Fluorescent Biomarker Discovery

Phage libraries permit the screening and recovery of bacterioviral clones expressing peptides that bind to unique antigens on a given tissue or cell. The phage clones each encode a single peptide (9-12 amino acids long) inserted into the coat of the phage by recombinant DNA methods. The small size of the peptides allows them to diffuse into neoplastic esophageal mucosa and to bind to markers specific for disease. Also, a peptide sequence of this length is able to generate a much higher degree of complexity than antibodies. The inserted peptides can be randomly generated, resulting in a library of millions of phage clones, each with a different peptide. This library can be screened for those clones whose peptides bind specifically to a given target, which may range from purified molecules to whole tissues. Recovery of phage that bind to a given tissue permits all the tools of recombinant DNA to be used for rapid determination of the sequence, and subsequent synthesis of the peptide for imaging or other purposes. Researchers in our group have undertaken to develop a pre-cleared sub-library by absorbing away phage that bind to a normal human epithelial cell line, as well as those that bind plastic or are inactivated by components of the culture conditions. This sub-library (referred to as the "cleared library") is useful for a variety of screens, including biopsies of Barrett's esophagus. The eventual goal of this work is to synthesize a peptide that binds to dysplastic esophageal tissue, and to link this peptide to a fluorescent molecule such as indocyanine green (ICG), Cy 7, and Alexa fluor 750 for endoscopy that images dysplasia.

Figure 1 Images by confocal fluorescence microscopy demonstrate (A) the relative absence of binding of peptide to non-tumorigenic esophageal cells, HET-1A and (B) significant binding to the periphery (cell surface) of Barrett's-derived adenocarcinoma cells, TE7.

As a proof-of-concept experiment in imaging with biomarkers, our group has undertaken to isolate phage that bind preferentially to the TE7 cell line, derived from Barrett's adenocarcinoma, and fail to bind to a normal epithelial cell line HET-1A. The M13 phage library (New England Biolabs) chosen for screening in this study, is well characterized and has been used broadly. For our purposes, 10 11 phage with a density of 10 9 , representing 100 copies of each clone, were panned initially on the normal epithelial cell line HET-1A, to adsorb the phage that bind normal cells. The pre-cleared phage from the resulting supernatant was then panned with a biopsy specimen of esophagus with dysplasia to identify 10 binding clones. Phage from each of the 10 binding clones were placed into separate wells containing HET-1A and TE7 cells. Monoclonal mouse Anti-M13 phage antibody, conjugated to FITC dye, was then administered into all the wells. The binding of the antibody to the phage was evaluated by standard confocal fluorescence microscopy (Bio-Rad MRC 1024). Figure 1A shows the relative absence of binding of the antibody for HET-1A with one clone, while Figure 1B shows the extensive binding for TE7 with the same clone, in particular around the cell surface. This data supports our hypothesis that small peptides that preferentially bind to cell surface antigens of neoplastic cells can be identified by phage display methods. These will be sequenced and the peptides that they encode will be synthesized and tested as imaging agents.