A CD8α⁻ subpopulation of macaque circulatory natural killer cells can mediate both antibody-dependent and antibody-independent cytotoxic activities

Abstract

Natural killer (NK) cells are important components of the innate immune system that mediate effector and regulatory functions. As effector cells, NK cells help control virus-infected cells through cell-mediated antibody-dependent mechanisms such as antibody-dependent cellular cytotoxicity (ADCC). Although macaques are an important and reliable animal model for the study of retrovirus-induced human diseases, and despite the crucial role played by NK cells in innate and adaptive immune responses against simian immunodeficiency virus (SIV), only a few studies have attempted to characterize different macaque NK cell subpopulations. In the present study, we identified a subpopulation of circulatory CD8α⁻ macaque NK cells that express NK lineage markers and exhibit cytotoxic potential. CD8α⁻ NK cells were phenotypically characterized as CD3⁻ CD14⁻ CD20⁻ CD8α⁻ cells that express NK cell markers including CD16, CD56, granzyme B, perforin, NKG2D and KIR2D. Based on their CD56/CD16 expression patterns, cells within the CD8α⁻ gate can be divided into four subpopulations: CD56^dim CD16^bright, CD56^dim CD16⁻, CD56^bright CD16⁻, and CD56⁻ CD16⁻ cells. In contrast, CD8α⁺ NK cells are 95% CD56^dim CD16^bright, which correlates with their high cytotoxic potential. Upon interleukin-15 activation, CD8α⁻ cells up-regulated CD69 expression and produced low levels of interferon-γ and tumour necrosis factor-α. Sorted CD8α⁻ NK cells were capable of killing MHC-I-devoid target cells and mediated ADCC responses against SIV gp120-coated target cells in the presence of macaque anti-gp120 antibodies. Taking into account CD8α⁻ myeloid dendritic cells, we show that about 35% of macaque CD8α⁻ cells represent a novel, functional population of circulatory NK cells that possesses cytotoxic potential and is capable of mediating anti-viral immune responses.

Introduction

Natural killer (NK) cells are a specialized subset of lymphocytes that navigate through the circulatory and lymphatic systems and provide a first line of defence against pathogen-infected and neoplastic cells. In humans, NK cells are phenotypically characterized as CD3⁻ CD56^dim/bright cells that account for up to 15% of peripheral blood lymphocytes.^1,2 NK cells, discovered in 1975,^3–5 are components of the innate immune system that protect host organisms against viral, bacterial and parasitic infections.⁶ They are also capable of directly killing tumour cells.^2,7 NK cells exert their function through two major effector mechanisms: direct killing of target cells, and production of inflammatory and regulatory cytokines.⁸ As cytotoxic effectors, NK cells are unique because they can kill certain target cells in vitro without previous sensitization.⁹ Unlike T cells, NK cells are not capable of antigen-specific receptor somatic recombination. Therefore, in vivo, NK cells rely on the surface recognition of MHC class I, class I-like molecules, and other ligands, by germline-encoded activating and inhibitory NK cell receptors (NKRs) to induce or arrest their cytotoxic activity against target cells.^10–12 Additionally, NK cells are capable of secreting a wide variety of cytokines and chemokines, which not only enhance innate immunity, but also shape downstream adaptive immune responses.^12–14 Human circulatory NK cells are phenotypically characterized in two subsets: cytolytic CD56^dim CD16⁺ NK cells (≥ 90%), and cytokine-producing CD56^bright CD16^−/dim NK cells (≤ 10%).⁷ Cytolytic CD56^dim CD16⁺ NK cells express high levels of killer cell immunoglobulin-like receptors (KIRs) and are capable of mediating potent antibody-dependent cellular cytotoxicity (ADCC). On the other hand, cytokine-producing CD56^bright NK cells express low levels of KIRs and mediate low ADCC and cytotoxic responses.²

Rhesus macaques (Macaca mulatta) are an important and reliable animal model for the study of retrovirus-induced human diseases. In fact, pre-clinical vaccine trials using macaque simian immunodeficiency virus (SIV) and simian/human immunodeficiency virus (SHIV) platforms are becoming gatekeepers for the advancement of candidate human immunodeficiency virus (HIV) vaccines into human trials.¹⁵ Even though the direct role played by NK cells during HIV infection remains undefined, there is strong evidence that these cells can provide some measure of protection against both initial infection and disease progression. Certain NKR phenotypes are associated with protection against HIV infection,¹⁶ and non-progressive HIV infections are associated with higher levels of NK cell cytotoxicity.¹⁷ Furthermore, vaccine-elicited non-neutralizing anti-envelope antibodies have been shown to contribute to protection against HIV, SIV and SHIV_89.6P challenges through cell-mediated activities including ADCC and antibody-dependent cell-mediated virus inhibition (ADCVI),^18–23 both of which are strongly associated with NK cell function.^24,25 An FcR-mediated activity of a broadly reactive HIV neutralizing monoclonal antibody (mAb) has also been shown to contribute to protective efficacy in a macaque challenge model,²⁶ further invoking a role of NK cells. Moreover, the recent modest success of the RV144 HIV clinical vaccine trial in Thailand²⁷ has been suggested to be partly the result of ADCC activity elicited by the vaccine regimen.²⁸ Hence, there is heightened interest in the HIV vaccine field in NK-cell-mediated effector functions.

Despite the potential role played by NK cells during innate and adaptive immune responses against HIV/SIV, and the utility of rhesus macaque models, the variety and function of roles of different macaque NK cell subpopulations have not been exhaustively explored. Previous reports have described macaque circulatory NK cells as CD3⁻ CD8α⁺ CD20^−/dim NKG2A⁺ cells that can be further divided into four subpopulations based on their CD56 and CD16 expression patterns.^29–31 However, CD8α expression on different human NK cell subsets is variable,^32,33 and therefore CD8α expression is not necessarily a requisite marker for NK cell phenotyping. In this regard, a minor subset of CD8α⁻ NK cells has been recently identified in healthy and HIV-infected chimpanzees.³⁴ Furthermore, it has been shown that peripheral blood mononuclear cells (PBMCs) from HIV-infected mothers and their infants that strongly respond to HIV-1 peptide stimulation [by up-regulating interferon-γ (IFN-γ) and interleukin-2 (IL-2) production in both CD3⁻ CD8⁻ and CD3⁻ CD8⁺ cells] are less likely to transmit and acquire infection, respectively.³⁵ For the reasons mentioned above, in the present study we evaluated the presence of NK cell lineage markers on macaque CD3⁻ CD14⁻ CD20^−/dim CD8α⁻ PBMCs, and the potential of these cells to mediate functional responses. Using multi-parametric flow cytometry, we identified a subpopulation of circulatory CD8α⁻ NK cells in naive and SIV-infected macaques that expressed the CD56 and/or CD16 NK cell lineage markers. A subset of these CD3⁻ CD14⁻ CD20^−/dim CD8α⁻ cells (from now on referred to as CD8α⁻ NK cells) also co-expressed granzyme B, perforin, NKG2D and KIR2D. Upon cytokine stimulation, CD8α⁻ NK cells up-regulated CD69 expression and IFN-γ mRNA transcription and produced low levels of tumour necrosis factor-α (TNF-α). Importantly, enriched CD8α⁻ NK cells were capable of mediating direct cell lysis as well as antibody-dependent killing, suggesting a potential for contributing to both innate and adaptive immune responses.

