Identifying CCR5 coreceptor populations permissive for HIV-1 entry and productive infection: implications for in vivo studies

JC10 cells (HeLa derivative, low CCR5 expression; ~ 2,000 mols/cell) and JC53 cells (HeLa derivative, high CCR5 expression; ~ 130,000 mols/cell) were generated and maintained as described in Platt et al. [47], and PBMCs were obtained from the MicroQuant Leukopaks at IHV UMSOM. JC10 cells, JC53 cells, and PBMCs were cultured in RPMI-1640 with 10% FBSΔ, l-glutamine, and antibiotics. PBMCs were stimulated with 5 µg/mL PHA and 50 UI/mL IL-2 for 3 days prior to an experiment. U87.CD4 cells were cultured in Opti-MEM with 15% FBSΔ and without phenol red.

Cells were transfected with plasmid expressing CCR5-GFP (or CCR5-HA), with GFP or HA fused to the C-terminal (CT) region of CCR5, using a Genepulsor (Bio-Rad) at 270 V (JC10, JC53) or 170 V (U87.CD4)/960µF. Post-transfection, cells were cultured for an additional day and tagged with mAbs against extracellular CCR5 regions. JC10 and JC53 cells treated with FLSC IgG1 received 5 µg/mL for 20 min at RT prior to surface primary antibody labeling. Cells were then permeabilized with eBioscience Permeabilization Solution (Cat. 00-8333-56) for 5 min at RT (JC53) or 0.01% Triton X-100 in PBS for 20 min at 4 °C (U87.CD4, PBMCs) prior to intracellular staining. Each group was stained with one of 1:10 T 21/8 (Invitrogen, Cat. 14-1957-82), CTC5 (R&D Systems, Cat. MAB1802), CTC8 (R&D Systems, Cat. MAB1801), 2D7 (BD Biosciences, Cat. 555991), 45523 (R&D Systems, Cat. MAB181), or 45531 (R&D Systems, Cat. MAB182) CCR5 primary antibodies in PBS with 1% FBSΔ for 20 min at RT, followed by 1:500 anti-mouse AlexaFluor 647 (abcam, Cat. ab150107) under the same conditions. PBMCs were additionally labeled with 1:150 anti-CCR5 C-terminus (abcam, Cat. ab63123) in PBS with 1% FBSΔ for 20 min at RT, followed by 1:500 anti-rabbit DyLight 488 (Invitrogen, Cat. 35552) under the same conditions after permeabilization. Samples were fixed with 3:4 IC Fixation Buffer (Invitrogen, Cat. 00-8222-49) in PBS for 30 min, and then nuclei were stained with 50 µg/mL DAPI for 1 min at RT. Cells were imaged in HBSS using a Zeiss LSM 800 confocal system and Airyscan super-resolution (Carl Zeiss Microscopy, Germany). Cells were analyzed for CCR5 CT labeling (green) and CCR5 extracellular region mAb labeling (red), including the NT and ECL2 using ZEN 2.3 Blue Software. The CT label allows us to track and quantify intracellular and surface expression of CCR5, while also allowing coincident fluorescence imaging of the molecule by mAbs [12]. Such dual staining in 3D provides much greater precision than conventional confocal methods. The dual staining concept is based on previous publications that employed numerous fluorescently tagged mAbs against CCR5 extracellular domains and a CCR5-GFP CT to determine how conformation, localization, internalization, trafficking, and G-protein association are related to permissivity for viral entry [12]. Statistical analysis for surface vs total data sets was by Student’s t-test.

U87.CD4.CCR5-GFP cells were temperature shifted from 4 to 37 °C and incubated for 0, 15, 30, 45, 60, 75, or 90 min in culture media with 100 nM native RANTES or volume-equivalent control before fixation as described above [48]. After fixation, cells were washed, stained, and imaged as described above. Colocalization coefficients (see “Confocal image acquisition and analysis” section) were calculated as a percentage of the 0-min time point.

U87.CD4.CCR5-GFP cells were temperature shifted from 4 to 37 °C and incubated for 90 min with 100 nM native RANTES [48]. U87.CD4.CCR5-GFP cells were then re-suspended in culture media with 0.45 M sucrose, 1 µM cytochalasin D, or volume-equivalent control for 0 (directly after RANTES removal), 30, 60, or 120 min at 37 °C [12, 48]. Cells were fixed, washed, and stained as described above. Colocalization coefficients (see “Confocal image acquisition and analysis” section) were calculated as a percentage of the volume-equivalent control before graphing.

