Natural cytotoxicity receptor
Signals leading to NK cell triggering are primarily mediated by natural cytotoxicity receptors (NCR) upon binding to as-yet-undefined cell surface ligand(s) on normal hematopoietic cells, pathogen-infected cells or tumor cells. In this study we tried to determine whether the decreased NK cell cytolytic function that is observed in HIV-1-infected patients may be related to a decreased expression of NCR. In HIV-1-infected patients, freshly drawn, purified NK cells expressed significantly decreased surface densities of NKp46 and NKp30 NCR. The low surface density of NKp46, NKp30 and NKp44 was also confirmed in in-vitro-activated NK cell populations and NK cell clones derived from HIV-1 patients compared with uninfected donors. This defective NCR expression in HIV-1 patients was associated with a parallel decrease of NCR-mediated killing of different tumor target cells. Thus, the present study indicates that the defective expression of NCR represents at least one of the possible mechanisms leading to the impaired NK cell function in HIV-1 infection and it can contribute to explain the relatively high frequency of opportunistic tumors reported in cohorts of untreated patients before the occurrence of profound immunosuppression (<200 CD4+ cells/mm3).
The main feature of HIV-1 infection is represented by a progressive decrease in CD4+ T-helper lymphocytes, leading to the impairment of adaptive immunity in the presence of active, chronic viral replication. This pattern is only partly affected by a vigorous immune response encompassing HIV-specific adaptive responses, both cell-mediated (e.g. by CD8+ CTL) and humoral (e.g. by neutralizing-antibodies and antibody-dependent cell cytotoxicity [ADCC] antibodies) 1, 2.
As HIV-1 infection progresses there is, however, also a decrease of innate immune mechanisms, including a defective NK cell function and number. This decreased NK cell activity has been demonstrated at all stages of HIV-1 progression, is present early after infection, and is particularly relevant in individuals with opportunistic infections or with subsequent progression to Kaposi'ssarcoma or to AIDS 3–9.
An impairment of NK cell function can also be detected at the clonal level 10, and is more pronounced for natural cytotoxicity than for ADCC 11, 12. In fact, in this respect, cross-linking of CD16 on peripheral NK cells in HIV-1-infected patients is still efficient in inducing considerable chemokine production in vitro and in inhibition of virus entry into CD4+ T cells 13. An altered expression of surface molecules such as CD16, CD56 and CD8, that are not unique to NK cells, as well as a decrease in the proportion of CD16+CD56+ and CD16+CD8+ cells, has been reported in most studies during the course of HIV-1 progression 5–12, 14, 15. In addition, either non-productive or productive infection of NK cells with HIV-1 has been shown to occur and has been suggested to represent a potential viral reservoir 16, 17. However, the relevance of the infection of a minor fraction of CD4+ NK cells is as yet unexplored and may not significantly affect the function of the remaining NK cells.
Indeed, despite these successful characterizations, the mechanism(s) underlying the defect in NK cell function during HIV-1 infection are still elusive. In this regard, qualitative rather than quantitative defects of NK cells have been suggested to occur in HIV-1-infected patients. 8, 15.
NK cell recognition and function in humans are regulated by multiple cell-surface receptors, including inhibitory NK receptors that bind to HLA class I and deliver inhibitory signals to NK cells to prevent NK-cell-mediated attack of autologous normal cells 18, 19. The "on" signals leading to NK cell triggering are primarily mediated by natural cytotoxicity receptors (NCR) upon binding to as-yet-undefined cell surface ligand(s) expressed on normal hematopoietic cells as well as on pathogen-infected or tumor cells 20, 21. Importantly, a strict correlation has been determined between NCR density and NK-cell-mediated cytolytic activity 22, 23. Thus, NK cells expressing high NCR surface density (NCRbright) displayed strong cytolytic activity, whereas those expressing low NCR surface density (NCRdull) were poorly cytolytic or even non-cytolytic against most target cells.
In view of the major role played by the NCR in the triggering of natural cytolytic activity of NK cells 20, 21, 24, we assessed whether the decreased NK cell function that is observed in HIV-1-infected patients could be related to a decreased expression of NCR molecules (i.e. NKp46, NKp30 and NKp44). For this purpose, we analyzed PBMC-derived purified NK cells, both unstimulated and after in vitro activation, for the expression of NCR molecules and for cytolytic activity against different tumor targets.
