Gliomas are the most common neoplasms of the central nervous system, comprising more than 60% of primary brain tumors. Glioblastoma multiforme (GBM) is the most malignant of these tumors, with an average survival time of less than 16 months and a 5-year survival time of approximately 5%.1 Despite aggressive multimodal therapy, the prognosis for patients with malignant glioma is still poor. Thus, the development of new therapeutic strategies is required.
The identification of human tumor antigens is important not only for the analysis of antitumor immune responses and the development of immunotherapy, but also for the development of diagnostic methods.2 Various methods for the identification of tumor antigens have recently been applied, including cDNA expression cloning using tumor-reactive T cells and IgG antibodies in patients' sera as well as a reverse immunology strategy that evaluates the induction of T cells against candidate molecules identified by various techniques, such as systematic gene expression analysis.
As of yet, several immunologic markers for glioma patients have been reported.3, 4 However, it is unclear how immune responses against these antigens could be induced in glioma patients or whether these immune responses are specific for glioma patients because these antigens were expressed not only in glioma tissue, but also in normal adult tissues. Previously, we reported SOX5 and SOX6 as glioma antigens detected by serological screening using a testis cDNA library and sera samples from allogenic glioma patients.5 The present study was designed to analyze the expression and immunoreactivity of SOX5 in human glioma. We also compared the survival of GBM patients who had antibodies against SOX5 to that of GBM patients without SOX5 antibodies to evaluate the role of the SOX5 antibody response as a prognostic factor for GBM patients.
Material and methods
This study was approved by the local ethical review board of Keio University (No. 12-21-2). Twenty-nine glioma patients were examined, after obtaining their informed consent. All the glioma patients had undergone a resection of their primary tumor. The glioma diagnoses were independently confirmed by 2 neuropathologists. Postoperative radiotherapy (60 Gy) and chemotherapy using MCNU and vincristine were performed for all GBM patients, 8 of the 9 anaplastic astrocytoma patients and all the anaplastic oligoastrocytoma patients. None of the low-grade glioma patients (5 diffuse astrocytomas and 1 oligoastrocytoma) received chemotherapy. Two diffuse astrocytoma patients received conventional radiotherapy (50 Gy), and a diffuse astrocytoma patient underwent stereotactic radiosurgery. Karnofsky performance status (KPS) scores were recorded at the time of surgery. Preoperative and early postoperative magnetic resonance (MR) imaging with contrast medium was used to determine the extent of tumor resection.
Sera and tissues
Tumor tissues and patient sera were obtained from the Department of Neurosurgery, Keio University School of Medicine. Sera from 29 patients with gliomas (12 glioblastomas, 9 anaplastic astrocytomas, 2 anaplastic oligoastrocytomas, 1 oligoastrocytoma, and 5 diffuse astrocytomas; patient age range, 3–77 years; average patient age, 43.8 years; 13 men and 16 women), 14 patients with other brain diseases (including meningioma, malignant lymphoma, metastatic brain tumor, Parkinson disease, craniopharyngioma, central neurocytoma, brain abscess and subarachnoid hemorrhage), 54 patients with various cancers and 37 healthy individuals (age range, 20–83 years; average patient age, 52.3 years; 20 men and 17 women) were analyzed. Tumor tissues (6 cases) out of these 29 patients were classified according to the recent World Health Organization (WHO) criteria6 as follows: GBM (WHO grade IV), anaplastic astrocytoma (WHO grade III), or diffuse astrocytoma (WHO grade II). We made paraffin-embedded tissue sections from 2 of the 6 glioma tissues and nonneoplastic tissues from the cerebral cortex. Normal tissues (brain, lung, heart, stomach, testis, esophagus, pancreas, spleen, liver, kidney, placenta, colon and skeletal muscle) were obtained from Clontech (Palo Alto, CA).
Glioma cell lines
The glioma cell line, Marcus, was purchased from Health Science Research Resources Bank (Osaka, Japan). Cell lines of U-87 (glioma) and T98G (glioma) were purchased from American Type Culture Collection (Manassas, VA).
Marcus, T98G and U87 were cultured in DMEM (Sigma) supplemented with 10% FBS, 100 IU/mL penicillin and 100 μg/mL streptomycin.
Construction of the cDNA library, immunoscreening of the testis cDNA library and characterization of immunoreactive clones
Construction of the cDNA library and the SEREX method were carried out as described previously.5 The testis cDNA library was screened with mixed sera from 4 glioma patients. The positive clones were picked up, and PCR was conducted by the Ex Taq kit (Takara, Kyoto, Japan). The PCR products were then sequenced on an ABI Prism 3100 sequencer (PerkinElmer, Branchburg, NJ).
Detection of antibodies against SOX5 by SEREX
Monoclonal phages from clones that reacted with the sera used for the immunoscreening were mixed with nonreactive phages of the cDNA library as internal negative controls at a ratio of 1:10 and used to transfect bacteria. IgG antibodies in the 1:100-diluted Escherichia coli-preabsorbed sera from glioma patients, other patients, and healthy controls were then tested with the above-described phage assay to determine antibody responses against SOX5.
Total RNAs from normal tissues were purchased from CLONTECH Laboratories. Total RNAs from glioma tissues were isolated using Trizol (Gibco BRL). The cDNA preparations used as templates in the quantitative reverse transcript-polymerase chain reaction (RT-PCR) reactions were synthesized by incubating total RNA template (5 μg), oligo (dT), and Avian myeloblastosis virus reverse transcriptase (Takara, Tokyo) in a total reaction volume of 200 μL at 42°C. Primers were designed as follows: forward primer, 5′-CAAGGCAATCATGCGCAACA-3′; reverse primer, 5′-TGCTAGACACGCTTGAGTGC-3′; and for β-actin, forward primer, 5′-GGCACCCAGCACAATGAAG-3′; reverse primer, 5′-GCCGATCCACACGGAGTACT-3′.
Quantitative RT-PCR analysis was performed using a fluorescent dye, SYBR Green, and the ABI prism 7700 Sequence Detection System (PerkinElmer, Foster City, CA), as previously described.7 The thermal cycler parameters were as follows: 10 min at 95°C, and 50 cycles of denaturation at 95°C for 30 sec, annealing at 60°C for 1 min, and extension at 72°C for 1 min. The relative SOX5 expression level for each glioma tissue was normalized to the β-actin level in each tissue and calculated as the threshold cycle (CT) value in each sample divided by the CT value in the normal brain. The CT value is defined as the value obtained in the PCR cycle when the fluorescence signal increases above the background threshold.
Western blot analysis of SOX5 in normal tissues, glioma tissues, and glioma cell lines
SOX5 expression in glioma and normal tissues was tested using a Western blot analysis. SOX5 protein exists in the nuclei of mammalian cells.8 To isolate nuclear extracts, normal tissues (purchased from CLONTECH Laboratories), glioma tissues, and 293T cells transfected with the full-length short form of SOX5 cDNA9 were homogenized in 0.25 M sucrose and centrifuged for 10 min. Pellets were resuspended in 0.25 M sucrose supplemented with a protease inhibitor cocktail (Sigma Aldrich). Twenty micrograms of total protein were mixed with an equal volume of sodium dodecyl sulfate (SDS)-sample buffer composed of 2% SDS, 10% β-mercaptoethanol, 10% glycerol, 1 mM EDTA, 40 mM Tris and 240 mM glycine at pH 8.5 and boiled for 3 min. The same amount of total protein was then loaded on each lane, and the proteins were transferred onto a nitrocellulose membrane (Hybond C+; Amersham Pharmacia) by electroblotting. After blocking with 5% skim milk for 2 hr, the sheets were incubated with a goat anti-human SOX5 polyclonal antibody (10 μg/mL, Santacruz), followed by 1:2,000-diluted peroxidase-conjugated mouse anti-goat immunoglobulin G (IgG) (Cappel, Aurora, Ohio). The proteins were visualized with the help of an ECL Western blot detection system (Amersham Biosciences, Buckinghamshire, UK).
