B7/CD28 Costimulation of T Cells Induces a Distinct Proteome Pattern*

Effective immune strategies for the eradication of human tumors require a detailed understanding of the interaction of tumor cells with the immune system, which might lead to an optimization of T cell responses. To understand the impact of B7-mediated costimulation on T cell activation comprehensive proteome analysis of B7-primed T cell populations were performed. Using this approach we identified different classes of proteins in T cells whose expression is either elevated or reduced upon B7-1- or B7-2-mediated CD28 costimulation. The altered proteins include regulators of the cell cycle and cell proliferation, signal transducers, components of the antigen processing machinery, transporters, cytoskeletal proteins, and metabolic enzymes. A number of differentially expressed proteins are further modified by phosphorylation. Our results provide novel insights into the complexity of the CD28 costimulatory pathway of T cells and will help to identify potential targets of therapeutic interventions for modulating anti-tumor T cell activation.

Effective T cell activation requires the engagement of two separate signals. The first is mediated by the interaction of the T cell receptor (TCR) 1 complex with MHC class I-peptide complexes. The second signal is dependent on the interaction of various cell surface receptors on T cells and their ligands typically expressed on antigen-presenting cells providing essential costimulatory signals. These costimulatory signals are pivotal in determining whether recognition of antigen by T cells leads to full T cell activation, T cell anergy, or apoptosis (1,2).
The best studied costimulatory pathway is represented by the CD28/CTLA4-B7 system (2,3) where B7-1 (CD80) and B7-2 (CD86) bind either to an activating (CD28) or to an inhibitory (CTLA4) receptor on T cells (4). A variety of studies demonstrate the significance of CD28 costimulation in the clonal expansion of naïve T cells, the regulation of long term survival of T cells, induction of cell cycle progression, and cytokine secretion (5). The elucidation of the CD28-mediated functions has facilitated the development of therapeutic strategies, particularly vaccination studies, that target this pathway. Most efforts to induce antigen-specific immunity by application of cancer vaccines have focused on the generation of tumor-reactive CD8 ϩ cytotoxic T cells (CTLs) (6). Whereas B7-negative tumor cells most often fail to induce an efficient immune response, B7-1 and/or B7-2 expression on tumor cells augments the anti-tumor response by inducing T cell proliferation, cytokine secretion, and an enhanced generation of tumor-specific CD8 ϩ CTLs (7-10). B7-transduced renal cell carcinoma cells are an example of the strategy to stimulate tumor-infiltrating T cells to promote an efficient antitumor response (11,12).
The efficacy of an antitumor reaction is usually monitored by in vitro determination of the number of antigen-specific T cells that secrete cytokines in response to antigen challenge. The monitoring can be achieved by enzyme-linked immunospot techniques (13,14), flow cytometry (15), multiplex-based immunoassays (16) or RT-PCR (17). Such assays, however, do not provide any information regarding the various signaling and metabolic pathways activated upon CD28 costimulation of the T cell. Particularly only limited information on the genetic expression programs linked to CD28 costimulation exists (18 -20), whereas to our best knowledge no study to date has assessed the protein expression patterns of both prestimulated and naïve T cells upon CD28 costimulation. To address the complexity of the T cell response directed against a defined, B7-transduced tumor cell vaccine, here we applied proteome-based technologies and identified a variety of differentially expressed proteins in T cells following B7-1 and B7-2 costimulation, respectively. Some of these proteins additionally exhibit altered phosphorylation patterns. The studies provide novel insights into the biology of the B7/CD28 signal transduction pathway as well as early steps in the induction of an antitumor T cell response.
Isolation of Phosphoproteins-Phosphoproteins were isolated from T cells using the Phosphoprotein Purification kit (Qiagen, Hilden, Germany) as recommended by the manufacturer. This kit is designed for the specific purification of all phosphorylated proteins from complex cell lysates. Briefly 2.5 ϫ 10 7 cells were lysed in lysis buffer containing detergent, nucleases, and protease inhibitors. Lysates were cleared by centrifugation (90 min at 13,000 rpm at 15°C), and the supernatants were loaded onto phosphoprotein-specific columns for affinity chromatography. The eluted proteins were concentrated using Nanosep ultrafiltration columns (Pall, Dreieich, Germany).
