|Home | About | Journals | Submit | Contact Us | Français|
Natural killer (NK) cell degranulation in response to virus-infected cells is triggered by interactions between invariant NK cell surface receptors and their ligands on target cells. Although HIV-1 Vpr induces expression of ligands for NK cell activation receptor, NKG2D, on infected cells, this is not sufficient to promote lytic granule release. We show that triggering the NK cell coactivation receptor NK-T-and -B cell antigen (NTB-A) alongside NKG2D promotes NK cell degranulation. Normally, NK cell surface NTB-A binds to NTB-A on CD4+ T cells. However, HIV-1 Vpu downmodulates NTB-A on infected T cells. Vpu associates with NTB-A through its trans-membrane region without promoting NTB-A degradation. Cells infected with HIV-1 Vpu mutant elicited at least 50% more NK cells to degranulate than wild-type virus. Moreover, NK cells have a higher capacity to lyse HIV-infected cells with a mutant Vpu. Thus, Vpu downmodulation of NTB-A protects the infected cell from lysis by NK cells.
NK cells respond to virus-infected cells without requiring prior exposure to viral antigens. Hence, they play a major role in managing virus-infected cells during the early stages of viral infection before the emergence of the virus-specific adaptive immune responses. The outcome of an NK cell response to a target cell is determined by intracellular signaling cascades initiated by interactions between germline encoded and invariant receptors on the NK cell and their ligands on the target cell (Lanier, 2005, 2008; Moretta et al., 2001; Moretta and Moretta, 2004). These receptor-ligand interactions are divided into three major categories: inhibiting, activating, and coactivating.
One set of inhibitory receptors on NK cells (iNKR) interact with the major histocompatibility complex class I molecules (MHC-I) (Ciccone et al., 1992; Dohring et al., 1996; Moretta et al., 1997; Natarajan et al., 2002). The three major MHC-I-specific families of iNKRs are the killer immunoglobulin (Ig)-like receptors (KIR), whose members have varying numbers of Ig domains and ligand specificities (recognizing HLA-A, -B or -C as ligands); the lectin-like heterodimer of NKG2A and CD94 (recognizing HLA-E); and interleukin-like transcript type 2 (which recognizes multiple MHC class I molecules). Furthermore, NK cells also express various inhibitory receptors that bind non-MHC-I ligands (reviewed in Borrego et al., 2002; Lanier, 2008).
While iNKR-ligand interactions provide a fail-safe mechanism by which NK cells avoid killing normal “self” cells, the lack of or impaired expression of iNKR ligands is insufficient to trigger an NK cell cytolytic response. The engagement of NK cell activation receptors (aNKRs) by ligands on infected cells is required to elicit NK cells to lyse their target cells. Triggering a variety of surface aNKR can induce NK cell activation (reviewed in Bryceson et al., 2006a; Lanier, 2008). The natural cytotoxicity receptors (NCRs) containing Ig-like domains; NKp30, NKp44, and NKp46 (Pende et al., 1999; Pessino et al., 1998); and the C-type lectin, NKG2D (Bauer et al., 1999), are the major aNKRs. Additionally KIRs with short intracellular tails (KIR2DS1/2 and -3, KIR3DS1) are also aNKRs (Dohring et al., 1996; Moretta et al., 1995). Another family of aNKRs is a lectin-like heterodimer consisting of CD94 associated with NKG2C, which binds HLA-E (Braud et al., 1998).
Although the activating NK receptors are necessary for NK-mediated lysis of target cells, they are insufficient to induce degranulation (Bryceson et al., 2006a) and require the concomitant triggering of coactivating receptors (caNKR) (Bryceson et al., 2006a; Moretta et al., 2001). The simultaneous engagement of both aNKR and caNKR by their ligands on target cells prompts resting NK cells to release their lytic granules (Bryceson et al., 2006b, 2009).
