Berberine showed greater efficacy in colitis treatment than sulfasalazine
Several studies have shown that BBR is effective in treating colitis in mouse models. To understand how BBR executes its therapeutic effect, we first compared BBR with SASP. We determined the optimal doses of BBR and SASP in treating DSS-induced colitis in mice by using 50, 100, 150, or 200 mg/kg BBR or SASP to treat adult male mice (via gavage administration) at day 1 after DSS treatment. We found that BBR showed maximal therapeutic effect at 100 mg/kg while SASP at 200 mg/kg, judged by the changes in body weight, colon length, histological changes, the number of goblet cells, and the colitis score (Fig. S1A–H). However, BBR showed a lesser effect at 150 or 200 mg/kg (Fig. S1A–D), suggesting that it might have a toxic effect at high doses.
We then compared the therapeutic effects of BBR (100 mg/kg) and SASP (200 mg/kg) head to head. We administrated BBR or SASP at day 1 or 7 after DSS treatment (Fig. 1A and Fig. S2A–E) and found that both BBR and SASP alleviated colitis phenotypes with BBR showing greater effects, under both conditions (Fig. 1B–E and Fig. S2A–E). We also compared BBR from two suppliers and found that they had similar effects on experimental colitis (Fig. S2F). Given that BBR has a similar molecular weight as SASP (372.822 vs 398.39), these results suggest that BBR is more potent than SASP in treating experimental colitis.
Transcriptome analysis revealed new pathogenic aspects of colitis
Both BBR and SASP have been shown to inhibit NF-κB to suppress inflammation [38, 57]. The above findings suggest that BBR may have additional targets. To test this, we analyzed the gene expression profiles of rectal samples of normal mice, colitis mice, normal mice receiving BBR, and colitis mice receiving BBR. The RNA was isolated from the rectal samples and used for bulk RNA-seq. We first compared the colitis samples to normal samples (Fig. 2A). GO analysis revealed an increase in the expression of genes in immune response, inflammatory response, innate immunity, and immune response to bacteria or virus in the samples of colitis mice (Fig. 2B), and KEGG analysis revealed enrichment in TNFα, NF-κB, cytokine-receptor interaction, IL17, NOD, IBD, TLR, p53, and the Jak-Stat pathway genes (Fig. 2C). These results validated the inflammation feature of the colitis model. On the other hand, GO analysis revealed down-regulated expression of genes in transmembrane and chloride transport, organic acid and retinoic acid metabolism, circadian rhythms, and circadian regulation of gene expression (Fig. 2D), and KEGG analysis uncovered down-regulated expression of mineral absorption, protein, carbon, and fatty acid degradation and absorption, and circadian rhythm genes (Fig. 2E). These results suggest that colitis is associated with suppression of metabolism, mineral absorption, and circadian gene expression in addition to sustained inflammation.
Analysis of the genes in mineral absorption, circadian rhythms, and inflammation confirmed the altered expression of many genes in these pathways or modules in colitis samples (Fig. 2F–H). Intriguingly, we found a decrease in the expression of Wnt-β-Catenin pathway genes and Wnt molecules, and an increase in pro-proliferation genes such as cyclin D and G and anti-proliferation genes such as Caspase 3 and 8 and Fas (Fig. 2I). These results suggest that colitis has complex effects on proliferation of the cells in the colorectal tract.
Gene expression profiling revealed pleiotropic effects of BBR on colitis
We then compared the gene expression profiles of rectal samples of colitis mice and colitis mice treated with BBR (Fig. 1A). GO analysis revealed that BBR increased the expression of genes in organic acid transport, oxidation-reduction, cell response to nutrient, and circadian genes in colitis mice (Fig. 3A), and KEGG analysis revealed that BBR promoted the expression of genes in mineral absorption, steroid hormone and retinol metabolism, and carbon metabolism, as well as circadian rhythm genes to a lesser extent (Fig. 3B). On the other hand, GO analysis showed that BBR suppressed the expression of genes in innate and adaptive immune response, inflammation, chemokine signaling pathway, chemotaxis, and cell response to bacteria (Fig. 3C), and KEGG analysis revealed that BBR suppressed the expression of genes in the pathways of cytokine-receptor, IL17, NF-κB, TNFα, and chemokine signaling (Fig. 3D). Further analysis confirmed that BBR suppressed the expression of many genes in inflammation and the Jak-Stat pathway (Fig. 3E, F). These results indicate that BBR has profound anti-inflammation effects and moreover, it rescues the expression of genes in circadian rhythm and mineral absorption.
We also compared the RNA-seq data of rectal samples of normal mice and normal mice treated with BBR (Fig. S3). GO analysis revealed that BBR up-regulated the expression of genes in a negative regulator of growth, protein folding, circadian genes, and retinol metabolism, and KEGG analysis showed that BBR up-regulated the expression of genes in mineral absorption, protein folding, amino acid metabolism, and circadian rhythms (Fig. S3A–E). On the other hand, GO analysis showed that BBR inhibited the expression of the cell cycle, cell division, and DNA repair genes, and KEGG analysis revealed that BBR suppressed cell cycle, cancer, and p53 signaling pathway genes (Fig. S3A–E). Further analysis confirmed that BBR increases the expression of several mineral absorption and circadian rhythm genes (Fig. S3E, G). The fact that BBR showed similar effects on gene expression profiles in normal mice and colitis mice suggests that many of the effects of BBR are not secondary to inflammation suppression.
The above studies indicate that colitis and BBR profoundly affect the expression of mineral absorption-related genes. qPCR analysis confirmed the reduction in the expression of several mineral absorption-related genes including Atp2b1 and Atp2b4, which were rescued by BBR (Fig. S4A). Western blot analysis of the colon samples confirmed a reduction in plasm membrane Ca2+ pump proteins (PMCA), encoded by Atp2b1, Atp2b2, Atp2b3, and Atp2b4 and recognized by the same antibody, in colitis mice, which was rescued by BBR (Fig. S4B). These results suggest that BBR may help mineral absorption in colitis mice.
Since disruption of circadian rhythms increases IBD risks in human and mice [17], we compared the colon samples of colitis mice and colitis mice treated with BBR at different time points of the day: 03 (3am), 09 (9am), 15 (3pm), 21 (9pm), and found no significant difference in colon length or histology scores among the four groups (Fig. S5A–C), suggesting that the colitis pathology itself does not show day-night rhythm. We also compared the expression of several key circadian genes including Arntl, Clock, Cry1, Cry2, Per1, Per2, Per3, Nr1d1, and Npas2 in colon samples collected at 03, 09, 15, and 21 of normal mice, colitis mice, normal mice receiving BBR, and colitis mice receiving BBR. We found that expression of many circadian genes showed day-night rhythms (Fig. 3G and Fig. S5D). In general, DSS greatly suppressed the expression of these circadian genes, which were largely rescued by BBR (Fig. 3G and Fig. S5D). On the other hand, DSS and BBR only slightly affected the day-night rhythms of circadian gene expression (Fig. 3G and Fig. S5D). These results suggest that circadian genes are targeted by BBR in colitis mice.
Transcriptome analysis confirmed that BBR activated Wnt and circadian gene expression compared to SASP
We also conducted RNA-seq on colon samples of normal mice, colitis mice, normal mice receiving SASP, and colitis mice receiving SASP. We found that compared to colitis samples, SASP suppressed the expression of genes regulating inflammation and chemotaxis (Ptgs2, IL1b, IL6, Ccl2, Ccl3, Cxcl1, Cxcl2, Cxcl3, Cxcl5, and Cxcl12), oxidative phosphorylation, IL17 signaling, and NF-kB signaling; and up-regulated genes regulating protein folding, RNA splicing, DNA replication and repair, amino acid metabolism, and some immune response genes (Irf1, H2-Q4, H2-T23, and Pamb8) (Fig. S6A–D). Overall, the findings support the anti-inflammation activity of SASP. Moreover, we compared the transcriptomes of BBR-treated colitis samples and SASP-treated colitis samples and found that BBR up-regulated the expression of genes involved in ECM, Calcium ion transport, wound healing, stromal cells, Wnt, Akt, MAPK, and circadian genes and suppressed the expression of genes involved in cell cycle, DNA repair, IBD, immune response (Fig. S6E–J). These results further support that BBR and SASP have different actions in colitis treatment and that BBR increases the expression of circadian genes and Wnt molecules.
BBR suppressed macrophages and granulocytes but not T cells in colitis models
Our RNA-seq data indicate that BBR suppressed inflammation. To verify this, we immunostained rectal sections for immune cell marker CD45 and found the signals were increased in DSS-treated samples compared to control samples, which were suppressed by BBR (Fig. 4A). qPCR analysis revealed that BBR suppressed the expression of IL6, IL1β, Ptgs2, IFNγ, and TNFα in rectal samples (Fig. 4B). Western blot analysis of TNFα and IL6 in colon samples confirmed an increase in these two proteins in colitis mice, which was suppressed by BBR (Fig. 4C). We then designed a panel to identify and quantify all leukocyte populations in the colorectal tract by flow cytometry and found that BBR inhibited the activation/infiltration of macrophages and granulocytes but not T cells (Fig. 4D, E and Fig. S7A, B), consistent with previous studies reporting suppression of macrophages but inconsistent with previous studies reporting suppression of T cells by BBR [42, 45, 46]. Moreover, it has been shown that BBR may inhibit inflammation via NF-κB and AMPKs [39]. Our RNA-seq data show that colitis activated whereas BBR suppressed Jak-Stat3 signaling (Fig. 2C, Fig. 3D, and F), a critical regulator of cytokine expression and inflammation [58], consistent with previous studies showing that Stat3 activation might be affected by BBR [31, 59].
BBR promoted ISC proliferation and epithelial repair in colitis models
Our RNA-seq data indicate that BBR affects not only cell division and cell cycle genes but also anti-proliferation genes (Fig. 2 and Fig. S3). Colorectal epithelia undergo rapid turnover, driven by Lgr5+ ISCs. To test the effects of BBR on ISCs, we induced colitis in Lgr5-CreERT-GFP; Rosa-tdTomato mice [5], which were treated with BBR. The genotypes of the mice were verified with PCR analysis of the genomic DNA (Fig. S8). While tamoxifen (TAM) administration led to Tomato labeling of the descendent cells of Lgr5+ ISCs, GFP staining detected ISCs (Fig. 5A). We found that BBR significantly increased the number of Lgr5+ ISCs and promoted the renewal activity of Lgr5+ ISCs, manifested by an increase in the number of GFP+ ISCs and Tomato+ epithelial cells, respectively (Fig. 5B, C). These results indicate that BBR promotes colorectal epithelial repair in experimental colitis models. Moreover, we found that in non-colitis mice, BBR modestly increased the colorectal epithelial turnover but not the number of GFP+ ISCs (Fig. 5B, C), suggesting that the epithelial phenotype is not secondary to inflammation suppression by BBR.
BBR promoted the expression of Wnt genes in colorectal stromal cells
The activity of ISCs is controlled by their niche, especially the resident stromal cells [6], which secrete Wnts, BMPs, Notch, and other signaling molecules to regulate ISC proliferation and differentiation. In particular, Wnts are critical for ISC proliferation [60]. Western blot analysis of major mitogenic pathways in rectal homogenates revealed that β-Catenin, a downstream transcription factor stabilized by Wnts, was greatly elevated by BBR treatment in colitis samples (Fig. 5D). On the other hand, BBR treatment did not affect ERK and Akt activation (Fig. 5D). We immunostained the rectal sections for β-Catenin and found that BBR increased β-Catenin signals in epithelial cells, associated with an increase in Ki67+ proliferating cells (Fig. 5E). These results suggest that Wnt-β-Catenin may play an important role in restitution of colorectal epithelia. We also detected increased p-Stat3 levels in the samples of colitis mice, which was suppressed by BBR (Fig. 5D). These results, together with our RNA-seq results (Fig. 2C, Fig. 3D, and F), suggest that Stat3 pathway plays a role in BBR-induced suppression of inflammation.
A recent study reported that BBR activated Wnt-β-Catenin signaling in colonic epithelial cells to protect the mucosal barrier and blocking the activation of this pathway with FH535, a β-Catenin/Tcf inhibitor, diminished the barrier-protection and immune-suppression activities of BBR [41]. We tested whether BBR directly acted on epithelial cells and activated the β-Catenin pathway using colonic epithelial cell line HCT116. We treated the cells with different doses of BBR and found that BBR activated ERK and Akt to minimal extents; however, it decreased the protein levels of β-Catenin (Fig. S9). These results suggest that increased activation of β-Catenin in colorectal epithelial cells of BBR-treated colorectal samples is unlikely to be cell-autonomous. Instead, BBR may act on other cells to stimulate the synthesis and secretion of Wnt molecules, which then activate β-Catenin in epithelial cells to promote the expression of tight junction proteins. Certainly, this warrants further investigation.
The colonic tissues contain immune cells (CD45+), endothelial cells (CD31+), and stromal cells besides epithelial cells (EpCAM+). Stromal cells do not express CD45, CD31, or EpCAM. Recent studies have shown that stromal cell express CD51 [61]. We isolated stromal cells (CD45-CD31-EpCAM-CD51+) from colorectal mononuclear cells by excluding immune cells, endothelial cells, and epithelial cells (Fig. 5F and Fig. S10A). We also analyzed the expression of stromal cell markers Vimentin and CD29 [62, 63] on sorted stromal cells by flow cytometry and found that more than 96.6% of cells were positive for the two markers (Fig. S10A). We then immunostained the sorted cells for Vimentin, EpCAM, CD31, and CD45 and found that all the sorted cells were positive for Vimentin but negative for EpCAM, CD31, and CD45 (Fig. S10B). These results verified the identity of the stromal cells.
qPCR analysis revealed that the stromal cells expressed many Wnt molecules (Fig. 5G). In DSS-induced colitis models, the expression of Wnt molecules including Wnt2, 4, 6, 8b, 9b, 10a, 10b, and 11 were suppressed in stromal cells, which was restored by BBR treatment (Fig. 5G), while expression of Wnt5a and Wnt16 was not significantly affected by DSS or BBR and other Wnt molecules were not detected (Fig. 5G). Western blot analysis of the colon samples for Wnt6 and Wnt10a expression confirmed a reduction in these two proteins in colitis mice, which was rescued by BBR (Fig. 5H). Overall, these results suggest that BBR might induce ISC expansion via stromal cell-produced Wnt molecules, consistent with the fact that Lgr5 is a target of Wnt signaling and that Lgr5 activation drives ISC proliferation [60].
Inhibiting Wnt secretion diminished the therapeutic effect of BBR
The above studies showed that Wnt molecules were down-regulated in DSS-induced colitis samples whereas BBR could restore their expression in resident stromal cells. To test the in vivo roles of Wnt-β-Catenin activation, we administrated IWP-2, an inhibitor of PORCN (Porcupine O-Acyltransferase) required for Wnt modification and secretion, to the mice via peritoneal injection at a dose that has been previously used [64]. We found that β-Catenin signals were significantly suppressed (Fig. 6A). Moreover, in the presence of IWP-2, the therapeutic effects of BBR on colitis mice including the body weight, the colon length, histological changes, the number of goblet cells, and the colitis score were largely diminished (Fig. 6B–E). The remaining effect is likely due to the immune suppression activity of BBR, as we found that IWP-2 did not affect Stat3 activation (Fig. 6A). Overall, these results suggest that Wnt secretion, likely by stromal cells, is required for BBR to execute its therapeutic effect on experimental colitis.
BBR restored circadian gene expression in stromal cells
Our RNA-seq data indicate that colitis was associated with a decrease in the expression of various circadian rhythm genes, which was restored by BBR (Figs. 2 and 3). Previous studies have implicated circadian clock in bone marrow/peripheral macrophages as a regulator of IBD pathogenesis [8]. To determine the contribution of various colorectal cell types to the change in circadian gene expression, we isolated CD45+ immune cells, EpCAM+ epithelial cells, and CD51+ stromal cells from the colon of normal mice, colitis mice, normal mice receiving BBR, and colitis mice receiving BBR, via FACS sorting (Fig. 5F). We isolated RNA from these cells and performed qPCR. We found that expression of major circadian genes was not altered by colitis or by BBR treatment in epithelial cells (Fig. S11A). In immune cells, colitis was associated with a decrease in expression of Per2 while the expression of other key circadian genes was not affected (Fig. S11B). However, BBR failed to rescue the expression of Per2 in colitis mice (Fig. S11B). These results suggest that colitis and BBR only have a minor effect on the expression of circadian genes in epithelial cells and immune cells of the colorectal tract.
On the contrary, we found that in colorectal stromal cells, colitis was associated with decreases in the mRNA levels of Arntl, Clock, Cry1, Cry2, Per2, Per3, Nr1d1, Npas2, and BBR restored the expression of these genes except Per2 (Fig. 7A). Furthermore, we found that BBR increased the expression of circadian genes including Arntl, Clock, Cry2, Per1, and Npas2 in stromal cells of normal mice (Fig. 7A). Western blot analysis confirmed a reduction in Arntl and Clock protein levels in the stromal cells of the colitis mice, which was rescued by BBR (Fig. 7B). These results suggest that colitis and BBR mainly affect the circadian rhythm in resident stromal cells rather than epithelial and immune cells.
Circadian rhythms play a role in BBR-induced Wnt gene expression
Recent studies have shown that circadian rhythms in ISCs are controlled by niche cells during regeneration [13, 65]. To determine whether the circadian rhythm is involved in BBR-induced Wnt expression in stromal cells, we used the primary colorectal stromal cells as a model. In cultured cells, we found that BBR promoted the expression of circadian rhythm genes including Arntl, Clock, and Nr1d1 but no other genes (Fig. 7C). Moreover, BBR promoted the expression of various Wnt genes including Wnt6, 8b, 9b, 10a, and 11 (Fig. 7D). These results suggest that BBR directly acts on stromal cells. Since not all circadian genes or Wnt genes are up-regulated by BBR in cultured stromal cells, we speculate that BBR also regulates the expression of circadian genes and Wnt genes in non-cell-autonomous manners.
We then knocked down the main circadian gene Arntl in cultured stromal cells with siRNA (Fig. 7E). We found that knock-down of Arntl inhibited the expression of Wnt6, 8b, 9b, and 10a but not Wnt11 in cultured stromal cells (Fig. 7F). Moreover, in Arntl knockdown cells, BBR failed to increase the expression of Wnt6, 8b, 9b, and 10a but not Wnt11 (Fig. 7F). Taken together, these results suggest that BBR promotes the expression of a number of Wnt genes in a circadian-dependent manner in stromal cells.