INTRODUCTION

Head and neck cancer accounts for ∼4% of all cancers in the United States (1). The current standard of care for management can involve surgery, radiation therapy, chemotherapy, targeted therapy or a combination of these treatment modalities depending on various factors including the stages of cancer, the surgical accessibility of the tumor, and morbidity associated with each modality (2). In nearly 70% of the patients undergoing radiation therapy due to location of many of these oral cancers, non-diseased adjacent tissues such as salivary parenchyma is irreparably damaged due to its high sensitivity to radiation. This often results in radiation-induced “objective” hyposalivation or a “subjective” clinical condition referred to as xerostomia or dry mouth (2, 3). Complications of long-term radiation-induced hyposalivation include impaired taste, difficulties in speech, mastication and denture retention, increased risk of dental caries and an overall compromised quality-of-life (4). Management of hyposalivation often involves the use of cholinergic drugs such as pilocarpine (5) or cevimeline (6). The effectiveness of these systemic sialagogues often depends on the presence of functionally intact glandular tissue units (5). Recently, putative stem cells, that can restore the salivary gland functions upon orthotopic transplantation in radio-ablated salivary glands of recipient mice, have been reported in the submandibular glands of mice (7–10) and humans (11). To compare the quality and quantity of regenerative outcomes in transplant experiments, a standardized in vivo assay needs to be developed using preclinical animal models. However, there are several limitations with currently available preclinical models of hyposalivation. For example, the radiation dose used for many of these preclinical models range widely from 2 Gy to 30 Gy (12). Although there appears to be some consensus on 15 Gy for immunocompetent C57BL/6 mice (7–9, 13–15), but the radiation doses are not well characterized for immunodeficient and radiation-sensitive models such as NSG mice that are more relevant to xenotransplantation studies.

Furthermore, several methods of saliva collection from preclinical models after cholinergic stimulation have been reported including direct cannulation of excretory salivary ducts (16), vacuum pump suction of saliva from oral cavity (17), glass capillaries (18), micropipette (19–21), gravimetry using pre-weighed filter paper strips (22) or a swab method (23). However, the methods such as direct cannulation are associated with risk of injury precluding repeated sampling of the gland, while gravimetry, glass capillary and micropipette-based methods show inter-operator variability (23). Therefore, an instantaneous method to measure kinetics of stimulated reflex saliva reliably and repetitively with little inter-operator variability would be advantageous.

The influence of age on the functionality of salivary glands is often debatable. Some studies have highlighted progressive and minor decline in saliva production with increasing age (24–26), while others show aging does not have an impact on functionality of salivary glands in healthy individuals (27, 28). However, gender and hormone status in women are known to have an impact on salivary flow rates (29–31). Nevertheless, age and gender specific differences in the context of stimulated salivary flow rates are not well understood in immunodeficient mouse models.

A small population c-KIT receptor expressing cells were identified in human salivary glands as putative stem cells and their transplantation restored salivary gland functions in NSG mice (11). Since the NSG mice do not express hc-KITL and therefore may not fully support the engraftment of hc-KIT expressing putative stem cells.

To address these lacunae, we comprehensively characterized a transgenic line of highly immunodeficient NSG-SGM3 mice that produces hc-KITL and identified that 20-week-old female NSG-SGM3 mice serves as a novel and robust model for chronic hyposalivation studies and preclinical assessment of experimental therapies. Further, we validated the Schirmer's test as a simple and reliable method to assess kinetics of saliva flow in mice. Moreover, characterization of glands at necropsy revealed highly complex differences in sex, age, radiation dose and salivary gland subtype-specific effects of salivary-ablative precision irradiation in NSG-SGM3 mice.