Cryopreserved PBMCs were thawed and cultured/activated as described above. Cells were washed and incubated with HIV_BaL (MOI = 0.00075) and 50 UI/mL IL-2 for 3 h at 37 °C, 5% CO₂, with control cells not challenged with HIV-1 [49]. Cells were washed and left in culture with 50 UI/mL IL-2 for 1 week with a half media change after 3 days. Cells were collected and stained as described above. Infected cells were examined for p24 expression via ELISA to verify the productive infection by HIV-1 (data not shown). Cell culture supernatant was collected and frozen at − 20 °C, and later thawed and added to a 96-well flat-bottom plate at a 1:200 dilution in 1× lysis buffer and culture medium. HIV-1 p24 was quantified using the PerkinElmer Alliance HIV-1 p24 Antigen ELISA Kit (Cat. No. NEK050001KT). Surface expression was measured as the colocalization coefficient and compared before and after infection to determine which conformations were primarily internalized by HIV-1 binding (Donor #7003). This assay was repeated using another donor (Donor #0872) with similar results (data not shown). Statistical analyses were evaluated by comparing relative surface expression (surface coefficient/mean total coefficient) using Student’s t-test to measure significance between infected and non-infected data sets.

Three Airyscan (~ 120 nm resolution in x–y direction) laser lines at 405 nm (DAPI, violet), 488 nm (GFP, green) and 647 nm (mAb, far red) were used to excite fluorescence signals in these imaging experiments, which was separated by the quad DAPI/FITC beam splitter and acquired by the EM-CCD camera [50] via the Carl Zeiss LSM 800 confocal system (Carl Zeiss, Germany). Individual fluorescence signals were further acquired using a Gasp detector (Carl Zeiss LSM 800, Germany). A Plan-Apochromat 63×/1.4 oil DIC objective (Carl Zeiss LSM 800, Germany) was used to visualize three-colored cell samples. All parameters were consistent in each experiment, including laser excitation power and detector, and offset gain. ZEN Blue 2.3 software (Carl Zeiss Microscopy) was used to generate and analyze original images. Non-transfected cells were stained and imaged with the same conditions as described above (data not shown) and used as a negative control to calculate the background. All the images were acquired using the same instrumental settings. Signal to noise ratio was assured by measuring the total intensity of each image and averaging data for each image set. Oversaturated signal was avoided by using the software’s algorithm. Images were analyzed to determine the percentage of CCR5 conformations. GFP labels (green pixels) were used to calculate total (all surface and intracellular) CCR5 and overlapping staining (yellow pixels) was used to calculate primary antibody staining. CCR5 colocalization coefficients were calculated using the primary staining (yellow) divided by total CCR5 (green and yellow), for each monoclonal antibody detecting a different conformation of CCR5 (red signal). Colocalization coefficients calculate the percentage of total CCR5 receptors labeled by the mAb. Surface colocalization measures only mAb-bound CCR5 at the cell surface, while total colocalization measures mAb-bound CCR5 both at the surface and inside the cell. The difference between surface and total colocalization coefficients is the amount of intracellular only mAb-bound CCR5. Color thresholds and background were determined using single color controls processed through ZEN Blue 2.3 software. Values for data analysis were collected by compiling data from at least 10 randomly picked sample fields. GFP and mAb signals were collected as voxel fluorescence intensities.

The abundant presence of CCR5 on the cell surface is essential for HIV-1 binding, entry, and infection. We investigated the quantities and distributions of the different subpopulations of CCR5 in selected cell lines. Figure 1 shows the dual staining methodology that allowed us to visualize CCR5 events within the cellular spaces of cell lines and donor cells using ZEN Blue 2.3 software with comparative images between Airyscan-based super resolution and standard confocal microscopy. Figure 1A schematically shows the double labeling strategy used in our imaging experiments. The figure depicts the C-terminus-GFP fusion protein (green) and NT/ECL2 tagged with an Alexa 647-conjugated mAb (red). This Airyscan-based super-resolution technique (Fig. 1B) enabled greater resolution of CCR5 localizations and staining puncta versus standard confocal microscopy (Fig. 1C), which failed to provide more comprehensive results for this type of study [50]. This level of resolution permits quantitation of the total amount of CCR5 bound by mAbs on a cell as well as internalized events. Further, mAb-bound CCR5 on the cell surface versus intracellular compartments can be effectively distinguished by imaging samples of cells with and without membrane permeabilization.

