceptor concentration was only computed as an average over a region of interest (i.e., single cells), and the data were analyzed as described below. 2.4. FRET imaging protocol and assembly of Eapp meta-histograms Two-photon micro-spectroscopic scans of yeast cells expressing Ste2-GFP2 and Ste2-YFP singly or in combination were performed on multiple fields of view; each field of view contained between 1 and 20 cells. A separate dish was used for each field of view. A general leastsquares (GLS) unmixing method was utilized to obtain the abundance vector, k (Eq. (S3) of Supplementary methods), for every pixel in a fluorescence image from samples expressing more than one fluorescent species, such as donor-tagged Ste2, kDA, acceptor-tagged Ste2, kAD, and endogenous fluorescent molecules. The GLS algorithm, written in Matlab and described in detail in Supplementary methods, was generalized from a previously published approach [41] to incorporate an unlimited number of fluorescent tags. Separately determined emission spectra of donors, acceptors, etc. were normalized by their respective emission intensity maxima to obtain corresponding elementary spectra. From these quantities, we computed the number of photons emitted by donors and acceptors in the samples coexpressing Ste2Δtail-GFP2 and

3

Ste2Δtail-YFP using the following formulas [8,26]: FDA(λex) = kDAwD, and FAD(λex) = kADwA; wD and wA represent the areas under the donor and acceptor emission spectra, respectively. The apparent efficiency of FRET occurring between multiple donors and acceptors within oligomeric complexes is defined as the ratio between the donor fluorescence lost due to FRET (FD , FRET) and the fluorescence of the donor in the absence of acceptor (FD(λex)) [8,26]: Eapp ¼

F D;FRET F D ðλex Þ

:

ð1Þ

In a typical micro-spectroscopic scan, neither of the two quantities on the right hand side of Eq. (1) are obtained directly, because in any given measurement both donor and acceptor fluorophores are present. Using a set of equations relating F DA and FAD to the amount of fluorescence lost due to FRET [8], Eq. (1) can be rewritten in terms of the fluorescence of the donor in the presence of the acceptor: Eapp ¼ 1−

F DA ðλex Þ F D ðλex Þ

:

ð2Þ

By choosing the first excitation wavelength such that the acceptor is not directly excited by laser light, in this work FA(λex, 1) was zero, and the FRET efficiency was determined using Eq. (2) [8,42] with just a single excitation wavelength (λex,1 = 800 nm). The pixel-level Eapp values were organized into a histogram plot of bin width equal to 0.01. Histograms with various widths and characteristic features were obtained. Broad histograms with distinctive peaks usually indicate the existence of donors and acceptors in different configurations within oligomeric structures [8], while narrower histograms correspond to cells containing only a few such oligomeric configurations and/or lower order oligomers. To take advantage of all Eapp histograms acquired and thereby create a statistical ensemble, we used a recently introduced strategy whereby the positions of the two largest or most dominant peaks in the individual E app histograms are accumulated and binned to form meta-histograms of peak positions [43,44] (using a bin of size 0.01). The “size” of a peak was defined as the difference in values between the height of the peak maximum and the height of the nearest local minima. These meta-histograms were further analyzed as described next. 2.5. Analysis of meta-histograms using quaternary structure models Conveniently, a theory exists describing FRET in oligomeric complexes of arbitrary size, n, according to which the ith donor in a complex of configuration q (see Fig. S1 in Supporting material) possesses a FRET efficiency, Ei,k,q, which is dependent on the number of donors, k, in the complex, as well as the distance, rij, of said donor with respect to the acceptors [31]. To remove any dependence on external calibration factors, such as spectral overlap between donor emission and acceptor excitation, E i,k,q may be written in terms of the pairwise FRET efficiency, Ep, and one of the distances in the complex, r1, as:

Ei;k;q


6 r1 Ep ∑ r ij j "
 # ¼ r1 6 1−Ep þ Ep ∑ r ij j

ð3Þ

where the summation index j represents all acceptors. Since all donors in a complex of configuration q emit signal from the same sample voxel (or image pixel), measurements do not discriminate

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

4

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

between signals from individual donors and therefore provide an average efficiency per donor, Ek;q ¼

1 ∑E : k i i;k;q

ð4Þ

According to Eq. (4), the FRET efficiency between a donor and an acceptor separated by a distance rij = r1 equals Ep, which provides a welcome constraint that relates tetramers to dimers. Compiling a metahistogram and modeling it in this manner avoids problems associated with stochastic FRET [45] because these interactions show up as a distinct signal in the individual cell histograms [46]. Fitting of theoretical models to the experimental data was achieved qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 by minimizing a fitting residual, Res ¼ ∑½Data−ModelFunction =f ,

