Polyfunctional natural killer cells with a low activation profile in response to Toll-like receptor 3 activation in HIV-1-exposed seronegative subjects

Natural killer (NK) cells are the main mediator of the cytotoxic response in innate immunity and may be involved in resistance to HIV-1 infection in exposed seronegative (ESN) individuals. Toll-like receptor (TLR) signalling is crucial for NK cell activation. Here, we investigated the polyfunctional NK cell response to TLR3 activation in serodiscordant couples. ESN subjects showed increased IFN-γ and CD107a expression in both NK subsets, CD56bright and CD56dim cells, in response to stimulation with a TLR3 agonist, while expression was impaired in the HIV-1-infected partners. TLR3-induced expression of IFN-γ, TNF and CD107a by polyfunctional CD56bright NK cells was more pronounced in ESN individuals than that in healthy controls. Activated NK cells, as determined by CD38 expression, were increased only in the HIV-1-infected partners, with reduced IFN-γ and CD107a expression. Moreover, CD38+ NK cells of the HIV-1-infected partners were associated with increased expression of inhibitory molecules, such as NKG2A, PD-1 and Tim-3, while NK cells from ESN subjects showed decreased NKG2A expression. Altogether, these findings indicate that NK cells of ESN individuals were highly responsive to TLR3 activation and had a polyfunctional NK cell phenotype, while the impaired TLR3 response in HIV-1-infected partners was associated with an inhibitory/exhaustion NK cell phenotype.

to produce IFN-γ after cytokine stimulation, proliferate in vivo during viral infection, and kill target cells upon engagement of activating receptors.
The NK cells of ESN drug users showed high cytolytic potential and produced IFN-γ, TNF-α, and β-chemokines when in contact with the K562 cell line 11 . In addition, NK cells from the ESN subjects produced high levels of IFN-γ in response to activation with phorbol myristate acetate and ionomycin 12 . The absence of human leukocyte antigen (HLA) binding to inhibitory killer-cell immunoglobulin-like receptors (KIRs), including KIR2DL2, KIR2DL3 and KIR3DL1, leads to a reduced activation threshold of NK cells from ESN individuals and has been associated with resistance to HIV-1 infection 13 .
In contrast, chronic HIV-1 infection alters the population distribution and functional capacity of NK cells. The chronic activation receptor CD38 is an ectoenzyme expressed on CD8 + T cells and is associated with progression to AIDS in chronically HIV-1-infected patients, even in those treated with antiretroviral therapy (ART) 14,15 . Moreover, NK cell activation through CD38 is increased in HIV-1-infected subjects progressing to AIDS, but not in elite and viremic controllers, and is associated with viremia and disease progression markers in both HIV-1 and HIV-2 infections 16,17 . Whether the NK CD38 + cells in HIV-1 infection are associated with altered expression of inhibitory/exhaustion molecules, such as NKG2A, PD-1, and Tim-3, is unclear. Few cytotoxic NK cells and high NKG2A expression have been observed in patients with late-stage HIV infection 18 . The expression of Tim-3, a type I transmembrane protein, has been implicated both in activation and inhibition of immune responses 19,20 , in the induction of apoptosis of Tim-3-bearing cells through interactions with galectin-9 21 and suppression of cell-mediated cytotoxicity 22 .
NK cells possess receptors allowing them to sense and respond to viral and bacterial patterns, including Toll-like receptors (TLRs) 23 . Viral nucleic acids, such as viral DNA, dsRNA, and ssRNA, can activate nucleic acid-sensing TLRs, including TLR3, TLR7, TLR8 and TLR9, to prevent viral invasion of cells via induction of type I IFN and production of antiviral factors 24 . TLR stimulation in ESN individuals induced a robust release of immunologic factors that can influence adaptive antiviral immune responses 25 . A TLR3 agonist, polyinosinic-polycytidylic acid [Poly(I:C)], a synthetic mimetic of viral RNA, significantly augmented NK cell-mediated cytotoxicity in healthy individuals 26 . A common TLR3 polymorphism in ESN individuals suggested the potential use of TLR3 triggering in HIV-1 immunotherapy 27 .
In this study, we evaluated the TLR3-induced activation of NK cells and assessed CD62L and CD38 expression in NK cell subsets to verify the polyfunctional response of NK cells in ESN individuals and their HIV-1-infected partners. Moreover, we examined whether CD38 + CD62L + NK cells were less responsive to TLR3 activation and could be associated with inhibitory/exhaustion markers, such as NKG2A, PD-1 and Tim-3. Elucidation of the underlying mechanisms of naturally manifested resistance against HIV-1 infection in ESN individuals has significant clinical implications for preventive and therapeutic strategies.

Materials and Methods
Study population. HIV-1-serodiscordant couples were enrolled from an outpatient clinic at the Emílio Ribas Infectious Diseases Institute in São Paulo, from the Ambulatory Service (ADEE/3002) of the Department of Secondary Immunodeficiency Clinic of the Clinical Hospital, University of São Paulo Medical School (HC/ FMUSP) and from the Reference Center and Treatment in STD-AIDS in São Paulo, Brazil. ESN individuals (n = 20), HIV-1-infected partners of ESN individuals (n = 20), and healthy donors not infected with HIV-1 (n = 20) were enrolled in this study. A mean relationship duration of 13 years with a single partner was reported by homosexual (n = 6) and heterosexual (n = 14) couples. Couples reported participating in vaginal, anal and oral sex, including 5 episodes of unprotected sexual intercourse at a frequency of 3-4 times per month 28,29 . Inclusion criteria consisted of an age over 18 years, reported participation in unprotected sex and having a single partner for over 1 year. The ESN cohort was seronegative at the studied time point. All HIV-1-infected individuals were receiving ART, and detectable viral loads (VLs) were found in 4/20 individuals. Exclusion criteria consisted of the use of immunosuppressant or immune-modifying drugs and pregnancy. This study was approved by the São Paulo University Institutional Use Committee (CAPPesq n° 0683/09), and informed consent was obtained from all subjects. All experimental protocols within this study were performed in accordance with the Ethics Committee of this institution. Statistical analysis. All cytokine measurements were background subtracted, taking into account the frequency of cells producing cytokines in the absence of antigenic stimulation. Kruskal-Wallis tests with Dunn's post-test were used to compare variables between HIV-1-infected individuals, ESN subjects and healthy controls. P ≤ 0.05 was considered statistically significant.

Results
Cytokine secretion by CD56 + CD62L + NK cells in serodiscordant couples. NK cells are an important component of innate immunity, and CD62L is differentially expressed during various stages of human NK cell maturation, with high CD62L expression on CD56 bright NK cells and heterogeneous CD62L expression on CD56 dim NK cells 7 . Therefore, we evaluated cytokine secretion by CD62L + NK cells following activation with TLR3 or TLR7/8 agonists in serodiscordant couples. The cohort of serodiscordant couples reported an average of 13 years in a single relationship, in which the majority of HIV-infected partners were successfully treated with ART, except for 4 individuals with 179; 50,930; 51; and 17,511 copies of RNA/mL (Supplementary Table 1). These serodiscordant couples had persistent sexual exposure to HIV-1, albeit with low viral loads. Figure 1B shows that in unstimulated conditions, a higher frequency of CD62L expression was detected in CD56 bright than that in CD56 dim NK cells in all groups. After activation of PBMCs with TLR3 agonists (Poly I:C) or TLR7/8 agonists (CL097), we observed similar expression of CD62L in NK cell subtypes in the ESN and HC groups, whereas CD62L expression was markedly decreased in HIV-1-infected partners ( Fig. 1C and D).
In evaluating cytokine secretion by NK cells, increased expression of IFN-γ and CD107a induced by TLR3 activation was detected in CD56 bright cells of the ESN group compared with that of the HC and HIV-1-infected groups, regardless of CD62L expression ( Fig. 2A and B). In addition, IFN-γ secretion induced by the TLR3 agonist in CD56 dim CD62L − cells from ESN individuals was increased compared with that of the other groups (Fig. 2C). A similar frequency of CD56 dim CD107a + NK cells was observed in the ESN and HC groups (Fig. 2D). Of note, HIV-1-infected partners showed decreased CD107a expression and IFN-γ production in both NK cell subtypes (Fig. 2). CD62L expression was associated with the functional aspects of NK cell subsets, where CD56 bright cells secreted IFN-γ and CD56 dim cells expressed CD107a.
CD56 bright CD62L + NK cells activated by the TLR3 agonist showed increased IFN-γ production and decreased CD107a expression in both the ESN and HC groups, while the opposite effect was observed in CD56 dim CD62L + NK cells (Fig. 2). In contrast, the CD62L + and CD62L − NK cell response was barely detectable in the HIV-1-infected group (Fig. 2).
ESN individuals exhibit higher polyfunctionality of CD56 bright CD62L + NK cells. Next, we evaluated the polyfunctional ability of CD56 bright NK cells to secrete TNF, IFN-γ and/or CD107a in response to TLR3 activation in the presence or absence of CD62L expression. The CD56 bright NK cell population was selected for analysis due to its higher expression of CD62L than that of CD56 dim cells. The gating strategy for polyfunctional CD56 bright NK cells is shown in Supplementary Figure 1. Figure 3 shows an increased frequency of CD56 bright CD62L + NK cells secreting CD107a, TNF and IFN-γ simultaneously in the ESN group compared with that of the HC and HIV-1-infected groups (Fig. 3A). These findings demonstrate that TLR3 activation enhanced the polyfunctional CD56 bright NK cell response in ESN individuals, principally with respect to CD62L + cells, in contrast to the reduced NK cell function observed in HIV-1-infected partners (Fig. 3B).
