Tankyrases maintain homeostasis of intestinal epithelium by preventing cell death

Lgr5+ intestinal stem cells are crucial for fast homeostatic renewal of intestinal epithelium and Wnt/β-catenin signaling plays an essential role in this process by sustaining stem cell self-renewal. The poly(ADP-ribose) polymerases tankyrases (TNKSs) mediate protein poly-ADP-ribosylation and are involved in multiple cellular processes such as Wnt signaling regulation, mitotic progression and telomere maintenance. However, little is known about the physiological function of TNKSs in epithelium homeostasis regulation. Here, using Villin-creERT2;Tnks1-/-;Tnks2fl/fl (DKO) mice, we observed that loss of TNKSs causes a rapid decrease of Lgr5+ intestinal stem cells and magnified apoptosis in small intestinal crypts, leading to intestine degeneration and increased mouse mortality. Consistently, deletion of Tnks or blockage of TNKS activity with the inhibitor XAV939 significantly inhibits the growth of intestinal organoids. We further showed that the Wnt signaling agonist CHIR99021 sustains the growth of DKO organoids, and XAV939 does not cause growth retardation of Apc-/- organoids. Consistent with the promoting function of TNKSs in Wnt signaling, Wnt/β-catenin signaling is significantly decreased with stabilized Axin in DKO crypts. Together, our findings unravel the essential role of TNKSs-mediated protein parsylation in small intestinal homeostasis by modulating Wnt/β-catenin signaling.

TNKSs are widely expressed in many human organs, including small intestine, colon and placenta [30,31]. Tnks1/2 double knockout leads to embryonic lethality in mice [19], while Tnks2 knockout mice exhibit mild phenotypes, such as decreased body weight, especially in male mice [32,33], indicating that these two genes have redundant functions. Therefore, the physiological functions of TNKSs in adult tissue homeostasis of mammals are still poorly understood. Recently, it has been shown that deletion of the Drosophila TNKS homolog caused no obvious macroscopic phenotypes under normal feeding condition, but increased ISC number surrounded by Tnks-deficient enterocyte [34].
To better understand the functions of TNKSs in intestinal physiology, we generated Villin-creERT2;Tnks1 -/-;Tnks2 fl/fl (DKO) mice. Our results revealed that TNKS DKO led to defective crypts with significant loss of both Lgr5 + ISCs and Ki67 + transient amplifying cells, resulting in an obvious increase of mortality in long term. Consistent with the promoting role of TNKSs in Wnt signaling, Axin level was increased while Wnt signaling activity was decreased in TNKS DKO intestinal epithelial cells. These data were confirmed with the TNKS inhibitor XAV939 in in vitro cultured organoids. Taken together, our findings indicate that TNKSs play an essential role in intestinal epithelium homeostasis by modulating Wnt/β-catenin signaling.

Double knockout of Tnks1 and Tnks2 causes lethal degeneration of intestinal crypts
It has been shown that TNKS1 and TNKS2 are highly expressed in both small intestine and colon [30,31]. We confirmed that both TNKS1 and TNKS2 were expressed in the intestinal epithelium (S1A Fig). Importantly, immunohistochemical analysis showed that TNKSs exhibited a gradually decreased expression from the crypt bottom to the villus tip in small intestine (S1A Fig). This expression pattern is consistent with the Wnt signaling activity gradient along the crypt-villus axis [7,8].
Conventional Tnks knockout in mice showed that TNKS1 and TNKS2 are functionally redundant, and Tnks double knockout resulted in embryonic lethality [19]. In order to investigate the role of TNKSs in the maintenance of intestinal epithelium homeostasis, we generated Villin-creERT2;Tnks1 -/-;Tnks2 fl/fl (DKO) mice, which harbored conventional Tnks1 deletion and inducible intestinal epithelium specific Tnks2 knockout. At 8-10 weeks of age, Villin-creERT2;Tnks1 -/-;Tnks2 fl/fl (DKO) mice and control (Ctrl) littermates (Tnks1 -/-;Tnks2 fl/fl ) were daily injected with tamoxifen (TAM) for 4 consecutive days to induce Tnks2 deletion, then sacrificed at different time points after their first injection (Fig 1A). By immunoblotting and immunohistochemical staining, we detected a near-complete loss of both TNKS1 and TNKS2 in DKO intestines at 4 day after the first TAM injection (Fig 1B; S1B and S1C Fig). Interestingly, DKO mice showed a rapid onset of weight loss from day 6 (S1D Fig). By day 10, DKO mice displayed diarrhea and small intestine bleeding (S1E and S1F Fig), and showed remarkably increased mortality ( Fig 1C). Overall, the small intestines from these DKO mice were shortened and appeared distended and necrotic (Fig 1D and S1F Fig). However, colons appeared normal.
In line with the mortality and gross abnormality phenotypes in DKO mice, although little changes were detected at day 4, severely degenerative crypts appeared in small intestine at day 5, and shortened and irregular villi were observed at day 10 ( Fig 1E and S1G Fig). Most crypts were disorganized and the stem cell zone consisting of Paneth cells and intestinal stem cells (ISCs) disappeared. At day 10, we could also detect some hyperplastic crypts, which might be the escaping crypts as such escaping crypts are frequently observed upon essential gene ablation [35,36]. Although a great amount of degenerative crypts with few escaping crypts were present throughout small intestine, but colon just showed a few degenerative crypts with mildly decreased viable crypts in day 10 DKO mice (S1H Fig). Together, Tnks deficiency caused the lethal degeneration of small intestinal crypts, indicating that, TNKSs are essential for small intestinal homeostatic self-renewal.