We here report that in HIV-1-infected patients, a significant proportion of fresh peripheral blood NK cells have decreased expression of the activating receptors NKp46 and NKp30. Further, in vitro NK cell activation does not lead to increased surface expression or to cytolytic activity comparable with healthy donors. Thus, the mechanism underlying the decreased NK cell activity in HIV-1-infected patients could reflect, at least in part, the down-regulation of the triggering receptors that are involved in natural cytotoxicity.
2.1 Expression of NCR on purified peripheral blood NK cells is reduced in HIV-1-infected patients
The majority of the work performed so far on NK cells in HIV-1 infection has analyzed relevant NK cell surface antigens and NK cell function in unseparated PBMC. However, this strategy suffers from some limitations due to the expression on other cell lineages (e.g. T lymphocytes and monocytes) of surface antigens that are typically present on NK cells (e.g. CD56, CD16 and inhibitory NK receptors). In the present study, we analyzed purified NK populations obtained after depletion of CD3+CD4+CD19+ and CD14+ cells from PBMC.
We excluded from our study patients on antiretroviral treatment (ART), to avoid both possible direct effects of antiretroviral drugs on NK cells and indirect effects of the treatment on the innate immune response, associated with the immune reconstitution after ART. Given these particular constraints, we limited the study to 10 patients. All the patients (8 male, 2 female) were ART-naive, viremic, and in different stages of HIV-1 infection (6 were CDC A1 and 4 were CDC A2), with a wide range of PBMC CD4+ T cell numbers (median 475, range 228–750/mm3), viral load (mean 3.53 log10, range 1.90–6.08 log10 cp/ml) and time from their estimated seroconversion (mean 3.4 years, range 1–8 years).
We first analyzed whether the expression of NCR (NKp46 and NKp30) on purified, peripheral blood NK cells was different in the HIV-1-infected patients as compared with a group of nine healthy, uninfected, control donors, as determined by cytofluorometric analysis using mAb specific for NKp46 or NKp30. Cytofluorometric analysis of infected patients showed a phenotype of NK cells generally comparable to uninfected donors, as far as CD16 and CD56 expression was concerned. However, a significant proportion of patient NK cells expressed low fluorescence intensity of NCR (Fig. 1A). Overall, when analyzing the results from all the patients and control donors as a group, the so-called NKp46bright phenotype was expressed on a median of 63.5% of NK cells in the healthy donors as compared with 10.4% of the NK cells of HIV-1-infected patients (p<0.05) (Fig. 1B). The staining of purified NK cells with anti-NKp30 mAb was also significantly lower in HIV-1-infected patients as compared with healthy, uninfected donors (4.25% vs. 30.70%, respectively; p<0.005). Functional analysis of purified NK cells from the HIV-1-infected patients confirmed previous reports of a reduced lytic activity when compared with NK cells obtained from uninfected donors in a 4-h 51Cr-release assay against K562 target cells (58 vs. 136 lytic units20/106 PBMC) 5, 6.
Thus, peripheral blood NK cells of HIV-1-infected patients express a remarkably decreased surface density of NKp46 and NKp30 receptors as compared with healthy uninfected donors. Indeed, at least in untreated patients, the median expression of the NKp46 and NKp30 "bright" phenotype may represent up to 16% and 14%, respectively, of what was found in uninfected donors.
2.2 Decreased expression of NCR on in-vitro-activated NK cells from HIV-1-infected patients
We next studied whether the observed decreased expression of NCR on purified peripheral blood NK cells of HIV-1-infected patients also occurred after NK cell activation and culture in vitro. Purified NK cell populations were derived from HIV-1-infected patients or from uninfected donors in the presence of rIL-2, as described previously 22–24. As determined by cytofluorometric analysis, the polyclonal NK cell populations contained <8% CD3+ cells and <2% CD4+ cells. In addition, all the cultured cells displayed the CD16+CD56+ phenotype. Importantly, the analysis of NKp46 and NKp30 showed that these receptors were expressed at low levels in all the in-vitro-activated NK cell cultures derived from the HIV-1-infected patients. Moreover, the expression of the third NCR, NKp44, which is usually expressed by cultured NK cells, was also strongly reduced. Fig. 2 shows a representative cytofluorometric panel of five patients. In all cases, NKp46 was expressed at a lower level of fluorescence intensity than NKp30. As compared with the defect observed in fresh NK cells, the expression of NCR after in vitro culture of NK cells resulted in a partial recovery, at least in some patients. Notably, however, after culture in vitro the proportion of NK cells expressing low NCR surface density (NCRdull) was still strongly increased, as compared with control cultures (Fig. 2).
To further analyze the NCR expression at the single-cell level, we studied 98 NK cell clones derived from five of the HIV-1-infected patients, and 36 NK cell clones from three healthy, uninfected donors. All the clones were characterized by the CD16+CD56+CD3– phenotype, as determined by cytofluorometric analysis. On the basis of the fluorescence intensity after staining with anti-NCR mAb, all the clones could be distinguished as NCRbright or NCRdull (Fig. 3A). A consistent and significant difference was observed between HIV-1-infected patients and normal donors. In fact, in HIV-1 patients 85% of the NK cell clones belonged to the NCRdull phenotype, whereas these represented only 15% of the NK cell clones derived from healthy donors (p<0.0001, Fisher's test).
With regard to the mean fluorescence intensity (MFI) for the clones derived from HIV-1-infected patients as compared with healthy normal donors, the difference in MFI was relevant for NKp30 (median MFI 24 vs. 59; p<0.0001), for NKp46 (median MFI 11 vs. 26; p<0.0001), and for NKp44 (median MFI 20 vs. 44; p<0.001). As far as the pattern and intensity of NKp44 expression are concerned, although this NCR had a significantly reduced expression, there seemed to be a less pronounced deficiency of NKp44 expression in the clones when compared with NKp46 and NKp30 (Fig. 3A, and MFI data). This is in agreement with what was observed in in-vitro-activated NK cell populations (Fig. 2), where expression of NKp44 in some patients appeared to be less deficient than was observed for NKp46 and NKp30.
In conclusion, these results show that in-vitro-activated NK cells in HIV-1-infected patients have a considerably decreased expression of triggering NCR, consistent with the decreased expression of NKp46 and NKp30 molecules observed on freshly drawn, purified, peripheral blood NK cells. The decrease in NCR expression is also present after in vitro activation, although it is more pronounced for NKp46 and NKp30. Moreover this phenotype is not the result of in vitro selection as it can also be observed at the clonal level. A reduced expression of NCR on NK cells could be responsible, at least in part, for the decreased lytic activity against virus-infected or tumor cells expressing the appropriate ligand(s).
2.3 In-vitro-activated NK cells from HIV-1-infected patients display a reduced cytotoxicity against tumor target cells
Some "opportunistic" neoplasms are one of the hallmarks of AIDS in HIV-1-infected patients, and in some cases (e.g. Kaposi's sarcoma, extracerebral lymphomas and cervical cancer) these tumors may develop independently of the usual threshold for significant CD4+ cell count suppression (e.g. <200/mm3). The finding of a decreased expression and surface density of NKp46, NKp30 and NKp44 NCR on in-vitro-cultured NK cells from infected patients prompted us to verify whether this could have a functional correlate as determined by the impairment of tumor-target lysis.
First, different NK cell populations derived from the 10 HIV-1-infected patients and from 9 uninfected control donors were analyzed in a redirected killing assay. In these experiments, the cytolytic activity of NK cell populations was assessed against the FcγR+ P815 murine target cell line, either in the absence or in the presence of mAb directed to CD16, NKp46, NKp30 and NKp44. In all instances, effector:target cell ratios varied from 5.00:1.00 to 1.25:1.00. As shown in Fig. 4, there was a significant overall decrease in the median NCR-mediated lysis of P815 targets in the redirected killing assay. As determined by box-plot analysis, the median lytic activity of NK cell populations derived from the 10 HIV-infected patients in the presence of NKp46-, NKp30- or NKp44-specific mAb represented 50% of what was observed for the NK cell populations obtained from healthy controls. On the other hand, the redirected cytolytic activity of NK cells in the presence of anti-CD16 mAb in HIV patients was only marginally decreased (Fig. 4).
These findings of a decreased NCR-mediated lysis of target cells are consistent with the cytofluorometric analysis of the various NK cell populations showing a deficient expression of triggering NCR (Fig. 2), as well as with the decreased lysis of P815 target cells by NCRdull NK cell clones in a redirected killing assay (Fig. 3B). Interestingly, the NKp44-mediated cytotoxicity was significantly decreased, but was more broadly overlapping with healthy donors. This finding is in agreement with the less severely impaired expression of NKp44 observed in the NK cell populations and clones.
The same set of NK cell populations were then analyzed against human tumor cell targets that are known to express the putative NCR ligand(s) 20, 22. In these experiments, the NK cell populations were first assessed in a 51Cr-release assay against the M14 melanoma cell line. As shown in Fig. 5A, NK cell populations derived from HIV-1-infected patients failed to lyse M14 tumor cells, whereas those derived from uninfected donors displayed a strong cytotoxicity. The cytolytic assay was performed in the presence of anti-HLA-class-I mAb (of IgM isotype) to disrupt the interaction between the inhibitory NK receptors expressed on NK cells and HLA class I molecules on the surface of the M14 tumor target cells. Thus, the possibility of a functional inhibition of NK cells in HIV-1-infected patients by overexpression of HLA-class-I-specific inhibitory NK receptors on their surface could be ruled out in this experimental setting. In addition, in all cases, tumor target cell lysis by NK cells could be prevented by the addition of a mixture of NCR-specific (i.e. anti-NKp46, anti-NKp30 and anti-NKp44) masking mAb (IgM isotype) (Fig. 5A).
Finally, the reduced lysis of tumor target cells was not confined to M14 melanoma cells, but could also be shown for B cell tumors, such as Daudi and the EBV-transformed cell line 221 (Fig. 5B). In these experiments, NK cell populations derived from viremic HIV-1-infected patients showed a decreased lysis of the Daudi and B-EBV-transformed target cells when compared with NK cell populations derived from uninfected donors.
Altogether these results show that the NCRdull phenotype of the NK cell populations that are derived from viremic HIV-1-infected patients results in a reduced capability to kill tumor target cells of different histotypes, and suggest that this could play, at least in part, a pivotal role in the increased incidence of HIV-associated tumors that is seen in infected patients.
Innate immunity, and in particular NK cells, plays an important role in immune defense against viral infections, in tumor control and in graft-versus-leukemia reactions 20, 21, 25–28. In allogeneic bone-marrow transplantation, allogeneic NK cells play a pivotal role in the effective removal of residual leukemic cells (tumor burden) via graft-versus-leukemia effects 26–28. Moreover, a defect in the expression of NCR has recently been described in patients with acute myelogenous leukemia (AML) 25. This defect was associated with a sharp reduction of the NK-mediated cytolytic activity against autologous and allogeneic AML blasts. With regard to HIV-1 infection, evidence has accumulated regarding a well-characterized NK cell dysfunction that occurs early after infection and that may be associated with rapid progression to opportunistic infections and AIDS in some patients 2–15. Interestingly, the absence of an NK cell dysfunction has also been associated with reduced sensitivity to HIV-1 infection of certain exposed but uninfected individuals 29, and with prolonged asymptomatic periods in untreated patients with AIDS 30. The molecular mechanisms of the described NK cell defect during HIV-1 infection have, however, so far eluded a precise characterization, despite active interest and repeated efforts.
In the present study we show that, in untreated, viremic HIV-1-infected patients, both circulating NK cells and in-vitro-cultured NK cell populations and clones display a significant reduction in the expression of NCR (NKp46, NKp30 and NKp44). More importantly, the low density of surface NCR was associated with functional impairment of NK cells, which showed a strong reduction in their ability to kill tumor target cells.
We enrolled previously untreated HIV-1-infected patients with a relatively wide range of immunological and clinical disease spectrum, to characterize NK cell function in the absence of any direct or indirect effect due to antiretroviral treatment (e.g. immune reconstitution or direct effects on NK cell function and phenotype). In fact, although in patient NK cells the median NCR expression was significantly lower than in the healthy donor group, there was a considerable inter-individual variability of NKp46 and NKp30 expression on purified CD3–CD14–CD19– freshly drawn NK cells. This may reflect differences in the immune status and possibly in the disease progression rate within the present group of patients. In this regard, a recent report by Kottilil et al. shows an inverse relationship between the level of viremia and the degree of impairment of NK cell function, as determined by CC-chemokine production 31. Thus, as far as NCR expression on NK cells and NK cell cytolytic function are concerned, larger groups of patients and stratification based on viral load and CD4+ T cell counts would be needed to determine whether these are similarly influenced by viremia.
The present finding that the defect in NK cell triggering via NCR was more compromised than that via CD16 is in line with previous observations indicating that the NK cell defect in HIV-1 patients affects natural cytotoxicity to a larger extent than antibody-dependent cell cytotoxicity 10, 12. Indeed, the decreased expression of NCR in NK cell populations of HIV-1-infected patients was associated with significantly reduced responses to stimuli acting on NCR. These findings were confirmed by the defective ability to lyse tumor cell lines (M14, Daudi and 221) that are predominantly killed by NK cells upon NCR-mediated recognition 20. Thus, the decreased expression of NCR, both in vivo and in vitro, together with the decreased ability of NK cells to lyse tumor cell lines could, at least in part, help to explain the relatively high frequency of some "opportunistic" tumors (Kaposi's sarcoma, non-Hodgkin's lymphoma and cervical carcinoma) in cohorts of untreated patients 32. Interestingly in this respect, these AIDS-defining opportunistic tumors may occur in HIV-1-infected patients before the onset of a relevant defect of adaptive immune responses (i.e. <200 CD4+ T cells/mm3) that heralds opportunistic infections. Thus, the occurrence of thesetumors in HIV-1-infected patients could be related predominantly to a reduced NK cell-triggering associated with the presently described defect in NCR expression.
It is worth considering that, from our findings and also on the basis of previous reports 22, 33, the NCRdull phenotype is rather rare in uninfected individuals and normally applies to a minor NK subset. In HIV-1-infected patients, the fraction of the NCRdull population significantly exceeds that of NCRbright cells, reaching high levels (>80%) in most patients analyzed. Based on these differences between infected viremic and uninfected individuals, it is likely that the NCRdull phenotype may be either directly or indirectly caused by the HIV-1 infection status. Whether this is the result of a selection of NCRdull NK cells at the NK cell precursor level, or rather of a down-regulation inducedon mature NK cells will be addressed in future studies.
In certain pathological conditions, a decreased surface expression of NCR could result from an effect of the microenvironment, possibly mediated by cytokines. Recently, it has been shown that TGF-β down-regulates the surface expression of NKp30 and in part of NKG2D but not that of other triggering receptors such as NKp46 34. Interestingly, TBF-β has been shown to decrease HIV-1 replication in vitro35, and increased TGF-β production by PBMC has been shown to be induced by HIV-1tat and HIVenv36–38. Thus, TGF-β could be involved in the presently reported decreased expression of NKp30 on NK cells in vivo. On the other hand, no information has been provided so far that other cytokines can modulate the surface expression of the other triggering NK receptors (e.g. NKp46 and NKp44). Alternatively, an effect of the microenvironment on NK cell precursors could determine a skewed maturation of NK cells during HIV-1 infection, leading to the prevalent expansion of mature NK cells with a "dull" expression of NCR that is persistent after in vitro activation and culture of cells from HIV-1-infected patients and possibly in other conditions where NCR have a decreased expression on NK cells (e.g. acute myelogenous leukemia).
In conclusion, the present finding of a defect in NCR expression and in NCR-mediated lysis provides an explanation for at least one of the possible mechanism(s) responsible for the defective NK cell function in HIV-1-infected patients that has long eluded identification. Since, however, other triggering NCR co-receptors have been described on NK cells, including NKp80 and 2B4, as well asthe triggering NKG2D molecule 20, further work needs to address whether some of these receptors may also be affected in HIV-1 infection. In addition it will be important to determine if and to what extent the impairment of NCR could be responsible for the impaired control of HIV-1 replication in these patients. Finally, further work will help to elucidate whether the defective NCR expression in HIV-1 infection is associated with viremia or with disease progression and whether it can be corrected by ART.
4 Materials and methods
Participants included 10 patients with HIV-1 infection, >250 CD4+ cells/mm3 and no previous history of ART, who were followed up at the Department of Internal Medicine, University of Genoa, and who gave their informed consent to the study according to the declaration of Helsinki. Human experimentation guidelines of the authors' Institution were followed in theconduct of clinical research. Peripheral blood samples were obtained during routine clinical visits and blood sampling for clinical purposes. Control samples were obtained from a group of 9 healthy, adult, volunteer blood donors from the local blood bank and from healthy laboratory workers.
4.2 Cell cultures and cytotoxicity assay
PBMC were obtained by density gradient centrifugation (Ficoll-Hypaque). PBMC were depleted of plastic-adherent cells and incubated with anti-CD3 (JT3A), anti-CD14 (LeuM3), and anti-CD19 mAb for 30 min at 4°C, followed by beads coated with goat anti-mouse antibody (Dynal, Oslo, Norway) for 30 min at 4°C 20, 32. After purification, NK cells were either immediately analyzed or were used to derive in-vitro-activated NK cell populations or NK cell clones as previously described 22–24. Briefly, after immunomagnetic depletion, CD3–CD19–CD14– cells were cultured on irradiated PBMC (5000 rad) used as feeder cells in the presence of rIL-2 at 100 U/ml (Proleukin, ChironCorp., Emeryville, CA, USA), and 1.5 ng/ml PHA (Gibco Ltd., Paisley, GB) to obtain activated polyclonal NK cell populations, or cultured under limiting dilution conditions in 96-well plates to obtain NK cell clones 22–24. The culture medium used was RPMI 1640 supplemented with 10% FCS, L-glutamine (2 mM/l) and 1% antibiotic mixture (penicillin 5 mg/ml, streptomycin 5 mg/ml, neomycin 10 mg/ml stock solution). In order to provide uniform analysis for all patients and controls, NK populations were analyzed between day 10 and 20 after purification/activation.
A series of FcγR+ and FcγR– target cell lines were used in the various cytolytic assays: P815 murine mastocytoma (FcγR+); K562 human erythroleukemia (FcγR–); M14 human melanoma (FcγR–); EBV-transformed 221 (FcγR–); and B lymphoma Daudi (FcγR–).
NK-cell-enriched populations and NK cell clones were tested for cytolytic activity in a 4-h 51Cr-release assay as previously described 22–24, either in the absence or in the presence of various mAb. The concentration of the various mAb was 10.0 μg/ml for the masking experiments and 0.5 μg/ml for the redirected killing experiments.
The following panel of anti-human mAb was used 22–24: JT3a (IgG2a) anti-CD3; HP2.6 (IgG2a) anti-CD4; B9.4 (IgG2B) anti-CD8; KD1 (IgG2a) and c127 (IgG1) anti-CD16; c218 (IgG1) and GPR165 (IgG2a) anti-CD56; 7A6 (IgG1) anti-NKp30; Z231 (IgG1) anti-NKp44; and BAB281 (IgG1) anti-NKp46.
4.4 Flow cytometry analysis
In all cases indirect staining was performed. Briefly, cells were incubated with the appropriate mAb followed by FITC- or PE-conjugated isotype-specific goat anti-mouse secondary reagent (Southern biotechnology Associated, Birmingham, AL, USA). Negative control reagents were isotype-matched mouse mAb directed against irrelevant surface molecules. Samples were analyzed by one- or two-color cytofluorometric analysis (FACScan, Beckton Dickinson, Mountainview, CA, USA). Data were analyzed using Cell Quest program. Cells were gated by forward- and side-scatter parameters based on low scatter and small size. For single-color analyses, results are expressed as logarithm of green fluorescence intensity (arbitrary units) vs. number of events. For two-color analyses, data areexpressed as logarithm of green fluorescence intensity (arbitrary units) vs. logarithm of red fluorescence intensity (arbitrary units). For each analysis 10,000 events were counted.
Sources of financial support were M.U.R.S.T. (to A. D. M.), and Istituto Superiore di Sanità, Programma Nazionale AIDS (to L. M.). None of the authors has a commercial or other association that might pose a conflict of interest.