Core clock gene BMAL1 and RNA-binding protein MEX3A collaboratively regulate Lgr5 expression in intestinal crypt cells

Abstract

The intestinal epithelium is highly regenerative. Rapidly proliferating LGR5⁺ crypt base columnar (CBC) cells are responsible for epithelial turnover needed to maintain intestinal homeostasis. Upon tissue damage, loss of LGR5⁺ CBCs can be compensated by activation of quiescent +4 intestinal stem cells (ISCs) or early progenitor cells to restore intestinal regeneration. LGR5⁺ CBC self-renewal and ISC conversion to LGR5⁺ cells are regulated by external signals originating from the ISC niche. In contrast, little is known about intrinsic regulatory mechanisms critical for maintenance of LGR5⁺ CBC homeostasis. We found that LGR5 expression in intestinal crypt cells is controlled by the circadian core clock gene BMAL1 and the BMAL1-regulated RNA-binding protein MEX3A. BMAL1 directly activated transcription of Mex3a. MEX3A in turn bound to and stabilized Lgr5 mRNA. Bmal1 depletion reduced Mex3a and Lgr5 expression and led to increased ferroptosis, which consequently decreased LGR5⁺ CBC numbers and increased the number of crypt cells expressing +4 ISC marker BMI1. Together, these findings reveal a BMAL1-centered intrinsic regulatory pathway that maintains LGR5 expression in the crypt cells and suggest a potential mechanism contributing to ISC homeostasis.

Introduction

A healthy intestine is essential for digestion, nutrient extraction, absorption and waste removal. The intestinal epithelium is organized into crypt-villus structures. In the crypts, there are two types of intestinal stem cells (ISCs) wedged between Paneth cells which protect and provide growth factors to the ISCs¹. One type of ISCs are LGR5-expressing (LGR5⁺) crypt base columnar (CBC) stem cells which sit at position +1 to +3 at the crypt base between Paneth cells. The other ISC type are ‘+4’ cells which express BMI1, HOPX and LRIG1¹. Proliferation of LGR5⁺ CBCs drives the rapid turnover and regeneration of intestinal epithelium to maintain intestinal homeostasis^2,3. However, fast-proliferating LGR5⁺ CBCs are sensitive to cytotoxic injury resulting from chemotherapy or irradiation (IR). Depletion of LGR5⁺ CBCs by IR or in a toxin receptor-mediated cell knockout mouse model resulted in severe crypt loss, irregular villus structure, and reduced epithelial turnover^4,5. In contrast to the stress sensitive LGR5⁺ CBCs, +4 cells are more resistant to IR⁶. It has been suggested that these +4 cells can reprogram into LGR5⁺ cells to replenish the crypt base when LGR5⁺ CBCs are depleted due to intestinal injury⁷. In addition, enterocyte progenitors⁸ and occasionally Paneth cells⁹ have also been shown to contribute to replenish the loss of LGR5⁺ CBCs. This robust ISC homeostasis requires interconnected developmental signaling pathways from the ISC niche, including Wnt, Notch, Hedgehog, Bone morphogenetic protein (BMP), Eph-ephrin and EGF¹⁰, to maintain continuous renewal of the intestinal epithelium.

Several studies have shown that LGR5⁺ CBC self-renewal, fate decisions and ISC conversion are regulated by extrinsic factors¹⁰. However, recent evidence has indicated that intrinsic signaling factors are also required to maintain the LGR5⁺ CBC pool and ensure adequate regeneration of intestinal epithelium¹¹. RNA-binding proteins regulate various aspects of RNA fate and function^12,13. In the intestine, RNA-binding proteins, including MEX3A, MSI1, HUR and IMP1, facilitate development and repair, as well as maintenance of normal mucosa structures^11,14. MEX3A expression has been used to identify a subset of slowly dividing LGR5⁺ CBCs¹⁵. Under chemotherapy-induced intestinal damage, these MEX3A^high cells contributed to intestinal epithelium repair and homeostasis¹⁵. Conversely, MEX3A knockout decreased the number of LGR5⁺ CBCs in the duodenum and delayed organoid formation¹¹, indicating that MEX3A is required for survival and stemness functions of LGR5⁺ CBCs. The factors controlling MEX3A expression and the underlying mechanism of MEX3A effect on LGR5⁺ CBCs are unknown.

Despite the increasing recognition that the circadian machinery impacts many physiological functions, it is unclear how the circadian machinery contributes to intestinal homeostasis. A few recent studies have suggested that circadian rhythm affects ISC proliferation and tissue regeneration^16,17. However, whether the core circadian machinery regulates expression of Lgr5, or other ISC genes, is unclear.

Among the core circadian clock genes, BMAL1 is the key transcription factor¹⁸. BMAL1 forms heterodimers with CLOCK¹⁹ to transcriptionally regulate genes required for the network of transcription-translation feedback loops (TTFL) that drive circadian rhythms. In the gastrointestinal (GI) tract, it has been shown that BMAL1 deletion inhibited drug export²⁰, prevented ghrelin secretion²¹, and promoted glucose uptake²². These observations demonstrated an indispensable role of BMAL1 in maintaining GI tract function. BMAL1 has been shown to regulate the timely division of intestinal cells as depletion of Bmal1 abolished oscillation of Wnt3a, a critical niche component for maintaining LGR5⁺ CBC proliferation, and resulted in disruption of circadian rhythm and cell cycle coupling¹⁷. In addition, BMAL1 was found to contribute to the inflammation response and cell proliferation during intestinal regeneration¹⁶. Despite such evidence suggesting that BMAL1 is essential to intestinal cell proliferation and regeneration, the mechanisms by which BMAL1 regulates ISC and intestinal homeostasis remain unclear. In any given mammalian cell or tissue, approximately 5–20% of the transcripts show circadian oscillations²³, and approximately 35% of the oscillating transcripts are modulated by post-transcriptional control²⁴. Several RNA-binding proteins participate in circadian regulation^25,26,27. It is possible that the circadian machinery incorporates RNA-binding proteins to mediate post-transcriptional regulation to maintain intestinal homeostasis.

