TIGAR/AP-1 axis accelerates the division of Lgr5− reserve intestinal stem cells to reestablish intestinal architecture after lethal radiation

During radiologic or nuclear accidents, high-dose ionizing radiation (IR) can cause gastrointestinal syndrome (GIS), a deadly disorder that urgently needs effective therapy. Unfortunately, current treatments based on natural products and antioxidants have shown very limited effects in alleviating deadly GIS. Reserve intestinal stem cells (ISCs) and secretory progenitor cells are both reported to replenish damaged cells and contribute to crypt regeneration. However, the suppressed β-catenin/c-MYC axis within these slow-cycling cells leads to limited regenerative response to restore intestinal integrity during fatal accidental injury. Current study demonstrates that post-IR overexpression of TIGAR, a critical downstream target of c-MYC in mouse intestine, mounts a hyperplastic response in Bmi1-creERT+ reserve ISCs, and thus rescues mice from lethal IR exposure. Critically, by eliminating damaging reactive oxygen species (ROS) yet retaining the proliferative ROS signals, TIGAR-overexpression enhances the activity of activator protein 1, which is indispensable for initiating reserve-ISC division after lethal radiation. In addition, it is identified that TIGAR-induction exclusively gears the Lgr5− subpopulation of reserve ISCs to regenerate crypts, and intestinal TIGAR-overexpression displays equivalent intestinal reconstruction to reserve-ISC-restricted TIGAR-induction. Our findings imply that precise administrations toward Lgr5− reserve ISCs are promising strategies for unpredictable lethal injury, and TIGAR can be employed as a therapeutic target for unexpected radiation-induced GIS.


Introduction
Unexpected radiation exposure during terrorist events (e.g., the use of "dirty bombs"), industrial or nuclear accidents (such as the nuclear disasters in Chernobyl and Fukushima) is a current and continuing threat to the future. Under homeostatic conditions, the rapid turnover of the intestinal epithelium is driven by leucine-rich repeat-containing G protein-coupled receptor 5 (Lgr5) high intestinal stem cells (ISCs), which are especially vulnerable to high-dose ionizing radiation (IR) 1,2 . A dose of 15 gray (Gy) of radiation is sufficient to abrogate the proliferative output of these mitotically active Lgr5 high ISCs, and thus causes severe acute damage of the epithelial integrity 2 . Within 7 days of high-dose IR exposure, mice suffered from diarrhea, malabsorption and weight loss always die with complications known as gastrointestinal syndrome (GIS). Although prophylactic administrations have demonstrated some desirable effects on preventing stem cell exhaustion and epithelial disintegration induced by high-dose IR exposure [3][4][5] , the current post-IR treatments based on natural products and antioxidants have shown very limited effects on reversing stem cell death and the deadly GIS [6][7][8] .
Besides the high-proliferating and radiosensitive Lgr5 high ISCs (i.e., crypt base columnar cells (CBCs)), a slow-cycling and injury-resistant pool of stem cells could be arisen to divide when the CBCs are depleted 9, 10 . These rare "+4" position cells mainly include the reserve ISCs marked by lineage tracing analysis with polycomb complex protein 1 (Bmi1)-creERT 11,12 and Lgr5 + label-retaining secretory progenitor cells which are regarded functionally distinct from reserve ISCs 13,14 . These radioresistant "+4" position cells are low-proliferative under homeostasis, while become proliferative from 3-4 days after high-dose radiation 15,16 . However, during lethal IR exposure, the CBCs are exhausted rapidly and the intestinal epithelium always disintegrates around 5 days after radiation, which happens even prior to effective "+4"-position-cell division and crypt regeneration. Hence, further elucidation of the mechanisms leading these quiescent cells to division after lethal IR-injury is required for mitigating fatal GIS.
The Wnt/β-catenin/c-MYC axis plays a central role in regulating the division of ISCs. However, the suppressed β-catenin/c-MYC pathway within "+4" position cells results in limited regenerative response within 3 days after lethal radiation 16 . Therefore, targeting β-catenin/c-MYC signal after lethal IR-injury may be potential countermeasures for accelerating the regeneration of these quiescent cells and the intestinal epithelium. TP53induced glycolysis and apoptosis regulator (TIGAR), a downstream target of c-MYC in mouse intestinal crypts 17 , has been indicated to be a critical scavenger of reactive oxygen species (ROS), which promotes DNA damage repair and cellular redox balance during genotoxic stress 18,19 .
In the present study, we demonstrate that overexpression of TIGAR may be promising in ameliorating the intestinal architecture and survival during unpredictable lethal injury. Mechanistically, TIGAR acts as a turnon switch that facilitates cell division of Lgr5 − reserve ISCs in an activator protein 1 (AP-1) dependent manner, which remedies the β-catenin/c-MYC-inhibited "defect" of these cells and gears crypt regeneration efficiently after lethal IR-injury.
TIGAR-induction exclusively gears Lgr5reserve ISCs to regenerate crypts By asymmetric division, a single reserve ISC could generate a daughter cell and an Lgr5 + CBC to replenish the active stem cell compartment 15 . The active CBCs then either generate transit-amplifying (TA) cells, which divide rapidly to produce large quantities of enterocytes, or differentiate into secretory progenitor cells which commit to Paneth cells, goblet cells, or enteroendocrine cells 10 .

TIGAR accelerates reserve ISCs toward division by limiting damaging ROS
It was noteworthy that redundant TIGAR activity failed to gear reserve-ISC division during homeostatic conditions. Upon a pulse of 4-OHT in vitro and TIGARoverexpression within reserve ISCs, the morphology and dynamics of TIGAR-overexpressing organoids remained the same as that of WT cohorts ( Supplementary Fig. 3a,   b). The reason might be attributed to another "initiating signal", which was essential for gearing reserve ISCs toward regeneration. It was reported that the proliferative ROS signal was pivotal for CBC division and proliferation 27 . With the administration of N-acetyl L-cysteine (NAC), a traditional antioxidant which indiscriminately scavenged the damaging ROS and pro-proliferating ROS, the regeneration of reserve ISCs was examined to determine whether IR-induced pro-proliferation ROS acted as "initiating signal" in accelerating reserve-ISC division. As shown in Fig. 4a-d, NAC treatment could only drive the Bmi1-creERT + cell division to a limited extent, which was far away from that of TIGAR-overexpression did. In vitro analysis also revealed that the AP-1 activity within Bmi1-creERT + cells after 12-Gy irradiation was modestly enhanced by the NAC administration, with the degree lagging far behind the TIGAR-overexpressing cohort's ( Fig. 4e, f). To confirm whether TIGAR-induced crypt regeneration was in a dose-dependent pattern, intestinal organoids derived from both homozygous Villin-creERT2; H11-Tigar +/+ (Villin-creERT2;H11-Tigar) mice (Fig. 4g) and heterozygous Villin-creERT2;H11-Tigar +/− mice (Fig.  4h) were irradiated and stimulated by 4-OHT immediately post-IR. Ki67-based immunofluorescence assay illustrated that both homozygous and heterozygous TIGAR-overexpressing miniguts moved to proliferative phases that were notable at 3-5 days after irradiation (Fig.  4i, j). The degree, however, in the homozygous TIGARoverexpressing organoids was considerably higher than that in heterozygous ones (Fig. 4i, j), whose TIGAR expression level was between WT organoids and homozygous organoids. These data further confirmed that IRinduced pro-proliferating ROS, which was not scavenged by TIGAR, might be a critical "initiating signal" for gearing reserve ISCs toward regeneration. This mechanism also explained why preclinical treatments simply based on traditional antioxidants had very limited effects on reversing intestinal disintegration and lethal GIS.
survival rate (Fig. 5f) of Villin-creERT2;H11-Tigar mice resembled those of Bmi1-creERT;H11-Tigar mice after lethal irradiation, indicating that TIGAR-induction failed to promote Lgr5 + secretory progenitor cell division to support crypt regeneration, even if the reserve ISCs were already accelerated to proliferation. Furthermore, the data also revealed that the contribution of other epithelial populations to crypt regeneration might be vanishingly small. Actually, although crypt cells such as Alpi-CreERmarked TA cells are reported to repopulate the crypt compartment upon IR-injury, little evidence supports their functional importance in epithelial regeneration 28 . Meanwhile, IR-evoked apoptosis of intestinal crypts within 24 h post-WAI was almost the same no matter TIGAR was overexpressed or not ( Fig. 5g-j), suggesting that the post-IR treatment applied in the current study failed to attenuate WAI-induced crypt cell death. Hence, it was summarized that the amelioration of intestinal integrity induced by TIGAR-overexpression after lethal IR-injury was predominantly attributed to the acceleration of Bmi1-creERT + reserve-ISC division.