Materials and methods

Animals, cell collection and sera

Rhesus macaques (n = 30, 17 naive and 13 chronically infected with SIV) used in this study were housed at the National Institutes of Health (NIH) Division of Veterinary Resources (Bethesda, MD), at Bioqual, Inc. (Gaithersburg, MD), and at Advanced BioScience Laboratories, Inc. (ABL; Kensington, MD), and maintained according to institutional Animal Care and Use Committee guidelines, and the NIH Guide for the Care and Use of Laboratory Animals. All animals were negative for SIV, simian T-cell leukaemia virus-type 1 and simian type D retrovirus except for the 13 subsequently infected with SIV. Blood samples were collected by venepuncture of anaesthetized animals into EDTA-treated collection tubes. The PBMCs were obtained by centrifugation on Ficoll-Paque PLUS gradients (GE Healthcare, Uppsala, Sweden). Cells were washed thoroughly and resuspended at 1 × 10⁶ cells/ml in R-10 medium (RPMI-1640 containing 10% fetal calf serum, 2 mm l-glutamine and penicillin/streptomycin [Gibco, Carlsbad, CA]). Serum samples obtained from previously immunized and SIV_mac251-challenged macaques³⁶ had been stored at −70° and were able to mediate potent ADCC activity, shown previously to correlate with reduction of post-challenge acute viraemia.¹⁸ Serum samples obtained before immunization were used as negative controls.

Flow cytometry and cell sorting

All fluorochrome-conjugated mAbs used in the present study were anti-human mAbs known to cross-react with rhesus macaque antigens. The following mAbs were purchased from BD Biosciences (San Jose, CA): FITC-conjugated anti-CD69 (FN50), anti-CD3 (SP34), and anti-CD20 (2H7); phycoerythrin (PE) -conjugated anti-CD8β (2ST8.5H7), and anti-CD20 (2H7); PE-Cy7-conjugated anti-CD56 (B159); allophycocyanin (APC) -conjugated anti-IFN-γ (B27), anti-TNF-α (MAb11) and anti-HLA-DR (TU36); Alexa Fluor 700-conjugated anti-CD3 (SP34-2); and APC-Cy7-conjugated anti-CD16 (3G8). The following reagents were purchased from eBiosciences (San Diego, CA): PE-conjugated anti-Perforin (deltaG9); peridinin chlorophyll protein-Cy5.5-conjugated anti-CD161/NKR-P1A (HP-3G10); and eFluor650NC-conjugated anti-CD20 (2H7). The following mAbs were purchased from Invitrogen (Carlsbad, CA): PE-TexasRed-conjugated anti-granzyme B (GB11); QDot605-conjugated anti-CD14 (TuK4); and Pacific Blue-conjugated anti-CD8 (3B5). Pacific Blue-conjugated anti-CD8 (RPA-T8) was purchased from BioLegend (San Diego, CA); APC-conjugated anti-CD159a/NKG2A (Z199) and PE-conjugated anti-CD335/NKp46 (BAB281) were purchased from Beckman Coulter (Miami, FL); PE-conjugated anti-CD337/NKp30 (AF29-4D12), APC-conjugated anti-CD314/NKG2D (BAT221), and anti-KIR2D (NKVFS1) were purchased from Miltenyi Biotec (Auburn, CA); and fluorescein-conjugated anti-CD11c (3.9) was purchased from R&D Systems (Minneapolis, MN). For multi-parametric flow cytometry analysis, approximately 1·5 × 10⁶ PBMCs were stained for specific surface molecules, fixed and permeabilized with a Cytofix/Cytoperm Kit (BD Biosciences), and then stained for specific intracellular molecules. The yellow LIVE/DEAD viability dye (Invitrogen) was used to gate-out the presence of dead cells. At least 300 000 singlet events were acquired on an LSR II (BD Biosciences) and analysed using FlowJo Software (TreeStar Inc., Ashland, OR). For all samples, gating was established using a combination of isotype and fluorescence-minus-one controls. For CD8α⁺ and CD8α⁻ NK cell sorting experiments, approximately 150 × 10⁶ PBMCs were stained with appropriate concentrations of FITC-conjugated anti-CD3, PE-conjugated anti-CD20 and Pacific Blue-conjugated anti-CD8 mAbs and passed through a FACSAria II Cell Sorter (BD Biosciences).

NK activation assays

Natural killer cells were activated using NK-cell-activating cytokines or by co-culture with NK-sensitive target cells. For the first approach, PBMCs were plated at 1 × 10⁶ cells/ml in 24-well plates and stimulated with recombinant macaque IL-15 (150 ng/ml) or recombinant macaque IL-2-Fc (a fusion of macaque IL-2 and IgG2 Fc, 400 ng/ml), both obtained from the NIH/NCRR funded Resource for Nonhuman Primate Immune Reagents, Emory University, Atlanta, GA, for 24 hr, with the last 6 hr of culture being in the presence of 1 μl/ml of GolgiPlug (BD Biosciences). As macaque IL-12 was not available from the Resource for Nonhuman Primate Immune Reagents, recombinant human IL-12 (100 ng/ml, Peprotech, Rock Hill, NJ) having 95% amino acid homology with the macaque protein³⁷ was also used as a stimulus. Cells were subsequently washed and expression of CD69, and IFN-γ/TNF-α production by CD8α⁺ and CD8α⁻ NK cells were measured by flow cytometry. For the second approach, PBMCs were initially cultured in the presence of IL-2 (400 ng/ml) or IL-15 (150 ng/ml) for 24 hr. Cells were then extensively washed and co-cultured with the HLA class I-defective B-cell line 721.221 at a 5 : 1 effector-to-target (E : T) ratio for 6 hr before flow cytometry analysis of CD69 and IFN-γ expression on CD8α⁺ and CD8α⁻ NK cells. In both approaches, non-stimulated PBMCs were used to determine the baseline levels of NK cell activation.

