Carbonic Anhydrase II Is a Tumor Vessel Endothelium–Associated Antigen Targeted by Dendritic Cell Therapy

Cancer immunotherapy makes use of the host immune response against tumor-associated antigens (TAA), some of which are mutated and others overexpressed in a tumor-specific manner. Identification of TAAs that elicit an efficient immune response is crucial in improving the quality of immunotherapy (1–3).

The TAAs of melanoma are generally classified into the following groups according to expression pattern: tumor-specific mutated antigens such as β-catenin (4); cancer-testis antigens such as MAGE-1 (5), HOM-MEL-40 (6), and NY-ESO-1 (7); and differentiation antigens such as tyrosinase (8). Although cancer vaccines using peptides of TAAs such as tyrosinase, MART-1, and NY-ESO-1 have been developed, their effects have been less than satisfactory thus far (9).

We previously conducted dendritic cell therapy on 10 melanoma patients and remarkable tumor reductions were observed in two patients. We then aimed to identify some of the unique antigens targeted by dendritic cell therapy in these patients. To identify the target antigens, we employed two-dimensional electrophoresis combined with Western blots analysis and matrix-assisted laser desorption ionization-time of flight/mass spectrometry (MALDI-TOF/MS) methods. Through this strategy, carbonic anhydrase II (CA-II) was identified as an antigen that elicited serum antibody response to dendritic cell therapy in the patient. To our knowledge, this is the first report showing the presence of serum anti–CA-II antibody in cancer patients although it has been reported that the antibody is prevalent in some autoimmune diseases (10). In this study, we also did the immunohistochemistry of CA-II in various cancer specimens and found that CA-II was expressed in tumor vessel endothelium. Taken together with the data of mRNA assay for CA-II in an in vitro angiogenesis model, we suggest that CA-II is a tumor vessel endothelium–associated antigen and discuss the relevance of this antigen in cancer immunotherapy.

Patients. Ten malignant melanoma patients were treated by dendritic cell therapy at our department from 2000 until 2002. The detailed protocol and results of the therapy are described elsewhere (11). In brief, dendritic cells were prepared by culturing autologous monocytes with granulocyte-macrophage colony-stimulating factor and interleukin-4, followed by pulsation with autologous tumor lysate combined with tumor necrosis factor α. All of the patients died by the end of 2002 and no one had manifested a complete response in accordance with the WHO criteria on solid tumors (12). However, three patients had shown favorable clinical outcomes to dendritic cell therapy. Specifically, shrinkage or disappearance of multiple metastatic tumors with massive necrosis occurred during the therapy in patients 8 and 9 and tumor growth ceased for 4 months after the therapy in patient 6 (stable disease condition).

Tumor specimens and the frozen sera were stored at −80°C and −20°C, respectively, until use. Pre-therapy serum samples were collected within 1 month before therapy and post-therapy serum samples were collected between 2 and 12 months after initiating therapy, depending on the survival period. As controls, sera of 10 melanoma patients who had not undergone dendritic cell therapy and sera of six thyroid cancer patients who had undergone dendritic cell therapy at our department were employed. Normal blood and serum were provided by healthy volunteers. Informed consent for enrollment in this study was obtained from each subject and the use of materials was approved by the Institutional Review Board of the Institute of Medical Science, University of Tokyo.

Cell lines, chemicals, and antibodies. Five melanoma cell lines, CRL1579, G361, HMV-I, HMV-II, and SK-MEL-28, were provided by Cell Resource Center for Biomedical Research, Tohoku University (Sendai, Japan). Culture media were DMEM (Invitrogen, Carlsbad, CA) for HMV-I and HMV-II, RPMI 1640 for G361 and CRL1579, and Eagle's MEM for SK-MEL-28 cells. All media were supplemented with 10% fetal bovine serum (Thermo Trace Ltd., Melbourne, Australia).

Purified human CA-II (Sigma-Aldrich, St. Louis, MO) and rabbit anti-human CA-II antibody (Chemicon International, Inc., Temecula, CA) were purchased.

