Use of an anti-CD16 antibody for in vivo depletion of natural killer cells in rhesus macaques

Abstract

Non-human primates serve as key animal models for a variety of viral infections. To evaluate the contribution of natural killer (NK) cells to the immune-mediated control of these viruses in macaque monkeys, we have described a method for depleting NK cells in vivo by administration of anti-human CD16 mouse monoclonal antibody. Using a fluorometric NK-cell cytotoxicity assay, we show that most NK-cell cytotoxicity in rhesus monkey peripheral blood mononuclear cells resides in the CD16⁺ and/or CD159A⁺ subset of lymphocytes. The anti-human CD16 antibody, 3G8, binds to subsets of rhesus monkey lymphocytes and monocytes but not to neutrophils. Intravenous administration of 10–50 mg/kg of 3G8 to normal rhesus monkeys resulted in anti-CD16 antibody persistence in the plasma for 1–3 weeks. This treatment also depleted 80–90% of CD3^– CD159A⁺ lymphocytes, putative NK cells, from blood for at least 1 week and was associated with the loss of NK-cell cytotoxicity when evaluated by in vitro assays. Using this method, transient depletion of NK cells from two rhesus monkeys chronically infected with simian immunodeficiency virus failed to cause changes in virus replication. These studies describe a non-human primate model for in vivo NK-cell depletion and suggest a limited role for cytotoxic CD16⁺ NK cells in controlling AIDS virus replication during chronic infection.

Introduction

Natural killer (NK) cells are a component of the innate immune system that is important in the control of viral infections and in host defence against tumour cells; it may also be involved in the rejection of transplanted tissues. Our understanding of the role of NK cells has been advanced through the clinical characterization of humans with deficiencies in NK-cell number or function.^1,2 More importantly, experimental data derived from mice treated with NK-cell-depleting antibodies,³ and from transgenic or knockout mice⁴ have helped to characterize NK-cell-mediated immunity and elucidate its role in viral infections. Unfortunately, easily manipulable rodent models do not exist for a number of infectious agents such as human immunodeficiency virus (HIV). Thus, non-human primates often serve as important animal models for studying pathogenesis and immunoprophylaxis of many infectious diseases. Since transgenic and knockout primates do not exist, we and others have turned to the use of monoclonal antibodies to target and deplete macaque monkeys of selected T- and B-lymphocyte subsets to define the contribution of specific components of the acquired immune response to the control of a number of viral infections.^5–10 We, therefore, strove to develop a similar model using antibody to deplete NK cells in rhesus monkeys.

NK cells represent a diverse lymphocyte population that lacks a single immunophenotypic marker. In addition, considerable variation exists between the immunophenotype of human and macaque NK cells, most notably a relative lack of CD56 expression on macaque NK cells.^11–14 However, previous studies have indicated that CD16, the Fc receptor FcγRIII, is expressed on 90% or more of all rhesus lymphocytes with NK-cell function,^13,14 suggesting that this cell surface molecule may be a useful NK-cell marker. We, therefore, explored the feasibility of using an anti-CD16 antibody to target and deplete NK cells in rhesus monkeys.

In the present study, we show that the anti-human CD16 antibody 3G8 is capable of transiently depleting up to 90% of NK cells in the blood of rhesus monkeys and reducing NK-cell lytic activity in the blood to near background levels. In a pilot study, transient depletion of NK cells from rhesus monkeys chronically infected with simian immunodeficiency virus (SIV) did not affect virus replication, suggesting that cytotoxic CD16⁺ NK cells may not play an important role in controlling chronic AIDS virus infections.

Materials and methods

Animals and viruses

Rhesus macaques (Macaca mulatta) used in this study were maintained in accordance with the guidelines of the Committee on Animals for the Harvard Medical School, federal and state law, American Association for Accreditation of Laboratory Animal Care regulations, and the Guide for the Care and Use of Laboratory Animals (DHHS Publication No. (NIH) 85-23 Revised 1996). Animals infected with SIV were inoculated intravenously with uncloned SIVmac251 12 months before this study.

