Tracking receptor motions at the plasma membrane reveals distinct effects of ligands on CCR5 dynamics depending on its dimerization status

  1. Fanny Momboisse
  2. Giacomo Nardi
  3. Philippe Colin
  4. Melanie Hery
  5. Nelia Cordeiro
  6. Simon Blachier
  7. Olivier Schwartz
  8. Fernando Arenzana-Seisdedos
  9. Nathalie Sauvonnet
  10. Jean-Christophe Olivo-Marin
  11. Bernard Lagane
  12. Thibault Lagache  Is a corresponding author
  13. Anne Brelot  Is a corresponding author
  1. Institut Pasteur, Université Paris Cité, CNRS UMR3569, Virus and Immunity Unit, France
  2. Institut Pasteur, Université Paris Cité, CNRS UMR3691, BioImage Analysis Unit, France
  3. Infinity, Université de Toulouse, CNRS, INSERM, France
  4. Institut Pasteur, Université Paris Cité, Dynamics of Host-Pathogen Interactions Unit, France
  5. Institut Pasteur, Université Paris Cité, INSERM U1108, Viral Pathogenesis Unit, France
  6. Institut Pasteur, Université Paris Cité, Group Intracellular Trafficking and Tissue Homeostasis, France

Abstract

G-protein-coupled receptors (GPCR) are present at the cell surface in different conformational and oligomeric states. However, how these states impact GPCRs biological function and therapeutic targeting remains incompletely known. Here, we investigated this issue in living cells for the CC chemokine receptor 5 (CCR5), a major receptor in inflammation and the principal entry co-receptor for Human Immunodeficiency Viruses type 1 (HIV-1). We used TIRF microscopy and a statistical method to track and classify the motion of different receptor subpopulations. We showed a diversity of ligand-free forms of CCR5 at the cell surface constituted of various oligomeric states and exhibiting transient Brownian and restricted motions. These forms were stabilized differently by distinct ligands. In particular, agonist stimulation restricted the mobility of CCR5 and led to its clustering, a feature depending on β-arrestin, while inverse agonist stimulation exhibited the opposite effect. These results suggest a link between receptor activation and immobilization. Applied to HIV-1 envelope glycoproteins gp120, our quantitative analysis revealed agonist-like properties of gp120s. Distinct gp120s influenced CCR5 dynamics differently, suggesting that they stabilize different CCR5 conformations. Then, using a dimerization-compromized mutant, we showed that dimerization (i) impacts CCR5 precoupling to G proteins, (ii) is a pre-requisite for the immobilization and clustering of receptors upon activation, and (iii) regulates receptor endocytosis, thereby impacting the fate of activated receptors. This study demonstrates that tracking the dynamic behavior of a GPCR is an efficient way to link GPCR conformations to their functions, therefore improving the development of drugs targeting specific receptor conformations.

Editor's evaluation

This manuscript seeks to push the frontiers of live-cell single-molecule imaging by tracking the diffusive movements of CCR5 receptors and CCR5 receptor complexes within the plasma membrane of living cells and how these motional behaviors change with physiological stimuli. The results will be important for researchers working at the interface of cell biology and biophysics on membrane-bound receptors.

https://doi.org/10.7554/eLife.76281.sa0

Introduction

G-protein-coupled receptors (GPCRs), also known as 7TM (seven transmembrane helical) receptors, represent the largest group of cell surface receptors in humans that transduce chemical signals from the extracellular matrix into the cell. They constitute one of the primary drug target classes (Pierce et al., 2002).

GPCRs exist in different subpopulations at the cell surface, in part due to differential post-translational modifications (Patwardhan et al., 2021; Scurci et al., 2021) and arrangements of receptor loops and transmembrane domains (Deupi and Kobilka, 2010). Receptor activation and G protein coupling indeed involves a series of conformational changes from an inactive to an active state (Ahn et al., 2021). Coupling to different G proteins or to other protein transducers (e.g. arrestins), as well as receptor oligomerization expand the diversity of conformational states (Seyedabadi et al., 2019; Sleno and Hébert, 2018). Molecular dynamics along with biophysical and structural studies brought to light this variety of GPCR arrangements and showed how binding of different ligands can stabilize or select different receptor conformations, which can in turn activate different signaling pathways (Ahn et al., 2021). This concept of ‘functional selectivity’ (or ‘biased agonism’) opens the possibility to develop therapies specifically targeting a selected receptor conformation, thereby increasing the effectiveness of drugs and reducing their adverse effects (Seyedabadi et al., 2019).

The nature and proportion of the different forms of GPCRs vary depending on their environment. This is likely to regulate the functional properties of the receptors (Colin et al., 2018; Patwardhan et al., 2021). Few studies, however, confirmed this diversity of receptors in living cells and investigated its regulation in time and space (Calebiro et al., 2012; Gormal et al., 2020; Kasai et al., 2018; Martínez-Muñoz et al., 2018; Sungkaworn et al., 2017; Veya et al., 2015). In this study, we tracked the chemokine receptor CCR5 at the particle level to access its dynamic behavior at the plasma membrane and identify the organization and the functional properties of the various receptor forms.

CCR5 is a class A GPCR expressed on the surface of hematopoietic and non-hematopoietic cells. It is a key player in the trafficking of lymphocytes and monocytes/macrophages and has been implicated in the pathophysiology of multiple diseases, including viral infections and complex disorders with an inflammatory component (Brelot and Chakrabarti, 2018; Flanagan, 2014; Vangelista and Vento, 2017). In addition, the CCL5/CCR5 axis represents a major marker of tumor development (Aldinucci et al., 2020). CCR5 binds several chemokines, including CCL3, CCL4, and CCL5. Binding of chemokines results in conformational change of the receptor, which then activates intracellular signaling pathways and leads to cell migration (Flanagan, 2014). CCR5 also binds the envelope glycoprotein of HIV-1, then acting as the major HIV-1 entry co-receptor (Alkhatib et al., 1996; Brelot and Chakrabarti, 2018). One CCR5 allosteric ligand, maraviroc (MVC), is part of the anti-HIV-1 therapeutic arsenal (Dorr et al., 2005), although emergence of MVC-resistant variants has been identified in some patients (Tilton et al., 2010).

We and others showed the existence of various CCR5 populations present at the cell surface (Abrol et al., 2014; Berro et al., 2011; Colin et al., 2013; Colin et al., 2018; Fox et al., 2015; Jacquemard et al., 2021; Jin et al., 2014; Jin et al., 2018; Scurci et al., 2021). Computational analysis predicts that CCR5 can adopt an ensemble of low-energy conformations, each of which being differentially favored by distinct ligands and receptor mutations (Abrol et al., 2014). CCR5 conformations display distinct antigenic properties, which vary depending on cell types (Colin et al., 2018; Fox et al., 2015). The multiple conformations interact differently with distinct ligands (agonist, antagonist, HIV-1 envelope glycoprotein) and differ in their biological properties, HIV co-receptor functions, and abilities to serve as therapeutic targets (Abrol et al., 2014; Colin et al., 2013; Colin et al., 2018; Jacquemard et al., 2021; Jin et al., 2014; Jin et al., 2018; Scurci et al., 2021). In particular, coupling to G proteins distinguishes CCR5 populations that are differently engaged by chemokines and HIV-1 envelope. This explains why HIV-1 escapes inhibition by chemokines (Colin et al., 2013). In this context, the improved capacity of chemokine analogs to inhibit HIV infection, as compared to native chemokines, is related to their ability to target a large amount of CCR5 conformations (Jin et al., 2014).

Like other receptors of this class, CCR5 forms homo- and heterodimers with other receptors, which contribute to the diversity of conformational states (Jin et al., 2018; Sohy et al., 2009). We identified three homodimeric organizations of CCR5 involving residues of transmembrane domain 5 (TM5) (Jin et al., 2018). Two dimeric states corresponded to unliganded receptors, whereas binding of the inverse agonist MVC stabilized a third state (Jin et al., 2018). CCR5 dimerization occurs in the endoplasmic reticulum, thereby regulating the receptor targeting to the cell surface (Jin et al., 2018). CCR5 dimerization also modulates ligand binding and HIV-1 entry into cells (Colin et al., 2018). MVC stabilizes CCR5 homodimerization, illustrating that CCR5 dimerization can be modulated by ligands (Jin et al., 2018), a feature shared with other chemokine receptors (Işbilir et al., 2020). Allosteric interaction within CCR2/CCR5 heterodimers is reported as well as cross-inhibition by specific antagonists (Sohy et al., 2009). This suggests that dimerization impacts therapeutic targeting.

To characterize the diversity of CCR5 subpopulations at the cell surface and to investigate the impact of CCR5 dynamics on its function, we tracked CCR5 fluorescent particles by total internal reflection fluorescence (TIRF) microscopy (Calebiro et al., 2012) and quantitatively classify their motion over time using a statistical method. We described CCR5 mobility patterns both at the basal state and after ligand binding (using two agonists, the inverse agonist MVC, and HIV-1 envelope glycoproteins) and under conditions that modulate CCR5 /G protein coupling, β-arrestin binding, and dimerization. This study provides novel insights into the organization of a GPCR at the cell surface and the mechanisms regulating its signaling and fate after activation.

Results

Statistical classification of receptor trajectories at the cell membrane

We studied CCR5 dynamics in two different models: eGFP-CCR5 and FLAG-SNAP-tagged-CCR5 (FLAG-ST-CCR5) expressing cells, in which we tracked either eGFP or receptor-bound fluorescent anti-FLAG antibodies. We used HEK 293 cell lines stably expressing a low density of eGFP-CCR5 or FLAG-ST-CCR5 at the cell surface (<0.5 particles/μm2), which is critical for single particle tracking on the surface of living cells (Calebiro et al., 2012). We chose HEK 293 cells because they do not express CCR5. Fusion of proteins to the N-terminus of CCR5 does not alter cell surface expression of the receptor or its intracellular trafficking (Boncompain et al., 2019; Jin et al., 2018).

To study the dynamics of CCR5 as a single particle at the plasma membrane of living cells, we used TIRF microscopy, which restricts the observation to the first 200 nm from the coverslip. The acquisitions were carried out at 37 °C. From the movies obtained, we tracked the motion of the particles over time using the Spot tracking plugin of the ICY software (Chenouard et al., 2013; de Chaumont et al., 2012; Figure 1A–C, Videos 14, see Materials and methods).

Figure 1 with 1 supplement see all
Single particle detection of eGFP-CCR5 using TIRF microscopy and analysis with the statistical method.

(A) Distribution of eGFP-CCR5 stably expressed in HEK 293 cells. Imaging was acquired at 30 Hz. The region of interest defined by the green line is used for A-C and F. Analysis of movies was performed using the ICY software and (B) the Spot detection and (C) the Spot tracking plugins. Scale bar 2 μm. (D) Single receptor tracks were partitioned into tracklets of five images each. (E) Analysis of tracks with the statistical method: tracklets were classified into confined, Brownian, and directed motion. (F) Results obtained from Matlab. (G) Pooled tracklets classification provided a global estimate of receptor dynamics and the number of motion changes along the track (transition rates). (Restricted motions: immobile and confined motions).

Video 1
TIRF movie of a cell stably expressing eGFP-CCR5-WT acquired at 30 Hz.

The region of interest was defined by the green line.

Video 2
TIRF movie of the same cell as in Video 1 analyzed using the Icy software.

Red circles correspond to the detection of bright spots using the Spot detection plugin.

Video 3
TIRF movie of the same cell as in Videos 1 and 2 analyzed using the Icy software and the Spot tracking plugin.

Colored lines correspond to the tracked spots.

Video 4
TIRF movie of a cell stably expressing FLAG-ST-CCR5-WT and stained with M2-Cy3.

Movie was acquired at 10 Hz.

The method generally used to evaluate the dynamics of a particle is based on Mean Square Displacement (MSD) analysis (Qian et al., 1991). However, MSD is a global analysis of particle trajectory that does not handle possible changes in particle motion. In particular, it indicates whether the observed motion is standard Brownian motion and computes the related diffusion coefficient of the trajectory, but it cannot characterize more complex stochastic motions as the frequency of motion changes. In addition, the MSD analysis does not provide a statistical significance of classified motion. More robust analysis using the Bayesian probabilistic framework have been proposed to classify single particle trajectories (Karslake et al., 2021; Monnier et al., 2015; Türkcan and Masson, 2013).

However, Bayesian inference is often associated with a high computational load and is not very robust for short trajectories. Therefore, to robustly characterize the complex stochastic motions of single receptors at the cell membrane, we chose to implement a statistical hypothesis testing method introduced in Briane et al., 2018. To mitigate the risk of tracking errors over long trajectories, and to detect potential motion changes between tracklets within each single particle trajectory, we partitioned single spot trajectories into small tracklets (with N=5 consecutive detections each; Figure 1D and Figure 1—figure supplement 1A). We first evaluated immobile objects and then used a robust statistical method to classify tracklet motion (see Materials and methods and Figure 1E–F). Briefly, for each tracklet X, we computed the statistics SX,N introduced in Briane et al., 2018 that evaluate the ratio between the maximal distance reached by the tracklet particle from the initial point and the motion standard deviation. We then used the statistics SX,N to classify each tracklet into one of the three following motion categories: confined, Brownian, or directed stochastic motion. For this, we computed S(X,N) for each tracklet and compared it to the quantiles (qα,q1-α), which are statistical reference values of Brownian motion at level α and 1-α . Quantiles of S(X,N) only depend on N and α (Briane et al., 2018), and can be evaluated independently of the characteristics of experimental trajectories. Finally, tracklets X were classified according to the associated stochastic motion: confined (if S(X,N)<q(α)), Brownian (if q(α)S(X,N)<q(1α)), and directed motion (if q(1α)S(X,N) (Figure 1—figure supplement 1)). Finally, to evaluate the robustness of tracklet classification to image noise and receptors’ density, we generated synthetic time-lapse sequences (Materials and methods) and measured the classification accuracy for different signal-to-noise ratio (from SNR = 2 to SNR = 10) and receptors’ spots density = 0.039, 0.16 and 0.63 spots/μm2 , the measured density being <0.5 spots/μm2 in most experiments. Our simulations showed that classification accuracy was maintained for SNR >6 (Figure 1—figure supplement 1B), the experimental SNR being ~10, and that classification was robust to spots’ density (Figure 1—figure supplement 1C).

After having implemented this statistical classification in the ICY software (processor Dynamics Classifier in the plugin Track Manager), we characterized the dynamics of CCR5 particles at the cell membrane.

CCR5 particles have different motions at the plasma membrane

We investigated CCR5 mobility in the basal state using the statistical method described above (Figure 1). The result of the classification of all the pooled tracklets provided a global estimate of the receptor dynamics, while the number of motion changes along the same trajectory gave us an estimate of the overall stability of the motion (Figure 1G).

In the basal state, the eGFP-CCR5 particles distributed homogeneously over the entire membrane surface (Figure 1A, Videos 13). However, the motions of eGFP-CCR5 particles were heterogeneous (Figure 2A). Eighty percent of the pooled CCR5 tracklets were mobile with Brownian motion, while 20% were classified as restricted motion (i.e. immobile and confined; Figure 2A). We observed almost no directed trajectories (<0.5 %). Around 50% of particles (52%) exhibited Brownian motion over the entire length of the path (Figure 2B). The other half fluctuated between Brownian and restricted motion (Figure 2B). This high degree of fluctuation between motions within one trajectory suggested the existence of transient conformations of CCR5 at the plasma membrane. Similarly, the motions of FLAG-ST-CCR5 particles were heterogeneous with high degree of fluctuation between motions (Figure 2—figure supplement 1). Note that compared to eGFP-CCR5, FLAG-ST-CCR5 exhibited a higher percentage of tracklet in restricted motion (50%), which we attributed to antibody binding (Harms et al., 2012).

Figure 2 with 1 supplement see all
In the basal state, eGFP-CCR5 exhibits different motions at the plasma membrane.

(A) Distribution of tracklets motion: restricted, Brownian, or directed (mean ± SEM, n=28,305 tracks from 19 cells, 3 independent experiments). (B) Distribution of tracklets motion changes along tracks (mean ± SEM, n=48,237 tracks from 45 cells, 7 experiments).

