Proteomic Comparison of 3D and 2D Glioma Models Reveals

Apr 7, 2014 - Department of Neurosurgery, General Hospital of People's Liberation Army Chengdu Military Region, Chengdu 610083, China. ‡. Department...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/jpr

Proteomic Comparison of 3D and 2D Glioma Models Reveals Increased HLA‑E Expression in 3D Models is Associated with Resistance to NK Cell-Mediated Cytotoxicity Weiqi He,† Yongqin Kuang,† Xuemin Xing,† Richard J. Simpson,‡ Haidong Huang,† Tao Yang,† Jingmin Chen,† Libin Yang,† Enyu Liu,† Weifeng He,*,§,∥ and Jianwen Gu*,† †

Department of Neurosurgery, General Hospital of People’s Liberation Army Chengdu Military Region, Chengdu 610083, China Department of Biochemistry, La Trobe Institute for Molecular Science, La Trobe University, Bundoora, Victoria, Australia § Chongqing Key Laboratory for Disease Proteomics and ∥State Key Laboratory of Trauma, Burns and Combined Injury, Institute of Burn Research, Southwest Hospital, Third Military Medical University, Chongqing 400038, China ‡

S Supporting Information *

ABSTRACT: Three-dimensional cell culture techniques can better reflect the in vivo characteristics of tumor cells compared with traditional monolayer cultures. Compared with their 2D counterparts, 3D-cultured tumor cells showed enhanced resistance to the cytotoxic T cell-mediated immune response. However, it remains unclear whether 3D-cultured tumor cells have an enhanced resistance to NK cell cytotoxicity. In this study, a total of 363 differentially expressed proteins were identified between the 2D- and 3D-cultured U251 cells by comparative proteomics, and an immune-associated protein−protein interaction (PPI) network based on these differential proteins was constructed by bioinformatics. Within the network, HLA-E, as a molecule for inhibiting NK cell activation, was significantly up-regulated in the 3D-cultured tumor cells. Then, we found that the 3D-cultured U251 cells exhibited potent resistance to NK cell cytotoxicity in vitro and were prone to tumor formation in vivo. The resistance of the 3D-cultured tumor cells to NK cell lysis was mediated by the HLA-E/NKG2A interaction because the administration of antibodies that block either HLA-E or NKG2A completely eliminated this resistance and significantly decreased tumor formation. Taken together, our findings indicate that HLA-E upregulation in 3D-cultured cells may result in enhanced tumor resistance to NK cell-mediated immune response. KEYWORDS: proteomics, three-dimensional culture, glioma, HLA-E, NK cell cytotoxicity, tumor surveillance



(tumor-associated antigen)11,12 expression in 3D cells was reported to play roles in their resistance to the T-cell-mediated immune response. Furthermore, tumor-infiltrating lymphocytes failed to recognize autologous bladder tumor cells in 3D culture, although they could effectively kill the targets when they were cultured in 2D.13 A CTL clone specific for a mutated α-actinin-4 peptide expressed by autologous lung cancer cells killed targets growing in 3D with a low efficiency; this result was related to a down-regulation in HSP70 expression.14 Another CTL clone specific for a TAA expressed by melanoma cells failed to recognize the targets when tumor cells were cultured in 3D.15,16 It is well known that in addition to CTLs the NK cell-mediated immune response is very important for tumor immune surveillance. However, whether tumor cells cultured in 3D have a higher resistance to NK cell-mediated immune response than tumor cells cultured in 2D and the

INTRODUCTION

Three-dimensional culture techniques have been developed in the past decade in the hope of modeling alterations in tissue architecture that are of critical importance for tumor development.1−3 Some scientific research focused on the tumor cells cultured in 3D model because tumor cells could grow as spheroids and colonies and were closer to in situ growth. Grun et al. compared the biological and morphological characteristics of 3D cell culture models with those of 2D models of human ovarian and endometrial cancers and found that 3D models were better than 2D models for studying the molecular and biological mechanisms of tumor progression.4 Lee et al. found that 3D models of EOCs (epithelial ovarian cancer) mimicked primary tumors in vivo better than 2D models; EOCs cultured in 3D or 2D also exhibited different sensitivity to chemotherapeutic agents.5 Tumor cells cultured in 3D exhibited the defective immune recognition by cytotoxic T lymphocytes (CTLs).6−8 The downregulation of human leukocyte antigen (HLA)9,10 or TAA © 2014 American Chemical Society

Received: October 9, 2013 Published: April 7, 2014 2272

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

Dimethyl Labeling, Separation, and Mass Spectrometry Analysis

mechanisms that underlie this response remain unclear. In the present study, we established a method for the 3D culturing of a glioma cell line (U251) and compared the biological features of 2D and 3D cultures, including the morphological characteristics and proteomic profiles. We revealed that the expression of HLA-E was up-regulated in 3D glioma cells and resulted in the enhanced resistance of the tumor cells to NK cell-mediated cytotoxicity.



For stable isotope dimethyl labeling, the digested peptide samples from the 2D and 3D cultures were reconstituted separately with 200 μL of CH3COONa, pH 5.9. Eight microliters of CH2O (light labeled) and CD2O (heavy labeled) were added to the 3D and 2D samples, respectively, and 8 μL of 0.6 M NaBH3CN was added to both samples, which were incubated at room temperature for 1 h. To quench the reaction, we added 32 μL of 1% (v/v) ammonia solution and 16 μL of 5% formic acid to both samples on ice.17 The light labeled 3D and the heavy labeled 2D samples were mixed at a 1:1 ratio and desalted before strong cation exchange (SCX) fractionation. For the SCX fractionation, the mixed labeled peptides were resuspended in buffer A containing 5 mM KH2PO4, 20% acetonitrile, pH 2.7. SCX was performed on a polysulfethyl column (2.1 × 50 mm, 5 μm × 200 Å) using a KCl gradient from 0 to 0.5 M to fractionate the peptides. Twelve fractions were collected and desalted with C18 ZipTips (Millipore) before MS analysis.

