Comparing development and regeneration in the submandibular gland highlights distinct mechanisms

Abstract A common question in organ regeneration is the extent to which regeneration recapitulates embryonic development. To investigate this concept, we compared the expression of two highly interlinked and essential genes for salivary gland development, Sox9 and Fgf10, during submandibular gland development, homeostasis and regeneration. Salivary gland duct ligation/deligation model was used as a regenerative model. Fgf10 and Sox9 expression changed during regeneration compared to homeostasis, suggesting that these key developmental genes play important roles during regeneration, however, significantly both displayed different patterns of expression in the regenerating gland compared to the developing gland. Regenerating glands, which during homeostasis had very few weakly expressing Sox9‐positive cells in the striated/granular ducts, displayed elevated expression of Sox9 within these ducts. This pattern is in contrast to embryonic development, where Sox9 expression was absent in the proximally developing ducts. However, similar to the elevated expression at the distal tip of the epithelium in developing salivary glands, regenerating glands displayed elevated expression in a subpopulation of acinar cells, which during homeostasis expressed Sox9 at lower levels. A shift in expression of Fgf10 was observed from a widespread mesenchymal pattern during organogenesis to a more limited and predominantly epithelial pattern during homeostasis in the adult. This restricted expression in epithelial cells was maintained during regeneration, with no clear upregulation in the surrounding mesenchyme, as might be expected if regeneration recapitulated development. As both Fgf10 and Sox9 were upregulated in proximal ducts during regeneration, this suggests that the positive regulation of Sox9 by Fgf10, essential during development, is partially reawakened during regeneration using this model. Together these data suggest that developmentally important genes play a key role in salivary gland regeneration but do not precisely mimic the roles observed during development.

tion. Salivary gland duct ligation/deligation model was used as a regenerative model.
Fgf10 and Sox9 expression changed during regeneration compared to homeostasis, suggesting that these key developmental genes play important roles during regeneration, however, significantly both displayed different patterns of expression in the regenerating gland compared to the developing gland. Regenerating glands, which during homeostasis had very few weakly expressing Sox9-positive cells in the striated/granular ducts, displayed elevated expression of Sox9 within these ducts. This pattern is in contrast to embryonic development, where Sox9 expression was absent in the proximally developing ducts. However, similar to the elevated expression at the distal tip of the epithelium in developing salivary glands, regenerating glands displayed elevated expression in a subpopulation of acinar cells, which during homeostasis expressed Sox9 at lower levels. A shift in expression of Fgf10 was observed from a widespread mesenchymal pattern during organogenesis to a more limited and predominantly epithelial pattern during homeostasis in the adult. This restricted expression in epithelial cells was maintained during regeneration, with no clear upregulation in the surrounding mesenchyme, as might be expected if regeneration recapitulated development. As both Fgf10 and Sox9 were upregulated in proximal ducts during regeneration, this suggests that the positive regulation of Sox9 by Fgf10, essential during development, is partially reawakened during regeneration using this model.
Together these data suggest that developmentally important genes play a key role in salivary gland regeneration but do not precisely mimic the roles observed during development.

K E Y W O R D S
adult, duct, Fgf10, ligation, remodelling, repair, Sox9

| INTRODUC TI ON
Regeneration of organs is often proposed to mimic development with re-activation of developmental signalling pathways and master regulators during repair (Cotroneo et al., 2010;Patel & Hoffman, 2014;Villanueva et al., 2006;Xu et al., 2008). To develop therapeutic strategies for organ regeneration, studies have therefore focused on understanding how organs are formed normally during embryogenesis. However, a common question that emerges is the extent that regeneration recapitulates embryonic development. Evidence from branching organs, including the pancreas and kidneys, indicates a re-activation of some developmental pathways and transcription factors during regeneration, linking these two processes (Villanueva et al., 2006;Xu et al., 2008).
In the developing pancreas both the endocrine and exocrine cell lineages derive from a common precursor (Akiyama et al., 2005), while during homeostasis, the endocrine and exocrine cells are derived by self-duplication (Desai et al., 2007;Dor et al., 2004). Pancreatic duct ligation model of injury, although still debatable (Yu et al., 2016), activates a population of adult endocrine progenitors located in the ducts. These adult progenitors re-express the transcription factor Neurogenin 3 (Ngn3), the earliest marker of islet precursors during development (Xu et al., 2008). In the case of exocrine cell regeneration, the remaining acinar cells dedifferentiate to an embryonic-like progenitor state during acute injury and activate developmental pathways (Gu et al., 2002;Jensen et al., 2005).
The kidney is another example where multiple developmental genes have been shown to be expressed during kidney regeneration after acute kidney injury (Bonventre, 2003;Devarajan et al., 2003;Little & Kairath, 2017;Villanueva et al., 2006). Among these upregulated genes is the transcription factor Sex Determining Region Y-Box 9 (Sox9), which is involved in branching morphogenesis during kidney development (Kumar et al., 2015;Reginensi et al., 2011). In the adult kidney, Sox9 is expressed at low levels in almost all of the tubule segments and is upregulated upon acute kidney injury (Kang et al., 2016;Kumar et al., 2015). The Sox9 expressing population is the main source of proliferating cells during the regeneration process and they are required for normal repair (Kang et al., 2016).
Given these similarities and differences between development and regeneration in these different branching organs, we have turned to salivary glands to investigate the links between development and repair. Salivary glands have a slow turnover ranging from 50 to 125 days (Aure et al., 2015), however, they have a remarkable ability to regenerate after certain types of tissue injury . Three pairs of major salivary glands exist in mammals; the parotid, the submandibular and the sublingual gland. These glands vary in the type of saliva produced, their branching pattern and therefore their function. They all contain the saliva producing acinar cells surrounded by myoepithelial cells and a well-developed ductal tree formed by intercalated, granular/striated and excretory ducts that drain saliva into the oral cavity (Tucker, 2007). Understanding the mechanism of salivary gland regeneration is important due to the therapeutic potential in enhancing organ regeneration. Salivary glands are often damaged by irradiation treatment in head and neck cancer patients (10% of malignant tumours) (Vissink et al., 2003), by autoimmune disorders in patients with Sjögren's syndrome (Young et al., 2001) and by ageing, affecting around 27% of the elderly (Gupta et al., 2006), and in these cases the glands do not naturally regenerate. Salivary gland dysfunction often leads to xerostomia, a feeling of dry mouth, which impairs speech, digestion and oral health (Plemons et al., 2014). Despite its significance, the current therapeutic options, such as the use of salivary stimulants, are inefficient as they only provide temporary relief from symptoms or rely on the pre-existence of some functional tissue (Plemons et al., 2014). Thus, there is a lot of interest in understanding what happens in the cases where salivary glands are able to regenerate after injury.
Salivary gland duct ligation is one of the most commonly used methods to study regeneration in rodents (Bozorgi et al., 2014;Carpenter et al., 2007). This method involves the obstruction of the main excretory duct which leads to saliva retention, increase tissue pressure and cell lysis (Cotroneo et al., 2010). As a result, the injured tissue undergoes extensive acinar cell loss, duct dilation and extensive fibrosis and inflammation (Cotroneo et al., 2010). Once the duct obstruction is released, acinar cells are produced by a process that involves self-duplication of the remaining acinar cells (Aure et al., 2015). Potential activation of an adult stem/progenitor cells found in the ducts has been suggested (Pringle et al., 2013), but this only appears to occur in extreme cases of damage (Weng et al., 2018).
Based on anatomical observations and activation of certain developmental pathways, it has been suggested that the process of salivary gland regeneration mimics embryonic development (Cotroneo et al., 2010;Patel & Hoffman, 2014). Anatomical observations include the appearance of embryonic-like branched structures ending in acinar cells during regeneration, suggesting a recapitulation of the branching process during repair (Cotroneo et al., 2010). Signalling pathways upregulated include Wnt signalling that is reactivated in the ductal cells of regenerating glands (Hai et al., 2010), similar to embryonic glands during differentiation (Patel et al., 2011). Signals provided by the parasympathetic nervous system that innervates the glands have also been shown to be necessary during embryonic salivary gland branching morphogenesis and regeneration (Knox et al., 2013;. In addition, pharmacological activation of Fibroblast Growth Factor (Fgf) and Ectodysplasin A (Eda) signalling, two pathways required for epithelial development of salivary glands (Entesarian et al., 2005;Jaskoll et al., 2005;May et al., 2015;Teshima et al., 2016;Wells et al., 2010), partially restore the function and the anatomy of irradiated salivary glands (Hill et al., 2014;Lombaert et al., 2008).
Salivary glands, like kidneys, pancreas and lung, are formed through branching morphogenesis, a process which is driven by the distal part of the epithelium (tips of the gland) (Chatzeli et al., 2017). Sox9 is a transcription factor expressed in the initial epithelial progenitors of the gland placode and later, during branching morphogenesis, in the distal epithelium (Chatzeli et al., 2017).
In the absence of Sox9, branching initiation is inhibited due to the depletion of the distal epithelial progenitors (Chatzeli et al., 2017). Sox9 has also been shown to be one of two transcription factors that can reprogramme mouse embryonic stem cells to a salivary gland fate (Tanaka et al., 2018). Fgf10 is expressed in the neural crest-derived mesenchyme at the initiation stage and throughout branching morphogenesis and is required both for the initial stages of bud formation and for branching morphogenesis (Jaskoll et al., 2005;Wells et al., 2013). Fgf10 signals to the epithelium through the receptor FgfR2 (Jaskoll et al., 2002;Ohuchi et al., 2000;Steinberg et al., 2005). Fgf10-null or FgfR2-null salivary glands develop a rudimentary prebud with no further growth while pharmacological inhibition of Fgf signalling inhibits branching morphogenesis (Jaskoll et al., 2005;Ohuchi et al., 2000). A similar phenotype is observed after conditional knockout of Fgf10 in the neural crest-derived mesenchyme, confirming the importance of Fgf10 produced from this tissue during development (Teshima et al., 2016). One of the roles of Fgf10 during development is to maintain Sox9 expression in epithelial progenitors (Chatzeli et al., 2017), thus these pathways are closely interlinked during early development.
Given the importance of Fgf10 and Sox9 in salivary gland development, we have examined the pattern of expression of Fgf10 and Sox9 during salivary gland regeneration and homeostasis in the adult and compared this to the pattern during development.
All procedures and culling methods were compliant with UK Home Office regulations and with the approval of the King's College London Biological Safety committee. Mice were aged between 6 and 10 weeks for all analysis of adult expression.

| Salivary gland duct ligation and deligation
Salivary gland ligation and deligation was performed in mice as previously described (Bozorgi et al., 2014). Female mice aged between 6 and 10 weeks were used for the surgery. Briefly, mice were anaesthetised with an intraperitoneal injection of Ketamine 75 mg/kg and Xylazine 15 mg/kg dissolved in PBS. A small incision was made at the neck region to expose the main excretory duct of the sublingual and the submandibular gland. Once the excretory duct was exposed, a titanium haemostatic micro clip (Mediplus) was applied using a clip applier (Mediplus). Ligation was performed only on the right side.
The incision was then sutured with absorbable Vicryl-coated sutures (Aston Pharma) and mice were left with the clip for 8 days. Mice were then either euthanised and their submandibular glands were removed or undergone a deligation procedure. For deligation, mice were re-anaesthetised, an incision made and the clip removed. To evaluate the ligation efficiency, a bigger incision was made in some mice in order to expose the operated and contralateral unoperated salivary glands. Efficiency was based on the size of the gland and the formation of fibrotic tissue. Mice were then sutured and left to recover for 4 days. After that time they were euthanised and submandibular glands were removed for further analysis. Unoperated mice were used as a control.

| X-gal staining
Salivary glands from Fgf10 nlacZ/+ mice where dissected and fixed in 4% PFA for 20 min (Kelly et al., 2001). Fgf10 +/+ littermates were used as controls. Salivary glands were then washed for 5 min in PBS with 2 mM MgCl 2 followed by another wash for 15 min. Glands were then incubated for 5 min in a solution containing 1 mM MgCl 2 , 0.2% NP-40 and 0.02% deoxycholic acid diluted in PBS (Solution B).
Glands were stained with Solution C made with 5 mM K 3 Fe(CN) 6 , 5 mM K 4 Fe(CN) 6 and 1 mg/ml x-gal diluted in Solution B. Staining was performed at 37°C for 4 h. Salivary glands were then washed in PBS 3 times for 5 min and processed for cryosectioning. Using the same protocol, cryosections on slides were additionally stained at 37°C for 4 days to increase the intensity of the signal, washed in PBS and re-fixed. Sections were then dehydrated in EtOH series, counterstained with alcoholic Eosin and mounted with Neo-Mount via Neo-clear. Slides were photographed on a Nikon microscope.

| Trichrome staining
Slides were dewaxed by incubating twice in Histoclear II (National Diagnostics) for 10 min, rehydrated in decreasing concentrations of IMS for 2 min and in dH 2 O for 2 min. For Alcian Blue staining slides were incubated in 1% Alcian Blue (Fluka) dissolved in 3% acetic acid (Analytical Reagents) pH2.5 for 10 min, rinsed briefly in dH 2 O and washed under running tap water for 10 min. For haematoxylin staining, slides were incubated in Ehrlich's haematoxylin (Solmedia) for 2 min, washed in running tap water for 10 min and rinsed in dH 2 O.
For staining differentiation, slides were incubated in 2.5% phosphomolybdic acid (Fisher) for 10 min and rinsed in dH2O. Sirius Red staining was performed with 0.5% Sirius Red (BDH) in saturated picric acid (Fluka Biochemica) for 20 min and slides were rinsed twice in 0.5% acetic acid. After blotting dry, slides were washed 3 times in 100% IMS for 2 min each and for 5 min twice in histoclear. Slides were mounted with Neomount (Merck), coverslipped and left to dry at 40°C overnight. Alcian blue stains muccopolysaccharides blue while Sirius red in pircric acide stains collagens for visualisation of the connective tissue. Haematoxylin stains the nuclei.

| Cell quantification and statistical analysis
Cells were quantified manually using the cell counter plug in of Fiji/ ImageJ (Schindelin et al., 2012). The mean of three different sections was calculated. Results were plotted and statistically analysed using Graph Pad Prism software. Statistical significance was calculated using unpaired t-tests. Significance was taken as p < 0.05 (*), p < 0.01 (**) or p < 0.001 (***).

| RE SULTS
3.1 | The expression of Sox9 in distal epithelium is maintained in adult submandibular glands but with altered levels of expression As a first approach to assess the role of Sox9 in the adult and embryonic salivary glands, the pattern of Sox9 expression in embryonic E15.5 and adult salivary glands was compared. At E15.5, a massive expansion of the epithelium through branching morphogenesis occurs along with the initiation of differentiation (Nelson et al., 2013). At E15.5, Sox9 was highly expressed at the tips of the epithelium where branching morphogenesis occurs and proacinar differentiation is initiated (Figure 1a), similar to that described previously (Chatzeli et al., 2017). In addition to expression in forming acini, Sox9 was also expressed in the intercalated ducts, adjacent to the acini, but was absent from the more proximally developing striated/granular and excretory ducts (Figure 1a′, a″) (Chatzeli et al., 2017). In the adult submandibular gland, three distinct populations of Sox9-positive cells were observed (Chatzeli et al., 2017).

| Fgf10 expression shifts from mesenchymal in embryonic glands to mainly epithelial in adult salivary glands
Having compared the pattern of Sox9 expression in embryonic and adult salivary glands we turned to investigate Fgf10, since Fgf10 and Sox9 are closely interlinked during development (Chatzeli et al., 2017). The mRNA of Fgf10 in the embryonic glands was detected  showed higher numbers of Fgf10-positive duct cells compared to female glands (Figure 6b,c), indicating that expression was mainly in the granular ducts, a specialised striated duct, which, due to sexual dimorphism, are more prominent in adult males (Gresik, 1975).
To investigate the origin of the cells expressed within the epithelium, we used the Wnt1 cre ;R26 Tom mouse to label the progeny of neural crest cells, which in submandibular glands include the mesenchyme and the parasympathetic nerves ( Figure 2i) (Ferreira & Hoffman, 2013;Jaskoll et al., 2002). No co-localisation was observed between red fluorescence protein (RFP) and Fgf10, highlighting that adult Fgf10-positive cells are not neural crest derived, and therefore a distinct lineage from the Fgf10-expressing embryonic population (Figure 2j,k).
In the embryo, Fgf10-expressing cells in the mesenchyme are located close to Sox9-expressing cells in the distal gland epithelium (Chatzeli et al., 2017). In the adult gland, a few epithelial cells showing expression of both genes but the majority of Sox9 cells were not associated with Fgf10 cells, and visa versa ( Figure S2). The relationship between these genes in the adult is therefore very different from the embryo.

| The percentage of Sox9-positive proliferating cells increases during regeneration
Having established the expression of Sox9 and Fgf10 in the adult, we then turned to investigate how their expression changed during regeneration. For this, we used the salivary gland duct ligation model (Bozorgi et al., 2014;Cotroneo et al., 2008;Takahashi et al., 2004). Salivary gland ligation was performed using female

| Ectopic expression of Sox9 in striated/ granular ducts during regeneration
Salivary gland development is characterised by elevated Sox9 expression in the highly proliferative distal epithelium (Chatzeli et al., 2017). To investigate whether regeneration was associated with changes to Sox9 expression, the number of Sox9-positive cells was Upregulation of Sox9 in these larger ducts therefore appears to be an important feature in injury and regeneration.
In addition to the change in expression in the large ducts, a change in the regenerating acini was also noted (Figure 5e-g).
In control glands, Sox9 was expressed at lower levels in the acini when compared to the neighbouring intercalated ducts (Figure 1d).
In the regenerating glands after deligation, two patterns of expression were observed in the acini. In some cases, the Mist1-positive

| Fgf10-positive cells increase during regeneration but maintain their expression in the gland epithelium
Having analysed the pattern of Sox9 expression in the homeostatic and regenerating salivary gland, we then moved on to investigate Fgf10 expression. Fgf10 is an important mitogen of epithelial progenitors during salivary gland embryonic F I G U R E 3 Degeneration of acinar cells after ligation and regeneration after deligation. (a) Experimental strategy to study submandibular gland regeneration after duct ligation. (b-g) Trichrome staining of an unoperated (b, e), ligated (c, f) and regenerating (d, g) submandibular gland. All mice were female 6-10 weeks old. (h-j) Immunofluorescence for the acinar cell marker Mist1 (green: nuclear stain) in an unoperated (h), ligated (i) and regenerating (j) submandibular gland. DNA is shown in blue (DAPI). Arrowheads in (

| DISCUSS ION
Salivary glands are important organs for dental health and wellbeing. Currently, around 27% of the elderly, and head and neck cancer patients (10% of malignant tumours) undergoing irradiation treatment suffer from salivary gland dysfunction, which could be treated by organ regeneration (Gupta et al., 2006;Vissink et al., 2003). Although salivary gland regeneration has been proposed to mimic embryonic development (Cotroneo et al., 2010;Patel & Hoffman, 2014), few studies have investigated the extent in which the developmental programme correlates with regeneration. Here, we compared the expression of two highly interlinked and essential genes for salivary gland development, Sox9 and Fgf10 during salivary gland development, homeostasis and regeneration. We found that although there is an elevated appearance of Fgf10 and Sox9-positive cells during regeneration, their function during regeneration may be different from that during organogenesis.