Together, these analyses revealed heterogeneity of CCR5 motion at the basal state consistent with the diversity of CCR5 forms described previously by other methods (Abrol et al., 2014; Colin et al., 2013; Fox et al., 2015; Jin et al., 2018; Scurci et al., 2021).

Multiple ligands impact CCR5 mobility differently

Since ligands modulate the conformation of CCR5 (Colin et al., 2018; Jacquemard et al., 2021; Jin et al., 2018), we investigated the impact of ligand binding on its spatiotemporal dynamic properties. We evaluated the effect of saturating concentration of ligands (two agonists with different efficacies and the inverse agonist MVC, i.e. a ligand with a negative efficacy) on CCR5 trajectories at the plasma membrane over time. We first incubated eGFP-CCR5-expressing cells in the presence of the native CCR5 chemokine CCL4 at a saturating concentration (>100 nM, kd = 0.4 nM; Colin et al., 2013) for the indicated time. The mobility of the receptor was then assessed immediately after addition of the ligand in a window of 1–12 min (Figure 3A). CCL4 triggered no significant change in CCR5 mobility after 10 min of stimulation (Figure 3B). However, a longer time of CCL4 stimulation (>12 min) increased the percentage of restricted CCR5 tracklets, indicating localized immobility of a small fraction of receptors (Figure 3—figure supplement 1). We also noted the formation of large and immobile spots after 12 min of stimulation (Video 5).

Figure 3 with 2 supplements see all
Different ligands, agonists and inverse agonist, impact eGFP-CCR5 mobility differently.

eGFP-CCR5-WT expressing cells were treated or not with a saturating concentration of agonists (CCL4, 200 nM or PSC-RANTES, 20 nM) or inverse agonist (maraviroc, 10 μM) and single particle tracking analysis was performed. (A) Percentage of restricted tracklets after treatment over time (n=tracks from 10, 4, and 3 cells for PSC-RANTES, CCL4, and MVC conditions respectively, at least three independent experiments). (B) Distribution of tracklets motion after 10 min of treatment (mean ± SEM, n=40,564, 15,421, 11,213, 9828 tracks for each condition from 35, 12, 12, and 9 cells, respectively, at least three independent experiments). Unpaired t test on restricted motions only: ns, nonsignificant; **p≤0.01; ***p≤0.001. (C) Distribution of tracklets motion changes along tracks after 10 min of treatment (mean ± SEM, n=48,237, 8954, 16,668, 9828 tracks from 45, 9, 17, and 9 cells for each condition respectively, at least three experiments). Unpaired t test on all restricted motions only: ns, nonsignificant; ****p≤0.0001. (D) (Left) Single particle detection of eGFP-CCR5-WT after 3 min of stimulation with PSC-RANTES (20 nM) from frame 1 of live-imaging movie (one representative image). (Right) Mean of the sum of fluorescence intensity under large immobile spots and small mobile spots after 3–10 min of stimulation (mean ± SEM, n=at least 40 spots from 12 cells, three experiments). (E) Percentage of restricted tracklets after successive stimulation with maraviroc (10 μM, 5 min) and PSC-RANTES (20 nM, 5–12 min; one representative experiment). (F) Distribution of tracklets motions after successive stimulation with maraviroc (10 μM, during 5 min) and PSC-RANTES (20 nM, during 6 min) (mean ± SEM, n=14,467, 3601, 2075 tracks from 14, 2, and 2 cells respectively, one experiment).

Video 5
TIRF movie acquired at 30 Hz of a cell stably expressing eGFP-CCR5-WT and treated by CCL4 (100 nM) for 14 min.

We compared the effect of CCL4 with that of an agonist targeting a greater proportion of receptor conformations and displaying a greater agonist efficacy, PSC-RANTES (Escola et al., 2010; Jin et al., 2014). We incubated the cells in the presence of a saturating concentration of PSC-RANTES (20 nM, Ki = 1.9 nM; Colin et al., 2013) and evaluated the motion of the receptors under the same conditions. PSC-RANTES triggered a progressive increase in the number of tracklets classified as restricted motion over time (Figure 3A). Ten minutes after stimulation with PSC-RANTES, about 50% of eGFP-CCR5 tracklets were in a restricted state (46 %) against 17% under basal conditions (Figure 3B). Consequently, the fraction of all Brownian trajectories decreased, while the fraction of fluctuated and all restricted trajectories increased (Figure 3C). Simultaneously, we observed the formation of large immobile spots (5–10 per cell) in PSC-RANTES-treated cells (Figure 3D, left). These large spots had a long lifespan (50–100 frames) (Video 6). The quantification of the fluorescence intensity of the spots from the frame 1 of live-imaging movies showed that the large spots had, on average, intensity four times higher than the other spots, indicating a clustering of at least four receptors per large spot (Figure 3D, right). These results revealed a change in CCR5 mobility upon activation toward receptor immobilization and clustering, supporting receptors trapping in nanodomains.

Video 6
TIRF movie acquired at 30 Hz of cells stably expressing eGFP-CCR5-WT and treated by PSC-RANTES (20 nM) for 3 min.

Unlike agonists, the inverse agonist MVC (10 µM, Kd = 1 nM) (Garcia-Perez et al., 2011) did not restrict receptor mobility (Figure 3A, B and C). On the contrary, the fraction of restricted eGFP-CCR5 tracklets at the surface of MVC-treated cells showed a slight decrease compared to untreated cells (Figure 3B). We verified the specificity of PSC-RANTES-induced CCR5 immobility by treating cells with MVC before PSC-RANTES stimulation. MVC treatment impaired PSC-RANTES-induced receptor immobilization (Figure 3E–F), indicating that CCR5 immobilization depended on PSC-RANTES binding to CCR5. We observed the same effect of ligand binding (using CCL4, PSC-RANTES, and MVC) on FLAG-ST-CCR5 mobility, supporting that our findings were independent of the model used (Figure 3—figure supplement 2).

These results showed that distinct ligands differently stabilize CCR5 in living cells, in accordance with our previous results (Colin et al., 2013; Colin et al., 2018; Jin et al., 2014; Jin et al., 2018). Interestingly, the amount of receptors immobilized correlates with the efficacy of ligands (PSC-RANTES >CCL4>MVC), suggesting a link between receptor activation and immobilization.

Gi coupling and β-arrestin association influence CCR5 motion differently under basal state and stimulated conditions

To further address the above hypothesis, we sought to determine whether the mobility of CCR5 is influenced by its coupling to Gi protein, which stabilizes the receptor in an activated state. We analyzed the pool of restricted CCR5 tracklets in the presence of pertussis toxin (PTX), which uncouples the receptor from Gi proteins (Figure 4A).

Figure 4 with 1 supplement see all
Gi coupling and β-arrestins association restrict eGFP-CCR5 mobility at basal state or after PSC-RANTES stimulation.

(A) Percentage of restricted tracklets in eGFP-CCR5-WT expressing HEK 293 cells pre-treated or not with 100 ng/ml of PTX for 3 hr (mean ± SEM, n=8614 and 11 377 tracks for each condition, 12 and 15 cells respectively, 3 independent experiments). Unpaired t test: p value 0.0083**. (B) Percentage of restricted tracklets over time of eGFP-CCR5-WT expressed on PSC-RANTES (20 nM) treated cells after incubation or not with PTX (100 ng/ml) (mean ± SD, n=3 independent experiments). (C) Proportion of restricted tracklets in eGFP-CCR5-WT expressing cells transfected with siRNA βarr1/2 (mean ± SD, n=6754 and 8854 tracks for each condition, from 7 and 8 cells, respectively). Unpaired t test: p value 0.46, ns. (D) Percentage of restricted tracklets over time of eGFP-CCR5-WT expressed on PSC-RANTES (20 nM) treated cells after siRNA βarr 1/2 transfection (n=1 representative experiment).

In the basal state, the fraction of restricted eGFP-CCR5 tracklets from cells pre-treated with PTX decreased compared to untreated cells (Figure 4A). Under this condition, PTX also inhibited chemotaxis, a process that depends on CCR5 coupling to Gi proteins (Figure 4—figure supplement 1). These results thus suggested that a small subset of CCR5 is in a Gi-protein-bound form in its basal state, which may contribute to the transient restriction of the motion of CCR5 at the cell surface.

After stimulation, receptor immobilization could be due to the recruitment of receptors in hub areas where the receptor meets the activation machinery and in particular the G protein (Sungkaworn et al., 2017). To evaluate the role of Gi coupling on receptor immobilization after PSC-RANTES stimulation, we analyzed tracks of TIRF movies of PSC-RANTES-stimulated cells pretreated or not with PTX. In this condition, the fraction of restricted tracklets increased over time after stimulation in the same proportion regardless of PTX treatment (Figure 4B). This suggested that Gi coupling was not involved in PSC-RANTES dependent immobilization of CCR5 after several minutes of stimulation. This result is actually consistent with our previous study showing high affinity interaction of PSC-RANTES with Gi protein uncoupled CCR5 (Colin et al., 2013).

After stimulation by PSC-RANTES, CCR5 follows a clathrin-dependent endocytosis pathway, involving β-arrestins, which bridge the receptor to AP2 and clathrin (Delhaye et al., 2007; Jin et al., 2014). We previously showed that silencing β-arrestin 1 and β-arrestin 2 endogeneous expressions with siRNA decreased CCR5 internalization after PSC-RANTES stimulation (Jin et al., 2014). Silencing β-arrestins in eGFP-CCR5 cells with siRNA did not impact eGFP-CCR5 motion in the basal state (Figure 4C) but inhibited PSC-RANTES-induced eGFP-CCR5 immobilization and clustering (Figure 4D). These experiments indicated that β-arrestins contributed to CCR5 immobilization after stimulation.

Together, these results pointed to the existence of a fraction of CCR5 in a transient pre-assembled signaling complex in the basal state, which is consistent with previous studies showing CCR5 constitutive activity (Garcia-Perez et al., 2011; Lagane et al., 2005). They also suggested that the fate of CCR5 several minutes after activation is independent of Gi coupling but dependent on β-arrestin recruitment, in accordance with receptor desensitization and uncoupling after activation (Flanagan, 2014).

Immobilization of CCR5 after stimulation depends on its oligomeric state

We previously showed by energy transfer experiments, molecular modeling, and a functional assay that a point mutation of CCR5 in TM5 (L196K) leads to a receptor, which has a reduced dimerization capacity compared to CCR5-WT (Jin et al., 2018). Functionally, this mutation alters CCR5 cell surface expression due to its intracellular retention in the endoplasmic reticulum (Jin et al., 2018). However, CCR5-L196K folding is not impacted: CCR5-L196K binds chemokines and HIV gp120s with the same affinity as CCR5-WT (Colin et al., 2018; Jin et al., 2018) and triggered ERK1/2 activation upon stimulation (Figure 5—figure supplement 1). To study the role of CCR5 dimerization on its mobility, we generated HEK 293 cells stably expressing eGFP-CCR5-L196K in the same proportion to the clone expressing eGFP-CCR5-WT.

We studied the molecular composition of both eGFP-CCR5-L196K and eGFP-CCR5-WT in these cells by analyzing the fluorescence intensity of eGFP per spot from the frame 1 of live-imaging movies. In a previous study, we calibrated the fluorescence intensity of eGFP while spotted on glass coverslip (Salavessa et al., 2021). We showed that most of eGFP spots bleached in a single step, suggesting that eGFP corresponds to 1 molecule, with an average fluorescence intensity of 300–500 au (Salavessa et al., 2021). In eGFP-CCR5 expressing cells, the fluorescence intensities were distributed in Gaussians, which we classified with the Akaike information criterion (AIC, see Materials and methods) (Akaike, 1974). We observed three types of Gaussians with double or triple mean intensities (300, 600, 900 au), which may correspond to spots comprising 1, 2, or 3 fluorescence entities relative to eGFP on coverslip (Figure 5A). This reflected the existence of a heterogeneous distribution of receptors. In this classification, the WT receptor distributed in 50% low, 40% medium, and 10% high fluorescence intensity forms at the plasma membrane, while eGFP-CCR5-L196K was mostly in a low fluorescence intensity form (75% low, 25% medium) (Figure 5B). These results revealed that eGFP-CCR5-L196K existed more as monomers or small-size oligomers compared to CCR5-WT at the surface of living cells. This is consistent with the role of Leu-196 in CCR5 oligomerization (Jin et al., 2018).

Figure 5 with 2 supplements see all
Dimerization through TM5 alters eGFP-CCR5 mobility.

(A) Distribution of the fluorescence intensity of spots detected at the surface of HEK 293 cells expressing eGFP-CCR5-WT or eGFP-CCR5-L196K. One representative experiment out of 6 (n=943 spots from 6 cells and 1207 spots from 8 cells for each condition); (B) Quantification of the fluorescent populations depending on the mean of the gaussian at the surface of cells treated or not with MVC (10 μM) (mean ± SD, nWT = 5171 spots from 47 cells, 11 experiments; nL196K=3144 spots from 30 cells, 7 experiments; nWT-MVC=3 055 spots from 25 cells, 4 experiments; nL196K-MVC=1 776 spots from 16 cells, 3 experiments). Unpaired t test on monomers: p value **p≤0.005; ****p≤0.0001; ns p≥0.05; (C) Distribution of pooled trackets motion of eGFP-CCR5-WT and eGFP-CCR5-L196K (mean ± SEM, n=11,321 tracks from 10 cells and 10,460 tracks from 12 cells in each condition; 2 independent experiments). Unpaired t test on the restricted tracklets: p value 0.0015**. (D) Percentage of restricted tracklets in eGFP-CCR5-L196K cells pre-treated or not with 100 ng/ml of PTX for 3 hr (mean ± SEM, n=5 cells). Unpaired t test: p value 0.15, ns. (E) Percentage of restricted tracklets over time of PSC-RANTES induced eGFP-CCR5-WT or eGFP-CCR5-L196K expressing cells (mean ± SD of 3 independent experiments). (F) Distribution of tracklets motion after 10 min of PSC-RANTES stimulation (20 nM) (mean ± SEM, n=11 218 tracks from 10 cells and 5 433 tracks from 4 cells for untreated and PSC-RANTES treated cells respectively, 2 independent experiments). Unpaired t test: p value 0.055, ns.

In the presence of MVC, both eGFP-CCR5-WT and eGFP-CCR5-L196K distribution exhibited 50% low, 40% medium, and 10% high fluorescence intensity forms (Figure 5B). The change of eGFP-CCR5-L196K fluorescence intensities distribution in the presence of MVC is consistent with our previous results showing that MVC stabilized CCR5 in a novel oligomeric form, which was not disrupted by the introduction of a lysine in TM5 (Jin et al., 2018).

To investigate the impact of CCR5 dimerization on its mobility, we compared the motion of eGFP-CCR5-L196K to eGFP-CCR5-WT at the cell surface. As for eGFP-CCR5-WT, eGFP-CCR5-L196K tracklets were predominantly classified as Brownian tracklets motion (85% of the tracklet motions are Brownian). However, we observed a decrease in the proportion of restricted tracklets for eGFP-CCR5-L196K compared to eGFP-CCR5-WT (Figure 5C). We observed the same decrease in the proportion of restricted tracklets for FLAG-ST-CCR5-L196K compared to FLAG-ST-CCR5-WT (Figure 5—figure supplement 2). These data suggested that the degree of receptor oligomerization contributed to the stability of CCR5 molecules at the cell surface, as previously proposed (Calebiro et al., 2012).

To test whether eGFP-CCR5-L196K coupling to Gi protein accounts in its restriction as shown for eGFP-CCR5-WT, we pre-treated cells with PTX. Contrary to eGFP-CCR5-WT, PTX treatment did not alter the proportion of the eGFP-CCR5-L196K restricted tracklets pool (Figure 5D), suggesting that most of eGFP-CCR5-L196K were not precoupled to the Gi protein at the basal state or that G protein precoupling induces differential effects on the dynamics of both receptors. Supporting the first hypothesis, previous biochemical and energy transfer experiments on a distinct GPCR showed that there could be a link between dimerization and Gi coupling at basal state (Maurice et al., 2010).

To investigate whether dimerization affected CCR5 mobility after stimulation, we analyzed single-particle movies of eGFP-CCR5-L196K cells after PSC-RANTES treatment (Figure 5E–F). Contrary to eGFP-CCR5-WT massive immobilization and clustering upon PSC-RANTES treatment (Figure 3A–B), eGFP-CCR5-L196K was only slightly immobilized after 10 min of treatment (Figure 5E–F), while large immobile spots were not detected (Video 7). This result indicated that CCR5 immobilization and clustering after stimulation depend on CCR5 dimerization.

