Effect of the intratumoral microbiota on spatial and cellular heterogeneity in cancer - PubMed (original) (raw)

. 2022 Nov;611(7937):810-817.

doi: 10.1038/s41586-022-05435-0. Epub 2022 Nov 16.

Hanrui Wu # 1, Kaitlyn D LaCourse # 1, Andrew G Kempchinsky 1, Alexander Baryiames 1, Brittany Barber 2, Neal Futran 2, Jeffrey Houlton 2 3, Cassie Sather 4, Ewa Sicinska 5, Alison Taylor 6, Samuel S Minot 7, Christopher D Johnston 8, Susan Bullman 9

Affiliations

Effect of the intratumoral microbiota on spatial and cellular heterogeneity in cancer

Jorge Luis Galeano Niño et al. Nature. 2022 Nov.

Abstract

The tumour-associated microbiota is an intrinsic component of the tumour microenvironment across human cancer types1,2. Intratumoral host-microbiota studies have so far largely relied on bulk tissue analysis1-3, which obscures the spatial distribution and localized effect of the microbiota within tumours. Here, by applying in situ spatial-profiling technologies4 and single-cell RNA sequencing5 to oral squamous cell carcinoma and colorectal cancer, we reveal spatial, cellular and molecular host-microbe interactions. We adapted 10x Visium spatial transcriptomics to determine the identity and in situ location of intratumoral microbial communities within patient tissues. Using GeoMx digital spatial profiling6, we show that bacterial communities populate microniches that are less vascularized, highly immuno‑suppressive and associated with malignant cells with lower levels of Ki-67 as compared to bacteria-negative tumour regions. We developed a single-cell RNA-sequencing method that we name INVADEseq (invasion-adhesion-directed expression sequencing) and, by applying this to patient tumours, identify cell-associated bacteria and the host cells with which they interact, as well as uncovering alterations in transcriptional pathways that are involved in inflammation, metastasis, cell dormancy and DNA repair. Through functional studies, we show that cancer cells that are infected with bacteria invade their surrounding environment as single cells and recruit myeloid cells to bacterial regions. Collectively, our data reveal that the distribution of the microbiota within a tumour is not random; instead, it is highly organized in microniches with immune and epithelial cell functions that promote cancer progression.

© 2022. The Author(s).

PubMed Disclaimer

Conflict of interest statement

S.B. has consulted for GlaxoSmithKline and BiomX. C.D.J. has consulted for Series Therapeutics and Azitra. S.B. is an inventor on US patent application no. PCT/US2018/042966, submitted by the Broad Institute and Dana-Farber Cancer Institute, which covers the targeting of Fusobacterium for the treatment of colorectal cancer. K.D.L. is currently employed by NanoString Technologies. The remaining authors declare no competing interests.

Figures

Fig. 1

Fig. 1. Assessing the spatial distribution of intratumoral bacteria throughout the tumour tissue.

a, Haematoxylin and eosin (H&E) staining (left), spatial distribution of total bacterial reads (centre) and total UMI transcripts (right) throughout the tumour tissue in the 10x Visium capture slides from human OSCC and CRC specimens. b, Pie chart of the top 10 most dominant bacterial genera detected in the 10x Visium RNA-sequencing data from the OSCC and CRC tumours. c, RNAscope-FISH imaging showing the distribution of bacteria across the tumour tissue in a sequential slide following the 10x Visium section. The F. nucleatum probe is red and the eubacterial probe is cyan. Scale bars, 1 mm. d, Spatial distribution of Parvimonas, Peptoniphilus and Fusobacterium UMIs detected in the 10x Visium OSCC specimen data. e, Spatial distribution of Fusobacterium, Bacteroides and Leptotrichia UMIs detected in the 10x Visium CRC specimen data.

Fig. 2

Fig. 2. Evaluating the effect of the tumour-associated microbiota in local microniches.

a, RNAscope-CISH images show the distribution of F. nucleatum (dark red) and other bacterial communities (eubacteria probe: cyan) in the tumour tissue; a sequential immunohistochemistry image shows the distribution of CD45+ (red) and PanCK+ (green) cells to identify the immune and epithelial compartments, respectively, in the tumour tissue. Inset images indicate representative AOIs that are positive and negative for bacteria and the corresponding UV exposure regions. b, Volcano plots from DSP data comparing the protein expression profiles in bacteria-positive AOIs and bacteria-negative AOIs from 8 OSCC (left) and 10 CRC (right) tumour specimens, referred to as microniche-level analysis. AOI comparative analysis, based on bacterial status, from immune (CD45+), epithelial (PanCK+) or combined (all AOIs) segmented data is shown. The number of AOIs per group is indicated. Dashed lines indicate the threshold of significant gene expression, defined as log2-transformed fold change ≥ 0.58 and ≤ −0.58 with −log10(P) ≥ 1.301 after linear mixed effect model (LMM) analysis and Benjamini–Hochberg multiple-correction testing. The p prefix indicates phosphorylation; ERK1/2 refers to ERK1 and ERK2; PR, progesterone receptor.

Fig. 3

Fig. 3. Effect of cell-associated intratumoral bacteria on transcriptomics in host single cells.

a, RNAscope-FISH (left) shows the distribution of intratumoral bacteria in a tumour from a patient with OSCC. Confocal images (right) show bacteria-associated single cells after tissue dissociation. Scale bars, 1 mm (left); 5 μm (right). b, Microbiome composition at the genus level after integration of tumour scRNA-seq data from seven patients with OSCC using the INVADEseq method. c, UMAP plots indicate host cell annotation and bacteria transcripts (UMI) from total bacteria and _Fusobacterium_- and _Treponema_-associated cells in integrated tumour single-cell data from seven patients with OSCC as indicated. Colour bars indicate the bacterial UMI transcripts for total bacteria and for each bacterial species as indicated. DCs, dendritic cells; MSCs, mesenchymal stem cells; Treg cells, regulatory T cells. d, GSEA analysis showing the signalling pathways that are differentially regulated in cells that contain ≥3 Fusobacterium UMI (High Fuso) or ≥3 Treponema UMI (High Trep) transcripts versus (vs) total bacteria-negative cells (Total Bac−) from the epithelial cell cluster. e, Volcano plots showing the differentially expressed genes between cell populations described in d. Dashed lines indicate the threshold of significant gene expression, defined as log2-transformed fold change ≤ −0.58 and ≥ 0.58 with −log10(P) ≥ 1.301. f, GSEA analysis showing the signalling pathways that are differentially regulated between total Fusobacterium (Total Fuso+) or total Treponema (Total Trep+) associated cells versus bacteria-negative cells (Total Bac−) in the monocyte-derived macrophage-v1 cell cluster. g, Volcano plots showing the differentially expressed genes between cell populations described in f. Dashed lines indicate the threshold of significant gene expression, defined as log2-transformed fold change ≤ −0.58 and ≥ 0.58 with −log10(P) ≥ 1.301. The normalized enrichment scores (NESs) in d,f were calculated using the Wilcoxon rank sum test. LMM analysis followed by Benjamini–Hochberg multiple-correction testing was used to calculate the fold change and P values for each gene in e,g.

Fig. 4

Fig. 4. F. nucleatum induces neutrophil swarming and the migration of cancer epithelial cells.

a, Live-cell confocal imaging showing neutrophil movements relating to CRC spheroids without (left) or with (right) F. nucleatum. Colour bars represent neutrophil cluster volume (µm3). Scale bars, 100 μm. b,c, Average speed (b) and cell displacement (c) of neutrophils migrating inside untreated control (lilac) and F. _nucleatum_-treated (red) spheroids. Red bars indicate mean. Data points represent individual tracks; n indicates the number of tracks per condition; three independent experiments. P values calculated by Mann–Whitney test. d, Neutrophil cell trajectory plots. e, Left, the log10-transformed fold change in volume over time of neutrophil clusters relative to the initial volume (T = 0 h). Data points represent average volume per time point, per condition. Right, quantification of the area under the curve for the fold change in volume. f, Confocal microscopy of HCT116 spheroid invasion capabilities without (left) or with (right) F. nucleatum over 19 h. Inset images represent differences in migration modes. Scale bar, 100 μm. g, The log10-transformed fold change in volume over time, representing the expansion rate of uninfected CRC spheroids. Error bars, s.d. h, Number of F. _nucleatum_-positive single cancer cells detaching from the spheroid over time. Error bars, s.d. i,j, Average speed (i) and cell displacement (j) of single cells escaping the F. _nucleatum_-infected spheroid. Red bars indicate mean. Data points represent individual tracks; n indicates the number of tracks per condition; three independent experiments. k, Cell trajectories of invading cancer cells escaping the F. _nucleatum_-infected spheroids. l, Signalling pathway analysis of CRC spheroids infected with F. nucleatum compared to uninfected control. The directed global significance (DGS) score was calculated as the square root of the mean squared _t_-statistic for genes in a gene set. ECM, extracellular matrix. m, Volcano plots of differential gene expression for selected pathways in F. _nucleatum_-infected spheroids compared to uninfected controls. Dashed lines indicate the threshold of significance, defined as log2-transformed fold change ≤ −0.58 and ≥ 0.58 and −log10(P) ≥ 1.301 after LMM analysis and Benjamini–Hochberg multiple-correction testing.

Extended Data Fig. 1

Extended Data Fig. 1. Heterogeneous distribution of the intratumoral microbiota throughout the tumour tissue.

a, Relative abundance of bacterial communities at the phylum and genus level for each tumour piece (n = 4 per patient) from 11 human CRC tumour specimens, identified via bulk 16S rRNA gene amplicon sequencing. Tumour tissue pieces (n = 4) are denoted as A, B, C and D from the 11 patients with CRC as indicated. b, Relative abundance of Fusobacterium genera in each tumour piece from 5 positive CRC specimens described in (a). c, Principal component analysis (PCoA) plot representing beta diversity clustering (Bray-Curtis Index) of bacterial communities at the genus level from each piece of CRC tumour tissue and PERMANOVA analysis. d, Dendrogram representing clustering of the microbiome composition at genus level in the tumour pieces as described in (a). The index of dissimilarity between samples was calculated using the Bray-Curtis test. Hierarchical cluster analysis was performed to detect clustering of patient specimens (colour bars) by using the Ward clustering algorithm. e, Top: RNAscope-FISH images indicating the spatial distribution of F. nucleatum and other bacterial communities (eubacteria) across the tumour tissue from a OSCC and CRC specimen. F. nucleatum probe is red and eubacterial probe is cyan. Middle: RNAscope-FISH images showing the negative control staining for the images. Bottom: Haematoxylin and eosin staining (H&E) of the RNAscope images. f. Validation of RNAscope probes. Left: PCR quantifying Fusobacterium nucleatum 16S rRNA gene in macro-dissected tissue based on RNAscope probe binding containing relatively high (Fuso. High) or low (Fuso. Low) F. nucleatum positivity as it is indicated. Right: Microbiome analysis using 16S rRNA gene sequencing in tumour areas that are “Fuso. High” or “Fuso. Low” as it is indicated. g, Schematic showing the experimental approach: RNAscope imaging was implemented to identify tumour areas positive for bacteria or F. nucleatum from OSCC and CRC tumours embedded in OCT blocks. Tumour tissues were trimmed to fit the capture area (6.5 mm x 6.5 mm) on the 10x Visium slide. Following tissue permeabilization, RNA is released from cells and bind to an array of probes that are attached to the surface of the slide within capture spots. Each probe has a unique molecular identifier (UMI) and a barcode sequence providing the spatial coordinates for each transcript. cDNA is generated from the captured RNA through a reverse transcription reaction. The barcoded cDNA is denatured and pooled and then further processed to generate cDNA libraries. All transcripts are aligned against the human transcriptome to map the human gene-expression profile across the sample. The unmapped reads are then aligned against microbial databases through GATK PathSeq to identify the microbiome composition. h, Distribution of the bacterial UMI count and bacterial reads for top bacterial genera detected in 10x Visium data from the OSCC and CRC cases as it is indicated.

Extended Data Fig. 2

Extended Data Fig. 2. The tumour-associated microbiota resides in highly immunosuppressive microniches with a low proliferation rate.

a, Experimental approach: GeoMx DSP was implemented to assess bacteria-associated microniches in one OSCC DSP cohort (n = 8) and two CRC DSP cohorts (RNAscope bacteria-positive cohort n = 10 and RNAscope bacteria-negative cohort n = 9). Sequential 4 µm-FFPE slides were prepared to identify spatial bacterial tumour distribution (RNAscope-CISH using F. nucleatum and eubacteria probes) and immunohistochemistry for immune (CD45+) and epithelial (PanCK+) compartments on the DSP slide treated with the 77-antibody panel. Segmented profiling for CD45+ and PanCK+ was performed on bacteria-positive AOIs (AOI_bac+) and bacteria-negative AOIs (AOI_bac-) per specimen, releasing photocleavable barcoded oligos for sequencing. Sequenced oligos provided the spatial information of the respective protein target in the bacteria positive or negative regions. b, RNAscope-CISH (left) showing the distribution of F. nucleatum (dark red), throughout the tumour tissue from a CRC specimen. A sequential slide (right) showed the distribution of the immune (CD45+; red) and epithelial (PanCK+; green) compartments by IHC staining. Inset images indicated the AOIs that were selected for DSP analysis from a bacteria-positive (Bac+) and bacteria-negative (Bac-) regions as it is indicate. Volcano plot showed the differential expression of genes from a single CRC sample comparing Bac+ with Bac- regions from the same tissue sample. Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≥0.58 and ≤−0.58 with a -Log10 p value ≥1.301 following LMM analysis and Benjamini–Hochberg multiple-correction testing. A 52-antibody panel were included here (this did not include the Cell Death and MAPK modules applied to DSP cohorts 1—3). c, Violin plots demonstrate the immuno‑suppressive microenvironment in bacteria-positive regions (Bac+) from the sample described in (b), highlighting the upregulation of ARG1 and CTLA4 and the enrichment of myeloid CD66b+ cells with lower expression of Ki67 and the T cell co-stimulatory molecule CD40 compared to bacteria-negative regions. p values calculated by t-test. d, Volcano plots indicate the differential gene-expression profile using the GeoMx DSP platform comparing AOIs from tumours (DSP cohort 2) that were RNAscope bacteria positive (Bac+; n = 120) against AOIs from tumours (DSP cohort 3) that were RNAscope bacteria negative (Bac-; n = 108). Using segmented analysis, the barcode oligos were collected either from the immune (CD45+) segment, epithelial (PanCK+) segment or both (All AOIs). Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≥0.58 and ≤−0.58 with a -Log10 p value ≥1.301 following LMM analysis and Benjamini–Hochberg multiple-correction testing.

Extended Data Fig. 3

Extended Data Fig. 3. The tumour-associated microbiota resides in T-cell-excluded areas, with lower proliferation capabilities.

a, RNAscope-FISH images showing bacteria-positive regions (Bac+) with the corresponding adjacent bacteria-negative region (Bac-) from a OSCC and CRC sample. A sequential IHC slide indicates the staining of CD8, PD-1, PanCK and Ki-67 in Bac+ and Bac- regions from the same tumour samples as it is indicated. Right panels indicate the CD8 fluorescent signal from a OSCC and CRC case comparing Bac+ vs Bac- regions from the same tumour tissue. b, Quantification of cell densities for PanCK, Ki-67 and PD-1, expressing cells in bacteria-positive regions in comparison to the contiguous bacteria-negative regions. p-values were calculated by Mann–Whitney test c, Representative RNAscope-FISH images showing bacteria-positive regions (Bac+) with the corresponding adjacent bacteria-negative region (Bac-) from a OSCC and CRC sample (dashed areas within images). A sequential IHC slide indicates the distribution of immune cell populations including myeloid CD66b+ or CD11b+ cells and CD4+ or CD8+ T cells. Magnified/inset images show the immune cell population that is more abundant in bacteria-positive (Bac+) and bacteria-negative (Bac-) regions for each tumour sample as it is indicated. d, Quantification of cell densities of CD66b, CD11b, CD4 and CD8 expressing cells in Bac+ regions compared to the contiguous Bac- region from the same (n = 4) OSCC (left) and (n = 4) CRC (right) tumour samples in two separate field of views. p-values were calculated by Mann–Whitney test. e, IHC images showing the CD45 fluorescent signal from the tumour samples described in Fig. 2a.

Extended Data Fig. 4

Extended Data Fig. 4. Detection of bacteria-associated single cells using the INVADEseq technique.

a, Schematic showing a modified gel bead emulsion (GEMs) by introducing a primer (1100R 16S) that targets a conserved region of the bacterial 16S rRNA, thus allowing cDNA generation of bacterial transcripts from the associated human single cells. In addition to the standard 10x genomics 5’ library preparation, bacterial cDNA was amplified with a nested conserved 16S rRNA gene primer and size selected libraries were sequenced and assessed through GATK PathSeq to identify bacterial taxa. Sequencing reads from the 16S rRNA amplified libraries retain the 10x genomics barcode sequence which facilitated mapping of annotated bacterial reads directly to the host single cells they are associated with. b, UMAP plots showing single-cell transcriptome of HT-29 cells with (orange dots) and without (blue dots) the 1100R 16S primer in the amplification mix before single-cell cDNA generation. UMAP plot inserts show the transcriptome for each condition, indicating no differences in the human gene-expression profile when the 1100R 16S primer was added. c, UMAP plots indicating single-cell transcriptome of HT-29 cells co-incubated with Bacteroides fragilis, Prevotella intermedia, Gemella haemolysans, Veillonella parvula and Escherichia coli DH5a for 3 h at MOI = 100. Insert table indicates the percentage of bacteria-associated single cells and total bacterial reads per cell per bacterial taxa. Note: Escherichia coli DH5a reads were not detected in human single cells. d, Experimental approach: HCT116 cancer cells were co-culture with either Fusobacterium nucleatum, Porphyromonas gingivalis or Prevotella intermedia at total multiplicity of infection (MOI) of 0, 100 and 500 for 3 hrs and processed for INVADEseq. e, Confocal images showing intracellular bacteria in HCT116 cancer cells after 3 h of incubation with F. nucleatum, P. gingivalis and P. intermedia as it is indicated. f, From top to bottom, UMAP plots from scRNA-seq data showing: Cell cluster distribution based on epithelial cell transcription, infected HCT116 cancer cells with F. nucleatum, P. gingivalis and P. intermedia, and expression level (UMI, bacterial load) of F. nucleatum and P. gingivalis transcripts in cancer cells following bacteria treatment for 3 h at multiplicity of infection (MOI) 0, 100 and 500 as it is indicated. Cluster ID indicates a unique transcriptional cellular group predicted by Seurat package (See methods). Colour bars indicate the expression level (UMI counts) of F. nucleatum and P. gingivalis transcripts as it is indicated. g, Top: Percentage of bacteria-associated cells positive for either F. nucleatum or P. gingivalis at MOI = 100 and 500 as it is indicated. Middle: Distribution of F. nucleatum and _P. gingivalis-_associated cells across all cell cluster annotated in (f) whereby all cell clusters combined, bacteria positive and negative, equal 100%. Bottom: Relative change in the percentages (Δ%) of cancer epithelial cell clusters between bacteria-associated cancer cells at MOI = 100 or 500 compared to the untreated control cell population (MOI = 0) for each cancer cell cluster annotated in (f). h, UMAP plots showing cancer epithelial cell clusters and detection of F. nucleatum and P. gingivalis transcripts following data integration from the experimental conditions described in (d). i, Dot plot showing the relative expression of gene markers for each ID cluster from CRC epithelial cells derived from the HCT116 cell line. Colour bars indicate the average expression level, and the dots represent the percentage expression level for each gene marker.

Extended Data Fig. 5

Extended Data Fig. 5. Differential gene expression and GSEA analysis comparing distinct CRC single-cell groups on the basis of bacteria association.

a, GSEA analysis indicating the signalling pathways that are differentially regulated in HCT116 cells co-incubated with Fusobacterium nucleatum at MOI = 500 for 3 h between different single-cell groups as follow: Top: Total _F. nucleatum_-associated cells (Total Fuso+) compared to total bacteria-negative cells (Total Bac-) in the entire sample “All clusters single cell analysis”. Middle: F. nucleatum_-associated cells that contain ≥3 F. nucleatum UMIs (High Fuso) compared to total bacteria-negative cells (Total Bac-) in the entire sample “_F. nucleatum single cell analysis”. Bottom: _F. nucleatum_-associated cells from cell cluster 5 compared to bacteria-negative cells from cluster 1 “Specific cell cluster analysis”. b, UMAP plots showing the cell population that are highlighted for each analysis described in (a), in which the F. nucleatum_-associated cells are coloured in red and the bacteria-negative cells are coloured in grey as it is indicated. Volcano plots indicate the differential gene-expression profile between the cell populations described in (a). Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≤−0.58 and ≥0.58 with a -Log10 p value ≥1.301. c, GSEA analysis indicates the signalling pathways that are differentially regulated in HCT116 cells co-incubated with Porphyromonas gingivalis at MOI = 500 for 3 h between different cellular groups as follow: Top: Total P. gingivalis -associated cells (Total Porph+) compared to total bacteria-negative cells (Total Bac-) in the entire sample “All clusters single cell analysis”. Middle: P. gingivalis -associated cells that contain ≥3 P. gingivalis UMIs (High Porph) compared to total bacteria-negative cells (Total Bac-) in the entire sample “_P. gingivalis single cell analysis”. Bottom: _P. gingivalis_-associated cells from cell cluster 6 compared to bacteria-negative cells from cluster 1 “Specific cell cluster analysis”. d, UMAP plots show the cell populations that are highlighted for each analysis described in (c), in which the _P. gingivalis_-associated cells are coloured in blue, and the bacteria-negative cell population are coloured in grey as it is indicated. Volcano plots indicate the differential gene-expression profile between the cell populations described in (c). Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≤−0.58 and ≥0.58 with a -Log10 p value ≥1.301. e, UMAP plots showing cell cluster distribution and the detection of bacterial transcripts in HT-29 cancer cells treated with or without F. nucleatum at MOI of 100 for 2 h. Colour bar indicates expression level (UMI counts) of F. nucleatum. f, GSEA analysis indicating the signalling pathways that are differentially regulated in HT-29 cancer cells treated with F. nucleatum compared to an uninfected control cancer cell population. g, Volcano plot showing the gene-expression profile in HT-29 cells treated with F. nucleatum relative to bacteria-negative cells. Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≥0.58 and ≤−0.58 with a -Log10 p value ≥1.301. Wilcoxon Rank Sum test was implemented to calculate the normalized enrichment score (NES) in panels (a), (c) and (f). A LMM analysis followed by Benjamini–Hochberg multiple-correction test was used to calculate the fold change and p-values for each gene in panels (b), (d) and (g).

Extended Data Fig. 6

Extended Data Fig. 6. Bacteria-associated single cells correlate with upregulation of cell inflammation and cancer progression pathways in patients with OSCC.

