Dysfunctional natural killer cells, in vivo, are governed by HIV viremia regardless of whether the infected individual is on antiretroviral therapy

Progression to AIDS during HIV-infection is associated with a decrease in the number and function of natural killer (NK) cells [1–4]. The reduction in NK cells expressing CD56 and CD16 correlates with low CD4+ T-cell count [5] and increasing viral load [6,7]. Moreover, higher viral loads are associated with an increased presence of a highly dysfunctional NK cell subpopulation lacking CD56 but retaining CD16 [7,8], whereas suppression of viral replication with highly-active antiretroviral therapy (HAART) results in improvements in both NK cell function and NK cell phenotype [7]. These observations have been used to propose that high level viral replication is a consequence rather than a cause of alterations in the NK cell phenotypic patterns [7], although this remains controversial [9]. Of note, most of these studies involved individuals who were either not controlling HIV replication or whose virus was fully controlled with HAART. No study has systematically included individuals whose virus was low in the absence of therapy, and presumably being controlled immunologically. The present study determines the characteristics of NK cell subsets across a broad range of chronically-infected patients. A total of 76 individuals from four well characterized groups of individuals were enrolled: (i) ‘elite’ controllers: plasma HIV RNA levels < 75 copies/ml in the absence of any therapy (n = 19) (Fig. 1a); (ii) ‘partial’ controllers on antiretroviral therapy (PCAT): plasma HIV RNA levels between 75 and 104 copies/ml (n = 19) (Fig. 1a); (iii) ‘HAART suppressed’: plasma HIV RNA levels < 75 copies on long-term HAART (n = 19) (Fig. 1a); and (iv) ‘noncontrollers’: plasma HIV RNA levels > 104 copies RNA/ml (n = 19) (Fig. 1a). All subjects were recruited from the UCSF Study of the Consequences of the Protease Inhibitor Era (SCOPE) cohort and all provided their written informed consent. The first two groups were selected because previous work from these individuals have the strongest evidence for immune-mediated virus control [10].Fig. 1: Viral load and percentage of natural killer (NK) cells and their subsets in viremic and aviremic HIV-infected individuals. (a) HIV viral load and percentage (b) natural killer cells, (c) CD56dim, (d) CD56neg and (e) CD56bright expressing natural killer cells (CD16+/CD3−) of aviremic and viremic HIV-infected individuals. CD69 expression was evaluated with all NK cell subpopulations (f). Error bars indicate SD. Dots represent outliers. * P < 0.05, ** P < 0.01 based on Wilcoxon test. PCAT, ‘Partial’ controllers on antiretroviral therapy; HAART, highly-active antiretroviral therapy.Peripheral blood lymphocytes from the study subjects were stained with flurochrome-conjugated CD56, CD16, CD3 and CD69 antibodies (BDIS, San Jose, California, USA). After staining, cells were washed twice with phosphate-buffered saline (PBS) containing 0.5% bovine serum albumin and 0.1% sodium azide. Cells were fixed with 2% paraformaldehyde in PBS, pH 7.2, acquired on a FACsCalibur (BDIS) flow cytometer and analyzed using Cell Quest Software, version 3.2.1 (BDIS). Plasma HIV RNA levels were determined by the bDNA amplification technique (Quantiplex HIV RNA; Chiron, Emeryville, California, USA), as described previously [11]. The virologic noncontrollers had the lowest total percentage of classically defined NK cells (CD56+/CD3−) (mean ± SD = 3.4 ± 2.3%). This percentage was significantly lower than that observed in the HAART-suppressed group (mean ± SD = 5.7 ± 3.9%; P = 0.03) and is consistent with prior observations [1–4]. The percentage of CD56+ NK cells was intermediate in the elite controllers and PCAT subjects, and did not differ significantly from the noncontroller group (Fig. 1b). Of the three different NK cell subsets, based on CD56 and CD16 expression, the CD56dim/CD16+ NK cell subset possesses the greatest capacity to mediate a cytotoxic response [6,7]. The two viremic groups (noncontrollers and PCAT) had the lowest percentage of these cells (Fig. 1c) and were not statistically different from one another. The differences in the percentage of CD56dim/CD16+ NK cells between the noncontrollers and either aviremic group were statistically significant (P = 0.004 in noncontrollers versus elite controllers, and P = 0.007 in noncontrollers versus HAART-suppressed). CD56neg/CD16+ subset of NK cells are thought to be largely dysfunctional [8]. In contrast to the trends observed above, we saw a striking association between high viral load and a high percentage of these cells. Subjects whose virus was controlled therapeutically (HAART) or presumably immunologically (elite controllers) had the lowest percentage of these cells overall compared with the viremic individuals. Since little if any of these cells exist in uninfected individuals [12], our findings suggest that the presence of this subpopulation of cells occurred as a consequence of virus infection and that virus control appears to maintain this population of NK cells to significantly lower levels than when virus is not controlled (Fig. 1d; P < 0.0001 for either elite or HAART suppressed compared to either viremic population). The percentage of CD56bright/CD16neg NK cells, which tend to be effector NK cells with the greatest capacity to produce cytokines [13], was similar across all groups tested (Fig. 1e). We next evaluated the activation status of NK cells in the different infected individuals studied, focusing on the expression of CD69 on CD56dim/CD3−/CD16+ and CD56neg/CD3−/CD16+ cells (Fig. 1f). In contrast to what has been observed from this cohort with T cell activation [11,14], patients who were viremic either in the presence or absence of therapy had lower levels of NK cell activation (P < 0.0001 for comparison between either aviremic group with either viremic group). In summary, we did not find any consistent NK cell phenotypic pattern in our subsets of individuals who are likely to be controlling their virus immunologically (i.e. elite controllers, PCAT subjects) as compared to either HAART-suppressed or the noncontroller populations. These data suggest that NK cells, at least as measured in the peripheral blood using classical cell surface markers, are not causally associated with control of HIV replication. We did, however, observe consistent trends that tracked more closely with the level of viremia. For example, undetectable viremia in either elite controllers or HAART-suppressed subjects was associated with a higher percentage of activated effector NK cells (CD56dim/CD16+) and lower numbers of presumably dysfunctional NK cells (CD56neg/CD16+). As these observations were independent of whether the virus was controlled immunologically or therapeutically, they suggest that the level of viral replication was a cause rather than a consequence of the observed NK cell phenotypic patterns. This conclusion is supported by our previous study performed in vitro indicating that NK cells do not kill HIV infected cells effectively [15]. Further work is needed to correlate these phenotypic changes with NK cell function in both peripheral blood and in tissues of infected individuals. Given that NK cells are important in control of tumors and virus infected cells, it would be important to understand how HIV replication leads to an alteration in NK cells.