Histochem Cell Biol (2007) 127:313–326 DOI 10.1007/s00418-006-0240-z
O RI G I NAL PAPE R
Clustered carbohydrates as a target for natural killer cells: a model system Elena I. Kovalenko · Elena Abakushina · William Telford · Veena Kapoor · Elena Korchagina · Sergei Khaidukov · Irina Molotkovskaya · Alexander Sapozhnikov · Pavel Vlaskin · Nicolai Bovin
Accepted: 25 September 2006 / Published online: 17 January 2007 © Springer-Verlag 2007
Abstract Membrane-associated oligosaccharides are known to take part in interactions between natural killer (NK) cells and their targets and modulate NK cell activity. A model system was therefore developed using synthetic glycoconjugates as tools to modify the carbohydrate pattern on NK target cell surfaces. NK cells were then assessed for function in response to synthetic glycoconjugates, using both cytolysis-associated caspase 6 activation measured by Xow cytometry and IFN- production. Lipophilic neoglycoconjugates were synthesized to provide their easy incorporation into the target cell membranes and to make carbohydrate residues available for cell–cell interactions. While incorporation was successful based on Xuorescence monitoring, glycoconjugate incorporation did not evoke artifactual changes in surface antigen expression, and had no negative eVect on cell viability. Glycoconjugates contained Lex, sulfated Lex, and Ley sharing the common structure motif trisaccharide Lex were revealed to enhance cytotoxicity mediated speciWcally by CD16 +CD56+NK cells. The glycoconjugate eVects were dependent on saccharide presentation in a polymeric form. Only polymeric, or clustered, but not monomeric glycoconjugates
Elena I. Kovalenko (&) · E. Abakushina · E. Korchagina · S. Khaidukov · I. Molotkovskaya · A. Sapozhnikov · P. Vlaskin · N. Bovin Shemyakin & Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences, ul. Miklukho-Maklaya, 16/10, Moscow 117997, Russia e-mail:
[email protected] W. Telford · V. Kapoor National Cancer Institute, National Institute of Health, 9000 Rockville Pike, Bethesda, MD 20892, USA
resulted in alteration of cytotoxicity in our system, suggesting that appropriate presentation is critical for carbohydrate recognition and subsequent biological eVects. Keywords NK cell · Neoglycoconjugate · Saccharide · Flow cytometry · Caspase Abbreviations DOPE Dioleoylphosphatidylethanolamine residue FBS Fetal bovine serum Fluo 5-[(5-Aminopenthyl) thioureidyl]Xuorescein residue Glyc Oligosaccharide residue PAA Poly(N-2-hydroxyethylacrylamide) PBMC Peripheral blood mononuclear cells Bdi Gal1–3GalBtri Gal1–3(Fuc1–2)GalGal1–4(Fuc1–3)GlcNAcLex Ley Fuc1–2Gal1–4(Fuc1–3)GlcNAcx HSO3Le HSO3–3Gal1–4(Fuc1–3)GlcNAcHSO3Lea HSO3–3Gal1–3(Fuc1–4)GlcNAcSiaLex Neu5Ac2–3Gal1–4(Fuc1–3)GlcNAcNeu5Ac2–8Neu5AcSia2 3⬘SL Neu5Ac2–3Gal1–4GlcT Gal1–3GalNAcTn GalNAcSiaTn Neu5Ac2–6GalNAcasialoGM1 Gal1–3GalNAc1–4Gal1–4Glcglucitol HOCH2(CHOH)4CH2-
Introduction Natural killer (NK) cells represent an important lymphoid component of the innate immune system. They
123
314
play a critical role in anti-tumor responses and in resistance to infections through production of regulatory cytokines and cytotoxic activity directed against tumor and infected cells. NK cells are able to rapidly lyse virus-infected, tumor and other aberrant cells without antigen-speciWc sensitization (Trinchieri 1989). The susceptibility of target cells to NK-mediated cytolysis is dependent on an expressed repertoire of both NK cell receptors and relevant target cell ligands that can mediate activating and inhibitory signals for NK cells. In part, NK cell cytotoxicity is dependent on MHC class I binding receptors, which can both suppress and activate NK cell cytotoxic function (Ljunggren and Karre 1990). Glycosylation-dependent cell adhesion and signaling may also contribute to NK cell activation. Malignancy results in signiWcant changes in cell surface carbohydrate expression; a correlation between the expression of certain saccharides including Lex (Zarcone et al. 1987), T or Tn antigens (Blottiere et al. 1992) on target cells and their sensitivity to NK cells may therefore have considerable relevance to their anti-tumor activity. Cell surface oligosaccharides are strongly implicated in the regulation of NK cell eVector functions, positive and negative eVects of saccharides on NK-cell-mediated cytotoxicity have been demonstrated in several experimental models (Ogata et al. 1992; Yoshimura et al. 1996; Sol et al. 1999). Some of earliest data on the inhibition of NK cell cytolytic function were obtained using substances containing or consisting of various saccharides capable of binding to NK cells, including sialo-oligosaccharides, glycosphingolipids, -glucans and others (Van Rinsum et al. 1986; Bergelson et al. 1989; Duan et al. 1994; Inverardi et al. 1997). Many studies have attempted to Wnd common features of oligosaccharides that inXuence NK cell activity. However, the results of these studies were sometimes paradoxical and diYcult to interpret. For example, surface sialylation has been suggested to provide protection from NK-cell-mediated cytolysis (Ogata et al. 1992). Nevertheless, while glycophorin A eVectively inhibited NK cell cytotoxicity, its protective action was not connected with the sialic part of this molecule (El Ouagari et al. 1995). More recently, the inhibitory receptor siglec-7 expressed on NK cells was found to recognize speciWcally Sia2–8Sia motif (Nicoll et al. 2003). Interestingly, N-linked but not O-linked carbohydrate chains modulated NK cell cytotoxic activity, and the eVect of the N-linked saccharide chains was also shown to depend on their structure (Yoshimura et al. 1996; Sol et al. 1999; Baba et al. 2000). In particular, the complex N-linked carbohydrate chain glycans of glycophorin A increased cell resistance to NK cell killing (El Ouagari et al. 1995)
123
Histochem Cell Biol (2007) 127:313–326
whereas high mannose-type glycans enhanced target cell sensitivity to natural killing (Ahrens 1993). The target cell susceptibility to NK cell lysis could also be augmented by (1,3)-galactosyltransferase, but reduced by (1,2)-fucosyltransferase (Artrip et al. 1999) and (1,4) N-acetylglucosaminyltransferase (Yoshimura et al. 1996). These results suggested that it is not only the carbohydrate itself but the pattern of saccharide expression on cell surface that determined the relative susceptibility of the cell to NK cell action. Saccharide presentation, including the natural glycosylation “backbone” and the surface saccharide density are likely to be critical for cell signaling and function mediated by glycosylated molecules (Hakomori 2003). Taken together, while carbohydrates undoubtedly play a role in the regulation of NK activity, the data collected thus far remains rather controversial and incomplete. In part, the variability in observations may be connected with diVerent methods and techniques applying for the study, including the use of carbohydrates presented on diVerent molecular backbones. This variability in technique could have a signiWcant impact on receptor-ligand recognition and on subsequent signal transduction. In order to reliably and reproducibly study the role of carbohydrates in regulation of cell–cell interactions, it is therefore necessary to be able to modify the cell surface carbohydrate repertoire in a predictable and measurable way. In the past, this has been done by several diVerent methods, including glycosidase treatment, modulation of expression of speciWc enzymes and incorporation of carbohydrate containing molecules. Our novel approach to solve this problem included the modiWcation of the cell surface carbohydrate repertoire with lipophilic neoglycoconjugates (Kovalenko et al. 1998, 2004). In this model, target cells were pre-incubated with lipophilic neoglycoconjugates, which are incorporated into the cell surface prior to addition of NK cells. By coupling Xuorescent labels to these neoglycoconjugates, their uptake and retention can be quantitatively monitored by Xow cytometry. In present paper, we advance this approach by determining new characteristics of the glycoconjugate incorporation; assess the dynamics of cell surface glycoconjugate retention using Xuorochrome-conjugated glycoconjugates; estimate physiological side eVects of glycoconjugate incorporation into cells and use the model for evaluation of eVects of a number of saccharides on NK cell cytolytic function. Synthetic polymerbased glycoconjugates were also incorporated in target cells of NK cells, providing a unique mode of multivalent oligosaccharide presentation that may eVectively mimic native carbohydrate recognition. This approach
Histochem Cell Biol (2007) 127:313–326
makes use of poly-acrylamide as a chemically inert presentation matrix, allowing us to study the role of saccharides in cell–cell interactions displayed on a consistent and standardized backbone. We then used this method for screening of a panel of glycoconjugates for their ability to modulate both NK cell cytotoxicity and cytokine production. EVects of both polymeric (clustered) and monomeric (monovalent) glycoconjugates were then compared for the eYcacy in modulating NK activity. We determined that polymeric Lex and related oligosaccharides increased NK cell cytotoxicity using this novel mode of presentation.
