CTLA-4 is constitutively expressed on tumor cells and can trigger apoptosis upon ligand interaction



CTLA-4 (CD152) is a cell surface receptor that behaves as a negative regulator of the proliferation and the effector function of T cells. We have previously shown that CTLA-4 is also expressed on neoplastic lymphoid and myeloid cells, and it can be targeted to induce apoptosis. In our study, we have extended our analysis and have discovered that surface expression of CTLA-4 is detectable by flow cytometry on 30 of 34 (88%) cell lines derived from a variety of human malignant solid tumors including carcinoma, melanoma, neuroblastoma, rhabdomyosarcoma and osteosarcoma (but not in primary osteoblast-like cultures). However, by reverse transcriptase-PCR, CTLA-4 expression was detected in all cell lines. We have also found, by immunohistochemistry, cytoplasmic and surface expression of CTLA-4 in the tumor cells of all 6 osteosarcoma specimens examined and in the tumour cells of all 5 cases (but only weakly or no positivity at all in neighbouring nontumor cells) of ductal breast carcinomas. Treatment of cells from CTLA-4-expressing tumor lines with recombinant forms of the CTLA-4-ligands CD80 and CD86 induced apoptosis associated with sequential activation of caspase-8 and caspase-3. The level of apoptosis was reduced by soluble CTLA-4 and by anti-CTLA-4 scFvs antibodies. The novel finding that CTLA-4 molecule is expressed and functional on human tumor cells opens up the possibility of antitumor therapeutic intervention based on targeting this molecule. © 2005 Wiley-Liss, Inc.

Cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4, CD152) is a homodimeric glycoprotein belonging to the human Ig gene superfamily originally described on the surface of murine and human activated T cells.1 The vast majority of in vitro and in vivo studies on CTLA-4 support its negative role on T-cell activation contributing to the physiologic termination of the immune response.2, 3 CTLA-4 inhibitory function occurs upon interaction with its ligands, CD80 (B7.1) and CD86 (B7.2), expressed on antigen-presenting cells (APCs), resulting in inhibition of IL-2, IFN-γ, IL-4 cytokines production, IL-2 receptor expression and cell cycle progression.4, 5 Several mechanisms of CTLA-4 function have been proposed including ligand competition with the positive T-cell costimulatory CD28 molecule,6 interference of TCR signalling7 and inhibition of cyclin D3 and cyclin-dependent kinases (cdk4/cdk6) production.8 A possible function of CTLA-4 in the regulatory role of suppressor CD4+CD25+ T cells has generated widespread interest indicating another mechanism by which CTLA-4 might downregulate immune responses9 and also promote peripheral tolerance.10

We and others have previously shown that CTLA-4 is also expressed on nonlymphoid cells including placental fibroblasts,11 cultured muscle cells,12 monocytes13 and a variety of leukemia cells,14 suggesting that this molecule might be involved in controlling functions other than the widely described T-cell response inactivation. For example, maintenance of pregnancy,15 autoimmune myositis development12 and regulation of monocyte function13 have been proposed. Although we have previously demonstrated that CTLA-4 is expressed on neoplastic cells of hematopoietic origin, its expression on solid tumor-derived cells has not yet been examined.

Tumor cells have developed multiple mechanisms to evade the immune system, including immunosuppressive properties and poor immunogenicity. Tumor cells can secrete soluble immunosuppressive factors16, 17 or promote the generation of “suppressor” T cells.18 Conversely, most tumor cells are poor antigen-presenting cells (APCs) due to the low (or absent) expression of HLA molecules19 or immunostimulatory cytokines20 as well as costimulatory molecules21 whose interaction with specific counter receptors on T cells is essential to efficiently elicit T-cell activation.22 To increase antitumor immunity, new approaches have been developed based on combination of CTLA-4 function blockade23, 24 with enhancement of tumor APCs function.25 To this regard, expression of costimulatory molecules has been induced on tumor cells by gene transfer, resulting in enhanced antitumor response and tumor rejection not only in animal models but also in a number of clinical trials.26, 27

Since CTLA-4 appears to have a physiologic role in inducing downregulation in responding T cells, it might be important to determine a possible expression in tumour cells and to investigate its role in initiating and maintaining the neoplastic process. In this perspective, we have analyzed the expression of CTLA-4 in a panel of human tumour cell lines as well as in tissues obtained from osteosarcomas and breast ductal carcinomas. The analysis on the cell panel was performed by flow cytometry using 2 anti-CTLA-4 human scFv antibodies previously described14 and by RT-PCR analysis using CTLA-4 full-length and extracellular primers.

We demonstrated that CTLA-4 is constitutively expressed on tumor cell lines at various degrees of intensity and can trigger apoptosis of CTLA-4-expressing tumor cells after interaction with soluble CD80 or CD86 recombinant ligands. The apoptosis induction is through a caspase-8-dependent mechanism. Moreover, CTLA-4 expression was detected in osteosarcoma as well as in breast tumor tissues by immunohistochemistry, whereas no or weak CTLA-4 staining was observed in breast nonmalignant tissues adjacent to tumors.


BR, breast; CTLA-4, cytotoxic T-lymphocyte-associated antigen-4; FITC, fluorescein isothiocyanate; HSSCs, human stromal stem cells; IHC, immunohistochemistry; OS, osteosarcoma; PBMCs, peripheral blood mononuclear cells; PE, phycoerythrin; PHA, phytohemagglutinin; PI, propidium iodide; PMA, phorbol ester; RT-PCR, reverse transcriptase-polymerase chain reaction; scFvs, single-chain antibody fragments.