Monoclonal antibodies

The mouse monoclonal antibodies, 3G8¹⁵ or a mouse immunoglobulin Gκ (IgG1κ) isotype-matched irrelevant control antibody, were grown in serum-free medium and purified to > 98% on protein A–Sepharose. The endotoxin level was < 1 EU/mg for both antibody preparations. Rhesus macaques were administered antibody solutions, diluted in phosphate-buffered saline (PBS), by slow intravenous bolus injection at various dosages. Blood leucocyte subsets were monitored for 6 weeks or until the NK-cell number returned to baseline values.

Antibodies, flow cytometry and cell counts

For immunophenotyping peripheral blood mononuclear cells (PBMC), ethylenediaminetetraacetic acid-anticoagulated rhesus blood specimens were incubated with anti-CD3 (SP34, BD Biosciences, San Jose, CA), anti-CD4 (L200; BD Biosciences), anti-CD20 (B1, Beckman Coulter, Fullerton, CA), anti-CD16 (3G8, BD Biosciences or DJ130c, Dako, Glostrup, Denmark), anti-CD159A (NKG2A, Z199, Beckman Coulter) and anti-CD8α (7PT-3F9), then red blood cells were lysed using an ImmunoPrep System (Beckman Coulter), washed with PBS and resuspended in PBS/1% formalin. Samples were analysed using a FACSCalibur and data were reanalysed using CellQuest software (Becton Dickinson).

Neutrophil and monocyte numbers in blood were determined from the complete blood count performed on an automated hematology analyser with rhesus monkey-specific automated leucocyte differential software (advia 120, Bayer, Tarrytown, NY).

Calcein release NK cytotoxicity assay

K562 cells (American Type Culture Collection, Manassas, VA) were loaded with calcein to serve as NK target cells. A 1 mm solution of calcein AM (Molecular Probes, Eugene, OR) was prepared by reconstituting 50 μg calcein AM in 25 μl of 20% dimethylsulphoxide and 25 μl Pluronic F-127 (Molecular Probes). Ten microlitres of this 1 mm calcein AM solution was added to 10⁶ target cells in 1 ml RPMI-1640 (without phenol red)/10% fetal bovine serum (FBS) and incubated at 37° in 5% CO₂ for 90 min. Target cells were then washed twice with RPMI-1640/10% FBS and resuspended in the same medium.

Effector cells, either PBMC or NK-92 cells (ATCC), were mixed with 10⁵ target cells at variable effector : target (E : T) ratios in a final volume of 200 μl using 0·3 ml eight-tube strips (Continental Lab Products, San Diego, CA). Each E : T ratio was tested in triplicate. After cell mixing, cell suspensions were centrifuged at 125 g for 3 min and then incubated at 37° in 5% CO₂ for 4 hr. The spontaneous release of calcein was determined by incubating loaded target cells in medium alone and maximal release was determined by adding 2% Triton-X to lyse all the target cells. After completion of incubation, tube strips were centrifuged at 400 g for 8 min, and 100 μl supernatant from each sample was transferred to a 96-well plate (Optiplate™ 96F, Perkin Elmer, Fremont, CA) and fluorescence was measured on a fluorometer (Victor-3, Perkin Elmer) at an excitation wavelength of 494 nm and emission wavelength of 517 nm. The median value for each triplicate was used in the calculation of cytotoxicity. Cytotoxicity, measured as per cent specific release of calcein, was calculated using the following formula:

