Quantitative encoding of a partial agonist effect on individual opioid receptors by multi-site phosphorylation and threshold detection

Phosphopeptide mapping of MOR phosphorylation from intact cells by MALDI MS

The reduced endocytic efficacy of morphine relative to the highly efficacious enkephalin analogue [D-Ala2, N-MePhe4, Gly-ol]-enkephalin (DAMGO) is clearly established in human embryonic kidney (HEK293) cells expressing Flag epitope-tagged MORs (15, 17). Differences in the endocytic effects of various opioids established in this cell model have proven to be relevant to differential regulation occurring in native neurons (e.g., (19, 20, 23, 25, 28)). We thus focused on HEK293 cells as a model for investigating agonist effects on MOR phosphorylation by mass spectrometry, after rapid immunopurification of receptors (Fig. 1A). Flag-tagged MORs were efficiently isolated from cell extracts, and exhibited electrophoretic mobility (Fig. 1B) consistent with that of the full-length receptor glycoprotein produced in cultured cells and native brain tissue (17, 41).

To focus specifically on differences in agonist efficacy rather than affinity (10), we applied each agonist at a saturating concentration (10 µM) sufficient to drive maximal receptor occupancy (42). Phosphorylation and endocytic trafficking of MORs through the recycling pathway occur dynamically and iteratively in the presence of opioid peptide (43), and considerable agonist-selectivity of net MOR internalization can be observed within 20 minutes after application in HEK293 cells (17); accordingly, we chose this time point for initial comparison. We first surveyed tryptic phosphopeptides covering the entire C-tail (Fig. 1C) by MALDI mass spectrometry (Fig. 1D). Metal ion affinity chromatography (IMAC), used to enrich phosphorylated peptides from the mixture (44), revealed prominent signals representing phosphorylated forms of a proximal region of the C-tail ('region 1' corresponding to phosphorylated EFCIPTSSTIEQQNSAR) as well as a middle region ('region 2' corresponding to phosphorylated IRQNTREHPSTANTVDR) (Fig. 1E). The distal portion of the C-tail ('region 3' corresponding to TNHQLENLEAETAPLP, or DRTNHQLENLEAETAPLP, using AspN), in contrast, was not detectably phosphorylated.

Quantitative analysis of C-tail phosphorylation using SILAC and LC-MS techniques

We quantified agonist effects on the abundance of phosphorylated species by LC-MS and stable isotope labeling in cell culture (SILAC) techniques (45), using isotope-tagged arginines (Fig. 2A) to achieve complete coverage of the C-tail (Fig. 2B, see Table S1 for corresponding m/z values). Additionally, this approach allowed us to assure co-elution of isoptomers (peptides containing the three different isotopic labels) in the LC separation step preceding peptide ionization (45). Two singly phosphorylated (+1Pi) forms of region 1 (peptides 1 and 2) were resolved by LC (Fig. 2C), and differed in whether phosphorylation was present in Ser363 (peptide 1, fig. S1) or in the 354TSST357 Ser/Thr cluster (peptide 2, fig. S2). Calculation of the isotope ratios for peptide 1 indicated relative abundances near unity for morphine vs no drug (M/ND), DAMGO vs no drug (D/ND), and DAMGO vs morphine (D/M) comparisons (Fig. 2D, Table S2), indicative of constitutive phosphorylation. By contrast, peptide 2, containing single phosphorylation in the 354TSST357 motif, was agonist-dependent (Fig. 2E), since its abundance was markedly increased by exposure of cells to both morphine (M/ND isotope ratio = 4.6) and DAMGO (D/ND = 9.3). Single phosphorylation in this region was also moderately agonist-selective, because DAMGO increased phosphorylated peptide abundance more effectively than morphine (D/M = 2.0). We also detected dual (+2Pi) phosphorylation in region 1, and phosphopeptides containing this modification were isolated from cells under all treatment conditions (peptide 3, Fig. 2F). Production of this species was agonist-dependent (M/ND = 2.8; D/ND = 2.9), but not agonist-selective (D/M = 1). In summary, none of the phosphorylated forms of region 1 was unique to a particular treatment condition, and the degree of agonist-selectivity observed in the production of any particular species was moderate (D/M ≤ 2).

Region 2 of the C-tail was represented by two overlapping proteolytic fragments ([368–382] and [372–383], Fig. 3A), suggesting that phosphorylation of a fraction of receptors occurred next to the protease cleavage site (46). MS/MS analysis of the singly phosphorylated [368–382] species (peptide 4) verified phosphorylation specifically at Thr370 (fig. S3). The abundance of this species was moderately increased by both morphine (M/ND = 1.4) and DAMGO (D/ND = 2.5), with DAMGO having a stronger effect (D/M = 1.8) (Fig. 3B). Single phosphorylation was also detected in the 375STANT379 Ser/Thr motif (peptide 5, Fig. 3C), revealing similar fragmentation patterns under the three different treatment conditions (fig. S4). The abundance of single phosphorylation in this motif was increased by morphine (M/ND = 1.6) and DAMGO (D/ND = 2.7), with DAMGO again having a somewhat stronger effect (D/M = 1.7). Dual phosphorylation in the fragment containing both Thr370 and the 375STANT379 Ser/Thr motif (peptide 6, Fig. 3D) was also observed under all treatment conditions. Quantification indicated that the abundance of this species was markedly increased by both morphine (M/ND = 3.5) and DAMGO (D/ND = 9.2), again with DAMGO having a moderately stronger effect (D/M = 2.6). Strikingly, the only phosphorylated species that was absolutely agonist-dependent (i.e., not detectable in untreated cells) included dual phosphorylation within the 375STANT379 Ser/Thr cluster (peptide 7, Fig. 3E). Further, this phosphopeptide species was the most highly agonist-selective in its production by DAMGO compared to morphine (D/M = 4.3). Thus, higher-order phosphorylation in region 2, involving the 375STANT379 Ser/Thr cluster, differentiates the effects of morphine and DAMGO in a disproportional manner relative to other phosphorylations observed.

Persistent agonist-selective phosphorylation in the 375STANT379 motif

We carried out a separate analysis to independently verify agonist-selective phosphorylation in region 2, and to examine if skewed production of higher order phosphorylation in this region could also be observed after longer periods of agonist exposure. We were particularly interested in a longer time point because morphine's duration of therapeutic action typically exceeds 20 min, and because agonist-selective trafficking is evident in cultured cells even after prolonged agonist exposure (16, 47). Thus, we carried out a separate labeling experiment comparing agonists after continuous application for 3 hours, approximating the clinical duration of morphine action in humans (47). To assure independent analysis, we focused on a distinct proteolytic fragment that includes the 375STANT379 motif of interest, but is generated by alternate cleavage at the proximal Arg residue in region 2 (Fig. 4A). Production of single phosphorylation (+1Pi) in this peptide (peptide 8) was moderately agonist-selective (D/M = 2.7, Fig. 4B), and the degree of agonist-selectivity increased progressively for production of higher-order phosphorylated forms (peptides 9 and 10, representing +2Pi and +3Pi species, respectively; Fig. 4C and D). The triple (+3Pi) phosphorylated form was completely undetectable in untreated cells and exhibited very pronounced agonist-selectivity in its production, as shown by a relative abundance ratio (D/M) of 14.8-fold between morphine and opioid peptide exposure (Fig. 4D, Table S2). These results independently verify agonist-selective phosphorylation in region 2, show that this phenomenon persists with longer agonist exposure, and emphasize the remarkable degree to which distinct agonists can drive disproportional higher-order phosphorylation of this region.

Multi-site phosphorylation in the 375STANT379 motif regulates endocytosis

We next investigated whether phosphorylation in either Ser/Thr cluster (354TSST357 or 375STANT379) present in C-tail regions 1 and 2, respectively, is functionally relevant to the control of MOR endocytosis. Mutating all phosphorylatable residues in the proximal cluster in region 1 (354AAAA357) had no detectable effect on steady state receptor localization or on agonist-selective internalization, as visualized by fluorescence microscopy. However, mutating the more distal cluster in region 2 (375AAANA379) resulted in a visibly reduced ability of DAMGO to induce receptor internalization (Fig. 5A). Complementary fluorescence flow cytometry experiments quantitatively verified the specificity and reproducibility of these results (Fig. 5B). Thus, MOR endocytosis is controlled specifically by phosphorylation within the C-tail involving the 375STANT379 Ser/Thr cluster, which, notably, represents a sequence motif conserved in MORs across vertebrates (48).