NK activation through CD38 of HIV-1-infected individuals is associated with viremia and disease progression markers 16,17 , although CD38 expression in NK cells of ESN subjects has not been previously assessed. Higher CD38 expression after TLR3 stimulation was detected in CD56 bright and CD56 dim NK cells of the HIV-1-infected group compared with that of HC and ESN individuals (Fig. 4A), as well as in NK cells co-expressing CD38 and CD62L (Fig. 4B). Although there was an increased frequency of CD62L + CD38 + NK cells in HIV-1-infected subjects, these cells showed a decreased functional ability compared with that of cells from HCs and ESN partners (Fig. 4C).

Activated NK CD56 bright/dim cells expressing inhibitory/exhaustion markers in HIV-1 infected
subjects. Since activated NK cells (CD38 + ) from HIV-1 individuals showed decreased IFN-γ/CD107a production induced by the TLR3 agonist, we next evaluated the expression of exhaustion/inhibition molecules, such as NKG2A, PD-1, and Tim-3, on NK cells. Figure 5B-E shows high frequencies of Tim-3, NKG2A and PD-1 in NK CD56 bright/dim CD62L + CD38 + cells in peripheral blood of HIV-1-infected individuals compared to those of the HC and ESN groups. Similar results were obtained by measuring the median fluorescence intensity (MFI) of these molecules (Supplementary Figure 2). Again, an increased frequency of CD56 bright/dim CD62L + CD38 + NK cells expressing Tim-3 was induced by TLR3 stimulation only in the HIV-1 infected partners (Fig. 5C).
Although there was a small percentage of CD56 + CD38 + NK cells expressing NKG2A in the uninfected groups, it was decreased in ESN individuals compared to those of the HC controls ( Fig. 5D and Supplementary  Figure 2).
These findings indicated that activated CD38 + NK cells express inhibitory/exhaustion molecules in HIV-1-infected partners, which may, in part, be responsible for the impaired TLR3 response.

Discussion
In this study, we found that NK cells from ESN individuals were more responsive to TLR3 signalling with respect to the production of IFN-γ and expression of CD107a than those from HIV-1-infected partners or healthy subjects. In response to TLR3 or TLR7/8 agonists, the percentage of CD62L + NK cells was similar between the ESN and HC groups, but was markedly reduced in the HIV-1-infected group. Moreover, an increased polyfunctional response induced by TLR3 activation was detected in CD56 bright CD62L + NK cells from ESN individuals. Along with the decreased percentage of NK cells in HIV-1-infected subjects and impaired IFN-γ and CD107a expression induced by TLR3 stimulation in CD38 + NK cells, we found high expression of inhibitory/exhaustion receptors, as Tim-3, PD-1 and NKG2A. In contrast, low activation and a strong response to TLR activation was detected in NK cells of ESN partners. The natural resistance to HIV-1 was mediated, in part, by NK cells with high TLR3 responsiveness, polyfunctionality and low inhibitory marker expression.
In the ESN and HC groups, CD62L expression in CD56 bright NK cells was associated with increased IFN-γ secretion, and CD107a production was increased in CD56 dim NK cells that also expressed CD62L. This profile was not observed in HIV-1-infected partners, in whom CD62L expression was markedly reduced even upon TLR3 or TLR7/8 activation. All the HIV-1-infected partners were receiving ART, and the majority (16 subjects) had an undetectable viral load, which was not sufficient to increase CD62L movement to the cell membrane. Interestingly, NK cells expressing receptors and co-receptors for HIV-1, including CD4, CCR5 and/or CXCR4,  were persistently infected with HIV-1 in ART-treated subjects 30 . Moreover, HIV-1 accessory proteins, including Nef and Vpu, retain CD62L in perinuclear compartments to prevent its export to the plasma membrane 30,31 .
The high TLR3 responsiveness of NK cells in ESN subjects was notable even with respect to the HC group. TLR3 induces the production of cytokines and cytotoxic molecules through a mechanism dependent on p38 MAPK activation in NK cells 32 . Although we did not study the signalling pathways downstream of TLR3, our data suggest that p38 MAPK may be activated more strongly in the ESN group. NK cells from ESN individuals produce IFN-γ and lyse K562 cells in response to activation with PMA and ionomycin 11,12 . Another study showed that a TLR3 polymorphism, in which the leucine residue at position 421 is substituted for phenylalanine and confers receptor activation, was predominant in a cohort of drug users in Spain 33 . These ESN drug users also showed an increased response to TLR3 stimulation (Poly I:C), indicating that a common TLR3 allele confers immunologically mediated protection from HIV-1 and suggesting the potential use of TLR3 triggering in HIV-1 immunotherapy 33 .