Effector cell preparation and fractionation

The PBMC were isolated from fresh, heparinized blood specimens obtained from normal rhesus macaques by density gradient centrifugation. They were either maintained unfractionated for use in cytotoxicity assays or were fractionated into NK-cell-enriched and NK-cell-depleted fractions by incubation with phycoerythrin (PE)-conjugated anti-CD16 (3G8, BD Biosciences) and anti-CD159A (NKG2A, Z199, Beckman Coulter) antibodies, washed and incubated with anti-PE magnetic beads. Cells were then sorted using an autoMACS (Miltenyi Biotechnology, Auburn, CA) into CD16/CD159A-enriched or CD16/CD159A-depleted cell fractions. Some PBMC were incubated with only anti-PE magnetic beads but otherwise processed similarly through the autoMACS system. These cells served as a sham-sorted control cell population. To confirm the size of the NK-cell subset in each cell fraction, cells were stained with anti-CD3-allophycocyanin (SP34, BD Biosciences) and anti-CD8-ECD (7PT-3F9) antibodies in addition to those described above.

Detection of circulating mouse antibody and anti-mouse immunoglobulin antibody

To detect the persistence of 3G8 in the blood, plasma specimens from antibody-treated monkeys were incubated with normal rhesus PBMC and then stained with a secondary goat anti-mouse PE-conjugated antibody (Jackson ImmunoResearch, West Grove, PA) to detect 3G8 binding to NK cells. Samples were analysed by flow cytometry as described above. This assay had a limit of 3G8 detection in plasma of 100 ng/ml.

To detect anti-mouse immunoglobulin antibodies in monkey plasma, 96-well enzyme-linked immunosorbent assay (ELISA) plates were coated with 3G8 and incubated overnight at 4°. Plates were blocked with blocking reagent buffer (Pierce, Rockford, IL) for 15 min at room temperature. Plasma samples from antibody-treated rhesus monkeys were diluted in dilution buffer (PBS/0·5% non-fat dry milk), applied to the wells in serial dilutions, incubated at room temperature for 1 hr and washed with washing buffer (PBS/0·5% non-fat dry milk/0·01% Tween-20). Goat anti-human IgG–horseradish peroxidase (Jackson ImmunoResearch), at 1:30 000 in dilution buffer, was added to each well, incubated at room temperature for 1 hr and washed with washing buffer. Tetramethylbenzidine microwell peroxidase substrate (Kirkegaard & Perry Laboratories, Gaithersburg, MD) was added to each well, incubated at room temperature for 10–30 min and then the reaction was stopped with H₂SO₄ (Stop Solution, Kirkegaard & Perry Laboratories). Optical density readings were measured on an ELISA reader at 450 nm. The titre of anti-mouse immunoglobulin antibody was determined as the dilution that gave an optical density reading 0·1 units above the pretreatment sample for each animal.

Plasma SIV RNA levels

Plasma SIV RNA levels were measured by an ultrasensitive branched DNA amplification assay with a detection limit of 125 copies per ml (Bayer Diagnostics, Berkeley, CA).

Results

Validation of a fluorescence-based in vitro assay of NK-cell function

We employed a fluorescence-based assay to assess NK-cell function in rhesus PBMC using a method similar to that previously described.^16–18 This assay was validated for specificity and sensitivity by first assessing the lysis of target K562 cells loaded with the fluorescent molecule calcein by the human interleukin-2-dependent NK-cell line, NK-92. As expected, we could detect NK-92-mediated lysis of K562 targets in a dose-dependent fashion and at low E : T ratios (Fig. 1a).

Figure 1.

Validation of fluorometric natural killer (NK)-cell cytotoxicity assay. (a) Calcein-loaded K562 target cells were lysed by NK-92 effector cells in a dose-dependent manner. (b) Rhesus peripheral blood mononuclear cells (PBMC) were fractioned into NK-cell-depleted or NK-cell-enriched using anti-CD16 and anti-CD159A antibodies with immunomagnetic beads. The efficiency of cell fractionation is shown. (c) Lysis of K562 target cells by unfractionated, sham-sorted or NK-cell-depleted PBMC. Depletion of cells expressing CD16 and CD159A eliminated K562 cell lytic activity. (d) Lysis of K562 cells by the NK-cell-enriched PBMC fraction. CD16⁺ and/or CD159A⁺ effector cells lysed target cells at low effector : target (E : T) ratios.

We and others have observed that a high proportion of rhesus non-T, non-B lymphocytes express CD16 and CD159A.^13,19 We next assessed the ability of unfractionated rhesus PBMC and PBMC fractionated to deplete or enrich NK cells to lyse calcein-loaded target cells. Using total rhesus PBMC (approximately 20% CD16⁺ CD159A⁺ cells; Fig. 1b), or sham-sorted PBMC that were processed through the same magnetic bead system without primary antibodies, we detected 35–45% specific lysis of target cells at E : T ratios of 25 : 1 and 50 : 1 (Fig. 1c). Rhesus monkey PBMC were depleted of putative NK cells using anti-CD16 and anti-CD159A antibodies followed by magnetic beads to produce a population of cells that was < 1% CD16⁺ CD159A⁺ (Fig. 1b). Depletion of CD16⁺ and CD159A⁺ cells resulted in an almost complete loss of the ability of rhesus PBMC to lyse target cells (Fig. 1c). In contrast, positively selected rhesus effector NK cells (enriched to 75% CD16⁺ CD159A⁺; Fig. 1b) showed substantial lysis of K562 cells at E : T ratios that were five- to 10-fold lower than unsorted or sham-sorted effector cells (Fig. 1d). These results validate the use of this fluorescence-based NK-cell assay and indicate that the cytotoxic rhesus monkey NK-cell subset is largely contained within the CD16/CD159A-expressing lymphocyte fraction.

In vivo depletion of NK cells using a murine anti-human CD16 monoclonal antibody

CD16, the Fc receptor, FcγRIII, is expressed in humans as a membrane-bound form on lymphocytes and monocytes, and as a glycosylphosphatidylinositol (GPI)-linked form on neutrophils.²⁰ However, in Old World monkeys, expression of CD16 in leucocyte subsets is more restricted.²¹ While the anti-CD16 antibody, 3G8, binds to a subset of rhesus lymphocytes and monocytes at proportions similar to those seen in humans, it fails to bind to rhesus neutrophils (Fig. 2). Thus, the more restricted cell specificity of this anti-CD16 antibody in rhesus monkeys made it a good candidate for specific in vivo targeting of NK cells.

Figure 2.

Anti-CD16 monoclonal antibody, 3G8, binds to a subset of rhesus lymphocytes and monocytes but not to neutrophils. Binding of 3G8 to blood leucocytes was assessed after staining with 3G8 using a whole blood lysis technique. Lymphocytes, monocytes and neutrophils were identified based on light-scatter characteristics. Antibody 3G8 bound to a subset of lymphocytes and monocytes but not to neutrophils.

To test the ability of this antibody to deplete rhesus NK cells in vivo, we administered escalating doses of 3G8 intravenously to normal rhesus monkeys. For this study, where anti-CD16 antibody was administered, NK cells in blood were defined as CD3⁻ CD159A⁺ lymphocytes to avoid cross-blocking of CD16 by the administered antibody. Doses as low as 10 mg/kg (divided between day 0 and day 2) depleted 80% of NK cells in blood within one day (Fig. 3). A single 20 mg/kg dose produced similar depletion kinetics whereas treatment with 50 mg/kg of anti-CD16 depleted approximately 90% of NK cells from blood (Fig. 3). These levels of NK-cell depletion persisted for 1 week after administration of the monoclonal antibody. In all animals, the NK-cell number in blood rebounded at variable rates during the second week post-treatment.

Figure 3.

Effect of anti-CD16 monoclonal antibody on natural killer (NK) cell number. Normal rhesus monkeys, one monkey per dose, were administered 10 mg/kg, 20 mg/kg, or 50 mg/kg of mouse monoclonal antibody, 3G8. NK-cell number, defined as CD3^– CD159A⁺, was determined in whole blood specimens. All dosages resulted in 80–90% depletion of NK cells for at least 1 week.