Material and methods

Monoclonal antibodies and recombinant fusion proteins

Two recombinant anti-CTLA-4 monoclonal antibody fragments, namely scFv#67 and #83, were obtained by selecting human scFv phage libraries with purified CTLA-4-Ig fusion protein as described previously.14 They were used either conjugated to fluorescein isothiocyanate (FITC) for direct immunofluorescence staining or unconjugated for immunohistochemical staining and apoptosis inhibition experiments. Other commercially available monoclonal antibodies (mAbs) were the following: anti-CTLA-4 mAb either FITC-conjugated (50.18.21 clone; Cymbus Biotechnology, Chandlers Ford, UK) or unconjugated (BN13 clone; BD Pharmingen, Milano, Italy), FITC-conjugated anti-CD80 mAb (MAB104 clone; Coulter Immunotech, Birmingham, UK), FITC-conjugated anti-CD28 mAb (CD28.2 clone; Coulter Immunotech) and phycoerythrin (PE)-conjugated anti-CD86 mAb (HA5.2B7 clone; Coulter Immunotech). Anti-HLA class I mAb W6/32 (ATCC HB-95) was used as primary antibody in indirect immunofluorescence staining. The following antibodies to caspases were used: rabbit polyclonal antibody to caspase-9 (Sigma-Aldrich, St. Louis, MO); mouse mAb to caspase-8 (clone 1C12; Cell Signaling Technology, Beverly, MA) and rabbit polyclonals to either caspase-10 or caspase-3 (Cell Signaling Technology). Anti-β-tubulin antibody (Sigma) was used for equalizing gel loading.

The human recombinant proteins (hereafter r-proteins) r-CD80, r-CD86, r-CD28 and r-CTLA-4 were prepared as full-length fusion proteins according to previously described procedures.28 Briefly, cDNA fragments for these molecules, cloned into pTEX7 in our previous study,28 were subcloned into pTEX2-eHis, whose multicloning site was slightly different from pTEX7 and carries nucleotide sequences for 6 histidine. By these constructs, r-CD80, r-CD86, r-CD28 and r-CTLA-4 were produced as fusion proteins with β-galactosidase at the N-terminal and with a histidine tag at the C-terminal in E. coli (POP2136). Production of the fusion proteins was induced by quick shift of the culturing temperature of the E. coli from 37°C to 42°C. Then the harvested cells were lysed and the recombinants were affinity-purified by Ni+-NTA columns (HiTrap Chelating HP, Amersham Bioscience, Piscataway, NJ) according to the manufacturer's guidelines.

Cells and culture conditions

A panel of tumor cell lines was selected according to histologic origin and tested for expression of CTLA-4 in addition to the CD80/86-CD28 costimulatory molecules. Some cell lines were obtained from American Tissue Culture Collection (ATCC, Rockville, MD) including 4 colorectal adenocarcinoma cell lines, HCT-8 (ATCC CCL-244), HT-29 (ATCC HTB-38), COLO 205 (ATCC CCL-222) and CACO-2 (ATCC HTB-37); 4 breast carcinoma cell lines, MCF-7 (ATCC HTB-22), MDA-MB-231 (ATCC HTB-26), T-47D (ATCC HTB-133), BT-20 (ATCC HTB-19); 3 lung carcinoma cell lines, CALU-1 (ATCC HTB-54), CALU-6 (ATCC HTB-56), A549 (ATCC CCL-185); 1 ovarian carcinoma cell line, SKOV-3 (ATCC HTB-77); and 1 uterine carcinoma cell line, C33A (ATCC HTB-31). The ovarian carcinoma cell line A2780 was provided by Dr. S. Canevari (Istituto Nazionale Tumori, Milano, Italy). Five neuroblastoma cell lines, NB100, SJNKP, CHP212, SY5Y, SKNBE-2C; 3 renal carcinoma cell lines, SKRC-10, SKRC-52, SKRC-59; 2 uterine carcinoma cell lines; TG, HELA; 1 bladder carcinoma cell line, T24; and 2 rabdomyosarcoma cell lines, RD/18, TE671, were obtained from local laboratories at the Department of Experimental Pathology, University of Bologna, Italy. Four osteosarcoma cell lines, HOS, MG-63, U2-OS, SaOS-2, were obtained from the Istituti Ortopedici Rizzoli (Bologna, Italy), and 3 melanoma cell lines, MEL-1, ALO-39, F0-1, were provided by Dr. M. Maio (Department of Medical Oncology, Centro di Riferimento Oncologico, IRCCS, Aviano, Italy). The nontumorigenic human breast epithelial cell lines MCF10A and HC11 were a kind gift from Dr. M. De Bortoli (Institute for Cancer Research and Treatment, Candiolo, Torino, Italy) and the B-lymphoblastoid cell line SWEIG was derived from the 12th International Histocompatibility Workshop cell panel deposited at the European Collection for Biomedical Research (ECBR) cell bank (Genova branch, www.biotech.ist.unige.it/ecbr/ecbrdescription.html).

All tumor cell lines were maintained in monolayer cultures using complete medium consisting of RPMI 1640 (Biochrom KG, Berlin, Germany) supplemented with 10% heat-inactivated fetal calf serum (FCS; Biochrom KG), antibiotics, 2 mM L-glutamine (Biochrom KG), at 37°C in a humidified 5% CO2 atmosphere and subcultured every 3–7 days. The confluent cells were harvested after treatment with trypsin/EDTA (Biochrom KG) solution and diluted with medium for further assays.

Peripheral blood mononuclear cells (PBMCs) were isolated, after informed consent, from the buffy-coats of healthy donors by density gradient centrifugation over Ficoll/Biocoll (Biochrom KG). PBMCs were activated by culturing them in complete RPMI 1640 medium in the presence of phorbol ester (PMA) (Sigma) at 5 ng/ml and phytohemagglutinin (PHA) (Life Technologies, Milano, Italy) at a final concentration of 2 μg/ml for 48 hr at 37°C.

Human stromal stem cells (HSSCs) were collected under general anesthesia, after informed consent and under a protocol approved by the Istituti Ortopedici Rizzoli review board. A 2 ml sample of bone marrow was aspirated into a 20 ml plastic syringe (containing 1 ml of saline with 1,000 units of heparin) from the posterior iliac crest of 10 adult donors (age 10–33 years, mean 18.3 ± 6.9). The marrow was collected by inserting the needle in more than a single site. Nucleated cells were isolated with a density gradient and resuspended in α-modified essential medium (α-MEM; Sigma Chemical, St. Louis, MO) containing 20% FCS (Euroclone, Wetherby, UK), 100 units/ml penicillin (Euroclone), 100 mg/ml streptomycin and 2 mM-glutamine (Euroclone). All the nucleated cells were plated in a 25 cm2 culture flask and incubated in a humidified atmosphere at 37°C with 5% CO2. Nonadherent cells were discarded after 1 week, and adherent cells were cultured for further expansion. When cultured dishes became near confluent, cells were detached by mild trypsinization and reseeded onto new plates at 1/3 density for continued passage. Medium was changed every 3 to 4 days. Cell viability was assessed for each experiment performed by Trypan blue exclusion and was always more than 98%.

