Ontogeny and function of the circadian clock in intestinal organoids.

Circadian rhythms regulate diverse aspects of gastrointestinal physiology ranging from the composition of microbiota to motility. However, development of the intestinal circadian clock and detailed mechanisms regulating circadian physiology of the intestine remain largely unknown. In this report, we show that both pluripotent stem cell-derived human intestinal organoids engrafted into mice and patient-derived human intestinal enteroids possess circadian rhythms and demonstrate circadian phase-dependent necrotic cell death responses to Clostridium difficile toxin B (TcdB). Intriguingly, mouse and human enteroids demonstrate anti-phasic necrotic cell death responses to TcdB. RNA-Seq analysis shows that ~3-10% of the detectable transcripts are rhythmically expressed in mouse and human enteroids. Remarkably, we observe anti-phasic gene expression of Rac1, a small GTPase directly inactivated by TcdB, between mouse and human enteroids, and disruption of Rac1 abolishes clock-dependent necrotic cell death responses. Our findings uncover robust functions of circadian rhythms regulating clock-controlled genes in both mouse and human enteroids governing organism-specific, circadian phase-dependent necrotic cell death responses, and lay a foundation for human organ- and disease-specific investigation of clock functions using human organoids for translational applications.

Introduction

3‐dimensional (3D) organoids are multicellular in vitro model systems possessing in vivo tissue architecture and function that is unattainable in homogenous single cell culture systems. Multiple tissues have been generated as organoids from a range of sources including pluripotent stem cells (PSC) and patient biopsies (Clevers, 2016). The source utilized for organoid generation influences the maturity of the end sample. For example, PSCs are differentiated through distinct developmental stages, definitive endoderm (DE) and midgut tube formation (MG), to generate human intestinal organoids (HIOs) (Fig 1A and B), a process mimicking the blastocyst, gastrula, somite, and fetal stages of intestinal development, respectively (Spence et al, 2011; Wells & Spence, 2014). The fetal nature of HIOs is highlighted by the absence of a full spectrum of markers for mature epithelial cell types (Watson et al, 2014; Finkbeiner et al, 2015). HIOs can be matured beyond their fetal stage by surgical implantation into the kidney capsule of immunocompromised non‐obese diabetic (NOD) scid gamma (NSG) mice. When transplanted, they become vascularized by the host circulatory system and develop a well‐defined crypt‐villus architecture. Transcriptome‐wide hierarchical clustering shows that transplanted samples are matured to an intermediate state between HIOs and adult intestinal tissue, containing a wider range of intestinal cell type markers compared to HIOs but not equivalent to adult intestinal tissue (Finkbeiner et al, 2015). Intestinal crypts from the transplanted human tissue can be isolated and cultured in vitro to generate mouse kidney capsule‐matured human intestinal enteroids (kcHIEs; Fig 1A). Analogously, crypts from human intestinal biopsies can be isolated and grown in vitro to generate biopsy‐derived HIEs (bHIEs; Mahe et al, 2015) (Fig 1A). Recent progress highlights the versatile utility of the organoid and enteroid model systems to investigate numerous complex biological questions including the molecular mechanisms maintaining the intestinal stem cell niche (Kretzschmar & Clevers, 2016), host–pathogen interactions (Forbester et al, 2015), cellular development (Sinagoga et al, 2018), and non‐pathogenic bacteria interactions (Hill et al, 2017). However, the development and physiological functions of circadian rhythms in human intestinal organoids/enteroids have not been investigated. In this report, we investigated the development of circadian rhythms from PSCs to kcHIEs and utilized bHIEs and mouse enteroids to characterize clock‐controlled genes (CCGs) and circadian functions in both mouse and human intestinal epithelium.

Figure 1

Maturation of HIOs is required for the development of robust circadian rhythms

Schematic summary for establishing 3D in vitro human intestinal organoids (HIOs—grey), kidney capsule matured human intestinal enteroids (kcHIEs—red), biopsy‐derived human intestinal enteroids (bHIEs—blue) or mouse enteroids (yellow).

Raw (top) and detrended (bottom) Bmal1‐luc bioluminescence data throughout directed differentiation of PSCs to HIOs, kcHIEs, and bHIEs.

Period length decreases with continual differentiation of PSCs into HIOs. Maturation of HIOs into kcHIEs leads to 24‐hour rhythms with greater reproducibility.

The amplitude of oscillations is significantly higher in kcHIEs/bHIEs compared to HIOs. Bmal1‐luc traces in B were generated by plotting the standard deviation of n ≥ 3 biological replicates.

Data information: Panels (C and D) depict mean ± S.D. of n ≥ 3 biological replicates. AU stands for arbitrary units.

Source data are available online for this figure.

Circadian rhythms exist throughout the body, providing temporal information to biological processes ranging from metabolism to cell proliferation. Molecularly, circadian rhythms are governed by a transcriptional–translational feedback loop (TTFL), which generates robust circadian oscillations at the single cell level (Nagoshi et al, 2004). Two basic helix–loop–helix (bHLH) PER‐ARNT‐SIM (PAS) domain containing proteins, CLOCK (circadian locomotor output cycles kaput) and BMAL1 (brain and muscle Arnt‐like protein‐1), form a heterodimeric circadian transcription factor that recognizes E‐box motifs on target gene promoters including Period (Per1, Per2, Per3) and Cryptochrome (Cry1, Cry2). PER and CRY bind together in the cytoplasm (Schmalen et al, 2014) and translocate into the nucleus where they interact with the CLOCK‐BMAL1 heterodimer to repress its activity (Lee et al, 2001). PER and CRY are then ubiquitinated and degraded, relieving CLOCK‐BMAL1 from negative feedback to re‐initiate the cycle (Reischl et al, 2007; Siepka et al, 2007; Yoo et al, 2013). CLOCK‐BMAL1 forms additional feedback loops with nuclear orphan receptor transcription factors. CLOCK‐BMAL1 drive rhythmic expression of positive (Rorα, Rorβ, Rorγ) and negative (Rev‐erbα, Rev‐erbβ) nuclear receptors that feedback on Bmal1 by competitively binding at ROR response elements (RORE) on the Bmal1 promoter (Preitner et al, 2002; Sato et al, 2004; Guillaumond et al, 2005).

The suprachiasmatic nucleus (SCN) is commonly referred to as the “master clock” of the mammalian system. Located in the hypothalamus, the SCN contains approximately 20,000 neurons that process photic signals to coordinate circadian rhythms in peripheral organs to external light/dark signals (Pando et al, 2002; Schibler et al, 2015). However, peripheral clocks are malleable and can be dissociated from SCN coordination by non‐photic circadian zeitgebers. For example, mice restricted to food during their inactive period (i.e., light phase or subjective day) realign the phase of clock and metabolic genes in the liver, but not SCN, to coincide with food availability (Damiola et al, 2000). Intriguingly, the intestinal microbiota changes over a circadian cycle (Thaiss et al, 2014), and perturbation of the microbiome may impact the function of peripheral circadian rhythms. Antibiotic‐induced depletion of microbiota induces constitutively high expression of Rev‐erbα that abolishes rhythmic Toll‐like receptor (TLR) expression (Mukherji et al, 2013). Loss of microbiota was also shown to change rhythmic gene expression within the mouse intestinal transcriptome while leaving behavioral rhythms intact (Thaiss et al, 2016). Notably, the recent development of tissue‐specific conditional Bmal1 rescue mice was instrumental in uncovering intact peripheral circadian rhythms in the skin epidermis and liver in the absence of circadian signaling from the SCN (Welz et al, 2019). These data highlight the importance of peripheral clocks and their ability to process input signals from local factors. More work is needed to uncover fundamental understanding of the conditions and mechanisms that integrate and dissociate the master and peripheral clocks. In this report, we utilized human intestinal organoids and enteroids to characterize the robustness of circadian rhythms during the development of intestinal tissue and investigated the functional role of the intestinal circadian clock in response to Clostridium difficile toxin B (TcdB) highlighting a critical function of the endogenous intestinal circadian clock.

Clostridium difficile is a spore‐forming, gram‐positive, toxin‐producing intestinal pathogen. C. diff. infection (CDI) manifests with diarrhea symptoms that can progress into pseudomembranous colitis and toxic megacolon (Kelly et al, 1994; Buffie et al, 2012). CDI pathogenesis includes the luminal release of two toxins, toxin A (TcdA) and B (TcdB), that bind to intestinal cellular receptors (Chen et al, 2018) and enter the cell via endocytosis (Papatheodorou et al, 2010). In the cytoplasm, autocatalytic cleavage releases the catalytic glucosyltransferase domain of the toxins (Reineke et al, 2007) that inactivate the small GTPases RHOA, RAC1, and CDC42, resulting in cellular cytoskeleton breakdown, disruption of epithelial barrier function, and cell death (Just et al, 1995; Abt et al, 2016). In addition, the NOX1 NADPH oxidase complex has also been implicated in promoting toxin‐mediated cell death at high concentrations of TcdB (Farrow et al, 2013). Several experimental models have been utilized to investigate C. difficile pathogenesis and each present limitation. In vivo rodent studies have been associated with low reproducibility attributed to confounding variables such as different C. diff. strains, antibiotic regimen to induce susceptibility, and cross contamination of animal housing (Best et al, 2012). The translatability of pathophysiological findings from rodent models to humans also cannot be guaranteed. In vitro and ex vivo studies are, comparatively, controllable but lack multicellular complexity or multi‐time point scalability, respectively. Human intestinal organoids/enteroids offer an opportunity to overcome these limitations by coupling in vitro experimental control with recapitulation of an in vivo human intestinal epithelium. 3D intestinal cultures have been successfully utilized to gain insights into the pathogenic response to human rotavirus (Saxena et al, 2016), E. coli (Noel et al, 2017 ), and Clostridium difficile toxins A/B (Leslie et al, 2015). These studies, however, did not investigate the influence of the circadian clock in the pathogenic response. Circadian clock‐mediated time of day dependency in the pathogenesis of Salmonella Typhimurium (Bellet et al, 2013) and herpes simplex virus (Edgar et al, 2016) has been observed in mice, but it is unknown whether humans show identical circadian time‐dependent responses via conserved molecular mechanisms. In this report, we utilized mouse and human intestinal enteroids to uncover potential differences with respect to circadian phase‐dependent necrotic cell death responses to TcdB.

Results

The development of a robust intestinal circadian clock requires maturation of HIOs

Mouse embryonic stem cells (ESC) do not possess circadian rhythms and have low expression of several canonical clock genes in comparison to NIH3T3 cells (Kowalska et al, 2010). Retinoic acid‐induced differentiation of mouse embryonic stem cells initiates autonomous circadian oscillations while reverting differentiated cells back to an induced pluripotent stem cell (iPSC) state abolishes circadian rhythms (Yagita et al, 2010). Subsequent studies revealed that during the early stages of cellular differentiation, PER1/PER2 are sequestered in the cytoplasm (Umemura et al, 2014) and the expression of CLOCK is downregulated by Dicer/Dgcr8‐dependent post‐transcriptional regulation of Clock mRNA (Umemura et al, 2017) disrupting autonomous oscillations of circadian rhythms. Identical to mouse ESCs and iPSCs, human ESCs and iPSCs lack circadian rhythms and show post‐transcriptional suppression of CLOCK protein expression (Dierickx et al, 2017; Umemura et al, 2019). Terminally differentiated cardiomyocytes from human ESCs demonstrate rhythmic gene regulation of CCGs, which regulate circadian time‐dependent responses to doxorubicin (Dierickx et al, 2017). Based on these data, we hypothesized that developmental maturation is a critical process for the development of robust circadian rhythms in 3D multicellular intestinal organoids.

To investigate the development of circadian rhythms in multicellular organoids, activity of circadian rhythms was assessed at different stages of development of human intestinal organoids and enteroids using Bmal1‐luciferase (Bmal1‐luc) reporter (Fig 1B). iPSCs were stably transduced with a plasmid containing Bmal1‐luc using lentivirus (Brown et al, 2005) and differentiated through two distinct stages: definitive endoderm (DE) and midgut tube (MG) formation (Spence et al, 2011). Bmal1‐luc reporter activity was recorded throughout differentiation with a luminometer (KronosDIOTM, ATTO). In agreement with previous reports (Umemura et al, 2019), iPSCs did not show circadian oscillations (Fig 1B (purple), Fig EV1A). Interestingly, we observe low‐intensity bioluminescence signals with long period lengths of about 50 and 34 h at the DE and MG stages, respectively (Fig 1B (green and orange), Fig EV1B and C). On the other hand, HIOs that were stably transduced with Bmal1‐luc plasmids demonstrated circadian oscillations with low bioluminescence output (Figs 1B (grey) and Fig EV1D). These data show that differentiation of iPSCs into multicellular HIOs promotes period length shortening, culminating with 24‐hour bioluminescent oscillations in HIOs (Fig 1C and D). However, circadian oscillations in these HIOs are not robust, lacking phase uniformity between independent experiments and unable to maintain oscillations longer than 48 h post‐synchronization of circadian rhythms (Figs 1B and EV1D). Although HIOs contain multiple intestinal epithelial cell types (Spence et al, 2011), they do not possess the full spectrum of epithelial cell markers, which reflect their immature, fetal‐like state. The stem and Paneth cell markers, Lgr5 and Lysozyme, respectively, are expressed in both HIOs and human tissue (Finkbeiner et al, 2015). The expression of secondary stem cell (Olfm4) and Paneth cell (Dll4, Reg3a) markers are, however, attenuated in HIOs, indicating they do not possess matured intestinal epithelial cell types (Finkbeiner et al, 2015) (Fig EV2A and B). Therefore, we transplanted HIOs into the kidney capsule of NSG mice and allowed them to mature over 3‐months (Watson et al, 2014) to test if maturation of HIOs could enhance the robustness of circadian rhythms.

Figure EV1

Detrended bioluminescent data from induced pluripotent stem cell (PSC), definitive endoderm (DE), midgut tube spheroids (MG), and HIOs with zoomed‐in Y‐axis

Induced pluripotent stem cells.

Definitive endoderm.

Midgut tube spheroids.

Human intestinal organoids.

Data information: AU stands for arbitrary units.

Figure EV2

Expression of Paneth cell markers, Dll4 and Reg3a, in HIO, kcHIE, and bHIE

Expression of Dll4.

Expression of Reg3a.

Source data are available online for this figure.

After 3 months of maturation, human intestinal crypts were isolated from the grafts to establish in vitro mouse kidney capsule‐matured human intestinal enteroids (kcHIEs; Mahe et al, 2015) (Fig 1A). In addition, we generated biopsy‐derived human intestinal enteroids (bHIEs) to compare circadian rhythms from matured human samples with HIOs and kcHIEs (Fig 1A). Both kcHIEs and bHIEs were transduced with the Bmal1‐luc plasmid and tested for rhythmic bioluminescent output. In contrast to HIOs, kcHIEs and bHIEs demonstrated robust circadian oscillations, demonstrated by sustained high amplitude oscillations of bioluminescent output from Bmal1‐luc reporter (Fig 1B). To validate the above bioluminescent data, we performed time course experiments collecting HIOs, kcHIEs, and bHIEs every 4 h over two circadian cycles for real‐time quantitative reverse transcription polymerase chain reaction (qRT–PCR). The expression of Bmal1, Rev‐erbα, and Per2 was significantly reduced and arrhythmic in HIOs (Fig 2A). In contrast, both kcHIEs and bHIEs demonstrated circadian rhythms with appropriate phase relationships between positive (Bmal1) and negative (Per2, Rev‐erbα) clock elements (Fig 2B and C). The average expression of these genes are not significantly different between kcHIEs and bHIEs except for Rev‐erbα (Fig 2D). Interestingly, the average amplitude of Bmal1, Per2, and Rev‐erbα are significantly lower in kcHIEs compared to bHIEs (Fig 2E), which suggest that bHIEs possess a more robust circadian clock compared to kcHIEs. Based on these data, we hypothesize that circadian time‐dependent responses to external perturbations will be observed in kcHIEs and bHIEs, but not in HIOs. As a proof of principle, we tested the circadian clock‐dependent response to Clostridium difficile toxin B (TcdB).

Figure 2

kcHIEs and bHIEs show circadian rhythms with appropriate phase relationships between canonical clock genes

mRNA expression of Bmal1 (black), Rev‐erbα (grey) and Per2 (red) in HIOs (A), kcHIEs (B), and bHIEs (C).

(D) Average expression of Bmal1 (black), Rev‐erbα (grey) and Per2 (red) in HIO, kcHIE, and bHIE. (E) Amplitude of Bmal1 (black), Rev‐erbα (grey) and Per2 (red). AU stands for arbitrary units.

Data information: The data in panels (A–C) are shown as mean (bold line) ± SD with replicates displayed as dashed lines. (D, E) Panels show mean ± SD of n = 3 biological replicates. *P < 0.05, **P < 0.01, ***P < 0.001 (Tukey’s multiple comparison).

Source data are available online for this figure.

The intestinal circadian clock in mouse and human enteroids regulate circadian phase‐dependent necrotic cell death responses to TcdB

Circadian clock‐dependent time‐of‐day variation in the pathogenic response has been demonstrated for Salmonella enterica serovar Typhimurium (Bellet et al, 2013), herpes and influenza A virus (Edgar et al, 2016) and Leishmania, in mice (Kiessling et al, 2017). However, circadian clock‐dependent responses to bacterial toxins in mouse or human enteroids have not been tested. It is critical to determine circadian clock‐dependent responses to different external perturbations (e.g., pathogens, toxins, and nutrients) in both mouse and human enteroids to compare and contrast their responses in order to develop potential translational applications using patient‐derived HIEs while taking advantage of various tools in transgenic mice. Therefore, we assessed circadian clock‐dependent responses following exposure to purified TcdB in both mouse and human enteroids.

Mouse enteroids were isolated from the tamoxifen‐inducible Bmal1 mouse (Yang et al, 2016). In vitro treatment of tamoxifen in Bmal1 enteroids successfully abolished circadian rhythms in contrast to control enteroids (Fig EV3A–D), which enabled us to use Bmal1‐floxed enteroids as arrhythmic‐negative controls. Circadian rhythms of Bmal1 enteroids were synchronized using dexamethasone, and 10 ng/ml of TcdB was added to the growth media at 24‐ (T24) and 36‐h (T36) post‐dexamethasone treatment (Fig EV4A). SYTOXTM Orange, a fluorescent dye that stains nucleic acids in cells with disrupted membranes, was added simultaneously with TcdB to quantify necrotic cell death at 2, 24, and 48 h post‐exposure of TcdB. In control Bmal1 enteroids, without tamoxifen treatments, we observe greater necrotic cell death in the T36 group compared to T24 at 48 h post‐TcdB treatments (Fig 3A and B). Identical data were observed in mouse enteroids derived from PER2::LUC mice (Fig EV4B and C). In contrast, tamoxifen‐treated Bmal1‐floxed enteroids showed a loss of circadian time‐dependent necrotic cell death responses to TcdB (Fig 3C and D), which indicates that the intestinal circadian clock regulates the observed circadian phase‐dependent phenotypes. Next, we used the aforementioned three human organoid models to test whether this phenotype from mouse enteroids is conserved in human organoids.

