Class I MHCs are clustered on the cell surface. In order to assess the functional consequences of clustering, we developed two methods to enhance the size of MHC clusters, cholesterol depletion and specific crosslinking of an engineered MHC molecule. Overall, clustering by cholesterol depletion and dimerization produced similar effects in presentation. However, at certain concentrations of peptide, the effective changes differed between treatments. These discrepancies may be due to a difference in the size and nature of clusters generated by the two treatments. They may also be due to technical reasons; we required different assays for the effects of the two treatments, therefore the data in and may not be equivalent. Nevertheless, clustering by either method enhanced recognition of peptide/MHC by T-cells, especially when agonist peptide was scarce.
Both acute and chronic depletion of target cell cholesterol enhanced granzyme, serine esterase, release by activated effector CTL. Enhanced T-cell responses also translated into enhanced lysis of target cells. Allo-reactive T-cells killed APCs with clustered MHC more efficiently than they killed control target cells. These effects were seen at all E:T ratios. Conversely, dispersion of clusters on cholesterol-depleted cells returned their presentation to control levels. As we mentioned earlier, we argue that the effects on class I mobility and clustering by cholesterol depletion are mediated through the actin cytoskeleton (10
). Stiffening and loosening the actin cytoskeleton using various actin filament drugs produced the same effects on presentation as predicted.
When we switched to a peptide-specific model of recognition, we could only enhance killing at low concentration of agonist peptide. This is consistent with the idea that allo-recognition involves only a subset of the peptides represented on the surface and that clustering may facilitate T-cell scanning for cognate peptide. If lysis was enhanced at all concentrations of agonist, one could argue that the functional consequences of cholesterol depletion were downstream of agonist recognition. Moreover, if lysis was also enhanced in the case of non-specific peptide or no added peptide, one could argue that recognition by T-cells did not play a role in the effects seen. However, we only saw enhancement at low agonist peptide concentrations, which argues strongly that the changes we found were due to changes in the efficiency of presentation.
Clustering class I MHC by crosslinking engineered H2-Kb molecules confirmed and extended our functional results on clustering of class I MHC molecules by cholesterol depletion. Microscopy shows that dimerizing MHC molecules stabilizes and amalgamates the small clusters of MHC present on APC surfaces.
Different peptides produced different effects when clustered. In the case of cells presenting high concentrations of SIY agonist, we saw weak enhancement of lysis after cholesterol depletion and saw some reduction after dimerizing Kb-1BP. We argue that at high levels of strong agonist, T-cells are efficiently activated; therefore clustering has a small effect on recognition. As we titrated down the levels of agonist peptide, recognition was less efficient, and clustering enhanced it. This was seen even more clearly when cells were loaded with the weak agonist peptide, p2CA, whose recognition by T-cells is inefficient. Clustering class I MHC on cells loaded with low concentrations of p2CA enhanced specific lysis to almost double that of controls while at high concentrations of p2CA, clustering actually inhibited recognition. In most cases, enhancement seemed to peak at nanomolar concentrations and would in some cases drop off at very low concentrations.
Clusters may enhance recognition by providing multiple local copies of some peptide/MHC or by enhancing T-cell residence time on a target cell. Normally, scarce agonist peptides are presented in a sea of self-peptide. Clustering of class I MHC molecules would cause a reduction in their rotational and lateral diffusion as compared to monomers on the plasma membrane. These large, slowly-rotating lattices of class I MHC could enhance recognition by providing T-cells with stable local densities of class I MHC + peptide combinations to sample. Since TcR can dock on a class I MHC in a variety of orientations and tilts (26
), cluster of class I may also facilitate proper docking and reduce scan time for TcR sampling. In addition, as some TcRs may exist as clusters as well, engaging clusters from both the T-cell and APC side may provide efficient activation.
Clusters of MHC may also function to provide simultaneous sampling of self- and non-self peptide by multiple TcR in a membrane domain. At high concentrations of strong agonist, the density of signal is more than sufficient and clustering may not be necessary for recognition, as was seen in our results with SIY peptide. However, in the absence of strong agonist at high surface concentrations, the sampling of bouquets of self and agonist may be a way T-cells can differentiate minute class I MHC peptide differences to make dramatically different outcomes for activation. It is important to note that in our peptide-loading model, APCs were loaded in the presence of serum. At high concentrations of peptide with high affinity for MHC, such as SIY or SIN, serum proteins probably had little effect on surface loading of MHC. However at low concentrations of agonist, serum was likely to contribute non-specific, non-activating peptides to the surface display. In the case of weak agonist, p2CA, with weak affinities and short koff
for the TcR (1
), the role of non-specific peptide may be particularly important for recognition. This may explain why at low concentrations, clustering had such a strong effect on p2CA recognition. It may also explain the negative effects clustering had on presentation of high concentrations of p2CA. Clustering high concentrations of weak agonist might produce a “bland” signal, and may read too similar to a null peptide when presented in the absence of null peptide. This also suggests that the choice of null peptide would play an important role in effective detection of weak agonists as others have suggested for class II MHC (27
From our results, we would not be surprised to find that professional APCs modulate class I clustering in order to modulate antigen presentation. Others have shown IFN-gamma treatment, known to upregulate class I levels, also reorganizes class I on the cell surface (28
). This change could enhance presentation to and activation of naïve T-cells. On the other hand, it may be that dispersion of clusters or their absence on some APCs (6
) leads to T-cell anergy rather than activation. It is unclear to what degree cells regulate class I clusters in terms of size and stability. But from their functional importance in presentation, clusters may likely be regulated in an inducible manner.
Taken together, these data suggest that MHC organization plays an important role in modulating T-cell sensitivity for agonist peptide. Clustering MHC may enhance recognition of weak or rare epitopes in cancer and viral vaccine models. One direction others have taken in cancer immunology has been to identify and expand the number of oncogenic epitopes recognized by ex-vivo stimulated tumor infiltrating lymphocytes (30
). Clustering class I on these ex-vivo stimulations may expand the repertoire of T-cell clones stimulated thereby enhancing the engendered response.