abundances of the fluorescent species. Since there is no physical interpretation for such negative numbers, a non-negative least squares (NNLS) iterative approach is generally implemented. However, our computer simulations for oligomer-forming fluorescent proteins indicated that the NNLS algorithm is detrimental to the shape of the Eave vs. XA curve (see Fig. S2 in Supporting methods). Therefore, as stated above, we have utilized the GLS unmixing algorithm (Eq. (S3) of Supplementary methods), which provides excellent results for data points which obey the condition: 0 b XA b 0.8. According to the kinetic theory of FRET [31], the expression for the apparent FRET efficiency for a population of oligomeric complexes in which each complex has the same size (n) and shape is: Eapp;n ¼

where f is the number of degrees of freedom. For the meta-histogram

n Xh i X 1 ð1−X A Þk X n−k Ek;q A nð1−X A Þ k¼0 q

ð6Þ

fitting, the function consisted of a sum of Gaussians, SðEapp Þ ¼ ∑ Aq exp

1 is the number of donors within a single complex when avwhere nð1−X AÞ

½

eraged over all complexes, and k is the number of donors in a particular combination of a single complex. The sum over q represents the sum over all combinations the complex can take by arranging k D's and
 n n! n − k A's in each complex; hence there exist ¼ k!ðn−kÞ! configurak tions for each particular value of k. The average FRET efficiency for entire regions of interest containing mixtures of complexes with varying sizes and shapes is obtained by summing over the apparent FRET efficiency of each type of complex, Eapp,n, contained within the mixture, i.e.:

−



q

ðEapp −Ek;q Þ2 , where Aq represents the amplitude of the qth Gauss2σ 2q

ian distribution and σq its standards deviation. The kinetic theory of FRET predicts the positions of each Gaussian (i.e., Ek, q in Eq. (4)), all of which relate to the same parameter, the pairwise FRET efficiency, Ep, which is common between dimers and tetramers (see Fig. S1 in Supplementary methods). 2.6. Analysis of average Eapp data using the FRET and concentration scans By adding a scan of the sample at a second excitation wavelength (960 nm), two additional quantities can be measured: the cellular average fluorescence of the donor and acceptor at the second excitation wavelength, i.e., FDA(λex,2) and FAD(λex,2), respectively. Because the fluorescence emissions of the two fluorophores at each excitation wavelength are linked by their respective excitation cross-sections at the two excitation wavelengths, no additional unknowns are added. By measuring the fluorescence emission following sequential excitation at the two different wavelengths of a sample expressing only the donor, as well as a sample expressing only the acceptor, the ratios of the excitation rate constants at the two excitation wavelengths (Γλex,1 and Γλex,2) can be easily determined, as described in detail in previous publications [32,47] and summarized in Supplementary methods. With the additional measured quantities extracted from the second excitation scan, sufficient information is obtained to quantify both the donor and acceptor fluorescence (FD(λex,1) and FA(λex, 2), respectively) as if determined in the absence of FRET, i.e., as a result of direct excitation with laser light. The total number of donors and acceptors residing in a given pixel (nD and nA, respectively) were obtained via a straightforward calibration of the pixel level values of FD(λex,1) and FA(λex,2) with fluorescence measurements on solutions of the respective fluorescent proteins at a number of known concentrations (see Eqs. (S8) and (S9) in Supporting materials for details on this method). The total receptor concentration could also be expressed as molecules/nm2 by dividing the total number of molecules per pixel by the area of the membrane, which occupies a cross-section of the excitation voxel corresponding to every pixel [47]. The voxel size was determined from previously derived equations [48]. The calculation of nD and nA enabled both a comparison of expression levels in various cells, and the calculation of the acceptor mole fraction, defined as: XA ¼

nA nA þ nD

ð5Þ

Pixel-level Eapp values were averaged over entire cells and plotted against the average XA corresponding to the same cells. While assembling the Eave vs. XA plot, we found that the method used for spectral unmixing occasionally provided negative values for very low

Eapp;theo ¼

1 ∑ ND;n Eapp;n ND;tot n

ð7Þ

where ND;tot ¼ ∑ ND;n is the total number of donors in the mixture, and n

ND,n = nμn(1− XA) is the total number of donors in oligomers of size n, with μn being the total number of complexes of size n within the mixture. With these notations, and for the case of a mixture of dimers and tetramers, Eq. (7) may be re-written in terms of the ratio of the number of protomers in tetrameric state vs. dimeric state, N4/N2, in the mixed population of oligomers,

Eapp;

theo

¼

Eapp;2 þ Eapp;4 1þ

N4 N2

N4 N2

ð8Þ

where N2 ≡ 2μ2 and N4 ≡ 4μ4 denote the total numbers of protomers within dimers and tetramers, respectively, where Eapp,2 and Eapp, 4 are derived from Eq. (6). Similar equations may be written for mixtures of other complex sizes. Fitting of the theoretical model to the experimental data was achieved by minimizing a fitting residual, Res ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ∑½Data−ModelFunction =f , where f is the number of degrees of freedom. Eq. (8) was used to simulate cellular averages of Eapp vs. XA (see Figs. 2b and 3b in the Results and discussion section), with Ek,q derived from Eq. (4) for n = 2 (dimers) and n = 4 (tetramers). 3. Results and discussion To determine the quaternary structure of Ste2 in living yeast cells (S. cerevisiae), we used optical micro-spectroscopy to acquire pixel-level FRET spectrograms for yeast cells expressing donor-tagged and acceptor-tagged Ste2. First, the elementary fluorescence spectra [26] of D (green fluorescent protein, GFP2) and A (yellow fluorescent protein, YFP) were determined from micro-spectroscopy measurements of cells expressing either Ste2 -GFP2 or Ste2 -YFP. These were then used

