Transient Surface CCR5 Expression by Naive CD8⁺ T Cells within Inflamed Lymph Nodes Is Dependent on High Endothelial Venule Interaction and Augments Th Cell–Dependent Memory Response

The key to a successful adaptive immune response requires the physical interaction between rare APCs bearing cognate Ag and rare Ag-specific T cells (1 in 10⁴–10⁶) within the secondary lymphoid organs, including the lymph nodes (LNs) (1). This interaction not only promotes the initial expansion of Ag-specific T cells but also creates a residual memory T cell population after the primary immune response has subsided. Development of these lymphocytes depends on helper activity provided by other immune cell types and soluble mediators within the inflammatory LN microenvironment. Although help from CD4⁺ T cells is not an absolute requirement to generate primary CD8⁺ T cell response, the presence of CD4⁺ Th cells enhances the magnitude of CD8⁺ memory T cell generation (2). We and others have previously shown that the initial surveillance by naive polyclonal CD8⁺ T cells of cognate Ags presented by dendritic cells (DCs) is facilitated by the local accumulation of CCL3 (MIP-1α) and CCL4 (MIP-1β), which are released by the complex between activated DCs and other Ag-specific CD4⁺ and CD8⁺ T cells (3, 4). This CCL3/CCL4–CCR5 chemokine interaction enhances the recruitment of nonantigen-specific CD8⁺ T cells to the site of activated DCs in the LN, and it increases potential Ag recognition by additional CD8⁺ T cells on DCs. Importantly, neutralizing the effects of CCL3/CCL4 during the early immune priming stage reduces the efficiency of polyclonal CD8⁺ T cell surveillance in a CCR5-dependent manner and abrogates the Th cell–enhanced long-term CD8⁺ memory T cell generation in vivo (3). The exact molecular mechanisms contributing to the efficacy of CCL3/CCL4–CCR5 interaction on naive CD8⁺ T cells with regard to memory T cell generation remain unknown.

The LN is positioned at a location where naive T cells and Ag-loaded DCs encounter each other. Circulating naive T cells first tether to the LN endothelium through the interaction of CD62L on T cells with peripheral node addressin (PNAd), a shared motif expressed on several proteins, including CD34 and Glycam-1 of the high endothelial venule (HEV) (5–10). These tethered T cells then roll on the endothelium, engaging surface CCR7 with CCL21 that is bound to heparan sulfate and collage IV on the luminal surface of the HEV (11–13). Engagement of both CD62L and CCR7 strengthens T cell adhesion to the HEV. It also results in a conformational change of CD11a on the T cell (14). This conformational change from low- to high-affinity CD11a/CD18 facilitates stronger adhesion through CD54 located on the HEV, thereby promoting transendothelial migration of T cells through the HEV (5). Upon entry into the inflamed LN, a subset of naive CD8⁺ T cells begins to navigate the complex LN microenvironment, guided by functional CCR5 molecule on the surface, for efficient cell–cell contact with activated DCs. Normally, only a minute number of naive CD8⁺ T cells express detectable levels of CCR5 on the cell surface in the blood and LN (3, 4). However, previous published data implicated the importance of the CCL3/CCL4–CCR5 chemokine signaling axis during vaccine-induced immune priming in the draining LN (DLN), suggesting that mechanisms exist for the expression and utilization of CCR5 by some naive CD8⁺ T cells in inflamed LNs that help to guide these cells to sites of activated T cell–DC complexes where high local concentrations of CCL3 and CCL4 exist.

In the present study, we find that a subset of circulating naive, CCR5⁻CD8⁺ T cells upregulates surface CCR5 protein expression early after entry into the inflamed LN in an Ag-nonspecific manner. Whereas engagement of increased ligands for CD62L and CD11a on the inflamed HEV promotes adhesion and entry of naive CD8⁺ T cells into the LN, the same molecular interactions rapidly promote a subset of naive CD8⁺ T cells to mobilize preformed intracellular CCR5 proteins from intracellular compartments to the cell surface. Furthermore, we found that the naive CCR5⁺CD8⁺ T cell subset developed more robust memory response upon Ag rechallenge and that this enhancement is associated with increased contact with Ag-bearing DCs in the inflamed LN.

The mice used in this study were purchased from either Taconic (Hudson, NY) or The Jackson Laboratory (Bar Harbor, ME). We used 8- to 12-wk-old male and female wild-type C57BL/6 mice, C57BL/6 mice congenic for CD45 (CD45.1), OT-I TCR transgenic mice (15) on the Rag2^−/− background, OT-II TCR transgenic mice (16), and OT-I × CCR5^−/−, MHC class II (MHC-II) knockout (KO), and β₂-microglobulin KO mice. All animals were housed and handled according to National Institutes of Health institutional guidelines under an approved protocol by Case Western Reserve University Institutional Animal Care and Use Committee (no. 2012-0126).

T cells from donor mice were isolated from LNs and spleen by passing tissue through a 40-μm filter to obtain a single-cell suspension. This cell suspension was treated with ACK lysis buffer to deplete RBCs. OT-I × CCR5^+/+ mice are on a Rag2^−/− background, and T cells isolated from these mice were 75–80% CD8⁺ without additional enrichment steps. CD8⁺ T cells from OT-I × CCR5^−/− and C57BL/6 mice were enriched using negative selection with B220 and CD4 Dynabeads (Life Technologies, Grand Island, NY). OT-II T cells were enriched using negative selection with B220 and CD8 Dynabeads.

Inflamed LNs were generated by injecting mice in the right footpad with either 10 μg LPS from Pseudomonas aeruginosa (Sigma-Aldrich, St. Louis, MO) or 20 μg CpG oligodeoxynucleotide (Klineman no. 1466, 5′-TCAACGTTGA-3′; Klineman no. 1555, 5′-GCTAGACGTTAGGT-3′; Life Technologies) admixed with aluminum hydroxide (Alum). PBS or PBS/Alum was injected into the contralateral footpad as controls for LPS/Alum or CpG/Alum, respectively. For in vivo activation, 1 × 10⁷ polyclonal T cells from C57BL/6 mice or monoclonal T cells from OT-I × Rag2^−/− mice were injected i.v. into recipient mice 24 or 48 h after footpad injections of LPS/Alum, CpG/Alum, or PBS/Alum. At different times both right (DLNs) and left (nondraining LNs; NDLNs) popliteal LNs were harvested, and single-cell suspensions were stained with Abs to CD8, CD45.2, and CCR5. For in vitro activation, both DLNs and NDLNs were removed 24 or 48 h after footpad injections and placed into 96-well V-bottom plates with 100 μl RPMI 1640–0.5% FBS to cover samples. For some experiments axial LNs were also collected as additional NDLNs. T cells were harvested from congenic mice as described above. One million T cells were added directly to the LN, which had been teased open with tweezers. To examine soluble effects of the inflamed LN, the T cells (1 × 10⁶ T cells) were added to the top well of a Transwell plate with 0.4-μm-diameter pores (Corning, Tewksbury, MA). Additionally, T cells (1 × 10⁶ T cells) were added to a 96-well plate with conditioned media that was collected from the inflamed LN after 24 h in culture with serum-free media. For Ab-coated beads, 50 μl sheep anti-rat IgG Dynabeads were suspended in 450 μl FACS buffer and 1 μg Ab (eBioscience rat IgG; CD62L, clone Mel-14; CD11a, clone 2D7) for 30–90 min at 4°C. Five micrograms rat IgG was added for an additional 30 min at 4°C. Beads were resuspended to a final volume of 500 μl FACS buffer and aliquots (10, 30, or 100 μl) were added to 2 × 10⁶ OT-I T cells resuspended in 500 μl FACS buffer, mixed at 4°C, and transferred to 37°C for different times. For restimulation experiments, 2 × 10⁶ cells were incubated with 100 μl anti-CD11a–coated beads for 20 min at 37°C and then placed on magnets to remove nonadherent cells. After washing, cells were flushed off beads and incubated with biotinylated anti-CCR5 Ab with anti-biotin microbeads (Miltenyi Biotec) to remove CCR5⁺ cells. CCR5⁻ cells (2 × 10⁶) were then incubated again with 100 μl anti-CD11a–coated beads for 20 min at 37°C and then examined for CCR5 expression by flow cytometry. To determine the role of protein synthesis or transport in CCR5 expression, OT-I T cells were cultured with specific inhibitors of protein synthesis and transport for 2 h and 100 μl beads was added for 30 min at 4°C, transferred to 37°C for 20 min, washed, and stained for expression of CCR5. To ensure that inhibitors were capable of blocking protein synthesis or transport, OT-I T cells were cultured with inhibitors for 2 h and then placed in anti-CD3/CD28–coated wells for 24 h and examined for surface CD69 expression by flow cytometry. To measure memory responses, OT-I T cells were cultured with inflamed LNs for 5 h, homogenized, and then sorted into CD45.2⁺CD8⁺CCR5⁺ or CD45.2⁺CD8⁺CCR5⁻ populations using FACSAria (BD Biosciences). One hundred thousand sorted T cells were then injected into naive CD45.1⁺ mice along with 3 × 10⁶ OT-II T cells. The following day, mice were immunized with Alum mixed with OVA_257–264 (1 μg) and CpG (20 μg) in the presence or absence of OVA_323–339 (10 μg). After 7 or 28 d, mice were sacrificed and the spleen and DLN were collected and restimulated with OVA_257–264 (1 μg) for 4 h at 37°C in the presence of brefeldin A. The total number of OT-I T cells and the number of IFN-γ–producing cells from both the spleen and LN were then enumerated based on FACS analysis.

Abs were purchased from eBioscience, BD Pharmingen, or BioLegend and are as follows: rat anti-mouse CD8α (clone 53.6.7), rat anti-mouse CD4 (clone GK1.5), biotinylated anti-mouse CCR5 (clone HM-CCR5), anti-mouse CD3e (clone 145.2C11), anti-mouse CD45.2 (clone A20), rat IgG2b, rat IgG2c, and hamster IgG-biotin isotype control. Additional CCR5 Ab, MC-68, was a gift from Dr. Matthias Mack (17). Between 2 × 10⁵ and 1 × 10⁶ cells were washed in ice-cold FACS buffer (PBS/2.5 mM EDTA/0.1% BSA), followed by incubation in blocking buffer (10% normal mouse serum in FACS buffer) for 15 min at 4°C. Abs were added for 30 min on ice, except for anti-CCR5 Ab, which was incubated for 1 h on ice. The samples were then washed twice with ice-cold FACS buffer and the samples were run on an Accuri C6 or FACSCalibur flow cytometer. Fold increase of CCR5 expression was determined by using the following formula: (% CCR5⁺ experiment − % CCR5⁺ isotype control)/(% CCR5⁺ control − % CCR5⁺ isotype control). To detect IFN-γ–producing T cells, after stimulation with anti-CD3, T cells were first stained with anti-CD8 and CD45.2 prior to fixing with Cytofix/Cytoperm (BD Biosciences). Cells were then washed with Perm/Wash buffer (BD Biosciences) and then stained with anti–IFN-γ Ab (clone XMG1.2). The data were then analyzed using FlowJo software.

