Naïve and Memory CD4 T Cells Differ in Their Susceptibilities to Human Immunodeficiency Virus Type 1 Infection following CD28 Costimulation: Implications for Transmission and Pathogenesis

Abstract

In vitro evidence suggests that memory CD4⁺ cells are preferentially infected by human immunodeficiency virus type 1 (HIV-1), yet studies of HIV-1-infected individuals have failed to detect preferential memory cell depletion. To explore this paradox, we stimulated CD45RA⁺ CD4⁺ (naïve) and CD45RO⁺ CD4⁺ (memory) cells with antibodies to CD3 and CD28 and infected them with either CCR5-dependent (R5) or CXCR4-dependent (X4) HIV-1 isolates. Naïve CD4⁺ cells supported less X4 HIV replication than their memory counterparts. However, naïve cells were susceptible to R5 viral infection, while memory cells remained resistant to infection and viral replication. As with the unseparated cells, mixing the naïve and memory cells prior to infection resulted in cells resistant to R5 infection and highly susceptible to X4 infection. While both naïve and memory CD4⁺ subsets downregulated CCR5 expression in response to CD28 costimulation, only the memory cells produced high levels of the β-chemokines RANTES, MIP-1α, and MIP-1β upon stimulation. Neutralization of these β-chemokines rendered memory CD4⁺ cells highly sensitive to infection with R5 HIV-1 isolates, indicating that downregulation of CCR5 is not sufficient to mediate complete protection from CCR5 strains of HIV-1. These results indicate that susceptibility to R5 HIV-1 isolates is determined not only by the level of CCR5 expression but also by the balance of CCR5 expression and β-chemokine production. Furthermore, our results suggest a model of HIV-1 transmission and pathogenesis in which naïve rather than memory CD4⁺ T cells serve as the targets for early rounds of HIV-1 replication.

Human immunodeficiency virus type 1 (HIV-1) infection is accompanied by depletion of CD4⁺ T lymphocytes and progressive loss of immune function (26). CD4⁺ T lymphocytes are a heterogeneous population, and controversy exists as to whether HIV-1 targets particular CD4⁺ subtypes for elimination (18, 53). In part, this controversy has centered on whether naïve or memory CD4⁺ cell subsets are preferentially depleted by HIV-1. Naïve CD4⁺ T lymphocytes have no previous antigen exposure; exposure to the cognate antigen is followed by proliferation and the acquisition of effector functions. A subset of the activated cells reverts to a resting state, at which point they are termed memory cells (61). Phenotypically, naïve cells are CD45RA⁺ CD45RO⁻ and respond to mitogenic stimuli with a greater calcium flux and proliferative ability, while memory cells are CD45RO⁺ CD45RA⁻ and have a much broader cytokine expression profile (7). Naïve cells are found almost exclusively in the secondary lymph organs, while memory cells have a much wider tissue distribution. These differing distributions are thought to be due to the higher level of adhesion molecule expression on memory cells (41).

In vitro, memory cells are more efficiently infected by HIV-1 (31, 55, 58, 60, 67) and they are more susceptible to HIV-induced cytopathic effects (15, 70). However, most studies of HIV-1 seroconverters either demonstrate no specific depletion of either subtype (14, 29, 42, 50, 51, 62) or indicate specific exhaustion of naïve cells (5, 6, 54). A major limitation of the in vitro studies is the almost exclusive use of CXCR4-dependent (X4) viruses. X4 viruses, also known as syncytium-inducing or T-cell-line-tropic viruses, use the α-chemokine receptor CXCR4 as a coreceptor (27). CXCR4-dependent viruses appear late in the course of HIV infection and they are more cytopathic than the CCR5-dependent (R5) viruses (21). R5 viruses, also known as non-syncytium-inducing or macrophage-tropic viruses, use CCR5 for a coreceptor (3, 13, 23–25). R5 viruses are essential for transmission and predominate during the early, asymptomatic phase of infection (45, 68). Thus, the virus isolates critical for transmission (R5 viruses) have been rarely used in in vitro acute infection model systems described to date.

While coreceptor expression is required for viral entry into CD4⁺ cells, productive HIV infection requires cellular activation and entry into the G₁b phase of the cell cycle (35, 69). T-cell activation and proliferation require at least two signals (9). Antigen presented in the context of major histocompatibility complex class II provides the first signal by triggering the T-cell receptor–CD3 complex. Delivery of a costimulatory signal is accomplished through ligation of the CD28 coreceptor on the CD4⁺ cell surface (33). Previously, we have shown that anti-CD3/CD28 stimulation results in exponential, polyclonal T-cell growth (37, 38). Furthermore, it renders the cells resistant to infection with R5 HIV isolates. This HIV-resistant state results from the increase in expression of the native CCR5 ligands (RANTES, MIP-1α, and MIP-1β) and the concomitant downregulation of CCR5 expression (12, 52). In this report, we sought to examine the HIV susceptibilities of naïve and memory cells activated by either CD3/CD28 costimulation or by mitogenic lectins. We report that susceptibility to R5 viruses is not governed solely by the level of CCR5 expression, but rather by the balance between CCR5 expression and β-chemokine expression. Together, these findings suggest a new mechanism concerning the role of coreceptor expression in HIV transmission and pathogenesis.

MATERIALS AND METHODS

Antibodies.

The following purified and azide-free antibodies were used for cell purification: anti-CD8 OKT8 (immunoglobulin G2a [IgG2a]), anti-CD11b OKM1 (IgG2b), anti-CD14 63D3 (IgG1), anti-CD16 3G8 (IgG1), anti-CD20 1F5 (IgG2a), and anti-HLA-DR 2.06 (IgG1). All of the hybridomas were obtained from the American Type Culture Collection, Manassas, Va., except 3G8, which was a kind gift from Stephen Shaw (National Institutes of Health, Bethesda, Md.). CD45RA and CD45RO monoclonal antibodies were obtained from Caltag (Burlingame, Calif.). Anti-CD3 (OKT3 [IgG2a]) (36) and anti-CD28 (9.3 [IgG2a]) (30) were used for cell stimulations. β-Chemokine-neutralizing antibodies (NAbs) were purchased from R&D Systems (Minneapolis, Minn.).

Cell separation and stimulation.

Peripheral blood lymphocytes were isolated by Percoll (Pharmacia Biotech, Uppsala, Sweden) gradient centrifugation of leukopacks obtained by apheresis of healthy donors. CD28⁺ CD4⁺ T cells were purified by negative selection using magnetic beads (Dynal, Lake Success, N.Y.) as described previously (34). Purified CD28⁺ CD4⁺ T cells, >98% CD3⁺, >98% CD28⁺, and <3% CD8⁺ as judged by flow cytometry, were separated into CD4⁺ CD45RO⁺ and CD4⁺ CD45RA⁺ subsets by negative selection as previously described (39). These preparations were routinely >95% pure. Recent reports indicate that naïve cells are further enriched by selecting for CD45RA⁺ CD62L⁺ cells (47). However, the donor cells used in this study were routinely <3% CD45RA⁺ CD62L⁻, obviating the need for further purification. Cells were cultured at 10⁶/ml in complete medium RPMI 1640 (Bio Whittaker, Walkersville, Md.) supplemented with 10% fetal bovine serum (Hyclone, Logan, Utah)–2 mM l-glutamine (Bio Whittaker)–20 mM HEPES (Bio Whittaker). Cells were stimulated by OKT3 and 9.3, which were covalently attached to magnetic beads (tosyl-activated M-450; Dynal) at ∼150 fg per bead (38) and added to cells in a 1:1 ratio. Alternatively, cells were stimulated with 5 μg of phytohemagglutinin (PHA) (Sigma Chemical Co., St. Louis, Mo.) and 100 U of interleukin 2 (IL-2) (Boehringer Mannheim, Indianapolis, Ind.) per ml. Cell volume was monitored on a Coulter Counter model ZM (Coulter, Hialeah, Fla.). Cells were fed at 2- to 3-day intervals and maintained at a concentration of 10⁶ to 2 × 10⁶ cells/ml.

Flow cytofluorometric analysis.

Samples were stained with anti-CD45RA–fluorescein isothiocyanate, anti-CD45RO–phycoerythrin (PE) (Becton Dickinson, San Jose, Calif.), anti-CXCR4–PE, and anti-CCR5–PE (Pharmingen, San Diego, Calif.) for 30 min at 4°C. After being washed in phosphate-buffered saline–0.01% sodium azide–0.05% bovine serum albumin (wash buffer), stained cells were fixed at 4°C with 0.1% paraformaldehyde. Anti-CCR5- and anti-CXCR4-stained cells were analyzed fresh without fixation. Samples were analyzed by flow cytometry on a Coulter Epics Elite after gating on live lymphocytes based on a standard light scatter histogram (integral forward scatter versus log 90°).

Acute infection and PCR/liquid hybridization procedures.

Six days poststimulation, cells were infected with either HIV-1_US-1 (43) or HIV-1_NL4-3 (1) as previously described (12, 52). Antibody-coated beads were removed immediately prior to the start of the infection. For each infection, 5 × 10⁶ cells were resuspended in 400 μl of 50% conditioned medium containing 10⁴ to 3 × 10⁴ 50% tissue culture infective doses of HIV-1. The cells were incubated at 37°C for 2 h, washed three times in complete medium to remove excess virus, and resuspended in 50% conditioned medium at 10⁶/ml. Where indicated, NAbs to the β-chemokines were preincubated with conditioned medium for 2 h before the start of the infection. The anti-RANTES NAb was used at 100 μg/ml, while anti-MIP-1α and anti-MIP-1β NAbs were used at a final concentration of 50 μg/ml. Goat IgG (Sigma) was used as a control antibody at a final concentration of 200 μg/ml. Antibody concentrations were maintained for the duration of the experiment. At designated time points, 10⁶ cells were pelleted by centrifugation and frozen at −70°C. The cell pellets were lysed and amplified by PCR using HIV gag-specific primers, and the amplified sequences were detected by hybridization to a radiolabelled internal probe (64, 65). The hybridized products were resolved by electrophoresis on 10% polyacrylamide gels, exposed to a phosphorimager screen for 1 h, and developed on a PhosphorImager 445 SI (Molecular Dynamics, Sunnyvale, Calif.). To ensure that the reactions were performed within the linear range of the assay, log increments of HIV gag plasmid standards were amplified at the same time (data not shown). Human β-globin sequences were PCR amplified to demonstrate that equivalent levels of input DNA were present in each PCR reaction mixture (64, 65). Figures were generated with ImageQuant software (Molecular Dynamics).

Chemokine measurements.

Levels of MIP-1α, MIP-1β, and RANTES in cell supernatants were measured with enzyme-linked immunosorbent assay kits from R&D Systems according to the manufacturer’s instructions.

Chemokine receptor RT-PCR assay.

Total RNA was isolated from cells with RNA STAT-60 (Tel-Test, Friendswood, Tex.), and cDNA was synthesized with the StrataScript reverse transcriptase PCR (RT-PCR) kit (Stratagene, La Jolla, Calif.). cDNA products were diluted in H₂O to predetermined optimal concentrations (1:3 for CCR5, 1:300 for CXCR4) and amplified by using the following program: 95°C for 30 s, 55°C for 30 s, and 72°C for 90 s (25 cycles). For CCR5-specific amplifications, the following primers were used: CCR5-42 (5′-GGG TGG AAC AAG ATG GAT TAT CAA GTG TCA-3′) and CCR5-640 (5′-ATG TCT GGA AAT TCT TCC AGA ATT GAT ACT-3′). For CXCR4-specific amplifications, the following primers were used: CXCR4-489 (5′-CCA CCA ACA GTC AGA GGC CAA GGA AGC TGT-3′) and CXCR4-1122 (5′-TCT GTG TTA GCT GGA GTG AAA ACT TGA AGA-3′). A portion of the PCR reaction mixture was hybridized as described previously (64) with end-labeled oligonucleotide probes specific for CCR5 (CCR5-81 [5′-GGG CTC CGA TGT ATA ATA ATT GAT GTC ATA-3′]) or CXCR4 (CXCR4-840 [5′-CCA GGA GGA TGA AGG AGT CGA TGC TGA TCC-3′]). The hybridized products were separated on 6% polyacrylamide gels, exposed to phosphorimager screens overnight, and developed on a PhosphorImager 445 SI (Molecular Dynamics). Figures were generated with ImageQuant software (Molecular Dynamics). To compare levels of 18S and 28S rRNA for each RNA sample, 1 μg was electrophoresed on a 1.5% denaturing agarose gel (56).

RESULTS

Isolation and differentiation of CD3/CD28-stimulated naïve and memory CD4⁺ cells.

CD4⁺ cells from healthy donors were separated into naïve and memory populations based on CD45 isoform expression. The cells were negatively selected to avoid any aberrant activation caused by the selection process and were routinely >95% pure (Fig. 1). The cells were judged to be resting based on a small mean cell volume and the absence of HLA-DR or CD25 expression (data not shown). Following stimulation with anti-CD3/anti-CD28-coated magnetic beads, both naïve (CD45RA⁺ CD45RO⁻) and memory (CD45RO⁺ CD45RA⁻) populations proliferated equivalently for the first 14 days (data not shown), although over longer time periods, naïve cells outgrow their memory siblings (66). Both subsets at least tripled their cell volumes following stimulation, but the memory cells consistently remained slightly larger than the naïve cells (data not shown). Upon activation, naïve cells convert from a CD45RA⁺ CD45RO⁻ phenotype to a CD45RA⁻ CD45RO⁺ phenotype (17, 59). Two-color fluorescence-activated cell sorter analysis revealed that 6 days poststimulation, 95% of the CD45RA⁺ CD45RO⁻ cells had become CD45RA⁻ CD45RO⁺ (Fig. 1). However, as indicated below, these cells remained functionally distinct from the original CD45RA⁻ CD45RO⁺ population.

FIG. 1.

Isolation and differentiation of naïve and memory CD4⁺ cells after CD3/CD28 costimulation. CD4⁺ T cells were isolated and fractionated by negative selection into CD45RO⁺ and CD45RA⁺ subsets. Two-color cytofluorometric analysis of naïve and memory cells was performed after separation (day 0) and after stimulation with CD3/CD28 for 6 days. Data shown are representative of at least three experiments.