To establish an imaging procedure for surface staining of CCR5, we first wanted to be able to visualize different levels of surface CCR5 conformations, as identified with different monoclonal antibodies, namely CTC8, labeling the CCR5 NT, and 2D7, labeling the CCR5 ECL2, in red color. The JC10 cell line expresses ~ 2,000 molecules per cell (Additional file 1: Fig. S1A) whereas the JC53 cell line expresses ~ 130,000 CCR5 molecules per cell (Additional file 1: Fig. S1B). There is a clear difference in the intensity levels of the surface CCR5 signal between the two cell lines in all sets of samples, as expected due to the difference in their natural surface CCR5 expression levels. Both cell lines showed a decrease in mAb binding to CCR5 when cells were pretreated with full length single chain IgG1 (FLSC IgG1), a chimeric protein containing HIV-1_BaL gp120 and the D1 and D2 domains of human CD4. FLSC IgG1 competitively binds the coreceptor at two different regions, the NT and ECL2 [17, 18]. As expected, the visualized CCR5 signal was correspondingly lower in the JC10 cell line due to lower CCR5 expression (Additional file 1: Fig. S1).

Figure 2 demonstrates how the ZEN analysis was used to quantify the colocalization of CCR5 events, allowing us to measure different CCR5 conformational subpopulations using our dual staining method. Figure 2A shows the pixel color spread and density along with the thresholds with which the program assigns color identity, determined using single color controls. The section labeled 1 is green color only (GFP), section 2 is red color only (Alexa 647-conjugated mAbs), section 3 is overlapping signal (high green and red, appears as yellow color), and section 4 denotes background (low green and red signals). Figure 2B is the Airyscan-based image of visualized dual CCR5 staining where the green signal shows the labeled CCR5 CT, while the red signal is extracellular and indicates NT or ECL2 labeling. The yellow signal corresponds to overlapping signals. The intracellular region with low/no labeling is the nucleus (not shown to better visualize and focus on the CCR5 events). Figure 2C is a color-coded image using the parameters set in Fig. 2A by the ZEN Blue 2.3 software and shows how it quantifies pixel count to calculate the colocalization coefficient. Red signal only (Alexa 647-conjugated mAbs) pixels are labeled in orange, green signal only (GFP) pixels in cyan and overlapping yellow color (single CCR5 molecule) pixels in dark blue. Figure 2 shows our method of analysis in JC53 cells, that was also used for all other cell lines and donor cells in the study.

We visualized the abundance, location, and distribution patterns of CCR5 conformations with different CCR5 epitopes exposed in the high CCR5-expressing JC53 cell line. To more precisely visualize and properly quantify CCR5 subpopulations with conformations permissive for viral binding and entry, we expressed CCR5-GFP fusion protein by plasmid electroporation into JC53 HeLa derivatives. Combined with the available CCR5 antibodies recognizing the NT and the ECL2 central epitopes, this allows two-color labeling to assess distances between epitopes. The C-terminus-GFP tag of CCR5 did not alter the intracellular and surface distribution of CCR5 (data not shown). The clear dynamics of CCR5 subpopulations and distinct pattern of inner (green signal; CT) versus outer (red signal; NT or ECL2) CCR5 membrane subpopulations were evident in the Airyscan-based imaging (Carl Zeiss, Germany). We visualized the surface versus total CCR5 subpopulations using six different mAbs. The images of panels containing mAbs against NT, ECL2, or against both [multiepitope (ME)] domains are presented in Fig. 3. Cultured cells were stained for surface CCR5 as described in “Materials and methods” section (Fig. 3A), then permeabilized prior to staining when measuring conformation frequency of internal CCR5 events to evaluate the presence or abundance of each conformation on the cell surface and within the cell relative to total CCR5 events (Fig. 3B). This yielded information on internal coreceptor conformations by comparing total and surface labeling percentages for each mAb of interest, that was later quantified. Visualizing total CCR5 subpopulations, we were able to determine whether most of each conformation is extracellular or intracellular. More detailed images of CCR5 conformational forms and divergent staining patterns are shown in Additional file 2: Fig. S2A with 2D7 (ECL2) visualized as a continuous pattern of puncta versus a distinctive clustered pattern, as seen with 45531 (ECL2, cholesterol-rich regions), while the quantified differences between surface densities are given in the graph of Additional file 2: Fig. S2B. Quantification of colocalization coefficients demonstrated that approximately 60% of 2D7 conformations were located on the surface whereas ~ 81% of 45531 conformations were located on the surface (Additional file 2: Fig. S2B), agreeing with the data in Fig. 4A. Punctuate staining patterns have been shown to be evident around dense cross-linked areas of actin filaments, which contribute to processes involved in virus entry, particularly endocytosis [51]. Airyscan based super resolution images in Fig. 3A, B show that the anti-ECL2 mAb 45531 has a more clustered staining pattern, most likely because it is binding to membrane regions that are rich in cholesterol [11, 12, 52], while the anti-NT mAb CTC5 yields more of a “capping” pattern, with CCR5 congregating on one side of the cell [38]. Interestingly, the patterns with the ME mAb 45523 or anti-NT mAbs are quite distinct from those of the anti-ECL2 epitope (2D7), which show a more uniform surface distribution pattern (Fig. 3A, B). These data agree with other findings by Flegler et al. [12].

We next quantified the colocalization coefficient of total versus surface CCR5 events in the JC53 (Fig. 4A) and U87.CD4.CCR5-GFP (Fig. 4B) cell lines to compare CCR5 cellular distributions in a more physiologically relevant (U87.CD4.CCR5) cell line, and ultimately, primary cells. CCR5 was stained as described above and colocalization coefficients analyzed as described in Fig. 2. The first three sets in each Fig. 4 panel correspond to the mAbs against the NT: CTC5, T 21/8, and CTC8; the last three sets correspond to the mAbs against the ECL2: 45523, 2D7, and 45531, with 45523 being a ME mAb against both ECL1 and ECL2. Logically, the number of total events were higher than surface only using all mAbs, except for the T 21/8 mAb. The quantified data in Fig. 4A complemented the visual data of surface and total mAb staining of CCR5 events in Fig. 3. For example, the lack of visualized cell surface events seen with mAb 2D7 in JC53 cells (Fig. 3) was observed in the quantitative analysis (Fig. 4A) by its percentage labeling of surface CCR5 being the least among all mAbs. A similar pattern of staining with mAb 2D7 was also seen in U87.CD4 cells (Fig. 4B). Additionally, mAbs T 21/8 and 45531 show the lowest total staining both visually and quantitatively. Graphs of JC53 (Fig. 4A) and U87.CD4.CCR5 (Fig. 4B) surface vs total colocalization coefficients clearly show that some CCR5 conformations were more prominent inside the cell (ex. CTC8, 45523) while others were primarily localized to the cell surface (ex. CTC5, 45531). Additionally, cell type seemed to play a role in how conformations were distributed, with 2D7 more prominent on the surface in U87.CD4.CCR5 cells compared to JC53, where it was more heavily internalized. The following surface versus total percentages in the case of JC53 clones (anti-NT mAbs—CTC5, CTC8, T 21/8; anti-ECL2—2D7, 45531, 45523) were: 81%, 52%, 100%, 62%, 100%, and 57% respectively (Fig. 4A). In the case of U87.CD4.CCR5 cells, the percentage of subpopulations expressed on the surface were: 91%, 61%, 93%, 87%, 87%, and 52% respectively (Fig. 4B). The limited cell membrane labeling seen with ME mAb 45523 was reflected in our analysis by its percentage labeling (Fig. 4A, B). Conformations detected by mAbs in Fig. 4 were abundant on the cell surface, with around 20–40% of all conformations bound by each mAb on the cell surface alone. These conformations are also found internally, although the proportions vary as presented in Fig. 4. For example, CTC8, an NT-specific mAb, recognizes conformations that appear ~ 50% internally, with a surface colocalization coefficient of ~ 20% compared to a total colocalization coefficient of ~ 40%. However, the 45531-epitope targeting ECL1/ECL2 was detected primarily on the surface (80+%). Some mAbs may have overlapping conformations depending on which epitopes are available for binding on each CCR5 molecule. There was a significant proportion of internal conformations with recognizable CTC8 or 45523 epitopes (Fig. 4B) with p = 0.002 and < 0.001 respectively in U87.CD4.CCR5 cells. Differences between surface and total frequency in all other data sets were not statistically significant.