Loss of TNKSs reduces proliferation and induces apoptosis in small intestinal crypts
We then examined cell proliferation in intestine. At day 2, Ki67 staining indicated that cell proliferation in crypts was normal (S2A Fig). At day 3 and 4 DKO mice, although the number of viable crypts had no obvious change, the proliferative cells in crypts marked by Ki67 were decreased in small intestine ( To further explore the degenerative crypt phenotype, we examined cell death in crypts of small intestine by FACS basing on propidium iodide (PI) staining. The result showed that Tnks deletion led to about 50% increase of cell death at day 4 mice, which was consistent with the TUNEL assay results (Fig 2C and 2D; S3A-S3C Fig). Similarly, more cell death was observed in DKO colon (S3D Fig). These data together suggest that the degenerative crypt phenotype in DKO mice is due to both reduced proliferation and enhanced apoptosis.

Tnks deficiency causes loss of Lgr5 + intestinal stem cells
We suspected that crypt degeneration is caused by loss of ISCs that fueling the self-renewal of crypts [6,37]. First, we performed in situ hybridization of Olfm4, a stem cell marker for Lgr5 + ISCs [14,38]. As shown in Figs 3A, S4A and S4B, Tnks DKO intestine showed dramatically decreased Olfm4 expression as early as at day 2. In order to quantify the number of ISCs, we generated Lgr5-EGFP-IRES-creERT2;Tnks1 -/-;Tnks2 fl/fl (DKO-Lgr5) mice, in which green fluorescent protein could mark Lgr5-positive ISCs. A marked decrease of GFP + cells was observed in day 4 Tnks DKO mice (Fig 3B), which was confirmed by FACS analysis (Fig 3C). We further employed quantitative reverse-transcription PCR (qRT-PCR) to assess the expression of the ISC markers and detected the down-regulation of Lgr5, Olfm4, Ascl2, EphB2 and Hopx1 mRNAs in Tnks deficient crypts as early as at day 2 ( Fig 3D and S3C Fig). Furthermore, loss of Lgr5 + ISCs in cultured organoids was also observed upon 4-hydroxytamoxifen (4-OHT)induced Tnks DKO ( Fig 3E). These data suggest that TNKSs are indispensable for the Lgr5 + ISC maintenance in adult small intestine.

Tnks deficiency reduces Paneth cells and goblet cells
As Paneth cells are a crucial component of the Lgr5 + ISC niche [39], we explored whether Tnks knockout affects Paneth cells by lysozyme immunohistochemical staining. As shown in Fig 4A, at day 4 and day 5 after first TAM injection, Paneth cells in crypt bottom were significantly diminished, and some of them were found to be mislocated in villi. Muc2 immunofluorescence staining showed that the goblet cell number was also significantly reduced in day 5 DKO mice ( Fig 4B), while the number of enteroendocrine cells (marked by chromogranin A), remained unchanged in control and DKO mice ( Fig 4C). Together, these data indicated that TNKSs play a critical role in the maintenance or differentiation of Paneth cells and goblet cells.

Ablation of TNKSs impairs Wnt/β-catenin signaling in intestinal epithelial cells
TNKSs play an important role in regulating Wnt/β-catenin signaling and are involved in many cellular processes, including mitosis, telomere maintenance and others. As cell proliferation was greatly reduced in DKO intestine and TNKS1 knockdown led to the formation of multipolar mitotic spindle or mitotic arrest, thus inhibiting the cell division [24,40], we analyzed the mitotic spindle morphology. Tubulin staining indicated that the bipolar mitotic spindle were normal in day 4 DKO intestinal crypt cells (S5A Fig). Although it was shown that the cell cycle arrest can lead to the formation of multinucleated cells shown by nuclear lamina protein Lamin B staining [41], we found no obvious multinucleated cells in small intestinal crypts of day 4 and day 5 DKO mice (S5B Fig). Together, these data suggested that mitosis was normal in DKO intestine.
Next, we examined Wnt/β-catenin signaling, which is required for the maintenance of proliferative crypt compartment and ISCs in intestine [5]. As Wnt signaling has been indicated to regulate apoptosis [42][43][44], the observed apoptosis increase in crypts may be caused by deficient Wnt signaling. Indeed, the expression of Wnt target genes started to be decreased at day 2, and dramatically reduced at day 4 ( Fig 5A and S5C Fig). Previous studies reported that TNKSs are positive regulators of Wnt signaling by controlling the stability of Axin [23]. Immunoblotting analysis showed that the protein level of Axin was greatly increased, while total β-catenin was significantly reduced, and GSK3β and CK1α remained the same ( Fig 5B). Consistently, parsylation of Axin was reduced in TNKSs DKO crypts ( Fig 5C). These data together demonstrate that TNKS deficiency results in Axin stabilization and thus attenuates Wnt signaling.

Activation of Wnt/β-catenin signaling overcomes TNKSs deficiencyinduced growth defect in organoids
To confirm that TNKSs regulate intestinal epithelium renewal mainly by modulating Wnt/βcatenin signaling, we employed the crypt organoid culture system to determine whether the Wnt signaling agonist GSK3β inhibitor CHIR-99021 could overcome Tnks deficiency. As shown in Fig 6A, CHIR-99021 restored the organoid survival and the bud formation in the 4-OHTinduced Tnks KO organoids. We further observed that CHIR99021 could partially recover Lgr5 + ISCs in the Tnks-deficient organoids (Fig 6B and 6C). We also examined the changes of apoptotic cells and proliferative cells in organoids under these culture conditions. As expected, apoptotic cells were obviously increased in Tnks DKO organoids, which were reversed by CHIR-99021 ( Fig 6D). In line with it, the expression of ISC markers and Wnt target genes in Tnks DKO organoids was partially restored by CHIR-99021 ( Fig 6E). In addition, we found that CHIR-99021 could counteract the detrimental effect of the TNKS inhibitor XAV939 on cell growth and apoptosis in Tnks1 -/organoids (S6A and S6B Fig). Lastly, we determined whether loss of Apc could abolish the negative effect of XAV939 on organoids as Apc mutation leads to the hyperactivation of Wnt signaling. We found that Apc mutant organoids derived from the polys of Villin-Cre; Apc +/fl mice could grow normally in the presence of XAV939 (S6A Fig). Similarly, XAV939 had no effect on cell proliferation, apoptosis and the expression of ISC markers and Wnt target genes of Apc mutant organoids (S6C and S6D Fig). Together, these data further support the note that TNKSs maintain intestinal homeostasis by controlling Wnt/β-catenin signaling.

Discussion
In this work, we highlight a previously unrecognized essential role for TNKSs in normal homeostatic self-renewal of adult murine intestine. TNKSs have been indicated to take part in many processes, such as controlling zebrafish gut and tail fin regeneration [45], murine kidney development [46] and murine bone mass maintenance [47]. As these studies used TNKS small molecule inhibitors, it is hard to rule out possible non-specific effects of the small molecules. Therefore, genetics approaches were used to explore their physiological functions. Due to functional redundancy, conventional knockout of single Tnks gene showed mild phenotypes    Tankyrases maintain intestinal homeostasis [29,32,33], while conventional Tnks double knockout (Tnks1 -/-;Tnks2 -/-) resulted in embryonic lethality [19]. Thus, we still know little about their in vivo roles, especially in tissue homeostasis of adult mammals. In this study, to address the function of TNKSs in intestine, we generated the mice harboring conventional Tnks1 KO and inducible Tnks2 conditional KO (Villin-creERT2;Tnks1 -/-;Tnks2 fl/fl ), and provided strong evidence that both TNKS1 and TNKS2 are essential for the homeostatic maintenance of intestinal epithelium in mice.
Tamoxifen-inducible loss of TNKSs caused a disorganized villi and disrupted crypt structure, leading to mouse death with shorter small intestines at day 10 after the TAM injection. At days 4-5, cell proliferation was reduced, while apoptosis significantly elevated along with a drastically decrease of ISCs in Tnks DKO mice, even at day 2 and day 3 after the first tamoxifen administration. This is consistent with a plethora of studies showing that TNKS inhibitors suppress cell proliferation in several cancer cell lines [48][49][50][51][52]. Although inhibition of TNKS activity has been shown to block Apc mutant colorectal cancer development [53,54], we found that the organoids derived from the polys of Villin-Cre;Apc +/fl mice showed normal cell proliferation and activation of Wnt/β-catenin signaling in the presence of XAV939. The discrepancy could be due to the following possibilities: Firstly, Apc mutant organoids are cultured in medium containing EGF, Noggin and R-spondin1, which can augment Wnt signaling to support cell growth, while the cells were cultured in a medium without these growth factors. Secondly, R-spondin in organoid culture may have other functions in addition to augment Wnt/β-catenin signaling [55]. Thirdly, the cancer cells may carry other gene mutations. Lastly, the organoids possess three-dimension structure and more than one cell types, while 2D cultured cancer cells only exist one cell type. Interestingly, removal of the only Tnks gene in Drosophila shows no obvious phenotypes under normal feeding condition, but ISCs are increased, especially in the posterior midgut [34]. The reduced Wnt signaling due to Tnks inactivation in enterocytes may lead to enhanced JAK/ STAT signaling in neighboring ISCs non-autonomously and thus their hyperproliferation. Therefore, TNKSs have distinct roles in maintenance of intestinal homeostasis in invertebrates and vertebrates. We also observed that some Paneth cells are dislocated in the villi, and the number of goblet cells decreases in day 5 Tnks DKO mice.
As a poly(ADP-ribose) synthase, TNKSs mediate parsylation on various proteins and thus regulate telomere length, mitosis and other cellular processes [20][21][22]. The degenerative phenotype observed in Tnks DKO intestine was associated with increased apoptosis and decreased cell proliferation. Interestingly, the degenerative phenotype was more obvious in small intestine than in colon of DKO mice, which is in agreement with the early reports [56,57]. It is likely to be caused by the difference in the turnover rate of colonic and small intestinal epithelia, and the pathology of the colon was not dramatic at the time of the death of Tnks DKO mice (10 d). By detailed analyses of the underlying mechanism, we found that Wnt signaling activity was greatly reduced due to accumulated Axin protein. Consistent with this, the GSK3β inhibitor CHIR-99021 could maintain the growth of Tnks DKO organoids with the recovery of ISCs. The essential role of TNKSs in the maintenance of intestinal epithelium homeostasis is in agreement of the function of Wnt signaling in promoting cell proliferation and inhibiting apoptosis [17]. The results from previous studies showed the high effectiveness of TNKS inhibitors in inhibiting various cancer cell proliferation including CRC [48,[51][52][53]. However, based on our genetic results and previous pharmacologic results [58], caution should be taken to the possible toxicity of TNKS inhibitors.

Ethics statement
Mice were maintained in the pathogen-free Laboratory Animal Facility of Tsinghua University. The facility has been accredited by AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care International), and the IACUC (Institutional Animal Care and Use Committee) of Tsinghua University approved all animal protocols used in this study by conferring the certificates 12-CYG-2 and 16-CYG-1.

Mice and in vivo studies
Villin-cre mice, Lgr5-EGFP-IRES-creERT2 and Apc fl/fl mice were obtained from Jackson Laboratory. Tnks1 +/-;Tnks2 +/fl mice have been described previously [19]. Villin-creERT2 mice were kindly provided by Dr. Sylvie Robine. Both male and female C57BL/6 mice aged 8-10 weeks were used. In general, 2-10 mice per genotype were used in different experiments. In Tnks2 deletion experiments, mice with the indicated genotype were intraperitoneally injected with 80 mg/kg tamoxifen in sunflower oil at 20 mg/ml for 4-5 consecutive days. Animals were then euthanized at different time points, and tissue was processed immediately. All animal experiments were conducted in accordance with the relevant animal regulations with approval of the Institutional Animal Care and Use Committee of Tsinghua University.