Video 7
TIRF movie acquired at 30 Hz of a cell stably expressing eGFP-CCR5-L196K and treated by PSC-RANTES (20 nM) for 2 min.

Because CCR5-WT immobilization involved β-arrestins (Figure 4D), an explanation for the lack of PSC-RANTES induced eGFP-CCR5-L196K immobilization is that eGFP-CCR5-L196K fails to recruit β-arrestins and therefore, is not desensitized and/or internalized after stimulation.

To test this hypothesis, we evaluated PSC-RANTES-induced β−arrestin 2 (βarr2) recruitment at the plasma membrane of cells expressing either FLAG-ST-CCR5-L196K or FLAG-ST-CCR5-WT (Jin et al., 2018). TIRF acquisitions were performed in fixed cells transiently expressing βarr2-GFP previously stained for FLAG detection. CCR5 activation drove rapid recruitment of βarr2-GFP into spots close to the plasma membrane (Figure 6A). The proportion of recruited βarr2-GFP at the plasma membrane was similar for CCR5-WT and CCR5-L196K (Figure 6B), suggesting that βarr2 recruitment is independent of the oligomeric status of the receptor. Note that we observed a slight decrease in the number βarr2-GFP spots that colocalize with fluorescent receptor spots in CCR5-L196K expressing cells compared to CCR5-WT expressing cells (Figure 6—figure supplement 1). We interpreted this as a consequence of the higher density of receptors per spot for CCR5-WT favoring the probability of βarr2 to colocalize with the receptor in our conditions. These results indicated that the lack of immobilization and clustering of activated CCR5-L196K (Figure 5E and F) is not due to a default of βarr2 recruitment.

Figure 6 with 1 supplement see all
Dimerization through TM5 unaffects β-arrestin 2 recruitment to CCR5 but alters its trafficking.

(A, B) TIRF microscopy on FLAG-ST-CCR5-WT and FLAG-ST-CCR5-L196K cells expressing βarr2-GFP. Cells were stained with M2-Cy3 for FLAG detection and treated or not with 3 nM PSC-RANTES for the indicated times. (A) βarr2-GFP spots were detected on TIRF images from untreated cells or cells treated 10 min with PSC-RANTES. Scale bar 2 μm. (B) Quantification of the βarr2-GFP spots detected over time using ICY software and spot detector plugin (mean +/-SEM, n=at least 6 cells), Unpaired t test: p≥0.05, ns. (C) CCR5 internalization. Cell surface expression of FLAG-ST-CCR5-WT or FLAG-ST-CCR5-L196K was monitored by flow cytometry in stable HEK 293 cell clones after stimulation with a saturating concentration of PSC-RANTES (20 nM) for the indicated time. The percentage of total bound anti-FLAG antibody was calculated from the mean fluorescence intensity relative to untreated cells (mean ± SD from two independent experiments).

We next evaluated PSC-RANTES-induced internalization of the dimerization-compromised mutant compared to the WT receptor in feeding experiments using FLAG-ST-CCR5 expressing cells (Delhaye et al., 2007; Jin et al., 2018). A saturating concentration of PSC-RANTES decreased cell surface expression of both receptors, but not in the same proportion (Figure 6C), suggesting that CCR5 dimerization impacted its internalization process. These results supported that dimerization regulated activated receptor mobility and internalization. Note that, while dimerization is a pre-requisite to the immobilization of the receptor, it was not essential for receptor internalization. This suggests that receptor massive immobilization is not an absolute requirement for receptor internalization.

Distinct HIV-1 envelope glycoproteins gp120 differently influenced CCR5 dynamics

Pharmacological studies suggested that distinct CCR5 conformations at the cell surface differentially engaged distinct HIV-1 envelope glycoproteins gp120 (Colin et al., 2018). Since we showed here that CCR5 mobility and ligand engagement are intrinsically linked, we used our mobility classification method to characterize the effect of different HIV-1 gp120s on CCR5 mobility and tested in living cells whether different gp120s engaged different conformational states of CCR5.

We tested the effect of two soluble gp120s, gp #25 and gp #34, described to induce distinct conformational rearrangements in CCR5 (Jacquemard et al., 2021), and to have different binding capacities to the receptor and fusogenic efficacies (Colin et al., 2018). Twenty min of gp120 exposure slightly modulated the mobility of eGFP-CCR5-WT (and FLAG-ST-CCR5-WT), although this trend was not statistically significant (Figure 7A and C) (Figure 7—figure supplement 1). However, and in contrast to what we observed using chemokines as ligands, the HIV-1 gp120s immobilized eGFP-CCR5-L196K, with gp #34 having the highest effect (Figure 7B and C). This suggested (i) that gp120s stabilized CCR5 conformations, which were different from those stabilized by chemokines, and (ii) that different envelopes also stabilized differently CCR5 conformations, in accordance with our previous result (Colin et al., 2013; Colin et al., 2018).

Figure 7 with 1 supplement see all
HIV-1 gp120s binding restricts eGFP-CCR5 mobility.

Soluble gp120s were incubated 30 min at RT in the presence of soluble CD4 (ratio sCD4/gp120>5) to allow their binding to CCR5. Then, gp120-sCD4 complexes were added to live eGFP-CCR5-WT or eGFP-CCR5-L196K expressing cells during at least 20 min before single particle analysis. The proportion of restricted tracklets after gp #25 and gp #34 treatment (100 nM) (in complex with sCD4) on eGFP-CCR5-WT (A, C) or eGFP-CCR5-L196K (B, C) expressing cells was represented (n=3 independent experiments). Unpaired t test: **p≤0.005; ***p≤0.0001; ns p≥0.05.

Discussion

In this study, we developed a statistical method to classify the motion of fluorescent particles at the cell surface. We applied this method to track eGFP-CCR5 or anti-FLAG Cy3 bound CCR5 under different stimuli and different conformations. We obtained the same results with the two models supporting that our findings are independent of the model used. We showed that the receptor fluctuates between Brownian and restricted motions at the cell surface, depending on (1) precoupling to Gi proteins at the basal state; (2) the type of ligand bound to the receptor, and in particular its efficacy on receptor activation and interaction with β-arrestins; and (3) receptor dimerization. Indeed, CCR5 mobility restriction following agonist stimulation were dependent on β-arrestins recruitment and receptor dimerization, but were independent of receptor interaction with Gi proteins. This study demonstrated that coupling receptor motion tracking to a statistical classification of trajectories is a powerful approach to characterize the dynamic behaviors of functionally different receptor populations at the plasma membrane.

Diversity of ligand-free forms of CCR5 at the cell surface

Quantitative analysis of the motion of CCR5 particles and their composition within the fluorescent spots present at the cell membrane of HEK 293 cells revealed in the basal state (i) two classes of receptor trajectories, Brownian and restricted (Figure 2) and (ii) different oligomeric states (Figure 5) with low (50 %), medium (40 %), and high fluorescence intensity (10 %). These features shared with other GPCRs (Gormal et al., 2020; Martínez-Muñoz et al., 2018; Sungkaworn et al., 2017; Tabor et al., 2016; Veya et al., 2015), established the existence of multiple CCR5 forms at the cell membrane.

In addition, our statistical method highlighted a fluctuation between Brownian and restricted states during the same trajectory, suggesting the existence of transient populations of receptors (Figure 2B). The change in mobility between periods of confinement separated by free diffusion could be attributed to the molecular organization of the receptor oscillating between different oligomeric forms at the cell surface (monomers, dimers, oligomers), as proposed for CCR5 (Jin et al., 2018) or other receptors (Möller et al., 2020; Kasai et al., 2018; Martínez-Muñoz et al., 2018; Tabor et al., 2016). In agreement with this, we observed differences in mobility between high and low order oligomeric forms of CCR5 (Figure 5C). Change in mobility could also be linked to a transient association of the receptor with the cytoskeleton regardless of its oligomeric status (Calebiro and Koszegi, 2019) and/or to transient coupling to G proteins, leading to a transient immobility of the receptor in the basal state. This latter hypothesis is supported by our data in the presence of PTX (Figure 4A) or in the presence of the inverse agonist MVC (Figure 3A and B), which both uncouple the receptor from G proteins and decreased the proportion of immobile receptors. These data are consistent with dual-color TIRF-M analysis of adrenergic receptor and G protein, showing that an active receptor-G protein complex is formed in a confined region of the plasma membrane at the basal state and lasts around 1 s (Sungkaworn et al., 2017). However, they contrast with a study on mGluR3 showing higher mobility of the receptor when complexed with G protein (Yanagawa et al., 2018). This suggested that dynamics of distinct GPCRs can be differently impacted by coupling to G proteins. Regarding β-arrestin association, we showed using siRNA that CCR5 was not precoupled to β-arrestins in its basal state (Figure 4C). This result suggests that CCR5 conformations, which bind to G proteins are not recognized by β-arrestins. This is consistent with the idea that the conformations of receptors interacting with G proteins and β-arrestins are different (Lagane et al., 2005).

Different ligands recognize/stabilize different sets of CCR5

We showed that CCR5 mobility is influenced differently according to the ligand it binds. Chemokine-induced activation of eGFP-CCR5-WT (or FLAG-ST-CCR5-WT) decreased receptor mobility and leads to clustering (Figure 3B and D), effects not observed with the inverse agonist MVC and abolished by MVC (Figure 3A and B and Figure 3E and F). This result reinforces the link between GPCR mobility and ligand binding proposed for GPCRs of different classes (Gormal et al., 2020; Möller et al., 2020; Veya et al., 2015; Yanagawa et al., 2018).

We also showed that two agonists with different efficacies, and targeting different subsets of receptors (CCL4 and PSC-RANTES) (Escola et al., 2010; Jin et al., 2018), restricted receptor motion in a different proportion (Figure 3). Therefore, characterizing ligands by their impact on receptor motion opens a new way to classify biased ligands.

Applied to viral envelope glycoproteins, our tracking approach revealed that HIV-1 gp120s displayed an agonist-like influence on CCR5 mobility, albeit to different extent according to the nature of the gp120 (Figure 7). This feature contrasts with the cryo-EM structure of the CD4-gp120-CCR5 complex, showing that CCR5 adopts inactive confomation (Shaik et al., 2019). However, it is in line with gp120s-induced CCR5 signaling (Brelot and Chakrabarti, 2018; Flanagan, 2014) and with recent MD simulations showing that gp120 binding reorients characteric microswitches involved in GPCR activation (Jacquemard et al., 2021). The fact that the fraction of immobilized receptors varied between gp120s could reflect that they do not bind to/stabilize the same CCR5 conformations, as previously shown (Colin et al., 2018; Jacquemard et al., 2021), and suggests that these gp120s behave themselves as biased agonists. These features of gp120s will help understand the determinants of HIV-1 tropism.

Receptor motion tracking analysis revealed that dimerization regulates the fate of activated CCR5

Our results suggest that receptor dimerization may regulate precoupling of CCR5 to Gi proteins. Indeed, the mobility of the dimerization-compromized mutant eGFP-CCR5-L196K was not affected by PTX treatment (Figure 5D), in contrast to the WT receptor (Figure 4A). This suggests that most eGFP-CCR5-L196K receptors that reside preferentially as monomers are not coupled to Gi proteins in the basal state, in agreement with previous conclusion on CXCR4 (Möller et al., 2020). Alternatively, but not exclusively, CCR5-L196K dimers might also be impaired in their ability to be precoupled to Gi proteins, contrary to WT receptor dimers.

Our analysis suggests that dimerization is a pre-requisite to receptor immobilization and clustering upon activation by chemokine agonists. Indeed, unlike eGFP-CCR5-WT, eGFP-CCR5-L196K receptors are only marginally immobilized in the presence of PSC-RANTES (Figure 5E). This result is not due to impaired binding of the chemokine, because we controlled that PSC-RANTES induced efficient ERK1/2 activation (Figure 5—figure supplement 1) and endocytosis of the mutant receptor (Figure 6C). Receptor immobility and clustering were independent of Gi protein coupling, as exemplified by unaffected CCR5 mobility after 10 min of agonist stimulation in PTX pre-treated cells (Figure 4B), but most likely related to uncoupled and desensitized form of CCR5 that accumulate in CCS (clathrin-coated structures), as proposed (Grove et al., 2014; Yanagawa et al., 2018). This hypothesis was strengthened with the essential role of β-arrestins in activated receptor immobility and clustering (Figure 4D; Markova et al., 2021) and with studies showing that β-arrestins recruitment depends on the efficiency of ligand to trigger CCR5 internalization (Jin et al., 2018; Tarancón Díez et al., 2014; Truan et al., 2013). We cannot rule out that activated receptor clustering may in addition correspond to an accumulation of receptor in early endosome for a second phase of activation (Irannejad et al., 2013).

In line with this, we showed that dimerization regulates endocytosis (Figure 6C). The lack of immobilization of the dimerization-compromised mutant leads to a suboptimal internalization of the receptor. This is not due to a default in βarr2 recruitment since CCR5-WT and CCR5-L196K similarly recruited βarr2-GFP to the plasma membrane (Figure 6A and B). Effective interaction of βarr2 with CCR5-L196K, which is mostly monomeric in the basal state, is consistent with structural studies showing GPCR-arrestin complexes in a 1:1 arrangement (Kang et al., 2015). We propose a model in which receptor oligomerization might be an essential requirement for β-arrestins to trigger receptor clustering and immobilization. A concerted self-association of arrestins may favor this process (Kim et al., 2011). Indeed, PSC-RANTES induces strong β-arrestins clustering (Tarancón Díez et al., 2014; Truan et al., 2013). We speculate that the co-clustering of β−arrestins with receptors may serve as a platform helping to concentrate cargo for optimal and productive internalization. Note that, while dimerization is a pre-requisite for receptor immobilization (Figure 5), it is not essential for receptor internalization (Figure 6C).

Differential effects of gp120 on immobilization of CCR5-WT and CCR5-L196K (Figure 7), compared to chemokines (Figure 5), could also be explained by differences in β-arrestins ability to cluster dimers, linked to differences in the stabilized conformations of receptors.

Finally, our study suggested that CCR5 can be activated whether monomeric or dimeric. We showed that eGFP-CCR5-L196K, while mostly monomeric in its basal state (Figure 5B), is able to activate ERK1/2 (Figure 5—figure supplement 1) and is still internalized after stimulation (Jin et al., 2018; Figure 6C). This is consistent with studies reporting that GPCR monomers can be active enough on their own to be functional (Whorton et al., 2007).

In summary, our receptor motion tracking analysis established that a diversity of CCR5 forms exists at the surface of living cells and that distinct ligands stabilize different receptors. This approach also revealed that receptor dimerization is involved in Gi protein-coupling in the basal state, and in the ability of β arrestin 2 to cluster receptors, therefore impacting the mobility of activated receptors. These findings, point out that receptor conformation regulates GPCRs signaling and fate after activation. In addition, our work suggested that the different receptor conformations likely engaged different ways of regulation, expanding GPCRs functions.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Cell line (Homo sapiens)HEK293 cellsATCCCRL-1573; RRID: CVCL_0045Human embryonic kidney (female)
Cell line (Homo sapiens)A3.01-R5Colin et al., 2013CEM T cell line derivated cells
Antibodyα-GFP
(mouse monoclonal)
Roche11814460001Flow cytometry dilution
(1: 100)
Antibodyα-CCR5
2D7
(mouse monoclonal)
BD-Biosciences555,991Flow cytometry dilution
(1: 500)
AntibodyFLAG tag M2
(mouse monoclonal)
SigmaCat# F3165Flow cytometry dilution
(1: 750)
AntibodyFLAG tag M2-Cy3
(mouse monoclonal)
SigmaCat# A9594TIRF microscopy dilution
(1: 1000)
AntibodyPhospho ERK1/2 (mouse monoclonal)Cell signalingCat# 9,106Western blot dilution
(1:2500)
AntibodyERK2 (Rabbit polyclonal)Santa-Cruz
Biotech
Cat# sc-154Western blot dilution
(1:750)
AntibodyGoat anti-mouse HRP (rat monoclonal)BD-BiosciencesCat# 559,751Western blot dilution
(1:120000)
AntibodyGoat anti-rabbit HRP (goat polyclonal)JacksonCat# 111-035-144Western blot dilution
(1:3500)
AntibodyGoat anti-mouse phycoerythrin (PE) (goat polyclonal)BD-BiosciencesCat# 550,589Flow cytometry dilution
(1:100)
Recombinant DNA reagentpmCherry-
(plasmid)
otherProvided by F. Perez (Institut Curie).
Recombinant DNA reagentpeGFP-CCR5
(plasmid)
otherProvided by F. Perez (Institut Curie).
Recombinant DNA reagentpeGFP-CCR5-L196K
(plasmid)
This paperContains a point mutation in CCR5 at position L196.
Recombinant DNA reagentpFLAG-SNAP-CCR5-WT
(plasmid)
Jin et al., 2018Provided by Cisbio
Recombinant DNA reagentpFLAG-SNAP-CCR5-L196K
(plasmid)
Jin et al., 2018Introduction of a lysine in position L196
Recombinant DNA reagentpβarr2-GFPStorez et al., 2005Provided by S. Marullo (Institut Cochin)
SiRNA reagentβarr1/2
(siRNA)
DharmaconSee Materials and methods for sequence
SiRNA reagentScrambled (siRNA)DharmaconSee Materials and methods for sequence
Soluble proteinHIV-1 gp120
#25, #34
Colin et al., 2018Gp120 from PBMCs of patients in early or late HIV-1 infection stage. See details in ‘cell culture and reagents’ section of 'Materials and methods'
Soluble proteinHuman sCD4Colin et al., 2018See details in ‘cell culture and reagents’ section of 'Materials and methods'
Chemical compound, drugMaravirocNIHCat# ARP-11580CCR5 inverse agonist
Chemical compound, chemokineCCL4This paperProvided by F. Baleux (Institut Pasteur)
Chemical compound, drugPSC-RANTESNIBSCCat# ARP973CCR5 agonist
Chemical compound, chemokineSDF-1PeprotecCat# 300–28 ACXCR4 agonist
Chemical compound, drugPertussis ToxinSigmaCat#179 A100 ng/ml
Software, algorithmPrismGraphPad8.1.1
Software, algorithmICYOpen accessVersion 2.4.0.0https://icy.bioimageanalysis.org/
Software, algorithmMATLABMathWorksR2017a

Cell culture and reagents

Request a detailed protocol

The HEK 293 cells stably expressing FLAG-SNAP tagged- CCR5-WT (FLAG-ST-CCR5-WT) and FLAG-SNAP tagged-L196K (FLAG-ST-CCR5-L196K) and the A3.01 human T cell line stably expressing CCR5 (A3.01-R5) were previously described (Colin et al., 2013; Jin et al., 2018). These cell lines were maintained in Dubelcco’s modified Eagle medium (DMEM) (Thermo Fisher Scientific) or RPMI 1640 medium supplemented with 10% Fetal Bovine Serum (FBS, GE Healthcare) and 100 µg/ml penicillin/streptomycin (Life technologies).

The CCR5 inverse agonist maraviroc (MVC) was obtained from the National Institutes of Health. The native chemokine CCL4 was chemically synthetized by F. Baleux (Institut Pasteur, Paris, France). The chemokine analog PSC-RANTES (N-α-(n-nonanoyl)-des-Ser(1)-[L-thioprolyl(2), L cyclohexylglycyl(3)] RANTES(4-68)) was obtained through the Center for Aids reagents, National Institute for Biological Standards and Control (NIBSC, UK). The primary antibodies used are the anti-GFP (Roche), the anti-CCR5 2D7 mAb (BD-Biosciences); the anti-FLAG monoclonal antibodies M1 or M2 or M2-Cy3 (Sigma-Aldrich), the phospho-ERK ½ (Cell Signaling) and ERK2 (Santa Cruz). Secondary antibodies used were a phycoerythrin (PE)-conjugated anti-mouse antibody (BD Biosciences), a horseradish peroxidase (HRP)-conjugated anti-mouse antibody (BD Pharmingen) and a horseradish peroxidase (HRP)-conjugated anti-rabbit antibody (Jackson). The toxin from Bordetella pertussis (PTX) used at a 100 ng/ml concentration were from Sigma. The βarr1/2 siRNA (5’-ACCUGCGCCUUCCGCUAUG-3’) and a scrambled siRNA (control, 5’-UGGUUUACAUGUCGACUAA-3’) (Dharmacon) were transfected by RNAimax (Invitrogen) according to the instructions of the manufacturer, as described (Jin et al., 2014). To select siRNA positive cells, cells were co-transfected with a plasmid coding the fluorescent protein mcherry (gift of F. Perez, Institut Curie). The construct encoding for GFP fusion of wild-type β-arrestin 2 (βarr2-GFP) have been described previously (gift of S. Marullo) (Storez et al., 2005). Soluble, monomeric HIV-1 glycoprotein gp120 was produced using a semliki forest virus (SFV) system as described (Benureau et al., 2016; Colin et al., 2018). The sequence coding for gp120 #25 and gp120 #34 were from PBMCs of patients collected early after seroconversion or in the AIDS stage of infection, respectively (Colin et al., 2018). Recombinant soluble CD4 (sCD4), produced in S2 cell lines, was purified on a strep-Tactin column using the One-STrEP-tag fused to the CD4 C-tail as a bait (production and purification of recombinant proteins technological platform, C2RT, Institut Pasteur).

Generation of cell lines

Request a detailed protocol

The eGFP-CCR5 plasmid was a gift of F. Perez (Institut Curie, Paris, France). eGFP-CCR5 was expressed from the CMV promoter. The mutant eGFP-CCR5-L196K (substitution of L196 with a lysine) was generated by site-directed mutagenesis using the QuickChange II Mutageneis kit (Agilent Technologies) according to the manufacturer’s instruction. This mutant was verified by sequencing (Eurofins). HEK 293 cells stably expressing eGFP-CCR5-WT and HEK 293 cells stably expressing eGFP-CCR5-L196K were generated by calcium phosphate transfection and cultured for several weeks in 1 mg/ml G418 (Geneticin, Invitrogen). Cell clones were screened and sorted by flow cytometry (Attune NxT flow cytometer, Thermo Fisher) using an anti-GFP monoclonal antibody.

Receptor cell surface expression levels and internalization measured by flow cytometry

Request a detailed protocol

Flow cytometry was used to quantitate the internalization of FLAG-ST-CCR5-WT compared to FLAG-ST-CCR5-L196K stably expressed in HEK 293 cells (Delhaye et al., 2007; Jin et al., 2018). We measured the levels of cell surface CCR5 stained with the anti-FLAG M2 antibody and with an anti-mouse coupled to phycoerythrin (PE) after chemokine treatment or not. Cells were incubated with a saturable amount of M2 for 45 min to label receptors present at the plasma membrane, then incubated in the presence (or not) of 20 nM PSC-RANTES for the indicated time at 37 °C. Cells were chilled to 4 °C and stained with a PE conjugated anti-mouse IgG. Mean values were used to compute the proportion of internalized receptors as indicated by a decrease of immune-reactive surface with PSC-RANTES compared with untreated cells. Cells were analyzed with Attune NxT flow cytometer (Thermo Fisher). At least 5000 cells were analyzed per experiment using Kaluza software. Background was subtracted using the fluorescence intensity obtained on the parental HEK 293 cells.

Chemotaxis

Request a detailed protocol

CCR5 expressing A3.01 cells (A3.01-R5, 1.5 × 105), pre-treated or not with PTX (100 ng/ml) during 3 hr, in prewarmed RPMI-1640 supplemented with 20 mM Hepes and 1% serum, were added to the upper chambers of HTS-Transwell-96 Well Permeable Supports with polycarbonate membrane of 5 μm pore size (Corning). PSC-RANTES (33.7 nM) or SDF-1 (control, 10 nM) was added to the lower chambers. Chemotaxis proceeded for 4 hr at 37 °C in humidified air with 5% CO2. The number of cells migrating across the polycarbonate membrane was assessed by flow cytometry with Attune NxT flow cytometer (Thermo Fisher). Specific migration was calculated by subtracting spontaneous migration from the number of cells that migrated toward the chemokine.

Phospho-ERK1/2 measurements

Request a detailed protocol

FLAG-ST-CCR5 expressing cells (1.5 × 105) were grown in 24-well plates pretreated with poly-D-lysine and rendered quiescent by serum starvation for 16 hr prior to incubation with or without CCL4, as indicated. Plates were placed on ice and the cells were then scraped into lysis buffer composed of 0.5% n-dodecyl-β-D-maltoside (NDM), 0.2% iodoacetamide, protease and phosphatase inhibitors in mTBS. After 30 min, samples were centrifuged and heated for 10 min at 60 °C before resolution of equal amounts of proteins on SDS-PAGE. The proteins were transferred to nitrocellulose membranes, and immunoblotting were carried out using the indicated antibodies. Immunoreactivity was revealed using a secondary antibody coupled to HRP. Band intensities on the same film were quantified by densitometry.

βarrestin 2 recruitment at the plasma membrane

Request a detailed protocol

FLAG-ST-CCR5 expressing cells, transfected with βarr2-GFP, were plated on MatTek plates 72 hr before imaging. Cells were stained with the anti-FLAG M2-Cy3 (5 min) and incubated in the presence or absence of 3 nM PSC-RANTES in DMEM/1%BSA medium for the indicated time. Cells were put on ice and fixed with paraformaldehyde (PFA) 4% at 4 °C for 40 min before three washes in PBS. Experiments were performed using a Elyra 7 microscope (Carl Zeiss Gmbh) equipped with two sCMOS cameras PCO Edge 4.2, and using an alpha Plan Apo 63 x/1.46 oil objective, a 488 nm (500 mW) and a 561 nm (500 mW) laser line, and a quad band filter coupled to BP 495–550 or BP 570–620 filters. All TIRF images analyses were performed using ICY software and the spot detector and the colocalization studio plugins. The number of spot detected per cell was normalized to the size of the cell surface.

Live cell TIRF imaging

Request a detailed protocol

Round 25 mm No. 01 glass coverslips (Fisher Scientific) were pre-cleaned with 70% ethanol followed by acetone, with three consecutive washes in ddH2O. 1.15×105 cells were plated onto pre-cleaned coverslips 72 hr before imaging. Cells were imaged in TIRF medium (25 mM HEPES, 135 mM NaCl, 5 mM KCl, 1.8 mM CaCl2, 0.4 mM MgCl2, 4.5 g/l glucose and 0.5% BSA, pH 7.4). For eGFP tracking, movies were acquired with an LSM 780 Elyra PS.1 TIRF microscope (Zeiss) equipped with an EMCCD Andor Ixon 887 1 K camera, and using an alpha Pin Apo 100 x/1.46 oil objective, a 488 nm (100 mW) HR solid laser line, and a BP 495–575+LP 750 filter to detect eGFP-CCR5. Image acquisition was done at 1 frame / 33 ms (30 Hz) (100–200 frames), with an illumination intensity <0.38 kW/cm2 (tracking) or 0.7 kW/cm2 (fluorescence intensity) at 37 °C. Under these conditions, the intensity of the spots is stable throughout the duration of the acquisition. Approximately 5–10 cells were acquired per condition, per experiment. For FLAG-ST-CCR5 tracking, movies were acquired with a TIRF microscope (IX81F-3, Olympus) equipped with a X 100 numerical aperture 1.45 Plan Apo TIRFM Objective (Olympus) and fully controlled by CellM (Olympus). Images were collected using an IxonEM camera (DU885, Andor). Image acquisition was done at 10 Hz with an illumination intensity of about 0.1 kW/cm2.

All live-imaging movies were analyzed using the open-source software Icy (Institut Pasteur).

Track analysis protocol

Tracking receptors in TIRF imaging with Icy software

Request a detailed protocol

To automatically detect eGFP-CCR5 tracks at the plasma membrane upon time, we used the software Icy (http://icy.bioimageanalysis.org) and the plugin Spot tracking, which reports their xy displacement and intensities, as previously described in Bertot et al., 2018. Spot tracking was set to detect spots with approximately 3 pixels, and a threshold of 135. All other parameters were as default. Tracks were analyzed with the Track manager plugin. All data was exported to Excel for further analysis.

Tracks containing more than 10% of virtual detections and more than three successive virtual detections were excluded from the track classification.

Splitting tracks into tracklets

Request a detailed protocol

We deal with trajectories that have very different lengths and we want to estimate motion variations along the trajectory. Thus, we split all long tracks into several tracklets in order to better classify local motions. According to Section 1, this is done by setting N=5 and considering only the tracks with length larger than 6. Then, the different successive tracklets are defined by using the position between the 5kth and 5(k+1)th frame with k0.

Detecting immobile receptors

Request a detailed protocol

To classify tracklets and identify distinct receptor dynamics, we first identified immobile receptors. In time lapse imaging, a tracklet X is defined by the vector of its successive positions at the different time frames X=X0,,XN-1 , with N the length of the tracklet. We considered that a receptor was immobile if.

maxij=0,,N1||Xi  Xj ||<2 l

where l is the size of the object (l=2 pixels typically). In other words, the previous criterion states that a tracklet is immobile if the maximal distance between two different positions is at most equal to the length of the diagonal of the square of edge l.

The three types of motion of mobile receptors

Request a detailed protocol

To classify the other tracklets corresponding to mobile receptors, we used the statistical method introduced in Briane et al., 2018, which allows to distinguish three main types of motions:

(i) Brownian motion: the object (receptor) evolves freely and its trajectory is denoted by σBt where σ is called the diffusion coefficient. The position of the object Xt at time t is given by Xt=X0+σBt . Brownian increments σdBt at each time are independent and normally distributed.

(ii) Directed motion: the object is actively transported by a deterministic force, and its motion can be modelled by the following stochastic differential equation:

dXt= μdt+ σdBt,

where μ is a 2D-vector called drift and represents the deterministic force, and σ is the diffusion coefficient modelling the random Brownian motion.

(iii) Confined motion: the object is confined in a domain or evolves in an open but crowded area. This kind of motion can be modeled by an Ornestein-Uhlenbeck process:

dXt= -λXt-μdt+ σdBt.

We refer to Durrett, 2018 for more properties about Brownian motion and stochastic calculus.

Statistical classification of mobile tracklets

Request a detailed protocol

The motion classification criterion defined in Briane et al., 2018 essentially considers the ratio between the maximal distance from the initial point and the length of the tracklets. This can be evaluated by defining the following statistics.

S(X,N)= maxi=0,,N|XtiXt0|[12i=1N|XtiXti1|2]12

where |.| denotes the 2D-Euclidean norm. The classification is made by using the quantiles of order α and 1-α (α=0.05) of such a statistic for Brownian tracklets.

These quantiles, denoted by q(α) and q(1-α) respectively, depend on α and N, and can be computed by Monte Carlo simulations (see Briane et al., 2018). This essentially consists in simulating a high number of Brownian tracklets, computing their statistics values and then evaluating the quantiles.

Then the tracklet motion is said to be confined if S(X,N)<q(α), directed if S(X,N)>q(1-α), and Brownian otherwise. For N=5 and α=0.05, we obtained q(α)=0.724 and q(1α)=2.464.

From local classification of tracklet motion to global analysis of receptors’ tracks

Request a detailed protocol

The above statistical classifier allows estimating the local motion of each receptor. In a second time, we analyzed the difference of tracklet motions along the same longer receptor track. In particular, we evaluated if a receptor changed its type of motion along its trajectory.

Finally, our statistical framework for classifying tracklets motion provided a two-scales picture of the receptors’ dynamic behavior: the classification of tracklets provided a global estimation of receptors’ motion, while the identified changes of receptors’ motion along their full trajectories indicated the stability of each receptor’s motion.