Earlier studies indicated that NK cells are ineffective at killing autologous, primary HIV-1-infected T cells (Bonaparte and Barker, 2003; Ruscetti et al., 1986; Zheng and Zucker-Franklin, 1992). However, examination of the infected cell surface revealed that virus infection leads to a decrease in surface expression of HLA-A and -B (Bonaparte and Barker, 2004) through the action of Nef (Cohen et al., 1999) and an increase in ligands for NKG2D (Ward et al., 2007) through the action of Vpr protein (Richard et al., 2010; Ward et al., 2009). The downregulation of inhibitory ligands combined with the upregulation of activating ligands should lead one to predict that HIV-1-infected cells could serve as ideal targets for NK cell-mediated destruction. However, the ability of NK cells from even healthy uninfected individuals to destroy HIV-1-infected cells has been consistently characterized to be weak at best (Bonaparte and Barker, 2003; Fogli et al., 2008; Ruscetti et al., 1986; Tomescu et al., 2007; Ward et al., 2007; Zheng and Zucker-Franklin, 1992). Therefore, additional factors must be involved in regulating NK activity in the presence of HIV-1 infection. One such factor that regulates NK cell cytolytic activity is NTB-A. NTB-A is found on all blood-derived NK, T, and B cells and is a member of the signaling lymphocytic activation molecule (SLAM) family of receptors (Bottino et al., 2001). It contains immunoreceptor tyrosine-based switch motifs that are docking sites for the SH2 domain of SLAM-associated proteins and the related Ewing's sarcoma-associated transcripts (EAT)-2 (Bottino et al., 2001; Falco et al., 2004; Flaig et al., 2004). NTB-A is a type I transmembrane protein of the Ig superfamily with an N-terminal IgV domain and a proximal truncated C-terminal IgC2 domain (Bottino et al., 2001). NTB-A functions as a homotypic ligand-caNKR pair (Falco et al., 2004; Flaig et al., 2004) and has been observed to induce NK cell cytotoxicity (Bottino et al., 2001; Flaig et al., 2004). Moreover, we have demonstrated that NTB-A facilitates NK cell-mediated killing of HIV-1-infected cells (Ward et al., 2007). However, HIV-1 downmodulates NTB-A on the infected cell surface (Fogli et al., 2008; Ward et al., 2007). The consequence that decreased NTB-A expression on HIV-1-infected cells will have on the ability of NK cells to lyse the infected target remains to be determined.
It is unclear if the surface downregulation of NTB-A by HIV-1 coincides with changes in the total NTB-A levels of the infected cell. We addressed this question by comparing the levels of NTB-A, after staining the surface of HIV-infected cells with anti-NTB-A, with the levels of NTB-A, after permeabilizing the infected cells and then staining them with anti-NTB-A. As a positive control, we evaluated CD4 expression, on and within infected cells. We observed that cell surface NTB-A expression was reduced during infection, but total NTB-A levels remained unchanged (Figures 1A). In contrast, HIV-1 infection of primary CD4 T cells led to a decrease in both cell surface and intracellular CD4 expression (Figures 1A). This indicated that, unlike CD4, downregulation of NTB-A on the surface by HIV-1 does not coincide with changes in the steady-state levels of NTB-A in the infected cells.
Since the HIV-1 accessory proteins, Nef and Vpu, induce cell surface downregulation of a number of host cell proteins (Garcia and Miller, 1991; Kerkau et al., 1997; Moll et al., 2010; Neil et al., 2008; Schwartz et al., 1996; Van Damme et al., 2008; Willey et al., 1992), we asked whether either or both of the viral gene products may play a role in NTB-A downregulation. We found that HIV-1ΔNef downregulated NTB-A from the infected cell surface to the same extent (MFI = 461 compared with MFI = 2088 for uninfected cells) (Figure 1B) as wild-type (WT) HIV-1 (MFI = 485 compared with MFI = 2032 for uninfected cells) (Figure 1B). In contrast, HIV-1ΔVpu was unable to downregulate NTB-A (MFI = 1998 compared with MFI = 2055 for uninfected cells) (Figure 1B).
To determine whether Vpu is sufficient to downregulate cell surface NTB-A, we determined cell surface NTB-A expression on both Vpu-transduced and untransduced cells. As a positive control, we also evaluated the cell surface expression of CD4. As expected, cell surface CD4 expression was downmodulated on Vpu-transduced cells (MFI = 1427) relative to CD4 on untransduced cells (MFI = 6094) (Figure 1C). We found NTB-A to be downmodulated on Vpu-expressing cells (MFI = 1485), relative to NTB-A on untransduced cells (MFI = 2962) (Figure 1C). These experiments indicate that Vpu is both necessary and sufficient to downregulate NTB-A.
Vpu enhances virus production by downmodulating CD4 and bone marrow stromal antigen 2 (BST-2) expression. Vpu serves as an adaptor protein linking CD4 and BST-2 to β-TrCP, a human F box protein that functions as a substrate recognition receptor for the multisubunit E3 ubiquitin ligase (Douglas et al., 2010; Margottin et al., 1998). To determine whether β-TrCP association with Vpu is required for NTB-A downmodulation, we asked if Vpu with both serine residues mutated to asparagines (Vpu [S52, 56N]) would be able to downmodulate NTB-A. As expected, Vpu (S52, 56N), failed to downregulate CD4 as compared to WT Vpu (Figure 2A). In contrast, we found that Vpu (S52, 56N) was able to downregulate NTB-A to comparable levels as WT Vpu (Figure 2A). Our studies therefore illustrate that Vpu downregulates cell surface NTB-A by a mechanism that is distinct from that of CD4 and BST-2 downregulation.
To rule out any role of the putative ubiquitination residues in the cytoplasmic tail of NTB-A on Vpu's ability to downmodulate NTB-A, we generated a mutant of NTB-A lacking a cytoplasmic tail (NTBA Δ250–331). Following transfection with Vpu, we found that removal of the cytoplasmic tail did not significantly alter the capacity of Vpu to downmodulate NTB-A (Figure 2B).
Because proteasomal degradation has been implicated in CD4 or BST-2 downmodulation (Douglas et al., 2010; Margottin et al., 1998), we asked what effect proteasomal inhibition would have on surface expression of NTB-A. As expected (Schubert et al., 1998), treatment with 100 nM epoxomicin or 10 μM MG-132 abrogated the capacity of Vpu to downregulate CD4 surface expression when compared to that in cells treated with vehicle only (Figure 2C). In contrast, treatment of NTB-A expressing cells with proteasome inhibitors failed to relieve NTB-A downregulation by Vpu relative to vehicle-treated cells (Figure 2C). Thus, Vpu-mediated NTB-A downregulation is independent of proteasome activity.
In the study shown in Figure 1, we demonstrated that Vpu did not change the steady-state levels of NTB-A even though surface levels of NTB-A were decreased. For this reason, we wanted to determine if Vpu was increasing internalization of NTB-A from the cell surface. As Figure 3A shows, NTB-A was internalized to the same extent over a course of 45 min regardless if Vpu was present or not.
Acidification of endocytic vesicles is a necessary step in various processes following endocytosis, including receptor recycling (Tycko et al., 1983). Bafilomycin A1 inhibited the capacity of Vpu to downregulate BST-2 from the cell surface and enhanced intracellular BST-2 levels by either blocking pH-dependent trafficking to late endosomes or inhibiting lysosomal degradation (Mitchell et al., 2009). We treated HeLa-NTB-A cells with 100 nM bafilomycin A1 and found no effect on Vpu's ability to modulate NTB-A (Figure 3B). The MFI of NTB-A on Vpu-transfected cells was 5,309 compared with an MFI of 22,578 for NTB-A on untransfected cells. The MFI for NTB-A on Vpu-transfected cells treated with 100 nM bafilomycin was 5,121 compared with 23,143 for bafilomycin A1-treated untransfected cells. This same concentration of bafilomycin A1 prevented Nef-dependent downmodulation of CD4 (see Figure S1 available online).
Although NTB-A downregulation by Vpu is mechanistically distinct from that of CD4 and BST-2, it remains to be determined if Vpu binds to NTB-A as it does to CD4 or BST-2. To evaluate this, we first asked if NTB-A binds directly to Vpu. As shown in Figure 4A, we observed a 37 kD band corresponding to Vpu-GFP only in the lane corresponding to NTB-A expressing cells transfected with Vpu; all other lanes did not contain this band. Taken together, the above findings indicated that Vpu binds to NTB-A.
Since Vpu binds to NTB-A, we proceeded to determine the region of Vpu that was critical for its interaction with NTB-A. Vpu-mediated downregulation of BST-2 points to Vpu's transmembrane region as being important for BST-2 downmodulation (Van Damme et al., 2008). To determine if Vpu interacted with NTB-A through the transmembrane region, we utilized a Vpu mutant that encodes a scrambled amino acid sequence of the transmembrane domain (URD) (Schubert et al., 1996). We immunoprecepitated NTB-A from lysates of NTB-A transfected with Vpu-GFP or URD-GFP followed by immunoblotting with anti-Vpu. As controls, we probed the input lysates with anti-Vpu and anti-β-actin and probed the IP with anti-NTB-A. Figure 4B illustrates that NTB-A does not interact with the transmembrane scrambled Vpu mutant URD but does bind to the WT Vpu. Vpu used in our study is part of a GFP fusion protein. Hence, it may be possible that detection of Vpu when immnoprecipitating NTB-A could be due to the GFP binding to NTB-A. However, this does not appear to be the case, since URD-GFP did not bind to NTB-A even though the Vpu Ab used in our study detected URD-GFP in the lysates. Thus, Vpu interacts with NTB-A through the transmembrane portion of Vpu.
Next we infected primary CD4 T cells with HIV-1NL4/3 and HIV-1NL4/3-URD. As shown in Figure 4C, we found that URD-infected cells expressed NTB-A on primary T cells to a similar extent as that on uninfected cells (MFI of 8960 for HIV-infected cells and 7313 for uninfected cells). However, URD downregulated CD4 indistinguishably from WT Vpu. On WT-Vpu-infected cells, both NTB-A and CD4 are downregulated. Thus the transmembrane portion of Vpu is critical for downmodulation of NTB-A but not of CD4.
