L. paracasei abundance was decreased in tumor tissue samples of GC patients
Our previous study based on 16s rRNA gene sequencing showed that the abundance of Lactobacillus in gastric cancer tumor tissues is significantly lower than that in normal tissues[8]. Here we validated the enrichment of Lactobacillus in normal gastric tissues with qPCR in paired tumor and adjacent normal tissues of 90 patients (Figure S1A). Then qPCR detection of major Lactobacillus species was also conducted, which revealed that L. paracasei, but not L. johnsonii or L. reuteri, had significantly higher relative abundance of this species in normal tissues than in tumor tissues (Fig. 1A,1B, S1B & S1C). Interestingly, L. paracasei is a known probiotics with reported antitumor capability[13][14]. The colonization of this species in the stomach was confirmed by fluorescence in situ hybridization (FISH) and PCR detection of bacterial DNA (Fig. 1C and Figure S1D). These findings suggest the potential protective role of L.paracasei in GC progression.
A novel Lactobacillus paracasei ZJUZ2-3 was isolated from normal gastric tissue
We isolated a strain of L. paracasei (designated ZJUZ2-3) from the normal tissue of a gastric cancer patient (Fig. 1D), and performed whole genome sequencing of the isolate. The complete genome of strain ZJUZ2-3 was determined and found to contain a circular chromosome with a size of 3080235 bp and a GC content of 46%. It contains 2850 protein coding genes, 77 non-coding genes. Genomic testing revealed that the strain contains two plasmids with sizes of 9535 bp and 10797 bp, respectively (Fig. 1E). After software analysis, the whole genome ANIb value of the relative Lactobacillus paracasei strain (NCBI genome registration number: GCA_029625355.1) was 99.6656, and the whole genome ANIb value of the relative Lactobacillus paracasei strain (NCBI genome registration number: GCA_030480425.1) was 99.6618. ANIb, Average nucleotide identity using BLAST, also known as average nucleotide identity using BLAST, is an indicator for evaluating the phylogenetic relationship of strains. It is generally believed that ≥ 95 indicates belonging to the same species. The genome of the above-mentioned strain ZJUZ2-3 is not completely consistent with the genome of Lactobacillus paracasei in the NCBI database. The species evolution tree showed that ZJUZ2-3 was a subspecies of Lactobacillus paracasei (Fig. 1F). Notably, it possessed a tryptophan metabolism pathway. We found the enzyme tryptophan decarboxylase (TrpD) which can decompose tryptophan into indole derivatives (Fig. 1G).
Figure1L. paracasei ZJUZ2-3 was isolated from a GC patient normal tissue
-
The abundance of Lactobacillus in patients with GC (N = 49)
-
The relative abundance of L. paracasei in tumor tissues and normal tissues of GC patients. *p < 0.05, ** p < 0.01, ***p < 0.001, **** p < 0.0001.
-
Representative FISH images of gastric tissue sections from GC patients (blue: nuclear, green: L.paracasei probe); scale bars, 100 µm.
-
Isolation of a novel L.paracasei strain from a GC patient’ normal tissue.
-
The whole genome map of L.paracasei ZJUZ2-3. The circle chart displays GC content, sequencing depth, gene element display, and COG function display from the outside to the inside.
-
Phylogenetic tree of L.paracasei ZJUZ2-3. A phylogenetic tree based on core single copy genes constructed using Neighbour joining clustering.
-
Functional annotation of L.paracasei ZJUZ2-3 genome using KEGG metabolic pathway.
L. paracasei ZJUZ2-3 and its conditioned medium inhibited the viability of gastric cancer cells
We next determined whether L. paracasei ZJUZ2-3 had anti-tumor effect in vitro. Gastric cancer cell lines AGS and HGC27 were cocultured with L. paracasei ZJUZ2-3 (1×107 CFU/ml) for 4 hours, while PBS was used as blank control. L. paracasei coculture significantly decreased GC cell viability (Fig. 2A). This growth-inhibitory effect was further confirmed by colony formation assay (Fig. 2B). Live L. paracasei ZJUZ2-3 was heat-killed by autoclaving and subsequently exposed to gastric cancer cells. The E. coli strain DH5ɑ was used as bacteria control. In contrast, neither heat-killed L. paracasei ZJUZ2-3 nor E. coli DH5ɑ induced the reduction of gastric cancer cell viability (Fig. 2C and D).
We next explored whether such function was attributed to L. paracasei ZJUZ2-3 itself or its metabolites. L. paracasei culture was centrifuged and the supernatant was then filtered through a 0.22 µm membrane to obtain a conditional culture medium (CM), which was used to treat AGS and HGC27 cells at the concentration of 1% (v/v) for 4 days. PBS and LBM were used as negative controls. Interestingly, the CM significantly inhibited AGS and HGC27 growth (Figure S2E and 2F). To further distinguish the type of molecules responsible for such effect, the L. paracasei conditional medium (Lp.CM) was pretreated with or without protease K (PK, 1 mg/mL). The inhibitory effect remained unchanged after PK treatment, indicating that the antitumor effect was dependent on non-protein metabolites in the Lp.CM.
Figure2 Living L. paracasei ZJUZ2-3 and its metabolites could inhibit GC cell growth.
(A)L.paracasei ZJUZ2-3 (MOI = 100) inhibits the cell viability of gastric cancer cells AGS and HGC27 as determined by CCK8 assays. (Lp: L. paracasei ZJUZ2-3)
(B)Colony formation of GC cells under the treatment of L.paracasei ZJUZ2-3.
(C, D) Cell viability (C) and colony formation (D)of GC cells under the treatment of L. paracasei ZJUZ2-3, heat-killed L. paracasei ZJUZ2-3 and E.coli DH5ɑ
(E, F) (E)Cell viability at OD450nm and (F)colony formation of human GC cells (AGS and HGC27) treated with PBS, Lp conditional medium (Lp.CM), Lp conditional medium with digestion by protease K (PK-Lp.CM) and lactobcillus broth medium (LBM).
L. paracasei ZJUZ2-3 exerted antitumor effects in GC xenograft mice
Mouse experiments were further conducted to assess the potential protective effect of L. paracasei ZJUZ2-3 against GC (Fig. 3A). Four-week-old nude mice were subcutaneously injected with HGC27 cells (3×106 cells). After seven days, L. paracasei ZJUZ2-3 (1×108 CFU/50 µl) or PBS with an equivalent volume (blank control) was intratumorally injected once in two days (Fig. 3B). L. paracasei ZJUZ2-3-treated mice had an obvious reduction in tumor size and weight compared to those injected with PBS (Fig. 3B, E and F). HE results indicated L. paracasei ZJUZ2-3 treated mice had a lower percentage of low-grade dysplasia as compared with control (Fig. 3G). L. paracasei ZJUZ2-3 treatment markedly reduced the number of Ki-67-positive cells in tumor tissues (Fig. 3H). Successful intratumor colonization of L. paracasei ZJUZ2-3 was confirmed by FISH (Fig. 3C) and PCR with customized primers (Supplementary Fig. 1E). In addition, L. paracasei ZJUZ2-3 injection did not alter mouse body weight (Fig. 3D).
We then assessed the efficacy of L. paracasei ZJUZ2-3 in an orthotopic mouse model of GC (Fig. 3I). Nude mice bearing orthotopic HGC27 tumors were oral gavaged with PBS or L. paracasei ZJUZ2-3 at a dose of 1×108CFU/100µl, for 10days. As shown in Fig. 3J, mice gavaged with L. paracasei ZJUZ2-3 significantly had a decreased growth of orthotopic tumor. HE and Ki-67 staining results confirmed L. paracasei ZJUZ2-3 inhibited the orthotopic tumor growth (Fig. 3K). Those results confirmed the protective effect of L. paracasei ZJUZ2-3 against gastric tumorigenesis.
Figure3L. paracasei ZJUZ2-3 inhibited tumor growth in the GC xenograft mice.
-
Schematic diagram shows the experimental design, time line and treatment of the GC xenograft mice (Lp: L. paracasei ZJUZ2-3).
-
The representative tumor morphologies.
-
L. paracasei ZJUZ2-3 colonization was confirmed by FISH.
(D-F) (D)The mouse body weight curve; (E)Tumor growth curve, and (F) tumor weight compared between L. paracasei ZJUZ2-3- or PBS-treated groups. Data are shown as mean ± SEM. ****p < 0.0001.
(G)Representative H&E image of tumor tissues from mice administrated with PBS, IAA and L. paracasei ZJUZ2-3. scale bars, 100 µm.
(H) Ki-67 staining showed decreased numbers of Ki-67+ cells in the tumor tissues of IAA and L. paracasei ZJUZ2-3 treated groups.scale bars, 100 µm
(I) Schematic diagram shows the experimental design, time line and treatment of orthotopic mouse model of GC.
(J) The representative stomach of mice.
(K) Representative H&E and Ki-67 staining images of stomach tumor tissues from mice oral gavaged with PBS and L. paracasei ZJUZ2-3. scale bars, 100 µm.
Protective effect of L.paracasei ZJUZ2-3 is associated with the production of indole-3-acetic acid (IAA)
L. paracasei has been shown to release metabolites, including the indole-derivative indole-3-acetic acid (IAA) to interact with the host [15]. Based on our previous untargeted metabolome analysis, we found that discriminative metabolites in tumor tissue and normal tissues of gastric cancer patients were enriched in the tryptophan pathway (Fig. 4A)[8]. The tryptophan metabolite IAA was the most significantly enriched in normal tissues (Fig. 4B). Correlation analysis revealed that Lactobacillus abundance was positively correlated with IAA, suggesting that IAA might play a role in the inhibitory tumor progression of L.paracasei ZJUZ2-3 (Fig. 4C). We then treated GC cells with L. paracasei ZJUZ2-3 and found IAA increased in the supernatant compared with PBS (Fig. 4D). Taken together, these data demonstrate that the antitumor effect of L. paracasei ZJUZ2-3 was likely mediated by IAA.
IAA inhibit GC development in vitro and in vivo
Upon establishing the involvement of IAA in L. paracasei-mediated tumor suppression, we assessed whether IAA alone is sufficient to induce an antitumor response. Remarkably, IAA significantly inhibited the viability of AGS (p < 0.01) and HGC27 (p < 0.01) cells in a concentration-dependent manner (Fig. 4E). IAA treatment also markedly restrained the colony formation of GC cells, compared with control (p < 0.01; Fig. 4F). Subsequently, we verified the impact of IAA on the nude mouse xenograft model (Fig. 3A). Consistently, IAA demonstrated significant inhibitory effects on mouse tumors. Compared to the PBS control, the IAA intratumoral injection group showed a significant reduction in tumor growth (Fig. 3B). HE and Ki-67 staining results indicated that IAA could suppress tumor growth in the GC xenograft model (Fig. 3G and 3H). The tumor size (p<0.001) and weight (p<0.001) in the IAA treatment group were significantly reduced compared to the control group (Fig. 3D and 3F). We then assessed the efficacy of IAA and L. paracasei ZJUZ2-3 in an orthotopic mouse model of GC (Fig. 4G). Nude mice bearing orthotopic HGC27 tumors were oral gavaged with PBS, IAA or L. paracasei ZJUZ2-3 at a dose of 100µM/100µl and 1×108CFU/100µl, for 10days. As shown in Fig. 4H, mice gavaged with IAA and L. paracasei ZJUZ2-3 significantly had a decreased growth of orthotopic tumor. HE and Ki-67 staining results confirmed IAA and L. paracasei ZJUZ2-3 inhibited the orthotopic tumor growth (Fig. 4I). These results confirmed that the GC-suppressive effect of L. paracasei ZJUZ2-3 was mediated by the tryptophan metabolite IAA.
Figure4 IAA inhibited tumor cell growth in vitro and in vivo.
-
KEGG pathway enrichment analysis of DEGs from RNA sequencing dataset of GC patients. Size of circles represents the ratio of DEGs in pathways. Red: upregulated, blue: downregulated.
-
The abundance of IAA in GC patients.
-
Correlation analysis between bacteria and indole derivatives.
-
The concentration of IAA in GC cells supernatant co-cultured with L. paracasei ZJUZ2-3 were detected by ELISA.
-
Cell viability at OD450nm of human GC cells (AGS, HGC27) treated with PBS or IAA.
-
Colony formation of GC cells under treatment of PBS or IAA (50 µM).
-
Schematic diagram shows the experimental design, time line and treatment of the GC xenograft mice.
-
The representative stomach morphologies of mouse.
-
Representative HE and Ki-67 staining images of tumor tissues from mice administrated with PBS, L. paracasei ZJUZ2-3 and IAA.
IAA mediates antitumor effects via activation of the aryl hydrocarbon receptor (AHR)
To investigate the mechanism triggered by IAA for GC inhibition, RNA sequencing was performed to profile gene expression alterations in GC cell treated with IAA, by which 100 deregulated genes were identified. Of note, the aryl hydrocarbon receptor (AHR) was significantly upregulated (Fig. 5A). AHR, a DNA binding protein existing in the cytoplasm as an inactive state AHR signal, is highly expressed in GC cells, so AHR is effective only activated by its ligand[16]. We further confirmed IAA treatment increased the protein levels and the activity of AHR in both GC cell lines (Fig. 5B, D and F). Accordingly, treatment with L. paracasei ZJUZ2-3 had the similar effect (Fig. 5C and E).
To investigate the direct link between the AHR pathway and the tumor-suppressive effect of L. paracasei ZJUZ2-3 and IAA, we used an AHR antagonist, CH223191, to block AHR activity. We observed that CH223191 treatment inhibit the antitumor effect of IAA via the CCK8 assay and colony formation test (Fig. 5G and H). Consistently, intratumoral injection of CH223191 abolished IAA-induced tumor suppression in vivo (Fig. 6A, B, D and E), these treatments did not impact the body weight of mouse (Figure S1F), and the ratio of Ki-67-positive cells was restored (Fig. 6F and G). Successful colonization of L. paracasei in tumor tissues of nude mice was also confirmed by FISH and PCR (Fig. 6C and Figure S1E). Meanwhile, knockdown of AHR in GC cells also abrogated the protective function of IAA in vitro (Figure S2A-C). These findings demonstrated a critical role of AHR in IAA-mediated the antitumor effect in GC.
Figure5L. paracasei ZJUZ2-3 and IAA could upregulate AHR, and AHR antagonists could abolish the antitumor effect of IAA.
-
Wiki pathway enrichment analysis of DEGs from RNA sequencing dataset of HGC27 cells treated with IAA. Size of circles represents the ratio of DEGs in pathways. Red: upregulated, blue: downregulated.
(B, C) Western blot analysis of the activation of AHR in (B) IAA- and (C) L. paracasei ZJUZ2-3- treated GC cells.
(D, E) The concentration of AHR in GC cells supernatant co-cultured with (D)IAA (50 µM) and (E) L. paracasei ZJUZ2-3.
-
The mRNA levels of AHR in GC cells treated with IAA (50µM).
(G,H) (G)Cell viability at OD450nm and (H)colony formation of human GC cells (AGS and HGC27) with PBS, IAA, CH223191(10µM), CH223191 + IAA.
IAA inhibits the activation of NF-κB signaling pathway
Numerous studies reported that the activation of AHR could inhibit the NF-κB signaling pathway, decrease the transcription of proinflammatory factors, including TNF-α and IL-6, and promote the release of anti-inflammatory factor IL-10, to prevent the proinflammatory process[17]. Pathway enrichment analysis identified that several oncogenic pathways including NF-κB, IL-17, and IL-22 pathway were downregulated in IAA-treated HGC27 (Fig. 5A). Enrichment score showed that NF-κB signaling was significantly inhibited by IAA (Fig. 6I). Indeed, protein expression level of pNF-κB and pIKB-ɑ were decreased in L. paracasei ZJUZ2-3-infected and IAA treated GC cells as shown by western blot (Fig. 6H). Our results collectively indicated that IAA can downregulate NF-κB signaling pathway, which in turn contributes to its antitumor effect in GC.
Figure6 IAA inhibites NF-kB signaling pathway through AHR
-
The representative tumor and mice morphologies.
-
Schematic diagram shows the experimental design, time line and treatment of the GC xenograft mice.
-
Representative FISH images of tumor tissues from PBS- and Lp-injected mice. (blue: nuclear, green: FITC, red: L. paracasei probe); scale bars, 100 µm.
-
Tumor growth curve of mice from mice injected PBS, IAA, CH223191 and CH223191 + IAA.
-
Tumor weight of HGC27 xenograft. Tumor weight was shown as mean ± SEM. ****p < 0.0001.
-
Representative images of H&E staining of tumor tissues of HGC27 xenograft injected PBS, IAA, CH223191 and CH223191 + IAA.
-
Ki-67 staining showed decreased proliferation rates in L. paracasei and IAA injected mice compaired with PBS, CH223191 and CH223191 + IAA. scale bars, 100µm. Lp, L. paracasei.
-
Western blot analysis of the inhibition of NF-kB of AGS and HGC27 treated with Lp and IAA.
-
Gene set enrichment analysis (GSEA) showing the enrichment of NF-kB signaling pathways.
IAA suppressed NF-κB signaling pathway through inhibiting IKB-mediated MTDH phosphorylation
To explore whether and how AHR suppresses NF-kB activation in GC cells, AHR-associated proteins were isolated from HGC27 cells (Fig. 7A) by Co-immunoprecipitation (Co-IP) assays and analyzed by liquid chromatography and high-throughput mass spectrometry (LC-MS/MS) (Fig. 7C). Over 10 candidates along with the known AHR-binding proteins HSPA7 and VIM were identified, but none of them were NF-kB subunits (Table S1), suggesting that AHR did not directly interact with NF-kB[18][19]. However, the oncoprotein Metadherin (MTDH), which binds to IKK-B to promote NF-kB activity in chemical HCC[20], was identified as a putative NF-kB-associated protein among the candidates (Fig. 7B). Immunofluorescence results showed the co-localization of AHR and MTDH in GC cells, further proving the interaction between AHR and MTDH (Fig. 7D).
The interaction between MTDH and IKK-B promotes NF-kB nuclear translocation induced by the tumor necrosis factor α (TNF-α)[21]. MTDH can be phosphorylated by IKK at Ser297 (murine) or Ser298 (human), and only phosphorylated MTDH acts as a co-activator for NF-kB on some of its target gene promoters[21] [22]. Therefore, we speculated that IAA could activate AHR, and then inhibit IKK-mediated MTDH phosphorylation to suppress NF-κB signaling pathway. We thus determined whether AHR affected the interaction between MTDH and IKB-ɑ in GC cells. MTDH phosphorylation (pMTDH) and IKB-ɑ phosphorylation (pIKB-ɑ) was increased in GC cells treated with siAHR or CH223191 (Fig. 7E). Conversely, IAA-treated groups alleviated the phosphorylation of MTDH and IKB-ɑ. Thus, AHR could inhibit IKB-mediated MTDH phosphorylation, thereby suppressing GC cells NF-kB activation.
To explore the direct binding between AHR and MTDH, HDock was employed to predict protein-protein interaction molecular docking. (Fig. 7F). The analysis of protein-protein interaction surfaces based on PDBePISA indicated that GLY237 and GLY240 were putative binding sites for AHR and MTDH (Figure S3A, B and C). Collectively, AHR inhibits MTDH phosphorylation by IKB and the inhibitory of AHR relieves this inhibition, thereby preventing MTDH to promote GC cells NF-kB activation.
Overall, these results suggested that IAA promotes the transcriptional expression of AHR, which may bind to MTDH to inhibit its phosphorylation, thereby suppressing the activation of the NF-κB signaling pathway.
Figure7 AHR complete with IKB-ɑ binding MTDH to suppress NF-kB activation
-
Co-immunoprecipitation (Co-IP) assays showed the protein binding with AHR.
-
Mass spectrometry analysis demonstrated the protein MTDH.
-
Immunoprecipitation of AHR and MTDH HGC27 cells lysates also showed positive binding between AHR and MTDH.
-
Immunofluorescence (IF) used to confirm the binding of AHR and MTDH.
-
Western blot analysis of the inhibition of NF-kB signaling pathway through suppressing the phosphorylation of MTDH.
(F, G)The protein-ligand interaction complex was generated by HDock. (F)ChimeraX was used to visualize the AHR-MTDH complex and (G)binding sites were displayed.