Phage display libraries, since their discovery in 1990 , have been used to identify high affinity ligands to a variety of molecules small and large. Recently, Tiwari and colleagues  have used two different peptide phage display libraries to identify potential peptides that mimic the antibody-binding epitopes of the extracellular domain of HER-2/neu antigen. Employing anti-HER-2/neu monoclonal antibodies as a bait and four rounds of bio panning, the three selected peptides were able to elicit humoral immune responses in mice and to inhibit the binding of the bait to HER-2/neu  antigen, thus illustrating that synthetic peptide mimics can elicit immune stimulatory activities.
The primary aim of this study was to test a "proof of principle", that functional synthetic mimics of native Hsp70-PCs could be identified by using two step bio-panning of random peptide phage display libraries. Previous studies have shown that Hsp-PCs from mice and humans can induce specific T-cell responses [7, 21]. The mimic peptides, in principle, should structurally resemble the Hsp70-PCs used as the initial bait and may have functional properties similar to Hsp70-PCs such as the ability to stimulate T-cells. The Hsp70-PCs preparation used in this study was obtained by affinity selection on an ADP-agarose column. Hsp70-PCs, isolated by this method, have been shown to contain a wide array of peptides and to possess immune-stimulatory activity [7, 43]. Thus, the starting material for the bio-panning was the pool of peptides complexed with Hsp70. This approach circumvents the need for the purification of individual peptides from the Hsp70-PC fraction and selects for the abundant ones. Similar bio-panning approaches have proved successful in identifying tissue-specific peptide ligands [(25); see ref.  for a review].
A variety of phages displaying unique peptides were identified by the bio-panning method. There appeared to be a high degree of enrichment of specific sequences following four rounds of bio-panning; certain peptides selected were common to at least two bio-pannings and one of the peptides (TMG) was recovered in all three bio-pannings. The diversity of peptides identified may be in part due to the complexity of the Hsp70-PCs fraction used as the bait. The peptide TMG was selected in three independent bio-panning experiments, suggesting that the motif recognised by TMG may be consistently abundant in the three pools of Hsp70-PCs.
Three of the 'recogniser' peptides, identified in multiple bio-panning experiments, were then used in a reverse bio-panning experiment to identify sequences that interact with these peptides. Two of these recogniser peptides, NNY and IER, selected a wide variety of binders, one of which, SVS, was common between the two selected pools. Subsequent bio-panning experiments with an unrelated peptide bait, consistently selected the SVS peptide (Arnaiz, James and Bond unpublished data). Interestingly, this same phage peptide had been identified in two unrelated bio-pannings for peptides interacting with (i) murine cerebellar granular neurons and (ii), a Japanese encephalitis virus envelope protein neutralizing antibody [40, 41]. The consistent selection of SVS with unrelated baits perhaps suggests that this peptide may recognise some common structure among all baits, for example the peptide bond backbone. Unlike the peptides NNY and IER, the TMG selected only a single displayed peptide after four rounds of bio-panning. Due to the high degree of selectivity of the TMG peptide and the fact that this peptide was identified in three independent bio-panning experiments, we focused our subsequent analysis on this TMG-DSP 'recogniser-mimic' pair.
The ability of the synthetic peptides to activate lymphocytes was examined using an in vitro assay. The results show that lymphocytes were stimulated but required two consecutive iDC-mediated exposures to either the mimic peptide DSP or the recogniser peptide TMG. Unlike LPS, these peptides did not stimulate iDCs to produce IL-12. Therefore, we can conclude that (a) the activation of lymphocytes must be dependent upon the uptake of the peptide by iDCs and its representation to lymphocytes and (b) it is not the result of any adjuvant-like contaminant present in the peptide preparation. The DSP peptide showed lymphocyte stimulatory activity while, another peptide WHK produced no stimulation. Surprisingly, we did observe lymphocyte stimulation with the recogniser peptide TMG. The reason why lymphocyte activation was not limited to only mimics is currently unclear. The observed differences in effectiveness between the peptides may be reflective of the different proteolytic processing and/or the preference of the different HLA class I molecules for presentation of 9-mer peptides with specific amino acids in anchor positions. A search of the comprehensive database SYFPEITHI  for HLA class I ligands with the peptide sequences revealed a higher likelihood for DSP and TMG than WHK to be presented by the HLA class I molecule (data not shown). Furthermore, based on our model of recognisers and mimics described above, it is quite possible that peptides structurally equivalent to recogniser peptides may also be present in the pool of Hsp70-PCs, for example, the EGF receptor (a possible 'recogniser' molecule) is activated by autocrine or paracrine growth factor loops and is known to be over-expressed in at least 50% of all epithelial malignancies  as is its ligand, EGF, (a possible 'mimic' molecule). Supporting this view, we find that both TMG and DSP peptides but not WHK specifically bind to MDA-MB-231 cells indicating that either peptide can interact with cellular components within these cells (Arnaiz and Bond, unpublished results).
To determine if the DSP peptide represents a mimic of true tumour antigens present in tumour cells and in particular tumour antigens in the Hsp70-PCs pool, we employed an assay in which lymphocytes are stimulated in two consecutive rounds with different antigen pools. We find that tumour cell extracts contain certain antigens in common with those present in the Hsp70-PCs fraction as they successfully stimulate T-cells previously primed with Hsp70-PCs from the same tumour cells. Additionally, we observe that Hsp70 alone, in the absence of any associated peptides, can prime T-cells to respond to tumour cell extracts. Thus Hsp70 in addition to chaperoning peptides into the antigen processing pathway of iDC, may also trigger IFN-γ production in a similar way to that of LPS [46–49]. These findings are in agreement with previous data showing that Hsp70 can enhance the ability of APCs for antigen uptake [50, 51] and can activate T cells in vitro an in vivo [10, 52]. Therefore, one could envisage a pool of antigens (peptides) being chaperoned by adjuvant molecules such as Hsp70 which can also facilitate their uptake by the APCs through Hsp-specific receptors (e.g., CD91) in the case of tissue damage or necrosis. These peptides may be then re-presented to T-cells, through the MHC class I antigen processing pathway (cross-priming). Thus, the reconstitution of peptides with Heat shock proteins such as Hsp70, might be an important strategy to ensure an enhancement of the T-cell response to peptides [16, 53].
Using the same technique, we also show that the DSP peptide resembles antigens present in total cell extracts from either tumourigenic (MDA-MB-231) or non-tumourigenic (MCF-12A) breast carcinoma cell lines. Thus, the DSP peptide may mimic a common antigen in both cell lines. The TMG peptide showed lower levels of lymphocyte stimulation following a second exposure to either extracts from MDA-MB-231 or MCF-12A cell lines. We also show that lymphocytes incubated initially with iDCs loaded with Hsp70-PCs can be re-stimulated with iDCs loaded the DSP peptide, again suggesting that this peptide resembles antigenic peptides associated with the Hsp70 in these cell lines.