Because the SILAC-LC/MS data revealed the most pronounced agonist selectivity in production of higher-order phosphorylation in the C-tail region that includes the 375STANT379 motif, we investigated whether the endocytic activity conferred by this motif specifically requires its phosphorylation at more than one residue. To do so, we mutated each potential phosphorylation site individually, and examined the effect on regulated endocytosis across multiple clones of stably transfected cell lines selected for similar surface receptor levels in untreated cells. Interestingly, mutating any individual Ser/Thr residue in the 375STANT379 motif retained the plasma membrane localization of receptors (Fig. 6A) and inhibited regulated endocytosis (Fig. 6B) to a similarly large degree as mutating all three in combination. These results suggest that full endocytic activity conferred by the 375STANT379 motif requires its phosphorylation at more than one residue.

β-arrestin recruitment to clathrin-coated pits functions as a specific detector of higher-order phosphorylation involving the 375STANT379 motif

We next investigated the mechanism by which multi-site phosphorylation is interpreted to produce differential endocytosis of MORs. We focused on arrestins because they are known to associate with preferentially with phosphorylated 7TMRs, including MORs, and act as regulated endocytic adaptors by binding both to the 7TMR and the assembled coat structure of clathrin-coated pits (32, 33). Previous studies have clearly established that regulated endocytosis of MORs is arrestin-dependent (22, 49, 50), and manipulations that increase phosphorylation of the MOR C-tail in intact cells enhance both arrestin recruitment and receptor endocytosis (21, 22, 37). However, it is not known whether agonist-selective engagement of this endocytic adaptor is determined simply by a proportional difference in net phosphorylation of the overall receptor pool, or if arrestins discriminate activated MORs according to specific higher-order phosphorylation in the C-tail.

To address this question, we investigated the potential role of 375STANT379 phosphorylation in controlling recruitment of β-arrestin-2 to discrete clathrin-coated pits. DAMGO-induced activation of MORs robustly recruits GFP-tagged β-arrestin-2 to coated pits, while morphine-induced recruitment is below the detection limit of this assay (22, 34, 51). To investigate this process with high spatiotemporal resolution, we thus focused on DAMGO and imaged a GFP-tagged version of β-arrestin-2 in living cells by total internal reflection fluorescence microscopy (TIR-FM). Activation of wild type MORs using 10 µM DAMGO elicited rapid and pronounced recruitment of β-arrestin to diffraction-limited spots on the plasma membrane (Fig. 7A, top row), shown previously to represent discrete clathrin-coated pits containing receptor-arrestin complexes (43, 52). Mutating all three phosphorylatable residues in the 375STANT379 motif (375AAANA379 mutant) visibly inhibited β-arrestin recruitment (Fig. 7A, bottom row), implicating the 375STANT379 motif in regulating this critical endocytic adaptor. This represented an effective blockade, rather than a kinetic delay, because β-arrestin clustering was reduced in continuous image series (Fig. 7B) spanning the typical surface residence time of individual coated pits (45, 47). This result was verified quantitatively by TIR-FM analysis across multiple fixed specimens, and was specific to the 375STANT379 motif because mutating all four residues in the 354TSST357 cluster did not prevent β-arrestin clustering (Fig. 7C). Interestingly, mutating any individual phosphorylatable residue in the 375STANT379 motif caused a pronounced inhibition of arrestin recruitment that was indistinguishable from mutating all three residues together (Fig. 7C), fully mirroring the effect of the same mutations on MOR endocytosis (Fig. 6). Thus, arrestins appear to function as a 'decoding' device for agonist-selective endocytosis of MORs, by selectively recognizing multi-phosphorylated receptor species involving the second Ser/Thr cluster that are produced in markedly disproportional amount by morphine relative to opioid peptide.

Materials

¹³C-labeled Arginine (¹³C L-Arg:HCl) and ¹³C¹⁵N-labeled Arginine (¹³C¹⁵N-L-Arg:HCl) were from Cambridge Isotope Laboratories (Andover, MA). Dulbecco’s modified Eagle’s medium (DME) and dialyzed FBS were from Invitrogen. Dulbecco’s modified Eagle’s medium (DME) deficient in L-Arginine was from Athena ES (Baltimore, MD). L-Arginine (L-Arg:HCl) and n-Dodecyl b-D-maltoside (DDM) were from Sigma (St. Louis, MO). Sequencing grade endoproteases Asp-N and modified bovine trypsin were purchased from Roche Applied Science (Indianapolis, IN). Handee minispin columns were from Pierce, and the microcon centrifugal filter devices and C18 zip tips were from Millipore (Bedford, MA). Reversed-phase packing, Oligo R3, and POROS 50 R2 were purchased from Applied Biosystems. All other chemicals, unless indicated otherwise, were obtained from Sigma (St. Louis, MO).

cDNA constructs

The Flag-tagged murine MOR1 cDNA was described previously (17). MOR mutants were generated by oligonucleotide-directed mutagenesis (Quikchange system, Strategene), and verified by dideoxy sequencing (Elim Biochemical). The rat β-arrestin-2-GFP construct was described previously (72).