Simulating synthetic receptors’ trajectories

Request a detailed protocol

To evaluate the robustness and accuracy of tracklet classification, we first simulated in Matlab n=100 confined trajectories (dXt=-λXt-μdt+σdBt) with length N+1, where N=5 or 10 is the length of used tracklets for classification. Diffusion coefficient σ was fixed to σ=2 and we varied the confinement parameter λ from λ=0 to λ=6 (step =0.2). We then measured the accuracy of classification with pN0={# tracklets classified as confined}{# simulated tracklets n} . As expected, the classification accuracy increases with tracklet length N (Figure 1—figure supplement 1). To account for the risk of mistracking or a change in receptor dynamics, that also increases with tracklet length, we then modeled a generic perturbation in receptor tracking with a standard exponential distribution with rate ρ. Therefore, the conditional probability pN(ρ) for a confined tracklet with length N to be correctly classified is given by pN(ρ)=pN(0)exp(ρN) where pN(0) is the classification accuracy when no risk of mistracking or dynamic change is considered.

In a second time, to measure the robustness of classification to image noise and particle (receptor) density, we simulated a mixture of n=1,000 of Brownian and Confined trajectories (the percentage of simulated confined trajectories was fixed to 10, 50, or 100%) and generated the associated synthetic fluorescence time-lapse sequences using a mixed Poisson-Gaussian model as described in Lagache et al., 2021. We implemented the simulator of synthetic tracklets of fluorescent spots in ICY (Plugin Dynamics Simulator). Using our simulator, we varied the signal-to-noise ratio (SNR) from 10 to 2 (we measured a mean SNR~10 in our experimental dataset). Concerning the density of receptors’ spots, we varied the size of the simulated sequence from xy = 1600 × 1600 pixels to 800 × 800 pixels and 400 × 400 pixels, corresponding respectively to spots’ density = 0.039, 0.16, and 0.63 spots/μm2 , the measured density being <0.5 spots/μm2 in most experiments.

Stoichiometry analysis

Request a detailed protocol

Icy software was used to determine the intensity distribution of eGFP-spots. Spots were detected using the Spot detector wavelet-based algorithm (Olivo-Marin, 2002), and then converted to ROIs with 2 pixels radius. Data was exported to Excel. We observed a multimodal distribution of eGFP spots’ intensities, and we decided to use the AIC criterion (Akaike information criterion; Akaike, 1974) to uncover the number of modes in intensity distribution. Each mode putatively corresponds to a number of molecules. Therefore, statistical characterization of the multimodal distribution of eGFP spots’ intensity will help to classify each spot with respect to its mode and, therefore, to its estimated number of molecules.

AIC analysis starts with the modeling of the empirical distribution e(x) of eGFP spots’intensities with a weighted sum of Gaussian laws,

ex=i=1pαiN(μi,σi)

where p is the number of Gaussian laws in the mixture, αi the weight of each law and (μi,σi) the corresponding mean and variance. For a fixed p, we first searched for the optimal parameters (αi,μi,σi), for i=1..p that maximize the likelihood L of the model to the data:

Lpα1,μ1,σ1,,αp,μp,σp=j=1ni=1pαi2πσiexp-xj-μi22σi

where (x1,x2,,xn) are the observed eGFP intensities in the considered frame of the time-lapse sequence.

This first step of the AIC analysis provides the calibrated parameters (αi, μi,σi)i=1..p when fitting a p-mixture model to data. Then, we computed the optimal number of modes p* that would describe the different populations of eGFP spots with respect to their estimated number of molecules by minimizing the AIC:

AICp=2kp-2log(Lp*)

where Lp* is the maximized likelihood the p-mixture model, and kp=3p-1 is the number of free parameters of the p-mixture model.

Data availability

The code used for data analysis in MatLab was provided as Figure 1—source code 1. The numerical data used to generate each figure and figure supplement were provided as source data files.

References

Decision letter

  1. Volker Dötsch
    Senior and Reviewing Editor; Goethe University, Germany
  2. Volker Dötsch
    Reviewer; Goethe University, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Single-molecule imaging reveals distinct effects of ligands on CCR5 dynamics depending on its dimerization status" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Volker Dötsch as the Reviewing Editor and Reviewer #3, and the evaluation has been overseen by a Senior Editor.

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

While everyone involved in the discussion of your manuscript was convinced that investigating the dynamic behavior of individual receptors on the cell surface is a very important and very ambitious task, concerns were raised whether the experimental data reported support the conclusions. The consensus was that more experimental data are necessary to fully support these conclusions. As these experiments will very likely take longer than 2 months, it is the policy of eLife to reject in such cases the manuscript. More details about the discussed problems you can find in the individual reviews below which will hopefully help you to revise the manuscript for submission to another journal.

Reviewer #1 (Recommendations for the authors):

The manuscript from Momboisse et al. provides a single molecule tracking (SMT) investigation of the CCR5 receptor on the surface of living cells.

Taking advantage on the ability of SMT to detect individual trajectories and hence monitor the behaviour of specific receptors over time, the authors approach aims at investigating fluctuations in the dynamic behavior of individual receptors in response to alterations in conformational changes of the receptor, either spontaneous, or upon stimulation with specific agonists or antagonists.

The main general hypothesis of the study, with GPCR-wide relevance, is that receptor dynamics can be correlated to its molecular conformation (active vs inactive, in the coarsest description). The specific hypothesis related to CCR5 is that gp120 (HIV-1 envelope glycoprotein) acts as an agonist, and that receptor dimerization has an influence on the fate of the activated receptors.

Although the introduction of the manuscript is very well written, and makes a very compelling case for the story, my assessment is that the experimental data reported support these conclusions only to a limited extent, and more experimental data should be gathered to fully support the conclusions. This can be ascribed to methodological considerations, data analysis and pharmacological reasons that I will discuss here in detail.

The manuscript articulates around 6 main figures: the first figure introduces the methodological approach (first developed by Briane et al) based on the segmentation of individual trajectories into tracklets, which are in turn classified into the three categories of free diffusion, confined diffusion and directed motion. Figure 2 displays the outcome of this approach when applied to approx. 20000 tracks of wt CCR5, in basal conditions, showing that most tracklets (80%) display Brownian motion, and only 20% display restricted motion. We can define the outcome of the breakdown and categorization of tracklet motions as the dynamic fingerprint of the receptor. Figure 3 illustrates the effect of ligands on the dynamic fingerprint.

The authors use two agonists (CCL4 and PSC-RANTES) and one antagonist (maraviroc, MVC), to show that PSC-RANTES (10 minutes incubation) is able to significantly shift the fingerprint towards more immobile CCR5, whereas the antagonist increases the percentage of freely diffusing tracklets. Given the timescale after ligand addition, these results are compatible respectively with an increase and a decrease of receptor endocytosis, which is not unexpected.

In Figure 4 the authors explore the effect of Gi (by using PTX to uncouple) and β arrestin (Si-RNA) in modulating their dynamic fingerprint. While the effect of PTX is harder to dissect, knocking down β-arrestin expression clearly impacts the ability of the receptor to become 'restricted' in its motion, again consistent with the expected role of β-arrestin in initiating receptor endocytosis. Figure 4 brings about the role of dimerization in modulating CCR5 dynamic fingerprint. By switching to the analysis of the intensity of individual spots, rather than of their dynamics, the authors now compare the wt CCR5 to a dimerization impaired mutant, the L196K. The notable finding here is that the application of the antagonist increases the dimeric population in the dimerization impaired mutant, which in turn displays also a more restricted dynamic behavior. Such restricted dynamic behavior is also increased by PTX treatment.

Here, in summary, it appears that G protein uncoupling (either by PTX treatment and/or antagonist treatment) leads to a more restricted dynamic fingerprint of CCR5, which is a novel and somewhat counterintuitive result.

Finally, the effect of distinct HIV-1 capsid glycoproteins is studied in their agonistic capacity against CCR5 (Figure 6), with one gp (#34) having a marked effect increasing the number of restricted tracks in both wt and monomerized receptor mutant.

Overall, these data are used to support several conclusions by the authors, namely that (i) receptors switch mode of motion over time in response to changes in their molecular conformation, (ii) that this process is ligand dependent, (iii) HIV-1 gps act as (biased) agonists and (iv) dimerization ultimately regulates agonist response and subsequent endocytosis.

This review aims at verifying if the experimental data provided by the authors confirm the hypotheses and support the authors conclusion that dimerization of CCR5 plays an important role in modulating precoupling of the receptor to G protein, as well as the fate of individual receptors upon activation.

Methods: the authors make the significant choice of using EGFP N-terminally tagged receptors, instead of using cell lines with SNAP-tagged receptors (although available to the authors). The use of EGFP as opposed to an organic dye limits the photon budget significantly, and it is honestly unclear how trajectories with more than a few frames can be obtained. It is experience of the reviewer and several of his colleagues that tracks with more than 5-10 frames using fluorescent proteins are quite uncommon. At the (rather high) particle density displayed by the authors in their SI movies, there is the likelihood that the long trajectories observed stem from relinking of trajectories belonging to distinct receptors. Also, the notion that receptors conformational transitions reflect on receptor dynamics is extremely suggestive, but would definitely require a more stringent validation, for example combining SMT with a conformational sensor, i.e. by single molecule FRET. There are several inconsistencies in the way the data are presented. As a prominent example, the effect of PTX in the barchart of Figure 4A does not appear to be consistent with the time 0 behavior of panel 4B.

Moreover, the density of the molecules in the examples displayed raises questions on how a pure intensity analysis of the spot may accurately separate dimers from monomers. It would be paramount to rule out other effects that could cause the difference in intensity values and shapes of the histrograms in Figure 5 A, such as receptor expression levels. Also, other approaches to validate dimerization/monomerization such as molecular brightness could be easily implemented and should be used to validate the results.

Pharmacology: The dynamic behaviour of the receptor (% of restricted tracklets) in response to ligands is generally not surprising, and consistent with the receptor becoming immobilized in clathrin pits and being internalized over a timescale of several minutes.

The key novel and counterintuitive finding concerns the increase in restricted motion of the L196K mutant upon antagonist stimulation, although there is no hard evidence that this effect is strictly caused by the basal monomeric behavior of the receptor. Regarding effects on G protein coupling, please see my comment above about inconsistencies within PTX data.

Overall, the manuscript has a very ambitious and interesting scope, that would be better served, in my humble opinion, by several manuscripts(!), each dealing with individual aspects of this story, but going at a higher level of detail. The authors could use several methods to validate key findings (see comment above, for example, concerning dimerization).

I have the following key recommendation for the authors:

1. Repeat the experiments (for validation) using SNAP-tagged receptors, at lower density, and in a cell line with better membrane adhesion to the coverslip than HEK293. An option would be the HEK293AD variant, which typically displays a nice and extended basolateral membrane.

2. Check thoroughly the PTX data, since the data in Figure 4A and 4B appear inconsistent

3. Validate the dimerization data using a second, independent method

Reviewer #2 (Recommendations for the authors):

The manuscript by Momboisse et al., "Single molecule imaging reveals distinct effects on ligands on CCR5 dynamics depending on its dimerization status" seeks to understand the mobility, oligomerization and internalization parameters of the CCR5 receptor on the surface of live cells using total internal reflection microscopy. In this context, the manuscript specifically seeks to examine how these parameters are affected by chemokines and other extracellular interacting partners relevant to receptor regulation. These variables include investigations of pertussis toxin (PTX), PCS-RANTES and two solubilized forms of the gp120 heterotrimer derived from patients. The manuscript also describes analogous/parallel studies in the context of a previously characterized 'dimerization-compromised mutant (L196K)' as a means to examine whether CCR5 dimerization plays a critical role in the aforementioned parameters, downstream signal propagation, including receptor internalization.

The motivation for the proposed investigations is quite clear – although some of the language used to delineate the contributions/importance of conformational diversity to regulation – intermingled with compositional diversity – are vague in nature such that it is difficult for the reader to understand the underlying complexities in meaningful detail. The experimental variables examined are quite appropriate and appropriately considered with respect to exploring the relationship between CCR5 track dynamics (Brownian, Fixed or Mixed) and CCR5 dimerization and downstream signaling propensities. For the dimerization component of the investigations, the reader is left to believe solely on the basis of previously literature that the L196K mutant specifically/only effects the dimerization propensity of CCR5 and has no impact on any aspect of agonist-induced conformational response or downstream signaling. More clarity on what this mutation specifically effects would be helpful to the reader. While it seems a number of clarifications are required, the overarching finding is that eGFP-tagged CCR5 exhibits distinct types of dynamics ranging from highly mobile to highly fixed (no mobility), where in the fixed cellular domains CCR5 appears to cluster in a manner that the authors speculate relates to endocytosis. This overall take home message seems consistent with prior literature and knowledge in the field.. The finding that CCR5 exhibits distinct signaling propensities and interaction partners dependent on these dynamic attributions and its conformational states seems logical given previous findings, but less well-supported by the data presented.

For their live-cell tracking studies, the authors use a previously described eGFP-CCR5 construct that was a gift from F. Perez, originally reported in Boncompain et al. 2019. While eGFP has been used by multiple groups in the past to track the oligomerization status and mobility of cellular protein components (see for instance Iino, Moerner et al. BPJ 2001; Ulbrich and Isacoff NMETH 2007 and; Cui et al. Molecular Plant 2018), investigations into the fluorescent properties of eGFP that govern its utility for the types of tracking experiments performed (see for instance Peterman et a. J. Phys Chem A 1999 and Vamosi et al. Scientific Reports 2016) have generally led to the realization that confidence in quantitative tracking outputs from such experiments are prone to compromises that have motivated the field to move in the direction of increasing the signal-to-noise ratio of this type of study to quantum dot (see for instance Veya et al. JBC 2015) or organic fluorophore-labeled species (see for instance, Sako and Yanagita et al. Nat Cell Biol. 2000; Hern et al. PNAS 2010; Calebiro et al. PNAS 2013; Moller et al. Nat Chem Biol. 2020; Asher et al. NMETH 2021). From my review of the literature, it seems that eGFP is not always immediately matured (ie. fluorescent) in the cell and at the cell surface, it is prone to intermittent blinking and rapid photobleaching under conditions of illumination and acquisition frame rates employed in the present investigations. eGFP is also quite dim by comparison to organic fluorophores due principally to a relatively low absorbance cross section. For all these reasons, the total photon budget of eGFP is relatively low and variable, factors that can make robust tracking quite challenging. For instance, in the references cited above, it appears that eGFP has a lifetime of roughly 180 ms at 5 kW/cm2 (the precise illumination intensity used in the present investigations in not clear). At 100 ms integration time eGFP brightness per frame is about 12-fold above background. A commensurately lower total photon budget and signal above background is therefore expected at 33 ms integration time used in the present investigation. At higher illumination intensities, the results of Vamosi et al. (Sci. Report (2016)) would contend that further enhancements in eGFP brightness and photo budget are unlikely. Based on the data provided in the manuscript it is not possible to know what the quality of monomeric eGFP-CCR5 are: how bright the fluorophore is during the tracks; the variance in brightness during the tracks; how long the tracks are both in terms of time and total photon budget. This information should be reported as photons as opposed to AU. The incident illumination intensity should be reported in kW/cm2 so that others can reproduce the conditions, rather than a percent of the potential power output at the laser head. All this information seems essential for reader comprehension. Without such information, the aforementioned considerations from prior literature give pause to the believability behind the fundamental assumption that monomeric species are being tracked and that robust quantitative results have been obtained. Supporting the idea that eGFP may present challenges as a fluorophore, a recent paper by Li et al. "Oligomerization-enhanced receptor-ligand binding revealed by dual-color simultaneous tracking in living cell membranes", (J Phys Chem Lett, 2021) (which used eGFP in their prior investigations), seemingly relevant to the present investigations replaced eGFP on their CCR5 expression construct with mNeonGreen, stating than mNeonGreen is about 3-fold brighter than eGFP and exhibits a nearly 3-fold shorter maturation time. These concerns speak to the need for careful consideration and delineation of the details underlying the experiments, which at present appear to be absent form this version of the manuscript. Consequently, my interpretation of the results started with examination of the construct examined to determine if it was an oligomeric eGFP construction being used.

After digging a bit through the Boncompain manuscript (the original paper describing the eGFP-CCR5 construct), I was not able to discern the precise nature of the eGFP-tagged CCR5 construct that was utilized in the present investigation. This consideration speaks to the overall paucity of information provided by the authors as to the underlying details of the experiments performed that would enable reproducibility. As far I can discern, the construct employed is based on a "RUSH" plasmid in which a streptavidin binding protein (SBP) is fused first to eGFP and then to CCR5 (N- to C-terminus) and co-expressed with an ER-resident protein (KDEL)-fused to streptavidin. In this assay, the plasmid containing this dual expressing system "Str-KDEL_SBP-EGFP-CCR5" is first transfected to obtain a stable cell line and then the stable cell line constitutively expresses the SBP-eGFP-CCR5 construction such that it is sequestered in the endoplasmic reticulum (ER) until the addition of biotin. Biotin releases the streptavidin-sequestered eGFP-CCR5 molecules trapped in the ER by outcompeting the SBP interaction. Biotin releases the SBP-eGFP-CCR5 protein to the cell surface through a Golgi apparatus-mediated secretion pathway. I was unable to find any mention of biotin in the manuscript so additional details about the stable cell line construction and protein expression seem relevant to include.

Given the data presented, it is not entirely clear what state of oligomerization the expressed eGFP-tagged CCR5 protein is in upon reaching the cell surface, or the tracks that being quantified. The authors state that CCR5 is naturally found in a variety of homo- and hetero-oligomeric complexes and hence, it is not at all clear what the precise composition is of the "particles" – homomeric, heteromeric or otherwise – being tracked. Are these considerations not important to the interpretations of their findings? It certainly seems relevant to take these considerations into account in regard to the conveyance of quantitative interpretations to the reader of the underlying biology.

In the vein of increased precision, the authors may also wish to temper their claims about the "original" or "novel" nature of their 'quantitative' tracking approach. If I am understanding things correctly, the acquisition of the tracks is based on the software provided in https://www.cell.com/cell-reports/pdf/S2211-1247(18)30071-8.pdf, where the classification of the tracklets (individual tracks computationally parsed into 5 frames) relies on a statistical tool initially developed in https://hal.inria.fr/hal-01416855/document and later published in https://arxiv.org/pdf/1804.04977.pdf. On this point, it is somewhat unclear to me that the 'innovation' of segmenting tracks into 5 frames using a software that was constructed and optimized using tracks of 30 frames (if I understand things correctly), is without hazard. Commentary on why the chosen analysis method(s) may be superior to other analysis approaches also seems worthy of comment.

In this light, it seems excessive to repeatedly mention the originality of their SPT tool. Moreover, as for the shortage of references to prior literature that have attempted to perform live-cell tracking of single-molecules in the membrane, the authors should be aware that the processing approaches employed have been implemented for many years by multiple groups. See for instance:

https://ani.stat.fsu.edu/~hycao/docs/pub/biometrika_15.pdf

https://www.nature.com/articles/nmeth.3483

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0082799

https://hal.inria.fr/hal-01416855/document

https://arxiv.org/pdf/1804.04977.pdf. (cited in the paper).

https://www.cell.com/biophysj/pdfExtended/S0006-3495(18)30132-2

https://doi.org/10.1016/j.ymeth.2020.03.008

https://iopscience.iop.org/article/10.1088/1478-3975/ab64b3

https://www.nature.com/articles/nmeth.2808

The specific novelty of the present investigations relates to the choice of parameters (N and α) that are required for the classification of the tracklets (pieces of tracks acquired via ICY software as it is introduced in https://arxiv.org/pdf/1804.04977.pdf). In the paper, the authors made the choice of N=5 and the choice of α = 0.05. The authors justify the use of these chosen parameter values in the supplementary material using statistical tests to examine the robustness/validity of these choices. However, the simulated tracks generated for this test, which exhibit the specific motion models that they aim to differentiate in live cells (Brownian, fixed or mixed), do not seem to be similar to the tracks actually examined in the authors' experiments. Specifically, the authors do not appear to have incorporated any measurement noise in their tracks, including background noise fluctuations, eGFP signal variances, blinking etc. At the particle densities shown in the SI movies, it also seems unavoidable that errors will be made when particle tracks cross paths. This validity test therefore seems to be quite unrealistic as it pertains to confidently discerning quantitative aspects of the underlying biology. For the reader to have confidence in the results presented, it would seem imperative to provide actual eGFP tracks to the reader in terms of photons/AU per frame as a function of time and provide some estimate of how relevant the chosen parameters are from their simulations to the actual experimental setting where the signal-to-noise ratio is much lower presumably and the tracks are potentially more crowded within the cell-surface area. What errors in the authors' measurements may be associated with such considerations?

Given these considerations, while the overall goals of the work are important and the results appear globally consistent with prior literature, without clarifications to issues delineated above, it is difficult to ascertain the validity of the authors' overall interpretation that CCR5 can dimerize in a manner that directly relates to G-protein/arrestin signaling and to what quantitative extent dimerization contributes to these events or the subsequent process of internalization. In the absence of significant efforts to address the concerns raised, the authors may wish to temper their claims and conclusions substantially and address their observations and qualitatively supportive of the models proposed, which do seem reasonable in nature and globally consistent with expectation.

Reviewer #3 (Recommendations for the authors):

Investigating the dynamic interaction of GPCRs on the surface of cells is of great importance for improving our understanding of signaling across the plasma membrane. Momboisse et al. use TIRF based single-molecule imaging to track the movement of individual CCR5 molecules under different conditions. They use a different tracking statistics method that enables the analysis of motion changes. They show that treating cells with agonists results in more restricted movements and clustering of receptors as well as receptor endocytosis. Responsible for these effects is the interaction with β-arrestin. Inverse agonist have the opposite effect while binding of the HIV-1 envelope glycoprotein gp120 shows agonist-like properties.

Important for the immobilization of the receptor is its dimerization (and to a small percentage formation of higher oligomers).

The reported results are a nice example of the quantitative analysis of the movement of cell surface receptors and will be of interest for the analysis of other GPCRs and cell surface receptors as well.

1) Heterodimerization with other GPCRs (e.g. CCR2) are documented. Can it be excluded that for example the percentage of receptors showing restricted movement in non stimulated cells is due to such oligomerizations? CCR2 or other receptors would not be fluorescence labeled and could therefore not be detected.

2) The clustering of arrestin has been documented and discussed in the two following manuscripts:

1) Coordinate-based co-localization-mediated analysis of arrestin clustering upon stimulation of the C-C chemokine receptor 5 with RANTES/CCL5 analogues.

Tarancón Díez L, Bönsch C, Malkusch S, Truan Z, Munteanu M, Heilemann M, Hartley O, Endesfelder U, Fürstenberg A. Histochem Cell Biol. 2014 Jul;142(1):69-77. doi: 10.1007/s00418-014-1206-1. Epub 2014 Mar 13. PMID: 24623038

2) Quantitative morphological analysis of arrestin2 clustering upon G protein-coupled receptor stimulation by super-resolution microscopy.

Truan Z, Tarancón Díez L, Bönsch C, Malkusch S, Endesfelder U, Munteanu M, Hartley O, Heilemann M, Fürstenberg A. J Struct Biol. 2013 Nov;184(2):329-34. doi: 10.1016/j.jsb.2013.09.019. Epub 2013 Sep 30. PMID: 24091038

These should be cited and the results being included in the discussion.

https://doi.org/10.7554/eLife.76281.sa1

Author response

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Reviewer #1 (Recommendations for the authors):

The manuscript from Momboisse et al. provides a single molecule tracking (SMT) investigation of the CCR5 receptor on the surface of living cells.

[…]Overall, these data are used to support several conclusions by the authors, namely that (i) receptors switch mode of motion over time in response to changes in their molecular conformation, (ii) that this process is ligand dependent, (iii) HIV-1 gps act as (biased) agonists and (iv) dimerization ultimately regulates agonist response and subsequent endocytosis.

This review aims at verifying if the experimental data provided by the authors confirm the hypotheses and support the authors conclusion that dimerization of CCR5 plays an important role in modulating precoupling of the receptor to G protein, as well as the fate of individual receptors upon activation.

Methods: the authors make the significant choice of using EGFP N-terminally tagged receptors, instead of using cell lines with SNAP-tagged receptors (although available to the authors). The use of EGFP as opposed to an organic dye limits the photon budget significantly, and it is honestly unclear how trajectories with more than a few frames can be obtained. It is experience of the reviewer and several of his colleagues that tracks with more than 5-10 frames using fluorescent proteins are quite uncommon. At the (rather high) particle density displayed by the authors in their SI movies, there is the likelihood that the long trajectories observed stem from relinking of trajectories belonging to distinct receptors. Also, the notion that receptors conformational transitions reflect on receptor dynamics is extremely suggestive, but would definitely require a more stringent validation, for example combining SMT with a conformational sensor, i.e. by single molecule FRET.

Observed long trajectories with eGFP tagged receptors do not necessarily correspond to particle relinking as we only considered tracks with relevant signal-to-noise ratio (SNR) and few missed detections and used a very moderate illumination intensity (0.4 kW/cm2) over short periods (3-6 seconds) to decrease bleaching.

That said, and at high particles’ density, there is indeed a risk of mistracking and trajectories’ relinking. To mitigate that risk, we fragmented receptors’ tracks into very short tracklets (N=5 frames). To assess the robustness of our classification method with respect to the density of tracked particles, we performed a novel set of simulations where we simulated n=1000 confined tracklets (N=5 frames) and generated the corresponding synthetic movies with a mixed Poisson-Gaussian noise model (see Material and Methods) for 3 different spots’ densities (d=0.039, 0.16 and 0.63 spots/μm2, the measured density being <0.5 spots/μm2 in most experiments). For different percentages of confined tracklets (10 % confined & 90% Brownian; 50 % confined & 50% Brownian and 100 % confined) we observed that the classification accuracy was robust to increased densities of spots (Author response image 1), which assessed the robustness of our algorithm in experimental conditions (Figure 1, figure supplement 1) (page 7).

Author response image 1
Relative accuracy of statistical classification for increasing spots’ density.

The proportion of confined trajectories was set to 10% (90% Brownian, blue), 50% (50% Brownian, red) and 100% (0% Brownian, green). Confinement parameter was fixed to λ = 2. For each condition (density and proportion of confined trajectories), we run n = 10 simulations with 1000 moving spots in a 1600x1600 pixels (density = 0.039 spots/μm2), 800 X 800 pixels (density = 0.16 spots/μm2) and 400 X 400 pixels (density = 0.63 spots/μm2).

In addition, to confirm our findings and to avoid a bias related to a unique model, we have added a new dataset carried out in another model, in which we tracked FLAG-SNAP-tagged proteins using fluorescent antibodies. The results obtained with this second model (FLAG-SNAP-tagged-CCR5) confirmed those obtained with the first one (eGFP-CCR5) in the different conditions tested (see details below in the “recommendation to authors” section).

There are several inconsistencies in the way the data are presented. As a prominent example, the effect of PTX in the barchart of Figure 4A does not appear to be consistent with the time 0 behavior of panel 4B.

We corrected Figure 4B (in presence of PSC-RANTES) by adding the time 0, which was missing. As for Figure 4A (basal state), we observed at t0 a small but significant decrease in CCR5 mobility restriction in the presence of PTX compared to untreated cells. This suggests that a small fraction of CCR5 is in Gi protein-bound form in the basal state

Moreover, the density of the molecules in the examples displayed raises questions on how a pure intensity analysis of the spot may accurately separate dimers from monomers. It would be paramount to rule out other effects that could cause the difference in intensity values and shapes of the histrograms in Figure 5 A, such as receptor expression levels. Also, other approaches to validate dimerization/monomerization such as molecular brightness could be easily implemented and should be used to validate the results.

We agree that a fluorescence intensity analysis of the spot may not be sufficient to accurately discriminate monomers from dimers. We attributed the fluorescence intensities of the 3 types of Gaussians observed in eGFP-CCR5-WT expressing cells to monomers/dimers/oligomers based on the fluorescence intensity of eGFP spotted on coverslip, which quenched in a single step (Salavessa, 2021).

However, we are aware that this fluorescence intensity measurement is relative to eGFP on coverslip. Therefore, we are now describing the different receptor forms as « low », « medium » or « high » fluorescence intensity forms instead of monomers/dimers/oligomers (page 17). As a consequence, we changed the title, by removing the term “single molecule tracking”.

However, regardless of the exact amounts of individual molecules in the “low”, “medium” or “high” fluorescence populations, their relative proportions differed between CCR5-WT and the dimerization-compromised mutant CCR5-L196K. We previously showed by alternative approaches (molecular modeling, HTRF, and a functional assay) (Jin, 2018) that this mutant formed less dimers and higher order oligomers than CCR5-WT. This is in agreement with the results in Figure 5A showing a reduction in the proportions of the “high” and “medium” fluorescence populations for eGFP-CCR5-L196K in favor of the “low” fluorescence population, relative to CCR5-WT. Figure 5B also showed that the CCR5 inverse agonist MVC increased the proportion of CCR5-L196K populations with “high or medium” fluorescence intensity. Again, this is in accordance with our previous work showing that MVC can stabilize CCR5-L196K in a dimeric conformation (Jin, 2018). Considered altogether, these data indicate that our classification method based on fluorescence intensity analysis of receptor populations is accurate to characterize the heterogeneity of receptor organization at the cell surface.

Characterizing the « exact » stoichiometry of molecules per spot is another question, highly challenging, which we will examine in a future study using molecular brightness and super-resolution approaches.

Of note, to minimize the influence of the amount of surface receptor on the stoichiometry of the molecule/spot, we worked with clones expressing similar amounts of cell surface receptors as mentioned page 17, line 460.

Pharmacology: The dynamic behaviour of the receptor (% of restricted tracklets) in response to ligands is generally not surprising, and consistent with the receptor becoming immobilized in clathrin pits and being internalized over a timescale of several minutes.

The key novel and counterintuitive finding concerns the increase in restricted motion of the L196K mutant upon antagonist stimulation, although there is no hard evidence that this effect is strictly caused by the basal monomeric behavior of the receptor. Regarding effects on G protein coupling, please see my comment above about inconsistencies within PTX data.

As mentioned by the reviewer, the dynamic behavior of eGFP-CCR5-WT upon activation supports receptor trapping in nanodomains. However, contrary to what is proposed by the reviewer, receptor clustering and massive immobilization is not an absolute prerequisite for receptor endocytosis. Indeed, we showed that CCR5-L196K, although impaired in agonist-dependent immobilization (Figure 5E, F), successfully recruits b-arrestins in response to agonist binding and is internalized (see below and new Figure 6). This suggests that receptor immobilization, triggered by barrestin recruitment, requires receptor oligomerization.

On one hand, our data showed a change in CCR5-WT mobility upon agonist activation towards immobilization, while treatment with an inverse agonist had the opposite effect (Figure 3). PTX treatment and b-arresting silencing suggested that this effect was not Gi-protein-dependent but dependent on b-arrestins recruitment (Figure 4).

On the other hand, results with the dimerization-compromised mutant CCR5-L196K showed that CCR5 clustering and immobilization was not a pre-requisite to its internalization. Indeed, contrary to eGFP-CCR5-WT massive immobilization, eGFP-CCR5-L196K was only slightly immobilized after 10 min of agonist treatment (Figure 5 E-F). In this condition, the internalization process was less efficient but not abrogated (Figure 6C), which is counterintuitive. We assumed that CCR5-L196K may lead to a conformation that recruits little or no b-arrestins or that interacts differently with b-arrestins. To test this hypothesis, we have now added a new set of experiments to evaluate agonist-induced barr2 recruitment. We showed that b-arrestins is correctly recruited to the plasma membrane upon CCR5-L196K activation (new Figure 6A, B). This suggested that the lack of immobilization and clustering of CCR5-L196K is not due to a default of b-arrestins recruitment. We propose that the conformational organization of the receptor and in particular its oligomeric status is necessary for the b-arrestin to trigger receptor clustering and optimize receptor internalization.

These findings are presented page 20 and discussed page 28.

We agree that multiple factors may come into play to explain the lack of immobilization of activated CCR5-L196K. However, we showed previously that CCL3 binds CCR5-L196K and CCR5-WT with similar affinities (Jin, 2018; Colin, 2018). In addition, we have added new results showing that activated CCR5-WT and CCR5-L196K similarly transduced ERK signaling (Figure 5—figure supplement 1). That is why we attributed the lack of CCR5-L196K immobilization after stimulation to its altered capacity to forms dimers, and not to an effect on ligand binding or signaling.

I have the following key recommendation for the authors:

1. Repeat the experiments (for validation) using SNAP-tagged receptors, at lower density, and in a cell line with better membrane adhesion to the coverslip than HEK293. An option would be the HEK293AD variant, which typically displays a nice and extended basolateral membrane.

