Sox9+ Cell is Required for Salivary Gland Regeneration after Radiation Damage via Wnt/β-Catenin Pathway

Background: Radiotherapy for head and neck cancer can cause serious side effects, including severe damage to the salivary glands, resulting in symptoms such as xerostomia, dental caries, oral infectious and so on. Due to lack of long-term treatment for the symptoms of saliva barren, current research has focused on nding endogenous stem cells that can differentiate into various cell lineage to replace lost tissue and restore function. Results: In our study, we identied Sox9 + cells can differentiate into various salivary epithelial cell lineages under homeostatic conditions. After ablating Sox9 + cells, the salivary glands of irradiated mice showed more severe phenotypes and reduced proliferative capacity. Analysis of online single cell RNA-sequencing data revealed enrichment of the Wnt/β-catenin pathway in Sox9 + cell population. Furthermore, treatment of Wnt/β-catenin inhibitor to irradiated mice inhibited the regenerative capability of Sox9 + cells. Finally, we showed that Sox9 + cells were able to form organoids in vitro and transplanting these organoids into salivary glands after radiation restored part of salivary gland function. Conclusions: In short, our research indicated that regenerative therapy targeting Sox9 + cells is a promising method to solve the radiation induced salivary gland injury.


Background
Head and neck cancer is the sixth most common cancer worldwide, with over 600,000 cases and around 300,000 deaths per year (1) .At present, surgical removal of the primary tumor is mainly used for head and neck cancer, combined with radiotherapy and chemotherapy. Unfortunately, radiotherapy of the head and neck inevitably leads to organic damage and dysfunction of the salivary glands including dry mouth, tooth decay, oral infection, and impaired taste and speech, seriously affecting the patient's quality of life (2). The current treatment methods for radiation-induced salivary gland damage mainly depend on short-term relief from the symptoms, such as commercial saliva substitutes, but short of long-term restorative therapies.
In clinical practice, radiotherapy may damage other organs, including lungs, heart, bladder(3), but there are few, if any, options available to restore lost tissue. Recently, emerging studies about radiation-induced damage concentrate on regenerative strategies such as stimulating the potential of endogenous stem cells or transplanting non-irradiated stem cells (4)(5)(6). However, the identi cation of stem cells under homeostatic or injury conditions is a roadblock in the application of regeneration strategies. Although recent studies have shown that acinar cells are derived through self-duplication following duct ligation injury, their cell subpopulation have not been con rmed (7). It is signi cant to identify whether the regeneration of acinar under radiation damage is self-duplication or derived from stem cells.
Over the last several years, the use of genetic lineage tracing to map fate of stem cells revealed the dynamics of cell fate-speci c patterns during development, tissue maintenance and repair(8). Salivary glands are composed of four main lineages, including the acinar, ductal, myoepithelial and basal cells.
Prior studies have demonstrated that the basal and myoepithelial cell populations widely expressing keratin 14 (K14) and keratin 5(K5) function as multipotent stem cells and contribute to all epithelial cell lineages during development (9,10). However, in adult salivary gland, K14 + cells remained lineage restricted and lost differentiation potential (11,12). Lineage tracing experiments con rmed that the sexdetermining Region Y (SRY) box (Sox) family members Sox9 contributes to acinar and ductal cells during the salivary gland development (13). However, in adult salivary glands, whether Sox9 + cells can remain lineage restrict and maintain the subpopulation of acinar cells under homeostatic or injury conditions remains unknown.
In this study, we performed in vivo genetic lineage tracing experiments in the mouse submandibular gland and unexpectedly discover that Sox9 + cells have the ability of repopulating lost functional compartment after radiation-induced damage. Analysis of single-cell RNA-sequencing (scRNA-seq) online dataset of salivary gland, show that Wnt/β-catenin pathway was enriched in cells with high Sox9 expression. Using inhibitor to block the Wnt/β-catenin pathway has indeed suppressed the repair and regeneration ability of the salivary glands. In addition, an emerging three-dimension organoid culture method that can recapitulate the structure, functionality and genetic characteristics of the original tissues has been established as an important tool for many basic biological applications, such as organoid transplantation (14). We transplanted Sox9 + cells into the submandibular glands of mice injured by radiation and found that successfully transplanted organoids can substitute part of the damaged salivary gland tissue. To sum up, our data revealed the extensive regenerative capacity of Sox9 + cells in salivary glands and suggested that Sox9 + organoid may be an effective treatment of radiation-damaged salivary glands.

Sox9 + cells mainly generate acinar cells in the SMG under homeostatic conditions
To characterize the expression of Sox9 in the salivary glands of adult mice, we rst performed immunohistochemical (IHC) staining on salivary gland tissues of normal mice at different anatomic sites. We found presence of Sox9 protein in all three major paired (parotid, submaxillary and sublingual) salivary glands (Fig. 1A). However, we detected no positive signal of Sox9 IHC staining in the minor salivary glands (Fig. 1B), suggesting possible different roles of Sox9 in salivary glands at different anatomic sites. Because submandibular gland (SMG) is the largest salivary secreting organ in mice, we then used SMG for our following study. Sox9 + cells have been shown to be a progenitor population that can produce acinar and ducts during embryonic developmental stage [15] . However, their roles in adult murine salivary gland remain unknown. To investigate that, we injected a single of tamoxifen to adult Sox9creER; Rosa-tdTomato mouse and traced Sox9 + cells for different periods of time under homeostatic conditions (Fig. 1C). Two days after the initial tamoxifen administration, we found that Tomato + cells were mainly overlapped with the Aqp5 + acinar cells, but not in the Krt18 + ductal cells, Acta2 + myoepithelial cells or Krt14 + basal cells (Fig. 1D). After 7 days, we detected expansion of Tomato + clones in acinar cell compartment (Fig. 1D). Notably, Tomato + cells were observed in different cell lineages, including Aqp5 + acinar cells, Krt18 + ductal cells, Acta2 + myoepithelial cells and Krt14 + basal cells 30 days after initial injection of tamoxifen (Fig. 1D). 90 days after tracing, we found an increased number of Aqp5 + /Tomato + double positive cells (58.75% ± 3.59%) Acta2 + /Tomato + (2.03% ± 0.81%), Krt14 + /Tomato + (3.64% ± 0.48%)as well as Krt18 + /Tomato + 4.25% ± 0.9% double positive cells in the SMG tissues (Fig. 1D, E). Overall, our genetic lineage tracing studies demonstrated that Sox9 + cells function as multipotent cells contributing to all epithelial cell lineages under homeostatic conditions, and produce most part of acinar cells in these cell types.

Sox9 + cells are required for radiation-induced regeneration of SMG
To study the role of Sox9 + cells in the wound-healing process of salivary gland, we generated Sox9creER, Rosa-tdTomato, Rosa26-DTA mice, which allow us to speci cally ablate Sox9 + cells upon tamoxifen administration. Our results demonstrated that Sox9 + cells underwent cell death as a result of diphtheria toxin A chain (DTA) expression ( Fig. 2A, 2E). We then perform a functional evaluation of the salivary gland after radiation exposure and found that there is a signi cant decrease in saliva secretion in the Sox9creER, Rosa-tdTomato, Rosa26-DTA compared with control Sox9creER, Rosa-tdTomato mice after a month (Fig. 2B). Furthermore, H&E and Masson's Trichrome staining also revealed that the experimental group suffered more severe morphological damage, such as massive destruction of acinar cells, and wider range of brosis than the control group after radiation (Fig. 2C, 2D). Results from TUNEL assay and Ki67 IHC staining indicated that ablating Sox9 + cells of SMG caused highly apoptosis and lower proliferation in SMG tissue (Fig. 2F, 2G). Moreover, compared to control group, the SMG in Sox9creER, Rosa-tdTomato, Rosa26-DTA displayed a greater loss of acinar cells and ductal cells and higher degree of tissue architectural disruption (Fig. 3A, 3C). 90 days after tracing, we found about 47.55% ± 3.18% Aqp5 + / Tomato + double positive cells than the control group (Fig. 3B). Surprisingly, Krt18 + /Tomato + (11.7% ± 1.27%)double positive cells increased dramatically compared with homeostatic conditions in the SMG tissues (Fig. 3D).In summary, Sox9 + cells play a critical role during regeneration and repair process of SMG after radiation injury.
scRNA-seq analysis revealed activation of Wnt/β-catenin pathway in Sox9 + cells To investigate the possible mechanism in regulating Sox9 + SMG cells, online single-cell RNA-sequencing (scRNA-seq) datasets regarding cell populations involved in salivary gland development and maintenance was downloaded and analyzed. Since our study mainly focused on the adult salivary gland tissue, we only utilized the dataset generated using adult mouse. We clustered cells from online dataset into six different categories, including acinar cells, basal cells, ductal cells, endothelial cells, monocytes and myoepithelial cells (Fig. 4A). Subsequently, we extracted acinar cells and reclustered them into Sox9 + cells and Sox9 − cells population according to the Sox9 expression level (Fig. 4B). Although cell cycle analysis showed no signi cant difference between these two populations ( Fig. 4C), KEGG enrichment analysis showed several pathways, including Wnt/β-catenin pathway, cytokine receptor interaction and extracellular matrix receptor interaction, were signi cantly enriched in Sox9 + cells (Fig. 4D). Since intensive studies have support a critical role Wnt/β-catenin pathway in the regulation of the stem cell fate and adult tissue regeneration(15), we postulated it might take part in the regulation of Sox9 + SMG cells.
We then examined the Wnt/β-catenin pathway targets Ccnd1, Tbx4 and found that they overlap with Sox9 + cells population (Fig. 4E). In conclusion, the results of single-cell sequencing analysis indicated that the Wnt/β-catenin pathway might be important in the maintenance of Sox9 + cell population.

Wnt/β-catenin pathway inhibitors aggravate salivary gland radiation damage
To test whether blocking Wnt/β-catenin pathway can affect the regenerative capacity of Sox9 + cells in salivary gland, a highly effective inhibitor (XAV-939) of Wnt/β-catenin pathway was applied (Fig. 5A). Saliva ow examination of the SMG exhibited a substantial decline in the XAV-939 treated group ( Fig. 5B). Histological analysis showed that the submandibular gland displayed large areas of acini loss and more serious brosis after treatment of XAV-939 (Fig. 5C, D). Lineage tracing results revealed smaller Sox9 + clones were formed after treatment of XAV-939, indicated regenerative ability compromised (Fig. 5E). The ability to proliferate of the salivary gland epithelium is also greatly reduced (Fig. 5F, G). In addition, the cell apoptotic rate was higher in the treatment group compared with the controls (Fig. 5H, I).
Thus, our data indicated that Wnt/β-catenin pathway is essential for the regenerative capacity of Sox9 + cells.
Transplant of Sox9 + organoids improved the regeneration of damaged salivary glands To test whether Sox9 + cells are able to form organoid in vitro, we isolated Tomato + cells from SMG. In the culture conditions, salivary gland organoids models derived from Sox9creER; Rosa-tdTomato mouse typically developed within 4 weeks (Fig. 6A). Organoids showing strong tomato signal also displayed different forms, either a cystic or a lumen-like phenotype (Fig. 6B). The organoids retained characteristic of differentiated salivary gland by immunostaining for the classical markers, including Aqp5, Krt18, Acta2 and Krt14 (Fig. 6C). To test whether these organoids were able to regenerate mouse salivary glands, we collected about 200 organoids and injected them into wild-type mice submandibular gland that suffered from radiation damage. Then, we indeed observed that tomato signal appeared in submandibular glands and distributed in the acinar and duct (Fig. 6D). What's more, the function evaluation experiment of salivary glands also showed the positive effect of transplanting Sox9 + organoids (Fig. 6E). The above data implies that Sox9 + organoid can be expanded in vitro and supplement salivary gland cells damaged by radiation.

Discussion
Most patients suffering from head and neck cancer requires radiation therapy in combination with other treatment, including surgical resection, radiotherapy and chemotherapy(16). However, radiotherapy around the head and neck will inevitably destroy salivary glands, resulting in impairment of saliva secretion and decline in quality of life. Our study using a murine model to characterize repair and regeneration revealed that Sox9 + cells as progenitors in the adult salivary gland was essential to the replenishment of acinar and ductal cells after radiation-induced damage. Hence, our study provides some insights for the feasibility of using stem cells to treat damaged salivary glands.
Over the last several years, effort has been made to identify stem/progenitor cell in salivary gland.
Studies have shown that these cell populations have an impact on all aspects of salivary gland biology, including cell fate speci cation during development, adult tissue regeneration and the role of stem cell in carcinogenesis(8). Previous research discovered that Sox9 + cells played an essential role as a biomarker for progenitor multipotent progenitor cells in the developing pancreas, mammary gland, lung, liver and duodenum (17)(18)(19)(20)(21)(22). To explore the role of Sox9 + cells in adult salivary gland, by labeling Sox9 + cells, we monitored the cell fate trajectory and nd that Sox9 + cells contribute all cell lineages of salivary gland under homeostatic conditions. We further investigated that mice with ablation of Sox9 + cells using diphtheria toxin A chain (DTA) show more severe phenotype including structural destruction, increased apoptosis and severe brosis, compared with control groups. What's more, the study unexpectedly showed that Sox9 + cells can differentiate into acinar cells and ductal cells after injury. In contrast to Aure and co-workers' report showing that acinar cells replenish through self-duplication of mature cells (7), our result con rmed that Sox9 + cells serving as a progenitor maintain both the acinar and ductal cell lineages.
Using the prevailing single-cell sequencing method, KEGG enrichment analysis discovered Wnt/β-catenin pathway, which is closely related to cell differentiation. Sox9 + cells can be used as progenitor cells to differentiate into various types of cells so as to restore missing structure. And our application of Wnt/βcatenin pathway inhibitors has also proved that blocking this pathway will make it di cult for the injured salivary glands to be repaired and then behave more serious phenotype.
Recently, many studies designed to exploit the potential of stem cell population to regenerate the missing radiation damage cell structures (23,24). Organoid technology, as an emerging independent tool, offer enormous possibilities for the basic research on tissue development, stem cell characteristics(25), regeneration medicine etc. (26,27). Studies have shown that intestinal stem cell organoid transplantation can restore the function of the intestinal mucosa and treat bile acid malabsorption in rats(28). Our experiment used the model to reconstruct the damaged structure and achieved good results. However, we also found that the organoids did not proliferate in the salivary glands one month after transplantation, which may be due to the insu cient recovery time. Anyway, the method, organoids transplantation, provides a good idea for the subsequent treatment of salivary gland radiation damage.

Conclusion
Taken together, our data highlight Sox9 + cells in the salivary gland, as a progenitor cell population, can occurs the expansion and differentiation capacity through Wnt/β-catenin pathway. Combined organoid transplantation can effectively restore the function of damaged salivary glands. Based on the above research, targeting Sox9 + cells have potential therapeutic bene ts against radiation-damaged salivary glands.

Mouse lines
All animal experiments and procedures were approved by sun yat-sen University Institutional Animal Care and Use Committee (IACUC) and complied with the guidelines for the use and care of laboratory animals. Sox9creER, Rosa-tdTomato and Rosa-DTA mice were purchase from the Jackson Laboratory.
Lineage tracing experiment 6-8week old adult Sox9creER, Rosa-tdTomato mice received tamoxifen (Sigma, T5648-1G) dissolved in corn oil by intraperitoneal injection with dose of 0.08 mg/g body weight every day for three consecutive days. Mice were anesthetized by 1% pentobarbital with 0.05mg/g body weight and submandibular glands were harvested at 1st day, 1st week, 1st month, 2nd months, and 3rd months post-injection. To exclude sexual dimorphism of murine salivary glands, all experiments in this study were performed in both sexes.