Real-time reverse transcription-PCR

Total RNA was isolated from sorted cells using Qiagen's RNeasy Plus Mini Kit according to the manufacturer's directions (Qiagen, Valencia, CA), followed by the immediate generation of cDNA using the Qiagen QuantiTect Kit with the following modification; the extension time was increased from 15 min to 1 hr at 42°. Primers (Table 1) were designed to be exon spanning and were tested against the rhesus macaque genome on the UCSC Genome Browser website (http://genome.ucsc.edu/) using Blat (University of California Santa Cruz, Santa Cruz, CA). Gene expression levels were normalized against 18s RNA as reference gene. For the calculation of expression levels we used the ΔΔ2^CT method. Samples were run in triplicate in a 96-well plate in 25 μl reaction volumes using SYBR green premix with ROX (Fermentas, Glen Burnie, MD) on an Applied Biosytems ABI7000 cycler (Life Technologies, Carlsbad, CA) under the following conditions: 2 min at 50°, 10 min at 95° and 40 cycles of 30 seconds at 95°, 15 seconds at 59° and 30 seconds at 72·5° followed by standard melting curve analysis.

Table 1.

Real-time PCR primer pairs for indicated target genes

Target	Forward primer 5′ to 3′	Reverse primer 5′ to 3′
Interferon-γ	GCAACAAAAAGAAACGGGATGAC	CTGACTCCTTTTTCGCTTCC
Interleukin-13	GTACTGTGCAGCCCTGGAAT	CCACCTCGATTTTGGTGTCT
Tumour necrosis factor-α	AGCCCATGTTGTAGCAAACC	GCTGGTTATCTGTCAGCTCCA
Tumour necrosis factor-β	CACCACACGCTCTTCTGTCTGCT	CAGGCTTGTCACTTGGGGTTCG
MxA	AGGAGTTGCCCTTCCCAGA	TCGTTCACAAGTTTCTTCAGTTTCA
Granzyme B	CCCCATCCAGCCTATAATCC	CTGGGCCTTGTTGCTAGGTA
Perforin	GAAGACCCACCAGGACCAGTA	GTGGAGGCATTGAAGTGGAG
18s	GCCCGAAGCGTTTACTTTGA	TCCATTATTCCTAGCTGCGGTATC

721.221 killing assay

To evaluate the cytotoxic potential of cytokine-treated PBMCs and sorted CD8α⁺ and CD8α⁻ NK cells we designed a flow cytometry-based killing assay. Briefly, 2 × 10⁶ target 721.221 cells were labelled with 5 μl of the DiO Vybrant cell-labelling solution (Molecular Probes, Carlsbad, CA) in 2 ml PBS for 15 min at room temperature. Target cells were washed twice and plated in R-10 at a final concentration of 25 000 cells per well in U-bottom 96-well plates. Effector cells (either cytokine-treated PBMCs or sorted CD8α⁺ and CD8α⁻ NK cells) were added at the indicated E : T ratios to a final volume of 200 μl. Plates were incubated at 37° for 4 h. After incubation, cells were labelled with 0·2 μl per well of the far red Live/Dead fixable dead cell stain kit (Invitrogen). Plates were washed twice with PBS and finally fixed in 200 μl of a 2% PBS-paraformaldehyde solution. Labelled cells were stored at −4° until acquisition on a FACSCalibur (BD Biosciences). At least 5000 target cells (FL1-DiO⁺ events) were acquired. Specific target cell killing was measured by incorporation of the far red LIVE/DEAD amine dye (FL4) in the DiO⁺ population. Target cells alone were used as controls to correct for background levels of cell killing.

Autologous ADCC assay

CD4⁺ T lymphocytes, to be used as target cells, were purified from naive macaque PBMCs using a non-human primate CD4⁺ T-cell isolation kit (Miltenyi Biotec), labelled with DiO (as described for the 721.221 killing assay), and then coated with 15 μg SIV₂₅₁ gp120 (ABL) at room temperature for 45 min in RPMI-1640. CD4⁺ target cells were then washed twice and plated in R-10 at a final concentration of 10 000 cells per well in U-bottom 96-well plates containing serial dilutions of macaque sera (known to mediate ADCC activity) and incubated for 15 min at room temperature to allow antibody–antigen interaction. Effector cells (autologous PBMCs or sorted CD8α⁺ and CD8α⁻ NK cells) were added at a 25 : 1 (PBMCs) or 12 : 1 (sorted cells) E : T ratios to a final volume of 200 μl. Plates were centrifuged for 3 min at 400 g to promote cell-to-cell interactions and then incubated at 37° for 4 hr. After incubation, cells were labelled and analysed as indicated for the 721.221 killing assay. SIV₂₅₁ gp120-coated target cells alone, ADCC-negative pre-immunization sera from the same macaques, and a no-serum target plus effector cell mixture were used as negative controls. To calculate results, non-specific killing (from target cells alone and from a no-serum target plus effectors mixture) was subtracted from all wells and an ADCC cut-off value was calculated as the mean of values from all dilutions of negative pre-immune sera plus three standard deviations. The ADCC killing was considered positive when killing percentages were higher than the cut-off value.

Phenotypic stability assay

To assess phenotypic stability of macaque NK cell subsets, PBMCs or sorted CD8α⁺ and CD8α⁻ NK cells were left untreated or were stimulated with IL-2 (400 ng/ml), IL-15 (150 ng/ml), or a combination of both for different time periods. After incubation, cells were stained with mAbs and the proportions of different subsets were analysed by polychromatic flow cytometry as described above.

Statistical analysis

Unless otherwise specified, all data reported were averaged from the number of macaques indicated in the figure legends. Results are shown as means ± SEM. Data were analysed using Prism (v5.03; GraphPad Software, La Jolla, CA). A P-value of ≤ 0·05 was considered statistically significant.

Results

Identification of CD8α⁻ and CD8α⁺ candidate NK cell subsets in peripheral blood of rhesus macaques

Previous studies have identified macaque NK cells as CD3⁻ lymphocytes that are positive for CD8α and CD159a, while lacking CD14 and CD8β expression.²⁹ However, expression of the NK cell-associated lineage markers CD16 and CD56, as well as perforin, have also been detected in CD8α⁻ NK cells of humans.^32,33 Given this, and in view of the increasing interest in elucidating NK effector mechanisms in SIV and SHIV macaque models, we investigated whether rhesus macaque CD3⁻ CD8α⁻ cells also included NK cells. Two candidate NK subpopulations, based on their CD8α expression patterns, were identified in rhesus macaque PBMCs as CD3⁻ CD14⁻ CD20^−/dim cells within a large side-scatter versus forward-scatter lymphocyte singlet gate (Fig. 1a). Cells in these two subsets were negative for the common lineage markers CD4, CD8β, CD123, γδTCR and CD19 (data not shown). Proportionally, CD3⁻ lymphocytes accounted for 28·62 ± 6·92% of CD14⁻ circulating lymphocytes (Fig. 1b).Within the CD3⁻ compartment, CD8α⁻ and CD8α⁺ cells represented 19·8 ± 7·1% and 34·3 ± 17·4% of CD3⁻ CD14⁻ CD20^−/dim cells, respectively (Fig. 1c).

Figure 1.

Identification of two natural killer (NK) cell subpopulations in the peripheral blood of rhesus macaques. Peripheral blood mononuclear cells (PBMCs) were isolated and stained with a cocktail of fluorochrome-conjugated monoclonal antibodies for multicolour flow cytometric analysis. (a) Two candidate NK cell subpopulations were identified as CD14⁻ CD3⁻ CD20⁻ CD8α⁺ and CD14⁻ CD3 CD20⁻ CD8α⁻ cells, within a broad side scatter versus forward scatter lymphocyte gate. (b, c) Frequencies (as percentage of parental populations) of CD3^−/+ cells and of each candidate NK cell subpopulation are shown. Horizontal bars in (b) and (c) represent the mean values obtained from 17 macaques. *P<0.05 indicates statistically significant differences between the compared groups by a paired Student's t-test.

Lineage marker expression in macaque circulatory CD8α⁻ and CD8α⁺ NK cell subsets

Natural killer cells can be identified by surface expression of the classical cell lineage markers CD16 and CD56, as well as a number of inhibitory/activating receptors and intracellular cytotoxic proteins.⁸ To determine if CD8α⁻ NK cells comprise a subpopulation of macaque NK cells, we used polychromatic flow cytometry to detect co-expression of NK cell-associated markers. As shown in the representative histograms (Fig. 2a), CD8α⁻ NK cells expressed CD16, CD56, granzyme B and perforin, but no expression of NKG2A, CD161, NKp46 and NKp30 was detected. On the other hand, CD8α⁺ NK cells stained positively for all of the above-mentioned molecules (Fig. 2a, bottom row). Further analysis revealed that CD8α⁻ and CD8α⁺ NK cells expressed comparable levels of the Integrin α-X (CD11c) on their surface; while NKG2D expression was more abundant on CD8α⁺ NK cells (approximately 85%) compared with CD8α⁻ NK cells (approximately 18%, Fig. 2b). Only CD8α⁻ NK cells expressed HLA-DR on their surface (Fig. 2b). Given the fact that granzyme B and perforin are crucial for NK cell cytolytic function,³⁸ we evaluated the co-expression of these two proteins in the NK cell subpopulations. Approximately 10% of CD8α⁻ NK cells co-expressed granzyme B and perforin (Fig. 2c), indicating cytolytic potential for this NK cell subpopulation. On the other hand, in agreement with their known cytolytic capability,³⁰ approximately 46% of macaque CD8α⁺ NK cells co-expressed these two proteins.

Figure 2.

Both subpopulations of macaque natural killer (NK) cells express lineage specific markers and cytotoxic molecules. Macaque peripheral blood mononuclear cells (PBMCs; n = 17) were stained and analysed by multicolour flow cytometric analysis for the expression of lineage markers. (a) Representative histograms showing the expression of NK cell lineage specific markers in CD8α⁻ and CD8α⁺ subpopulations. Fluorochrome-matched isotype controls are represented by open histograms. (b) Mean levels ± SEM of expression of all markers evaluated in all macaques are shown. (c) Representative staining and mean ± SEM percentages of perforin and granzyme B double-positive cells in both CD8α⁻ and CD8α⁺ subpopulations. (d,e) Representative flow cytometry dot plots showing the distribution of CD16 and CD56 subsets within the CD8α⁻ (d) and CD8α⁺ (e) subpopulations. Graphs on the right of each plot show the mean percentage ± SEM of each subset (as a fraction of the total NK cell subpopulation) from all macaques. GrzB, granzyme B.

In humans, circulatory NK cells have been classified in two subsets based on their CD56 and CD16 expression patterns.⁷ Cytolytic CD56^dim CD16⁺ NK cells comprise 90% of circulatory NK cells, whereas, cytokine-producing CD56^bright CD16^−/dim NK cells represent about 10%. Examining the CD56 and CD16 expression patterns of macaque CD8α⁻ NK cells, we found that these cells could be divided into four subpopulations (Fig. 2d): double-negative cells (CD56⁻ C16⁻) accounted for 22·2 ± 10·6%, 34·2 ± 15·9% of cells were CD56^dim CD16⁺, and CD56^dim/+ CD16⁻ cells together represented approximately 39·4 ± 19·3% of CD8α⁻ NK cells. On the other hand, 90 ± 7·9% of CD8α⁺ NK cells were CD56^dim CD16⁺, but only two other minor populations could be detected: CD56^dim CD16⁻ (1·5 ± 1·1%) and CD56⁺ CD16⁻ (2·1 ± 3·7%) (Fig. 2e).

Activation and cytokine production by CD8α⁻ NK cells

Given the fact that NK cells exert their function through direct cytotoxicity and by producing inflammatory and regulatory cytokines,³⁹ we investigated whether CD8α⁻ NK cells could become activated and produce cytokines upon stimulation with the known NK cell activating cytokines, IL-2, IL-15 and IL-12. After 24 hr of incubation with IL-15, we detected an up-regulation of the early activation antigen CD69 on the surface of CD8α⁻ and CD8α⁺ NK cells (P<0·01, Fig. 3a). As for cytokine production potential, CD8α⁺ NK cells were capable of producing IFN-γ and TNF-α in response to 24 hr stimulation with IL-15, whereas CD8α⁻ NK cells showed an upward trend for TNF-α production, but did not produce IFN-γ (Fig. 3b,c). Of note, neither CD8α⁻ nor CD8α⁺ NK cells significantly up-regulated CD69, IFN-γ or TNF-α in response to IL-12 (data not shown).

Figure 3.

Activation of macaque natural killer (NK) cell subpopulations in response to cytokine stimuli. Macaque peripheral blood mononuclear cells (PBMCs) were isolated and stimulated for 24 hr in the presence of interleukin-2 (IL-2; 400 ng/ml) or IL-15 (150 ng/ml). The last 6 hr of culture were in the presence of Golgi Plug. Cells were then washed and stained for FACS analysis of the indicated molecules. Up-regulation of CD69 (a), interferon-γ (IFN-γ) (b), and tumour necrosis factor-α (TNF-α) (c) was measured in CD8α⁻ and CD8α⁺ NK cells. Data shown are the mean ± SEM from 5–10 macaques. *P<0.05, **P<0.01, and ***P<0.001 indicate statistically significant differences between the non-stimulated (NS) and stimulated groups by non-parametric one-way analysis of variance.