Proteome Analyses-Protein lysates (300 g of total protein lysates and 100 g of phosphoprotein fraction, respectively) were loaded onto IPG strips (Immobiline dry strips, pH 3-10, non-linear, Amersham Biosciences). Isoelectric focusing, second dimension SDS-PAGE separation, silver staining, and gel documentation was performed as described previously (23). The gels were analyzed using the Proteomweaver software package (Version 2.0, Definiens, Munich, Germany). Proteins were subjected to further analyses when the given protein spot was found to be altered in at least three of five gels and the differences in the expression levels was at least 2-fold. Furthermore spots were analyzed when a different position compared with the corresponding spot from T cells coincubated with MZ1257RC cells was observed. Protein spots of interest were excised from the respective gel, digested with the endoproteinase Lys-C (Roche Diagnostics), and subjected to mass analysis using positive reflector mode with a deflection cutoff range of m/z 800 on a 4700 Proteomics Analyzer (Applied Biosystems, Framingham, MA). Subsequently high energy MALDI-TOF/TOF CID spectra were recorded on selected ions from the same sample spot. For the unambiguous identification of the proteins, tandem mass spectrometric analyses were performed on one or two peptides of the corresponding peptide mass fingerprints and annotated by MASCOT using the National Center for Biotechnology Information (NCBI) protein database (www.matrixscience.com). For protein classification into functional clusters the website ncbi. nlm.nih.gov/entrez/querey.fcgi?dbϭPubMed was used.
RT-PCR and Western Blot Analyses-Total RNA from T cells was isolated using the RNeasy kit (Qiagen) according to the manufacturer's instructions, and the RNA was subjected to RT-PCR analyses (Titan kit, Roche Applied Science) using the sets of gene-specific primer oligonucleotides listed in Table I.

T Cells Express an Altered Pattern of Proteins upon B7
Costimulation-We wanted to identify T cell proteins whose expression is altered upon contact of activated T cells with tumor cells that express B7-1 and B7-2, respectively. T cells obtained from two unrelated HLA-A2 ϩ volunteers were preactivated by incubation with the anti-CD3 monoclonal anti-

Proteomic Approach of B7-mediated Stimulation of T Cells
body OKT3 and IL-2. T cells were coincubated with paraformaldehyde-fixed wild-type (wt) MZ1257RC cells, B7-1-and B7-2-modified MZ1257RC cells, or MZ1257RC cells transfected with "empty" vector DNA. Cells of the B7-transduced MZ1257RC clones express about 1.25 ϫ 10 6 B7-1 and B7-2 molecules/cell, respectively (22). Upon coincubation with either B7-1 ϩ or B7-2 ϩ MZ1257RC cells T cells increased proliferation and cytokine secretion (interferon-␥, granulocyte macrophage colony-stimulating factor, tumor necrosis factor-␣, and IL-10), whereas coincubation with wt and mocktransfected MZ1257RC cells did not (data not shown). T cells were lysed and subjected to two-dimensional gel electrophoresis (2-DE) as described under "Experimental Procedures" after 48-h costimulation with B7-transduced and con-trol cells. The data were obtained from two independent sets of proteome analyses and are based on five gels per sample. Representative 2-D gel spot patterns are shown in Fig. 1. Routinely about 1,500 protein spots per gel were detected. In samples from T cells costimulated via B7-1 or B7-2, respectively, 35 protein spots with quantitatively altered expression patterns were found from which 15 spots are increased while 20 spots are decreased in intensity. It is noteworthy that seven proteins are altered in expression upon B7-1 and 13 proteins are altered in expression upon B7-2 costimulation, whereas 15 of the differentially expressed proteins are similarly altered in expression upon both B7-1 and B7-2 costimulation. Additionally 10 proteins show a modified expression upon both B7-1 and B7-2 costimulation as characterized by a  Table II. The differentially expressed proteins include metabolic enzymes, components of the antigen processing machinery, transporters, and cytoskeletal proteins as well as cell cycle and cell proliferation regulators. In addition, proteins with yet unknown functions in T cells were identified.

FIG. 2. Representative mass spectrometry analysis.
A protein spot of interest was excised from a gel, processed as described under "Experimental Procedures," and digested with endoproteinase Lys-C. The resulting peptide fragments were subsequently analyzed by mass spectrometry. The resulting fingerprint comprised several fragments in the range of 899 -4013 Da is shown in A. As indicated the x axis represents the mass to charge ratios (m/z), whereas the y axis represents the given peak intensities. The interpretation of the fragment pattern led to the identification of the glia maturation factor-␥. To verify the peptide mass fingerprinting of the glia maturation factor-␥ the fragment with the molecular mass of 2194 Da was further characterized by MS/MS analysis. Its fragmentation led to the identification of the protein sequence VFEIRTTDDLTEAWLQEK covering the stretch of amino acids 120 -137 within the target protein (B).

TABLE II T cell proteins with altered expression upon B7-1-and B7-2-mediated costimulation
T cells were coincubated for 24 h with paraformaldehyde-fixed MZ1257RC cells and MZ1257RC cells transfected with B7-1 and B7-2, respectively. Protein lysates from T cells were subjected to 2-D gel electrophoresis as described under ЈExperimental Procedures.Ј Protein expression patterns of T cells coincubated with B7-1 ϩ and B7-2 ϩ MZ1257RC cells were compared with those of T cells coincubated with MZ1257RC cells lacking B7 expression. The identities of the individual protein spots were determined by mass spectrometry. N.A., expression level not altered compared to stimulation with wt_MZ1257RC cells; Ͼ, expression level increased; Ͻ, expression level decreased; phospho, protein is phosphorylated; modified, protein spot is at a different position compared to the corresponding spot from T cells coincubated with MZ1257RC cells. ER, endoplasmic reticulum; ERK, extracellular signal-regulated kinase; PKC, protein kinase C; GDI, GDP dissociation inhibitor. Altered Phosphorylation Pattern of Differentially Expressed Proteins in T Cells upon B7 Costimulation-We monitored alterations in the phosphorylation patterns of proteins in T cells upon B7-1 and B7-2-mediated costimulation after 15 min, 60 min, and 24 h. Upon 60 min of stimulation 2-D gel analyses of phosphoprotein subproteome revealed 288 phosphorylated protein spots (Fig. 3A). From these, 248 protein spots are phosphorylated upon both B7-1-and B7-2-mediated costimulation. In contrast, five proteins are only phosphorylated upon B7-1 costimulation alone, and 35 proteins are phosphorylated upon B7-2 stimulation of T cells (Fig. 3B). From these differentially phosphorylated protein spots, 24 proteins could be identified by mass spectroscopy (Table II). Most of these phosphorylated proteins belong to components of the antigen processing machinery in particular to subunits of the proteasome complex.
Validation of the Differentially Expressed Proteins-RT-PCR and Western blot analyses were used to validate a selected number of differentially expressed proteins in T cells upon B7 costimulation. Some of the validation results are shown ex-   Fig. 4. For all proteins analyzed, the B7 costimulation-associated expression alterations detected by proteomics were confirmed by Western blot and/or RT-PCR. For example, the PCNA was confirmed by both RT-PCR and Western blot analysis to be increased in expression upon both B7-1 and B7-2 costimulation. The expression of stathmin, a ubiquitously expressed, highly conserved cytoplasmic protein (24,25), was increased upon B7-1 and B7-2 costimulation by 2-DE analysis, which was confirmed by RT-PCR and Western blot analyses (Fig. 4). Phosphoproteome analysis revealed increased stathmin phosphorylation upon B7 costimulation ( Table II). Loss of expression of the platelet-derived growth factor (PDGF)-binding protein in T cells upon stimulation with B7 was identified by 2-DE analysis and confirmed by RT-PCR (Fig. 4). Nudix hydrolase-5 (nudt5), which catalyzes the hydrolysis of ADP-ribose and ADP-mannose (26), is significantly down-regulated in expression upon costimulation with B7-1 but not with B7-2. RT-PCR analysis confirmed reduced amounts of nudt5 mRNA in B7-1 but not in B7-2 costimulated T cells (Fig. 4). DISCUSSION T cell activation by costimulatory pathways is a key issue in T cell immunology and in T cell-based immunotherapy. The detailed knowledge of their signal transduction pathways has a significant impact on the design of therapies for autoimmune diseases, transplantation, and tumors. T cells stimulated by antigen receptors, cytokines, or chemokines trigger an independent array of signaling molecules including the activation of different kinases as well as other molecules involved in cellular metabolism, cell growth and maintenance, differentiation, and other cellular processes. Some novel components of the signaling network have been identified recently by transcriptome and proteome analysis of activated CD4 ϩ and CD8 ϩ T lymphocytes, undifferentiated T