Figure EV3

Tamoxifen‐treated Bmal1 mouse enteroids shows arrhythmic clock gene expression

Control mouse enteroids, Bmal1 / ‐EsrCRE without tamoxifen treatment, show rhythmic Bmal1 (black) and Rev‐erbα (grey) gene expression with out‐of‐phase profiles.

Bmal1 (black) and Rev‐erbα (grey) gene expression are low and arrhythmic in tamoxifen‐treated Bmal1 / ‐EsrCRE mouse enteroids. Thick lines are mean ± SD of n = 3 biological replicates depicted as dotted lines.

Source data are available online for this figure.

Figure EV4

The circadian phase‐dependent necrotic cell death response to TcdB is observed in control PER2::LUC mouse enteroids

Schematic summary of the experimental design to test for a circadian phase‐dependent response to TcdB in vitro. Separate samples were synchronized either 36 h or 24 h prior to TcdB exposure to generate two sample groups 12 h out‐of‐phase at the time of TcdB addition.

Representative images of necrotic cell death (SYTOX orange—red fluorescence) at 2, 24, and 48 h post‐exposure to 10 ng/ml TcdB in PER2::LUC enteroids.

Quantitative analysis of fluorescent intensity from SYTOX orange in PER2::LUC mouse enteroids. PER2::LUC enteroids show greater necrotic cell death in the 36‐hour group compared to the 24‐hour group. Data represented as mean ± S.D. of n = 4 biological replicates (i.e., enteroids derived from different mice). All data were normalized to time‐ and phase‐matched vehicle controls set to 1.**P < 0.01. Scale bar = 250 μM.

Source data are available online for this figure.

Figure 3

Mouse enteroids show circadian phase‐dependent necrotic cell death response to Clostridium difficile toxin B

Representative images of necrotic cell death (SYTOX orange ‐ red fluorescence) at 2‐, 24‐ and 48 h post‐exposure to 10 ng/ml TcdB in control Bmal1 enteroids. Scale bar = 250 μm.

Quantitative analysis of fluorescent intensity from SYTOX orange in control Bmal1 enteroids without tamoxifen treatment.

Representative images of necrotic cell death (SYTOX orange) at 2, 24, and 48 h post‐exposure to 10 ng/ml TcdB in tamoxifen‐treated circadian arrhythmic Bmal1 enteroids. Scale bar = 250 μm.

Quantitative analysis of fluorescent intensity from SYTOX orange in tamoxifen‐treated circadian arrhythmic Bmal1 enteroids.

Data information: Data represented as mean ± SD of n = 4 biological replicates (i.e., enteroids derived from different mice). All of the data were normalized to time‐ and phase‐matched vehicle controls set to 1. **P < 0.01 (two‐way ANOVA).

Source data are available online for this figure.

HIOs, kcHIEs, and bHIEs were tested for circadian clock phase‐dependent responses following the same protocol used for mouse enteroids. HIOs showed identical necrotic cell death responses between the T24 and T36 groups (Fig 4A and B) consistent with their lack of robust circadian rhythms (Fig 1B). In contrast, both kcHIEs and bHIEs demonstrated circadian phase‐dependent necrotic cell death responses to TcdB with greater amount of cell death in the T24 compared to the T36 group for both kcHIEs and bHIEs (Fig 4C–F). These findings indicate that a functional, robust circadian clock is required for circadian phase‐dependent responses to TcdB. Intriguingly, mouse and human enteroids demonstrated anti‐phasic necrotic cell death responses to TcdB. Mouse enteroids showed a higher necrotic cell death in the T36 group (Fig 3B), whereas HIEs were more responsive in the T24 group (Fig 4D and F).

Figure 4

A functional circadian clock is needed for a circadian phase‐dependent response to TcdB in human organoids

Representative images of necrotic cell death (SYTOX orange) at 2, 24, and 48 h post‐exposure to 0.1 μg/ml TcdB in HIOs. Scale bar = 250 μm.

Quantification of necrotic cell death in HIOs.

Representative images of necrotic cell death (SYTOX orange) at 2, 24, and 48 h post‐exposure to 0.5 μg/ml TcdB in kcHIEs. Scale bar = 250 μm.

Quantification of necrotic cell death in kcHIEs.

Representative images of necrotic cell death (SYTOX orange) at 2, 24, and 48 h post‐exposure to 0.5 μg/ml TcdB in bHIEs. Scale bar = 250 μm.

Quantification of necrotic cell death in bHIEs.

Data information: All samples were exposed to TcdB at two circadian time points, 24 h (T24, grey) or 36 h (T36, red) post‐synchronization. Data are shown as mean ± SD of n ≥ 3 biological replicates (i.e., HIOs are generated from different PSCs and bHIEs are derived from three different patients’ samples). All data were normalized to time‐ and phase‐matched vehicle controls set to 1. *P < 0.05, **P < 0.01 (two‐way ANOVA).

Source data are available online for this figure.

Necrotic cell death responses measured by SYTOXTM Orange do not provide quantitative data. Therefore, we used fluorescence‐activated cell sorting (FACS) analysis to measure both necrotic and apoptotic cells using propidium iodide and Annexin V, respectively (Natarajan et al, 2019), in mouse enteroids and bHIEs. Consistent with the results obtained with SYTOXTM Orange, we observed anti‐phasic necrotic cell death responses between mouse enteroids and bHIEs (Fig EV5A–C,F). In contrast, we did not observe circadian time‐dependent changes in the number of apoptotic cells (Fig EV5D,E,G,H). We speculated that anti‐phasic behavioral rhythms between diurnal humans and nocturnal mice could be hardwired within the peripheral circadian clocks determining distinct phases of CCGs between mouse and human enteroids. To uncover CCGs in HIEs and mouse enteroids and identify potential target genes for the observed anti‐phasic necrotic cell death responses in mouse vs. human enteroids, we performed a 48‐h time course experiment with 2‐h resolution extracting RNA from bHIEs and mouse enteroids for RNA‐Seq.

Figure EV5

Circadian time‐dependent necrotic cell death responses to TcdB in mouse enteroids and bHIEs

All samples were exposed to TcdB at two time points, 24 h (T24, grey) or 36 h (T36, red) post‐synchronization of circadian rhythms with dexamethasone.