To evaluate the specificity of the anti-SOX5 antibody, the antibody was incubated with 293T cells expressing SOX5 at a final concentration of 10 μg/mL in 1% PBS, then applied to a Western blot as described earlier.
Paraffin-embedded tissue sections (10 μm) were deparaffinized in xylene and rehydrated. The sections were then treated with a heat-based antigen retrieval method using a citrate solution (pH 6.0, 10 mM). Endogenous peroxidase was blocked by incubation in 0.3% hydrogen peroxide in methanol, and the nonspecific binding of antibodies was blocked by incubation in 5% BSA (bovine serum albumin) in 0.02 M PBS and 0.1% Triton X-100 in PBS for 1 hr. The slides were then incubated with anti-human SOX5 antibody (3 μg/mL) diluted in the same blocking solution overnight at 4°C. The slides were incubated with a secondary antibody (Universal Immunoperoxidase Polymer, Anti-Goat; Histofine Simple Stain MAX PO (G); NICHIREI CORPORATION, Japan) for 30 min at 37°C, and the HRP labeling was visualized using DAB. The sections were lightly counterstained with hematoxylin. Each step was followed by 3 washes in PBS.
The Fisher exact test was used to assess proportional differences among the 29 glioma patients. The survival periods of the 12 GBM patients were measured as the time from the date of the surgery for the primary tumor until the date of death from any cause. The overall survival curves were estimated using the Kaplan-Meier method and were compared using the log-rank test. Overall survival was also assessed using a Cox regression analysis adjusted for potential confounding factors. The statistical analysis was performed using SPSS version 14.0 (SPSS, Chicago, Illinois). P-values less than 0.05 were considered statistically significant, and all tests were 2-sided.
Isolation of glioma antigen, SOX5, recognized by IgG antibodies in glioma patients' sera
Approximately 1 × 106 clones from a human testis cDNA library were screened by SEREX using the mixed sera of 4 glioma patients (a 41-year-old man with anaplastic astrocytoma, a 21-year-old woman with glioblastoma, a 7-year-old girl with diffuse astrocytoma and a 3-year-old girl with diffuse astrocytoma). In this screening, 11 clones encoding 4 gene products were isolated (a SOX5-encoding clone, 6 SOX6-encoding clones, 3 M-phase phosphoprotein 1-encoding clones and a Zinc-finger homeobox 1B-encoding clone). Sry-related, HMG box-containing, SOX5-encoding clones showed strong immunoreactivity with the glioma patient sera (Fig. 1). Thus, SOX5 was further analyzed for its immunogenicity and expression. Sequence analysis of the isolated cDNA clones revealed that the SOX5 cDNA contained the full-length (1 kb, 347 amino acids) SOX5 cDNA sequence (Gene accession number: NM006940) with an open reading frame. No mutations were detected among the SOX5 clones.
SEREX analysis of IgG antibodies against SOX5 in sera from healthy individuals and patients with glioma or various other cancers
To examine the relationship between the antibody response against SOX5 and glioma, sera from various patients and healthy individuals was screened. Sera from 29 patients with glioma (12 glioblastomas, 9 anaplastic astrocytomas, 2 anaplastic oligoastrocytomas, 1 oligoastrocytoma, 5 diffuse astrocytomas; 3–77 years old, average 43.8 years; 13 men and 16 women), 14 patients with other brain diseases (including meningioma, malignant lymphoma, metastatic brain tumor, Parkinson disease, craniophalyngioma, central neurocytoma, brain abscess and subarachnoid hemorrhage), 54 patients with various cancers and 37 healthy individuals (20–83 years old, average 52.3 years; 20 men and 17 women) were diluted 1:100 and used to screen for IgG antibodies to the SOX5 antigen, which was expressed by λ phage-infected E. coli, on a nitrocellulose membrane. Serological responses against SOX5 were observed in samples from 8/29 (27.6%) glioma patients, whereas 0/14 (0%) samples from patients with other brain diseases, including inflammatory diseases, and 1/54 (1.9%) samples from patients with other cancers exhibited serological responses. In the sera from 37 normal adults, IgG antibody against SOX5 was not detected except in one sample from an elderly woman (Table I). Most of the SOX5-positive sera from glioma patients also showed positive reactions when diluted 1:1,000 (Fig. 1), suggesting that a high titer of IgG antibodies against SOX5 were present in the glioma patients' sera.
Table I. The Presence of IgG Antibodies Against SOX51
Values inside parentheses indicate percentages.
The presence of IgG antibodies against SOX5 was evaluated by screening λ phage infected E. coli on nitrocellulose membranes with sera at 1:100 dilution.
2Other brain diseases included 4 meningiomas, 2 malignant lymphomas, 2 metastatic brain tumors, 2 Parkinson diseases, craniophalyngioma, central neurocytoma, brain abscess and subarachnoid hemorrhage.
To evaluate the expression of the SOX5 gene in normal tissues, we performed a quantitative RT-PCR analysis using SYBR Green. The PCR product was confirmed the specificity by sequencing (data not shown). The SOX5 gene was highly expressed in adult testis tissues, compared with the level in adult brain tissue, but not in other normal adult tissues (Fig. 2).
Preferential expression of SOX5 protein in glioma
To investigate the specificity of the anti-SOX5 polyclonal antibody used in this study, we performed a Western blot analysis after neutralizing the antibody with recombinant SOX5 protein. The 40-kDa band detected in 293T cell transfected with SOX5 cDNA and in glioma tissues disappeared in this blocking experiment (Fig. 3*). In addition, the anti-SOX5 antibody did not recognize the same Sox Group D protein, SOX6 (data not shown). These results indicate that the antibody specifically recognizes SOX5 and confirming the expression of SOX5 in glioma tissues. Then, the expression of the SOX5 protein in normal tissues, glioma tissues, and glioma cells was evaluated using the anti-SOX5 antibody. The SOX5 protein was expressed in glioma tissues (GB1, GB5, AA1 and DA2: 4 of the 6 glioma patients who were analyzed) and glioma cell lines (U87 and Marcus), but not in normal adult tissues, except the testis (Fig. 3). Moreover, SOX5 expression was confirmed in tumor tissues from glioma patients (cases GB1, AA1, and DA2) whose sera reacted with SOX5, whereas the SOX5 protein was not expressed in tumor tissue from a glioma patient (case GB6) whose serum was SOX5 IgG-negative (data not shown).
Immunohistochemical analysis of SOX6 in glioma
Sectioned materials from formalin-fixed tumors were analyzed using the antibody against SOX5. Two glioma tissues (a glioblastoma and a diffuse astrocytoma) expressed SOX5 in the nuclei of the tumor cells, but only a few SOX5-positive cells were detected in non-neoplastic tissue from the cerebral cortex (Fig. 4).
Presence of serum SOX5 antibodies in glioma patients
We examined the relationship between the presence of SOX5 IgGs in sera and the clinicopathological features of glioma patients (Table II). Glioma patients under 30 years of age had SOX5 IgG antibodies significantly more frequently than those over 30 years of age. These results suggest that immune responses against SOX5 may occur in younger glioma patients. Although we also evaluated the possibility of a relationship between the presence of SOX5-specific IgGs and KPS score, completeness of tumor resection, or gender, no apparent correlation was found. Regarding the serological responses in different WHO grades of gliomas, IgG antibodies for SOX5 were detected in 4 of the 12 (33.3%) patients with WHO grade IV glioma, 3 of the 11 (27.3%) patients with grade III glioma and 1 of the 6 (16.6%) patients with WHO grade II glioma (Table I). Although sera from malignant glioma patients (WHO grades III and IV) showed a higher rate of SOX5-reactivity than those from WHO grade II glioma patients, no statistically significant correlation was observed between the immunogenicity of SOX5 and the grade of malignancy in gliomas (Table II).
Table II. Relationship Between the Presence of SOX5 IgG and Clinical Variables IN 29 Patients with Glioma
Prognostic value of SOX5 immune response in GBM patients
To evaluate whether SOX5 antibody responses might predict the outcome of GBM patients (12 GBM patients among the 29 glioma patients who were analyzed; patient age range, 7–69 years; average patient age, 44.3 years; 4 men and 8 women), we compared the overall survivals of patients with or without antibodies against SOX5 IgG. The SOX5 IgG-(+) patients had a longer survival period than the SOX5 IgG-(−) group; this difference was statistically significant (P = 0.028) (Fig. 5). We next examined the associations between the presence of SOX5-specific IgGs and overall survival in GBM patients using the Cox regression analysis. When adjusted for the KPS score, sex, and the completeness of tumor resection, anti-SOX5 IgG was statistically significant (P= 0.034), with a hazard ratio of 27.8 (95%CI, 1.3–596.5) (Table III). This model could not be adjusted for age because anti-SOX5 IgG was highly correlated with age. Therefore, the regression coefficient estimates were unstable and not accurate measures, a phenomenon known as colinearity.
Table III. Results of Cox Regression Analysis of Favorite Outcome in Patients with GBM
Absence of anti-SOX5 IgG
KPS score ≤60
Gross total resection
Although several glioma antigens have recently been isolated using SEREX,3, 4, 10 these isolated antigens have shown broad gene expressions; thus, it is unclear how immune responses against these antigens could be induced in glioma patients or whether these immune responses are specific for glioma patients. In this study, the presence of IgG antibodies against SOX5 in sera was predominantly observed in glioma patients; in fact, seroreactivity against SOX5 was observed in over one-fourth of the glioma patients. A variety of mechanisms that may cause an immune response against tumor antigens have been discussed, including overexpression, mutations and post-translational modifications of the protein.11, 12, 13 The present study revealed that SOX5 expression was very weak in normal adult tissues, except for the testis––an immunologically privileged anatomical site, but was highly expressed in gliomas. In addition, no mutations in the SOX5 gene were observed by sequence analysis (data not shown). These results suggest that the selective, high expression of SOX5 in gliomas may cause specific immune responses in glioma patients.
The SOX family of high-mobility group (HMG) domain proteins has recently been recognized as a key player in the regulation of embryonic development and in the determination of cell fate.14, 15, 16, 17, 18, 19, 20, 21, 22 Because SOX HMG domains share a number of conserved amino acid residues, SOX proteins were initially classified by the deduced amino acid sequence of the HMG domain and were grouped into groups A–G.16, 21 Within an individual group, the amino acid sequence identity of the HMG domain is 90%, although it decreases to 60% between distant groups. Because SOX5 and SOX6 are grouped in the same Group D, considerable homology occurs between the HMG domain of SOX5 (SOX5-HMG) and that of SOX6 (SOX6-HMG). We previously demonstrated that SOX6-HMG may be an immunogenic epitope recognized by IgG antibodies in sera from glioma patients.5 These results suggest that SOX5-HMG also may be an immunogenic epitope recognized by glioma patients' IgG.
To confirm that SOX5 expression was distinct from SOX6 expression, we designed a PCR product sequence for SOX5 with a low homology (42.3%) to SOX6 and used an anti-human SOX5 polyclonal antibody raised against a synthesized peptide with a low homology (33.3%) to SOX6. Indeed, we demonstrated that the anti-SOX5 antibody recognized SOX5 protein, but not SOX6 protein. In this study, we showed that human SOX5 was expressed in gliomas, but not in adult brain, similar to the expression pattern of SOX6. An individual SOX protein appears to interact selectively with and to regulate a unique repertoire of target genes by pairing off with specific partners to regulate gene transcription. The partner factors of Group D SOX proteins appear to be the SOX proteins in the same group. Dimerization of SOX5 and SOX6 greatly increases the binding efficiency of the 2 SOX proteins to DNA that contains adjacent SOX sites.8, 23 Indeed, mouse Sox5, Sox6 and Sox9 are expressed simultaneously and at high levels from the early stages of chondrogenesis in all cartilaginous sites in the mouse embryo, but expression significantly decreases in adult tissues.8 Mouse Sox5 and Sox6 are expressed together during the development of the CNS.24 These results suggest that this partnering might allow Group D SOX proteins to act in a cell-specific manner, which would be key to their role in cell differentiation or tumorgenesis in the CNS.
To evaluate tumor antigens for diagnostic and therapeutic purposes, accurate information on the protein expression pattern is essential. We examined SOX5 expression using Western blot analysis in glioma tissues, glioma cells and normal tissues. SOX5 expression was frequently observed in gliomas, but no apparent expression was detected in the normal adult tissues, except testis. Moreover, SOX5 expression was confirmed in tumor tissues from patients whose sera contained IgG antibodies against SOX5 (Fig. 3). The fact that a glioma antigen, SOX5, induces IgG responses in over one-fourth of glioma patients implies the coexistence of cognate T-cell help. Indeed, several antigens, including MAGE, tyrosinase and NY-ESO-1, that were isolated using SEREX were originally identified by CTL responses.25, 26, 27 The specific IgG responses against SOX5 are likely to be caused by its predominant expression in gliomas. These results suggest that SOX5 may be a useful target for immunotherapy directed against gliomas.
So far, limited information has been available concerning the utility of serum antibodies against these tumor antigens as diagnostic and prognostic markers. In the present study, we evaluated the clinical features of SOX5 IgG-positive glioma patients. No apparent correlation between the immunoreactivity of SOX5 and the grade of glioma malignancy was seen. In contrast, a statistically significant relationship was observed between patient age and the presence of SOX5 IgGs in glioma patients (p= 0.024). It has been reported that an increase in CD4+ CD25high regulatory T cells in peripheral blood is associated with ageing.28 The titers of influenza-specific antibodies after immunization in a mouse model have also been shown to be lower in old mice, compared to that in young mice.29 Furthermore, thymic T cell output is reportedly reduced with aging, and this reduction is associated with impaired antitumor immunity in glioblastoma patients.30 Stronger immune responses against SOX5 might occur in younger glioma patients, compared to those in elderly patients, because of the low suppression of immunity in the young.
Little is known about prognostic factors for glioma patients, other than clinical and histological features like patient age, Karnofsky performance status and completeness of resection.31, 32 Several molecular markers have been proposed, including LOH of chromosome 10 and the overexpression of EGFR––both of which are correlated with a poor prognosis.33, 34, 35 To learn more about the roles of SOX5 antibody responses in GBM patients, we compared the survival of antibody-positive patients with that of antibody-negative patients. Kaplan–Meier plotting showed that the survival of GBM patients with SOX5 IgG was significantly prolonged, compared to that of SOX5 IgG-negative patients. A Cox regression analysis also revealed the presence of anti-SOX5 IgG as an independent prognostic factor, when adjusted for the KPS score, sex, and the completeness of tumor resection, although this model could not be adjusted for age because of the colinearity. These results suggest that an immune response against SOX5 may play a role in the maintenance of a tumor-free status after treatment. Further analysis regarding immune responses against SOX5 in glioma patients are needed to confirm this hypothesis. If immune responses against gliomas are associated with a good prognosis, SOX5 may be one such target antigen. Therefore, the glioma antigen, SOX5, may be useful not only as a diagnostic or a prognostic marker, but also as a good candidate antigen for immunotherapy, particularly for GBM patients with sera that are positive for SOX5 antibodies.
The authors thank Dr. Toru Takebayashi (Department of Preventive Medicine and Public Health, Keio University School of Medicine, Tokyo) and Dr. Yukinori Kurokawa (Statistics and Cancer Control division, National Cancer Center, Tokyo) for performing the statistical analysis.