Re: Lactobacillus johnsonii, 면역 항암제 효과 2024 Cell!!

[문형철]

ArticleVolume 187, Issue 7p1651-1665.e21March 28, 2024Open Archive

Download Full Issue

Microbial metabolite enhances immunotherapy efficacy by modulating T cell stemness in pan-cancer

미생물 대사산물이 범암종에서 T 세포 줄기세포 특성을 조절함으로써 면역치료 효능을 증진시킨다

Dingjiacheng Jia1,2,11 ∙ Qiwen Wang2,3,11 ∙ Yadong Qi1,2,11 ∙ … ∙ Yuhao Wang10 ∙ Shujie Chen2,3,4 chenshujie77@zju.edu.cn ∙ Liangjing Wang1,2,4,12 wangljzju@zju.edu.cn … Show more

Affiliations & NotesArticle Info

• Download PDF
• Cite
•  
• Set Alert
• Get Rights
• Reprints

Download Full Issue

Previous articleNext article

Show Outline

Highlights

•

L. johnsonii enhances ICB efficacy via CD8+ T cells

•

L. johnsonii cooperates with C. sporogenes to produce IPA

•

IPA activates progenitor exhausted CD8+ T cells by H3K27 acetylation modification

•

IPA improves ICB responsiveness at the pan-cancer level

Summary

The immune checkpoint blockade (ICB) response in human cancers is closely linked to the gut microbiota. Here, we report that the abundance of commensal Lactobacillus johnsonii is positively correlated with the responsiveness of ICB. Supplementation with Lactobacillus johnsonii or tryptophan-derived metabolite indole-3-propionic acid (IPA) enhances the efficacy of CD8+ T cell-mediated αPD-1 immunotherapy. Mechanistically, Lactobacillus johnsonii collaborates with Clostridium sporogenes to produce IPA. IPA modulates the stemness program of CD8+ T cells and facilitates the generation of progenitor exhausted CD8+ T cells (Tpex) by increasing H3K27 acetylation at the super-enhancer region of Tcf7. IPA improves ICB responsiveness at the pan-cancer level, including melanoma, breast cancer, and colorectal cancer. Collectively, our findings identify a microbial metabolite-immune regulatory pathway and suggest a potential microbial-based adjuvant approach to improve the responsiveness of immunotherapy.

요약

인간 암에서의 면역 체크포인트 차단(ICB) 반응은

장내 미생물군과 밀접하게 연관되어 있다.

immune checkpoint blockade (ICB)

본 연구에서는 공생균인 Lactobacillus johnsonii의 풍부도가

ICB 반응성과 양의 상관관계가 있음을 보고한다.

Lactobacillus johnsonii 또는 트립토판 유래 대사산물 인돌-3-프로피온산(IPA)의 보충은

CD8+ T 세포 매개 αPD-1 면역치료의 효능을 향상시킨다.

기전적으로,

Lactobacillus johnsonii는 Clostridium sporogenes와 협력하여

IPA를 생성한다.

IPA는 Tcf7의 슈퍼-강화자 영역에서 H3K27 아세틸화를 증가시켜

CD8+ T 세포의 줄기세포성 프로그램을 조절하고,

선조 소진 CD8+ T 세포(Tpex)의 생성을 촉진한다.

IPA는

흑색종, 유방암, 대장암을 포함한 범암종 수준에서 ICB 반응성을 향상시킵니다.

immune checkpoint blockade (ICB)

종합적으로,

본 연구 결과는 미생물 대사산물-면역 조절 경로를 규명하며,

면역요법 반응성 향상을 위한 잠재적 미생물 기반 보조접근법을 제시합니다.

Graphical abstract

Keywords

1. cancer immunotherapy
2. immune checkpoint blockade
3. TCF-1
4. progenitor-exhausted CD8+ T cells
5. indole-3-propionic acid
6. Lactobacillus johnsonii
7. gut microbiota
8. tryptophan metabolism

Introduction

Gut microbiota is one of the major environmental factors that affects the efficacy of immune checkpoint blockade (ICB).1,2 The composition of the gut microbiota differs in patients,3 and fecal microbiota transplantation (FMT) has been shown to improve ICB responsiveness.4 Specific strains, such as Lactobacillus rhamnosus GG,5 Bifidobacterium breve,6 and Fusobacterium nucleatum,7 have been found to improve the efficacy of ICB therapy by modulating the innate and adaptive immune response processes. However, specific bacterial components and detailed underlying mechanisms remain elusive.

Infiltrating T cells in the tumor microenvironment play a key role in ICB therapy.8 Multiple CD8+ T cell subsets have been revealed to influence anti-tumor response.9 Recent studies have shown that progenitor exhausted CD8+ T cells (Tpex), which are characterized by high expression‎ of the transcription factor T cell factor 1 (TCF-1, encoded by TCF7), are the main subset responding to anti-programmed cell death protein 1 antibody (αPD-1) immunotherapy.10,11,12,13,14 Tpex cells exhibiting stemness characteristics demonstrate the ability to self-renew, proliferate, and differentiate into effector CD8+ T cells (Teff) to limit tumorigenesis.15 Moreover, the expression‎ level of TCF7 could be served as a positive prognostic indicator for patients treated with ICB.16 Because a clear correlation between the deficiency of Tcf7 and a disrupted balance of the gut microbiota has been observed in mice,17 further investigation is needed to determine whether the gut microbiota and its metabolites could regulate the stemness program of CD8+ T cells.

서론

장내 미생물군은

면역 체크포인트 차단(ICB)의 효능에 영향을 미치는 주요 환경적 요인 중 하나이다.1,2

환자마다 장내 미생물 군집의 구성이 다르며,3

분변 미생물 이식(FMT)이

ICB 반응성을 개선하는 것으로 나타났습니다.4

Lactobacillus rhamnosus GG,5  

Bifidobacterium breve,6 및

Fusobacterium nucleatum7과 같은 특정 균주는

선천적 및 적응성 면역 반응 과정을 조절하여

ICB 치료의 효능을 향상시키는 것으로 밝혀졌습니다.

그러나

구체적인 세균 성분과 상세한 기전은

여전히 불분명하다.

종양 미세환경 내 침윤 T 세포는

ICB 치료에서 핵심적인 역할을 한다.8

여러 CD8+ T 세포 하위집단이

항종양 반응에 영향을 미치는 것으로 밝혀졌다.9

최근 연구에 따르면

전사인자 T 세포 인자 1(TCF-1, TCF7 유전자에 의해 암호화됨)의 높은 발현을 특징으로 하는

전구 소진 CD8+ T 세포(Tpex)가 (TCF-1, TCF7 유전자에 의해 암호화됨)의 높은 발현을 특징으로 하는

선조 소진 CD8+ T 세포(Tpex)가 항-프로그래밍된 세포사멸 단백질 1 항체(αPD-1)

면역요법에 반응하는 주요 하위 집합임이 밝혀졌다.10,11,12,13,14

줄기세포 특성을 보이는 Tpex 세포는

자가 재생, 증식 및 효과기 CD8+ T 세포(Teff)로의 분화 능력을 보여 종양 발생을 제한한다.15

또한, TCF7 발현 수준은 ICB 치료 환자의 양호한 예후 지표로 활용될 수 있다.16 Tcf7 결핍과 장내 미생물군집 균형 교란 사이의 명확한 상관관계가 생쥐에서 관찰되었기에,17 장내 미생물군집 및 그 대사산물이 CD8+ T 세포의 줄기세포성 프로그램을 조절할 수 있는지 확인하기 위한 추가 연구가 필요하다.

In this study, we discovered that the abundance of commensal Lactobacillus johnsonii (L. johnsonii) is positively associated with the proportion of Tpex cells, as well as improved the responsiveness to ICB. The tryptophan metabolite indole-3-propionic acid (IPA), derived from L. johnsonii, could enhance H3K27 acetylation at the super-enhancer (SE) of Tcf7 gene and thus promote the differentiation of CD8+ T cells into Tpex cells. Importantly, we found that the synthesis of IPA requires cooperation between L. johnsonii and another symbiotic bacterium, Clostridium sporogenes (C. sporogenes, C. s). Overall, our findings point to a microbial-based adjuvant approach to improve the effectiveness of immunotherapy.

ResultsL. johnsonii sensitizes αPD-1 immunotherapy by activating tumor-infiltrating CD8+ T cells

When mice are given the same dose of αPD-1, their tumor growth could exhibit either a good or poor response, similar to what is observed in clinical practice.18,19,20 Using the Mc38 cell line (high microsatellite instability, MSI-H),21,22 we modeled αPD-1 therapy for colorectal cancer (CRC) and divided the mice into two groups, namely, the “poor-responder” group and the “responder” group, based on tumor volume (Figures 1A and 1B). We then separately transplanted feces from each group into recipient mice that were pretreated with antibiotics (Abx) (Figure 1C). Mice transplanted with the poor-responder feces (PR-recipient) showed lower responsiveness to αPD-1 therapy than mice transplanted with the responder feces (R-recipient) (Figure 1D), suggesting that the gut microbiota has a direct impact on the reactivity of ICB therapy.

본 연구에서 우리는

공생균 Lactobacillus johnsonii (L. johnsonii)의 풍부도가

Tpex 세포의 비율과 양의 상관관계를 보일 뿐만 아니라

ICB에 대한 반응성을 향상시킨다는 사실을 발견했습니다. 

L. johnsonii에서 유래된 트립토판 대사산물 인돌-3-프로피온산(IPA)은 

Tcf7 유전자의 슈퍼 인핸서(SE)에서 H3K27 아세틸화를 증강시켜

CD8+ T 세포의 Tpex 세포 분화를 촉진할 수 있습니다.

중요한 점은

IPA 합성에 

L. johnsonii와 공생 세균인 Clostridium sporogenes(C. sporogenes, C. s) 간의

협력이 필요하다는 사실을 발견했다는 것이다.

종합하면,

우리의 연구 결과는

면역치료 효과를 향상시키기 위한 미생물 기반 보조접근법을 제시한다.

결과 

L. johnsonii는 종양 침윤성 CD8+ T 세포를 활성화하여

αPD-1 면역요법의 감작 효과를 높입니다.

https://pmc.ncbi.nlm.nih.gov/articles/PMC10609197/

락토바실러스 존소니는 수십 년간 다양한 질환 치료를 위한 프로바이오틱스로 사용되어 왔으며, 여러 질환 치료에 특정한 이점이 있는 것으로 밝혀졌습니다. 우리는 PubMed 및 Web of Science 데이터베이스를 기반으로 다양한 질환에서 L. 존소니의 잠재적 치료 효과와 기전을 검토했습니다. NCBI(2023년 2월 14일 기준)에서 149종의 L. johnsonii 정보를 확보하고, 포함된 플라스미드 정보를 포함한 포괄적 메타데이터를 검토했습니다. 본 검토는 다양한 L. johnsonii 균주의 기본적 특성과 여러 질환에 대한 잠재적 치료 특성을 제시합니다. 작용 기전은 아직 완전히 규명되지 않았으나, 향후 연구에 일부 근거를 제공할 수 있기를 기대합니다. 또한 L. johnsonii의 다양한 균주의 항생제 내성 여부는 명확하지 않으며, 보다 완전하고 심층적인 연구가 필요합니다. 요약하면, L. johnsonii는 질병 치료 또는 예방에 상당한 연구 잠재력을 지니고 있으나, 치료적 적용을 정당화하기 위해서는 더 많은 증거가 필요합니다. 항생제 내성 유전자 확산을 줄이기 위해 이 균주가 보유한 항생제 내성 유전자에 대한 추가 연구도 필요합니다.

마우스에게 동일한 용량의 αPD-1을 투여했을 때,

임상에서 관찰되는 것과 유사하게 종양 성장에 대한 반응이 양호하거나 불량한 양상을 보였습니다.18,19,20 

Mc38 세포주(고도 미세위성 불안정성, MSI-H)를 사용하여21,22 대장암(CRC)에 대한 αPD-1 치료를 모델링하고 종양 부피에 따라 마우스를 “저반응군”과 “반응군”으로 나누었습니다(그림 1A 및 1B). 그런 다음 각 군의 분변을 항생제(Abx)로 사전 처리된 수혜 마우스에 각각 이식했습니다 (그림 1C). 반응이 낮은 분변을 이식받은 마우스(PR-수혜자)는 반응이 높은 분변을 이식받은 마우스(R-수혜자)보다 αPD-1 치료에 대한 반응성이 낮았으며(그림 1D), 이는 장내 미생물군이 면역관문 차단제(ICB) 치료의 반응성에 직접적인 영향을 미친다는 것을 시사합니다.

Figure 1 L. johnsonii sensitizes αPD-1 immunotherapy by activating tumor-infiltrating CD8+ T cells

Show full captionFigure viewer

We then performed metagenomics and 16S ribosomal RNA (rRNA) sequencing on feces from both the poor-responder and the responder mice to compare the microbial communities. There were differences in β-diversity observed through principal co-ordinates analysis (PCoA) (Figures 1E and S1A). Colony composition analysis at the genus level revealed a decrease in the abundance of Lactobacillus in the poor-responder group (Figures 1F, S1B, and S1C). At the species level, L. johnsonii was found to be the most preval‎ent (Figure 1G). We confirmed that the lower abundance of L. johnsonii in the poor-responder group negatively correlated with tumorigenesis (Figures S1D and S1E). Apcmin/+ or azoxymethane (AOM)/dextran sodium sulfate (DSS)-induced CRC mice also showed decreased L. johnsonii colonization (Figures S1F and S1G). Likewise, compared with adjacent normal tissues, the abundance of L. johnsonii was significantly lower in human CRC tissues (Figure S1H) and negatively correlated with the advanced stage of human CRC (Figure S1I).

우리는 L. johnsonii 균주를 분리하여 그 기능적 역할을 탐구했다.

L. johnsonii의 경구 투여는

Mc38 종양에서 αPD-1 치료에 대한 반응을 향상시켰다(그림 1H).

항종양 반응의 주요 면역 효과 세포인 CD8+ T 세포는

L. johnsonii 경구 투여 후 빈도가 증가하는 것으로 나타났다(그림 1I 및 1J).

이러한 증가는 인터페론(IFN)-γ 및 종양 괴사 인자 알파(TNF-α)와 같은 Teff 세포의 사이토카인 분비와 동반되었다(그림 S1J 및 S1K).

Figure S1 L. johnsonii reduces in colorectal cancer and is related to ICB responsiveness through activating IFN-γ+ and TNF-α+ CD8+ T cells, related to Figure 1

Show full captionFigure viewer

We isolated a strain of L. johnsonii and explored its functional role. Oral administration of L. johnsonii enhanced the response to αPD-1 therapy in Mc38 tumor (Figure 1H). CD8+ T cells, which are the main immune effector cells for anti-tumor response,23 were found to increase in frequency after oral administration of L. johnsonii (Figures 1I and 1J). This increase was accompanied by the release of cytokines from Teff such as interferon (IFN)-γ and tumor necrosis factor alpha (TNF-α) (Figures S1J and S1K).

L. johnsonii promotes ICB responsiveness via tryptophan metabolism

To explore whether such a function was attributed to L. johnsonii itself or its secreted products, we administered mice with L. johnsonii that had been high-temperature inactivated (heated group), ultrasonic disrupted (ultrasound group), the original medium (de Man, Rogosa, and Sharpe [MRS] group), and the conditional medium (CM) (Lj. CM group) in addition to live L. johnsonii. Interestingly, only Lj. CM group (Lj. CM + αPD-1) had a comparable immunotherapeutic effect to live L. johnsonii group (L. j + αPD-1), indicating that the antitumor effect of L. johnsonii was primarily through the molecules secreted by the bacteria (Figures 2A and 2B).

L. johnsonii는 트립토판 대사를 통해 ICB 반응성을 촉진한다

이러한 기능이 L. johnsonii 자체에 기인하는지 아니면 분비물에 기인하는지 탐구하기 위해, 우리는 생체 내 L. johnsonii를 고온 불활성화 처리한 균주(가열군), 초음파 분쇄 처리된 균(초음파 처리군), 원배지(de Man, Rogosa, Sharpe [MRS] 배지군), 조건부 배지(CM)를 처리한 균(Lj. CM군)을 생균 L. johnsonii와 함께 투여하였다. 흥미롭게도, Lj. CM 그룹(Lj. CM + αPD-1)만이 생체 L. johnsonii 그룹(L. j + αPD-1)과 유사한 면역치료 효과를 나타냈으며, 이는 L. johnsonii의 항종양 효과가 주로 박테리아가 분비하는 분자를 통해 이루어짐을 시사한다(그림 2A 및 2B).

Figure 2 L. johnsonii promotes ICB responsiveness via tryptophan metabolism

Show full captionFigure viewer

Plasma liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis revealed distinct differences in the metabolite composition between the αPD-1 alone group and the live L. johnsonii group (L. j + αPD-1) (Figure 2C). The live L. johnsonii group showed a significant enrichment of the “tryptophan metabolism” pathway (Figure 2D). We then confirmed the necessity of tryptophan in L. johnsonii-promoted ICB responsiveness by subjecting mice to a tryptophan deficiency diet (Trpneg diet).24 As expected, L. johnsonii did not enhance the effectiveness of αPD-1 therapy or increase the frequency CD8+ T cells when tryptophan substrates were absent (Figures 2E–2G).

L. johnsonii-derived IPA promotes ICB responsiveness via CD8+ T cells

We examined the plasma tryptophan-related metabolites that were upregulated by L. johnsonii and found that IPA exhibited a most pronounced increase (Figure 3A). IPA, an indole analog that is specifically produced by the gut microbiota,25,26 has been shown to promote axonal regeneration27 and protect against radiation toxicity.28 Consistent with L. johnsonii administration, mice transplanted with feces from the responder group showed higher plasma IPA levels (Figures 3B, S2A, and S2B). However, supplementation with L. johnsonii did not increase plasma IPA levels in mice that were fed with Trpneg diet (Figures 3C and S2C).

L. johnsonii 유래 IPA는 CD8+ T 세포를 통해 ICB 반응성을 촉진한다

우리는 L. johnsonii에 의해 상향 조절된 혈장 트립토판 관련 대사물을 조사한 결과, IPA가 가장 현저한 증가를 보였음을 발견했다(그림 3A). 장내 미생물에 의해 특이적으로 생성되는 인돌 유사체인 IPA는25,26 축삭 재생을 촉진하고27 방사선 독성으로부터 보호하는 것으로 알려져 있다.28 반응군 분변을 이식받은 마우스는 L. johnsonii 투여와 일관되게 혈장 IPA 수치가 더 높았다 (그림 3B, S2A, S2B). 그러나 트립토판 결핍 식이를 섭취한 생쥐에서는 L. johnsonii 보충이 혈장 IPA 농도를 증가시키지 않았다(그림 3C, S2C).

