Influence of Glycosylation Inhibition on the Binding of KIR3DL1 to HLA-B*57:01

Viral infections can affect the glycosylation pattern of glycoproteins involved in antiviral immunity. Given the importance of protein glycosylation for immune function, we investigated the effect that modulation of the highly conserved HLA class I N-glycan has on KIR:HLA interactions and NK cell function. We focused on HLA-B*57:01 and its interaction with KIR3DL1, which has been shown to play a critical role in determining the progression of a number of human diseases, including human immunodeficiency virus-1 infection. 721.221 cells stably expressing HLA-B*57:01 were treated with a panel of glycosylation enzyme inhibitors, and HLA class I expression and KIR3DL1 binding was quantified. In addition, the functional outcomes of HLA-B*57:01 N-glycan disruption/modulation on KIR3DL1ζ+ Jurkat reporter cells and primary human KIR3DL1+ NK cells was assessed. Different glycosylation enzyme inhibitors had varying effects on HLA-B*57:01 expression and KIR3DL1-Fc binding. The most remarkable effect was that of tunicamycin, an inhibitor of the first step of N-glycosylation, which resulted in significantly reduced KIR3DL1-Fc binding despite sustained expression of HLA-B*57:01 on 721.221 cells. This effect was paralleled by decreased activation of KIR3DL1ζ+ Jurkat reporter cells, as well as increased degranulation of primary human KIR3DL1+ NK cell clones when encountering HLA-B*57:01-expressing 721.221 cells that were pre-treated with tunicamycin. Overall, these results demonstrate that N-glycosylation of HLA class I is important for KIR:HLA binding and has an impact on NK cell function.


Introduction
Natural killer (NK) cells are part of the innate immune system, and serve as a first line of defense against intracellular pathogens and malignantly transformed cells. NK cells can lyse target cells via directed secretion of perforin and granzyme, antibody-dependent-cell-mediated-cytotoxicity and Fas/FasL-mediated cytotoxicity [1], but can also activate and recruit other immune cells via secretion of pro-inflammatory cytokines and chemokines. The importance of NK cell function is highlighted in the setting of impaired NK cell function or NK cell deficiencies, which are associated with increased susceptibility to infections with viruses and intracellular bacteria [1][2][3][4]. NK cell activity is regulated via a number of activating and inhibitory receptors, including killer cell immunoglobulin-like receptors (KIRs) [1,5], which recognize human leukocyte antigen (HLA) proteins. Binding of KIRs to HLA class I molecules is determined by the respective HLA class I and KIR allotypes, as well as the sequence of the HLA class I-presented peptides [6,7]. While binding of inhibitory KIRs to HLA class I molecules results in NK cell inhibition, engagement of activating KIRs or loss of inhibitory KIR binding can lead to NK cell activation [8,9]. The KIR-HLA interaction has been demonstrated to have an important impact on viral infections, including human immunodeficiency virus-1 (HIV-1). In HIV-1 positive individuals, KIR3DL1 + NK cells expanded preferentially in the presence of HLA-Bw4-80I, the ligand class for KIR3DL1 [10,11].
Protein glycosylation is a post-translational modification occurring in the endoplasmatic reticulum (ER) and Golgi apparatus. In multiple enzymatic steps, a complex oligosaccharide (i.e. glycan) is synthesized in the ER, transferred to a specific receptor sequence on its target protein, and subsequently cropped and remodeled in the ER and Golgi [12]. In the case of asparagine N-glycosylation, the target amino acid sequence on the respective proteins is N-X-S/T, where the glycan is bound to the asparagine. It is estimated that over 7,000 distinct glycan structures can be generated in mammals [13], suggesting a wide range of functional properties, which can be modulated by differential glycosylation. On HLA-class I, the glycan at position N86, which is located close to the Bw4 motive critical for binding of KIR3DL1 to its ligands, is highly conserved and present on practically all HLA-class I allotypes [14,15]. Here we use a panel of glycosylation inhibitors to examine the effects of glycosylation modifications on KIR3DL1:HLA-B Ã 57:01 interactions and their potential effect on NK cell function, with a more in-depth characterization of the effect tunicamycin (TUN)-an antibiotic that can completely block N-glycosylation in eukaryotic cells by preventing the linkage of the glycan to the asparagine [16,17]-and castanospermine (CSP)-a α-glucosidase I and II inhibitor able to inhibit the glycan trimming in the ER necessary for the assembly of a fully functional glycan [17,18].

