APCs size up antigens

Intracellular antigens—from viruses, tumors, bacteria and other sources—pose a challenge to the immune response and to vaccine development. How do such antigens enter antigen-presenting cells? Three new studies examine this process, implicating whole proteins instead of small peptides as key intermediaries.

Many of the pathogens that threaten human health do so from within the protected environment of our own cells and tissues. An effective immune response against such invaders often necessitates the drastic step of killing the host cells that harbor them. Cytotoxic CD8⁺ T lymphocytes (CTLs) are especially suited to this task, with an impressive range of methods for inducing the death of target cells. CTLs recognize target cells through major histocompatibility (MHC) class I molecules occupied with 8–10-amino-acid peptides derived from endogenous proteins. Initiation of CTL responses against infected target cells generally requires that the antigens presented at their surface must first be acquired and presented to CTLs by host antigen-presenting cells (APCs), such as dendritic cells. This poorly understood pathway of antigen trafficking and presentation is called 'cross-priming' and is essential for the generation of CD8⁺ T-cell responses against viruses, intracellular bacteria, tumors and tissue antigens^1,2.

A central question in the immunobiology of cross-priming concerns the nature of the antigenic material transferred from donor cells to APCs; information about this would significantly influence vaccine design, immunotherapy and transplantation. Three recent papers reveal that cellular proteins, as opposed to the smaller peptides they contain, constitute the 'molecular currency' of cross-priming and have begun to shed light on the rules that govern this process^3,4,5.

APCs acquire foreign antigens by migrating to sites of inflammation; there they sample their microenvironment before moving on to lymphoid organs where their cargo of acquired antigens is displayed to responding T cells as smaller peptides bound to surface class I MHC molecules. One attractive theory holds that peptides are transferred to APCs as extracellular complexes bound to cellular heat shock proteins (HSPs)⁶. In support of this idea, HSP complexes containing specific antigenic peptides can be isolated from variety of cell types and taken up by specific receptors into APCs in vitro; on in vivo administration, these APCs can cross-prime CD8⁺ T cells⁷.

Another hypothesis states that cellular antigens are transferred to APCs as intact proteins or large fragments taken up through phagocytosis⁸. Understanding the relative contribution of these pathways to cross-priming is more than an academic matter; cell-associated antigen activates CD8⁺ T cells in vivo much more efficiently than does soluble antigen, raising the possibility that vaccination strategies based on exploiting the cross-priming pathway could promote superior T-cell responses⁹.

The new studies, by Norbury et al.⁴, Shen et al.⁵ and Wolkers et al.³, take a common approach to this question, examining the ability of donor cells expressing different forms of a precursor antigen to generate CD8⁺ T-cell responses through cross-priming. Norbury et al. used recombinant vaccinia virus to transduce MHC-mismatched tumor cells with various forms of ovalbumin. These forms included a minimal OVA_257–264 epitope targeted to either the cytoplasm or the endoplasmic reticulum and expressed within either a larger chimeric fusion protein or the full-length OVA protein itself. Each construct could prime OVA_257–264-specific CD8⁺ T cells after direct infection of mice. When the cells expressing the various constructs were used for immunization, however, only the larger forms of the source antigen (intact OVA and the OVA_257–26e-containing fusion protein) could cross-prime transgenic OVA_257–264-specific CD8⁺ T cells.

Norbury et al. then examined cross-priming of T cells by a rapidly degradable form of the epitope-containing fusion protein. The fusion protein could cross-prime T cells only if donor cells were first treated with a proteasome inhibitor, although the same treatment did not boost the ability of cells expressing stable OVA_257–264-containing proteins to cross-prime OVA-specific CTLs.

The results of Norbury et al. indicate that cross-priming is not mediated by exogenous or endogenous peptides but rather through the transfer of intact proteins (possibly through phagocytosis) or protein fragments from donor cells to APCs. These findings dovetail with those of Shen et al., who found that the ability of an antigen to cross-prime CD8⁺ T cells did not depend on colocalization with heat shock proteins but rather on the amount of intact source protein and its subcellular location (cytoplasm, endoplasmic reticulum or plasma membrane).

Wolkers et al. asked whether the location of an antigenic peptide within its source protein could influence cross-priming of CD8⁺ T cells. Using green fluorescent protein (GFP) as a model antigen, the authors engineered several chimeric proteins. They placed one of two distinct CD8⁺ T-cell epitopes either near the N-terminus, within a signal sequence, or near the C-terminus of GFP. Using both T-cell recognition and biochemical assays, the authors showed that when these constructs were expressed in wild-type tumor cells, both epitopes were presented with comparable efficiency irrespective of their position within the chimeric GFP molecules.

The authors then expressed these constructs in tumor cells lacking the transporter associated with antigen processing (TAP), a heterodimeric complex that translocates cytosolic peptides into the endoplasmic reticulum. Immunization of mice with these TAP-deficient cells, a setting that depends on cross-priming, could elicit antigen-specific CD8⁺ T cells only when the relevant epitope was located in the C-terminal position of the chimeric protein. These data support the idea that cross-priming favors antigenic peptides located within mature proteins and reveal an unexpected bias against epitopes located within functional signal peptides.

Each of these studies square nicely with two recent reports describing a pathway in antigen-presenting dendritic cells for endocytosed material (presumably including intact proteins); this pathway leads from the phagosome to the cytosol and results in protein degradation and presentation via the endogenous class I MHC machinery (Fig. 1)^10,11. The finding that the location of an epitope within a larger protein can profoundly influence cross-priming of CD8⁺ T cells may help to clarify some of the controversy surrounding the ability of different antigens to participate in this process¹².

Figure 1: The long and short of cross-priming.

Deborah Maizels

Parenchymal cells (left) can generate surface complexes of class I MHC molecules occupied with virtually any cellular peptide of the correct length and sequence, including those found within a mature protein (red fragment) or derived from a signal sequence (green fragment). Cross-priming involves the uptake of cellular antigens by bone marrow–derived APCs (right), which are then directed to an endogenous class I MHC presentation pathway. The three new studies indicate that it is mature proteins, instead of smaller peptides, that are transferred from donor cell to APCs during cross-priming and suggest a bias against epitopes contained within signal sequences in this process. ER, endoplasmic reticulum.

Taken together, these three new studies provide a useful theoretical foundation for understanding the cross-priming pathway, and offer a practical framework for its exploitation. If cellular proteins and not peptides are the source material for cross-priming of CD8⁺ T cells, vaccination strategies based on maximizing steady-state expression levels and stability in donor cells would be predicted to generate superior CD8⁺ T cell responses. Further studies will certainly shed light on this and perhaps allow the meaningful translation of this new understanding from bench to bedside.