65 British volunteers have rolled up their sleeves to test a new approach to vaccination, for the first time in people. In this issue, McConkey et al. report success in the initial safety and immunogenicity studies of a malaria vaccine consisting of a DNA 'prime' followed by a 'boost' with a replication-defective poxvirus, modified vaccinia Ankara (MVA)1.

New approaches to immunization have sprung from the understanding of DNA and the ability to construct expression plasmids, recombinant viruses and recombinant bacteria. The first genetically engineered vaccine, a hepatitis B virus vaccine, used recombinant protein to raise protective antibodies. More recently, genetically engineered vaccines have been developed to elicit cell-mediated as well as humoral immunity. The ability to raise cell-mediated immunity has been necessary to combat such scourges as HIV/AIDS, malaria and tuberculosis.

Particularly powerful among these T-cell vaccines have been combinations of DNA and live viral vectors, in which a DNA vaccine is used to prime a T-cell response and a recombinant viral vaccine is used to boost the response, or in which one recombinant viral vector is used for priming and a second viral vector for boosting2.

These heterologous prime-boost immunizations elicit immune responses of greater height and breadth than can be achieved by priming and boosting with the same vector. The first immunogen initiates memory cells; the second immunogen expands the memory response. Outside of the immune response to the common vaccine insert, which undergoes a tremendous boost, the two agents do not raise responses against each other and thus do not interfere with each other's activity. In preclinical models, this strategy has reinvigorated efforts to construct vaccines for AIDS3,4, malaria5, tuberculosis6 and cancer7.

The malaria vaccine described by Hill and colleagues is the first human evaluation of a heterologous prime-boost vaccine. The DNA-MVA combination not only proved safe, but also elicited large numbers of T cells and provided partial protection against a malaria challenge, extending to humans the promise that such vaccines have shown in monkeys and mice.

To get at just the right vaccine combination, the investigators evaluated 20 different conditions of priming and boosting by rolling volunteers from priming studies into boosting studies. The scope and breadth of this initial trial staked boundaries for future trials, conducting dose escalations and testing different numbers of immunizations. Under the most favorable conditions, single modality immunizations elicited fewer than 100 responding T cells per million white blood cells. In contrast, immunizations combining DNA priming with MVA boosting resulted in frequencies of responding cells that were ten times higher.

Impressively, T cells were elicited against all regions of the 557-amino acid malaria protein expressed in the immunogen. This contrasted with the low response against a 232-amino acid multiepitope string that had been fused to the C terminus of the expressed protein. The more robust response to the natural protein than the multiepitope string is good news. Despite the power of genetic engineering, expressing proteins is far easier than defining the histocompatibility types in a target population, identifying immunodominant epitopes for a pathogen and constructing a string of epitopes.

For those of us developing vaccines for which T cells, not antibody, will be the correlate for protection, the study adroitly pioneers the screening of volunteers for vaccine-elicited T cells. White blood cells were harvested and placed directly in enzyme-linked immunospot analyses to allow an ex vivo count of responding T cells. Assays were conducted on fresh, rather than frozen, cells to facilitate the scoring of proapoptotic T cells in peak effector responses. Consistent with the temporal pattern of T-cell responses, points were taken at one, rather than at two, weeks after boosting, when antibody responses would have been at their peak. The ability to detect enzyme-linked immunospot analysis responses was optimized by accepting, rather than attempting to zero, the nonspecific chatter of peptide stimulations. And finally, data were presented as geometric means as well as medians, to better represent patient-to-patient variability in the numbers of elicited T cells.

This heterologous prime-boost protocol is but one of several currently in human trials. McMichael and Hanke, supported by the International AIDS Vaccine Initiative, have obtained promising results in DNA-MVA trials for HIV and AIDS, being conducted in England and Kenya8. In these trials, as in the current trial, the protein portion of the immunogen seems more effective than the multiepitope string at eliciting T cells. Approximately 800 volunteers are participating in trials for an AIDS vaccine developed by Merck that uses DNA priming and replication-defective adenovirus boosting and, more recently, replication-defective adenovirus priming followed by boosting with an avian poxvirus developed by Aventis. Many more modalities of priming and boosting are being tested in preclinical models. Provocatively, different protocols are raising different patterns of CD4+ and CD8+ T-cell responses9 and potentially different levels of T cells to dominant and subdominant epitopes. Thus, the rules of heterologous prime-boost immunizations are just beginning to be understood. Interesting science, as well as the potential to control major human pathogens, awaits those of us pursuing this powerful new approach to vaccination.