Although neutrophil and monocyte numbers in blood varied considerably from day-to-day, neutrophil numbers showed no consistent change during the course of treatment (Table 1), as predicted by the lack of neutrophil binding by 3G8. Monocyte number did decline by 25% or less at the 10 mg/kg and 20 mg/kg dosages, and by approximately 50% at the 50 mg/kg dose (Table 1). Neither CD4⁺ and CD8⁺ T-cell subsets nor B cells showed significant changes in the blood in response to 3G8 treatment (data not shown).

Table 1.

Changes in leucocyte number during 3G8 administration

Time-point	NK cells	Lymphocytes	Neutrophils	Monocytes
	Median cells/μl (range)1
10 mg/kg
Pretreatment	262 (111–413)	5470 (3490–5530)	2250 (1556–3622)	424 (284–430)
During depletion	55 (35–67)	3397 (3289–4335)	3199 (1703–4965)	320 (281–446)
Recovery	342 (N/A)	3734 (3190–4279)	2695 (2593–2797)	388 (308–469)
20 mg/kg
Pretreatment	568 (531–727)	3430 (2920–3670)	3502 (2290–3832)	361 (282–580)
During depletion	127 (66–221)	2333 (1997–3034)	4259 (1668–7238)	258 (204–842)
Recovery	468 (350–701)	3107 (2640–3935)	3383 (2566–3880)	472 (207–678)
50 mg/kg
Pretreatment	655 (409–983)	3910 (3690–4130)	2024 (1872–2175)	441 (415–467)
During depletion	33 (20–64)	2637 (1900–2946)	4150 (1839–5252)	244 (228–275)
Recovery	115 (86–216)	3374 (2689–3980)	1672 (1045–5011)	201 (175–311)

¹For all experiments, numbers of observations used to calculate median (range) were: pretreatment, n = 2 to n = 4; during depletion, n = 4; recovery n = 2 to n = 6.

We measured the persistence of 3G8 in the plasma of monkeys using a cell-binding assay to ensure that circulating antibody detected retained ligand-binding ability. As shown in Fig. 4, the length of time that 3G8 was detectable in plasma was dose-dependent. At the highest dose, 3G8 capable of binding to target cells was detectable for 3 weeks.

Figure 4.

Persistence of 3G8 in plasma and emergence of anti-mouse immunoglobulin antibodies. Antibody persistence in plasma was dose-dependent. Anti-mouse immunoglobulin antibodies first appeared at approximately 1 week post-treatment irrespective, of dosage.

Because we reasoned that the duration of target cell depletion would most likely be limited by the monkey's immune response to foreign mouse IgG, we also assessed the anti-mouse immunoglobulin antibody response. The emergence of this anti-mouse immunoglobulin response was detectable at about 1 week post-treatment at all dosages (Fig. 4) and coincided with the recovery of NK-cell numbers in the blood (Fig. 3). However, clearance of 3G8 was only observed at higher anti-mouse IgG titres. The detection of higher anti-mouse immunoglobulin titres was substantially delayed at the higher antibody dosages.

Loss of NK-cell function in vitro correlation with in vivo NK-cell depletion

To confirm that depletion of CD16⁺ lymphocytes abrogated NK-cell functional activity, we evaluated the in vitro NK-cell cytotoxicity in PBMC of a monkey treated with 20 mg/kg anti-CD16 monoclonal antibody. NK-cell cytolytic activity in the blood of this 3G8-treated monkey was substantially reduced at day 2 and day 9 post-treatment compared to pretreatment levels (Fig. 5). This corresponded to the period of most pronounced NK-cell depletion in this animal (Fig. 3). The NK-cell cytolytic activity in the blood returned to baseline levels by day 23 (Fig. 5), at which time the NK-cell population had also recovered (Fig. 3). These results confirmed the feasibility of using an anti-human CD16 monoclonal antibody to transiently deplete NK cells and suppress NK-cell lytic activity from the blood of rhesus monkeys.