Materials and methods Cell lines Erythroblastoid leukemia K562 and B-lymphoma Raji target cell lines were cultivated in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 1 mM sodium pyruvate, and 0.05 mM 2-mercaptoethanol (all from Sigma-Aldrich, St Louis, MO), hereafter referred to as complete medium. The NK-like cell line NK-92 was maintained in RPMI-1640 medium containing 10% FBS, 2 mM L-glutamine, 100 U/ml penicillin, 100 g/ml streptomycin, 200 U/ml recombinant human IL-2 (HoVmann-La Roche) and 10 ng/ml recombinant human IL-15 (Sigma-Aldrich) (Hodge et al. 2002). For 4 h before prior to the cytolytic reaction, NK-92 cells were grown at a density of 1 £ 106 per ml in medium lacking IL-2 and IL-15. All cells were cultured in a humid atmosphere containing 5% CO2 at 37°C. Glycoconjugates Synthetic polymeric glycoconjugates of several types were used in this work: (1) lipophilic glycoconjugates Glyc (15%)-PAA-DOPE (5%). In this construct, Glyc was the oligosaccharide residue, PAA was poly(N-2hydroxyethylacrylamide), and DOPE was dioleoylphosphatidylethanolamine (in this and subsequent cases, the molar portion of acrylate units modiWed by a ligand is designated as a percentage); (2) non-lipophilic glycoconjugates Glyc (15%)-PAA; (3) Xuorescein (Fluo—5-[(5-aminopenthyl)thioureidyl]Xuorescein residue)-labeled glycoconjugates contained diVerent amount of DOPE and a constant, 1% molar amount of Fluo. Glyc was represented by the saccharide residues Bdi, Btri, Lex, SiaLex, Ley, HSO3Lex, HSO3Lea, Sia2, 3⬘SL, T, Tn, SiaTn, asialoGM1, and glucitol. Monomeric lipophilic glycoconjugates (Glyc-DOPE) con-
315
taining Bdi, Lex, SiaLex, Ley, 3⬘SL, T, SiaTn, and asialoGM1 were used as indicated. Spacer oligosaccharides were from Lectinity (Moscow, Russia). Synthesis of polymeric glycoconjugates (MW»30 kDa) was performed as described earlier (Bovin et al. 1993). Incorporation of glycoconjugates into cells Target cells were harvested, washed with PBS, resuspended in PBS with 2% FBS containing 100 g/ml of polymeric glycoconjugate, shaken continuously and incubated for 45 min at 37°C unless otherwise speciWed. The cells were then washed twice and analyzed immediately or after indicated timepoints by Xuorescence microscopy and/or Xow cytometry (for Xuorochrome-conjugated lipids) or were immediately incorporated into a cytotoxicity assay. Fluorescent microscopy Fluorescent microscopy was performed with an Olympus AX-70 epiXuorescence microscope equipped with Xuorescein excitation/emission Wlters. About 105 K562 cells were treated with Xuorescein-labeled glycoconjugates as described, resuspended in GelMount mounting/anti-fade media (Vector Laboratories, Burlingame, CA), mounted on microscope slides and imaged within 4 h. Antibodies and immunolabeling FITC-conjugated monoclonal antibodies against human CD15, CD14, CD16, and CD45, PE-conjugated antiCD56 and PE-Cy5-conjugated anti-CD3 were obtained from Invitrogen Caltag (Burlingame, CA). FITC-conjugated anti-CD20 and anti-CD36 antibodies were obtained from BD Pharmingen (San Diego, CA). Anti-human MICA/MICB monoclonal antibody was purchased from Immatics Biotechnologies GmbH (Germany) and hsp72speciWc antibody BRM-22 and F(ab⬘)2 anti-mouse Ig FITC conjugates for secondary labeling were obtained from Sigma-Aldrich (St Louis, MO). NK cell labeling with CD surface markers was done simultaneously for 30 min at 4°C, followed by immediate analysis. For HSP70 and MICA/MICB labeling, cells were incubated with the primary antibody for 30 min at 4°C, washed twice and stained with secondary antibody (30 min at 4°C), then washed twice again followed by analysis. NK cell isolation Peripheral blood mononuclear cells (PBMC) were separated from heparinized whole blood by density gradient centrifugation. A group of 15 healthy donors took
123
316
part in the investigation. Some donors were sampled for their blood more than once. The cells were washed twice with PBS containing 2% FBS. Then NK cells were isolated by magnetic separation technique using negative selection (NK Isolation Kit, Miltenyi Biotec). Cells were labeled with a mixture of biotin-conjugated lineage-speciWc antibodies to remove most non-NK cells (CD3, CD4, CD14, CD15, CD19, CD36, CD123, and CD235a), washed and labeled with streptavidinconjugated magnetic particles. Non-labeled cells were then separated using a Miltenyi Biotec AutoMACS magnetic cell separator. NK cells were routinely puriWed to >80% by this method. Flow cytometry and cell sorting Glycoconjugate uptake analysis was measured on a Coulter EPICS Elite Xow cytometer equipped with a 488 nm laser and a detector with a Xuorescein bandpass Wlter. In these measurements, the amount of Fluolabeled glycoconjugate molecules bound to the cells was estimated by comparing Xuorescence of the cells with a standard Xuorescent microsphere (IMMUNOBRIT Level II, Coulter Electronics Inc., Hialeah, FL). Multicolor NK cell immunophenotyping was carried out on a FACSCalibur Xow cytometer (BD Biosciences, San Jose, CA). A minimum of 10,000 events was collected for each sample. Data were analyzed using WinMDI version 2.8 (Dr. Joe Trotter, Scripps Institute) or CellQuest version. 3.3 (BD Biosciences) Xow cytometry analysis programs. Multicolor NK cell sorting was done by labeling cells with the NK-speciWc antibodies described above, followed by separation of the NK subsets on a FACSVantage DiVa Xuorescence activated cell sorter (BD Biosciences). Cytotoxicity assay NK-mediated cytotoxicity was estimated by measuring caspase 6 activity in target cells using a cell-permeant Xuorogenic caspase 6 substrate (CyToxiLux kit, Oncoimmunin, Gaithersburg, MD). K562 target cells were labeled with a proprietary tracking dye (TFL2) and coincubated with NK cells at the indicated ratios for 1 h. The cell mixture was subsequently loaded with the caspase 6 Xuorogenic substrate for 45 min at 37°C, washed and analyzed on a Becton Dickinson FACScan equipped with a 488 nm air-cooled argon laser. Caspase 6 substrate Xuorescence was measured through the FL1 channel (530/30 nm Wlter) and the tracking dye through the FL2 channel (575/26 nm Wlter). Caspase 6 activity could be measured speciWcally in the TFL2positive (target) fraction by “gating” on the cells posi-
123
Histochem Cell Biol (2007) 127:313–326
tive for the tracking dye. The proportion of cells with spontaneously activated caspase 6 did not exceed 3% (and on average was
