Vangl2 coordinates cell rearrangements during gut elongation

Abstract

Background.

The embryonic gut tube undergoes extensive lengthening to generate the surface area required for nutrient absorption across the digestive epithelium. In Xenopus, narrowing and elongation of the tube is driven by radial rearrangements of its core of endoderm cells, a process that concomitantly opens the gut lumen and facilitates epithelial morphogenesis. How endoderm rearrangements are properly oriented and coordinated to achieve this complex morphogenetic outcome is unknown.

Results.

We find that, prior to gut elongation, the core Wnt/PCP component Vangl2 becomes enriched at both the anterior and apical aspect of individual endoderm cells. In Vangl2-depleted guts, the cells remain unpolarized, downregulate cell-cell adhesion proteins and, consequently, fail to rearrange, leading to a short gut with an occluded lumen and undifferentiated epithelium. In contrast, endoderm cells with ectopic Vangl2 protein acquire abnormal polarity and adhesive contacts. Consequently, endoderm cells also fail to rearrange properly and undergo ectopic differentiation, resulting in guts with multiple torturous lumens, irregular epithelial architecture and variable intestinal topologies.

Conclusions.

Asymmetrical enrichment of Vangl2 in individual gut endoderm cells orients polarity and adhesion during radial rearrangements, coordinating digestive epithelial morphogenesis and lumen formation with gut tube elongation.

Introduction

The Wnt/Planar Cell Polarity (PCP) pathway plays fundamental patterning roles in embryogenesis. In a wide variety of contexts, PCP signaling organizes the polarity of cell features within a plane of developing tissue, such as aligning actin hairs along the proximodistal axis of the Drosophila wing (see Experimental Procedures; Adler 1992;Gubb 1993;Gubb and Garcia-Bellido 1982), or arranging stereocilia across the sensory epithelium of the mammalian cochlea (Wang et al. 2005). The ability of PCP signaling to orient tissue-level patterning results from polarized distribution of the “core” components of the pathway within individual cells. For example, the Drosophila protein Van Gogh and its binding partner Prickle form a membrane-bound complex selectively enriched at the proximal borders of wing cells, with complementary enhancement of a complex of Frizzled and Disheveled at the distal membranes (Davey and Moens 2017). These and other PCP components act within and between cells to establish and coordinate cell polarity across the tissue. Similar segregation patterns and interactions are exhibited by PCP proteins in vertebrate organs (Davey and Moens 2017). The conservation of PCP asymmetries across species and developmental contexts suggests this pathway plays a fundamental role in aligning the polarities of neighboring cells to achieve tissue-level patterning.

Interestingly, Wnt/PCP activity is also required to align cell polarities during dynamic tissue-level morphogenetic events. A prime example of this is convergent extension (CE), a process in which the polarized behaviors of multiple individual cells become aligned with respect to larger-scale embryonic axes in order to effectively narrow and elongate a tissue (Keller et al. 2000). During CE of the vertebrate body, cells first become elongated and aligned perpendicular to the embryo’s anterior-posterior (AP) axis. They then intercalate between each other, along the embryo’s mediolateral axis, narrowing the field of cells, while preferentially separating anterior and posterior neighbors to drive tissue lengthening (Keller et al. 1992;Shih and Keller 1992b;Shih and Keller 1992a). As observed in the fly wing, core members of the PCP pathway are asymmetrically distributed in converging and extending cells (Roszko et al. 2015;Davey and Moens 2017)—e.g., Pk homologs are enriched at anterior membranes (Ciruna et al. 2006;Gao et al. 2011;Yin et al. 2008) while Dvl is localized to posterior membranes (Sepich et al. 2011;Yin et al. 2008). Perturbing PCP signaling impairs cell polarity and intercalation during CE (Goto and Keller 2002;Yen et al. 2009;Topczewski et al. 2001;Marlow et al. 2002;Yin et al. 2008;Wallingford et al. 2000;Shih and Keller 1992a;Keller 2002;Jessen et al. 2002) suggesting that, in addition to coordinating tissue-level patterning, PCP asymmetries also orient cell properties and behaviors across tissues, in order to direct large-scale morphogenetic events.

As in body axis elongation, coordinated cell rearrangements are also required for the primitive gut tube to achieve its requisite length. In Xenopus, the overall topology of the gut changes dramatically during development, from a short, nearly solid cylinder of unpolarized endoderm cells (Figure 1A) to a long, hollow tube lined by columnar digestive epithelium (Figure 1C). To accomplish this, the endodermal cells first elongate along the radii of the gut tube, as if aligning along the spokes of a wheel (Figure 1B; Reed et al. 2009). They then intercalate radially (blue arrows in Figure 1B), opening up the central lumen of the gut while condensing into a single layer of epithelium (Figure 1C). During this process, the entire gut tube narrows and lengthens along its AP axis (Chalmers and Slack 2000;Chalmers and Slack 1998), suggesting that gut morphogenesis also involves CE (red and black arrows in Figure 1B), i.e., the radially intercalating cells preferentially separate anterior and posterior neighbors along the length of the tube (Reed et al. 2009).

Figure 1. Model of endoderm dynamics during the morphogenesis of the Xenopus primitive gut tube.

Between NF32 and 42 (A-B), the endoderm cells of the prospective midgut gradually elongate and become oriented along the radii of the primitive gut tube. Subsequent radial intercalation (blue arrows in B) facilitates lumen expansion and the morphogenesis of a single-layered digestive epithelium by NF46. This convergence of the rearranging endoderm also drives longitudinal tissue extension (black arrows in B), as the cells preferentially intercalate between anterior and posterior neighbors within the walls of the gut (red arrows), ultimately facilitating tube narrowing and elongation (C). (Adapted from Reed et al. 2009)

The similarities of gut morphogenesis with body axis elongation, i.e., coordinated cell rearrangements driving tissue-level shape change, suggests that Wnt/PCP signaling may also orchestrate the cell rearrangements of gut development. Indeed, gut lengthening is known to be dependent on the activity of several Wnt/PCP components, including Wnt5a, Ror2, and Sfrps (Cervantes et al. 2009;Li et al. 2008;Yamada et al. 2010). Moreover, key endoderm cell properties, including cell-cell adhesion and microtubule (MT) architecture, are tightly regulated by Wnt/PCP signaling mediators (e.g., Rho, ROCK, non-muscle Myosin, JNK); loss of function of these factors profoundly disrupts endoderm intercalation, epithelial morphogenesis and gut tube elongation (Reed et al. 2009;Dush and Nascone-Yoder 2013). Nonetheless, the functional roles of “core” PCP molecules (e.g., Vangl2, Pk, Dsh) in gut morphogenesis remain largely unknown.

Here, we describe the role of the core PCP component Vangl2 in Xenopus gut elongation. We find that Vangl2 expression is enriched at both the apical tips and anterior surfaces of endoderm cells as they become radially oriented. Perturbations of Vangl2 activity, either loss-of-function or overexpression, severely disrupt both epithelial morphogenesis and tissue elongation. We show that Vangl2 regulates key cell properties involved in endoderm rearrangement, including cell polarity, adhesion, and MT architecture. These results suggest that apical and anterior asymmetries in Vangl2 cellular localization orchestrate the complex multi-dimensional endoderm rearrangements that drive gut morphogenesis.

Results

Vangl2 becomes progressively apically and anteriorly enriched during gut morphogenesis.

To determine the role of Vangl2 in Xenopus gut morphogenesis, we first asked whether Vangl2 is asymmetrically localized in endoderm cells. Others have found it difficult to visualize endogenous PCP proteins during dynamic morphogenetic processes (e.g., CE) without employing mosaic overexpression strategies (Davey and Moens 2017). We therefore injected a low dose of Xenopus vangl2 mRNA (approximately 25% of the amount required to elicit a gain of function phenotype; see Experimental Procedures), to overexpress the protein in a mosaic manner. Embryos were then harvested at various stages for immunolocalization.

Early in gut morphogenesis, when endoderm cells still have an unpolarized, amorphous morphology [Nieuwkoop and Faber stage 35; NF35; (Nieuwkoop and Faber 1994)], Vangl2 protein is present throughout the plasma membrane and cytoplasm as discrete puncta, without any obvious polarity (Figure 2A,C,E). In contrast, by NF39, when endoderm cells are beginning to elongate along the radial axes of the gut tube, Vangl2 is detected primarily in the membrane, where it is enriched at both the apical tips and anterior surfaces of the spindle-shaped cells (Figure 2B,D,F). Temporally, the asymmetrical localization of Vangl2 corresponds with the acquisition of cell polarity, as indicated by microtubule (MT) bundles aligning along the guts’ radial axes (compare Figure 2G and H).

Figure 2. Vangl2 localization becomes apically and anteriorly polarized during Xenopus gut morphogenesis.

Low levels of Vangl2 mRNA were targeted to the gut endoderm to generate mosaic overexpression (see Experimental Procedures). Injected embryos (NF35 and NF39) were then sectioned frontally (as indicated by the dashed line in the cartoons) and immunostained to detect integrin (Int, green cell outlines, A-F), Vangl2 (red, A-F), β-catenin (βcat; red, G-H) and/or alpha-tubulin (αTub; green, G-H). The boxed region in A is magnified in C and E, while the boxed region in B is magnified in D and F. At NF35, Vangl2 (red) is localized in puncta throughout endoderm cell membranes (A,C,E); the disorganized (non-parallel) microtubule architecture at this stage (G) indicates endoderm cells are still largely unpolarized. However, by NF39 (B,D,F), when cells are elongating along the radial axis of the gut tube, and the parallel alignment of microtubule bundles reveals the cells’ radial polarization (arrows, H), Vangl2 has become localized predominately at the apical (left) end and anterior (top) face of each cell (arrows, F). Scale bars in A-B = 75 μM; C-H = 25 μM.

Vangl2 is required for proper gut elongation

The intriguing apico-anterior polarization of Vangl2 in gut endoderm cells suggests it may influence gut morphogenesis. To test the requirement for Vangl2 in this process, we first employed CRISPR-Cas9 genome editing to generate loss of function. While many embryos injected with Vangl2 gRNA + Cas9 mRNA did not survive to gut morphogenesis stages, presumably due to an early requirement for Vangl2 during gastrulation, on average, 23.9% (n=149) of surviving F0 embryos developed severely shortened or malrotated gut tubes (Figure 3B). In contrast, none of the control embryos, i.e., injected with Cas9 (n=14) or Vangl2 gRNA (n=12) alone (Figure 3A), exhibited such phenotypes. Importantly, the gut phenotypes correlated with mutations at the Vangl2 locus (42%, n=12; Figure 3C).

Figure 3. Endoderm-specific knockdown of Vangl2 activity causes gut morphogenesis defects.

While the gut has substantially elongated in control embryos injected with vangl2 gRNA alone (A; NF44), embryos injected with vangl2 gRNA plus Cas9 mRNA (B; NF44) are severely shortened and/or malrotated, correlating with indels at the vangl2 locus on both L (Δ4, Δ3, Δ11, Δ5) and S chromosomes (Δ8). Likewise, while embryos injected with control morpholino (CoMO; D; NF46) develop long, coiled gut tubes, embryos injected with Vangl2 morpholino (Vangl2 MO) have abnormally short and malrotated guts (E). F) Western blotting for Xenopus Vangl2 protein confirms that Vangl2 is depleted in Vangl2 MO injected embryos, compared to CoMO injected controls (NF41); GAPDH was detected as a loading control. G-H) Co-injection of Vangl2 MO with a morpholino-resistant WT vangl2 mRNA partially rescues the short gut phenotype (compare to E). H) Annotations (a,b,c) indicate injected groups significantly different from each other p<0.05 (error bars indicate standard deviation).

To confirm this short gut phenotype reflected a specific requirement for Vangl2 in gut morphogenesis, we next injected a translation-blocking Vangl2 morpholino (Vangl2 MO) into the blastomeres of the early embryo that are fated to give rise primarily to the midgut endoderm (Reed et al., 2009). Targeting Vangl2 knockdown specifically to the developing gut tube elicited gut phenotypes identical to those observed with loss of Vangl2 activity by CRISPR-Cas9, while avoiding gastrulation defects. Whereas injection of a control morpholino (CoMO) rarely affected gut morphogenesis (11%, n=143; Figure 3D), injection of Vangl2 MO resulted in severe inhibition of gut elongation in 89% (n=160; p<0.01; Figure 3E) of injected embryos, accompanied by knockdown of Vangl2 protein (Western blot; Figure 3F).

The specificity of the phenotype was confirmed by co-injecting Vangl2 MO with a morpholino-resistant vangl2 mRNA (see Experimental Procedures). While most guts injected with Vangl2 MO were abnormal, fewer guts co-injected with Vangl2 MO and vangl2 mRNA exhibited abnormal morphology (65%, n=37; Figure 3G–H), although this result was only marginally significant (p=0.051). Attempts to increase the frequency of rescue by injecting higher amounts of vangl2 mRNA elicited a confounding GOF phenotype, as might be expected for a PCP molecule (not shown, but see Figure 7 below). Overall, the identical gut morphogenesis phenotypes obtained with two different loss of function strategies confirmed that Vangl2 activity is required for gut elongation.

Figure 7. Ectopic Vangl2 expression disrupts tissue-level endoderm orientation during early gut morphogenesis.

Embryos were injected with GFP mRNA alone (A, H-I, L-O) or GFP plus vangl2 mRNA (B, D-G, J-K, P-S), and allowed to develop to NF35 (D-G) or NF41 (H-S). By NF41, the gut tubes of embryos injected with vangl2 mRNA are short and wide, with a bulging topology (B) compared to controls (A). Frontal sections (as exemplified by cartoon diagram, C) were immunostained for the indicated proteins at either NF35 (D-G) or NF41 (H-S). The boxed regions in F, H, J, M and Q are magnified in G, I, K,N-O and R-S, respectively. At NF35, vangl2 mRNA injected endoderm cells, indicated by GFP expression (green in D), exhibit ectopic Vangl2 localization at the membrane (E-G), compared to neighboring un-injected cells. By NF41-42, GFP mRNA injected cells (red in H and L) are uniformly radially oriented, with apically enriched E-cadherin (I), and parallel arrays of microtubules (arrows, O). In contrast, vangl2 mRNA injected cells (red in J and P) are not aligned with respect to the gut axes, exhibit variably localized E-cadherin (K) and β-catenin (R), and form unusual tissue-level configurations, including rosettes (marked by arrowheads in Q, asterisks in R; arrows in S indicate abnormal microtubule orientations). Scale bars in D, E, F, H, I, L, M, P and Q = 75 μM; G, I, K, N, O, R and S = 25 μM.

Vangl2 is required for proper polarity and adhesion of endodermal cells

To determine the effects of Vangl2 knockdown at the cellular level, we first examined Vangl2 morphants at stages just prior to intercalation (NF 39/41). Vangl2 deficient guts were noticeably straighter than controls, in some cases with loose cells escaping from the hindgut (Figure 4A–B). In contrast to CoMO-injected guts, in which the endoderm cells are spindle-shaped and elongated along the gut’s radial axes (Figure 4D, F), with an average length to width (L:W) ratio of 4.7 (Figure 4H), Vangl2 MO injected endoderm cells remain relatively amorphous (Figure 4I, K), with a significantly reduced L:W of 1.6 (p<0.01; Figure 4H). Moreover, while CoMO-injected cells have parallel arrays of MTs (Figure 4E, G), the majority of which are co-aligned within 0-20° of the gut’s radial (i.e., future apicobasal) axes (Figure 4M), Vangl2 MO injected cells exhibit absent or disorganized MT arrays (Figure 4J, L) that are randomly distributed with respect to the gut radius (Figure 4M).

Figure 4. Vangl2 is required early in gut morphogenesis for endoderm cell shape, adhesion and MT organization.

Control morpholino (CoMO; A, D-G, N-O) or Vangl2 morpholino (Vangl2 MO; B, I-L, P-Q) was co-injected with mRNA encoding GFP (green, D,I), targeting the gut endoderm. Compared to control guts (NF41-42; A), Vangl2 MO injected guts are shorter, straighter and less cohesive, occasionally with loose cells escaping from the gut tube (arrowhead in B). Transverse sections through injected guts (as indicated by the dashed line in the cartoon, C) were immunostained to reveal cell outlines (red, β-catenin, βcat; D-F, I-K) and microtubules (green, αTub; E,G,J,L). The boxed region in E is magnified in F-G, while the boxed region in J is magnified in K-L. Vangl2 MO injected cells exhibit rounder cell shapes, as indicated by decreased length to width (L:W) ratios of individual cells (displayed as box and whiskers plot, H). In addition, βcat levels in Vangl2 MO injected cells (K) are decreased compared to control cells (F). Moreover, whereas microtubules (MTs) in CoMO-injected cells are oriented parallel to the radial axis of the gut tube (arrows, G), they appear randomly oriented in Vangl2 MO injected cells (arrows, L). M) Most MTs in CoMO injected cells are distributed within 0-20° angle of the apico-basal axis of the epithelium; in contrast, the MTs in Vangl2 MO injected cells are distributed more broadly, deviating from parallel. Endoderm cells were dissociated from CoMO (N-O) or Vangl2 MO (P-Q) injected gut tubes, then challenged to re-aggregate for 30’ (see Experimental Procedures). While CoMO injected cells are able to condense into large clumps of cells that reform adherens junctions (O), Vangl2 MO injected cells remain almost completely dissociated (Q). R) The extent of endoderm adhesion is quantified as the percent reduction of the area of the culture dish still covered by loose cells after re-aggregation challenge; error bars represent standard error. *, p<0.05; **, p<0.01. Scale bars in D, E, I and J = 75 μM; F, G, K and L = 25 μM.

In addition to affecting cell polarity and MT architecture, Vangl2 also appears to be required to maintain cell-cell adhesion. For example, whereas CoMO-injected endoderm cells exhibit levels of membrane-localized β-catenin (a key component of adherens junctions) comparable to adjacent un-injected cells, Vangl2 MO injected cells display decreased or absent β-catenin (compare Figure 4F to 4K). Indeed, in an ex vivo cell dissociation/re-aggregation assay designed to evaluate calcium-dependent (i.e., adherens-junction mediated) intercellular adhesion (see Experimental Procedures; Pickett et al. 2017), we detected significantly less re-aggregation (p<0.05; Figure 4R) of cells derived from Vangl2 MO injected gut tubes than from CoMO-injected guts (compare Figure 4N–O to 4P–Q), suggesting morphant cells have fewer and/or less stable adherens junctions. Despite their decreased adhesion, Vangl2-deficient cells remain viable, as we do not detect apoptotic markers in the Vangl2 MO injected population (not shown).

To determine the effects of Vangl2 knockdown on lumen formation and digestive epithelial morphogenesis, we then examined Vangl2 morphants at NF46, when endodermal cells have normally completed intercalation and formed a single layer epithelium comprised of columnar cells (L:W = 5.1), as seen in CoMO injected control embryos (Figure 5A, C, E). In Vangl2 morphants, the injected cells are unable to properly rearrange or undergo epithelial morphogenesis. Instead, Vangl2 morphant cells fill the lumen (asterisks, Figure 5B) and possess more rounded shapes (L:W = 1.4; p<0.01) and abnormal apicobasal polarity, as indicated by decreased apical localization of aPKC (Figure 5D, F). Moreover, while CoMO injected embryos display apically-nucleated MT bundles (Figure 6C, E), and well-established epithelial architecture, as indicated by robust E-cadherin expression (Figure 6 G, I, K, M), Vangl2 MO injected cells exhibit decreased MT polarization and bundling (Figure 6D, F) and loss of cell-cell adhesion, as suggested by severely diminished levels of E-cadherin (Figure 6H, J, L, N).

Figure 5. Vangl2 is required for gut lumen formation and apicobasal polarity.

Embryos were injected with CoMO (A,C,E) or Vangl2 MO (B,D,F), plus mRNA encoding mCherry (mCh, red in A-B; as a lineage tracer), targeting the gut endoderm. Frontal sections (as indicated by the dashed line in the cartoon diagram, NF46) were immunostained for apical (Par3, green in A-B; aPKC, red in C-F) and basolateral (Integrin, Int, green in C-F) proteins to reveal cell polarity. Boxed regions in C and D are magnified in E and F, respectively. Unlike controls (A,C), Vangl2 MO injected guts do not form a central lumen; the unintercalated cells (asterisks in B; boxed region in D) exhibit abnormally rounded shapes and decreased apical markers (e.g., aPKC; F), compared to controls (E). Nuclei, TO-PRO-3 (blue). Scale bars in A-D = 75 μM; E-F 25 μM.

Figure 6. Vangl2 is required during late gut morphogenesis for epithelial polarity and microtubule architecture.

