A SEREX collaborative group was established in 1996 by the Ludwig Institute for Cancer Research, involving investigators at the University of Saarland (Homburg, Germany), Ludwig Institute Branches in New York, Melbourne, and London (University College), Aichi Cancer Center (Japan), Krankenhaus Nordwest (Frankfurt, Germany), and Moscow State University (Russia). A database (150) incorporating SEREX data has been organized by Victor Jongeneel (Director of Information Technology, Ludwig Institute for Cancer Research), and 2593 sequences derived from 2169 clones have been deposited at last count (February 2004). Through the efforts of this group and that of many other independent researchers, multiple SEREX studies have been performed since its introduction in 1995 (19). The following section is an attempt to summarize all SEREX data available to date in an "all-inclusive" manner, so as to help construct a comprehensive immunological profile of human cancer (cancer immunome). It is almost inevitable that certain studies will be inadvertently missed and I would appreciate notification so that these studies can be included in future updates of this review.

Melanoma

Several SEREX studies have focused on melanoma tumor specimens, melanoma cell lines, and sera from melanoma patients (19, 22, 25, 89, 90, 91). In their initial analysis, Sahin et al. (19) screened a melanoma library of 1.0x10⁶ clones and isolated 40 positive clones representing ten different gene products. These included tyrosinase and two CT antigens, MAGE-1 and HOM-MEL-40/SSX2. To extend the search for CT antigens recognized by melanoma patient sera, a testicular cDNA library was screened by Güre et al. (22). This resulted in the isolation of eight gene products, including six universally expressed genes and two members of the SSX family, SSX2 and a new member SSX3. Southern blot analysis confirmed SSX as a multigene family, and subsequent PCR-based cDNA cloning led to the identification of two additional SSX members, SSX4 and SSX5, both transcribed in testis and in malignancy.

To further extend this search for CT antigens, a cDNA library was constructed from SK-MEL-37, a melanoma cell line expressing multiple CT antigens, and the library was screened with an allogeneic melanoma patient serum known to have antibodies to two CT antigens, NY-ESO-1 and MAGE-1 (25). Sixty-one positive clones were identified, including four CT antigen genes: MAGE-4a, NY-ESO-1, LAGE-1 [a gene closely related to NY-ESO-1 (28, 41)] and a new CT antigen CT7. Additionally, the most abundantly represented clones in this screening, 33 of 61 positive clones, were derived from three closely related genes. One of these three genes, called KOC (KH domain-containing gene overexpressed in cancer), had been previously identified as a gene overexpressed in pancreatic cancer (42). These three genes were designated KOC1 (the original KOC gene), KOC2, and KOC3, of which KOC3 was the most abundantly expressed (30 of 33 isolated clones). Northern blot analysis of KOC3 expression showed variable levels of mRNA expression in melanoma cell lines and no detectable mRNA in any normal tissue examined. Subsequently, KOC3 was also cloned from a bladder cancer line by antibody screening using serum from a hepatocellular carcinoma patient (43) (see below).

Three additional SEREX analyses of melanoma have appeared in the literature since the original publication of this review (2000). Jäger et al. (89) isolated 43 genes through autologous screening of a melanoma cell line, including a novel gene that encoded a melanocyte-specific rab GTP-binding protein, designated Rab38, and a chromosome condensation protein, hCAP-G. Mollick et al. (90) performed SEREX using serum from a melanoma patient who had been vaccinated with autologous whole cell vaccine, and identified an antigen that was a putative opioid growth factor receptor (OGFr). It is of interest that OGFr contains a MUC1-like tandem repeat, raising the possibility that this might have been related to its immunogenicity. Ehlken et al. (91) also utilized post-vaccination sera from patients that have received cytokine-transduced whole cell vaccines. A testicular cDNA library was screened, and 27 genes were identified, including two known CT antigens SCP-1 and PLU-1.