a, Dot plot showing the relative expression of gene markers for the T cell, myeloid and epithelial / mesenchymal cell compartment from single-cell data from patients with OSCC (n = 7 patients) as it is indicated in the cell clusters annotated in the UMAP plot from Fig. 3c. Colour bars indicate the average expression level, and the dots represent the percentage of expression for each gene marker. b, UMAP plot indicates the distribution of aneuploid and euploid cells in samples from patients with OSCC. c, Heat map shows the copy number variations (CNV) across 41,723 cells from the epithelial cell cluster as it is indicated. d, Percentage of aneuploid cells for each cell type annotated in Fig. 3c. Insert tables indicate the percentage of aneuploid cells and bacterial transcripts (UMIs) for each sub-cell clusters in the epithelial cell cluster as it is indicated. e, GSEA analysis showing the signalling pathways that are differentially regulated in cells from the sub-cell cluster 3, which contains the majority of aneuploid cancer cells, relative to other sub-cell clusters (All others) from the epithelial cell cluster. f, UMAP plot highlighting the epithelial sub-clusters that are contained in the epithelial cell cluster detected in Fig. 3c. g, Volcano plot showing the gene-expression profile in cancer cells from sub-cell cluster 3 in comparison to other sub-cell clusters from the epithelial cell cluster. Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≥0.58 and ≤−0.58 with a -Log10 p value ≥1.301. h, GSEA analysis showing the signalling pathways that are differentially regulated comparing total bacteria-associated cells (Total Bac+) vs bacteria-negative cells (Total Bac-) in the epithelial single cells as described in Fig. 3c. I, UMAP plot highlighting the transcriptional cellular group that is associated with bacterial invasion (Total Bac+; red cells) along with the bacteria-negative cell population (Total Bac-; grey cells). j, Volcano plot indicates the gene-expression profile in total bacteria-associated cell (Total Bac+) relative to total bacteria-negative cells (Total Bac-) in the epithelial single-cell cluster described in Fig. 3c. k, GSEA analysis showing the signalling pathways that are differentially regulated comparing total bacteria-associated cells (Total Bac+) vs bacteria-negative cells (Total Bac-) in the macrophage single cells as described in Fig. 3c. l, Volcano plot indicates the gene-expression profile in total bacteria-associated cell (Total Bac+) relative to total bacteria-negative cells (Total Bac-) in the macrophage cell cluster described in Fig. 3c. Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≥0.58 and ≤−0.58 with a -Log10 p value ≥1.301. Wilcoxon Rank Sum test was implemented to calculate the normalized enrichment score (NES) in panels (e), (h) and (k). A LMM analysis followed by Benjamini–Hochberg multiple-correction test was used to calculate the fold change and p-values for each gene in panels (g), (j) and (l).

Extended Data Fig. 7

Extended Data Fig. 7. Expression levels of genes that are correlated or anticorrelated with bacterial load (UMI) in patients with OSCC.

a, Expression level of genes as a function of single-cell-associated bacterial load (UMI) in seven patients with OSCC. The colour bars (top) indicate a gradient of bacterial load across the samples; the bacterial UMI values are designated for each patient. The correlation values were measured for each gene based on mean gene expression and bacterial UMI counts as it is indicated.

Extended Data Fig. 8

Extended Data Fig. 8. Differential gene expression, GSEA and microbiome analysis for each individual patient with OSCC.

a, UMAP plots showing host cell annotation and detected bacterial transcripts (UMI) from each patient with OSSC as indicated. b, Volcano plots indicate the differential gene-expression profile comparing total bacteria-associated cells against total bacteria-negative cells in the entire sample for the dominant bacterial genera for each patient with OSCC; a ≥ 3 bacteria UMI cut-off was applied for samples OSCC_11, OSCC_12, OSCC_13 and OSCC_14. Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≥0.58 and ≤−0.58 with a -Log10 p value ≥1.301 following LMM analysis and Benjamini–Hochberg multiple-correction testing. c, GSEA analysis showing the top 10 signalling pathways that are differentially regulated comparing total bacteria-associated cells against total bacteria-negative cells in the entire sample for the respective dominant bacterial species for each patient with OSCC as indicated. A ≥ 3 bacteria UMI cut-off was applied for samples OSCC_11, OSCC_12, OSCC_13 and OSCC_14. No differentially regulated pathways were detected in OSCC_15 based on Streptococcus positive and negative cells. d, Microbiome analysis using the INVADEseq technique (SC INVADEseq) on single cells and bulk 16S rRNA gene sequencing (Bulk 16S seq) on tissue homogenate from the same dissociated tissue piece for 5 of the patients with OSCC. Genera with ≥ 1% relative abundance in either SC INVADEseq or Bulk 16S seq data are shown. The coloured bar (right) indicates the bacterial load (UMI count) in which the OSCC samples were arranged from the highest to lowest based on the bacterial UMI count e, Percentages of bacteria-associated cells in the aneuploid and euploid enriched epithelial cell clusters for each patient with OSCC (data points), 6 patients that contained both aneuploid and euploid epithelial cells are included. Box-whiskers indicate medians and the interquartile range. The box represents the middle 50% of scores for each group. Red bars indicate the mean of the combined data. p-values indicates statistical significance calculated by two-tailed paired t test.