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

5

Fig. 1. Typical results obtained from imaging yeast (S. cerevisiae) cells co-expressing Ste2-GFP2 and Ste2-YFP. Spectral unmixing provided separate maps of the fluorescence signals of donors in the presence of acceptors (kDA) and acceptors in the presence of donors (kAD). Apparent FRET efficiency (Eapp) values were determined for each image pixel from the pixellevel values of kDA and kAD, as described in the Materials and methods section. Contours of regions of interest (ROI) were drawn based on the kDA images and then transferred to the Eapp maps (shown as yellow lines). Histograms of Eapp were generated for each ROI using a bin size of 0.01. The positions of the two most dominant peaks (see Materials and methods) of these histograms (indicated by vertical red lines) were used to generate the meta-histograms shown in Fig. 2.

to unmix the composite spectra obtained, at image pixel level, for cells co-expressing Ste2-GFP2 and Ste2-YFP, as described in Materials and methods and Supplementary methods. A spectral unmixing procedure (see Materials and methods) provided fluorescence intensity maps of donors, kDA, and acceptors, kAD, at every image pixel (see Fig. 1). These maps were used to compute pixel-level apparent FRET efficiency, Eapp, from which FRET spectrograms (or histograms) were obtained. The random selection of cells shown in Fig. 1 reveals histograms with different widths and levels of detail (i.e., peak location and number). It has been shown previously that for Ste2 at relatively low expression levels, broad histograms with distinctive peaks are common [8,26]. Such histograms typically indicate the existence of higher order oligomers (such as rhombus tetramers for Ste2) and have been used previously to determine the Ste2 quaternary structure (or oligomer type and size) [8,26]. This same method has been used with success to investigate the quaternary structure of the M3 muscarinic acetylcholine receptor in living cells [9]. Narrower histograms are usually obtained from cells containing either (a) dimers only, (b) one or just a few different donor-acceptor configurations within a large oligomeric structure, or (c) extremely high concentrations of oligomers of any size [26,43,47]. To take advantage of all types of Eapp histograms acquired for a certain protein and thus create a statistical ensemble of an entire population of cells, a strategy has been proposed whereby the positions of the dominant peaks in the individual Eapp histograms are accumulated and binned to form meta-histograms of peak positions [43], as also described in the Materials and methods section. This method effectively relies on measurements over a large number of cells to filter out the numerous additional peaks that result from combinations of oligomers

with different donor and acceptor configurations [26] within each histogram, which are generated when the receptor expression levels are higher than 1 molecular complex per pixel. Herein we imaged a large number of cells (N 250 per experiment) to generate meta-histograms of the positions of the two most dominant peaks in each histogram in order to determine the quaternary structure of the oligomeric complexes, as well as the parameter that relates the meta-histograms to the protein complex geometry: the pairwise FRET efficiency (see Materials and methods). To determine the relative abundance of the various oligomeric complexes, or the ratio of concentrations of protomers within higher order oligomers and those within dimers (if they exist), we turned to the FRET ensemble approach. This method relates oligomer abundances through the pairwise FRET efficiency to the average FRET efficiency, Eave, over an entire region of a cell (e.g., the cell membrane or the cytoplasm) corresponding to a certain acceptor mole fraction, XA. The mole fraction was determined by D and A excitation at two different laser wavelengths followed by correction of their respective emissions for FRET (see the Materials and methods section) and a concentration calibration using pure solutions of GFP2 and YFP, respectively (see Supplementary methods). We have recently shown that combining the meta-histogram approach with the ensemble approach enabled the determination of the relative abundance of tetramers and dimers in living cells for another GPCR (opsin) as a function of its expression level [47]. In this work, we modified the previous method by simultaneously rather than serially fitting the meta-histogram and the Eave vs. XA curves to theoretical models, in order to apply tighter constraints on the abundance of oligomers higher than tetramers. We also analyzed meta-histogram features arising from mixing of different

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

6

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

quaternary structures, allowing for information extraction that was not possible previously. We began our analysis with the data from cells expressing on average b25 receptors per pixel (i.e., b3.5 · 10− 5 molecules/nm2). The meta-histogram generated by extracting the position of the tallest peaks for a population of histograms of the type illustrated in Fig. 1 is shown in Fig. 2a, while the average Eapp vs. XA is shown in Fig. 2b. Several peaks are visible in our meta-histogram, which indicate that several different pathways for FRET exist in Ste2 complexes. The kinetic theory of FRET (see Materials and methods and Supplementary methods) predicts the FRET efficiency for each donor-acceptor configuration q of an oligomeric complex of size n, number of donors, k, and