To promote ostoblastic differentiation, a proportion of cultures were incubated in complete medium additionally supplemented with 100 nM dexamethasone (Dex, Sigma-Aldrich), 2 mM β-glycerophosphate (Sigma Chemical) and 0.05 mM ascorbic acid (Sigma Chemical).

Tissue samples

A total of 6 formalin-fixed, paraffin-embedded osteosarcoma (OS) tumor specimens were selected for our study from the Istituti Ortopedici Rizzoli (Bologna, Italy) files, in accordance with the informed consent and local ethics committee approval. All the specimens were biopsies from previously untreated patients. The histology of the primary tumors was reviewed by pathologists experienced in bone tumors. All of the tumor samples were grade 4 osteoblastic osteosarcomas.29 Tumor tissue samples from 5 invasive breast (BR) ductal carcinomas, formalin-fixed and paraffin-embedded, and their nonmalignant tissue counterpart adjacent to tumor were taken at the time of surgery at the National Institute for Cancer Research in Genova (Italy) in accordance with the informed consent and local ethics committee approval. Four of the 5 neoplastic tissues were moderately differentiated carcinomas (G2 grade), and 1 tissue was a well-differentiated carcinoma (G1 grade).

Immunofluorescence and flow cytometry

A direct immunofluorescence was performed for analyzing surface and cytoplasmic expression of CTLA-4 in tumor cells or PBMCs. Briefly, a pellet of 4 × 105 cells, without or with fixation in 2% paraformaldehyde followed by permeabilization with 0.5% saponin, was incubated for 30 min at room temperature (RT) with FITC anti-CTLA-4 scFvs, #67 or #83, FITC anti-CTLA-4 BN13 mAb, or with FITC-anti-BSA scFv #26 and mouse IgG1 mAb as negative controls. A direct immunofluorescence was also performed to analyze surface expression of CD80 and CD86 CTLA-4 ligands in addition to CD28. Indirect immunofluorescence was performed by incubating cells for 30 min at 4°C with anti-HLA-class I W6/32 mAb as primary antibody followed by incubation for a further 30 min at 4°C with an FITC-conjugated goat-anti-mouse IgG (Perbio Science, Tattenhall, UK) as secondary antibody.

The fluorescence intensity was measured on a Coulter flow cytometer (EPICS Elite Coulter Electronics, Hialeah, FL). At least 15,000 cells/sample were counted.

cDNA synthesis and PCR

Total cellular RNA was used to synthesize cDNA by oligo(dT) priming with a Retrotranscript kit (Ambion, Austin, TX) as previously described.30 PCR reactions were carried out in 50 μl volume, using 1/10 of the reverse transcriptase (RT) mixture (500 ng RNA). Specific amplification of CTLA-4 full-length transcript was performed on each cDNA samples using the set of primers previously described.31 PCR reaction was run after the denaturation at 94°C for 1 min, annealing temperature (AT) at 60°C for 1 min, elongation at 72°C for 1 min and a total of 35 cycles. The reaction was initially hot started (94°C for 3 min) and terminally extended at 72°C for 5 min. As internal control, G3PDH gene amplification (G3PDH forward primer: 5′-AACGGATTTGGTCGTATTGGGC-3′; G3PDH reverse primer: 5′-AGGGATGATGTTCTGGAGAGCC-3′) was carried out for each cDNA sample using 60°C of AT and the same reaction conditions as for CTLA-4 amplification. The obtained PCR products were analysed by electrophoresis on a 2% agarose gel. The size and specificity of CTLA-4 PCR products were confirmed after direct sequencing analysis in both directions using an ABI-PRISM 377 Perkin-Elmer DNA Sequencer.

A second round of nested PCR was performed amplifying 1 μl of CTLA-4 full-length first PCR products with CTLA-4 extracellular domain primers as inner primers.32 Twenty-five more cycles were carried on at 58°C AT, as previously described.14 Negative as well as positive results were confirmed by repeating the assay with a second aliquot of each original total RNA sample. Reproducibility was almost 100% in negative cases and >90% in positive cases. Adequate precautions to prevent cross-contamination and negative control reactions were performed routinely. The obtained nested-PCR products were analyzed and sequenced as described above.


Immunohistochemical (IHC) staining was performed using the biotin-streptavidin complex/HRP method (DAKO ARK, DAKO, Milano, Italy) according to the manufacturer's instructions. Briefly, 5 μm-thick, formalin-fixed, paraffin-embedded tissue sections were deparaffinized, rehydrated and treated with 0.3% H2O2 in PBS for 30 min at RT to block endogenous peroxidase activity. OS and BR tissue sections were equally processed except that OS sections underwent decalcification with the solutions 910 CC H2O2, 50 CC formic acid 99%, 40 CC hydrogen chloride 37% and antigen unmasking by treatment with type I collagenase for 15′ at RT.

After rinsing in PBS, pH 7.4, sections were incubated in 4% low-fat milk for 1 hr at 4°C to reduce nonspecific binding. Anti-CTLA-4 scFv #83 was used as primary antibody at a final concentration of 10 μg/ml, mixed in solution with an equal amount of the mAb 9E10, recognizing the c-myc peptide tag linked to the scFv #83,14 and with 100 μl of Biotinylation Reagent (biotinylated anti-mouse Ab) per ml of antibody solution. Immunocomplexes were allowed to stabilize for 30 min at RT. The Blocking Reagent was then added to the solution according to the manufacturer's instructions, and incubation was carried out for 10 more min at RT. The mixture was then applied to the sections and incubated for 1 hr at RT. After thorough PBS washes, horseradish peroxidase (HRP)-conjugated streptavidin was applied for 15 min at RT. Antibody binding was detected after reaction with 3-amino-9-ethylcarbazole (AEC)/hydrogen peroxide as chromogen-substrate for 20 min at RT. Slides were counterstained with Mayer's hematoxylin before microscopical evaluation.