Since NTB-A was downregulated by Vpu, we next wanted to determine the consequences modulating this ligand on the infected cells surface had on the NK cells’ cytolytic function. To begin, we asked whether NTB-A engagement, in the context of NKG2D engagement, could trigger NK cells to degranulate.
Ligands for the activation receptor NKG2D, such as ULBP-1 and -2, are able to trigger NK cell lysis of HIV-infected cells (Ward et al., 2007). However, it has also been demonstrated that triggering NKG2D alone on primary NK cells is insufficient to induce optimal degranulation (Bryceson et al., 2006b). Whether NTB-A is required for triggering NK cell degranulation when NKG2D is also triggered remains to be determined. To address this issue, we utilized a redirected degranulation assay in which only these two specific receptors can be triggered using agonist antibodies as ligand mimics. We found that very few (1.6%) fresh blood derived-NK cells spontaneously express CD107a (Figure 5Ai). When NKG2D was triggered alone, only 1.3% of NK cells degranulated (Figure 5Aii). Similarly, engagement of NTB-A alone caused only 1.8% of NK cells to express CD107a (Figure 5Aiii). However, when NKG2D and NTB-A were triggered together, the percent of degranulating NK cells increased to 19.3% (Figure 5Aiv). These results demonstrate that triggering NTB-A together with NKG2D induces the release of granules from NK cells even though triggering either receptor alone has a minimal effect.
Our previous studies demonstrated that HIV-1-infected cells have a modest sensitivity to lysis by NK cells (Bonaparte and Barker, 2004; Fogli et al., 2008; Ward et al., 2007). Given that NTB-A is required as a costimulatory signal to trigger NK cells to degranulate (Figure 5A), and that NTB-A is downmodulated by Vpu (Figures 1B and 1C), we wished to determine whether the failure of NK cells to lyse infected cells effectively occurs at the level of degranulation per se or at the prior step of activation. For this purpose we evaluated the proportion of NK cells expressing on their surface the activation marker, CD69 (Ziegler et al., 1994), and the degranulation marker, CD107a (Peters et al., 1991), after a 4 hr exposure to HIV-infected cells (Figure 5B). As a positive control target cell, we utilized the NK-sensitive cell line, K562 cells, and as a negative control, uninfected CD4 T cells.
Very few NK cells expressed CD69 or CD107a spontaneously (1.2% and 1.0%, respectively) when freshly isolated from the blood (Figures 5Bi and 5Bv). Following exposure to uninfected CD4 T cells, 7.6% of NK cells expressed CD69 and 6.3% of NK cells expressed CD107a (Figures 5Bii and 5Bvi). When exposed to K562 cells, 34.8% of NK cells expressed CD69 and 35.2% expressed CD107a (Figures 5Biii and 5Bvii), indicating vigorous activation and degranulation. When exposed to HIV-1-infected cells, 43.2% of NK cells expressed CD69 (Figure 5Biv), indicating potent activation; however, only 11.6% of the cells expressed the degranulation marker CD107a (Figure 5Bviii). Therefore, we concluded that the failure of NK cells to lyse HIV-1-infected target cells occurs at or near the stage of degranulation, but not at an earlier activation step.
The results in Figures 5Bi–5Bviii were confirmed on a per-cell basis by comparing NK cell activation and degranulation simultaneously. We found that 66% of the CD69+ NK cells degranulated in response to K562 cells (Figure 5Bix). In contrast, only 19% of CD69+ NK cells degranulated in response to HIV-1-infected T cells (Figure 5Bx). Our observations demonstrate that NK cells become activated in response to HIV-1-infected T cells but fail to degranulate to the same extent.
Since Vpu downmodulates NTB-A (Figure 1), and NTB-A is required for NK cells to degranulate when NKG2D is triggered (Figure 5), we predicted that introducing the URD mutation in HIV-1 Vpu, which fails to downregulate NTB-A (Figure 4), would enable NK cells to degranulate in response to HIV-1-infected cells. As controls, we evaluated the percentages of NK cells degranulating when cultured alone, with uninfected CD4 T cells, with K562 cells, and with CD4 T cells infected with HIV-1 WT. As shown in Figure 6A, regardless of the iNKRs expressed, NK cells degranulated ~3-fold higher against URD-infected cells (63.34% of total NK cells evaluated were CD107a+) than against WT infected cells (22.6% of total NK cells evaluated were CD107a+). Of the NK cells exposed to K562 cells, 71.85% of the cells degranulated. Less than 7% of NK cells degranulated when exposed to uninfected CD4+ T cells.
