High Efficacy but Low Potency of δ-Opioid Receptor-G Protein Coupling in Brij-58-Treated, Low-Density Plasma Membrane Fragments

Principal Findings HEK293 cells stably expressing PTX-insensitive δ-opioid receptor-Gi1α (C351I) fusion protein were homogenized, treated with low concentrations of non-ionic detergent Brij-58 at 0°C and fractionated by flotation in sucrose density gradient. In optimum range of detergent concentrations (0.025–0.05% w/v), Brij-58-treated, low-density membranes exhibited 2-3-fold higher efficacy of DADLE-stimulated, high-affinity [32P]GTPase and [35S]GTPγS binding than membranes of the same density prepared in the absence of detergent. The potency of agonist DADLE response was significantly decreased. At high detergent concentrations (>0.1%), the functional coupling between δ-opioid receptors and G proteins was completely diminished. The same detergent effects were measured in plasma membranes isolated from PTX-treated cells. Therefore, the effect of Brij-58 on δ-opioid receptor-G protein coupling was not restricted to the covalently bound Gi1α within δ-opioid receptor-Gi1α fusion protein, but it was also valid for PTX-sensitive G proteins of Gi/Go family endogenously expressed in HEK293 cells. Characterization of the direct effect of Brij-58 on the hydrophobic interior of isolated plasma membranes by steady-state anisotropy of diphenylhexatriene (DPH) fluorescence indicated a marked increase of membrane fluidity. The time-resolved analysis of decay of DPH fluorescence by the “wobble in cone” model of DPH motion in the membrane indicated that the exposure to the increasing concentrations of Brij-58 led to a decreased order and higher motional freedom of the dye. Summary Limited perturbation of plasma membrane integrity by low concentrations of non-ionic detergent Brij-58 results in alteration of δ-OR-G protein coupling. Maximum G protein-response to agonist stimulation (efficacy) is increased; affinity of response (potency) is decreased. The total degradation plasma membrane structure at high detergent concentrations results in diminution of functional coupling between δ-opioid receptors and G proteins.

The prototypical detergent used for preparation of DRMs is Triton X-100. The cell homogenate or membrane preparations are treated with Triton X-100 for 60 min at 0-4°C and fractionated by flotation in density gradients [7,8,18]. Unfortunately, these procedures using the high detergent concentrations (0.5-1%), result in a loss of functional coupling between GPCR and their cognate signaling molecules, trimeric G proteins and adenylyl cyclase (AC) [19]. Membrane domains/caveolae were also prepared by detergent-free method using 1M Na 2 CO 3 (pH 11) and sonication [20,21], however, due to the highly alkaline pH of Na 2 CO 3 , the responsiveness of G proteins and AC activity to GPCR agonist stimulation was also diminished [19,22]. Isolation of the highly purified caveolae from human fibroblasts by detergent-free and alkaline-free procedure was reported by Smart et al. [23], but the amount of protein (10 μg) recovered in final preparation represented 0.13-0.14% of total protein originally present in cell homogenate (7-8 mg). Obviously, the low amount of protein in final preparation of membrane domains/caveolae is not sufficient for characterization of GPCR by radioligand binding assays using 3 H-labelled ligands  Ci/mmol) and measurements of G protein response to agonist stimulation by high-affinity [ 32 P]GTPase or [ 35 S]GTPγS binding assays. Therefore, we tried to improve the methodological conditions for isolation of DRMs and tested different detergents at varying concentrations and detergent/protein ratios with the aim to prepare membrane domains with preserved functional coupling between -opioid receptors (δ-OR) and G proteins. We also tried to find some reasonable compromise between purity and amount of protein recovered in final preparation of DRMs. When using Brij-58 at low concentrations and in optimum range of protein amount in cell homogenate (10-15 mg × ml -1 ), detergent-treated membranes floated up to the low-density end of sucrose gradient and were functional in terms of receptor-G protein coupling. Surprisingly, the efficacy of δ-OR when increasing the high-affinity [ 32 P]GTPase activity or [ 35 S]GTPγS binding was 2-3-fold higher than in membranes prepared in the absence of detergent. The potency of G protein response to agonist stimulation, however, was decreased.  30 Ci/mmol, 35007) was from MP Biomedicals (Santa Ana, CA, USA). Complete protease inhibitor cocktail was from Roche Diagnostic (cat. no. 1697498, Mannheim, Germany), geneticin and hygromycin were from Life Technologies-Gibco (Paisley, Scotland). All other chemicals and drugs were purchased from Sigma-Aldrich.

Cell culture
HEK293T cells stably expressing δ-OR-G i 1α (C 351 I) fusion protein were cultivated in Dulbecco's modified Eagle's medium (Sigma) supplemented with 2 mM (0.292 g × l -1 ) L-glutamine and 10% v/v newborn calf serum at 37°C as described previously [24]. Geneticin (800 μg) or hygromycin (200 μg) were included in the course of cell cultivation. The cells were grown to 80% confluency before harvesting and the beginning of experiments. When indicated, the cells were incubated with (+PTX) pertussis toxin (10 ng × ml -1 ) for 24 h before harvesting.
Preparation of post-nuclear fraction δ-OR-G i 1α (C 351 I)-HEK293 cells (δ-OR-G i 1α cells for short) were harvested from 15 flasks (80 cm 2 ) by centrifugation for 10 min at 300 × g (0-4°C) and homogenized in 50 mM Tris-HCl, pH 7.6, 3 mM MgCl 2 , 1 mM EDTA (TME buffer) containing 1 mM fresh PMSF plus Complete protease inhibitors cocktail in Teflon-glass homogenizer for 7 min at 1360 rpm on ice. The cell homogenate was centrifuged for 7 min at 3500 rpm (1160 × g) to separate cell debris and nuclear fraction (remaining in the sediment) from post-nuclear supernatant, PNS.
Isolation of low-density membranes (LDM) in the presence of low concentrations of non-ionic detergent Brij-58 δ-OR-G i 1α (C 351 I)-HEK293 cells were washed in PBS, harvested from totally 60 flasks (80 cm 2 ) by centrifugation for 10 min at 300 × g (0-4°C). Homogenization was performed in TME buffer containing 1 mM fresh PMSF plus Complete protease inhibitors cocktail in Teflon-glass homogenizer for 7 min at 1360 rpm on ice. The cell homogenate was divided into four identical portions having exactly the same volume. The first portion was diluted with 1/10 volume of water (no detergent). The second, third and fourth portion was diluted with 1/10 volume of water plus 10% w/v Brij-58 to a final concentration of 0.025, 0.05 and 0.1% w/v Brij-58, respectively. Exactly 2 ml volumes of these four samples were transferred into Beckman SW41 tube, mixed with 2 ml of 80% w/v sucrose, overlaid with 35%, 30%, 25%, 20% and 15% w/v sucrose (1.5 ml each) and centrifuged for 24 hours at 187,000 × g (0-4°C). Centrifugation resulted in the separation of two clearly visible layers. The lower, optically dense layer represented the plasma membrane fraction (PM). The low-density membranes (LDM) were visible as a hazy area in the upper part of sucrose gradient. Sucrose fractions 1-12 (1 ml) were collected from top to bottom of the centrifuge tube, snap frozen in liquid nitrogen and stored at -80°C.
Isolation of PM fraction in Percoll density gradient δ-OR-G i 1α (C 351 I)-HEK293 cells were cultivated in 60 flasks (80 cm 2 ), washed in PBS and harvested by centrifugation for 10 min at 1800 rpm (300 × g) in 50 ml conical vials. The cell sediment was homogenized in 250 mM sucrose, 20 mM Tris-HCl, 3 mM MgCl 2 , 1 mM EDTA, pH 7.6 (STEM medium) plus fresh 1 mM PMSF and protease inhibitor cocktail in loosely-fitting, Teflon-glass homogenizer for 7 min at 1360 rpm on ice. The cell homogenate was centrifuged for 7 min at 3500 rpm (1160 × g) to separate cell debris and nuclear fraction (remaining in the sediment) from post-nuclear supernatant, PNS. A 3 ml volume of PNS was applied on the top of 20 ml of 30% v/v Percoll in thick polycarbonate Beckman Ti70 tubes. Centrifugation for 35 min at 30,000 rpm (66,200 × g) resulted in separation of two clearly visible layers. The upper layer represented the plasma membrane-enriched fraction (PM), the lower layer represented mitochondria. The upper layer was diluted 1:4 in STEM medium and centrifuged in Beckman Ti70 rotor for 1.5 hours at 50,000 rpm (175,000 × g). The PM sediment was removed from the compact, gel-like sediment of Percoll, re-homogenized by hand in a small volume of 50 mM Tris-HCl, 1 mM EDTA, pH 7.4 (TE buffer), frozen in liquid nitrogen and stored at -80°C.

[ 32 P]GTPase activity
When screening the density gradient profiles, constant volume aliquots (30 μl) of sucrose gradient fractions were incubated with or without 100 μM DADLE in reaction mix containing in a final volume of 100 μl 10 mM creatine phosphate, 5 units of creatine kinase, 0.5 μM GTP, 1 mM ATP, 0.1 mM adenosine-5´-O-(3-imidotriphosphate (App(NH)p), 1 mM ouabain, 100 mM NaCl, 5 mM MgCl 2 , 2 mM dithiothreitol, 0.1 mM EDTA, 40 mM Tris-HCl, pH 7.5 and [γ-32 P]GTP (about 200,000 dpm per assay) for 30 min at 37°C. The enzyme reaction was terminated by transferring the reaction mix into the ice-cold water bath and adding 0.9 ml of 5% w/v charcoal in 10 mM phosphoric acid. After 5 min at 0°C, the charcoal was removed by centrifugation for 10 min at 10,000 × g and 0.5 ml of clear supernatant was used for determination of radioactivity by liquid scintillation.

[ 3 H]naltrindole binding
Post-nuclear supernatant (PNS), LDM and 0.025% Brij-58-treated LDM were incubated in TME buffer with increasing concentrations of [ 3 H]naltrindole (0.1-10.8 nM). Non-specific binding was determined in the presence of 100 μM naloxone. Binding reaction was continued for 60 min at 30°C, the bound and free radioactivity separated by filtration through GF/B filters in Brandel cell harvestor, the filters were washed three times with 3 ml ice-cold TME and the bound radioactivity was determined by liquid scintillation. Binding data were fitted with the rectangular hyperbola and K d and B max values calculated by GraphPad Prism4. Steady-state anisotropy of fluorescence of hydrophobic membrane probe DPH Percoll-purified PM were labeled with diphenylhexatriene (DPH) by the fast addition (under mixing) of 1 mM stock solution of DPH in freshly distilled acetone to the membrane suspension (0.1 mg protein × ml -1 ; 1 μM final concentration); after 30 min at 25°C, which was allowed to ensure the optimum incorporation of the probe into the membrane interior [26], the anisotropy of DPH fluorescence was measured at Ex 365 nm/Em 425 nm wavelengths. Under these conditions, the fluorescence intensity of the membrane-bound DPH was %500x higher than that of the unbound free probe in aqueous medium; light scattering problems could be therefore omitted. Steady-state fluorescence anisotropy, r DPH , was calculated according to the formula: r DPH = (I vv − I vh )/(I vv + 2I vh ) as described before [27,28].
The time-resolved fluorescence and dynamic depolarization of DPH Fluorescence lifetime and polarization experiments were performed in a time correlated single photon counting (TCSPC) spectrofluorometer IBH 5000 U equipped with a cooled Hamamatsu R3809U-50 micro-channel plate photomultiplier detector as described before [29]. The sample was excited at 373 nm with a diode laser (IBH NanoLED-375L, FWHM 80 ps, 1 MHz repetition rate). The emission monochromator was set to 450 nm with slits set to 16 nm. A 400 nm cut-off filter was placed in the emission part of the instrument to reduce a parasitic light from the laser source. The sample was continuously mixed and kept at the constant temperature of 25°C. Measurement consisted in recording of fluorescence decays at 4 different orientations of the emission and excitation polarizers. The anisotropy free decay I (t), which we used for the lifetime determination, was calculated according to Eq (1): The G-factor (G) was determined by measuring a standard solution of POPOP and calculated according to Eq (2): where I hv corresponds to the signal measured with the horizontally polarized excitation and vertically polarized emission, and I hh to the excitation and emission both polarized horizontally. In order to obtain fluorescence lifetimes, the I(t) was fitted with a two-exponential decay (Eq 3): yielding lifetimes τ 1 and τ 2 and corresponding amplitudes B 1 and B 2 .
The decay of the anisotropy r(t) was determined according to Eq (4): and fitted with the formula (Eq 5): rðtÞ ¼ ðrð0Þ À rð1ÞÞ:expðÀt=Þ þ rð1Þ; ð5Þ where r(0), and r(1) stands for the limiting and residual anisotropy, respectively and ϕ is the rotational correlation time. The anisotropy decays were fitted by the non-linear least squares method including the impulse re-convolution with the instrumental response function (fwhm~100 ps). χ2 generated by the IBH software package served as goodness of fit criterion. The anisotropy data were analyzed according to the "wobble in cone" model introduced by [30,31]. Wobbling diffusion constant (D w ) and S-order parameter S were calculated according to Eqs 6 and 7, where r(0), and r(1) stand for limiting and residual anisotropy, ϕ is the rotational correlation time and σ s is the relaxation time which is a function of the S-order parameter.
detergent. Subsequently, these 4 parts of homogenate, containing exactly the same protein concentration, were fractionated in parallel by flotation in sucrose density gradients for 24 hours and the basal and DADLE-stimulated, high-affinity [ 32 P]GTPase was measured in gradient fractions (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) collected from the top to bottom of the centrifuge tube as described in Methods. The high level of total [ 32 P]GTPase activity per fraction (pmol × ml -1 ) and specific [ 32 P] GTPase activity (pmol × min -1 × mg -1 ) was detected in "floating", low-density fractions 1-6 ( Fig 1). Membranes recovered in these fractions were collectively termed (in all 4 gradients) the low-density membranes, LDM. In LDM prepared in the absence of detergent (no detergent), DADLE-stimulation represented 180-240% of the basal level. In 0.025% Brij-58-treated GTPase activity. Cell homogenate prepared from δ-OR-G i 1α cells was divided by volume into 4 identical parts. The three parts were treated with increasing concentrations of 0.025, 0.05 and 0.1% w/v Brij-58, respectively; the fourth part represented the control sample containing no detergent. Subsequently, these 4 parts of homogenate, containing exactly the same protein concentration, were fractionated by centrifugation in 15/20/25/30/35/40% w/v sucrose density gradient as described in Methods. Upper panels. Basal () and DADLE-stimulated (•), high-affinity [ 32 P]GTPase activity was measured in gradient fractions 1-12 collected from top to the bottom of the centrifuge tube and expressed as total activity (pmol × min -1 × ml -1 ) or specific enzyme activity (pmol × min -1 × mg -1 protein) in each fraction. Lower panels. The ratio between the agonist-stimulated (DADLE) and basal level of specific enzyme activity was expressed as % of agonist stimulation over the basal level; the basal level represented 100%. The amount of protein in fractions was determined by Lowry method and expressed as mg per fraction (mg × ml -1 ). Results represent the average of 4 experiments ± SEM. The significance of difference between the specific DADLE-stimulated and basal [ 32 P]GTPase activity in fractions 1-12 collected from sucrose density gradient 1 (no detergent), 2 (0.025% Brij-58), 3 (0.5% Brij-58) and 4 (0.1% Brij-58) was determined by Student´s t-test [see (A) in S1 Table]. Furthermore, net increment of agonist stimulation (Δ DADLE ) was calculated as the difference between specific DADLE-stimulated and basal [ 35 P]GTPase activity in fractions 1-6. The significance of difference of Δ DADLE values in the four types of sucrose density gradients was determined by one-way ANOVA followed by Bonferroni´s multiple comparison test using GraphPad Prism4 [see (B) in S1 Table].
LDM, agonist-stimulation was increased to 220-370%. The highest increase of DADLE stimulation (%400%) was observed in 0.05% Brij-58-treated LDM. The further increase of detergent concentration to 0.1% was reflected in diminution of both basal and DADLE-stimulated [ 32 P] GTPase activity to the zero level (Fig 1, right-hand panels). Net increment of DADLE-stimulated [ 32 P]GTPase activity was also higher in 0.025% and 0.05% Brij-58-treated LDM when compared with LDM [see (B) in S1 Table].
Similar density gradient profiles were obtained by measurement of the high-affinity [ 35 S] GTPγS binding in fractions 1-12 (Fig 2). The 0.025% Brij-58-treated LDM recovered in lowdensity fractions 1-6 exhibited % 2-fold higher % of DADLE stimulation over the basal level when compared with the same fractions prepared from homogenate containing no detergent, GTPγS binding. Cell homogenate prepared from δ-OR-G i 1α cells was divided by volume into 3 identical parts. The two parts were treated with 0.025% and 0.1% w/v Brij-58; the third part represented the control sample containing no detergent. Subsequently, these 3 parts of homogenate, containing exactly the same protein concentration, were fractionated by centrifugation in 15/20/25/30/35/40% w/v sucrose density gradients as described in Methods. Upper panels. Basal () and DADLE-stimulated (•) [ 35 S]GTPγS binding was measured in fractions 1-12 and expressed as total binding (pmol × ml -1 ) or specific binding in each fraction (pmol × mg -1 protein). Lower panels. The ratio between the agonist-stimulated (DADLE) and the basal level of [ 35 S]GTPγS binding was calculated from specific binding data (pmol × mg -1 protein) and expressed as % of agonist stimulation over the basal level; the basal level represented 100%. Results represent the average of 3 experiments ± SEM. The significance of difference between the specific DADLE-stimulated and basal [ 35 S]GTPγS binding in fractions 1-12 collected from the three types of sucrose density gradients was determined by Student´s t-test [see (A) in S2 Table]. Net increment of agonist stimulation (Δ DADLE ) was calculated as the difference between specific DADLE-stimulated and basal [ 35 S]GTPγS binding in fractions 1-6. The significance of difference of Δ DADLE values in the three types of sucrose density gradients was determined by one-way ANOVA followed by Bonferroni´s multiple comparison test using GraphPad Prism4 [see (B) in S2 Table].
LDM. The LDM isolated from homogenate treated with 0.1% Brij-58 exhibited the zero level of both basal and DADLE-stimulated [ 35 S]GTPγS binding (Fig 2, right-hand panels). Net increment of DADLE-stimulated [ 35 S]GTPγS binding was also significantly higher in 0.025% than in LDM [see (B) in S2 Table].