Representative plots demonstrating Annexin V‐FITC/PI staining in mouse enteroids (A) and bHIEs (B). Four different populations of cells are measured: unlabeled (viable cells‐Q1), stained with PI (necrotic‐Q2), cells that are labeled with both Annexin V‐FITC and PI (late apoptotic/necrotic cells‐Q3), and cells that are labeled with Annexin V FITC only (early apoptotic‐Q4).

Quantification of necrotic (C), early apoptotic (D), and total apoptotic (E) cell death at 24 and 48 hours post 20 ng/ml TcdB in C57B6 mouse enteroids.

Quantification of necrotic (F), apoptotic (G), and total apoptotic (H) cell death at 24 and 48 h post‐1 μg/ml TcdB in bHIEs. The experiment was performed in triplicates and presented with mean ± SEM (*P < 0.05; **P < 0.01; Student's t‐test, n ≥ 4).

Source data are available online for this figure.

Approximately 3–10% of detectable transcripts in mouse and human intestinal enteroids show rhythmic gene expression

For the analysis of our RNA‐Seq data, we used RAIN algorithm (Thaben & Westermark, 2014) to identify CCGs and performed phase set enrichment analysis (PSEA; Zhang et al, 2016) to determine signaling pathways that are under the control of intestinal circadian clocks. For RAIN analysis, we excluded low expressing transcripts with mean expression below the lowest 25%, and transcripts with null expression in more than 2/3 of the samples. Rhythm cutoff was set to RAIN BH q‐value < 0.15. The analysis uncovered 445 and 1,728 CCGs out of 14,342 and 17,898 detectable transcripts, which are 3.1% and 9.6% of total genes above the cutoff, in mouse enteroids and bHIEs, respectively (Fig 5A). This is in sharp contrast to less than a dozen cycling genes in other in vitro systems (i.e., NIH3T3 and U2OS cells) (Hughes et al, 2009). Importantly, heatmap of Spearman’s rho for circadian clock and clock‐associated genes show appropriate correlation of each gene against the rest in both mouse and human intestinal enteroids (Fig 5B), which confirmed that the circadian clock machinery is intact and functional with proper phase relationships in these enteroids.

Figure 5

3–10% of transcripts show circadian oscillations in mouse and human enteroids

Heatmap of circadian transcripts over 48 h in mouse and human enteroids.

Spearman’s rho for circadian clock and clock‐associated genes.

8 conserved rhythmic signaling pathways identified by phase set enrichment analysis (PSEA).

Source data are available online for this figure.

Circadian time‐dependent necrotic cell death responses may be due to rhythmic expression of TcdB receptor genes (i.e., Cspg4, Pvrl3, and Fzd1,2,7), TcdB‐targeted small GTPases (i.e., Rac1, Cdc42 and RhoA), and/or NOX1,2 NADPH oxidase complexes (i.e., Nox1, Noxo1, Noxa1, p22, Nox2, p47, p67, p22, and p40). RAIN algorithm did not detect rhythmicity in most of the above transcripts. Specifically, RAIN detected two (Rac1 and Nox1) and three (Cspg4, Rac1, and Noxo1) cycling transcripts in mouse and human enteroids, respectively, identifying Rac1 as the only conserved rhythmic transcript in this category (Table EV1, Appendix Fig S1A and B).

Phase set enrichment analysis uses prior knowledge to identify signaling pathways demonstrating temporally coordinated expression profiles. PSEA revealed that the intestinal epithelial circadian clocks regulate temporal organization of 8 conserved signaling pathways in both mouse and human intestinal enteroids including metabolism, cell cycle, and circadian clock (Fig 5C, Table EV1). Six out of eight signaling pathways including circadian rhythms showed similar phases between mouse and human enteroids. Intriguingly, two signaling pathways, cell–cell communication and neurotrophin signaling pathways, showed dramatically difference phases. To our surprise, these two signaling pathways included a single TcdB target gene, Rac1, which showed anti‐phasic expression between human and mouse enteroids. Hence, we validated the anti‐phasic expression of Rac1 between mouse and human enteroids using qRT–PCR (Fig 6A, Appendix Fig S2A). We observe arrhythmic expression of Rac1 in both HIOs and Bmal1‐floxed mouse enteroids, which is consistent with lack of circadian phase‐dependent necrotic cell death responses (Appendix Fig S2B and C). Lastly, disruption of Rac1 abolished circadian time‐dependent necrotic cell death responses to TcdB in both mouse enteroids and bHIEs. We used Rac1‐KD bHIEs and tamoxifen‐inducible Rac1‐KO mouse enteroids derived from Rac1; CAGGCre‐ERTM mice. Both Rac1‐KO mouse enteroids and Rac1‐KD bHIEs demonstrated a loss of circadian time‐dependent necrotic cell death responses measured by SYTOXTM Orange (Fig 6B–D, Appendix Fig S3A–D). Together, our data strongly suggest that the rhythmic expression of Rac1 determines the observed anti‐phasic necrotic cell death response against TcdB in mouse vs. human enteroids.

Figure 6

Disruption of Rac1 abolishes circadian time‐dependent necrotic cell death responses

Validation of anti‐phasic Rac1 expression in mouse (mRac1, red) and human enteroids (hRac1, black) by qRT–PCR.

Quantitative analysis of fluorescent intensity from SYTOX orange in Rac1‐KD bHIEs.

Quantitative analysis of fluorescent intensity from SYTOX orange in control Rac1 ;CAGGCre‐ERTM without tamoxifen (C) or Rac1 ;CAGGCre‐ERTM with tamoxifen‐treated (D) enteroids.

Data information: Data represented as mean ± S.D. of n = 5 technical replicates. All of the data were normalized to time‐ and phase‐matched vehicle controls set to 1. *P < 0.05 (two‐way ANOVA).

Source data are available online for this figure.

Discussion

Circadian rhythms regulate CCGs in a tissue‐specific manner throughout the body (Ruben et al, 2018). Circadian rhythms in peripheral organs are affected by numerous inputs including light (Buhr et al, 2019; Nayak et al, 2020), nutrients (Damiola et al, 2000; Stokkan et al, 2001; Crosby et al, 2019), and systemic cues from SCN (Silver et al, 1996; Guo et al, 2005; Barca‐Mayo et al, 2017; Brancaccio et al, 2017; Tso et al, 2017). Previous studies characterizing CCGs in different organs from in vivo mice uncovered organ‐specific regulation of circadian rhythms integrated by both peripheral clocks and systemic cues from SCN. Recent development of tissue‐specific conditional Bmal1 rescue mice will be instrumental to discover organ‐specific endogenous CCGs independent of SCN (Welz et al, 2019; Sinturel et al, 2021). Characterization of organ‐specific endogenous CCGs is necessary to systematically compare and contrast changes of circadian gene expression profiles in the presence of systemic and external cues, and control and disease states (e.g., cancer). However, the discovery of organ‐specific endogenous CCGs in humans is difficult, creating a major bottleneck hindering organ‐ and disease‐specific investigation of circadian rhythms and CCGs for translational applications.

Human organoids are complex, multicellular in vitro systems that are devoid of systemic signaling factors, making them an appealing system to evaluate endogenous, tissue‐specific functions of peripheral circadian rhythms. The function of circadian rhythms, however, will depend on the maturity of organoids, because previous studies indicated lack of circadian rhythms in PSCs and mouse fetus during the early stages of its development (Reppert & Schwartz, 1986; Dierickx et al, 2017; Umemura et al, 2019). Therefore, we first characterized robustness of circadian rhythms from iPSCs to HIOs demonstrating the lack of robust circadian rhythms in HIOs. Differentiation of mouse embryonic or induced pluripotent stem cells results in the onset of circadian oscillations after 28 days (Umemura et al, 2017). Interestingly, a subsequent study from the same group found that human iPSCs require up to 90 days of differentiation to initiate circadian oscillations (Umemura et al, 2019). On the other hand, differentiation of human embryonic stem cells (ESC) into cardiomyocytes results in rhythmic cultures oscillations after 30 days. However, it is worth indicating that circadian oscillations of Bmal1‐dLuc and Per2‐dLuc reporter activities in the differentiated cardiomyocytes show weak oscillations (Dierickx et al, 2017) compared to HIEs (Fig 1B) and differentiated cells from iPSCs after 90 days of differentiation (Umemura et al, 2019). The lack of robust circadian rhythms in HIOs after 35‐day differentiation protocol is consistent with the lack of circadian oscillations at day 42 of in vitro differentiation of human iPSCs by Umemura and colleagues (Umemura et al, 2019), which indicate that further maturation is necessary for the development of circadian rhythms.

In contrast to HIOs, we observe robust circadian rhythms in both kcHIEs and bHIEs based on their identical bioluminescence outputs and oscillations reflecting the activity of Bmal1‐luc reporter (Fig 1B). However, we observe lower amplitude of Bmal1, Per2, and Rev‐erbα transcripts in kcHIEs compared to bHIEs (Fig 2E). These findings suggest that kcHIEs represent an intermediate state between HIOs and bHIEs, which is in agreement with previous results (Watson et al, 2014; Finkbeiner et al, 2015). Transcriptomic analysis comparing HIOs, kcHIEs, and adult intestinal tissue demonstrated that kcHIEs possess intermediate expression of brush border enzymes, sucrase isomaltase and trehalase, between HIOs and adult intestinal tissue (Watson et al, 2014). Similar findings were discovered for expression profiles of Paneth and stem cell markers in HIOs, kcHIEs, vs. adult intestinal tissue (Finkbeiner et al, 2015). It would be interesting to perform RNA‐Seq using time course samples from kcHIEs to compare and contrast CCGs between kcHIEs and bHIEs characterizing the reorganization of CCGs during development.

Approximately 3–10% of transcriptome show circadian oscillations in mouse enteroids and bHIEs regulating diverse signaling pathways ranging from cell cycle to immune response. It is not clear why we observe a lower number of CCGs in mouse enteroids compared to bHIEs, but the number of cycling genes in both mouse and human enteroids indicate that these in vitro systems possess functional circadian rhythms regulating a large number of tissue‐specific CCGs. In contrast, NIH3T3 and U2OS cells show less than a dozen cycling genes despite the presence of robust core clock gene oscillations (Hughes et al, 2009). On the other hand, 3–16% of cycling genes are observed depending on peripheral organs of mice in vivo (Zhang et al, 2014 ). We hypothesize that intestinal epithelial cells in vivo will show a greater number of CCGs compared to enteroids, because those CCGs will be a result of both systemic cues and peripheral circadian clocks. In our future work, we plan to compare CCGs in mouse enteroids and mouse intestinal tissue to systematically dissect CCGs that are controlled by systemic cues and intestinal clocks.

The intestinal circadian clock possesses robust circadian rhythms (Moore et al, 2014) and is responsible for promoting circadian variation in a wide range of biological processes (Rosselot et al, 2016), including intestinal motility (Kumar et al, 1986), nutrient absorption (Pan & Hussain, 2009), innate immunity (Froy & Chapnik, 2007), timing of cell division (Matsu‐ura et al, 2016), and the pathogenic response (Bellet et al, 2013). In patients, nocturnal diarrhea is an “alarm” symptom that should prompt clinicians to consider further evaluation, including testing for C. difficile infection (Cohen et al, 2010). However, whether a disrupted circadian clock predisposes to or is a result of gastrointestinal infections remains largely unknown. Insights to such questions will be needed to determine whether or not targeting circadian rhythms to treat such diseases would be efficacious. Our findings demonstrate robust operation of canonical clock genes regulating specific signaling pathways, and species‐dependent anti‐phasic expression of Rac1 between mouse and human enteroids explaining the observed anti‐phasic necrotic cell death response in mouse vs. human enteroids against TcdB. It is intriguing that we observe anti‐phasic expression of Rac1 between mouse and human enteroids despite similar phases of canonical clock genes (Fig 5C). We speculate that this could be due to either different accessibility and/or transcriptional regulation of the Rac1 promoter between mouse and human enteroids. In the future, we plan to evaluate the importance of circadian‐phase specific expression of Rac1 and other genes that are involved in signaling by RHO GTPases between mouse and human enteroids, and design chronotherapeutic regimens for treatments of C. difficile infection. Furthermore, our work demonstrates a proof of concept that human organoids derived from different organs and disease states could provide a unique opportunity to: (i) characterize function of peripheral circadian clocks in different organs, (ii) identify circadian biomarkers for human diseases, and (iii) leverage the information to develop novel circadian treatment regimens for translational applications.

Materials and Methods

Animals

Mouse enteroids were established and cultured from 3‐ to 6–month‐old C57BL/6J (Jackson Laboratories), PER2::LUCIFERASE (Yoo et al, 2004), and Bmal1 (Yang et al, 2016). We generated a conditional Rac1‐KO mouse (Rac1;CAGGCre‐ERTM) by crossing Rac1 (Stock No: 005550, Jackson Laboratory) and tamoxifen‐inducible CAGGCre‐ERTM (Stock No: 034795, Jackson Laboratory) mice. Immune‐deficient NSG mice (Jackson Laboratories) were utilized for kidney capsule transplantation of HIOs (Watson et al, 2014). Animal handling was performed according to IACUC protocols #2016‐0014 (PI: Michael Helmrath—Cincinnati Children’s Hospital Medical Center) and #17‐01‐30‐01 (PI: Christian Hong—University of Cincinnati).

