Synaptotagmin 13 orchestrates pancreatic endocrine cell egression and islet morphogenesis

Epithelial cell egression is important for organ development, but also drives cancer metastasis. Better understandings of pancreatic epithelial morphogenetic programs generating islets of Langerhans aid to diabetes therapy. Here we identify the Ca2+-independent atypical Synaptotagmin 13 (Syt13) as a key driver of endocrine cell egression and islet formation. We detected upregulation of Syt13 in endocrine precursors that correlates with increased expression of unique cytoskeletal components. High-resolution imaging reveals a previously unidentified apical-basal to front-rear repolarization during endocrine cell egression. Strikingly, Syt13 interacts with acetylated tubulin and phosphatidylinositol phospholipids and localizes to the leading-edge of egressing cells. Knockout of Syt13 impairs endocrine cell egression and skews the α- to-β-cell ratio. Mechanistically, Syt13 regulates endocytosis to remodel the basement membrane and cell-matrix adhesion at the leading-edge of egressing endocrine cells. Altogether, these findings implicate an unexpected role of Syt13 in regulating cell polarity to orchestrate endocrine cell egression and islet morphogenesis.


Introduction
The pancreatic endocrine cells (α-, β-, δ-, Ɛ-, and PP-cells) regulate glucose homeostasis. How endocrine lineage segregation in interlinked with morphogenesis is not well understood. During endocrinogenesis, a subset of multi-/bipotent progenitors, that reside in the epithelium, gradually express the transcription factor (TF) Neurogenin 3 (Neurog3; hereafter called Ngn3) to sequentially generate endocrine progenitors (Ngn3 low ) and precursors (Ngn3 high ) 1 . The Ngn3 high precursors give rise to different endocrine cell types via a transient intermediate expressing the transcription factor (TF) Fev 2 ( Figure 1A). The stepwise lineage formation is tightly connected with tissue morphogenesis, in which endocrine cells egress from the ductal epithelium (also referred to endocrine cell delamination) and cluster into the proto-islets [3][4][5] . These morphogenetic events are initiated after Ngn3 induction and followed by the reduction of the apical plasma membrane (PM) area to form a tether structure, which ultimately undergoes abscission [6][7][8] . Cytoskeletal dynamics and small GTPases such as Cdc42 and Rac1 are involved in these stepwise events and coordinate local endocrine cell egression to form proto-islets described as the peninsular model [9][10][11] . However, whether asymmetric cell division 4,12 , endocrine cell migration 13,14 and epithelial-to-mesenchymal transition (EMT) 8,15 mediate endocrine cell egression from the pancreatic epithelium is still a matter of debate. Moreover, the upstream morphogenetic drivers orchestrateing cell and cytoskeletal dynamics during endocrine cell egression have not been identified. Further, the molecular mechanisms enforcing egressing cells to push or break through the basal lamina are unknown. Deciphering these mechanisms aids to tissue engineer islets for cell-replacement therapy but also to identify novel targets to intervene on epithelial cell dissemination during cancer metastasis.
Here, we report Synaptotagmin 13 (Syt13) ( Figure 1B) as a major regulator of endocrine cell egression and islet morphogenesis during endocrinogenesis. Syt13 is an atypical member of the Synaptotagmin (SYT) family of membrane trafficking proteins, which are known to be involved in intracellular vesicle trafficking and exocytosis. This protein family consists of 17 members and each possesses a transmembrane sequence connected to two lipid binding cytoplasmic domains (C2A and C2B), responsible for docking and fusion of the carrier vesicles to the target membrane 16,17 . Compared to well-known members, such as Syt1 and Syt2, Syt13 does not have an N-terminal sequence preceding the transmembrane region, and does not interact with membranes in a calcium (Ca 2+ )-dependent manner 18,19 . Further, although classical Syt proteins are involved in vesicle exocytosis, the cellular and molecular function of atypical members, such as Syt13, is less described.
By combining mouse genetics, high-resolution cell biology and single cell mRNA profiling, we identified a critical function of Syt13 in pancreatic endocrine morphogenesis. Our data identified apical-basal to front rear polarization during endocrine cell egression, which was accompanied by upregulation of molecular components involved in cell motility and cytoskeletal dynamics. Importantly, we uncovered Syt13 as a leading-edge protein that remodels the basement membrane and cell-matrix adhesion during endocrine morphogenesis. Together, our findings provide a detailed mechanistic description of how endocrine precursors leave the pancreatic epithelium and discovered Syt13 as a polarity protein and morphogenetic regulator during islet formation.