We carried out another set of experiments using a second model in which we tracked FLAG-SNAP-tagged proteins using fluorescent antibodies. The results obtained with this second model (FLAG-SNAP-tagged-CCR5) confirmed those obtained with the first one (eGFP-CCR5) in the different conditions tested: after ligand binding (using two agonists, an inverse agonist, and HIV-1 envelope glycoproteins) and after modulation of receptor dimerization (using the eGFP-CCR5-L196K mutant). They validated our tracking analysis and our conclusions, supporting that our findings are independent of the model used. We added 4 Figures as Figure supplements and 1 video (Video 4, Figure 2—figure supplement 1, Figure 3, figure supplement 2, Figure 5—figure supplement 2, Figure 7—figure supplement 1). We mentioned the results in the manuscript pages 6, 10, 13, 18, 22.

2. Check thoroughly the PTX data, since the data in Figure 4A and 4B appear inconsistent.

As mentioned above, we corrected Figure 4B by adding the time 0, which was missing.

3. Validate the dimerization data using a second, independent method.

As mentioned above, we agree that the fluorescence intensity of the spots cannot accurately validate the stoichiometry of the molecules per spot. We are now describing the different receptor forms by their relative fluorescent intensity (« low », « medium » or « high » fluorescent intensity forms) compared to eGFP spotted on coverslip instead of monomers/dimers/oligomers (page 17).

This approach, even relative, confirmed that CCR5-L196K forms fewer high order oligomers than CCR5-WT. This is fully consistent with our previous work using alternative approaches (molecular modeling, FRET, and a functional assay), and showing that the L196K mutation in the CCR5 dimerization interface impairs the receptor in its capacity to form di-/oligo-mers (Jin, 2018).

Reviewer #2 (Recommendations for the authors):

[…]

The experimental variables examined are quite appropriate and appropriately considered with respect to exploring the relationship between CCR5 track dynamics (Brownian, Fixed or Mixed) and CCR5 dimerization and downstream signaling propensities. For the dimerization component of the investigations, the reader is left to believe solely on the basis of previously literature that the L196K mutant specifically/only effects the dimerization propensity of CCR5 and has no impact on any aspect of agonist-induced conformational response or downstream signaling. More clarity on what this mutation specifically effects would be helpful to the reader.

We are now describing in more details the influence of the mutation L196K on the functional properties of CCR5 (page 17).

Previously, we showed that Leu-196 in TM5 is part of the CCR5 dimerization interface. Its substitution by Lys in CCR5-L196K inhibits the capacity of the receptor to dimerize, as revealed by complementary approaches such as molecular modeling, energy transfer, and a functional export assay (Jin, 2018).

Functionally, the L196K mutation decreases cell surface expression of the receptor due to retention in the endoplasmic reticulum (Jin, 2018). However, CCR5-L196K folding is not altered as shown by its ability to bind chemokines and HIV-1 gp120s with the same affinity as CCR5-WT (Jin, 2018; Colin, 2018). Additionally, in cells expressing similar amounts of CCR5-WT or CCR5-L196K, we now showed that both receptors similarly activate ERK1/2 after agonist stimulation (Figure 5—figure supplement 1).

Finally, our manuscript indicated (i) that receptor immobilization depended on b-arrestin recruitment (Figure 4) and (ii) that only receptors able to oligomerize were immobilized upon activation (Figure 5). We now added a set of data revealing that CCR5-L196K and CCR5-WT similarly recruited b-arrestins (Figure 6A, B) and was internalized upon activation (Figure 6C). This suggested that b-arrestin-induced receptor immobilization requires receptor oligomerization and that receptor immobilization is not a prerequisite for receptor endocytosis. However, CCR5-L196K internalization was less efficient compared to CCR5-WT, suggesting that receptor immobilization and clustering are important for an optimal endocytosis.

While it seems a number of clarifications are required, the overarching finding is that eGFP-tagged CCR5 exhibits distinct types of dynamics ranging from highly mobile to highly fixed (no mobility), where in the fixed cellular domains CCR5 appears to cluster in a manner that the authors speculate relates to endocytosis. This overall take home message seems consistent with prior literature and knowledge in the field.

As explained above, our experiments suggest that massive immobilization may not be a prerequisite for endocytosis. Indeed, CCR5-L196K remains highly mobile after agonist stimulation, while being able to be internalized and to recruit b-arrestins. As such, we believe that our results provide an additional degree of information compared to the current literature.

The finding that CCR5 exhibits distinct signaling propensities and interaction partners dependent on these dynamic attributions and its conformational states seems logical given previous findings, but less well-supported by the data presented.

In the present manuscript, we showed that CCR5-WT mobility at the plasma membrane upon PSC-RANTES activation depends on b-arrestins (and not on Gi-protein) (Figure 4), a feature only recently reported after MOR stimulation (Markova, 2021).

The use of a dimerization-compromised mutant added that the mobility of the receptor and its optimal internalization depend on receptor dimerization. In particular, we propose that b-arrestin-induced receptor immobilization requires receptors oligomerization.

For their live-cell tracking studies, the authors use a previously described eGFP-CCR5 construct that was a gift from F. Perez, originally reported in Boncompain et al. 2019. While eGFP has been used by multiple groups in the past to track the oligomerization status and mobility of cellular protein components (see for instance Iino, Moerner et al. BPJ 2001; Ulbrich and Isacoff NMETH 2007 and; Cui et al. Molecular Plant 2018), investigations into the fluorescent properties of eGFP that govern its utility for the types of tracking experiments performed (see for instance Peterman et a. J. Phys Chem A 1999 and Vamosi et al. Scientific Reports 2016) have generally led to the realization that confidence in quantitative tracking outputs from such experiments are prone to compromises that have motivated the field to move in the direction of increasing the signal-to-noise ratio of this type of study to quantum dot (see for instance Veya et al. JBC 2015) or organic fluorophore-labeled species (see for instance, Sako and Yanagita et al. Nat Cell Biol. 2000; Hern et al. PNAS 2010; Calebiro et al. PNAS 2013; Moller et al. Nat Chem Biol. 2020; Asher et al. NMETH 2021). From my review of the literature, it seems that eGFP is not always immediately matured (ie. fluorescent) in the cell and at the cell surface, it is prone to intermittent blinking and rapid photobleaching under conditions of illumination and acquisition frame rates employed in the present investigations. eGFP is also quite dim by comparison to organic fluorophores due principally to a relatively low absorbance cross section. For all these reasons, the total photon budget of eGFP is relatively low and variable, factors that can make robust tracking quite challenging. For instance, in the references cited above, it appears that eGFP has a lifetime of roughly 180 ms at 5 kW/cm2 (the precise illumination intensity used in the present investigations in not clear). At 100 ms integration time eGFP brightness per frame is about 12-fold above background. A commensurately lower total photon budget and signal above background is therefore expected at 33 ms integration time used in the present investigation. At higher illumination intensities, the results of Vamosi et al. (Sci. Report (2016)) would contend that further enhancements in eGFP brightness and photo budget are unlikely. Based on the data provided in the manuscript it is not possible to know what the quality of monomeric eGFP-CCR5 are: how bright the fluorophore is during the tracks; the variance in brightness during the tracks; how long the tracks are both in terms of time and total photon budget. This information should be reported as photons as opposed to AU. The incident illumination intensity should be reported in kW/cm2 so that others can reproduce the conditions, rather than a percent of the potential power output at the laser head. All this information seems essential for reader comprehension. Without such information, the aforementioned considerations from prior literature give pause to the believability behind the fundamental assumption that monomeric species are being tracked and that robust quantitative results have been obtained. Supporting the idea that eGFP may present challenges as a fluorophore, a recent paper by Li et al. "Oligomerization-enhanced receptor-ligand binding revealed by dual-color simultaneous tracking in living cell membranes", (J Phys Chem Lett, 2021) (which used eGFP in their prior investigations), seemingly relevant to the present investigations replaced eGFP on their CCR5 expression construct with mNeonGreen, stating than mNeonGreen is about 3-fold brighter than eGFP and exhibits a nearly 3-fold shorter maturation time. These concerns speak to the need for careful consideration and delineation of the details underlying the experiments, which at present appear to be absent form this version of the manuscript. Consequently, my interpretation of the results started with examination of the construct examined to determine if it was an oligomeric eGFP construction being used.

We thank the reviewer for all of these constructive comments.

We have now specified the illumination intensities (< 0.4 kW/cm2 for tracking; < 0.7 kW/cm2 for fluorescence intensity) (page 37). We measured the power at the output of the microscope related to the smallest illumination disk corresponding to the field of the camera. Note that the sample being larger than the field of the camera, the measured values were probably overestimated. The intensity values measured were 10 times lower than those mentioned in the article cited by the reviewer. The camera detector used in our experiments is likely much more sensitive than that used several years ago in the cited article.

Under this relatively low illumination power, the intensity of the spots is stable throughout the duration of the acquisition (200 frames) (Author response image 2 “mean intensities / frames”) allowing the tracking of long trajectories. In addition, we only considered tracks with relevant SNR and few missed detections. To mitigate a risk of mistracking and trajectories relinking, we also fragmented receptors’ tracks into very short tracklets (N=5 frames) (Materials et Methods pages 40-41).

Author response image 2

As mentioned above in response to the reviewer 1, we considered that the fluorescence intensity analysis of the spot is relative to eGFP spotted on coverslip and is not sufficient to accurately discriminate monomers from dimers. Therefore, we are now describing the different receptor forms as « low », « medium » or « high » fluorescence intensity forms instead of monomers/dimers/oligomers (pages 17). Since we are not quantifying single particles, the photons/AU parameter is not relevant here.Finally, as mentioned in response to the reviewer 1, we carried out another set of experiments using a second model in which we tracked FLAG-SNAP-tagged proteins using fluorescent antibodies. The results obtained with this second model (FLAG-SNAP-tagged-CCR5) confirmed those obtained with the first one (eGFP-CCR5) in the different conditions tested. They validated our tracking analysis and our conclusions, supporting that our findings are independent of the model used (Video 4, Figure 2—figure supplement 1, Figure 3, figure supplement 2, Figure 5—figure supplement 2, Figure 7—figure supplement 1).

After digging a bit through the Boncompain manuscript (the original paper describing the eGFP-CCR5 construct), I was not able to discern the precise nature of the eGFP-tagged CCR5 construct that was utilized in the present investigation. This consideration speaks to the overall paucity of information provided by the authors as to the underlying details of the experiments performed that would enable reproducibility. As far I can discern, the construct employed is based on a "RUSH" plasmid in which a streptavidin binding protein (SBP) is fused first to eGFP and then to CCR5 (N- to C-terminus) and co-expressed with an ER-resident protein (KDEL)-fused to streptavidin. In this assay, the plasmid containing this dual expressing system "Str-KDEL_SBP-EGFP-CCR5" is first transfected to obtain a stable cell line and then the stable cell line constitutively expresses the SBP-eGFP-CCR5 construction such that it is sequestered in the endoplasmic reticulum (ER) until the addition of biotin. Biotin releases the streptavidin-sequestered eGFP-CCR5 molecules trapped in the ER by outcompeting the SBP interaction. Biotin releases the SBP-eGFP-CCR5 protein to the cell surface through a Golgi apparatus-mediated secretion pathway. I was unable to find any mention of biotin in the manuscript so additional details about the stable cell line construction and protein expression seem relevant to include.

We did not use the SBP-eGFP-CCR5 plasmid used in Boncompain, 2019, but a different plasmid expressing eGFP-CCR5 from the CMV promoter. We have now clarified this point in the material and method section and have deleted the reference boncompain, 2019, so as not to mislead the reader (page 34).

Given the data presented, it is not entirely clear what state of oligomerization the expressed eGFP-tagged CCR5 protein is in upon reaching the cell surface, or the tracks that being quantified. The authors state that CCR5 is naturally found in a variety of homo- and hetero-oligomeric complexes and hence, it is not at all clear what the precise composition is of the "particles" – homomeric, heteromeric or otherwise – being tracked. Are these considerations not important to the interpretations of their findings? It certainly seems relevant to take these considerations into account in regard to the conveyance of quantitative interpretations to the reader of the underlying biology.

On one hand, we tracked receptor populations over time, and on the other hand, we analyzed the intensity of spots from frame 1. However, we did not make a link between fluorescence intensity and tracking over time. We therefore do not know how the populations evolve over time. This would require associating AIC algorithm with the tracking one. We will implement such a link in a near future. We consider here that studying the dimerization-compromised mutant, which we well characterized in our previous work (Jin, 2018), is sufficient to support a role of receptor dimerization in its dynamics behavior.

Fluorescence intensity analysis revealed different CCR5 homo-oligomeric organization at the cell surface. Studying CCR5 heterodimerization, which require to follow receptors expressing two different fluorescent proteins, goes beyond the scope of this study. Performing such a study will help understand the impact of the heterodimerization on receptor functions, which is of particular interest.

In the vein of increased precision, the authors may also wish to temper their claims about the "original" or "novel" nature of their 'quantitative' tracking approach. If I am understanding things correctly, the acquisition of the tracks is based on the software provided in https://www.cell.com/cell-reports/pdf/S2211-1247(18)30071-8.pdf, where the classification of the tracklets (individual tracks computationally parsed into 5 frames) relies on a statistical tool initially developed in https://hal.inria.fr/hal-01416855/document and later published in https://arxiv.org/pdf/1804.04977.pdf. On this point, it is somewhat unclear to me that the 'innovation' of segmenting tracks into 5 frames using a software that was constructed and optimized using tracks of 30 frames (if I understand things correctly), is without hazard. Commentary on why the chosen analysis method(s) may be superior to other analysis approaches also seems worthy of comment.

In this light, it seems excessive to repeatedly mention the originality of their SPT tool. Moreover, as for the shortage of references to prior literature that have attempted to perform live-cell tracking of single-molecules in the membrane, the authors should be aware that the processing approaches employed have been implemented for many years by multiple groups. See for instance:

https://ani.stat.fsu.edu/~hycao/docs/pub/biometrika_15.pdf

https://www.nature.com/articles/nmeth.3483

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0082799

https://hal.inria.fr/hal-01416855/document

https://arxiv.org/pdf/1804.04977.pdf. (cited in the paper).

https://www.cell.com/biophysj/pdfExtended/S0006-3495(18)30132-2

https://doi.org/10.1016/j.ymeth.2020.03.008

https://iopscience.iop.org/article/10.1088/1478-3975/ab64b3

https://www.nature.com/articles/nmeth.2808

The specific novelty of the present investigations relates to the choice of parameters (N and α) that are required for the classification of the tracklets (pieces of tracks acquired via ICY software as it is introduced in https://arxiv.org/pdf/1804.04977.pdf). In the paper, the authors made the choice of N=5 and the choice of α = 0.05. The authors justify the use of these chosen parameter values in the supplementary material using statistical tests to examine the robustness/validity of these choices. However, the simulated tracks generated for this test, which exhibit the specific motion models that they aim to differentiate in live cells (Brownian, fixed or mixed), do not seem to be similar to the tracks actually examined in the authors' experiments. Specifically, the authors do not appear to have incorporated any measurement noise in their tracks, including background noise fluctuations, eGFP signal variances, blinking etc. At the particle densities shown in the SI movies, it also seems unavoidable that errors will be made when particle tracks cross paths. This validity test therefore seems to be quite unrealistic as it pertains to confidently discerning quantitative aspects of the underlying biology. For the reader to have confidence in the results presented, it would seem imperative to provide actual eGFP tracks to the reader in terms of photons/AU per frame as a function of time and provide some estimate of how relevant the chosen parameters are from their simulations to the actual experimental setting where the signal-to-noise ratio is much lower presumably and the tracks are potentially more crowded within the cell-surface area. What errors in the authors' measurements may be associated with such considerations?

We thank the referee for having pointed out several relevant papers to our study. Many of them are actually based on the probabilistic framework of Bayesian analysis to determine the most likely repartition of trajectories into classes depending on their diffusion coefficient. While Bayesian analysis can be used to classify trajectories, we chose to rather implement a method based on statistical hypothesis testing as it is more efficient for short tracklets, does not requires an important computational load and can be easily implemented. We have now better discussed the choice of our classification method and added references on page 7.