CD3/CD28-stimulated naïve cells are susceptible to R5 HIV-1 infection.

The susceptibilities of naïve and memory CD4⁺ T lymphocytes to HIV-1 infection were examined by infecting cells 6 days poststimulation with either R5 or X4 HIV-1 isolates. PHA/IL-2-stimulated naïve and memory cells were robustly infected with X4 HIV-1 (Fig. 2A). Similarly, both CD4 cell subsets were infected with R5 virus after PHA/IL-2 stimulation (Fig. 2B). HIV gag DNA was detected 72 h postinfection, and a spreading infection was revealed by an analysis of later time points in both subsets of cells. Similarly, when PHA/IL-2-stimulated naïve and memory CD4 cells were combined immediately prior to infection, the pooled population was also sensitive to both R5- and X4-dependent viruses.

FIG. 2.

Distinct R5 and X4 replication kinetics in CD4 subsets. Purified CD4⁺ (CD4), CD45RO⁺ (RO), and CD45RA⁺ (RA) cells were stimulated for 6 days with either PHA/IL-2 or CD3/CD28 and infected with either HIV_NL4-3 (CXCR4-dependent) (A) or HIV_US-1 (CCR5-dependent) (B) strains of HIV-1 as described in Materials and Methods. Samples were taken at 0, 2, 72, and 144 h postinfection, and cell pellets were analyzed for gag DNA by a quantitative PCR assay using liquid hybridization. Data shown are representative of four experiments.

CD3/CD28-stimulated naïve and memory cell populations, as well as pooled naïve and memory cells, were susceptible to infection with CXCR-4-dependent viruses (Fig. 2A). However naïve CD3/CD28-stimulated cells were much less permissive to R4-dependent infection than memory cells, in accordance with previous reports (15, 31, 58, 60, 67). In marked contrast, CD3/CD28-stimulated naïve and memory cell populations infected with R5 viruses experienced two entirely different outcomes (Fig. 2B). Surprisingly, naïve cells supported a weak R5-dependent virus infection, while memory cells remained completely resistant to infection with R5-dependent isolates. As previously reported with unfractionated CD3/CD28-stimulated CD4⁺ cells (12, 52), pooling of CD3/CD28-stimulated naïve and memory cells before infection with R5-dependent HIV isolates generated a completely resistant cell population. Thus, when CD3/ CD28-stimulated naïve CD4⁺ cells are separated from memory cells, they become susceptible to low-level R5 virus infection. Furthermore, this suggested that the previously reported absence of infection in the unfractionated CD4 cells was most likely the result of the protective effect conferred on naïve cells by memory cells.

A quantitative analysis of HIV-1 gag DNA accumulation for the experiment discussed above is shown in Fig. 3. After CD3/CD28 stimulation only the naïve CD4⁺ cell subset was susceptible to infection with R5 isolates of HIV. The kinetics of infection for the RA subset were delayed compared to those for PHA/IL-2 stimulated cells, while little or no HIV was detected in the memory subset and the recombined CD4 cell population. The memory and recombined CD4⁺ cells supported a high-level infection with X4 virus after CD3/CD28 costimulation; however, the naïve cells produced 10- to 50-fold-less gag DNA than the other populations. In contrast, both the X4 (NL4-3)- and R5 (US-1)-dependent viruses produced high-level infections of CD4 cell subsets after PHA stimulation. Together, the results shown in Fig. 2 and 3 further indicate that infection and replication of HIV-1 in primary CD4 cell subsets are critically dependent on the strain of HIV used as well as the form of T-cell activation.

FIG. 3.

Quantitative analysis of DNA HIV-1 gag content from the experiment shown in Fig. 2. The numbers of copies per 100,000 cells were determined by using plasmid standards that contained gag sequences, and these values were normalized to β-globin values. The top panel shows the infection data from Fig. 2B, and the bottom panel shows the infection data from Fig. 2A.

Coreceptor expression levels do not correlate with susceptibilities of costimulated CD4⁺ cell subsets to R5 HIV-1 isolates.

The differential susceptibilities of CD3/CD28-stimulated naïve CD4⁺ cells to R5 and X4 HIV-1 isolates suggested that in these cells, CXCR4 expression was decreased and CCR5 expression was enhanced compared to the memory cell population. Previously, we showed that CD3/CD28 costimulation inhibited CCR5 mRNA accumulation but not CXCR4 mRNA expression in unfractionated CD4⁺ cell populations (12). To examine in more detail the steady-state CXCR4 and CCR5 transcript levels in naïve and memory cell populations, a semiquantitative chemokine receptor RT-PCR assay was developed as described in Materials and Methods. Memory and naïve cells were activated with PHA/IL-2 or CD3/CD28 for 6 days, the sixth day having been chosen to coincide with the day of infection for the above experiments, and coreceptor expression was examined by RT-PCR. Twofold dilutions of the RT product were tested to ensure that the PCR was in the linear range. CXCR4 mRNA was readily detectable in both PHA/IL-2- and CD3/CD28-stimulated cells (Fig. 4). Marked differences in CXCR4 mRNA were not observed in the resting and activated CD4 cell subsets, consistent with the previously reported constitutive expression of this coreceptor (4, 8, 28). However, after activation, transcript levels were lower in the CD3/CD28-stimulated naïve population (by approximately fourfold) than in CD3/CD28-stimulated memory cells, consistent with the infection and viral replication data (Fig. 2 and 3).

FIG. 4.

Chemokine receptor expression in CD4 cells, and in CD45RA⁺ (RA) and CD45RO⁺ (RO) subpopulations. Total RNA was isolated from purified CD4⁺, CD45RO⁺, and CD45RA⁺ cells that were either resting (day 0), or were stimulated with PHA/IL-2 or with CD3/CD28 immunobeads for 6 days. (A) cDNA was synthesized and diluted to the optimal level (1:3 for CCR5 and 1:300 for CXCR4), and either 2.5, 5, or 10 μl of the RT product was used in the subsequent PCR and liquid hybridization reaction. NoRT indicates a control lane where 10 μl of an RT product was run with reverse transcriptase omitted. Data shown are representative of three experiments. (B) One microgram of total RNA for each sample was electrophoresed on a 1.5% denaturing agarose gel.

In contrast to those of CXCR4, CCR5 mRNA levels varied widely between CD4 cell subsets and as a consequence of the method of cellular activation (Fig. 4). Low levels of CCR5 mRNA were detected in resting CD4⁺ cells, and these transcripts were largely confined to the memory cell population. PHA/IL-2 stimulation resulted in a large increase in steady-state CCR5 transcript levels in both memory and naïve cell populations, as predicted by the infection data. Following stimulation with CD3/CD28 for 6 days, the levels of CCR5 mRNA decreased to below that of resting cells. Trace levels of CCR5 mRNA were detected in memory cell populations, but, surprisingly, CCR5 mRNA levels were below the limit of detection in naïve cells, despite their sensitivity to R5 viruses. Thus, CCR5 mRNA levels in naïve and memory cells did not correlate with sensitivity to infection.

The discrepancy between CCR5 mRNA levels and susceptibility to R5 HIV-1 isolates was further explored by examining CCR5 and CXCR4 surface expression in naïve and memory cell populations (Table 1). CXCR4 expression was much more prominent on resting cells, and preferential expression on naïve cells was observed, consistent with previous reports (8, 28). PHA/IL-2 stimulation resulted in an upregulation of CXCR4 expression levels as judged by the mean fluorescent intensities of all subsets (data not shown), although the fraction of cells expressing CXCR4 changed only modestly. CD3/CD28 stimulation resulted in much less surface CXCR4 expression of all CD4⁺ cells, and this was most pronounced in naïve cells. In the four donors examined after CD3/CD28 costimulation, 38% (standard error of the mean [SEM] = 2%) of the memory cells were expressing CXCR4 whereas only 21% (SEM = 4%) of the naïve cells were expressing CXCR4. This suggests a possible explanation for why naïve cells are less susceptible to infection by X4 viruses.

TABLE 1.

Effects of activation on HIV-1 coreceptor cell surface expression on CD4 cell subsets^a

Cell condition (day of staining)	% of cells with phenotype:
	CD4+ expressing:		CD4/45RA+ expressing:		CD4/45RO+ expressing:
	CXCR4	CCR5	CXCR4	CCR5	CXCR4	CCR5
Resting (0)	81	5	75	2	68	9
PHA/IL-2 stimulated (6)	77	10	87	6	87	28
CD3/28 stimulated (6)	32	1	29	1	41	1

^aCD4 T cells and the CD45RA⁺ (CD4/45RA⁺) and CD45RO⁺ (CD4/45RO⁺) subsets were stained for CXCR4 or CCR5 expression immediately after isolation or after culture for 6 days with PHA/IL-2 or with anti-CD3/CD28-coated beads as described in Materials and Methods. Values of 2% or less are considered background. These data are generated from one donor and are representative of data from four others. 

Resting CD4⁺ cells expressed low surface levels of CCR5, and this expression was confined to the memory subset, in agreement with recent observations by Bleul et al. (8). As predicted by the infection and RNA data, CCR5 expression increased in both memory and naïve-cell subsets after PHA/IL-2 stimulation. In contrast, CD3/CD28 costimulation resulted in the reduction of CCR5 expression to background levels in both memory and naïve-CD4 cell subsets, consistent with the mRNA data. In concurrence with the mRNA expression data, CD3/CD28-stimulated naïve cells, though susceptible to R5 viruses, were negative for CCR5 expression. Thus, the levels of CCR5 coreceptor expression cannot account for the differential susceptibilities of CD3/CD28-stimulated naïve and memory cells to R5 HIV-1 isolates.

CD3/CD28 costimulation does not trigger equivalent β-chemokine production in memory and naïve CD4⁺ cells.

The β-chemokines MIP-1α, MIP-1β, and RANTES, the native ligands for CCR5 (57), block infection of susceptible cells by R5 HIV-1 isolates (19). In human lymphocytes, CD28 stimulation upregulates RANTES promoter activity (46), and in mouse cells, CD28 stimulation is essential for MIP-1α but not RANTES secretion (32). Additionally, we have demonstrated that unfractionated CD4⁺ cells generally produce 50- to 100-fold more of these CCR5 ligands in response to CD3/CD28 costimulation than after PHA/IL-2 stimulation (52). The discordance between CCR5 expression and HIV susceptibility in CD3/CD28-stimulated naïve and memory cells prompted us to examine β-chemokine production in these cell subsets 6 days post-CD3/CD28 stimulation (Table 2). Although the absolute levels of MIP-1α, MIP-1β, and RANTES varied between donors, in every instance, the accumulation of β-chemokines in the supernatants after costimulation was substantially higher in memory cells than in naïve cells. As yet we do not know whether the differences in the magnitude of secretion between samples reflects donor-to-donor variation or differential contamination of naïve-cell populations with memory cells. Previously, it was noted that naïve CD4⁺ and CD8⁺ cells stimulated by anti-CD3 and phorbol ester plus ionomycin produce significantly less β-chemokines than similarly stimulated memory cells (20). Thus, in subsets of CD3/CD28-stimulated cells, resistance to infection by R5 HIV-1 isolates correlates not with CCR5 expression but rather with CCR5 ligand (β-chemokine) production.

TABLE 2.

β-Chemokine secretion after CD3/CD28 costimulation in CD4 cell subsets^a

Donor and β-chemokine	β-Chemokine production (pg/ml) by:
	CD4+ cells	CD4/45RA+ cells	CD4/45RO+ cells
Donor 1
RANTES	19,768	4,545	32,899
MIP-1α	93,775	33,471	215,000
MIP-1β	196,000	31,339	270,000
Donor 2
RANTES	22,935	975	54,773
MIP-1α	180,000	682	180,000
MIP-1β	468,000	6,637	484,000

^aCD4⁺ T cells and CD45RA⁺ (CD4/45A⁺) and CD45RO⁺ (CD4/45RO⁺) subsets were purified and stimulated with immunobeads coated with anti-CD3 and anti-CD28 for 6 days. β-Chemokine levels were measured by enzyme-linked immunosorbent assay. These data are generated from two separate donors. Donor 1 showed the least differences in production between naïve and memory cells, whereas donor 2 showed the greatest differences. Four other donors were examined, and their production values lie between those of donors 1 and 2. 

Neutralization of β-chemokine function renders CD3/CD28-stimulated memory cells susceptible to R5 virus infection.

The analysis of CCR5 surface expression and β-chemokine production in CD3/CD28-stimulated cells suggested that memory cells, despite their higher levels of CCR5 expression, remain resistant to R5 viruses because their high level of β-chemokine production restricts the ability of infecting viruses to use CCR5 as a coreceptor. Therefore, we hypothesized that neutralization of β-chemokine binding to CCR5 should render memory cells readily susceptible to infection by R5 viruses. This hypothesis was tested by incubating CD3/CD28-stimulated cells with neutralizing antibodies to RANTES, MIP-1α, and MIP-1β prior to infection with R5 HIV-1 isolates (Fig. 5). Preincubation with β-chemokine-neutralizing antibodies resulted in a modest increase in HIV susceptibility in CD3/CD28-stimulated naïve cells compared to the same cells treated with control IgG (data not shown). In contrast, when previously resistant CD3/CD28-stimulated memory cells were pretreated with β-chemokine-neutralizing antibodies, a vigorous infection ensued, as indicated by HIV-1 gag DNA accumulation. The neutralization of β-chemokines to CD3/CD28-costimulated memory cells resulted in a nearly 300-fold enhancement of p24 production in culture supernatants (data not shown). This observation demonstrates that the resistance of CD3/CD28-stimulated memory cells to R5 viruses is due to their ability, through β-chemokine production, to prevent coreceptor utilization by R5 viruses. Furthermore, this observation suggests that the resistance of unfractionated CD3/CD28-stimulated CD4⁺ cells to R5 HIV isolates arises from memory cell-driven β-chemokine production sufficient to block coreceptors on both memory and naïve cells.

FIG. 5.

