Two new papers in Science this month have shed light on the mechanism of cross-priming, by indicating the types of antigen that are favoured by this process. Both groups suggest that this has important implications for vaccine design.

Peptides presented on MHC class-I molecules to CD8+ cytotoxic T lymphocytes (CTLs) are normally derived from endogenous proteins, such as viral proteins produced by infected host cells. In some cases, exogenous proteins can also be taken up by MHC class-I-positive cells for cross-presentation. This might be particularly important for naive CTL priming, which requires professional antigen-presenting cells (pAPCs) and cannot be achieved by other tissue cells.

Ton Schumacher and colleagues looked at whether the location of a CTL epitope within a peptide affects cross-presentation. Signal peptides — the amino-terminal part of a protein that targets it to the endoplasmic reticulum and is then cleaved — are an important endogenous source of MHC class-I-restricted antigens. But, in their system, an epitope present in the functional signal sequence of a GFP fusion protein could not be cross-presented to CTLs at any significant level, whereas the same epitope or a different epitope close to the carboxyl terminus of the mature GFP protein could be efficiently cross-presented. They showed that this difference was due to the efficiency of uptake by pAPCs. When naive mice were challenged with cells transfected with a GFP construct containing two different epitopes (one in the signal sequence and one in the mature protein), T-cell priming in vivo was skewed towards the epitope in the mature protein. This shows that the exogenous pathway is dominant for the priming of naive CTLs in this system.

The authors suggest that cross-priming is biased towards unprocessed antigens and against sequences that are degraded rapidly after synthesis, such as signal peptides. This ties in nicely with the second study by Jonathan Yewdell and colleagues, which shows that cross-priming favours proteasome substrates rather than pre-processed peptides.

MHC class-I-negative cells, which cannot present endogenous antigen, were infected with recombinant vaccinia virus encoding full-length ovalbumin (OVA) and then introduced into mice that had received ovalbumin-specific CD8+ T cells. When the proteasome inhibitor lactacystin was added to the virus-infected cells, activation of the CD8+ T cells was not prevented, indicating that cross-priming can occur in the absence of proteasomal degradation of the protein in the donor cells. By contrast, when donor cells were infected with recombinant vaccinia virus encoding a particular OVA epitope, CTLs specific for the OVA epitope could not be detected, showing that minimal peptides cannot be cross-presented. Similarly, a chimeric protein designed to be rapidly degraded by the proteasome could not be cross-presented unless the cells were first treated with lactacystin to prevent degradation. The authors therefore suggest that in their in vivo set-up, cross-priming is based on the transfer of proteins rather than peptides, which casts doubt on the previously suggested role of peptide-binding molecular chaperones in this process. Together, the two studies show that for efficient cross-presentation, vaccine antigens should be engineered for maximum half-life.