Embryos were injected with control morpholino (CoMO; A,C,E,G,I,K,M) or Vangl2 MO (B,D,F,H,J,L,N), plus mRNA encoding mCherry (mCh, red in A-B; as a lineage tracer), targeting the gut endoderm. Frontal sections (NF46) were immunostained for the apical marker MHCB (green in A-B), α-Tubulin [αTub, green in C-F, to reveal microtubule (MT) architecture] and Ecadherin (Ecad, green in G-N, to outline cell surfaces and reveal adherens junctions). Boxed regions in C, D, G, H, K and L are magnified in E, F, I, J, M and N, respectively. Unlike controls (E), Vangl2 MO injected cells are unable to form properly oriented MT arrays (F). Likewise, intercellular adhesion is reduced dramatically in Vangl2 morphants (compare Ecad staining in I to J). Immunostaining for the mitotic marker phosphohistone H3 (pHH3, red in G-J) suggests that the abnormal epithelial morphogenesis observed in Vangl2-deficient guts is independent of early cell inviability. However, immunostaining for activated caspase (Casp, red in K-N) shows that un-intercalated cells in the Vangl2-deficient gut lumen eventually die by apoptosis (L,N), an event rarely observed in controls (K,M). Nuclei, TO-PRO-3 (blue). Scale bars in A-D, G-H, K-L= 75 μM; E-F, I-J, M-N = 25 μM.

We observed phospho-histone H3 (pHH3) positive nuclei in both un-injected (Figure 6 G, I) and Vangl2 MO injected tissues (Figure 6H, J), indicating most Vangl2-deficient cells remain viable and proliferative throughout gut morphogenesis. However, some un-intercalated Vangl2-deficient cells, i.e., those that did not become incorporated into epithelia, do undergo cell death by NF46 (Figure 6 L, N), perhaps as a result of the decreased cell-cell adhesion. This outcome is rarely observed in CoMO injected tissue (Figure 6 K, M), and suggests that the Vangl2 morphant phenotype may be partly attributable to a decreased number of endoderm cells participating in morphogenesis. Taken together, these results suggest that Vangl2 is required to establish or maintain cell properties necessary to execute proper epithelial morphogenesis and gut elongation, including polarization, MT architecture and cell-cell adhesion.

Vangl2 gain of function causes a loss of coordinated cell polarity and adhesion

To further characterize the role of Vangl2 in gut morphogenesis, we conducted over-expression studies. Synthetic Xenopus vangl2 mRNA was targeted to the gut endoderm by microinjection as above, along with mRNA encoding GFP or mCherry (mCh; Figure 7) as lineage tracers. Immunostaining confirmed that this procedure results in upregulation of membrane-localized Vangl2 in injected cells (see Figure 7D–G).

By the onset of radial intercalation (NF 39/41), guts with ectopic Vangl2 activity appear shorter and wider than controls (compare Figure 7A and B). This abnormal gross morphology also reflects a severely disrupted internal endoderm organization. For example, whereas guts injected with GFP mRNA alone are comprised largely of radially aligned, spindle-shaped cells with apically enriched E-cadherin (Figure 7H–I), vangl2 mRNA injected guts contain cells with atypical shapes and orientations, and variable E-cadherin distribution (Figure 7J–K). In contrast to control cells (Figure 7L–O), Vangl2 overexpressing cells form unusual configurations, such as rosette arrangements (Figure 7P–S), suggestive of variable polarity and inappropriate adhesive contacts. Moreover, while GFP mRNA (only) injected control cells have uniformly distributed β-catenin at the cell surface (Figure 7N), and radially oriented MTs (Figure 7O), cells co-injected with vangl2 mRNA have varying levels of β-catenin (Figure 7R), e.g., concentrated at the center of rosette clusters, often associated with mis-oriented MT bundles (Figure 7S).

In this context, endoderm cells are not properly oriented for radial intercalation or CE, as required for proper gut lumen formation, epithelial morphogenesis and tube elongation. To evaluate the effects of Vangl2 overexpression on these later features of gut morphogenesis, we also analyzed vangl2 mRNA injected embryos at end stage (NF46). In contrast to control embryos injected with mCh mRNA alone (Figure 8A), which rarely exhibit defective gut morphologies (8%, n=169), the majority (58%, n=155) of vangl2 mRNA co-injected embryos have variably thickened and shortened guts, often with irregular topological features, e.g., indentations and/or bulges (arrowheads, Figure 8E, I).

Figure 8. Ectopic Vangl2 expression disrupts tissue-level gut epithelial organization.