MATERIALS AND METHODS

Ethics Statement

All human materials used in this study were acquired after approval by the hospital’s Institutional Review Board. All samples were encoded to avoid patient identification. It is important to note that a written, informed consent form, through which the patients agreed to have samples taken for research purposes, was obtained in all cases. Patients and Tissue Samples

Three glioma patients were enrolled in this study. All tissue samples were from de novo pediatric, high grade gliomas that were obtained ante mortem at the General Hospital of People’s Liberation Army Chengdu Military Region; all diagnoses were confirmed by central pathological review. The peripheral blood of these patients was collected for NK cell isolation when the 2D- and 3D-cultured glioma cells were ready for use.

LC−MS/MS and Data Analysis

A mass spectrometer (Applied Biosystems) coupled to a nanoflow HPLC system (TempoTM, Applied Biosystems) was used for the relative quantitation of the whole cell lysate proteins from the 2D and 3D cell cultures. The LC−MS method was the same as that used by Chen et al.18 and Zhang et al.19 Raw data from QSTAR ELITE were analyzed using the MASCOT Daemon software (version 2.2.2, Matrix Science). The data were searched against the Swiss-Prot database using in house mascot search engine. Carbamidomethylation of cysteine was set as a fixed modification, and the oxidation of methionine and the phosphorylation of serine, threonine, and tyrosine were set as variable modifications. The peptide mass tolerance was set at 200 ppm and 0.4 Da. The peptide charge was set to 2+ and 3+, and up to two missed cleavage sites were allowed. The significance threshold was set at P < 0.05. After the Mascot search, the raw data obtained from the database search were quantified using Mascot Distiller (Version 2.3.2.0, Matrix Science). For the quantitation analysis, we set the fraction, correlation, and standard error at 0.5, 0.9, and 0.2, respectively. The false-positive rates of the peptide spectrum matches were determined by data searches against a decoy database. The peptide ratios were calculated as the weighted average ratios (ion intensity versus ratio) if several spectra were available for the same peptide. The median of all quantitative data from all peptides was used to normalize the peptide ratios. Protein ratios were also calculated as the geometric mean of the peptide ratios for a protein. Student’s t test was performed, and P values were calculated using the SPSS 13.0 software (SPSS, Chicago, IL).

Cell Lines: Two- and Three-Dimensional Cultures

U251 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) with 10% (v/v) fetal bovine serum, 100 IU/ mL penicillin, and 100 μg/mL streptomycin. The same media were used for each cell line in 2D and 3D culture, and all cultures took place in a humidified incubator at 37 °C with a 5% CO2 atmosphere. 3D culture utilized the Rotary Cell Culture System (RCCS) (Synthecon, Houston), which was situated within a cell culture incubator. Culture began when 1 × 106 cells were introduced into a 10 mL culture vessel as a single cell suspension, and the cells were initially rotated at 15 rpm (rpm). Once a visible aggregate formed, the revolution speed was adjusted to balance the Coriolis force against gravity to maintain the aggregate in stationary freefall. The aggregates were harvested after 3 weeks, and the media were changed twice each week. Protein Extraction and Protein Determination

The cells cultured in 2D or 3D models were washed three times with PBS and centrifuged at 1000 rpm for 5 min. The supernatants were removed, and 200 μL of lysis buffer was added (7 M urea, 2 M thiourea) to the cell pellet. After sonication for 3 min (5 s intervals with 2 s sonication), the sonicated lysate was centrifuged at 20 000g for 20 min. The Bradford method was adapted for use with the lysis buffer and was used to determine the protein concentration in the samples. The protein solution was reduced with 5 mM DTT at 37 °C for 45 min, alkylated with 15 mM iodoacetamide at room temperature for 45 min and then digested with Lys-C (1:30) at room temperature for 4 h. After six-fold dilution with H2O, the solution was subsequently digested overnight with trypsin (1:50) at 37 °C. Note that the pH value should be 8.0, and the pH was checked before each digestion step; 5% NH3 in H2O was used to adjust the pH. For the following steps, 200 μg of protein lysate was used.

Immune System Process Analysis Using Cytoscape with the ClueGO+Cluepedia Plugin

We used ClueGO+Cluepedia, an easy to use Cytoscape plugin,20 to integrate the Gene Ontology (GO) terms and the immune system process and to create a functionally organized GO term/pathway network of the differentially expressed proteins. The enrichment tests for the terms and groups were two-sided (enrichment/depletion) tests based on a hypergeometric distribution, and all terms that were significant with P < 0.01 (after correcting for multiple term testing using the Benjamini and Hochberg false discovery rate corrections) were selected as over-represented. The kappa score threshold was set 2273

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

The percentage of 51Cr release was calculated using the following equation: 100 × ([experimental release − spontaneous release]/[maximum release − spontaneous release]).

to 0.5. ClueGO+Cluepedia visualized the selected terms in a functionally grouped annotation network that reflected the relationships between the terms based on the similarity of their associated proteins. Finally, these associated proteins were selected to construct immune-associated protein−protein interaction (PPI) networks after the removal of the separated proteins.

Tumor Models

BALB/c-nude mice were injected intraperitoneally with 20 μL of antiasialo GM1 (anti-ASGM1; Wako Chemicals, Richmond, VA) on Day −3 to deplete the NK cells of the mice; on Day −1, 1 × 106 MACS-purified human peripheral NK cells were intravenously injected into the nude mice. On Day 0, 5 × 105 U251 cells cultured in 2D or 3D were subcutaneously injected into the left flanks of the nude mice (two groups: 2D and 3D; n = 10 per group). To check the function of HLA-E on tumor cells, following 2D or 3D tumor cells injection, on Day 0 and 3, an anti-HLA-E antibody (200 μg/mouse) or an isotype control IgG (200 μg/mouse) was intraperitoneally injected into the mice (Four groups: 2D+control IgG, 2D+anti-HLA-E antibody, 3D+control IgG, and 3D+anti-HLA-E antibody; n = 15 per group). To check the function of NKG2A on NK cells, following 2D or 3D tumor cells injection, on Day 0 and 3, an anti-NKG2A antibody (200 μg/mouse) or an isotype control IgG (200 μg/mouse) was intraperitoneally injected into the mice (Four groups: 2D+control IgG, 2D+anti-NKG2A antibody, 3D+control IgG, and 3D+anti-NKG2A antibody; n = 15 per group). Tumor growth was monitored and recorded daily for 3 weeks.

Construction of PPI Network

PPIs of the up- and down-regulated proteins were predicted using the Search Tool for the Retrieval of Interacting Genes/ Proteins (STRING) database v9.0 (http://www.string-db.org/ ). The proteins were linked based on six criteria: neighborhood, gene fusion, co-occurrence, coexpression, experimental evidence, and existing databases.21 Quantitative Real-Time PCR

Total RNA was extracted from cells using the RNeasy Mini kit (Qiagen, USA) and reverse-transcribed using the Strata Script First Strand Synthesis System (Stratagene, USA). PCR was performed on an iCycler (Bio-Rad). The cycling conditions were 12 min at 95 °C, followed by 40 cycles of 95 °C for 15 s and 60 °C for 60 s. Analysis was performed using the sequence detection software that was supplied with the instrument. HLA-E was analyzed concurrently on the same plate with HPRT, and cytokine transcripts were normalized to GAPDH transcripts using the following primers: GAPDH: 5′-CTCTCTGCTCCTCCTGTTCGAC-3′ (450−469), 5′-TGAGCGATGTGGCTCGGCT-3′ (636−619); and HLA-E: 5′-GGGACACCGCACAGATTTT-3′ (266−284), 5′-CTCA-GAGGCATCATTTGACTTTT (519−497).

Statistics

Statistical significance was evaluated using either a two-tailed unpaired Student’s t test or nonparametric analysis if the SDs were significantly different between the two groups being tested; analyses were performed using the Instat version 2.03 for Macintosh (GraphPad). The incidence of tumor development was compared and analyzed using the log rank test with the GraphPad Prism version 3.0a software for Macintosh (GraphPad). Throughout the text, figures, and legends, the following symbols are used to denote statistical significance: *, P < 0.05; **, P < 0.01.

Flow Cytometry

For surface HLA-E staining, the cells were labeled with PEconjugated antihuman HLA-E antibody (eBioscience, USA) for 30 min on ice; the cells were then washed, fixed with 4% PFA, and finally resuspended in FACS EDTA buffer. The cells prepared for flow cytometry were analyzed after gating on the live cells. Independent experiments were performed a minimum of three times.



RESULTS

Cell Culture

Western Blot Assay

Upon cell culture in RCCS, the U251 cells under the 3D microgravity condition start to generate spheroid after 48 h, and the size of the spheroid became 400 to 500 μm diameter containing on average 4000 cells in the next 6 days (Figure S1A in the Supporting Information). Proliferation kinetics of U251 cells cultured in 2D or 3D were sharply different. 2D cultures showed much more rapid cell proliferation than 3D cultures (Figure S1B in the Supporting Information). Comparable results were obtained by direct count of trypsinized cells.

The whole cell lysate protein extracts from the 2D or 3D cell cultures were prepared using a lysis buffer containing protease inhibitor cocktail 8340. For Western Blot analysis, ∼20 μg urinary protein was separated by 12% SDS-PAGE, then transferred to a PVDF membrane (Millipore, USA) and probed with polyclonal rabbit antihuman HLA-E antibody (1: 500, Santa Cruz, USA), respectively. The blots were labeled with horseradish peroxidase-conjugated secondary antibodies (1:10000) and visualized with enhanced chemiluminescence (ECL) detection system (Pierce Biotech, Rockford, IL).

Proteomics Analysis of the Differentially Expressed Proteins between 3D- and 2D-Cultured U251 Cells

NK Cell Cytotoxicity Assay

It is well known that cells cultured under 3D and 2D conditions display different biological characteristics. However, a global differential analysis between spheroid and monolayer cultured cells at the protein level has not yet been performed. Herein, we analyzed the differentially expressed proteins between 3Dand 2D-cultured U251 cells, a human glioma cell line, using quantitative proteomics. The strategy is outlined in Figure 1. In brief, the total proteins from U251 cells in 3D- and 2D-cultured conditions were extracted, and the protein samples were used for quantitative proteome analysis. Twelve SCX fractions were collected and labeled with stable isotope dimethyl labels for

Cytotoxicity was assessed using the standard 51Cr-release method. In brief, U251 cells cultured in 3D or 2D were labeled with 51Cr. Human NK cells purified by MACS and glioma cells were added in triplicate into 96-well plates at various effector/target (E/T) ratios for 4 h at 37 °C. In some experiments, freshly isolated human NK cells and U251 cells were cocultured in the absence or presence of a specific antibody against HLA-E or NKG2A. Spontaneous 51Cr release was determined by incubating the target cells with the medium alone. Maximum release was determined by adding 2% NP-40. 2274

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

proteins (Figure S3 in the Supporting Information). Of them, four proteins were related to 4-hydroxyproline metabolic process, and a total of three proteins were annotated within “4-hydroxyproline metabolic process” (Table S2 in the Supporting Information). Moreover, in the KEGG pathway analysis, the glycolysis-associated terms “glycolysis/gluconeogenesis”, “glycosaminoglycan degradation”, and “glycosphingolipid biosynthesis” were enriched in the 3D up-regulated proteins (Figure S4 in the Supporting Information). Construction of an Immune-Associated PPI Network for the Up-Regulated and Down-Regulated Proteins