Cell culture, transfection and isotope labeling

Stably transfected HEK293 cells from a single clone expressing amino-terminally Flag-tagged murine µ-opioid receptor (MOR1), as described previously (17), were propagated in Dulbecco’s modified Eagle’s medium (DME) deficient in L-Arginine containing 10% fetal bovine serum, penicillin (100 U/ml), streptomycin (100 ug/ml), and supplemented with L-Arginine, ¹³C-labeled Arginine, or ¹³C¹⁵N-labeled Arginine, to a final concentration of 0.398 mmol/L. Cells were maintained in specific isotope conditions for a minimum of 6 doublings with frequent medium changes. Uniform isotopic labeling of receptors was confirmed at >99% by MALDI-TOF-MS of Arg-containing peptides. The stably transfected clone was selected on the basis of fluorescence intensity. This clone exhibited robust internalization to DAMGO, and normal growth rate and morphology typically seen in Flag-MOR transfected HEK293 cells.

For cell biological analysis of mutant receptors, HEK293 cells were transfected with the indicated mutations of Flag-MOR cloned in pcDNA3, using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. For assays of transiently transfected cells, experiments were carried out 48 hours later. For experiments with stably transfected cells, clones were selected in the presence of 50 µg/ml G418 conditions (Geneticin, Invitrogen). Clones used on subsequent studies were chosen based on similar levels of receptor surface levels, as assessed by fluorescence flow cytometry.

Agonist stimulation and cell harvest for mass spectrometry samples

Cells were grown to ~70% confluence in 15cm round cell culture dishes containing 20ml of appropriate culture medium. For non-SILAC experiments, 6 dishes of HEK293 cells were used for each agonist condition (10 µM morphine or 10 µM DAMGO). For SILAC experiments, two dishes were used for each condition. Cells were grown in arginine-depleted media which was supplemented with either regular arginine, ¹³C arginine, or ¹³C¹⁵N arginine. Regular arginine supplemented cells acted as the control. ¹³C arginine supplemented cells were treated with morphine. ¹³C¹⁵N arginine supplemented cells were treated with DAMGO. Dishes were incubated with agonist for 20 minutes at 37°C, the medium was removed and replaced with ice-cold PBS with 1 mM EDTA and incubated on ice for 10 min. Intact cells were harvested by gentle aspiration while on ice. The cell suspension of each condition was pelleted for 2 minutes at 500×g at 4°C, and aspirated pellets were quickly frozen at −80°C. To allow for efficient incorporation of isotopic arginine, cells were allowed to double 5–6 times prior to agonist application. Subsequent analysis of arginine levels showed that approximately 99 percent of arginines within a peptide contained the heavy isotope.

Receptor purification

The protocol for Flag-MOR purification was adapted from procedure described previously for isolation of Flag-tagged adrenergic receptors (71). For non-isotopically-labeled experiments, each cell pellet from a specific agonist condition was purified separately. For isotopically labeled experiments, cell pellets from three different isotope agonist conditions to be compared were solubilized together in buffer A (20 mM Tris pH 7.4, 100 mM NaCl) and 1% n-dodecylmaltoside containing protease inhibitors (Complete cocktail, Roche), 2.5 mg/ml leupeptin) and phosphatase inhibitors (50 mM NaF, 0.1 mM sodium orthovanadate, 80 mM glycerol-2-phosphate) and 10 µM Naloxone to stabilize the solubilized receptor. Pellets were processed in a Dounce homogenizer and extracts were supplemented with 1 mM CaCl₂ and centrifuged at 18000 rpm in a SS34 centrifuge rotor for 30 minutes. Clarified supernatant was applied to M1-Flag affinity resin (Sigma) equilibrated in low salt washing buffer (buffer A + 0.1% n-dodecylmaltoside) supplemented with 1 mM CaCl₂. The volume of resin used was 0.05 ml/ml extract. The column was washed three times with 3 column volumes of low salt washing buffer, and then washed with 5 sets of alternating high salt buffer (20 mM Tris pH 7.4, 500 mM NaCl, 0.1% n-dodecylmaltoside, 1 mM CaCl₂) and low salt washing buffer. After an additional three column volume washes with low salt washing buffer, the receptor was eluted with buffer A supplemented with 1mM EDTA and 200 µl/ml Flag peptide (Sigma). Receptor-containing fractions were identified by Bradford assay, combined with SDS-PAGE sample buffer, and electrophoresed on 4–20% polyacrylamide gels. SYPRO Ruby staining reagent (Molecular Probes) was applied per the manufacturer’s instructions, and imaged using a Typhoon 9400 (Amersham Biosciences) with excitation at 610 nm to detect the SYPRO signal. Images were processed and staining intensities quantified using ImageQuant software (Amersham Biosciences).