Since HLA-C and -E remain on the infected cell surface and may suppress NK cell response to the infected cells (Bonaparte and Barker, 2004; Cohen et al., 1999), we analyzed the capability of NK cells lacking iNKRs to HLA-C and -E to degranulate. We have also performed analysis with NKG2A+ (CD159a, part of the HLA-E-specific receptor) as well as KIR2DL1 and 2/3+ (CD158a and -b, two HLA-C-specific receptors) CD56+ CD3− cells. This is important since NK cells are variegated with regards to expression of these receptors (Bonaparte and Barker, 2004). As seen in Figure 6A, only 1.3% percent of fresh NK cells lacking HLA-C and -E iNKRs degranulate, and similarly, when exposed to uninfected CD4 T cells, 4.89% of NK cells degranulated. When exposed to K562 cells, 34.30% of NK cells degranulated. However, when exposed to HIV-1 WT 15.21% of NK cells degranulated. Over twice as many NK cells degranulated (39.89%) when exposed to HIV-1 URD-infected cells. Thus, Vpu's ability to downmodulate NTB-A limits the capacity of NK cells to degranulate by at least 2-fold. Although the HLA-C and -E on infected cells decreased NK cell degranulation by 2-fold, Vpu prevented NK cells expressing iNKRs to HLA-C and -E from lysing infected cells by 2-fold (Figure 6A).
Studies by Drs. Bryceson and Long (Bryceson et al., 2005, 2006b, 2009) indicated that “resting” NK cells required engagement of two activation receptors in order to degranulate. Therefore we asked whether stimulation of NK cells with IL-2 prior to exposure to target cells would still require that two activation receptors be engaged to get maximal degranulation. Stimulation with IL-2 leads NK cells to express the activation marker CD69 (see Figure S2) regardless of target cells used to stimulate the NK cells. When exposed to 200 U/ml of IL-2, overall, NK cells degranulated to a greater extent than NK cells not exposed to cytokines (Figure 6B), even though the pattern of degranulation remained similar. Thus even “activated” NK cells require engagement of two activation receptors in order to achieve maximal degranulation.
Next we wanted to determine whether the URD-infected cells triggered NK cell degranulation to a greater extent because of its inability to downmodulate NTB-A. To accomplish this, we blocked the NTB-A on the NK cells with anti-NTB-A antibodies prior to adding URD-infected T cells. As seen in Figure 6C, NK cells had a decreased ability to degranulate when exposed to URD cells after they were treated with anti-NTB-A blocking Ab prior to adding the target cells (Figure 6C).
Since degranulation of NK cells equates to its capacity to lyse target cells (Alter et al., 2004), we wanted to determine if Vpu's ability to downmodulate NTB-A decreased the responding NK cells’ ability to lyse the infected cells. In Figure 6D, NK cells had a 2-fold greater capacity to lyse HIV-infected cells with URD compared to target cells infected with WT virus (p < 0.05 for both E:T ratios; Student's t test). The level of lysis seen against URD-infected cells was similar to the capability of NK cells to lyse the NK-sensitive K562 cell line (Figure 6D) (p > 0.1 for both E:T ratios; Student's t test). Thus, these studies indicate that Vpu's ability to prevent NTB-A downmodulation protects the infected cell from lysis by NK cells.
We had previously demonstrated the importance of NKG2D and NTB-A, separately, in the destruction of HIV-infected cells by NK cells (Ward et al., 2007). However, the connection between these receptors in the NK cell cytolytic response has never been made. Moreover, it has never been reported that NTB-A is capable of triggering degranulation in combination with NKG2D engagement. This is important to know, since the proper activating receptor pairings are essential for NK cells to degranulate (Bryceson et al., 2006b), and here we begin to evaluate which activation/coactivation receptor pairs may be functional against HIV-1-infected cells. In this study, we demonstrate that simultaneous engagement of NKG2D and NTB-A is essential to elicit NK cell degranulation. This is not only true of “resting” NK cells but of activated ones as well. In contrast, triggering NKG2D and the activation coreceptor DNAM-1 simultaneously was insufficient to induce degranulation even after exposure of NK cells to cytokines (data not shown). This is important since HIV-infected cells express CD155, the ligand for DNAM-1 (Z.B.D. and E.B., unpublished data).
By downregulating NTB-A, Vpu may protect HIV-1-infected cells from NK killing. We have noted that the URD-infected cells were lysed by NK cells 2-fold higher than lysis of HIV-1-infected cells expressing WT Vpu. Furthermore, the level of lysis of URD-infected cells approached the level of NK cell lysis of K562 cells. This was important to note, since K562 cells are sensitive to NK cell-mediated lysis. It is noteworthy that HIV-2 does not encode a Vpu protein (Bour and Strebel, 1996) even though it does encode Vpr and Nef, and furthermore, we have found that HIV-2-infected cells do not downregulate NTB-A and are killed more efficiently by NK cells than HIV-1-infected cells (J.P.W. and E.B., unpublished data).
