Serotonin and MucXS release by small secretory cells depend on Xpod, a SSC specific marker gene

Mucus secretion and ciliary motility are hallmarks for muco‐ciliary epithelia (MCE). Both, mammalian airways as well as the less complex epidermis of Xenopus embryos show cilia‐driven mucus flow to protect the organism against harmful effects by exogenous pathogens or pollutants. Four cell types set up the epidermal MCE in Xenopus. Multi‐ciliated cells (MCCs) generate an anterior to posterior flow of mucus. Ion secreting cells (ISCs) are characterized by the expression of ion transporters, presumably to maintain a favorable homeostasis. The largest cell type is represented by goblet cells, which cover most of the epidermis and exhibit secretory properties. Additionally, small secretory cells (SSCs) release mucus, antibiotic compounds, and the monoamine serotonin (5‐hydroxytryptamine; 5‐HT). We have recently shown that serotonin regulates flow velocity by acting on ciliary beat frequency. Here, we describe the identification and functional characterization of Xenopus polka‐dots (Xpod). No homologous genes or proteins were found in other vertebrates, including Xenopus tropicalis. We demonstrate that Xpod serves as an SSC‐specific marker, starting to be expressed shortly after SSC specification at neurula stages. Overexpression of a tagged Xpod protein resulted in the localization of secretory granules. Notch signaling induced SSC cell fate, in contrast to its repressing effect on MCC and ISC specification. Xpod loss‐of‐function revealed that mucus and 5‐HT release by SSCs was severely diminished, which impaired the ciliary beating of MCCs. In summary, Xpod specifically marked SSCs and was required for muco‐ciliary secretion in Xenopus laevis.

goblet cells. Upon secretion, changes in pH and ionic composition in the extracellular space unfolds and hydrates the secreted mucins to establish the viscous, gel-like mucus . A second highly specialized MCE cell type is multi-ciliated cells (MCCs), each projecting up to 300 motile cilia from the apical surface. Ciliary beating direction of MCCs aligns, resulting in a constant and directional mucus movement (Meunier & Azimzadeh, 2016).
The lining of the airways represents the most relevant MCE for human health. Under normal conditions, the mucus-cilia interplay removes inhaled pathogens and hazardous substances (i.e., tobacco smoke) from the airways. Particles get trapped by mucus, which is then cleared from the airways through cilia motion. If this process is impaired, microbes are able to accumulate and cause severe airway infections. In human primary ciliary dyskinesia (PCD), mutations specifically affect motile cilia function (Afzelius, 2004;Bustamante-Marin & Ostrowski, 2017). The clinical presentation of PCD comprises male infertility, alteration in left-right body axis formation, and chronic airway infections, underscoring the importance of cilia-driven mucus movement (Tilley et al., 2014). Mutations impairing mucus functions are known as well. The case best studied is cystic fibrosis (CF), where luminal mucus becomes hyperviscous and thus unable to be moved by cilia. CF patients suffer from clogged airways and chronic infections. Mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, encoding a chloride-channel, underlie CF (Livraghi & Randell, 2007). CFTR and other ion-channels and -pumps are required to maintain and regulate ion homeostasis, which is required for maintenance of a normal airway surface liquid layer. If this system is perturbed, luminal water concentrations drop and subsequently affect the dissolution of mucus (Boucher, 2007;Rubin, 2007). Airway clearance by cilia-driven mucus flow thus depicts a rather simple first line of defense mechanism for the organism, before major immune responses are active. However, this simplicity is embedded into a highly complex regulatory and cellular network of interactions to ensure airway functionality, which has to integrate environmental parameters such as humidity as well. Therefore, it is rather difficult to unravel basic mechanisms, which is apparently true for mammalian airways with 50 or even more cell types (BéruBé, Prytherch, Job, & Hughes, 2010).
