Mycobacterium tuberculosis (Mtb), the intracellular bacterium that causes tuberculosis (TB), is believed to have infected 25% of the world's population, making it a top public-health priority. On average, 5–10% of infected individuals develop active TB, leading to 10 million new cases yearly1. Disparities in health infrastructure have caused TB to remain endemic in many countries. Furthermore, the emergence of antibiotic resistance has made TB increasingly difficult to treat, and few new antibiotics are available. Vaccines represent the most cost-effective strategy to eliminate TB. The current TB vaccine, bacille Calmette–Guérin (BCG), is the most widely dispensed vaccine in the world. However, its efficacy continues to be debated. Clinical trials of BCG have shown up to 80% efficacy in the prevention of active pulmonary disease, particularly in infants and in temperate climate countries with low TB incidence. However, there is considerable geographic variation in BCG efficacy, and the vaccine performs poorly in TB-endemic countries2. While there is agreement that a more effective vaccine is needed, developing such a vaccine has been unpredictably difficult. In this issue, Hansen et al.3 vaccinate rhesus macaques with a rhesus CMV vector expressing six or nine Mtb antigens (RhCMV/TB) and show an unprecedented level of protection from TB.

Mtb is able to infect macrophages and occupies an intracellular niche following infection, requiring the activity of T cells to control infection. T cell production of cytokines such as tumor necrosis factor (TNF) and interferon (IFN)-γ enables macrophages to restrict Mtb growth. However, T cell activation also changes the nature of the inflammatory response to infection. IFN-γ-producing T cells decrease the accumulation of neutrophils, a cell type that, when infected, is permissive for bacterial replication, and enhance the influx of monocytes and macrophages, which are then better equipped to control Mtb replication4. Hence, a major focus of vaccine development has been to elicit large numbers of memory T cells that recognize and activate infected cells before Mtb can establish a niche in which to evade host immunity. However, protecting the lung against a bacterium that infects alveolar macrophages is no small feat. The adaptive immune response must locate the very small numbers of bacteria among a billion macrophages distributed among 480 million alveoli.

It is likely that the best vaccine will elicit T cells that are able to encounter the pathogen immediately after infection. While central memory T (TCM) cells maintain a presence in secondary lymphoid tissue, such as lymph nodes, effector memory T (TEM) cells are thought to circulate and patrol peripheral tissues5 (Fig. 1a). Tissue-resident memory T (TRM) cells, non-circulating memory cells that reside in parenchymal tissues, are positioned for early recognition of infected cells6. Currently, few vaccine candidates stimulate strong TEM responses and most elicit a mixture of TCM and TEM cells. Although vaccine-induced protection against Mtb correlates with a TCM phenotype in mice7, there are limited data. Mouse studies have shown that memory T cells are first activated in the draining lymph nodes of the lung, which slows the onset of memory-recall responses and reduces the potential benefit of vaccination8,9,10. However, even if T cells are not able to prevent infection outright, their ability to enforce latency and delay or prevent the emergence of clinical disease would reduce Mtb transmission and eventually reduce disease incidence. Thus, the preclinical benchmark for previous vaccines has been reducing the bacterial load.

Figure 1: A new strategy for developing a tuberculosis vaccine.
figure 1

Marina Corral Spence/Springer Nature

(a) While a vaccine that prevents infection altogether is the preferred goal of vaccination, sterilizing immunity appears possible using a vaccine that elicits TRM and TEM cell responses, as compared with TCM or naive T cell responses in unvaccinated individuals. (b) In comparison with prior vaccine strategies that primarily aim to elicit TCM cell responses, Hansen et al.3 use a CMV-base vaccine to elicit a response composed primarily of TEM cell responses, resulting in sterilizing immunity to TB in Mtb-challenged rhesus macaques. The experimental strategy used by Hansen et al including disease endpoints and possible mechanisms of RhCMV/TB vaccine action is shown. CFU, colony-forming units; Ab, antibody.