CD56 bright NK cells are high producers of cytokines and constitute the subtype that predominantly expresses CD62L. Therefore, we evaluated the polyfunctional CD56 bright NK cell response induced by a TLR3 agonist in the presence and absence of CD62L expression. We verified that ESN individuals possess CD56 bright NK cells with an increased ability to secrete multiple cytokines, mainly cells that are CD62L + . The high polyfunctionality of NK cells expressing CD62L in ESN individuals could be related to resistance to HIV-1 infection. Characteristics of HIV-specific T cell responses, such as specificity, breadth, magnitude, polyfunctionality, and phenotype, may underlie the control of viral replication in HIV-1 infected individuals 34 . In addition, a high frequency of polyfunctional CD8 + NK cells is associated with slower disease progression in chronic HIV-1 infection 35 .
We previously demonstrated that PBMC activation with PMA and ionomycin increased the co-expression of IFN-γ and CD107a in CD56 bright and CD56 dim NK cells in ESN individuals, whereas expression was not affected in the HIV-1 group compared with expression in the HCs 28 . These data support the hypothesis that impairment of NK signalling in HIV-1-infected partners is related to TLR activation instead of the PKC pathway.
Usually, the CD38 ectoenzyme in T cells is an indicator of chronic cell activation during HIV-1 infection 36,37 . CD38 and CD16 are functionally dependent and physically associated, and therefore, CD38 ligation promotes target cell killing by human NK cells 38 . Additionally, disease markers, including the CD4 + T cell percentage, viral load, and CD4 + T cell activation, have been associated with CD38 expression levels in both iNKT and NK cells 17 . Elevated levels of iNKT cell and NK cell activation are associated with viremia and disease progression markers in both HIV-1 and HIV-2 infections 17 . Moreover, in HIV-1-infected patients who achieve viral suppression following ART treatment, NK cell activation persists 39 . We showed that CD38 expression in NK cell subsets in the HIV-1-infected group was associated with an impaired ability to produce IFN-γ and CD107a. The co-expression of CD62L and CD38 in NK cells was also pronounced in HIV-1-infected individuals, suggesting that even immature NK cells can display an activated profile during infection. Whether the CD38 + NK cells are associated with immune activation as are CD4 + and CD8 + CD38 + T cells in the chronically HIV-1-infected subjects 40,41 has yet to be determined. Curiously, low percentages of CD38 + NK cells were found in HC as well as ESN subjects, which showed a low activation profile.
The CD38 expression in NK cells subsets was associated with Tim-3, NKG2A and PD-1 expression and impaired functional ability in the chronically infected HIV-1 partners. NKG2A expression in the activated NK cells was decreased in the uninfected groups, predominantly in the ESN individuals; it was evaluated in NK cells expressing CD38, a population that is poorly represented in ESN subjects. In contrast, NK cells of a Vietnamese population of highly exposed uninfected drug users had a similar distribution of NKG2A and NKG2C expression among CD107a + NK cells compared to that of HIV-1-infected individuals 42 . The difference in NKG2A expression in NK cells could be due to the intensity of HIV-1 exposure or route among the ESN cohorts. Our cohort is considered chronic, with low HIV-1 exposure, since the majority of HIV-1 infected partners were receiving ART treatment. Moreover, the decreased NKG2A expression in a small percentage of NK cells of ESN individuals may be correlated with the high TLR3 responsiveness since NKG2A is associated with suppression of cytotoxicity in HIV-1 patients with advanced clinical status 18 . Other exhaustion molecules, such as the PD-1 and Tim-3, were barely detected in NK CD38 + cells from ESN subjects and healthy controls. ESN and HIV-controllers have lower expression of CD69, LAG-3, PD-1, and Tim-3 in NK cells and CD4 + T cells compared to that of HIV progressors 43 . Our results showed for the first time that the low activation profile was due to a low frequency of NK CD38 + cells in ESN individuals and inhibitory/exhaustion molecule expression.
Together, our results show that ESN individuals possess NK cells with pronounced responsiveness to TLR3 activation and the ability to develop a polyfunctional response, as determined by cytokine production and degranulation. Our data support the hypothesis that the immune activation in chronically HIV-infected individuals affects the NK cells, leading for exhausted/inhibited profile. Protective immunity against HIV-1 infection in ESN individuals is associated with a low activation NK profile. Understanding the mechanisms underlying the activation of NK cells with TLRs agonists in naturally controlled HIV-1 infection may contribute to developing strategies that harness NK cells to treat HIV infection.