cells, and antigen receptor-activated T cells as well as Jurkat cells (18 -20, 27-31). In contrast, only little information exists on the effects of CD28 costimulation on the global gene transcription program of T cells (18 -20). Using cDNA microarray analysis of T cells triggered via TCR and/or costimulatory receptors, a large series of differentially expressed mRNAs has been identified (20), whereas the protein expression profile of T cells under CD28 costimulatory conditions has not yet been investigated. Therefore, the CD28 costimulation-mediated alterations in the protein expression pattern and their contributions to T cell stimulation are so far poorly understood. Because T cell activation involves changes in the signal transduction that are often associated with posttranscriptional modifications such as phosphorylation, both the global proteome and phosphoproteome analyses represent the first step toward elucidating the mechanisms by which B7-1 and B7-2 costimulation alters the antitumor response. Due to the high number of T cells required for proteome analysis and the few changes found in the mRNA expression pattern following stimulation of naïve T cells (20), the use of naïve T cells in this approach is experimentally challenging. Therefore, here we focused on the analysis of prestimulated T cells upon coincubation with wt, mock-transfected, and B7-1-or B7-2-transfected MZ1257RC cells using proteome-based technologies to identify proteins with altered protein expression patterns under these conditions. However, it is noteworthy that this state of the art analysis still has some practical limitations mainly based on the scarce material available for proteomics. The analyses indicate that upon incubation with B7-1 ϩ and B7-2 ϩ MZ1257RC cells CD28 costimulation of T cells induces distinct alterations in the global protein expression as well as in the protein phosphorylation patterns. 45 proteins were differentially expressed upon B7 costimulation, and 288 proteins exhibit an altered phosphorylation pattern (Table II). In this study, we found 15 proteins that are up-regulated and 20 proteins that are down-regulated upon B7 costimulation indicating a distinct translational program associated with CD28 costimulation of T cells. Although the majority of the differentially expressed proteins are similarly regulated upon B7-1 and B7-2 stimulation, certain proteins are exclusively altered in response to either the B7-1-(seven proteins identified) or B7-2 (13 proteins identified)-mediated CD28 costimulation, also further underlining the distinct function of both molecules in the costimulation process (32,33). In addition, phosphoproteome analyses revealed a distinct and time-dependent alteration in the protein phosphorylation pattern of T cells stimulated with wild-type or B7-transduced MZ1257RC cells. These results suggest that changes in the phosphorylation of a broad set of proteins are required for efficient T cell activation. The majority of proteins, however, are phosphorylated in a similar fashion 60 min after onset of B7-1 and B7-2 costimulation. However, in this context it is even noteworthy that the paraformaldehyde-fixed tumor cells may cause an underestimation of the effect of B7-2 because B7-2 but not B7-1 is affected by paraformaldehyde fixation (34).
The proteins differentially expressed and/or phosphorylated upon CD28 costimulation of T cells are involved in controlling cell proliferation, apoptosis, antigen processing, transport, signal transduction, cellular metabolism including various energy pathways, protein degradation, cytoskeleton formation, and the immunological synapse (Table II). In addition, proteins with yet unknown functions were altered in their expression under these conditions. Their identification might give further insights into the B7/CD28 signal transduction pathway. In comparison to cDNA microarray data, there exists an overlap with our proteome data regarding the differentially expressed gene/protein families. However, the changes identified on a single gene/protein are mainly not overlapping. This might be due to the limited number of differentially expressed proteins analyzed as well as to the often observed posttranscriptional modifications that appear to be important for T cell signal transduction.
Because T cell activation is known to be closely associated with enhanced proliferation and consequently with increased PCNA levels (35), the detection of increased amounts of PCNA protein in T cells upon stimulation with B7-1-and B7-2-transduced MZ1257RC cells by proteome analysis served as an internal standard and, on the other hand, indicates the proof of principle for the experimental procedure. Regarding proteins involved in metabolism, our data are in accordance with Frauwirth et al. (36) who demonstrated that CD28 costimulation increases the glucose metabolism allowing T cells to meet their energetic and biosynthetic needs associated with a sustained response.
