(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



The epithelial polarity genes frazzled and GUK-holder adjust morphogen gradients to coordinate changes in cell position with cell fate specification [1]

['Yongqiang Xue', 'Department Of Biology', 'Case Western Reserve University', 'Cleveland', 'Ohio', 'United States Of America', 'Aravindan Krishnan', 'Juan Sebastian Chahda', 'Robert Allen Schweickart', 'Rui Sousa-Neves']

Date: 2023-03 

Morphogenetic gradients specify distinct cell populations within tissues. Originally, morphogens were conceived as substances that act on a static field of cells, yet cells usually move during development. Thus, the way cell fates are defined in moving cells remains a significant and largely unsolved problem. Here, we investigated this issue using spatial referencing of cells and 3D spatial statistics in the Drosophila blastoderm to reveal how cell density responds to morphogenetic activity. We show that the morphogen decapentaplegic (DPP) attracts cells towards its peak levels in the dorsal midline, whereas dorsal (DL) stalls them ventrally. We identified frazzled and GUK-holder as the downstream effectors regulated by these morphogens that constrict cells and provide the mechanical force necessary to draw cells dorsally. Surprisingly, GUKH and FRA modulate the DL and DPP gradient levels and this regulation creates a very precise mechanism of coordinating cell movement and fate specification.

Funding: This work was supported by National Institutes of Health grant numbers 1R21EB016535 and 5R33AG049863 (to CMM and RSN) and 5R01AG061390 (to RSN). YX, AK, JSC and RAS were supported by the College of Arts and Sciences of Case Western Reserve University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The discovery of global stereotyped cell movements in the blastoderm opens the possibility of investigating the mechanisms by which morphogens coordinate cell fate specification relative to cell position. This exquisitely simple, bidimensional embryonic patterning allows for referencing each cell in space and in this way asks how morphogens control the stereotyped cell trajectories. In addition, this model allows for testing possible molecular effectors that coordinate this process. To address these issues, we focused on the stereotyped cell movements along the D/V axis, where the mesoderm, neuroectoderm, and ectoderm are formed. The formation of these domains depends on a ventral-to-dorsal nuclear gradient level of the transcription factor DL/NFκ-B and an opposing dorsal-to-ventral gradient of secreted decapentaplegic/bone morphogenetic 4 (DPP/BMP-4) [ 13 – 19 ]. DL activates mesodermal genes such as snail (sna) and neuroectodermal genes, such as short gastrulation (sog) and intermediate nervous system defective (ind), and represses ectodermal genes, such as dpp. The DL gradient is established by a maternal signaling pathway that regulates the graded activation of the Toll receptor by Spätzle ligand along the DV axis leading to the selective transport of DL into the nucleus. DPP activates ectodermal genes, such as rhomboid (rho) and race and represses neuroectodermal genes, such as muscle segment homeotic (msh) and ind. The DPP gradient is established zygotically and depends on the interaction of various extracellular modulators. Here, we show that DPP, and not DL, directly attracts cells to the dorsal region and that DL stalls cells ventrally by excluding DPP expression. Through an in silico screen, we identified GUK-holder (gukh) and frazzled (fra) as the effector genes that respond to DPP and/or DL and control the stereotyped cell movements. We show that both genes are required for the correct formation of cell density patterns and the specification of the mesodermal, neuroectodermal, and ectodermal expression domains. Finally, we show that fra and gukh modulate the shapes of the gradients of DL and DPP. Together, our results provide evidence that DPP instructs cells about their fate in a dosage-dependent manner and adjusts these thresholds by positioning cells in space.

