Report

Programmed cell death along the midline axis patterns ipsilaterality in gastrulation

See allHide authors and affiliations

Science  10 Jan 2020:
Vol. 367, Issue 6474, pp. 197-200
DOI: 10.1126/science.aaw2731

Apoptosis prevents left-right crossing

Animals generally display a bilateral body plan, with symmetry between the left and right sides. Maya-Ramos and Mikawa examined the mechanism that prevents the crossing of cells between sides. Using chick embryos, they show that cellular mixing between sides is prevented by a barrier at the embryonic midline that involves programmed cell death and the extracellular matrix. This work demonstrates the dependence of normal development on programmed cell death during the gastrulation stage.

Science, this issue p. 197

Abstract

Bilateral symmetry is the predominant body plan in the animal kingdom. Cells on the left and right sides remain compartmentalized on their ipsilateral side throughout life, but with occasional variation, as evidenced by gynandromorphs and human disorders. How this evolutionarily conserved body plan is programmed remains a fundamental yet unanswered question. Here, we show that germ-layer patterning in avian gastrulation is ipsilateral despite cells undergoing highly invasive mesenchymal transformation and cell migration. Contralateral invasion is suppressed by extracellular matrix (ECM) and programmed cell death (PCD) along the embryonic midline. Ipsilateral gastrulation was lost by midline ECM and PCD inhibition but restored with exogenously induced PCD. Our data support ipsilaterality as an integral component of bilaterality and highlight a positive functional role of PCD in development.

Bilaterians have two symmetrical sides and constitute 99% of all animal species (1). Human disorders and naturally occurring bilateral gynandromorphs reveal that progenitor cells on the left and right sides give rise to cells that will primarily remain on the same (ipsilateral) side of the organism (2, 3). Although this topological cell-fate restriction is evolutionarily conserved across many animal species, ipsilaterality is not well understood (4). It is particularly intriguing how left–right cellular mixing is prevented in birds and mammals considering that in these species, gastrulating cells undergo an epithelial–mesenchymal transition (EMT), resulting in highly migratory and invasive mesenchymal cells (5).

Using the chick as a model system, we analyzed gastrulation cell-movement patterns using whole-embryo live imaging. The chick allows for high-precision control of spatiotemporal transfection parameters. The primitive streak (PS) is the conduit for germ-layer formation and defines left–right compartments in the embryo. Fluorescently tagged left and right epiblast cells moved toward the PS and underwent EMT (Fig. 1, A and B; fig. S1, A to C; and movie S1), with most ingressed cells migrating away from the PS to the ipsilateral side of transfection (left 96.1 ± 1.2%, right 96.1 ± 0.3%, p = 0.9123, n = 14) (Fig. 1C) (6). There is minimal mixing between left and right mesenchymal populations, effectively creating two compartments and patterning the ipsilateral sides.

Fig. 1 Gastrulation is ipsilateral with PS midline PCD persistence.

(A) Experimental flow diagram. Left and right side epiblast cells were electroporated with Flag:2A:H2B-RFP (red) and Flag:2A:H2B-GFP (blue), respectively, at the early PS stage; tagged cells were traced during gastrulation. L, left; A, anterior; R, right; P, posterior; NS, nonsignificant. (B) Dorsal views of embryos at stages Hamburger–Hamilton (HH) 3 (25) (left) and HH 4+ (right). Arrowheads indicate the PS midline. (C) Ipsilateral versus contralateral distribution of tagged cells at end point. (D) Cleaved caspase 3+ cells (red) at the PS midline. Cyan is DAPI staining. (E) Dorsal view of propidium iodide–stained gastrulating embryo. Note the array of stained cells along the PS midline. (F and G) SEM images of persisting cellular debris along the PS midline (arrowheads). (H) Quantification of blebbing cells along the mediolateral axis (n = 4).

To assess whether ipsilateral gastrulation is governed by inherent epiblast cell information or by non–cell-autonomous forces, an epiblast segment from green fluorescent protein (GFP)–transgenic chick embryo was grafted to the contralateral side of a stage-matched wild-type (WT) embryo (fig. S1, D and F). In a cell-autonomous process, donor cells grafted isochronously at a contralateral side would be expected to move back to their originating side. However, grafted implants displayed ipsilateral gastrulation, identical to control grafts (heterotopic 91.3 ± 4.7%, homotopic 91 ± 2.4%, p = 0.9893, n ≥ 3) (fig. S1, E, G, and H). These data point to an environmentally regulated process.

In both GFP-tagged and grafted embryos, labeled epiblast cells migrated without detectable obstruction until reaching the PS midline, where a sharp boundary was formed by their distribution along the PS, as if a midline barrier was preventing their movement (Fig. 1B and fig. S1, B, E, and G). This precise movement restriction hinted at a midline-specific program deterring cell crossing.

To examine the midline in depth, epiblast cells were tagged with membrane-tethered GFP for multiphoton live imaging. Results showed increased apoptotic cellular membrane blebbing as cells approached the midline (Fig. 1H, fig. S2Q, and movie S2), which is consistent with our previous report (7). TUNEL assay and cleaved caspase 3 immunofluorescent (IF) staining revealed broad epiblast programmed cell death (PCD) distribution in early gastrulation that became enriched at the midline in later stages (Fig. 1D and fig. S2).

Midline PCD did not follow classical scavenger cell–mediated cellular corpse clearance (8, 9) as continued presence of PCD at the midline was morphologically evident (Fig. 1, E to G, and figs. S2 and S3). These data indicate that PCD is enriched and persistent at the midline. Thus, we suspected that midline PCD could act as a barrier directing ipsilateral gastrulation.

To test this hypothesis, the ontogeny of midline PCD was traced with the early apoptosis reporter Annexin V (10). Yellow fluorescent protein–tagged secreted Annexin V was preferentially detected at the midline in cells undergoing blebbing compared with lateral epiblast cells (figs. S2R and S2S and movie S4). Additionally, Alexa Fluor 488–conjugated Annexin V labeling of epiblast cells revealed that PS midline cells undergoing PCD originated in the PS posterior region (Fig. 2, A to C, and movie S4). Therefore, to suppress midline PCD, the molecular pan-caspase inhibitor P35 was introduced posteriorly (Fig. 2D) (11). In embryos electroporated with P35:2A:H2B-GFP at the posterior region and H2B:RFP at lateral epiblast, many red fluorescent protein–positive (RFP+) cells invaded the contralateral side (36.0 ± 1.0%, p < 0.0001, n = 4) (Fig. 2, E to G; fig. S4; and movies S5 and S6). In comparison, lateral epiblast expressing P35 ingressed identically to WT cells (fig. S5). These results strongly suggest that midline PCD is required for ipsilateral gastrulation.

Fig. 2 PS midline PCD is necessary for ipsilateral gastrulation.

(A) Diagram of Annexin V epiblast cell labeling at stage HH 2 and tracking to stage HH 4+. (B) Time-lapse images of labeled cell trajectories. Color code, original x-axis position; arrows, PS midline. (C) Distribution of Annexin V+ cells along the embryonic axes at HH 4+. (D) Diagram of control and P35 (green)–mediated midline PCD suppression with tracking of lateral epiblast cells (red). (E) Control ipsilateral gastrulation of H2B-GFP+–tagged cells (red) with Lissamine oligo (green) along the PS. (F) Induced contralateral invasion of H2B-RFP+–tagged cells (red) with P35-expressing cells (green) along the PS. (G) Quantification of P35-induced contralateral invasion.

We surveyed several factors that could be involved in directing ipsilateral gastrulation (fig. S6). FGF and Eph signals are prominent in other midline systems, yet FGFR1-FGF8 and Eph3-EphrinB1 did not mimic the PCD midline barrier function (fig. S7) (12, 13). Therefore, we examined the ECM, which can suppress cellular movement and be involved in PCD (14, 15). Although the ECM was highly fenestrated in the lateral PS, it was uniquely enriched along the PS midline (Fig. 3, A, C, and D). Inversely, the membrane-bound matrix metalloproteinase 15 (MMP-15) had more expression in the lateral PS than in the midline PS (Fig. 3B and fig. S8). This complementary expression pattern suggested ECM regulation by MMP-15.

Fig. 3 PS midline ECM is necessary for ipsilateral gastrulation.

(A) Whole-mount laminin IF staining. (B) Whole-mount MMP-15 in situ hybridization. (C, E, and G) Transverse sections of HH4 WT (C), LAMA1 morpholino (E), and hMMP15-treated (G) embryos. Shown is IF staining for laminin (Lm), fibronectin (Fn), and collagen IV (Col IV). Arrows indicate the PS midline. (D, F, and H) Measurement of IF staining intensity of ECM proteins (indicated) at various distances from the PS midline in (C), (E), and (G), respectively. AU, arbitrary units. (I) Diagram of ECM-targeted proteins and subsequent gastrulation pattern analysis. (J and K) Induced contralateral gastrulation by LAMA1 morpholino (J) and hMMP15 (K). (L) Quantification of contralateral invasion. Shown are the LAMA1 control morpholino (n = 3), the LAMA1 morpholino (p = 0.0029, n = 6), and the hMMP-15-treated embryo (p < 0.0001, n = 3).