Esophageal Cancer

An esophageal cancer cDNA library derived from a squamous cell carcinoma has been analyzed (28). Thirteen positive clones were identified, derived from eight different genes, designated NY-ESO-1 through NY-ESO-8. Among these, NY-ESO-1 showed a characteristic cancer-testis expression pattern, and NY-ESO-5 appeared to be preferentially expressed in squamous epithelium. Other genes were found to be universally expressed, including two genes coding for autoimmune antigens in the U1 small nuclear ribonucleoprotein family. NY-ESO-1 was subsequently isolated from SEREX studies of melanoma, breast cancer, prostate cancer, ovarian cancer, and sarcoma (see related sections in this review) and has been recognized as one of the most immunogenic tumor antigens known to date.

Colon Cancer

SEREX analysis of four colon cancer cDNA libraries has been carried out by three groups to date (29, 92, 93). In the initial study of Scanlan et al. (29), a total of 234 immunoreactive cDNA clones encoding 48 different antigens were identified. Sequence analysis revealed 17 novel genes and 31 known genes coding for a vast array of cellular components. A large proportion of the antigens were nuclear proteins, such as transcription factors, mRNA splicing factors, and DNA-binding proteins, and a smaller percentage represented metabolic enzymes, molecular chaperones, signaling molecules, cytoskeletal proteins, and membrane-associated proteins. In addition to these common cellular constituents, however, several of the colon cancer antigens stood out as having a known or suspected etiologic association with human cancer; the most obvious example of this being the isolation of a mutated version of p53 tumor-suppressor gene (NY-CO-13). RT-PCR analysis and Northern blotting showed that transcripts encoding 3 of the 48 antigens were differentially expressed, whereas the other 45 genes were universally expressed. NY-CO-27 is identical to the S-type lectin galectin-4 (44), and is expressed primarily in normal colon and small intestine. The other two differentially expressed antigens, NY-CO-37 and NY-CO-38, are related isoforms encoded by a previously unknown gene that maps to chromosome 11p15.4-p15.1 (45). These two antigens are members of a group of differentially expressed isoforms that are characterized by a variable number of PDZ domains, the presence or absence of a PEST protein degradation motif, a large coiled-coil domain, and the existence of a putative C-terminal module that may function as a binding site for PDZ domains (indicating that it may form homomeric or heteromeric protein complexes). In general, PDZ domains function as scaffolding sites for the organization of signal transduction complexes, cell junctions, and cytoskeletal-plasma membrane linkages (46). A yeast two-hybrid screen of proteins that interact with the Coxsackie and adenovirus receptor identified NY-CO-38 (or a related isoform) as an interacting protein (152). One of the NY-CO-38 isoforms is expressed in some normal tissues but is not found in normal colon. In contrast, this isoform is expressed in colon cancer, and this aberrant expression of a normally expressed isoform in cancer may be the basis for its immunogenicity in cancer patients (see Mutational Antigens).

Two additional studies on colon cancer have been performed by Line et al. (92) and Ishikawa et al. (93). Eight gene products were isolated by Line et al., including one gene (AD034) showing a 32 bp frameshift insertion, and another gene (RHAMM) that have also been identified in other SEREX studies, e.g. in leukemia (see below). This gene appeared to be overexpressed in tumor tissue. The study of Ishikawa et al. is intriguing in that they specifically targeted colon cancer cell lines with known microsatellite instability (MSI+), screening with serum from a patient with MSI(+) colon cancer. Sixty-four genes were isolated, one of them, CDX2, carried a frameshift mutation in the microsatellite sequences within its coding region, indicating that the immune response was raised against tumor-specific amino acid sequences generated through MSI.

Gastric Cancer

SEREX analysis of five cases of gastric carcinoma derived from tumors of various histologic types and grades has been carried out by Obata et al. (34). The screening of each cDNA library with autologous serum was continued until approximately 50 positive clones were isolated from each library. In total, 297 positive clones were obtained, representing 135 distinct genes, 87 of which were previously known. Of the 135 gene products, 21 were identified in two or more gastric cancer libraries and 24 were identified in SEREX analysis of other tumor types. No CT antigens were isolated, consistent with the low frequency of CT gene expression in gastrointestinal cancer (28, 47).