The “novelty” of our approach relies on the choice of a classification pipeline, which consists here of splitting receptors’ tracks into very short tracklets (N=5 frames) to mitigate the risk of mis-tracking and dynamics’ change along each receptor’s trajectory, and implementing a robust statistical classifier initially developed by Kervrann and co-authors. We have therefore soften our claims throughout the manuscript by removing all the inappropriate “novel”… terms. In addition, and to motivate our approach, we have now performed new sets of simulations to validate our choice of very short tracklets (N=5), and assess the robustness of our statistical classifier with respect to image SNR and spots density. (Figure 1—figure supplement 1)

First, the accuracy of tracklet classification is expected to decrease with tracklet length. By using Monte-Carlo simulations of confined motion for an increasing confinement parameter λ (see Material and Methods), we compared the accuracy obtained for N=5 and N=10 tracklets. We observed that, indeed, the classification accuracy for N=5 is almost 2-fold decreased compared to N=10. Then, to account for the risk of mistracking, or a change in receptor dynamics, we modeled a generic perturbation in receptor tracking with a standard exponential distribution with rate ρ. Therefore, the conditional probability pN(ρ) for a confined tracklet with length N to be correctly classified is given by pN(ρ)=pN(0)exp(ρN), where pN(0) is the classification accuracy when no risk of mistracking or dynamic change is considered. We observed that for ρ=0.1, i.e. for a mean time of correct tracking/absence of dynamic change of 10 frames, the conditional probability of accurate classification for N=5 or N=10 converged to very similar values, and that for higher rate such as ρ=0.2, the conditional accuracy became greater for very short tracklets (N=5). In other words, when the rate of mistracking or dynamics change is quite high, which is likely the case in our experimental dataseet (low SNR and high density of particles), it is more accurate to consider very short tracklets.

Second, to assess the robustness of our classification method with respect to the density of tracked particles, we performed a novel set of simulations where we simulated n=1000 confined tracklets (N=5 frames) and generated the corresponding synthetic movies with a mixed Poisson-Gaussian noise model (see Material and Methods) for 3 different spots’ densities (d=0.039, 0.16 and 0.63 spots/μm2, the measured density being <0.5 spots/μm2 in most experiments). For different percentage of confined tracklets (10 % confined / 90% Brownian ; 50 % confined / 50% Brownian and 100 % confined) we observed that the classification accuracy was robust to increased densities of spots, which validated the robustness of our algorithm.

Third, the noise (background noise, fluctuating intensity of particles…) within experimental movies could, indeed, affects the accuracy of tracklets classification because of the potential mis-localization of receptors’ spots (the localization accuracy depends on the SNR as discussed in Ober, R. J., Ram, S., & Ward, E. S. (2004). Localization accuracy in single-molecule microscopy. Biophysical journal, 86(2), 1185-1200.). To assess the robustness of our classification method to noise, we performed a third set of simulations where we simulated confined tracklets and generated the corresponding synthetic movies for decreasing SNR. We observed that our method remained very accurate for SNR6, before decreasing due to spots’ mis-detection and localization. In our experimental dataset, we measured a mean SNR10, validating the robustness of our statistical classifier.

Reviewer #3 (Recommendations for the authors):

Investigating the dynamic interaction of GPCRs on the surface of cells is of great importance for improving our understanding of signaling across the plasma membrane. Momboisse et al. use TIRF based single-molecule imaging to track the movement of individual CCR5 molecules under different conditions. They use a different tracking statistics method that enables the analysis of motion changes. They show that treating cells with agonists results in more restricted movements and clustering of receptors as well as receptor endocytosis. Responsible for these effects is the interaction with β-arrestin. Inverse agonist have the opposite effect while binding of the HIV-1 envelope glycoprotein gp120 shows agonist-like properties.

Important for the immobilization of the receptor is its dimerization (and to a small percentage formation of higher oligomers).

The reported results are a nice example of the quantitative analysis of the movement of cell surface receptors and will be of interest for the analysis of other GPCRs and cell surface receptors as well.

1) Heterodimerization with other GPCRs (e.g. CCR2) are documented. Can it be excluded that for example the percentage of receptors showing restricted movement in non stimulated cells is due to such oligomerizations? CCR2 or other receptors would not be fluorescence labeled and could therefore not be detected.

Any interaction with a partner is likely to influence the movement of the receptor. It is therefore not excluded that receptor heterodmerization can slow down CCR5 motion. In our study, we used HEK293 cells, which do not express CCR2 at their surface (no functional response upon activation, doi: 10.1189/jlb.0802415; doi: 10.1126/scisignal.aai8529). The influence of CCR5/CCR2 heterodimerization is not relevant here.

2) The clustering of arrestin has been documented and discussed in the two following manuscripts:

1) Coordinate-based co-localization-mediated analysis of arrestin clustering upon stimulation of the C-C chemokine receptor 5 with RANTES/CCL5 analogues.

Tarancón Díez L, Bönsch C, Malkusch S, Truan Z, Munteanu M, Heilemann M, Hartley O, Endesfelder U, Fürstenberg A. Histochem Cell Biol. 2014 Jul;142(1):69-77. doi: 10.1007/s00418-014-1206-1. Epub 2014 Mar 13. PMID: 24623038

2) Quantitative morphological analysis of arrestin2 clustering upon G protein-coupled receptor stimulation by super-resolution microscopy.

Truan Z, Tarancón Díez L, Bönsch C, Malkusch S, Endesfelder U, Munteanu M, Hartley O, Heilemann M, Fürstenberg A. J Struct Biol. 2013 Nov;184(2):329-34. doi: 10.1016/j.jsb.2013.09.019. Epub 2013 Sep 30. PMID: 24091038

These should be cited and the results being included in the discussion.

We have added the two references and briefly discussed them page 28.

Indeed, a novelty of our article is that even though arrestins can contribute to its own clustering, this is not obverved when the receptor cannot dimerize. The clustering of arrestins would therefore be directly dependent on the structural properties of the receptor. This clustering and the resulting immobilization lead to the optimization of receptor endocytosis.

https://doi.org/10.7554/eLife.76281.sa2

Article and author information

Author details

  1. Fanny Momboisse

    Institut Pasteur, Université Paris Cité, CNRS UMR3569, Virus and Immunity Unit, Paris, France
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Validation, Writing - original draft
    Competing interests
    The authors declare that no competing interests exist
  2. Giacomo Nardi

    Institut Pasteur, Université Paris Cité, CNRS UMR3691, BioImage Analysis Unit, Paris, France
    Contribution
    Conceptualization, Formal analysis, Methodology, Software, Validation, Writing - original draft
    Competing interests
    No competing interests declared
  3. Philippe Colin

    Infinity, Université de Toulouse, CNRS, INSERM, Toulouse, France
    Contribution
    Formal analysis, Investigation, Validation
    Competing interests
    The authors declare that no competing interests exist
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5552-8908
  4. Melanie Hery

    Institut Pasteur, Université Paris Cité, CNRS UMR3569, Virus and Immunity Unit, Paris, France
    Contribution
    Validation
    Competing interests
    The authors declare that no competing interests exist
  5. Nelia Cordeiro

    Institut Pasteur, Université Paris Cité, CNRS UMR3569, Virus and Immunity Unit, Paris, France
    Contribution
    Validation
    Competing interests
    The authors declare that no competing interests exist
  6. Simon Blachier

    Institut Pasteur, Université Paris Cité, Dynamics of Host-Pathogen Interactions Unit, Paris, France
    Contribution
    Validation
    Competing interests
    The authors declare that no competing interests exist
  7. Olivier Schwartz

    Institut Pasteur, Université Paris Cité, CNRS UMR3569, Virus and Immunity Unit, Paris, France
    Contribution
    Funding acquisition, Resources, Supervision
    Competing interests
    The authors declare that no competing interests exist
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0729-1475
  8. Fernando Arenzana-Seisdedos

    Institut Pasteur, Université Paris Cité, INSERM U1108, Viral Pathogenesis Unit, Paris, France
    Contribution
    Funding acquisition, Resources, Supervision
    Competing interests
    The authors declare that no competing interests exist
  9. Nathalie Sauvonnet

    Institut Pasteur, Université Paris Cité, Group Intracellular Trafficking and Tissue Homeostasis, Paris, France
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Resources, Writing - original draft
    Competing interests
    The authors declare that no competing interests exist
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0306-3376
  10. Jean-Christophe Olivo-Marin

    Institut Pasteur, Université Paris Cité, CNRS UMR3691, BioImage Analysis Unit, Paris, France
    Contribution
    Funding acquisition, Resources, Supervision
    Competing interests
    No competing interests declared
  11. Bernard Lagane

    Infinity, Université de Toulouse, CNRS, INSERM, Toulouse, France
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Resources, Supervision, Writing - original draft
    Competing interests
    The authors declare that no competing interests exist
  12. Thibault Lagache

    Institut Pasteur, Université Paris Cité, CNRS UMR3691, BioImage Analysis Unit, Paris, France
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Software, Writing - original draft
    For correspondence
    thibault.lagache@pasteur.fr
    Competing interests
    The authors declare that no competing interests exist
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7033-4677
  13. Anne Brelot

    Institut Pasteur, Université Paris Cité, CNRS UMR3569, Virus and Immunity Unit, Paris, France
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing - original draft
    For correspondence
    anne.brelot@pasteur.fr
    Competing interests
    The authors declare that no competing interests exist
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8095-4909

Funding

Agence Nationale de la Recherche (ANR-10-LABX-62-IBEID post-doctoral fellowship)

  • Fanny Momboisse

Agence Nationale de la Recherche (ANR-16-CONV-0005-INCEPTION Post doctoral fellowship)

  • Giacomo Nardi

Agence Nationale de Recherches sur le Sida et les Hépatites Virales (Post-doctoral fellowship)

  • Philippe Colin

Agence Nationale de Recherches sur le Sida et les Hépatites Virales

  • Olivier Schwartz
  • Bernard Lagane

Agence Nationale de la Recherche (ANR-10-LABX-62-IBEID)

  • Olivier Schwartz
  • Fernando Arenzana-Seisdedos
  • Nathalie Sauvonnet
  • Jean-Christophe Olivo-Marin
  • Bernard Lagane
  • Thibault Lagache
  • Anne Brelot

Agence Nationale de la Recherche (ANR-10-INBS-04)

  • Jean-Christophe Olivo-Marin

Agence Nationale de Recherches sur le Sida et les Hépatites Virales (ANRS-AP19-1)

  • Anne Brelot

Agence Nationale de la Recherche (ANR-16-CONV-0005-INCEPTION)

  • Jean-Christophe Olivo-Marin

Agence Nationale de la Recherche (ANR-21-CE44-0030-GET-REDI)

  • Thibault Lagache
  • Anne Brelot

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful to Françoise Baleux (Institut Pasteur), Stefano Marullo (Institut Cochin) and Franck Perez (Institut Curie) for the gifts of chemokines and plasmids. We acknowledge Oliver Hartley (University of Genova) and the Programme EVA Centre for AIDS Reagents for the chemokine derivative PSC-RANTES. We acknowledge Stéphane Petres from the Production and Purification of Recombinant Proteins (PPRP) platform (C2RT, Institut Pasteur) for sCD4 production. We thank Audrey Salles from the Photonic BioImaging (PBI) platform (Imagopole) of Institut Pasteur for microscope maintenance and technical help. We thank Vannary Meas-Yedid Hardy (Institut Pasteur), Stéphane Dallongeville (BioImage Analysis Unit, Institut Pasteur), the Image Analysis Hub (C2RT, Institut Pasteur), Gael Moneron (Synapse and Circuit Dynamics), and VizionR (Paris) for help with the image and data analysis. This work was supported by grants from Agence National de Recherche sur le SIDA et les hepatitis virales (ANRS), the French Government’s Investissement d’Avenir program, Laboratoire d’excellence “Integrative Biology of Emerging Infectious Diseases’ (grant ANR-10-LABX-62-IBEID), INCEPTION (ANR-16-CONV-0005) and GET-REDI (ANR-21-CE44-0030). UTechS PBI is part of the France–BioImaging infrastructure network (FBI) supported by the French National Research Agency (ANR-10-INBS-04; Investments for the Future), and acknowledges support from Institut Pasteur, ANR/FBI, the Région Ile-de-France (program ‘Domaine d'Intérêt Majeur-Malinf’ and DIM1HEALTH), and the French Government Investissement d'Avenir Programme—Laboratoire d'Excellence ‘Integrative Biology of Emerging Infectious Diseases’ (ANR-10-LABX-62-IBEID) for the use of ELYRA PS1 LSM780 and ELYRA7 microscopes. FM was the recipient of ANR-10-LABX-62-IBEID fellowship, GN of INCEPTION (ANR-16-CONV-0005) fellowship and PC of an ANRS fellowship.

Senior and Reviewing Editor

  1. Volker Dötsch, Goethe University, Germany

Reviewer

  1. Volker Dötsch, Goethe University, Germany

Publication history

  1. Received: December 10, 2021
  2. Preprint posted: December 20, 2021 (view preprint)
  3. Accepted: June 17, 2022
  4. Version of Record published: July 22, 2022 (version 1)
  5. Version of Record updated: July 25, 2022 (version 2)

Copyright

© 2022, Momboisse et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 392
    Page views
  • 125
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Fanny Momboisse
  2. Giacomo Nardi
  3. Philippe Colin
  4. Melanie Hery
  5. Nelia Cordeiro
  6. Simon Blachier
  7. Olivier Schwartz
  8. Fernando Arenzana-Seisdedos
  9. Nathalie Sauvonnet
  10. Jean-Christophe Olivo-Marin
  11. Bernard Lagane
  12. Thibault Lagache
  13. Anne Brelot
(2022)
Tracking receptor motions at the plasma membrane reveals distinct effects of ligands on CCR5 dynamics depending on its dimerization status
eLife 11:e76281.
https://doi.org/10.7554/eLife.76281

Further reading

    1. Cell Biology
    Willa Wen-You Yim et al.
    Tools and Resources Updated

    Monitoring autophagic flux is necessary for most autophagy studies. The autophagic flux assays currently available for mammalian cells are generally complicated and do not yield highly quantitative results. Yeast autophagic flux is routinely monitored with the green fluorescence protein (GFP)-based processing assay, whereby the amount of GFP proteolytically released from GFP-containing reporters (e.g. GFP-Atg8), detected by immunoblotting, reflects autophagic flux. However, this simple and effective assay is typically inapplicable to mammalian cells because GFP is efficiently degraded in lysosomes while the more proteolytically resistant red fluorescent protein (RFP) accumulates in lysosomes under basal conditions. Here, we report a HaloTag (Halo)-based reporter processing assay to monitor mammalian autophagic flux. We found that Halo is sensitive to lysosomal proteolysis but becomes resistant upon ligand binding. When delivered into lysosomes by autophagy, pulse-labeled Halo-based reporters (e.g. Halo-LC3 and Halo-GFP) are proteolytically processed to generate Haloligand when delivered into lysosomes by autophagy. Hence, the amount of free Haloligand detected by immunoblotting or in-gel fluorescence imaging reflects autophagic flux. We demonstrate the applications of this assay by monitoring the autophagy pathways, macroautophagy, selective autophagy, and even bulk nonselective autophagy. With the Halo-based processing assay, mammalian autophagic flux and lysosome-mediated degradation can be monitored easily and precisely.

    1. Cell Biology
    2. Computational and Systems Biology
    Théo Aspert et al.
    Tools and Resources

    Automating the extraction of meaningful temporal information from sequences of microscopy images represents a major challenge to characterize dynamical biological processes. So far, strong limitations in the ability to quantitatively analyze single-cell trajectories have prevented large-scale investigations to assess the dynamics of entry into replicative senescence in yeast. Here, we have developed DetecDiv, a microfluidic-based image acquisition platform combined with deep learning-based software for high-throughput single-cell division tracking. We show that DetecDiv can automatically reconstruct cellular replicative lifespans with high accuracy and performs similarly with various imaging platforms and geometries of microfluidic traps. In addition, this methodology provides comprehensive temporal cellular metrics using time-series classification and image semantic segmentation. Last, we show that this method can be further applied to automatically quantify the dynamics of cellular adaptation and real-time cell survival upon exposure to environmental stress. Hence, this methodology provides an all-in-one toolbox for high-throughput phenotyping for cell cycle, stress response, and replicative lifespan assays.