Jurkat reporter cell assay
Jurkat cells (clone E6.1; ATCC) were cultivated in R10 medium. 24 h before performing experiments, 10% fetal bovine serum was added to the culture to reduce background activation. Jurkat cells stably expressing KIR3DL1z, a chimeric receptor composed of the extracellular and transmembrane domains of KIR3DL1 linked to the cytoplasmic tail of CD3z, which we denote as 'KIR3DL1z', were produced via lentiviral transduction/transfection. Briefly, gene constructs were designed and ordered via GeneArt (Life Technologies), and cloned into a lentiviral transfer vector containing an SFFV promotor and IRES-driven puromycin resistance. HEK293T cells (ATCC) were transfected with a VSV-G envelope vector (pHEF-VSVG, NIH AIDS Reagent Program), HIV-1 gag-pol packaging vector (psPAX2, NIH AIDS Reagent Program), and the transfer vector containing KIR3DL1z. Lentivirus-containing supernatants were harvested 72 h after transfection and used to transduce Jurkat cells, which were subsequently selected in R10 containing 1 μg/mL puromycin and sorted for gene expression by flow cytometry. In these KIR3DL1z + Jurkat cells, ligand engagement by KIR3DL1z results in an activating signal that triggers CD69 expression, making them a suitable reporter cell system. Prior to coincubation with target cells, Jurkat cells were cultured for 24 h in R10 medium supplemented with an additional 10% fetal bovine serum. 221 cells expressing HLA-B Ã 08:01, HLA-B Ã 57:01, or no HLA class I underwent glycosylation inhibition and were washed twice. Target cells were washed to remove glycosylation inhibitors and then co-incubated with either KIR3DL1z + or untransduced Jurkat cells at a reporter-to-target cell ratio of 1:1 (1 × 10 5 each) for 2.5 h at 37°C/5%CO 2 . Cells were then washed, stained with anti-CD3-PerCP-Cy5.5, anti-CD69-BV421, and anti-KIR3DL1-APC (all from BD) for 30 min at 4°C, fixed with 4% paraformaldehyde/PBS, and analyzed on an LSR-Fortessa (BD). KIR3DL1z-transduced Jurkat cells were gated for KIR3DL1 expression, classifying them either as KIR3DL1z low , KIR3DL1z dim or KIR3DL1z bright .

NK cell cloning and degranulation assay
Primary human NK cells isolated from peripheral blood mononuclear cells (PBMCs) of healthy donors were cloned by limiting dilution in the presence of feeders and maintained in NK cell cloning medium consisting of R10 supplemented with 5% human serum (Sigma-Aldrich), 1X MEM-NEAA (Gibco), 1X sodium pyruvate (Gibco), 100ug/mL Kanamycin, 200 U/mL IL-2 (AIDS Reagent Program, NIH) using a protocol adapted from a previously reported method [22]. Briefly, NK cells were isolated from peripheral blood mononuclear cells (PBMCs) from a KIR3DL1 + HLA-Bw4 + donor via magnetic negative selection (NK cell isolation kit from Miltenyi), added to a mix of irradiated feeders consisting of freshly isolated allogeneic (PBMCs) combined with log-phase-growth RPMI 8866 cells (Sigma-Aldrich) at a 10:1 ratio in cloning medium containing 1 μg/mL phytohaemagglutinin (PHA; Fisher) and mixed thoroughly before plating at 100 μL/well (0.5 NK cell/well) in 96-well plates and incubated for 14 days at 37°C/5% CO 2 . After 14 days, wells that had outgrowth of cells were transferred to 48-well plates and maintained in NK cell cloning medium with frequent media exchange (approximately every 3 days). Cells were phenotyped by flow cytometry to assess NK cell marker (e.g. CD56 and CD16) and KIR3DL1 expression. Degranulation assays were performed by co-incubating 5 × 10 4 NK cells with 2.5 × 10 5 target cells (effector-to-target cell ratio of 1:5) in 200 μL of R10 containing 3 μL of anti-CD107a-PE-Cy7 antibody (BioLegend) for 2 h at 37°C/5% CO 2 , and subsequently staining with anti-CD56-BV605 and anti-CD16-BV785 antibodies (both from BioLegend) for 30 min at 4°C, and then fixed with 4% paraformaldehyde/PBS. Flow cytometric analysis was performed on a BD LSRFortessa.

