Persistent Decreases in Blood Plasmacytoid Dendritic Cell Number and Function Despite Effective Highly Active Antiretroviral Therapy and Increased Blood Myeloid Dendritic Cells in HIV-Infected Individuals

Patients and PBMC isolation

Consenting chronically HIV-seropositive patients (n = 65) at various stages of disease were enrolled in this study. All patients were from the Jonathan Lax Immune Disorder Clinic (Philadelphia Field Initiation Group for HIV Trials). CD4 T cell counts range from 9 to 1,347 (mean of 501 cells/mm³), and viral load from <50 to 199,000 copies/ml (mean of 15,091 copies/ml). Twenty-one patients were not receiving therapy, and 44 were under HAART. Ninety-six percent of virally suppressed patients (viral load <50 copies/ml) were under HAART. None of the patients presented any O.I. or HIV-related neoplasms. Healthy HIV-1-seronegative donors from the Wistar Institute Blood Donor Program were included as control subjects. Institutional Review Board approval (from the Wistar Institute and Philadelphia Field Initiation Group for HIV Trials) and informed consent were obtained before blood donation. Blood was processed within 2–3 h from drawing. All reagents used were selected for their low levels of endotoxin contamination. PBMC were separated on Ficoll-Paque (Amersham Pharmacia Biotech, Uppsala, Sweden) density gradient and resuspended in RPMI 1640 medium supplemented with 10% heat-inactivated FBS, l-glutamine, and antibiotics.

Whole blood phenotypic analysis of MDC and PDC/IFN-α-producing cell subsets

A total of 200 μl blood was lysed in ammonium chloride (Pharml; BD PharMingen, San Diego, CA) at room temperature and washed twice in FACS buffer (PBS, 2% FBS, 0.5% BSA, and 0.02% sodium azide). Cell pellets were incubated in 200 μl FACS buffer with adequate amounts of mAb at room temperature for 20 min. The following mAbs were used: lineage mixture (Lin1-FITC), HLA-DR-PerCP, CD123-PE, and CD11c-allophycocyanin or PE, from BD Immunocytometry Systems (San Jose, CA). PE-conjugated BDCA-2 Ab was purchased from Miltenyi Biotec (Auburn, CA). Irrelevant isotype-matched (IgG1, IgG2b) mAbs were used in each experiment. MDC were identified by Lin⁻HLA-DR⁺CD11c⁺ and PDC by Lin⁻HLA-DR⁺CD123⁺ and by the newly described mAb BDCA-2 (HLA-DR⁺BDCA-2⁺) that identify a novel human DC Ag on peripheral blood PDC (23). Cells were washed twice with FACS buffer, fixed, and analyzed in a FACSCalibur with CellQuest software (BD Immunocytometry Systems). The fluorescence of 150,000–200,000 events was accumulated for analysis. When indicated, the absolute number of circulating blood DC (MDC and PDC) was calculated using the percentage of cells with respect to the lymphocyte and monocyte absolute counts, as determined by an automated differential blood count.

IFN-α production

PBMC (2.5 × 10⁶/well) from controls and patients were cultured in 24-well plates with irradiated cell-free viral supernatant (5HA Influenza-PR8 or 10⁵ PFU HSV-1-NS) for 18 h. As control for Influenza-PR8, allantoid fluid was used and did not induce any detectable amount of IFN-α (data not shown). Cell-free supernatants were harvested and tested by a commercial ELISA using matched Ab pair for IFN-α and performed according to the manufacturer (Endogen, Woburn, MA). Absorbance was measured on an automatic ELISA reader. Sensitivity of the assay was ∼8–12 pg/ml.

Mixed lymphocyte reaction

MLR was performed as described by Young and Steinman (24). A total of 2 × 10⁵ plastic nonadherent mononuclear cells isolated from the umbilical cord blood of healthy term neonates by Ficoll-Paque density gradient centrifugation was used as responders. Responding cells were cultured in triplicate for 5 days with 1 × 10⁵/well irradiated allogeneic stimulator cells in 96-well plates in 200 μl medium. Cultures were pulsed with [³H]TdR (1 μCi/well; Amersham, Arlington Heights, IL) for 16 h. Nuclei were collected into fiberglass filter paper (Packard Instrument, Meriden, CT). Radioactivity in the filter paper was quantified as cpm by use of a direct beta counter 9600 (Packard Instrument). Results were expressed as stimulation index (SI) determined as follows: mean cpms in cultures containing responding and stimulator cells/mean cpms in culture containing responder cells alone. A SI of 3 and above was considered positive.

Statistical analysis

This was performed with JMP 4.0 software (SAS Institute, Cary, NC). Data were compared using paired or unpaired nonparametric tests, as appropriate. Statistical significance between HIV⁻ and HIV⁺ was performed by the Wilcoxon tests. Correlation and association between CD4 counts, viral load, and other parameters were performed by the Spearman’s rank test. When needed, patients were stratified according to their CD4 T cell counts (<250, 250–500, >500), viral load (<50 copies/ml, 50–5000 copies/ml, and >5000 copies/ml), and combination antiretroviral therapy. All p values are two sided.

PDC and MDC subsets in HIV infection

We analyzed the phenotype and functions of the two major peripheral blood DC subsets (MDC and PDC) in a cross-sectional cohort of healthy uninfected controls (n = 53) and chronically HIV-infected individuals (n = 65). Cells were quantified by multiparametric flow cytometry on whole blood (Fig. 1⇓ shows a representative analysis). Our patient cohort consisted of HAART-treated (n = 44) and untreated individuals (n = 21), with a CD4⁺ T cell number ranging from 9 to 1,347/mm³ and a viral load from <50 to 199,000 copies/ml. The median frequency of PDC in control individuals was 0.29% (7,250 cells/ml) for CD123⁺ subset and 0.37% (8,250 cells/ml) for BDCA-2⁺ subset, whereas that of MDC was 0.27% (6,625 cells/ml). PDC were significantly (p < 0.001) decreased in patients (0.10%, 3,288 CD123⁺ cells/ml; 0.09%, 2,612 BDCA-2⁺ cells/ml) as compared with healthy controls (Fig. 2⇓ and Table I⇓). The proportions of PDC (CD123⁺ and BDCA-2⁺) were highly correlated in both control (r = 0.55, p = 0.007, n = 21) and HIV-infected subjects (r = 0.64, p < 0.001, n = 33), indicating that these two phenotypes (CD123⁺ and BDCA-2⁺) identify largely overlapping cell subsets. Unlike PDC, the median frequency of MDC in HIV-infected patients (0.27%, 6,100 cells/ml) was similar to that observed in uninfected controls (0.27%, 6,625 cells/ml; Fig. 2⇓ and Table I⇓). Additional DC subsets analyzed include the newly described BDCA-3⁺ subset (23), which was found to be decreased in patients as compared with uninfected controls (0.03%, 1,000 cells/ml vs 0.07%, 1,750 cells/ml, p = 0.0034, data not shown). However, because the nature of the BDCA-3⁺ subset is still poorly understood, further functional characterization of this subset is needed before its role in the context of an immune response can be interpreted.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Representative phenotypic analysis of MDC and PDC subsets. Whole blood (200 μl) assay is used. Multicolor staining for MDC (Lin⁻HLA-DR⁺CD11c⁺) and for PDC (Lin⁺HLA-DR⁺CD123⁺; HLA-DR⁺BDCA-2⁺) is shown. After excluding dead cells based on light scatter gating, 150,000–200,000 events were acquired. A, PBMC from HIV⁻ and HIV⁺ donors gated on HLA-DR^high, and of Lin-1⁻, DC are identified based on CD123 (PDC) or CD11c (MDC) expression. B, The same donors stained with HLA-DR and BDCA-2.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Phenotypic analyses of MDC and PDC subsets. Whole blood (200 μl) assay is used, and multicolor staining is performed as described in Materials and Methods. Each dot represents a single donor, and horizontal bars represent the median. Statistical significance between controls and patients is shown at the top of the figure.

IFN-α secretion and MLR response as functional correlates of PDC and MDC subsets

To gain further insight into the functional outcome of the decrease in circulating PDC, we evaluated the production of IFN-α by healthy uninfected controls (n = 36) and HIV-infected individuals (n = 48) in response to Influenza-PR8 or HSV-1-NS strain exposure. IFN-α has a potent antiviral and antitumor activity and, within peripheral blood, it is selectively produced by PDC following enveloped viral stimulation (7, 11, 20). Consistent with a decreased number of circulating PDC in HIV infection, PBMC from HIV-infected patients produced on average 20-fold less IFN-α than PBMC from control donors when stimulated in vitro with either Influenza-PR8 or HSV-1-NS (p < 0.0001, Fig. 3⇓). Moreover, a positive correlation was found between PDC frequencies and IFN-α production by PBMC in response to either virus (Rho = 0.8, p < 0.0001 for CD123, and Rho = 0.699, p < 0.0001 for BDCA-2, respectively).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

IFN-α production. A total of 2.5 × 10⁶ PBMC from HIV⁻ (n = 36) and HIV⁺ (n = 48) was exposed to irradiated Influenza-PR8 (6HA) or HSV-1-NS (10⁵ PFU) for 18 h. IFN-α was measured in the cell-free supernatants by IFN-α-specific ELISA. Each dot represents a single donor, and horizontal bars represent the median. Statistical significance between controls and patients is shown.

We analyzed whether in HIV-infected patients APC-specific activities associated with MDC were affected by testing the ability of PBMC to stimulate proliferation of allogeneic T cells. We compared in PBMC from 14 HIV-infected patients and in 8 healthy controls the number of both DC subsets by phenotypical analysis and their associated functional properties. As shown in Table II⇓, even though the PDC compartment was impaired in the HIV-infected patients (as indicated by their decreased frequency and IFN-α secretion), MDC frequency and the allostimulatory function of PBMC from the patients were comparable with those of the controls. Moreover, no significant difference was observed in the MLR assays between virally suppressed (<50 copies/ml) and viremic patients (>50 copies/ml). Our results, indicating that MDC from HIV-infected patients are not impaired in their ability to stimulate allogeneic T cells, are in agreement with earlier reports (25, 26) and with our previous observation using enriched population of total DC (27). Interestingly, we found a positive correlation between the frequency of the MDC subset and the MLR response (r = 0.70, p = 0.0073) consistent with a role for the MDC subset in this functional response. Taken together, these results strongly suggest a dichotomy between PDC and MDC in our cohort of HIV-infected individuals: the depletion of PDC is associated with impairment of IFN-α production, while MDC are maintained and retain functional allogeneic responses.

Lack of correlation between CD4 counts, viral load, and PDC number

We compared patients grouped according to CD4 count or viral load, as described in Materials and Methods, to address the question of whether the PDC and MDC subsets were affected by CD4 T cell counts, viral load, or antiretroviral therapy. Statistical anal-ysis revealed no correlation between CD4 T cell counts or viral load and PDC frequency (defined by expression of CD123⁺ or BDCA-2⁺), albeit their number in the patients was significantly decreased in comparison with healthy uninfected controls. Importantly, analysis of PDC in virally suppressed patients as compared with uninfected controls revealed a sustained decrease in both cell frequency and IFN-α secretion (Tables II⇑ and III⇓), indicating that the loss of PDC is not reversed by antiretroviral treatment. IFN-α production was significantly associated with CD4 counts only when the analysis was restricted to viremic patients with >50 HIV copies/ml (data not shown). In contrast to a recent report in which increased levels of IFN-α secretion were described following HAART (28), impaired IFN-α production was observed in treated patients with complete viral suppression (<50 copies/ml) as well as in patients with persistent viral replication (>50 copies/ml, p < 0.0001 when compared with controls; Table III⇓). The differences in the results reported in this study and the one mentioned above (28) could depend on specific characteristics of the patient cohorts studied: e.g., none of the patients used in this study had a history of O.I. or oncological diseases. It should also be noted that our interpretation of a sustained decrease in IFN-α secretion (by ELISA) rests on a direct comparison with uninfected controls, while the above mentioned study addressed recovery based on investigator-assigned thresholds of IFN-α secretion (by bioassay) within specific HIV⁺ populations (28). Taken together, our results suggest that a loss of PDC may be an early event in HIV infection with a limited capacity for reconstitution following antiretroviral therapy, as reflected by persistent low frequency of PDC in subjects under treatment with <50 HIV copies/ml and high CD4 T cell count.

Although the exact mechanism responsible for the PDC deficiency remains to be defined, our observations argue against a predominant role of PDC in the control of HIV replication in chronic infection, as proposed by others (20, 21, 28). Of interest, we noticed that six of the HIV-infected subjects on HAART had PDC frequencies in the normal range (≥0.29%) and only two of these individuals, with CD4 counts of 1246 and 651 cells/mm³, had normal IFN-α responses. These data suggest that retention of the PDC subset may be possible in a small subset of patients highly responsive to antiretroviral therapy, as it was previously described for the long-term nonprogressor patients (20). Additional experiments isolating the PDC will be needed to determine whether a decrease in IFN-α secretion is due to a decrease in frequency of PDC or whether a functional impairment of this subset is also present.

Our data confirm a recent report by Feldman et al. (29) and support the association in HIV-infected patients between a decrease in IFN-α secretion and a decrease in the CD123⁺ DC subset, and extend it with the use of additional markers such as BDCA-2.

CD4 counts, viral load, and frequency of MDC

The overall frequency of MDC in patients remained stable with similar cell percentage as compared with uninfected controls (Fig. 2⇑ and Table I⇑). No correlation was found between MDC frequency and CD4⁺ T cell counts (data not shown). However, virally suppressed patients were found to have a significantly higher number of MDC (0.31%; 7020 cells/ml) than patients with a viral load of >5000 copies/ml (0.13%; 2636 cells/ml, p < 0.0032; Table III⇑). An association between a decrease of MDC frequency with an increase of viral load was further indicated by a significant negative correlation between these variables (Rho = −0.5, p < 0.0001, n = 65). The inverse association between MDC and viral load was also sustained when the analysis was restricted to a subset of 21 patients not receiving treatment (Rho = −0.68, p = 0.0007). It is important to note that the extensive use of antiviral treatment in our cohort may have biased our study against the possibility to detect the previously reported decreases in MDC during HIV infection (18), specifically the prevalence of HIV-infected subjects with <5000 copies/ml that did not show a significant decrease in MDC frequency compared with healthy controls. Overall, these results are consistent with a decreased MDC frequency only at high viral loads that can be reversed following antiretroviral treatment in contrast to the early and persistent deficiency in the PDC subset measured in the same patients, as described above. A longitudinal analysis of the consequences of HAART-mediated suppression on each DC subset (i.e., recovery of number and function as a correlate of viral suppression) is needed to confirm our cross-sectional findings.

We show for the first time a differential depletion between the frequency and function of MDC and PDC subsets when analyzed in parallel under suppressive therapy. Our data complement previous observations describing the depletion of MDC and PDC in viremic or end-stage HIV-infected patients (20, 21, 28). However, our studies indicate a lack of full DC reconstitution upon HAART that may contribute to a sustained immune impairment in these subjects. While we document a phenotypic and functional preservation in the MDC subset in the virally suppressed patients and HAART-responding patients, alteration in the PDC subset appears to be an early event and associated with a sustained impairment of IFN-α production. The consequence of the lack of recovery of IFN-α production and PDC number on the innate and adaptive immune functions of the patients remains to be defined. For example, a decrease in viral-induced IFN-α secretion may also affect in part the immune reconstitution of NK cell function. Additional experiments should assess whether the loss of this subset from circulation reflects a cytopathic effect caused by the virus through direct infection of the PDC subset (30) or through viral-induced apoptosis, as described with measles virus infection (31, 32). A role for interactions between HIV-1 and PDC subsets is supported by the susceptibility of IL-3-cultured Lin⁻HLA-DR⁺CD11c⁻ cells to be infected in vitro by HIV-1 (30). It is also of interest to relate our findings to current treatment guidelines for HIV infection, as a delay in treatment of patients with viral loads above 5000 copies/ml, irrespective of CD4 T cell count, would not prevent a decrease of MDC and PDC that may bear on immune function in general.