MATERIALS AND METHODS

Salivary-Ablative Precision-Irradiation Set-up

Each mouse was anesthetized via isoflurane for the duration of the procedure via a nose cone delivery system on a bite block and placed in a feet-first prone (FFP) position in the irradiation chamber. A commercially available stereotactic stage (Model 900M, Kopf Instruments, Tujunga, CA) was modified to facilitate mounting to the X-RAD small animal irradiator system (SmART) (Precision X-ray Inc., North Bradford, CT) (32) (Fig. 1A). Opposing beam stereotactic radiation doses of a total 0 (or sham), 2.5, 5 or 7.5 Gy was delivered at 5 Gy/min dose rate using X-RAD SmART irradiator to the head and neck in a brain-sparing manner. The stereotactic X-RAD SmART irradiator set-up doesn't require additional blocking shield. Each mouse was volumetrically imaged with cone beam computed tomography (CBCT) using aluminum filter (60 kVp, 0.3 mA, 2.0-mm Al filter, 256 projections 0.2 mm 3-voxel size) and target was defined to ensure complete coverage of major salivary glands. Subsequently, the filter was switched to high-energy copper (irradiation) filter and beam is focused on the target using 10-mm-circular collimator for brain-sparing salivary-ablative radiation treatment.

FIG. 1

Establishment of preclinical humanized animal model of precision-radiation-induced salivary ablation. Panel A: Stereotactic X-RAD SmART irradiator stage set-up. Panel B: Dosimetry and angle of precision dose. Panel C: CBCT-guided target placement at the midline of the mouse neck region; (i) axial, (ii) sagittal, and (iii) temporal views. Panel D: Survival plot. Panel E: Surviving fractions of young (n = 19 male and 20 female) and old (n = 19 female) mice at 6 months after precision irradiation. Panel F: Change in body weight (in grams) of young and old mice over chronic timeline. Panel G: Chronic alopecia in neck region associated with precision doses above 5 Gy young female NSG-SGM3 mice.

RESULTS

To generate a preclinical model of chronic hyposalivation, we utilized 3D image-guided stereotactic X-RAD SmART to radioablate the major salivary glands of immunodeficient NSG-SGM3 mice (33). Details of the functionality of this irradiator are described elsewhere (34–36). Dosimetric calculations were performed with an in-house 1D “point-dose calculator” (PDC) tool developed by our medical physics group to enable efficient dose calculations (single prescription reference point on the central beam axis) (37). This simple 1D PDC was verified using the vendor supplied 3D Monte Carlo treatment planning system, SmART-ATP version 1.1 (SmART Scientific Solutions B.V., Maastricht, Netherlands), which is based on the open-source Monte Carlo code (EGSnrc/BEAMnrc)(38, 39) (Fig. 1A–C). The chosen beam arrangement was parallel opposed beams using a 10-mm-circular collimator, with tipped laterals (89° and 271°) aligned to match beam divergence along the brain edge to maximize brain sparing and achieve specific targeting of major salivary glands along the midline of the mouse in the neck region [Fig. 1C (i-iii)]. The individual young (approximately 20-week-old male or female) or old (more than 1-year-old female) mouse was restrained on a bite block using an anesthesia nose cone and imaged with CBCT. Subsequently, CBCT image-guidance defined target and single radiation dose ranging from 0 (sham) to 10 Gy (old mice) or 7.5 Gy (young mice) was delivered at target site, i.e., the three major salivary glands (Fig. 1B). The high accuracy of CBCT guided targeting in each mouse and the beam arrangement enabled precision dose delivery to whole-salivary-gland structures. Both sham-treated and precision-radiation treated mice were followed for 6 months.

After evaluating all strains of mice and irradiated anatomical sites, we found many differences on the used radiation dose, sex, age and/or duration of the study period ( Supplementary Table S1 (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE-21-00237.1.S1). For example, a single 15 Gy radiation dose on immunocompetent female C57BL/6 mice reduced ∼50% of saliva levels after 3 months postirradiation (40), and showed inflammatory lesions characteristic of human gland pathology (41). In contrast, Pringle et al. (11) showed that local irradiation with only a single 5 Gy dose was needed to achieve the same 50% reduction in saliva in immunodeficient NSG mice, implying that radiosensitivity between immunocompetent and immunodeficient mice is different. It is also important to note that a 5 Gy whole-body dose of radiation is lethal to NSG mice. While immunocompetent animals are informative and desirable models for studying damage to radio-ablated salivary glands, there is still an urgent need for an immunodeficient model that optimally supports the survival of human cells as part of regenerative therapies. In this study, we characterized the effect of precision radiation in a preclinical immunodeficient NSG-SGM3 model. Humanized NSG-SGM3 mice (33) are one of the most immunodeficient mouse strains. They were originally derived by crossing NOD/LtSz-scid IL2RG (NSG) mice (42) with NOD/LtSz-scid/scid (NOD/ SCID) mice, are engineered to express three human cytokines (43), and are extremely radiosensitive owing to the presence of the Prkdc^scid mutation (44). Using our CBCT precision dose of radiation, we found that NSG-SGM3 mice (young and old) tolerated doses up to 7.5 Gy (Fig. 1D–F). Increased mortality at 10 Gy in older mice precluded further life-style studies using this dose in NSG-SGM3 mice. We also observed that doses above 5 Gy led to permanent alopecia in the mandibular region, whereas 5 Gy itself showed transient alopecia. The dose of 2.5 Gy had no effect on hair follicles (Fig. 1G).

Next, we evaluated differences in saliva flow among the various groups. Saliva flow stimulation is a nerve-mediated reflex action modulated by the central nervous system, and this activity is an important indicator of functional salivary glands (45). In hyposalivation models, cholinergic drugs such as pilocarpine or carbachol are used to stimulate saliva secretion (11, 46, 47). Although there are several methods that have been reported to collect stimulated reflex saliva and evaluate changes in flowrate in vivo or ex vivo ( Supplementary Table S1 (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE21-00237.1.S1), we found that the field lacked an independent, easy and objective method for instantaneous kinetic measurement of stimulated-saliva flow in model organisms. Therefore, we adopted the Schirmer's test strip method to assess quantitative changes in salivary flowrate in NSG-SGM3 mice. Schirmer's test strips are routinely used in clinical practice to measure eye dryness (48) and hyposalivation in patients (49). Thus, we tested this method against the commonly used gravimetric method in our mice ( Supplementary Table S1 (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE21-00237.1.S1). We injected 2 mg/kg of pilocarpine subcutaneously into the dorsal flanks of the animal and started continuously collecting saliva 5 min after injection from the floor of mouth using pre-weighed filter paper or Schirmer's test strips for 30 min while changing the strips every 5 min. Saliva migration under the capillary action (mm) was recorded on Schirmer's test strips by making a permanent pencil mark, and weight of blotted filter paper strips was determined gravimetrically (Fig. 2A). Measurements of saliva fraction over the chronic timeline demonstrated a high degree of correlation between the Schirmer's test strip method and the conventionally used gravimetric method in both young male (R² = 0.90, P < 0.0001; Fig. 2B) and female (R² = 0.88, P < 0.0001; Fig. 2C) NSG-SGM3 mice. Thus, we continued with the Schirmer's test for further analysis. The cumulative percent distribution analysis demonstrated a significant difference in saliva flowrate between young- and old-female mice (Fig. 2D), but not between young male and female mice (Fig. 2E), suggesting age-dependent changes in salivary physiology. Indeed, our kinetic analysis of flowrate in sham-treated mice suggested that reflex-saliva flowrate peaked a few minutes earlier in old females compared to young females, but no such difference was noted in young males vs. females after stimulation (Fig. 2F and G). In all mice, regardless of sex and age, the saliva flowrate declined after peaking approximately 10 min, but the decline was more dramatic in old-female mice (Fig. 2F–I). Schirmer's test demonstrated age-specific (Fig. 2F) and sex-specific (Fig. 2G) differences in kinetics of stimulated-saliva flow captured over 30 min, as well as characteristic differences in flowrate at 10- and 25-min after stimulation (Fig. 2H and I). This trend was consistent throughout the chronic timeline in sham-treated mice (Fig. 3A and B). Taken together, our results suggest that Schirmer's test strips offer a rapid and reliable quantitative method to assess real-time changes in saliva flowrate in mice. Additionally, sexual dimorphism, as previously reported in submandibular glands due to the unique presence of granular convoluted tubular (GCT) structures in male mice (50), was also noticeable in NSG-SGM3 mice ( Supplementary Fig. S1 (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE-21-00237.1.S1).

FIG. 2

A novel method for kinetic study identifies age and sex specific differences in reflex-saliva flow in preclinical model. Workflow to measure pilocarpine-stimulated saliva flow by Schirmer's test strips and gravimetry (panel A). X-Y plot showing correlation between gravimetric method vs. Schirmer's test based detection of pilocarpine-stimulated saliva fraction in 19 young male (panel B) and 20 young female (panel C) precision-irradiated NSG-SGM3 mice. Cumulative Gaussian plot comparing young and old female mice (panel D) and age-matched male and female (young NSG-SGM3) mice (panel E). ***P < 0.0001, ns = not significant. Kinetics of pilocarpine-stimulated saliva flowrate (µg/min) in 3 week after sham-treated 5 young and 4 old female NSG-SGM3 mice (panel F) and in prior-to-sham/radiation-treated 20 male and 20 female young NSG-SGM3 mice (panel G). Plot showing sex- and age-specific differences in means of saliva flowrate at 10 and 25 mins after stimulation in young and old female mice shown in panel F (panel H) and in male and female mice in panel G (panel I). * P < 0.01.

FIG. 3

Precision irradiation establishes chronic hyposalivation in preclinical NSG-SGM3 model. Panel A: Kinetics of pilocarpine-stimulated saliva flowrate (µg/min) in mice treated with different precision- doses [19 males (n = 4–5 per group), 20 females (n = 5 per group) and 16 old females (n = 4 per group) were used]. Panel B: Sex- and age-specific significant differences in pilocarpine-stimulated saliva fraction in mice treated with different precision doses. Data in each sex group is calculated relative to sham-treated animals. Panel C: Precision-high-dose-irradiated mice displaying delayed onset of stimulated saliva flow in time (0, 2, 4, 6 months). Panel D: Barplot showing significant changes in saliva fraction at +3, +6 and +24 weeks in irradiated young male, female and old female mice relative sham-treated anim als.*P < 0.01; **P < 0.001 and ***P < 0.0001.

Then, we compared the kinetics of pilocarpine-stimulated saliva flow over the course of 30 min using Schirmer's test strips in young-male and -female mice, and old-female mice precision irradiated with 0, 2.5, 5 and 7.5 Gy doses. At 6 months postirradiation, we observed significantly reduced salivary flowrates at 10 min in mice precision irradiated with doses of 5 and 7.5 Gy compared to sham-treated controls (Fig. 3A and B). At 5 min after stimulation, saliva flowrate after 2.5 Gy precision irradiation was significantly lower in young males and old females, but not in young females compared to sham-treated mice. However, this difference disappeared at 10 min in males (Fig. 3A and B), suggesting important sex- and age-specific complex salivary physiology linked with different doses of precision radiation.

To define the chronic hyposalivation timeline in NSG-SGM3 model, we determined changes in stimulated-saliva fraction in precision-irradiated mice compared to sham-treated mice over 6 months postirradiation. We observed marked reduction in stimulated-saliva fraction 3 weeks after 2.5 Gy precision dose in young males and old females, but not in young females when compared to sham-irradiated animals (Fig. 3C). In young mice treated with 5 and 7.5 Gy, after about 12 weeks of fluctuations, saliva fraction stabilized at 50% compared to sham-treated animals, and this was sustained until the endpoint of 6 months (Fig. 3C and D). Interestingly, 2.5 Gy precision-irradiated mice showed a decline in stimulated-saliva fraction from 3 (males and old females) to 9 weeks (young females) after precision irradiation, but then showed partial restoration (Fig. 3C and D). In the case of old females, the stimulated-saliva fraction started marginally declining again after approximately 9 weeks (Fig. 3C and D). Of note, the sex-specific early responses of 2.5 Gy doses were not noticeable when the secreted fraction of saliva data from young males and females were combined ( Supplementary Figure S2 (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE-21-00237.1.S1).

Radiation-induced salivary gland injury is also associated with glandular shrinkage, loss of acinar cell area, inflammation, fibrosis, microvascular injury and atrophy (51). Therefore, at the endpoint of 6 months, we sacrificed precision-irradiated as well as sham-treated mice and harvested major salivary glands. In general, the weight of major salivary glands in old-female mice was significantly greater relative to their body weight than young mice (Fig. 4A). The weight of major salivary glands relative to their body weight reduced significantly in young- and old-female mice with increasing precision-radiation dose (Fig. 4B and C). Interestingly, we did not observe such correlation between precision-radiation dose and salivary gland weight in male mice across the radiation doses (Fig. 4B). Next, we prepared tissue sections from harvested parotid (PG), submandibular (SMG) and sublingual (SLG) glands of sham-treated old- and young-female mice or 7.5 Gy precision irradiated and performed histological and immunohistochemical analysis. H&E staining revealed gross changes in tissue morphology and loss of acinar and ductal cells in all major salivary glands of precision-irradiated mice compared to sham-treated animals [Fig. 4D (i);  Supplementary Fig. 3-(i) (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE-21-00237.1.S1]. Similarly, Sirius red [Fig. 4D (ii);  Supplementary Fig. S3 (ii) (243_rare-198-03-06_s01.pdf)] and trichrome [Fig. 4D (iii);  Supplementary Fig. S3 (iii) (243_rare-198-03-06_s01.pdf)] staining showed relative thickening of collagen fibers and fibrosis in tissue sections from precision-irradiated mice compared to sham-treated animals. Moreover, to assess the extent of radiation-induced damage to salivary epithelium, we stained whole salivary tissue sections for epithelial cell adhesion molecule (EpCAM) by immunostaining (10). We observed reduced expression of EpCAM in salivary epithelium of precision-irradiated mice compared to sham controls, which could be attributed to the loss of functional ducts and acini [Figure 4D (iv);  Supplementary Figure S3 (iv) (243_rare-198-03-06_s01.pdf)]. To validate this observation, we performed immunostaining for aquaporin-5, a membrane protein that facilitates water movement across the basal/lateral/apical membrane in acinar cells (52). Consistently, our results revealed reduction in expression of aquaporin-5 in acinar cells of precision-irradiated mice compared to sham controls [Fig. 4D (v);  Supplementary Fig. S3 (v) (243_rare-198-03-06_s01.pdf)]. Interestingly, we observed higher expression of β-catenin in ductal cells of precision-irradiated mice relative to controls [Fig. 4D (v);  Supplementary Figure s3 (v) (243_rare-198-03-06_s01.pdf)], suggesting ongoing tissue remodeling activity (10, 53, 54). In addition, we performed immunostaining of sodium-potassium-chloride cotransporter 1 (NKCC1), which is highly expressed in baso-lateral membrane of acinar cells. We observed reduced expression of NKCC1 in SMG of precision-irradiated mice when compared to controls [Fig. 4D (vi);  Supplementary Fig. s3 (vi) (243_rare-198-03-06_s01.pdf)]. However, we did not observe any changes in NKCC1 expression in PG and SLG between precision-irradiated and control mice.

FIG. 4

Necropsy reveals precision-dose-irradiation associated changes on salivary tissue, stroma, epithelial cell polarity, and integrity in preclinical model. Panel A: Baseline differences in wet weight of total salivary gland tissue (mg/g body weight) harvested at study endpoint (6 months postirradiation) from young male, female and old female mice exposed 0-7.5 Gy doses of radiation [19 male (n = 4–5 per group/radiation dose), 20 female (n = 5 per group/radiation dose) and 16 old female (n = 4 per group/radiation dose) were used]. ns = not significant; **P < 0.001. Panel B: Precision-dose-radiation-specific percent changes in the body weight adjusted wet weight of total salivary tissue in mg. Panel C: Illustration of precision-radiation-induced salivary tissue volume reduction compared to sham-treated young female mouse. Panel D: Representative photomicrograph of stained parotid, submandibular and sublingual glands of 3 young female NSG-SGM3 mice at 6 month postirradiation (0 and 7.5 Gy). Sections were stained with H&E (i), Sirius red (ii), trichrome (iii), and immunostained for EpCAM (iv), Aquaporin 5/beta-catenin (v) and NKCC1 (iv). Scale bars = 200µm.

