E98: Identification of Fusobacterium nucleatum-associated immune features and proteomic signature in colorectal cancer.

The association between the microbiome and colorectal cancer (CRC) development and prognosis has been firmly established, but the underlying mechanisms remain poorly characterized. Technical challenges to this work require novel methods for assessing microbiome content in archival tissue, where microbiology, cancer biology and immunobiology can be simultaneously studied. We hypothesized that a metaproteomics technique could illuminate important correlations between microbial taxa and tumor or immune parameters of interest.

Methods:

We performed metaproteomics to simultaneously quantify Fusobacterium nucleatum and human proteins in a pilot series of 21 archival CRCs using liquid chromatography-tandem mass spectrometry. Spectra were searched against a custom human/bacterial proteome database using ProLuCID, followed by manual validation and label-free quantification to assign peptide intensities for correlation with clinical features and human protein levels using parametric (linear regression) and non-parametric (Spearman’s test) methods. F. nucleatum–associated human CRC proteins were then searched using the Metascape platform to identify enriched ontology clusters.

Results:

F. nucleatum proteins (2,3-bisphosphoglycerate-dependent phosphoglycerate mutase and DUF4145 domain-containing protein) were repeatedly identified in CRC and significantly associated with tumor size (p < 0.01), stage (p < 0.05), lymphovascular invasion (p < 0.05), and CD3+/CD8+ intra-tumoral lymphocyte (TIL) density (p=0.001 and p=0.031, respectively). Among 396 F. nucleatum-associated human proteins, top enriched pathways included cellular stress response (p=10^-38), cytokine signaling (p=10^-36), viral infection (p=10^-30), and interferon signaling (p=10^-22); as well as proteins regulated by key immunologic transcription factors such as IRF1 (p=10^-5.2) and STAT1 (p=10^-3.4). STAT transcription factors themselves correlated strongly with F. nucleatum proteins (Figure).

Conclusions:

We used a customized tissue metaproteomics approach to demonstrate quantitative association of F. nucleatum with TIL density and critical immunologic pathways in CRC. The impact of this finding on CRC biology and prognosis remains to be determined in larger datasets. Quantitative assessment of bacterial proteins, such as those described here from F. nucleatum, offers hope to illuminate the specific impact of bacterial factors within tumors on oncologic or immunologic variables, and may enable the development of microbiome-based immunomodulatory therapies in CRC.