Syt13 function is required for endocrine cell egression and migration
In a global gene expression profiling during pancreas development, we previously identified Syt13 to be expressed during the peak of endocrinogenesis 20 . To further pinpoint the temporal and cell-type specific expression pattern of Syt13 mRNA, we leveraged single-cell RNA-sequencing (scRNA-seq) data from mouse embryonic pancreatic epithelial cells that were sampled from E12.5 to E.15.5 21 . Syt13 expression was specific to endocrine progenitors/precursors (EPs) and lineage cells ( Figure 1C), and together with Syt7 first expressed of all SYT family members during endocrinogenesis ( Figure S1A). Likewise, we found SYT13 and SYT7 to be early-onset SYT genes in human in vitro stem cell differentiation toward endocrine lineage 22 ( Figure S1B). Next, we confirmed Syt13 protein synthesis and localization in cells expressing high levels of TF Foxa2 (Foxa2 high ) leaving the ductal epithelium ( Figure 1D, blue dashed lines) or clustered in proto-islets ( Figure 1D, yellow dashed line). Further analysis indicated that Syt13 protein expression was restricted to the endocrine lineage and it was synthesized in a major fraction of embryonic α-and β-cells, but only in a minor fraction of δ-cells ( Figure 1E and 1F). To uncover the cellular and molecular function of Syt13, we generated a Sy13 knockout (KO) mouse line using the genetrap targeted mESCS (EUCOMM), in which the critical exon 2 was removed. This strategy resulted in the generation of a Flox (F)-deleted (FD) allele and whole-body Syt13 FD/FD (Syt13 KO) mice (Figure S1C-F). Mendelian ratio of heterozygous (Syt13 +/FD ) intercrosses indicated no lethality of Syt13 FD/FD at embryonic but at prenatal stages ( Figure S1G).
As the heterozygous mice showed similar phenotype to the wild type (WT) animals, we refer to both genotypes hereafter as controls. We detected similar size and weight of control and Syt13 FD/FD embryos at E18.5 ( Figure S1H and S1I). Gross morphology of control and Syt13 FD/FD pancreata were comparable at E18.5 ( Figure S1J). Moreover, we found no evident histological alterations in the Syt13 FD/FD pancreatic epithelium at E12.5 and E13.5 ( Figure S2A and S2B). Further analysis revealed no changes in apical-basal polarity in Syt13 FD/FD ductal epithelial and acinar cells ( Figure S2C and S2D). These data reveal normal ductal and acinar differentiation and morphogenesis upon Syt13 deletion and indicate that Syt13 expression and function is highly specific to the endocrine compartment.
Starting from E14.5, striking alterations in epithelial organization appeared in the Syt13 KO pancreata.
While the pancreatic epithelium of control mice consisted of a single-layer of Foxa2 low cells, the Syt13 KO pancreata contained a multi-layer epithelium. The extra layers were mainly Foxa2 high cells ( Figure 1G and S2E) and negative for the bipotent/ductal cell marker, Sox9 ( Figure 1H), suggesting a defect during endocrine lineage acquisition. In support of this, we found an increased number of retained EPs/endocrine cells (Nkx6-1 high /Sox9 -) within the Syt13 FD/FD compared to the control epithelium ( Figure 1I and 1J).
Moreover, we detected an increased direct attachment between the proto-islets and the epithelium in Syt13 FD/FD compared to control pancreata ( Figure 1K and 1L). Additionally, the typical proto-islet rearrangement, in which α-cells are at the periphery and β-cells are at the core, was disrupted in Syt13 KO pancreata ( Figure S2F). To dissect at which cell state Syt13 regulates endocrine cell egression, we generated two lineage and cell type-specific Syt13-conditional KO (CKO) mice using Ngn3 Cre/+ and Ins1 Cre/+ Credriver lines to delete Syt13 specifically in EPs and the β-cell lineage, respectively. We found an impairment in β-cell egression/migration in Ngn3 Cre/+ ; Syt13 F/FD , but not in Ins1 Cre/+ ; Syt13 F/FD animals ( Figure S2G and S2H). We conclude that Syt13 function is crucial for endocrine cell egression in EPs, but not in newly formed β-cells.
High expression levels of Syt13 in endocrine precursors and Fev + cells prime β-cell fate We next dissected the expression dynamics of Syt13 in endocrine lineages using the scRNA-seq data 21 .
Syt13 expression levels increased from Ngn3 low progenitors to Ngn3 high precursors ( Figure 2A). Moreover, Ngn3 high precursors from E14.5 and E15.5 had higher levels of Syt13 compared to those from E12.5 and E13.5 ( Figure 2B). To investigate the possible link between Syt13 expression levels and different endocrine cell fates, we divided Ngn3 high precursors from E12.5-15.5 into two clusters expressing high levels (Syt13 high ) and no or low levels of Syt13 (Syt13 low/-) ( Figure 2C). The fraction of Syt13 high cells increased from E12.5 to E15.5 ( Figure 2D). Differential gene expression analysis between Syt13 high and Syt13 low/-EPs identified several previously reported EP-signature genes 21 ( Figure 2E and Table S2). The majority of signature genes , which likely associate with the β-cell fate and include Neurod2, Sult2b1 and Upk3bl, were highly expressed in the Syt13 high precursors, whereas genes, which are possibly linked to the α-cell fate including Dll1, Rsad2 and Rasgrp3, were highly expressed in Syt13 low/precursors ( Figure 2F). Furthermore, the expression levels of Pax4, Gck, Nkx6-1 and Nkx2-2 were higher in Syt13 high than Syt13 low/precursors, suggesting that Syt13 high Ngn3 high cells are primed for β-cell fate allocation ( Figure 2G).
We then analyzed Syt13 expression in Fev + cells and found higher expression levels in Fev + β compared to other Fev + endocrine subtypes ( Figure 2H). Additionally, the percentage of Syt13 high cells was higher in Fev + β cells than in Fev + α, Fev + δ and Fev + ε and Fev + Pyy cells. ( Figure 2I). Next, we splitted all Fev + cells into Syt13 high and Syt13 low/cells and performed differential gene expression analysis ( Figure S3A and Table   S2). Pax4 and Pcsk2 were upregulated in Syt13 high , while Arx, Isl1 and Pou3f4 were upregulated Syt13 low/-Fev + cells ( Figure 2J). These data indicate that higher expression levels of Syt13 in Fev + cells associate with β-cell program. Consistent with Syt13 expression in EPs and Fev + cells, the fraction of Syt13 high cells was highest in β-cells of all different hormone + endocrine subtypes ( Figure 2K). Moreover, partition-based graph abstraction (PAGA) analysis showed high connectivity between Syt13 high precursors/Fev + cells, which indicates that Syt13 high precursors/Fev + cells transition to β-cells ( Figure S3B). In line with these findings, in scRNA-seq data of human fetal pancreas 23 we found higher SYT13 expression in fetal β-cells and their precursors compared to fetal α-cells and their precursors ( Figure S3C). To support the findings from scRNA-seq analysis, we performed qPCR analysis of FACS isolated Ngn3 + cells from E15.5 NVF; Syt13 FD/FD pancreata, which were obtained through crossing Syt13 +/FD mice with homozygous Ngn3-Venus fusion (NVF) reporter mouse line. We found comparable levels of Ngn3 in the control and Syt13 FD/FD Ngn3 + cells. Furthermore, an increased expression level of Arx but not Pax4 was observed in the Syt13 FD/FD Ngn3 + cells compared to the controls ( Figure 2L). Collectively, these analyses demonstrate that high physiological levels of Syt13 mRNA expression in endocrine precursors and Fev + cells correlates with β-cell fate acquisition.

Lack of Syt13 is dispensable for endocrine induction but reduces β-cell specification
To validate the results from the scRNA-seq data, we examined the impact of Syt13 knockout on endocrine lineage formation. We found comparable numbers of Ngn3 + cells from control and Syt13 KO mice at E13.5-15.5 ( Figure 3A and 3B). FACS analysis also revealed similar percentage of Ngn3 + cells in NVF; Syt13 FD/FD and control pancreata at E15.5 ( Figure 3C). In addition, Syt13 KO and control pancreata contained a comparable fraction of cells expressing the pan-endocrine cell marker, Chromogranin A (ChgA) ( Figure   3D and 3E). Next, we quantified the number of α-and β-cells at E14.5-16.5 and E18.5. A significant increase in the number of α-cells at the expense of β-cells in Syt13 KO pancreata was detected ( Figure 3F and 3G). Importantly, we found that the deletion of Syt13 in EPs but not in the newly generated insulinexpressing cells resulted in an increased α-to β-cell ratio ( Figure 3H-J and S3D). Collectively, these results support the findings from the in vivo scRNA-seq data and further indicate that Syt13 acts downstream of Ngn3 and its function in EPs is essential for β-cell fate allocation. Therefore, Syt13 regulates endocrine cell egression and allocation at EP/Fev + state but not after β-cell specification.