Discussion
A two-stem cell model is supported by burgeoning studies from the small intestine, involving an actively cycling but radiosensitive stem cell and a long-lived, injury-resistant reserve pool of ISCs which is regarded to reside upstream of the high-proliferating CBCs 11,24,29,30 . Classic theories of radiobiology demonstrate that cell's radiosensitivity is positively correlated with the proliferative activity. Indeed, relieving the proliferative suppression of the reserve pool of ISCs before irradiation can result in enhanced epithelial radiosensitivity and aggravated GIS 16 . Conversely, if the suppressed cell division of low-proliferating stem-like cells cannot be released in time, the intestinal integrity may also fail to be regenerated after lethal IR-injury. Using "cre-loxp" mouse model, the present study indicates the possibility of TIGAR-based post-IR treatment in accelerating reserve-ISC division and ameliorating mouse survival under grievous GIS.
To establish the involvement of TIGAR in driving the intestinal regeneration, we applied mouse models of which TIGAR could be efficiently induced 18 h after stimulation in vivo ( Supplementary Fig. 4a-f). During WAI, the head, neck, thorax, and extremities were shielded to protect the bone marrow (Fig. 1b), thus inducing predominant GIS 3,4 . Single intraperitoneal injection of tamoxifen was performed immediately after 15-Gy WAI to induce TIGAR expression timely. After lethal WAI exposure, Bmi1-creERT;H11-Tigar and Villin-creERT2; H11-Tigar mice revealed equivalent amelioration of intestinal epithelial integrity (Fig. 5c-e) and mouse survival (Fig. 5f), indicating that TIGAR-overexpression primarily accelerated reserve ISCs toward division to reestablish the intestinal architecture after lethal irradiation. It is worth noting that biomarkers of "quiescent" reserve ISCs are also found in a subpopulation of Lgr5 + crypt cells, while around 20% of Lgr5 + intestinal cells are largely quiescent 13,24 . This quiescent Lgr5 + population is mainly comprised of the Lgr5 + (label-retaining) secretory progenitors and a subpopulation of the Bmi1-creERT + reserve ISCs. Critically, by lineage cell tracing analysis, the feasibility of TIGAR-overexpressing quiescent Lgr5 + cells in rescuing mice from lethal GIS was ruled out (Fig. 2i-k; and Supplementary Fig. 1f-h). The mechanisms might be roughly attributed to the following two reasons. On the one hand, when compared with the exact quiescent Bmi1-creERT + reserve ISCs, quiescent Lgr5 + cells were reported to demonstrate much fewer tracing events in response to injury 13,14 , which made effective crypt regeneration incapable after lethal WAI exposure. On the other hand, the Lgr5 + characteristics endowed these cells with higher radiosensitivity 31 , which made them already lose viability or undergo apoptosis before TIGAR was introduced (Fig.  5h, j). However, the present study does not eliminate the indispensability of the de novo-generated Lgr5 high CBCs in intestinal regeneration after lethal IR-injury 32 .
Based on the lineage tracing analysis, a recent study indicated that the Bmi1 + cancer stem cells possessed an increased AP-1 activity that drove tumor recurrence 33 , suggesting that AP-1 played critical roles in endowing Bmi1-creERT + stem cells with proliferative potential. In the present study, a classical inhibitor of AP-1, 3-PA, was used to demonstrate the mechanism of TIGAR-induced proliferation after lethal IR. A significant abrogation of Bmi1-creERT + cell division, especially the first asymmetric division at the early stage (1 day) post-IR, was observed when the transcriptional activity of AP-1 was inhibited by 3-PA (Fig. 3j, k). Interestingly, AP-1 abolishment only dramatically abrogated the Bmi1-creERT + lineage after irradiation, but did not affect the proliferative activity of CBCs during homeostatic conditions (Supplementary Fig. 2c, d). This finding suggests that the AP-1 activity is dispensable for CBC-like stem cells during homeostasis, which might attribute to the high proliferative activities of Lgr5 high CBCs endowed by the Wnt/β-catenin signals 31,[34][35][36] . This also suggested that TIGAR-induction remedied the β-catenininactivated "defect" of the low-proliferating reserve ISCs, which facilitated the acceleration of cell division and crypt regeneration after lethal IR-injury (Fig. 6). Mechanically, TIGAR-induced activation of c-Fos/AP-1 might be attributed to the increased phosphorylation of c-Fos, but not the upregulation of c-Fos expression. In conclusion, the current study indicates that during unexpected disasters, quiescent Lgr5 − reserve ISCs can be awakened timely by TIGAR/AP-1 activation to reestablish intestinal architecture and ameliorate mouse survival. Meanwhile, our work reveals an unexplored role of TIGAR in accelerating reserve ISCs toward regeneration, and the capability of TIGAR-induction in activating AP-1 demonstrates its significant advantage over traditional antioxidant treatments (Fig. 6). If not otherwise stated, only male mice were used and littermates were randomly and blindly allocated to experimental groups. All experiments were conducted at 8-10 weeks of age. Genotyping was performed following the protocols of Jackson Laboratory. The study was conducted in compliance with local animal welfare laws, guidelines, and policies. All procedures were approved by the ethic committee of Soochow University (Approval No. ECSU-2019000150).