Ethics
PBMCs from healthy donors were obtained and frozen after obtaining written informed consent following procedures approved by the Partners Human Research Committee (ethics committee) and the Institutional Review Board of Massachusetts General Hospital. HLA-I types of selected human samples were determined prior to this study by high-resolution HLA-I typing performed at the HLA-typing laboratory of the National Cancer Institute, National Institutes of Health. KIR types were determined prior to this study by Sanger sequencing in the laboratory of Mary Carrington.

Statistics
Flow data was analyzed in FlowJo v10 (FlowJo LLC, Ashland, Oregon). Statistical analyses were performed using GraphPad Prism 6.0c (GraphPad Software Inc., La Jolla, California, USA). Normality tests (Kolmogorov-Smirnov test with Dallal-Wilkinson-Lilliefor P value) were performed on all data sets, statistical significance was calculated via one-way ANOVA, and multiple comparisons were corrected using Tukeys's test if not stated otherwise. All graphs show mean ± SD if not stated otherwise in the figure legends. Significance levels were defined as Ã <0.05, ÃÃ <0.01, ÃÃÃ <0.001 and ÃÃÃÃ <0.0001.

Effects of glycosylation enzyme inhibitors on HLA expression and KIR-Fc binding
In order to examine the effect of different commercially available glycosylation inhibitors on KIR:HLA interactions, we examined their effect on HLA-B Ã 57:01 expression on HLA-B Ã 57:01transduced 221 cells and on KIR3DL1-Fc binding to those cells. Inhibitors that targeted Golgiresident α-mannosidase enzymes (KIF, SWA) increased HLA-B Ã 57:01 expression, which resulted in minimally increased (MFI of 32220 vs. MFI of 36900; p = 0.9456) KIR3DL1 binding (Fig 1A and 1B). The Golgi-resident sialyl transferase inhibitor (STI) tested, however, did not result in any significant change in HLA-B Ã 57:01 expression or KIR3DL1 binding. Inhibitors that targeted ER-resident α-glucosidase II (CSP, DNM), which is an enzyme necessary for the last removal of glucose from N-glycans to allow anterograde transport to the Golgi, dramatically reduced HLA-B Ã 57:01 expression and abrogated KIR3DL1 binding. The ER-resident α-glucosidase I-specific inhibitor (AUS) minimally increased HLA-B Ã 57:01 expression or KIR3DL1 binding (Fig 1A). However, in the case of TUN, we observed a major decrease in KIR3DL1-Fc binding while HLA-B Ã 57:01 surface expression was increased slightly (Fig 1B). Thus, after preliminary testing of the effects that several different glycosylation inhibitors had on HLA class I expression and KIR binding, we decided to focus on TUN, the only inhibitor tested that did not decrease HLA class I surface expression but dramatically reduced KIR3DL1-Fc binding, and that had the unique function of completely inhibiting that addition of N-glycans.
We subsequently performed a titration assay in order to identify the minimal amount of glycosylation inhibitors required to observe a strong functional effect on the binding of  Co-incubation of 221-HLA-B Ã 57:01 cells with TUN (+T) for 24 h significantly increased HLA-B Ã 57:01 expression as measured by anti-HLA-Bw4 antibody staining (1.5-fold HLA-Bw4 MFI increase, p = 0.02; 2.0-fold W6/32 MFI increase, p < 0.0001) (Fig 2A, S1 Fig). In contrast, co-incubation with CSP (+C) significantly decreased HLA expression (2.1-fold HLA-Bw4 MFI decrease, p = 0.02; 1.7-fold W6/32 MFI decrease, p < 0.0001). Since both the staining with the anti-HLA-Bw4 antibody as well as the staining with the anti-pan-HLA class I antibody, which only recognizes HLA class I proteins that are properly folded and bound to β 2 -microglobulin, were similarly affected my the inhibition of glycosylation, we believe that the HLA-B Ã 57:01 proteins expressed at the cell surface were indeed properly folded HLA-B Ã 57:01 complexes and not free heavy chains.