Figure 3 L. johnsonii-derived IPA promotes ICB responsiveness via CD8+ T cells

Show full captionFigure viewer

We then explored the potential impact of IPA on tumor immunotherapy. Administering 60 mg/kg/day of IPA through gavage could enhance the effectiveness of αPD-1 and increase plasma or intratumoral IPA levels (Figures 3D and S2D–S2G). Similar enhancement could also be achieved by direct intratumoral injection (i.m.) of 5 μΜ IPA (Figures S2H and S2I). IPA supplementation increased the frequency of infiltrating CD8+ T cells and the production of their effector cytokines in tumors (Figures 3E–3G and S2Q). However, no significant alterations were observed for natural killer (NK) cells, B cells, dendritic cells, and macrophages (Figures S2J–S2P).

Figure S2 L. johnsonii and its metabolite, IPA, sensitize αPD-1 immunotherapy, and affect associated immunocytes, related to Figure 3

Show full captionFigure viewer

To determine whether the immunotherapeutic effect is dependent on T cell response, L. johnsonii or IPA was administered to Rag1−/− mice. Notably, both treatments failed to reduce tumor growth (Figures 3H and 3I). Moreover, treating mice with CD8 neutralizing antibody abrogated the antitumor effect of both L. johnsonii and IPA (Figure 3J). We conjectured that CD8+ T cells treated with IPA could be maintained in an activated state and confirmed this by adoptively transferring pretreated CD8+ T cells or B cells to Rag1−/− mice (Figure 3K). Mice receiving IPA-pretreated CD8+ T cells (IPA T) showed better αPD-1 responsiveness compared with mice receiving IPA-untreated CD8+ T cells (wild-type [WT] T). However, mice that received both B cells and CD8+ T cells pretreated with IPA (IPA B + T) showed no better tumor inhibitory effect compared with the IPA T group (Figure 3L). Together, these findings indicate that L. johnsonii-derived metabolite IPA is dependent on CD8+ T cells to promote the responsiveness to ICB therapy.

IPA sustains CD8+ T cells response by promoting Tpexcells

CD8+ T cells consist of various subsets of cells that have distinct functions and work collaboratively to regulate the immune microenvironment of tumors.8,29 To further reveal the role of IPA on CD8+ T cell subsets, we performed single-cell RNA sequencing (scRNA-seq), single-cell T cell receptor sequencing (scTCR-seq), and single-cell assay for targeting accessible-chromatin with high-throughput sequencing (scATAC-seq) on CD8+ T cells sorted from tumors in mice that have been treated with either αPD-1 alone (WT group) or combination with IPA (IPA group) (Figure 4A). To exclude the interference of macrophages, neutrophils, and fibroblasts (Figures S3A–S3C), we performed the uniform manifold approximation and projection (UMAP) analysis to obtain relatively pure CD8+ T cells (Figure S3D). We identified 13 cell clusters and 10 CD8+ T cell subsets: naive (Lef1hi, Sellhi, and Ccr7hi), progenitor exhausted CD8+ T cells (Tpex, Tcf7hi and Pdcd1hi), activated (Cd69hi and Isg15hi), Teff (Prf1hi, Gzmbhi, and Klrd1hi), natural killer T cells (NKT, Cd160hi and Xcl1hi), Trbv3hi, Cotl1hi, Bcl2hi, Mki67hi, and Mcm3hi (Figures 4B and S3E–S3H).

Figure 4 IPA sustains CD8+ T cells response by promoting Tpex cells

Show full captionFigure viewer

Figure S3 Single-cell RNA sequencing analysis shows IPA promotes the differentiation of CD8+ T cells by upregulating progenitor exhausted CD8+ T cells, related to Figure 4

Show full captionFigure viewer

IPA treatment reduced naive CD8+ T cells but increased Tpex cells and Teff cells in tumors (Figure 4C). To comprehensively map immune cell function,30,31 we performed VISION analysis to project the total score of the Gene Ontology (GO) biological process gene sets and calculate the average signature score for each cluster (Figure 4D).

“GO_ALPHA_BETA_T_CELL_DIFFERENTIATION” was enriched in 9 of 13 cell clusters after IPA treatment, particularly in Tpex cells (Figures 4E–4G). Also, “GO_ALPHA_BETA_T_CELL_ACTIVATION” was enriched in 7 of 13 cell clusters after IPA treatment, particularly in Teff cells (Figures S3I–S3K). In addition, IPA treatment could increase the proportion of TCF-1+ stem-like CD8+ T cells in ex vivo culture (Figures 4H, 4I, and S4E) and in tumor-draining lymph node (TDLN) or spleen (Figures S4A–S4D). Because Tcf7 expression‎ marks the activation of Tpex cells, we knocked out Tcf7 in mice (Tcf7−/−) and found that IPA only enhanced the efficacy of immunotherapy in WT littermates but not in Tcf7−/− mice (Figure 4J). Collectively, these findings suggest that IPA exerts its immunotherapy-enhancing effects by elevating Tpex cells in tumors.

Figure S4 Flow cytometry analysis, monocle analysis, and single-cell TCR sequencing analysis are performed on CD8+ T cells after IPA treatment, related to Figure 4

Show full captionFigure viewer

To further examine the differentiation route of CD8+ T cells after IPA treatment, we performed trajectory analysis of four classical subsets of CD8+ T cells. Monocle analysis revealed that naive CD8+ T cells developed into Tpex cells and activated CD8+ T cells and Teff cells step by step (Figure 4K). This differentiation process was enhanced by IPA treatment as a higher frequency of Tpex cells was enriched during the late stage of development (Figures S4F and S4G). Using scTCR-seq, we identified clonotypes between WT group and IPA group (Figure S4H). The subsets of CD8+ T cells shared TCR profiles and underwent clonal expansion with each other. The predominant TCR clonotypes in Teff cells primarily originated from the Tpex cells, and IPA treatment increased the proportion of conservative transmission (Figures 4L and 4M). In short, IPA increased the frequency of Tpex cells in the tumor microenvironment and promoted their differentiation to Teff cells to enhance ICB responsiveness.

IPA activates Tpex cells by modifying the H3K27 acetylation of Tcf7 gene at its SE region

Tpex cells are mainly regulated by histone modifications,32,33,34 we sought to explore whether IPA activates Tpex cells through histone remodeling of Tcf7 gene. We mapped the results of scRNA-seq to scATAC-seq and identified 10 common CD8+ T cell subsets (Figures 5A, S5A, and S5B). Consistent with scRNA-seq results, IPA treatment increased the proportions of Tpex cells and Teff cells (Figure 5B). Further differential peak analysis revealed that open chromatin regions of Tpex cells were mainly enriched in the promoter and distal intergenic regions (Figure S5C), and IPA treatment increased the chromatin opening of Tcf7 gene at its super-enhancer (SE) region (Figures 5C and 5J). In addition, we noticed that IPA treatment primarily enhanced histone acetylation but not histone methylation in Tpex cells (Figures 5D and S5D), suggesting that IPA may activate Tpex cells by modifying the histones of Tcf7 gene at its SE through acetylation.

Figure 5 IPA activates Tpex cells by modifying the H3K27 acetylation of Tcf7 gene at its SE region

Show full captionFigure viewer

Figure S5 IPA affects chromosome accessibility in progenitor exhausted CD8+ T cells, related to Figure 5

Show full captionFigure viewer

Nucleosome, the basic unit of chromatin, is an octamer composed of four core histones (H2A, H2B, H3, and H4).35 scRNA-seq analysis revealed that histone H3 acetylation was most significantly enriched in Tpex cells after IPA treatment (Figures 5E, S5E, and S5F). We performed western blot analysis with antibodies specific to common histone modifications, including H2AK5ac,36 H2BK5ac,37 H3K27ac, and H4K20me1,38 to validate this result. Consistently, only H3K27ac exhibited a dose-dependent increase in response to IPA, along with an increase in TCF-1 expression‎ (Figure 5F). Moreover, we performed chromatin immunoprecipitation (ChIP), cleavage under targets and release using nuclease (CUT&RUN), and cleavage under targets and tagmentation (CUT&Tag) on H3K27ac in sorted CD8+ T cells. Results confirmed that IPA treatment increases H3K27 acetylation at the SE of Tcf7 gene (Figures 5G–5J).39

Taken together, above results prove that IPA promotes the stemness program of CD8+ T cells by modifying the H3K27 acetylation of Tcf7 gene at its SE region.

L. johnsonii cooperates with C. sporogenes to produce IPA

Tryptophan could be metabolized by the gut microbiota into indole-3-pyruvate acid (IPYA), indole-3-lactic acid (ILA), indole-3-acrylic acid (IA), and IPA in a sequential manner,26,40 but how L. johnsonii metabolizes tryptophan to produce IPA remains elusive. Surprisingly, we were unable to detect IPA or IA in the supernatant of the CM after culturing L. johnsonii. However, ILA was detected in the CM. This suggests that L. johnsonii could only metabolize tryptophan into ILA and not further into IA or IPA (Figures 6A, 6B, and S6A).

Figure 6 L. johnsonii cooperates with C. sporogenes to produce IPA

Show full captionFigure viewer

To confirm‎ that ILA lacks the ability to enhance ICB responsiveness if it is not further converted to IPA by gut microbes, we administered either ILA or IPA to Mc38 tumor-bearing mice. As expected, when ILA was given to mice receiving Abx pretreatment, it did not improve the efficacy of αPD-1 therapy and no longer raised plasma IPA levels, which were in contrast to IPA administration group (Figures 6C, 6D, and S6B). Similarly, intraperitoneal injection of ILA, which avoids further catabolism by gut microbiota,41 did not improve αPD-1 effectiveness or increase plasma IPA (Figures 6E, 6F, and S6C). These findings lead to the hypothesis that L. johnsonii requires collaboration with other commensal microbes to produce IPA.

Because C. sporogenes is known to convert ILA to IPA,42,43 and oral gavage of L. johnsonii increased the fecal abundance of C. sporogenes (Figure S6D), we selected it as a representative bacterium and tested our hypothesis in germ-free mice (Figure 6G). In comparison with single-bacteria (L. johnsonii or C. sporogenes) administration, mixed administration of L. johnsonii and C. sporogenes further enhanced efficacy of αPD-1 (Figure 6H). LC-MS/MS analysis verified that L. johnsonii alone was unable to increase plasma IPA, despite being able to produce large amounts of ILA. In contrast, C. sporogenes or mixture of L. johnsonii and C. sporogenes significantly increased plasma IPA (Figures 6I and S6E). These results confirmed that ILA produced by L. johnsonii needs to be further metabolized to IPA by C. sporogenes in order to enhance the efficacy of immunotherapy.

Figure S6 The cooperation of L. johnsonii with D-2-hydroxyacid dehydrogenase and C. sporogenes promotes IPA production, related to Figure 6

Show full captionFigure viewer

We then proceeded to explore how L. johnsonii produces ILA. We performed whole-genome sequencing for our isolated L. johnsonii and used the protein basic local alignment search tool (BLAST) to search for sequences that were similar to fldH, a gene from C. sporogenes that encodes a key enzyme for converting tryptophan to ILA25 (Figure 6J). We found that ldhA, a gene present in the genome of L. johnsonii that encodes a 337 amino acid D-2-hydroxyacid dehydrogenase, may be essential for L. johnsonii to produce ILA (Figure S6F). The secondary structure of the ldhA-encoded enzyme was predicted by 3D modeling (Figure S6G). To test the function of this enzyme, we transformed a ldhA expression‎al vector into E. coli (E. c-ldhA) and induced protein expression‎ (Figure S6H). Administration with E. c-ldhA to Mc38 tumor-bearing mice increased plasma IPA levels and improved αPD-1 reactivity (Figures 6K, 6L, and S6I), confirm‎ing that L. johnsonii relies on ldhA to encode enzyme for ILA production.

IPA promotes ICB responsiveness in pan-cancer and CRC-derived organoids

We then tested whether microbiota-derived IPA improves the responsiveness of ICB therapy at the pan-cancer level. IPA effectively promoted the efficacy of αPD-1 immunotherapy (Figures 7A and 7C) and increased the frequency of infiltrating CD8+ T cells and the expression‎ of TCF-1 (Figures 7B, 7D, S7A, and S7B) in both breast cancer and melanoma transplantable tumor model. We also verified the sensitizing effect of IPA for immunotherapy in mammary fat pad orthotopic implantation model, murine mammary tumor virus-polymavirus middle T antigen (MMTV-PyMT) spontaneous breast cancer model,44,45 and cecum orthotopic implantation model (Figures 7E–7I, S7C, and S7D). Immunofluorescence staining verified that IPA treatment increased the infiltration of Tpex cells in the tumor microenvironment of breast cancer, melanoma, and CRC (Figures 7J and 7K).

Figure 7 IPA promotes ICB responsiveness in pan-cancer and CRC-derived organoids

Show full captionFigure viewer

We further investigated whether IPA could act on human CD8+ T cells to improve the responsiveness of ICB therapy. In fresh CRC tissues, although the majority of CD8+ T cells in the tumor microenvironment are TIM-3+ PD-1+ terminally exhausted cells, there is also a specific proportion of Tpex cells present (Figures S7G and S7H). Considering the limitation of preserving immune cells in the tumor microenvironment using conventional patient-derived organoids (PDOs),46 we established an air-liquid interface (ALI)-PDOs system. This system contains a more complete immune microenvironment and additional matrix components, allowing for a more accurate representation of immunotherapy47,48 (Figure 7L). Both IPA-treated and mock-treated PDOs successfully preserved the CRC structures (Figure 7M). IPA-treated PDOs exhibited increased the infiltration of CD8+ T cells (Figures S7E and S7F) and elevated the expression‎ of Teff cells effector proteins (Figure 7N). Collectively, these results demonstrate that modulating the stemness program of CD8+ T cells through microbiota-derived IPA may be a promising approach to reinforce the effectiveness of tumor ICB therapy in clinical settings.

Figure S7 IPA modulates ICB responsiveness in pan-cancer and CRC-derived organoids, related to Figure 7

Show full captionFigure viewer

Discussion

In this study, we found that L. johnsonii, a commensal bacterium in the intestine, can enhance the efficacy of ICB therapy mainly by increasing the synthesis of tryptophan-derived metabolite IPA. Many metabolites derived from bacteria cannot be synthesized by a single bacterium alone but require participation of other microbes. Recent studies have reported that C. scindens collaborates with its closely related strains to transform primary bile acids into secondary bile acids.49 Bacteroides collaborates in a relaying manner with Lactobacillus plantarum to metabolize host tyrosine.50 We present an additional example in which L. johnsonii requires collaboration with C. sporogenes to convert dietary tryptophan into IPA.

A variety of tryptophan-indole metabolites derived from bacteria, including ILA, IA, IPA, indole-3-acetic acid (IAA), and indole-3-acetamide (IAM), have been shown to have a wide range of biological effects.26,51 IPA, specifically, could improve atherosclerosis,52 mitigate radiotoxicity,28 and prolong the posttraumatic survival time.53 The impact of IPA on modulating cancer immunotherapy has not been reported. In this study, we discovered that IPA mainly promotes the infiltration of CD8+ T cells into tumors and activates Tpex cells during immunotherapy. Dodd and colleagues found that mutant C. sporogenes unable to produce IPA colonized germ-free mice could affect intestinal permeability and increase the number of memory T cells in peripheral blood.25 The interactions between the bacteria and host immunity may regulate the intestinal barrier, and this mechanism needs to be further explored.

A number of studies have reported that the responses to anti-PD-1 are mediated by Tpex cells, which rely on the reactivation of the stemness program of CD8+ T cells for their differentiation.10,11 Several strategies, including activation of IFN gene signaling pathway,54 elevation of extracellular potassium levels,55 and overexpression‎ of c-Myb56 have been shown to reprogram the stemness of therapeutic CD8+ T cells. Our study revealed that microbiota-derived IPA is able to modulate the stemness program of CD8+ T cells and promote the differentiation of Tpex cells by increasing H3K27 acetylation level at the SE of Tcf7 gene.

In conclusion, our study establishes that L. johnsonii can boost the therapeutic efficacy of ICB by increasing the synthesis of IPA and enhancing Tpex cell activity. L. johnsonii and bacterial-derived IPA may potentially serve as an effective drug adjuvant for patients receiving personalized cancer immunotherapy.

토론

본 연구에서 우리는

장내 공생균인 L. johnsonii가

주로 트립토판 유래 대사산물 IPA의 합성을 증가시켜

ICB 치료의 효능을 향상시킬 수 있음을 발견했다.

박테리아에서 유래된 많은 대사산물은

단일 박테리아만으로는 합성될 수 없으며

다른 미생물의 참여가 필요하다.

최근 연구에 따르면 C. scindens가

근연 균주와 협력하여 1차 담즙산을 2차 담즙산으로 전환한다는 사실이 보고되었다.

Bacteroides는

Lactobacillus plantarum과 계주 방식으로 협력하여

숙주의 티로신을 대사한다.50

본 연구는

L. johnsonii가 식이 트립토판을 IPA로 전환하기 위해

C. sporogenes와의 협력이 필요하다는 추가 사례를 제시한다.

세균에서 유래한 ILA, IA, IPA, 인돌-3-아세트산(IAA), 인돌-3-아세트아미드(IAM) 등

다양한 트립토판-인돌 대사산물은

광범위한 생물학적 효과를 나타내는 것으로 밝혀졌다.26,51

특히 IPA는 죽상경화증을 개선하고,52

방사선 독성을 완화하며,28

외상 후 생존 시간을 연장할 수 있다.53

IPA가 암 면역 치료 조절에 미치는 영향은 보고된 바 없다.

본 연구에서 우리는

IPA가 주로 면역치료 중 CD8+ T 세포의 종양 내 침윤을 촉진하고

Tpex 세포를 활성화한다는 사실을 발견했다.

Dodd 등은 IPA를 생산하지 못하는 돌연변이 C. sporogenes가 무균 마우스에 정착할 경우 장 투과성에 영향을 미치고 말초혈 내 기억 T 세포 수를 증가시킬 수 있음을 발견했다.25 박테리아와 숙주 면역 간의 상호작용은 장 장벽을 조절할 수 있으며, 이 메커니즘은 추가 연구가 필요하다.

다수의 연구에서 항-PD-1 반응이 Tpex 세포에 의해 매개되며, 이는 CD8+ T 세포의 줄기세포성 프로그램 재활성화에 의존하여 분화한다는 점을 보고하였다.10,11 IFN 유전자 신호전달 경로의 활성화,54 세포외 칼륨 농도 상승,55 c-Myb 과발현 등 여러 전략이 치료용 CD8+ T 세포의 줄기세포성을 재프로그래밍하는 것으로 밝혀졌다. 본 연구는 IPA가 Tpex 세 인터페론 유전자 신호전달 경로 활성화,54 세포외 칼륨 농도 상승,55 c-Myb 과발현56 등이 치료용 CD8+ T 세포의 줄기세포성을 재프로그래밍하는 것으로 밝혀졌다. 본 연구는 미생물군집 유래 IPA가 Tcf7 유전자 SE 부위의 H3K27 아세틸화 수준을 증가시켜 CD8+ T 세포의 줄기세포성 프로그램을 조절하고 Tpex 세포 분화를 촉진할 수 있음을 확인하였다.