Mouse irradiation and tamoxifen administration
Mice weighing between 24 and 27 g were anesthetized and treated with a single dose of 15-Gy WAI at a dose rate of 1.6 Gy/min using an X-RAD 320iX Biological Irradiator (Precision X-ray, North Branford, CT, USA). A 3-cm area of the mice containing the gastrointestinal tract was irradiated (irradiation field), shielding the head, neck and upper thorax as well as lower and upper extremities and protecting a significant portion of the bone marrow. Immediately after irradiation, mice were injected with tamoxifen intraperitoneally. Tamoxifen (Sigma, Cat#T5648) was dissolved in corn oil (Sigma, Cat#C8267) at a final concentration of 20 mg/ml. Cre enzyme was induced by single injection of tamoxifen at a dose of 4.5 mg per 20 g body weight. The schedules for tamoxifen administration and radiation as well as mouse grouping were provided in the relevant figure legends. If not otherwise stated, H11-Tigar mice were used as controls and were given similar doses of tamoxifen.

Survival rate and small-intestine length
After 15-Gy WAI, the survival rate of the mice was monitored every day for up to 30 days. For intestinal length measurement, mice died before day 5 post-WAI were excluded, and the mice alive were administered with euthanasia on day 5 after 15-Gy WAI to analyze the small intestinal length.

Histology of small intestine
At day 1, 3, 5 after 15-Gy WAI, mice were administered with euthanasia, and the proximal small intestines were excised for histology. Small intestine tissues were fixed in 10% neutral-buffered formalin overnight. After embedded in paraffin, tissues were cut into 5-μm sections for haematoxylin and eosin (H&E) staining and observation.

Crypt isolation
Mouse small intestine was cut open longitudinally and washed with cold PBS. The villi were collected by scraping the intestine with a microscope slide and stored for protein analysis. While the remaining small intestine was cut into 5-mm pieces and incubated in PBS with 2 mM EDTA for 30 min at 4°C. After incubation, the tissue fragments were separated by vigorous shaking. The supernatant enriched for crypts was passed through a 70-μm cell strainer (BD Falcon, Cat#352350). After centrifugation, the final fraction of intestinal crypts was used for further culture.

Organoid irradiation and transfection
After 12-Gy irradiation in vitro, transfection of Tigaroverexpressing adenovirus (Adv-CMV-Tigar-3flag) was performed by mechanical separating the organoids from Matrigel, and 4-OHT (10 nM) was added into the medium soon after replanting the organoids. After 24 h, the medium was replaced by normal organoid culture medium.

Fluorescence-activated cell sorting (FACS)
The intestine was cut open longitudinally and incubated with 2 mM EDTA solution at 4°C for 30 min to isolate intestinal crypts. To generate a single cell suspension, cells were incubated with Accutase (BD Biosciences, Cat#561527) at 37°C for 10 min. Flow cytometry analysis was performed with CELL SORTER SH800S (SONY, Japan). Cells were gated for single cell based on the profiles of Forward-scatter area versus backward-scatter area (FSC-A vs. BSC-A) and forward-scatter height versus forward-scatter width (FSC-H vs. FSC-W). The size of the nozzle for all sorting is 100 µm.

Analysis of AP-1 activation
The DNA-binding activity of AP-1 was measured using the Trans AM kits (Active Motif, Cat#44096). Nuclear extracts of Bmi1-creERT + cells containing c-Fos/AP-1 factors were added into the multi-well plates precoated with consensus double-stranded DNA oligomers. After incubation, the transcription factor bound to DNA sequences was detected by using antibodies against c-Fos according to the manufacturer's protocol. The absorbance was examined by a Microplate Reader (BioTek, Synergy2, Winooski, VT, USA).

3-PA and NAC treatments
3-PA (MedChem Express, Cat#HY-12270) was dissolved in polyvinylpyrrolidone. For in vivo administration, mice were administrated with 3-PA (120 mg/kg body weight, i.g.) daily for 4 consecutive days before 15-Gy WAI and the subsequent tamoxifen induction. For in vitro administration, 3-PA (10 μM) was added into the organoid growth medium 1 day before 12-Gy irradiation. For NAC treatment, the organoids were treated with NAC (Sigma, Cat#A8199) soon after the irradiation at concentrations of 1.5 mM or 4.0 mM, respectively.

Statistical analysis
Data were expressed as mean ± SD from three independent determinations. Differences between groups with similar variance were analyzed by Student's t test. Kaplan-Meier survival analysis and log-rank comparison were performed for survival studies. Asterisks represent the p values as follows: *p < 0.05, **p < 0.01 and ***p < 0.001.