Apoptosis assay

Monolayer cultures of tumor cell lines were harvested by trypsinization and cultured for 48 hr in 24-well plates (Costar, Cambridge, MA) at a concentration of 5 × 105/ml in complete medium in the presence or absence of recombinant CTLA-4 ligands, namely r-CD80 and r-CD86, at 25 μg/ml. This final concentration was selected after preliminary titration experiments in which the ligands were tested in the range of 6.25–50 μg/ml. Other recombinant fusion proteins, r-CD28 and r-CTLA-4, were used at the same final concentration as negative controls. PBMCs were incubated with ligands in similar conditions except that the incubation was performed in combination or not with PMA/PHA stimuli.

For coculture experiments, cells from SWEIG B-lymphoblastoid cell line, growing in suspension, were seeded on a monolayer of HOS osteosarcoma cells at different ratios (1:2, 1:4 and 1:8, HOS:SWEIG, respectively), and the cocultures were continued for 48 hr.

For inhibition of apoptosis studies, cells from HOS osteosarcoma cell line were incubated with r-CD80 or r-CD86 ligands in the presence or absence of anti-CTLA-4 scFvs #67 or #83 or with r-CTLA-4 fusion protein at different concentrations (25–100 μg/ml). All the apoptosis assays were performed in triplicate.

Evaluation of apoptosis

Cell apoptosis was evaluated by analyzing DNA content, cell viability and nuclear morphology. For DNA content analysis, adherent and nonadherent cells treated with r-proteins were first washed with PBS and then harvested by brief trypsinisation. HOS adherent cells from cocultures were collected after removal of SWEIG cell suspension. After centrifugation, cells were fixed in 70% ethanol for at least 1 hr. After washing with PBS, cells were resuspended in 500 μl of propidium iodide (PI, Sigma) solution containing PI at 50 μg/ml and RNase (Invitrogen, Milano, Italy) at 0.5 mg/ml in PBS for 30 min in the dark. DNA content was analyzed using a Coulter flow cytometer (EPICS Elite Coulter Electronics). Cell viability was analyzed by the Trypan blue dye exclusion assay. For morphologic evaluation, cells were collected after being washed once with ice-cold PBS and fixed using 0.25% solution of paraformaldehyde in PBS. Cells were attached to the slide by cytospin (130g for 6 min at 4°C), air-dried and stained with PI solution. The stained cells were examined using fluorescence microscopy.

Western blot analysis

Cells from HOS osteosarcoma cell line were washed twice with PBS and then resuspended in lysis buffer (10 mM TRIS-HCL pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Nonidet P-40, aprotinin 3 μg/ml, leupeptin 3 μg/ml) for 30 min on ice. Cell debris was removed by centrifugation, and protein concentration from HOS cell lysates was determined according to the instruction of the manufacturer using the Bio-Rad Protein Assay (detergent compatible; Bio-Rad, Milano, Italy). Equal amounts of protein (40 μg) were loaded on a reducing 12% (wt/vol) SDS-polyacrylamide gel, electrophoresed at 100 V for 2 hr and blotted onto nitrocellulose membrane by electrotransfer at 300 mA for 3.5 hr in 190 mM glycine 25 mM Tris, pH 8.3. After blocking with 0.1% nonfat dry milk, the membrane was incubated overnight at 4°C with 1:1,000 dilution of specific antibodies to caspase-8, -9, -10 and -3. After washing, the membrane was incubated with HRP-conjugated goat anti-mouse or anti-rabbit IgG for 1 hr at RT followed by treatment with ECL reagents (Amersham Life Science, Buckinghamshire, UK) and exposure to Hyperfilm ECL (Amersham Life Science).


CTLA-4 expression in tumor cell lines by flow cytometry

CTLA-4 expression was investigated by flow cytometry in a panel of 34 established tumor cell lines of different istotypes, including carcinoma, melanoma, osteo- and rabdomio-sarcoma in addition to neuroblastoma cell lines.

Cells were stained with 2 human anti-CTLA-4 FITC-scFvs #67 and #83 antibody fragments whose CTLA-4 specificity was previously well defined by enzymatic assay (ELISA), Western blot and immunofluorescence.14 ScFvs were used in parallel to the commercially available anti-CTLA-4 BN13 mAb either in surface or cytoplasmic direct immunofluorescence.

Although high levels of cytoplasmic CTLA-4 expression were observed in all the cell lines tested, surface expression was detected at various intensities on the majority of cell lines (30 of 34) showing the highest levels of expression on osteosarcoma (4/4) and breast carcinoma (3/4) cell lines, whereas the lowest levels were observed on renal (2/3) and uterine (3/3) carcinoma cell lines and melanoma (2/3) cell lines. Exceptions were SKRC-52 renal carcinoma and BT20 breast carcinoma cell lines in addition to F0-1 melanoma and SJNKP neuroblastoma cell lines that did not express surface CTLA-4 at all. Other tumor cell lines exhibited intermediate levels of CTLA-4 expression (Table I).

Table I. Analysis of CTLA-4 Expression in Human Tumor Cell Lines by Flow Cytometry and RT-PCR
Tumor cell linesFlow cytometry1RT-PCR2
CTLA-4Costimulatory moleculesCTLA-4 transcripts
scFv#67scFv#83BN13CD80CD86CD28Full lengthExtracellular
  • 1

    Surface reactivity of different human tumor cell lines and resting or activated PBMCs with two FITC-conjugated anti-CTLA-4 scFvs (#67, #83) and the commercial BN13 mAb by flow cytometry. Fluorescence intensity was scored as follows: −, negative; +/−, weak; and + to ++, positive, with grading from 1 log to more than 1 log of histogram shift relative to the negative control. The FITC-conjugated anti-BSA scFv#26 and a FITC-conjugated mouse IgG1 mAb were used as negative controls for anti-CTLA-4 scFvs and BN13 mAb, respectively (data not shown). Activation of PBMCs was obtained by incubation with PMA at 5 ng/ml and PHA at 2 μg/ml final concentrations for 48 h at 37°C as described in Materials and Methods.

  • 2

    RT-PCR analysis of CTLA-4 transcripts performed with primers specific for either the full length CTLA-4 coding region (conventional RT-PCR) or the extracellular domain (nested PCR). Reaction was scored as −, absence or +, presence of specific size of PCR products.