To determine expression of CD54 and PNAd in LNs, DLNs or NDLNs were isolated from mice 48 h after footpad injection of CpG/Alum or PBS/Alum, fixed in periodate-lysine-paraformaldehyde buffer overnight, transferred to 30% sucrose for at least 6 h, frozen at −80°C in OCT, and then sectioned into 5-μm slices and transferred to microscope slides. Slides were treated with Ag retrieval buffer prior to being blocked with mouse serum, stained against PNAd (eBioscience, clone MECA-79) Ab and hamster anti-CD54 (Abcam, Cambridge, MA, ab171118) using PELCO BioWave Pro (Ted Pella, Redding, CA), washed, and stained with secondary FITC mouse anti-rat IgG Ab (eBioscience) and Alexa Fluor 594 goat anti-hamster IgG (Life Technologies). Slides were examined on a Leica SP5 confocal microscope (Leica Microsystems, Buffalo Grove, IL), and images were analyzed on LAS software using the same settings (Leica Microsystems). Regions of interest (ROIs) were drawn around PNAd⁺ regions, and the relative intensities were determined for both the PNAd⁺ and CD54⁺ channels. Background intensities were determined by selecting ROIs in unstained areas and subtracting these values from both the PNAd and CD54 channels. The relative expression of CD54 in DLNs and NDLNs was determined using the following formula: (average intensity CD54 − background intensity)/(average intensity PNAd − background average intensity). For examining CCR5 localization, freshly isolated OT-I T cells or cells isolated from the DLN and NDLN were added on top of poly-l-lysine–coated coverslips, spun down at low speed (600 rpm) for 30 s, washed, and then fixed in 2% paraformaldehyde. Cells were then incubated with FACS buffer containing 0.05% Tween 20 for 20 min at room temperature to permeabilize cells. Cells were stained with anti–CD8-FITC, CD45.2-PerCp-Cy5.5, and anti–CCR5-biotin for 30 min at room temperature, washed three times, and then incubated with streptavidin–Alexa Fluor 594 (Invitrogen) for an additional 30 min at room temperature. Slides were washed and air dried for 10 min before adding DAPI Fluoromount-G (SouthernBiotech, Birmingham, AL) followed by placement of a coverslip and sealing with nail polish. Images were collected on a Leica SP5 confocal microscope, and CD45.2⁺ cells were selected and examined for expression of CD8 and CCR5. The amount of CCR5 that colocalized with CD8 was measured using Imaris software (Bitplane). To measure DC–T cell interaction frequency, 1 × 10⁶ bone marrow–derived DCs were stained with CellTracker Blue (10 μM), pulsed with 1 μg/ml OVA_257–64 and 10 μg/ml LPS for 1 h at 37°C, washed, and then injected into the right footpad of C57BL/6 mice. Twenty-four hours later equal numbers of 5–10 × 10⁶ SNARF-labeled OT-I⁺ T cells and CFSE-labeled CCR5 KO OT-I⁺ T cells were coinjected via tail vein. Five hours later DLNs and NDLNs were removed, fixed in formalin overnight, and then incubated in 30% sucrose overnight. LNs were then sliced into 5-μm sections and examined via two-photon microscopy. The number of DC–T cell contacts per slide was determined using Imaris software and averaged over multiple imaging fields in multiple LN samples.

Total RNA was isolated using TRIzol reagent per the manufacturer’s protocol (Gibco BRL, Carlsbad, CA) and purified using an RNeasy Mini kit (Qiagen). RNA quality was assessed by spectrophotometer absorption at 260/280 nm. RNA was converted to cDNA using 200 U Moloney murine leukemia virus reverse transcriptase (Invitrogen Life Technologies, Grand Island, NY) for 60 min at 37°C in the presence of 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl₂, 10 mM DTT, 100 ng of oligo(dT) 12–18, 0.5 mM 2′-deoxynucleoside 5′-triphosphate, and 40 U recombinant RNase inhibitor. cDNA was amplified in the presence of FAM-labeled gene-specific primers and TaqMan universal master mix (Applied Biosystems, Foster City, CA) in a 96-well microtiter plate format on an ABI Prism 7300 sequence detection system (Applied Biosystems). Each PCR reaction was performed in triplicate using the following conditions: 2 min at 50°C and 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 60 s at 60°C. Relative levels of RNA were determined using the cycle threshold (C_t) (18). C_t values from target gene were subtracted by the C_t value of GAPDH to determine relative expression (ΔC_t = C_t target − C_t GAPDH). The concentrations of gene-specific mRNA in samples were compared with untreated LN and were calculated by subtracting normalized values from treated groups from untreated LN (ΔΔC_t = ΔC_t test − ΔC_t control) and then determining relative concentration (2^−ΔΔCt). For microarray analysis, naive OT-I T cells were exposed to CpG-inflamed CD45.1 LNs for 6 h at 37°C, stained for surface expression of CCR5, and sorted into CCR5⁺ and CCR5⁻ populations. Total RNA was isolated from both populations and subjected to Affymetrix GeneChip Mouse Gene 1.0 ST Array analysis to identify differentially expressed gene transcripts as described (Affymetrix, Santa Clara, CA). Heat maps were generated and examined using a robust multichip analysis algorithm to determine the criteria of significance (fold change ≤ −1.5 or ≥ 1.5). All microarray data have been deposited in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo) with accession no. GSE76820.

All error bars represent SD from the mean, and the statistical analysis was performed using either a Student t test or one-way ANOVA with InStat software (GraphPad Software).

Although expression of CCR5 is classically associated with post-TCR–activated T cells and considered absent on the surface of naive T cells under steady-state (19), available data implicate an important functional role for CCR5 on a small subset of naive T cells during the initiation of inflammation-mediated primary immune responses in inflamed LNs (3, 4). To examine whether the observed functional role for CCR5 is the result of a small number of previously activated T cells or an upregulation of surface CCR5 molecule on recent T cell immigrants in the LN, naive polyclonal T cells (CD44^loCD62L^hiCCR5⁻) from C57BL/6 (CD45.2⁺) mice were injected via tail vein into congenic (CD45.1⁺) recipients that were induced to undergo regional tissue and draining popliteal LN (DLN) inflammation upon injection of CpG/Alum in the right rear footpad 48 h prior to T cell inoculation. As controls, the contralateral footpad received either PBS/Alum injection or sham injection, thereby creating NDLNs in the contralateral popliteal LN. At various times after adoptive T cell transfers, both the DLN and NDLN were removed, and isolated cells were stained for CD45.2, CD8, and CCR5. Whereas there was some modest increase of surface CCR5 expression on transferred naive T cells isolated from the NDLN 4 h after adaptive transfer, we observed a >10-fold increase in CCR5 expression on the same transferred T cells in the DLN (Fig. 1A). Multiple experiments using either polyclonal C57BL/6 CD8⁺ T cells or monoclonal OT-I CD8⁺ T cells showed similar results, although there was a wide range (∼5- to 45-fold increase) in the absolute magnitude of CCR5 induction (Fig. 1B). Similar results of CCR5⁺ T cell accumulation among transferred naive CD8⁺ T cells in the DLN were also seen using either LPS/Alum (Supplemental Fig. 1) or LPS-treated bone marrow–derived DCs (data not shown). Although to a lesser magnitude, there is also an enhanced accumulation of endogenous CCR5⁺CD45.1⁺CD8⁺ T cells in the DLN (Supplemental Fig. 2). Interestingly, we found that the surface expression of CCR5 was transitory, as it peaked between 2 and 4 h in the DLN, began to decrease by 6 h (Fig. 1C), and returned to basal expression level by 24 h (data not shown). These data suggest an accumulation of naive CCR5⁺CD8⁺ T cells in the DLN; however, it does not discriminate whether this observation was the result of preferential in vivo recruitment and retention of contaminating CCR5⁺ donor memory T cells from the circulation or the induction of CCR5⁺ T cells from the CCR5⁻ naive donor T cell pool. The former scenario is unlikely due to several reasons. First, the total number of CCR5⁺CD45.2⁺CD8⁺ T cells in the DLN far exceeds the total number of CCR5⁺ cells in the initial pool of transferred T cells. Second, the transfer of OT-I T cells isolated from Rag2^−/− mice, which lack memory T cells (20), resulted in the comparable appearance of CCR5⁺CD8⁺ T cell accumulation in the DLN (Fig. 1B). To provide further evidence for the direct conversion of CCR5⁻CD8⁺ T cells into CCR5⁺CD8⁺ T cells in the DLN, as well as to determine whether such conversion is dependent on either direct cell–cell contact or mediated by some soluble factor from inflamed LNs, DLNs and NDLNs were removed from CD45.1 congenic mice 48 h after injection of CpG/Alum and teased apart in a 96-well plate. Naive CCR5⁻CD45.2⁺CD8⁺ T cells were added either directly to the LNs ex vivo or placed on top of a Transwell with small pores (0.4 μm) to prevent direct contact with the LN but allow for soluble factors to pass through. This ex vivo experimental setup also prevented the potential ongoing recruitment of circulating CCR5⁺CD8⁺ memory T cells as a cause of the observed accumulation in vivo. Four hours following incubation, CD45.2⁺CD8⁺ T cells were stained for surface CCR5 expression. Interestingly, only direct contact between the DLN tissue and T cell resulted in greatest upregulation of surface CCR5 expression within the CD45.2⁺CD8⁺ T cell population (Fig. 1D), whereas the same T cells failed to upregulate CCR5 upon direct contact with the NDLN ex vivo or following coincubation with the DLN through a Transwell. To interrogate whether this direct DLN contact-mediated CCR5 upregulation was dependent on MHC expression, naive CCR5⁻CD45.1⁺CD8⁺ T cells were injected into C57BL/6, MHC-II^−/−, or β₂-microglobulin^−/− mice. We found that expression of either MHC class I or MHC-II was not necessary for the rapid induction of surface CCR5 on CD8⁺ T cells (Fig. 1E).