Immune escape is important for tumor development and progression in the body.22 To understand what differences were observed in the immune-associated proteins between the 3Dand 2D-cultured U251 cells, we used the Cytoscape plugin ClueGO+Cluepedia to analyze the differentially expressed proteins involved in the immune system process. On the basis of the 363 differentially expressed proteins, the immuneassociated annotations network was created as a group of functionally organized GO terms (Figure 2A). It comprised four specific, significantly overrepresented-terms: “antigen processing and presentation of peptide antigen via MHC class I”, “antigen processing and presentation of exogenous peptide antigen”, “antigen processing and presentation of exogenous peptide antigen via MHC class I, TAP-dependent”, and “antigen processing and presentation of exogenous peptide antigen via MHC class I, TAP-independent”. The information in the immune term associated network is described in detail in Table 1. A total of 15 differentially expressed proteins were involved in these 4 immune-associated terms. Because the PPI network better reflected the biological status of the cells than single proteins, the 15 immune-associated proteins were used to construct PPI networks using STRING analysis. Single nodes and small components initially assembled in the PPI network were removed, and only the largest component was saved as a new PPI network. A PPI network including 13 proteins (PSMC2, HLA-B, HLA-A, PSMA1, CALR, PSMB1, PSMD2, PSMD3, HLA-C, HLA-E, PSMA2, AP1G1, and PDIA3) was constructed into the immuneassociated PPI network (Figure 2B, Table 2). The detailed information in the immune-associated PPI network is described in Table S3 in the Supporting Information. Within this network, five proteins (CALR, HLA-B, HLA-C, HLA-E, and PDIA3) were up-regulated and eight proteins (AP1G1, HLA-A, PSMA1, PSMA2, PSMB1, PSMC2, PSMD2, and PSMD3) were down-regulated in the 3D-cultured U251 cells. These results suggested that the function of antigen processing and presentation may be significantly different in the 2D- and 3Dcultured U251 cells.

Figure 1. Overview of the procedure used to analyze the proteomes of the 2D- and 3D-cultured U251 cells. LC, liquid chromatography; MS, mass spectrometry; SCX, strong cation exchange.

quantitation. After the data sets were searched against the Swiss-Prot database, the false-positive rates of the peptidespectrum matches determined by searching against a decoy database were below 1%. A total of 1118 high-confidence proteins were identified. Comparison and statistical analysis between these two groups were then performed. The stable isotope-labeling technique that we employed provided relative quantitative information, and the light/heavy (L/H) ratio from each peptide provided information about the protein levels. We analyzed the L/H ratio for the total proteins, and 649 proteins were quantified. The following criteria were required to consider a protein to be differentially expressed: two or more high-confidence (>95%) unique peptides were identified, and the peptides exhibited a fold change >2 according to Protein Quant. Testing for multiple comparisons from the quantitative information eventually led to 363 differentially expressed proteins. Among them, compared with 2D models, a total of 68 proteins were up-regulated and 295 were down-regulated in 3D models (Table S1 in the Supporting Information). Characterization of the Differentially Expressed Proteins via Gene Ontology Annotation

3D-Cultured U251 Cells Exhibit Enhanced HLA-E Expression and Resistance to NK Cell Cytotoxicity

The 363 differentially expressed proteins were functionally categorized based on universal Gene Ontology (GO) annotation terms using the ClueGO+Cluepedia program. The information from the GO analysis is described in detail in Table S2 in the Supporting Information. The results showed that in the biological process category 45 GO terms were enriched in the 3D-down-regulated proteins (Figure S2 in the Supporting Information). Of them, 23 proteins were related to metabolic process, 14 were related to cellular process, 6 were related to the regulation of protein metabolic process, and 2 were related to the regulation of ligase activity. In the biological process category, 14 GO terms were enriched in the 3D up-regulated

CTLs and NK cells, two major subsets of effector cells in tumor immune surveillance, play critical roles in tumor growth.23 It has been demonstrated that the 3D growth of tumor cells is able to affect antigen recognition by specific CTLs, resulting in a reduction of CTL-mediated cytotoxicity.6−8 However, the effects of 3D culture on NK cell-mediated cytotoxicity in tumor cells remain unknown. HLA-E is a well-known ligand of CD94/ NKG2A that inhibits NK cell function.24 It is reasonable to assume that the up-regulation of HLA-E could enhance the resistance of tumor cells to NK cell-mediated cytotoxicity. Consistent with the results of the proteomic quantification, 2275

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

Figure 2. Immune-associated PPI network was constructed using a bioinformatics approach based on the differentially expressed proteins. (A) Immune-associated terms were analyzed using the Cytoscape software with the ClueGO+Cluepedia plug-in. Four immune-associated enriched immune process terms were obtained and are shown. (B) Immune-associated PPI network composed of 13 differentially expressed proteins associated with the four immune-associated terms was constructed using STRING.

Table 1. Differentially Expressed Proteins in the Immune Procession Terms GO Term antigen processing and presentation of peptide antigen via MHC class I antigen processing and presentation of exogenous peptide antigen antigen processing and presentation of exogenous peptide antigen via MHC class I, TAP-dependent antigen processing and presentation of exogenous peptide antigen via MHC class I, TAP-independent

associated genes BCAP31, CALR, HLA-A, HLA-B, HLA-C, HLA-E, PDIA3, PSMA1, PSMA2, PSMB1, PSMC2, PSMD2, PSMD3 ACTR1A, AP1G1, CALR, DYNC1H1, HLA-A, HLA-B, HLA-C, HLA-E, PDIA3, PSMA1, PSMA2, PSMB1, PSMC2, PSMD2, PSMD3 CALR, HLA-A, HLA-B, HLA-C, HLA-E, PDIA3, PSMA1, PSMA2, PSMB1, PSMC2, PSMD2, PSMD3 HLA-A, HLA-B, HLA-C, HLA-E

cells by MACS as effector cells and cocultured them with either 2D- or 3D-cultured U251 cells as target cells to examine the capacity of tumor cell resistance to NK cell lysis in vitro. Compared with the 2D-cultured U251 cells, the 3D-cultured tumor cells showed significantly enhanced resistance to NK cell cytotoxicity (Figure 4A).

HLA-E expression in the 3D-cultured U251 cells was shown to be markedly higher than in the 2D-cultured U251 cells using real-time PCR (Figure 3A, P < 0.01), FACS (Figure 3B), and Western Blot (Figure 3C). To further assess the NK cell-mediated cytotoxicity of 2Dand 3D-cultured tumor cells, we purified human peripheral NK 2276

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

Table 2. Differentially Expressed Proteins in the Immune-Associated PPI Network UniprotKB AC

gene name

gene name 1

ratio (L/H)

P01892 P20618 P35998 O43747 P25787 O43242 Q13200 P25786 P10321 P13747 P27797 P30101 P30480

HLA-A PSMB1 PSMC2 AP1G1 PSMA2 PSMD3 PSMD2 PSMA1 HLA-C HLA-E CALR PDIA3 HLA-B

HLA-A proteasome subunit beta, type 1 proteasome 26S subunit, ATPase 2 adaptor related protein complex 1, gamma 1 subunit proteasome subunit alpha type 2 proteasome 26S subunit, ATPase 3 proteasome 26S subunit, ATPase, 2 proteasome subunit alpha type 1 HLA-C HLA-E calreticulin protein disulfide isomerase A3 HLA-B

0.222 0.333 0.335 0.353 0.407 0.451 0.455 0.493 2.106 2.106 2.308 2.421 4.935

stronger resistance to NK cell-mediated cytotoxicity than the 2D-cultured tumor cells. Up-Regulation of HLA-E in the 3D-Cultured Cells Enhances Tumor Resistance to NK Cell-Mediated Cytotoxicity by Binding to CD94/NKG2A

Similar to other MHC class I molecules, HLA-E is a heterodimer consisting of an α heavy chain and a light chain (β-2 microglobulin). HLA-E has a very specialized role in cell recognition by NK cells. NK cells recognize the HLA-E-peptide complex using the heterodimeric inhibitory receptor CD94/ NKG2A.24 When CD94/NKG2A is stimulated, it produces an inhibitory effect on the cytotoxic activity of the NK cells to prevent cell lysis.24 Moreover, HLA-E has been reported to have increased expression in primary glioma tissues compared with normal CNS tissues in vivo.25 Using our system, we tested whether 3D-cultured U251 cells were resistant to NK cellmediated cytotoxicity via HLA-E. Indeed, the addition of an anti-HLA-E-blocking antibody could reverse the resistance of the 3D-cultured U251 cells to NK cell-mediated lytic activity, resulting in levels equivalent to the 2D-cultured tumor cells in vitro (Figure 5A). We further performed in vivo experiments using the tumor model, where nude mice with human NK cells were inoculated with U251 cells, together with the control IgG or an anti-HLA-E antibody. As expected, by the addition of the anti-HLA-E antibody, the enhanced tumor development of the 3D-cultured U251 cells was almost completely eliminated, resulting in tumor formation comparable to the tumor incidence of the 2D-cultured cells (Figure 5B). Because HLA-E−peptide complexes are ligands of NGK2A, we used an anti-NKG2A antibody to block the interaction of HLA-E on glioma cells and NKG2A on NK cells. Similar to the results of blocking HLA-E, NKG2A blockage also reduced the resistance of the 3D-cultured U251 cells to NK cytotoxicity in vitro, resulting in levels comparable to the 2D-cultured cells (Figure 5C). Similar to HLA-E blocking experiment in vivo, with the addition of the anti-NKG2A antibody, the enhanced tumor development of the 3D-cultured U251 cells was almost completely abolished, resulting in tumor formation comparable to the tumor incidence of the 2D-cultured cells (Figure 5D). Taken together, these data strongly support the idea that glioma cells cultured in 3D could facilitate tumor development via HLA-E-dependent suppression of NK cytotoxicity.

Figure 3. Validation of HLA-E expression in the 2D- and 3D-cultured U251 cells using real-time PCR, FACS, and Western Blotting. (A) Transcription of HLA-E in the 2D- and 3D-cultured U251 cells. 2Dand 3D-cultured cells were evaluated using real-time PCR analysis, and the HLA-E transcripts were normalized against HPRT. (B) Expression of HLA-E on the 2D- and 3D-cultured U251 cells was analyzed by flow cytometry. The isotype IgG-labeled 2D-cultured U251 cells were used as a control. (C). Expression of HLA-E on the 2D- and 3Dcultured U251 cells was analyzed by Western Blotting; intracellular βactin served as a loading control. One example from three independent experiments is shown.

Sex- and age-matched nude mice (n = 20), which were depleted of mouse NK cells and reconstituted with human NK cells, were subcutaneously inoculated with U251 cells from 3D (n = 10) or 2D culture (n = 10) (2 × 105 cells/mouse), respectively, to further examine the effect of 3D or 2D culture on tumor cell immunity in vivo. Tumor growth was monitored and recorded daily as previously described. As expected, the 3D-cultured tumor cells exhibited significantly increased tumor occurrence (Figure 4B) compared with 2D-cultured tumor cells. These data indicated that the 3D-cultured tumor cells had 2277

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

recurrence. The tumor lysis capacity of NK cells from patients was ordinarily evaluated using K562 cells, which are easily killed by NK cells as they lack the MHC complex required to inhibit NK activity, or cells from primary tumor tissues as target cells. The latter, which directly reflects the interaction between NK cells and tumor tissues in patients, was better than the former, which could reflect the functions of the NK cells themselves in the patient. However, as previously mentioned, NK cells have significantly enhanced cytotoxicity toward 2D-cultured tumor cells compared with 3D-cultured tumor cells, which more closely resemble their in vivo equivalents. Herein, we randomly selected three patients bearing gliomas for further analysis. Similar to the results previously mentioned, 3D-cultured glioma cells have significantly enhanced expression of HLA-E compared with their 2D counterparts (Figure S5A in the Supporting Information). Interestingly, there was no significant difference in the NK cell-mediated cytotoxicity to either K562 cells (P > 0.05) or 2D-cultured glioma cells (P > 0.05) among the three patients (Figure S5B in the Supporting Information). However, the NK cells from patient 1 and patient 2 displayed significantly enhanced cytotoxicity to the 3D-cultured tumor cells compared with the NK cells from patient 3 (patient 1 or patient 2 vs patient 3 P < 0.01; Figure S5B in the Supporting Information). It is reasonable to assume that the 3D-cultured tumor cells, as target cells, may better reflect the status of NK cell tumor killing in patients.



DISCUSSION Cells cultured in 3D better reflect in vivo cell behavior compared with cells cultured in 2D. The cellular/molecular morphology and gene-expression profiles of 3D-cultured cells are significantly different from those of 2D-cultured cells.1−3,8 In this study, we analyzed the altered proteomes of 2D- and 3D-cultured glioma cells, and a total of 1118 high-confidence proteins (Table S1 in the Supporting Information) were identified, among which 363 proteins were differentially expressed between the 3D- and 2D-cultured U251 cells (Table S1 in the Supporting Information). It is reported that the proliferation of tumor cells in spheroid culture was significantly slower than that in traditional flat culture, and the 3D-cultured tumor cells displayed significantly different metabolic characteristics compared with their 2D counterparts, 27 and the increased glycolysis appears to be a predominant feature.28 Consistent with these reports, the 363 differentially expressed proteins between 2D- and 3D-cultured U251 cells were clustered in metabolic process-associated terms; the 68 up-regulated proteins in 3D models were enriched in glycolysis-associated terms. Furthermore, it is well known that tumor cells grown in 3D architectures have enhanced resistance to drugs.29 Interestingly, among these differentially expressed proteins, prolyl 4-hydroxylase beta polypeptide (P4HB), which was reported to enhance temozolomide resistance in malignant gliomas via the endoplasmic reticulum stress response (ERSR) pathway,30 was confirmed to be significantly up-regulated in 3D culture compared with 2D culture using Western blot analysis (Figure S6 in the Supporting Information). These results demonstrated that we have provided a reference proteome that can facilitate future research regarding the 3D cell culture model for U251. Immune escape is an important mechanism for tumor growth in the body,22 and cytotoxic T cells and NK cells are two types of effector cells for tumor immune surveillance.23 Tumor cells cultured in 3D, which mimic their in vivo

Figure 4. Evaluation of NK cell cytotoxicity on the 2D- and 3Dcultured U251 cells. (A) Compared with the 2D-cultured U251 cells, the 3D-cultured U251 cells displayed an enhanced resistance to NK cell-mediated lysis in vitro. MACS-purified human NK cells were cocultured with either 2D- or 3D-cultured U251 cells at various ratios. The data are shown as the mean percentages of 51Cr release (±SD) after 6 h of in vitro culture and are representative of three independent experiments. (B) 2D-cultured U251 cells showed a lower incidence of tumor development in vivo than the 3D-cultured U251 cells. Sex- and age-matched BALB/c-Nude mice were injected intraperitoneally with 20 μL of antiasialo GM1 (anti-ASGM1; Wako Chemicals, Richmond, VA) on day 3 to deplete the mice of NK cells; on day 1, 1 × 106 MACS-purified human peripheral NK cells were intravenously injected into the nude mice. On day 0, either 5 × 105 2D- or 3D-cultured U251 cells were subcutaneously injected into the left flanks of the nude mice (n = 10 per group). Tumor incidence was monitored and recorded daily for over 3 weeks, and tumors that were >4 × 4 mm2 were considered positive. The data represent twp independent experiments. **, P < 0.01.

Glioma Cells from Primary Tumor Tissues Cultured in 3D Have Increased HLA-E Expression and Enhanced Resistance to NK Cell-Mediated Cytotoxicity Compared with Those Grown in 2D

Gliomas are the most common tumors of the central nervous system in adults, and they account for 40−50% of all intracranial tumors; however, these tumors often have postoperative relapses.26 The capacity of NK cell-mediated tumor killing is very important for tumor development, metastasis, and 2278

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

Figure 5. Resistance of the 3D-cultured U251 cells to NK cell cytotoxicity was HLA-E-dependent. (A) Anti-HLA-E antibody abrogated the resistance of 3D-cultured U251 cells to NK cell cytotoxicity in vitro. The data are representative of three independent experiments. (B) Anti-HLA-E antibody could reverse the resistance of the 3D-cultured U251 cells to NK function in vivo. Nude mice that were depleted of NK cells and reconstituted with human NK cells (n = 15/group) were intravenously injected with 5 × 105 2D- or 3D-cultured U251 cells, followed by intraperitoneally injection of 200 μg anti-HLA-E on days 0 and 3. Tumor formation in each animal was recorded. The data represent two independent experiments. (C) Anti-NKG2A antibody abrogated the resistance of 3D-cultured U251 cells to NK cell cytotoxicity in vitro. The data are representative of three independent experiments. (D) Anti-NKG2A antibody could reverse the resistance of the 3D-cultured U251 cells to NK function in vivo. Nude mice that were depleted of NK cells and reconstituted with human NK cells (n = 15/group) were intravenously injected with 5 × 105 2D- or 3D-cultured U251 cells, followed by intraperitoneally injection of 200 μg anti-HLA-E on days 0 and 3. Tumor formation in each animal was recorded. The data represent two independent experiments.

counterparts, have significantly enhanced resistance to cytotoxic T cell-mediated immune response compared with those grown in 2D.6−8 Here, based on differentially expressed proteins, an immune-associated PPI network, which was composed of 13 proteins involved in antigen processing and presentation, was constructed by bioinformatics analysis. Within the network, HLA-A was shown to be an important molecule for T-cellmediated tumor immunity. Several studies showed that the down-regulation of HLA9,10 or TAA11,12 expression and the high production of lactic acid in 3D cells might play important roles in the resistance to T-cell-mediated immune response. Unexpectedly, another two major histocompatibility complex molecules, HLA-B and HLA-C, were up-regulated in 3D culture (Table 1), and these changes were confirmed by Western blot analysis (Figure S6). This finding indicated that there were other mechanisms for the enhanced resistance to Tcell-mediated lysis, in addition to the down-regulation of HLAA. Notably, in addition to HLA-A, the PSMA1, PSMA2, PSMB1, PSMC2, PSMD2, and PSMD3 proteins in the network are subunits of the proteasome and are down-regulated in 3D culture. The proteasome plays an important role in immune surveillance, as it degrades proteins that originate from invading

pathogens or tumor antigens; the resulting peptides are displayed by the major histocompatibility complex class I (MHC) proteins.31 These data demonstrated that tumor cells cultured in 3D had an impaired antigen presentation, which resulted in enhanced resistance to the CD8 T-cell-mediated immune response. Additionally, the down-regulation of proteasome function may be an underlying mechanism for gliomas immune escape. Whether 3D-cultured tumor cells on their own exhibit enhanced resistance to NK cell-mediated cytotoxicity compared with their 2D counterparts still remains unclear. In this study, HLA-E was identified as an up-regulated protein in 3D-cultured U251 cells and confirmed to render the 3D-cultured tumor cells resistant to the NK cell-mediated immune response. NKmediated killing of target cells depends on the balance of interactions from activating and inhibitory ligands expressed on the surface of target cells with their respective receptors expressed on the surface of NK cells.32 The balance of signals emanating from both activating and inhibitory receptors determines the outcome of NK cell cytolytic function.33 Gliomas highly express the classical MHC class I molecules (HLA-ABC) and nonclassical MHC class I molecules (HLA-E 2279

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research



and HLA-G) that act to protect them from NK lysis.34,35 Because HLA-E is a ligand of CD94/NKG2A that inhibits NK cell activation,24 as expected, we demonstrated here that the 3D-cultured tumor cells’ enhanced resistance to the NK cell cytotoxicity was mediated by the HLA-E/NKG2A interaction. However, why is the level of HLA-E expression on glioma cells significantly increased in 3D system? What’s the exact relationship between HLA-E expression and 3D architecture? These interesting questions are still unknown. Revealing the underlying mechanisms will provide a novel strategy for glioma clinical treatment in future. Malignant gliomas are deadly brain tumors with a median survival only 12 months. NK cells are very important for tumor surveillance and can kill gliomas efficiently. Using NK cells as therapeutic cells to treat patients with gliomas are very promising, and several preclinical experimental studies have been performed.34,36,37 The efficiency of therapeutic NK cells cytotoxicity to gliomas in vivo is critical for therapeutic outcome. However, the therapeutic NK cells are ordinarily amplified to the large-scale by using several tumor feeder cells including K562 cells in 2D model,37 which are clearly different from their in vivo equivalents. Indeed, either K562 cells or 2Dcultured primary glioma cells were much more sensitive to NK cell-mediated lysis than the 3D-cultured primary glioma cells. It is reasonable to assume that NK cells expanded in 3D model might be much better in immunotherapy of glioma than in traditional 2D model. In summary, we revealed for the first time that 3D-cultured tumor cells have enhanced resistance to NK cell-mediated lysis due to the up-regulation of HLA-E expression; in addition, the 3D culture model will have good application prospects for expanding therapeutic NK cells to treat gliomas.



ACKNOWLEDGMENTS This work was supported by grants from the National Natural Science Foundation of China (No. 81071037 and No. 81271395) and the special funding of Chongqing Key Laboratory.



REFERENCES

(1) Sutherland, R. M. Cell and environment interactions in tumor microregions: the multicell spheroid model. Science 1988, 240, 177− 184. (2) Cukierman, E; Pankov, R; Stevens, D. R.; Yamada, K. M. Taking cell-matrix adhesions to the third dimension. Science 2001, 294, 1708− 1712. (3) Lee, J; Cuddihy, M. J.; Kotov, N. A. Three-dimensional cell culture matrices: state of the art. Tissue Eng., Part B 2008, 14, 61−86. (4) Grun, B; Benjamin, E; Sinclair, J; et al. Three-dimensional in vitro cell biology models of ovarian and endometrial cancer. Cell Prolif. 2009, 42, 219−228. (5) Lee, J. M.; Mhawech-Fauceglia, P; Lee, N; et al. A threedimensional microenvironment alters protein expression and chemosensitivity of epithelial ovarian cancer cells in vitro. Lab Invest. 2013, 93, 528−542. (6) Ochsenbein, A. F.; Klenerman, P; Karrer, U; et al. Immune surveillance against a solid tumor fails because of immunological ignorance. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 2233−2238. (7) Ochsenbein, A. F.; Sierro, S; Odermatt, B; et al. Roles of tumour localization, second signals and cross priming in cytotoxic T-cell induction. Nature 2001, 411, 1058−1064. (8) Feder-Mengus, C; Ghosh, S; Reschner, A; Martin, I; Spagnoli, G. C. New dimensions in tumor immunology: what does 3D culture reveal? Trends Mol. Med. 2008, 14, 333−340. (9) Chang, C. C.; Ferrone, S. Immune selective pressure and HLA class I antigen defects in malignant lesions. Cancer Immunol. Immunother. 2007, 56, 227−236. (10) Marincola, F. M.; Jaffee, E. M.; Hicklin, D. J.; Ferrone, S. Escape of human solid tumors from T-cell recognition: molecular mechanisms and functional significance. Adv. Immunol. 2000, 74, 181−273. (11) Jager, E; Ringhoffer, M; Altmannsberger, M; et al. Immunoselection in vivo: independent loss of MHC class I and melanocyte differentiation antigen expression in metastatic melanoma. Int. J. Cancer 1997, 71, 142−147. (12) Khong, H. T.; Wang, Q. J.; Rosenberg, S. A. Identification of multiple antigens recognized by tumor-infiltrating lymphocytes from a single patient: tumor escape by antigen loss and loss of MHC expression. J. Immunother. 2004, 27, 184−190. (13) Dangles, V; Validire, P; Wertheimer, M; et al. Impact of human bladder cancer cell architecture on autologous T-lymphocyte activation. Int. J. Cancer. 2002, 98, 51−56. (14) Dangles-Marie, V; Richon, S; El-Behi, M; et al. A threedimensional tumor cell defect in activating autologous CTLs is associated with inefficient antigen presentation correlated with heat shock protein-70 down-regulation. Cancer Res. 2003, 63, 3682−3687. (15) Feder-Mengus, C; Ghosh, S; Weber, W. P.; et al. Multiple mechanisms underlie defective recognition of melanoma cells cultured in three-dimensional architectures by antigen-specific cytotoxic T lymphocytes. Br. J. Cancer 2007, 96, 1072−1082. (16) Ghosh, S; Rosenthal, R; Zajac, P; et al. Culture of melanoma cells in 3-dimensional architectures results in impaired immunorecognition by cytotoxic T lymphocytes specific for Melan-A/MART-1 tumor-associated antigen. Ann. Surg. 2005, 242, 851−857 , discussion 858. (17) Boersema, P. J.; Raijmakers, R; Lemeer, S; Mohammed, S; Heck, A. J. Multiplex peptide stable isotope dimethyl labeling for quantitative proteomics. Nat. Protoc. 2009, 4, 484−494. (18) Chen, C; Wu, D; Zhang, L; Zhao, Y; Guo, L. Comparative phosphoproteomics studies of macrophage response to bacterial virulence effectors. J. Proteomics 2012, 77, 251−261.

ASSOCIATED CONTENT

S Supporting Information *

The proliferation of U251 cells cultured in 2D or 3D conditions. The down-regulated proteins in the 3D-cultured U251 cells were analyzed using the BinGO plugin in the Cytoscape software. The up-regulated proteins in the 3Dcultured U251 cells were analyzed using the ClueGO +Cluepedia plugin in the Cytoscape software. The up-regulated proteins in the 3D-cultured U251 cells were analyzed using the BinGO plugin in the Cytoscape software. Glioma cells from primary tumor tissues cultured in 3D have increased HLA-E expression and enhanced resistance to NK cell-mediated cytotoxicity compared to those grown in 2D. Expression of P4HB, HLA-B and HLA-C in the 2D- and 3D-cultured U251 cells. Up-regulated and down-regulated differentially expressed proteins in 3D models. Details of the GO analysis. Details of the immune-associated PPI network. This material is available free of charge via the Internet at http://pubs.acs.org.



Article

AUTHOR INFORMATION

Corresponding Authors

*W.H.: Tel/Fax: +86-2365461677. E-mail: heweifeng7412@ aliyun.com. *J.G.: Tel/Fax +86-2989159139. E-mail: gujianwen5000@ gmail.com. Notes

The authors declare no competing financial interest. 2280

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281

Journal of Proteome Research

Article

(19) Zhang, L. K.; Chai, F; Li, H. Y.; Xiao, G; Guo, L. Identification of host proteins involved in Japanese encephalitis virus infection by quantitative proteomics analysis. J. Proteome Res. 2013, 12, 2666− 2678. (20) Bindea, G; Mlecnik, B; Hackl, H; et al. ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics 2009, 25, 1091−1093. (21) Franceschini, A; Szklarczyk, D; Frankild, S; et al. STRING v9.1: protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Res. 2013, 41, D808−815. (22) Walker, P. R.; Dietrich, P. Y. Immune escape of gliomas. Prog. Brain Res. 2001, 132, 685−698. (23) McCann, F. E.; Suhling, K; Carlin, L. M.; et al. Imaging immune surveillance by T cells and NK cells. Immunol. Rev. 2002, 189, 179− 192. (24) Braud, V. M.; Allan, D. S.; O’Callaghan, C. A.; et al. HLA-E binds to natural killer cell receptors CD94/NKG2A, B and C. Nature 1998, 391, 795−799. (25) Wischhusen, J; Friese, M. A.; Mittelbronn, M; Meyermann, R; Weller, M. HLA-E protects glioma cells from NKG2D-mediated immune responses in vitro: implications for immune escape in vivo. J. Neuropathol. Exp. Neurol. 2005, 64, 523−528. (26) Bello, L; Giussani, C; Carrabba, G; et al. Suppression of malignant glioma recurrence in a newly developed animal model by endogenous inhibitors. Clin. Cancer Res. 2002, 8, 3539−3548. (27) Santini, M. T.; Rainaldi, G; Romano, R; et al. MG-63 human osteosarcoma cells grown in monolayer and as three-dimensional tumor spheroids present a different metabolic profile: a (1)H NMR study. FEBS Lett. 2004, 557, 148−154. (28) Fischer, K; Hoffmann, P; Voelkl, S; et al. Inhibitory effect of tumor cell-derived lactic acid on human T cells. Blood 2007, 109, 3812−3819. (29) Fischbach, C; Chen, R; Matsumoto, T; et al. Engineering tumors with 3D scaffolds. Nat. Methods 2007, 4, 855−860. (30) Sun, S; Lee, D; Ho, A. S.; et al. Inhibition of prolyl 4hydroxylase, beta polypeptide (P4HB) attenuates Temozolomide resistance in malignant glioma via the endoplasmic reticulum stress response (ERSR) pathways. Neuro-Oncology 2013, 15, 562−577. (31) Ossendorp, F; Eggers, M; Neisig, A; et al. A single residue exchange within a viral CTL epitope alters proteasome-mediated degradation resulting in lack of antigen presentation. Immunity 1996, 5, 115−124. (32) Ogbomo, H; Cinatl, J, Jr.; Mody, C. H.; Forsyth, P. A. Immunotherapy in gliomas: limitations and potential of natural killer (NK) cell therapy. Trends Mol. Med. 2011, 17, 433−441. (33) Moretta, A; Locatelli, F; Moretta, L. Human NK cells: from HLA class I-specific killer Ig-like receptors to the therapy of acute leukemias. Immunol. Rev. 2008, 224, 58−69. (34) Castriconi, R; Daga, A; Dondero, A; et al. NK cells recognize and kill human glioblastoma cells with stem cell-like properties. J. Immunol. 2009, 182, 3530−3539. (35) Mittelbronn, M; Simon, P; Loffler, C; et al. Elevated HLA-E levels in human glioblastomas but not in grade I to III astrocytomas correlate with infiltrating CD8+ cells. J. Neuroimmunol. 2007, 189, 50− 58. (36) Alizadeh, D; Zhang, L; Brown, C. E.; Farrukh, O; Jensen, M. C.; Badie, B. Induction of anti-glioma natural killer cell response following multiple low-dose intracerebral CpG therapy. Clin. Cancer Res. 2010, 16, 3399−3408. (37) Ishikawa, E; Tsuboi, K; Saijo, K; et al. Autologous natural killer cell therapy for human recurrent malignant glioma. Anticancer Res. 2004, 24, 1861−1871.

2281

dx.doi.org/10.1021/pr500064m | J. Proteome Res. 2014, 13, 2272−2281