Of the genes isolated, two showed possible etiologic significance for gastric cancer. One was derived from a fusion gene product between E-cadherin (E-Cad) and a novel gene designated as GeneY. E-Cad is an adhesion molecule involved in the regulation of various cellular functions, including normal differentiation and tumor invasion (48). Mutations in E-Cad have been identified in gastric and other cancers and inherited mutations have been related to familial gastric cancer (49, 50). Nine independent cDNA clones encompassing the fusion gene were isolated and the combined sequencing data indicated that the 5' end of E-Cad was fused to the 3' end of GeneY. RT-PCR using a 5' primer derived from E-Cad and a 3' primer derived from GeneY showed that the amplification product was restricted to the tumor and was not detected in autologous nonneoplastic tissue, or allogeneic normal or tumor tissues. This finding indicates a somatic translocation event involving E-Cad and GeneY in the cancer, an event possibly contributing to the cancer's origin or progression. Although this translocation event was not found in five other cases of gastric cancer in this small series or reported in the literature, a larger panel of gastric cancer specimens should be evaluated to assess the frequency and significance of this genetic alteration.

A second gene, recognized by the serum of a different gastric cancer patient and also related to gastric carcinogenesis, is the AKT1 (PKB) oncogene. The AKT1 gene is thought to promote cell survival by modulating antiapoptotic signals, and AKT1 gene amplification has been reported in a primary gastric cancer (51). AKT1 expression was elevated in five of eight gastric cancers, and one of the five patients with amplified AKT1 expression had an anti-AKT1 antibody response. Overexpression of this gene presumably forms the basis for its immunogenicity in cancer patients.

Another SEREX study was performed by Cho et al. (147). In this study, cDNA libraries from testis and two gastric cancer cell lines were screened, leading to the identification of a CT antigen CAGE. This gene, encoding a DEAD box helicase protein, was found to be an intronless gene on Xp22, sharing homology with the helicase proteins p72, p68, and HAGE. HAGE, identified by Martelange et al. (148) by RDA of genes specifically expressed in sarcoma, is also a CT gene, localized to chromosome 6.

Renal Cancer

In the initial study by Sahin et al., five different antigens were isolated from a single renal cell carcinoma cDNA library screened with autologous serum (19). Of these, HOM-RCC-3.1.3 is a type I integral-membrane protein (40 kDa) representing a novel tumor-associated carbonic anhydrase (CA XII). The coding gene for CA XII maps to chromosome 15q22 (52). CA XII is expressed in normal kidney, small intestine, and renal cancer. In 10% of the renal cancer examined, HOM-RCC-3.1.3 mRNA was expressed at a much higher level than in adjacent normal kidney tissue (19, 52).

In a subsequent study by Türeci et al. (23), serum from a patient with renal cancer was used to screen a testicular library subtracted with a range of normal tissues. The aim of the screening was to uncover new members of the CT family. Five positive clones representing three genes were isolated, one encoding synaptonemal complex protein 1 (SCP1), a protein involved in chromosome reduction during meiosis. The expression of SCP1 is normally restricted to germ cells and was not previously known to be expressed in cancer. However, Türeci et al. found SCP1 mRNA and protein expression in a subset of malignant gliomas, breast cancers, renal cancers, and ovarian cancers. Thus, SCP1 belongs to the CT antigen category.

Four additional cases of renal cancer were analyzed in SEREX by Scanlan et al. (33) using autologous patient sera. These studies yielded a total of 169 clones representing 65 different gene products. Sequence analysis showed that 36 were coded by known genes and 29 were novel gene products. Four of the antigens, NY-REN-9, -10, -19, and -26, have a known association with human cancer. NY-REN-9 (LUCA-15) and NY-REN-10 (gene 21) map to the p21.3 locus on chromosome 3, a region presumably containing tumor suppressor gene(s), which is often deleted in small-cell and non-small-cell lung cancer (53) and renal cancer (53, 54). NY-REN-19 is identical to the LKB/STK11 serine-threonine kinase. Defects in the LKB/STK11 gene are responsible for Peutz-Jeghers syndrome, a disease characterized in part by an increased risk of gastrointestinal cancer (55). Somatic mutations in the LKB1/ STK11 gene have also been reported in cases of colon cancer (56), breast cancer (57), and gastric cancer (58). NY-REN-26 is encoded by the bcr gene involved in the t(9;22) bcr/abl translocation associated with chronic myelogenous leukemia (59). No evidence of gene mutation was detected in the cDNA sequences of clones defining NY-REN-9, -10, -19, or -26. In addition to these four genes, two others, NY-REN-33 and NY-REN-65, are related to less well-characterized potential tumor suppressor genes. NY-REN-33 is identical to a possible tumor suppressor, SNC6 (GenBank Accession No. U28918), which maps to 22q13.1, a region known to be deleted in cases of breast cancer (60). NY-REN-65 maps to 11q12 and is a possible human homologue of the rat HREV107 tumor suppressor gene (GenBank Accession No. X92814).

With regard to tissue mRNA expression, transcripts encoding all 65 antigens are expressed in a range of normal tissues as determined by expressed sequence tag analysis, RT-PCR, and/or Northern blotting. However, three antigens, NY-REN-3, NY-REN-21, and NY-REN-43, have a differential mRNA expression pattern. NY-REN-3 is identical to NY-CO-38 (described above), belonging to a family of differentially expressed isoforms with PDZ domains. Both NY-REN-21 and NY-REN-43 transcripts are expressed at higher levels in testis than in other normal tissues. NY-REN-21 represents a possible human homologue of mouse zinc finger protein-38 (61), a DNA-binding protein thought to be involved in meiosis, and NY-REN-43 cDNA encodes a putative protein containing a ring-zinc finger domain and a polyserine domain.

More recently, another SEREX study has appeared in the literature, in which 66 genes were isolated (94). Few details, however, are known regarding the nature of these genes.

Lung Cancer

Several SEREX studies of lung cancer have been reported, both on non-small cell lung cancer (30, 31, 32, 95, 96, 97) and on small cell lung cancer (98). Brass et al. (30, 31) derived 35 positive clones representing 19 genes from the autologous screening of a squamous cell carcinoma cDNA library. Six clones coded for eIF-4 gamma, a eukaryotic translation initiation factor, which maps to chromosome 3q26.1-3q26.3, a region that is amplified in 30% of squamous lung cancer (62). By comparative PCR analysis, the authors showed that eIF-4 gamma and eight other genes were amplified in the tumor of origin, including DnaJ heat shock protein and Jk-recombinant signal binding protein, two genes also found in the SEREX analysis of renal cancer (NY-REN-14, NY-REN-30) (33). Three of the amplified genes were located on chromosome 3, and two of these mapped to the 3q region. The authors suggested that immune recognition is a consequence of amplified gene expression, and that amplified eIF-4 gamma may be important in the development of squamous cell carcinoma.

In a separate study, Güre et al. (32) analyzed a moderately differentiated adenocarcinoma of the lung with autologous serum and isolated 20 positive clones representing 12 genes. One of these was aldolase A, a gene known to be expressed at high levels in most lung cancer (63, 64). Aldolase A has subsequently been isolated in SEREX analysis of breast cancer (35), indicating a possible linkage of aldolase A overexpression to adenocarcinomas of various tissue origin. Another SEREX-defined gene of special interest, NY-LU-12, maps to the tumor suppressor gene locus on chromosome 3p21.3, as do NY-REN-9 and NY-REN-10 from renal cancer (see above). NY-LU-12 encodes a nuclear zinc finger protein with two RNA-binding domains. To date, no mutations have been found in the NY-LU-12, NY-REN-9, or NY-REN-10 coding genes. What is the basis for the immunogenicity of these 3p antigens? Further sequencing is necessary to exclude the possibility of mutation as the immune stimulus. Another possibility is that loss or downregulated expression of a gene product such as allelic 3p deletion can result in an immune response, just as gain or overexpression can be an immunogenic event.