Costimulated memory cells treated with neutralizing antibodies to β-chemokines are rendered susceptible to R5 HIV-1 infection. Purified CD45RA⁺ (RA) and CD45RO⁺ (RO) cells were stimulated with CD3/CD28 for 6 days and infected with HIV_US-1 (R5 virus strain) in the presence or absence of NAbs to the β-chemokines. Samples were taken at 0, 2, 72, and 144 h postinfection, and cell pellets were analyzed for HIV-1 gag DNA by a quantitative PCR liquid hybridization assay. Data shown are representative of two experiments.

DISCUSSION

The identification of α- and β-chemokine receptors as coreceptors for X4 (27) and R5 HIV-1 isolates (3, 13, 23–25), respectively, has spurred rapid advances in the understanding of HIV-1 infection and pathogenesis. In this report, we demonstrate that R5 viruses, the isolates critical for transmission (22, 45, 68), infect CD3/CD28-stimulated naïve CD4⁺ cells. In contrast, R5 isolates were unable to infect either CD3/CD28-stimulated memory cells or unfractionated CD4⁺ T cells. Given that naïve cells express lower levels of CCR5 than memory cells, this susceptibility was surprising. However, after costimulation memory cells were found to secrete significantly higher levels of the native CCR5 ligands RANTES, MIP-1α, and MIP-1β than naïve cells. Neutralization of these β-chemokines rendered memory cells highly susceptible to infection with R5 isolates. These observations suggest, therefore, that susceptibility to R5 isolates at the cellular level is governed not by the absolute level of CCR5 expression but rather by the level of unoccupied CCR5. This assertion is supported by our previous observation that CD3/CD28 costimulation renders unfractionated CD4⁺ cells resistant to R5 HIV-1 isolates through upregulation of β-chemokine expression (52) and inhibition of CCR5 upregulation (12).

In contrast to the distinct susceptibilities of costimulated CD4 cell subsets to infection with R5 strains, we found that both subsets were susceptible to infection with CXCR4-dependent strains. However, viral replication was at least 10-fold less efficient in costimulated naïve cells than in memory cells. These results confirm those of other studies indicating that the ability of CXCR4-dependent strains to replicate in naïve cells is impaired relative to their ability to replicate in memory cells (55, 60, 67). Indeed, it is likely that our studies have underestimated the magnitude of this effect as our naïve-cell populations were contaminated with approximately 3% memory cells since we did not exclude CD62L⁻ cells from our populations (55). Spina and coworkers have found that viral replication in naïve cells is inefficient and particularly dependent on nef function (60).

The use of CD3/CD28 costimulation reveals distinct facets of HIV-1 coreceptor regulation of CD4 cells. R5 isolates were unable to infect either CD3/CD28-stimulated memory cells or unfractionated CD4⁺ T cells. This form of natural immunity appears to act by both a paracrine mechanism mediated by enhanced β-chemokine secretion and an autocrine mechanism involving costimulation-mediated downregulation of CCR5 expression. In fact, CD3/CD28 costimulation exerts a generally repressive effect on all β-chemokine receptor transcripts examined to date: CCR1, CCR2b (40), CCR3, and Bonzo/Strl33 (52a). The susceptibility of CD3/CD28-stimulated naïve cells to R5 virus despite the lack of detectable CCR5 expression is surprising considering the essential role that CCR5 plays in virus entry (3). However, Platt and colleagues recently reported that in the presence of optimal CD4 levels only minute levels of CCR5 are needed for viral entry, suggesting that undetectable levels of CCR5 on the surfaces of naïve cells may be sufficient for entry (48). While it is possible that R5 isolates infect naïve cells through an alternate coreceptor, the global β-chemokine receptor downregulation caused by CD3/CD28 costimulation makes this an unlikely possibility as the addition of exogenous recombinant RANTES and MIP-1α to naïve cells blocked infection (data not shown).

In contrast to that of CCR5, CXCR4 expression was less variable between donors and after cellular activation. However, we did note discordance between CD28-mediated upregulation of CXCR4 mRNA and the decrease in surface CXCR4 levels for both memory and naïve cells (Table 1). Forster and coworkers have provided a potential explanation for these observations by demonstrating that there is a large intracellular store of CXCR4 (28). They found that cell signaling by phorbol ester treatment could lead to rapid downregulation of CXCR4 surface expression by internalization. Even though CD3/CD28 stimulation appears to lead to a downregulation of CXCR4 surface expression, our previous studies of unfractionated CD4 cells have shown that substantial amounts of functional CXCR4 remain on the cells, as indicated by the ability of the cells to mediate fusion with cells expressing X4 envelopes (12).

We found that CD3/CD28-stimulated naïve cells rapidly acquire the “memory” marker of CD45RO expression. The kinetics of the CD45RA-to-CD45RO transition were similar to those previously reported for cord blood naïve cells (47). However, the naïve cells still retain their naïve characteristics in their impaired ability to sustain HIV-1 replication, and in our previous studies, we found that they remain unable to secrete gamma interferon after CD3/CD28 stimulation (39). Thus, the conversion of naïve cells to memory cells is a complex, multistep process, and the conversion from CD45RA to CD45RO probably represents one of the earliest changes.

The implications from our present studies may suggest modifications to current models of HIV transmission and pathogenesis. While there is general agreement that dendritic cells and/or macrophages (dendriphages) are the initial targets of incoming virus (10, 11), the mechanism by which HIV-1 is subsequently introduced into the CD4⁺ T-cell population is unclear. Based on the demonstration of Bleul et al. (8) showing the predominant memory distribution of CCR5 expression, Unutmaz and Littman proposed that memory T cells are the first targets of infected dendriphages (63). In this model, dendritic cells and macrophages are the first cells infected by the transmitting virus. Although dendriphages can transmit X4 viruses to circulating T cells, this occurs at a very low frequency, and R5 viruses are predominantly passed to the new host (10, 11, 49). Based on the higher CCR5 surface levels of memory cells and on the propensity of resting memory cells to migrate to sites of inflammation (41), it was proposed that R5 would become established first in the memory CD4⁺ cell population. Additional evidence for this model came from in vitro studies showing that memory cells serve as better targets for HIV-1, although only X4 viruses were used (31, 55, 58, 60, 67). Here, we show that the generalization of data generated with X4 viruses to all strains of HIV-1 was oversimplified and that R5 and X4 viruses have different tropisms in regard to naïve and memory cells. We have also found that the level of total CCR5 expression is not correlated with the susceptibility of costimulated lymphocytes to infection with R5 isolates of HIV-1. Thus, several aspects of the Unutmaz and Littman model are not supported by our results derived from costimulated cells which have been activated in a presumably more physiologic manner than that associated with stimulation by PHA/IL-2. It is important to note that our results are not likely to pertain to resting lymphocytes, as CCR5 is not expressed in detectable amounts at the RNA level (Fig. 4) or on the surfaces (8) of resting naïve cells and β-chemokine secretion is not detected in resting memory cells (data not shown).

Our assertion that the level of coreceptor unbound by native ligand governs HIV susceptibility leads us to propose an altered model of viral transmission. In this model, shown in Fig. 6, dendriphages are the initial targets of HIV infection and naïve T cells are the targets of initial rounds of viral replication. Furthermore, we propose that successful early transmission events occur not in the periphery but in the regional secondary lymph organs. Dendritic cells have been reported to migrate to lymph nodes (16) and preferentially interact with naïve T cells via DC-CK1, a recently described chemokine that attracts naïve cells to dendritic cells (2). In this setting of antigen presentation and cell activation without a concomitant increase in β-chemokine production, activated naïve cells serve as ideal targets for HIV-1 infection. Memory cells, by virtue of their wider tissue distribution, are not concentrated in the lymph nodes and hence may be underrepresented during the initial transmission events. More importantly, despite their higher level of CCR5 expression, memory cells are resistant to infection by R5 viruses due to their high-level β-chemokine production. Furthermore, due to this high level of β-chemokine production, memory cells may be able to offer protection from an R5 virus to nearby naïve cells. In a physiologic situation it is difficult to estimate the concentration of β-chemokines available to bind CCR5; however, CD3/CD28 stimulation results in optimal chemokine secretion, which is probably not duplicated by natural ligand interactions (B7 family). Nonetheless, the relative differences of chemokine production between naïve and memory cells are maintained upon B7-1 costimulation (52a). As the infection matures and X4 viruses emerge, memory cells become susceptible to infection while naïve cells remain susceptible to infection.

FIG. 6.

Model of HIV-1 transmission. In this model, R5 viruses preferentially infect dendritic cells (DC) in the periphery and the DC transport the virus to the lymph nodes, where there is a high concentration of activated naïve CD4 T cells and low levels of β-chemokines. Naïve cells are preferentially depleted but occasionally can be protected from R5 virus by nearby activated memory cells.

One implication of this model is that naïve cells must die more rapidly after infection than memory cells, in order to account for the observation first made by Schnittman and coworkers that circulating memory cells preferentially harbor HIV-1 (58). Another prediction of the model is that while naïve cells could serve as targets for CCR5-dependent virus, memory cells would be expected to serve as the primary producers of the viral load because naïve cells presumably die quicker and cannot produce as much progeny virus as memory cells (31, 55, 58, 60, 67). Finally, this model, if correct, could serve to explain the previously reported rapid depletion of naïve cells in adults with HIV-1 infection (14, 44, 54, 70). Thus, the distinction between total CCR5 expression and levels of unoccupied CCR5 revealed by examination of costimulated naïve and memory CD4 lymphocytes has important implications for viral transmission and pathogenesis.