To test the role of ECM in ipsilateral gastrulation, laminin was depleted with the laminin subunit alpha-1 (LAMA1) morpholino (Fig. 3I and fig. S9). The resulting embryos exhibited higher contralateral invasion than scrambled morpholino controls (21.8% versus 4.3%, p = 0.0029, n = 6, n = 3) (Fig. 3, J and L, and movie S7). As there was more contralateral invasion after caspase 3 inhibition, we suspected that additional ECM proteins were involved in preventing contralateral invasion (Fig. 3, E and F). Indeed, introduction of MMP-15:2A:mCherry largely diminished midline ECM (Fig. 3, I, G, and H, and fig. S4), leading to robust contralateral invasion of GFP+ lateral cells (39.3 ± 1.5% p < 0.0001, n = 3) (Fig. 3, K and L, and movie S8). These data point to a spatial regulation of ECM enrichment along the midline that is necessary for ipsilateral gastrulation.

Because the above data suggested that both PCD and ECM are required for ipsilateral gastrulation patterning, we suspected a potential interplay between these signaling axes (Fig. 4A). Midline ECM was apparent before PCD (fig. S10, A to K). Furthermore, P35-mediated PCD inhibition did not diminish midline ECM expression (fig. S10, L to O). By contrast, LAMA1 morpholino–mediated midline laminin knock-down decreased the amount of cleaved caspase 3 signal at the midline (Fig. 4B). Given that PCD occurs throughout the embryonic disc and that midline PCD originates in cells from the posterior PS, ECM may be required for the alignment of PCD along the midline, although we cannot rule out ECM as a direct PCD inducer through anoikis (14, 16). In both scenarios, we expect that the resulting atypically persisting midline PCD functions as a barrier to prevent contralateral invasion of gastrulating cells.

Fig. 4 Induced PCD restores ipsilateral gastrulation in ECM-depleted embryos.

(A) Transverse optical section, 10-μm thickness, double stained for cleaved caspase 3 (top) and laminin (middle). Bottom is a 3D reconstruction. C. casp 3, cleaved caspase 3. (B) Quantification of cleaved caspase 3+ clusters at various distances from the PS midline in WT embryos (yellow), control morpholino (blue), and LAMA1 morpholino (aqua). Ratios represent the total number of clusters at the midline and non–midline-lateral regions. (C) Diagram of ipsilateral gastrulation rescue by staurosporine-induced PCD. (D) Electroporated LAMA1 morpholino and staurosporine-injected embryo. Gastrulation pattern is traced by Flag:2A:H2B-GFP (white). Contralateral invading cells are pseudocolored in green. Blue arrows indicate the PS midline. (E) As in (D) but with microinjected vehicle only (dotted circle). (F) Quantification of contralateral invasion suppressed by induced PCD. Shown is the injection zone versus the noninjection zone (p < 0.0001, n = 3). (G) Model of PCD-mediated regulation for ipsilateral gastrulation.

To test this possibility directly, we introduced PCD using two independent methods in midline laminin–depleted embryos (Fig. 4C) (17, 18). Induced PCD was sufficient to suppress contralateral invasion locally, thereby restoring ipsilateral ingression in a midline ECM–depleted background (Fig. 4, D to F; figs. S11 and S12; and movies S9 and S10). Thus, both ECM and PCD are required to pattern ipsilateral gastrulation, but only PCD is sufficient to restore ipsilaterality. In this scenario, we propose that ECM aligns PCD along the midline, which prevents the contralateral invasion of mesenchymal cells, thereby patterning two ipsilateral body compartments (Fig. 4G).

We have uncovered a functional and developmental role for PCD. Although developmental PCD is classically considered a process required for unnecessary cell removal and tissue morphogenesis (19), our data establish a positive role for PCD: the patterning of ipsilateral gastrulation. Ipsilaterality could ensure a symmetrical balance by preventing large disparities between left and right sides, resulting in proper laterality development (7), viability, and fitness in bilaterians (20). These findings in the avian model will stimulate comparative work in mammals, amphibians, and other bilaterian species as new live-imaging technologies become widely available (21). PCD is also a mechanism to curtail pathological EMT in certain malignancies (22). Our work adds another layer to this concept by showing that cellular mesenchymal migration directionality can be instructed by PCD in a physiological process.

Supplementary Materials

science.sciencemag.org/content/367/6474/197/suppl/DC1

Materials and Methods

Figs. S1 to S12

References (2325)

Movies S1 to S10

References and Notes

Acknowledgments: We thank E. G. Baylon, J. Hyer, and T. Kornberg for their comments and assistance with statistical analysis and past and present Mikawa lab members for their suggestions regarding this work. Funding: This work was supported by NIH grants R01 HL122375, R37 HL078921, R01 HL132832, and R01 HL148125. L.M.-R. was awarded a Howard Hughes Medical Institute Gilliam Fellowship. Author contributions: T.M. and L.M.-R. conceptualized the project and wrote the manuscript. L.M.-R. performed all experiments. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the main text, supplementary materials, or auxiliary files.

Stay Connected to Science

Subjects

Navigate This Article