Embryos were injected with mCherry (mCh) mRNA alone (A-C) or mCherry plus vangl2 mRNA (two examples in E-G and I-K, respectively) to target the prospective gut tube, and allowed to develop to NF46. Compared to the long, rotated guts of control embryos injected with mCh mRNA alone (A), the gut tubes of embryos injected with vangl2 mRNA (E, I) are shorter and wider than controls, and often have unusual bulges and/or indentations (arrowheads in E, I). To analyze tissue architecture in injected embryos, transverse sections (e.g., section plane approximated by horizontal line in A) were immunostained for various proteins, as indicated: β-catenin (βcat; green) to reveal cell shape/adhesion, aPKC (red) or MHC (green) to reveal the apical/lumenal cell surface, and Integrin (Int, green), which is enriched at basement membranes. Serial sections from the same embryo are shown sequentially in F-F” and J-J”, with section planes approximated by the horizontal lines in E and I, respectively. B) In controls, the segments of the gut tube are comprised of a central lumen surrounded by a single layer of apicobasally polarized epithelium, as summarized in the cartoon [D; blue arrows represent the orientation of cell polarity with respect to basement membrane (green) and apical surface (red)]. C) Higher magnification views (from serial sections) of the boxed region in B reveal uniform basolateral distribution of β-catenin, apically localized MHC/aPKC, and parallel arrays of apically nucleated microtubules (as indicated by αTub). F-F”) In contrast to controls, the segments of a vangl2 mRNA injected gut appear thickened, with walls composed of multiple layers of disoriented epithelial tissue forming a torturous, branching and/or non-contiguous lumen, as exemplified in the cartoon (H). G) Higher magnification views (from serial sections) of the boxed region in F’ reveal that vangl2 mRNA overexpressing cells retain β-catenin expression, although microtubule bundles are often short and/ or obliquely oriented, consistent with the complex, multilayered epithelial organization created by the non-contiguous apical/lumenal surface delineated by MHC/aPKC. J-J”) In another vangl2 mRNA injected gut, basement membrane, as indicated by Int expression, is detected internally, as represented in the cartoon (L). K) Higher magnification views (from serial sections) of the boxed region in J show both the discontinuous apical/lumenal surface (indicated by MHC/aPKC expression), and presumed basement membrane (Int) in the center of the gut tube, revealing the perturbed polarity of epithelial organization in the context of Vangl2 overexpression. Nuclei, TO-PRO-3 (blue). Scale bars in B, F-F” and J-J” = 75 μM; C, G and K = 25 μM.

As observed at earlier stages, the internal architecture of later stage vangl2 mRNA injected guts is strikingly abnormal. Whereas the endoderm cells in mCh mRNA (only) injected control guts have completed intercalation and formed an apicobasally-polarized epithelium surrounding a single, central lumen (Figure 8B–D), much of the epithelium of vangl2 mRNA co-injected guts (Figure 8F–H and J–L) remains highly disorganized. Interestingly, although many Vangl2 overexpressing cells have not properly intercalated, they have nonetheless acquired apicobasal polarity, albeit in a manner uncoordinated with respect to their neighbors, resulting in the formation of multiple layers of differentiated tissue with aberrant (and often opposing) apicobasal orientation, and disconnected, branching and/or convoluted lumens (compare Figures 8F–H to Figures 8B–D). In some cases, basement membrane—normally only generated around the circumference of the gut tube, between the endoderm and the surrounding mesoderm layer (indicated by intense Integrin expression)—can be detected deep in the center of the gut tube, suggesting internal epithelial layers became oriented with “inside out” apicobasal polarity (Figures 8J–L). Moreover, while Vangl2 overexpressing cells in contact with the outer basement membrane (i.e., near the gut circumference), may acquire normal apicobasal polarity, the innermost Vangl2 overexpressing cells are often obliquely or orthogonally-oriented with respect to the gut radii, e.g., with their long axis parallel to the anterior-posterior axis of the tube (Figures 8G, K).

Consistent with these severe disruptions of polarity, MT orientation in vangl2 mRNA injected cells is also abnormal, corresponding with the cells’ stochastic apicobasal polarity (compare alpha-tubulin, αTub, expression in Figure 8G,K to 8C). However, cell-cell adhesion, as indicated by the distribution of β-catenin (Figure 8C,G,K) and E-cadherin (not shown), appears largely unaffected by this stage of epithelial differentiation. These results suggest that endoderm cells ectopically expressing vangl2 mRNA are no longer properly oriented with respect to the radial and/or AP axes of the gut tube; they consequently fail to intercalate or differentiate in a tissue-level coordinated manner.

Discussion

During morphogenesis of the primitive gut tube, complex cell rearrangements simultaneously open the central lumen of the future digestive tract, generate its epithelial lining, and narrow and elongate the entire tube. To properly execute this multifaceted process, endoderm cells must be polarized and oriented with respect to both the radial/apicobasal and anterior-posterior axes of the gut tube. Our results suggest that this multidimensional orientation is achieved by asymmetrical localization of Vangl2 at both apical and anterior edges of individual endoderm cells—dual polarities that coordinate epithelial morphogenesis with convergent extension.

Vangl2 is required for required for endoderm rearrangements during gut morphogenesis

We observed enrichment of Vangl2 at the apical ends of individual endoderm cells as they elongate along the gut’s radii, i.e., corresponding to the apicobasal axis of the future digestive epithelium. Interestingly, Vangl2 has been implicated in controlling apicobasal polarity and radial intercalation in other contexts. For example, Prickle1 synergizes with Vangl2 in establishing or maintaining the apicobasal polarity of epiblast cells of the early mouse embryo (Tao et al. 2009). Likewise, cells of both the Xenopus neuroectoderm and skin have an apical enrichment of Vangl2 required to undergo radial intercalation to a more superficial position within their respective tissue; Vangl2 is found at the tips of these intercalating cells to drive their invasion into the overlying cell layer (Davey and Moens 2017;Ossipova et al. 2015a;Ossipova et al. 2015b).

We found that Vangl2 expression is also enhanced at the anterior faces of individual endoderm cells. This asymmetry is reminiscent of the anterior localization of Vangl2 during tissue-elongating cell rearrangements like CE (Davey and Moens 2017). Just as CE is driven by mediolateral cell intercalations achieving preferential separation of anterior and posterior neighbors, the narrowing and elongation of the gut tube is thought to be driven by endoderm cells preferentially separating anterior and posterior neighbors during radial intercalation (Reed et al. 2009). To date, the mechanisms by which this polarity might be globally coordinated throughout the gut tube have remained elusive.