In the new study, Hansen et al.3 use CMV-vectored vaccines to prevent TB. CMV, like other herpesviruses, has a latency phase that is followed by periodic reactivation and stimulation of T cell responses. Thus in infected individuals an enormous part of the immune bandwidth is directed toward CMV. The choice of CMV as a vaccine vector makes use of the virus's natural periodic reactivation, leading to intermittent re-stimulation of Mtb-specific T cells and maintenance of a population of TEM cells.

The authors vaccinated rhesus macaques with three different RhCMV/TB vaccine constructs encoding Mtb proteins expressed during latency, early reactivation and active disease: the original '68-1' RhCMV backbone used previously in their simian immunodeficiency virus (SIV) studies11, containing nine Mtb antigens, which was found to generate non-classically restricted CD8+ T cells, and a new '68-1.2' RhCMV backbone containing either nine or six Mtb antigens, which does not generate non-classically restricted CD8+ T cells. When rhesus macaques were vaccinated, underwent vaccine boosting and were challenged with virulent Mtb nearly a year later, all three vaccines reduced the burden of TB similarly, as compared with unvaccinated controls, and prevented detectable infection in >40% of animals after Mtb challenge (Fig. 1b). This level of protection is impressive given the great susceptibility of rhesus macaques to Mtb relative to other nonhuman primate models, such as cynomolgus macaques, and also to humans. The ability of the RhCMV/TB vaccine generated by Hansen et al.3 to induce sterilizing immunity could represent a crucial leap forward in TB vaccine development. However, there remain some areas that require further clarification.

The authors' conclusion of achieving sterilizing immunity is based on the observation in inoculated macaques of an absence of radiological disease and negative bacterial cultures from stereotactic punch biopsies, which are done in a reproducible manner but do not necessarily target focal disease. The second experiment, which compared the efficacy of the 68-1 and 68-1.2 RhCMV vectors, used a low inocculum. In the absence of noticeable disease, a concern is that the animals 'cleared' of Mtb never had a productive infection. The generation of T cell responses to Mtb antigens that were not part of the vaccine is presented as evidence that all macaques were infected, but measuring bacterial chromosome equivalents by PCR might validate the initial infection12.

Another fundamental question is how RhCMV-based vaccines confer protection. Antibodies to Mtb antigens in RhCMV/TB were not detected after vaccination; instead, the authors attribute the protective immunity elicited by RhCMV/TB to high frequencies of TEM cells generated by the re-stimulation of Mtb-specific T cells from RhCMV/TB vaccination11. However, classical 'recall' T cell responses were observed in neither 68-1- nor 68-1.2-vaccinated animals upon Mtb challenge, indicating that these Mtb-specific memory TEM cells might not proliferate upon antigen exposure and may persist in a manner poised to confer immunity. An interesting facet of the 68-1 RhCMV vector is that it induces MHC-E-restricted and unconventional class II MHC–restricted CD8+ T cells, although these do not seem to be responsible for prevention of infection.

The authors were also able to detect a neutrophil-specific transcriptional signature in the blood after vaccination in protected animals, which may suggest a role for innate responses in vaccine-elicited protection. Thus, an alternative explanation for the conferred immunity is that RhCMV stimulated, or trained, innate immune cells that limit or prevent Mtb infection irrespective of the vaccine antigens, a contribution the authors also suggest. Therefore, the conclusion that vaccine-elicited TEM cells can prevent TB would be strengthened by the analysis of a group vaccinated with an empty RhCMV vector studied in parallel with either of the RhCMV/TB vaccines.

Finally, whether these principles can be used to design a vaccine for use in people remains to be determined. While RhCMV has considerable homology with human CMV, RhCMV/TB is a replication-competent virus. Such a virus could potentially be pathogenic in vulnerable people, whereas all macaques in this study were previously RhCMV seropositive. Thus, it is essential to determine whether an attenuated or replication-defective vector could elicit the type of T cell responses that are capable of mediating sterilizing immunity against Mtb. While the ultimate mechanism of protection deserves additional investigation, this report by Hansen et al.3 will no doubt reinvigorate TB vaccine research.