To further investigate the cellular mechanisms responsible for this transient surface CCR5 expression in vivo, we injected increasing numbers of OT-I T cells in vivo to see whether the CCR5⁻ → CCR5⁺ conversion is a saturable and limiting event. When LNs were analyzed 4 h after adoptive transfer of CCR5⁻ OT-I T cells, we found that the percentage of CCR5⁺ T cells decreased dramatically in the DLN with an increasing number of T cell inocula (Fig. 1F), further suggesting that CCR5 expression was not caused by selective retention of contaminating CCR5⁺ donor T cells because we would expect the percentage of CCR5⁺ T cells to either remain the same or increase with increasing number of injected T cells. Instead, we observed that equivalent total numbers of CCR5⁺ donor T cells were found in the DLN regardless of the sizes of the T cell inocula (Fig. 1G). These data suggest that there were limited sites within the inflamed LN that were capable of inducing CCR5 expression in newly arrived T cells within the 4 h following the adoptive T cell transfer.

To identify factors that may influence surface CCR5 expression in newly arrived naive CD8⁺ T cells in the inflamed LN, we performed quantitative PCR analysis of selected target proteins that are important for initial T cell binding to the HEV within the inflamed DLN. Owing to the huge influx of immune cells into LNs during inflammation, we compared the gene expressions of our targeted genes with the expression of CD31 mRNA as a surrogate marker for endothelial cells. We examined CD34 and Glycam-1, two competing molecules that possess the same PNAd motif important in binding to CD62L on incoming naive T cells. We found that whereas the DLN exhibited a slight reduction in CD34 mRNA 48h after CpG/Alum stimulation, the level of mRNA for Glycam-1 was significantly reduced in the DLN relative to steady-state LNs as compared with NDLNs (100-fold reduction versus 2- to 3-fold reduction; Fig. 2A). As Glycam-1 exists as a soluble molecule that may inhibit T cell interaction with the HEV, a reduction in Glycam-1 may help facilitate binding of circulating T cells to HEV (8–10). Alternatively, the mRNA for CD54, which binds to activated CD11a/CD18 complexes on T cells and provides the final step for T cell binding to the HEV, was significantly increased in the DLN 48 h after CpG/Alum injection (Fig. 2A). Taken together, the molecular changes in the HEV during inflammation favors homing of circulating T cells to the DLN.

Next, we examined tissue sections of steady-state LNs as well as both DLNs and NDLNs for the expression of PNAd and CD54. Although we observed a reduction in mRNA for Glycam-1 and slight reduction in CD34 mRNA expression (Fig. 2A), we did not see a change in the overall PNAd expression in DLN HEV (Fig. 2B). The PNAd expression within the LN is reflective of CD34 expression, as Glycam-1 exists as a soluble form of PNAd and decreases during inflammation (21). We also observed an increase in CD54 staining along the HEV, in agreement with the increased mRNA abundance (Fig. 2A). We found that most CD54 found in the DLN was associated with PNAd⁺ regions (Fig. 2B). When we analyzed ROIs around PNAd⁺ areas, we observed a 2-fold increase in CD54 protein relative to PNAd in the DLN as compared with that in the NDLN (Fig. 2C).

Next, we examined whether these HEV-associated ligands may be responsible for the rapid surface CCR5 expression on incoming naive CCR5⁻CD8⁺ T cells. To test whether interactions with CD62L or CD11a on T cells can promote CCR5 expression, we cultured OT-I T cells with Ab-coated beads to cross-link either CD62L or CD11a to induce surface CCR5 expression as determined by flow cytometry. We found that both anti-CD62L– and anti-CD11a Ab–coated beads could induce CCR5 expression within the first 10 min after cross-linking (Fig. 2D). As ligation of CD62L promotes conformational change of CD11a/CD18, which allows for binding to and signaling through CD54, either direct activation through CD11a or cross-linking of CD62L is sufficient to induce surface CCR5 expression on naive CCR5⁻CD8⁺ T cells in vitro. This induction occurred rapidly and was dependent on the degree of stimulation (Fig. 2E). Not all naive CCR5⁻ cells were capable of expressing CCR5, as only a subset of cells that failed to express CCR5 after primary exposure to anti-CD11a–coated beads could upregulate CCR5 after a secondary exposure to anti-CD11a–coated beads (Fig. 2F).

CCR5 protein has been shown to exist in preformed intracellular vesicles in human T cells (22, 23). Therefore, we examined naive OT-I T cells for the presence of intracellular CCR5 protein using confocal microscopy. We injected naive surface CCR5⁻ OT-I T cells into mice that received footpad injection of CpG/Alum 48 h before and isolated NDLNs and DLNs 3 h later to detect CCR5 protein expression. Immunofluorescence showed that CCR5 could be found in vesicles below the surface of a subset of unstimulated naive OT-I T cells (Fig. 3A), with a baseline of ∼15% CCR5 signal colocalized with CD8 expression (Fig. 3B). Similarly, newly arrived transferred OT-I cells in the NDLN also exhibited ∼10% colocalization of CD8 and CCR5 proteins. The extent of CCR5 and CD8 colocalization was significantly increased to 30% in OT-I cells isolated from the DLN (Fig. 3A, 3B). By fixing and permeabilizing OT-I T cells, we were able to detect intracellular CCR5 protein expression in a small number of T cells as determined by flow cytometry (Fig. 3C). To interrogate whether increased surface CCR5 expression was caused by the mobilization of a pre-existing protein pool or through de novo synthesis, OT-I cells were incubated in the presence of inhibitors that block synthesis or transport of protein prior to incubation with anti-CD11a–coated beads (Fig. 3D, 3E). We found that surface CCR5 was still induced by CD11a cross-linking in the presence of either actinomycin D or cyclohexamide, albeit at reduced levels. The addition of brefeldin A or monensin, inhibitors of transport within the trans-Golgi network, similarly caused only a slight reduction in CCR5 expression after CD11a stimulation. We found that all treatments failed to completely eliminate the appearance of surface CCR5, supporting the notion that CCR5 was being transported to the surface, at least in part, via pre-existing vesicles and does not require de novo synthesis. This is in contrast to CD8⁺ T cells stimulated with anti-CD3/CD28, where these same inhibitors significantly reduced expression of the activation marker CD69 (Supplemental Fig. 3).

We have previously identified that the chemokines CCL3 and CCL4, which were produced in high concentrations locally at sites of activated DC–CD4⁺ T cell interactions in the DLN, were essential for the enhancement of the magnitude of Th cell–dependent memory CD8⁺ T cell responses (3). To test whether the ability of naive T cells to rapidly express surface CCR5 in the DLN can modulate the magnitude of Ag-specific memory CD8⁺ T cell response, naive CCR5⁻ OT-I T cells were added directly to exposed CpG/Alum-inflamed CD45.1⁺ LNs in vitro for 4 h prior to sorting the T cells into CD45.2⁺CD8⁺CCR5⁺ or CD45.2⁺CD8⁺CCR5⁻ fractions (Fig. 4A). OT-I T cells (1 × 10⁵) from each cohort were injected along with 3 × 10⁶ naive OT-II T cells into separate recipients (Fig. 4A). Twenty-four hours later mice were immunized with OVA_257–264 and CpG/Alum in the right footpad with or without OVA_323–339 to provide CD4⁺ T cell help. At 7 and 28 d after immunization, mice were sacrificed, spleens and LNs were removed, and the cells were restimulated with OVA_257–264 for 4 h in the presence of brefeldin A. The total number of OT-I T cells and the number of IFN-γ–producing OT-I cells in both the spleen and LN were enumerated. Although we did not find significant differences in the total number of OT-I cells (Fig. 4B) or in the number of IFN-γ–producing OT-I cells in the LN on day 7 (Fig. 4C), we observed significantly elevated total numbers of OT-I T cells on day 28 in the spleen and LN of mice that received CCR5⁺ sorted OT-I T cells when compared with those that received CCR5⁻ sorted OT-I T cells (Fig. 4D). Importantly, more IFN-γ–producing OT-I T cells were found in mice that received the naive CCR5⁺ OT-I cohort (Fig. 4E). The level of enhancement in memory OT-I cell number and functional recall in the CCR5⁺ sorted OT-I cohort was the same or slightly higher than bulk, unsorted naive OT-I cell transfer. Importantly, the magnitude of the total number of OT-I T cells and IFN-γ–producing OT-I T cells isolated from mice that received CCR5⁻-sorted OT-I T cells was similar to that found in mice receiving naive OT-I cells with homozygous germline CCR5 KO. This supports our earlier work that showed CCL3 and CCL4 signaling through naive CCR5⁺CD8⁺ T cells during the very early phase of a primary immune response is critical in augmenting the magnitude of the late CD8⁺ memory T cell recall response (3, 4). Naive CCR5⁺ OT-I T cell fraction did not exhibit increased CD69 or CD44 expression following in vitro exposure to the inflamed LN prior to sorting, indicating that OT-I T cells in the CCR5⁺-sorted population were not otherwise activated prior to transfer (Supplemental Fig. 4A, 4B). In further support of this, we used gene array analysis to examine expression of genes associated with development of memory phenotype in freshly isolated cells prior to adoptive transfer in vivo. We found that there was no significant difference in expression of genes in the sorted CCR5⁻ or CCR5⁺ T cells (Supplemental Fig. 4C, 4D), again confirming that CCR5 expression alone does not confer a memory phenotype and that our CCR5⁺ fraction did not contain contaminating endogenous memory T cells.

To gain additional insight into how early CCR5 expression on naive CD8⁺ T cells affects access to potential antigenic stimulation by LN DCs, we transferred differentially fluorescent-labeled naive wild-type and CCR5 KO OT-I T cells into mice that were footpad injected 24 h prior with fluorescent-labeled LPS and OVA_257–264-pulsed bone marrow–derived DCs. Seven hours later the DLNs were harvested and examined by static two-photon microscopy (Fig. 4F). The numbers of CD8⁺ T cell–DC contacts were determined for multiple LN samples (Fig. 4G). As expected, significantly more WT OT-I T cells were able to gain access and form contacts with Ag-bearing DCs as compared with cotransferred naive CCR5 KO OT-I T cells, suggesting a distinct advantage for the CCR5⁺ fraction of the naive T cells to have early and ready access for priming by newly arrived Ag-bearing DCs from the periphery.

Under steady-state, CD4⁺ T cells and CD8⁺ T cells exhibit distinct LN surveillance behavior, with CD8⁺ T cells traversing the LN almost twice as long as that of CD4⁺ T cells (∼18–21 versus ∼12 h, respectively) (24). During their LN transit, CD8⁺ T cells were estimated to interact with an average of ∼313 APCs compared with ∼160–200 APC contacts made by CD4⁺ T cells (24). In generating a primary immune response, therefore, the adaptive immune system must rely on additional biochemical or physical cues to facilitate the necessary physical contacts between the rare Ag-bearing APCs and the even rarer Ag-specific T cells to combat the offending agent in a timely manner. We have previously identified that inflammatory chemokines, CCL3 and CCL4, are central in enhancing this Ag surveillance by naive polyclonal CD8⁺ T cells to sites of productive CD4⁺ T cell–DC interactions during the early LN priming phase (3). The ability of these naive T cells to use the CCL3/CCL4–CCR5 chemokine axis correlates with the magnitude of helper CD4⁺ T cell–enhanced memory CD8⁺ T cell generation. Hugues et al. (4) subsequently confirmed this critical chemokine signaling axis by demonstrating that productive Ag-specific CD8⁺ T cell–DC interaction also enhances polyclonal CD8⁺ T cell surveillance of the Ag-bearing DCs, again in a CCL3/CCL4–CCR5-dependent manner. These data imply that although not expressed on the surface of naive T cells in circulation or under noninflamed states, CCR5 is functional in at least a fraction of naive CD8⁺ T cells in the inflamed LN (13). In the present study, we presented molecular mechanisms that shed new insights into how naive CD8⁺ T cells use CCR5 to accomplish this feat in vivo and in vitro.

We found that a fraction of naive cells among both polyclonal and TCR transgenic CD8⁺ T cells could be induced rapidly to express CCR5 on the cell surface after entering an inflamed LN that is induced by TLR agonists (CpG and LPS; Fig. 1, Supplemental Fig. 1) or TLR-activated DC vaccines (data not shown). This expression was transient and reached a maximum density between 2 and 4 h upon T cell arrival in the inflamed LN before returning to basal levels by 24 h (Fig. 1C and data not shown). This rapid induction of surface CCR5 expression does not depend on TCR engagement with either cognate/MHC or self-peptide/MHC complexes, as similar magnitude of CCR5 upregulation was observed in MHC class I–deficient hosts as compared with wild-type or MHC-II–deficient hosts (Fig. 1E). Neither does this process require nascent CCR5 gene transcription or protein synthesis, as cyclophosphamide and actinomycin treatments failed to completely inhibit CCR5 protein expression on the naive CD8⁺ T cell surface (Fig. 3D). These observations agree well with previously published intravital imaging reports in which the chemotactic behavior of naive CD8⁺ T cells toward the CCL3/CCL4 gradient was evident in a small fraction (<5%) of recently arrived naive CD8⁺ T cells in LNs that were undergoing productive cognate Ag-mediated DC–CD4⁺ T cell or DC–CD8⁺ T cell interactions (3, 4). As the duration of surface CCR5 expression lasts only for a few hours on naive CD8⁺ T cells (Fig. 1C), and as ligand-mediated receptor downregulation further reduces CCR5 expression in the absence of cognate peptide/MHC–TCR engagement (25), the relative proportion of naive CD8⁺ T cells that are capable of responding to local CCL3/CCL4 gradient would be expected to diminish over time. Furthermore, during inflammation, HEVs undergo structural changes that favor increased recruitment of T cells from the circulation (Fig. 2B). As CD8⁺ T cells have prolonged intranodal transit time (∼18–21 h) as compared with that of CD4⁺ T cells (∼12 h), these data suggest that, within the available CD8⁺ T cell pool, the fraction of naive CD8⁺ T cells capable of responding to CCR5 ligands would decrease as LN inflammation ensues over time. These data also predict that the highest impact of CCL3/CCL4–CCR5 signaling on the intranodal naive CD8⁺ T cell fate resides in the early phases of an inflammation-induced primary LN response.

We found similar maximum numbers of donor CD44^loCD62^hiCCR5⁺CD8⁺ T cells that were acutely converted from a CCR5⁻ state in the DLN irrespective of the sizes of transferred T cell inocula (Fig. 1G), suggesting that CCR5 expression is not intrinsically uniform in every naive CD8⁺ T cell arriving in the LN from the general circulation, but is rather promoted on a specific subset of CD8⁺ T cells that have access to limited interaction sites within the LN. Our in vitro studies show a requirement for direct contact with the internal contents of an inflamed LN, further supporting the idea that CCR5 is being induced in CCR5⁻ naive CD8⁺ T cells in anatomically distinct sites within the LN (Fig. 1D). These intriguing data, therefore, beg the question: What is the unique feature of the inflamed LN environment that could promote CCR5 expression in naive CD8⁺ T cells in the absence of cognate Ag/MHC–TCR engagement?

Three sequential steps are necessary for T cell entry into the LN through the HEV (Fig. 5). The first step involves the rolling and tethering of the T cell via CD62L interaction with a group of sialomucins on the HEV that contains the PNAd motif. The PNAd motif is required for binding of l-selectin and is expressed in a few proteins, including Glycam-1, CD34, Sgp200, and on a subset of mucosal addressin cell adhesion molecule-1 (10). The second step is ligation of chemokine receptor CCR7 on T cells with luminal CCL21 on HEV. Whereas both CCL19 and CCL21 bind to CCR7, only CCL21 contains a hydrophobic region that forms complexes with heparan sulfate, gp38, and collagen IV on the surface of the HEV to enhance rolling of T cells (11, 13). Binding of both CD62L and CCR7 on T cells is important, because inhibition of either pathway greatly reduces T cell adhesion and transendothelial migration through the HEV (26–28). Whereas interaction between CCR7 and CCL21 results in a strong interaction between the T cell and HEV, signals from both CD62L and CCR7 are capable of promoting a conformation change of CD11a/CD18 on the T cell through inside–out signaling (29). This conformational change of CD11a/CD18 allows for the third step to take place, which is to allow a strong binding of T cells to CD54 on the HEV. This final step results in outside–in signaling in T cells, allowing for transmigration of T cells into the LN through the HEV (30, 31).

Although CD34 mRNA expression did not change significantly in the inflamed LN, Glycam-1 dramatically decreased by >100-fold at 48 h after activation of inflammation. Both CD34 and Glycam-1 contain the PNAd motif; however, Glycam-1 is secreted and has been postulated to inhibit LN homing of T cells by competing with CD34 for binding to CD62L in the HEV (8–10). Elevated plasma levels of Glycam-1 have been observed shortly after inflammation induction, and the Glycam-1 level decreases after 12 h (21). This suggests that the loss of soluble Glycam-1 production during inflammation enhances effective CD62L-mediated homing of circulating naive T cells through binding to CD34 on the HEV. Concurrently, CD54 protein expression dramatically increased in PNAd⁺ regions of the HEV 48 h after stimulation, which further enhances homing of circulating naive CD8⁺ T cells to the inflamed LN. LN inflammation induced by either CpG oligodeoxynucleotide or LPS can promote surface CCR5 expression in naive CD8⁺ T cells. These pathogen-associated molecular patterns are known to promote the release of inflammatory cytokines such as TNF-α and IL-6, and they could upregulate CD40 expression in DCs (32). IL-6 has also been shown to increase CD54 expression in PNAd⁺ areas in the HEV (33, 34). Together, the molecular changes within the PNAd⁺ regions of the HEV result in preferential T cell recruitment to the inflamed LN and enhance CCR5 expression on T cells that arrive in such an LN.

Engagement of TCR and CD11a/CD18 results in the binding of T cells to target cells as well as the polarization and release of lytic granules and cytokine-containing granules (35–38). In this context, CD11a/CD18 plays an important role in directing the granules toward the target cell (39). Whereas wild-type OT-I T cells transferred into CD54^−/− mice exhibit normal activation, proliferation, and effector function acquisition upon primary Ag stimulation, they are unable to respond to Ag rechallenge (40). Using bone marrow chimera approaches, it has been shown that memory CD8⁺ T cells had reduced ability to respond to secondary Ag challenge in mice lacking CD54 in nonhematopoietic cells (41). These data further support our present observation that CD54 expression in the HEV contributes to memory CD8⁺ T cell development.

Chemokine receptors are membrane-bound molecules composed of seven transmembrane-spanning domains, which are coupled to G proteins (12). These receptors bind to specific chemokines and promote cellular migration by chemotaxis. Although G protein–coupled receptors are mostly exported directly to the plasma membrane, their intracellular trafficking pathway may be modulated by specific proteins (42). In particular, CCR5 has been shown to reside in intracellular vesicles associated with the CD4 molecule in primary human T lymphocytes (22, 23). Ligation of CD62L either by Ab cross-linking or by using the CD62L ligands (fucoidan or sulfatide) can promote translocation of intracellular CXCR4 to the cell membrane surface in human T cells presumably from intracellular stores (43). In the present study, we observed surface CCR5 expression in mouse CD8⁺ T cells in as little as 10 min after cross-linking with Ab against either CD62L or CD11a. The exact molecular signaling responsible for translocation of intracellular CCR5 to the cell membrane remains to be studied.

Although some naive polyclonal CD8⁺ T cells express surface CCR5 in the inflamed LN, not all naive T cells do. This holds true for naive CD8⁺ T cells from either wild-type animals or TCR transgenic animals on a Rag2^−/− background. Most strikingly, exposure of this naive CCR5⁺ T cell cohort to cognate Ag was able to recapitulate the same magnitude of Th cell–enhanced memory CD8⁺ T cell generation, whereas exposure of the CCR5⁻ T cell cohort to the same cognate Ag produced long-term memory CD8⁺ T cells to the same extent as observed in germline CCR5-deficient CD8⁺ T cells (Fig. 4). An insight into this phenomenon came from the observation that the CCR5⁻ T cell cohort has a reduced ability to be induced into CCR5⁺CD8⁺ T cells upon additional inflammatory exposure (Fig. 2F). These data argue for differential epigenetic or metabolic regulation of thymic-derived CD8⁺ T cells in deciding which T cell subpopulations are better endowed to be precursor memory T cells. It also supports the notion that the presence of CCR5 alone is not associated with constitutive expression of genes required for the development of memory CD8⁺ T cells (Supplemental Fig. 4). In this context, CCR5 serves as a target for cell entry by HIV. Based on the present study, we speculate that the early CCR5⁺ converters among naive CD8⁺ T cells (hence the robust memory-forming subpopulation) in the inflamed LN may be targeted by the virus for elimination during early HIV infection, leading to the observed clinical development of robust anti-HIV effector T cells with subsequent collapse of the memory T cell pool (44–47). More studies are needed to understand the genetic and epigenetic regulation of the early CCR5⁺ converters among the naive T cell pool to fully understand the intriguing observations in the present study.

We thank Dr. Matthias Mack for providing the anti-CCR5 Ab, MC-68.

The authors have no financial conflicts of interest.