Sample preparation for mass spectrometric analyses of proteolytic digests

To separate Flag peptide (DYKDDDDK) from the MOR, gel filtration was performed using Handee minispin columns filled with Sephadex G-50 fine resin, essentially as described previously (71). Briefly, purified receptor (~10 pmol) was loaded and eluted from the resin using a Tris buffer (20 mM, pH 7.5) containing NaCl (100 mM). Reduction, alkylation (iodoacetamide), and in solution proteolytic digests (with modified-trypsin and Asp-N) of the MOR was accomplished using minor modifications of methods described previously in greater detail (73). All samples were depalmitoylated on the microconcentrator membrane prior to enzymatic digestion by adding 1 M hydroxylamine (pH to ~8 with NaOH) and leaving samples in the dark for ~20 min at room temperature. Reduction of disulfides with tris(carboxyethyl)phosphine (TCEP) pH 7 were carried out for ~20 min at room temperature instead of 37°C to minimize receptor aggregation. IMAC procedures for phosphopeptide enrichment were carried out using the manufacturer's protocol (Phosphopeptide kit, Pierce Biotechnology) except that an extra elution step with 1 M EDTA pH 8 was included. Phosphopeptides were subsequently purified using a step-elution reversed-phase procedure [Oligo R3: POROS R2 (1:2) loaded on top of zip tips] (71, 74). For LC-ESI MS procedures, samples were dried using a SpeedVac concentrator (Thermo Electron) and redissolved in 0.1% formic acid.

MALDI mass spectrometry

Comprehensive analysis of available proteolytic peptides pre and post-IMAC enrichment were analyzed using the MALDI orthogonal time-of-flight instrument prOTOF 2000 (PerkinElmer Life Sciences) as previously described (75). The MALDI matrix used was 4-hydroxy-a-cyanocinnamic acid (4HCCA) mixed with a mixture of trifluoroacetic acid (0.1%), acetonitrile (33%) and water (67%). All MALDI- prOTOF 2000 spectra for preparation of figures used the program M-over-Z data (Genomic Solutions, Ann Arbor, MI).

Nano-LC-ESI-QqTOF mass spectrometry

IMAC-enriched or non-enriched peptide digests were separated using a 75-mm×15-cm reversed-phase C18 column (LC Packings) at a flow rate of 350 nl/min, running a 3–32% acetonitrile gradient in 0.1% formic acid over 1 h on nano-HPLC equipped with an autosampler (Eksigent). The LC eluent was coupled to a Nanospray II source attached to a QSTAR Elite mass spectrometer running Analyst QS 2.0 software (Applied Biosystems). The mass spectrometer was set up to an MS survey scan for 500 milliseconds followed by up to 12 MS/MS spectra at fixed m/z values corresponding to peptides observed from the MOR (200 milliseconds per MS/MS). The Q1 precursor selection resolution was set to “low”, with a full width half maximum of approximately five Da. Collision-induced dissociation energies were calculated using an empirically determined ramp based upon the optimal collision energies for a standard set of peptides. ESI peptides were identified using Protein Prospector version 5.2 (prospector.ucsf.edu). For quantification on the QSTAR, extracted ion chromatograms were generated for selected fragment ions. The MS/MS data was manually submitted to A-score (www.ascore.med.harvard.edu) for assessing the identified sites of phosphorylation for each peptide.

LC-ESI QTRAP mass spectrometry

IMAC-enriched or non-enriched peptides were separated using a 75-mm × 15-cm reversed-phase C18 column (LC Packings) at a flow rate of 350 nl/min, running a 3–32% acetonitrile gradient in 0.1% formic acid over 1 h on nano-HPLC equipped with an autosampler (Eksigent). The LC eluent was coupled to a Nanospray II source attached to a QTRAP 4000 triple quadrupole mass spectrometer running Analyst QS 1.4 software (Applied Biosystems). Selective reaction monitoring (SRM) transitions were based upon informative fragment ions observed in the QTOF experiments. The collision energies were identical to those used on the QTOF. Individual SRM measurements were measured for 20 msec each. For precursor ion isolation, the quadrupole resolution was set to “low”.