The epidermis of Xenopus embryos represents a bona fide MCE of low complexity, which protects against bacterial infections in much the same way as the mammalian airways. The embryonic Xenopus skin is composed of only four cell types. Ion secreting cells (ISC) are characterized by the expression of ion channels, which are thought to ensure homeostasis and probably mucus deflation. Both, goblet cells and small secretory cells (SSCs) produce and release mucus. The main cell type of mucus secretion, however, is the SSCs. A Xenopus specific mucin, called MucXS, constitutes the major mucus component. In addition, antibacterial factors and the monoamine 5-hydroxytryptamine/serotonin (5-HT) have been identified to be secreted by SSCs as well (Dubaissi et al., 2018;Dubaissi et al., 2014;Walentek et al., 2014). We have previously demonstrated that the forkhead transcription factor foxA1 is required and sufficient for SSC formation. Loss of foxA1 function impaired SSC specification and thus resulted in the absence of SSCs. Embryos lacking SSCs and consequently major secretion properties disintegrated due to bacterial infection. The protective nature of SSCs was further demonstrated by MucXS gene knockdown through antisense morpholino oligomers (MO), where loss of MucXS facilitated bacterial infections (Dubaissi et al., , 2018. Besides their protective capacity, SSCs are able to regulate ciliary beat frequency of neighboring MCCs via serotonin release. MCCs express the 5-HT receptor Type 3, a ligand-gated ion channel, which impacts on ciliary beating. Therefore, 5-HT adjusts liquid flow velocity along the anterior to posterior axis of the embryo (Walentek et al., 2014). Epidermal MCCs have been successfully used to investigate the multiple stages of ciliogenesis. This is due to the accessibility and paramount embryological manipulation techniques of the Xenopus system (Schweickert & Feistel, 2015).
So far, a comparable in-depth analysis of SSC development and function has not been conducted. Yet, the question of transferability to human MCEs has not been solved either. Although SSC secretory products can be visualized by IF (serotonin and MucXS; Dubaissi et al., 2014;Walentek et al., 2014), an SSC specific marker gene with strong mRNA expression is still missing. Genes found to be expressed in SSCs during embryonic development like foxA1, MucXS, enzymes of 5-HT synthesis (tryptophan hydroxylase/tph1; dopamine decarboxylase/ddc) were not SSC-specific and are transcribed in other cell types as well. In addition, most showed rather low expression levels in SSCs when used in in situ hybridization (ISH) experiments. Identification of a bona fide marker gene, therefore, would allow us to follow SSC development and function more precisely.
Here, we show that Xenopus polka-dots (Xpod) serves as a reliable, strong and specific SSC marker. Xpod represents a secreted protein, which so far has no homologous counterparts in other vertebrates.
Xpod mRNA was used to visualize SSC development starting at neurula stages and to demonstrate SSC fate induction by Notch signaling.
Functionally, Xpod knockdown inhibited MucXS and 5-HT release, suggesting a role in the secretory pathway of SSCs.

| RESULTS
To identify robust and-importantly-SSC-specific marker genes, a database and literature search was performed, focusing on a regular punctate expression pattern in the larval epidermis. A publication by Yoshii et al. (2011) about an epidermally expressed gene termed Xenopus polka-dots (Xpod) drew our attention, because subcellular Xpod localization showed a high degree of similarity to SSC secretory granules (cf. Figure 3h and i; Yoshii et al., 2011;Dubaissi et al., 2014;Walentek et al., 2014). Unexpectedly, Xpod was not annotated in the Xenopus database Xenbase (Karimi et al., 2018). By homology search we obtained multiple expressed sequence tags (ESTs) but no relevant genomic hits (data not shown). Although EST sequences corresponded to the published Xpod cDNA, we detected a 5 0 extended reading frame, which added 32 amino-acids to the N-terminus of the published protein, giving rise to a final size of 107 amino-acids ( Figure 1a).
In silico domain searches revealed the presence of a signal peptide, suggesting that Xpod was a secreted protein. No additional functional domains could be identified. Surprisingly, homologous gene products were not present in genomes of other vertebrates, including its sister species Xenopus tropicalis (data not shown). To identify the nature of Xpod expressing epidermal cells, ISH followed by peanut agglutinin (PNA) staining was performed. The lectin PNA detects sugar residues on MucXS, the mucin secreted by SSCs (Dubaissi et al., , 2018. To investigate whether related sequences were present in Xenopus tropicalis, an ISH using the X. laevis Xpod probe was performed on Xenopus tropicalis embryos. Based on the close evolutionary distance of both species, many probes can be used interchangeably. However, no Xpod positive staining was found in Stage 32 Xenopus tropicalis embryos (data not shown). To verify that Xpod was a secreted protein, a myc-tag was inserted C-terminally of the predicted signal peptide (mycXpod, Figure 1a). mycXpod mRNA was injected into the epidermal lineage, that is, in the animal-ventral region of 4-8 cell embryos, which were cultured until Stage 33.
Detection of mycXpod and SSC granules was performed by immunofluorescence (IF) using an anti-myc antibody and PNA. As expected, mycXpod was found in PNA positive granules (Figure 1c,c 0 ). Next, we analyzed, if mycXpod co-localized with 5-HT as well by using an antiserotonin antibody. In non-SSC cells, mycXpod positive signals showed a vesicle-like appearance (Figure 1d), while mycXpod and 5-HT co-localized in secretory SSC granules ( Figure 1d,d 0 ,d 00 ). We rarely found co-localization in more than one or two granules, which might indicate qualitative differences between granules. Taken together, the IF analysis suggests that Xpod was secreted via SSC granules.
Next, we followed Xpod mRNA expression during SSC development using histological sections. As published, Xpod transcription started at mid-neurula stages in the deep layer of the epidermis and thus marked SSC precursor cells ( Figure S1A; Yoshii et al., 2011). During later neurula stages, Xpod signals and the number of SSC precursor cells increased. By early tadpole (Stage 22), SSC intercalation into the superficial cell layer was first observed. At Stage 24, most SSCs had entered the superficial layer, as reported recently (Figure S1B-D; Walentek et al., 2014). Up to Stage 38, Xpod expression in nonswimming tadpoles was restricted to SCCs (data not shown). From these data, we conclude that-beginning with SSC specification-Xpod mRNA expression reliably follows all steps of SSC differentiation.
SSC cell fate specification has been shown to be executed by the FoxA1 transcription factor . Upstream signaling pathways, however, have not been investigated. Inhibition of Notch signaling has been described as key to MCC and ISC fate determination (Deblandre, Wettstein, Koyano-Nakagawa, & Kintner, 1999;Quigley, Stubbs, & Kintner, 2011). We, therefore, wondered whether  Table S1) and because co-injection of rescXpod restored the frequency of loaded granules per SSC to almost wild type levels (10 MucXS and 8 5-HT; Figure 3c,d,e; Table S1). In agreement with our marker gene analysis, we did not detect a significant change in SSC numbers in  (atp6ve1) were analyzed. In addition, MCCs and SSCs were identified by IF using antiacetylated tubulin (blue) and anti-serotonin (red) antibodies, respectively. Actin staining by phalloidin (green) marks cell border Xpod morphants (Figure 3f). Based on these experiments, we concluded that Xpod played a role during the secretion of MucXS and 5-HT by SSCs.
We, therefore, wondered what consequences Xpod depletion had for MCE function. Loss of SSCs by foxA1 knockdown resulted in the disintegration of tadpoles due to bacterial infections. This lethal effect was potentially due to impaired muco-ciliary clearance, as mucus and antibacterial active substances, which are secreted by SSCs, were missing . In Xpod morphants, however, we did not observe any impact on embryonic survival (data not shown).
Because 5-HT levels were substantially reduced upon Xpod loss, the ciliary beat frequency of MCCs should be reduced, in accordance with our recent work (Walentek et al., 2014). We, therefore, analyzed cili-  Table S1). This effect was specific, because of (a) a dose-dependent MO effect (Figure 4d

| DISCUSSION
The work presented here establishes Xpod as the sole SSC-specific marker gene to date, which should be a valuable tool to unequivocally identify SSCs in wildtype and manipulated Xenopus laevis embryos.  Table S1 for p values Besides this more descriptive aspect, additional questions on Xpod function and structure arise.
The apparent reduction in MucXS and serotonin release in Xpod morphants suggests that Xpod either plays a role in their synthesis processes or during secretion. Although our data are insufficient to definitely discern between these two possibilities, we favor a secretory function of Xpod because of two arguments: (a) We show that mycXpod was incorporated into secretory granules, which demonstrates that the sole identified structure, the signal peptide, is functional. In addition, Yoshii et al. (2011) using IF detected Xpod in apical blebs, strongly resembling SSC granules. Unfortunately, this antibody was no longer available (Dr. Kinoshita-personnel communication) to conduct a high-resolution subcellular localization study; (b) MucXS and serotonin synthesis differ substantially, which renders an Xpod activity on both processes highly unlikely. Like any other mucin, MucXS mRNA gets translated at the endoplasmic reticulum (ER), the protein modified at the Golgi and subsequently loaded into secretory vesicles. Serotonin, however, is synthesized in the cytoplasm by enzymatic conversion (ddc and tph) of tryptophan to 5-HT, which is shuttled into secretory vesicles via vesicular monoamine transporters (vmat). We have recently shown that vmat1 is uniformly expressed in the epidermis, starting at early neurula stages (cf. Figure S2A,B in Walentek et al., 2014). Vmat1 serves as a monoamine (i.e., 5-HT) hydrogen-ion exchanger, which actively fills the exocytotic vesicle.
Therefore, vesicle loading with 5-HT by this canonical pathway should be active in SSCs (Lawal & Krantz, 2013;Yaffe, Forrest, & Schuldiner, 2018). The only compartment that all three components (MucXS, 5-HT, and Xpod) have in common is the secretory vesicles. The most parsimonious explanation, therefore, is that Xpod acts at the level of secretion.
At first glance, it is surprising that Xpod has no vertebrate homolog, not even in its sister species Xenopus tropicalis. The basic process of mucus release via exocytic vesicles, however, is conserved throughout the vertebrates and beyond (Knoop & Newberry, 2018;Perez-Vilar, 2007;. This raises the question on the evolutionary origin of Xpod. All amphibian genomes, including both Xenopus species, are far from being completely assembled, suggesting that Xpod-like sequences might still be revealed. The mammalian databases have been completed at high resolution, which might indicate that Xpod could be restricted to lower vertebrates. In evolutionary terms, this scenario has the highest probability. On the other hand, our failure to identify homologs could also indicate that Xpod might be a gene specific to Xenopus laevis. Surely, it is hard to imagine how such a species-specific protein could evolve and obtain a crucial function in secretion just once. Although the precise Xpod genomic loci are currently not available, Xpod cDNA fragments match to Xenopus laevis chromosomes 9_10L and S. In Xenopus tropicalis, two homologous but distinct chromosomes 9 and 10 exist, thus chromosome 9_10 reflects an initial fusion result (Matsuda et al., 2015;Session et al., 2016), which potentially could create a species-specific gene product. However, in most cases genomic shuffling results in loss of gene function or in fusion of previously separated gene products.
Even if genomic sequences would accidentally become under the control of an SSC specific promotor, a functional impact on a specific cellular process is still highly unlikely. To our knowledge, de novo gene emergence in a single species has not been reported before.
Interestingly, MucXS is specific to both Xenopus species, although the commonly used mammalian mucins (i.e., mucin5A) are present in  In addition, we demonstrated that activation of Notch signaling induces SSC formation in the epidermis of frog embryos, confirming a recent report (Walentek, 2018). Balanced (repressed or activated) Notch signaling, therefore, accounts for stem cells positioning in the superficial layer, which is differentiating toward MCC, ISC, or SSC fate. Interestingly the different states of Notch signaling induce cell type-specific transcription factors of the forkhead family in each case, that is, foxj1 (MCCs), foxi1e (ISCs), and foxa1 (SSCs). This regulatory code for cell identities is conserved in mammalian tissues (Besnard, Wert, Kaestner, & Whitsett, 2005;Dubaissi et al., 2014;Esaki et al., 2009;Montoro et al., 2018;Thomas et al., 2010;Wan et al., 2004;Yu, Ng, Habacher, & Roy, 2008), indicating that an SSCs counterpart must exist. Foxa1 and Foxa2 are known inducers of goblet cell identity in mammalian mucus-producing tissues, that is, the gastrointestinal tract (van der Sluis et al., 2008;Ye & Kaestner, 2009

| RNA in situ hybridization
Embryos were fixed in MEMFA for 2 hr and processed following standard protocols (Sive et al., 2000). Digoxigenin-labeled (Roche) RNA probes were prepared from linearized plasmids using SP6 or T7 RNA polymerase (Promega). ISH was according to Belo et al. (1997).

| mRNA synthesis and microinjections
For in vitro synthesis of mRNA using the Ambion sp6 message machine kit, the plasmid was linearized with NotI. Embryos were injected at the 4-to 8-cell stage, using a Harvard Apparatus.

ACKNOWLEDGMENTS
We are grateful to Tsutomu Kinoshita for sharing Xpod probe, to Peter Walentek and Martin Blum for critical reading of the manuscript, Tim Ott for technical help, and all members of the Blum lab for continuous support and discussions.