CCR7 signalosomes are preassembled on tips of lymphocyte microvilli in proximity to LFA-1

Abstract

Leukocyte microvilli are elastic actin-rich projections implicated in rapid sensing and penetration across glycocalyx barriers. Microvilli are critical for the capture and arrest of flowing lymphocytes by high endothelial venules, the main lymph node portal vessels. T lymphocyte arrest involves subsecond activation of the integrin LFA-1 by the G-protein-coupled receptor CCR7 and its endothelial-displayed ligands, the chemokines CCL21 and CCL19. The topographical distribution of CCR7 and of LFA-1 in relation to lymphocyte microvilli has never been elucidated. We applied the recently developed microvillar cartography imaging technique to determine the topographical distribution of CCR7 and LFA-1 with respect to microvilli on peripheral blood T lymphocytes. We found that CCR7 is clustered on the tips of T cell microvilli. The vast majority of LFA-1 molecules were found on the cell body, likely assembled in macroclusters, but a subset of LFA-1, 5% of the total, were found scattered within 20 nm from the CCR7 clusters, implicating these LFA-1 molecules as targets for inside-out activation signals transmitted within a fraction of a second by chemokine-bound CCR7. Indeed, RhoA, the key GTPase involved in rapid LFA-1 affinity triggering by CCR7, was also found to be clustered near CCR7. In addition, we observed that the tyrosine kinase JAK2 controls CCR7-mediated LFA-1 affinity triggering and is also highly enriched on tips of microvilli. We propose that tips of lymphocyte microvilli are novel signalosomes for subsecond CCR7-mediated inside-out signaling to neighboring LFA-1 molecules, a critical checkpoint in LFA-1-mediated lymphocyte arrest on high endothelial venules.

Significance

Leukocyte microvilli are highly elastic actin-rich projections. T lymphocytes enter lymph nodes in search of antigens via venules. Upon encountering specific ligands presented on these venules, the T cells undergo rapid activation mediated by the chemokine receptor CCR7, which switches on their main adhesion receptor LFA-1. CCR7 must encounter its ligands within a subsecond time frame. Because these proteins are embedded inside a thick layer of glycoproteins, CCR7 must be topographically available on the T cell surface. Here, using super-resolution microscopy, we found that CCR7 and its key signaling partners are preassembled on tips of T cell microvilli in proximity to a subset of LFA-1 molecules. We suggest that tips of lymphocyte microvilli are signaling hubs for rapid encounter and transmission of CCR7-mediated signals.

Introduction

Microvilli are submicron elastic cellular membrane protrusions involved in a wide variety of functions, including absorption, secretion, cellular adhesion, and mechanotransduction (1, 2, 3). These protrusions are a subgroup of protrusive actin structures (4) that in nonhematopoietic cells, such as professional sensory neurons, serve to transmit information to different neighboring cells (5). The functions of leukocyte microvilli, on the other hand, have been attributed mainly to leukocyte interactions with blood vessels (6). Leukocyte microvilli were traditionally implicated in the initial capture of circulating leukocytes by lectin endothelial adhesion molecules, selectins, under shear forces (7,8). Recent findings allude to new functions of microvilli in leukocyte-leukocyte communications (9). For instance, T cells localize their T cell receptors (TCRs) on microvilli and seem to use these projections to facilitate T cell scanning of rare antigenic signals presented on dendritic cells and, potentially, on other antigen-presenting cells (10, 11, 12). In addition, leukocyte microvilli are thought to facilitate the accessibility of various receptors found on these projections to counter-receptors masked by the thick glycocalyx layers of different types of cells, including endothelial cells and other leukocytes (13,14). The dense glycocalyx of all mammalian cells is rich in negatively charged proteoglycans that create a barrier for receptor-ligand interactions. Leukocyte microvilli have been implicated in force-driven penetration of this dense barrier (15).

Our earlier studies suggested yet another critical role of microvilli as potential platforms for ultrafast signaling between chemokine receptors and integrins taking place in actively rolling leukocytes (16, 17, 18). Our studies on integrin activation by surface-bound chemokines suggested that various T cell integrins can undergo in situ activation by surface-presented chemokines within subsecond time frames (19,20). We therefore speculated that the specific G-protein-coupled chemokine receptors (GPCRs) specialized in transmitting rapid inside-out activation signals to integrins might be enriched on microvilli. Because α4 integrins are also enriched on these lymphocyte projections (21,22), these GPCRs would be expected to activate α4 integrins on subsecond time frames (19). Nevertheless, it has been unclear how the integrin LFA-1, found to be largely excluded from microvilli (21), also undergoes GPCR-triggered activation within a 0.3–0.5 s time frame (20).

In this study, we have applied microvillar cartography, a technique that combines variable-angle (VA)-total internal reflection fluorescence (TIRF) microscopy and stochastic localization nanoscopy (SLN) (10,12) to decipher the topographical distribution of the canonical T cell GPCR CCR7 on the surface of human peripheral blood T lymphocytes. Our analysis suggests that CCR7 molecules are highly enriched on tips of microvilli, which may facilitate their interaction with endothelial glycocalyx-embedded CCL21 and CCL19 ligands. In an attempt to address the standing question regarding how this GPCR encounters endothelial-displayed ligands, such as CCL21 and CCL19, and transmits rapid signals to near LFA-1, we also determined the location of this integrin near CCR7 molecules. We identified a small subset of LFA-1 that are distributed within T cell microvilli near CCR7. Our data argue that individual microvilli are functional CCR7 hubs in which small groups of LFA-1 undergo in situ activation by signals from proximal CCR7 molecules. Indeed, the CCR7 hubs are highly enriched with clustered RhoA and Janus kinase (JAK) 2, key activators of LFA-1 involved in rapid signaling between chemokine-occupied CCR7 to LFA-1. These are critical for lymphocyte arrest on ICAM-1 and ICAM-2 on the surface of high endothelial venules (HEVs), the key blood vessels used by circulating T cells to enter lymph nodes (23, 24, 25, 26).

Materials and methods

Antibodies and reagents

LEAF-purified anti-human LFA-1 (clone HI111; catalog number 301214), Alexa 647-conjugated anti-human LFA-1 (clone HI111; catalog number 301217), LEAF-purified anti-human CD62L (clone DREG-56; catalog number 304812), Alexa 647-conjugated anti-human CD62L (clone DREG-56; catalog number 304818) were purchased from BioLegend (San Diego, CA). Alexa Fluor 568 (catalog number A20103) and Alexa Fluor 647 (catalog number A20106) were purchased from Thermo Fisher Scientific (Waltham, MA). Alexa Fluor 647 AffiniPure F(ab′)₂ Fragment Goat Anti-Human IgG, Fcγ fragment-specific antibodies (catalog number 109-606-098), R-Phycoerythrin AffiniPure F(ab′)₂ Fragment Goat Anti-Human IgG, Fcγ Fragment Specific (catalog number 109-116-098) and Biotin-SP-conjugated AffiniPure F(ab′)2 Fragment Goat Anti-Human IgG, Fcγ Fragment Specific (catalog number 109-066-098) secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). FM143Fx and polyclonal rabbit anti-RhoA antibodies (catalog number OSR00266W) were purchased from Invitrogen (Waltham, MA). Monoclonal rabbit anti-JAK2 antibodies were purchased from Abcam (catalog number ab108596; Cambridge, UK). Goat anti-rabbit IgG H&L (Alexa Fluor 568) ab175471 was purchased from Abcam. Alexa Fluor 488 dye-labeled streptavidin 10 nm colloidal gold conjugates (catalog number A-32361) were purchased from ThermoFisher (Waltham, MA). fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse antibody was from Sigma-Aldrich (St. Louis, MO). Recombinant human CCL19 (catalog number 300-29B) and recombinant human CXCL12 (catalog number 300-28A) were purchased from PeproTech (East Windsor, NJ). CCL19, from R&D Systems (361-MI-025; Minneapolis, MN) was used for LFA-1 activation experiments. CCL19-and CCL21 Fc fusion proteins were prepared as described (27,28). The monoclonal anti-β2-integrin affinity reporter monoclonal antibody KIM127 was from American Type Culture Collection (Manassas, VA).

Human T cells

T cells were isolated from both male and female healthy volunteer donors. The Weizmann Institute of Science Institutional Review Board, appointed by the president of the Weizmann Institute of Science, has reviewed all the experimental protocols involved in this study in accordance with the Israeli law, National Institutes of Health guidelines and the Common Rule of ethics regarding biomedical and behavioral research involving human subjects in the United States (Title 45 CFR 46). To isolate human peripheral T lymphocytes, whole blood from healthy donors was citrate-anticoagulated, followed by dextran sedimentation and Ficoll (Sigma-Aldrich) gradient separation. To remove B cells, a nylon wool column (Uni-sorb) was first used followed by negative depletion human T lymphocyte enrichment set (IMAg catalog number 557874; BD Biosciences, San Jose, CA), resulting in ∼99% CD3 plus T lymphocytes.

Flow cytometry

For analysis of surface expression of various proteins, T cells were labeled with fluorescent primary antibodies (10 μg/mL), washed, resuspended in fluorescence-activated cell sorting buffer (phosphate buffered saline without Ca²⁺ and Mg²⁺ (PBS^−/−), 1% BSA, 5 mM EDTA, and 0.01% sodium azide), and analyzed on a CytoFLEX S cell sorter (Beckman Coulter, Brea, CA). Data were acquired with the FACSDiva software (Becton Dickinson, Franklin Lakes, NJ). Postacquisition analysis was performed using the FlowJo software (Tree Star, Ashland, OR).

Static adhesion assays

Primary T lymphocytes were suspended at 5 × 10⁶/mL in standard adhesion buffer (PBS 1× plus FBS 10% plus CaCl₂ 1 mM plus MgCl₂ 1 mM). Adhesion assays were performed as previously described (29) on 12-well glass slides coated with human ICAM-1 at 1 μg/mL in PBS. Where indicated, lymphocytes were pretreated with JAK2 inhibitors in dimethyl sulfoxide (DMSO) for 1 h at 37°C. 20 μL of cell suspension were added to the wells and stimulated for 2 min at 37°C with CCL19 at 0.25 μM. After rapid washing, adherent cells were fixed in ice-cold 1.5% glutaraldehyde in PBS, and still images of adherent cells in 0.2-mm² fields were acquired at 20× magnification, NA 0.40, with a charge-coupled device camera (ICD-42B; Ikegami, Tokyo, Japan) connected to an inverted microscope (IX50; Olympus Life Science, Tokyo, Japan) operating in the phase-contrast mode. Image acquisition and computer-assisted enumeration of adherent cells were performed with ImageJ (National Institutes of Health, Bethesda, MD).

Determination of CCL19-stimulated LFA-1 affinity states on T cells

T lymphocytes suspended in standard adhesion buffer (PBS 1× plus FBS 10% plus CaCl₂ 1 mM plus MgCl₂ 1 mM) at 2 × 10⁶/mL were stimulated for 2 min at 37°C with 0.25 μM CCL19 in the presence of 10 μg/mL KIM127 β2-integrin reporter mAb that detects the intermediate-affinity extended state of LFA-1 (20,23,24). After rapid washing, the cells were stained for 30 min with FITC-conjugated secondary antibody at 4°C and analyzed by cytofluorimetric quantification.