Recently, a revised phenotypic analysis of chimpanzee CD8α⁻ NK cells showed that approximately 80% of CD8α⁻ CD16⁺ cells are myeloid dendritic cells (mDCs) that express CD11c and HLA-DR on their surface. This suggests that in chimpanzees, CD8α⁻ NK cells represent only approximately 20% of the cells present in the CD8α⁻ CD16⁺ fraction.⁴⁰ Based on this recent report, we re-evaluated our population of macaque CD8α⁻ NK cells for expression of CD11c and HLA-DR. As shown in Fig. S1 (see Supplementary material) we found that, similarly to what was observed in chimpanzees, only approximately 35% (37·1 ± 10·7) of the cells within the CD8α⁻ gate were negative for CD11c and HLA-DR expression and therefore could be considered true CD8α⁻ NK cells. These CD8α⁻ NK cells still showed four clear subpopulations based on their CD56 and CD16 expression patterns (see Supplementary material, Fig. S1c), but with slightly different proportions compared with those described in Fig. 2(d). Contaminating mDCs represented approximately 60% (61·7 ± 10·9%) of cells in the CD8α^– CD16⁺ population, and were mostly CD56^dim CD16⁺ and double-negative cells (see Supplementary material, Fig. S1d). These findings are in agreement with the small proportion of macaque CD8α⁻ NK cells that expressed cytotoxic markers (Fig. 2b,c) and became activated in response to IL-2 and IL-15 stimulation (Fig. 3a). Furthermore, upon repeating our cytokine stimulation experiments (and gating out contaminating mDCs) we observed a low but significant increase in the production of IFN-γ by CD8α⁻ NK cells in response to IL-15 (3·3 ± 1·7%, data not shown), not previously observed (Fig. 3b) because of the abundance of mDCs within the same gate.

An alternative ex vivo approach to induce NK cell activation and cytokine production is through co-culture with NK-sensitive target cells. First, using a flow cytometry-based killing assay, we confirmed the ability of unstimulated, as well as IL-2-stimulated and IL-15-stimulated, macaque PBMCs to kill the MHC-devoid human cell line 721.221. As shown in Fig. 4(a), treatment with both IL-2 and IL-15 significantly increased the killing capacity compared with non-stimulated PBMCs at different E : T ratios ranging from 40 : 1 to 5 : 1 (P<0·001 for both cytokines at a 40 : 1 E : T ratio). Second, using the 721.221-based NK cell activation assay, we analysed the effect of E : T cell co-culture on the activation status of CD8α⁻ and CD8α⁺ NK cells. To accomplish this, IL-2-treated and IL-15-treated PBMCs were cultured at a 5 : 1 E : T ratio with 721.221 cells for 6 hr before mAb staining and flow cytometry analysis (which included CD11c and HLA-DR to gate out mDCs in both NK cell subpopulations). Co-culture of IL-15-treated PBMCs with 721.221 cells induced the expression of CD69, CD107a and IFN-γ on the surface of CD8α⁺ NK cells. CD8α⁻ NK cells up-regulated the expression of CD69 and IFN-γ (Fig. 4b,c), while showing a modest trend for up-regulation of CD107a (Fig. 4d).

Figure 4.

Interleukin-2 (IL-2) and IL-15 enhance the cytotoxic potential of macaque natural killer (NK) cells. Macaque peripheral blood mononuclear cells (PBMCs) were purified and stimulated for 24 hr in the presence of IL-2 (400 ng/ml) or IL-15 (150 ng/ml). After stimulation, cells were washed and cultured with the HLA class I-defective B-cell line 721.221 for 4 hr at the indicated effector : target cell ratios. (a) Measurement of target cell killing by flow cytometry. CD69 (b), interferon-γ (IFN-γ) (c) and CD107a (d) expression in the effector cells (CD14⁻ CD3⁻ CD20⁻ CD11c⁻ HLA-DR⁻ CD8α⁻ or CD14⁻ CD3⁻ CD20⁻ CD11c⁻ HLA-DR⁻ CD8α⁺ cells) after 6 hr culture with 721.221 cells at a 5 : 1 effector-to-target ratio. Data shown are the mean ± SEM from four macaques. *P<0.05, **P<0.01 and ***P<0.001 indicate statistically significant differences between the non-stimulated (NS) and stimulated groups by non-parametric one-way analysis of variance.

Functional capacity of purified CD8α⁻ NK cells

Having found that CD8α⁻ NK cells express some NK cell lineage markers and become activated upon cytokine and target cell stimulation, we directly investigated the cytokine-producing and cytolytic potential of the entire population of CD8α⁻ NK cells which included the mDCs. CD8α⁻ and CD8α⁺ NK cells were sorted by FACS using fluorochrome-conjugated anti-CD3, anti-CD20 and anti-CD8 mAbs. The CD8α⁻ NK cells were enriched to a 95% pure population. CD8α⁺ NK cells (97% pure) and CD8⁻ CD20⁺ B cells (97% pure) were used as positive and negative controls, respectively (Fig. 5a). As described above, only approximately 35% of enriched CD8α⁻ NK cells were negative for CD11c and HLA-DR expression. However, further purification of CD8α⁻ NK cells to exclude mDCs was not possible because of limitations on the amount of blood allowed to be drawn from individual rhesus macaques. Because contaminating mDCs would not interfere in the functional assays, we proceeded to characterize the activities of NK cells present in the highly enriched CD8α⁻ NK cell population.

Figure 5.

Purified CD8α⁻ and CD8α⁺ macaque natural killer (NK) cells produce cytokines and can mediate cytotoxic responses. Macaque peripheral blood mononuclear cells (PBMCs) were isolated, labelled and passed through a FACSAriaII to obtain highly pure CD8α⁻ and CD8α⁺ NK cell populations. (a) Dot plots indicating pre-sort gates and post-sort purity obtained after a representative sort experiment. B cells (CD8⁻ CD20⁺) sorted under similar conditions were used as controls. (b) Sorted cells were stimulated with interleukin-2 (IL-2; 400 ng/ml) plus IL-15 (150 ng/ml) for 5 hr before RNA extraction for reverse transcription-PCR analysis of target gene expression. Results are shown as the mean fold increase ± SEM in cytokine-treated versus non-stimulated (NS) groups, normalized to 18s RNA expression and statistically compared with B-cell expression levels. (c) For 721.221 killing experiments, sorted cells were washed and cultured with target cells for 4 hr at the indicated effector-to-target cell ratios. Target cell killing was measured by flow cytometry as described in the Material and methods section. Killing mediated by each NK cell subpopulation was statistically compared with killing mediated by B cells (indicated by asterisks). ***P<0.001 indicates statistically significant differences between the killing mediated by CD8α⁻ and CD8α⁺ NK cells at the 16 : 1 E : T ratio and all other E : T ratios examined by two-way analysis of variance (anova). (d) Percentage of antibody-dependent cellular cytotoxicity (ADCC) -mediated killing of simian immunodeficiency virus gp120-coated CD4⁺ T cells by autologous PBMCs at a 25 : 1 effector-to-target cell ratio in the presence of sequential dilutions of an ADCC-positive serum. (e) Autologous ADCC assay using sorted CD8α⁻ or CD8α⁺ NK cells as effector cells at a 12 : 1 effector-to-target ratio in the presence of a 10⁻³ serum dilution (same serum as in d). Dotted lines represent the threshold values determined by the killing mediated by a negative serum sample (pre-immunization) obtained from the same macaque. Data shown represent pooled data from three independent sorts for (b), and two independent sorts (for each experiment in c, d and e). *P<0.05, **P<0.01 and ***P<0.001 indicate statistically significant differences between the compared groups by two-way anova.