In addition to the published studies, Güre et al. (153) have screened three additional non-small cell lung cancers. These three screenings yielded only eight positive clones from a total of more than 2x10⁶ bacteriophages, and the clones represented two known genes and two unknown genes, all of them widely expressed in normal tissues. Low-yield SEREX screening is, in fact, not uncommonly encountered, most likely reflecting low tumor immunogenicity or low host immune reactivity, or both. Such essentially negative studies should be recognized because they are unlikely to be published and are therefore underrepresented in the literature.

Following these earlier studies, non-small cell lung cancer has been examined by a few groups (95, 96, 97). These studies have defined a new CT antigen CAGE-1 (96) and defined previously unknown transcript variants of another CT antigen, XAGE-1 (97). Two other genes known to be associated with oncogenesis, namely TP53BP and LBC (lymphoid blast crisis oncogene), were also isolated (95).

Regarding small cell lung cancer (SCLC), Güre et al. (98) analyzed two SCLC cell line cDNA libraries with a pool of 5 SCLC sera. Fourteen genes were defined, including 4 SOX group B genes (SOX1, SOX2, SOX3, and SOX21) and ZIC2. SOX and ZIC2 genes are all transcriptional factors that are expressed at early developmental stages in the embryonic nervous system, but are downregulated in the adult. These genes were found to be expressed in high frequency in SCLC, and were often associated with the emergence of high-titer antibodies in these patients. Whether these strong immune responses are clinically relevant is intriguing, and studies to evaluate this are ongoing.

Breast Cancer

Several SEREX studies have been performed on breast cancer (35, 36, 37, 99, 100, 101, 102). In the first of their series of three studies, Jäger et al. (35) analyzed one breast cancer library with two separate serum sources - autologous serum and a pool of seven allogeneic sera from breast cancer patients. Additionally, a testicular library was also screened with the autologous serum. These analyses led to the isolation of 89 clones derived from 30 different genes, 27 known and 3 unknown. Among the known genes were two encoding CT antigens, NY-ESO-1 and SSX2, and these were prominently represented, constituting 31 (14 NY-ESO-1, 17 SSX2) of the 89 positive clones. In addition to these two genes, a candidate breast tumor suppressor gene, ING1, was isolated from both the breast cancer library and the testicular library. ING1 was initially identified in the cloning of genes preferentially expressed in a normal breast epithelial cell line but not in breast cancer lines; transfection of a breast cancer cell line with the ING1 gene was shown to lead to growth inhibition (65). No mutation has been found in the SEREX-defined ING1 gene from the breast cancer specimen. By analyzing cDNA clones obtained by hybridization screening of the testicular library, however, three splice variants of ING1 (variants A, B, and C) were identified. These three variants differed in their 5' sequences, encoding putative products with different amino terminal sequences. By RT-PCR analysis, only variant A showed universal expression in all tissues examined. Variants B and C were expressed in a tissue-specific fashion, whereas variant C showed expression almost exclusively in the testis. Among tumor cell lines, variant C was silent in six breast cell lines tested, and variant B, although not expressed in normal breast, was weakly expressed in four of six breast cancer cell lines tested. As discussed above in relation to NY-CO-37/38, perturbation in the tumor expression of ING1 splice variants may account for immune recognition in the human host.

In their second study, Jäger et al. (99) identified seven genes, including a new breast differentiation antigen NY-BR-1. This gene was found to be expressed only in breast and testis, but not in any other normal tissues examined. Among breast cancer specimens, NY-BR-1 was expressed in 21 of 25 tumors examined, making NY-BR-1 a potential cancer vaccine target in this patient group. This gene was also isolated in the third study from this group (100), further confirming its immunogenicity. In this last study, the authors also found the SCP-1 gene, as well as another gene with strong expression in testis, designated NW-BR-3. This gene, however, showed weak expression in 5 of 16 somatic tissues tested by RT-PCR, and was thus described as a "CT-like" gene.

A large scale SEREX on breast cancer was performed by Scanlan et al. (36). Multiple rounds of SEREX were performed, leading to a total of 94 distinct gene products. The serological profiles of these antigens against sera from cancer patients and normal individuals were evaluated, and 40 of the antigens reacted exclusively with sera from cancer patients. Genes identified in this series included NY-ESO-1, MAGE-3, MAGE-6, p53, and the oncogene product HER2/neu, as well as several new genes with tissue-restricted expression pattern, e.g., kinesin 2, NY-BR-62, and NY-BR-85.