(A, B) Cell density heatmaps of dorsal and ventral sides of wild-type embryos at early (A) and late (B) cellularization stages. “Average heatmaps” combine data of 9 embryos for all genotypes, except for gd 7 ;dpp- (n = 5). “Representative heatmap” shows a single embryo with A/P marker ftz (black dots). Note the higher dorsal cell density in early stage that increases in late stage. (C) gd 7 mutants with no DL gradient loose early asymmetry of cell densities. (D) Dorsal and ventral cell count differences of embryos shown in (A–C). (E) anti-E-CAD membrane labeling of dorsal cells in wild type during cellularization. (F) Quantification of cell constriction over time. Dorsal cells show significant decrease in segmented surface area from early to mid- to late stages, while ventral cells of same embryos do not. (G, H) DPP increases cell density in the absence of DL. (G) Average cell density heatmaps in late stage gd 7 embryos without DL gradient (top left). The high-density levels in ventral and dorsal surfaces is abolished when DPP is also removed (gd 7 ;dpp- embryos, top center) and restored by the ectopic expression of dpp (gd 7 ;dpp- st2-dpp embryos, top right; dots indicate margins of rho domain). Changes in the expression of DPP-target rho in these genetic backgrounds (bottom) confirm the manipulations. (H) Dorsal and ventral cell counts in wild type, gd 7 and gd 7 ;dpp- embryos at late cellularization stage. Error bars, standard deviation. N.S., not significant. Asterisks, threshold values based on p-values calculated with two-tail Mann–Whitney test (D, F, H, comparison of different genotypes and stages) or two-tailed Wilcoxon signed-rank test (H, comparison of D/V surfaces of same genotypes and stages). *p < 0.05, ***p < 0.0001. Scale bars, 10 μm (E) and 60 μm (G). Metadata for the graphs shown in D, F, and H can be found at Supporting information S1 Metadata . A/P, antero-posterior; DL, dorsal; DPP, decapentaplegic; D/V, dorso-ventral.

To that end, we analyzed the Drosophila blastoderm, a phase of embryonic development in which the syncytium nuclei behave as compartmentalized open cells [ 5 – 8 ] that move in a single plane on the surface of embryos as they acquire distinct fates. For a long time, cells at this embryonic stage were treated as a static field. However, this view changed with the discovery that not only cells move but they actually follow stereotyped trajectories that ultimately establish well-defined patterns of cell densities along the antero-posterior (A/P) and dorso-ventral (D/V) axes [ 9 – 11 ] ( Fig 1 ). In particular, cells from the lateral sides of the embryo and from the anterior and posterior poles move towards the dorsal midline, whereas cells within the ventral region remain mostly immotile with only slight movements towards the posterior region. Noteworthy, these stereotyped cell movements are disrupted in mutant embryos without either bicoid (BCD) or dorsal/NFκB (DL/NFκ-B) gradients, indicating that they are regulated by morphogens that pattern the A/P and D/V embryonic axes, respectively [ 9 ]. Additional evidence of a genetic control of these movements comes from the fact that different lineages of Drosophilids display species-specific patterns of cell density [ 12 ].

The “French Flag model” of positional information is a central model in developmental biology that explains how cell fates are specified within broad regions of developing embryos, limbs, and other organs. In this model, cells are thought to interpret threshold levels of a morphogen according to their spatial position and acquire distinct fates through the activation of specific gene expression programs [ 1 – 4 ]. Despite its immense power to explain diverse phenomena such as embryonic axial patterning and segregation of different neuronal populations in the nervous system, a major simplification is frequently overlooked in this model. Namely, it is assumed that morphogen thresholds reach and modify a static cell population. However, this condition is seldom if ever met in most developmental contexts, which generally involve a dynamic displacement of cells at the same time they read the instructive thresholds that determine their fates. Here, we sought to understand how tissues coordinate cell movements with cell fate specification during pattern formation.

Results

Cells move towards the dorsal midline during cellularization in a stereotyped fashion We began our study by first analyzing the formation of cell density patterns resulting from cell movements during the blastoderm stage. To quantify these patterns along the D/V axis, we analyzed ventral and dorsal halves of individual embryos precisely oriented in each position according to D/V and A/P markers (rhomboid (rho), snail (sna), intermediate neuroblasts defective (ind), even-skipped (eve), and fushi-tarazu (ftz) (see Methods). We focused on early and late cellularization stages defined by initial and fully extended membrane invagination on the ventral side, respectively. Images obtained from these stages were then segmented to obtain the centroids of each nucleus, and the 3D cell features were imported into geographic information system (GIS) to generate average cell density heatmaps for the dorsal and ventral regions and to analyze the data with spatial statistics. In addition to heatmaps and hot/cold spot maps, we also tested for differences in cell count numbers (see Methods). Our results show that at the cellularization onset, there is already a slightly higher number of dorsal cells than ventral cells (Fig 1A). These differences are established prior to cellularization by the DL gradient as evidenced by the fact that embryos without nuclear DL have equal cell densities in the ventral and dorsal sides at the beginning of cellularization (Fig 1C and 1D). In agreement with previous reports [9,12], by the end of cellularization, the asymmetric pattern of high density of cells within the dorsal region versus low density in the ventral region becomes evident in density heatmaps of late stage embryos (Fig 1B and 1D). Since the total cell number in the embryo remains constant during cellularization [20,21] and apoptosis is absent during this stage [22], the emergence of a dorsal region of high cell density reflects the movement of cells towards the dorsal midline [9]. To visualize these movements, we analyzed time-lapse videos of live embryos expressing the cell membrane protein E-Cadherin/Shotgun-GFP (E-CAD/Shg-GFP) [23] as well as E-CAD in fixed embryos throughout cellularization (Fig 1D). The data obtained with E-CAD-GFP agree with previous analyses using Histone-GFP [9], but since E-CAD-GFP labels cell contours, our analyses rule out the possibility of nuclear movement within cells and reveal that the cell movements involve cell constriction with no evident intercalation between cells (S1 Video). We measured the segmented surface areas of individual dorsal cells expressing E-CAD-GFP in time-lapse images (S2 Video and S1 Fig) and from dorsal and ventral cells of same fixed embryos at 3 distinct stages (Fig 1E and 1F). These results show that the apical surface of dorsal cells becomes constricted and tightened as cells move dorsally, resulting in a significant decrease in cell size from early to mid and to late stage. In contrast, cells located in the ventral side of the embryo do not constrict over time (Fig 1F). Thus, these experiments confirm the stereotyped cell movements from the lateral regions and poles towards the dorsal midline center, resulting in an increase in cell density on the dorsal surface of the embryo compared to the ventral surface.