Breast carcinoma
Colon carcinoma
 COLO 205++++/−+
Renal carcinoma
Lung carcinoma
Ovarian carcinoma
Uterine carcinoma
Bladder carcinoma
 NB 100+++++++
Resting PBMCs+++
Activated PBMCs+++++/−++++

Similar reactivity patterns were obtained with all the antibody reagents, although BN13 mAb showed a weaker or absent reactivity with some cell lines, due to a different expression of surface epitopes. All the lines were negative (i.e., <5% reactive) for isotype control antibodies reactivity (data not shown).

In agreement with previous reports, no surface CTLA-4 expression could be detected on freshly isolated peripheral PBMCs, but it was induced upon 48 hr activation with PMA/PHA stimuli.14, 33 The flow cytometric profiles of surface and cytoplasmic CTLA-4 expression in 9 tumor cell lines, representative of different expression levels, are shown in combination to the profiles of resting and activated PBMCs (Fig. 1). The cytoplasmic CTLA-4 expression in resting PBMCs is consistent with the known T-cell intracellular compartmentalization of CTLA-4 that is mainly localized in vesicles of the Golgi apparatus and is released to the cell surface during T-cell activation.34, 35

Figure 1.

Flow cytometric profiles of CTLA-4 expression in 9 representative human tumor cell lines and resting or activated PBMCs. Cells were tested either on the surface (black histograms) or in the cytoplasm (gray histograms) after permeabilization as described in Material and Methods. Activation of PBMCs was obtained by incubation with PMA at 5 ng/ml and PHA at 2 μg/ml final concentrations for 48 hr at 37°C. Cells were stained with FITC-conjugated anti-CTLA-4 scFv #83 and analysed by flow cytometry. The FITC-conjugated anti-BSA scFv#26 was used as negative control for anti-CTLA-4 scFv #83 (not shown). Results are expressed as percentage of stained cells. Data are representative of 2 independent experiments.

Correlation of CTLA-4 with CD80, CD86 and CD28 expression in tumor cell lines

Surface expression of CTLA-4-specific ligands, CD80 and CD86, as well as the CTLA-4-structural homologue CD28, was also investigated by flow cytometry in the same cell panel as above. CD86 was the only other costimulatory molecule consistently expressed on the tumor cell lines (23 of 34) with few exceptions compared to CD80, which was weakly expressed in few samples (10 of 34) (Table I). By contrast, only 1 cell line weakly expressed CD28 costimulatory molecules on the surface. Additional flow cytometry analysis was carried out to investigate HLA-class I expression in the same tumor cell panel. HLA-A, -B and -C molecules were expressed in all of the cell lines, except in the ovarian A2780 and colon HCT-8 carcinoma cell lines, that resulted CTLA-4-positive, as well as in breast BT-20 and renal SKRC-52 carcinoma cell lines and melanoma F0-1 cell line that resulted all negative for CTLA-4 staining (data not shown).

Detection of CTLA-4 transcripts in tumor cell lines by RT-PCR

To confirm the CTLA-4 expression in tumor cell lines, total RNA was extracted from the cells and evaluation by RT-PCR was carried out using 2 sets of primers specific for the entire CTLA-4 coding region and for the extracellular domain.

Few tumor cell lines (5 of 34) revealed detectable RT-PCR products (Fig. 2a) obtained with the first set of primers that have been reported to amplify both the full-length (672 bp) and the alternatively spliced (550 bp) CTLA-4 transcripts in resting PBMCs,31 but none of them showed the splice variant transcript (Fig. 2a). In contrast, both transcripts were detectable in resting PBMCs as already described14 with the spliced transcript disappearing after PBMCs activation (Fig. 2a). The specificity of all the amplified bands was confirmed by sequencing.

Figure 2.

RT-PCR analysis of CTLA-4 transcripts in 8 representative human tumor cell lines and resting or activated PBMCs. Total RNA from tumor cells was reverse transcribed and PCR-amplified with primers specific for the CTLA-4 full-length coding sequence (exons 1–4) (black arrows) or for the extracellular domain (exon 2) (white arrows). (a) Five of 8 cell lines exhibited the full-length PCR product (672 bp) without the spliced variant (550 bp) corresponding to the deleted isoform, which lacks exon 3. (b) Nested PCR was performed on the first-round CTLA-4 full-length PCR product as template with CTLA-4 extracellular domain inner primers (exon 2 forward–exon 3 reverse). All cell lines exhibited the extracellular PCR product (369 bp). (c) As internal control, G3PDH gene amplification (599 bp) was carried out.

In further studies, a nested PCR assay was developed in which the first-round CTLA-4 full-length PCR product was amplified with CTLA-4 extracellular domain inner primers, resulting in a sharp band (369 bp) in all the cell lines tested (Fig. 2b). The failure in detecting CTLA-4 transcripts in tumor cells by conventional RT-PCR is probably due to the low amount of CTLA-4-specific RNA molecules present in these cells that requires a more sensitive nested RT-PCR method to be detected, as we have previously demonstrated in haematopoietic cell lines.14

CTLA-4 expression in tumor cells and their nonmalignant counterparts

CTLA-4 expression levels of tumor cells were compared to that of cells exhibiting a phenotype as close as possible to that of their normal counterparts. Osteosarcoma is a high-grade tumor composed of mesenchymal cells producing osteoid and immature bone. Human stromal stem cells (HSSCs) are multipotential cells obtained from a subset of clonogenic adherent marrow-derived cells that undergo replication in culture. HSSCs can be stimulated to differentiate toward lineages of the mesenchymal tissue, including bone, cartilage, fat, muscle, tendon and marrow stroma.36, 37, 38

Therefore, CTLA-4 surface expression of HOS osteosarcoma cell line was compared to that of HSSCs stimulated to differentiate toward the osteogenic lineage by adding the synthetic glucocorticoid Dex, a potent inducer of osteogenic differentiation.39

No expression of CTLA-4 could be detected by flow cytometry on control, as well as on the Dex-induced, HSSCs primary cultures, while HOS expression levels were high as opposite to the isotype control (Fig. 3a).

Figure 3.