Antibody labeling

Unlabeled antibodies were labeled with Alexa Fluor 568 or Alexa Fluor 647 according to need. To this end, antibody molecules in PBS buffer were reacted with the N-hydroxysuccinimide (NHS) ester of the corresponding dye in a 1:10 molar ratio in presence of 0.1 M sodium bicarbonate buffer for 1 h at room temperature in the dark. A micro Bio-Spin column packed with Bio-Gel P-30 (Bio-Rad Laboratories, Hercules, CA) was used to remove the unreacted dye molecules.

Cellular labeling with antibodies

T cells were washed with 5 mM EDTA/PBS for 5 min by centrifugation at 4°C. The cells were incubated in a blocking solution (1% bovine serum albumin (BSA), 5 mM EDTA, 0.05% N₃Na, PBS) on ice for 10 min. For labeling of surface molecules, the cells were treated with fluorescently conjugated primary antibodies to either CCL21-Fc or CCL19-Fc (both at 5 μg/mL), followed by Alexa Fluor 647-conjugated anti-immunoglobulin G (IgG) antibodies, for 20–30 min on ice. Labeling of T cells at 4°C was previously verified not to lead to any artifact; see the section “Labeling at 4°C captures the bona fide resting state” in (12). After washing the cells twice with 5 mM EDTA/PBS by centrifugation at 4°C, they were fixed with a fixation buffer (4% (wt/v) paraformaldehyde, 0.2–0.5% glutaraldehyde, 2% (wt/v) sucrose, 10 mM EGTA, and 1 mM EDTA, PBS) in suspension for 2 h on ice. Fixing the cells in solution prevents activation changes of membrane topography and redistribution of surface molecules potentially encountered by settling live cells on charged agents such as poly-l-lysine (PLL). The fixative was washed twice with PBS by centrifugation. The cell membrane was then stained with FM143Fx (5 μg/mL; Invitrogen) for 30 min on ice. We chose this dye because it is known to label membranes homogeneously (30, 31, 32), which we verified (12). After staining with FM143FX, another fixation step for 30 min on ice was performed, and fixatives were washed twice at 4°C with PBS by centrifugation. Cells were suspended in PBS and were kept at 4°C. These cells were then directly used for super-resolution mapping of the molecule of interest. For intracellular labeling of RhoA and JAK2, immediately after fixation and two washes with PBS as described above, cells were permeabilized with 0.05% saponin and 1% BSA in PBS with RhoA- or JAK2-specific antibodies added at 1:250 dilution overnight at 4°C. Cells were then washed twice with PBS, incubated with Alexa 568 goat anti-rabbit antibodies for 1 h at room temperature, and washed two more times with PBS before imaging.

Sample preparation for super-resolution microscopy

Glass-bottom Petri dishes were cleaned with 1 M NaOH (Fluka) for 40 min, then coated with PLL (0.01%; Sigma-Aldrich) for 40 min and washed with PBS. 100 mL of the labeled cell solution in PBS was placed in the Petri dish, and cells were allowed to settle on the PLL surface for 10 min. The PBS buffer was then exchanged with a freshly prepared “blinking buffer” (50 mM cysteamine, 0.5 mg/mL glucose oxidase, 40 μg/mL catalase, 10% (wt/v) glucose, 93 mM Tris/HCl, PBS buffer (pH 7.5–8.5), all from Sigma Aldrich), and the sample was kept for 30 min before imaging was conducted.

TIRF setup

The microscopy setup was previously described in detail (10,12). Briefly, the setup contained three different lasers sources (405 nm, Toptica iBeam smart 405-S; 532 nm, Cobalt Samba 50 mW; and 642 nm, Toptica iBeam smart 640-S), out of which we used the latter two here. The power of each laser was computer-controlled. An achromatic focusing lens (f = 500 mm; LAO801; CVI Melles Griot, Albuquerque, NM) was employed to focus the expanded laser beams at the back focal plane of the microscope objective lens (UAPON 100XOTIRF; N.A., 1.49; Olympus Life Science). To achieve total internal reflection at the sample, we shifted the position of the focused beam from the center of the objective to its edge. A quad-edge super-resolution laser dichroic beam splitter (Di03-R405/488/532/635-t1-25x36) was utilized to separate the emitted fluorescence from the excitation. Notch filters (NF01-405/488/532/635 StopLine Quad-notch filter and ZET635NF; Semrock, Rochester, NY) were used to block the residual scattered laser light. A single EMCCD camera (iXonEM+ 897 back-illuminated; Andor, Oxford Instruments) was employed in a dual-view mode. Spectrally separated images were projected onto the two halves of the CCD chip. Our setup had a final magnification of 240×, which resulted in an effective pixel size of 66.67 nm.

Reconstruction of three-dimensional cell surfaces and detection of microvilli

Three-dimensional surface reconstruction and detection of microvilli from TIRF images taken at a series of angles were performed as previously described in detail (10,12). Briefly, we recorded TIRF images of cells at a series of angles of incidence under weak illumination of a 532 nm laser. The relative axial distance of each point (δz) was calculated from its intensity relative to the pixel with maximal intensity at a particular angle of incidence (I_max(θ)). To determine the location of the tips of microvilli, an algorithm was developed in MATLAB (The MathWorks, Natick, MA) (12).

SLN

The dyes Alexa Fluor 647 and Alexa Fluor 568 were used for SLN measurements. These dyes can be converted to a dark state upon illumination with a very high intensity laser of the appropriate wavelength. In the presence of a buffer containing β-mercaptoethylamine together with glucose oxidase/catalase (an oxygen-scavenging system (33,34)), a small fraction of the dye molecules may be switched back at any moment in time. Thus, individual molecules appear separately in consecutive recording frames. This phenomenon assists in the detection of the location of a molecule with precision well below the diffraction limit. For all SLN measurements, the laser beam incident angle was 66.8°. For each focal plane, 28,000 frames were recorded in seven series of videos (each containing 4000 frames). A piezo stage (PInano Piezo-Z Slide Scanner; PI, Auburn, MA) was used to record SLN videos at the 0 nm and −400 nm focal planes. The measurement at two different planes ensured that proteins localized on membrane regions situated somewhat further away from the surface, due to the length of microvillar protrusions, would not be overlooked. Given the significant depth of focus and the high laser power used in the super-resolution measurements, focusing on the −400 nm plane enabled close observation of cell body regions and included signals from molecules on microvilli as well. The intensity of the evanescent field decreased because of the penetration depth but was verified to be high enough to detect molecules with equal probability at the 0 nm plane and the −400 nm plane (see details in (12), particularly the “TIRF setup” section and Fig. S1, M–O).

Localization analysis

Detailed molecular localization procedures were provided in (10). A thresholding step was used to identify individual emitters whose pattern was then fitted to a two-dimensional Gaussian function to obtain their x and y coordinates and the widths of their point-spread functions in the x and y directions. The average localization uncertainty in x-y was estimated to be ∼11 nm full width at half maximum (FWHM), whereas the image resolution, calculated based on Fourier ring correlation analysis of images (35), was ∼30 nm. (Note that the resolution depends both on the localization uncertainty and on the density of labeled molecules.) It should be noted here that the number of localizations in the −400 nm plane was always smaller than that in the 0 nm plane because 28,000 super-resolution frames were first recorded from the 0 nm plane, so that during the latter −400 nm plane recordings, multiple molecules had already photobleached. We report the average number of localizations observed for each molecule studied in Table S1. As expected, when the order of acquisition of data from the two planes was reversed, the relative number of molecules identified in the two planes was inverted. This had no effect on our findings because we did not count the absolute number of molecules, but rather, we followed their spatial distribution. Therefore, fluorophore blinking also did not affect our conclusions.

Drift correction

Because a stack of SLN videos was recorded in each experiment, a correction for microscope drift during the measurement was required. To this end, TIRF reference images of the cell membrane before and after each SLN video were recorded under weak illumination of a 532 nm laser (10–20 mW). Next, a two-dimensional cross-correlation of each reference TIRF image with the previous reference image was performed, and the identified drift was used to register images by shifting them as necessary. The drift typically found within one SLN video was not larger than 35 nm. We estimated that the accuracy of the registration over time (i.e., the error introduced after correction by the above method) was as good as 4 nm.

To register the 532 and 647 nm channels in experiments in which the membrane was stained, we took advantage of the fact that some leak occurred from the 532 nm channel into the 647 nm channel (this leak was not present when we measured the true 647 nm images because the 532 nm laser was then off). The leaked image in the 647 nm channel was cross-correlated with the original 532 nm image, and this cross-correlation was used to determine the shift between the two images and correct for it. We estimated an excellent registration accuracy of ∼5 nm, based on this method. For dual-color SLN, fluorescent nanodiamonds were used as fiducial points for drift correction with an accuracy of ∼12 nm. Importantly, after drift correction, blinking artifacts could only generate artificial clusters over length scales compared with the spread of single-molecule localizations (12).

Quantitative analysis of molecular distribution

To analyze the distribution of each membrane protein, the membrane area of each cell was segmented into microvilli regions and nonmicrovilli or cell body regions using the analysis described in our previous work (10). The percentage of molecules on microvillar regions and on cell body regions of each cell was then calculated. Next, the cumulative fractional increase of the number of molecules of a specific protein on each cell as a function of the distance from the tip region of each microvillus was determined. To this end, the tip region was defined as the region that is not more than 20 nm from the microvillar tip, which is the pixel of the minimal δz-value. The “boundary” function of MATLAB was utilized to encapsulate this tip region with a geometric shape that could take different forms, depending on the shape of the specific microvillus and on where it was situated with respect to the surface. Concentric closed curves of a similar shape and increasing size were plotted, and the number of molecules in each concentric structure was calculated to obtain the cumulative fractional increase. This value was normalized by the cumulative fractional increase of the area, to obtain the δCount/δArea plot. The more a protein was localized to the microvillar region, the steeper was the slope of this plot. Finally, the cumulative increase in the percentage of molecules as a function of distance from the tip regions was plotted.

Colocalization probability analysis for dual-color super-resolution images

Dual-color super-resolution microscopic experiments were performed to find the degree of colocalization of protein molecules in pairs on resting human T cells with nanometer resolution. An analysis scheme was developed to obtain the fraction of molecules that are colocalized on microvilli (12). To this end, a distance-dependent colocalization probability (CP) was introduced, defined as the probability that a molecule of one type will have at least one partner of the other type within a specified interaction distance. Thus, the CP for molecules i and j within distance R is given by N_ij(R)/N_i, where N_i is the total number of detected points of molecule i, and N_ij(R) is the number of points of molecule i that have at least one point of molecule j within R.

Density-based cluster analysis

The single-molecule localization maps of different proteins were analyzed using a custom-written MATLAB code based on the algrorithm density-based spatial clustering of applications with noise (DBSCAN) (36,37). Each localization map was segmented into separate clusters and noise points using the “dbscan” function of MATLAB. The boundary of each cluster was defined by the “boundary” function of MATLAB. The “centroid” of each cluster was then found. Finally, the distance of the centroid of each cluster of LFA-1 from its nearest microvillar tip was determined.