Radiation-induced damage experiment
Sox9creER, Rosa-tdTomato mice and Sox9creER, Rosa-tdTomato and Rosa-DTA mice were anesthetized before irradiation. They were placed in a radiation protection lead box and only exposed neck and part of the head. Each mouse received a single dose of 15Gy from a Cs137 gamma radiation source (Rs2000, USA). All mice were given soft diet and clear water after recovery from anesthesia. Their submandibular glands were harvested at 1st,7th,15th,30th,60th and 90th. For Wnt/β-catenin pathway inhibitor treatment, each irradiated Sox9creER, Rosa-tdTomato mice were intraperitoneally injected with Wnt/β-catenin pathway inhibitor (XAV-939 5ul/g) (X3004-5MG, Sigma-Aldrich) for ve consecutive days. The control group was injected with corn oil at the same time. These submandibular gland samples were collected at 1st, 3rd, 7th, 15th, 30th, 60th, 90th.

Saliva collection and analysis
All mice were intraperitoneally injected with 1% pentobarbital with 0.05mg/g body weight to keep the mouse still. About two minutes later, pilocarpine (0.5mg/kg i.p.) was injected into their bodies by intraperitoneal injection. Fifteen minutes later, saliva was collected via cotton ball in pre-weighed Ep tubes. 1mg saliva can be calculated as 1ul. Finally calculate the salivary gland ow rate(ul/min).
The submandibular gland samples were xed in 4% paraformaldehyde (biosharp, BL539A) at 4°C overnight. After xation, salivary glands were processed for OCT and para n embedding. 8-um frozen sections were cut and store at -20°C. 5um para n sections were cut using a microtome (Leica) and stored at room temperature.

Hematoxylin and Eosin Staining and Masson's trichrome staining
After depara nization and hydration, para n sections were stained with hematoxylin and eosin according to standard protocol (Solarbio, G1120-100). Masson's trichrome staining was performed following the standard protocol (G1343, Solarbio). Then carry out ethanol dehydration and neutral resin sealing.

Immunohistochemical staining
For immunohistochemical studies, 5 um para n sections were depara nized and rehydrated. They are incubated in a 3% hydrogen peroxide solution for 10 minutes to remove endogenous peroxidase activity. Then antigen retrieval was performed with sodium citrate buffer (pH 6.0) using Microwave oven. After incubating in peroxide blocking buffe (BSA) for 30 minutes in 37°C thermostat, sections were incubated with rabbit monoclonal anti-Sox9 antibody (1:200, CST, 82630S) at 4°C overnight, followed by 1 hour incubation with the goat anti-rabbit kit (SA1022, BOSTER). Finally, use DAB (AR1022, BOSTER) to develop the color under the microscope. For peroxidase reactions, color was developed for 5 mins in DAB solution and counterstained with hematoxylin for1 minute.
Single-cell RNA-sequencing (scRNA-seq) analysis The single-cell sequencing data comes from the Gene Expression Omnibus (GEO) database, and the number is GSE150327. The data was analyzed using Seurat R package.
Organoid culture and transplantation Salivary gland samples from Sox9creER, Rosa-tdTomato mice were used for isolation and development of ex vivo organoid culture. After washed gently in cold PBS, tissues were cut into approximately 1-mm pieces and washed again with cold PBS. Clean tissue was then incubated in Type IV Collagenase for 30 minutes with shaking from time to time in a 37℃ thermostat. Then the pieces were passed through 70µm lters (BD Biosciences) and centrifuged at 270g for 5 min at 4°C. Precipitate was resuspended in 1 mL cold DMEM-F12(C11330500BT-1) and centrifugated at 200×g for 5 min. Precipitate were resuspended in cold Matrigel (Corning #356231) and cold media in a 50:50 ratio. Then plant the mixture in a 24-well plate and place it in a 37 ℃ incubator for 15 minutes to solidify the Matrigel. Finally, stem cell culture media was added from the side of the well. For organoid transplantation, 100-150 salivary gland organoid were collected and transplanted into wild type mice exposed to neck irradiation. Samples were collected one month after transplantation.