Obata et al. (37) performed autologous SEREX screenings of two breast cancers in a Japanese patient group and isolated 94 reactive clones. These clones represented 55 different genes, 34 known and 21 unknown. Eighteen (33%) of the 55 were also identified by previous SEREX analysis of other breast cancers or other cancers. Among the gene products identified in these two screenings were the CT antigen SSX2, the heat shock protein hsp105 (nine clones), and the guanylate binding protein isoform II (five clones).

Two more recent studies were performed by Forti et al. (101) and Minenkova et al. (102). Genes of potential interest, as highlighted by the authors in the respective studies, included fibulin-1 and thyroid hormone-binding protein (101), as well as topoisomerase IIB, MUC3, Golgin p245, etc. (102). Possible CD4+ T cell responses to fibulin-1 were subsequently described by the same group (see below).

Prostate Cancer

SEREX analyses of prostate cancer with autologous sera have been carried out by Chen et al. (38) and by Obata et al. (39). By autologous screening of prostate cancer, Chen et al. isolated 19 positive clones derived from 18 genes (10 known and 8 unknown), including genes encoding DNA replication licensing factor beta subunit, acid finger protein, and scaffold attachment factor. The same serum was also used to screen a testicular library, and 15 genes were isolated. One of the genes was hMLH1, human DNA mismatch repair protein homolog, which had been shown to be mutated in various human cancers. cDNA cloning of this gene from the original tumor specimen, however, failed to identify any mutations. Using the same autologous screening approach, Obata et al. isolated more than 50 positive clones. Among the known gene products were eIF-4gamma and NF-kappaB p105. Another gene of interest in this group was the HERV-K gag gene. HERV-K endogenous retroviral sequences are abundant in the human genome, and some of them have been shown to be expressed in normal and tumor tisues (see Viral Antigens).

SEREX on prostate cancer has also been performed by Zhou et al. (103) and Fossa et al. (104). In the latter study, 13 genes were isolated by autologous SEREX, including three CT antigens - NY-ESO-1, LAGE-1, and XAGE-1. It is worth mentioning that this study used the pJuFo phage surface display system, allowing display of recombinant proteins on M13 filamentous phage. The screening is thus a combined biopanning and immunoscreening approach, making future high throughput analysis a feasible possibility.

Hepatic Cancer

Three autologous SEREX studies have been performed on hepatocellular carcinoma (105, 106, 107). Stenner-Liewen et al. (105) isolated 19 genes, all ubiquitously expressed in tumor and normal tissues, including albumin, LDH, and kinectin. Wang et al. (106) analyzed 4 sera from hepatocellular carcinoma patients in China, and 55 gene products were identified. Of these, the two most interesting genes were HCA587 and HCA661, both showing a cancer/testis restricted expression pattern. HCA587 was indeed found to be identical to CT10/MAGE-C2, a CT gene previously identified by RDA, thus further confirming the immunogenicity of this CT antigen. Uemura et al. (107) analyzed two hepatocellular carcinoma patients in Japan, and several tumor-associated genes were isolated, including SART1, ROCK-1, gamma catenin, heat-shock protein, etc.

A variation of the general SEREX approach was taken by Zhang et al. (43), who carried out an analysis of the T24 bladder cancer cell line with serum from a patient with hepatocellular carcinoma. The KOC3 gene, a previously identified SEREX-defined antigen in melanoma, was isolated (see above). This indicates that KOC-related gene products are overexpressed in more than one tumor type and are often immunogenic.

Ovarian Cancer

Naora et al. (108) identified the homeobox gene HOXA7 as an immunogenic tumor antigen in epithelial ovarian tumors by SEREX. It was found to be associated with tumors with mullerian-like differentiation, and its expression often led to antibody production in these patients. This antibody response appeared to be highly specific in ovarian cancer patients, and its possible diagnostic value was raised by the authors.