Isolation and maintenance of mouse enteroids

Mouse jejunal enteroids were generated following established protocols (Nalapareddy et al, 2017). Briefly, crypt domains from small intestinal tissue were isolated, added to 40 µl of Matrigel (Corning) membrane matrix, suspended as a single 3D Matrigel dome within a 24‐well plate, and provided 400 µl mouse enteroid media (Advanced DMEM/F‐12 (Gibco), L‐Glutamine (Gibco), Penicillin/Streptomycin (Gibco), N‐2 (Gibco), B‐27 (Gibco), 10 mM HEPES solution (Millipore‐Sigma), 50 ng/ml recombinant murine EGF (Peprotech), and R‐Spondin/Noggin conditioned media generated in‐house. Enteroids were passaged every 4–7 days via a 27G syringe (BD). CRE‐driven Bmal1 exon 4 or Rac1 exon 1 excisions were initiated in vitro by adding 1 µM 4‐OH tamoxifen (Cayman) for 24‐h to Bmal1 or Rac1CAGGCre‐ER™ enteroids. All samples were maintained in cell culture incubators set at 37°C and 5% CO2. Enteroids isolated from separate mice were considered biological replicate samples.

Experiment design to test for a circadian phase‐dependent response to TcdB

1–2 HIOs and ≈15–20 mouse/human enteroids were plated per well to control for density. 100 nM dexamethasone (DEX) was added to treatment groups either 36 h (T36) or 24 h (T24) prior to TcdB addition. Samples were then exposed to a vehicle buffer (50 mM Tris Base Ultrapure (US biological), 100 mM NaCl (Fisher Scientific), pH 7.5) or Clostridium difficile toxin B (List Laboratories) by addition to cell culture media. SYTOX Orange (500 nM) (Invitrogen) was added with Vehicle/TcdB for quantification of necrotic cell death. Four wells of a 24‐well plate were seeded with twenty enteroids/well at the onset of each experiment. Five enteroids within each well were randomly selected for imaging. These same five enteroids were imaged at 2, 24, and 48 h post‐vehicle/TcdB addition on a Dmi8 inverted microscope (Leica Microsystems) to minimize the confounding variable of imaging different enteroids at each time point. When quantifying fluorescence from SYTOX™ Orange, an area of interest was manually traced around each enteroid using LasX software from Leica microsystems. An independent area of interest was generated for each enteroid at each time point to account for enteroid growth during the experiment. Fluorescence intensity within the area of interest was reported by the LasX imaging software. This approach was selected to minimize the variability in overall size of each enteroid. A total of 20 individual enteroids (5 technical replicates from 4 biological replicates) were analyzed for each treatment group. All TcdB data were set relative to time‐ and phase‐matched vehicle controls set to 1.

HIO generation

Human intestinal organoids were generated following previously established protocols (Spence et al, 2011) using multiple pluripotent stem cell (PSC) lines (Table EV2). Briefly, PSCs were differentiated through the definitive endoderm stage via the addition of Activin A (Cell Guidance Systems) between day‐0 (D0) and D2, BMP4 (R&D Systems) at D1 and CHIR99021 (Stemgent)/FGF4 (R&D Systems) from D3 to D6. The addition of CHIR99021/FGF4 prompted the formation of midgut tube spheroids from the definitive endoderm monolayer. At D7 midgut tube spheroids were collected and suspended in a Matrigel dome (Corning). Spheroids were then grown in gut media: DMEM/F12 (Gibco) supplemented with N‐2 (Gibco), B‐27 (Gibco), HEPES solution (Millipore‐Sigma), recombinant murine epidermal growth factor (Peprotech), L‐Glutamine (Gibco), Penicillin/Streptomycin (Gibco). At D21, sample density was reduced to ≈1–3 per 50 µl Matrigel bubble. Samples were grown for an additional 14 days in gut media, resulting in terminal HIO cultures (Spence et al, 2011). Individual HIOs were considered as biological replicates for all experiments. All experimentation using human tissues described herein was approved by an IRB at CCHMC (IRB#2014‐0427). Informed consent for tissue collection, storage, and use of the samples was obtained from the donors at CCHMC.

Generation and maintenance of kidney capsule‐matured and biopsy‐derived human intestinal enteroids

Human intestinal organoids were transplanted into the kidney capsule of immune‐deficient NOD‐SCID IL‐2Rγnull (NSG) (Jackson Laboratories) mice, as previously described (Watson et al, 2014). Crypt domains from transplanted HIOs were isolated and suspended in Matrigel domes to form kidney capsule‐matured human intestinal enteroids (kcHIEs; Mahe et al, 2015). Biological replicates for kcHIEs are derived from the human crypts harvested from three independent mice. Four independent frozen duodenal patient biopsy‐derived human intestinal enteroids (bHIEs) were provided by the lab of Dr. Noah Shroyer (Baylor College of Medicine, Table EV2). Both bHIEs and kcHIEs were cultured following identical protocols. Enteroids were expanded by suspending pelleted enteroids in Matrigel and plating as three separate 10 µl domes/well in a 24‐well plate. Intesticult organoid growth medium (StemCell Technologies) was used for enteroid expansion. Human enteroid differentiation was initiated by replacing expansion media with differentiation (Intesticult Component A (StemCell Technologies) mixed at a 1:1 ratio with DMEM/F12 supplemented with 15 mM HEPES) media 2 days post‐passaging. Human enteroids were maintained in differentiation media at least four days before being used for experiments. kcHIEs derived from separate transplanted mice were counted as biological replicates. bHIEs isolated from separate human patient biopsies were counted as biological replicates.

Bmal1‐luciferase lentiviral transduction and monitoring of bioluminescence

All human samples from PSCs to HIOs, kcHIEs and bHIEs were transduced with the same pABpuro‐BluF (Addgene) plasmid DNA (Brown et al, 2005). Plasmid DNA was packaged into lentiviral vectors by the Viral Vector Core at Cincinnati Children’s Hospital Medical Center (CCHMC). PSC transduction was carried out following a previously published lentiviral transduction protocol (Matsu‐ura et al, 2016). Transduced PSCs were Puromycin selected (2 µg/ml) (Invivogen) and differentiated into definitive endoderm (DE) and midgut tube (MG) cultures, generating PSC, DE, and MG clock reporting samples. Midgut tube samples were re‐transduced following the same protocol used for PSC transduction and differentiated into HIOs to generate Bmal1‐luc reporting HIOs. The bHIE and kcHIE transduction protocol was designed by merging previously reported enteroid digestion (Zou et al, 2017) and transduction (Koo et al, 2012) protocols. After transduction, digested enteroids were suspended in Matrigel and plated following normal culture protocol. 350 µl of Intesticult Component A/B (1:1) media supplemented with 10 µM ROCK inhibitor and 10 µM CHIR99021 was added to the culture. After 24 h, the media was replaced with 350 µl Intesticult Component A/B (1:1) supplemented with 10 µM ROCK inhibitor. On day two post‐infection, media was transitioned to Intesticult Component A/B (1:1) supplemented with 2 µg/ml Puromycin. Puromycin selection was maintained for 2 weeks before starting experiments. Bmal1‐luc transduced samples were monitored for clock output over 4‐days with a KronosDIO luminometer (ATTO). PSC, DE, MG, and HIO samples were maintained in normal culture media throughout recording. kcHIEs and bHIEs were differentiated for 4 days prior to bioluminescence recording. On day 4, samples were re‐plated to 35 mm dishes, synchronized and added to the KronosDIO to begin recording. All samples were provided a 1‐hour clock synchronization stimulus with 100 nM DEX (Millipore‐Sigma) and given new, non‐DEX containing, media prior to recording. 200 µM Luciferin (GoldBio) was added to each culture for luminescence detection.

Time course sample collection

Rhythmic gene expression was determined via time course sample collection. Samples were collected every 4 h over a 48‐h period, two full circadian cycles. At each collection time point, cells were pelleted, suspended in 1 ml TRI reagent (Molecular Research Center) and snap‐frozen in liquid N2. After completion of the time course, RNA was isolated via a RNeasy mini‐kit column (Qiagen) and cDNA were generated using the Superscript III Reverse transcriptase kit (Invitrogen). FAST SYBR green (Applied Biosystems) and gene‐specific primer pairs were used to quantify gene expression via real‐time quantitative reverse transcriptase polymerase chain reaction (qRT–PCR) with TATA‐Box Binding protein (Tbp) expression used as a housekeeping gene.

FACS analysis

A quantitative assessment of apoptosis and necrosis in mouse enteroids and bHIEs was performed using fluorescein isothiocyanate (FITC) Annexin V Apoptosis Detection Kit I (BD Pharmingen). Enteroids were dissociated using TrypLE for ~5 min at 37°C to get single cells (0.5 ~ 1 × 106 cells). The crypt suspension was washed with 1 ml of DMEM/F12 medium and centrifuged down. Pellets were resuspended in 100 μl PBS by mixing 5 μl of FITC‐conjugated Annexin V (Annexin V‐FITC) and 5 μl propidium iodide (PI) according to manufacturer’s instructions. The cells were incubated for 15 min at room temperature in the dark. The stained cells were diluted with 400 μl binding buffer and immediately analyzed by the BD FACSCalibur flow cytometer (BD Biosciences, Heidelberg, Germany). 10,000 cells were analyzed per measurement. Data were processed using the CellQuestPro Software (Kwan et al, 2016; Natarajan et al, 2019). Dose–response curves were used to determine different concentrations of TcdB for mouse and human intestinal enteroids (Appendix Fig S4A and B).

RNA‐sequencing analysis

To perform RNA‐sequencing, we collected C57BL/6J mouse enteroids and human bHIEs derived from patient sample #103 (Table EV2) as a time course. Mouse enteroids were pooled from two separate mice to mitigate single mouse biasing and bHIEs were from a single patient (female, 24‐years old) biopsy. Samples were collected every two hours for 46 h starting at 24 h post‐DEX synchronization. Isolated RNA samples were shipped on dry ice for further processing and sequencing analysis using paired‐end reads with 50 million reads per sample by Novogene. Raw RNA‐sequencing FASTQ files were directly used to generate expression files using the Kallisto (Bray et al, 2016) embedded algorithm inside the Altanalyze module (Emig et al, 2010). Kallisto pseudo‐aligned each transcript with the Ensemble reference transcriptome (Ensembl version 72) and calculated TPM (transcripts per million) by Altanalyze (Emig et al, 2010). For PSEA, we used the reference gene sets in c2.cp.v5.0.symbols.gmt downloaded from the Molecular Signatures Database (MSigDB).

Statistical analysis

All data are displayed as mean values of at least three biological replicates with error bars representing standard deviation around the mean. KronosDIO bioluminescent traces were quantified for clock parameters via fast Fourier transformation (Matsu‐ura et al, 2016). qRT–PCR and RNA‐seq transcript expression profiles were tested for significant rhythms using the R package MetaCycle 2D (Wu et al, 2016; P‐value cutoff of 0.05; Thaben & Westermark, 2014). All between‐group comparisons were made using a two‐way analysis of variance (two‐way ANOVA) with Tukey’s post hoc analysis. Normality of the residuals was confirmed with Shapiro’s test of residuals at a significance of P < 0.05. All statistical tests were performed in R Studio version 1.2.1335.

Author contributions

AER, MP, and MK designed the experiments, collected the data, generated, and compiled figures and wrote the manuscript. TM‐U and SL assisted in designing and performing experiments. KRS, KC, and NSa received and aligned RNA‐sequencing data to reference transcriptomes. DEF, GW, and JBH analyzed RNA‐sequencing data and generated the associated figures. NSu and JAH performed surgical transplantation of HIOs into the kidney capsule of mice and provided kcHIEs. TRB provided HIOs and assisted in the design of HIO related experiments. HAM provided kcHIE samples. NFS supplied bHIEs and reviewed the manuscript. MAH, JMW, and JBH were involved in the study design and review of this manuscript. SRM and CIH designed and directed the overall progress of studies, wrote and provided final approval of the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Appendix

Click here for additional data file.

Expanded View Figures PDF

Click here for additional data file.

Table EV1

Click here for additional data file.

Table EV2

Click here for additional data file.

Source Data for Expanded View and Appendix

Click here for additional data file.

Source Data for Figure 1

Click here for additional data file.

Source Data for Figure 2

Click here for additional data file.

Source Data for Figure 3

Click here for additional data file.

Source Data for Figure 4

Click here for additional data file.

Source Data for Figure 5

Click here for additional data file.

Source Data for Figure 6

Click here for additional data file.