The intestine, particularly the upper small intestine, is the major organ that absorb various chemicals, including nutrients and xenobiotics. Moreover, oncology drugs and radiation therapy for gastrointestinal cancers or para-aortic nodes from gynecological cancers have been reported to cause high toxicity in the small intestine, particularly duodenum^28,29. Thus, to cope with such frequent stress and toxic insult, the small intestine must rely on tissue regeneration driven by intestinal stem cells to maintain homeostasis and fitness. We found that BMAL1 and MEX3A act together to regulate Lgr5 expression and may contribute to LGR5⁺ CBC homeostasis in duodenal crypts. In this mechanism, BMAL1 upregulated expression of Mex3a, while MEX3A further enhanced LGR5 expression by directly binding and stabilizing Lgr5 mRNA. Bmal1 depletion and Mex3a downregulation elevated lipid peroxidation level and ferroptosis stress. In mouse models, Bmal1 knockout in either LGR5⁺ CBCs or in villus and crypt epithelial cells downregulated Mex3a and Lgr5 expression, reduced CBC prevalence and increased the prevalence of BMI1 expressing cells in crypts. These findings suggest a BMAL1-centered intrinsic regulatory pathway contributing to LGR5⁺ CBC homeostasis.

Results

BMAL1 maintains LGR5 expression and ISC homeostasis in crypts

To determine the role of BMAL1 in LGR5⁺ CBCs, we established a tamoxifen (TAM) induced-LGR5-expressing cell-specific Bmal1 knockout mouse model [Lgr5-Cre⁺;Bmal1^fl/fl (LC⁺B^fl/fl)] by crossing B6.129P2-Lgr5^{tm1(cre/ERT2)Cle}/J and B6.129S4(Cg)-Arntl^tm1Weit/J mice (Fig. 1A). In duodenal crypts, Bmal1 knockout led to a 3.8-fold reduction of MEX3A expression (Fig. 1B–D). Quantification of EGFP⁺ cells at the crypt base found a 2.5-fold decrease of LGR5⁺ CBC-containing crypts in the LC⁺B^fl/fl mice compared to the LC⁺B^wt/wt control (Fig. 1B and E). It is well established that BMI1⁺  + 4 cells can convert into LGR5-expressing cells upon intestinal damage and LGR5⁺ CBC loss^6,7. Consistent with this notion, we also found a three-fold increase of Bmi1 RNA in the crypts of TAM-treated LC⁺B^fl/fl mice (Fig. 1F and G).

Figure 1

BMAL1 knockout reduces MEX3A expression and decreases LGR5⁺ CBC numbers but increases BMI1⁺ cells in the crypts. (A) Diagram of the LGR5⁺ CBC specific BMAL1 knockout mouse model. Lgr5CreERT2⁺; Bmal1^wt/wt (LC⁺B^wt/wt) and Lgr5CreERT2⁺; Bmal1^fl/fl (LC⁺B^fl/fl) male mice were fed with a tamoxifen (TAM) diet for 3 weeks to specifically knockout exon8 of the Bmal1 gene in LGR5-EGFP⁺ CBCs. (B–E) Immunofluorescence staining (B), quantification of BMAL1 and MEX3A expression in duodenum crypts (C, D) and prevalence of EGFP⁺ crypts (E) from TAM fed male mice at ZT5 (zeitgeber time (ZT), ZT0 corresponds to lights on; ZT12 corresponds to lights off). Scale bar indicates 10 μM. (n = 3 mice, ≥ 50 crypts/n). (F) RNAscope analysis of Lgr5 (pink dots) and Bmi1 (red dots) mRNA in duodenum crypts of TAM fed LC⁺B^wt/wt and LC⁺B^fl/fl male mice at ZT5. Scale bar indicates 10 μM (n = 3 mice). (G) Quantification of Bmi1 mRNA dots from (F). (n = 3 mice, ≥ 50 crypts/n). (H–L) Immunohistochemistry (IHC) of BMAL1 and RNAscope analysis of Lgr5, Mex3a and Bmi1 mRNA in intestinal crypts of Vil-Cre⁺B^wt/wt and Vil-Cre⁺B^fl/fl male mice at ZT5 (H). Scale bar indicates 60 μM. Quantification of BMAL1⁺ cell prevalence (I), mRNA expression of Mex3a (J) and Lgr5 (K), and prevalence of BMI1⁺ cells (L) in intestinal crypts of Vil-Cre⁺B^wt/wt and Vil-Cre⁺B^fl/fl male mice at ZT5. (n = 3 mice, 100 crypts/n). All quantification data are means ± SD, significant differences are based on unpaired T-tests. * P < 0.05, ** P < 0.01, *** P < 0.001, ****P < 0.0001.

To confirm that this was not a LC⁺B^fl/fl mouse specific phenomenon, we further generated an intestinal villus and crypt epithelial cell-specific Bmal1 knockout mouse model [Vil-Cre⁺;Bmal1^fl/fl (Vil-Cre⁺B^fl/fl)] by crossing B6.Cg-Tg(Vil1-cre)1000Gum/J and B6.129S4(Cg)-Arntl^tm1Weit/J mice. Consistent with the results from the LC⁺B^fl/fl mouse model, Bmal1 knockout significantly reduced Mex3a and Lgr5 expression (Fig. 1H–K) and increased Bmi1 expression in intestinal crypts of the Vil-Cre⁺B^fl/fl mice (Fig. 1H and L). These observations indicated that, even without external stress or tissue damage, loss of BMAL1 perturbed the homeostasis between LGR5⁺ CBCs and BMI1⁺  +4 cells in the intestine.

We previously found that human MEX3A inhibited ferroptosis in epithelial ovarian cancer cells³⁰. Consistent with those observations, Mex3a depletion in an immortalized mouse intestinal epithelial cell line (mIEC) significantly increased lipid peroxidation (Fig. 2A) and intracellular Fe²⁺ levels (Fig. 2B), indicators of ferroptosis. Bmal1-downregulated intestinal cells also showed significant increase in both lipid peroxidation (Fig. 2C) and ferroptosis (Fig. 2D), consistent with BMAL1 regulation of Mex3a expression. This stress phenotype in Bmal1 and Mex3a downregulated cells was likely due to reduced level of Glutathione peroxidase 4 (GPX4), a phospholipid hydroperoxidase that protects cells against membrane lipid peroxidation³¹ (Fig. 2E). Thus, Bmal1-depletion mediated ferroptosis which in turn contributed to the decrease in LGR5⁺ CBCs. Together, these results suggest a critical role of BMAL1 in maintaining the ISC homeostasis.

Figure 2

Mex3a or Bmal1 depletion in mIECs elevated lipid peroxidation and ferroptosis stress. (A) Lipid peroxidation level in the sh-Ctrl or sh-Mex3a mIEC cells detected by BODIPY-C11 staining. Relative BODIPY-C11 mean fluorescence intensity is presented as percent of control. Three independent experiments were performed and data are means ± SD from one representative experiment (n = 5). Significant differences are based on unpaired T-test (**P < 0.01, ****P < 0.0001). (B) Representative images of FerroOrange staining using the sh-Ctrl or sh-Mex3a mIEC cells. Scale bar indicates 20 μm. Relative FerroOrange mean fluorescence intensity is presented as fold change of control. Three independent experiments were performed and data are means ± SD from one representative experiment (n = 3). Significant differences are based on unpaired T-test (****p < 0.0001). (C) Lipid peroxidation level in the sh-Ctrl or sh-Bmal1 mIEC cells detected by BODIPY-C11 staining. Data formatting is as described for (A). (D) Representative images of FerroOrange staining using the sh-Ctrl or sh-Bmal1 mIEC cells. Scale bar indicates 20 μm. Data formatting is as described for (B). (E) IB of GPX4, SLC7A11 and SOD1 protein expression using sh-Ctrl, sh-Bmal1 or sh-Mex3a knockdown mIEC cells. GAPDH was used as a loading control. Blots shown are from one representative experiment of two replicates. The original blots and additional experiments with similar results are shown in the “Supplementary information” file.

BMAL1 and MEX3A co-regulate Lgr5 expression in intestinal epithelial cells

To further examine the relationship between BMAL1, Mex3a and Lgr5 expression in intestinal cells, Bmal1-depleted mIEC cells were subjected to immunoblotting assays. A decrease in MEX3A and LGR5 protein level was observed upon Bmal1 knockdown (Fig. 3A and B). Consistent with the observation that Mex3a knockout decreased LGR5⁺ CBCs¹¹, we also found that depletion of Mex3a with shRNA decreased LGR5 protein levels (Fig. 3A and C). Also, overexpression of either BMAL1 in Mex3a-depleted cells or MEX3A in Bmal1-depleted cells (using pBMAL1-His or pMEX3A-Flag plasmids) partially recovered the reduction of LGR5 caused by Bmal1 or Mex3a knockdown (Fig. 3D). Together, these data indicated that both BMAL1 and MEX3A contribute to LGR5 expression and that BMAL1 regulates LGR5 expression via at least two pathways. In one of these pathways, BMAL1 acts via Mex3a expression while in the other BMAL1 may directly regulates Lgr5 transcription (Fig. 3E).

Figure 3

BMAL1 controls LGR5 expression in part by regulating MEX3A. (A) Immunoblot of BMAL1, MEX3A and LGR5 in mIEC transduced with sh-Bmal1 or sh-Mex3a lentivirus. GAPDH was used as an internal control. Blots shown are from one representative experiment of three independent experiments. The original blots and additional experiments with similar results are shown in the “Supplementary information” file. (B, C) Immunoblot of LGR5 in sh-Bmal1 (B) or sh-Mex3a (C) transduced mIEC. GAPDH was used as an internal control. Blots shown are from one representative experiment of two independent experiments. Numbers below the blot are the quantification of band intensity relative to the sh-Ctrl. The original blots and additional experiments with similar results are shown in the “Supplementary information” file. (D) Immunoblot of BMAL1, MEX3A and LGR5 in sh-Bmal1 or sh-Mex3a transduced mIEC cells with MEX3A or BMAL1 overexpression, respectively. GAPDH was used as an internal control. Blots shown are from one representative experiment of two independent experiments. Numbers below each blot are the quantification of band intensity relative to the sh-Ctrl control. The original blots and additional experiments with similar results are shown in the “Supplementary information” file. (E) qRT-PCR of Lgr5 in mIEC cells transduced with sh-Ctrl or sh-Bmal1 lentiviruses. Three independent experiments were performed. Data show means ± SD based on one-way ANOVA. *** P < 0.001.

BMAL1 directly upregulates Mex3a transcription

Given the close relationship between Bmal1 depletion and MEX3A protein and mRNA downregulation (Figs. 3A, 4A), we analyzed the Mex3a promoter using the JASPAR database³² and found a putative BMAL1 binding site [GCTTTCCAC; -678~-669 nt upstream of the transcriptional start site (TSS)] (Fig. 4B). Even though this site is not a complete match to the BMAL1 consensus binding site [E-box (E1, CACGTG) or E1-E2 (E2, AACGTG) tandem sites]³³, chromatin-immunoprecipitation (ChIP) assay using mIEC cells found BMAL1 associated with this Mex3a promoter region (Fig. 4C). Biotinylated-oligonucleotide-pull-down assay demonstrated that BMAL1 directly and specifically bound to this site, as BMAL1 bound an oligo probe containing this putative site but not a mutated probe (Fig. 4D). Transient reporter assay showed that mutation of this site abolished BMAL1 transactivation of the Mex3a promoter in HEK-293T cells (Fig. 4E). Consistent with previous reports that BMAL1-CLOCK heterodimers¹⁹ enhance their transcriptional activity by interacting with chromatin-modifying proteins such as JARID1A³⁴ and CBP/p300³⁵, BMAL1 interacted with these factors in mIEC cells (Fig. S1). The BMAL1-CLOCK-JARID1A-p300 complex was on the Mex3a promoter to activate Mex3a transcription (Fig. 4F–H). Together, these results indicated that BMAL1 directly activates Mex3a expression.

Figure 4

BMAL1 transcriptionally upregulates Mex3a. (A) qRT-PCR analysis of Mex3a in sh-Ctrl or sh-Bmal1 mIEC cells. GusB was used as an internal control. Three independent experiments were performed and data are means ± SD from one representative experiment. Significant differences are based on unpaired T-test. (n = 3, ****P < 0.0001). (B) Diagram shows one predicted BMAL1 binding site on the mouse Mex3a promoter using the JASPAR database. (C) ChIP analysis of BMAL1 occupancy on the Mex3a promoter in mIEC cells. Mouse IgG and far site control region (FSC; Chr 3, 88,521,891–88,522,294) were used as negative controls. Gel images of BMAL1 occupancy on P1 and P2 regions are from one representative experiment of two independent experiment. The original gel images and additional experiments with similar results are shown in the “Supplementary information” file. (D) Biotinylated oligo pull-down of BMAL1 using WT or mutant predicted BMAL1 binding sequence and mIEC nuclear extract. Blots shown are from one representative experiment of three independent experiments. The original blots and additional experiments with similar results are shown in the “Supplementary information” file. (E) Top: Schematic diagram of pGL4.10-Mex3a promoter-Luc reporter with WT BMAL1 binding sequences. Bottom: Luciferase reporter assays were conducted using 293T cells co-transfected with pcDNA3.0-BMAL1 and the WT (black bars) or mutant (MUT, blue bars) Mex3a promoter constructs. Three replicate experiments were performed. Data show means ± SD based on two-way ANOVA. ** P < 0.01, *** P < 0.001. (F–H) ChIP analysis of CLOCK (F), JARID1A (G) and p300 (H) on the Mex3a promoter using nuclear extract from mIEC cells. IgG and far site region (FSC) were used as an IP control. Gel images of CLOCK, JARID1A and p300 occupancy on P2 region are from one representative experiment of two independent experiments. The original gel images and additional experiments with similar results are shown in the “Supplementary information” file. (I) RNA-IP assay of MEX3A-Lgr5 complex using mIEC lysates. IgG was used as an antibody control for IP. Three independent experiments were performed. Data are presented as means ± SD, significant differences are based on Student’s T test. * P < 0.05. (J) Top: Schematic of mouse Lgr5 mRNA and 11 biotinylated RNA probes (P1–P11) spanning nearly whole Lgr5 mRNA sequence used for RNA pull-down assay. Bottom: RNA pull-down assay of MEX3A using 11 biotinylated RNA probes that covered the Lgr5 coding regions (CDS) and 3′-UTR. Beads only and the Gapdh RNA probe were used as pull-down controls. Blot shown is from one representative experiment of three independent experiments. The original blots and additional experiments with similar results are shown in the “Supplementary information” file. (K) 5-bromouridine (BrU) IP chase assay (BRIC) of Lgr5 mRNA in mIEC cells transfected with pcDNA3.0 control vector or pMex3a-3X flag (MEX3A OE). Lgr5 levels at 0 h was set as 100%. Two independent experiments were performed. Data show means ± SD, significant differences are based on two-way ANOVA. ** P < 0.01, *** P < 0.001. (L) qRT-PCR analysis of Lgr5 levels in mIEC transduced with sh-Ctrl or sh-Mex3a lentiviruses. Three independent experiments were performed. Data show means ± SD, significant differences are based on one-way ANOVA. ** P < 0.01, *** P < 0.001.

MEX3A binding stabilizes Lgr5 mRNA

Even in BMAL1 expressing mIEC cells, Mex3a depletion still reduced LGR5 level (Fig. 2C), suggesting that MEX3A can affect LGR5 levels independently of BMAL1. RNA immunoprecipitation (RNA-IP) assay showed that MEX3A bound to Lgr5 mRNA (Fig. 4I). This suggested that MEX3A may regulate Lgr5 expression post-transcriptionally. Since the RNA motif and secondary structure recognized by MEX3A is not known, we used eleven biotinylated-RNA probes (400–500-nt in length per probe) spanning the Lgr5 mRNA including the 3′-UTR (Fig. 4J) to perform biotinylated-RNA pull-down assay. The results showed that MEX3A interacted with Lgr5 mRNA at multiple sites, including two in the coding region (CDS, probes 3, 5) and one in the 3′-UTR (probe 9) (Fig. 4J). MEX3A binding stabilized Lgr5 mRNA as MEX3A overexpression significantly increased the half-life of Lgr5 mRNA from 4 to 22.9 h in mIEC (Fig. 4K). Consistent with these observations, depletion of MEX3A reduced Lgr5 mRNA level in mIEC cells (Fig. 4L). Together, these data indicated that BMAL1 could directly activate Lgr5 transcription and indirectly stabilize Lgr5 mRNA by upregulating MEX3A.

Discussion

The circadian clock modulates several aspects of intestinal physiology including IEC regeneration³⁶, microbiota-IEC crosstalk³⁷, and intestinal permeability³⁸. BMAL1 is essential for maintaining rhythmic cell division within intestinal organoids¹⁷ and for coordinating intestinal regeneration¹⁶. Despite some controversy of whether LGR5⁺ CBCs are dispensable in intestinal tissue regeneration and tissue homeostasis⁷, convincing evidence has shown that LGR5⁺ CBCs are required to maintain a healthy intestinal epithelium⁵. Our results show that BMAL1 is essential for Lgr5 expression in the crypt cells and therefore may contribute to maintaining intestinal stem cell homeostasis. Using two Bmal1 knockout mouse models, we showed that ablation of Bmal1 reduced Mex3a and Lgr5 expression as well as LGR5⁺ CBC numbers but increased BMI1⁺ cells in the crypts (Fig. 5). We further demonstrated that BMAL1 increased MEX3A to stabilize Lgr5 mRNA in mIEC cell line. Our results, together with the known role of MEX3A in maintaining the LGR5⁺ CBC pool¹¹, suggest a possible role of BMAL1 in regulating intestinal stem cell succession, and thereby contributing to intestinal homeostasis. Intriguingly, we did not observe growth deficiency in BMAL1 knockout intestinal epithelium despite the significant loss of LGR5⁺ CBCs (Fig. 1). This phenotypic observation is consistent with the previous report that BMAL1 knockout mice had no obvious difference in tissue growth but did have reduced numbers of intestinal LGR5⁺ CBC and crypts¹⁶. Interestingly, we observed some LGR5/GFP^- crypt cells expressing BMAL1 in the LC⁺B^fl/fl intestine (Fig. 1B). A possible explanation is that BMI1⁺  +4 cells turned on BMAL1 expression to facilitate reprogramming and thus replenish the loss of LGR5⁺ CBCs in the crypts. However, the mosaicism in the Lgr5-Cre mouse model in which GFP expression does not reflect the actual LGR5 expression level in the crypts³⁹ makes it difficult to determine whether BMAL1 detected in the crypt cells was due to BMI1⁺  +4 cells which underwent reprograming to replenish the loss of LGR5⁺ cells or due to incomplete knockout in the LGR5⁺ CBCs. A careful lineage tracing of LGR5⁺ CBCs and BMI1⁺  +4 cells after Bmal1 is conditionally knocked out, either in vivo or using enteroids, will be required to more conclusively investigate this point and to further evaluate if BMAL1 and MEX3A promote BMI1⁺  +4 cell reprogramming to LGR5-expressing cells. Nevertheless, our mechanistic findings have multiple implications for how circadian time may affect intestinal damage and recovery in response to various stimuli.

Figure 5

Model of BMAL1 function in maintaining intestinal homeostasis. In intestinal crypts, BMAL1 is crucial to maintain homeostasis between the fast-proliferating LGR5⁺ CBCs and damage-resistant BMI1⁺  +4 cells. BMAL1, together with CLOCK, JARID1A and p300, upregulated expression of Mex3a; while MEX3A further enhanced LGR5 expression by directly binding and stabilizing Lgr5 mRNA. Ablation of Bmal1 not only reduced both Mex3a and Lgr5 expression, but also increased lipid peroxidation (LPO) and ferroptosis, resulting in decreased LGR5⁺ CBC prevalence but increased BMI1 expressing cells in crypts.

In addition to regulating Mex3a and Lgr5 expression, BMAL1 may also contribute to crypt homeostasis by protecting crypt cells from lipid peroxidation and ferroptosis. Several lines of evidence have shown that BMAL1 dysregulation results in ferroptosis; however, the underlying mechanisms vary among different cell types. For example, in fibrosarcoma and non-small-cell lung cancer cell lines, autophagy-mediated BMAL1 degradation facilitates EGLN2 expression to destabilize HIF1A which ultimately increases lipid peroxidation⁴⁰. In pancreas, conditional knockout of Bmal1 inhibits antioxidant defense genes such as SLC7A11 (an antiporter mediating the uptake of extracellular cystine in exchange for glutamate), GPX4 and superoxide dismutase 1 (SOD1) and consequently increases ferroptosis-associated pancreatitis⁴¹. Consistent with these reports, we also found that Bmal1 depletion in mIEC cells significantly elevated lipid peroxidation and intracellular Fe²⁺ levels (Fig. 2C and D), likely via downregulation of GPX4 (Fig. 2E). In a parallel manner, Mex3a knockdown also led to GPX4 suppression and ferroptosis elevation (Fig. 2E). Whether BMAL1 and MEX3A directly regulate GPX4 or other targets in ferroptosis pathways remain to be elucidated. Interestingly, Mex3a deficiency has been shown to upregulate the peroxisome proliferator‐activated receptor γ (PPARγ) pathway¹¹. The PPAR family, including PPARα, PPARδ and PPARγ, have all been suggested to decrease ferroptosis^42,43,44. As ferroptosis is regulated by several pathways⁴⁵, it is unclear how GPX4, the primary enzyme that prevents ferroptosis³¹, and PPAR or other lipid peroxidation regulatory pathways interact in the intestinal crypt cells under Bmal1 depleted conditions. This topic requires further investigation. Moreover, depletion of Bmal1 or Mex3a can result in apoptosis^46,47 or alter tissue inflammatory status⁴⁸ which consequently may lead to pyroptosis⁴⁹. Whether BMAL1 also contributes to crypt homeostasis by preventing these cell death pathways will also be of interest for further investigation.

It should also be noted that the role of BMAL1 in colonic homeostasis has been described using intestinal tissue specific Bmal1 knockout and Apc^min/+ mouse models^50,51. BMAL1 is critical for colonic epithelial regeneration which contributes to colitis recovery⁵⁰. In Apc^+/+ and Apc^min/+ mice lacking Bmal1, BMAL1 was found to regulate intestinal stem cell pathways, such as Hippo signaling⁵¹. Loss of Bmal1 perturbed the homeostasis of stem cell population and increased tumor initiation⁵¹. However, to our knowledge, the regulatory relationship between BMAL1, MEX3A and Lgr5 has not been described in the colon. Whether the BMAL1-MEX3A-Lgr5 regulatory mechanism identified in duodenal crypts is also applicable in colonic crypts remains to be investigated. Intriguingly, colorectal cancer stem cells have high Lgr5 expression⁵². Whether these cancer stem cells are able to upregulate Lgr5 expression in a BMAL1 independent manner and how BMAL1, MEX3A and Lgr5 regulatory relationship in colorectal cancer stem cells differs from that in non-cancerous intestinal tissue will also be of interest for further investigation.

Consistent with previous observations that MEX3A is essential for the survival of LGR5⁺ cells and organoid forming abilities¹¹, our study further demonstrated that MEX3A is transcriptionally upregulated by BMAL1 (Fig. 4) and plays an essential role to co-regulate ISC homeostasis with BMAL1 (Figs. 1 and 2). Moreover, LGR5⁺ CBCs can be divided into two subpopulations with different proliferation rate and stress response based on MEX3A expression level¹⁵. The MEX3A^high-LGR5⁺ cells are more stress resistant and critical for intestinal epithelium repair and homeostasis upon toxic insults¹⁵. Since BMAL1 drives rhythmic gene expression and regulates biological functions under circadian control⁵³, it is possible that the expression of MEX3A, and therefore the prevalence of the MEX3A^high-LGR5⁺ cells, also fluctuate with BMAL1 oscillation through the day-night cycle. Interestingly, in C57BL/6J mice, BMAL1 protein oscillated in a period shorter than 24 h in the duodenum crypt cells and Mex3a expression oscillated with a pattern that coincided with BMAL1 expression [p-value 0.038 using cosinor analysis (https://cosinor.online, period length = 12 h)] (Fig. S2). However, Lgr5 mRNA did not have a statistically significant oscillation (p = 0.1, period length = 12 h). How and why the crypt cells exhibited a shorter period of BMAL1 oscillation remains to be determined. Recently, a few studies have used enteroid culture and colon tumor organoids from transgenic mice to examine circadian rhythms and BMAL1 regulated genes. In these published RNA-seq data, Mex3a rhythmic expression was not detected^51,54. Enteroid culture is currently the best in vitro model to mimic intestinal epithelium. However, each enteroid contains multiple epithelial cell types and the proportion of stem cells is low. Thus, the oscillation of Mex3a transcript level in stem cells is likely to have been masked due to the high stoichiometry of other cell types. As for colon tumor organoids, they have no self-autonomous clock function⁵¹ and thus no rhythmic Mex3a was detected. Therefore, techniques such as single cell RNA sequencing using intestinal crypts at different times of day will be required to evaluate whether Mex3a and Lgr5 oscillate in CBCs. Nevertheless, consistent with our mechanistic findings, tumor organoids derived from an Apc^min/− mouse model lacking Bmal1 (Apc^min/−; Bmal1^−/−) found that both Lgr5 and Mex3a were significantly downregulated compared to organoids derived from Apc^min/−; Bmal^+/+ tumors⁵¹ indicating that Lgr5 and Mex3a are regulated by BMAL1.

In metastatic colon cancer patients, several random trials have tested time-modulated combination treatment of 5-FU, oxaliplatin and leucovorin/folinic acid and found that drug delivery between 2 and 4 o’clock in the morning, instead of constant infusion, showed higher tumor objective response rates (> 50%)^55,56 and milder mucositis⁵⁶. The main goal of such chrono-therapeutic strategy is to deliver chemotherapeutic agents at specific circadian time points to maximize drug efficacy and minimize adverse effects⁵⁷. Our results indicate that drug delivery at the time of maximal BMAL1 and MEX3A expression in intestinal stem cells may allow intestinal side-effects of drugs such as 5-FU to be minimized. More detailed mechanistic studies are needed to determine whether this BMAL1-MEX3A-Lgr5 regulation is circadian dependent which may provide insights into the mechanisms of chrono-therapeutic response and could facilitate development of more effective treatment strategies for intestinal cancer and other diseases.

Materials and methods

Mouse breeding and maintenance

All animal breeding and methods were carried out in accordance with relevant guidelines and regulations, and were approved by the Academia Sinica Institutional Animal Care and Utilization Committee (AS IACUC# 17-12-1165). All mice were maintained under standard 12:12 h light/dark (LD) condition, ZT0 (7:00 AM) corresponded to lights-on and ZT12 (7:00 PM) corresponded to lights-off. To reduce animal use to fulfill the 3Rs (Replacement, Reduction and Refinement) principle, we examined BMAL1 protein level and observed a 24-h oscillation in the duodenum tissue (containing both crypt cells and differentiated epithelial cells in the villus) of C57BL/6J mice kept under LD condition (Fig. S3). BMAL1 protein expression was high during the light phase (peak at ZT5), but low during the dark period in the duodenum epithelium. We therefore decided to choose ZT5 for our animal experiments. To specifically knockout Bmal1 expression in LGR5⁺ CBCs in the intestine, Lgr5-EGFP-ires-CreERT2 (referred as Lgr5-Cre; JAX stock no. 008875) mice² were crossed with Bmal1 floxed mice with two loxP sites flanking exon8 of mouse Bmal1 (Bmal1^fl/fl, JAX stock no. 007668)⁵⁸. To improve animal welfare, repetitive intraperitoneal administration or gavage which can cause stress, injury or local inflammation to the mice^59,60 was avoided. The progeny (Lgr5-Cre⁺; Bmal1^fl/fl, Lgr5-Cre⁺; Bmal1^wt/wt) were fed with tamoxifen (TAM) diets (ENVIGO, Cat#TD130856) for 3 weeks to activate the CreERT2 recombinase (Fig. 1A). To knockout Bmal1 expression in intestinal epithelial cells, B6.Cg-Tg(Vil1-cre)1000Gum/J (referred as Vil-Cre, JAX stock no. 021504) mice⁶¹ were crossed with Bmal1^fl/fl mice to generate Vil-Cre⁺;Bmal1^fl/fl (Vil-Cre⁺B^fl/fl) mice. All experiments were performed using duodenum (approximately 1 cm distal to pyloric sphincter) of male mice at 8–12 weeks of age. Euthanasia was performed by carbon dioxide inhalation. All methods are reported in accordance with ARRIVE guidelines (https://arriveguidelines.org).

Cell lines

Small intestinal epithelial cell line (mIEC) from mouse E19.5 fetuses [⁶², Insreenex Cellular Screening Technologies, Cat# INS-CI-1007] was maintained in muINTEPI medium (Cat# INS-ME-1005) with basal supplement (Cat# ME1005BS) in Type I collagen coated plates. Cells were cultured at 37 °C in a 5% CO₂ humidified incubator.

Plasmids and reagents

The lentiviral pLKO-puro-shRNA expression vectors sh-Ctrl (TRC005), sh-Bmal1#1 (TRCN282303), sh-Bmal1#2 (TRCN282305), sh-Mex3a#1 (TRCN255045) and sh-Mex3a#2 (TRCN255048) were from the National RNAi Core Facility (Academia Sinica).

pcDNA3.Bmal1-His was made by inserting Bmal1-His cDNA from pBMPC3 (pBmal1-His) (Addgene, Cat# 31367). pMex3a-3X Flag was from GeneCopoeiaTM (Cat# EX-Mm32128-M14). All constructs were verified by DNA sequencing.

Immunoblot assay

Whole cell lysate was prepared using RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris pH8.0, SIGMAFAST™ Protease Inhibitors (Sigma-Aldrich, Cat# S8830-20TAB) and PhosSTOP EASYpack phosphatase inhibitor cocktail (Roche, Cat# 4906837001). Protein concentration was determined by Bradford assay (Bio-Rad, Cat# 5000006). 50–30 μg total proteins were separated by 7.5% or 10% tris–glycine polyacrylamide gel, with overnight incubation with primary antibodies (Table S1) and followed by a 1:10,000 dilution of horseradish peroxidase (HRP)-conjugated secondary antibodies. HRP signals were detected using Western Lightning® Ultra Chemiluminescence Substrate (PerkinElmer INC., Cat# NEL113001EA) and images captured by a UVP ChemStudio Plus BioImaging system. The densitometry of blot bands was quantified using Image Lab 6.0 software.

Immunohistochemical (IHC) and Immunofluorescence (IF) staining

Tissues were fixed in 10% formalin overnight at 4 °C and embedded in paraffin. Sections were cut into 4 μm slices, dewaxed with xylene and rehydrated with descending ethanol series to water. Antigen retrieval was performed using the citrate-based antigen unmasking solution (Target Retrieval Solution (10X), Dako, Cat# S1699) for 20 min under high pressure condition. Slices were stained with primary antibodies (Table S1) overnight at 4 °C, followed by incubation with DAKO REAL™ EnVision™ HRP Labeled Polymer Anti-Rabbit (Cat# K4003) at room-temperature for 30 min. The 3.3′-diamiobenzidine (DAB) substrate was used to detect the peroxidase activity, and the slices were counterstained with hematoxylin. Slides were photographed under 40× magnification by the Aperio scanner machine (Leica Biosystems, Singapore). For IF staining, tissues were hybridized with mouse anti-BMAL1 (1:50), rabbit anti-MEX3A (1:50) or chicken anti-GFP (1:500) primary antibodies overnight at 4 °C after de-paraffined, rehydration and antigen retrieval. After washed by PBST three times, the slices were incubated with anti-mouse (1:1000), anti-rabbit polymer-HRP (1:1000) or anti-chicken CF488A (1:200) secondary antibodies at room-temperature for 1 h. To enhance the fluorescence signal, the anti-BMAL1 and anti-MEX3A hybridized slides were further labeled with TSA™ plus cyanine 3 (1:1500, PerkinElmer INC., Cat# NEL744B001KT) or TSA™ plus cyanine 5 (1:1500, Cat# NEL745B001KT), respectively. Images were obtained by a laser scanning confocal microscope (LSM770, Carl Zeiss MicroImaging). The fluorescence intensity was quantified using the QuPath software.

RNAscope in situ hybridization

Lgr5, Mex3a and Bmi1 mRNA expression were detected using either RNAScope® Multiplex Fluorescence Detection Reagent V2 kit (Advanced Cell Diagnostics, Cat# 323110) or RNAScope® 2.5 HD Detection Reagent-Brown kit (Cat# 322310) with Mm-Lgr5 (Cat# 312171), Mm-Mex3a-E2-CDS (Cat No. 318551) and Mm-Bmi1-O1 (Cat No. 466021) probes according to the manufacturer’s instructions. Quantification of mRNA positive spots were performed using the trainable Weka segmentation classifier in Fiji software according to the recommendation of TECHNICAL NOTE from Advanced Cell Diagnostics.

Flow cytometry (FACS) for lipid peroxidation analysis

For lipid peroxidation detection, cells were stained with 0.75 μM BODIPY 581/591 C11 (Thermo Fisher Scientific, D3861) for 30 min at 37 °C. Oxidation of the polyunsaturated butadienyl structure in BODIPY-C11 was measured by fluorescence emission at 600 nm with excitation at 570 nm. FACS analysis was performed using a FACS Canto II system (BD Biosciences).

FerroOrange staining

One million cells were seeded in a 35 mm plate and cultured overnight before subjected to FerroOrange staining according to the manufacturer’s instructions. In brief, cells were washed with HBSS (Gibco, Cat# 14175-095) three times and then stained with 1 µmol/L FerroOrange (Dojindo, Cat# F374) in serum free medium for 15 min at 37 °C. Images were acquired using an Andor Dragonfly 202 high speed confocal microscope system and the level of labile iron (II) ions (Fe2⁺) was measured by fluorescence emission at 572–615 nm with excitation at 561 nm. The fluorescence intensity was quantified using the Imaris 9.9 software.

Quantitative real-time PCR (qRT-PCR)

RNA was extracted using TRI Reagent (Sigma-Aldrich, Cat# T9424) according to the manufacturer’s instructions. For mRNA expression detection, 3 µg total RNA from each sample was reverse-transcribed using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Cat# K1622). 10 ng 1st-strand cDNA was used for real-time PCR with appropriate primer sets and SYBR® Green PCR Master Mix (Thermo Fisher Scientific, Cat# 4309155) using ABI-7900 thermocycler. The mRNA relative quantities were determined using comparative cycle threshold methods with Beta-glucuonidase (GusB) as an internal control. Primers for qRT-PCR were 5′- GCAGGCAAGGCTGCAAGATT -3′ (forward) and 5′- ACTTGTTGCGTGAGGCTCTT -3′ (reverse) for Mex3a, 5′- CCTTCCCCAGGTCCCTTCAA -3′ (forward) and 5′-GAACACGGTCAAAGCCACCA-3′ (reverse) for Lgr5, and 5′- CCGACCTCTCGAACAACCG -3′ (forward) and 5′- GCTTCCCGTTCATACCACACC -3′ (reverse) for GusB.

Chromatin immunoprecipitation (ChIP) assay

Cells were crosslinked with 1% formaldehyde (Sigma-Aldrich, Cat# F8775) for 10 min at 37 °C and the crosslinking was stopped by 0.125 M glycine. After PBS wash, the crosslinked cells were lysed in 1 mL cell lysis buffer (10 mM Tris–HCL pH8.0, 1 mM EDTA, 0.5% NP-40, 1 mM PMSF, SIGMAFAST™ protease inhibitor cocktail, and Roche PhosStop) on ice for 20 min. Nuclei pellet was then resuspended and incubated in 1 mL of nuclear lysis buffer (50 mM Tris–HCL pH8.0, 100 mM EDTA, 1% SDS, 1 mM PMSF, SIGMAFAST™ protease inhibitor cocktail, and Roche PhosStop) on ice for 30 min and 3 freeze–thaw cycles to release chromatin. After centrifugation, the crosslinked chromatin was resuspended in 80 µL nuclear lysis buffers and sonicated using a Diagenode Bioruptor®pico (15 cycles of 30 s on and 30 s off) to 200–500 bases. A 10 µL aliquot of the fragmented chromatin was used as an input control. For immunoprecipitation assay, pre-washed protein A/G magnetic beads (Thermo Fisher Scientific, Cat# 88803) were incubated with 4 µg anti-BMAL1, 9 µg anti-CLOCK, 9 µg anti-p300, 6 µg anti-JARID1A, or control IgG antibody (Table S1) at 4 °C overnight. 25 µg sonicated chromatin was diluted to a final volume of 1 mL with dilution buffer (20 mM Tris–HCL pH 8.0, 150 mM NaCl, 1 mM EDTA, 0.01% SDS, 1% Triton X-100), and pre-cleaned with pre-washed protein A/G magnetic beads and 10 µg mouse-IgG isotype control at 4 °C overnight. The antibody-coupled beads were then incubated with the sonicated chromatin for 4 h at 4 °C. After washed several times, the chromatin-protein complex was reversed crosslinked by addition of 15 µL solution containing 2.5 M NaCl, 2 µL 50 mg/mL RNaseA and 6 µL 10 mg/mL proteinase K. DNA was extracted by MinElute® PCR purification kit (Qiagen, Cat# 28004). Semi-quantitative PCR was performed to detect BMAL1 associated promoter regions using primers listed in Table S2.