As CD8α⁻ NK cells only minimally up-regulated the expression of IFN-γ (Fig. 4c) but did not up-regulate expression of TNF-α significantly (Fig. 3c), we further investigated expression of these and other cytokines by evaluating mRNA transcription of both genes in the enriched cell populations after 5 hr of IL-2 plus IL-15 treatment. In addition we assessed transcription levels of IL-13, TNF-β, MxA, granzyme B and perforin. Figure 5 (b) illustrates gene transcription relative to the level in non-stimulated cells, where a fold increase of 1·5 or more is considered positive. The figure further shows gene expression profiles of CD8α⁻ and CD8α⁺ sorted cells in comparison to sorted B cells. Increased transcription of IFN-γ (P<0·001), IL-13, TNF-α, TNF-β and MxA genes was observed for IL-2 + IL-15-stimulated sorted CD8α⁺ cells. A similar gene transcription profile was seen for CD8α⁻ cells. In these cells, increased transcription of IFN-γ (P<0·05), IL-13, TNF-α and TNF-β was seen. Under the conditions tested, B cells used as negative controls did not exhibit increased transcription of IFN-γ, IL-13, TNF-α or perforin, and only displayed marginally positive transcription levels for TNF-β, MxA and granzyme B (all with values of 1·6-fold increase).

To evaluate antibody-independent cytolytic function of CD8α⁻ NK cells, we used the flow cytometry-based 721.221 killing assay. As shown in Fig. 5(c), enriched CD8α⁻ NK cells were capable of killing target cells at E : T ratios of 16 : 1, 8 : 1 and 4 : 1 (P<0·001, when compared with the killing mediated by B cells at similar E : T ratios). On the other hand and as expected, enriched CD8α⁺ NK cells were capable of killing target cells at E : T ratios as low as 0·5 : 1 (P<0·001 versus B cells, Fig. 5c).

Given the demonstrated contributions of vaccine-elicited non-neutralizing antibodies to control of HIV/SIV viraemia and disease progression by cell-mediated effector mechanisms such as ADCC and ADCVI,^19,21 we evaluated whether CD8α⁻ NK cells could mediate ADCC. An autologous ADCC assay was established using SIV₂₅₁ gp120-coated macaque CD4⁺ T cells as targets and matched PBMCs as effectors. Serum-dependent ADCC activity was observed using a known antibody-positive serum when compared with a negative serum from the same animal (Fig. 5d). Subsequently, FACS-enriched CD8α⁻ and CD8α⁺ NK cells were used as effectors. The numbers of sorted CD8α⁻ and CD8α⁺ NK cells were limiting, so the effector activity of these cells was tested only at a single E : T ratio using a 1 : 1000 serum dilution. The ADCC activity was observed in both subsets (P<0·01 and P<0·001, for CD8α⁻ and CD8α⁺ NK cells, respectively), indicating that CD8α⁻ NK cells are capable of mediating functional ADCC responses (Fig. 5e).

Phenotypic stability of CD8α⁻ and CD8α⁺ NK cells

After determining that macaque CD8α⁻ NK cells can become activated and exert functional activity, we wanted to examine whether CD8α⁻ and CD8α⁺ NK cells are unique subsets, or if CD8α expression distinguishes members of the same cell population in different activation/differentiation stages. Initially, we conducted phenotypic stability studies using macaque PBMCs. As shown in Fig. 6(a), CD3⁻ cells (which include not only NK cells, but also B cells and mDCs) declined progressively over time, and this was not prevented by the addition of IL-2 or IL-15 in the culture media. Furthermore, CD8α⁻ NK cells also declined steadily throughout the 3-day observation period (Fig. 6b), and once again the addition of IL-2 or IL-15 did not preserve this subpopulation. On the other hand, survival of CD8α⁺ NK cells (Fig. 6c) was maintained over the 3 days, and was modestly, although not significantly, enhanced by the addition of IL-2 and IL-15. Most interestingly, we detected the appearance of a CD8α^dim population (minimally present at day 0, Fig. 1a), which was most abundant in untreated PBMCs, but still observed in IL-2-treated and IL-15-treated PBMCs (Fig. 6d). To explore which NK cell subpopulation contributed to the appearance of CD8α^dim cells, we performed phenotypic stability assays using sorted CD8α⁻ and CD8α⁺ NK cells. Sorted cells were left untreated or were stimulated with a combination of IL-2 and IL-15 to monitor their CD8α expression patterns. In unstimulated CD8α⁻ cells, we detected a subset of CD8α⁻ CD20^dim cells after 1 day of culture, which declined in proportion by day 2 (Fig. 6e, left panel). The addition of IL-2/IL-15 did not alter the proportion of CD8α⁻ CD20^dim cells when compared with the unstimulated controls. On the other hand, cultured CD8α⁺ NK cells progressively gave rise to a CD8α^dim CD20⁻ subpopulation over time (Fig. 6e, right panel) when left unstimulated. This ‘down-regulation’ of CD8 expression was prevented when IL-2 and IL-15 were added to the culture media. Taken together, our data suggest that macaque CD8α⁻ NK cells do not represent a differentiation stage of the CD8α⁺ population. Rather, CD8α⁻ NK cells are a unique and functional population of circulatory NK cells with cytotoxic potential, capable of mediating anti-viral immune responses.

Figure 6.

Phenotypic stability of CD8α⁻ and CD8α⁺ macaque natural killer (NK) cells. Macaque peripheral blood mononuclear cells (PBMCs) were purified and left untreated or cultured in the presence of interleukin-2 (IL-2; 400 ng/ml), or IL-15 (150 ng/ml). CD3 (a) and CD8α (b–d) expression patterns were evaluated at the indicated time-points by flow cytometry. Data shown are the mean ± SEM from three macaques. (e) Phenotypic stability assay using sorted CD8α⁻ and CD8α⁺ NK cells. Sorted cells were left untreated or cultured in the presence of IL-2 (400 ng/ml) plus IL-15 (150 ng/ml). At the indicated time-points, cells were washed and stained with monoclonal antibodies for flow cytometric analysis of CD8 expression. Representative results from one of two experiments are shown. NS, non-stimulated.

