CoChat
Sign Up
Cited by 163Immunotherapy and Immune ResponsesCancer Immunotherapy and BiomarkersT-cell and B-cell Immunology

The DOI matches its Crossref record; 62 of the 62 references it cites were found in a registry or an index.

Microbial peptides activate tumour-infiltrating lymphocytes in glioblastoma

Open access
Naghavian, Reza, Faigle, Wolfgang, Oldrati, Pietro, Wang, Jian, Toussaint, Nora C., Qiu, Yuhan, Medici, Gioele, Wacker, Marcel, Freudenmann, Lena K., Bonté, Pierre-Emmanuel, Weller, Michael, Regli, Luca, Amigorena, Sebastian, Rammensee, Hans-Georg, Walz, Juliane S., Brugger, Silvio D., Mohme, Malte, Zhao, Yingdong, Sospedra, Mireia, Neidert, Marian C., Martin, Roland·2023·Nature
DOI 10.1038/s41586-023-06081-wPMC PMC10208956
Open this paper in CoChat — free accountOpen PDF

Abstract

Abstract Microbial organisms have key roles in numerous physiological processes in the human body and have recently been shown to modify the response to immune checkpoint inhibitors1,2. Here we aim to address the role of microbial organisms and their potential role in immune reactivity against glioblastoma. We demonstrate that HLA molecules of both glioblastoma tissues and tumour cell lines present bacteria-specific peptides. This finding prompted us to examine whether tumour-infiltrating lymphocytes (TILs) recognize tumour-derived bacterial peptides. Bacterial peptides eluted from HLA class II molecules are recognized by TILs, albeit very weakly. Using an unbiased antigen discovery approach to probe the specificity of a TIL CD4+ T cell clone, we show that it recognizes a broad spectrum of peptides from pathogenic bacteria, commensal gut microbiota and also glioblastoma-related tumour antigens. These peptides were also strongly stimulatory for bulk TILs and peripheral blood memory cells, which then respond to tumour-derived target peptides. Our data hint at how bacterial pathogens and bacterial gut microbiota can be involved in specific immune recognition of tumour antigens. The unbiased identification of microbial target antigens for TILs holds promise for future personalized tumour vaccination approaches.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41586-023-06081-w

Cites

62 of 62 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Elution of HLA class II-presented bacterial peptides from 19 glioblastomas and the stimulatory effect of IPdBPs on bulk TILs. a, Schematic of the immunopeptidome analyses of glioblastoma tissue, from which eluted peptides were annotated using the UniProt bacteria database (1). TILs or peripheral memory T cells were stimulated with IPdBPs (2 and 3). LC–MS, liquid chromatography–mass spectrometry; PBMC, peripheral blood mononuclear cell. The images in panel a were created using BioRender (https://biorender.com). b, Number of unique IPdBPs in patients with glioblastoma (n = 19) (data from patient 1635WI contains both primary and recurrent tumours together), with 344 unique IPdBPs identified (left). These peptides originated from 255 and 199 unique proteins and bacteria, respectively (3 technical replicates). The taxonomy of these bacteria shows that they primarily stem from Proteobacteria and Firmicutes at the phylum level (right). c, IPdBPs from primary and recurrent tumours (patient 1635WI) (top). As a control, peptidome data were also annotated using human protein database to identify eluted self-peptides. As shown, most of HLA class II-bound peptides are self-(human) peptides. IPdBPs of primary (bottom left, n = 18) and recurrent (bottom right, n = 23) tumours were separately annotated to their source bacteria at the phylum level. d, Bulk TILs (patient 1635WI) were stimulated with IPdBPs (See Supplementary Table 8), each in five replicate wells, and proliferation was measured after 5 days via 3H-thymidine incorporation. Proliferation is shown as a stimulatory index (SI), and peptides with SI ≥ 2 are considered stimulatory (the dotted line shows SI = 2 throughout the paper). The data are expressed as the mean value of all wells ± s.e.m. SIN3A* peptide was used as a positive control as TILs recognize this tumour antigen. T cells with APCs without any stimuli were used as a negative control. e, Heatmap of cytokine and chemokine secretion of TILs stimulated with three IPdBPs, which weakly stimulated proliferation. The secretion of each cytokine and chemokine was calculated by subtracting the corresponding cytokine and chemokine amount (pg ml−1) of the negative control. TILs with APCs without any stimuli were used as a negative control. Also see Extended Data Fig. 1. Source data
Fig. 1 Elution of HLA class II-presented bacterial peptides from 19 glioblastomas and the stimulatory effect of IPdBPs on bulk TILs. a, Schematic of the immunopeptidome analyses of glioblastoma tissue, from which eluted peptides were annotated using the UniProt bacteria database (1). TILs or peripheral memory T cells were stimulated with IPdBPs (2 and 3). LC–MS, liquid chromatography–mass spectrometry; PBMC, peripheral blood mononuclear cell. The images in panel a were created using BioRender (https://biorender.com). b, Number of unique IPdBPs in patients with glioblastoma (n = 19) (data from patient 1635WI contains both primary and recurrent tumours together), with 344 unique IPdBPs identified (left). These peptides originated from 255 and 199 unique proteins and bacteria, respectively (3 technical replicates). The taxonomy of these bacteria shows that they primarily stem from Proteobacteria and Firmicutes at the phylum level (right). c, IPdBPs from primary and recurrent tumours (patient 1635WI) (top). As a control, peptidome data were also annotated using human protein database to identify eluted self-peptides. As shown, most of HLA class II-bound peptides are self-(human) peptides. IPdBPs of primary (bottom left, n = 18) and recurrent (bottom right, n = 23) tumours were separately annotated to their source bacteria at the phylum level. d, Bulk TILs (patient 1635WI) were stimulated with IPdBPs (See Supplementary Table 8), each in five replicate wells, and proliferation was measured after 5 days via 3H-thymidine incorporation. Proliferation is shown as a stimulatory index (SI), and peptides with SI ≥ 2 are considered stimulatory (the dotted line shows SI = 2 throughout the paper). The data are expressed as the mean value of all wells ± s.e.m. SIN3A* peptide was used as a positive control as TILs recognize this tumour antigen. T cells with APCs without any stimuli were used as a negative control. e, Heatmap of cytokine and chemokine secretion of TILs stimulated with three IPdBPs, which weakly stimulated proliferation. The secretion of each cytokine and chemokine was calculated by subtracting the corresponding cytokine and chemokine amount (pg ml−1) of the negative control. TILs with APCs without any stimuli were used as a negative control. Also see Extended Data Fig. 1. Source dataCC BY · PubMed Central
Fig. 2 Isolation and characterization of tumour-specific TCCs by stimulating TILs with SIN3A* peptide. a, Bulk TILs were labelled with CFSE and stimulated with irradiated autologous PBMCs as APCs pulsed with SIN3A* peptide. After 10 days, 60 replicate wells were pooled and CFSEdimCD3+CD4+ T cells were single-cell sorted and expanded using PHA and human IL-2. The images in panel a were created using BioRender (https://biorender.com). b, All isolated TCCs were tested against the cognate antigen (SIN3A* peptide) for proliferation and IFNγ production. After 3 days, proliferation was measured using 3H-thymidine incorporation (SIs), and ELISA was used to measure IFNγ secretion. The response of each TCC was compared with the stimulation of the same TCC with APCs without peptide. c, TCRB V-gene and J-gene and functional phenotypes of several of the TCCs. d, Several TIL-derived TCCs with reactivity to tumour antigen (SIN3A*) were tested for reactivity against IPdBPs. TCCs were stimulated with peptides and irradiated PBMCs as APCs for 3 days. Proliferation was measured using 3H-thymidine incorporation. SIN3A* peptide served as the positive control. Mean SIs are shown in a heatmap format. SI < 2 (non-stimulatory) are shown with an empty field, and only SI ≥ 2 for stimulatory peptides are shown. Also see Extended Data Fig. 3. Source data
Fig. 2 Isolation and characterization of tumour-specific TCCs by stimulating TILs with SIN3A* peptide. a, Bulk TILs were labelled with CFSE and stimulated with irradiated autologous PBMCs as APCs pulsed with SIN3A* peptide. After 10 days, 60 replicate wells were pooled and CFSEdimCD3+CD4+ T cells were single-cell sorted and expanded using PHA and human IL-2. The images in panel a were created using BioRender (https://biorender.com). b, All isolated TCCs were tested against the cognate antigen (SIN3A* peptide) for proliferation and IFNγ production. After 3 days, proliferation was measured using 3H-thymidine incorporation (SIs), and ELISA was used to measure IFNγ secretion. The response of each TCC was compared with the stimulation of the same TCC with APCs without peptide. c, TCRB V-gene and J-gene and functional phenotypes of several of the TCCs. d, Several TIL-derived TCCs with reactivity to tumour antigen (SIN3A*) were tested for reactivity against IPdBPs. TCCs were stimulated with peptides and irradiated PBMCs as APCs for 3 days. Proliferation was measured using 3H-thymidine incorporation. SIN3A* peptide served as the positive control. Mean SIs are shown in a heatmap format. SI < 2 (non-stimulatory) are shown with an empty field, and only SI ≥ 2 for stimulatory peptides are shown. Also see Extended Data Fig. 3. Source dataCC BY · PubMed Central
Fig. 3 Bacteria-derived and microbiota-derived peptides can stimulate TCC88 as well as the tumour antigen. a, Peptides with the highest predicted stimulatory scores were identified by searching various databases (see Methods). Numbers of unique 10-mer peptides within each database, synthesized peptides, TCC88 stimulatory peptides and the corresponding percentage of stimulatory peptides within each database are shown. Proliferation was measured using 3H-thymidine incorporation (see Supplementary Table 10). *The immunopeptidome is annotated using human proteins in which 11 peptides score over 80%. Hence, the rest of the synthesized peptides were chosen from 60–80% of maximum score. RNA-seq, RNA sequencing. b, TCC88 was stimulated with bacteria-derived (top) and microbiota-derived (bottom) peptides for 3 days and proliferation was measured by 3H-thymidine incorporation (data represent the mean SI value of two replicate wells). HB, UniProt bacteria database; HGM, human gut microbiota34. c, Pie chart displaying the number of peptides from all databases that stimulated TCC88 equally well or better than the cognate antigen (SIN3A* tumour antigen) at a concentration of 10 µM. d, TCC88 was tested with several strongly stimulatory bacteria-derived and microbiota-derived and SIN3A* peptides at decreasing concentrations including 20, 2, 0.2, 0.02 and 0.002 µM. Irradiated BLS-DRB3*02:02 was pulsed with the peptides and used to stimulate TCC88 (data represent the mean SI value of five replicate wells ± s.e.m.). e, Supernatants were collected from TCC88 after stimulation with SIN3A*, HGM3 and HB84 peptides. Cytokine and chemokine secretion of TCC88 was examined and compared upon stimulation with these three different antigens. A specific amount of each cytokine and chemokine (pg ml−1) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). f, Cytotoxicity of TCC88 against the autologous tumour cell line after stimulation with 2 µM of SIN3A* and HGM3 (data represent the mean percentage of two replicate wells). E:T, effector to target. g, TCC88 was tested with HGM3 (10 µM) or SIN3A* (10 µM) peptides or both together (each 10 µM), and 3H-thymidine incorporation was measured after 3 days (data represent the mean SI value of three replicate wells ± s.e.m.). Also see Extended Data Figs. 3 and 4. Source data
Fig. 3 Bacteria-derived and microbiota-derived peptides can stimulate TCC88 as well as the tumour antigen. a, Peptides with the highest predicted stimulatory scores were identified by searching various databases (see Methods). Numbers of unique 10-mer peptides within each database, synthesized peptides, TCC88 stimulatory peptides and the corresponding percentage of stimulatory peptides within each database are shown. Proliferation was measured using 3H-thymidine incorporation (see Supplementary Table 10). *The immunopeptidome is annotated using human proteins in which 11 peptides score over 80%. Hence, the rest of the synthesized peptides were chosen from 60–80% of maximum score. RNA-seq, RNA sequencing. b, TCC88 was stimulated with bacteria-derived (top) and microbiota-derived (bottom) peptides for 3 days and proliferation was measured by 3H-thymidine incorporation (data represent the mean SI value of two replicate wells). HB, UniProt bacteria database; HGM, human gut microbiota34. c, Pie chart displaying the number of peptides from all databases that stimulated TCC88 equally well or better than the cognate antigen (SIN3A* tumour antigen) at a concentration of 10 µM. d, TCC88 was tested with several strongly stimulatory bacteria-derived and microbiota-derived and SIN3A* peptides at decreasing concentrations including 20, 2, 0.2, 0.02 and 0.002 µM. Irradiated BLS-DRB3*02:02 was pulsed with the peptides and used to stimulate TCC88 (data represent the mean SI value of five replicate wells ± s.e.m.). e, Supernatants were collected from TCC88 after stimulation with SIN3A*, HGM3 and HB84 peptides. Cytokine and chemokine secretion of TCC88 was examined and compared upon stimulation with these three different antigens. A specific amount of each cytokine and chemokine (pg ml−1) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). f, Cytotoxicity of TCC88 against the autologous tumour cell line after stimulation with 2 µM of SIN3A* and HGM3 (data represent the mean percentage of two replicate wells). E:T, effector to target. g, TCC88 was tested with HGM3 (10 µM) or SIN3A* (10 µM) peptides or both together (each 10 µM), and 3H-thymidine incorporation was measured after 3 days (data represent the mean SI value of three replicate wells ± s.e.m.). Also see Extended Data Figs. 3 and 4. Source dataCC BY · PubMed Central
Fig. 4 Response of bulk TILs to bacteria-derived and microbiota-derived peptides. a, Bulk TILs were stimulated with bacteria-derived and microbiota-derived peptides and irradiated autologous PBMCs as APCs. Proliferation was measured by 3H-thymidine incorporation after 5 days (data represent the mean value of five replicate wells ± s.e.m.). b, Overview of proliferative responses of several TCCs and bulk TILs to bacteria-derived and microbiota-derived peptides shown in a heatmap. SI < 2 are shown in white and SI ≥ 2 are shown in varying shades of grey. c, Bulk TILs were labelled with CFSE and separately stimulated with a pool of bacteria-derived and microbiota-derived peptides and a pool of vaccine or tumour peptides that includes peptides from somatic mutations and genes that were overexpressed in the autologous tumour (see Supplementary Table 13). Peptide-pulsed, irradiated autologous PBMCs were used as APCs, and bulk TILs were incubated for 10 days. Sixty replicate wells were pooled and proliferating CD4+ T cells (CFSEdim) were sorted. FSC-H, forward scatter height. d, Expanded CFSEdimCD4+ TILs that responded to the bacteria-derived and microbiota-derived peptide pool were stimulated with several tumour antigens, a bacteria-derived and microbiota-derived peptide pool and a pool of vaccine peptides. After 5 days, supernatants were harvested and proliferation was measured using 3H-thymidine incorporation (data represent the mean SI value of four replicate wells ± s.e.m.). GM-CSF was measured by ELISA and is depicted in (pg ml−1) after subtraction of the negative control. e, Part of the CFSEdimCD4+ TILs that responded to a bacteria-derived and microbiota-derived peptide pool and a vaccine peptide pool were used for TCRVβ sequencing. The percentage and frequency of TCR overlap of unique productive TCRVβ sequences of CFSEdim TILs responding to the vaccine peptide pool and TCRVβ sequences of CFSEdim TILs responding to the bacteria-derived and microbiota-derived peptide pool are shown. f, TCR sequence overlap of cross-reactive T cells in TILs recognizing that both the vaccine-derived peptide pool and the bacteria-derived and microbiota-derived peptide pool comprise 1% of entire TILs. Also see Extended Data Figs. 5 and 6. Source data
Fig. 4 Response of bulk TILs to bacteria-derived and microbiota-derived peptides. a, Bulk TILs were stimulated with bacteria-derived and microbiota-derived peptides and irradiated autologous PBMCs as APCs. Proliferation was measured by 3H-thymidine incorporation after 5 days (data represent the mean value of five replicate wells ± s.e.m.). b, Overview of proliferative responses of several TCCs and bulk TILs to bacteria-derived and microbiota-derived peptides shown in a heatmap. SI < 2 are shown in white and SI ≥ 2 are shown in varying shades of grey. c, Bulk TILs were labelled with CFSE and separately stimulated with a pool of bacteria-derived and microbiota-derived peptides and a pool of vaccine or tumour peptides that includes peptides from somatic mutations and genes that were overexpressed in the autologous tumour (see Supplementary Table 13). Peptide-pulsed, irradiated autologous PBMCs were used as APCs, and bulk TILs were incubated for 10 days. Sixty replicate wells were pooled and proliferating CD4+ T cells (CFSEdim) were sorted. FSC-H, forward scatter height. d, Expanded CFSEdimCD4+ TILs that responded to the bacteria-derived and microbiota-derived peptide pool were stimulated with several tumour antigens, a bacteria-derived and microbiota-derived peptide pool and a pool of vaccine peptides. After 5 days, supernatants were harvested and proliferation was measured using 3H-thymidine incorporation (data represent the mean SI value of four replicate wells ± s.e.m.). GM-CSF was measured by ELISA and is depicted in (pg ml−1) after subtraction of the negative control. e, Part of the CFSEdimCD4+ TILs that responded to a bacteria-derived and microbiota-derived peptide pool and a vaccine peptide pool were used for TCRVβ sequencing. The percentage and frequency of TCR overlap of unique productive TCRVβ sequences of CFSEdim TILs responding to the vaccine peptide pool and TCRVβ sequences of CFSEdim TILs responding to the bacteria-derived and microbiota-derived peptide pool are shown. f, TCR sequence overlap of cross-reactive T cells in TILs recognizing that both the vaccine-derived peptide pool and the bacteria-derived and microbiota-derived peptide pool comprise 1% of entire TILs. Also see Extended Data Figs. 5 and 6. Source dataCC BY · PubMed Central
Fig. 5 Peripheral memory T cells recognize several tumour antigens after activation with bacteria-derived and microbiota-derived peptides. a, CD45RA− peripheral T cells were isolated and labelled with CFSE. Cells were then stimulated with the bacteria-derived and microbiota-derived peptide pool for 10 days. Next, 60 replicate wells were pooled and proliferative (CFSEdim) CD4+ T cells were sorted and short-term expanded. b, Bacteria-reactive or microbiota-reactive peripheral CD4+ T cells responded to several tumour antigens from the autologous tumour when stimulated with peptide presented by irradiated autologous PBMCs as APCs (top). Proliferation was measured after 5 days by 3H-thymidine incorporation (data represent the means of five replicate wells ± s.e.m.). Peripheral blood memory T cells were isolated and directly tested with tumour antigens (bottom). Proliferation was measured after 5 days by 3H-thymidine incorporation (data represent the means of five replicate wells ± s.e.m.). c,d, Part of CFSEdim (that is, proliferating) peripheral CD4+ T cells that responded to the bacteria-derived and microbiota-derived peptide pool was used for TCRVβ sequencing. The percentage and frequency of TCR overlap of unique productive TCRVβ sequences of bacteria-reactive and microbiota-reactive peripheral CD4+ (CFSEdim) and TCRVβ sequences of bulk TILs (c), and the overlap of unique productive TCRVβ sequences of bacteria-reactive and microbiota-reactive peripheral CD4+ (CFSEdim) and TCRVβ sequences of (CFSEdim) bacteria-reactive and microbiota-reactive TILs (d) are shown. Also see Extended Data Figs. 6 and 7. Source data
Fig. 5 Peripheral memory T cells recognize several tumour antigens after activation with bacteria-derived and microbiota-derived peptides. a, CD45RA− peripheral T cells were isolated and labelled with CFSE. Cells were then stimulated with the bacteria-derived and microbiota-derived peptide pool for 10 days. Next, 60 replicate wells were pooled and proliferative (CFSEdim) CD4+ T cells were sorted and short-term expanded. b, Bacteria-reactive or microbiota-reactive peripheral CD4+ T cells responded to several tumour antigens from the autologous tumour when stimulated with peptide presented by irradiated autologous PBMCs as APCs (top). Proliferation was measured after 5 days by 3H-thymidine incorporation (data represent the means of five replicate wells ± s.e.m.). Peripheral blood memory T cells were isolated and directly tested with tumour antigens (bottom). Proliferation was measured after 5 days by 3H-thymidine incorporation (data represent the means of five replicate wells ± s.e.m.). c,d, Part of CFSEdim (that is, proliferating) peripheral CD4+ T cells that responded to the bacteria-derived and microbiota-derived peptide pool was used for TCRVβ sequencing. The percentage and frequency of TCR overlap of unique productive TCRVβ sequences of bacteria-reactive and microbiota-reactive peripheral CD4+ (CFSEdim) and TCRVβ sequences of bulk TILs (c), and the overlap of unique productive TCRVβ sequences of bacteria-reactive and microbiota-reactive peripheral CD4+ (CFSEdim) and TCRVβ sequences of (CFSEdim) bacteria-reactive and microbiota-reactive TILs (d) are shown. Also see Extended Data Figs. 6 and 7. Source dataCC BY · PubMed Central
Extended Data Fig. 1 Confirming the presence of bacteria in tumor tissues (see also Fig. 1, Extended Data Fig. 2 and Supplementary Table 4). a, HLA-II immunopeptidome analyses for bacterial peptides in control including brain lesions of 3 MS patients and brain tissue of 6 healthy donors. A total of 55 bacterial peptides was identified from all the samples. Identical bacterial peptides in the control cohort were removed from the GBM tumor-specific bacterial peptides. b, 16S rRNA gene sequencing confirmed bacteria in tumor tissues of 10 GBM patients, from whom immunopeptidome analyses were available. OTUs relative abundance in read counts were shown at the phylum level for each tumor (Supplementary Tables 5 and 6). * 16S rRNA data from patient 1635WI includes both primary and recurrent tumors of the patient in triplicates. OTUs relative abundance was calculated based on mean OTU of the triplicates divided by total OTUs in each tumor. c, Deduced core binding motif all HLA-DR-derived peptides annotated to human/tumor proteins and UniProt bacterial peptides from primary and recurrent tumors (patient 1635WI) were compared using iceLogo (p = 0.05, two-tailed independent t-test by IceLogo58). Light blue squares show the known anchor positions for corresponding HLA-DR molecules62.
Extended Data Fig. 1 Confirming the presence of bacteria in tumor tissues (see also Fig. 1, Extended Data Fig. 2 and Supplementary Table 4). a, HLA-II immunopeptidome analyses for bacterial peptides in control including brain lesions of 3 MS patients and brain tissue of 6 healthy donors. A total of 55 bacterial peptides was identified from all the samples. Identical bacterial peptides in the control cohort were removed from the GBM tumor-specific bacterial peptides. b, 16S rRNA gene sequencing confirmed bacteria in tumor tissues of 10 GBM patients, from whom immunopeptidome analyses were available. OTUs relative abundance in read counts were shown at the phylum level for each tumor (Supplementary Tables 5 and 6). * 16S rRNA data from patient 1635WI includes both primary and recurrent tumors of the patient in triplicates. OTUs relative abundance was calculated based on mean OTU of the triplicates divided by total OTUs in each tumor. c, Deduced core binding motif all HLA-DR-derived peptides annotated to human/tumor proteins and UniProt bacterial peptides from primary and recurrent tumors (patient 1635WI) were compared using iceLogo (p = 0.05, two-tailed independent t-test by IceLogo58). Light blue squares show the known anchor positions for corresponding HLA-DR molecules62.CC BY · PubMed Central
Extended Data Fig. 2 HLA-II expression of tumor tissue and cancer cell line. a, RNA-seq analysis of the tumor tissue (1635WI) was conducted by TCGA-RNAseqv2 pipeline (https://webshare.bioinf.unc.edu/public/mRNAseq_TCGA/UNC_mRNAseq_summary.pdf)25. Normalized gene counts of HLA-DR were compared with The Cancer Genome Atlas (TCGA) GBM cohort. White square represents normal expression of the gene in the tumor (Supplementary Table 14). b, Cancer cells isolated from the tumor tissue (patient 1635WI) and expanded in vitro were stained with HLA-DR antibody. c, Paraffin-embedded recurrent tumor tissue (patient 1635WI) was stained for GFAP, CD31, CD3, CD68 and HLA-DR25. HLA-DR is present on the surface of immune cells (CD3+), endothelial cells (CD31) and GFAP+ astrocytes (6 different regions in n = 1 section of the tumor were used for this analysis). Scale bar: 200 µm.
Extended Data Fig. 2 HLA-II expression of tumor tissue and cancer cell line. a, RNA-seq analysis of the tumor tissue (1635WI) was conducted by TCGA-RNAseqv2 pipeline (https://webshare.bioinf.unc.edu/public/mRNAseq_TCGA/UNC_mRNAseq_summary.pdf)25. Normalized gene counts of HLA-DR were compared with The Cancer Genome Atlas (TCGA) GBM cohort. White square represents normal expression of the gene in the tumor (Supplementary Table 14). b, Cancer cells isolated from the tumor tissue (patient 1635WI) and expanded in vitro were stained with HLA-DR antibody. c, Paraffin-embedded recurrent tumor tissue (patient 1635WI) was stained for GFAP, CD31, CD3, CD68 and HLA-DR25. HLA-DR is present on the surface of immune cells (CD3+), endothelial cells (CD31) and GFAP+ astrocytes (6 different regions in n = 1 section of the tumor were used for this analysis). Scale bar: 200 µm.CC BY · PubMed Central
Extended Data Fig. 3 T cell cloning using SIN3A* tumor antigen and graphical summary of TCC isolation, response of TCC88 to ps-SCL, and depiction of database screening for antigen discovery approach (see also Figs. 1 and 2 and Extended Data Fig. 4). a, Gating strategy for isolation of TCCs from TILs responding to SIN3A* peptide. 60 replicate wells of CFSE-labeled TILs were pooled after 10 days stimulation with SIN3A* tumor antigen. Cells were stained using live-dead, CD3, CD4 and were next single-cell sorted. b, Schematic depiction of the steps for discovering target antigens for a TCC. TCCs are tested with positional scanning combinatorial peptide libraries (200 mixtures of decamer peptides) presented by BLS cells expressing a single HLA-DR molecule, which had been previously established as the restriction element for each TCC. Proliferative or cytokine responses of TCC to each of the decamer mixtures, i.e. 200 compounds are measured. Data are then used to create a scoring matrix. Antigens with predicted high stimulatory potential are identified by searches of several databases including human proteome, autologous tumor transcriptome and peptidome, pathogenic bacteria and viruses as well as human gut microbiota. Candidate peptides originating from pathogenic bacteria and gut microbiota strongly stimulated TCC88. Further, these peptides were used to stimulate other tumor-reactive TCCs as well as bulk TILs and peripheral memory CD4+ T cells. The schematics in panel b were created using BioRender (https://biorender.com).
Extended Data Fig. 3 T cell cloning using SIN3A* tumor antigen and graphical summary of TCC isolation, response of TCC88 to ps-SCL, and depiction of database screening for antigen discovery approach (see also Figs. 1 and 2 and Extended Data Fig. 4). a, Gating strategy for isolation of TCCs from TILs responding to SIN3A* peptide. 60 replicate wells of CFSE-labeled TILs were pooled after 10 days stimulation with SIN3A* tumor antigen. Cells were stained using live-dead, CD3, CD4 and were next single-cell sorted. b, Schematic depiction of the steps for discovering target antigens for a TCC. TCCs are tested with positional scanning combinatorial peptide libraries (200 mixtures of decamer peptides) presented by BLS cells expressing a single HLA-DR molecule, which had been previously established as the restriction element for each TCC. Proliferative or cytokine responses of TCC to each of the decamer mixtures, i.e. 200 compounds are measured. Data are then used to create a scoring matrix. Antigens with predicted high stimulatory potential are identified by searches of several databases including human proteome, autologous tumor transcriptome and peptidome, pathogenic bacteria and viruses as well as human gut microbiota. Candidate peptides originating from pathogenic bacteria and gut microbiota strongly stimulated TCC88. Further, these peptides were used to stimulate other tumor-reactive TCCs as well as bulk TILs and peripheral memory CD4+ T cells. The schematics in panel b were created using BioRender (https://biorender.com).CC BY · PubMed Central
Extended Data Fig. 4 (see also Fig. 3). HLA-DR restriction and response to autoantigens of TCC88. a, Several of the isolated TCCs were stimulated with irradiated BLS cells expressing single HLA-DR molecules of patient 1635WI (BLS-DRB1*03:01, BLS-DRB1*04:02, BLS-DRB3*02:02 and BLS-DRB4*01:01) as APCs and dose titration of SIN3A* peptide (10, 1, 0.1, 0.01 and 0.001 µM). Proliferation was measured using 3H-thymidine incorporation after 3 days. Data represents mean SI of 5 replicate wells ± SEM). Several TCCs did not respond to peptides presented on any of HLA-DR molecules (data not shown). b-c, Median IFN-γ response of three independent experiments (transformed to log base 10 values) of TCC88 stimulated via ps-SCL was measured and depicted for 20 L-amino acids fixed in all positions of a 10-mer peptide. c, IFN-γ values were depicted in a table format (scoring matrix) and represent the relative stimulatory potency of each L-amino acid in different positions of the decamers. d, Relationship between predicted score and stimulatory index (SI) was established by testing potential cognate antigens within the predicted score range of 60–100% of the maximal score with TCC88. e–f, TCC88 was tested with pools of overlapping peptides (including 260 peptides) covering brain, pancreas and skin autoantigens (Supplementary Table 11) including peptide pools covering myelin basic protein (MBP pools), myelin oligodendrocyte glycoprotein (MOG pools), myelin proteolipid protein (PLP pool contains 5 immunodominant PLP peptides), RAS guanyl-releasing protein 2 (RASGRP2 pools),GDP-L-fucose synthase (TSTA3 pools), MELAN-A, tyrosinase (TYR), premelanosome protein (PMEL), insulin, SLC30A8 (ZnT8) and islet amyloid polypeptide (IAPP). TCC88 was stimulated with peptide-pulsed irradiated BLS-DRB3*02:02 cells for 3 days. Proliferation was measured using 3H-thymidine incorporation (data represents mean value of 3 replicate wells ±SEM). f, Response of TCC88 to unmutated SIN3A peptide was compared to SIN3A* at different concentrations including 10, 1, 0.1, 0.01 and 0.001 µM. Source data
Extended Data Fig. 4 (see also Fig. 3). HLA-DR restriction and response to autoantigens of TCC88. a, Several of the isolated TCCs were stimulated with irradiated BLS cells expressing single HLA-DR molecules of patient 1635WI (BLS-DRB1*03:01, BLS-DRB1*04:02, BLS-DRB3*02:02 and BLS-DRB4*01:01) as APCs and dose titration of SIN3A* peptide (10, 1, 0.1, 0.01 and 0.001 µM). Proliferation was measured using 3H-thymidine incorporation after 3 days. Data represents mean SI of 5 replicate wells ± SEM). Several TCCs did not respond to peptides presented on any of HLA-DR molecules (data not shown). b-c, Median IFN-γ response of three independent experiments (transformed to log base 10 values) of TCC88 stimulated via ps-SCL was measured and depicted for 20 L-amino acids fixed in all positions of a 10-mer peptide. c, IFN-γ values were depicted in a table format (scoring matrix) and represent the relative stimulatory potency of each L-amino acid in different positions of the decamers. d, Relationship between predicted score and stimulatory index (SI) was established by testing potential cognate antigens within the predicted score range of 60–100% of the maximal score with TCC88. e–f, TCC88 was tested with pools of overlapping peptides (including 260 peptides) covering brain, pancreas and skin autoantigens (Supplementary Table 11) including peptide pools covering myelin basic protein (MBP pools), myelin oligodendrocyte glycoprotein (MOG pools), myelin proteolipid protein (PLP pool contains 5 immunodominant PLP peptides), RAS guanyl-releasing protein 2 (RASGRP2 pools),GDP-L-fucose synthase (TSTA3 pools), MELAN-A, tyrosinase (TYR), premelanosome protein (PMEL), insulin, SLC30A8 (ZnT8) and islet amyloid polypeptide (IAPP). TCC88 was stimulated with peptide-pulsed irradiated BLS-DRB3*02:02 cells for 3 days. Proliferation was measured using 3H-thymidine incorporation (data represents mean value of 3 replicate wells ±SEM). f, Response of TCC88 to unmutated SIN3A peptide was compared to SIN3A* at different concentrations including 10, 1, 0.1, 0.01 and 0.001 µM. Source dataCC BY · PubMed Central
Extended Data Fig. 5 Response of TILs and TCCs to bacteria/microbiota-derived peptides (see also Figs. 3 and 4). a, The source bacterium of bacteria/microbiota-derived peptides that stimulate TCC88 is shown at the phylum and class levels. Source proteins of the peptides were used to check within UniProt databases and taxonomy data was then extracted. b, Heatmap result analyzing all peptides that were stimulatory or non-stimulatory for TCC88 with the aim to examine if amino acid motifs are present in stimulatory peptides (Green, black and red represent higher, average and lower prevalence of distinct amino acids in the positions of the tested 10-mer peptides. IceLogo58 utilizes the z-score to calculate amino acid specific p-values at each position via an error function58). c, Supernatants were collected from bulk TILs stimulated with bacteria/microbiota-derived peptides after 5 days. IFN-γ secretion of TILs was measured by ELISA (data represents mean value of 5 replicate wells ± SEM). d, Several TIL-derived TCCs were stimulated with bacteria/microbiota-derived peptides and irradiated autologous PBMCs as APCs. Proliferation was measured by 3H-thymidine incorporation after 3 days (data represents mean value of 2 replicate wells). e, Supernatants were collected after stimulation of TCC75 with SIN3A* and HGM27 peptides for 3 days. Cytokine/chemokine secretion of TCC75 was examined for the different antigens and depicted in a heatmap. Specific amount of each cytokine/chemokine (pg/ml) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). Source data
Extended Data Fig. 5 Response of TILs and TCCs to bacteria/microbiota-derived peptides (see also Figs. 3 and 4). a, The source bacterium of bacteria/microbiota-derived peptides that stimulate TCC88 is shown at the phylum and class levels. Source proteins of the peptides were used to check within UniProt databases and taxonomy data was then extracted. b, Heatmap result analyzing all peptides that were stimulatory or non-stimulatory for TCC88 with the aim to examine if amino acid motifs are present in stimulatory peptides (Green, black and red represent higher, average and lower prevalence of distinct amino acids in the positions of the tested 10-mer peptides. IceLogo58 utilizes the z-score to calculate amino acid specific p-values at each position via an error function58). c, Supernatants were collected from bulk TILs stimulated with bacteria/microbiota-derived peptides after 5 days. IFN-γ secretion of TILs was measured by ELISA (data represents mean value of 5 replicate wells ± SEM). d, Several TIL-derived TCCs were stimulated with bacteria/microbiota-derived peptides and irradiated autologous PBMCs as APCs. Proliferation was measured by 3H-thymidine incorporation after 3 days (data represents mean value of 2 replicate wells). e, Supernatants were collected after stimulation of TCC75 with SIN3A* and HGM27 peptides for 3 days. Cytokine/chemokine secretion of TCC75 was examined for the different antigens and depicted in a heatmap. Specific amount of each cytokine/chemokine (pg/ml) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). Source dataCC BY · PubMed Central
Extended Data Fig. 6 TILs depleted of TCC88 and also other TCCs respond to bacteria/microbiota-derived peptides (see also Figs. 3 and 4). a, Gating strategy for bulk TILs using CD4, CD8, PE- and FITC-TCRVβ antibodies. Two populations of CD4+ PE-TCRVβ+ and CD4+ FITC-TCRVβ+ were sorted and later mixed together. Unstained TILs were used as control to demonstrate specific staining of TILs with TCRVβ antibodies of interest. b, Sorted CD4+ PE and FITC TCRVβ+ were mixed and rested. Cells were then stimulated with irradiated autologous PBMCs pulsed with several bacteria/microbiota-derived peptides at 10 µM and proliferation was measured after 3 days using 3H-thymidine incorporation (data represents mean value of 3 replicate wells ± SEM). c, Supernatants were tested for secretion of a broader range of cytokines/chemokines. Cytokine/chemokine secretion of bacteria/microbiota-reactive TILs was measured after stimulation with vaccine and bacteria/microbiota-derived peptide pools (left panel). Cytokine/chemokine secretion of bacteria/microbiota-reactive peripheral CD4+ T cells (selected peripheral CD4+) and CD45RA− peripheral T cells (unselected peripheral CD4+) in response to a tumor antigen (LRP6) was measured (right panel). Specific amount of each cytokine/chemokine (pg/ml) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). Source data
Extended Data Fig. 6 TILs depleted of TCC88 and also other TCCs respond to bacteria/microbiota-derived peptides (see also Figs. 3 and 4). a, Gating strategy for bulk TILs using CD4, CD8, PE- and FITC-TCRVβ antibodies. Two populations of CD4+ PE-TCRVβ+ and CD4+ FITC-TCRVβ+ were sorted and later mixed together. Unstained TILs were used as control to demonstrate specific staining of TILs with TCRVβ antibodies of interest. b, Sorted CD4+ PE and FITC TCRVβ+ were mixed and rested. Cells were then stimulated with irradiated autologous PBMCs pulsed with several bacteria/microbiota-derived peptides at 10 µM and proliferation was measured after 3 days using 3H-thymidine incorporation (data represents mean value of 3 replicate wells ± SEM). c, Supernatants were tested for secretion of a broader range of cytokines/chemokines. Cytokine/chemokine secretion of bacteria/microbiota-reactive TILs was measured after stimulation with vaccine and bacteria/microbiota-derived peptide pools (left panel). Cytokine/chemokine secretion of bacteria/microbiota-reactive peripheral CD4+ T cells (selected peripheral CD4+) and CD45RA− peripheral T cells (unselected peripheral CD4+) in response to a tumor antigen (LRP6) was measured (right panel). Specific amount of each cytokine/chemokine (pg/ml) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). Source dataCC BY · PubMed Central
Extended Data Fig. 7 Stimulatory effect of IPdBPs on peripheral memory T cells (see also Fig. 5). a, CD45RA− PBMCs from patient 1635WI were CFSE-labeled and stimulated with a pool of IPdBPs (see Supplementary Table 13). 60 replicate wells were pooled after 10 days and stained with CD3, CD4 and CD8. Proliferating (CFSEdim) CD4+ T cells were sorted and short-term expanded. b, Expanded and IPdBP-pre-stimulated peripheral CD4+ T cells were tested with 10 µM of autologous tumor peptides derived from over-expressed or mutated proteins (Supplementary Table 13). Proliferation was measured using 3H-thymidine incorporation (data represents mean value of 4 replicate wells ±SEM). GM-CSF ELISA was used as another read-out for T cell activation. GM-CSF was measured by ELISA and is depicted in (pg/ml) after subtraction of the negative control. The peptide pool that was used to pre-stimulate peripheral blood CD4+ T cells (IPdBP pool) served as a positive control (see material and methods section). c, Cytokine/chemokine response of peripheral memory CD4+ T cells that had been pre-stimulated with the pool of IPdBPs and short-term expanded in vitro, to the autologous tumor peptide (LRP6) as well as the IPdBP pool was measured. Specific amount of each cytokine/chemokine (pg/ml) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). d–e, Percentage and frequency overlap of unique productive TCRVβ sequences of CFSEdim peripheral memory CD4+ T cells responding to IPdBP pool with TCRVβ sequences of bulk TILs. Overlap of unique productive TCRVβ sequences of IPdBP-reactive peripheral CD4+ (CFSEdim) (d) and bacteria/microbiota-reactive peripheral CD4+ (CFSEdim) (e). Source data
Extended Data Fig. 7 Stimulatory effect of IPdBPs on peripheral memory T cells (see also Fig. 5). a, CD45RA− PBMCs from patient 1635WI were CFSE-labeled and stimulated with a pool of IPdBPs (see Supplementary Table 13). 60 replicate wells were pooled after 10 days and stained with CD3, CD4 and CD8. Proliferating (CFSEdim) CD4+ T cells were sorted and short-term expanded. b, Expanded and IPdBP-pre-stimulated peripheral CD4+ T cells were tested with 10 µM of autologous tumor peptides derived from over-expressed or mutated proteins (Supplementary Table 13). Proliferation was measured using 3H-thymidine incorporation (data represents mean value of 4 replicate wells ±SEM). GM-CSF ELISA was used as another read-out for T cell activation. GM-CSF was measured by ELISA and is depicted in (pg/ml) after subtraction of the negative control. The peptide pool that was used to pre-stimulate peripheral blood CD4+ T cells (IPdBP pool) served as a positive control (see material and methods section). c, Cytokine/chemokine response of peripheral memory CD4+ T cells that had been pre-stimulated with the pool of IPdBPs and short-term expanded in vitro, to the autologous tumor peptide (LRP6) as well as the IPdBP pool was measured. Specific amount of each cytokine/chemokine (pg/ml) was calculated by subtracting the cytokines in the negative control cultures (no peptide control). d–e, Percentage and frequency overlap of unique productive TCRVβ sequences of CFSEdim peripheral memory CD4+ T cells responding to IPdBP pool with TCRVβ sequences of bulk TILs. Overlap of unique productive TCRVβ sequences of IPdBP-reactive peripheral CD4+ (CFSEdim) (d) and bacteria/microbiota-reactive peripheral CD4+ (CFSEdim) (e). Source dataCC BY · PubMed Central
Extended Data Fig. 8 IFN-γ fluorospot testing of expanded TILs from patient GBM-40. a, 64 bacteria/microbiota-derived peptides that were stimulatory for TILs from patient 1635WI were tested with TILs from GBM-40 patient. TILs were seeded with irradiated autologous BLCLs primed with bacteria/microbiota-derived peptides and in triplicates. ΔSFU (spot forming units) for each peptide was calculated after subtraction of the mean number of spots in negative control (data represents mean value of 3 replicate wells ± SEM). b, Positive and negative responses of TILs to several bacteria/microbiota-derived peptides are shown on the lower panel. CD2-CD3-CD28 beads are used as positive control. Source data
Extended Data Fig. 8 IFN-γ fluorospot testing of expanded TILs from patient GBM-40. a, 64 bacteria/microbiota-derived peptides that were stimulatory for TILs from patient 1635WI were tested with TILs from GBM-40 patient. TILs were seeded with irradiated autologous BLCLs primed with bacteria/microbiota-derived peptides and in triplicates. ΔSFU (spot forming units) for each peptide was calculated after subtraction of the mean number of spots in negative control (data represents mean value of 3 replicate wells ± SEM). b, Positive and negative responses of TILs to several bacteria/microbiota-derived peptides are shown on the lower panel. CD2-CD3-CD28 beads are used as positive control. Source dataCC BY · PubMed Central
Extended Data Fig. 9 Summary of T cell specificities, TCR sharing and comparison of bacterial species identified via immunopeptidome analyses and ps-SCL screening of TCC88. a, Overview of proliferative responses of all tested TCCs and bulk T cell populations to different antigens. Stimulatory antigens are depicted in red, non-stimulatory antigens in blue (N/A means that T cells were not tested with the respective antigens, * Peripheral memory CD4+ T cells selected by IPdBPs, ** Peripheral memory CD4+ T cells selected by bacteria/microbiota-derived peptides). b. Overview of shared TCRs between T cells from different compartments. Green displays the presence of shared TCRs (N/A means that samples were not examined for presence of shared TCRs, * Peripheral memory CD4+ T cells selected by IPdBPs, ** Peripheral memory CD4+ T cells selected by bacteria/microbiota-derived peptides). c. Comparison of peptides from bacterial species identified via HLA-II immunopeptidome analysis versus those derived from unbiased ps-SCL screening of TCC88. Bars of blue and red for Bacillus sp. and Bacteroides ovatus mean that these species were found via both approaches.
Extended Data Fig. 9 Summary of T cell specificities, TCR sharing and comparison of bacterial species identified via immunopeptidome analyses and ps-SCL screening of TCC88. a, Overview of proliferative responses of all tested TCCs and bulk T cell populations to different antigens. Stimulatory antigens are depicted in red, non-stimulatory antigens in blue (N/A means that T cells were not tested with the respective antigens, * Peripheral memory CD4+ T cells selected by IPdBPs, ** Peripheral memory CD4+ T cells selected by bacteria/microbiota-derived peptides). b. Overview of shared TCRs between T cells from different compartments. Green displays the presence of shared TCRs (N/A means that samples were not examined for presence of shared TCRs, * Peripheral memory CD4+ T cells selected by IPdBPs, ** Peripheral memory CD4+ T cells selected by bacteria/microbiota-derived peptides). c. Comparison of peptides from bacterial species identified via HLA-II immunopeptidome analysis versus those derived from unbiased ps-SCL screening of TCC88. Bars of blue and red for Bacillus sp. and Bacteroides ovatus mean that these species were found via both approaches.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 61 references

Cited by

Sorted by citations

Related

See all

Metadata

Authors
Naghavian, R., Faigle, W., Oldrati, P., Wang, J., Toussaint, N. C., Qiu, Y., Medici, G., Wacker, M., Freudenmann, L. K., Bonté, P.-E., Weller, M., Regli, L., Amigorena, S., Rammensee, H.-G., Walz, J. S., Brugger, S. D., Mohme, M., Zhao, Y., Sospedra, M., Neidert, M. C., & Martin, R.
Affiliations
  • University of Zurich (CH)
  • Inserm (FR)
  • University of Science and Technology of China (CN)
  • SIB Swiss Institute of Bioinformatics (CH)
  • German Cancer Research Center (DE)
  • Universität Hamburg (DE)
  • National Institutes of Health (US)
ORCID
Venue
Nature
Publisher
Springer Science and Business Media LLC
Published
2023-05-17
Volume
617
Issue
7962
Pages
807-817
ISSN
  • 0028-0836 (ISSN-L)
  • 1476-4687
DOI
10.1038/s41586-023-06081-w
PMC
PMC10208956
OpenAlex
W4376959488
Cited by
163
References
61
Citations by year
2026: 38 · 2025: 60 · 2024: 49 · 2023: 15
FWCI
17.76
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Repository for Publications and Research Data (ETH Zurich) (submittedVersion, cc-by)
  • RePEc: Research Papers in Economics (submittedVersion)
  • Zurich Open Repository and Archive (University of Zurich) (submittedVersion, cc-by)
  • ETH Zürich Research Collection
Funders
  • National Science Foundation
  • Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (153213)
  • Universität Zürich
  • FP7 Ideas: European Research Council (340733)
Topics
  • Life Sciences › Immunology and Microbiology › Immunology › Immunotherapy and Immune Responses
  • Health Sciences › Medicine › Oncology › Cancer Immunotherapy and Biomarkers
  • Life Sciences › Immunology and Microbiology › Immunology › T-cell and B-cell Immunology
Keywords
glioblastoma, tumor-infiltrating lymphocytes, bacterial peptides, HLA class II, tumor antigen presentation, gut microbiota, commensal bacteria, pathogenic bacteria, CD4+ T cells, personalized cancer immunotherapy, immune checkpoint inhibitors
MeSH terms
Gastrointestinal Microbiome, Histocompatibility Antigens Class II, Antigens, Neoplasm, Bacteria, Bacterial Proteins, Glioblastoma, HLA Antigens, Humans, Peptide Fragments, Symbiosis, CD4-Positive T-Lymphocytes, Lymphocytes, Tumor-Infiltrating, Cancer Vaccines, Cell Line, Tumor

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review