Preparation of cell lysate. The frozen tumor block was homogenized in PBS supplemented with proteinase inhibitor cocktail (Complete Mini, Roche Diagnostics, Mannheim, Germany) and 5 mmol/L iodoacetamide using a Polytron homogenizer and centrifuged at 2,000 × g for 10 minutes. The pellet was incubated in lysis buffer [20 mmol/L Tris-HCl (pH 8), 140 mmol/L sodium chloride, 2% Triton X-100, 1% sodium deoxycholate, 5 mmol/L iodoacetamide, Complete Mini protein inhibitor cocktail] by rotating for 2 hours at 4°C. The supernatant after centrifuging for 1 hour at 18,000 × g and 4°C was then harvested and stored at −80°C.

Freshly drawn blood was mixed with 5 mmol/L EDTA and fractionated by centrifugation with Ficoll-Paque Plus (Amersham Biosciences, Uppsala, Sweden). Erythrocyte membrane (ghosts) and cytoplasmic fraction were separated by hemolysis in 5 mmol/L sodium phosphate (pH 8) and centrifugation (13). To obtain an erythrocyte cytoplasmic 10- to 50-kDa protein fraction, the hemolytic solution was filtered using a filtration device limiting 50-kDa protein (Centriplus YM-50, Millipore, Bedford, MA) and the filtrate was concentrated using a filtration device limiting 10-kDa protein (Centriplus and Microcon YM-10, Millipore).

Electrophoresis and Western blots. Conventional SDS-PAGE was done based on the method of Laemmli (14). After electrophoresis, the gel was blotted on Protran nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany). The membrane was incubated with blocking buffer composed of TBS-T [20 mmol/L Tris-HCl (pH 7.6), 137 mmol/L sodium chloride, 0.05% polyoxyethylenesorbitan monolaurate] supplemented with 5% skim milk for 2 hours at room temperature. To detect serum reactivity, the membrane was incubated with serum diluted in blocking buffer for 2 hours at room temperature, followed by incubation with horseradish peroxidase–conjugated antihuman immunoglobulin G Fc portion antibody (Sigma-Aldrich). The signal was detected by using enhanced chemiluminescence system (Amersham Pharmacia Biotech UK, Buckinghamshire, United Kingdom).

For two-dimensional electrophoresis analysis, protein samples were prepared in 8.5 mol/L urea, 2% NP40, 2% Ampholine (Amersham Pharmacia Biotech AB, Uppsala, Sweden), and 5% 2-mercaptoethanol. Isoelectric electrophoresis was done using immobilized pH gradient tube gel (Daiichi Kagaku, Tokyo, Japan) according to the instruction of the manufacturer. After electrophoresis, the gel was soaked in an equilibration buffer [125 mmol/L Tris-HCl (pH 6.8), 4.3% SDS, 10% 2-mercaptoethanol, 0.01% bromophenol blue, 30% glycerol] and loaded on polyacrylamide slab gel (Daiichi Kagaku). After electrophoresis, the gel was stained with Coomassie brilliant blue R-250 (ICN Biomedicals, Aurora, OH) or blotted on nitrocellulose membrane followed by Western blots as described above.

Protein identification. Spots of Coomassie brilliant blue–stained gel corresponding to positive spots on the Western blot were cut out and decolorized in a solution of 50 mmol/L ammonium hydrogen carbonate and 50% methanol and were subjected to reduction in a mixture of 10 mmol/L DTT and 0.1 mol/L ammonium hydrogen carbonate, followed by alkylation in a mixture of 40 mmol/L iodoacetamide and 0.1 mol/L ammonium hydrogen carbonate. Subsequently, the gel was incubated in 50 nmol/L trypsin for 14 hours at 37°C and digested peptides were subjected to the 4700 Proteomics Analyzer MALDI-TOF/MS (Applied Biosystems, Foster City, CA). Protein identification was carried out using the Mascot Search system (Matrix Science, London, United Kingdom).