Fig. 2. Determination of the quaternary structure of Ste2 from FRET measurements of yeast (S. cerevisiae) cells expressing Ste2-GFP2 and Ste2-YFP at relatively low concentrations (i.e., b25 receptors per pixel). (a) Eapp meta-histograms (empty circles) were obtained by collecting the dominant peak positions of Eapp histograms from individual cells as shown in Fig. 1. Solid red lines represent simulated meta-histograms (see Materials and methods) using a rhombus model for the oligomer (see panel c), while individual Gaussian peaks correspond to particular configurations of dimers or rhombus-shaped tetramers. Positions of all the meta-histogram peaks depend only on Ep, which was determined from fitting to be 0.187. (b) Cellular averages of Eapp vs. acceptor mole fraction (XA) obtained from 264 cells averaged over XA intervals of 0.05. Experimental data (open circles; vertical bars are standard deviations) were fitted to a model (solid red line) consisting of mixtures of dimers and tetramers of the type shown in (c). Structural parameters were set equal to those in (a). The best-fit was obtained for N4/N2 N 1000 (Res = 1.23). (c) Structural model consisting of rhombus tetramers with different configurations of donors and acceptors (see Supplementary methods for details). The first row shows single oligomers per pixel, which generate zeroth-order peaks, the second shows first-order mixing, while the third accounts for incomplete labelling. Higher order mixing of oligomers only contributes to the broad histogram background and is therefore not discernible experimentally. For single complex configurations or combinations predicting the same peak position, only one such configuration is shown.

acceptors, n − k, in the complex, and distance of each donor with respect to the acceptors [31]. The discrete set of such FRET efficiencies provides a sum of Gaussian functions whose positions are determined by the oligomer model chosen. Differing D-A configurations corresponding to rhombus tetramers [8] at concentration levels of b 1 complex/pixel are shown in the first row of Fig. 2c (see also the Materials and methods section and Supplementary Table 1 for more details on the theory of this model). We termed these, the “zeroth-order peaks” of the rhombus tetramer model. To remove any dependence on calibration factors, such as the Förster radius (R0) [20], the apparent FRET efficiency was written in terms of the pairwise FRET efficiency, Ep (which incorporates R0) and the distance between two adjacent fluorescent proteins. This provides welcome constraints for the positions of the five predicted peaks. Straightforward extension of the kinetic theory of FRET (see Materials and methods) also provided a simple expression for the average FRET efficiency, Eave vs. XA, in terms of the apparent efficiencies that would be possible for the particular complex model tested. As described in the Materials and methods section, Eave depends on both Ep (as the meta-histogram also does), and on the ratio of protomers comprising the tetramers and those in dimeric form, i.e., N4/N2. We simultaneously fitted a model consisting of a mixture of rhombus tetramers and dimers to both the Eapp meta-histogram and Eave by constraining Ep to be the same for both plots, while allowing the Gaussian peaks' widths and heights for the meta-histogram and the N4/N2 ratio for the Eave plot to vary independently. The FRET-productive dimers, if they existed, would contribute to a peak in the meta-histogram corresponding to Ep for rhombus (labelled as 2 in Fig. 2), which was nevertheless absent at low concentrations (see Fig. 2a). The best-fit curve (with Res = 4.54) obtained with the mixture model, proposed previously for the quaternary structure of Ste2 [8,26], captured some but not all of the features of the meta-histogram (see Supplementary results Fig. S7). Most conspicuously, a series of peaks is present in the experimental meta-histograms at Eapp values of 0.1 or less, which are not predicted for pixels containing only a single rhombus tetramer. Since, although relatively low, the receptor concentrations in these experiments still exceeded one complex per pixel, we reasoned that combinations of two, three, or more complexes per pixel would account for the additional peaks in the meta-histogram. This does not have any effect on the fitting of the cellular Eave (Fig. 2b) as that model already takes combinations of configurations into account. The different possible combinations of two complexes per pixel are listed in the second row of Fig. 2c. We termed these specific combinations “first-order mixing” and refer hereafter to peaks generated in the meta-histogram due to these combinations as “first-order peaks”. The number of additional peaks in the meta-histograms is expected to increase rapidly with increasing the number of complexes per pixel and quickly grow to the level of a broad, almost featureless distribution of values. We were able to test this simple hypothesis for mixtures of several complexes per pixel (from 1 to 8 complexes/pixel), by computing the various combinations of FRET efficiencies, using the FRET algebra described elsewhere [26] as well as their probabilities of occurrence for several different XA values. When we added them all together, we indeed observed relatively broad distributions of Eapp (meta-) histograms “decorated” by zeroth- and first-order peaks (see Fig. S6 in Supplementary results). We then used the zeroth- and first-order model-predicted peaks to fit the meta-histogram simultaneously with the Eave vs. XA plot and obtained an improved overall fit, both visually and as judged by the reduced fitting residual (Res = 2.41) for combination of zeroth and first order peaks, as compared to 4.54 for the zeroth order only fitting. Finally, we noticed that the highest peak in the meta-histogram did not actually coincide with any of the predicted zeroth- and first-order peaks, or any other higher order peaks for that matter (see Supplementary Fig. S6). Although this remaining discrepancy is actually minute when compared to the overall capability of the model to represent the data, we sought an explanation for it. Of the several possibilities considered, the only one that correctly predicted that peak was the possible