Luciferase reporter assay

Mus muscles strain C57BL/6JNarl chromosome 3 (88,530,849–88,533,098 bp) containing the proximal promoter of Mex3a, Mex3a exon1 and part of the coding sequence, was amplified by PCR from mIEC genomic DNA and cloned into pGL4.10 vector (Promega, Cat#E6651) (referred as pGL4.Mex3aP.WT). The putative BMAL1 binding site predicted by the JASPAR database was mutated from 5′-GCCTTTCCAC-3′ to 5′-TTTTTTTTAT-3′ (referred as pGL4.Mex3aP.MUT). pGL4.74 Renilla luciferase plasmid (pGL-4.74[hRluc/TK], Promega, Cat# E692A) was used as a transfection control. For HEK-293T cells, 6 × 10⁵ cells were seeded and co-transfected with 500 ng of pGL4.Mex3aP.WT or pGL4.Mex3aP.MUT, 6.25 ng of pGL-4.74[hRluc/TK], pcDNA3 or pcDNA3.Bmal1-His (range from 300 to 600 ng) using TransIT-LT1 transfection reagents (Mirus Bio LLC., Cat# MIR2300). Cell extracts were collected 48 h after transfection and the luciferase activity was measured using Dual-Luciferase® Reporter (DLR™) Assay System (Promega, Cat# E1910).

Co-immunoprecipitation (Co-IP) assay

Nuclear extraction was done using Subcellular Protein Fractionation Kit (Thermo Fisher Scientific, Cat# 78840) according to the manufacturer's instruction. For IP, 300 µg nuclear extract was incubated with 4 µg anti-BMAL1, 9 µg anti-CLOCK, 9 µg anti-p300, 6 µg anti-JARID1A, or control IgG antibody at 4 ℃ overnight. The antibody-protein complexes were precipitated using 20 µL pre-washed and pre-blocked (10% BSA) protein A/G agarose beads (Thermo Fisher Scientific, Cat# 20422) at 4 ℃ for 1 h. After washed with NP-40 TNE buffer (10 mM Tris–HCl pH7.5, 150 mM NaCl, 0.5 mM EDTA, 0.1% NP-40, and protease inhibitor cocktail) five times, interacting proteins were eluted with SDS-PAGE and detected by western blotting assay.

RNA immunoprecipitation (RIP) assay

Whole-cell lysate from 2 × 10⁷ mIEC was used for RIP assay using Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (Millipore, Cat# 17-700) according to the manufacturer’s instruction. 4 µg anti-MEX3A antibody was used for the immunoprecipitation at 4 ℃ overnight, and rabbit IgG was used as a pull-down control. MEX3A bound RNAs were then isolated following the procedures of proteinase K digestion and phenol–chloroform-isoamyl alcohol extraction. The precipitated RNAs were resuspended in 12 µL DEPC-ddH₂O. Quantitative PCR was performed to determine the level of MEX3A-bound Lgr5 mRNA.

RNA pull-down assay

Biotin-14-CTP (Thermo Fisher Scientific, Cat# 19519016) labeled RNA was in vitro transcribed using MEGAshortscript™ T7 High Yield Transcription Kit (Thermo Fisher Scientific, Cat# AM1354) according to the manufacturer's instruction. After purification with RNeasy Plus Mini Kit (Qiagen, Cat# 74136), 2 µg biotinylated RNAs was incubated with 20 µL pre-washed Dynabeads™ M-270 Streptavidin (Thermo Fisher Scientific, Cat# 65305) in 1 mL buffer (20 mM Tris–HCl (pH7.5), 100 mM KCl, 5 mM MgCl₂, and 0.5% NP-40), and incubated at 4 °C for 1 h with gentle agitation. 450 µg mIEC cytoplasmic protein lysate was added to the washed RNA-probe coupled dynabeads, and incubated in 1 × TENT buffer (10 mM Tris–HCl (pH8.0), 1 mM EDTA (pH 8.0), 250 mM NaCl, and 0.5% Triton X-100) with SIGMAFAST™ protease inhibitor cocktails and RNaseOUT™ Recombinant Ribonuclease Inhibitor (Thermo Fisher Scientific, Cat# 10777019) at 4 °C for 8 h. After washed with 1 × TENT buffer, the proteins interacted with mouse lgr5 probes were eluted using 40 µL SDS buffer and detected by immunoblot assay. Biotinylated Gapdh RNA was used as a pull-down control. All RNA-probes are listed in Table S3.

BRIC assay

mIEC was transfected with pcDNA3.0 or pMex3a-3X Flag and then incubated in the 5′-bromo-uridine (BrU, Sigma-Aldrich, Cat# 850187) containing medium to a final concentration of 1 mM in a humidified incubator with 5% CO₂ for 8 h. After removal of BrU, RNA was harvested at indicated time points using the TRI Reagent®. 15 µL pre-washed Dynabeads™ Protein G (Thermo Fisher Scientific, Cat# 10004D) was blocked with 2% BSA and 5 mg heparin (Santa Cruz, Cat# 203075) at 4 °C for 1 h, and then incubated with 4 µg anti-BrdU Ab at 4 °C for 1 h. The BrdU Ab coupled dynabeads were further washed using BrU-IP buffer and 250 mM NaCl three times and resuspended in 15 µL BrU-IP buffer supplemented with RNaseOUT™ Recombinant Ribonuclease Inhibitor. To normalize the immunoprecipitation of BrU-labeled RNAs among different time points, the in vitro transcribed BrU-labeled luciferase RNA was used as a spike-in control for internal standard of quantitative PCR assay. 50 μg BrU-labeled RNAs from mIEC and 100 ng BrU-labeled luciferase RNA were mixed in 1 mL BrU-IP buffer with RNaseOUT, and the RNA mixture was denatured by heating at 80 °C for 3 min. Subsequently, 15 µL BrdU Ab conjugated dynabeads were added into RNA mixture and incubated at room temperature for 1 h. After washed four times with BrU-IP buffer (20 mM Tris–HCL pH7.5, 250 mM NaCl, and 20 μL RNaseOUT), the BrU-labeled RNAs were isolated using RNeasy Plus Mini Kit. The half-life of lgr5 mRNA was determined by RT-qPCR assay.

Statistical analysis

Quantitative results were represented as Mean ± SD. Statistical analysis was conducted using the Graph Pad Prism 9.0 (GraphPad Software). Non-parametric Mann–Whitney test was used to compare control and experimental genotypes or treatment groups. Statistical significance among more than three groups was analyzed using one-way ANOVA with Dunn’s multiple comparison test. For the BRIC assay, two-way ANOVA with Sidak multiple comparison test was performed to detect significance among different groups. Asterisk (*, **, ***) indicates statistical significance with p-value < 0.05, 0.01, 0.001, respectively.