CD8α⁻ and CD8α⁺ NK cells in SIV-infected macaques

Having observed that CD8α⁻ NK cells are a functional subpopulation of NK cells in healthy rhesus macaques, we sought to determine if these cells were also present in SIV-infected macaques. Proportionally, CD8α⁻ NK cells were present at similar percentages in naive and SIV-infected macaques; whereas the percentage of CD8α⁺ NK cells was decreased in the blood of SIV-infected macaques (P<0·05, Fig. 7a). When assessing CD16 and CD56 expression patterns in both subpopulations of NK cells, we observed that CD56⁻ CD16⁺ cells were significantly decreased within CD8α⁺ NK cells of SIV-infected macaques (P<0·001, Fig. 7b). In contrast, the proportion of CD56⁻ CD16⁻ CD8α⁺ NK cells was significantly increased in SIV-infected macaques (P<0·001, Fig. 7b). Similar trends were observed in CD8α⁻ NK cells of SIV-infected macaques although they lacked statistical significance (Fig. 7c, CD56^dim CD16⁺ and CD56⁻ CD16⁻ subpopulations). Similar expression patterns for CD161, NKG2A, perforin and granzyme B within CD8α⁻ NK cells were observed in naive and SIV-infected macaques (data not shown). Collectively, our data suggest that macaque CD8α⁻ NK cells are a phenotypically unique and functional subpopulation of NK cells, capable of mediating antibody-dependent and antibody-independent cytotoxic activities.

Figure 7.

CD8α⁻ and CD8α⁺ natural killer (NK) cells are present in simian immunodeficiency virus (SIV) -infected macaques. Peripheral blood mononuclear cells (PBMCs) were isolated from naive (n = 17) and SIV-infected (n = 13) macaques and stained with a cocktail of fluorochrome-conjugated monoclonal antibodies for multicolour flow cytometric analysis. (a) Frequencies of CD8α⁻ or CD8α⁺ NK cells in naive and SIV-infected macaques. (b, c) Distribution of CD16 and CD56 subsets within the CD8α⁺ (b) and CD8α⁻ (c) subpopulations in naive and SIV-infected macaques. Graphs show the mean percentage ± SEM of each subset from all tested macaques. *P<0.05 and ***P<0.001 indicate statistically significant differences between the compared groups by two-way analysis of variance.

Discussion

Non-human primate models provide an invaluable tool for understanding and dissecting immune responses associated with lentivirus infection.¹⁵ The rhesus macaque in particular has been invaluable in both SIV and SHIV vaccine and pathogenesis studies. The most effective use of the macaque model requires detailed knowledge of the cells that make up the immune system, including phenotypic identification and functional analysis of individual cell populations, and elucidation of the role they play during innate and adaptive immune responses. This knowledge enhances our understanding of both protective and non-protective immune mechanisms during viral exposure and on-going infection and contributes to the design of candidate prophylactic and therapeutic regimens.^41,42

Natural killer cells are important for both the innate and adaptive lines of defence, and therefore represent a cell population of great interest. They have been shown to contribute to the control of both HIV and SIV infections,^35,43–46 most likely because of their presence at mucosal effector sites.^29,31,47 Despite their importance, only minimal efforts have been made to phenotypically identify and functionally characterize macaque NK cell subpopulations. In humans, NK cells can be categorized in multiple subsets by their surface expression patterns of CD56 and CD16 and by the expression of different types of NKRs.^2,7,48 Recent reports have described rhesus macaque NK cells as CD3⁻ CD8αα⁺ NKG2A⁺ lymphocytes present in the blood and tissues.^29,30 However, study of NK cells in non-human primates has proven to be technically challenging for several reasons. First, CD56 in macaques is not only expressed by NK cells, but also by monocytes.⁴⁹ Yet it has been recently shown that tissue NK cells are mostly CD16⁻ CD56⁺,²⁹ which indicates that CD56 is the most reliable marker for tissue NK cells. Therefore, use of anti-CD16 mAbs for depletion of NK cells in HIV/SIV in vivo studies may not be providing correct information regarding the role played by these cells in control of infection and the overall mucosal immune responses.⁵⁰ Additionally, the presence of other CD3⁻ cell subsets within the lymphocyte gate (B cells and monocytes), requires the use of specific lineage markers for the correct identification of NK cells.⁵¹ In the present study, our consistent gating strategy which eliminated dead cells, monocytes, T cells and B cells (Fig. 1a), left two distinct NK cell candidate populations based on their CD8α expression patterns. We subsequently found that a subset of the CD3⁻ CD14⁻ CD20^−/dim CD8α⁻ cells expressed NK-cell-associated lineage and activation markers, and responded to NK-cell-stimulating cytokines, making them a candidate macaque NK cell population.

As mentioned, not all cells within the CD8α⁻ gate were candidate NK cells because, as shown in Fig. 2, only a fraction of these cells expressed CD16, CD56, granzyme B and/or perforin. Similar to our findings, Rutjens et al. identified a minor CD8α⁻ NK cell population present in the blood of naive and HIV-infected chimpanzees. These CD8α⁻ chimpanzee NK cells not only co-expressed CD16 on their surface, but also were partially positive for a variety of cytotoxicity (such as NKG2D and NKp46) and co-activatory receptors.³⁴ We were able to confirm the presence of mDCs in the candidate population of CD8α⁻ NK cells as has been described in chimpanzees (see Supplementary material, Fig. S1).⁴⁰ Interestingly, once mDCs were accounted for within the CD8α⁻ gate, four subpopulations of CD8α⁻ NK cells were still distinguishable based on their CD16 and CD56 expression patterns (see Supplementary material, Fig. S1c). Similar to previous reports, macaque mDCs were mostly CD56^dim CD16⁺ and CD56⁻ CD16⁻.^51,52 This observation explains the low proportion of cells within the CD8α⁻ gate that co-expressed perforin and granzyme B (Fig. 2b). It may also explain the relatively poor response of the CD8α⁻ cells to IL-2 and IL-15 stimulation in the phenotypic stability study (Fig. 6b–e), which is characterized by the persistence of CD8α^dim cells. Finally, given that only approximately 35% of the cells present in the CD8α⁻ gate are in fact NK cells, there would be a clear impact on the E : T ratios of cytotoxic assays. This might explain why killing with CD8α⁻ NK cells was only observed at higher E : T ratios (Fig. 5c,e).

The fact that macaque CD8α⁻ NK cells represent a small population with only about 50% expressing CD56 or CD16 (see Supplementary material, Fig. S1c), suggests that these cells may have an immediate lineage relationship with CD8α⁺ NK cells. Although the cells became activated in response to IL-15 stimulation (Fig. 3a), they exhibited low cytokine production in response to cytokine stimuli (Figs 3b,c and 4c). Despite this, CD8α⁻ NK cells also expressed significant levels of CD56, NKG2D, granzyme B, perforin and KIR2D, giving them all the requirements for cytotoxic activity. This activity was demonstrated unequivocally with functional experiments performed on enriched CD8α⁻ NK cells (Fig. 5c,e). Furthermore, as shown in Fig. 6, their stable phenotypic signature and the absence of any shift in CD8α expression with cytokine stimulation clearly supports the contention that CD8α^– NK cells represent a distinct cell population rather than one that simply evolves from CD8α⁺ cells.