One of the earliest observations regarding HIV-1 evasion of NK cell responses made by Drs. David Baltimore's and Jack Strominger's laboratories over 10 years ago was that HIV-1 Nef downmodulated HLA-A and -B but left -C and -E on the infected cell surface (Cohen et al., 1999). The latter two HLA-class I molecules were believed to prevent NK cell killing of HIV-infected cells (Cohen et al., 1999). However, we have found that NK cells are variegated with regards to expression of HLA-C and -E iNKRs, and a significant number of NK cells could not be controlled by HLA-C or -E (Bonaparte and Barker, 2004). In fact, we found that NK cells lacking HLA-C and -E iNKRs lysed HIV-infected cells, though modestly, relative to the ability of the NK cells to lyse the NK-sensitive cell line, K562 cells (Bonaparte and Barker, 2004). In the present study, we demonstrate that HLA-C and -E control NK cells ability to degranulate. Although they reduce the ability of NK cells to release lytic granules by 2-fold, they still do not prevent degranulation entirely. Moreover, our current study shows that Vpu's impact on the ability of NK cells to degranulate appears to be far greater than HLA-C and -E.
Although our studies centered specifically on Vpu derived from HIV-1NL4/3, we have observed NTB-A downmodulation on cells infected with other strains such as HIV-1SF162 and HIV-1SF128A (Ward et al., 2007). Moreover, we have even observed downmodulation of NTB-A on infected cells derived from HIV-1-infected patients (Fogli et al., 2008).
NTB-A downmodulation by Vpu reveals a similarity with Vpu-mediated downmodulation of BST-2 (Neil et al., 2008; Van Damme et al., 2008), in that the transmembrane region of Vpu is important for not only BST-2 downmodulation (Van Damme et al., 2008) but also for NTB-A downmodulation. However, aside from this similarity in Vpu-mediated downmodulation of BST-2 and NTB-A, there are a couple of notable differences. First, downmodulation of NTB-A does not require the phosphorylation of serines in the 52nd and 56th positions of Vpu while it appears to be required for BST-2/tethein downmodulation (Mitchell et al., 2009). Second, Vpu-mediated downmodulation of BST-2 is sensitive to proteasome inhibitors and inhibitors of endosomal/lysosomal acidification (bafilomycin A1) (Mitchell et al., 2009), but Vpu-mediated downmodulation of NTB-A is not.
From our studies a model is generated showing the steps leading to NK cells’ response to HIV-infected cells and how the HIV-infected cell evades lysis (see Figure 7). Following HIV entry, reverse transcription, and integration into the host chromosome, the virus begins to express one of the early gene products, Nef. This viral protein downmodulates HLA-A and -B. While the decrease of the two MHC class I molecules lowers the capacity of HLA-A and -B restricted HIV-1 peptide antigen specific cytotoxic T-lymphocytes from lysing the infected cells (Collins et al., 1998), they also reduce the threshold for inhibiting KIR3DL1-bearing NK cells. Another HIV-1 gene product, Vpr, enhances HIV virion production through its ability to activate ATR. However, the activation of ATR also led to the expression of NKG2D ligands (i.e., ULBP-1 and -2) on the infected cells. Thus, the combination of Vpr and Nef activities should increase the susceptibility of infected cells to lysis by NK cells. Although HLA-C and -E remain on the infected cells surface, they only affect a subpopulation of NK cells. However, the downmodulation of NTB-A on the infected cell decreases NK cells’ ability to degranulate even though NKG2D ligands are present and HLAA and -B are downmodulated. Without engaging NTB-A on NK cells, triggering NKG2D is insufficient for NK cells to degranulate. Thus, despite the activation of NK cells by HIV-infected cells, the lysis of the target cell does not occur because Vpu prevents the release of lytic granules by NK cells by downmodulating the ligand to NTB-A responsible for the second signal needed for degranulation (Bryceson et al., 2006b).
All primary cells used in this study were isolated from peripheral blood obtained from healthy HIV-1 uninfected donors after informed written consent was acquired in accordance with the Declaration of Helsinki and the policies of the Institutional Review Board at Rush University Medical Center, Chicago, IL, USA. CD4 T cells were isolated and stimulated in vitro as described (Ward et al., 2009). The 293 FT cell line was used in the generation of viruses and vectors and maintained according to the manufacturer's (Invitrogen) specifications. K562 cells, P815 cells, and HeLa and Jurkat E6-1 cell lines were obtained from ATCC and maintained according to distributors’ recommendations. HeLa-CD4 cells were obtained from Dr. Richard Axel (Columbia University, NY, USA) through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH (HeLa-T4+) and maintained in medium containing 500 μg/ml G418 (Invitrogen). HeLa-NTBA and HeLa-NTBA Δ250–331 cells were generated by transduction of HeLa cells with retroviral particles encoding a bicistronic mRNA for NTB-A or NTB-A Δ250–331 and puromycin resistance. Stably transduced cells were selected in 300 ng/ml puromycin (Invitrogen) for 60 days and subjected to three rounds of cell sorting (FACS Aria; BDIS) for cell surface NTB-A expression. Sorted cells were maintained in medium containing 200 ng/ml puromycin.
