11 Serological antigen selection of phage displayed colorectal tumour cDNA libraries

There has been a resurgence of interest in immunotherapies for cancer, particularly concerning the isolation of novel antigens which distinguish cancer cells from normal cells. Recently i t was demonstrated that the humoral immune response can be used to isolate novel tumour antigens. This involved the probing of a tumour specific cDNA expression library in h phage with autologous patient serum (serological cDNA screening). Using this method Sahin et al., (1%5)[1] have been able to identify new tumour antigens in melanoma, renal cell carcinoma, Hodgkin's disease and astrocytoma. We are applying serological cDNA screening to colorectal cancer (CRC) but are also extending this approach by allying it to phage display technology through the serological selection of phage displayed CRC cDNA libraries. Both approaches require the use of the patient's anti-tumour cell response to identify those tumour cDNA expression products that naturally evoke an IgG immune response, and as such may be interesting candidates for immunotherapy. We have made two colon tumour cDNA libraries in I-ZAP. The size of the libraries was greater than 106 recombinants with an average insert size of 1 kb, and with 1% positive GAPDH positive plaques. The quality of our libraries has also been tested by DNA probing for CEA which has shown approximately 0.022% CEA positive recombinants. One of our tumour cDNA h expression libraries has been screened using autologous paticnt serum. h phage recombinants were plated at the desired plaque density for screening and expression of the cDNA products [ 11. The library was seen to have a high proportion of IgG transcripts (0.1%) which are detected by the secondary antibody alone (antihuman IgG alkaline phosphatase conjugate) and complicate the screening procedure significantly. This high proportion of immunoglobulin transcripts in the cDNA library is most likely caused by high B-cell infiltration into the tumour. This in turn may be associated with a high humoral immune response. Serological cDNA screening has proven difficult due to the high proportion of IgG transcripts present and as such we are pursuing alternative approaches for the isolation of tumour antigens in CRC. We are using serological selection of cDNA libraries displayed on filamentous phage. The filamentous phage links the expressed and displayed protein product to the gene, cloned into the phage genome. This allows the selection and enrichment of those cDNA products that interact with a chosen immobilised, or labelled ligand, for example, in our case, autologous patient serum IgG. Selection from phage cDNA display has potential advantages over serological screening of cDNA libraries in I phage. (a) The gene product and its genetic information are physically linked through the display on thc surface of filamentous phage so retrieval of positive clones is much quicker and less labour intensive. (b) By using phage display we can enrich for specific genes via interactions between gene product and ligand. (c) The correct secretion and processing of the cDNA products is a pre-requisite for the display on filamentous phage. This means that phage displayed cDNA libraries will bias for cell surface tumour antigens (i.e naturally secreted antigens that can move through the E.coli secretion apparatus). (d) Proteins displayed on phage surfaces are likely to have an authentic three-dimensional structure so serum reactivity with conformational epitopes will operate in the selection procedure unlike in serological cDNA screening. (e) The conditions set for proteinlprotein interactions can be controlled during the selection procedure as all selections are done in vitro . The presence of stop codons in cDNA has suggested that direct display by fusion to the N-terminus of p3 by established methods is limited. As such an indirect display approach has been developed by Crameri and Suter (1994)[2] which involves the linking of cDNA libraries to the phage surface via the heterodimerisation of the Jun and Fos leucine zippers. A Jun-p3 fusion and a Fos-cDNA fusion are expressed in the same E.cofi cell and both products are exported to the periplasm where viral assembly occurs. The Jun-p3 fusion is incorporated into the M I 3 phage coat and the Fos-cDNA product can heterodimerise with the Jun leucine zipper such that it is displayed on the phage surface. This approach has been successfully applied to selection of allergenic proteins from a cDNA library prepared from Aspergillusfumigatus using serum from affected patients [2,3]. This vector (pJuFo) was made available by Dr. M. Suter for our own work. cDNA libraries have also been displayed on p6 [4] which connects the phage body to the host binding p3 and, unlike p3, it is thought to have a solvent exposed C-terminus available for the attachment of foreign proteins. We have constructed our own p6 display vectors, pSP6A, pSP6B and pSP6C which have a versatile multiple cloning site for the display of cDNA libraries in all three reading frames. We have displayed the CH3 domain of human IgG and alkaline phosphatase on p6 and compared the display efficiency with that of display on p3. Our results are in agreement with those of Jespers (1995)[4] in that p6 display efficiency is lower than p3 display (Manuscript in preparation). The two cDNA libraries that we have generated in h phage for display systems. We have cloned these libraries into both the