The human tumor microenvironment is colonized by cancer type–specific bacteria and fungi

DOI N°1: 10.1126/science.aay9189

DOI N° 2: 10.1016/j.cell.2022.09.005

Several studies have shown the importance of the gut microbiota in cancer progression and treatment response. Although bacteria were first detected in human tumors more than 100 years ago, the tumor microbiome has been overlooked, mainly because its characterization remains challenging. In 2020, Nejman et al. assessed the bacteriome of 1,526 tumors and their normal adjacent tissues (NATs) across seven cancer types, including breast, lung, ovary, pancreas, melanoma, bone, and brain tumors. Later, in 2022, Narunsky-Haziza et al. profiled the mycobiome from 1,183 tumors and NATs across eight tissue types, namely breast, lung, melanoma, ovary, colon, brain, bone, and pancreas.

Bacteria and fungi are detected in every tumor type analyzed

In their studies, Nejman et al. and Narunsky-Haziza et al. focused on solid tumors that represent either common cancers or for which the tumor microbiome is largely unknown such as melanoma, bone, and brain tumors. Bacterial and fungal DNA were quantified using quantitative PCR with universal primers targeting the bacterial 16S rRNA gene or the fungal 5.8 rRNA gene. Levels of bacterial and fungal DNA were detected in all tumor types, including in solid tumors that have no direct connection with the external environment. Microbial loads were higher in tumors than the corresponding NATs. The different cancer types varied in the levels of microbial DNA, with breast, bone, and pancreatic having the highest levels of bacterial and fungal DNA. Bacterial and fungal loads correlated across tumor types, but bacteria predominated over fungi in the tumor microbiome.

Bacteria and fungi are localized within the tumor microenvironment, either within immune cells or tumor cells

To validate the presence of bacteria, hundreds of additional tumors from different cancer types were stained for bacterial detection using fluorescence in situ hybridization (FISH) targeting the 16S rRNA gene and immunohistochemistry using antibodies against lipopolysaccharide (LPS) and lipoteichoic acid (LTA) to detect Gram-negative and Gram-positive bacteria, respectively. Bacteria were mainly localized in cancer or immune cells. Interestingly, LTA, indicative of Gram-positive bacteria, was the highest in melanoma tumors. Most of the skin commensals are Gram-positive bacteria (Corynebacterium, Cutibacterium, Staphylococcus, Streptococcus…), suggesting the recruitment of local bacteria by the tumors.

The localization of fungal cells was more challenging as no staining method can detect all fungi, so four staining methods with varying levels of sensitivity and specificity were used: (1) a fungal cell wall-specific anti-β-glucan antibody, (2) an anti-Aspergillus antibody that also binds several additional fungal species, (3) FISH against three conserved fungal 28S rRNA sequences, and (4) fungal cell wall-specific Gomori methenamine silver. Fungi showed cancer type-specific localization patterns: they were mainly within cancer cells in pancreatic, breast, and ovarian cancer, while they were mainly within macrophages in melanoma and lung tumors.

The tumor microbiome is comprised of living microorganisms

To determine whether live bacteria are present in the tumor microenvironment (TME), five fresh breast tumors were cultivated under a broad span of growth conditions to accommodate a high diversity of bacteria. Over 1,000 colonies were grown per tumor from four of the tumors, and 37 colonies were grown from the 5^th tumor. Whole-genome sequencing of 474 representative colonies from all five tumors identified 37 different bacterial species. For 105 of the colonies, bacteria could not be identified at species level. Overall, bacterial isolates were from three main phyla: Proteobacteria, Firmicutes, and Actinobacteria. To further validate the presence of living and metabolically active bacteria in human tumors, four freshly resected human breast tumors were cultured ex vivo with fluorescently labeled D-alanine or dimethyl sulfoxide (DMSO) control. D-alanine is used by bacteria to generate peptidoglycan but is not used by mammalian cells. Intracellular labeling was detected in all four tumors, thus supporting the hypothesis that the tumors harbor live intracellular bacteria.

The tumor microbiome composition is cancer-type-specific and shares similarities with the microbiome profile of the corresponding NATs

The tumor microbiome was sequenced targeting the 16S rRNA gene or the ITS2 to describe bacterial and fungal composition, respectively. Breast tumors had the most diverse bacteriome than all other tumor types. Beta-diversity analyses revealed that the tumor bacteriomes of the same cancer type tend to be more similar to each other than they are to the bacteriomes of other tumor types. A distinct microbiome across subtypes of the same tumor type was also observed. For example, the comparison of different subtypes of breast cancer according to their estrogen receptor, progesterone receptor, and HER2 status showed differences in the prevalence of multiple bacterial taxa. The overall bacterial composition of the different tumor types was relatively similar to their corresponding NATs, suggesting again the recruitment of local bacteria by the tumors. Similar conclusions could be drawn from the fungal profiling, where beta-diversity analyses showed the co-clustering of tumor and corresponding NAT samples, thus suggesting the possible seeding of the tumors from tissue-specific microbial ecologies. ITS2 sequencing analyses portrayed ubiquitous, low-abundance, and cancer-type-specific mycobiomes.

The tumor microbiome is associated with the tumor immune profile and cancer outcome

As bacteria can be found inside CD45+ immune cells, they might influence or reflect the immune state of the TME. In patients with metastatic melanoma, multiple bacterial taxa (46) were differentially abundant in ICI responders compared to non-responders. Taxa that were more abundant in tumors of responders included Clostridium, whereas Gardnerella vaginalis was more abundant in tumors of non-responders. Notably, this is consistent with the differential abundances observed in the gut microbiome of melanoma patients responding to ICI. Fungi are known to interact with bacteria through physical and biochemical mechanisms. Most significant fungi-bacteria co-occurrences presented in breast cancer, which had the most samples, potentially reflecting less power in other cancer types. Unsupervised analyses revealed three distinct fungi-bacteria-immune clusters driven by fungal co-occurrences, herein called “mycotypes,” named F1 (Malassezia-Ramularia-Trichosporon), F2 (Aspergillus-Candida), and F3 (multi-genera including Yarrowia). Narunsky-Haziza and colleagues then tested whether mycotypes were associated with immune responses. Log-ratios of immune cells co-occurring with F1, F2, or F3-clustered fungi significantly separated immune response, suggesting that different intratumoral mycobiomes may elicit distinct host responses. Two of the three significant comparisons were associated with higher inflammatory responses which also have the best survival prognosis.

The comprehensive analysis of thousands of human tumors demonstrated that the tumor microenvironment is colonized by cancer-type-specific bacteria and fungi. Whether or not microbes play a causal role in tumorigenesis, it is of interest to further explore the effects that intratumor bacteria and fungi may have on the different phenotypes of cancer cells and on immune responses. The manipulation of the tumor microbiome may also affect tumor immunity and the response to ICI. Thus, a better understanding of these effects may pave the way for novel treatment options for cancer patients.

The human tumor microbiome is composed of tumor type–specific intracellular bacteria

Nejman D, Livyatan I, Fuks G, Gavert N, Zwang Y, Geller LT, Rotter-Maskowitz A, Weiser R, Mallel G, Gigi E, Meltser A, Douglas GM, Kamer I, Gopalakrishnan V, Dadosh T, Levin-Zaidman S, Avnet S, Atlan T, Cooper ZA, Arora R, Cogdill AP, Khan MAW, Ologun G, Bussi Y, Weinberger A, Lotan-Pompan M, Golani O, Perry G, Rokah M, Bahar-Shany K, Rozeman EA, Blank CU, Ronai A, Shaoul R, Amit A, Dorfman T, Kremer R, Cohen ZR, Harnof S, Siegal T, Yehuda-Shnaidman E, Gal-Yam EN, Shapira H, Baldini N, Langille MGI, Ben-Nun A, Kaufman B, Nissan A, Golan T, Dadiani M, Levanon K, Bar J, Yust-Katz S, Barshack I, Peeper DS, Raz DJ, Segal E, Wargo JA, Sandbank J, Shental N, Straussman R. Science. 2020 May 29;368(6494):973-980. doi: 10.1126/science.aay9189. PMID: 32467386; PMCID: PMC7757858.

Abstract :

Keywords : –

Pan-cancer analyses reveal cancer-type-specific fungal ecologies and bacteriome interactions

Narunsky-Haziza L, Sepich-Poore GD, Livyatan I, Asraf O, Martino C, Nejman D, Gavert N, Stajich JE, Amit G, González A, Wandro S, Perry G, Ariel R, Meltser A, Shaffer JP, Zhu Q, Balint-Lahat N, Barshack I, Dadiani M, Gal-Yam EN, Patel SP, Bashan A, Swafford AD, Pilpel Y, Knight R, Straussman R. Cell. 2022 Sep 29;185(20):3789-3806.e17. doi: 10.1016/j.cell.2022.09.005. PMID: 36179670; PMCID: PMC9567272.

Abstract :

Keywords : tumor mycobiome, tumor microbiome, cancer biomarkers, fungi, 
microbial interactions, liquid biopsy, metagenomics, metatranscriptomics.

DOI: 10.1016/j.cell.2022.09.005

Expand your knowledge on the different microbiomes: