Tryptophan metabolite norharman secreted by cultivated Lactobacillus attenuates acute pancreatitis as an antagonist of histone deacetylases

Sprague–Dawley (SD) rats and wild-type (WT) C57BL/6 mice were purchased from the Experimental Animal Center of Dalian Medical University (Dalian, China). Rftn1[flox/flox, lyz2-cre] mice were purchased from Cyagen Biosciences, Inc. (Guangzhou, China). All experimental procedures were conducted in compliance with the People’s Republic of China Legislation Regarding the Use and Care of Laboratory Animals. We constructed rat and mouse AP models as previously reported [19, 20].

After DNA extraction and PCR amplification, the samples were sequenced on an Illumina MiSeq platform according to our previous method [3].

ChIP assays were performed using a ChIP kit (CST#9003, USA) according to the manufacturer’s instructions. Eluted purified DNA was used for quantitative reverse transcription (qPCR) assays by specific primers (Additional file 1: Table S1).

Additional experimental details including AP model establishment, bacterial culture, LC–MS/MS analysis, cell viability and cytotoxicity, cell transfection, RNA sequencing, confocal microscopy, lipid metabolomics, luciferase assay, molecular docking, chromatin immunoprecipitation (ChIP) assay, ELISA, isolation of bone marrow-derived macrophages (BMDMs), infection of BMDMs with recombinant Rftn1 lentiviral vector, agarose gel electrophoresis, serum enzyme assays, hematoxylin and eosin and double-color immunofluorescence staining, flow cytometry analysis, qPCR, western blotting, and statistical analysis are provided in the Supplementary Information.

Gut microbiota and host metabolism have been a research hotspot in recent years. To explain the relationship of AP-induced metabolism disorder, we performed 16S rDNA gene sequencing and untargeted metabolomics of fresh fecal samples obtained from rats with AP, and sham operation was used as a control. Consistent with previous research, AP has a characteristic gut microbiota, mainly manifested as a decrease in beneficial bacteria and an increase in pathogenic bacteria (Fig. 1A, B). The relative abundance of Lactobacillus species, including Lactobacillus oris, Lactobacillus crispatus, and Lactobacillus helveticus, was decreased in the AP group compared with the Ctrl group (Fig. 1C). According to the results of untargeted metabolomics, we found 798 metabolites were upregulated but 826 metabolites were downregulated in positive model, and 1117 metabolites were upregulated but 1207 metabolites were downregulated in negative model (Fig. 1D). This result indicated metabolite disorder was observed in the feces from AP. As shown in Fig. 1E, the differentially abundant metabolites were significantly enriched in multiple metabolic pathways. Literature review revealed that gut Lactobacillus-mediated tryptophan metabolism modulated host susceptibility and disease development [21, 22]. Particularly, the relative abundance of tryptophan was absolutely downregulated in the AP group compared with the Ctrl group (Fig. 1F). Tryptophan was regarded as an essential substance in host physiology, and the three tryptophan metabolism pathways were regulated by gut microbiota, involving 5-hydroxytryptamine (5-HT) production, ligands of the aryl hydrocarbon receptor (AhR), and kynurenine-producing indoleamine 2,3-dioxygenase pathways [23]. Moreover, metabolites of kynurenine pathway (kynurenic acid and xanthurenic acid) and indole and indole derivatives were declined in the AP group, suggesting both endogenous and gut microbiota tryptophan metabolism were indeed inhibited in AP (Fig. 1G). We speculated that Lactobacillus-mediated tryptophan metabolism may play a central role in AP. To further verify the relationship between Lactobacillus and tryptophan metabolism. We conducted the correlation analysis and confirmed that the abundance of Lactobacillus was positively correlated with the level of tryptophan in both negative and positive models (Fig. 1H), indicating the dysfunction of tryptophan metabolism occurs in microbiota-host crosstalk in AP. Taken together, these findings indicated that AP induced gut microbiota dysfunction and metabolic imbalance, resulting in the suppression of Lactobacillus-mediated tryptophan metabolism pathway.

Although various studies have expounded the relationship between gut microbiota dysfunction and metabolic imbalance in AP, there is a lack of in-depth mechanistic research focusing on Lactobacillus-mediated tryptophan metabolism in AP. Based on AP-induced decreases in the abundance of Lactobacillus and tryptophan, we cultured Lactobacillus with tryptophan to further clarify how Lactobacillus regulates tryptophan metabolism. A total of three Lactobacillus strains, L. oris, L. crispatus, and L. helveticus, were cultured in tryptophan-rich medium in vitro, and a schematic diagram of LC–MS/MS was displayed in Fig. 2A. Absorbance detection showed that several low concentrations of tryptophan could not promote the proliferation of Lactobacillus strains, but Lactobacillus growth was significantly inhibited by 1% tryptophan (Additional file 1: Fig. S1A). Thus, Lactobacillus strains were cultured in a medium with 0.1% tryptophan for the detection of differential metabolites. As shown in Fig. 2B, a total of five Lactobacillus metabolites of tryptophan were identified as indoles and their derivatives, including norharman, tryptophan, 5-HT, indole-3-lactic acid and indole-3-acrylic acid, based on both negative and positive models of LC–MS/MS.

Regarding classically activated (M1) macrophages exhibiting proinflammatory phenotypes in AP, we explored the roles of five Lactobacillus metabolites in macrophage activation. LPS is a classic inducer of M1 macrophages in vitro, and RAW264.7 cells were stimulated with three increasing concentrations of LPS: 0.1, 1, and 10 μg/ml. Then, LPS at 1 μg/ml was validated as the optimal concentration for M1 macrophage activation by qPCR and flow cytometry (Additional file 1: Fig. S1B-C). In addition, RAW264.7 cells were treated with above five metabolites for 24 h to evaluate the cytotoxicity, and finally, 10 μM was determined to be the maximum nontoxic concentration in vitro (Additional file 1: Fig. S1D). Interestingly, two metabolites (norharman and 5-HT) could inhibit macrophage M1 polarization, as shown by the detection of Cd86 and inducible nitric oxide synthase (iNOS) expression (markers of M1 macrophages), and norharman had the strongest inhibitory effect (Fig. 2C). To explore whether norharman was exogenous or endogenous, we examined the level of norharman in RAW264.7 cells and culture supernatant of tryptophan broth medium with or without Lactobacillus strains by targeted LC–MS/MS. The results showed that there was no intracellular norharman in RAW264.7 cells, whereas the concentration of norharman was increased in the culture supernatant of tryptophan broth medium with Lactobacillus strains compared with those without Lactobacillus strains, suggesting that norharman was mainly from tryptophan metabolism regulated by Lactobacillus (Fig. 2D). Norharman also significantly reduced the M1 macrophage ratio after LPS stimulation, as shown by flow cytometry (Fig. 2E, F).

On the basis of significant inhibition of M1 macrophage activation in vitro, we also validated the effect of the enterobacterial metabolite norharman in suppressing macrophage M1 activation in AP mouse models. Toxicological studies showed that the serum levels of pancreatic toxicity, hepatotoxicity, and nephrotoxicity markers, including AMS, ALT and AST, BUN, and CRE, were unaffected by increasing doses of norharman in mice (Additional file 1: Fig. S2A-E). Thus, the maximum dose of norharman (100 mg/kg for 24 h) was chosen for the pharmacodynamics experiments to confirm the effect of norharman in vivo. As shown in Fig. 3A, B, the levels of serum AMS and lipase, as well as the mRNA expression of Tnf-α and Il-1β in pancreatic and intestinal tissue, were significantly higher in the AP group than that in Ctrl group but markedly decreased after norharman treatment. HE staining demonstrated obvious pancreatic edema, inflammatory infiltration, and destruction of intestinal villi in the AP group compared with the Ctrl group. However, norharman reduced pancreatic lesions and inflammation, maintained the integrity of intestinal villi structure, and reduced intestinal damage (Fig. 3C). As shown in Fig. 3D, E, an increased number of CD68⁺/iNOS⁺ M1 macrophages (red and green fluorescent regions) was observed in the pancreas from the AP group compared to the Ctrl group, but the fluorescence intensity was obviously reduced in the norharman group. Flow cytometric analysis of the spleen presented similar results. Compared with the Ctrl mice, the mice with AP had a higher M1 macrophage ratio, which was decreased after norharman treatment (Fig. 3F, G). These results confirmed that norharman ameliorated the pathology of the mice with AP by inhibiting M1 macrophage activation.

To further investigate the targets of norharman that regulate macrophage polarization, we used RNA sequencing to analyze gene alterations at the transcriptome level. The heatmap in Fig. 4A shows the top 10 DEGs between the LPS group and the LPS + Nor group, including placenta-expressed transcript 1 (Plet1), histone cluster 1 (Hist), casitas B-lineage proto-oncogene like 1 (Cbll1), threonine aldolase (Tha1), interleukin-19 (Il-19), endothelial lipase (Lipg), chemokine (C–C motif) ligand 6 (Ccl6), ski sarcoma viral oncogene homolog (Ski), tissue inhibitor of metalloproteinase 1 (Timp1) and Rftn1; these genes were verified by qPCR in vitro (Fig. 4B). According to a literature review, we found that Rftn1 encodes the Raftlin protein, which can control clathrin-associated adaptor protein-2 in clathrin-mediated endocytosis of TLR4 ligands in macrophages, as well as modulate cell entry of poly(I:C), which is important for activation of TLR3 in human myeloid dendritic cells and epithelial cells [24, 25].

To determine whether Rftn1-mediated macrophage polarization, we designed Rftn1 knockdown by using siRNA and shRNA to detect the change in polarization phenotype. After transfection of three siRNAs/shRNAs into RAW264.7 cells, the siRNA/shRNA with the best knockdown efficiency was chosen for the follow-up studies (Additional file 1: Fig. S3A-B). Compared with the NC treatment, knockdown of Rftn1 increased the mRNA expression of M1 macrophage markers, whereas norharman decreased the mRNA expression of M1 macrophage markers by upregulating Rftn1 expression (Fig. 4C, D). Moreover, the Rftn1 knockdown groups had a higher ratio of M1 macrophages than the NC group, which was significantly reduced in the siRftn1/shRftn1 + Nor group (Fig. 4E, F). As shown in Fig. 4G, H, the mRNA expression of Rftn1 and M1 macrophage markers was increased in the AP group compared with that in the Ctrl group, whereas decreased in the AP + Nor group, except the mRNA expression of Cd86 without significant difference. In addition, the protein levels of Rftn1 in the pancreas and intestine were significantly decreased in the AP group compared with the Ctrl group, whereas they were increased in the AP + Nor group (Fig. 4I, J). Furthermore, we have examined the change of intestinal tight junction proteins (Claudin-1, Occludin-1 and ZO-1), and protective mucins (Mucin-2). As shown in Additional file 1: Fig. S4, the expression of intestinal tight junction proteins (Claudin-1, Occludin-1, and ZO-1) was decreased in the AP group compared with the Ctrl group, which was increased in the AP + nor group. Similarly, we observed less expression of protective mucins (Mucin-2) in the AP groups compared with that in the Ctrl group, and norharman could reverse this phenotype. These results revealed that Rftn1 is a crucial regulator of norharman in restraining the M1 polarization of macrophages.

As a novel major protein in lipid rafts, Raftlin is localized exclusively in lipid rafts by fatty acylation of N-terminal Gly2 and Cys3, influencing the structural integrity of lipid rafts and participating in lipid metabolism [26]. To determine the structure of lipid rafts affected by Rftn1, we used cholera toxin B (CTB) as a classic marker of membrane lipid rafts as previously reported [27, 28]. As shown in Fig. 5A, CTB-positive macrophages appeared round or fusiform in shape with dense fluorescence in the NC group by laser confocal microscopy. However, the destruction of lipid rafts structures characterized by scattered fluorescence was observed following siRftn1/shRftn1 transfection in vitro, but this effect was reversed by norharman treatment.

Based on these results, we hypothesized that Rftn1 may influence lipid metabolism. Therefore, we detected the expression of lipid metabolism-related genes and notably found that Rftn1 knockdown affected the expression of lipid metabolism-related genes, including acetyl coenzyme A carboxylase alpha (Acaca), acetyl coenzyme A carboxylase beta (Acacb), carnitine palmitoyltransferase 1a (Cpt1a), carnitine palmitoyltransferase 1b (Cpt1b), fatty acid binding protein 4 (Fabp4), fatty acid transport protein 1 (Fatp1), fatty acid synthase (Fas), insulin-induced gene (Insig), and lipoprotein lipase (Lpl) (Fig. 5B, Additional file 1: Fig. S5). Subsequently, lipid metabolomics further confirmed the imbalance of lipid metabolism in macrophages after Rftn1 knockdown, which mainly manifested as increased production of sphingomyelin (SM) and phosphatidylcholine (PC), but norharman treatment reversed these phenotypes (Fig. 5C, D). Recent research has suggested that SM protects against dysfunctional lipid metabolism, gut dysbiosis, antigen presentation, and inflammation [29, 30]. At present, the anti-inflammatory properties of PC have attracted increased attention. PC could benefit systemic inflammation via the gut-brain axis [31], decrease the expression of other lipases and the lipogenic marker peroxisome proliferator-activated receptor-γ, and induce lipolysis in adipose tissue [32]. Together, these findings indicated that norharman could inhibit the M1 polarization of macrophages by restoring the imbalance of lipid metabolism.

Given the strong relationship between lipid metabolism and the inflammatory response in macrophage polarization, we applied RNA sequencing to identify the key candidate gene changes and analyzed downstream inflammatory signaling pathways after Rftn1 knockdown and norharman treatment. We found that various proinflammatory factors were upregulated in the shRftn1 group, including C-X-C motif ligand 2 (Cxcl2), C-X-C motif ligand 10 (Cxcl10), chemokine (C–C motif) ligand 2 (Ccl2), myeloid differentiation primary response 88 (Myd88), janus kinase 2 (Jak2), and interleukin-23a (Il23a). In contrast, the expression of proinflammatory factors, such as Cxcl2, Il23a, Tnf, and toll-like receptor 2 (Tlr2), was downregulated in the shRftn1 + Nor group. Furthermore, KEGG enrichment analysis revealed that norharman could block multiple pathways related to the inflammatory response induced by Rftn1 knockdown, such as the TNF signaling pathway, NF-κB pathways, MAPK signaling pathway, PI3K signaling pathway, and JAK-STAT signaling pathway (Fig. 6A, B). To validate these DEGs in vitro, we examined the mRNA expression levels of NF-kappaB (Nfkbia), signal transducer and activator of transcription 1 and 2 (Stat1 and 2), Ccl2, Jak2, Tlr2, Cxcl2, Il23a, and interleukin-4 receptor alpha (Il4ra), which were consistent with the RNA sequencing results (Fig. 6C).

Inspired by the above data, we determined that norharman inhibited M1 macrophage activation by promoting Rftn1 expression, and we also validated the mRNA and protein levels of Rftn1 after norharman treatment. The results showed that norharman could significantly upregulate Rftn1 expression at both the mRNA and protein levels in RAW264.7 cells (Fig. 7A, B). To gain insight into the molecular mechanisms by which norharman promoted Rftn1 expression, we hypothesized that norharman could directly bind to the Rftn1 promoter region and promote its transcription, and we performed dual-luciferase reporter assays. As shown in Fig. 7C, HEK293T cells were transfected with a luciferase plasmid containing the sequence of the Rftn1 promoter or control sequence, and we found that norharman could not increase Rftn1 promoter-driven luciferase activity, suggesting that norharman promoted Rftn1 transcription not by directly binding to its promoter region but through other regulatory means.

Through a literature review, we found that the norharman analog β-carboline tethered cinnamoyl 2-aminobenzamides could serve as class I selective HDAC inhibitors, and a series of novel β-carboline-based hydroxamate derivatives exhibited profound HDAC1/3/6 inhibitory effects [33, 34]. In addition, acetylation on histones is regulated by histone acetyltransferases, while HDACs can remove acetyl groups from the N-terminal tail of histones [35]. To verify the relationships between norharman and HDACs, we measured the enzymatic activity of HDAC1-4 after norharman treatment and found that norharman could inhibit the enzymatic activity of HDAC1-4 (Fig. 7D).

Subsequently, molecular docking analysis was used to confirm the interactions of HDAC1-4 and norharman. As a result, the docking scores (Kcal/mol) of HDAC1-4 showed a decreasing trend; thus, HDAC4 possessed the best binding ability with the norharman compounds. The representative three-dimensional patterns of docking combinations are shown in Fig. 7E. To determine whether H3K9/14 acetylation was recruited directly to the Rftn1 gene, we performed ChIP and qPCR analyses of RAW264.7 cells. For determination of the specific binding sites, the Rftn1 promoter region (− 2000/ + 500) was truncated into 9 segments (Fig. 7F, Additional file 1: Table S1). ChIP experiments showed enrichment of H3K9/14ac signal binding around all elements of the Rftn1 promoter in the LPS group, while increased H3K9 acetylation signals were also enriched around special elements (− 1736/ − 1488, − 614/ − 433, − 357/ − 54, and + 201/ + 499 bp) after norharman treatment, and increased H3K14 acetylation signals enriched around (− 882/ − 736 and − 186/ − 54 bp) after norharman treatment (Fig. 7G, H). Taken together, these findings demonstrated that norharman promoted Rftn1 transcription by repressing the activity of HDACs to increase H3K9/14 acetylation in RAW264.7 cells.

Given the crucial role of Rftn1 in M1 macrophages aggravating AP, myeloid-specific Rftn1 knockout mice were constructed (Fig. 8A, Additional file 1: Fig. S6). The concentrations of AMS in serum were significantly higher in the AP group than in the Ctrl group and markedly increased in the Rftn1^−/− + AP group but showed no change after norharman treatment in the Rftn1^−/− + AP + Nor group (Fig. 8B). Relative to the mice with AP, the Rftn1^−/− + AP mice showed more severe AP, as evidenced by aggravated acinar edema and necrosis in the pancreas (Fig. 8C) and an elevated M1 macrophage ratio in the spleen (Fig. 8D, E). However, these indexes were not changed after norharman treatment. These results confirmed that norharman ameliorated the pathology of AP partly by directly targeting Rftn1 in mice.

To verify the necessity of norharman-increased H3K9/14ac binding at the promoter of Rftn1, we used substitution promoters by infection with Rftn1 lentiviral vectors. Recombinant Rftn1 lentiviral vectors contained the CMV promoter, which was completely different from the Rftn1 self-contained promoter. Bone marrow-derived macrophages (BMDMs) were isolated and cultured for further investigation. Rftn1 expression in BMDMs was hard to detect in the Rftn1^−/− group compared with the WT group, while Rftn1 was overexpressed after recombinant Rftn1 lentiviral vector infection in the Rftn1^−/−  + Rftn1^OE group, but no significant change was observed after norharman treatment in the Rftn1^−/−  + Rftn1^OE + norharman group by using qPCR and agarose gel electrophoresis (Fig. 8F, G). In addition, obvious destruction of lipid rafts structures of BMDMs was observed by laser confocal microscopy in the Rftn1^−/− group compared with the WT group; this parameter was rescued in the Rftn1^−/−  + Rftn1^OE group, while no lipid rafts structures were observed after norharman treatment (Fig. 8H). Together, these data verified that norharman targeted HDAC1-4 to increase H3K9 acetylation depending on the promoter region of Rftn1, which promoted Rftn1 transcription.