Quantification of MS data

To examine agonist effects on MOR phosphorylation, light (i.e. ¹²C and ¹⁴N) and two different heavy MOR cell preparations (representing three different treatment conditions) were mixed in equal amounts prior to protein purification to ensure identical handling. After separation and analysis with LC-MS, we calculated peptide abundance ratios using the MS signal intensity areas for each treatment condition. These signal intensity areas were calculated by first averaging the mass signals at each LC elution time. Next, the total number of signal counts of 2–3 monoisotopic peak tops per peptide was summed per condition. Subsequently, these areas were adjusted by normalizing all MS signal counts with an internal unmodified MOR proteolytic peptide to account for small variations in mixing. The counts from the MS signal of the Asp-N MOR peptide, [164–176], [DRYIACHPVAL], fulfilled this role as the internal standard for quantitation experiments (both Trypsin and Asp-N digests). For determining quantitative ratios when there was overlap between light and heavy or the different heavy (¹³C,¹⁴N vs ¹³C,¹⁵N) isotope distributions, an isotopic correction factor was applied as described previously by Ong et al (76).

Quantification of receptor internalization by fluorescence flow cytometry

Endocytosis of Flag-tagged MOR either stably or transiently transfected HEK293 cells in response to DAMGO or morphine was assessed by fluorescence flow cytometry with minor modifications to a previously described method (77). Briefly, cells were incubated in the absence or presence of the indicated agonist for 20 minutes at 37°C, then quickly chilled on ice. Cells were dissociated and surface-accessible receptors were labeled with Alexa 488-conjugated M1 antibody, and surface receptor immunoreactivity was measured using a FACScan instrument (BD Biosciences). 10,000 cells were collected for each sample. Triplicate samples were analyzed for each condition in each experiment.

Visualization of receptor trafficking by fluorescence microscopy

Flag-tagged MOR constructs were visualized by an antibody-feeding assay previously described (22). Briefly, HEK293 cells either stably or transiently transfected with the indicated construct(s) (as indicated) were plated on glass coverslips. Surface-accessible receptors were labeled in intact cells by addition to the culture medium of Alexa 488 –conjugated M1 anti-Flag antibody (2 µg/ml) for 30 minutes at 37°C, then cells were incubated for an additional 20 minutes in the presence or absence of 10 µM morphine or 10 µM DAMGO as indicated. After a quick wash with PBS, cells were fixed in 4% formaldehyde in PBS for 15 minutes, then washed 4 more times with PBS and mounted onto glass slides with Vectashield (Vector Laboratories, Burlingame, CA). Epifluorescence microscopy was performed with an inverted Nikon Diaphot microscope equipped with a 60X/numerical aperture (NA)1.4 objective, mercury arc lamp illumination, and standard dichroic filter sets (Omega Optical). Images were collected with a CCD camera (Princeton Instruments) interfaced to a PC running Micromanager software (www.micro-manager.org).

TIR-FM imaging of β-arrestin-2 recruitment in HEK293 cells

Total internal reflection fluorescence microscopy (TIR-FM) was carried out using a system described previously (52) based on a Nikon TE-2000 inverted microscope equipped with a 60x / NA 1.49 objective. Evanescent illumination was generating by focusing a fiber-coupled 488 nm argon ion laser (Melles Griot, Rochester, NY) onto the outer back focal plane of the objective using a micrometer-guided illuminator (Nikon). Temperature was controlled at 37°C using a thermoelectric stage (Bioscience Tools, San Diego, CA) and objective warmer (Bioptechs, Butler, PA). Time-lapse sequences were collected in living cells using a deep-cooled electron-multiplying CCD camera (Andor) operated in frame transfer mode, and in the linear range of detection. Image analysis was carried out using ImageJ (imagej.nih.gov). For analysis across multiple cells and experiments, specimens were fixed 7 min after agonist application using 4% formaldehyde freshly dissolved in phosphate-buffered saline (PBS) supplemented with 150 mM sucrose. Samples were washed in PBS after exposure to fixative for 10 min, and TIRF imaging was carried out in PBS.

Statistical analysis

The significance of differences between experimental groups was analyzed using Student's t-test or ANOVA with Bonferroni's multiple comparison test, depending on whether two or more groups were being compared, calculated using Prism software (GraphPad, La Jolla CA), using p < 0.05 as threshold for significance.