DISCUSSION

Here, we report a novel and robust immunodeficient NSG-SGM3 mouse model of precision-radiation-induced chronic hyposalivation suitable for assessment of safety and efficacy of experimental human salivary cell therapies.

The precision dose of 5 Gy used in NSG-SGM3 mice not only selectively ablated salivary function in the long term, but also served to partially eliminate endogenous salivary cells to make room for cell engraftment. Interestingly, doses higher than 5 Gy did not guarantee a more complete salivary ablation, and only increased other complications, such as chronic alopecia or morbidity. Our kinetic analysis revealed that a 5 Gy dose was optimal for a long-term ∼50% reduction in saliva of NSG-SGM3 mice, consistent with observations reported elsewhere with NSG mice (11). Additionally, the effect of sex, age and radiation doses on reflex saliva kinetics had not been previously reported in NSG mice. Aging is a natural physiological phenomenon that has functional impact on most organs including salivary glands. The aging salivary glands of mice are characterized by acinar cell atrophy (26), apoptosis and loss of salivary gland function (55). Consistently, in our kinetic analysis we demonstrate that the reflex-saliva flowrate peaks a few minutes earlier in old females, compared to young females, but drops sharply to levels significantly lower than that of young female mice. Therefore, the older female mice may not be a preferred choice to measure human salivary regeneration and saliva production.

We also compared precision dose specific reflex saliva flowrate by gravimetric and Schrimer's strip methods over a period of 6 months. The two methods highly correlated with each other in both sexes. Each Schrimer's strip costs approximately $0.30 (U.S.). Single kinetic measurement of a flowrate over 30 min in a single animal requires 6 strips. The reading can be noted on-site in real time, or a pencil mark is sufficient to store and analyze the strips later with no concerns of evaporation of saliva.

SMG gland is the largest of three major salivary glands in rodents and is relatively more radiosensitive than PG (50, 56, 57), and therefore precision-radiation induced net loss in saliva may be due to higher loss of SMG function compared to PG or SLG. This is supported by more structural abnormalities in the SMG observed in our histological analysis. However, it is to be noted that this contrasts with human salivary glands, where the PG is the largest and most radiosensitive than the SMG and SLG (51). In fact, SMG contributes to the majority of produced saliva in rodents and irradiation of salivary glands may cause imbalance in overall salivary composition. However, these gland-specific dynamics did not affect our overall salivary readout. Sexual dimorphisms in SMG in mice are also well documented and differ with humans. There are larger GCT ducts in male SMG compared to female SMG, and such sex specific differences do not exist in human SMG (58–60). In fact, the SMG of female mouse morphologically resembles to that of human SMG due to their lack of GCT.

We also noted that after 5 and 7.5 Gy precision doses of radiation, reflex saliva levels spontaneously, but transiently, improved before it declined again. Interestingly, the onset, duration and level of this transient host response appeared to vary based on age, sex and radiation dose.

Currently, the models available for human salivary stem cell testing are limited ( Supplementary Table S1 (243_rare-198-03-06_s01.pdf);  https://doi.org/10.1667/RADE-21-00237.1.S1). The strains with incomplete immunodeficiency due to residual NK cell activity after single genetic alteration include Nude (61), SCID (62, 63), Rag1^–/– (64), and Rag2^–/– mice (65), and those with multiple genetic alterations include non-obese diabetic (NOD)-SCID mice (66). SCID mice carry a spontaneous Prkdc mutation that affects V(D)J recombination during T- and B-cell maturation and NHEJ DNA repair in all cells rendering derivative mice like NSG and NSG-SGM3 sensitive to radiation (44, 67). Whereas Rag2^–/– mice derived with a germline deletion of the recombination-activating gene 2 (Rag2), an enzyme specifically required for V(D)J recombination during T- and B-cell maturation enables the cells to tolerate “normal” murine related radiation doses compared to NSG mice (68). While highly immunodeficient mice lacking NK cells was made possible with loss-of-function genetic alteration of gamma chain/ Il2rg gene (69–71) in NSG and Rag2/Il2rg (R2G2) double knockout (68, 72), only NSG mice model, to our knowledge, has been reported for human salivary stem cell transplants (11). Future comparative investigations of the extent of host salivary dysfunction after radiation injury and efficiency of human salivary stem cell engraftment in Il2rg^–/– mice with SCID or Rag2^–/– are needed.