| During regeneration Sox9 is turned on in striated/granular ducts, a location with low Sox9 levels during development and homeostasis
If regeneration recapitulates embryonic development, it could be speculated that genes expressed during development would be expressed in similar locations during regeneration. During branching morphogenesis in the embryo, Sox9 was highly expressed at the tips of the developing epithelium and absent from the proximal structure that will form the main duct network (Chatzeli et al., 2017). Similar to the developing gland, Sox9 was expressed in the most distal epithelium of adult submandibular glands, the acini and intercalated ducts, which are the progeny of the embryonic distal epithelium (Matsumoto et al., 2016) and was mainly absent from the striated/granular and excretory ducts. Sox9 has recently been suggested to regulate a putative stem cell population of CD133 expressing cells in the adult gland (Tanaka et al., 2019). Contrary to homeostasis, in regenerating salivary gland, Sox9 was upreg- ductal cells during regeneration that is distinct from its role during development and homeostasis. This upregulation might be linked to the activation of a suggested population of adult progenitors within the ducts that contributes to the production of distally located ductal cells (Pringle et al., 2013).
In contrast to the novel expression in the ducts, high levels of expression of Sox9 were also observed in a few acinar cells in the regenerating glands, suggesting that Sox9 upregulation is involved in formation of new acini during repair, potentially recapitulating its role in acini development during embryogenesis. High and low levels of Sox9 expression have previously been described in the developing pancreas, with the low state correlating with lower levels of Fgfr2 expression (Seymour et al., 2012).

| Mesenchymal Fgf10 expression does not recur in the regenerating salivary gland
Salivary gland embryonic development is characterised by high levels of Fgf10 expression in the neural crest-derived mesenchyme of the gland during initiation and branching morphogenesis (Jaskoll et al., 2005;Teshima et al., 2016;Wells et al., 2013).
This reliance on mesenchymal expression of Fgf10 is highlighted by the fact that the conditional Fgf10 knockout in the neural crest (Wnt1-cre driven) has the same early arrest of the salivary gland placode as the full Fgf10 knockout (Teshima et al., 2016). However, in the lobes of the adult submandibular gland, Fgf10 expression was limited and confined to the epithelium. Some Fgf10-positive cells were located in the interlobular connective tissue and capsule, which has a mesenchymal origin, however, these cells were very rare. During regeneration, the number of Fgf10 expressing cells was upregulated from 1% to 4% of the total gland, but still represented a small number of cells overall. The upregulation was most obvious in the striated/granular ducts but no new expression was observed in the mesenchyme around the regenerating acini, as might be predicted if regeneration directly recapitulated development. The role of Fgf10 in regeneration therefore appears distinct from development.

| Fgf10 and Sox9 are both upregulated in striated ducts during regeneration
During salivary gland development, Sox9 expression is regulated by Fgf10 (Chatzeli et al., 2017), this relationship being conserved during the development of a number of other branching organs (Abler et al., 2009;Chang et al., 2013;Chen et al., 2014;Seymour et al., 2012). In the developing gland, their spatial and temporal expression F I G U R E 7 Schematic of the expression pattern of Fgf10 and Sox9 during salivary gland organogenesis, homeostasis and repair. Fgf10 is shown in green while Sox9 is shown in pink. During development, mesenchymal Fgf10 signals to epithelial Sox9 and directs expression in the distal region of the gland. During homeostasis, the expression of Fgf10 shifts to the ductal epithelium, while Sox9 remains expressed in the distal part of the glands (acini and intercalated ducts). During regeneration Sox9 and Fgf10 are upregulated in the ducts and may interact.
Black arrows indicate a regulatory mechanism between Fgf10 and Sox9. This has been confirmed during development and needs to be tested during regeneration [Colour figure can be viewed at wileyonlinelibrary.com] is tightly linked from the initiation stage of salivary gland development, with Fgf10 expressed in the mesenchyme and Sox9 in the epithelium (Chatzeli et al., 2017;Jaskoll et al., 2005;Wells et al., 2013).
As branching morphogenesis proceeds, Sox9 is highly expressed at the tips of the epithelium that will generate the acinar cells and smaller ducts while it is reduced at cells that form the excretory duct (Chatzeli et al., 2017;Matsumoto et al., 2016), with Fgf10 expressed in the mesenchyme that surrounds these distal structures ( Figure 2).
Interestingly, given their close association in the developing gland, the expression pattern of Fgf10 and Sox9 in the adult submandibular gland was largely distinct, with Fgf10 concentrated in more proximal epithelial, while Sox9 was largely excluded from this region ( Figure   S2). This suggests that the regulation of Sox9 by Fgf10 reported in the embryonic salivary gland (Chatzeli et al., 2017) is not maintained in the adult gland.
During regeneration, however, both Fgf10 and Sox9 were upregulated in the granular/striated ducts, suggesting a re-awakening of this regulatory relationship during regeneration. The changing pattern of expression and potential regulatory relationships of Fgf10 and Sox9 during regeneration is illustrated in Figure 7. A number of duct progenitor cells markers have previously been identified (Keratin 5, Keratin 14, Axin 2, c-Kit, Ascl 3), which appear to label distinct populations within different ducts (Bullard et al., 2008;Kwak et al., 2016;May et al., 2018;Rugel-Stahl et al., 2012;Weng et al., 2018). Using the duct ligation model, Keratin 5 and Axin 2 duct progenitor cells have been shown to be able to regenerate duct cells, but not cells of the acinar lineage (Weng et al., 2018). It would therefore be interesting to investigate the relationship of Sox9 and Fgf10 expressing cells to these other markers of duct stem/progenitors cells, and to test their potential using lineage labelling studies.
The experiments performed here rely on the duct ligation model to trigger gland atrophy and regeneration. An alternative model used to study regeneration is the irradiation model, which more closely recapitulates the injury observed after radiation therapy in head and neck cancer patients. The two models have different degrees of regeneration in mouse models, with irradiation leading to a more significant, long-term, damage to the gland and less robust regenerative capacity (Weng et al., 2018). The irradiated gland, however, does undergo some regeneration, with Sox2 cells replenishing the acinar compartment of the sublingual gland (Emmersen et al., 2018). The two methods share some similarities, for example, p63 is upregulated after both techniques (Ikai et al., 2020). It would therefore be interesting to study whether the changes in Fgf10 and Sox9 expression observed here after gland ligation also occur after irradiation.
Overall our data indicate that Sox9 and Fgf10 regulation during regeneration in adult glands does not directly mimic development.
We see no widespread upregulation of Fgf10 in regenerating gland mesenchyme, while we observe the induction of Sox9 and Fgf10 in tissues that do not express these genes in the embryo. The regulatory relationship between these two factors established in the embryo also does not appear to be conserved during homeostasis but may be re-established during regeneration (Figure 7). It seems, therefore, that regeneration uses similar genes to development but does not directly recapitulate the embryonic programme, perhaps due to differences in the cell types generated and the microenvironment of the tissue. Learning about development can therefore shed light on the genes that are likely to be involved during regeneration but regeneration should not be presumed to always follow similar rules.

CO N FLI C T O F I NTE R E S T
Abigail Tucker is a council member of the Anatomical Society.

FU N D I N G
LC was funded by the Anatomical Society from a PhD studentship awarded to AT and GP.

DATA AVA I L A B I L I T Y S TAT E M E N T
All mouse lines used in this research are available for sharing. All reagents are commercially available.