Statistical Analysis
Positive signals are counted with ImageJ. All statistical analyses were performed using GraphPad Prism 8.0 software. Quantity results were reported as mean ± standard deviation (S.D.) of three or more independent experiments.

Declarations Ethics Statement
All animal experiments and procedures were approved by sun yat-sen University Institutional Animal Care and Use Committee (IACUC) and complied with the guidelines for the use and care of laboratory animals.

Consent for publication
Not applicable

Availability of data and materials
The single-cell RNA-seq data analysed during the current study is available at the NCBI's Gene Expression Omnibus (GEO) data repository with the accession ID GSE150327. https://www.ncbi.nlm.nih.gov/search/all/?term=GSE150327 Author Contributions DC, CW, YB and TJL conceived and designed this project. XX and GX conducted experiments and interpreted results of experiments. MZ, XJX and SC drafted paper, edited and revised manuscript. All authors read and approved the nal manuscript. Figure 1 A. H&E staining images of normal mice salivary glands includes parotid glands, sublingual glands, submandibular glands and minor salivary glands. Scale bar:50um B. Images of normal mice minor salivary glands, parotid glands, sublingual glands, and submandibular glands immunostained for Sox9. Scale bar:50um C. Sox9creER; Rosa-tdTomato mice were injected with tamoxifen for three days. Collect submandibular gland at 1st,3rd,7th,15th,30th,60th,90th. D. Representative images of Sox9 lineage-traced submandibular gland under homeostasis condition. At 2th,7th,30th,90th, immunostaining for acinar markers (Aqp5), ductal cell markers(K18), myoepithelial marker (Acta2) and basal cell marker (K14) respectively. Nuclei were stained with Dapi (blue) Scale bar:50um E. Aqp5+Sox9+cells, K18+Sox9+cells, Acta2+Sox9+ cells, K14+Sox9+ cells account for the percentage of total cells at 2nd,7th,30th,90th Figure 3 A. Representative immuno uorescence image of co-staining with acinar markers (Aqp5) after radiation damage at different time point between control and DTA groups. Scale bar, 50μm B. Quanti cation analyses of Aqp5+Tomato-cells, Aqp5+Tomato+ cells, Aqp5-Tomato+cells, other cells at the 90th day between control and DTA groups. C. Representative immuno uorescence image of co-staining with ductal cell markers (K18) after radiation damage at different time point between control and DTA groups.

Figure 4
A. t-SNE plot of normal mice salivary gland cells reveal consistent clusters. Clusters are assigned to indicated cell types by differentially expressed genes. B. UMAP plots that describe clustering results based on the Sox9 expression level of acinar cells C. Cell cycle score between Sox9+ cells clusters and Sox9-cells cluster. D. KEGG enrichment analysis indicated the statistically signi cant enriched pathways between Sox9+ cells clusters and Sox9-cells cluster. E. FeaturePlots that describe the expression level of Ccnd1, Tbx3, which are the target of Wnt/β-catenin pathway, between Sox9+ cells cluster and sox9-cells cluster.

Figure 5
A. After 15Gy radiation, Sox9creER; Rosa-tdTomato mice were injected with tamoxifen and XAV-939 for three days and ve days respectively (n = 3 per group). Collect submandibular gland at 1st, 3rd, 15th, 30th, 60th, 90th. B. Measure the salivary gland ow rate of the two groups (control and XAV-939 group) of mice at different time points. Data were analyzed using a multiple-tests or one-per row analysis. *p<0.05, **p<0.01, ***p<0.001 C. Representative images of H&E staining in the two groups at different time points. Scale bar:50um D. Representative images of Masson's trichrome staining in the two groups at different time points. Scale bar:50um E. Immunostaining images of Sox9+ cell-driven lineage tracing in mice injected with XAV-939 and control group. Scale bar, 50 μm. F. H Immunostaining of Ki67 and TUNEL in mice injected with XAV-939 and control group. Scale bar, 50 μm. G. I Immunostaining scores of Ki67 and TUNEL were counted respectively. * p<0.05, * * p<0.01, and * * * p<0.001.