The NSG-SGM3 mice described here has the following features; first, it expresses physiological levels of 3 human cytokines (hIL-3, hGM-CSF and hc-KITL) and has been previously shown to promote better engraftment of hc-KIT receptor expressing hematopoietic stem cells than NSG mice (73, 74). Second, NSG-SGM3 mice has all the attributes of NSG mice including radiosensitivity and immunodeficiency intact. However, it is important to underline that the expression of hc-KIT receptor is not exclusive to a sub-population of salivary gland cells but is widely expressed by other tissue cells including hematopoietic cells, germ cells, mast cells, melanoma cells, gastrointestinal cajal cells, breast luminal cells and small neurons of brain (75, 76). Therefore, our planned future studies involve comparison of salivary transplants in NSG and NSG-SGM3 mice to delineate the functionality of hc-KIT in salivary stem cell engraftment. Indeed, hc-KIT as salivary stem cell marker remains disputed and the number of c-Kit expressing cells was extremely low and decreases with age in salisphere cultures (11). Further research into the phenotype to prospectively isolate salivary stem cells is urgently needed.

The use of isoflurane anesthesia during saliva measurement is a limitation of the current study, as it has been reported to reduce pilocarpine-induced saliva production (77). In fact, most other methods of saliva measurement use ketamine anesthesia that is known to increase salivary secretions through sympathetic stimulation (78). However, since the isoflurane is used similarly for sham-treated and radiation-treated groups of mice in this study, any influence of anesthesia on saliva secretion is normalized. Although, the method that involves the use of vacuum pump suction for saliva collection does not require anesthesia (17), but in the absence of anesthesia animals tend to struggle and this may induce varying level of anxiety that in turn may have an influence on secreted saliva. In the current study, we utilized precision stereotactic doses to radio-ablate salivary glands in NSG-SGM3 young male, female and old female mice to model hyposalivation. The high accuracy of CBCT-guided targeting and the beam arrangement enabled precision dose delivery to whole salivary gland structures in each mouse in a brain sparing manner. This method may overcome issues associated with head and neck radiotherapy which may lead to cognitive impairment (79).

In conclusion, we propose 5 Gy precision-irradiated female immunodeficient NSG-SGM3 mice as a suitable preclinical model for chronic hyposalivation studies, and the Schirmer's test as a reliable quantitative assay for salivary regenerative outcome. We expect the new preclinical model to facilitate rapid progress towards development of salivary regenerative therapies.

SUPPLEMENTARY INFORMATION

 Supplementary Fig. S1 (243_rare-198-03-06_s01.pdf): H&E staining of the submandibular gland of male and female NSG-SGM3 mice showing unique presence of GCT structures indicated by yellow arrowheads. Male GCT is markedly hypertrophic composed of number of large secretory granules compared to females. Scale bars = 200 µm. H&E staining showed absence of GCT structures in male and female human submandibular glands. Scale bars = 20 µm.

 Supplementary Fig. S2 (243_rare-198-03-06_s01.pdf): Plot showing data from young NSG-SGM3 mice (male and female combined) from FIG. 3C. Data is plotted as mean±SEM.

 Supplementary Fig. S3 (243_rare-198-03-06_s01.pdf): Necropsy reveals precision-IR-associated changes on salivary tissue, stroma, epithelial cell polarity, and integrity in old female NSG-SGM3 mice. Representative photomicrograph of stained parotid, submandibular and sublingual glands of old female (n=3) NSG-SGM3 mice at 6 month post-IR (0 and 7.5 Gy). Sections were stained with H&E (i), Sirius red (ii), trichrome (iii), and immunostained for EpCAM (iv), Aquaporin 5/beta-catenin (v) and NKCC1 (iv). Scale bars = 20µm.

 Supplementary Table S1 (243_rare-198-03-06_s01.pdf): Summary of methods reported for modelling radiation induced hyposalivation in animals.

 Supplementary Table S2 (243_rare-198-03-06_s01.pdf): Details of antibodies used for immunohistochemistry.