In the following step of our work, PNS, LDM and 0.025% Brij-58-treated LDM were compared by measurement of [ 35 S]GTPγS/GTPγS competitive inhibition curves determined in the presence or absence of 100 μM DADLE (Fig 6). Purification of LDM by flotation in sucrose gradient and increase of efficacy by detergent was documented again. The DADLE-stimulated [ 35 S]GTPγS binding level at zero concentration of GTPγS (B 0 ), was increased from 0.22 ± 0.01 pmol × mg -1 in PNS to 1.08 ± 0.02 and 1.19 ± 0.02 pmol × mg -1 in LDM and 0.025% Brij-58-treated LDM, respectively. The net increment of agonist stimulation (ΔB 0 ), calculated as the difference between B 0 (+DADLE) and B 0 (-DADLE), was increased from 0.11 ± 0.01 pmol × mg -1 in PNS to 0.61 ± 0.03 pmol × mg -1 and 0.98 ± 0.02 pmol × mg -1 in LDM and Brij-58-treated LDM, respectively ( Table 4). Comparison of IC 50 values in PNS, LDM and , low-density membranes (LDM) and 0.025% Brij-58-treated LDM prepared from δ-OR-G i 1α cells as described in Methods. LDM and 0.025% Brij-58-treated LDM represent the mixed pool of membranes in combined fractions 1-6 prepared in the absence or presence of detergent, respectively. The net increment of agonist-stimulation (Δ DADLE ) and DADLE concentration inducing the half-maximum response (EC 50 ) were calculated by GraphPad Prism4. Lower panels. The ratio between the agonist-stimulated (DADLE) and basal level of specific enzyme activity was expressed as % of agonist stimulation over the basal level; the basal level represented 100%. Results represent the average of 3 experiments ± SEM. The significance of difference of EC 50 and Δ DADLE in PNS versus LDM versus 0.025% Brij-58-treated LDM was determined by one-way ANOVA followed by Bonferroni´s multiple comparison test using GraphPad Prism4 (see S4 Table). Brij-58-treated LDM indicated no significant difference between the three types of membranes in this type of experiment (Fig 6).
We have also determined the number of δ-opioid receptors in PNS, LDM and 0.025% Brij-58-treated LDM by saturation binding assay with antagonist [ 3 H]naltrindole (Fig 7). As before, purification of δ-OR-G i 1α protein was evidenced again because B max of [ 3 H]naltrindole binding in LDM (3.61 ± 0.21 pmol × mg -1 ) was 7-fold higher than in PNS (0.56 ± 0.21 pmol × mg -1 ). However, in 0.025% Brij-58-treated LDM, number of receptor sites was less than half of this amount (B max = 1.74 ± 0.11 pmol × mg -1 ). This result may be interpreted as the detergent-  Tables 3 and 4). These results may be simply interpreted as a loss of about half of receptor sites in 0.025% Brij-58-treated LDM when compared with LDM.
It may be therefore concluded that Brij-58-treated LDM prepared at a low detergent concentrations exhibit the high efficacy when activating their cognate G proteins, but the affinity of agonist response, expressed as EC 50  δ-OR-G protein coupling in low-density membranes prepared from PTXtreated δ-OR-G i 1α (C 351 I)-HEK293 cells at 0°C; effect of non-ionic detergent Brij-58 Virtually the same results were obtained in analysis of 0.025% Brij-58-treated membranes prepared from δ-OR-G i 1α cells exposed to pertussis toxin (10 ng × ml -1 ) for 12 hours. The screening of density gradient profiles indicated that the peak of DADLE-stimulated [ 35 S]GTPγS binding was shifted to lower densities: from fractions 5+6, which were combined together and corresponded to plasma membrane fractions, to fraction 4. The net increment of agonist  (Fig 8). The average net increment of agonist stimulation in combined fractions 4+5+6 was 1.8-fold higher than in the same fractions prepared in the absence of detergent. In this set of experiments, we have also determined theδ-OR number by agonist binding assay using[ 3 H]DADLE. As expected from the previous results of antagonist binding (Fig 7), the level of specific [ 3 H]DADLE binding in 0.025% Brij-58-treated fractions 5-6 was %2-fold lower than in the same fractions prepared in the absence of detergent (Fig 9). The detergentinduced shift of receptors to lower densities was noticed as an increase of [ 3 H]DADLE binding in fraction 4 from 0.60 ± 0.02 to 1.21 ± 0.03 pmol × mg -1 proceeding in parallel with the decrease of agonist binding in fractions 5-6 from 1.86 ± 0.03 to 1.00 ± 0.03 pmol × mg -1 . The sum of [ 3 H]DADLE binding in fractions 4, 5 and 6 was decreased by Brij-58 from 4.32 ± 0.04  50 and Δ DADLE in PNS versus LDM versus 0.025% Brij-58-treated LDM was determined by one-way ANOVA followed by Bonferroni´s multiple comparison test using GraphPad Prism4 (see S5 Table).  Table 5).