Pacific Blue-conjugated anti-CD3, -CD20, -CD14, and -CD4; FITC-conjugated anti-CD69; and PE-conjugated anti-CD107a and staining controls were obtained from BD PharMingen. APC-conjugated anti-NTBA was obtained from R&D Systems. AF700-conjugated anti-CD56 and PECy7 anti-CD16 were obtained from Biolegend. FITC- and PE-conjugated anti-HIV-1 p24 was obtained from Beckman Coulter. Purified anti-NKG2D (BD PharMingen) and anti-NTB-A (R&D Systems) were used in the redirected degranulation assay. HIV-1NL4-3 Vpu antiserum was obtained from Dr. Frank Maldarelli and Dr. Klaus Strebel (NCI, NIH) through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH. HRP-conjugated mouse anti-rabbit, rabbit anti-goat, and donkey anti-mouse antibodies were obtained from Jackson Immunoresearch.
Simultaneous detection of surface antigens and intracellular HIV-1 p24 antigen (Ag) was done as previously described (Ward et al., 2009). For GFP-expressing cells or for HeLa-NTBA cell sorting, only surface antigen staining was performed. For determination of total NTB-A or CD4 cellular content, simultaneous cell surface and intracellular labeling was performed with the same antibody. Samples were acquired using FACS LSRII. Analysis of flow cytometric data was performed with FloJo Software (TreeStar, Inc).
To eliminate possible differences in replication kinetics due to the presence or absence of Vpu and Nef in the studies described in Figures 1, ,5,5, and and6,6, we used a defective HIV-1 construct, DHIV-3, that has a deletion in the env gene. We then provided the VSV-G glycoprotein in trans to form pseudotyped virions as described (Andersen et al., 2006; Bosque and Planelles, 2009). The HIV-1NL4/3 reagent was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH; HIV-1NL4/3-URD was a direct gift of Dr. Klaus Strebel, NIH. Viral particles were produced and titered as previously described (Ward et al., 2009).
Human codon-optimized Vpu sequence of HIV-1NL4/3 (pcDNA-hVpu) was obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH; pcDNA-hVpu was from Dr. Stephan Bour and Dr. Klaus Strebel (NIH). The human codon-optimized Vpu sequence was PCR amplified with appropriate restriction enzyme extensions and cloned into pAcGFP-N1 (Clontech) to generate Vpu with a C-terminal GFP fusion protein. The vector pAcGFP-Vpu S52,56N was constructed by site-directed mutagenesis (Stratagene) using the primer pair shown in Figure S3A. The Vpu sequence from pNL4.3-Urd was used as the template to generate the Urd-GFP fusion protein. NTB-A was cloned by RT-PCR of mRNA obtained from Jurkat cells as described (Ward et al., 2009). NTB-A primer pair sequence is shown in Figure S3B. NTB-A D250–331 was constructed by site directed mutagenesis (Stratagene) using the primer pair shown in Figure S3C. NTB-A encoding retroviral constructs were generated by subcloning NTB-A into pQCXIP (Clontech) bicistronic retroviral vector. Vpu-GFP and GFP encoding retroviral constructs were generated by subcloning either Vpu-GFP or GFP (from pACGFP-N1) into pQCXIP (Clontech). Retroviral particles were generated according to the manufacturer's instructions (Clontech). The Vpu and NTB-A coding regions of all plasmids were sequenced (University of Chicago Sequencing Facility) to verify accuracy of the coding sequence.
Infections and transductions of primary cells were done as described in Ward et al. (2009).
HeLa-NTBA or HeLa-CD4 cells were seeded at a density of 1.25 × 105 cells/well in a 6-well tissue culture plate (BD) and incubated overnight at 37°C, 5% CO2. Cells were transfected with 1.5 μg of Vpu expression plasmids using HeLa Monster Trans-IT reagent (Mirus Bio) and incubated for 24–36 hr. For inhibition of proteasome activity, proteasome inhibitors epoxomicin and MG-132 (Sigma-Alrdrich) were reconstituted in DMSO (Mediatech) prior to its addition to medium 24 hr after transfection. For inhibition of vesicle acidification, bafilomycin A1 (Sigma-Aldrich) was reconstituted in DMSO (Mediatech) prior to its addition to medium 24 hr after transfection. Immunoblots with mutant Vpu-GFP constructs were done in the absence of epoxomicin. All HeLa cells were harvested with 0.05% trypsin (Mediatech) prior to analysis by flow cytometry or immunoblots.