Scanning electron microscopy

T cells were washed with 5 mM EDTA/PBS for 5 min by centrifugation at 4°C. The cells were incubated in a blocking solution (1% BSA, 5 mM EDTA, 0.05% N₃Na, PBS) on ice for 10 min. Cells were treated with CCL19-Fc (5 μg/mL) followed by Biotin-SP-conjugated anti-human IgG, Fcγ antibodies (1:200), for 30 min on ice. After washing the cells twice with 5 mM EDTA/PBS by centrifugation at 4°C, cells were fixed with a fixation buffer 4% (wt/v) paraformaldehyde, 0.2–0.5% glutaraldehyde, 2% (wt/v) sucrose, 10 mM EGTA, and 1 mM EDTA, in PBS for 1 h on ice. The fixative was washed twice with PBS by centrifugation. Another fixation step for 30 min on ice was performed, and fixatives were washed twice at 4°C with PBS. Finally, 10 nm colloidal gold streptavidin conjugates (30 μg/mL) were added for 30 min at room temperature and washed in excess volume of PBS. Silicon wafers were plasma cleaned (EVACTRON Combiclean Decontaminator; XEI Scientific, Redwood City, CA) and incubated in 1 mg/mL poly-l-lysine (PLL) (catalog number P1524-25MG; Sigma-Aldrich) for 15 min. Cells were allowed to adhere to the PLL-coated silicon plates overnight at 4°C. Cells were then dehydrated in a graded series of ethanol concentrations (30, 50, 70, 90, 95% ethanol, two rinses for 10 min for each step), followed by three rinses with 100% ethanol for 20 min. Samples were then dried in a CPD 030 (Bal-Tec, Balzers, Liechtenstein) critical point drying machine. The bottom of the silicon plates was then adhered to scanning electron microscopy stubs using a double-sided carbon tape. Observation was conducted in a GeminiSEM 500 scanning electron microscope (Zeiss, Oberkochen, Germany) using an in-lens backscattered electron detector as well as a mixed mode of in-lens backscattered electron and secondary electron detectors at an acceleration voltage of 1–1.5 kV using a 20 μm aperture as previously described (10,12).

Quantification and statistical analysis

Statistical details of experiments can be found in the figure legends. Standard error of the mean (SEM) calculation was performed using MATLAB.

Data and code availability

Distribution of raw data supporting the study is shown in Fig. 2 A, and any other raw data and MATLAB code used for analysis are available from the corresponding authors upon a reasonable request.

Figure 2.

Quantitative measures for the distribution of CCR7 and LFA-1 proteins on the T cell surface at the plane of the microvilli tips (0 nm plane). (A) Percentage of CCR7 and LFA-1 molecules on microvillar regions of the membrane. The values for individual cells are shown as dots in the plot. (B) Cumulative increase in the percentage of CCR7 (blue) and LFA-1 (red) molecules as a function of the distance from the tip region of the microvilli. (C) Cumulative increase in the fraction of total molecules on each cell as a function of the distance from the tip region of the microvilli, normalized by the cumulative increase in the fraction of area (δCount/δArea) as a function of distance from the tip region of the microvilli. Error bars represent SEMs. The tip region of each microvillus was defined as the region that is less than 20 nm distance from the microvillar tip, defined as the pixel of minimal δz-value. The analysis is based on 10 resting T cells imaged for CCR7 and 9 resting T cells imaged for LFA-1.

Results

CCR7 is highly enriched on the tips of T cell microvilli, whereas LFA-1 is primarily macroclustered on the T cell body

We previously demonstrated that microvillar cartography can reliably map the distribution of functional surface proteins, including TCRs and L-selectin, with respect to microvilli (10). This approach also allowed us to further map the distribution and proximity between the TCR and its various signaling partners (12). We therefore chose this method to determine the spatial localization and distribution of the CCR7 chemokine receptor and of the integrin LFA-1 on the surface of resting, blood-derived human T lymphocytes.

To label CCR7, we used a common approach to stain this cell surface GPCR with a functional ligand-IgG probe, composed of CCL19 fused with an Fc moiety (27). We first confirmed the staining specificity of CCL19-Fc. T cells were incubated with the CCL19 fusion protein followed by fluorescently labeled secondary anti-IgG antibodies, and full blockade of this staining was achieved by preincubating the T cells with soluble CCL19, but not with the CXCR4 ligand CXCL12 (Fig. S1). In contrast to CCL19-Fc, CC21-Fc fusion proteins bound T cells with poor CCR7 specificity (data not shown). We then similarly stained the T cells with the CCL19 fusion protein and secondary antibodies, fixed the lymphocytes, and stained them with the membrane dye FM143Fx to image the entire T cell surface. These differentially labeled cells were then imaged using combined VA-TIRF and SLN microscopy. Strikingly, nearly all CCR7 molecules were found as clusters on tips of T cell microvilli (Fig. 1 A; Fig. S2 A). Extensive CCR7 clustering was also confirmed on the tips of T cell microvilli by immunogold labeling and scanning electron microscopy (Fig. S3), although in this experiment, only a fraction of the T cell microvilli contained CCR7 clusters, possibly due to impaired microvillar integrity resulting from sample processing. In agreement with previous results (6), L-selectin was found evenly distributed on the entire microvillar surface (Fig. S4). These results suggest that CCR7 is organized in a specialized microvillar compartment, which likely facilitates T cell encounter of the HEV-presented CCR7 ligands CCL21 and CCL19 during L-selectin-mediated rolling.

Figure 1.

Mapping the spatial distribution of CCR7 and LFA-1 in relation to microvilli. Localization maps of CCR7 (A) and LFA-1 (B) are shown. Positions of protein molecules obtained from SLN (black dots) at the 0 nm plane on resting human T cells are superimposed on membrane topography maps obtained by VA-TIRFM. The color bars represent distance from the glass in nanometers. Scale bar, 1 μm. The corresponding images for the −400 nm planes are shown in Fig. S4, A and B.

Next, we used the same approach with a fluorescently labeled anti-αL mAb that binds the I domain of the integrin headpiece in its closed conformation on both bent and extended LFA-1 molecules (38). Accordingly, 99% of the total surface LFA-1 was labeled with this mAb (data not shown). Super-resolution analysis of the mAb-stained LFA-1 on the surface of T cells revealed two major subsets. One large subset comprised LFA-1 macroclusters excluded from microvilli and enriched on the T cell body, whereas scattered LFA-1 molecules, possibly individual heterodimers or oligomers, were randomly distributed on the T cell body as well as on the entire surface of the T cell microvilli (Fig. 1 B; Fig. S2 B). As can be appreciated (Fig. S6), LFA-1 distribution on microvilli and nonmicrovillar compartments was comparable in the absence or presence of Ca²⁺ and Mg²⁺ during the staining process. Interestingly, under both conditions, a linear correlation was found between the size of LFA-1 macroclusters (size > 20,000 nm²) and their distance from microvillar tips (Pearson’s R as high as 0.75) (Fig. S7). A potential functional role for differently clustered LFA-1 subsets remains to be determined. It should be also noted that, in a drift-corrected super-resolved image, the artifactual appearance of nanoclusters due to reblinking may occur only on length scales compared with the spread of single-molecule localization events (39, 40, 41); in other words repeated events of localization of the same molecule may appear approximately within a circle of the radius of the localization accuracy, which is ∼15 nm in our experiments (Fig. S5 in (12)). Notably, the sizes of the macroclusters of LFA-1 (>20,000 nm²) are significantly larger than that. Moreover, the T cell labeling with CCR7 and LFA-1-specific mAbs was performed at low temperature (4°C). Thus, the macroclusters of LFA-1 and of CCR7 were unlikely to be the outcome of antibody-mediated integrin ligation.

A small fraction of LFA-1 resides in close proximity to CCR7 clusters on tips of microvilli

Further quantitative analysis revealed that 98% of CCR7 but only 20% of LFA-1 resided on microvilli (Fig. 2 A; Fig. S4, C–E). Whereas over half of the total surface CCR7 (55%) were localized on tips of microvilli, only 5% of the total surface LFA-1 localized to these tip regions (Fig. 2, B and C). Notably, CCR7 was clearly distinguishable in this respect from any other surface molecules previously studied by us using microvillar cartography (10,12), including TCRs, which are enriched on microvilli yet not necessarily on their tips (Fig. S5).

We next used a method recently introduced (12) to determine the fraction of CCR7 and LFA-1 molecules that are colocalized on microvilli. In particular, we defined the distance-dependent CP as the probability that a molecule of one type will have at least one partner of the other type within a specified distance. As expected, the highest CP-values were found for CCR7 and L-selectin (Fig. 3). Surprisingly, high CP-values were also detected for CCR7 and LFA-1 (Fig. 3). Interestingly, although both TCR and L-selectin molecules are expressed on microvilli, their CP-values were in fact lower than that of the CCR7/LFA-1 pair, suggesting that these molecules reside in more separated compartments within individual microvilli. Taken together, CCR7 is exclusively enriched on the microvilli, with over half of these chemokine receptors on the tips of microvilli, and this subset resides near a subset of LFA-1 molecules that comprises ∼5% of the total surface LFA-1.

Figure 3.

Colocalization probability (CP) analysis reveals significant proximity of CCR7 to L-selectin and a subset of LFA-1. (A) A super-resolution image of a resting human T cell labeled with Alexa Fluor-568-conjugated HI111 antibodies (shown in green), which stain LFA-1. (B) Super-resolution image of the same cell labeled with Alexa Fluor 647-anti-IgG antibodies (shown in red), which stain CCL19-Fc-bound CCR7. (C) Same as (B), but CCL19-Fc-IgG-Alexa-647-tagged CCR7 molecules that have at least one LFA-1 molecule within a radius of 30 nm are marked yellow. The number of these molecules divided by the total number of CCR7 molecules is the CP-value for this pair at 30 nm. Dual-color overlaid colocalization maps of these molecules are presented in Fig. S8 i. (D) Super-resolution image of a resting human T cell labeled with Alexa Fluor-568-conjugated anti-CD62L antibodies (green), which stain L-selectin. (E) Super-resolution image of the same cell labeled with Alexa Fluor 647-Fc antibodies (red), which stain CCL19-Fc bound with CCR7. (F) Same as in (E), but CCR7 molecules that have at least one L-selectin molecule within a radius of 30 nm are marked in yellow. The number of these molecules divided by the total number of CCR7 molecules is the CP-value for this pair at 30 nm. Scale bars, 0.5 μm. Dual-color overlaid colocalization maps of these molecules are presented in Fig. S8 ii. (G) CP as a function of distance for the pairs of molecules indicated in the legend, obtained from the analysis of five double-labeled cells for each pair. Error bars represents SEMs.

The LFA-1 inside-out effector RhoA resides in close proximity to CCR7 on tips of microvilli

Because CCR7 can signal within a fraction of a second to LFA-1 via a signalosome consisting of Gi-proteins and RhoA (23,24), we next tested the distribution of this key GTPase on microvillar regions of T cells. Notably, in a major fraction of T cells, RhoA was found in microclusters highly enriched on the tips of T cell microvilli (Fig. 4, A and B). Strikingly, the RhoA clusters were highly colocalized with CCR7 (Fig. 4, C–E) and the distance-dependent CP of RhoA and CCR7 reached its maximal value at a shorter distance compared with all microvillar proteins studied here (Fig. 4 F).

Figure 4.

RhoA is enriched on microvilli in proximity to CCR7. (A) Positions of RhoA molecules (black dots) at the 0 nm plane on resting human T cells are superimposed on membrane topography maps. The color bars represent distance from the glass in nanometers (nm). Scale bars, 0.5 μm. (B) Cumulative increase in the fraction of total RhoA molecules on each cell as a function of the distance from the tip region of the microvilli, determined as outlined in Fig. 2C, obtained from the analysis of six cells. (C) Super-resolution image of a resting human T cell labeled with Alexa Fluor 647-anti-IgG antibody (shown in red), which stains CCL19-Fc-bound CCR7. (D) Super-resolution image of the same cell labeled with Alexa Fluor-568-conjugated anti-rabbit antibody (green), which stains anti-human RhoA-polyclonal-rabbit-IgG-antibody-tagged RhoA. (E) Same as (C), but CCL19-Fc-IgG-Alexa-647-tagged CCR7 molecules that have at least one RhoA molecule within a radius of 30 nm are marked in yellow. The number of these molecules divided by the total number of CCR7 molecules is the CP-value for this pair at 30 nm. Dual-color overlayed colocalization maps of these molecules are presented in Fig. S8iii. (F) CP as a function of distance for the CCR7 RhoA pair that was obtained from the analysis of five double-labeled cells. Error bars represent SEMs.

The JAK2 kinase controls LFA-1 inside-out activation by CCR7 signaling and is also enriched on tips of microvilli

JAK2 and 3 were previously implicated in CXCL12-triggered high LFA-1 affinity (29). We therefore hypothesized that JAK2 may be also involved in CCR7-mediated LFA-1 affinity triggering and firm adhesion to ICAM-1. Importantly, a global JAK blocker, AG490, a cell-permeable peptide with a specific JAK2 inhibitory activity (29), significantly inhibited both CCL19-triggered T cell adhesion to ICAM-1 as well as CCL19-induced transition of LFA-1 to the intermediate-affinity LFA-1 conformation critical for firm T cell adhesion to ICAM-1 (20,24) (Fig. 5, A and B). Similar to RhoA, JAK2 was also highly enriched on T cell microvilli (Fig. 5 C). Taken together, our results identify, to our knowledge, a new signaling hub, a CCR7 signalosome, consisting of RhoA and JAK2 and located on tips of microvilli.

FIGURE 5.

JAK2 mediates rapid CCR7 signaling to LFA-1 and is enriched on tips of T cell microvilli. (A) Effect of JAK inhibition on LFA-1-mediated T cell adhesion to ICAM-1 in a static adhesion assay. Lymphocytes were pretreated at 37°C for 1 h with DMSO (vehicle), the JAK inhibitor AG490 (100 μM), the cell-permeable JAK2-specific peptide blocker P1-TKIP (40 μM), or its Penetratin-1 control P1 peptide (40 μM) and stimulated with 0.25 μM CCL19 while settled on ICAM-1-coated substrates. Significance was calculated by two-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. ^∗∗∗∗p < 0.0001 vs. P1 peptide or vehicle. n = 4. Error bars show SDs. (B) Effect of JAK inhibition on LFA-1 affinity triggering by CCL19. Lymphocytes were pretreated and stimulated with CCL19 as described in (A). LFA-1 conformational activation was detected with the KIM127 mAb specific for the extended intermediate-affinity state of the heterodimer. n = 4. Error bars show SDs. Significance was calculated by two-way ANOVA followed by Tukey’s post hoc test. ^∗∗∗∗p < 0.0001 vs. P1 peptide or vehicle. n = 4. (C) Positions of JAK2 molecules (black dots) at the 0 nm plane on resting human T cells are superimposed on membrane topography maps. The color bars represent distance from the glass in nanometers (nm). Scale bars, 0.5 μm. (D) Cumulative increase of the fraction of total JAK2 molecules on each cell as a function of the distance from the tip region of the microvilli determined as outlined in Fig. 2C and obtained from the analysis of seven cells. Error bars show SEMs.

Discussion

Scanning and transmission electron microscopy have revealed that resting T cell microvilli are 70–150 nm in diameter and from 100 nm to several μm in length (42). These microvilli are considered essential for leukocyte recognition of selectins or selectin ligands under shear stress conditions because of their dimensions and highly elastic properties (1,6). Our recent results also established microvilli as key signaling hubs in extravascular spaces devoid of shear stress (10,12). Using super-resolution microscopy of immunolabeled surface molecules on the surface of fixed T cells, we and others found that the TCR complex and its proximal signaling molecules and adaptors are preassembled on microvilli before activation by antigenic peptides (10, 11, 12,43).

We showed here that essentially all (98%) of the T cell CCR7 molecules reside within highly compartmentalized regions on microvilli (see Fig. S9 for a schematic description). This finding allowed us to address a long-standing question in the field of chemokine-mediated LFA-1-dependent lymphocyte arrest on the key endothelial ligand ICAM-1 expressed by HEVs: how do endothelium-bound CCL21 and CCL19 and their GPCR CCR7 rapidly transmit signals via Gi-protein machineries to the integrin LFA-1? LFA-1 has been shown to reside primarily on the cell body of leukemic cells (21). Our finding that the majority of T cell LFA-1 is excluded from microvilli is in line with these earlier findings. It has been postulated that during lymphocyte rolling on HEVs, which is mediated by L-selectin enriched on microvilli and interacting with glycoprotein surface ligands (44), the collapse of these microvilli (45, 46, 47) brings that pool of LFA-1 to the vicinity of ICAM-1 on endothelial cells. This model has been challenged, however, by the extremely fast kinetics of CCR7-mediated LFA-1 activation, which was shown by several labs to take place within a time frame of 0.1–0.3 s, much faster than microvilli collapse. Our new findings suggest that a small fraction (20%) of LFA-1 resides on microvilli and that about one-fourth of this pool (5% of the total LFA-1) resides in very close proximity to CCR7. We thus postulate that it is this subset of LFA-1 that is targeted by the ultrafast signals transmitted by CCR7 once it has been occupied with CCL21 and CCL19 during T cell rolling on HEVs.

T cells captured by and rolling on HEV-expressed L-selectin ligands were proposed to encounter CCR7 chemokines like CCL21 and CCL19 presented on their surface of as their microvilli engage these L-selectin ligands. Because endothelial surfaces, in general, and the surface of HEVs, in particular, express thick glycocalyx, which masks the endothelial surface-expressed L-selectin glycoprotein ligands (13,14), it was assumed that CCR7 signals from CCL21 and CCL19 are likely transmitted by a glycocalyx-tethered chemokine. Indeed, the HEV glycocalyx is enriched with CCL21 scaffolds, primarily endothelial surface-expressed heparan sulfate proteoglycans (48). In contrast, in the model suggested by our findings, the T cell microvilli would allow lymphocyte capture under flow by their penetration through the glycocalyx, bringing L-selectin in close contact with its ligands embedded inside the glycocalyx. CCR7 distribution on microvilli would then serve a similar function to L-selectin, allowing this GPCR to encounter CCL21 and CCL19 presented on endothelial scaffolds embedded in the thick HEV glycocalyx.

Our results suggest for the first time, to our knowledge, that all the key molecules that participate in the multistep cascade of lymphocyte rolling, activation, and arrest, in fact, reside on microvilli (Fig. S9). We propose that each T cell microvillus is a signaling hub specializing in transmitting CCR7 signals that trigger LFA-1 inside-out activation (17,18,49). This hub likely performs its task by recruiting to the immediate vicinity of the integrin heterodimer two key activators, a conformationally activated talin-1 and Kindlin-3 (20,50,51). These signals are likely transmitted within a fraction of a second along individual microvilli after their encounter with endothelial CCL21, CCL19, and juxtaposed ICAM-1 (20). Another direct prediction of our data is that lymphocyte arrest can take place abruptly via an individual microvillus, although firm arrest probably requires that several simultaneous CCR7-mediated LFA-1 activations take place independently on several microvilli out of the dozens that project from the basal surface of a rolling lymphocyte into the HEV surface during rolling.

Accumulating results predicted that rapid GPCR signaling to LFA-1, including CCR7, involves a cascade of Gi-protein signals transmitted to Rho GTPases, primarily RhoA (23,24). Basal activity of RhoA not only maintains microvillar stability by keeping ERM proteins (ezrin, radixin, and moesin) at phosphorylated active conformations, but once this protein is in situ activated by CCR7, it may activate talin-1 by elevating PIP2 near CCR7 and LFA-1 within the microvilli tips (24). The proximity between CCR7, RhoA, and LFA-1 on tips of T cell microvilli suggests that CCR7 likely signals to this GTPase within a fraction of a second to trigger local rises in PIP2 levels and local activation of talin-1. It is also likely that the CCR7-Gi-proteins/RhoA signalosomes identified by us in this work and LFA-1, ERM proteins, and talin-1 assemblies are preorganized at some level within lymphocyte microvilli (23,24). Another component of this signalosome is JAK2, previously implicated in CXCL12-triggered high LFA-1 affinity (29). Our new results suggest that JAK2 is also critical for CCL19/CCL21/CCR7-mediated LFA-1 affinity triggering and firm adhesion to ICAM-1, and similar to RhoA, this LFA-1-activating tyrosine kinase is also enriched on T cell microvilli (Fig. S9). JAK2 can phosphorylate and activate guanine -exchange factors that in turn activate RhoA (29). Future super-resolution studies could help identify which of these guanine exchange factors are preassembled within CCR7 signalosomes on circulating T cells.

Once arrested, T cells can propagate additional and much slower CCR7 signals (on a timescale of seconds) to the large nonmicrovillar LFA-1 pools identified by us (Fig. S7). Some of these signals may give rise to small scattered focal adhesions, previously characterized by us, and termed focal contacts or millipede-like adhesions (52). CCR7 can also activate PI3K machineries involved in LFA-1 mobility and adhesion strengthening (53,54). Other signals may involve high mobility LFA-1 molecules and PKCζ activated by CCR7, possibly by Gβγ-dependent signaling (23). A previous study in peripheral blood T lymphocytes and a Jurkat T cell line suggested that LFA-1 occupancy with clustered ICAM-1 can further increase its anchorage (55). This pool of LFA-1, therefore, likely participates in adhesion strengthening of arrested T cells as well as in focal contact-millipede type of crawling over endothelial ICAM-1 (52).

In conclusion, our super-resolution microscopy opens up a new direction for follow-up studies on the potential distribution of CCR7 signaling machineries along specific compartments within T cell microvilli. Future studies should address the possibility that other GPCRs involved in subsecond integrin activation and lymphocyte arrest are also organized in assemblies enriched on the tips of microvilli. We speculate that such GPCRs might share similar effector molecule neighbors with CCR7, such as RhoA, JAK2, and specific talin-activating kinases (24). In that regard, although the chemokines involved in these rapid events were previously termed “arrest chemokines” (16,56) because they are also involved in postarrest leukocyte polarization, adhesion strengthening, and motility processes, a more accurate term to define chemokine receptors specialized in rapid inside-out integrin activation should be “arrest GPCRs”. Whether such specialized GPCRs are also enriched on microvilli of other leukocytes is an interesting and open question. A similar elucidation of additional GPCR-integrin machineries in relation to microvilli in different leukocytes by super-resolution microscopy may help identify distinct and shared players that mediate rapid and efficient integration of chemokine signals critical for integrin-mediated leukocyte arrest on vascular endothelium under shear flow.

Author contributions

S.G. performed all the staining experiments and analysis. S.W.F. performed the CCL19-Fc and antibody validation studies and assisted in cytoskeletal staining and preparing the manuscript. A.M. performed the lymphocyte adhesion and LFA-1 affinity-triggering studies. E.S. performed the scanning electron microscopy experiments. F.R. assisted with cytoskeletal stainings. D.F.L. produced CCL19 and CCL21 fusion proteins for CCR7 labeling. C.L. supervised the adhesion and LFA-1 activation studies. R.A. and G.H. supervised the experiments and wrote the manuscript.

Editor: Sarah Veatch.