Immunohistochemistry. Immunohistochemical staining of tissue sections with anti–CA-II antibody was done using a peroxidase technique. Briefly, deparaffinized sections of formalin-fixed, paraffin-embedded tissues were heated in 10 mmol/L citrate buffer (pH 7.0) at 95°C for 30 minutes for antigen retrieval. After washing, sections were treated by H₂O₂ solution to quench endogenous peroxidase activity followed by blocking reagent (K5006, DakoCytomation Japan, Kyoto, Japan) to block nonspecific binding sites. The sections were then incubated with primary antibody at a dilution of 1:4,000 for 45 minutes at room temperature. DAKO Envision system (EnVision+, peroxidase, rabbit, DakoCytomation) was used for the detection of immune complexes.

In vitro angiogenesis model. Human umbilical vein endothelial cells (HUVEC; Clonetics, San Diego, CA) were normally cultured in endothelial growth medium (Clonetics) containing 2% fetal bovine serum at 37°C under atmospheric conditions of 95% air and 5% CO₂. To make an in vitro angiogenesis model, the wells of a six-well tissue culture plate were coated with Matrigel basement membrane matrix (BD Biosciences, Bedford, MA). HUVECs (2 × 10⁵) were seeded in a Matrigel-coated well and cultured in experimental conditions for 18 hours. Cells were harvested with Cell Recovery Solution (BD Biosciences) and subjected to reverse transcription-PCR assay.

Quantitative reverse transcription-PCR. Total RNA was extracted from tumors or cultured cells using Trizol (Invitrogen). First-strand cDNA was transcribed from total RNA using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). Real-time PCR was done using QuantiTect SYBR Green PCR kit (Qiagen, Hilden, Germany) and iCycler (Bio-Rad) with 45cycles of 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 30 seconds. One microgram of total RNA was subjected to reverse transcription reaction in a 20-μL solution and a 1-μL aliquot after the reaction was used in the subsequent real-time PCR. The primers for CA-II were caatggtcatgctttcaacg and tccatcaagtgaaccccagt and those for glyceraldehyde-3-phosphate dehydrogenase were gctcatttcctggtatgacaac and ttcctcttgtgctcttgctg. The quantity of CA-II mRNA copies relative to glyceraldehyde-3-phosphate dehydrogenase (internal control) was calculated on the basis of standard reactions.

Statistical analysis. The correlation between two factors was evaluated by using the χ² analysis. The nonparametric Mann-Whitney U test was used to evaluate any significant differences between two groups. P < 0.05 was considered statistically significant.

29-kDa antigen eliciting serum antibody. To identify the serum antibodies augmented after dendritic cell therapy in patient 9 who experienced massive tumor necrosis following therapy as was described in Materials and Methods, tumor lysate from the patient was subjected to Western blots with the autologous sera of pre- and post-therapy (6 months after the start point). As depicted in Fig. 1A, a 29-kDa band appeared only in the post-therapy sample whereas 47-kDa bands were detected in both the pre- and post-therapy sera. We then aimed to identify the 29-kDa antigen, a possible target of dendritic cell therapy. We tried to separate the tumor lysate by two-dimensional electrophoresis followed by Western blots with the serum but it was difficult to obtain the 29-kDa protein purely and abundantly enough for subsequent MALDI-TOF/MS analysis. Therefore, we searched the 29-kDa protein in melanoma cell lines, HUVECs, and blood cells by Western blots with the serum and found that the 29-kDa protein was abundant in cytoplasmic fraction of erythrocytes (data not shown).

Two-dimensional electrophoresis and matrix-assisted laser desorption ionization-time of flight/mass spectrometry. The 29-kDa protein was isolated from erythrocytes as follows. After hemolysis and removal of membranes (ghosts), the hemolytic solution was size fractionated to collect a 10- to 50-kDa protein fraction so that the sample did not contain any hemoglobin α₂β₂ tetramer (68 kDa), the isomers of which would cause contamination in degenerative electrophoresis. The fraction was subjected to two-dimensional electrophoresis followed by Western blots with the post-therapy serum of patient 9. Three positive spots at 29 kDa, which similarly reacted with the serum, were detected (Fig. 1B). These three spots of the Coomassie brilliant blue–stained gel (Fig. 1B) were separately retrieved and analyzed by MALDI-TOF/MS. As a result, the three spots were revealed to represent the same protein, CA-II. MS data for spot 1 are shown in Table 1. The reason for the differences in isoelectric point among these three was unclear. To confirm that the 29-kDa protein in the lysate recognized by the serum is CA-II, tumor lysate and purified CA-II were loaded in parallel on SDS-PAGE and Western blotting with the post-therapy serum, the same serum precleared by purified CA-II, or anti–CA-II antibody was done. As shown in Fig. 1C, 29-kDa bands in both lanes of the tumor lysate and purified CA-II appeared in Western blots with the post-therapy serum and anti–CA-II antibody in a similar fashion, and the bands diminished in Western blots with the same serum precleared by purified CA-II. These results indicated that the post-therapy serum specifically reacted with CA-II in the tumor lysate. The serum reactivity level with CA-II was assessed by Western blots and the serum was positive at 1:400 dilution against 1 μg purified CA-II (Fig. 1D).

Anti–carbonic anhydrase II antibody in the patients. We assessed anti–CA-II antibody response by dendritic cell therapy in 10 melanoma and 6 thyroid cancer patients who underwent dendritic cell therapy at our department. One microgram of purified CA-II was subjected to Western blots with the patients' sera. Anti–CA-II antibody was positive in 6 of 10 melanoma patients and in none of the six thyroid cancer patients at a serum dilution of 1:100. All of three normal sera were negative at the same dilution. For the six positive patients, Western blots at serum dilutions of 1:100 and 1:200 were shown in Fig. 1E. Clinical courses and status of the antibody levels in melanoma patients are summarized in Table 2. A marked antibody response to dendritic cell therapy in patient 9 (from none to 1:400 dilution) as well as a weak elevation in antibody levels in patients 6 and 8 (from 1:100 to 1:200 dilution) was observed whereas the antibody levels remained low or diminished in patients 1, 3, and 7. When the patients are divided to clinical responders (patients 6, 8, and 9) and nonresponders (other patients), the relationship between clinical response and antibody response to dendritic cell therapy was significantly correlated by the χ² analysis (P < 0.05, Fisher's exact test). As we were interested in the fact that anti–CA-II antibody was positive in five melanoma patients before dendritic cell therapy, we further investigated the antibody status in 10 other melanoma patients who had not been enrolled in dendritic cell therapy. Three of them were positive at 1:100 dilution and negative at 1:200 dilution. Accordingly, 8 of 20 (40%) melanoma patients were positive for anti–CA-II antibody independent of dendritic cell therapy.

Immunohistochemistry for carbonic anhydrase II. To examine localization of CA-II antigen expression in the tissue, immunostaining was done using a rabbit anti-human CA-II antibody (Fig. 2). Consistent with previous report (15), CA-II could be detected in the cytoplasm of the tubular epithelium (Fig. 2B). In the metastatic tumor of the kidney in patient 9, expression of CA-II was found to be positive in the membrane of the tumor vessel endothelium (Fig. 2D). Of note was that the vessel endothelium as well as glomerular endothelium of the normal kidney was negative for CA-II (Fig. 2B, arrows). In patient 8, however, tumor vessel endothelium was negative for CA-II (data not shown). To address whether CA-II expression in the tumor vessel endothelium is melanoma specific, we did CA-II immunostaining in other cancer tissues. Positive CA-II expression in tumor vessel endothelium was also detected in some, but not all, cancer tissues including esophageal cancer, renal cell carcinoma, and lung cancer (Fig. 2E-H). Thus, CA-II expression in vessel endothelium was limited in tumors including melanoma and was not detected in the normal tissues as far as we examined.