To explore the potential of CD8α⁻ cells for functional activity, we evaluated cytokine production by both flow cytometry and transcription of cytokine genes by real-time PCR. The results for TNF-α were modestly positive by both methods, showing an upward trend for TNF-α production by flow cytometry (Fig. 3c) and increased transcription of the TNF-α gene following cytokine stimulation (Fig. 5b). Results for IFN-γ, however, showed different outcomes by the two methods. Reverse transcription-PCR analysis for the expression of the IFN-γ gene on both CD8α⁻ and CD8α⁺ sorted NK cells showed significant up-regulation in response to 5 hr of IL-2 plus IL-15 stimulation (Fig. 5b). However, by FACS analysis, CD8α⁻ NK cells exhibited only a modest up-regulation of IFN-γ production following co-culture with target cells (Fig. 4c). The rapidity of IFN-γ gene transcription is consistent with reports showing that unlike T cells, which exhibit a delay in T-cell activation and function, NK cells are designed for a very rapid response. In the murine system, IFN-γ production is observed after only 4 hr of cytokine stimulation.⁵³ The difference observed here by flow cytometry in the two NK subpopulations suggests a difference in kinetics of IFN-γ protein expression that will require further investigation. It is important to mention that although a significant proportion of mDCs is present in the enriched CD8α⁻ NK cells used for the reverse transcription-PCR assays, mDCs do not up-regulate IFN-γ production even in the presence of strong chemical stimulation such as PMA and ionomycin.⁴⁰

In terms of cytotoxic potential, both NK cell subsets were positive for perforin and granzyme B expression, although to different degrees (Fig. 2) and both exhibited transcription of perforin and granzyme B mRNA (Fig. 5b). The increased transcription observed between un-treated and cytokine-treated cells, however, was very low. Both perforin mRNA and protein have been reported to be constitutively present in human NK cells and other types of CTLs, with gene transcription only up-regulated under long-term stimulating conditions.⁵⁴ Therefore, it appears that perforin mRNA transcripts were constitutively present in both the CD8α⁻ and CD8α⁺ NK cells of macaques, but were absent from B cells. Moreover, the stimulation approach used in the present study did not further increase perforin gene transcription. With regard to granzyme B, no increase in transcription relative to that of B cells was observed (Fig. 5b). However, human B cells can produce granzyme B in response to cytokine stimulation,⁵⁵ which may be the case for macaque B cells as well. Overall, NK cells rely on pre-formed granules of perforin and granzymes to respond rapidly and exert cytotoxic function.^56,57 The co-expression of these two cytotoxic proteins in approximately 10% of CD8α⁻ NK cells (Fig. 2c) provides the potential for cytotoxic activity. In fact, the CD8α⁻ NK cells exhibited reduced, albeit significant, killing capacity when compared with similarly purified CD8α⁺ NK cells, both by direct lysis of cells lacking MHC class I expression (Fig. 5c) and by antibody-dependent killing (Fig. 5e). This decreased capacity to mediate cytotoxic function probably reflects the relatively large proportion of mDCs present in the enriched CD8α⁻ NK cell population, which significantly alters the E : T ratios.

Interestingly, certain subsets of mouse and human DCs have recently been reported to be capable of exerting direct in vitro and in vivo cytotoxic activity against cancer cells (killer DCs or kDCs), which is mediated through TRAIL and/or Fcγ receptor up-regulation on their cell surface.⁵⁸ Although kDCs are capable of cytotoxic function, their differentiation into a killer phenotype is largely dependent on the presence of stimulatory factors such as lipopolysaccharide, IL-15, IFN-α or IFN-γ,^59,60 which were not used in any of our cytotoxic functional studies using enriched CD8α⁻ and CD8α⁺ NK cells (Fig. 5c,e). Given this, we believe that the capacity of CD8α⁻ NK cells to mediate modest (albeit significant) cytotoxic function is in direct correlation to their activation profile and expression of cytotoxic proteins, and not to the potential acquisition of a killer phenotype by mDCs.

Evaluation of PBMCs from SIV-infected macaques for CD8α⁻ NK cells showed that these cells, and their CD16/CD56 subpopulations, are present at frequencies similar to those in naive animals (Fig. 7a,c). On the other hand, we detected a significant decrease in the frequency of CD8α⁺ CD16⁺ NK cells, which was accompanied by a significant increase in the proportion of CD8α⁺ CD56⁻ CD16⁻ NK cells (Fig. 7b). Interestingly, when comparing CD16/CD56 subpopulations within CD8α⁻ NK cells of naive and SIV-infected macaques, we also observed a decrease in the proportion of CD8α⁻ CD16⁺ cells and a concomitant rise in the proportion of CD8α⁻ CD56⁻ CD16⁻ NK cells, although these changes did not reach statistical significance (Fig. 7c). This observation suggests that during SIV infection, loss of CD3⁻ CD16⁺ cells affects both CD8α⁻ and CD8α⁺ NK cell subsets. Our results are in line with previous descriptions of HIV patients, where CD3⁻ CD8⁺ CD16⁺ NK cells are depleted despite an overall increase in CD8⁺ lymphocytes.^61,62

The ability of CD8α⁻ NK cells to mediate ADCC activity during adaptive immune responses when anti-viral antibodies are present, could contribute significantly to disease prevention and control.^19,21,24 Stratov et al.⁶³ have shown that robust ADCC responses, targeted mainly towards the Env protein, are observed in HIV-infected subjects. Importantly, the effector cells identified were of the CD3⁻ CD4⁻ CD8⁻ CD14⁻ CD2⁺ CD56^+/− phenotype, which is strikingly similar to the phenotype we describe here for macaque CD8α⁻ NK cells. Despite the significant presence of mDCs in the CD8α⁻ NK cell gate, our results are in line with those reported by Rutjens et al.³⁴ and Reeves et al.,⁴⁰ and confirm the presence and functional capacity of a CD8α⁻ NK cell population in rhesus macaque PBMCs.

Natural killer cells express a wide variety of chemokine receptors and tissue-homing molecules that influence their tissue distribution and migratory potential.²⁹ Chronic SIV infection has been shown to enhance the expression of the gut-homing marker α₄/β₇ in different subsets of NK cells.⁴⁷ It will be of interest to analyse the chemokine-receptor and tissue-homing molecule expression profiles of this novel subpopulation of circulatory CD8α⁻ NK cells in naive and SIV-infected macaques. Together with their potential for antibody-dependent and antibody-independent killing, evaluation of the distribution and function of CD8α⁻ NK cells in different tissues and mucosal effector sites will help to clarify their role in innate and adaptive immune responses.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Figure S1. CD8_α⁻ macaque NK cells represent 35 percent of CD3⁻CD8^α− lymphocytes and express both CD56 and CD16.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than about missing material) should be directed to the corresponding author for the article.