Syt13 localizes at the leading-edge of egressing endocrine cells
To decipher the cellular processes by which Syt13 coordinates endocrine cell egression and islet morphogenesis, we first explored the intracellular localization of the protein. The onset of Syt13 mRNA expression occurs in the epithelium-residing EPs ( Figure 1C). Therefore, we first assessed the localization of this protein in Syt13-overexpressing Madin-Darby Canine Kidney (MDCK) cells cultured in a 3D condition as a model for an apical-basal polarized epithelium. Remarkably, Syt13 was specifically localized at the apical domain of polarized epithelial cysts ( Figure 4A). Proximity-dependent biotin identification (BioID) further identified the close proximity of Syt13 with several apical polarity determinants including aPKC, Ezrin, EBP50 and Merlin ( Figure 4B). Surface biotinylation assay indicated the incorporation of Syt13 within the PM ( Figure 4C). Further, structure-function study unveiled the requirement of both cytoplasmic C2A and C2B domains for integration and apical PM localization of the Syt13 protein ( Figure   4D, E and S4A). This PM localization was different to Syt1 and Syt2, which are mainly enriched in the synaptic vesicles and mediate exocytosis. This suggests a distinct function of Syt13 compared to typical Syt proteins.
Next, we explored the cellular localization of Syt13 during endocrinogenesis. In newly generated endocrine cells, which were still fully integrated within the pancreatic epithelium, Syt13 was detected at the apical PM domain ( Figure 4F and S4B). Importantly, during endocrine cell egression, Syt13 was mainly accumulated at the basal side of the ductal epithelium, which becomes the front of the precursors that leave the epithelium ( Figure 4G and 4H). This localization pattern was also identified in cells, which appear to migrate out of the epithelium towards the proto-islet clusters ( Figure 4I). Along this line, we found increased expression levels of several lamellipodium-related genes including Vasp, Cotl1, Vil1, Arpc2, Cyfip2 and Marcks in Syt13 high compared to Syt13 low/precursors ( Figure S4C). Expression of these genes started in EPs and peaked in Fev + cells ( Figure 4J), which stresses that these cell states are highly motile. Additionally, we identified close proximity of Syt13 with several proteins including Actin, Vasp, Cortactin, the Arp2/3 regulators WAVE2 and N-WASP, but not with Cofilin ( Figure 4K), further indicating leading-edge localization of Syt13.
One hallmark of directed cell migration is increased phosphatidylinositol triphosphate (PtdIns(3,4,5)P3 or PIP3) levels at the cell leading-edge. To investigate whether Syt13 interacts with phosphatidylinositol phospholipids, we performed an in vitro lipid-binding analysis of purified recombinant Syt13 protein variants ( Figure S4D) to large unilamellar vesicles (LUVs). While C2A domain showed low binding to LUVs, C2B and C2AB domains exhibited binding preference towards PIP3, PtdIns(3,5)P2, PtdIns(4,5)P2 and with lower degree with PtdIns(3,4)P2 ( Figure 4L and S4E-H). These data indicate that Syt13 directly interacts with PIP2 and PIP3 phosphoinositides mainly through its C2B cytoplasmic domain. Next, we tested if inhibition of PIP3 formation or function mimics the Syt13 action during endocrine cell egression.
Administration of both inhibitors resulted in striking impairments in the endocrine cell egression ( Figure   4M, N and S4I). Overall, these data demonstrate that Syt13 localizes at the leading-edge of egressing cells to regulate endocrine cell repolarization along the front-rear axis ( Figure 1).

Syt13 levels correlate with the expression of unique molecular components of cell motility and cytoskeleton in endocrine precursors
To identify the molecular components linked with Syt13 functions during endocrine cell egression, we first analyzed the expression dynamics of EMT-related genes. We found low and transient expression of Snail1/2 in EPs and a transient downregulation of E-cadherin (Cdh1) in Fev + cells that did not correlate with increased expression of N-cadherin (Cdh2) at this cell state. Together with the absence of other classical EMT marker genes ( Figure S5A), this suggests that endocrine cell egression is not mediated by a complete EMT program 24 . Pathway enrichment analysis of differentially expressed genes between Syt13 high and Syt13 low/precursors showed that increased Syt13 expression in precursors coincided with upregulation of cell migration, actin and microtubule (MT) cytoskeletal organization, membrane protrusions and intracellular transport pathways ( Figure 5A and Table S2). In Syt13 high precursors several genes that have been previously shown to regulate cell migration in different cellular systems and include Tgfbr1, Cd40, Fgf18, Ret, Fer, Gfra3, Arhgef7, Slit1, Dok4, Olfm1 and Lingo1 were highly expressed ( Figure 5B and Table S2). Notably, the expression of most of these genes peaked at EPs or Fev + cells, which indicates increased cell motility at these cell states ( Figure S5B). Also several genes involved in actin and MT cytoskeletal rearrangement were differentially expressed. Tubulins (Tuba1a, Tuba4a, Tubb3), MT organizers (Mtcl1, Mapre3, Kif26a), MT-associated trafficking proteins (Map1a, Map1b, Dynll2, Dync1i1, Kif5b), actin remodeling proteins (Tmsb4x, Dbn1, Scin, Vil1, Mical2) and the cytoskeletal linker protein dystonin (Dst) were increased in Syt13 high compared to Syt13 low precursors ( Figure 5C and Table S2).

Syt13 interacts with the acetylated-tubulin, Map1b and dystonin cytoskeletal proteins
To explore which of the identified molecular components in the endocrine precursor and lineage cells are interlinked with Syt13 function, we performed two complementary interactome analyses. Mass spectrometry identified 79 and 69 proteins as potential Syt13 interaction partners using affinity purification and BioID proximity labeling, respectively ( Figure 6A, B and Table S3). Pathway analysis revealed terms associated with cytoskeleton, vesicle trafficking, protein internalization and degradation, cell adhesion and movement ( Figure S6A and S6B). Importantly, we identified α-tubulin (Tuba1a) in the direct interactome and dystonin (Dst) in both interactome lists. We confirmed the interaction of Syt13 with α-tubulin that was not hampered by deletion of Syt13 C2A or C2B domains ( Figure 6C and S6C, D). α-tubulin is the major form of tubulins that can be acetylated (Ac-tub) to produce stable long-lived MTs. IF analysis disclosed the increased Ac-tub content concomitant with increased Syt13 protein levels during endocrinogenesis ( Figure   6D and S6E). Moreover, the transcripts of enzymes involved in α-tubulin acetylation (Atat1) and deacetylation (Hdac6 and Sirt2) were increased during endocrinogenesis ( Figure 6E). Remarkably, these genes were not differentially expressed in Syt13 high and Syt13 low/precursors ( Figure S6F), and Ac-tub was still detected in Syt13 KO cells ( Figure S6G). Yet, Syt13 was accumulated in the intracellular area enriched for Ac-tub and directly associated with this tubulin in epithelial cell lines ( Figure 6F-H and S6H).
The interconnection between Ac-tub, Map1b and dystonin and the direct interaction between Map1b and dystonin have been previously shown [25][26][27] . We further found co-expression of Ac-tub with both Map1b and dystonin in mouse and human endocrine cells ( Figure 6I-K and S7A). Furthermore, we revealed that both Map1b and dystonin localized at similar cellular domains to Syt13 in endocrine cells ( Figure 6L and 6M).
Yet, no apparent alterations in the expression levels nor the localization of both proteins was observed in Syt13 KO cells ( Figure S7B and S7C). Overall, these data indicate specific upregulation of Actub/Map1b/dystonin protein complex during endocrinogenesis and that Syt13 interacts with this protein complex for its trafficking and localization.

Syt13 regulates vesicle endocytosis and remodels basal lamina
Syt13 is a leading-edge protein ( Figure 4) and its KO reduces endocrine cell egression ( Figure 1). Moreover, Syt13 interacts with a unique cytoskeletal protein complex ( Figure 6). Since Syt13 is also a member of a vesicle trafficking protein family, these findings suggest that Syt13 might be involved in leading-edge turnover of egressing endocrine cells. In support of this idea, we identified several Syt13-interacting proteins involved in endocytosis (Caveolin-1 (Cav-1), Hgs (Hrs), Stam2, Scyl2 and Snx1) and membrane and endosomal trafficking (Ank3, Dync1h1, Dynll1, Myo6, Rab5, Rack1, Tuba1a, Vps26b and Dst), ( Figure 6A, B and Table S3). Among these, we confirmed the interaction of Syt13 with Cav-1, HGS and Rab5 ( Figure 7A). Furthermore, Syt13 was colocalized with Cav-1 at the PM of MDCK cells, while a fraction of this protein was localized to the intracellular vesicles ( Figure 7B). Finally, we confirmed the interaction of Syt13 with Rab7 and Rab11 in IP and its colocalization with Lamp1 and Rab7 in IF, revealing the engagement of Syt13 in protein endocytosis, trafficking, degradation and recycling ( Figure 7A, C and S7D).
The endocytosis function of Syt13 and its localization at the leading-edge (basal domain of precursors before or upon cell repolarization) of egressing endocrine cells, suggested its possible function in removal or recycling of the integrin-based cell-matrix adhesion complexes. To find out which integrin subunits are expressed during endocrinogenesis, we analyzed the scRNA-seq data indicating α3, α5, α6, α9, αV, β1, β4, β5 and β6 with different expression levels in epithelial cells ( Figure 7D). Most of these integrin subunits were downregulated at EP and Fev + states ( Figure 7D), suggesting the necessity of their remodeling during endocrine cell egression. Among these, an increased expression of α6 integrin (CD49f) in acinar cells compared to lower levels in ductal and endocrine lineages has been previously reported 28  Similar increased levels were also observed for β4 integrin subunit ( Figure 7J). These data demonstrate a function of Syt13 in the elimination of integrin-based adhesion structures at the leading-edge of egressing endocrine cells.

Discussion
Here we present the detailed cellular and molecular underpinnings of endocrine cell egression. Setting our findings in the context of previous studies, we propose a model, in which endocrine cell egression and migration occur at EP and Fev + cell states. The onset of Ngn3 expression in EPs triggers a partial EMT program 8,15 and results in the expression of a set of unique cytoskeletal and trafficking components, such as the polarity regulator and morphogenetic driver Syt13. Beginning at this stage, endocrine lineages condense their apical PM 6 marked by Syt13 protein, gradually remodel adherens junctions and reduces the apical-basal polarity components 29 . The relocalization of Syt13 from the apical side to the leading-edge switches the apical-basal polarity to a front-rear axis essential for EPs to leave the ductal epithelium. This process is likely mediated by intracellular vesicle trafficking and requires a stable MT network consisting of proteins, such as acetylated and β3-tubulin. At the rear part of the egressing cells, the apical domain continuously narrows, which results in a formation of a tether structure connecting the cells with the epithelial plane 7 . At the front domain, Syt13 remodels the basal lamina and modulates cell-matrix adhesion.
Syt13 in concert with membrane protrusion proteins possibly generate an active and dynamic leading-edge that after abscission of the rear tether structure 7 enforce the detachment of the egressing cells from the epithelium.
The identity of the cellular processes, which coordinate endocrine cell egression is still a matter of debate.
We found a previously unknown cell repolarization process during endocrine cell egression. Prior studies have shown reduction of the apical domain that ultimately leads to loss of apical-basal polarity in endocrine cells 6,29 . Here, we provide evidence that after loss of apical-basal polarity, egressing and migrating endocrine cells acquire a front-rear polarity. Importantly, we found Syt13 as the first protein marking this repolarization event. The direct interaction of Syt13 with phosphoinositide phospholipids indicates the involvement of these lipids in apical-basal to front-rear repolarization. Therefore, it is likely that Syt13 is recruited to the apical or leading-edge domain through its direct interactions with PIP2 and PIP3, respectively, as it has been shown for Syt1 in neurons 30 . Cell repolarization and the expression of actinbased membrane protrusion components indicate the existence of an active leading-edge, driving required forces for endocrine egression and directed cell migration toward the proto-islets. Additionally, we found that the ECM components are not degraded but likely remodeled during endocrine cell egression. Therefore, the newly differentiated cells that leave the epithelium join the nearby clusters with whom they share the basement membrane resulting in a local islet formation as described before 9 .
A set of unique cytoskeleton-related genes were differentially expressed between Syt13 high and Syt13 low/precursors including several actin-remodeling proteins. Among these were actin depolymerizing and sequestering genes such as Tmsb4x, Scin, Cfl1 and Mical2, and several genes involved in actin capping, bundling and elongation such as Dbn1, Vasp, Marcks and Vil1. Along this line, the remodeling of actin cytoskeleton has been shown to regulate endocrine differentiation 31,32 . Our analysis uncovered the molecular factors that mediate actin network rearrangements for endocrine egression. Further, we identified the increased expression of a set of particular tubulins and MT-associated proteins in Syt13 high precursors including Ac-tub, Map1b and dystonin. Strikingly, Syt13 directly interacted with Ac-tub, suggesting the dependency of Syt13 to these stable MTs for its intracellular trafficking and leading-edge localization.
Moreover, the association of Syt13 with Map1b and dystonin indicates the requirement of a unique cytoskeletal railroad for Syt13-mediated vesicle trafficking. This notion together with the restricted expression of Syt13 in certain cell types, suggests that Syt13 mediates internalization and trafficking of a subset of specific membrane proteins.
Deletion of Syt13 impaired endocrine cell egression and islet morphogenesis. Furthermore, lack of this protein resulted in a shift in β-to α-cell fate. It is possible that the absence of Syt13 impacts endocrine cell egression, which consequently influences their fate decision. Along this line, lack of p120ctn in EPs has resulted in faster egression that subsequently increases their differentiation into α-cells 33 . Contrary to this, our findings show that the delay in cell egression is linked with increased EP specification towards α-cells.
Therefore, activation of a β-cell program might require a precise timing of EP occupancy within the epithelium and that faster or slower egression process induce α-cell fate. Alternatively, lack of Syt13 might directly affect β-cell differentiation. This notion is supported by the correlation between increased Syt13 expression levels in endocrine precursors and Fev + cells with β-cell programs. As Syt13 is involved in endocytosis, it may trigger β-cell programs through controlling the involved signaling pathways and TF networks by modulating the surface receptors. Furthermore, Syt13 high precursors expressed higher levels of F-actin depolymerizing and sequestering genes. Two previous studies have shown the positive impact of F-actin network reduction on β-cell differentiation 31,32 . Thus, it will be interesting to explore in the future whether Syt13 is directly involved in modulating actin cytoskeleton for β-cell fate decision.
We uncovered an unexpected role for the atypical Syt13 protein as endocrine cell polarity regulator and morphogenetic driver of progenitor egression. Different to the typical Syt members such as Syt1 and Syt2, deletion results in the earliest lethal phenotype among all Syt members, highlighting the unique and critical role of this protein for organ formation and function. In support of this, Syt13 is expressed in the brain 36 , and its associated cytoskeletal proteins are also expressed in neuronal cells such as sensory neurons 37 . In addition, a protective function of Syt13 in motor neurons of patients with neurological disorders has been shown 38 . Thus, our findings will help to dissect the molecular action of Syt13 during neurogenesis and neurological disorders. Remarkably, SYT13 is also upregulated in several cancer cell types and its inhibition reduces cancer cell metastasis [39][40][41] . These studies support a polarity and morphogenetic function of Syt13 and highlight the importance of our findings in the context of organ formation but also cancer cell dissemination.  (Table S1). To generate Syt13 tissue-specific conditional knockout mice, Syt13 F/F mice were crossed with constitutive Tg (Neurog3-cre)C1Able/J (Ngn3 +/Cre ) 42 and Ins1 Cre (Ins1 tm1(cre)Thor ) (Ins1 +/Cre ) mice 43 . Syt13-Venus fusion mouse line, in which endogenous Syt13 is fused with the fluorescent protein Venus has been recently generated by us and will be described elsewhere.

Cloning, cell culture and transfection
Cloning was performed using standard protocols. Different Syt13 constructs were generated using the pCAG mammalian expression vector (Table S1) The differentiation of human iPSCs into pancreatic endocrine cells in vitro was performed using previously described protocol 45 with slight modifications 29 .

Immunostaining and imaging
Dissected embryonic pancreata or explant culture samples were fixed in 4% PFA in PBS for 2 h overnight at 4 °C. The tissues were merged in 10% and 30% sucrose-PBS solutions at RT (2 h each solution) followed by 1:1 solution 30% sucrose:tissue-freezing medium (Leica 14020108926). Afterwards, they were embedded in cryoblocks using tissue-freezing medium and sections of 20 μm thickness were cut using a cryostat. Next, the samples were permeabilized (0.1% Triton, 0.1 M Glycine) for 30 min and incubated in blocking solution (10% FCS, 3% Donkey serum, 0.1% BSA and 0.1% Tween-20 in PBS) for 1 h at room temperature (RT). Then, the primary antibodies (Table S1) diluted in the blocking solution were added to the samples overnight at 4 °C. After washing with PBS they were stained with secondary antibodies (Table   S1) diluted in the blocking solution for 3-5 h at RT. The samples were then incubated with 4', 6-diamidin-2-phenylindol (DAPI), followed by washing with PBS and embedding in commercial medium (Life Tech., ProLong Gold).
2D and 3D cultures of cell lines or primary pancreatic cells were fixed in 4% PFA (12 min at 37 °C) followed by 10 min permeabilization (100 mM Glycine and 0.2% Triton X-100) at RT. After 3x washing, cells were incubated with blocking solution for 30 min at RT and then incubated with primary antibodies (Table S1) for 1-3 h at RT. After washing, cells were incubated with secondary antibodies (Table S1) for 1 h at RT and DAPI was added and samples were embedded.
All images were obtained with a Leica microscope of the type DMI 6000 using the LAS AF software.
Images were analyzed using LAS AF and ImageJ software programs.

Cell sorting and quantitative PCR (qPCR) analysis
Embryonic pancreata from NVF; Syt13 KO at E15.5 were dissected. Next, individual pancreata were kept in 0.25% Trypsin for 5 min on ice and then incubated at 37 °C for 10 min. The single-cell samples were then centrifuged at 1700 rpm for 5 min at 4 °C. 5 µl anti-mouse CD326 (EpCAM) PE (eBioscience, 12-5791-81) and rat IgG2a K isotype control (eBioscience, 12-4321-42) were used for 1x10 6 (Table S1) followed by 3x washing and then incubating with the secondary antibodies (Table S1)  Supernatants were discarded and the beads were used for western blotting or mass spectrometry as follows.
For western blotting, the beads were washed 3x with the washing buffer (50 mM Tris buffer pH 7.5, 0.5% Sodiumdesoxycolate, 150 mM NaCl, 1% NP40, 0.5% SDS) and 2x with 1x TBS. The beads were then resuspended in 90 µL sample buffer, heated for 5 min at 95 °C. After cooling down on ice the samples were centrifuged and the supernatants were collected as the IP sample, which together with the input samples were applied to western blotting.
For mass spectrometry, 500 µl TBS was added to the beads and mixed rigorously. After centrifugation at 7000 x g for 30 second supernatants were removed and the washing was repeated two more times. 60 µl elution buffer (2 M urea, 50 mM Tris-HCl pH 7.5 and 5 μg/mL Trypsin (SIGMA; T6567-5X20UG)) (twice as the bead volume) was added and samples were incubated for 1 h at 27 °C under constant agitation (800 rpm in a thermo mixer) followed by centrifugation. Supernatants were collected in a fresh Eppendorf tube.
The beads were then resuspended with 60 µl of the elution buffer (containing 2 M urea, 50 mM Tris-HCl pH 7.5 and 1 mM DTT), centrifuged and the supernatant was collected and pooled with the previous one. This last step was repeated and 180 µl total volume per each reaction was obtained. Samples were left at RT to continue to digest overnight and were stored at -80 °C for further mass spectrometry procedure.

Mass spectrometry
Affinity purified eluates were precipitated with chloroform and methanol followed by trypsin digestion as described before 46 . LC-MS/MS analysis was performed on Ultimate3000 nanoRSLC systems (Thermo Scientific) coupled to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific) by a nanospray ion source. Tryptic peptide mixtures were injected automatically and loaded at a flow rate of 30 μl/min in 0.1% trifluoroacetic acid in HPLC-grade water onto a nano trap column (300 μm i.d. × 5 mm Pre column, packed with Acclaim PepMap100 C18, 5 μm, 100 Å; Thermo Scientific). After 3 minutes, peptides were eluted and separated on the analytical column (75 μm i.d. × 25 cm, Acclaim PepMap RSLC C18, 2 μm, 100 Å; Thermo Scientific) by a linear gradient from 2% to 30% of buffer B (80% acetonitrile and 0.08% formic acid in HPLC-grade water) in buffer A (2% acetonitrile and 0.1% formic acid in HPLC-grade water) at a flow rate of 300 nl/min over 117 minutes. Remaining peptides were eluted by a short gradient from 30% to 95% buffer B in 5 minutes. Analysis of the eluted peptides was done on an LTQ Fusion mass spectrometer. From the high-resolution MS pre-scan with a mass range of 335 to 1500, the most intense peptide ions were selected for fragment analysis in the orbitrap depending by using a high speed method if they were at least doubly charged. The normalized collision energy for HCD was set to a value of 27 and the resulting fragments were detected with a resolution of 120,000. The lock mass option was activated; the background signal with a mass of 445.12003 was used as lock mass 47 . Every ion selected for fragmentation was excluded for 20 seconds by dynamic exclusion. MS/MS data were analyzed using the MaxQuant software (version 1.6.1.0) 48,49 . As a digesting enzyme, Trypsin/P was selected with maximal 2 missed cleavages. Cysteine carbamidomethylation was set for fixed modifications, and oxidation of methionine and N-terminal acetylation were specified as variable modifications. The data were analyzed by label-free quantification with the minimum ratio count of 3. The first search peptide tolerance was set to 20, the main search peptide tolerance to 4.5 ppm and the re-quantify option was selected. For peptide and protein identification the human subset of the SwissProt database (release 2014_04) was used and contaminants were detected using the MaxQuant contaminant search. A minimum peptide number of 2 and a minimum length of 7 amino acids was tolerated. Unique and razor peptides were used for quantification. The match between run option was enabled with a match time window of 0.7 min and an alignment time window of 20 min. The statistical analysis including ratio, t-test and significance A calculation was done using the Perseus software (version 1.6.2.3) 50 .
Interaction candidates from mass spectrometry were used to gain insights into the biological processes involved with Syt13. We used Metascape 51 to perform the pathway enrichment analysis.

Electrochemiluminescence-based immunoassay
The binding assay was performed using Meso Scale Discovery 384 well high bind plates as described previously 53 . In brief, all binding experiments were carried out at 22 °C. Liposomes (2 μl) were passively adsorbed on the electrode surface for 1 h, and residual sites on the surface were blocked for 1 h with 0.25% porcine gelatin (Sigma-Aldrich) in TRIS buffer (50 mM Tris, 150 mM NaCl, pH 8.0). After three washing steps with TRIS buffer, serial dilutions of recombinant protein in blocking buffer was added to the myc antibody (clone -c -respective wells and incubated for 2 h. Unbound protein was removed, and anti was applied for 1 hnology) at 1.25 µg/ml concentration in 0.25% porcine gelatin 9E10, Santa Cruz Biotec h followed by three subsequent wash steps with TRIS buffer. mouse -For detection, a secondary anti g/ml in blocking buffer for TAG (Meso Scale Discovery) was used at 1.25 μ -antibodies labeled with Sulfo free reading buffer -1 h in the dark. Free secondary antibody was washed off, and reading buffer (surfactant from Meso Scale Discovery) was added. The readout was performed on a Meso Scale Discovery SECTOR scence reader. Imager 6000 chemilumine Data were analyzed with GraphPad Prism 6.07. First, signal from PC/cholesterol (70/30 mol %) vesicles was subtracted as a background. Next, a non-linear curve fitting was applied, and the binding kinetics were calculated using one site -specific binding fitting. Processed and normalized scRNA-seq data and cell annotations of human in vitro stem cell differentiation 22 were downloaded from GEO (accession number GSE114412). The human data contains cells sampled from stage 5 of the differentiation protocol.

Code availability
Custom notebooks for all analyses of scRNA-seq data will be made available in a github repository upon publication.

Statistical analysis
All statistical analysis was performed on GraphPad prism 9 and presented in figure legends.

Image quantification
Endocrine egression in vivo: Quantification was performed using confocal images from randomly selected pancreas areas. The quantification was done by counting the ratio (as percentage) of Nkx6-1 high /Sox9cells that were residing within or in the direct contact with the epithelium to the total Sox9 + cells. For each condition, more than 11 confocal images from randomly selected pancreas areas from 4 animals were used.
The analysis was conducted using LAS-AF software.
Endocrine cluster direct attachment to the epithelium: Quantification was performed using confocal images from randomly selected pancreas areas. We divided the area of epithelium-proto islets direct attachment to the total area of proto-islet periphery Areas were detected via "wand tracing tool", supervised by adjusting the tolerance level based on the quality of the staining. The surface of contact was manually drawn. The analysis was conducted using Fiji ImageJ.
Ngn3 quantification: Quantification of the endocrine progenitor (Ngn3 + ) cells was performed on confocal images from randomly selected pancreas areas by using automatic nuclei counts with IMARIS software Syt13 apical localization: Signal intensity of the recombinant Syt13 protein variants at the apical domain were divided to the total cytosolic signal. 32 total epithelial cysts from each condition were quantified from 4 independent experiments. The analysis was conducted using LAS-AF software.
Endocrine delamination in the explant culture: Quantification was performed using confocal images from randomly selected pancreas areas. The quantification was done by counting the percentage of GFP + cells that were residing within the epithelium of the total epithelial cells. For each condition, more than 65 epithelial domains from 3 independent experiments were quantified. The analysis was conducted using LAS-AF software.
α6 and β4 integrin subunit: Quantification was performed using confocal images from randomly selected pancreas areas. The signal intensity of the respective proteins at the basal domain of endocrine cells was divided to the signal at the basal domain of epithelial cells. For each protein, n ≥ 20 images from n ≥ 3 independent experiment were quantified. The analysis was conducted using LAS-AF software.    Figure   S7.