The results presented herein suggest that the apical and anterior localization of Vangl2 is critical for orienting and coordinating the rearrangement of endoderm cells during gut morphogenesis. We found that inhibition of Vangl2 expression prevents proper polarization of these cells, such that they are devoid of radial or anterior-posterior positional information. Consequently, the cells remain unpolarized and, therefore, unable to intercalate. Instead of forming a single-layer epithelial lining, the poorly cohesive, rounded cells fill the lumen, resulting in an abnormally stratified gut lining, failed lumen formation, and an abnormally shortened gut tube.

In contrast, cells with ectopic Vangl2 expression are able to acquire apico-basal polarity; however, they do so in a highly uncoordinated manner. Vangl2 overexpressing cells are no longer aligned perpendicular to the anterior-posterior axis of the gut tube or along the gut’s radial axes. Instead, they orient stochastically, independent of radial or anterior-posterior positional information, forming abnormal multicellular configurations (e.g., rosettes) inconsistent with productive cell rearrangement in the gut tube. For example, the orthogonally oriented, often opposing, apicobasal polarity of neighboring cells inhibits the thinning of the epithelium into a contiguous layer surrounding a single central lumen. Likewise, the randomly oriented polarity also impedes the coordinated CE rearrangements that narrow and lengthen the larger tube. Consequently, unregulated Vangl2 expression results in variable gut tube topology, with surface indentations and bulges reflecting the highly disorganized epithelial mass inside.

Overall, this study suggests that apical and anterior asymmetry of Vangl2 expression in the gut endoderm cells is a key factor coordinating lumen formation and epithelial morphogenesis with the narrowing and elongation of the entire tube.

Vangl2 regulates endoderm intercellular adhesion and microtubule architecture.

Dynamic regulation of intercellular adhesive contacts is a critical component of all multicellular morphogenesis (Pinheiro and Bellaiche 2018). In the developing gut tube, intercellular adhesion must be properly modulated to allow endoderm cells to move relative to one another while maintaining sufficient contact to preserve their polarized shape and overall tissue cohesivity. We previously showed that this dynamic is carefully balanced by downstream mediators of Wnt/PCP signaling (Reed et al. 2009;Dush and Nascone-Yoder 2013); the current results suggest that the core PCP molecule Vangl2 plays a critical role in orienting and coordinating this process on the tissue level.

The reduced cell-cell adhesion seen in Vangl2 morphants, and inconsistent/inappropriate adhesion seen at early stages in guts overexpressing Vangl2, suggests Vangl2 activity modulates adhesive contacts during endoderm intercalation. In other contexts, Vangl2 has been shown to directly affect the endocytosis of cadherins (Nagaoka et al. 2014a;Lindqvist et al. 2010;Nagaoka et al. 2014b), suggesting that Vangl2 could regulate polarized cell adhesion in the gut endoderm by direct physical interaction with adherens junction components. As we also observed reduced MT bundling in cells deficient in Vangl2, and our previous results showed that depleting polymerized MTs disrupts endoderm cell-cell adhesion (Dush and Nascone-Yoder 2013), the effect of Vangl2 on endoderm cell adhesion may also be an indirect consequence of effects on microtubule architecture. Detailed studies of the cellular properties and behaviors controlled by Vangl2 in the gut endoderm (e.g., co-localization and interaction of Vangl2 with adherens junction components or microtubules, and/or functional studies of the effects of Vangl2 and other core PCP components on cadherin localization and/or endocytosis) could shed light on PCP-driven cell rearrangement dynamics in the gut and other organs.

Experimental Procedures

Embryos

Xenopus laevis embryos were obtained by in vitro fertilization, de-jellied with 2% cysteine-HCl pH 7.9, sorted to eliminate individuals with developmental anomalies and cultured in 0.1× Marc’s Modified Ringers (MMR) at 16° or 22°C (Sive et al. 1998). Staging was according to (Nieuwkoop and Faber 1994). Tadpoles at stages NF35-46 were anesthetized in 0.05% MS222 in 0.1× MMR for morphological analysis, dissections and/or fixation.

CRISPR

All microinjections were performed at the one-cell stage, using established methods (Sive et al. 1998). Cas9 mRNA and guide RNA (gRNA) were synthesized as described (Guo et al. 2014). The vangl2 target site used was CCCGGGACAAGAACTACC, within exon 6, in the region between transmembrane domain 3 and 4; the gRNA used in this study targets both the L and S chromosomes. Primers used to PCR amplify the vangl2 genomic locus targeted by the gRNA were:

fwd 5’- GGCCTTCAAGCTCCTCATACTTCTAC - 3’

rev 5’- GAGCACGGAAGAGCTTTGAAGTGGTG - 3’.

PCR products were subcloned into pCRII vector (ThermoFisher Scientific) and individual clones were sequenced with M13R primer (5’-CAGGAAACAGCTATGAC-3’) to determine mutation frequency. Sequencing was performed at Eton Bioscience (Research Triangle Park, NC, USA).

Morpholino Knockdown and mRNA overexpression

All reagents were injected into ventrovegetal blastomeres of 8-16-cell stage embryos as described (Reed et al. 2009), along with lineage-tracer mRNA encoding membrane-associated variants of GFP or mCherry. For loss of function, a previously validated morpholino oligonucleotide (5.6 ng/injection) that targets the translation start site of X. laevis Vangl2 [5’- ACTGGGAATCGTTGTCCATGTTTTC-3’ (Mitchell et al. 2009)] and a standard control sequence (Gene Tools) were employed.

For ectopic expression of wild type Vangl2, the mMessage Machine kit (Ambion) was used to synthesize mRNA encoding full length X. laevis Vangl2 from a pCS2 construct, which was then purified using NucAway columns (Ambion; as previously described, Dush and Nascone-Yoder 2013). Approximately 560 pg of vangl2 mRNA was injected into 1 ventrovegetal blastomere of 8-16-cell embryos.

To assess morpholino specificity, embryos were co-injected with the Vangl2 morpholino (5.6 ng) plus a morpholino-resistant vangl2 mRNA (60-174pg) in which the morpholino binding site was mutated by PCR, using the following oligo to introduce mutations:

5’-ggcctctcgagcctGCCACCATGGATAACGACTCCCAATATTCGGGCTACTCCTATAAG-3’

Endoderm adhesion assays

Stage NF41 gut tubes were dissected from both CoMO and Vangl2 MO injected embryos and the foregut and pancreas removed with an insect pin. The remaining midgut region was filleted open, then transferred to an agarose coated 6-well dish containing 9 ml of calcium- and magnesium-free medium (CMFM; Sive et al. 2007). Individual endoderm cells were allowed to dissociate over a period of thirty minutes, after which additional endodermal cells were removed from the disintegrating gut tube by gentle dissociation with an eyelash hair dissecting tool. The CMFM was removed and replaced with Modified Barth’s Saline (Sive et al. 2007) and the endoderm cells were allowed to re-aggregate over a period of thirty minutes.

Immunohistochemistry

Whole embryos were fixed for 45 minutes in Dent’s fixative (80% methanol/20% DMSO), or in 4% PFA made up in HEPES buffered saline (Fagotto and Gumbiner 1994), followed by 8 washes in Dent’s fixative. After overnight storage at −20°C, embryos were rinsed and incubated for 1 hour in Tris-NaCl (100 mM Tris HCl pH 7.3, 100 mM NaCl) before transfer to sucrose/gelatin (15% sucrose /25% cold water fish gelatin) overnight. Embryos were embedded in OCT (Tissue-Tek), sectioned at 10 μm, picked up on coated slides (Fisher Superfrost), and allowed to air dry overnight. Sections were post-fixed in acetone for 1 minute, rinsed in PBS, then antigen retrieval was performed by incubating the slides in 1% SDS for 3 minutes. Following three washes in PBS, the slides were blocked for 30 minutes as described (Reed et al. 2009).

Immunohistochemical staining was achieved by overnight incubation at 4°C in blocking buffer containing combinations of the following primary antibodies: Vangl2 (Sigma, HPA 027043; 1:200), E-cadherin (DSHB, 5D3; 1:200), β-catenin (SCBT, H-102; 1:100), aPKC (SCBT, sc216; 1:200-1:250), α tubulin (Sigma, T9026; 1:1000), cleaved Caspase-3 (Abcam, ab13847; 1:200), mCherry (Clontech, 632543; 1:1000), GFP (Invitrogen, A6455; 1:1000), and phospho-Histone H3 (Ser10; Upstate, 06-570; 1:1000). Slides were then washed twice in PBT for 5 minutes, incubated for 3 hours in blocking buffer containing Alexa 488-conjugated goat anti-mouse IgG (Invitrogen, A11029; 1:2000) and/or Alexa 555-conjugated goat anti-rabbit IgG (Invitrogen, A11035; 1:2000), and washed in PBT. Slides were washed in PBS, mounted in Prolong Gold, cured overnight in the dark, and sealed with nail polish. Fluorescence was visualized on a Leica SPEII confocal microscope.

Morphometric measurements and statistics

For morpholino studies, individual embryos were injected in each control versus experimental group (n=18-102), and gut morphology was scored at end stage (NF45-46). The experiment was repeated three times with different clutches of embryos with similar results, and the data were then used to calculate the average percent of injected embryos with abnormal gut morphology from the three experiments (total n=143-160 per condition). The average length to width (L:W) ratio of individual endoderm cells was calculated from measurements (Photoshop) of the maximum length and width of representative injected cells (n=10) in both control and morphant gut tubes (NF41, 46), from three different experiments (total n=30 per condition). To assess proliferation, the numbers of phospho-Histone H3 (pHH3) positive cells (n=69-81 per gut) were counted (Image J) in both control and morphant gut tubes (NF46) from three different experiments. The extent of endoderm re-aggregation was quantified as the percent reduction of the area of the culture dish covered by control versus morphant cells (Image J) after thirty minutes of allowing them to condense into aggregates (loose cells cover more area than aggregated clumps). The experiment was performed twice (i.e., using injected embryos from different clutches) with similar results, and the data were used to calculate the average percent reduction from both experiments. In all cases, significant differences between control and experimental values were determined by one way ANOVA.

Differences in MT architecture were quantified by measuring the angle of deviation of individually discernable MT bundles (n= 48-72) with respect to the gut radial (apical-basal) axes in both control and morphant gut tubes (NF41). The data were plotted in a histogram to show the frequency distribution in control versus experimental groups (i.e., the percent of total MTs falling within specific 10° intervals from the apicobasal axis).