Figure 5.

In vitro depletion of CD3^– CD159A⁺ lymphocytes after administration of 3G8 correlated with loss of K562 lytic activity. Natural killer (NK) -cell activity was measured in peripheral blood mononuclear cells (PBMC) of a rhesus monkey treated with 3G8 (20 mg/kg). Loss of NK-cell lysis of K562 cells was associated with depletion of the CD3^– CD159A⁺ subset (Fig. 3). The pretreatment value represents the mean of two assays performed during the 4 weeks before antibody treatment.

NK-cell depletion in chronic SIV infection

To directly examine the role of NK cells in controlling virus replication in AIDS in chronic infection, we performed a pilot study to assess the effect of NK-cell depletion on rhesus monkeys chronically infected with SIV. The rhesus monkeys used in this study had been infected with SIVmac251 for 12 months before this study and had stable levels of plasma virus and CD4⁺ T cells. Two of the four animals studied had active virus replication and detectable levels of plasma viraemia. The other two animals had controlled their virus replication and plasma virus was < 125 RNA copies/ml. Animals were administered a single 10 mg/kg dose of anti-CD16, or control monoclonal antibody, on day 0. CD4 lymphocyte count and plasma virus levels were monitored. Both anti-CD16-treated monkeys showed rapid depletion of 80–90% of NK cells from blood (Fig. 6a). However, in contrast to the uninfected rhesus macaques, the duration of NK-cell depletion was shorter in these SIV-infected animals. Monkeys that received the control antibody showed no depletion of the NK-cell subset (Fig. 6a). Surprisingly, in both SIV-infected animals that received 3G8, mouse immunoglobulin could not be detected in the plasma after days 3 to 4, although anti-mouse immunoglobulin titres first emerged more than 7 days post-treatment (data not shown). This was in contrast to the uninfected rhesus macaques, in which anti-CD16 monoclonal antibody was detectable in the plasma for at least 8 days. This more rapid clearance of 3G8 from SIV-infected animals may have been responsible for the shorter duration of NK-cell depletion.

Figure 6.

Natural killer (NK) cell depletion of macaques chronically infected with simian immunodeficiency virus (SIV) did not affect virus replication. (a) Change in CD3^– CD159A⁺ NK-cell number following treatment with 3G8 or control antibody. (b) Plasma virus level in chronically SIV-infected macaques during NK-cell depletion.

Plasma levels of SIV (Fig. 6b) and CD4 T-cell counts (data not shown) measured during transient NK-cell depletion showed no changes from baseline measurements. Plasma SIV levels remained undetectable in the NK-cell-depleted monkey whose viraemia was below detection before treatment. These results are consistent with a limited role for CD16⁺ NK cells in the control of chronic SIV viraemia.

Discussion

Natural killer cells play a key role in the host defence response against many viral infections and malignancies. The use of mouse or recombinant monoclonal antibodies capable of targeting and depleting T- and B-lymphocyte subpopulations in non-human primates has proven to be an effective method for defining the contribution of these cells in the control of various infections.^5–10 In the present studies, we have validated the use of an anti-CD16 monoclonal antibody to target and deplete NK cells and eliminate NK-cell lytic function in the blood of rhesus monkeys.

We implemented and validated a non-radioactive, fluorometric in vitro NK-cell cytotoxicity assay which was sensitive and specific for NK-cell-mediated target cell lysis. We showed that in vitro depletion of CD16⁺ CD159A⁺ cells could eliminate NK effector cell cytotoxicity from rhesus monkey PBMC, which is consistent with earlier studies correlating NK-cell function with cell immunophenotype.^13,14,20

We next evaluated the anti-human CD16 monoclonal antibody 3G8 as a candidate for depleting NK cells in vivo in rhesus monkeys. Unlike humans, rhesus neutrophils were not recognized by 3G8. A search of a rhesus genome database²² found open reading frames encoding proteins with 95% amino acid homology to both the transmembrane and GPI-linked forms of human CD16. The transmembrane and GPI-linked forms of rhesus CD16 differed by only one amino acid between their common regions, suggesting that the same anti-CD16 antibody would probably recognize both forms of rhesus CD16. This finding implies that, unlike humans, rhesus neutrophils may not express a GPI-linked form of CD16. This limited specificity of 3G8 for rhesus blood cells resulted in neutrophils being preserved following anti-CD16 treatment.