Reverse transcription-PCR for carbonic anhydrase II in melanomas. Immunohistochemistry analysis had revealed that CA-II expression was positive in the tumor vessel endothelium but not in the melanoma cells or normal vessel endothelium. CA-II–positive tumor vessel endothelium was seen in patient 9 but not in patient 8. To better understand these findings, CA-II expression in tumors and melanoma cell lines was assessed by quantitative real-time reverse transcription-PCR. The quantity of CA-II mRNA copies relative to glyceraldehyde-3-phosphate dehydrogenase was evaluated. CA-II expression in the tumor of patient 9 was higher than those in the tumor of patient 8 and in the five melanoma cell lines (Fig. 3A). These data suggest prominent expression of CA-II in nonmelanoma cells in the tumor of patient 9 and are consistent with the results of immunohistochemistry analysis showing that CA-II expression was limited in the tumor vessel endothelium of patient 9.

Carbonic anhydrase II expression in in vitro angiogenesis. To assess CA-II expression of HUVECs in an in vitro angiogenesis model, the cells were cultured on a plastic plate (two-dimensional culture, Fig. 3B) or Matrigel (three-dimensional culture, Fig. 3C) for 18 hours and CA-II mRNA expression was evaluated using quantitative real-time reverse transcription-PCR. CA-II expression was significantly up-regulated in the three-dimensional culture compared with two-dimensional culture in the normal media and atmosphere. When cells were exposed to acidic media (pH 6.8), hypoxia (2% oxygen), or both in the three-dimensional culture, CA-II expression was significantly up-regulated in the acidic and hypoxic conditions (Fig. 3D).

By using two-dimensional electrophoresis combined with Western blots and MALDI-TOF/MS methods, we have identified CA-II as an antigen which elicited humoral immune responses in melanoma patients. A similar strategy has been employed to explore TAAs in renal cell carcinoma (16, 17), neuroblastoma (18), and lung cancer (19), the methods of which these researchers have named “serological proteome analysis: SERPA” (20) or “proteome-based target evaluation combined with immunoreactive target structure identification explored by sera: PROTEOMEX” (21) or “serological and proteomic evaluation of antibody responses: SPEAR” (22). Another powerful way of detecting TAAs is the serologic identification of antigens by recombinant expression cloning method (SEREX; ref. 23). In this study, we employed the proteomic strategy to identify a target protein that we found in the preliminary Western blots, the results of which have been successful.

It was thought that TAAs with gene mutations such as β-catenin are specific to the tumor and recognized as “nonself” by the host immune system. On the other hand, the majority of TAAs in humans do not have mutations and are expressed in certain normal tissues, as is the case with cancer-testis antigens that are expressed in normal testis and with differentiation antigens that are expressed in normal melanocytes. Furthermore, some TAAs have also been reportedly expressed in a wide range of normal tissues. For instance, ATP6S1, a putative accessory unit of the vacuolar H⁺-ATPase complex that is expressed in a broad range of normal tissues, has been reported to elicit potent humoral responses associated with immune-mediated tumor destruction (24). Therefore, this ubiquitous expression in normal tissues may not exclude the possibility that CA-II is a TAA.

CA-II was identified as a TAA through reaction with the serum of patient 9. Actually, a marked antibody response (from negative to 1:400) was detected only in patient 9. Additional two patients who had antibody response (from 1:100 to 1:200) were patient 6 (stable disease) and patient 8, who had tumor reductions along with patient 9, whereas no antibody response was detected in the remaining patients who had no clinical response to dendritic cell therapy as depicted in Table 2. Anti–CA-II antibody status during therapy may be useful as a marker of good clinical response to dendritic cell therapy, although additional studies are required to conclude the notion as the number of patients who responded to dendritic cell therapy was small.

The presence of anti–CA-II antibody has been reported in some systemic autoimmune diseases, such as systemic lupus erythematosus (prevalence of 24.1%), primary Sjögren's syndrome (20.0%), progressive systemic sclerosis (16.7%), dermatomyositis (25.0%; ref. 10), and in other autoimmune-related diseases including Sjögren-associated chronic pancreatitis (61.9%; ref. 25), autoimmune-related pancreatitis (58.8%; ref. 26), primary biliary cirrhosis [8% (27) to 18% (28)], type 1 diabetes (50.0%; ref. 29), and ulcerative colitis (27.8%; ref. 30). The mechanism of antibody acquisition or the role of the antibody in pathophysiology of these diseases has not been elucidated. In this study, 8 of 20 (40%) melanoma patients had anti–CA-II antibody independent of dendritic cell therapy although the antibody level was relatively moderate compared with the augmented antibody level after dendritic cell therapy in the responders. As seen in patient 3 whose antibody diminished after dendritic cell therapy, anti–CA-II antibody may fluctuate independent of dendritic cell therapy in some melanoma patients. Because no thyroid cancer patient had anti–CA-II antibody in this study and no cancer patient with anti–CA-II antibody has been reported thus far, melanoma may be a unique cancer in this regard.

CA-II has not generally been regarded as a tumor-associated protein because it is expressed in a wide variety of normal cells including erythrocytes, pancreatic epithelial cells, as well as those of the kidney and gastrointestinal tract. A few examples of cancers with high CA-II expression are brain tumors (31) and leukemia (32). Likewise, in this study, CA-II expression was low in melanoma cells. However, we showed that CA-II is expressed in the tumor vessel endothelium instead of normal vessels not only in melanoma but also in esophageal cancer, renal cell carcinoma, and lung cancer. Therefore, we propose that CA-II is a tumor vessel endothelium–associated antigen.

To support the notion that CA-II is specifically expressed in the tumor vessel endothelium, we investigated CA-II mRNA expression in an in vitro angiogenesis model. Compared with the two-dimensional culture, CA-II was significantly up-regulated in the three-dimensional culture, suggesting that CA-II is required in angiogenesis of normal endothelial cells. We further addressed CA-II expression in tumor angiogenesis where we set up cultures in acidic media (pH 6.8) and hypoxic conditions (2% oxygen) to mimic a tumor environment (33, 34). In the acidic and hypoxic conditions, CA-II was significantly up-regulated compared with normal conditions, supporting the specific CA-II expression in tumor vessel endothelium. In our study, however, not all of the tumors had CA-II–positive tumor vessel endothelium. Tumor endothelial CA-II expression may depend on the degree of acidity or hypoxia within the tumor. Another possibility is that CA-II expression is a phenotype of tumor vessel endothelium limited to some tumors. Although tumor vessel endothelium–specific genes were explored (35), to our knowledge, CA-II has not been reported thus far.

How anti–CA-II immunity recognizes CA-II expressed in the tumor vessel endothelium is a critical question because CA-II is generally known to be a cytoplasmic protein. In the immunohistochemistry analysis, CA-II expression was localized to the membrane of the tumor endothelium. CA-II may be attached to the inner leaflet of the membrane as shown in a pancreatic cancer cell line (36). However, it is still possible that CA-II expressed in the endothelium is targeted by the immune system because mouse models have shown that immunization with CA-II induced sialoadenitis (37), cholangitis (38), and pancreatitis (39) by targeting CA-II expressed in each organ.

The discovery of CA-II expression specific to tumor vessel endothelium may lead to another therapeutic strategy. CA inhibitors have been shown to inhibit growth and invasion of cancers including melanoma (40, 41) although its precise mechanism has yet to be uncovered. Given that CA-II has a critical role in tumor angiogenesis, CA inhibitors may deteriorate tumor angiogenesis and cause tumor reduction or inhibition of tumor growth.

Further investigation is needed to clarify the role of CA-II in tumor angiogenesis and to explore the therapeutic effects of vaccination with CA-II and CA inhibitors against tumor angiogenesis.