Flow cytometric profiles of CTLA-4 expression in human HSSCs. (a) Top: Human stromal stem cells (HSSCs) were tested for surface CTLA-4 expression either as untreated HSSCs (empty histogram) or Dexamethasone (Dex)-treated (gray histogram) primary cultures compared to HOS osteosarcoma cell line (black histogram). Cells were stained with FITC-conjugated anti-CTLA-4 scFv #83. Bottom: Untreated HSSCs (gray histogram) and HOS (black histogram) were stained with FITC-conjugated anti-BSA scFv#26 used as isotype control. Results are expressed as percentage of stained cells. Data are representative of 2 independent experiments. (b) RT-PCR analysis of CTLA-4 transcript in HSSCs. (Lane A) Nested PCR was performed with CTLA-4 extracellular domain primers. Both untreated and Dex-treated HSSCs exhibited the extracellular PCR product (369 bp) as well as HOS cell line. Specific PCR products of (lane B) osteocalcin (303 bp) and (lane C) alkaline phosphatase (478 bp) markers were increased in Dex-treated HSSCs compared to control HSSCs. (Lane D) G3PDH gene amplification (599 bp) was carried out as internal control.

Although not expressed at surface level, CTLA-4 protein was detected at cytoplasmic level (data not shown) as well as CTLA-4 transcripts either in HSSCs or Dex-treated HSSCs (Fig. 3b, lane A). CTLA-4 transcriptional analysis was carried out in combination with that of osteocalcin (Oc) (Fig. 3b, lane B) and alkaline phosphatase (ALP) (Fig. 3b, lane C), 2 markers of the osteoblastic phenotype.40 The genes used as control for HSSC differentiation, Oc and ALP, confirmed the differences of expression at RNA level in Dex-treated HSSCs compared to untreated HSSCs, suggesting that these cells have indeed the characteristics of human osteoblasts.

The expression of CTLA-4 was evaluated by immunohistochemistry (IHC) in 6 human osteosarcoma (OS) tissue samples (all grade 3 tumors according to the FNCLCC grading system). IHC was carried out on formalin-fixed, paraffin-embedded tissues using anti-CTLA-4 scFv#83 in combination with the anti-tag 9E10 mAb. Peroxidase staining revealed cytoplasmic and surface expression in all the OS samples of either neoplastic bone with entrapped tumor cells or tumor cells growing in intratrabecular space (Fig. 4d–f ).

Figure 4.

Immunohistochemical staining of CTLA-4 in osteosarcoma (OS) tumor tissues. Formalin-fixed, paraffin-embedded tissue sections from a representative OS were stained using the biotin-streptavidin complex/HRP method (DAKO ARK). Anti-CTLA-4 scFv# 83 (10 μg/ml) was used as primary antibody in combination with the 9E10 mAb anti-c-myc peptide tag, as described in Material and Methods. Antibody reactivity was detected by addition of HRP-streptavidin and AEC/hydrogen peroxide. Slides were counterstained with Mayer's hematoxylin before microscopical evaluation (Microscope Nikon Eclipse E600W). Cytoplasmic and membrane-positive staining is shown in OS samples (d–f ) of either neoplastic bone with entrapped tumor cells (arrow n) or tumor cells growing in intratrabecular space (arrow t). No staining is detected when the primary antibody is omitted (a–c). Original magnifications are indicated in each panel.

CTLA-4 expression was also examined in human breast (BR) neoplastic and normal tissues derived from 5 ductal carcinoma patients (G1 and G2 grade). IHC was carried out according to the same protocol used for OS samples excluding the collagenase treatment for antigen unmasking. Immunostaining revealed membrane and cytoplasmic CTLA-4 localization in tumor BR cells of either the invasive or noninvasive part of the ductal carcinoma contained on the same slide (Fig. 5d) or on different slides (Fig. 5e,f ). Similar intensity of CTLA-4 staining was observed in the 2 BR tumour areas, whereas CTLA-4 positivity was absent or very weak in BR nonmalignant parenchimal ductal epithelial cells adjacent to tumor (Fig. 5a–c). Similar staining results were obtained in all 5 BR carcinoma samples.

Figure 5.

Immunohistochemical staining of CTLA-4 in breast (BR) tumor and normal tissues. Formalin-fixed, paraffin-embedded tissue sections from 3 representative BR invasive ductal carcinomas were stained using the biotin-streptavidin complex/HRP method (DAKO ARK). Anti-CTLA-4 scFv# 83 (10 μg/ml) was used as primary antibody in combination with the 9E10 mAb anti-c-myc peptide tag, as described in Material and Methods. Antibody reactivity was detected by addition of HRP-streptavidin and AEC/hydrogen peroxide. Slides were counterstained with Mayer's hematoxylin before microscopical evaluation (Microscope Nikon Eclipse E600W). Cytoplasmic and membrane-positive staining is shown in tumor cells of either the invasive (arrow i) or noninvasive (arrow ni) part of the ductal BR carcinoma contained on the same slide (d, G2 grade) or on different slides (e, G2 grade and f, G1 grade), whereas negative staining is shown in the normal BR parenchymal ductal epithelial cells (p) (a–c). Original magnifications are indicated in each panel.

CTLA-4 surface expression was also analyzed by flow cytometry on the nontumorigenic breast epithelial cell lines MCF10A and HC11 of human and murine origin, respectively. These lines expressed lower levels of CTLA-4 compared to the breast carcinoma cell line MCF7 with HC11 cell line exhibiting the lowest intensity of staining (data not shown).

CD80 and CD86 ligands can trigger apoptosis of tumor cells via CTLA-4 interaction

CTLA-4 engagement by its ligands CD80 and CD86 results in different inhibitory effects on T-cell proliferation including cell-cycle arrest.4, 5 Therefore, experiments were performed to investigate whether incubation of CTLA-4-expressing tumor cell lines with soluble recombinant ligands, namely r-CD80 and r-CD86, might exert a similar effect on tumor cell proliferation. To this end, tumor cell lines were incubated for 48 hr with medium alone or with addition of r-CD80, r-CD86, r-CD28, r-CTLA-4 or β-galactosidase, at various concentrations, and tested for analysis of cell cycle status by propidium iodide staining. R-CD28, r-CTLA-4 and β-galactosidase (the fusion partner of all recombinant proteins) were used as negative controls. As revealed by the DNA profiles of HOS osteosarcoma cell line, a percentage of cells in the sub-G1 area (indicative of apoptosis) was induced only by the r-CD80 (65%) or r-CD86 (57%) treatment (Fig. 6a, lane A.1). The apoptosis induction phenomenon occurred in a dose-dependent manner with optimal effect at the concentration of 25 μg/ml for both r-CD80 and r-CD86 (data not shown) and, not surprisingly, according to data from literature,41, 42 a more evident effect was exerted by r-CD80.

