Antecedentes de la investigación 1 Regional.

In the present study, a stlec-fi\/c approach was adopted in order to rapidly identify CTL

peptide epitopes from the CMV protein pp65. Potential antigenic peptides were first selected from the known sequence of pp65 (Ruger et al., 1987), by the presence of HLA-A*0201 anchor residues (Falk et al., 1991), and then a further selection was made by two different methods of their capacity to bind to HLA-A2 molecules.

The results described in this chapter show that, although 17 peptides from the CMV pp65 protein from AD 169 carried the appropriate HLA-A2 MHC binding motif, only 7 of them were able to stabilise HLA-A2 molecules as demonstrated by the T2 stabilisation assay. Previous studies had suggested that, besides the dominant anchor residues that comprise a peptide binding motif, the presence of hydrophobic and aromatic residues in secondary anchor positions can be relevant for preferential peptide binding, and that negatively or positively charged residues in the same positions can be associated with no or poor binding (Drijfhout et al., 1995; Ruppert et al., 1993). These latter observations agree with some of the results obtained here, when the pp65 peptides bearing different anchor motifs (L2-V9,

I2-L9, L2-I9, L2-L9 or I2-V9) were examined in the T2 stabilisation assay. Peptides AE42, AE44 and AE45 bearing anchor residues Leucine and Valine at positions 2 and 9 respectively, demonstrated particularly good HLA class I stabilisation. These peptides also have several hydrophobic residues in their sequences, such as Valine and Alanine in positions 6 and 7 in peptide AE42. In contrast, peptides APS8 , AE47, AE48 and AE49

showed minimal or no HLA-A2 stabilisation, although they possessed the same anchor residues. However, these peptides contain at least one charged residue in secondary positions, which may explain their poor binding. For example, although peptide AF8 8 has

an aromatic residue (Phenylalanine) in position 5, which has been shown to be linked with preferential binding (Drijfhout et al., 1995), its lack of stabilisation of HLA-A2 molecules on T2 cells could be explained by the presence of two negatively charged residues

(Glutamic acid) in positions 3 and 7, since it has been reported that the presence of negatively charged residues at position 3 is common in non-binding peptides (Ruppert et al., 1993).

Certain inconsistencies were found when pp65 derived peptides with the L2-V9 motif were tested in a peptide-binding competition assay. In this assay, all of the peptides, with the exception of peptide AP8 8 , showed either a high or intermediate binding affinity for

HLA-A2. Interestingly, the nonamer AE48, and the decamers AE47 and AE49, which did not stabilise cell surface HLA-A2 in the T2 stabilisation assay, appeared to have a high or intermediate relative affinity for the HLA-A2 molecule in the peptide binding competition assay. It is possible that positions 7 and 8 , which are secondary residues in

HLA-A2 binding decamers, may be influencing peptide binding. For example, AE47 contains a positively charged residue. Histidine, at position 7, and a uncharged Valine residue at position 8 . Whilst positively charged residues such as Arginine, Lysine or

Histidine are more frequent in HLA-A2 non-binding peptides, uncharged residues at position 8 in decamers have been shown to be associated with high affinity binding to

HLA-A2 (Ruppert et al., 1993). Therefore this observation, together with the fact that AE47 possesses preferred anchor residues, may override the detrimental effects of the residue at position 7. In contrast, peptide AE49 contains only one residue in its sequence known to be associated with poor binding, this being an Arginine in position 4 (Ruppert et al., 1993). However, it also possesses an Isoleucine in position 1 which has been associated with good binding (Drijfhout et al., 1995). In the particular case of peptide AE48, no secondary residues were found that would be expected to either decrease or promote the binding of this peptide to HLA-A2. However, when a computer-driven algorithm which analyses the sequence of a given protein in the search for potential T cell peptide epitopes (Parker et al., 1994; Parker et a i, 1995), was applied to investigate the possible HLA-A2 binder nonamers or decamers in pp65, peptide AE48 emerged as the

third best binding peptide to HLA-A2. This computer analysis therefore correlated with the results from the peptide-binding competition assay for this peptide.

In general, the pp65 peptides predicted to be good HLA-A2 binding peptides by the computer algorithm, corresponded to peptides which showed a good binding affinity by either of the peptide-binding assays tested in the present study. Two interesting exceptions were observed, peptide AE92 was predicted by Parker's method to be the 2nd best HLA-A2 binding peptide, but this peptide did not show any stabilisation of HLA-A2 molecules when tested in the T2 stabilisation assay. This peptide would be an interesting candidate to test using the peptide binding competition assay, however, due to time limitations it was not tested. In contrast, peptide AE44, which in both the peptide binding assays utilised in the present study showed an intermediate affinity for the HLA-A2 molecule, appeared only at 11th place on the list of the best HLA-A2 binding peptides from the computer algorithm. A similar situation was found with the HIV-derived peptide ILKEPVGH, which is an HLA-A2 restricted CTL epitope. This peptide "theoretically" ranks 45th of the 1007 possible nonamers in the HIV polymerase, which would place it only in the top 5%. Nevertheless, this peptide binds to HLA-A2 much better than expected (Parker et al., 1994), and this peptide was used as a positive control in the peptide-binding

competition assay employed in the present study, where it also showed a high binding affinity for HLA-A2. In contrast, the influenza matrix peptide Flu-M (58-66) GILGFVTL, which is a known HLA-A2 CTL peptide epitope and also showed consistently good binding in the assays tested here, ranks first among all possible nonamers from the matrix protein and therefore conforms to the expectations. These observations suggest that the prediction of good binding peptides following either the presence of peptide-binding motifs or algorithms like the one used here, offer a fast and efficient strategy, but may be relatively imprecise.

Inconsistency between the T2 stabilisation assay and the peptide-binding competition assay could possibly be attributed to poor sensitivity of the first method. The latter proposal is supported by a report showing that where the T2 stabilisation assay has been compared to other peptide binding assays that employ either fluoresceinated or radiolabelled peptides to measure affinity, and it was found that they required a lower concentration of peptide to demonstrate a positive effect compared to the T2 stabilisation assay (Zeh III et a l, 1994). However, there are other factors to consider when correlating the results obtained with the peptide-binding competition assay and the T2 stabilisation assay, i.e. the fact that two peptide binding assays were performed at different temperatures and in different cell types. This may influence the stability and expression levels of empty HLA class I molecules on the cell surface (Baas et a l, 1992; Schumacher et al., 1990)

Discrepancies similar to the ones observed between the T2 binding assay and the competition assay were found when a number of the pp65 derived peptides (AE44, AE45, AE47, AE48 and AE49) tested in this study were analysed in a separate study at the Anthony Nolan Research Institute using an optical biosensor lASYS™ system. The five peptides tested in the latter study significantly increased ^2 ^ 3 exchange kinetics in that

system (Morgan et al., 1998) which, although not being a direct measurement of the affinity of peptides for the HLA-A2 molecule, suggested that such peptides facilitated the stabilisation of peptide-MHC-p2in complexes. These observations highlight the increased

sensitivity of the peptide binding competition assay and also the lasys™ over the T2

stabilisation assay.

In summary, utilising peptide binding motif predictions and the binding assays described here, it was possible to identify a number of peptide sequences from CMV pp65 that stabilised HLA-A2 molecules present on the surface or in the T2 cell line, and which showed either high or intermediate affinity for the HLA-A2 molecule in peptide-binding

competition assays. However, at the present time, the binding predictions are imprecise, and it is known that MHC binding is necessary but not sufficient for a peptide to be immunogenic for CD8^ T cells. It is therefore essential to confirm that the predicted CTL

peptide epitopes are able to be generated by endogenous antigen presentation pathways in

vivo, and that they are also capable of eliciting a CTL response. This point will be crucial if the synthetic peptides in question are to be considered useful as vaccines or for any other form of therapy. Thus it would be useless to generate a CTL response in vitro to a predicted peptide fragment of a viral antigen, if that peptide was never generated during a viral infection in vivo and did not stimulate a CTL response. Therefore, 5 of the pp65- derived peptides identified in the present study which showed high or intermediate affinity for the HLA-A2 molecule were tested for their ability to generate a CMV-specific CTL response in vitro, and this work forms the basis for the results described in Chapter 5.