DL regulates cell movements indirectly through DPP From the data presented above and previous reports, it is clear that cells are attracted to the dorsal region and their density increases in this region over time. To a large extent, this pattern of cell clustering is the mirror image of the DL gradient, which decreases continuously towards the dorsal side. In addition, the asymmetric dorsal clustering of cells was shown to be abrogated in embryos with no nuclear DL [9]. However, if DL regulated this dorsal-bound cell movement directly, then this would imply that it does so by creating cell repulsion. In this case, we should expect repulsion to reach its maximum in the ventral region where the levels of DL peak. Nevertheless, in this region, the cell movement is virtually inexistent [9]. Thus, DL cannot be the morphogen that directly governs these cell movements, but rather it must do so indirectly through repression of another morphogen that attracts cells dorsally. The natural candidate to exert this activity is DPP, which meets the requirements of being repressed ventrally by DL and achieving peak levels in the dorsal midline, the site where cells are attracted to. To understand the relationship between cell clustering and the DL and DPP gradients, we tested the effects caused by the removal of DL and DPP individually and simultaneously. First, we analyzed embryos without the DL gradient using the maternal mutation gastrulation defective (gd), which prevents the processing of Spätzle and the activation of Toll. In those embryos, the DL gradient is not formed because DL is absent from the nuclei. In addition, the DPP gradient is not formed because the relief of DL repression allows DPP to expand across the DV axis as can be seen by the ubiquitous activation of its target rho (Fig 1G). As expected, we observe a high cellular density across the D/V axis within the center region of these embryos (Fig 1B, 1G and 1H) [9]. The anterior and posterior poles have a low density as in the wild type, indicating that the movements controlled by the A/P coordinates are maintained, but these anterior/posterior-most cells move towards the center of the embryo without noticeable dorso-ventral differences (i.e., the directionality to the dorsal midline is lost in gd embryos). Next, to test if this cellular density packing stems from the indirect ubiquitous activation of DPP, we removed DPP from embryos without DL gradient (note loss of rho activation in Fig 1G). In contrast to the loss of DL only, embryos without DL and DPP gradients have a much lower cell density across the D/V axis (Fig 1G and 1H). Thus, DPP is required for cell clustering. To further test the ability of DPP to attract cells, we analyzed embryos without DL and DPP gradients expressing DPP orthogonally by using the even-skipped stripe 2-dpp (st2-dpp) construct [24]. These experiments show that DPP expressed in this position activates ectopically its target gene rho and indeed attracts cells to the center of the embryo across its entire circumference (Fig 1G and 1H). Several other pieces of evidence unambiguously demonstrate that DPP attracts cells. First, the cell density in the dorsal region never increases over time in embryos without DPP (Fig 2A–2C). Second, this phenotype can be rescued by the addition of st2-dpp. The rescued embryos have a broad area of high cell density on the dorsal surface beyond the sites of dpp RNA expression and rho activation (Fig 2D). Third, cell counts along the A/P axis of these embryos show a higher cell density anteriorly than posteriorly, showing a reorganization in the cell density along the A/P axis in response to the localized ectopic DPP source (Fig 2E and 2F). Finally, Getis-Ord Gi* spatial statistics show that while wild-type embryos have large hot spots in the central dorsal region flanked by cold spots in the poles (Fig 2G), the hot and cold spots are smaller in dpp embryos and there is an increase in randomly distributed cells (Fig 2H). The expression of st2-dpp reverts this phenotype similar to the wild type, with hot spots shifted more anteriorly (Fig 2I). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. DPP pulls cells over long distances. Cell density heatmaps of dpp- embryos at early (A) and late (B) cellularization stages. The dorsal surface has slightly higher density than ventral surface that does not change over time. Average heatmap shown on left (n = 9). Representative heatmap on right shows ftz stripes (black dots). (C) Cell number differences between ventral and dorsal sides in dpp- at early and late stages. (D) Cell density heatmaps of the dorsal surface of late stage dpp- embryos ectopically expressing dpp (st2-dpp) (gray dots, stripe position). Black dots in representative heatmap show rho-expressing cells. (E) Selected regions for cell counts using ftz as an A/P landmark for nearby (st-2) and distant (st-6~7) regions from ectopic dpp. (F) Cell counts in late-stage wild type and dpp- and dpp-, st2-dpp embryos. Note that st2-dpp expression increases the density of cells in ftz stripe 2 and reduces the density of cells between stripes 6 and 7 compared to wild type. Error bars, standard deviation. N.S., not significant. Asterisks, threshold values based on p-values (**p < 0.01) calculated with two-tail Mann–Whitney test (different stages in C; genotype comparisons in F) or two-tailed Wilcoxon signed-rank test (st-2 and st-6~7 comparisons of same genotypes in F). (G) Hot spot analysis of dorsal surface of wild type, dpp and dpp, st2-dpp embryos. Legend on left side indicates color-code for confidence intervals of hot spots (red shades) and cold spots (blue shades), and nonsignificant regions (NS, yellow). Scale bars, 60 μm (D and E, top) and 20 μm (E, bottom). Metadata for the graphs shown in C and F can be found at Supporting information S1 Metadata. A/P, antero-posterior; DPP, decapentaplegic. https://doi.org/10.1371/journal.pbio.3002021.g002