The anti-CD16 antibody 3G8 does bind to a subset of rhesus monocytes. A subsequent study has shown that treatment with 50 mg/kg of 3G8 resulted in 30–70% loss of CD16-bearing monocytes when assessed by a non-cross-blocking anti-CD16 antibody. The CD16⁺ monocyte subset in primates is most probably the equivalent of mouse monocytes that express high levels of CXCR1.²³ This monocyte subset does not express receptors for inflammatory chemokines and lacks classical inflammatory effector function. Nevertheless, the transient loss of CD16⁺ monocytes must be considered when interpreting data resulting from the use of this antibody in vivo.

Dosages of 10–50 mg/kg depleted 80–90% of NK cells and resulted in a loss of NK-cell cytolytic activity from blood. However, evidence suggests that the highest dose may deplete target cells more efficiently and for a longer duration. A small subset of phenotypically defined NK cells appeared resistant to depletion and may represent the minor CD16⁻ NK-cell fraction previously reported.¹³ In humans, the CD16^lo NK-cell subset has less cytolytic activity but increased production of cytokines and chemokines when compared to the CD16^hi NK cells.²⁴ This difference in NK-cell effector function based on CD16 expression is probably present in macaques as well.¹³ No adverse effects were observed in any monkeys treated with this antibody, confirming the safety of this treatment.

As expected, anti-mouse immunoglobulin antibody responses arose during the second week after treatment and were associated with recovery of NK cells. However, the magnitude of the anti-immunoglobulin response was temporarily overcome at the highest 3G8 dosage. Mouse/rhesus recombinant forms of 3G8 hold the potential to be less antigenic than mouse immunoglobulin and, therefore, have an extended duration of in vivo effect. Recombinant forms of 3G8 are currently being evaluated in monkeys.

Numerous qualitative and quantitative defects in NK-cell function have been described in HIV-infected humans.^25–27 However, the precise role that these cells play in controlling virus replication remains poorly defined. We, therefore, undertook a pilot study to determine whether NK-cell depletion in monkeys chronically infected with SIV could impact virus replication. Although NK-cell depletion was not optimal in this study, treatment with 3G8 did deplete 80% of NK cells in treated monkeys and levels remained below 50% for 3 days. The mechanism responsible for the shorter NK-cell depletion in these two SIV-infected monkeys is unclear. Our data suggest that anti-CD16 antibody was cleared more rapidly from the plasma of SIV-infected monkeys than from normal monkeys and that this clearance was not immune-mediated. However, immune complexes in the serum of SIV-infected monkeys have also been shown to block binding of 3G8 to CD16 in vitro.²⁸ Thus, a similar binding inhibition in vivo could potentially affect 3G8-mediated depletion of target cells as well.

Despite a brief but substantial reduction in CD16⁺ NK-cell number following 3G8 treatment, virus replication remained unchanged. However, NK cells possess both cytolytic and non-cytolytic effector function. C–C chemokines secreted by NK cells have shown potent inhibition of HIV replication in vitro.^29,30 It is possible that CD16^lo NK cells, which have poor cytotoxic activity but enhanced ability to secrete cytokines and chemokines, are preserved during 3G8 treatment. The persistence of this minor CD16^lo NK-cell subset could exert an antiviral effect through a chemokine-mediated mechanism despite the depletion of CD16^hi NK cells. However, these findings are consistent with a minimal contribution of cytolytic CD16⁺ NK cells in the control of virus replication in chronic virus infection in AIDS.