In a separate study, Stone et al. (109) extensively studied sera from 25 late-stage ovarian cancer patients, screening against 3 cDNA libraries. Nine antigens were identified in more than one patient, and these included p53, NY-ESO-1, TOP2A, HOXB6, among others. One intriguing observation was that 4 antigens identified in this screening mapped to chromosome 17q, suggesting that cytogenetic changes involving this region might be the basis for immunogenicity.

Sarcoma

Lee et al. (110) selected two synovial sarcoma patients with serological response to NY-ESO-1, and used their sera to screen two sarcoma cell line libraries. In addition, a fibrosarcoma patient serum was used to screen a testis library. This resulted in the identification of 113 genes, including CT antigens LAGE-1 and SSX1, 2, 3, and 4. A new CT antigen NY-SAR-35 was also defined, which mapped to chromosome Xq28.

The only other sarcoma that has been evaluated to date is osteosarcoma (111). In this study, Nabeta et al. identified two proteins by autologous screening, namely HLA-Cw and smooth muscle myosin light chain.

Neuroblastoma

In collaboration with M. K. Brenner's group (Baylor College of Medicine, Houston, TX), our group has performed an allogeneic SEREX screening on a neuroblastoma cell line library with serum from a neuroblastoma patient that had received autologous tumor vaccine. Eleven genes were identified, four of which were found to cluster on chromosome 17q21-23, mirroring the findings in ovarian cancer (109). Since 17q21 gain has been found in 63-83% of neuroblastomas (112, 113, 114), gene amplification and subsequent overexpression is the likely mechanism for the immunogenicity of these genes. Among these four genes, TOP2A (topoisomerase II alpha) represented 38 of the 43 SEREX-reactive clones, suggesting that it is abundantly expressed in this tumor. It is known that chromosome 17q gain is a poor prognostic indicator in neuroblastoma, but the culprit gene has been elusive. The abundance of TOP2A transcripts in neuroblastoma, the independent isolation of TOP2A in ovarian cancer (109), as well as the clinical effectiveness of TOP2A inhibitors on neuroblastoma and other tumor types, all point to the possibility that TOP2A may be a biologically crucial gene in neuroblastoma and possibly other tumors, and a critical evaluation on this subject is warranted.

Autologous SEREX has also been performed on this tumor type by Behrends et al. (115). In this study, 10 antigens were defined, including neuronal antigens derived from the Hu genes and NNP-1. A 100-fold increase in the anti-Hu titer was observed in a patient that preceded the clinical diagnosis of recurrent disease.

Other Solid Tumors

In addition to the tumor types discussed above, other solid tumors have been studied, and one study each was found in the literature for astrocytoma (116), pancreatic cancer (117), mesothelioma (118), head and neck cancer (119), and medulloblastoma (120).

In the study of astrocytoma by Pfreundschuh and his colleagues (19, 116), sera from 18 astrocytoma and other gliomas were evaluated, and only 10 antigens were isolated, including testis-enhanced gene transcript (Tegt) (66), glial fibrillary acidic protein (GFAP), Bax-inhibitor 1, etc. This relatively low yield in comparison to other SEREX studies led the authors to conclude that astrocytomas only rarely elicited antibody responses and postulated that this might be related to immunosuppressive effects of these tumors.

A study of seminoma was also carried out by the same group of researchers (24). In that study, Türeci et al. (24) screened a cDNA library enriched for testis-specific transcripts with serum from a seminoma patient. Nine clones were isolated, derived from four known and two unknown genes. The known genes included human hook-1 protein and a mitotic protein. One of the two unknown genes, CT8/HOM-TES-85, was found to encode a new CT antigen. Sequence analysis revealed that the CT8 gene product is a leucine zipper protein. Of interest, the leucine zipper region of the CT8 protein shows an atypical amphipathy, a feature that has only been observed in the N-myc proto-oncogene (24).

Pancreatic cancer was studied by Nakatsura et al. (117). Eighteen genes were isolated. One of them, coactosin-like protein (CLP), was subsequently shown by the same group to be a target for cell-mediated immunity mediated through CD8+ T cells (121).

In mesothelioma, Robinson et al. (118) identified 8 antigens from an allogeneic screening of a mesothelioma cell line. One of the antigens, topoisomerase II beta (TOP2B), was shown to react with 13 of 14 sera from mesothelioma patients, in contrast to other antigens to which only one or two serum samples were shown to be positive. TOP2B has also been identified by SEREX of breast cancer (102).

Monji et al. (119) evaluated head and neck squamous cell carcinoma patient sera against a tumor cDNA library and a testicular library. The 19 genes identified included CT antigen MAGE-4, as well as two genes that might represent new CT genes. The characteristics of these two genes, KM-KN-1 and KM-HN-3, however, have not been described in detail.

In addition to neuroblastoma, another pediatric malignancy, medulloblastoma, was also studied by Behrends et al. (120). Fifteen gene products were isolated, including a proto-oncogene (rab18) and two candidate tumor suppressor genes (BAP1, PRDM13).

Hodgkin's and Non-Hodgkin's Lymphoma

A case of Hodgkin's disease was analyzed in the initial SEREX study of Sahin et al. (19). Of 1.0x10⁶ clones screened, 14 positive clones were identified, representing 4 genes. Of most interest was HOM-HD-21, which encodes a new member of the galactoside binding protein family, designated galectin-9 by Türeci et al. (21). This gene contains two lectin domains with galactoside binding capacity, and its expression was found to be restricted to peripheral blood leukocytes and lymphoid tissues. Serum reactivity to galectin-9 has been found to be restricted to patients with Hodgkin's disease, supporting a close association of galectin-9 with the disease. Another clone identified in this initial SEREX screening was a splice variant of the intermediate filament protein restin. Unlike seroreactivity to galectin-9, antibody to the restin splice variant was frequently found in normal individuals as well as in patients with Hodgkin's disease (19).

The same research group, headed by Pfreundschuh, also analyzed immune responses in non-Hodgkin's lymphoma (122) and in multiple myeloma (123) patients. Twenty-three genes were found in the lymphoma study, including genes homologous to NY-ESO-1 and SCP-1. In the myeloma study, myeloma proteins from 42 patients were screened against a testicular library, and a monoclonal IgA from a female patient was shown to target sperm-specific cylicin II.

Among other types of non-Hodgkin's lymphoma, Eichmuller et al. (124) analyzed cutaneous T cell lymphoma (CTCL) with SEREX. This study identified 15 genes, including the known CT gene SCP-1, as well as a new CT antigen cTAGE-1. cTAGE-1 mRNA was detected in 35% of CTCL tumors and 6/18 sera were found to be reactive to this SEREX clone.

Leukemia

In chronic myeloid leukemia, a SEREX screening with autologous serum identified 8 positive clones, corresponding to 6 known and 2 unknown genes (40). One of the two unknown genes, referred to as clone #4, was initially suggested to be a possible CT antigen. Additional data (125), however, indicated that this gene was also expressed in several somatic tissues, albeit at lower levels.

Greiner et al. have also published a series of SEREX studies on myeloid leukemias (126, 127, 128), including acute and chronic forms. The genes highlighted in their studies included the receptor for hyaluronic acid mediated motility RHAMM (127), M phase phosphoprotein 11 (MPP11), BRAP, and RBPJkappa (128). These genes have been isolated in many other SEREX screenings, as can be seen in the Cancer Immunome Database (150).

The application of SEREX to B cell neoplasms such as chronic lymphocytic leukemia is complicated by the fact that the neoplastic cells produce immunoglobulins (Igs), and cDNA clones that encode Ig sequences thus need to be distinguished from true sero-positive clones. Despite this difficulty, a study was performed by Krachhardt et al. (129) leading to the isolation of 14 antigens.

Adult T cell leukemia (ATL) has also been analyzed (130). In this study, Itoh et al. identified 10 genes, including a gene that was homologous to HTLV-1 U5RE binding protein 1 (HUB1). Serological response to this gene, however, was observed in ATL patients as well as healthy donors, indicating that HUB1 is an autoantigen.