Figure 6.

Apoptosis induction of human tumor cell lines by CTLA-4 ligands. (a, A.1) Cell cycle analysis of HOS cells after treatment for 48 hr with CTLA-4 recombinant ligands, r-CD80 and r-CD86, at the optimal concentration of 25 μg/ml. R-CD28 and r-CTLA-4 recombinant proteins were used as negative controls at a similar concentration. At the end of the incubation period, the cells were harvested, washed once in PBS and analyzed for propidium iodide fluorescence by flow cytometry. Gated regions correspond to subdiploid quantities (left bar), G0/G1 (middle bar) and S/G2 phases (right bar), respectively. In the graphic of r-CD80- and r-CD86-treated cells, events are indicated as a percentage of the total number of events (5,000 as 100%). For r-CD28- and r-CTLA-4-treated cells, percentage was constantly below 5% (not shown). (A.2) Cell nuclear morphology of the above HOS cells analyzed by staining of cytocentrifuged samples with PI 50 μg/ml, RNAse 0.5 mg/ml, and fluorescence microscopy. (b) Cell-cycle analysis of HOS cells (adherent) after coculture for 48 hr with SWEIG EBV-B cell line (suspension) at different ratios. (c) Cell-cycle analysis of the other 8 representative tumor cell lines and PMA/PHA activated or resting PBMCs. Data are representative of 3 independent experiments.

Induction of apoptosis was further demonstrated by observation of cell nuclear morphology after PI-staining of HOS cell line and use of fluorescence microscopy (Fig. 6a, lane A.2).

To address the question of whether CTLA-4-expressing tumor cells can undergo apoptosis even after interaction with CD80/CD86-expressing cells, besides with soluble ligands, HOS osteosarcoma cells were coincubated with the CD80+ SWEIG B-lymphoblastoid cell line at different ratios. After separation of the 2 cell lines according to their growth properties, suspension (SWEIG) or adherence (HOS), the DNA content was analysed. The number of HOS apoptotic cells in the sub-G1 area increased, in a dose-dependent manner, according to the HOS:SWEIG ratio from 2% of untreated HOS cells to 12%, 18% and 26% of HOS-treated cells with SWEIG cells at the ratio of 1:2, 1:4 and 1:8, respectively (Fig. 6b).

Flow cytometry profiles of DNA content are also shown for 8 other representative cell lines and resting or activated PBMCs (Fig. 6c). Apoptotic cells, ranging from 12–70% depending on the ligand analyzed, were induced by r-CD80 and r-CD86 in CTLA-4-expressing cell lines. In contrast, the CTLA-4 surface negative breast BT-20 and renal SKRC-52 carcinoma cell lines were unaffected by ligands treatment as well as resting PBMCs, showing spontaneous apoptosis rate of 8% and 6%, respectively. However, activated PBMCs exhibited a small fraction of apoptotic cells in accordance with the low expression levels of CTLA-4 on activated T cells. Moreover, CTLA-4-expressing tumor cell lines treated with r-CD80/r-CD86 ligands resulted in a significant reduction of viable cells (evaluated by Trypan blue exclusion) (data not shown).

Specificity and mechanism of CD80/CD86-induced apoptosis

To confirm that specific binding of CD80/CD86 to surface CTLA-4 was involved in apoptosis induction, we performed competitive inhibition experiments.

First, to block the CTLA-4/CD80 interaction, HOS cells were incubated for 48 hr with r-CD80 alone or in combination with anti-CTLA-4 scFvs #67, #83 or BN13 mAb. As assessed by flow cytometry, addition of both scFvs exhibited a decrease in the percentage of apoptotic cells by 83% and by 71% with scFv #67 and scFv #83, respectively, both at a concentration of 100 μg/ml, in relation to the r-CD80 treatment alone of the control cultures (Fig. 7a). No further reduction was observed with the combination of both scFvs, and no competitive effect was exerted by the BN13 mAb. Treatment of HOS cells with scFv #67 or scFv #83 alone or with their combination had no effect on cell growth (data not shown).

Figure 7.

Apoptosis competitive inhibition and caspases activation. (a) HOS cells were cultured for 48 hr with medium alone (black bar) or r-CD80 (gray bars) or r-CD86 (white bars) ligands in the presence or absence of anti-CTLA-4 scFvs #67, #83, BN13 mAb (100 μg/ml) or with r-CTLA-4 (50 μg/ml). At the end of the culture, the cells were harvested, washed once in PBS and analyzed for apoptosis induction by propidium iodide fluorescence and flow cytometry. Results are expressed as mean ± SD of 3 different experiments. (b) Activation of caspase-8 and -3 in r-CD80- and r-CD86-treated HOS osteosarcoma cell line detected by Western blot. Proteins from HOS cell lysates were subjected to SDS-PAGE on a reducing 12% (wt/vol) polyacrylamide gel, transferred to nitrocellulose membrane and probed with antibodies to caspases-8, -9, -10 and -3. Forty micrograms were loaded in each lane. Immunoreactive bands were visualized by ECL technique. Immunostaining with a monoclonal antibody to β-tubulin confirmed equal loading. Blots are representative of 3 separate experiments.

We then incubated HOS cells with r-CD80 and r-CD86 in combination with r-CTLA-4 as competitive inhibitor. This resulted in a decrease in the percentage of apoptotic cells by 65% and 47%, respectively, in CD80- and CD86-treated cells, at optimal r-CTLA-4 concentration of 50 μg/ml (Fig. 7a).