HeLa-NTB-A cells were transiently transfected with plasmids encoding either GFP or Vpu-GFP for 24 hr. Cells were harvested with 0.05% trypsin and labeled with monoclonal anti-human NTB-A (Biolegend). Cells were washed with PBS, aliquoted into sterile tubes containing DMEM complete media, and incubated at 37°C with 5% CO2. At indicated times, cells were fixed with 1% paraformaldehyde, washed with PBS, and labeled with APC conjugated goat anti-mouse IgG2 (Jackson Immunoresearch).
For coimmunoprecipitation of Vpu and NTB-A, transfected HeLa and HeLa-HA-NTB-A cells were harvested 48 hr posttransfection and lysed overnight with 0.5% NP40 with 150 mM NaCl, 50 mM Tris-HCl, PMSF (Sigma) in methanol and a protease inhibitor cocktail (Roche). Protein concentration was determined by performing Bradford assay (Pierce). Cell lysates (200 μg) were precleared using recombinant protein G agarose (Invitrogen) for 30 min at 4°C with rotation, and then incubated with anti-HA (Santa Cruz) at 4°C overnight with rotation. Precipitation was accomplished with protein G agarose for 30 min at 4°C. Immunoprecipitates and 20 μg of whole-cell lysates were analyzed for the presence of Vpu on a 10% SDS-PAGE gel using a Mini Protean 3 Cell (Bio-Rad) for 2 hr at 100 V and 500 mA in 1× SDS (Fisher Scientific) running buffer. The gel was transferred to a methanol-soaked PVDF membrane in Towbin transfer buffer for 2 hr at 250 Volts and 200 mA. Membranes were blocked overnight at 4°C in 5% skim milk (Fisher Scientific) in Tris-buffered saline with 0.1% Tween 20 (Fisher Scientific) with agitation. Membranes were probed with appropriate primary antibodies for 2 hr in 5% milk solution followed by appropriate secondary antibodies for 1.5 hr in 5% milk solution. Antibodies were detected by enzymatic chemiluminescence using Pierce ECL western blotting substrate (Thermo) for 5 min on each membrane. Membranes were then exposed to autoradiography film (Denville Scientific) and developed in an autoprocessor (Konica).
HIV-infected cells (106) were incubated with fresh PBMC(106) from the same donor for 4 hr at 37°C, 5%CO2. After the coincubation, cells were treated with 1 mg/ml of human γ-globulin to block Fc receptors and then stained with fluorochrome-conjugated anti-CD3, CD14, CD20, CD56, CD107a, and CD69 . Following surface staining, dead cells were selected using the LIVE/DEAD Fixable Dead Cell Stain Kit (Invitrogen). Viable CD3−CD20−CD14− and CD56+ cells (2 × 104) were evaluated for the percentage of CD107a and/or CD69.
To compare the ability of NK cells to degranulate when exposed to T cells infected with VSV-G pseudotyped DHIV-3 expressing WT Vpu or URD, purified NK cells left untreated or stimulated with 200 U/mL rhIL-2 were used and stained as described above, except 2 × 105 infected cells were mixed with 105 NK cells. In addition, the cells were stained with anti-CD158a, CD158b, and CD159a fluorochrome-conjugated antibodies (BDIS). All three Abs had the same fluorochrome. Twenty-thousand viable CD3−CD20−CD14− and CD56+ cells that possessed CD158a, CD158b, and/or CD159a or lacked all three were evaluated for the percentage of CD107a surface expression. NTB-A blocking was achieved by treating the NK cells with 5 μg/mL purified anti-NTB-A (clone ON56) for 20 min at 4°C. All unbound antibody was removed by washing once with RPMI complete. NK cells were then treated with 100 μg/mL goat anti-mouse F(ab′)2 (Jackson Immuno) and incubated at 4°C for 20 min prior to incubation with targets.
Fresh PBMC (106) was exposed to 5 μg/ml purified anti-NKG2D and/or anti-NTB-A . The PBMC were then mixed with P815 cells (105) and incubated for 4hr at 37°C, 5%CO2. After 4 hr the cells were stained as described above in “Degranulation and Activation Assay.” Ten-thousand viable CD3−CD20−CD14− and CD56+ cells were evaluated for the percentage of CD107a.
The authors would like to thank Dr. Alessandro Moretta, University of Genoa, Genoa, Italy, for providing us with the ON56 clone of anti-NTB-A antibody. This work was supported by grants from NIH/NIAID: AI65361 and AI81684 for E.B. and AI49057 and AI81684 for V.P.
Supplemental Information includes three figures and can be found with this article online at doi:10.1016/j.chom.2010.10.008.