Sustained expression of HLA-B Ã 57:01 on 221-B Ã 57:01 cells following TUN treatment enabled assessment of the consequences of HLA class I de-glycosylation on KIR3DL1 binding to HLA-B Ã 57:01. Despite enhancement of overall HLA-B Ã 57:01 expression, TUN abrogated KIR3DL1-Fc binding. Cells treated with TUN bound KIR3DL1-Fc significantly less than untreated cells (4.7-fold KIR-Fc MFI decrease, p < 0.0001). 221-HLA-B Ã 57:01 cells treated with CSP also exhibited reduced KIR3DL1-Fc binding (6.9-fold KIR-Fc MFI decrease, p < 0.0001) (Fig 2B), which was expected given the reduced surface expression of HLA-B Ã 57:01 on CSP-treated cells. Taken together, these data demonstrate that the presence of N-glycosylation is critical for KIR3DL1 binding to HLA-B Ã 57:01. HLA class I N-glycan is necessary for functional signaling through KIR3DL1 in KIR3DL1ζ + Jurkat cells A Jurkat cell model stably expressing a chimeric receptor of the extracellular and transmembrane domain of KIR3DL1 fused to the cytoplasmic tail of CD3z, referred to as 'KIR3DL1z', was employed to assess the functional consequences of glycosylation inhibition. KIR3DL1z + ligand engagement by HLA-B Ã 57:01 results in an activating signal that triggers CD69 expression. Jurkat cells were gated by size, granularity, CD3 expression and KIR3DL1z expression. Cells were classified as either KIR3DL1z bright , KIR3DL1z dim , KIR3DL1z low (+/~/-) (Fig 3A). KIR3DL1z -Jurkat cells were compared to untransduced Jurkat cells (data not shown) and no difference in activation after coincubation with HLA class I-expressing target cells was observed. Thus, KIR3DL1z -Jurkats were used as a negative control (Fig 3B and 3C).

Discussion
N-glycosylation is a critical step in post-translational protein modification with functional importance that is not yet fully understood. In the case of all HLA class I allotypes, a single, highly conserved N-glycosylation site is located at position N86 [14,15], in close proximity to the highly polymorphic motif spanning from amino acids 77 to 83, which constitutes the Bw4 motif in HLA-Bw4 alleles and is critical for KIR3DL1 binding to these [25]. We therefore hypothesized that HLA class I N-glycosylation can modulate binding between HLA-Bw4 proteins and KIR3DL1. To test this hypothesis, we employed a panel of glycosylation enzyme inhibitors to modify or eliminate N-glycan structures in HLA-B Ã 57:01-expressing cells and assessed HLA-B Ã 57:01 expression and binding to KIR3DL1. We demonstrate that presence of the N-glycan on HLA-B Ã 57:01 is critical for binding to KIR3DL1, and that inhibition of N-glycosylation in target cells can have functional consequences on KIR3DL1 + cell lines and primary KIR3DL1 + NK cells.