These findings demonstrated that specific binding of r-CD80 and r-CD86 ligands to their natural receptor CTLA-4 expressed on HOS cells can trigger apoptotic effects via this receptor. We next investigated whether the apoptosis phenomenon induced by r-CD80/r-CD86 was caspase-dependent. To this end, Western blot analysis of caspase activation was carried out in HOS cell line treated with CTLA-4 ligands or r-CTLA-4 and r-CD28 for 48 hr. Cleavage of procaspase-8 into the characteristic 43/41 KDa and 18 KDa proteolytic fragments was observed after treatment with both r-CD80 and r-CD86 ligands (Fig. 7b). No cleavage was detected after incubation with either r-CD152 or r-CD28. In contrast, there was no activation of the other apical caspases, -9 and -10, even in response to r-CD80 and r-CD86 ligands. We then investigated the effector caspase-3, which was found to be cleaved into the 17–20 KDa proteolytic fragment by r-CD80 and r-CD86 treatments (Fig. 7b). R-CD152 and r-CD28 were ineffective. In addition, incubation of ligand-treated HOS cells with caspase-3 and caspase-8 selective inhibitors (Z-DEVD-FMK and Z-IETD-FMK, respectively) as already described43 markedly reduced the percentage of apoptotic cells, as assessed by flow cytometry (data not shown). This last finding strongly suggests that caspases are required for CTLA-4 triggered apoptosis.


Up to now, several costimulatory molecules have been analyzed in human solid tumor-derived cells including CD80, CD86, PD-1L, CD40, B7H2, OX40L and 4-IBBL,44, 45 but no results are available as to the expression of CTLA-4 on such tumor cells. In our study, we demonstrated that CTLA-4 is constitutively expressed in several types of tumor-derived cell lines including breast, colon, renal, lung, ovarian, uterine, bladder carcinoma cell lines, osteo/rabdomyosarcoma, neuroblastoma and melanoma cell lines. Analysis by flow cytometry revealed expression of CTLA-4 at different densities on 88% of cell lines examined, with higher intensity of staining on osteosarcoma and breast carcinoma cell lines. The finding that most CTLA-4-expressing tumor cell lines were also positive for CD86 ligand expression suggests that tumor cells may interact with each other and with APCs upon cell-to-cell contact.

HLA molecule expression was also investigated because simultaneous HLA-class I downregulation19 and constitutive expression of CTLA-4 immunosuppressive molecules might be expected to further contribute to the tumor immune surveillance escape. No correlation of CTLA-4 with tissue origin of cell lines was found as well as with HLA class I downregulation.

Expression of CTLA-4 in tumor cells was confirmed by nested RT-PCR analysis, showing that CTLA-4 is ubiquitously transcribed in all tumor cell lines tested, indicating a possible wider role of CTLA-4, as also testified by its high degree of conservation between species.46 However, posttranscriptional and/or posttranslational control may be responsible for the different expression levels of CTLA-4 at the cell surface.

This is the first study to our knowledge where CTLA-4 is shown to be expressed in a nonimmunogenic context on tumor cells, expanding its possible role from the negative control of immune response to the onset and progression of neoplastic process.

Immunohistochemical staining of OS and BR tumor tissues revealed cytoplasmic and surface expression in all of the samples investigated. A similar level of CTLA-4 staining intensity was observed in bone and intratrabecular space of OS as well as in the invasive and noninvasive areas of BR carcinomas.

We demonstrated that CTLA-4 found on tumor cells is functional in that it is able to specifically transduce an apoptotic signal after incubation with soluble CD80 and CD86 ligands or, to a lesser extent, with CD80-expressing B-EBV cells. Because it was not possible to completely reproduce with natural ligands (expressed on cell surface), the apoptotic phenomenon observed with recombinant ligands, it is reasonable to assume that optimal apoptosis induction condition mediated by CTLA-4 is achieved with a high ligand dosage that may not be the one corresponding to the physiologic state. Moreover, although significant spontaneous induction of cell apoptosis was not observed in tumour cell lines (CALU-1, HOS, MG-63, U2-OS) co-expressing CTLA-4 and CD80/CD86 ligands, the apoptotic effect might occur in vivo as a phenomenon resulting from the sum of more receptorial interactions.

The apoptosis was markedly reduced by the addition of blocking anti-CTLA-4 scFv antibodies or soluble CTLA-4 to the cultures, confirming that binding of CD80/CD86 was indeed responsible for the apoptosis induction. The apoptotic-inducing effects of CTLA-4 ligands have also been demonstrated by our group on primary neoplastic cells derived from acute myeloid leukaemia patients (data not shown).

At present, the involvement of CTLA-4 in apoptosis induction is not well defined and relies on controversial reports. Initial studies observed apoptosis after CTLA-4 crosslinking of a long-term human alloreactive T-cell clone,47 whereas other studies demonstrated that cross-linking of resting murine T cells blocks cell-cycle progression without inducing apoptosis.48 More recently, it has been reported that cross-linking of the CTLA-4 receptor by mAb on the surface of murine-activated T cells induces apoptosis in a Fas-independent manner that may involve a novel pathway.49 Therefore, it is conceivable that CTLA-4 induces different biochemical signals in resting vs. activated T cells, and tumor cells can be functionally regarded as “activated,” concerning cell-cycle machinery. Our findings demonstrate that upon binding of CD80 or CD86, there is cleavage of apical caspase-8 but not of either apical caspase-9 or -10. Nevertheless, the apoptotic process elicited through CTLA-4 is caspase-dependent, since there is also an activation of effector caspase-3. Caspase-8 is an effector of the death-receptor-mediated apoptotic signaling pathway, initiated by ligands such as Fas ligand (FasL), tumor necrosis factor-α (TNF-α) or TRAIL.50 As highlighted above, the Fas system does not seem to be involved in CTLA-4-mediated apoptosis of activated T cells. However, at present we could not rule out that the Fas/FasL system is not upregulated in cancer cells upon stimulation with CD80/CD86 or that other death ligands might cause cleavage and activation of procaspase-8. Further studies are necessary to elucidate this issue.

In conclusion, our results suggest that killing of tumor cells expressing CTLA-4 may be obtained upon triggering via this receptor, and this can have important clinical application especially in view of the broad distribution pattern of CTLA-4 that provides this receptor the potential of being a generic tumor antigen for targeted therapy.