A genome-wide search identifies GUK-holder and frazzled as candidate genes to regulate cell movements The results above show that DPP attracts cells dorsally and DL stalls them ventrally by excluding DPP expression. However, since DL encodes a transcription factor and DPP encodes a secreted signaling protein, these cell movements must be enabled by downstream genes. To identify these effectors, we searched for genes likely to respond to DL and/or DPP gradients and that encode proteins with a role consistent with the regulation of cell movement. We screened the Drosophila genome for genes with similar developmental expression patterns to the DPP receptor thickveins using the existing modENCODE mRNAseq development database. Out of 101 genes identified, we selected those with predicted functions in cell migration, cell adhesion, and/or cytoskeleton regulation, as well as asymmetric expression along the D/V axis (see Methods for details). This search led to the identification of 2 genes, frazzled (fra) and GUK-holder (gukh). Both genes have been implicated in migration in various developmental contexts but were not previously associated to either DPP or DL, and their early embryonic functions are unknown. fra is the Drosophila homolog of Deleted in Colorectal Cancer gene (DCC) and encodes a protein belonging to the immunoglobulin subfamily that functions as the receptor of Netrin [25]. FRA/DCC was previously implicated in glial and axonal migration, and migration of various cell types including cardiac, salivary gland, and mesenchymal cells in Drosophila [25–30]. gukh encodes a protein with a SCAR-WAVE domain predicted to act on the nucleation of actin filaments [31]. In addition, GUKH physically interacts with membrane-associated guanylate kinases (MAGUKs), such as DISCS LARGE (DLG) [31–33] and is required for the correct subcellular localization of the planar cell polarity proteins DLG and Scribble (SCRIB) [31,34]. In Danio rerio, its ortholog has been shown to regulate cell migration during craniofacial development [35]. Furthermore, the human orthologue of GUKH, the Nance–Horan syndrome gene (NHS), nucleates actin filaments [36]. Noteworthy, emerging work link both fra and gukh to epithelial polarity functions, morphogenesis, and regulation of adherens junctions along the cell apico-basal axis [29,34,37–39].

fra and gukh are required for stereotyped cell movements We next asked if the response of fra and gukh to the D/V gradients of DL and/or DPP is required for the stereotyped cell movements in the embryo. To address this issue, we analyzed cell density heatmaps and hot/cold spots of the null mutants fra3 and gukhL1. The heatmaps show that the cell density decreases in the dorsal region of mutant embryos and increases in the ventral region (Fig 3L). These findings are confirmed by direct cell counts that yield a significantly lower D/V cell count differences in fra and gukh embryos compared to the wild type (Fig 3M). The hot spot analyses in the dorsal region can still identify hot spot clusters in the mutants, and those are established with less cell numbers (Fig 3N). Noteworthy, we observe the emergence of cell density hot spots within the ventral midline of mutant embryos (Fig 3N). This finding contrasts to wild-type embryos, which typically show random cell distribution in the ventral region, and occasionally have few hot spots in more lateral regions but not in the ventral midline (Fig 3N). Finally, density heatmaps in the lateral region of the embryo corresponding to the neuroectoderm show a higher cell density in fra3 and gukhL1 than the wild type (S3 Fig), in agreement with a decreased attraction of lateral cells to the dorsal region. Taken together, these results show that fra and gukh are required for the stereotyped cell movements in the blastoderm.

The cell movement towards the dorsal side of embryos deploys changes in cell area that are regulated by FRA and GUKH Our results show that there is a noticeable decrease in the size of dorsal cells from early to late stage (Fig 1D and 1E). We also observe that in late stage, dorsal cells are on average 34% smaller than ventral cells (Fig 4A–4D). Since the shape of these cells often approximates a regular hexagon (Figs 1D and 4A), a single dorsal row that shrinks 34% of its area could dislodge 20% of the diameter of a ventral cell or 24% of a dorsal cell. Thus, the constriction of 5 cell rows can roughly pull 1 cell diameter of ventral-sized cells. To test whether the expression of fra and gukh in dorsal cells are required to constrict and pull lateral cells dorsally, we analyzed if mutants for these genes changed the area of dorsal cells. These experiments reveal that dorsal cells of fra mutants are 26% larger than wild-type dorsal cells, whereas gukh dorsal cells are almost 40% larger than the wild type (Fig 4D). Thus, we conclude both fra and gukh are required for the constriction of dorsal cells. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. FRA and GUKH are required for cell constriction through apical localization of E-CAD at cell membranes. (A–C) Image segmentation for cell surface measurements. (A) Filtered image of anti-E-CAD staining. (B) Image segmentation shows outlines of membranes bordering E-CAD. (C) Merge of A, B. (D) Quantifications of surface areas from wild-type ventral and dorsal cells, and dorsal cells from fra and gukh mutant at late cellularization stage. In the wild type, dorsal cells are significantly smaller than ventral cells. In fra and gukh, dorsal cells are not constricted as dorsal cells in wild type. Error bars, standard deviation; p-values calculated with two-tail Mann–Whitney test (***p < 0.001). (E, F) Late-stage wild-type embryo stained for anti-E-CAD showing lower levels in ventral (E) than dorsal cells (F). Dorsal cells from fra (G) and gukh (H) stained for E-CAD. Note larger cell sizes with diffuse E-CAD signal compared to wild type, and low E-CAD membrane signal in fra (see also S3 Fig). (I–K) Segmented dorsal cells from wild type, fra and gukh stained with E-CAD. The low levels of E-CAD in the mutants result in less cells with high circularity (red outlines) and more cells with lower circularity (green outlines) compared to the wild type. FRA-GFP (L, O, green) and E-CAD (M, P, red) in dorsal cells shown in sagittal (L and M) and apical-lateral surface views (O, P). (N, Q) Signal co-localization in merged images (arrows) (see also S4 Fig). Scale bar, 10 μm. Metadata for the graph shown in D can be found at Supporting information S1 Metadata. E-CAD, E-Cadherin. https://doi.org/10.1371/journal.pbio.3002021.g004