Crypt isolation and organoid culture
Small intestinal crypts were isolated and cultured as previously described [59]. In XAV939 treatment experiments, organoids were cultured in the medium containing EGF, Noggin and R-spondin1 (ENR) for 24h followed by passaging, then cultured in ENR medium with DMSO or XAV939 (50μM) for different time spans. To achieve in vitro Tnks2 knockout, organoids were cultured in ENR medium for 24h followed by passaging, then incubated in ENR medium with EtOH or 4-OH-tamoxifen (4-OHT, 1μM) for the indicated time. For organoid passaging, cultured organoids were vigorously suspended in cold PBS after removing culture medium and were collected by centrifugation (400g, 3min, 4˚C). Then the pelleted organoids were mixed fully with fresh Matrigel, seeded on plate. After Matrigel polymerization, crypt culture medium was added.

Flow cytometry
To obtain single cell suspension, fresh intestinal crypts isolated from Lgr5-EGFP-IRES-creERT2 and Lgr5-EGFP-IRES-creERT2;Tnks1 -/-;Tnks2 fl/fl mice were placed in TrypLE (Invitrogen) for 20 min at 37˚C. Then the dissociated cells were filtered through a 40μm cell strainer. The EGFP positive cells were subjected to the FACS analysis (MoFlo XDP, Beckman). The acquired data were analyzed by using Summit software.

mRNA isolation and quantification
RNA was isolated from fresh crypts or cultured organoids by using TRIzol Reagent (Life Technologies). Total RNA yield was determined by using NanoDrop 2000 (Thermo Fisher Scientific). cDNAs were generated using Revertra Ace (Toyobo). Quantitative real-time PCR (qRT-PCR) were carried out in triplicates on the LightCycler 480 (Roche). Primers used were listed in S1 Table. TUNEL assay Using in situ cell death detection kit (Roche), apoptosis was assessed by TdT-mediated dUTP nick end labelling (TUNEL) assay in intestinal paraffin section (5μm) according to the manufacturer's recommendations.

Immunofluorescence and immunohistochemistry
For immunofluorescence, intestine and organoids were fixed for overnight with 4% formaldehyde at room temperature. Then paraffin intestine or organoid sections were de-paraffinized in isopropanol and dehydrated by a graded alcohol series, followed by antigen retrieval. Next, the section was permeabilized for 10 min in PBS with 0.1% Triton X-100 at room temperature. Then the sections were blocked in 5% BSA/0.1% Triton-X/PBS (blocking buffer) for 1 h at room temperature followed by incubating with the primary antibody at 4˚C overnight. The fluorescein-labeled secondary antibodies (1:300, Life Technologies) were added for 1 h at room temperature. Samples were visualized by an Olympus FV1200 Laser Scanning Microscope.
For immunohistochemistry, the paraffin section was prepared as immunofluorescence. Then endogenous peroxidase was quenched by H 2 O 2 . Next, the sections were blocked in blocking buffer for 30 min and incubated overnight with primary antibody at 4˚C. Horseradish peroxidase-conjugated secondary antibody (Invitrogen, 1:200) was added for 2 h at room temperature followed by developing with substrate DAB. Then sections were counterstained with hematoxylin and eosin, dehydrated and mounted with resinene.

Immunoblotting and immunoprecipitation
Protein samples were prepared from fresh intestinal crypts and villi. These two assays were carried out as previously described [59].

In situ hybridization
For in situ hybridization experiments, the paraffin sections were de-waxed, rehydrated and hybridized with digoxigenin-labeled probe at 65˚C for overnight. After several rounds of wash, sections were placed in blocking solution for 2 h followed by incubating with alkaline phosphatase-conjugated anti-digoxigenin antibody (1:2000; Roche) at 4˚C overnight. Then the sections incubated with AP substrate (Roche) after washing several times. Olfm4 probe (2-1622) was generated with in vitro transcription kit (Roche).

Statistical analysis
All experiments were performed at least three biological replicates. Data are expressed as mean± SEM or SD, and the statistic significance determined by nonparametric Student's t-test (Mann-Whitney test) or Two-way ANOVA with GraphPad Prism 6 software. The p value < 0.05 was considered significant and the differences were annotated with asterisks: Ã (p < 0.05), ÃÃ (p < 0.01), ÃÃÃ (p < 0.001). The numerical data for statistical analysis in each figure are in S2 Table. Supporting information