We next quantified rates of CCR5 internalization and recycling to better understand how different conformations are trafficked and redistributed (Fig. 5). CCR5 internalization was triggered by treating cells with native RANTES at 37 °C and comparing CCR5 surface conformation appearances over time to those without RANTES treatment. The dynamics of CCR5 population cycling (both on the surface and internally within the cells) should give insight into HIV-1’s entry potential depending upon the presence of selected CCR5 subpopulations. To measure RANTES-induced CCR5 endocytosis, we first quantified the localization of CCR5 conformations using mAbs against different CCR5 domains and their rates of internalization (Fig. 5A). For anti-NT, anti-ECL2, and anti-ME mAbs, RANTES caused strong internalization (solid symbols) compared with a 37 °C temperature trigger alone control experiment (open symbols). The mAbs specific for ME ECL1/2 (45523) and the NT (CTC8) showed a great increase in internalization in response to RANTES (red and blue symbols respectively). Interestingly, this was not the case with mAb 2D7, which recognizes a conformational epitope in ECL2, suggesting either a conformational change that occurs during internalization, or that this conformation more readily binds RANTES, leading to faster internalization and subsequent recycling. Cell surface CCR5 was reduced by ~ 50% after 30 min of a saturating concentration (100 nM) of RANTES (anti-NT and anti-ECL2 mAb, the blue and green symbols respectively), while anti-ME mAb (red symbols) required approximately 60 min to reach its maximum internalization (Fig. 5A). These differences need further investigation and may be related to different rates of CCR5 endocytosis, recycling, and binding affinity of RANTES to the cell surface [12, 53], but the data support the classical model for chemokine binding to CCR5 [7] that describes the chemokine core binding to extracellular regions of CCR5, while the N-terminal tail of RANTES interacts with the transmembrane domains. CCR5 was redistributed upon RANTES application. The analysis did not focus on competitive binding between RANTES and mAbs due to RANTES inducing internalization of CCR5, and only surface CCR5 events being measured. Next, we monitored recycling rates upon RANTES removal and CCR5 reappearance rates on the cell surface (Fig. 5B), using the same set of mAbs and staining procedure that were used in Fig. 5A. Figure 5B is a direct kinetics experiment continuation of the kinetics experiment presented in Fig. 5A.

Knowing that CCR5 internalization and recycling rates are regulated by actin polymerization and activation of small G proteins [54], we measured the present CCR5 epitopes available for mAb binding on the cell surface upon cytochalasin D (depolymerizes actin) addition (Fig. 5B), which prevents CCR5 internalization by blocking endocytosis [12, 54], and sucrose [12], which depletes cytoplasmic pools of clathrin by causing an accumulation of clathrin microcages on the inner cellular membrane as percentages of the volume-equivalent control (without treatment) as a continuation of the timepoints in Fig. 5A. Our rationale for this experimental effort was that actin filaments contribute to processes such as CCR5 endocytosis [51, 54], and we wanted to know which CCR5 populations were primarily affected by the blockage of these endocytic pathways. We selected antibodies targeting conformations that showed a high level of naturally internalized molecules in U87.CD4.CCR5 cells. Based on the results of Fig. 5, there did not seem to be any major differences between the untreated control sample (black dotted 100% line) and conformations labeled for the CTC8 or 2D7 epitopes, which indicates that recycling is not significantly dependent on these trafficking pathways (p > 0.05 at each time point). There also did not seem to be any major difference between the control and conformations labeled for the 45523-epitope treated with sucrose (p > 0.05 at each time point taken). However, there was a significant reduction in CCR5 surface recovery for conformations labeled with the ME 45523 antibody and treated with cytochalasin D until 120 min (p ≤ 0.001 at 0 min, 0.006 at 30 min, 0.02 at 60 min) (Fig. 5B).

To address which CCR5 epitopes (and by extension, which CCR5 conformations) are primarily used for HIV-1 infection, we identified a high CCR5 expressing PBMC donor and infected the cells with HIV-1_BaL for 1 week before staining and measuring CCR5 surface conformation frequency using the six mAbs previously used in the three cell line experiments. Colocalization coefficients were calculated with or without HIV-1 infection and compared to determine how HIV-1 affects CCR5 surface subpopulations (Fig. 6). Those conformations that had a greater reduction in surface density after HIV-1 challenge should be those more readily bound (and internalized) by the virus, resulting in a reduction in their surface density. For the most part, there was not a major change in the percentage of surface conformations after HIV-1 challenge, with only a notable drop (p = 0.033) in conformations with the CTC8 epitope. We confirmed this for a second donor (data not shown) that exhibited an even larger drop in surface CTC8 conformations (p < 0.001), while also exhibiting a drop in relative surface 45523 conformations (p = 0.001), although this could not be replicated in the first donor. Other differences between surface CCR5 conformation densities with or without HIV-1 infection were not significant. Total (permeabilized) colocalization remained relatively the same between infected and non-infected cells except for the T 21/8 epitope in the first donor, which showed lower total CCR5 compared to non-infected cells, although this could not be replicated in the second donor. The second donor showed no significant differences in total CCR5 with or without infection for any detectable conformations. PBMC culture supernatants were additionally checked for p24 via ELISA to ensure the cells were productively infected (data not shown). This was repeated with a second donor to validate the pattern (data not shown).