FRA increases the levels of E-CAD in adherens junctions and GUKH increases F-Actin bundles While quantifying the cellular areas of fra and gukh mutants, we noticed that E-CAD at spot adherens junctions (SAJs) appeared much weaker, ill-defined, and with gaps in fra mutants, indicating that fra is required for maintaining correct E-CAD levels and the integrity of the SAJs (Fig 4F–4G). In contrast, E-CAD levels at the membrane do not appear to change in gukh mutants, though there is more diffuse signal within the cells compared to the wild type, as well as abnormal SAJs with fewer interruptions than those seen in fra mutants (Fig 4H). In segmented images of E-CAD in the mutants, we note the presence of cells with jagged membranes and fused cells, as opposed to the smooth contours of well-separated cells in the wild type. These differences can be quantified by measuring the circularity of cell contours in fra and gukh mutants and wild-type embryos (red and green outlines in Fig 4I–4K). The SAJs are apical constrictions containing E-CAD with catenin among other proteins that interact with actin and regulate cell adhesion and signaling [41–43]. Our analyses also revealed that like FRA, E-CAD is more abundant dorsally than ventrally and its expression is greatly reduced in the ventral presumptive mesoderm (Fig 4E and 4F) [44]. This asymmetry of E-CAD is regulated directly or indirectly by DL since dorsalized embryos that lack nuclear DL have high E-CAD levels, and ventralized embryos with ubiquitous nuclear DL have low E-CAD levels (S3 Fig). Previous work on wound healing also found that E-CAD is either directly or indirectly regulated by DL [45]. However, in the case shown here, at least part of this regulation is indirect and mediated by FRA because in the absence of fra, the levels of E-CAD throughout the embryo drop to levels found in the ventral region (Figs 4E, 4G and S4). This similar distribution of E-CAD and FRA is consistent with the fact that E-CAD was also identified in our genomic screening for targets of DPP and/or DL. In agreement with these results, we show that both FRA and E-CAD co-localize at SAJs (Figs 4L–4Q and S5). In addition, we show that E-CAD regulation is independent from DPP (S6A Fig). Together, these data indicate that FRA is necessary to maintain high levels of E-CAD at the membrane. Since GUKH is required for apical cell constriction and its protein contains the SCAR-WAVE domain implicated in the nucleation of actin in filaments [31,46,47], we asked if the distribution of filamentous actin (F-Actin) in the cell perimeter was affected in gukh mutants. These analyses show that the loss of GUKH reduces the thickness of the actin bundles localized right below the cell membranes (S7 Fig). Consistent with this result, we show that dpp mutants, which lack gukh expression in the dorsal midline (Fig 3B), also have thinner actin bundles in the dorsal region (S6B Fig). Thus, from these experiments, we conclude that GUKH constricts cells by increasing F-actin. The ability of GUKH to regulate cell area appears to be highly conserved since proteins of the SCAR/WAVE family regulate cell morphology from plants to humans by promoting actin filament nucleation [46,47]. For example, the gukh human ortholog NHS was shown to maintain cells constricted, and its removal leads to a cell spreading phenotype [36].

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002021

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/