결론적으로, 본 연구는 L. johnsonii가 IPA 합성 증가와 Tpex 세포 활성 증진을 통해 ICB의 치료 효능을 향상시킬 수 있음을 입증한다. L. johnsonii 및 박테리아 유래 IPA는 맞춤형 암 면역치료를 받는 환자에게 효과적인 약물 보조제로 활용될 잠재력을 지닌다.

Limitations of the study

There are several limitations to our study. The exact mechanisms by which IPA induces acetylation modifications need to be further investigated. The role of ldhA enzyme in the production of ILA by L. johnsonii could be better confirmed by genetic engineering. Whether IPA can be used as a stemness regulator of CD8+ T cells for adjuvant immunotherapy needs to be validated in larger clinical cohorts. Additionally, improving IPA bioavailability to increase intratumoral concentration is another potential future research direction.

STAR★MethodsKey resources table

REAGENT or RESOURCESOURCEIDENTIFIER

Antibodies
InVivoMAb rat IgG2a isotype control (Rat, Clone 2A3)	BioXcell	Cat# BE0089; RRID: AB_1107769
InVivoMAb anti-mouse PD-1 (CD279) (Rat, Clone 29F.1A12™)	BioXcell	Cat# BE0273; RRID: AB_2687796
InVivoMAb anti-mouse CD8α (Rat, Clone 53-6.7)	BioXcell	Cat# BE0004-1; RRID: AB_1107671
Anti-mouse CD45 (Cell sorting) (Rat, Clone30-F11), PerCP-Cy™5.5 conjugated	BD Biosciences	Cat# 561869; RRID: AB_394003
Anti-mouse CD3 (Cell sorting) (Rat, Clone 17A2), FITC conjugated	BioLegend	Cat# 100203; RRID: AB_312660
Anti-mouse CD8α (Cell sorting) (Rat, Clone 53-6.7), APC/Cyanine7 conjugated	BioLegend	Cat# 100714; RRID: AB_312753
Anti-mouse/human TCF-1 (Flow cytometry) (Mouse, Clone S33-966), BV421 conjugated	BD Biosciences	Cat# 566692; RRID: AB_2869822
Mouse IgG1, κ Isotype Control (Flow cytometry) (Mouse, Clone X40), BV421 conjugated	BD Biosciences	Cat# 562438; RRID: AB_11207319
Anti-mouse CD45 (Flow cytometry) (Rat, Clone 30-F11), Alexa Fluor 700 conjugated	BioLegend	Cat# 103128; RRID: AB_493715
Anti-mouse CD3e (Flow cytometry) (Armenian Hamster, Clone 145-2C11) PerCP-Cy™5.5 conjugated	BD Biosciences	Cat# 551163; RRID: AB_394082
Anti-mouse CD4 (Flow cytometry) (Rat, Clone RM4-5), BV605 conjugated	BD Biosciences	Cat# 563151; RRID: AB_2687549
Anti-mouse CD8α (Flow cytometry) (Rat, Clone 53-6.7), APC-Cy™7 conjugated	BD Biosciences	Cat# 561967; RRID: AB_396769
Anti-mouse NK1.1 (Flow cytometry) (Mouse, Clone PK136), BV711 conjugated	BioLegend	Cat# 108745; RRID: AB_2563286
Anti-mouse CD45R/B220 (Flow cytometry) (Rat, Clone RA3-6B2), PerCP-Cy™5.5 conjugated	BioLegend	Cat# 103236; RRID: AB_893354
Anti-mouse Ly-6G (Flow cytometry) (Rat, Clone 1A8), PE/Dazzle™ 594 conjugated	BioLegend	Cat# 127648; RRID: AB_2566319
Anti-mouse CD11c (Flow cytometry) (Armenian Hamster, Clone N418), PE/Cyanine7 conjugated	BioLegend	Cat# 117318; RRID: AB_493568
Anti-mouse I-A/I-E (Flow cytometry) (Rat, Clone M5/114.15.2), PE conjugated	BD Biosciences	Cat# 557000; RRID: AB_396546
Anti-mouse F4/80 (Flow cytometry) (Rat, Clone BM8), BV421 conjugated	BioLegend	Cat# 123132; RRID: AB_11203717
Anti-mouse CD11b (Flow cytometry) (Rat, Clone M1/70), BV605 conjugated	BioLegend	Cat# 101257; RRID: AB_2565431
Anti-mouse PD-1 (Flow cytometry) (Rat, Clone 29F.1A12), FITC conjugated	BioLegend	Cat# 135213; RRID: AB_10689633
Anti-mouse TNF-α (Flow cytometry) (Rat, Clone MP6-XT22), PE/Dazzle™ 594 conjugated	BioLegend	Cat# 506346; RRID: AB_2565955
Anti-mouse IFN-γ (Flow cytometry) (Rat, Clone XMG1.2), BV421 conjugated	BioLegend	Cat# 505830; RRID: AB_2563105
Anti-human CD45 (Flow cytometry) (Mouse, Clone HI30), FITC conjugated	BioLegend	Cat# 304006; RRID: AB_314394
Anti-human CD3 (Flow cytometry) (Mouse, Clone UCHT1), PerCP-Cy™5.5 conjugated	BD Biosciences	Cat# 560835; RRID: AB_2033956
Anti-human CD8a (Flow cytometry) (Mouse, Clone HIT8a), APC conjugated	BioLegend	Cat# 300911; RRID: AB_314115
Anti-human TIM-3 (Flow cytometry) (Mouse, Clone 7D3) PE conjugated	BD Biosciences	Cat# 563422; RRID: AB_2716866
Anti-human PD-1 (Flow cytometry) (Mouse, Clone EH12.1) BV605 conjugated	BD Biosciences	Cat# 563245; RRID: AB_2738091
anti-human PD-1 Antibody (nivolumab, ALI-PDOs)	Bristol Myers Squibb	N/A
Ultra-LEAF™ Purified anti-human CD3 Antibody (ALI-PDOs) (Mouse, Clone OKT3)	BioLegend	Cat# 317326; RRID: AB_11150592
Ultra-LEAF™ Purified anti-human CD28 Antibody (ALI-PDOs) (Mouse, Clone CD28.2)	BioLegend	Cat# 302934; RRID: AB_11148949
Ultra-LEAF™ Purified anti-mouse CD3 Antibody (Cell activation) (Rat, Clone 17A2)	BioLegend	Cat# 100239; RRID: AB_2810313
Ultra-LEAF™ Purified anti-mouse CD28 Antibody (Cell activation) (Syrian Hamster, Clone 37.51)	BioLegend	Cat# 102115; RRID: AB_11150408
Acetyl-Histone H2A-K5 Rabbit pAb (western blot)	ABclonal	Cat# A15620; RRID: AB_2763027
Histone H2A Rabbit mAb (western blot) (Clone ARC2072)	ABclonal	Cat# A3692; RRID: AB_2863118
Acetyl-Histone H2B-K5 Rabbit pAb (western blot)	ABclonal	Cat# A15621; RRID: AB_2763028
Histone H2B Rabbit mAb (western blot) (Clone ARC2337)	ABclonal	Cat# A19812
MonoMethyl-Histone H4-K20 Rabbit pAb (western blot)	ABclonal	Cat# A2370; RRID: AB_2764330
Histone H4 Rabbit pAb (western blot)	ABclonal	Cat# A1131; RRID: AB_2758500
Histone H3 Rabbit pAb (western blot)	ABclonal	Cat# A2348; RRID: AB_2631273
TCF-1/TCF7 (C63D9) Rabbit mAb (western blot, Immunofluorescence)	Cell Signaling Technology	Cat# 2203S
Acetyl-Histone H3-K27 Rabbit mAb (CUT & Tag, CUT & RUN, ChIP, western blot) (clone ARC54943)	ABclonal	Cat# A22264
CD8 Antibody (Immunofluorescence)	Hubei BIOSSCI Biotech Co., LTD	Cat# za-0508
TruStain FcX™ (anti-mouse CD16/32) Antibody (Fc block) (Rat, clone 93)	BioLegend	Cat# 101320; RRID: AB_1574975
Bacterial and virus strains
Lactobacillus johnsonii	This paper	N/A
Clostridium sporogenes	ATCC	Cat# 11437
E. coli BL21 (DE3)	Yeasen Biotechnology	Cat# 11804ES80
Biological samples
Colorectal cancer tumor tissues and adjacent normal tissues	Sir Run Run Shaw Hospital, Zhejiang University School of Medicine	N/A
Feces from healthy individuals and adenoma and colorectal cancer patients	Sir Run Run Shaw Hospital, Zhejiang University School of Medicine	N/A
Chemicals, peptides, and recombinant proteins
De Man, Rogosa and Sharpe (MRS) Medium	hopebio, China	Cat# HB0384-5
Reinforced Clostridium Medium (RCM)	hopebio, China	Cat# HB0316
Luria broth (LB) medium	Sangon Biotech, China	Cat# A507002
Ampicillin	meilunbio, China	Cat# MB1507
Metronidazole	meilunbio, China	Cat# MB2200
Neomycin	meilunbio, China	Cat# MB1716
Vancomycin	meilunbio, China	Cat# MB1260
Matrigel	Yeasen Biotechnology	Cat# 40183ES10
Penicillin-Streptomycin	Gibco	Cat# 15140122
Fetal Bovine Serum	Gibco	Cat# 10437028
7-AAD Viability Staining Solution (Cell sorting)	BioLegend	Cat# 420404
Isopropyl-β-d-thiogalactoside (IPTG)	Selleck	Cat# S6826
Recombinant Mouse IL-2 Protein (Cell activation)	BioLegend	Cat# 575404
Recombinant Human IL-2 Protein (ALI-PDOs)	ABclonal	Cat# RP01039
Indole-3-propionic acid (IPA)	Sigma	Cat# V900491
Indole-3-lactic acid (ILA)	Bide Pharmatech Ltd.	Cat# BD13033
Collagenase IV	Worthington	Cat# LS004189
Fixable viability Stain 510 (Flow cytometry)	BD Biosciences	Cat# 564406; RRID: AB_2869572
FluoroFix Buffer	BioLegend	Cat# 422101
Intracellular Staining Permeabilization Wash Buffer	BioLegend	Cat# 421002
Transcription Factor Staining Buffer	Invitrogen™	Cat# 00-5523-00
Cell Activation Cocktail with Brefeldin A	BioLegend	Cat# 423303
Solution A ((Cellmatrix I-A)	Nitta Gelatin	N/A
Solution B (10× concentrated sterile culture medium, Ham’s F-12)	Sigma-Aldrich	Cat# D8900
Organoid Medium (ALI-PDOs)	Shanghai Bioheb Biomed Technology Co., Ltd.	Cat# I-ALI-OC-Medium-20221011
Critical commercial assays
MojoSort™ Mouse CD8 T Cell Isolation Kit	BioLegend	Cat# 480035
MojoSort™ Mouse Pan B Cell Isolation Kit	BioLegend	Cat# 480051
ChIP Assay kit	Thermo Scientific™	Cat# 26156
Hyperactive pG-MNase CUT&RUN Assay Kit for PCR/qPCR	Vazyme	Cat# HD101
Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro	Vazyme	Cat# TD904
TruePrep Index Kit V2 for Illumina	Vazyme	Cat# TD202
TIANamp Stool DNA Kit	TIANGEN	Cat# DP328-02
TIANamp Genomic DNA Kit	TIANGEN	Cat# DP304-02
ABScript III RT Master Mix	ABclonal	Cat# RK20429
SYBR Green Fast qPCR Mix	ABclonal	Cat# RK21203
SteadyPure RNA extraction kit	Accurate Biology	Cat# AG21017
Deposited data
Raw sequencing data: scATAC-seq	This paper	PRJCA023433 GSA: CRA014884
Raw sequencing data: scTCR-seq	This paper	PRJCA023433 GSA: CRA014885
Raw sequencing data: scRNA-seq	This paper	PRJCA023433 GSA: CRA014886
Experimental models: Cell lines
Mus: Mc38	BMCR	Cat# 1101MOU-PUMC000523
Mus: B16-F10	ATCC	Cat# CRL-6475
Mus: 4T1	ATCC	Cat# CRL-2539
Experimental models: Organisms/strains
Mouse: C57BL/6J	Shanghai SLAC Laboratory Animals	N/A
Mouse: BALB/c	Shanghai SLAC Laboratory Animals	N/A
Mouse: Germ-Free: C57BL/6JGpt	GemPharmatech Co., Ltd.	Strain NO. N000295
Mouse: Rag1-/-: C57BL/6JGpt-Rag1em1Cd3259/Gpt	GemPharmatech Co., Ltd.	Strain NO. T004753
Mouse: Tcf7-/-: C57BL/6Smoc-Tcf7em1Smoc	Shanghai Model Organisms Center, Inc.	Strain NO. NM-KO-190688
Mouse: C57BL/6-JMMTV-PyMT	Cyagen Biosciences Inc.	Strain NO.C001212
Oligonucleotides
Lactobacillus johnsonii forward: TCGAGCGAGCTTGCCTAGATGA	TsingkeBiotechnologyCo., Ltd.	N/A
Lactobacillus johnsonii reverse: TCCGGACAACGCTTGCCACC	TsingkeBiotechnologyCo., Ltd.	N/A
Clostridium sporogenes forward: AAGCTTCCTTCGGGAAGTGG	TsingkeBiotechnologyCo., Ltd.	N/A
Clostridium sporogenes reverse: CCTTTCGGAAGGCTATCCCC	TsingkeBiotechnologyCo., Ltd.	N/A
Universal Eubacteria 16S forward: CGGCAACGAGCGCAACCC	TsingkeBiotechnologyCo., Ltd.	N/A
Universal Eubacteria 16S reverse: CCATTGTAGCACGTGTGTAGCC	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse β-actin forward: ACACCCGCCACCAGTTCGC	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse β-actin reverse: ATGGGGTACTTCAGGGTCAGGATA	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse Tcf7 SE forward: GGTTGTCTGGAGGTCAGTGG	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse Tcf7 SE reverse: GAACTTGCTCATCCCAGCA	TsingkeBiotechnologyCo., Ltd.	N/A
Human PRF1 forward: GTGGGACAATAACAACCCCAT	TsingkeBiotechnologyCo., Ltd.	N/A
Human PRF1 reverse: TGGCATGATAGCGGAATTTTAGG	TsingkeBiotechnologyCo., Ltd.	N/A
Human GZMB forward: TGGGGGACCCAGAGATTAAAA	TsingkeBiotechnologyCo., Ltd.	N/A
Human GZMB reverse: TTTCGTCCATAGGAGACAATGC	TsingkeBiotechnologyCo., Ltd.	N/A
Human IFNG forward: TCGGTAACTGACTTGAATGTCCA	TsingkeBiotechnologyCo., Ltd.	N/A
Human IFNG reverse: TCGCTTCCCTGTTTTAGCTGC	TsingkeBiotechnologyCo., Ltd.	N/A
Human β-ACTIN forward: AGAGCTACGAGCTGCCTGAC	TsingkeBiotechnologyCo., Ltd.	N/A
Human β-ACTIN reverse: AGCACTGTGTTGGCGTACAG	TsingkeBiotechnologyCo., Ltd.	N/A
Software and algorithms
GraphPad Prism software 9.0	GraphPad Software, Inc.	https://graphpad.com/scientificsoftware/prism/
FlowJo-10.4	Tree Star Inc.	https://www.flowjo.com/solutions/flowjo
BD FACSDiva 9.0.1	BD Biosciences	https://www.bdbiosciences.com/en-us/products/software/instrument-software/bd-facsdiva-software
Cell Ranger-7.0.0	10 X Genomics	http://10xgenomics.com/
VISION-3.0.1	DeTomaso et al.30	https://github.com/YosefLab/VISION
Seurat-4.1.1	Stuart et al.57	https://satijalab.org/seurat/
scRepertoire 2.0.0	Borcherding et al.58	https://doi.org/10.12688/f1000research.22139.2
Monocle 2/Monocle 3	Qiu et al.59	http://cole-trapnell-lab.github.io/monocle-release/docs/
BWA-0.7.12	Vasimuddin et al.60	https://github.com/lh3/bwa
Macs2-2.1.0	Zhang et al.61	https://github.com/macs3-project/MACS
Bedtools-2.30.0	Quinlan laboratory	https://bedtools.readthedocs.io/
bcl2fastq-5.0.1	Illumina, Inc.	https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html
RStudio Server	RStudio, PBC	https://www.rstudio.com/products/
R-4.1.2	R-project	https://www.r-project.org/
Other
70 μm cell strainer	biosharp	Cat# BS-70-XBS
MojoSort™ Magnet	BioLegend	Cat# 480019
Millicell dish	Millipore	Cat# PICM01250
Amino acid control feed (2 g/kg L-tryptophan)	Readydietech Co., Ltd.	N/A
Tryptophan-free feed (0 g/kg L-tryptophan)	Readydietech Co., Ltd.	N/A

• Open table in a new tab

Resource availabilityLead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Liangjing Wang (wangljzju@zju.edu.cn).

Materials availability

Mouse and microbial strains used in this study are available from the lead contact upon request.

Data and code availability

•

The raw sequencing data that support the findings of this study are deposited (PRJCA023433, https://ngdc.cncb.ac.cn/) under the supervision and control of the Genome Sequence Archive of the Beijing Institute of Genomics, Chinese Academy of Sciences, under the accession number: CRA014884, CRA014885, CRA014886 (accessible at GSA, https://ngdc.cncb.ac.cn/gsa/). Publicly available databases and software used in this work are noted in the STAR Methods and the key resources table.

•

This paper does not report original code.

•

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant detailsCell lines

All cell lines were described in the key resources table. A 10 % fetal bovine serum (Gibco, 10437028) and 1 % penicillin/streptomycin solution (Gibco, 15140122) were used to culture the cells. They were kept at 37 °C in a humidified atmosphere containing 5 % CO2.

Microbe strains

All microbe strains were described in the key resources table. Referring to our previous method,62 Lactobacillus johnsonii was isolated independently and verified using 16S rRNA sequencing (V4 sequences). The complete genome sequence was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The bacteria were cultured in MRS medium (hopebio, HB0384-5) at 37 °C for 24 h. To get high-temperature inactivated Lactobacillus johnsonii, the bacteria were heated in the 100 °C metal bath for 2 h. To get ultrasonic broken Lactobacillus johnsonii, the bacteria were lysed using the Fisher Scientific™ Model 50 Sonic Dismembrator. The conditions of ultrasonic fragmentation were 300 w, 20 s start-up and 10 s pause, 20 min. As a representative bacterium for IPA production, Clostridium sporogenes was purchased from ATCC (ATCC 11437) and cultured in RCM (hopebio, HB0316) under an atmosphere of 10 % CO2, 10 % H2 and 80 % N2 for 72 h at 37 °C.

Human participants

Tissue samples from 92 patients with CRC (Cohort 1), fecal samples from 67 healthy control, 40 patients with colon adenoma, 108 patients with CRC (Cohort 2) were obtained. After surgical resection, CRC tissue samples and their adjacent normal mucosa were immediately frozen in liquid nitrogen and stored at -80 °C. Fresh CRC tissues from 10 patients using for ALI-PDOs or flow cytometry were obtained. The patients were without antibiotics and probiotics in the past one month. Informed consent was obtained from all participants, and the experimental protocol was approved by the Clinical Research Ethics Committee of Sir Run Run Shaw Hospital, Zhejiang University School of Medicine (20211103-35).

Animals

All experiments used mice were described in the key resources table. Unless otherwise stated, 6-8 weeks males or females were used in all experiments; no significant sex-dependent differences were found in the experiments reported. They were under specific pathogen-free (SPF) environment, water and food were provided adequately. The temperature was constant, and a 12 h circadian rhythm was maintained every day. All animals were done under the guidelines of the Animal Experimentation Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine (AIRB-2022-0336), Zhejiang Chinese Medical University (20221212-07, 20230918-14) and IACUC of GemPharmatech (GPTAP20231010-3). For the germ-free mice model, feces were microbiologically tested weekly to confirm‎ sterility or specific microbial colonization status.

Method detailsTumor models and treatments

Antibiotics (Abx) treatment was performed to normalize the gut microbiota in animal experiments. Mice were given 0.2 g/L Ampicillin (meilunbio, MB1507), 0.2 g/L Metronidazole (meilunbio, MB2200), 0.2 g/L Neomycin (meilunbio, MB1716) and 0.1 g/L Vancomycin (meilunbio, MB11260) quadruple antibiotics for one week.63 For transplantable tumor models, 106 Mc38, 105 B16-F10, 105 4T1 and Matrigel (Yeasen Biotechnology, 40183ES10) were co-injected into the loose subcutaneous tissues of the mice back, 100 μl of cell suspension each mouse. For orthotopic tumor models, 4T1 or Mc38 were implanted into the mammary fat pad or the cecum respectively. On the 8th, 11th, and 14th days, each mouse was intraperitoneally injected with 100 μg of IgG isotype control antibody (BioXcell, BE0089) or αPD-1 antibody (BioXcell, BE0273) for immunotherapy. Transgenic MMTV-PyMT mice (Cyagen Biosciences Inc., C001212) at 16 weeks were used as an autochthonous breast cancer model for survival experiments. Each mouse was given 100 μg αPD-1 antibody for immunotherapy every three days and removed from the study as soon as an individual tumor reached a volume of 1000 mm3.44 For CD8+ T cells neutralization experiments, 200 μg of αCD8 antibody (BioXcell, BE0004-1) were delivered by intraperitoneal injection, once every three days. For bacterial treatment, each mouse was given 200 μl of PBS per day by gavage containing 109 CFU of bacteria. For metabolites treatment, each mouse was gavaged at a dose of 60 mg/kg (or at concentration gradients as indicated) per day.28 In some experiments, ILA (Bide Pharmatech Ltd., BD13033) and IPA (Sigma, V900491) were administered by intraperitoneal or intratumoral injection.

Fecal microbiota transplantation

Referring to published studies,64 fresh feces from Responder Donors (n=5) and Poor-responder Donors (n=5) mice were collected and resuspended in sterile PBS solution at a ratio of 40 mg/ml. Subsequently, grinding beads were added to the mixture and filtered through a 70 μm cell strainer (Biosharp, BS-70-XBS). 10 % volume of glycerol was added to the fecal suspension, and it was sub packaged and frozen at -80°C. For each Abx-treated mouse, 200 μl of fecal suspension was orally administered three times each week.

Adoptive cell transfer therapy

Referring to previous work,65,66 mice were sacrificed to separate tumor-draining lymph nodes or spleen, and cell clumps were removed by grinding and filtering. After PBS washing and cells counting, cells were resuspended in culture medium. CD8+ T cells or B cells were collected by magnetic cell separation system as the manufacturer’s protocol (BioLegend, 480019, 480035, 480051). The sorted cells were counted. 106 CD8+ T cells, 106 B cells, or 106 CD8+ T cells and 106 B cells were transferred into per Rag1-/- mice by tail vein injection.

Heterologous expression‎ of ldhA

Standard molecular cloning techniques were used to clone the DNA fragments encoding full-length ldhA into the pRSFDuet vector. After overexpression‎ of ldhA in E. coli BL21 (Yeasen Biotechnology, 11804ES80), the bacteria were grown at 37 °C to an OD600 of 0.6. Afterwards, ldhA protein expression‎ was induced with 200 μΜ IPTG (Selleck, S6826) for 6 h at 37 °C. E. coli and ldhA-overexpressing E. coli (E. c-ldhA) were incubated in LB medium (Sangon Biotech, A507002) for 24 h at 37 °C.

Ex vivo CD8+ T cells differentiation

CD8+ T cells were purified from lymph nodes or spleen of mice using MojoSort™ Mouse CD8 T cell Isolation Kit (BioLegend, 480035). Purified CD8+ T cells were activated with αCD3 (5 μg/ml, BioLegend, 100239), αCD28 (5 μg/ml, BioLegend, 102115) and 50 U/ml Il-2 (BioLegend, 575404). In specific experiments, CD8+ T cells were treated with 0, 5 μM or 500 μM IPA for 48 h. For flow cytometry, CD8+ T cells were stained with anti-CD8, anti-PD-1 and anti-TCF-1 antibodies (BD Biosciences, 561967, 566692, BioLegend, 135213). For western blot, total protein from CD8+ T cells was extracted and blocked with anti-TCF-1, anti-H2AK5ac, anti-H2BK5ac, anti-H3K27ac, anti-H4K20me1 antibodies (Cell Signaling Technology, 2203S, ABclonal, A15620, A15621, A22264, A2370) at 4 °C overnight. Histone H2A, H2B, H3 and H4 (ABclonal, A3692, A19812, A2348, A1131) served as the loading controls.

ChIP-qPCR assay and analysis

ChIP was conducted following the manufacturer’s protocol (Thermo Scientific™, 26156). Briefly, PBS or 500 μM IPA-treated CD8+ T cells were cross-linked with 1 % formaldehyde at room temperature for 10 min, which was terminated by Glycine Solution. After centrifugation, the acquired pellet was lysed by the indicated lysis buffer with a protease inhibitor cocktail. Then the extracted genomic DNA was digested enzymatically to achieve DNA fragmentation using micrococcal nuclease. Transfer 5 μl of the supernatant containing the digested chromatin to a 1.5 ml tube and store at -20 °C as the 10 % total input sample from one ChIP. Then the samples were subjected to immunoprecipitation at 4 °C overnight with anti-H3K27ac (ABclonal, A22264). Add 20 μl ChIP Grade Protein A/G Plus Agarose to each IP and incubate for 1 h at 4 °C on a rocking platform. After resin incubation, the precipitated protein-DNA complexes were eluted, reversal of crosslinking, DNA clean-up and subjected to qPCR.

CUT & RUN assay and analysis

CUT & RUN assay was conducted following the manufacturer’s protocol (Vazyme, HD101). Briefly, PBS or 500 μM IPA-treated CD8+ T cells were incubated with ConA Beads Pro at room temperature for 10 min, anti-H3K27ac antibody (1:50, ABclonal, A22264) was added and rotated at room temperature for 2 h, then washed twice, added pG-MNase Enzyme and incubated at 4 °C for 1 h, then washed twice, Cacl2 was added and incubated for 1 h on ice, added stop buffer and incubated at 37 °C for 30 min. DNA was extracted and qPCR was used to detect the acetylation of Tcf7 SE. Spike in DNA derived from the λDNA of E. coli was used for uniform correction.

CUT & Tag assay and analysis

PBS or 500 μM IPA-treated CD8+ T cells were collected with six replicates per group for CUT & Tag assay (Vazyme, TD904). The sequencing process was commissioned to chi-biomedicine (Guangdong, China). Briefly, CD8+ T cells were resuspended with wash buffer, incubated with bead, incubated with primary anti-H3K27ac antibody (1:50, ABclonal, A22264) and incubated with secondary antibodies. Then the samples were incubated with Hyperactive pA/G-Transposon Pro, fragmented with Mgcl2. The DNA was extracted and PCR amplification was performed using indexing primers (Vazyme, TD202). CUT & Tag library was purified and assessed on the Agilent Bioanalyzer 2100 system. The clustering of the index-coded samples was performed on a cBot Cluster Generation System and the library preparations were sequenced on the Illumina Novaseq platform with 150 bp paired-end reads. Sequencing data was further processed by bioinformatics pipelines including raw data cleaning (removing containing, low-quality reads), reference mm10 mouse genome mapping (MAPQ >=13), peak calling (with the q-value threshold of 0.05), peak annotation (ChIPseeker) and different peak analysis (fold change of RPM >= 2). Bam files were visualized using IGV.

Air-liquid interface PDOs

Referring to previous work,47,48 air-liquid interface patient-derived organoids (ALI-PDOs) were established. Briefly, fresh human CRC tissues were clipped in pre-chilled PBS and washed in medium containing antibiotics. Collagen cocktail solution was constructed by mixing Solution A (Nitta Gelatin), Solution B (Sigma-Aldrich, D8900) and Solution C (Sterile reconstitution buffer containing 2.2 g NaHCO3 in 100 ml of 0.05 N NaOH and 200 mM HEPES) at a ratio of 8:1:1 on ice, and 300 μl of collagen cocktail solution was added to the Millicell dish (Millipore, PICM01250) to form the bottom gel layer, which was then solidified at 37 °C for 30 min. The excised tumor tissues were then resuspended with a further 300 μl of collagen cocktail solution and added to the top of the pre-solidified gel layer. 12-well plates were filled with outer wells supplemented with organoid medium (Shanghai Bioheb Biomed Technology Co., Ltd., I-ALI-OC-Medium-20221011). On this basis, additional 50 U/ml human recombinant IL-2 protein (Abclonal, RP01039) and 5 μg/ml human anti-CD3/CD28 antibody (BioLegend, 317326, 302934) were added for the transient culture of T cells. On top of the above medium, 10 μg/ml αPD-1 antibody (nivolumab, Bristol Myers Squibb) was added at the same time. 5 mM IPA or equal amounts of DMSO (Mock) were added every 2 days. The external medium was changed every 2 days. For immunofluorescence, after about 7 days of culture, the up-layer gel was removed with forceps and paraffin-embedded tissue were stained with anti-CD8 antibody (1:4000, Hubei BIOSSCI Biotech Co., LTD, za-0508), anti-TCF-1 antibody (1:50, Cell Signaling Technology, 2203S).

Flow cytometry analysis

At the end of modeling, the subcutaneous tumors were collected and quickly immersed in cold medium. After cutting them into 1 mm pieces, collagenase IV (Worthington, LS004189) was added for 30 min to further digest. The tumor-draining lymph node and spleen were gently grounded and filtered through a 70 μm cell strainer (biosharp, BS-70-XBS). Splenocytes were resuspended in RBC Lysis buffer and incubated for 15 minutes at 4 °C. Then, 1 × 106 cells were incubated with 1 μg of anti-CD16/32 antibody (BioLegend, 101320) for 10 min to block non-specific binding of immunoglobulin to the Fc receptors. Then, cells were stained for live cells (BD Biosciences, 564406) for 30 minutes. Afterwards, the appropriate amount of pre-diluted fluorescent labeled antibody (BioLegend, 103128, 108745, 103236, 127648, 117318, 123132, 101257, 135213, BD Biosciences, 551163, 563151, 561967, 557000) was added to each tube as recommended by the manufacturer. Cells were incubated in the 4 °C refrigerator for 30 min and fixed with Fixation Buffer (BioLegend, 422101) for 30 min at room temperature. After washing, the membranes were ruptured (BioLegend, 421002) and intracellular fluorescent antibodies were added to each tube and incubated for 30 min at room temperature. The effectors IFN-γ and TNF-α (BioLegend, 505830, 506346) were detected by adding 2 μl Activation Cocktails (BioLegend, 423303) to 1 ml of cell suspension and incubating for 6 h at 37 °C. For TCF-1 staining, cells were incubated with Fixation work solution for 40 minutes at room temperature, washed with Transcription Factor Staining Buffer (Invitrogen™, 00-5523-00), and stained with antibody (BD Biosciences, 566692) for 40 minutes at room temperature.

Metabolomic analysis

For plasma samples, 5-10 times the volume of pre-cooled methanol was added to plasma, vortexed, and shaken, and incubated for 1 h at -20 °C. After centrifuging samples for 10 min at 14,000 g, the supernatant was collected and centrifuged again repeatedly. The supernatant was filtered and tested.42 For culture supernatant samples, after centrifuging for 10 min at 14,000 g to remove the organisms and impurities, the supernatant was acidified to pH=2.5 using hydrochloric acid and extracted twice with a double volume of ethyl acetate. Then, air-dried and redissolved in one-tenth volume of methanol. The supernatant was filtered and assayed.

For Untargeted LC-MS/MS, prepared plasma samples as described above. The supernatant was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Briefly, metabolites were detected on the Thermo UHPLC-Q Exactive system. The analytical conditions follow: buffer A was 95 % ddH2O plus 5 % acetonitrile (containing 0.1 % formic acid) and buffer B was 47.5 % acetonitrile with 47.5 % isopropanol and 5 % ddH2O (containing 0.1 % formic acid). The gradient conditions were as follows: Positive ion mode: 0 to 3 min, gradient to 20 % buffer B; 3 to 4.5 min, gradient to 35 % buffer B; 4.5 to 5 min, gradient to 100 % buffer B; 5 to 6.3 min, 100 % buffer B; 6.3 to 6.4 min, gradient to 0 % buffer B; 6.4 to 8 min, 0 % buffer B. Negative ion mode: 0 to 1.5 min, gradient to 5 % buffer B; 1.5 to 2 min, gradient to 10 % buffer B; 2 to 4.5 min, gradient to 30 % buffer B; 4.5 to 5 min, gradient to 100 % buffer B; 5 to 6.3 min, 100 % buffer B; 6.3 to 6.4 min, gradient to 0 % buffer B; 6.4 to 8 min, 0 % buffer B. The flow rate was 0.40 ml/min. HMDB (http://www.hmdb.ca/) and Metlin (https://metlin.scripps.edu/) were used to match the mass spectrometry data. On the basis of the KEGG database, metabolic enrichment and pathway analysis were performed on the differential metabolites.

IPYA, ILA, IA and IPA were detected as described above. Metabolites were detected on the 5500 QTRAP triple quadrupole mass spectrometer (SCIEX, Framingham, MA, USA). The analytical conditions follow: buffer A was ddH2O plus 0.1 % formic acid and buffer B was acetonitrile plus 0.1 % formic acid. The gradient conditions were as follows: 0 to 1 min, 5 % buffer B; 1 to 7 min, gradient to 95 % buffer B; 7 to 10 min, 95% buffer B; 10 to 13 min, gradient to 5 % buffer B; 13 to 14 min, 5 % buffer B. The flow rate was 0.40 ml/min.

16S rRNA sequencing

Feces from both ‘Responder’ (n=5) and ‘Poor-responder’ (n=5) groups of mice were collected. Genomic DNA from fecal samples was extracted and tested for concentration and purity using electrophoresis and NanoDrop 2000. The subsequent library construction and sequencing process was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Briefly, full-length PCR amplification of the 16S rRNA gene was performed using 27F (5’-AGRGTTYGATYMTGGCTCAG-3′) and 1492R (5′-RGYTACCTTGTTACGACTT-3′) primers, followed by library construction and sequencing. UPARSE 7.1 was used to cluster the sequencing results into operational taxonomic units (OTUs), with 97 % sequence similarity and chimeras were removed. Taxonomic annotation of OTUs using the RDP classifier ratio against the Silva 16S rRNA gene database with a 70 % confidence threshold. The similarity of microbial community structure between samples was examined using PCoA based on the Bray-Curtis distance algorithm.

Shotgun metagenomic sequencing

Total genomic DNA was extracted from both ‘Responder’ (n=5) and ‘Poor-responder’ (n=5) groups of feces with the PF Mag-Bind Stool DNA Kit according to the manufacturer’s instructions. The subsequent library construction and sequencing process was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Briefly, DNA extract was fragmented to an average size of about 400 bp for paired-end library construction. The paired-end library was constructed using NEXTFLE Rapid DNA-Seq. Paired-end sequencing was performed on Illumina Novaseq 6000 (Illumina Inc., San Diego, CA, USA) according to the manufacturer’s instructions. Metagenome sequencing data was further processed by bioinformatics pipelines on the online platform of Majorbio Cloud Platform (www.majorbio.com) with data quality-filtering (length<50 bp or with a quality value <20 or having N bases), assembly, genomic contamination elimination and taxonomic compositions identification.

Single-cell RNA and TCR sequencing

To perform single-cell RNA sequencing analysis, we treated mice under each treatment condition (WT (αPD-1) and IPA-treated (IPA+αPD-1), n=2 biological replicates/group). Flow sorting was used to obtain tumor-infiltrating CD8+ T cells (BD Biosciences, 561869, BioLegend, 420404, 100203, 100714) and cell suspensions were dissociated and loaded into the 10x Chromium instrument. After library construction, the NovaSeq 6000 sequencing platform was used for RNA sequencing and a sequencing depth of 20,000 reads per cell was required. Illumina bcl2fastq software was used to demultiplex and convert the sequencing data into FASTQ format files. TCR library construction and TCR V(D)J targeted enrichment were performed with the Chromium Single Cell V(D)J Enrichment Kit (10x Genomics) according to the manufacturer’s user guide. For dimension reduction, clustering, and analysis of the scRNA-seq data, the Cell Ranger output was loaded into Seurat.57 All cells were filtered by quality control conditions (Expression‎ of all genes was detected in at least 3 cells, with the number of genes expressed in a single cell between 500 and 5000, the number of UMIs greater than or equal to 500, and the ratio of mitochondrial gene expression‎ less than 25 %). The filtered cells were analyzed by UMAP downscaling using R software to obtain visualization results. Using the scRepertoire (v.2.0.0) package,58 sample-specific consensus annotation files were consolidated into a list of TCR sequencing results and then integrated with the Seurat object for visualization. Enrichment analysis was performed by Vision package (v2.1.0).30 Monocle 3 was applied to perform trajectory analysis.59 The other Single Cell Data Analysis was performed using the OmicStudio tools created by LC-BIO Co., Ltd (Hangzhou, China) at https://www.omicstudio.cn/cell.

Single-cell ATAC sequencing

To perform single-cell ATAC-seq, we treated mice under each treatment condition (WT (αPD-1) and IPA-treated (IPA+αPD-1), n=2 biological replicates/group). Flow sorting was used to obtain tumor-infiltrating CD8+ T cells (BD Biosciences, 561869, BioLegend, 420404, 100203, 100714) and cell suspensions were dissociated and loaded into the 10x Chromium instrument. Single-cell ATAC-seq libraries were prepared according to the Chromium Single Cell ATAC Library Kit from 10x Genomics following the manufacturer’s instructions. 10x Genomics Cell Ranger ATAC pipeline (version 2.1.0) was used for scATAC-seq analyses of alignment, deduplication, and identification of transposase cut sites (https://support.10xgenomics.com/). The trimmed read-pairs are aligned to a specified reference using BWA-MEM with default Parameters.60 Fragments were identified as read pairs with mapping quality (MAPQ)>30. Reads were counted across the genome, using 500-bp bins (tiles) to generate a genome-wide tile-count-matrix. Latent semantic indexing (LSI), Louvain clustering and Harmorny algorithm were applied. Gene activity scores were computed as the summed local accessibility of promoter-associated count-tiles in the proximity of each gene, using a distance-weighted accessibility model. Finally, the above-weighted sum was multiplied by the aggregated Tn5 insertions in each tile. Gene scores were then scaled to 10,000 counts and log2-normalized. To enhance the visual interpretation of gene activity scores, smoothing was applied using the MAGIC algorithm. Downstream bioinformatics pipelines analysis was conducted with R (R version 4.1.2) package ArchR (version 1.0.2) and Seurat (version 4.1.1). ScATAC-seq cluster was identified by gene activity scores and scRNA-seq gene expression‎. MACS2 was applied to identify a robust merged peak.61 Then peaks were annotated according to their respective genomic position (promoter, intronic, exonic, distal etc). Differential accessibility analysis between cells was performed to identify celltype-specific and condition-specific marker peaks (|Log2FC| > 0.5 and p-value < 0.01).

Quantitative real-time PCR

Bacterial DNA was extracted from mouse and human feces using the TIANGEN Fecal Kit (TIANGEN, DP328-02) and bacterial genomic DNA from human tumor and adjacent normal mucosa using the TIANGEN DNA Kit (TIANGEN, DP304-02). RNA from ALI-PDOs was extracted using the SteadyPure RNA extraction kit (Accurate Biology, AG21017) and reverse transcribed using ABScript III RT Master Mix (ABclonal, RK20429). qPCR was performed in Light Cycler® 480 real-time PCR system (Roche) using SYBR Green Fast qPCR Mix (ABclonal, RK21203). cDNA was amplified by PCR under the following conditions: 95 °C for 3 min, followed by 45 cycles of 95 °C for 5 s and 60 °C for 30 s. The specific Primer sequences were listed in the key resources table. The universal Eubacteria 16S or β-ACTIN was used as internal reference genes.

Quantification and statistical analysis

GraphPad Prism was used for all statistical analyses. The experiments were designed to use a minimum of 3 samples/replicates per experiment or per group. Representative immunofluorescent staining, LC-MS/MS and flow cytometry images are presented. Each experiment was repeated in triplicate independently. The data are expressed as the mean ± standard deviation (SD) or mean ± standard error of mean (SEM). Differences between groups were analyzed by two-tail ratio paired t test, unpaired t test, Wilcoxon rank-sum test, Mann Whitney test, one-way ANOVA with Sidak’s correction for multiple comparisons, two-way ANOVA with Sidak’s correction for multiple comparisons, Spearman correlation analysis and log-rank test. Statistically significant were P values < 0.05.

Acknowledgments

The authors would like to express their gratitude to all colleagues who contributed to this work, in particular Prof. Jianmin Si, Prof. Di Wang, Prof. Yongqun Zhu, Prof. Lie Wang, Dr. Dong Cen (Zhejiang University), and Prof. Zheng Kuang (Carnegie Mellon University). Thanks to Yanwei Li (the Core Facilities, Zhejiang University School of Medicine) and Zhanglian He (Biomedical Research Center, Sir Run Run Shaw Hospital, Zhejiang University) for technical support in flow cytometry. Thanks to Zhongjing Zhou and Yi Teng (Zhejiang Academy of Agricultural Sciences) for technical support in LC-MS/MS analysis. This project was financially supported by the National Foundation of Natural Science of China (82273269, 82072623 to Liangjing Wang; 82270573 to S.C.), the Key program of Natural Science Foundation of Zhejiang Province (LZ22H160002 to Liangjing Wang), and the National Key Research and Development Program of China (2022YFC2505100 to Liangjing Wang).

Author contributions

Conceptualization, Liangjing Wang and S.C.; methodology, D.J., Q.W., Y.Q., Y.J., and J.H.; software, Y.Q. and Q.W.; formal analysis, D.J., Q.W., Y.Q., Y.J., and J.H.; investigation, Y.L., Y.S., J.X., W.C., L.F., R.Y., C.X., and Q.G.; resources, W.Z., G.R., Lan Wang, W.L., F.X., and P.W.; writing – original draft, D.J.; writing – review & editing, Liangjing Wang, S.C., and Y.W.; funding acquisition, Liangjing Wang and S.C.; supervision, Liangjing Wang.

Declaration of interests

The authors declare no competing interests.

References

1.

Fernandes, M.R. ∙ Aggarwal, P. ∙ Costa, R.G.F. ...

Targeting the gut microbiota for cancer therapy

Nat. Rev. Cancer. 2022; 22:703-722

Crossref

Scopus (85)

PubMed

Google Scholar

2.

Jiang, Y. ∙ Jia, D. ∙ Sun, Y. ...

Microbiota: A key factor affecting and regulating the efficacy of immunotherapy

Clin. Transl. Med. 2023; 13, e1508

Crossref

Google Scholar

3.

Wang, F. ∙ He, M.M. ∙ Yao, Y.C. ...

Regorafenib plus toripalimab in patients with metastatic colorectal cancer: a phase Ib/II clinical trial and gut microbiome analysis

Cell Rep. Med. 2021; 2, 100383

Google Scholar

4.

Davar, D. ∙ Dzutsev, A.K. ∙ McCulloch, J.A. ...

Fecal microbiota transplant overcomes resistance to anti-PD-1 therapy in melanoma patients

Science. 2021; 371:595-602

Crossref

Scopus (807)

PubMed

Google Scholar

5.

Si, W. ∙ Liang, H. ∙ Bugno, J. ...

Lactobacillus rhamnosus GG induces cGAS/STING- dependent type I interferon and improves response to immune checkpoint blockade

Gut. 2022; 71:521-533

Crossref

Scopus (134)

PubMed

Google Scholar

ArticleVolume 187, Issue 7p1651-1665.e21March 28, 2024Open Archive

Download Full Issue

Microbial metabolite enhances immunotherapy efficacy by modulating T cell stemness in pan-cancer

Dingjiacheng Jia1,2,11 ∙ Qiwen Wang2,3,11 ∙ Yadong Qi1,2,11 ∙ … ∙ Yuhao Wang10 ∙ Shujie Chen2,3,4 chenshujie77@zju.edu.cn ∙ Liangjing Wang1,2,4,12 wangljzju@zju.edu.cn … Show more

Affiliations & NotesArticle Info

• Download PDF
• Cite
•  
• Set Alert
• Get Rights
• Reprints

Download Full Issue

Previous articleNext article

Show Outline

Highlights

•

L. johnsonii enhances ICB efficacy via CD8+ T cells

•

L. johnsonii cooperates with C. sporogenes to produce IPA

•

IPA activates progenitor exhausted CD8+ T cells by H3K27 acetylation modification

•

IPA improves ICB responsiveness at the pan-cancer level

Summary

The immune checkpoint blockade (ICB) response in human cancers is closely linked to the gut microbiota. Here, we report that the abundance of commensal Lactobacillus johnsonii is positively correlated with the responsiveness of ICB. Supplementation with Lactobacillus johnsonii or tryptophan-derived metabolite indole-3-propionic acid (IPA) enhances the efficacy of CD8+ T cell-mediated αPD-1 immunotherapy. Mechanistically, Lactobacillus johnsonii collaborates with Clostridium sporogenes to produce IPA. IPA modulates the stemness program of CD8+ T cells and facilitates the generation of progenitor exhausted CD8+ T cells (Tpex) by increasing H3K27 acetylation at the super-enhancer region of Tcf7. IPA improves ICB responsiveness at the pan-cancer level, including melanoma, breast cancer, and colorectal cancer. Collectively, our findings identify a microbial metabolite-immune regulatory pathway and suggest a potential microbial-based adjuvant approach to improve the responsiveness of immunotherapy.

Graphical abstract

Keywords

1. cancer immunotherapy
2. immune checkpoint blockade
3. TCF-1
4. progenitor-exhausted CD8+ T cells
5. indole-3-propionic acid
6. Lactobacillus johnsonii
7. gut microbiota
8. tryptophan metabolism

Introduction

Gut microbiota is one of the major environmental factors that affects the efficacy of immune checkpoint blockade (ICB).1,2 The composition of the gut microbiota differs in patients,3 and fecal microbiota transplantation (FMT) has been shown to improve ICB responsiveness.4 Specific strains, such as Lactobacillus rhamnosus GG,5 Bifidobacterium breve,6 and Fusobacterium nucleatum,7 have been found to improve the efficacy of ICB therapy by modulating the innate and adaptive immune response processes. However, specific bacterial components and detailed underlying mechanisms remain elusive.

Infiltrating T cells in the tumor microenvironment play a key role in ICB therapy.8 Multiple CD8+ T cell subsets have been revealed to influence anti-tumor response.9 Recent studies have shown that progenitor exhausted CD8+ T cells (Tpex), which are characterized by high expression‎ of the transcription factor T cell factor 1 (TCF-1, encoded by TCF7), are the main subset responding to anti-programmed cell death protein 1 antibody (αPD-1) immunotherapy.10,11,12,13,14 Tpex cells exhibiting stemness characteristics demonstrate the ability to self-renew, proliferate, and differentiate into effector CD8+ T cells (Teff) to limit tumorigenesis.15 Moreover, the expression‎ level of TCF7 could be served as a positive prognostic indicator for patients treated with ICB.16 Because a clear correlation between the deficiency of Tcf7 and a disrupted balance of the gut microbiota has been observed in mice,17 further investigation is needed to determine whether the gut microbiota and its metabolites could regulate the stemness program of CD8+ T cells.

In this study, we discovered that the abundance of commensal Lactobacillus johnsonii (L. johnsonii) is positively associated with the proportion of Tpex cells, as well as improved the responsiveness to ICB. The tryptophan metabolite indole-3-propionic acid (IPA), derived from L. johnsonii, could enhance H3K27 acetylation at the super-enhancer (SE) of Tcf7 gene and thus promote the differentiation of CD8+ T cells into Tpex cells. Importantly, we found that the synthesis of IPA requires cooperation between L. johnsonii and another symbiotic bacterium, Clostridium sporogenes (C. sporogenes, C. s). Overall, our findings point to a microbial-based adjuvant approach to improve the effectiveness of immunotherapy.

ResultsL. johnsonii sensitizes αPD-1 immunotherapy by activating tumor-infiltrating CD8+ T cells

When mice are given the same dose of αPD-1, their tumor growth could exhibit either a good or poor response, similar to what is observed in clinical practice.18,19,20 Using the Mc38 cell line (high microsatellite instability, MSI-H),21,22 we modeled αPD-1 therapy for colorectal cancer (CRC) and divided the mice into two groups, namely, the “poor-responder” group and the “responder” group, based on tumor volume (Figures 1A and 1B). We then separately transplanted feces from each group into recipient mice that were pretreated with antibiotics (Abx) (Figure 1C). Mice transplanted with the poor-responder feces (PR-recipient) showed lower responsiveness to αPD-1 therapy than mice transplanted with the responder feces (R-recipient) (Figure 1D), suggesting that the gut microbiota has a direct impact on the reactivity of ICB therapy.

Figure 1 L. johnsonii sensitizes αPD-1 immunotherapy by activating tumor-infiltrating CD8+ T cells

Show full captionFigure viewer

We then performed metagenomics and 16S ribosomal RNA (rRNA) sequencing on feces from both the poor-responder and the responder mice to compare the microbial communities. There were differences in β-diversity observed through principal co-ordinates analysis (PCoA) (Figures 1E and S1A). Colony composition analysis at the genus level revealed a decrease in the abundance of Lactobacillus in the poor-responder group (Figures 1F, S1B, and S1C). At the species level, L. johnsonii was found to be the most preval‎ent (Figure 1G). We confirmed that the lower abundance of L. johnsonii in the poor-responder group negatively correlated with tumorigenesis (Figures S1D and S1E). Apcmin/+ or azoxymethane (AOM)/dextran sodium sulfate (DSS)-induced CRC mice also showed decreased L. johnsonii colonization (Figures S1F and S1G). Likewise, compared with adjacent normal tissues, the abundance of L. johnsonii was significantly lower in human CRC tissues (Figure S1H) and negatively correlated with the advanced stage of human CRC (Figure S1I).

Figure S1 L. johnsonii reduces in colorectal cancer and is related to ICB responsiveness through activating IFN-γ+ and TNF-α+ CD8+ T cells, related to Figure 1

Show full captionFigure viewer

We isolated a strain of L. johnsonii and explored its functional role. Oral administration of L. johnsonii enhanced the response to αPD-1 therapy in Mc38 tumor (Figure 1H). CD8+ T cells, which are the main immune effector cells for anti-tumor response,23 were found to increase in frequency after oral administration of L. johnsonii (Figures 1I and 1J). This increase was accompanied by the release of cytokines from Teff such as interferon (IFN)-γ and tumor necrosis factor alpha (TNF-α) (Figures S1J and S1K).

L. johnsonii promotes ICB responsiveness via tryptophan metabolism

To explore whether such a function was attributed to L. johnsonii itself or its secreted products, we administered mice with L. johnsonii that had been high-temperature inactivated (heated group), ultrasonic disrupted (ultrasound group), the original medium (de Man, Rogosa, and Sharpe [MRS] group), and the conditional medium (CM) (Lj. CM group) in addition to live L. johnsonii. Interestingly, only Lj. CM group (Lj. CM + αPD-1) had a comparable immunotherapeutic effect to live L. johnsonii group (L. j + αPD-1), indicating that the antitumor effect of L. johnsonii was primarily through the molecules secreted by the bacteria (Figures 2A and 2B).

Figure 2 L. johnsonii promotes ICB responsiveness via tryptophan metabolism

Show full captionFigure viewer

Plasma liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis revealed distinct differences in the metabolite composition between the αPD-1 alone group and the live L. johnsonii group (L. j + αPD-1) (Figure 2C). The live L. johnsonii group showed a significant enrichment of the “tryptophan metabolism” pathway (Figure 2D). We then confirmed the necessity of tryptophan in L. johnsonii-promoted ICB responsiveness by subjecting mice to a tryptophan deficiency diet (Trpneg diet).24 As expected, L. johnsonii did not enhance the effectiveness of αPD-1 therapy or increase the frequency CD8+ T cells when tryptophan substrates were absent (Figures 2E–2G).

L. johnsonii-derived IPA promotes ICB responsiveness via CD8+ T cells

We examined the plasma tryptophan-related metabolites that were upregulated by L. johnsonii and found that IPA exhibited a most pronounced increase (Figure 3A). IPA, an indole analog that is specifically produced by the gut microbiota,25,26 has been shown to promote axonal regeneration27 and protect against radiation toxicity.28 Consistent with L. johnsonii administration, mice transplanted with feces from the responder group showed higher plasma IPA levels (Figures 3B, S2A, and S2B). However, supplementation with L. johnsonii did not increase plasma IPA levels in mice that were fed with Trpneg diet (Figures 3C and S2C).

Figure 3 L. johnsonii-derived IPA promotes ICB responsiveness via CD8+ T cells

Show full captionFigure viewer

We then explored the potential impact of IPA on tumor immunotherapy. Administering 60 mg/kg/day of IPA through gavage could enhance the effectiveness of αPD-1 and increase plasma or intratumoral IPA levels (Figures 3D and S2D–S2G). Similar enhancement could also be achieved by direct intratumoral injection (i.m.) of 5 μΜ IPA (Figures S2H and S2I). IPA supplementation increased the frequency of infiltrating CD8+ T cells and the production of their effector cytokines in tumors (Figures 3E–3G and S2Q). However, no significant alterations were observed for natural killer (NK) cells, B cells, dendritic cells, and macrophages (Figures S2J–S2P).

Figure S2 L. johnsonii and its metabolite, IPA, sensitize αPD-1 immunotherapy, and affect associated immunocytes, related to Figure 3

Show full captionFigure viewer

To determine whether the immunotherapeutic effect is dependent on T cell response, L. johnsonii or IPA was administered to Rag1−/− mice. Notably, both treatments failed to reduce tumor growth (Figures 3H and 3I). Moreover, treating mice with CD8 neutralizing antibody abrogated the antitumor effect of both L. johnsonii and IPA (Figure 3J). We conjectured that CD8+ T cells treated with IPA could be maintained in an activated state and confirmed this by adoptively transferring pretreated CD8+ T cells or B cells to Rag1−/− mice (Figure 3K). Mice receiving IPA-pretreated CD8+ T cells (IPA T) showed better αPD-1 responsiveness compared with mice receiving IPA-untreated CD8+ T cells (wild-type [WT] T). However, mice that received both B cells and CD8+ T cells pretreated with IPA (IPA B + T) showed no better tumor inhibitory effect compared with the IPA T group (Figure 3L). Together, these findings indicate that L. johnsonii-derived metabolite IPA is dependent on CD8+ T cells to promote the responsiveness to ICB therapy.

IPA sustains CD8+ T cells response by promoting Tpexcells

CD8+ T cells consist of various subsets of cells that have distinct functions and work collaboratively to regulate the immune microenvironment of tumors.8,29 To further reveal the role of IPA on CD8+ T cell subsets, we performed single-cell RNA sequencing (scRNA-seq), single-cell T cell receptor sequencing (scTCR-seq), and single-cell assay for targeting accessible-chromatin with high-throughput sequencing (scATAC-seq) on CD8+ T cells sorted from tumors in mice that have been treated with either αPD-1 alone (WT group) or combination with IPA (IPA group) (Figure 4A). To exclude the interference of macrophages, neutrophils, and fibroblasts (Figures S3A–S3C), we performed the uniform manifold approximation and projection (UMAP) analysis to obtain relatively pure CD8+ T cells (Figure S3D). We identified 13 cell clusters and 10 CD8+ T cell subsets: naive (Lef1hi, Sellhi, and Ccr7hi), progenitor exhausted CD8+ T cells (Tpex, Tcf7hi and Pdcd1hi), activated (Cd69hi and Isg15hi), Teff (Prf1hi, Gzmbhi, and Klrd1hi), natural killer T cells (NKT, Cd160hi and Xcl1hi), Trbv3hi, Cotl1hi, Bcl2hi, Mki67hi, and Mcm3hi (Figures 4B and S3E–S3H).

Figure 4 IPA sustains CD8+ T cells response by promoting Tpex cells

Show full captionFigure viewer

Figure S3 Single-cell RNA sequencing analysis shows IPA promotes the differentiation of CD8+ T cells by upregulating progenitor exhausted CD8+ T cells, related to Figure 4

Show full captionFigure viewer

IPA treatment reduced naive CD8+ T cells but increased Tpex cells and Teff cells in tumors (Figure 4C). To comprehensively map immune cell function,30,31 we performed VISION analysis to project the total score of the Gene Ontology (GO) biological process gene sets and calculate the average signature score for each cluster (Figure 4D). “GO_ALPHA_BETA_T_CELL_DIFFERENTIATION” was enriched in 9 of 13 cell clusters after IPA treatment, particularly in Tpex cells (Figures 4E–4G). Also, “GO_ALPHA_BETA_T_CELL_ACTIVATION” was enriched in 7 of 13 cell clusters after IPA treatment, particularly in Teff cells (Figures S3I–S3K). In addition, IPA treatment could increase the proportion of TCF-1+ stem-like CD8+ T cells in ex vivo culture (Figures 4H, 4I, and S4E) and in tumor-draining lymph node (TDLN) or spleen (Figures S4A–S4D). Because Tcf7 expression‎ marks the activation of Tpex cells, we knocked out Tcf7 in mice (Tcf7−/−) and found that IPA only enhanced the efficacy of immunotherapy in WT littermates but not in Tcf7−/− mice (Figure 4J). Collectively, these findings suggest that IPA exerts its immunotherapy-enhancing effects by elevating Tpex cells in tumors.

Figure S4 Flow cytometry analysis, monocle analysis, and single-cell TCR sequencing analysis are performed on CD8+ T cells after IPA treatment, related to Figure 4

Show full captionFigure viewer

To further examine the differentiation route of CD8+ T cells after IPA treatment, we performed trajectory analysis of four classical subsets of CD8+ T cells. Monocle analysis revealed that naive CD8+ T cells developed into Tpex cells and activated CD8+ T cells and Teff cells step by step (Figure 4K). This differentiation process was enhanced by IPA treatment as a higher frequency of Tpex cells was enriched during the late stage of development (Figures S4F and S4G). Using scTCR-seq, we identified clonotypes between WT group and IPA group (Figure S4H). The subsets of CD8+ T cells shared TCR profiles and underwent clonal expansion with each other. The predominant TCR clonotypes in Teff cells primarily originated from the Tpex cells, and IPA treatment increased the proportion of conservative transmission (Figures 4L and 4M). In short, IPA increased the frequency of Tpex cells in the tumor microenvironment and promoted their differentiation to Teff cells to enhance ICB responsiveness.

IPA activates Tpex cells by modifying the H3K27 acetylation of Tcf7 gene at its SE region

Tpex cells are mainly regulated by histone modifications,32,33,34 we sought to explore whether IPA activates Tpex cells through histone remodeling of Tcf7 gene. We mapped the results of scRNA-seq to scATAC-seq and identified 10 common CD8+ T cell subsets (Figures 5A, S5A, and S5B). Consistent with scRNA-seq results, IPA treatment increased the proportions of Tpex cells and Teff cells (Figure 5B). Further differential peak analysis revealed that open chromatin regions of Tpex cells were mainly enriched in the promoter and distal intergenic regions (Figure S5C), and IPA treatment increased the chromatin opening of Tcf7 gene at its super-enhancer (SE) region (Figures 5C and 5J). In addition, we noticed that IPA treatment primarily enhanced histone acetylation but not histone methylation in Tpex cells (Figures 5D and S5D), suggesting that IPA may activate Tpex cells by modifying the histones of Tcf7 gene at its SE through acetylation.

Figure 5 IPA activates Tpex cells by modifying the H3K27 acetylation of Tcf7 gene at its SE region

Show full captionFigure viewer

Figure S5 IPA affects chromosome accessibility in progenitor exhausted CD8+ T cells, related to Figure 5

Show full captionFigure viewer

Nucleosome, the basic unit of chromatin, is an octamer composed of four core histones (H2A, H2B, H3, and H4).35 scRNA-seq analysis revealed that histone H3 acetylation was most significantly enriched in Tpex cells after IPA treatment (Figures 5E, S5E, and S5F). We performed western blot analysis with antibodies specific to common histone modifications, including H2AK5ac,36 H2BK5ac,37 H3K27ac, and H4K20me1,38 to validate this result. Consistently, only H3K27ac exhibited a dose-dependent increase in response to IPA, along with an increase in TCF-1 expression‎ (Figure 5F). Moreover, we performed chromatin immunoprecipitation (ChIP), cleavage under targets and release using nuclease (CUT&RUN), and cleavage under targets and tagmentation (CUT&Tag) on H3K27ac in sorted CD8+ T cells. Results confirmed that IPA treatment increases H3K27 acetylation at the SE of Tcf7 gene (Figures 5G–5J).39

Taken together, above results prove that IPA promotes the stemness program of CD8+ T cells by modifying the H3K27 acetylation of Tcf7 gene at its SE region.

L. johnsonii cooperates with C. sporogenes to produce IPA

Tryptophan could be metabolized by the gut microbiota into indole-3-pyruvate acid (IPYA), indole-3-lactic acid (ILA), indole-3-acrylic acid (IA), and IPA in a sequential manner,26,40 but how L. johnsonii metabolizes tryptophan to produce IPA remains elusive. Surprisingly, we were unable to detect IPA or IA in the supernatant of the CM after culturing L. johnsonii. However, ILA was detected in the CM. This suggests that L. johnsonii could only metabolize tryptophan into ILA and not further into IA or IPA (Figures 6A, 6B, and S6A).

Figure 6 L. johnsonii cooperates with C. sporogenes to produce IPA

Show full captionFigure viewer

To confirm‎ that ILA lacks the ability to enhance ICB responsiveness if it is not further converted to IPA by gut microbes, we administered either ILA or IPA to Mc38 tumor-bearing mice. As expected, when ILA was given to mice receiving Abx pretreatment, it did not improve the efficacy of αPD-1 therapy and no longer raised plasma IPA levels, which were in contrast to IPA administration group (Figures 6C, 6D, and S6B). Similarly, intraperitoneal injection of ILA, which avoids further catabolism by gut microbiota,41 did not improve αPD-1 effectiveness or increase plasma IPA (Figures 6E, 6F, and S6C). These findings lead to the hypothesis that L. johnsonii requires collaboration with other commensal microbes to produce IPA.

Because C. sporogenes is known to convert ILA to IPA,42,43 and oral gavage of L. johnsonii increased the fecal abundance of C. sporogenes (Figure S6D), we selected it as a representative bacterium and tested our hypothesis in germ-free mice (Figure 6G). In comparison with single-bacteria (L. johnsonii or C. sporogenes) administration, mixed administration of L. johnsonii and C. sporogenes further enhanced efficacy of αPD-1 (Figure 6H). LC-MS/MS analysis verified that L. johnsonii alone was unable to increase plasma IPA, despite being able to produce large amounts of ILA. In contrast, C. sporogenes or mixture of L. johnsonii and C. sporogenes significantly increased plasma IPA (Figures 6I and S6E). These results confirmed that ILA produced by L. johnsonii needs to be further metabolized to IPA by C. sporogenes in order to enhance the efficacy of immunotherapy.

Figure S6 The cooperation of L. johnsonii with D-2-hydroxyacid dehydrogenase and C. sporogenes promotes IPA production, related to Figure 6

Show full captionFigure viewer

We then proceeded to explore how L. johnsonii produces ILA. We performed whole-genome sequencing for our isolated L. johnsonii and used the protein basic local alignment search tool (BLAST) to search for sequences that were similar to fldH, a gene from C. sporogenes that encodes a key enzyme for converting tryptophan to ILA25 (Figure 6J). We found that ldhA, a gene present in the genome of L. johnsonii that encodes a 337 amino acid D-2-hydroxyacid dehydrogenase, may be essential for L. johnsonii to produce ILA (Figure S6F). The secondary structure of the ldhA-encoded enzyme was predicted by 3D modeling (Figure S6G). To test the function of this enzyme, we transformed a ldhA expression‎al vector into E. coli (E. c-ldhA) and induced protein expression‎ (Figure S6H). Administration with E. c-ldhA to Mc38 tumor-bearing mice increased plasma IPA levels and improved αPD-1 reactivity (Figures 6K, 6L, and S6I), confirm‎ing that L. johnsonii relies on ldhA to encode enzyme for ILA production.

IPA promotes ICB responsiveness in pan-cancer and CRC-derived organoids

We then tested whether microbiota-derived IPA improves the responsiveness of ICB therapy at the pan-cancer level. IPA effectively promoted the efficacy of αPD-1 immunotherapy (Figures 7A and 7C) and increased the frequency of infiltrating CD8+ T cells and the expression‎ of TCF-1 (Figures 7B, 7D, S7A, and S7B) in both breast cancer and melanoma transplantable tumor model. We also verified the sensitizing effect of IPA for immunotherapy in mammary fat pad orthotopic implantation model, murine mammary tumor virus-polymavirus middle T antigen (MMTV-PyMT) spontaneous breast cancer model,44,45 and cecum orthotopic implantation model (Figures 7E–7I, S7C, and S7D). Immunofluorescence staining verified that IPA treatment increased the infiltration of Tpex cells in the tumor microenvironment of breast cancer, melanoma, and CRC (Figures 7J and 7K).

Figure 7 IPA promotes ICB responsiveness in pan-cancer and CRC-derived organoids

Show full captionFigure viewer

We further investigated whether IPA could act on human CD8+ T cells to improve the responsiveness of ICB therapy. In fresh CRC tissues, although the majority of CD8+ T cells in the tumor microenvironment are TIM-3+ PD-1+ terminally exhausted cells, there is also a specific proportion of Tpex cells present (Figures S7G and S7H). Considering the limitation of preserving immune cells in the tumor microenvironment using conventional patient-derived organoids (PDOs),46 we established an air-liquid interface (ALI)-PDOs system. This system contains a more complete immune microenvironment and additional matrix components, allowing for a more accurate representation of immunotherapy47,48 (Figure 7L). Both IPA-treated and mock-treated PDOs successfully preserved the CRC structures (Figure 7M). IPA-treated PDOs exhibited increased the infiltration of CD8+ T cells (Figures S7E and S7F) and elevated the expression‎ of Teff cells effector proteins (Figure 7N). Collectively, these results demonstrate that modulating the stemness program of CD8+ T cells through microbiota-derived IPA may be a promising approach to reinforce the effectiveness of tumor ICB therapy in clinical settings.

Figure S7 IPA modulates ICB responsiveness in pan-cancer and CRC-derived organoids, related to Figure 7

Show full captionFigure viewer

Discussion

In this study, we found that L. johnsonii, a commensal bacterium in the intestine, can enhance the efficacy of ICB therapy mainly by increasing the synthesis of tryptophan-derived metabolite IPA. Many metabolites derived from bacteria cannot be synthesized by a single bacterium alone but require participation of other microbes. Recent studies have reported that C. scindens collaborates with its closely related strains to transform primary bile acids into secondary bile acids.49 Bacteroides collaborates in a relaying manner with Lactobacillus plantarum to metabolize host tyrosine.50 We present an additional example in which L. johnsonii requires collaboration with C. sporogenes to convert dietary tryptophan into IPA.

A variety of tryptophan-indole metabolites derived from bacteria, including ILA, IA, IPA, indole-3-acetic acid (IAA), and indole-3-acetamide (IAM), have been shown to have a wide range of biological effects.26,51 IPA, specifically, could improve atherosclerosis,52 mitigate radiotoxicity,28 and prolong the posttraumatic survival time.53 The impact of IPA on modulating cancer immunotherapy has not been reported. In this study, we discovered that IPA mainly promotes the infiltration of CD8+ T cells into tumors and activates Tpex cells during immunotherapy. Dodd and colleagues found that mutant C. sporogenes unable to produce IPA colonized germ-free mice could affect intestinal permeability and increase the number of memory T cells in peripheral blood.25 The interactions between the bacteria and host immunity may regulate the intestinal barrier, and this mechanism needs to be further explored.

A number of studies have reported that the responses to anti-PD-1 are mediated by Tpex cells, which rely on the reactivation of the stemness program of CD8+ T cells for their differentiation.10,11 Several strategies, including activation of IFN gene signaling pathway,54 elevation of extracellular potassium levels,55 and overexpression‎ of c-Myb56 have been shown to reprogram the stemness of therapeutic CD8+ T cells. Our study revealed that microbiota-derived IPA is able to modulate the stemness program of CD8+ T cells and promote the differentiation of Tpex cells by increasing H3K27 acetylation level at the SE of Tcf7 gene.

In conclusion, our study establishes that L. johnsonii can boost the therapeutic efficacy of ICB by increasing the synthesis of IPA and enhancing Tpex cell activity. L. johnsonii and bacterial-derived IPA may potentially serve as an effective drug adjuvant for patients receiving personalized cancer immunotherapy.

Limitations of the study

There are several limitations to our study. The exact mechanisms by which IPA induces acetylation modifications need to be further investigated. The role of ldhA enzyme in the production of ILA by L. johnsonii could be better confirmed by genetic engineering. Whether IPA can be used as a stemness regulator of CD8+ T cells for adjuvant immunotherapy needs to be validated in larger clinical cohorts. Additionally, improving IPA bioavailability to increase intratumoral concentration is another potential future research direction.

STAR★MethodsKey resources table

REAGENT or RESOURCESOURCEIDENTIFIER

Antibodies
InVivoMAb rat IgG2a isotype control (Rat, Clone 2A3)	BioXcell	Cat# BE0089; RRID: AB_1107769
InVivoMAb anti-mouse PD-1 (CD279) (Rat, Clone 29F.1A12™)	BioXcell	Cat# BE0273; RRID: AB_2687796
InVivoMAb anti-mouse CD8α (Rat, Clone 53-6.7)	BioXcell	Cat# BE0004-1; RRID: AB_1107671
Anti-mouse CD45 (Cell sorting) (Rat, Clone30-F11), PerCP-Cy™5.5 conjugated	BD Biosciences	Cat# 561869; RRID: AB_394003
Anti-mouse CD3 (Cell sorting) (Rat, Clone 17A2), FITC conjugated	BioLegend	Cat# 100203; RRID: AB_312660
Anti-mouse CD8α (Cell sorting) (Rat, Clone 53-6.7), APC/Cyanine7 conjugated	BioLegend	Cat# 100714; RRID: AB_312753
Anti-mouse/human TCF-1 (Flow cytometry) (Mouse, Clone S33-966), BV421 conjugated	BD Biosciences	Cat# 566692; RRID: AB_2869822
Mouse IgG1, κ Isotype Control (Flow cytometry) (Mouse, Clone X40), BV421 conjugated	BD Biosciences	Cat# 562438; RRID: AB_11207319
Anti-mouse CD45 (Flow cytometry) (Rat, Clone 30-F11), Alexa Fluor 700 conjugated	BioLegend	Cat# 103128; RRID: AB_493715
Anti-mouse CD3e (Flow cytometry) (Armenian Hamster, Clone 145-2C11) PerCP-Cy™5.5 conjugated	BD Biosciences	Cat# 551163; RRID: AB_394082
Anti-mouse CD4 (Flow cytometry) (Rat, Clone RM4-5), BV605 conjugated	BD Biosciences	Cat# 563151; RRID: AB_2687549
Anti-mouse CD8α (Flow cytometry) (Rat, Clone 53-6.7), APC-Cy™7 conjugated	BD Biosciences	Cat# 561967; RRID: AB_396769
Anti-mouse NK1.1 (Flow cytometry) (Mouse, Clone PK136), BV711 conjugated	BioLegend	Cat# 108745; RRID: AB_2563286
Anti-mouse CD45R/B220 (Flow cytometry) (Rat, Clone RA3-6B2), PerCP-Cy™5.5 conjugated	BioLegend	Cat# 103236; RRID: AB_893354
Anti-mouse Ly-6G (Flow cytometry) (Rat, Clone 1A8), PE/Dazzle™ 594 conjugated	BioLegend	Cat# 127648; RRID: AB_2566319
Anti-mouse CD11c (Flow cytometry) (Armenian Hamster, Clone N418), PE/Cyanine7 conjugated	BioLegend	Cat# 117318; RRID: AB_493568
Anti-mouse I-A/I-E (Flow cytometry) (Rat, Clone M5/114.15.2), PE conjugated	BD Biosciences	Cat# 557000; RRID: AB_396546
Anti-mouse F4/80 (Flow cytometry) (Rat, Clone BM8), BV421 conjugated	BioLegend	Cat# 123132; RRID: AB_11203717
Anti-mouse CD11b (Flow cytometry) (Rat, Clone M1/70), BV605 conjugated	BioLegend	Cat# 101257; RRID: AB_2565431
Anti-mouse PD-1 (Flow cytometry) (Rat, Clone 29F.1A12), FITC conjugated	BioLegend	Cat# 135213; RRID: AB_10689633
Anti-mouse TNF-α (Flow cytometry) (Rat, Clone MP6-XT22), PE/Dazzle™ 594 conjugated	BioLegend	Cat# 506346; RRID: AB_2565955
Anti-mouse IFN-γ (Flow cytometry) (Rat, Clone XMG1.2), BV421 conjugated	BioLegend	Cat# 505830; RRID: AB_2563105
Anti-human CD45 (Flow cytometry) (Mouse, Clone HI30), FITC conjugated	BioLegend	Cat# 304006; RRID: AB_314394
Anti-human CD3 (Flow cytometry) (Mouse, Clone UCHT1), PerCP-Cy™5.5 conjugated	BD Biosciences	Cat# 560835; RRID: AB_2033956
Anti-human CD8a (Flow cytometry) (Mouse, Clone HIT8a), APC conjugated	BioLegend	Cat# 300911; RRID: AB_314115
Anti-human TIM-3 (Flow cytometry) (Mouse, Clone 7D3) PE conjugated	BD Biosciences	Cat# 563422; RRID: AB_2716866
Anti-human PD-1 (Flow cytometry) (Mouse, Clone EH12.1) BV605 conjugated	BD Biosciences	Cat# 563245; RRID: AB_2738091
anti-human PD-1 Antibody (nivolumab, ALI-PDOs)	Bristol Myers Squibb	N/A
Ultra-LEAF™ Purified anti-human CD3 Antibody (ALI-PDOs) (Mouse, Clone OKT3)	BioLegend	Cat# 317326; RRID: AB_11150592
Ultra-LEAF™ Purified anti-human CD28 Antibody (ALI-PDOs) (Mouse, Clone CD28.2)	BioLegend	Cat# 302934; RRID: AB_11148949
Ultra-LEAF™ Purified anti-mouse CD3 Antibody (Cell activation) (Rat, Clone 17A2)	BioLegend	Cat# 100239; RRID: AB_2810313
Ultra-LEAF™ Purified anti-mouse CD28 Antibody (Cell activation) (Syrian Hamster, Clone 37.51)	BioLegend	Cat# 102115; RRID: AB_11150408
Acetyl-Histone H2A-K5 Rabbit pAb (western blot)	ABclonal	Cat# A15620; RRID: AB_2763027
Histone H2A Rabbit mAb (western blot) (Clone ARC2072)	ABclonal	Cat# A3692; RRID: AB_2863118
Acetyl-Histone H2B-K5 Rabbit pAb (western blot)	ABclonal	Cat# A15621; RRID: AB_2763028
Histone H2B Rabbit mAb (western blot) (Clone ARC2337)	ABclonal	Cat# A19812
MonoMethyl-Histone H4-K20 Rabbit pAb (western blot)	ABclonal	Cat# A2370; RRID: AB_2764330
Histone H4 Rabbit pAb (western blot)	ABclonal	Cat# A1131; RRID: AB_2758500
Histone H3 Rabbit pAb (western blot)	ABclonal	Cat# A2348; RRID: AB_2631273
TCF-1/TCF7 (C63D9) Rabbit mAb (western blot, Immunofluorescence)	Cell Signaling Technology	Cat# 2203S
Acetyl-Histone H3-K27 Rabbit mAb (CUT & Tag, CUT & RUN, ChIP, western blot) (clone ARC54943)	ABclonal	Cat# A22264
CD8 Antibody (Immunofluorescence)	Hubei BIOSSCI Biotech Co., LTD	Cat# za-0508
TruStain FcX™ (anti-mouse CD16/32) Antibody (Fc block) (Rat, clone 93)	BioLegend	Cat# 101320; RRID: AB_1574975
Bacterial and virus strains
Lactobacillus johnsonii	This paper	N/A
Clostridium sporogenes	ATCC	Cat# 11437
E. coli BL21 (DE3)	Yeasen Biotechnology	Cat# 11804ES80
Biological samples
Colorectal cancer tumor tissues and adjacent normal tissues	Sir Run Run Shaw Hospital, Zhejiang University School of Medicine	N/A
Feces from healthy individuals and adenoma and colorectal cancer patients	Sir Run Run Shaw Hospital, Zhejiang University School of Medicine	N/A
Chemicals, peptides, and recombinant proteins
De Man, Rogosa and Sharpe (MRS) Medium	hopebio, China	Cat# HB0384-5
Reinforced Clostridium Medium (RCM)	hopebio, China	Cat# HB0316
Luria broth (LB) medium	Sangon Biotech, China	Cat# A507002
Ampicillin	meilunbio, China	Cat# MB1507
Metronidazole	meilunbio, China	Cat# MB2200
Neomycin	meilunbio, China	Cat# MB1716
Vancomycin	meilunbio, China	Cat# MB1260
Matrigel	Yeasen Biotechnology	Cat# 40183ES10
Penicillin-Streptomycin	Gibco	Cat# 15140122
Fetal Bovine Serum	Gibco	Cat# 10437028
7-AAD Viability Staining Solution (Cell sorting)	BioLegend	Cat# 420404
Isopropyl-β-d-thiogalactoside (IPTG)	Selleck	Cat# S6826
Recombinant Mouse IL-2 Protein (Cell activation)	BioLegend	Cat# 575404
Recombinant Human IL-2 Protein (ALI-PDOs)	ABclonal	Cat# RP01039
Indole-3-propionic acid (IPA)	Sigma	Cat# V900491
Indole-3-lactic acid (ILA)	Bide Pharmatech Ltd.	Cat# BD13033
Collagenase IV	Worthington	Cat# LS004189
Fixable viability Stain 510 (Flow cytometry)	BD Biosciences	Cat# 564406; RRID: AB_2869572
FluoroFix Buffer	BioLegend	Cat# 422101
Intracellular Staining Permeabilization Wash Buffer	BioLegend	Cat# 421002
Transcription Factor Staining Buffer	Invitrogen™	Cat# 00-5523-00
Cell Activation Cocktail with Brefeldin A	BioLegend	Cat# 423303
Solution A ((Cellmatrix I-A)	Nitta Gelatin	N/A
Solution B (10× concentrated sterile culture medium, Ham’s F-12)	Sigma-Aldrich	Cat# D8900
Organoid Medium (ALI-PDOs)	Shanghai Bioheb Biomed Technology Co., Ltd.	Cat# I-ALI-OC-Medium-20221011
Critical commercial assays
MojoSort™ Mouse CD8 T Cell Isolation Kit	BioLegend	Cat# 480035
MojoSort™ Mouse Pan B Cell Isolation Kit	BioLegend	Cat# 480051
ChIP Assay kit	Thermo Scientific™	Cat# 26156
Hyperactive pG-MNase CUT&RUN Assay Kit for PCR/qPCR	Vazyme	Cat# HD101
Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro	Vazyme	Cat# TD904
TruePrep Index Kit V2 for Illumina	Vazyme	Cat# TD202
TIANamp Stool DNA Kit	TIANGEN	Cat# DP328-02
TIANamp Genomic DNA Kit	TIANGEN	Cat# DP304-02
ABScript III RT Master Mix	ABclonal	Cat# RK20429
SYBR Green Fast qPCR Mix	ABclonal	Cat# RK21203
SteadyPure RNA extraction kit	Accurate Biology	Cat# AG21017
Deposited data
Raw sequencing data: scATAC-seq	This paper	PRJCA023433 GSA: CRA014884
Raw sequencing data: scTCR-seq	This paper	PRJCA023433 GSA: CRA014885
Raw sequencing data: scRNA-seq	This paper	PRJCA023433 GSA: CRA014886
Experimental models: Cell lines
Mus: Mc38	BMCR	Cat# 1101MOU-PUMC000523
Mus: B16-F10	ATCC	Cat# CRL-6475
Mus: 4T1	ATCC	Cat# CRL-2539
Experimental models: Organisms/strains
Mouse: C57BL/6J	Shanghai SLAC Laboratory Animals	N/A
Mouse: BALB/c	Shanghai SLAC Laboratory Animals	N/A
Mouse: Germ-Free: C57BL/6JGpt	GemPharmatech Co., Ltd.	Strain NO. N000295
Mouse: Rag1-/-: C57BL/6JGpt-Rag1em1Cd3259/Gpt	GemPharmatech Co., Ltd.	Strain NO. T004753
Mouse: Tcf7-/-: C57BL/6Smoc-Tcf7em1Smoc	Shanghai Model Organisms Center, Inc.	Strain NO. NM-KO-190688
Mouse: C57BL/6-JMMTV-PyMT	Cyagen Biosciences Inc.	Strain NO.C001212
Oligonucleotides
Lactobacillus johnsonii forward: TCGAGCGAGCTTGCCTAGATGA	TsingkeBiotechnologyCo., Ltd.	N/A
Lactobacillus johnsonii reverse: TCCGGACAACGCTTGCCACC	TsingkeBiotechnologyCo., Ltd.	N/A
Clostridium sporogenes forward: AAGCTTCCTTCGGGAAGTGG	TsingkeBiotechnologyCo., Ltd.	N/A
Clostridium sporogenes reverse: CCTTTCGGAAGGCTATCCCC	TsingkeBiotechnologyCo., Ltd.	N/A
Universal Eubacteria 16S forward: CGGCAACGAGCGCAACCC	TsingkeBiotechnologyCo., Ltd.	N/A
Universal Eubacteria 16S reverse: CCATTGTAGCACGTGTGTAGCC	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse β-actin forward: ACACCCGCCACCAGTTCGC	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse β-actin reverse: ATGGGGTACTTCAGGGTCAGGATA	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse Tcf7 SE forward: GGTTGTCTGGAGGTCAGTGG	TsingkeBiotechnologyCo., Ltd.	N/A
Mouse Tcf7 SE reverse: GAACTTGCTCATCCCAGCA	TsingkeBiotechnologyCo., Ltd.	N/A
Human PRF1 forward: GTGGGACAATAACAACCCCAT	TsingkeBiotechnologyCo., Ltd.	N/A
Human PRF1 reverse: TGGCATGATAGCGGAATTTTAGG	TsingkeBiotechnologyCo., Ltd.	N/A
Human GZMB forward: TGGGGGACCCAGAGATTAAAA	TsingkeBiotechnologyCo., Ltd.	N/A
Human GZMB reverse: TTTCGTCCATAGGAGACAATGC	TsingkeBiotechnologyCo., Ltd.	N/A
Human IFNG forward: TCGGTAACTGACTTGAATGTCCA	TsingkeBiotechnologyCo., Ltd.	N/A
Human IFNG reverse: TCGCTTCCCTGTTTTAGCTGC	TsingkeBiotechnologyCo., Ltd.	N/A
Human β-ACTIN forward: AGAGCTACGAGCTGCCTGAC	TsingkeBiotechnologyCo., Ltd.	N/A
Human β-ACTIN reverse: AGCACTGTGTTGGCGTACAG	TsingkeBiotechnologyCo., Ltd.	N/A
Software and algorithms
GraphPad Prism software 9.0	GraphPad Software, Inc.	https://graphpad.com/scientificsoftware/prism/
FlowJo-10.4	Tree Star Inc.	https://www.flowjo.com/solutions/flowjo
BD FACSDiva 9.0.1	BD Biosciences	https://www.bdbiosciences.com/en-us/products/software/instrument-software/bd-facsdiva-software
Cell Ranger-7.0.0	10 X Genomics	http://10xgenomics.com/
VISION-3.0.1	DeTomaso et al.30	https://github.com/YosefLab/VISION
Seurat-4.1.1	Stuart et al.57	https://satijalab.org/seurat/
scRepertoire 2.0.0	Borcherding et al.58	https://doi.org/10.12688/f1000research.22139.2
Monocle 2/Monocle 3	Qiu et al.59	http://cole-trapnell-lab.github.io/monocle-release/docs/
BWA-0.7.12	Vasimuddin et al.60	https://github.com/lh3/bwa
Macs2-2.1.0	Zhang et al.61	https://github.com/macs3-project/MACS
Bedtools-2.30.0	Quinlan laboratory	https://bedtools.readthedocs.io/
bcl2fastq-5.0.1	Illumina, Inc.	https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html
RStudio Server	RStudio, PBC	https://www.rstudio.com/products/
R-4.1.2	R-project	https://www.r-project.org/
Other
70 μm cell strainer	biosharp	Cat# BS-70-XBS
MojoSort™ Magnet	BioLegend	Cat# 480019
Millicell dish	Millipore	Cat# PICM01250
Amino acid control feed (2 g/kg L-tryptophan)	Readydietech Co., Ltd.	N/A
Tryptophan-free feed (0 g/kg L-tryptophan)	Readydietech Co., Ltd.	N/A

• Open table in a new tab

Resource availabilityLead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Liangjing Wang (wangljzju@zju.edu.cn).

Materials availability

Mouse and microbial strains used in this study are available from the lead contact upon request.

Data and code availability

•

The raw sequencing data that support the findings of this study are deposited (PRJCA023433, https://ngdc.cncb.ac.cn/) under the supervision and control of the Genome Sequence Archive of the Beijing Institute of Genomics, Chinese Academy of Sciences, under the accession number: CRA014884, CRA014885, CRA014886 (accessible at GSA, https://ngdc.cncb.ac.cn/gsa/). Publicly available databases and software used in this work are noted in the STAR Methods and the key resources table.

•

This paper does not report original code.

•

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant detailsCell lines

All cell lines were described in the key resources table. A 10 % fetal bovine serum (Gibco, 10437028) and 1 % penicillin/streptomycin solution (Gibco, 15140122) were used to culture the cells. They were kept at 37 °C in a humidified atmosphere containing 5 % CO2.

Microbe strains

All microbe strains were described in the key resources table. Referring to our previous method,62 Lactobacillus johnsonii was isolated independently and verified using 16S rRNA sequencing (V4 sequences). The complete genome sequence was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The bacteria were cultured in MRS medium (hopebio, HB0384-5) at 37 °C for 24 h. To get high-temperature inactivated Lactobacillus johnsonii, the bacteria were heated in the 100 °C metal bath for 2 h. To get ultrasonic broken Lactobacillus johnsonii, the bacteria were lysed using the Fisher Scientific™ Model 50 Sonic Dismembrator. The conditions of ultrasonic fragmentation were 300 w, 20 s start-up and 10 s pause, 20 min. As a representative bacterium for IPA production, Clostridium sporogenes was purchased from ATCC (ATCC 11437) and cultured in RCM (hopebio, HB0316) under an atmosphere of 10 % CO2, 10 % H2 and 80 % N2 for 72 h at 37 °C.

Human participants

Tissue samples from 92 patients with CRC (Cohort 1), fecal samples from 67 healthy control, 40 patients with colon adenoma, 108 patients with CRC (Cohort 2) were obtained. After surgical resection, CRC tissue samples and their adjacent normal mucosa were immediately frozen in liquid nitrogen and stored at -80 °C. Fresh CRC tissues from 10 patients using for ALI-PDOs or flow cytometry were obtained. The patients were without antibiotics and probiotics in the past one month. Informed consent was obtained from all participants, and the experimental protocol was approved by the Clinical Research Ethics Committee of Sir Run Run Shaw Hospital, Zhejiang University School of Medicine (20211103-35).

Animals

All experiments used mice were described in the key resources table. Unless otherwise stated, 6-8 weeks males or females were used in all experiments; no significant sex-dependent differences were found in the experiments reported. They were under specific pathogen-free (SPF) environment, water and food were provided adequately. The temperature was constant, and a 12 h circadian rhythm was maintained every day. All animals were done under the guidelines of the Animal Experimentation Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine (AIRB-2022-0336), Zhejiang Chinese Medical University (20221212-07, 20230918-14) and IACUC of GemPharmatech (GPTAP20231010-3). For the germ-free mice model, feces were microbiologically tested weekly to confirm‎ sterility or specific microbial colonization status.

Method detailsTumor models and treatments

Antibiotics (Abx) treatment was performed to normalize the gut microbiota in animal experiments. Mice were given 0.2 g/L Ampicillin (meilunbio, MB1507), 0.2 g/L Metronidazole (meilunbio, MB2200), 0.2 g/L Neomycin (meilunbio, MB1716) and 0.1 g/L Vancomycin (meilunbio, MB11260) quadruple antibiotics for one week.63 For transplantable tumor models, 106 Mc38, 105 B16-F10, 105 4T1 and Matrigel (Yeasen Biotechnology, 40183ES10) were co-injected into the loose subcutaneous tissues of the mice back, 100 μl of cell suspension each mouse. For orthotopic tumor models, 4T1 or Mc38 were implanted into the mammary fat pad or the cecum respectively. On the 8th, 11th, and 14th days, each mouse was intraperitoneally injected with 100 μg of IgG isotype control antibody (BioXcell, BE0089) or αPD-1 antibody (BioXcell, BE0273) for immunotherapy. Transgenic MMTV-PyMT mice (Cyagen Biosciences Inc., C001212) at 16 weeks were used as an autochthonous breast cancer model for survival experiments. Each mouse was given 100 μg αPD-1 antibody for immunotherapy every three days and removed from the study as soon as an individual tumor reached a volume of 1000 mm3.44 For CD8+ T cells neutralization experiments, 200 μg of αCD8 antibody (BioXcell, BE0004-1) were delivered by intraperitoneal injection, once every three days. For bacterial treatment, each mouse was given 200 μl of PBS per day by gavage containing 109 CFU of bacteria. For metabolites treatment, each mouse was gavaged at a dose of 60 mg/kg (or at concentration gradients as indicated) per day.28 In some experiments, ILA (Bide Pharmatech Ltd., BD13033) and IPA (Sigma, V900491) were administered by intraperitoneal or intratumoral injection.

Fecal microbiota transplantation

Referring to published studies,64 fresh feces from Responder Donors (n=5) and Poor-responder Donors (n=5) mice were collected and resuspended in sterile PBS solution at a ratio of 40 mg/ml. Subsequently, grinding beads were added to the mixture and filtered through a 70 μm cell strainer (Biosharp, BS-70-XBS). 10 % volume of glycerol was added to the fecal suspension, and it was sub packaged and frozen at -80°C. For each Abx-treated mouse, 200 μl of fecal suspension was orally administered three times each week.

Adoptive cell transfer therapy

Referring to previous work,65,66 mice were sacrificed to separate tumor-draining lymph nodes or spleen, and cell clumps were removed by grinding and filtering. After PBS washing and cells counting, cells were resuspended in culture medium. CD8+ T cells or B cells were collected by magnetic cell separation system as the manufacturer’s protocol (BioLegend, 480019, 480035, 480051). The sorted cells were counted. 106 CD8+ T cells, 106 B cells, or 106 CD8+ T cells and 106 B cells were transferred into per Rag1-/- mice by tail vein injection.

Heterologous expression‎ of ldhA

Standard molecular cloning techniques were used to clone the DNA fragments encoding full-length ldhA into the pRSFDuet vector. After overexpression‎ of ldhA in E. coli BL21 (Yeasen Biotechnology, 11804ES80), the bacteria were grown at 37 °C to an OD600 of 0.6. Afterwards, ldhA protein expression‎ was induced with 200 μΜ IPTG (Selleck, S6826) for 6 h at 37 °C. E. coli and ldhA-overexpressing E. coli (E. c-ldhA) were incubated in LB medium (Sangon Biotech, A507002) for 24 h at 37 °C.

Ex vivo CD8+ T cells differentiation

CD8+ T cells were purified from lymph nodes or spleen of mice using MojoSort™ Mouse CD8 T cell Isolation Kit (BioLegend, 480035). Purified CD8+ T cells were activated with αCD3 (5 μg/ml, BioLegend, 100239), αCD28 (5 μg/ml, BioLegend, 102115) and 50 U/ml Il-2 (BioLegend, 575404). In specific experiments, CD8+ T cells were treated with 0, 5 μM or 500 μM IPA for 48 h. For flow cytometry, CD8+ T cells were stained with anti-CD8, anti-PD-1 and anti-TCF-1 antibodies (BD Biosciences, 561967, 566692, BioLegend, 135213). For western blot, total protein from CD8+ T cells was extracted and blocked with anti-TCF-1, anti-H2AK5ac, anti-H2BK5ac, anti-H3K27ac, anti-H4K20me1 antibodies (Cell Signaling Technology, 2203S, ABclonal, A15620, A15621, A22264, A2370) at 4 °C overnight. Histone H2A, H2B, H3 and H4 (ABclonal, A3692, A19812, A2348, A1131) served as the loading controls.

ChIP-qPCR assay and analysis

ChIP was conducted following the manufacturer’s protocol (Thermo Scientific™, 26156). Briefly, PBS or 500 μM IPA-treated CD8+ T cells were cross-linked with 1 % formaldehyde at room temperature for 10 min, which was terminated by Glycine Solution. After centrifugation, the acquired pellet was lysed by the indicated lysis buffer with a protease inhibitor cocktail. Then the extracted genomic DNA was digested enzymatically to achieve DNA fragmentation using micrococcal nuclease. Transfer 5 μl of the supernatant containing the digested chromatin to a 1.5 ml tube and store at -20 °C as the 10 % total input sample from one ChIP. Then the samples were subjected to immunoprecipitation at 4 °C overnight with anti-H3K27ac (ABclonal, A22264). Add 20 μl ChIP Grade Protein A/G Plus Agarose to each IP and incubate for 1 h at 4 °C on a rocking platform. After resin incubation, the precipitated protein-DNA complexes were eluted, reversal of crosslinking, DNA clean-up and subjected to qPCR.

CUT & RUN assay and analysis

CUT & RUN assay was conducted following the manufacturer’s protocol (Vazyme, HD101). Briefly, PBS or 500 μM IPA-treated CD8+ T cells were incubated with ConA Beads Pro at room temperature for 10 min, anti-H3K27ac antibody (1:50, ABclonal, A22264) was added and rotated at room temperature for 2 h, then washed twice, added pG-MNase Enzyme and incubated at 4 °C for 1 h, then washed twice, Cacl2 was added and incubated for 1 h on ice, added stop buffer and incubated at 37 °C for 30 min. DNA was extracted and qPCR was used to detect the acetylation of Tcf7 SE. Spike in DNA derived from the λDNA of E. coli was used for uniform correction.

CUT & Tag assay and analysis

PBS or 500 μM IPA-treated CD8+ T cells were collected with six replicates per group for CUT & Tag assay (Vazyme, TD904). The sequencing process was commissioned to chi-biomedicine (Guangdong, China). Briefly, CD8+ T cells were resuspended with wash buffer, incubated with bead, incubated with primary anti-H3K27ac antibody (1:50, ABclonal, A22264) and incubated with secondary antibodies. Then the samples were incubated with Hyperactive pA/G-Transposon Pro, fragmented with Mgcl2. The DNA was extracted and PCR amplification was performed using indexing primers (Vazyme, TD202). CUT & Tag library was purified and assessed on the Agilent Bioanalyzer 2100 system. The clustering of the index-coded samples was performed on a cBot Cluster Generation System and the library preparations were sequenced on the Illumina Novaseq platform with 150 bp paired-end reads. Sequencing data was further processed by bioinformatics pipelines including raw data cleaning (removing containing, low-quality reads), reference mm10 mouse genome mapping (MAPQ >=13), peak calling (with the q-value threshold of 0.05), peak annotation (ChIPseeker) and different peak analysis (fold change of RPM >= 2). Bam files were visualized using IGV.

Air-liquid interface PDOs

Referring to previous work,47,48 air-liquid interface patient-derived organoids (ALI-PDOs) were established. Briefly, fresh human CRC tissues were clipped in pre-chilled PBS and washed in medium containing antibiotics. Collagen cocktail solution was constructed by mixing Solution A (Nitta Gelatin), Solution B (Sigma-Aldrich, D8900) and Solution C (Sterile reconstitution buffer containing 2.2 g NaHCO3 in 100 ml of 0.05 N NaOH and 200 mM HEPES) at a ratio of 8:1:1 on ice, and 300 μl of collagen cocktail solution was added to the Millicell dish (Millipore, PICM01250) to form the bottom gel layer, which was then solidified at 37 °C for 30 min. The excised tumor tissues were then resuspended with a further 300 μl of collagen cocktail solution and added to the top of the pre-solidified gel layer. 12-well plates were filled with outer wells supplemented with organoid medium (Shanghai Bioheb Biomed Technology Co., Ltd., I-ALI-OC-Medium-20221011). On this basis, additional 50 U/ml human recombinant IL-2 protein (Abclonal, RP01039) and 5 μg/ml human anti-CD3/CD28 antibody (BioLegend, 317326, 302934) were added for the transient culture of T cells. On top of the above medium, 10 μg/ml αPD-1 antibody (nivolumab, Bristol Myers Squibb) was added at the same time. 5 mM IPA or equal amounts of DMSO (Mock) were added every 2 days. The external medium was changed every 2 days. For immunofluorescence, after about 7 days of culture, the up-layer gel was removed with forceps and paraffin-embedded tissue were stained with anti-CD8 antibody (1:4000, Hubei BIOSSCI Biotech Co., LTD, za-0508), anti-TCF-1 antibody (1:50, Cell Signaling Technology, 2203S).

Flow cytometry analysis

At the end of modeling, the subcutaneous tumors were collected and quickly immersed in cold medium. After cutting them into 1 mm pieces, collagenase IV (Worthington, LS004189) was added for 30 min to further digest. The tumor-draining lymph node and spleen were gently grounded and filtered through a 70 μm cell strainer (biosharp, BS-70-XBS). Splenocytes were resuspended in RBC Lysis buffer and incubated for 15 minutes at 4 °C. Then, 1 × 106 cells were incubated with 1 μg of anti-CD16/32 antibody (BioLegend, 101320) for 10 min to block non-specific binding of immunoglobulin to the Fc receptors. Then, cells were stained for live cells (BD Biosciences, 564406) for 30 minutes. Afterwards, the appropriate amount of pre-diluted fluorescent labeled antibody (BioLegend, 103128, 108745, 103236, 127648, 117318, 123132, 101257, 135213, BD Biosciences, 551163, 563151, 561967, 557000) was added to each tube as recommended by the manufacturer. Cells were incubated in the 4 °C refrigerator for 30 min and fixed with Fixation Buffer (BioLegend, 422101) for 30 min at room temperature. After washing, the membranes were ruptured (BioLegend, 421002) and intracellular fluorescent antibodies were added to each tube and incubated for 30 min at room temperature. The effectors IFN-γ and TNF-α (BioLegend, 505830, 506346) were detected by adding 2 μl Activation Cocktails (BioLegend, 423303) to 1 ml of cell suspension and incubating for 6 h at 37 °C. For TCF-1 staining, cells were incubated with Fixation work solution for 40 minutes at room temperature, washed with Transcription Factor Staining Buffer (Invitrogen™, 00-5523-00), and stained with antibody (BD Biosciences, 566692) for 40 minutes at room temperature.

Metabolomic analysis

For plasma samples, 5-10 times the volume of pre-cooled methanol was added to plasma, vortexed, and shaken, and incubated for 1 h at -20 °C. After centrifuging samples for 10 min at 14,000 g, the supernatant was collected and centrifuged again repeatedly. The supernatant was filtered and tested.42 For culture supernatant samples, after centrifuging for 10 min at 14,000 g to remove the organisms and impurities, the supernatant was acidified to pH=2.5 using hydrochloric acid and extracted twice with a double volume of ethyl acetate. Then, air-dried and redissolved in one-tenth volume of methanol. The supernatant was filtered and assayed.

For Untargeted LC-MS/MS, prepared plasma samples as described above. The supernatant was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Briefly, metabolites were detected on the Thermo UHPLC-Q Exactive system. The analytical conditions follow: buffer A was 95 % ddH2O plus 5 % acetonitrile (containing 0.1 % formic acid) and buffer B was 47.5 % acetonitrile with 47.5 % isopropanol and 5 % ddH2O (containing 0.1 % formic acid). The gradient conditions were as follows: Positive ion mode: 0 to 3 min, gradient to 20 % buffer B; 3 to 4.5 min, gradient to 35 % buffer B; 4.5 to 5 min, gradient to 100 % buffer B; 5 to 6.3 min, 100 % buffer B; 6.3 to 6.4 min, gradient to 0 % buffer B; 6.4 to 8 min, 0 % buffer B. Negative ion mode: 0 to 1.5 min, gradient to 5 % buffer B; 1.5 to 2 min, gradient to 10 % buffer B; 2 to 4.5 min, gradient to 30 % buffer B; 4.5 to 5 min, gradient to 100 % buffer B; 5 to 6.3 min, 100 % buffer B; 6.3 to 6.4 min, gradient to 0 % buffer B; 6.4 to 8 min, 0 % buffer B. The flow rate was 0.40 ml/min. HMDB (http://www.hmdb.ca/) and Metlin (https://metlin.scripps.edu/) were used to match the mass spectrometry data. On the basis of the KEGG database, metabolic enrichment and pathway analysis were performed on the differential metabolites.

IPYA, ILA, IA and IPA were detected as described above. Metabolites were detected on the 5500 QTRAP triple quadrupole mass spectrometer (SCIEX, Framingham, MA, USA). The analytical conditions follow: buffer A was ddH2O plus 0.1 % formic acid and buffer B was acetonitrile plus 0.1 % formic acid. The gradient conditions were as follows: 0 to 1 min, 5 % buffer B; 1 to 7 min, gradient to 95 % buffer B; 7 to 10 min, 95% buffer B; 10 to 13 min, gradient to 5 % buffer B; 13 to 14 min, 5 % buffer B. The flow rate was 0.40 ml/min.

16S rRNA sequencing

Feces from both ‘Responder’ (n=5) and ‘Poor-responder’ (n=5) groups of mice were collected. Genomic DNA from fecal samples was extracted and tested for concentration and purity using electrophoresis and NanoDrop 2000. The subsequent library construction and sequencing process was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Briefly, full-length PCR amplification of the 16S rRNA gene was performed using 27F (5’-AGRGTTYGATYMTGGCTCAG-3′) and 1492R (5′-RGYTACCTTGTTACGACTT-3′) primers, followed by library construction and sequencing. UPARSE 7.1 was used to cluster the sequencing results into operational taxonomic units (OTUs), with 97 % sequence similarity and chimeras were removed. Taxonomic annotation of OTUs using the RDP classifier ratio against the Silva 16S rRNA gene database with a 70 % confidence threshold. The similarity of microbial community structure between samples was examined using PCoA based on the Bray-Curtis distance algorithm.

Shotgun metagenomic sequencing

Total genomic DNA was extracted from both ‘Responder’ (n=5) and ‘Poor-responder’ (n=5) groups of feces with the PF Mag-Bind Stool DNA Kit according to the manufacturer’s instructions. The subsequent library construction and sequencing process was commissioned to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Briefly, DNA extract was fragmented to an average size of about 400 bp for paired-end library construction. The paired-end library was constructed using NEXTFLE Rapid DNA-Seq. Paired-end sequencing was performed on Illumina Novaseq 6000 (Illumina Inc., San Diego, CA, USA) according to the manufacturer’s instructions. Metagenome sequencing data was further processed by bioinformatics pipelines on the online platform of Majorbio Cloud Platform (www.majorbio.com) with data quality-filtering (length<50 bp or with a quality value <20 or having N bases), assembly, genomic contamination elimination and taxonomic compositions identification.

Single-cell RNA and TCR sequencing

To perform single-cell RNA sequencing analysis, we treated mice under each treatment condition (WT (αPD-1) and IPA-treated (IPA+αPD-1), n=2 biological replicates/group). Flow sorting was used to obtain tumor-infiltrating CD8+ T cells (BD Biosciences, 561869, BioLegend, 420404, 100203, 100714) and cell suspensions were dissociated and loaded into the 10x Chromium instrument. After library construction, the NovaSeq 6000 sequencing platform was used for RNA sequencing and a sequencing depth of 20,000 reads per cell was required. Illumina bcl2fastq software was used to demultiplex and convert the sequencing data into FASTQ format files. TCR library construction and TCR V(D)J targeted enrichment were performed with the Chromium Single Cell V(D)J Enrichment Kit (10x Genomics) according to the manufacturer’s user guide. For dimension reduction, clustering, and analysis of the scRNA-seq data, the Cell Ranger output was loaded into Seurat.57 All cells were filtered by quality control conditions (Expression‎ of all genes was detected in at least 3 cells, with the number of genes expressed in a single cell between 500 and 5000, the number of UMIs greater than or equal to 500, and the ratio of mitochondrial gene expression‎ less than 25 %). The filtered cells were analyzed by UMAP downscaling using R software to obtain visualization results. Using the scRepertoire (v.2.0.0) package,58 sample-specific consensus annotation files were consolidated into a list of TCR sequencing results and then integrated with the Seurat object for visualization. Enrichment analysis was performed by Vision package (v2.1.0).30 Monocle 3 was applied to perform trajectory analysis.59 The other Single Cell Data Analysis was performed using the OmicStudio tools created by LC-BIO Co., Ltd (Hangzhou, China) at https://www.omicstudio.cn/cell.

Single-cell ATAC sequencing

To perform single-cell ATAC-seq, we treated mice under each treatment condition (WT (αPD-1) and IPA-treated (IPA+αPD-1), n=2 biological replicates/group). Flow sorting was used to obtain tumor-infiltrating CD8+ T cells (BD Biosciences, 561869, BioLegend, 420404, 100203, 100714) and cell suspensions were dissociated and loaded into the 10x Chromium instrument. Single-cell ATAC-seq libraries were prepared according to the Chromium Single Cell ATAC Library Kit from 10x Genomics following the manufacturer’s instructions. 10x Genomics Cell Ranger ATAC pipeline (version 2.1.0) was used for scATAC-seq analyses of alignment, deduplication, and identification of transposase cut sites (https://support.10xgenomics.com/). The trimmed read-pairs are aligned to a specified reference using BWA-MEM with default Parameters.60 Fragments were identified as read pairs with mapping quality (MAPQ)>30. Reads were counted across the genome, using 500-bp bins (tiles) to generate a genome-wide tile-count-matrix. Latent semantic indexing (LSI), Louvain clustering and Harmorny algorithm were applied. Gene activity scores were computed as the summed local accessibility of promoter-associated count-tiles in the proximity of each gene, using a distance-weighted accessibility model. Finally, the above-weighted sum was multiplied by the aggregated Tn5 insertions in each tile. Gene scores were then scaled to 10,000 counts and log2-normalized. To enhance the visual interpretation of gene activity scores, smoothing was applied using the MAGIC algorithm. Downstream bioinformatics pipelines analysis was conducted with R (R version 4.1.2) package ArchR (version 1.0.2) and Seurat (version 4.1.1). ScATAC-seq cluster was identified by gene activity scores and scRNA-seq gene expression‎. MACS2 was applied to identify a robust merged peak.61 Then peaks were annotated according to their respective genomic position (promoter, intronic, exonic, distal etc). Differential accessibility analysis between cells was performed to identify celltype-specific and condition-specific marker peaks (|Log2FC| > 0.5 and p-value < 0.01).

Quantitative real-time PCR

Bacterial DNA was extracted from mouse and human feces using the TIANGEN Fecal Kit (TIANGEN, DP328-02) and bacterial genomic DNA from human tumor and adjacent normal mucosa using the TIANGEN DNA Kit (TIANGEN, DP304-02). RNA from ALI-PDOs was extracted using the SteadyPure RNA extraction kit (Accurate Biology, AG21017) and reverse transcribed using ABScript III RT Master Mix (ABclonal, RK20429). qPCR was performed in Light Cycler® 480 real-time PCR system (Roche) using SYBR Green Fast qPCR Mix (ABclonal, RK21203). cDNA was amplified by PCR under the following conditions: 95 °C for 3 min, followed by 45 cycles of 95 °C for 5 s and 60 °C for 30 s. The specific Primer sequences were listed in the key resources table. The universal Eubacteria 16S or β-ACTIN was used as internal reference genes.

Quantification and statistical analysis

GraphPad Prism was used for all statistical analyses. The experiments were designed to use a minimum of 3 samples/replicates per experiment or per group. Representative immunofluorescent staining, LC-MS/MS and flow cytometry images are presented. Each experiment was repeated in triplicate independently. The data are expressed as the mean ± standard deviation (SD) or mean ± standard error of mean (SEM). Differences between groups were analyzed by two-tail ratio paired t test, unpaired t test, Wilcoxon rank-sum test, Mann Whitney test, one-way ANOVA with Sidak’s correction for multiple comparisons, two-way ANOVA with Sidak’s correction for multiple comparisons, Spearman correlation analysis and log-rank test. Statistically significant were P values < 0.05.

Acknowledgments

The authors would like to express their gratitude to all colleagues who contributed to this work, in particular Prof. Jianmin Si, Prof. Di Wang, Prof. Yongqun Zhu, Prof. Lie Wang, Dr. Dong Cen (Zhejiang University), and Prof. Zheng Kuang (Carnegie Mellon University). Thanks to Yanwei Li (the Core Facilities, Zhejiang University School of Medicine) and Zhanglian He (Biomedical Research Center, Sir Run Run Shaw Hospital, Zhejiang University) for technical support in flow cytometry. Thanks to Zhongjing Zhou and Yi Teng (Zhejiang Academy of Agricultural Sciences) for technical support in LC-MS/MS analysis. This project was financially supported by the National Foundation of Natural Science of China (82273269, 82072623 to Liangjing Wang; 82270573 to S.C.), the Key program of Natural Science Foundation of Zhejiang Province (LZ22H160002 to Liangjing Wang), and the National Key Research and Development Program of China (2022YFC2505100 to Liangjing Wang).

Author contributions

Conceptualization, Liangjing Wang and S.C.; methodology, D.J., Q.W., Y.Q., Y.J., and J.H.; software, Y.Q. and Q.W.; formal analysis, D.J., Q.W., Y.Q., Y.J., and J.H.; investigation, Y.L., Y.S., J.X., W.C., L.F., R.Y., C.X., and Q.G.; resources, W.Z., G.R., Lan Wang, W.L., F.X., and P.W.; writing – original draft, D.J.; writing – review & editing, Liangjing Wang, S.C., and Y.W.; funding acquisition, Liangjing Wang and S.C.; supervision, Liangjing Wang.

Declaration of interests

The authors declare no competing interests.

References

1.

Fernandes, M.R. ∙ Aggarwal, P. ∙ Costa, R.G.F. ...

Targeting the gut microbiota for cancer therapy

Nat. Rev. Cancer. 2022; 22:703-722

Crossref

Scopus (85)

PubMed

Google Scholar

2.

Jiang, Y. ∙ Jia, D. ∙ Sun, Y. ...

Microbiota: A key factor affecting and regulating the efficacy of immunotherapy

Clin. Transl. Med. 2023; 13, e1508

Crossref

Google Scholar

3.

Wang, F. ∙ He, M.M. ∙ Yao, Y.C. ...

Regorafenib plus toripalimab in patients with metastatic colorectal cancer: a phase Ib/II clinical trial and gut microbiome analysis

Cell Rep. Med. 2021; 2, 100383

Google Scholar

4.

Davar, D. ∙ Dzutsev, A.K. ∙ McCulloch, J.A. ...

Fecal microbiota transplant overcomes resistance to anti-PD-1 therapy in melanoma patients

Science. 2021; 371:595-602

Crossref

Scopus (807)

PubMed

Google Scholar

5.

Si, W. ∙ Liang, H. ∙ Bugno, J. ...

Lactobacillus rhamnosus GG induces cGAS/STING- dependent type I interferon and improves response to immune checkpoint blockade

Gut. 2022; 71:521-533

Crossref

Scopus (134)

PubMed