SEREX was developed to combine serological analysis with antigen cloning techniques to identify human tumor antigens eliciting high-titer immunoglobulin G (IgG) antibodies. The SEREX technique is shown schematically in Figure 1. Although the concept behind SEREX is straightforward, a number of technical challenges needed resolution. One of the most crucial involved eliminating antibodies in human sera that react with bacterial or phage components. This step, usually done by repeated absorption of the diluted serum with bacterial and phage lysates, is absolutely essential because such contaminating antibodies would completely obscure the detection of other classes of antibodies. The second challenge is the presence of B cells in tumors, sometimes in quite large numbers. These B cells give rise to IgG complementary DNA (cDNA), which is expressed and detected in SEREX. The frequency of such IgG clones varies in each library but can represent a substantial percentage of the clones. When present in high numbers, these clones have to be distinguished before the human serum screening step, and Türeci et al. (21) have devised a prescreening procedure that serves this purpose quite effectively. The third challenge, at least initially, was the suspicion that the majority of antibodies detected in SEREX would be autoantibodies with little or no relevance to cancer. Although known autoantigens do constitute a fraction of the SEREX antigens identified to date, such antigens have not been overrepresented. Part of the reason may be attributed to the initial decision of Pfreundschuh and his colleagues to exclude IgM from the analysis (by using secondary antibodies specific to IgG) and to focus on high-titered IgG antibodies (by performing the immunoscreening at serum dilutions of 1:100 to 1:1,000).

After the initial selection of antibody-reactive clones in SEREX analysis, subsequent steps in SEREX analysis are directed to answering the following questions:

1. Why is the antigen immunogenic?
2. What role, if any, does the antigen have in the malignant process?
3. Is the immune response to the antigen cancer related?

To address these issues, the first step is to sequence the clone to determine its relationship to known genes or motifs and to search for structural or functional modifications in the gene (e.g., mutations, translocations, and gene amplification). The messenger RNA (mRNA) expression pattern of the gene in panels of normal tissues and malignancies by reverse transcriptase-polymerase chain reaction (RT-PCR) and Northern analysis represents the next step to determine whether the gene has a restricted or nonrestricted pattern of expression. This expression pattern of the gene product is then extended to the protein level by generating polyclonal or mAb from animals immunized with recombinant SEREX-defined proteins and using them for biochemical characterization of the cellular protein and for immunohistochemical analysis. If not previously known, chromosomal mapping of the SEREX-defined gene is another important aspect of characterizing SEREX antigens. In evaluating the importance of SEREX-defined antigens in the context of cancer, the serological screening of sera from normal individuals, cancer patients, and patients with nonneoplastic diseases is of critical importance. These surveys of antibody responses to SEREX-defined antigens define whether the immune response is cancer restricted or whether the antigen is recognized by humans in a non-cancer-restricted fashion. To initiate such serological surveys, a small panel of sera from normal individuals and patients with cancer are screened against the SEREX-defined bacteriophage clone, a process called petit serology. SEREX isolates showing a cancer-restricted seroreactivity can then be tested in larger-scale serological surveys (grand serology) using purified recombinant protein as the antigenic target in enzyme-linked immunoassay (ELISA). Modifications of the petit serology have been devised by Scanlan et al. (36, 141) and by Lagarkova (142), and given the names SADA and SMARTA, respectively. These two techniques are fundamentally identical and involve arranging the phage clones on filters in a "dot blot" fashion, thus allowing simultaneous testing of a large panel of antigens. This technique works very well for clones that show strong sero-reactivity. However, the distinction between a weakly-positive signal and a high background is often arbitrary, if not impossible.

At this stage in the development of SEREX, not all SEREX-defined antigens have been through this battery of steps in SEREX analysis. However, it is already known that the vast majority of SEREX-defined antigens appear to show universal or near universal mRNA expressions in normal tissues. Examples of differentially expressed SEREX genes have been found, but these are uncommon. Also, mutations and other structural modifications are rare, but more extensive sequencing should be carried out to exclude structural changes in the coding genes. What is encouraging is that SEREX has identified a number of gene products that have known or suspected relevance to cancer development (e.g., oncogenes and suppressor genes) and other products that have potential as targets for cancer vaccines. Additionally, the finding of SEREX-defined gene products that appear to be recognized by the humoral immune system of subsets of cancer patients but not normal individuals further emphasizes the potential of SEREX. The fact that a number of these genes are widely expressed in normal tissues indicates that cancer-specific recognition can occur in the absence of cancer-specific expression. The basis for this cancer-specific immunogenicity is one of the central challenges that needs resolving.

Modifications in the SEREX Technique

Although SEREX analysis was initially developed to analyze the autologous humoral response to cancer - that is, using tumor and serum from the same patient - a number of modifications on the original SEREX design have been introduced or are being considered (Table 1). The first variation, developed to discover new antigens belonging to the cancer-testis (CT) category - that is, antigens with restricted expression in normal testis and in cancer - involves the screening of sera from cancer patients on allogeneic testicular cDNA libraries, either unsubtracted or after subtraction with cDNA derived from nontesticular normal tissue. This approach has led to the recognition of several CT antigens, including new members of the SSX family (22), SCP-1 (23), HOM-TES-85/CT8 (24), and CAGE-1 (96). Another modification involves established tumor cell lines rather than fresh tumor as the source of cDNA for SEREX analysis. Although one of the great advantages of SEREX is that it circumvents the requirement for cultured tumor cells, cell lines can be a useful alternative target source of cDNA. This is illustrated by the analysis of SK-MEL-37, an established melanoma cell line that was chosen for SEREX because it expressed most, if not all, of the known CT antigens (25). A cDNA library from SK-MEL-37 was screened with allogeneic sera from a patient having high titers of antibody to two CT antigens, NY-ESO-1 and MAGE-1. This screening resulted in the isolation of the two known CT antigens, NY-ESO-1 and MAGE-4, and the discovery of a new member of the CT family, CT7. The use of cell lines for SEREX may be particularly useful in the case of tumors in which tumor specimens are difficult to obtain (e.g., small-cell lung cancer). A third modification involves a combination of representational difference analysis (RDA) (26) and SEREX. This approach is not SEREX-based cloning in the strictest sense, but rather uses SEREX-based serology to identify seroreactivity to the products of genes cloned by the RDA approach. An example of this was the cloning of CT10 (27). CT10, a gene sharing homology to CT7 and members of the MAGE family, was initially cloned by RDA as a gene abundantly expressed in SK-MEL-37 but not in normal skin. After initial RDA cloning, the cDNA was subcloned into the same expression vector used for SEREX, and the resulting phage clone was tested against a panel of absorbed serum samples from tumor patients. In this manner, the immunogenicity of RDA-identified gene products can be analyzed and documented.

In addition to extending the antigenic targets for SEREX analysis beyond the autologous tumor (e.g., testis, tumor cell lines, and RDA) different sources of antibodies other than autologous and allogeneic sera are being explored. Human oligoclonal or mAb generated from B cells (directly after isolation or after expansion in vitro) from peripheral blood, tumor, or draining lymph nodes of cancer patients are promising sources that need investigation. The screening using IgA myeloma protein (123) represented an example of a natural monoclonal antibody in this regard. Additional serological probes could come from cDNA cloning of IgG sequences expressed by tumor-infiltrating B cells of cancer patients. Another interesting modification is the screening of urine samples from patients. This notion was proposed by Jäger et al. (143), based on their observations that anti-NY-ESO-1 antibodies are present in the urine of patients with high-titer NY-ESO-1 antibody. Urine gave very low background in SEREX screening, eliminating the need for pre-absorption of serum with E. coli and phage lysates.

Another area of modification involves the expression systems used for the recombinant cDNA library. Although most studies used the lambda ZAP phagemid system, one recent modification used the pJuFo phage surface display system in which recombinant proteins are expressed on the surface of M13 filamentous phage (104). This allowed the combination of immunoscreening with the biopanning procedure, potentially enabling the establishment of a high-throughput technique capable of screening a large number of sera against many libraries. Another approach aims to improve on the lack of post-transcriptional processing of E. coli. In this approach, recombinant libraries were expressed in yeast cells, allowing a certain degree of glycosylation and other modifications (149).