Human tumours are colonized by microorganisms1, collectively called the tumour microbiota, that can affect the microenvironment of the tumour — for example, by causing inflammation or local immune suppression2. This can lead to changes in how the body’s immune system responds to the tumour, and can alter responses to therapy3. But are intratumoral bacteria themselves recognized by the immune system? Writing in Nature, Kalaora et al.4 show that bacterial protein fragments called peptides are presented to the immune system on the surface of tumour cells, and are recognized by immune cells called T cells. This discovery might be exploited for cancer immunotherapeutics.
Molecules called tumour antigens enable the immune system to differentiate tumour cells from healthy cells. Every cell contains antigen-processing machinery, which enables antigen-derived peptides to be presented to the immune system by specialized molecules called human leukocyte antigens (HLAs) on the cell surface. HLA-presented peptides that are recognized by immune cells are called epitopes.
Tumour antigens come in two main categories: tumour-associated and tumour-specific5. Tumour-associated antigens are expressed in normal tissues as well as in tumours, and so do not readily activate immune responses. But if immune responses are mounted, there is a risk of harmful autoimmune reactions against the normal tissues that express the antigen. Nonetheless, because tumour-associated antigens are often found in multiple types of tumour and in many people who have cancer, they can be good targets for broadly applicable immunotherapies. By contrast, tumour-specific antigens are expressed solely on tumour cells and so are ideal targets for mounting a specific immune attack against tumours. One subtype, neoantigens, arises from tumour-specific gene mutations, and so neoantigens are typically tumour- and patient-specific. Targeting neoantigen-derived peptides (called neoepitopes) therefore necessitates truly personalized immunotherapies.
A skin cancer called melanoma has three known classes of tumour-associated antigen, and its cells typically carry many genetic mutations, resulting in a high likelihood of neoantigens6. It has therefore been at the forefront of tumour-antigen discovery and cancer-immunotherapy development7–9. It is thus fitting that Kalaora et al. use melanoma samples to describe another potential class of tumour antigen.
The authors set out to investigate the bacterial composition of 17 melanoma metastases (tumours formed when a cancer spreads from its original site to other regions of the body) from 9 people. They found that the composition of bacteria was highly similar in different metastases from the same individual, and sometimes in samples from different people. This finding indicates that particular bacterial species are common to melanoma, in line with a previous study reporting tumour microbiota specific to different types of cancer1. The authors also confirmed that these bacteria were present in melanoma cells, rather than in the surrounding extracellular microenvironment.
Kalaora and colleagues went on to investigate whether peptides from these intracellular bacteria are presented to the immune system in the same way as are other intracellular antigens. To this end, they used an approach based on mass spectrometry called immunopeptidomics, which allows direct detection of HLA-presented peptides. They found close to 300 peptides in their samples, from 33 bacterial species. Several peptides were found in more than one tumour from the same person and in tumours from different people.
The authors next asked whether the bacterial peptides are truly presented by melanoma cells, rather than by immune cells called antigen-presenting cells (APCs) that detect, take up and present pathogens to other cells of the immune system. The authors used an immune-cell marker protein to separate cells from two melanoma samples into APCs and tumour cells. Immunopeptidomics revealed that both groups of cell presented bacterial peptides. A subset of the peptides was presented by both APCs and tumour cells, indicating that the same peptide can both initiate an immune response through presentation on APCs and be a target for immune attack on tumour cells. The researchers then showed that T cells (which recognize HLA-presented peptides) isolated from the melanomas reacted to the identified bacterial peptides, including some of the peptides shared between tumours and individuals.
Taken together, Kalaora and colleagues’ results point to the possibility that tumour-displayed bacterial peptides are a previously unidentified class of tumour antigen (Fig. 1). However, several questions remain. To be bona fide tumour antigens, the identified bacterial species should not invade non-tumour tissue and their peptides should not be presented on HLAs on non-tumour cells. If this presentation were detected, the peptides would not qualify as immunotherapy targets. In addition, the bacterial peptides seem to be quite abundant (at least, compared with the numbers of identified melanoma neoepitopes7), so why does the body not mount an effective immune response against melanomas? Further studies of tumour-displayed bacterial peptides in combination with patient information will be needed to elucidate the potential clinical role of the peptides. Such data might help researchers to select suitable bacterial targets for cancer-immunotherapy approaches.
In conclusion, the bacterial peptides identified by Kalaora et al. could be attractive targets for immunotherapy. As bacterial peptides are ‘non-self’, it should be comparatively easy to elicit strong immune responses against them, and there would be no concerns about autoimmunity if it could be ascertained that they are not presented on any normal tissue. Thus, tumour-displayed bacterial peptides could serve as tumour-specific antigens shared between people — a rare and useful combination for therapeutics, so far seen only in virally induced tumours, in which epitopes can be derived from viral cancer-causing proteins5. Recent data indicate that tumour-invading bacteria might be a common phenomenon1,2. Kalaora and colleagues’ work could therefore lay the foundation for identifying shared tumour-specific antigens in a wide range of tumour types.
Nature 592, 28-29 (2021)