To validate some of the differentially expressed proteins, e.g. PCNA, stathmin, nudix hydrolase-5, and PDGF-binding protein, we applied both Western blot and RT-PCR analyses. Stathmin was found to be one of the up-regulated proteins in T cells upon both B7-1-and B7-2-mediated CD28 costimulation. Stathmin integrates a series of intracellular signals (24,25). Its expression and phosphorylation are regulated by extracellular signals mediating proliferation and/or differentiation. In line with our results, a distinct CD28 costimulation-dependent pattern of stathmin phosphorylation has been reportedinthecontextofTcellactivation (37).Ca 2ϩ /calmodulindependent kinase II is associated with costimulatory signals of T lymphocyte activation and phosphorylation of stathmin on Ser-16 (37). In contrast to stathmin, expression of the PDGF-binding protein is decreased upon CD28 costimulation. PDGF-binding protein was found to be differentially expressed upon B7/CD28 costimulation. It is a physiological inducer of the nuclear factor of activated T cells (NFAT) that is substantially involved in Ca 2ϩ -dependent gene transcription processes in immune cells (38). Furthermore the CD28 pathway is important for IL-4 production, which involves NFAT members. The PDGF-binding protein enhances the PDGF-Amediated stimulation of cell proliferation but inhibits the mitogenic effect of PDGF-B (39,40). Thus, decreased PDGFbinding protein expression upon B7-mediated T cell activation may be linked to increased T cell proliferation by withdrawal of the inhibitory activity toward PDGF-B.
Nudix hydrolase-5 is found to be down-regulated in T cells after CD28 costimulation exclusively via B7-1 but not via B7-2. Proteins of the superfamily of nudix hydrolases catalyze the hydrolysis of a panel of nucleoside-diphosphate derivatives and are therefore involved in the cellular nucleotide metabolism. To our best knowledge nudix hydrolase has never been described in the context of T lymphocytes. Cyclin D1 (41)(42)(43)(44) is in the center of a network involved in cell cycle control, interacting directly with PCNA and GST Pi (45)(46)(47). PCNA, the RNA-binding subunit (48 -51), and the Ras family protein Ran (52)(53)(54)(55) interact with the androgen receptor, modulating its signal transduction activities (56 -60). The Rho GDP dissociation inhibitor (61,62), the Raf kinase inhibitor (63-66), prefoldin-5 interacting with c-Myc (67,68), the glia maturation factor (67)(68)(69), and stathmin are linked to the Ras/Raf/mitogen-activated protein kinase pathway involved in c-Myc acti-vation, suggesting that these proteins are involved in the synapse formation. However, it has been noted that no cytokines have been identified by this approach. This might be due to the low intracellular levels and might be enhanced by brefeldin A treatment of T cells.
The functional consequences of this distinct protein expression pattern need further investigation. The expression pattern, however, clearly points to shared but also distinct activities of B7-1-and B7-2-mediated CD28 costimulation as reported earlier (70 -72). The differentially expressed proteins identified to be altered in expression and/or phosphorylation upon B7-1-and B7-2-mediated stimulation suggest their involvement in T cell activation in general, whereas the differentially expressed proteins point to differences in the B7-1 and B7-2 costimulatory pathway. Interestingly the majority of posttranslational modifications are the same upon B7-1 and B7-2 costimulation.
Although most of the proteins associated with B7 costimulation have been found in T lymphocytes, there is only limited information available on the network of the potential protein/ protein interactions and their regulations. A putative model illustrating the signaling pathways that describe the costimulatory pathways should include the alterations in protein expression and phosphorylation as described here. The results of the proteome-based analysis enable a more solid interpretation of the alterations in protein expression during costimulation-mediated T cell activation. Despite this overall screen, our analysis identified a distinct set of proteins with potential significance in B7-mediated costimulation. It is noteworthy that the protein informations obtained by this strategy may also have some drawbacks because the protein expression pattern may additionally differ in particular T cell repertoires and the specific TCR/MHC engagement. The experimental approach described here and the knowledge drawn from the survey, however, may also contribute to identify new therapeutic targets within the CD28 costimulatory pathway and to develop therapeutic agents capable to manipulate T cell responses.