Inhibition of different glycosylation enzymes led to different effects in terms of HLA-B Ã 57:01 expression and KIR3DL1 binding. The most striking effect was observed for TUN-induced inhibition of N-glycosylation, which increased cell surface expression of HLA-B Ã 57:01 while abrogating KIR3DL1 binding. This was in contrast to all other inhibitors, which either increased (KIF, SWA), decreased (CSP, DNM), or did not significantly alter (AUS, STI) HLA-B Ã 57:01 expression in a manner that correlated with increased, decreased, or unchanged KIR3DL1 binding, respectively. TUN-induced increase in HLA-B Ã 57:01 expression was confirmed using two different antibodies-anti-HLA-Bw4 (clone: Bw4) and anti-pan-HLA class I (clone: W6/32)-demonstrating that the Bw4 motif, which is situated next to the N-glycosylation site, is still recognized, and that the HLA class I molecule is expressed as a complex bound to β 2 -microglobulin. We further demonstrated that loss of KIR3DL1-Fc binding to TUN-treated HLA-B Ã 57:01 + target cells matched functional readouts using KIR3DL1z + Jurkat reporter cells and primary human KIR3DL1 + NK cells. KIR3DL1z + Jurkat cells were stimulated by co-incubation with 221-HLA:B Ã 57:01 cells, yet this stimulation was almost completely abrogated by treating target cells with TUN prior to co-incubation. In line with this, primary KIR3DL1 + NK cells were potently inhibited by 221-HLA-B Ã 57:01 cells, but were 'de-repressed' when 221-HLA-B Ã 57:01 cells were pre-treated with TUN. These data indicates that N-glycosylation of HLA-B Ã 57:01 plays a critical and functional role in KIR3DL1 binding and can modulate NK cell function, a finding that may extend to numerous other KIR:HLA interactions that have not yet been studied in the context of N-glycolysation.
Reports of the importance of N-glycosylation for KIR:HLA binding are scarce, save for one study suggesting that the HLA class I N-glycan does not influence KIR binding [26]. In contrast, several studies reported the consequences of HLA class I N-glycosylation modifications on T-cell receptor (TCR) binding [23,[27][28][29]. Those reports all concluded that TCR binding to HLA class I (and MHC class I in mice) is independent of HLA class I glycosylation, given the relatively long distance between the TCR binding site and the HLA class I N86 glycan. KIR: HLA crystal structures, however, reveal a very close proximity between the KIR binding site on HLA class I and position N86, which in these structures is aglycosylated due to the production of these proteins in E. coli (Fig 5). This suggests that the HLA class I N86 glycan may be contacting KIR and influencing binding avidity. Of note, the one study that concluded that HLA class I glycosylation was not necessary for KIR binding was based on a generally assumed interaction between HLA-B Ã 08:01 and an undiscovered inhibitory KIR, which later was found to not exist and only be the effects of the interaction between the inhibitory receptor NKG2A and HLA-E, which was not discovered at the time of the study [30]. Thus, to the best of our knowledge, our study is the first to implicate the HLA class I N-glycan as being critical for KIR: HLA binding, which may serve as another means of modulating the interaction between NK cell receptors and target cell ligands.
It has been demonstrated that the glycosylation pattern of several immune receptor-ligand pairs can be influenced in the setting of infection. In HIV-1 infection, a global shift in the glycosylation pattern of IgG has been observed, with HIV-1-specific antibodies displaying the most distinct glycosylation patterns {Ackerman, 2013 #357} [31,32]. This shift in IgG glycosylation patterns can alter Fc receptor binding and is associated with improved antiviral activity and control of HIV-1, but has also been described for other viral and bacterial infections [31,33]. Indeed, the N-glycan structure found on IgG is very similar to the HLA class I N86 glycan, and can be modified similarly by the addition of fucose, bisecting N-acetyl glucosamine, galactose, or sialic acid [34][35][36]. Furthermore, HIV-1 infection has been shown to alter glycosylation in host cells, and it is conceivable that HIV-1 might affect HLA class I glycosylation, either as a host response mechanism or a direct immunevasive tactic depending on whether HLA class I binding to NK cell receptors is enhanced or diminished by the altered glycosylation pattern. It has been suggested that other viruses have taken advantage of this level of regulation, as in the case of hepatitis C virus, which downregulates HLA class I expression in order to escape immune pressure, a process that is hypothesized to be due to altered glycosylation [37][38][39][40].
While much about the role of modified glycosylation patterns remains to be elucidated, our data demonstrates the importance of glycosylation in KIR:HLA binding and that removal of the glycan has a functional effect on the activation of NK cells. The extent to which pathogens and the immune system can exploit this mechanism to their advantage or whether this mechanism can be harnessed for therapeutic purposes means remains to be determined.