Thus, determination of functional response of G proteins to agonist stimulation in membranes prepared from PTX-treated δ-OR-G i 1α (C 351 I)-HEK293 cells and normalization of this response to δ-OR number determined by agonist binding assay confirmed our previous results obtained in analysis of membranes prepared from PTX-untreated δ-OR-G i 1α (C 351 I)-HEK293 cells.

Direct effect of Brij-58 on δ-OR-G protein coupling in isolated plasma membranes at 30°C
As noticed before in studies of baclofen-stimulated [ 35 S]GTPγS binding in Percoll-purified PM prepared from rat brain cortex [29], the non-ionic detergents Brij-58, Triton X-100 and digitonin were able, within the narrow range of detergent concentrations, to increase GABA B -R agonist-stimulated [ 35 S]GTPγS binding when added directly to the assay medium. Exposure to higher detergent concentrations resulted in drop of G protein activity to zero level. In the case of Brij-58, this range was between 0.006 and 0.013% (w/v). Similar result was observed in PM isolated from δ-OR-G i 1α (C 351 I)-HEK293 cells: the increase of DADLE-stimulated [ 35 S] GTPγS binding was detected at 0.006% Brij-58 (Fig 12). Importantly, activation of G proteins at this detergent concentration was detected in PM isolated from both PTX-untreated and  Fig 12). Therefore, the effect of Brij-58 on [ 35 S]GTPγS binding was manifested for G i 1α present within δ-OR-G i 1α (C 351 I) fusion protein as well as for PTX-sensitive G proteins endogenously expressed in HEK293 cells.

PTX-treated δ-OR-G i 1α cells (compare upper and lower panels in
In accordance with the data obtained in analysis of LDM isolated at 0°C, comparison of the dose-response curves of DADLE-stimulated [ 35 S]GTPγS binding in plasma membranes exposed to 0.006% Brij-58 at 30°C indicated a highly significant decrease in potency of agonist response: EC 50 value was increased from 24 ± 5 nM (no detergent) to 120 ± 10 nM in PM exposed to 0.006% Brij-58 (Fig 13, S13 Table). The agonist binding to δ-OR was not affected at this detergent concentration (Fig 14A, S14 Table) and 0.02% concentration of Brij-58 was needed to achieve the half-maximum inhibition of [ 3 H]DADLE binding (Fig 14B). It should be noticed that at this detergent concentration, DADLE-stimulated [ 35 S]GTPγS binding was completely diminished (Fig 12). Therefore, agonist binding to δ-OR was substantially less  50 and ΔB 0 values in PNS versus LDM versus 0.025% Brij-58-treated LDM was determined one-way ANOVA followed by Bonferroni´s multiple comparison test using GraphPad Prism4 (see S6  Table).
doi:10.1371/journal.pone.0135664.g006 δ-OR-G Protein Coupling in Brij-58 PM Fragments sensitive to deteriorating effect of detergent than receptor ability to activate G proteins. This conclusion was in our experiments demonstrated for PTX-insensitive G i 1α (C 351 I) covalently bound to δ-OR as well as for PTX-sensitive G proteins endogenously expressed in HEK293 cells.

The effect of Brij-58 on hydrophobic plasma membrane interior
Measurement of steady-state anisotropy of hydrophobic membrane probe DPH indicated that Brij-58 exhibited a strong "fluidization" of hydrophobic plasma membrane interior (Fig 15). When increasing Brij-58 above the critical micelle concentration (CMC = 0.0086% w/v), the highly polarized signal of DPH observed in detergent-untreated PM (r DPH = 0.224 ± 0.003) was gradually decreased to the highly depolarized signal at high detergent concentrations (r DPH = 0.087 ± 0.003), which was close to that of Brij-58 alone in aqueous medium at 25°C (r DPH = 0.072 ± 0.001).
More detailed understanding of the detergent effect on organization of PM microenvironment was obtained by determination of the lifetime values of DPH fluorescence and analysis of the time-resolved DPH anisotropy decay by "wobble in cone" model [30,31]. As shown in Table 6, the average lifetime of DPH fluorescence (τ av ) was continuously decreased by increasing detergent concentrations. The S-order parameter decreased as well and this decrease proceeded in parallel with the increase of wobbling diffusion constant D w (Fig 16). Accordingly, the values of rotational correlation time (ϕ) were decreased. Thus, the lower friction of DPH molecules within plasma membrane microenvironment and higher rate of DPH rotation was detected.

Discussion
According to the collision-coupling model of receptor-G protein interaction [32], the level of G protein activated is independent of the concentration of agonist-bound receptors, but the rate of G protein activation is proportional to the concentration of agonist-bound receptors. Thus, reduction of receptor density should cause no reduction in the maximal level of G protein  δ-OR-G Protein Coupling in Brij-58 PM Fragments activation but it should reduce the rate of G protein activation and decrease the EC 50 values of agonist-stimulated G protein activity. It was also suggested that diffusion of receptors and effectors is slow, so that receptors will only activate effectors that are in their vicinity at the time of agonist occupation [33,34]. The rate of G protein activation was found to be dependent on the agonist binding frequency relative to the encounter time of G protein and receptor [35,36]. The test of validity of collision-coupling model for opioid receptors was performed in C6 cells expressing different amounts of μand δ-OR and G proteins [37]. The time-course of full or partial agonist-stimulated [ 35 S]GTPγS binding did not vary in cells containing widely different amounts of μ-OR or δ-OR and the 10-fold reduction of the functional complement of G proteins by pertussis toxin caused the decrease of maximum agonist-stimulated [ 35 S]GTPγS binding. The association rate of [ 35 S]GTPγS with agonist-activated G protein was unchanged. Results of this study were not compatible with the random collision-coupling model but Saturation binding studies of specific δ-opioid antagonist [ 3 H]naltrindole sites in PNS, LDM and 0.025% Brij-58-treated LDM was measured as described in Methods. The B max and K d values were calculated by GraphPadPrism4. The data represent the average of 3 experiments ± SEM. Significance of difference of B max and K d values in PNS versus LDM versus 0.025% Brij-58-treated LDM was determined one-way ANOVA followed by Bonferroni´s multiple comparison test using GraphPad Prism4 (see S7 Table).
doi:10.1371/journal.pone.0135664.g007 δ-OR-G Protein Coupling in Brij-58 PM Fragments supported a compartmentalization model [38][39][40] in which receptors and G proteins are "in some way" associated together and compartmentalized within plasma membrane.  S8 Table). The association of GPCRs and G proteins with the plasma membrane makes them susceptible to their lipid environment so that lipid-protein interactions are important for their function [41]. In direct relevance to theoretical models of receptor-G protein coupling, plasma membrane domains/rafts were suggested as devices concentrating the signaling molecules, i.e. bringing them into a closer spatial organization than just the random order arrangement within the fluid mosaic of proteins and lipids in the bulk of plasma membranes [1][2][3][4][5][6][9][10][11][12]16]. The content of trimeric G proteins and GPCR in these PM compartments responded dramatically to agonist stimulation [22,23,[42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59].  S9 Table). More specifically, κ-OR were found to be localized in lipid rafts [60] and analysis of μ-ORand δ-OR-G protein coupling indicated the compartmentalization of these signaling molecules in PM [37]. Cholesterol reduction by methyl-β-cyclodextrin was found to attenuate δ-OR-initiated signaling in neuronal cells but enhance it in non-neuronal cells [57]. In HEK293 cells stably expressing δ-OR-G i 1α fusion protein, cholesterol depletion by β-cyclodextrin resulted in decrease of potency of G protein response to agonist DADLE stimulation, but the maximum response was unchanged [26]. Adenylyl cyclase super-activation induced by prolonged exposure to morphine was found to be dependent on receptor localized within lipid rafts and independent of receptor internalization [61]. Palmitoylation and membrane cholesterol stabilized the homo-oligomerization of μ-OR and coupling with G proteins [62].
Furthermore, experiments based on plasmon-waveguide resonance spectroscopy [63] indicated that the unbound, inactive δ-OR receptors were enriched in 1, 2-palmitoyl-oleoyl-sn-glycero-3-phosphoserine (POPC)-rich domain of lipid bilayer composed of 1:1 mixture of POPC and sphingomyelin (SM). The agonist-and antagonist-bound receptors were concentrated in SM-rich component (with a two-fold greater propensity in the case of the agonist) characterized by a greater thickness. Since it is known that receptor activation involves changes in the orientation of trans-membrane helices with an increase in receptor vertical length, i.e. perpendicular to the plane of the plasma membrane [64], it was hypothesized that the driving force for receptor redistribution comes from the increased hydrophobic match of the elongated receptor for the thicker SM-enriched part of the bilayer, a sorting mechanism that has previously been observed in other subcellular compartments [65,66].
Literature data mentioned above justify the effort to analyze in details the interaction of detergents with various plasma membrane preparations containing the functional GPCR and G proteins with the aim to find "some correlation" between the structural and dynamic state of the membrane lipid bilayer and parameters of GPCR-G protein coupling. Obviously, this type of analysis requires the step by step determination of concentration dependency of detergent effect.
Comparison of LDM isolated from PTX-untreated and PTX-treated δ-OR-G i 1α (C 351 I)-HEK293 cells in the absence or presence of low concentrations of Brij-58 at 0°C indicated the decrease of δ-OR number in Brij-58-treated LDM when compared with detergent-untreated LDM (Figs 7 and 9). The decrease of δ-OR was measured by both agonist and antagonist binding assays and accompanied by a decrease in potency of agonist response (Figs 4, 5 and 10). This was demonstrated by comparison of the dose-response curves of both DADLE-stimulated  4) and [ 35 S]GTPγS binding (Figs 5 and 10). However, the maximum G protein response (efficacy) was significantly increased ( Tables 1-5). Similar data were obtained in analysis of the direct effect of increasing concentrations of Brij-58 on isolated plasma membranes at 30°C (Figs 12-14).
When interpreting these results, it is logical to assume that the increase of detergent concentration at hydrophobic interface of δ-OR (exposed to the bulk lipid phase of the membrane) removes lipids attached to the outer surface of protein molecule (lipid annulus) and in this way perturbs the optimum lipid environment of δ-OR needed for its optimum function. This is reflected first as a transitional increase of efficacy of G protein response to agonist stimulation accompanied by decrease of potency. Subsequently, at high detergent concentration, receptor is unable to find the proper, agonist-responding conformation and G protein response in completely diminished.
It may be also visualized that limited degradation of organized PM structure by detergent in critical range of low concentrations (0.01-0.1% Brij-58 at 0°C) might detach G proteins from the inhibitory effect of caveolin [15,50,67,68] or that the detergent-induced change of structural organization of lipid annulus surrounding δ-OR alters the oligomerization state of OR and, consequently, parameters of functional coupling between OR and their cognate G proteins [69][70][71][72][73][74][75][76][77].
Our previous results [29] indicated that DPH anisotropy (r DPH ) represents a useful parameter for analysis of detergent effect on membranes isolated from brain. In δ-OR-G i 1α (C 351 I)-HEK293 cells, the decrease of r DPH by Brij-58 (Fig 15) was 5-fold higher than that induced by  [26]. The highly polarized signal of DPH, measured in PM in absence of detergent (r DPH = 0.224), was decreased to the highly depolarized signal (r DPH = 0.087) detected at high detergent concentrations. The application of "wobble in cone" model enabled us to retrieve the information about membrane dynamics characterized by wobbling diffusion constant (D w ). This model also enabled us to gain the "static" information about the degree of the orientation constrains of rotational movements of DPH dye by calculation of S-order parameter (S). The time-resolved analysis of DPH fluorescence by the "wobble in cone model" of DPH motion ( Table 6 and Fig 16) indicated that the exposure of PM to the increasing  S12 Table).  The existence of "critical range" of detergent concentrations inducing the limited membrane damage before the total collapse of plasma membrane structure and diminution of membrane protein function was also detected for GABA B -receptors in plasma membranes isolated from rat forebrain cortex [29]. In close analogy to the data presented in this work (Fig 12), the clearly defined stimulatory [Brij-58, 0.006-0.013% w/v; Triton X-100, 0.013-0.025%) and   Conclusions 1) This work describes the simple method for preparation of intermediate forms of low-density, detergent-treated plasma membrane fragments (Brij-58-treated LDM) which are distinct from detergent-untreated PM fragments (LDM) exhibiting the same density but, on the other hand, do not represent the typical detergent-resistant membrane domains (DRMs) isolated in 0.5-1% Triton-X100 at 0°C. The "typical" DRMs contain less than 1% of original PM protein, large portion (30-40%) of trimeric G proteins, caveolin and flotilin but do not exhibit the functional coupling between δ-OR and G proteins due to deleterious effect of high detergent concentrations.
2) In optimum range of detergent concentrations (0.025-0.05% w/v), Brij-58-treated, lowdensity membranes exhibited 2-3-fold higher efficacy of DADLE-stimulated, high-affinity [γ-32 P]GTPase and [ 35 S]GTPγS binding than membranes of the same density prepared in the absence of detergent. The potency of agonist DADLE response was significantly decreased. At high detergent concentrations (>0.1%), the functional coupling between δ-OR and G proteins was completely diminished. The same detergent effects were measured in LDM isolated from PTX-treated cells. Therefore, the effect of Brij-58 on δ-OR-G protein coupling was not δ-OR-G Protein Coupling in Brij-58 PM Fragments restricted to the covalently bound G i 1α within δ-OR-G i 1α fusion protein, but it was also valid for PTX-sensitive G proteins of G i /G o family endogenously expressed in HEK293 cells.