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

existence of dark acceptors in a small fraction of oligomers. Dark acceptors could be generated through one of following well known mechanisms: (1) presence of a fraction of protonated dark acceptors in cells with lower pH [49], (2) photo-bleaching via repeated excitation by light and/or FRET [50], and (3) photo-bleaching due to higher order photon interactions [51]. Regardless of the mechanism responsible for their generation, dark acceptors may lead to an additional type of oligomer, illustrated by the bottom row of Fig. 2c. The FRET efficiencies for these oligomers which present incomplete labelling are predicted to occupy identical or very close to identical positions with some of the other peaks in the meta-histograms, with the exception of a single one, labelled IL in Fig. 2. Having established the origin of all the significant meta-histogram peaks, we further refined the simultaneous fitting of the meta-histogram and Eave vs. XA curve and obtained the best overall fit, both visually (see Fig. 2a and b) and according to the reduced residual (Res = 1.23). The resulting ratio of receptor concentrations in tetrameric vs. dimeric form (N4/N2) was N1000 (i.e., dimeric fraction b 0.1% and tetramer fraction N 99.9%), which indicates that dimers of Ste2 are not present even at the comparatively low concentrations of 2 to 25 complexes per pixel (i.e., 2.8 · 10−6 to 3.5 · 10−5 molecules/nm2). Since, within the experimental uncertainties, dimers were not present in any significant number, monomers were also highly unlikely to exist. We also ruled out trimers because they would require monomers (which do not exist) in order to form. We then turned to the analysis of cells expressing higher concentrations of between 25 and 100 molecules per pixel (i.e., 3.5 · 10− 5 to 1.4 · 10−4 molecules/nm2), in order to probe the effect of concentration on oligomerization state. The experimental data for that subset of cells and its best fit to the same mixture model used for data in Fig. 2 are presented in Fig. 3. The predicted Eave vs. XA plot (dashed red line of Fig. 3b) did not fit the experimental data very well (Res = 1.76), but the fit improved dramatically after octamers were incorporated into the theoretical model (thick red line of Fig. 3b) used for this curve (Res = 1.66). Although octamers were not added to the model used to simulate the meta-histogram, due to the complications generated by the large number of peaks without improving the fit dramatically, Fig. 3b does present some additional peaks, albeit with low amplitudes, at Eapp values in excess of 0.4, which are not predicted by tetramers (or even hexamers). Based on the numerical values presented in the caption to Fig. 3, we computed the numeric fractions of dimers, tetramers, and octamers to be b0.1%, ~ 27%, and ~ 73%, respectively, at these expression levels. Since we did not detect dimers in significant amounts, we assumed that hexamers were not present either, as their formation would require interactions between dimers and tetramers. As above, monomers are even more unlikely to exist; this also rules out the existence of trimers, which would require monomers (which do not exist) in order to form. Overall, therefore, our data seem to indicate that the yeast pheromone receptor Ste2 associates into tetramers at low concentrations and octamers, and higher order oligomers, at higher concentrations. 4. Conclusion In summary, using a combination of FRET spectrometry and ensemble FRET approaches, we determined the proportion of different oligomeric structures that the yeast pheromone receptor Ste2 forms in living cells. Future studies will likely fill in the mechanistic details of the picture describing the shift from tetramer-dominated to dimerdominated receptor populations. We observed that at comparatively low expression levels, the receptors associate into tetramers, and that no detectable amount of dimers exists. At higher expression levels, a significant contingent of octamers needed to be added in order to explain the results. This invites the possibility that Ste2 oligomerization depends on expression level, possibly following the Law of Mass action. The detected oligomers are real and not due to any experimental artifacts, such as dimerization induced by the fluorescent tags or

7

Fig. 3. Determination of the quaternary structure of Ste2 from FRET measurements of yeast (S. cerevisiae) cells expressing Ste2-GFP2 and Ste2-YFP at high concentrations (i.e., 25 to 100 receptors per pixel). (a) Meta-histograms of FRET efficiency (empty circles) were obtained by collecting and binning the dominant peak positions of the Eapp histograms of individual cells as shown in Fig. 1. Solid red lines represent simulated meta-histograms (see Materials and methods) using a rhombus model for the oligomer, while individual Gaussian peaks correspond to particular configurations of dimers or rhombus-shaped tetramers. The positions of all the meta-histogram peaks depend only on the parameter Ep, which was determined according to the fitting to be: 0.182. (b) Cellular averages of Eapp vs. acceptor mole fraction (XA). Experimental data (open circles; vertical bars are standard deviations) were fitted to a model (dashed red line) consisting of mixtures of dimers and tetramers. Structural parameters were set equal to those in (a). Structural parameters were set equal to those in (a). The best-fit was obtained for values of N4/N2 well in excess of 1000 for the dashed line. Same cellular averages of Eapp were also fitted to a model consisting of the tetramers shown in Fig. 2 (c) to which a contingent of octamers was added (solid red line). The best-fit (Res = 1.66) in this case was obtained for: N4/N2 N 1000, N8/N4 = 2.75.

stochastic FRET – a process of energy transfer between randomly distributed donors and acceptors that are expressed in membranes at densities N 10−3 molecules/nm2 [45]. The first possible artifact has been eliminated by using monomeric forms of GFP2 and YFP (see Materials and methods), while the occurrence of the second has been prevented by capping the receptor density to less than ~ 10− 4 molecules/nm2 in the present work. Hence, the tetramers and octamers formed by Ste2 must bear biological significance. Previous studies have suggested that several residues in at least two transmembrane helices of Ste2 may be involved in oligomerization [27– 30]. We considered the binding interfaces reported by one of those studies [28] and employed a relaxed form of a model GPCR (rhodopsin) fused to a fluorescent protein [47] to support the notion that a Ste2 tetramer thus formed could indeed obey the geometrical constraints provided by FRET spectrometry in the present study (see Supplementary Fig. S9). If the crystal structure of Ste2 were available, it would become possible to determine the structure of the entire Ste2 oligomer with atomic resolution by using parameters derived from FRET spectrometry and molecular dynamics simulations of the fluorescently tagged protomers within an oligomer [26]. At this juncture in time, however, we must be content with the realization that crosslinking studies [28] are consistent with our present results, and together they are consistent with the recent proposal that GPCRs present at least two binding sites

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

8

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx

that facilitate formation of dimers, tetramers, or higher order oligomers [9,47]. A remarkable difference between the oligomerization behaviour of Ste2 described above and other investigated GPCRs, such as the M3 muscarinic receptor [9] and human rhodopsin [47], is that the former forms tetramers but not dimers even at relatively low concentrations of receptor. Details regarding the physiological role played by such differing inter-protomer binding strengths will likely be filled in by future studies.

[15]

[16] [17] [18] [19]

Competing financial interests Dr. Raicu is a co-funder of Aurora Spectral Technologies LLC (AST), which commercializes OptiMiS technology used in this work. Dr. Stoneman was employed by AST in the past and occasionally consults for them. All the other authors declare no competing financial interests.

[20] [21]

[22]

[23]

Transparency document The Transparency document associated with this article can be found, in online version. Acknowledgements

[24] [25]

[26]

The optical microspectroscopy imaging facility used for this research was developed with support of the National Science Foundation, Major Research Instrumentation Program (Grant No. PHY-1126386 awarded to V.R.). The research was funded by the National Science Foundation, Physics of Living Systems Program (Grant No. PHY-1058470 awarded to V.R.). We thank Amber Grupe for help with the experimental protocol for determination of the concentration calibration curves, and Ionel Popa for stimulating discussions.

[27]

Appendix A. Supplementary data

[31]

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbamem.2016.12.008.

[32]

References

[34] [35]

[1] D.M. Rosenbaum, S.G. Rasmussen, B.K. Kobilka, The structure and function of G-protein-coupled receptors, Nature 459 (2009) 356–363. [2] W. Huang, et al., Structural insights into micro-opioid receptor activation, Nature 524 (2015) 315–321. [3] Y. Kang, et al., Crystal structure of rhodopsin bound to arrestin by femtosecond Xray laser, Nature 523 (2015) 561–567. [4] S.G.F. Rasmussen, et al., Crystal structure of the beta(2) adrenergic receptor-Gs protein complex, Nature 477 (2011) 549–555. [5] T. Warne, R. Moukhametzianov, J.G. Baker, R. Nehme, P.C. Edwards, A.G.W. Leslie, G.F.X. Schertler, C.G. Tate, The structural basis for agonist and partial agonist action on a beta(1)-adrenergic receptor, Nature 469 (2011) 241–244. [6] P.S. Park, J.W. Wells, Oligomeric potential of the M2 muscarinic cholinergic receptor, J. Neurochem. 90 (2004) 537–548. [7] D. Maurel, et al., Cell-surface protein-protein interaction analysis with time-resolved FRET and snap-tag technologies: application to GPCR oligomerization, Nat. Methods 5 (2008) 561–567. [8] V. Raicu, et al., Determination of supramolecular structure and spatial distribution of protein complexes in living cells, Nat. Photonics 3 (2009) 107–113. [9] S. Patowary, E. Alvarez-Curto, T.R. Xu, J.D. Holz, J.A. Oliver, G. Milligan, V. Raicu, The Muscarinic M3 Acetylcholine receptor exists as two differently sized complexes at the plasma membrane, Biochem. J. 452 (2013) 303–312. [10] K. Herrick-Davis, E. Grinde, A. Cowan, J.E. Mazurkiewicz, Fluorescence correlation spectroscopy analysis of serotonin, adrenergic, muscarinic, and dopamine receptor dimerization: the oligomer number puzzle, Mol. Pharmacol. 84 (2013) 630–642. [11] G. Milligan, The prevalence, maintenance, and relevance of G protein-coupled receptor oligomerization, Mol. Pharmacol. 84 (2013) 158–169. [12] S. Ferre, The GPCR heterotetramer: challenging classical pharmacology, Trends Pharmacol. Sci. 36 (2015) 145–152. [13] H.F. Vischer, M. Castro, J.P. Pin, G protein-coupled receptor multimers: a question still open despite the use of novel approaches, Mol. Pharmacol. 88 (2015) 561–571. [14] M.R. Whorton, M.P. Bokoch, S.G.F. Rasmussen, B. Huang, R.N. Zare, B. Kobilka, R.K. Sunahara, A monomeric G protein-coupled receptor isolated in a high-density

[28] [29]

[30]

[33]

[36]

[37] [38]

[39] [40]

[41]

[42]

[43]

[44]

[45]

[46]

lipoprotein particle efficiently activates its G protein, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 7682–7687. J.A. Hern, A.H. Baig, G.I. Mashanov, B. Birdsall, J.E. Corrie, S. Lazareno, J.E. Molloy, N.J. Birdsall, Formation and dissociation of M1 muscarinic receptor dimers seen by total internal reflection fluorescence imaging of single molecules, Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 2693–2698. T.H. Lan, Q. Liu, C. Li, G. Wu, J. Steyaert, N.A. Lambert, BRET evidence that beta2 adrenergic receptors do not oligomerize in cells, Sci. Rep. 5 (2015) 10166. V.V. Gurevich, E.V. Gurevich, GPCR monomers and oligomers: it takes all kinds, Trends Neurosci. 31 (2008) 74–81. J.E. Reneke, K.J. Blumer, W.E. Courchesne, J. Thorner, The carboxy-terminal segment of the yeast alpha-factor receptor is a regulatory domain, Cell 55 (1988) 221–234. C. Xue, Y.P. Hsueh, J. Heitman, Magnificent seven: roles of G protein-coupled receptors in extracellular sensing in fungi, FEMS Microbiol. Rev. 32 (2008) 1010–1032. P.R. Selvin, The renaissance of fluorescence resonance energy transfer, Nat. Struct. Mol. Biol. 7 (2000) 730–734. N.K. Lee, A.N. Kapanidis, H.R. Koh, Y. Korlann, S.O. Ho, Y. Kim, N. Gassman, S.K. Kim, S. Weiss, Three-color alternating-laser excitation of single molecules: monitoring multiple interactions and distances, Biophys. J. 92 (2007) 303–312. D.R. Singh, Q. Cao, C. King, M. Salotto, F. Ahmed, X.Y. Zhou, E.B. Pasquale, K. Hristova, Unliganded EphA3 dimerization promoted by the SAM domain, Biochem. J. 471 (2015) 101–109. A. Guzman, M. Zelman-Femiak, J.H. Boergermann, S. Paschkowsky, P.A. Kreuzaler, P. Fratzl, G.S. Harms, P. Knaus, SMAD versus non-SMAD signaling is determined by lateral mobility of bone morphogenetic protein (BMP) receptors, J. Biol. Chem. 287 (2012). S. Sarabipour, K. Hristova, Mechanism of FGF receptor dimerization and activation, Nat. Commun. 7 (2016) 10262. D.R. Singh, E.B. Pasquale, K. Hristova, A small peptide promotes EphA2 kinase-dependent signaling by stabilizing EphA2 dimers, Biochim. Biophys. Acta 1860 (2016) 1922–1928. V. Raicu, D.R. Singh, F.R.E.T. Spectrometry, A new tool for the determination of protein quaternary structure in living cells, Biophys. J. 105 (2013) 1937–1945. M.C. Overton, S.L. Chinault, K.J. Blumer, Oligomerization, biogenesis, and signaling is promoted by a glycophorin A-like dimerization motif in transmembrane domain 1 of a yeast G protein-coupled receptor, J. Biol. Chem. 278 (2003) 49369–49377. H.X. Wang, J.B. Konopka, Identification of amino acids at two dimer interface regions of the alpha-factor receptor (Ste2), Biochemistry 48 (2009) 7132–7139. H. Kim, B.K. Lee, F. Naider, J.M. Becker, Identification of specific transmembrane residues and ligand-induced interface changes involved in homo-dimer formation of a yeast G protein-coupled receptor, Biochemistry 48 (2009) 10976–10987. G.K. Umanah, L.Y. Huang, J.M. Maccarone, F. Naider, J.M. Becker, Changes in conformation at the cytoplasmic ends of the fifth and sixth transmembrane helices of a yeast G protein-coupled receptor in response to ligand binding, Biochemistry 50 (2011) 6841–6854. V. Raicu, Efficiency of resonance energy transfer in homo-oligomeric complexes of proteins, J. Biol. Phys. 33 (2007) 109–127. C. King, M. Stoneman, V. Raicu, K. Hristova, Fully quantified spectral imaging reveals in vivo membrane protein interactions, Integr. Biol. UK 8 (2016) 216–229. T. Zimmermann, J. Rietdorf, A. Girod, V. Georget, R. Pepperkok, Spectral imaging and linear un-mixing enables improved FRET efficiency with a novel GFP2-YFP FRET pair, FEBS Lett. 531 (2002) 245–249. R.Y. Tsien, The green fluorescent protein, Annu. Rev. Biochem. 67 (1998) 509–544. M.C. Overton, K.J. Blumer, G-protein-coupled receptors function as oligomers in vivo, Curr. Biol. 10 (2000) 341–344. L. Hicke, B. Zanolari, H. Riezman, Cytoplasmic tail phosphorylation of the alpha-factor receptor is required for its ubiquitination and internalization, J. Cell Biol. 141 (1998) 349–358. C.I. Chang, K.A. Schandel, D.D. Jenness, Interaction among Saccharomyces cerevisiae pheromone receptors during endocytosis, Biol. Open 3 (2014) 297–306. D.A. Zacharias, J.D. Violin, A.C. Newton, R.Y. Tsien, Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells, Science 296 (2002) 913–916. C.J. Stefan, K.J. Blumer, A syntaxin homolog encoded by VAM3 mediates down-regulation of a yeast G protein-coupled receptor, J. Biol. Chem. 274 (1999) 1835–1841. G. Biener, M.R. Stoneman, G. Acbas, J.D. Holz, M. Orlova, L. Komarova, S. Kuchin, V. Raicu, Development and experimental testing of an optical micro-spectroscopic technique incorporating true line-scan excitation, Int. J. Mol. Sci. 15 (2014) 261–276. V. Raicu, D.B. Jansma, R.J. Miller, J.D. Friesen, Protein interaction quantified in vivo by spectrally resolved fluorescence resonance energy transfer, Biochem. J. 385 (2005) 265–277. S. Patowary, L.F. Pisterzi, G. Biener, J.D. Holz, J.A. Oliver, J.W. Wells, V. Raicu, Experimental verification of the kinetic theory of FRET using optical microspectroscopy and obligate oligomers, Biophys. J. 108 (2015) 1613–1622. D.R. Singh, M.M. Mohammad, S. Patowary, M.R. Stoneman, J.A. Oliver, L. Movileanu, V. Raicu, Determination of the quaternary structure of a bacterial ATP-binding cassette (ABC) transporter in living cells, Integr. Biol. UK 5 (2013) 312–323. A.K. Mishra, T. Mavlyutov, D.R. Singh, G. Biener, J. Yang, J.A. Oliver, A. Ruoho, V. Raicu, The sigma-1 receptors are present in monomeric and oligomeric forms in living cells in the presence and absence of ligands, Biochem. J. 466 (2015) 263–271. C. King, S. Sarabipour, P. Byrne, D.J. Leahy, K. Hristova, The FRET signatures of noninteracting proteins in membranes: simulations and experiments, Biophys. J. 106 (2014) 1309–1317. D.R. Singh, V. Raicu, Comparison between whole distribution- and average-based approaches to the determination of fluorescence resonance energy transfer efficiency in ensembles of proteins in living cells, Biophys. J. 98 (2010) 2127–2135.

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008

M.R. Stoneman et al. / Biochimica et Biophysica Acta xxx (2016) xxx–xxx [47] A.K. Mishra, M. Gragg, M. Stoneman, G. Biener, J.A. Oliver, P. Miszta, S. Filipek, V. Raicu, P. Park, Quaternary structures of opsin in live cells revealed by FRET spectrometry, Biochem. J. 473 (2016) 3819–3836. [48] W.R. Zipfel, R.M. Williams, W.W. Webb, Nonlinear magic: multiphoton microscopy in the biosciences, Nat. Biotechnol. 21 (2003) 1369–1377. [49] T.B. McAnaney, et al., Protonation, photobleaching, and photoactivation of yellow fluorescent protein (YFP 10C): a unifying mechanism, Biochemistry 44 (2005) 5510–5524.

9

[50] X.X. Kong, E. Nir, K. Hamadani, S. Weiss, Photobleaching pathways in single-molecule FRET experiments, J. Am. Chem. Soc. 129 (2007) 4643–4654. [51] G.H. Patterson, D.W. Piston, Photobleaching in two-photon excitation microscopy, Biophys. J. 78 (2000) 2159–2162.

Please cite this article as: M.R. Stoneman, et al., Quaternary structure of the yeast pheromone receptor Ste2 in living cells, Biochim. Biophys. Acta (2016), http://dx.doi.org/10.1016/j.bbamem.2016.12.008