Extended Data Fig. 9

Extended Data Fig. 9. F. nucleatum induces the formation of cell clusters in immune and cancer cells.

a, Confocal images showing cluster formation of differentiated neutrophils derived from human HL-60 cells (green) co-cultured with F. nucleatum (pink) at different multiplicity of infection (MOI). Top micrographs display the raw imaging data. Bottom micrographs display the corresponding mask surfaces for each experimental condition by using Imaris software. Colour bar indicates the size (volume µm3) of the objects. b, Left: Violin plot indicates the quantification of volume of individual neutrophil clusters (data points) in the present of F. nucleatum as shown in (a); combined data from 4 independent experiments. Right: Dot plot shows the number of neutrophil clusters (objects) per field of view as indicated in (a). Data points represent the number of cell objects for each independent experiment (n = 4). p-values were calculated by one-way ANOVA followed by Bonferroni multiple comparison test. c, Flow cytometry plots show the levels of phosphorylation of ERK and p38 MAPK in neutrophils treated with F. nucleatum. Corresponding dot plots indicate the level of phosphorylation for each independent experiment (data points; n = 4). d, Spheroids derived from a mouse CRC cell line CT26WT were treated with or without F. nucleatum for 12 h and then embedded in collage matrices. The cell invasion capabilities for both conditions were evaluated using live-cell confocal imaging over a period of 19 h. Amplified images show the difference in the migration mode from both conditions. e, Log10 fold change volume over time of uninfected CRC spheroids revealing the expansion rate of uninfected control spheroids. Data points indicate the average values from three independent experiments. Error bars indicate the standard deviation (SD). f, Number of _F. nucleatum_-positive cancer cells that detached from the spheroid mass as single motile cells from three independent experiments. Errors bars indicate the SD. g-h, Distribution of the average speed and cell displacements of single cells that escape the spheroid mass infected with F. nucleatum. Combined data from three independent experiments. Red bars indicate the mean. Data points represents individual tracks; “n” indicates number of tracks per condition. i, Cell trajectories from an origin point of invading cancer cells that escape the spheroids infected with F. nucleatum. j, Signalling pathway analysis of CRC spheroids infected with F. nucleatum in comparison to uninfected control spheroids. k, Volcano plots showing the regulation of genes in selected signalling pathways in spheroids infected with F. nucleatum in comparison to uninfected spheroids. Dashed lines indicate the threshold of significant gene expression defined as the Log2 fold change ≤−0.58 and ≥0.58 with a -Log10 p value ≥1.301 following LMM analysis and Benjamini–Hochberg multiple-correction testing. l, Confocal images showing cell cluster formation of cancer cells derived from the human cell line HCT116 treated with F. nucleatum at different MOI as it is indicated. m, Left: Violin plot indicates the quantification of volume of individual cancer cell clusters (data points) in the present of F. nucleatum as shown in (l); combined data from 4 independent experiments. Right: Dot plot shows the number of cancer cell objects per field of view as indicated in (l). Data points represent the number of clusters for each independent experiment (n = 4). p-values were calculated by one-way ANOVA followed by Bonferroni multiple comparison test. n, Flow cytometry plots show the levels of phosphorylation of ERK and p38 MAPK in cancer cells treated with F. nucleatum. Corresponding dot plots indicate the level of phosphorylation for each independent experiment (data points; n = 4).

Comment in

Similar articles

Cited by

References

    1. Sepich-Poore GD, et al. The microbiome and human cancer. Science. 2021;371:eabc4552. doi: 10.1126/science.abc4552. - DOI - PMC - PubMed
    1. Nejman D, et al. The human tumor microbiome is composed of tumor type-specific intracellular bacteria. Science. 2020;368:973–980. doi: 10.1126/science.aay9189. - DOI - PMC - PubMed
    1. Kostic AD, et al. Genomic analysis identifies association of Fusobacterium with colorectal carcinoma. Genome Res. 2012;22:292–298. doi: 10.1101/gr.126573.111. - DOI - PMC - PubMed
    1. Rao A, Barkley D, Franca GS, Yanai I. Exploring tissue architecture using spatial transcriptomics. Nature. 2021;596:211–220. doi: 10.1038/s41586-021-03634-9. - DOI - PMC - PubMed
    1. Tang F, et al. mRNA-seq whole-transcriptome analysis of a single cell. Nat. Methods. 2009;6:377–382. doi: 10.1038/nmeth.1315. - DOI - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources