Fig 1.
Progressive sharpening of selector gene expression and the formation of epithelial folds in the EAD.
(A) Correspondence of developmental fields between late third instar (l-L3) EAD and the adult head. Red: eye; blue, green, yellow, and pink: proximal to distal (A1, A2, A3, and Ar) antennal segments; orange: ocelli; grey: head cuticle. (B) Lim1 (red) and Dll (blue) expressions are separated in the l-L3 EAD by an epithelial fold (F-actin, green). (C-E) w1118 EAD analyzed for morphological changes (F-actin, green), and gene expression patterns (Lim1, red; Dll, blue) in m-L2 (26-32h AEH, C-C”‘), l-L2 (38-44h AEH, D-D”‘) and e-L3 (48h AEH, E-E”‘). Z-axis projections at the yellow lines show the epithelial morphology in lateral view. (C) Dashed lines indicate m-L2 cells coexpressing Lim1 and Dll. (E) Arrows point to the fold at the Lim1/Dll expression border. (F-H) Dll (blue) and Lim1 (red) expression levels—based on pixel intensities (Y axis) and individually normalized against background—were quantified from the bracketed region (X axis, total 60 μm) in C’, D’ and E’. The center (0 in the X axis) was manually positioned at the fold (e-L3) or at the Dll-Lim1 overlapping regions (L2). All images in this and subsequent figures are oriented as dorsal-face up and posterior end to the right, with cross-sections oriented with the apical surface of the disc proper to the right or top. Scale bars: 50μm.
Fig 2.
Lineage restriction coincides with fold formation.
(A-D) TSM clones were induced at the indicated time and examined in l-L3 EAD. For lineage restriction analysis, only those clones located at or spanning the respective borders were scored. Clones were examined along XZ and YZ optical sections to determine whether the clone crossed the boundary at the bottom of the fold. (A-D) Apical XY planes. (A’, B’, C’ and D’) Basal XY planes. (A”, B”, C” and D”) The Z-axis projection images along the yellow line are shown. Red and blue arrows indicate clone that crossed or were restricted by the boundary, respectively. (E) Summary of the locations of all meaningful clones examined. Clones crossing or restricted by the boundary are indicated by a red cross or blue triangle, respectively. (F) The percentages of clonal patterns at the three stages are plotted; numbers of analyzed clones: m-L2 = 34, l-L2 = 37, e-L3 = 35. Scale bars: 50μm.
Fig 3.
Apical constriction and cell shape changes during A1 fold formation.
(A) Time-lapse images of Sqh-GFP in ex vivo cultured l-L2 EAD for 5 hours. Imaris filament, and surface-tracing modules were used for individual cell segmentation and tracing. (B) Cells are classified into three groups based on their dynamic changes in the apical area and are shown for Tfinal (4:52:01). Cells that showed a significant reduction in apical size (from 30–40μm2 to <10μm2) are labeled red. Cells that showed a fluctuating apical area, defined as δAreai (Amaxi-Amini) ≥ 10μm2 over time, are labeled blue. Otherwise, cells were considered to have a constant apical area (labeled grey). The yellow dashed line marks the A1 fold at Tfinal. (C) Cells in each group are plotted according to by their proportional change in apical area ((Ati−Aavgi) /Aavgi) over time. Mean ± stdev are shown as a solid line and lightly shaded area, respectively. The ranges of stdev for the constant and the fluctuating groups are 0.05–0.4, and 0.3–0.9, respectively. Analyzed cells in each group are indicated. (D) The apical areas of cells located at the A1 fold were significantly reduced 2h prior to fold formation (at 180min). Mean ± stdev are shown. Cells away from the fold do not show a significant reduction in apical area. (E) Quantitation of cell height and volume from l-L2 and e-L3 EAD. Stack contours from aPKC (apical, magenta) and FasIII (basolateral, white) staining for volume rendering are illustrated in the middle panel. In addition to cells at the A1 fold, cells one row (perifold-1) and two rows (perifold-2) away from the A1 fold were also scored. (F) Sqh-mCherry accumulates periodically in the apical-medial region (yellow arrow) of constricting cells (red dashed line). Original time-lapse images: see S1 Movie. (G) During e-L3, the cells at the fold (marked by stars) exhibit an even distribution of Sqh-GFP (green) in a 2–3μm stack image projection (N = 4). (H) Mitotic cells (pH3, magenta) are observed along the A1 fold (arrow) during e-L3 (DAPI, blue; F-actin, green). EAD presented 3–4 mitotic cells (N = 7). Scale bars: 5μm, except in H for which it is 50μm. Time indicated as hh:mm:ss.
Fig 4.
Myosin activity underlies formation of the epithelial fold to ensure lineage restriction.
(A-B) zip2 MARCM clone (GFP-positive) and sqaf01512 clone (GFP-negative) show mixing of Lim1 and Dll cells (A” and B’, 18/23 in zip2, 11/15 in sqaf01512, numbers indicates discs with Lim1-Dll mixing/total disc). (A’) Disc morphology is revealed by Coracle (white) staining. White arrow indicates the A1 fold position. (C-E) In adults carrying zip2 or sqaf01512 clones, mislocalized ommatidia (highlighted red in SEM) can be detected in antennal segments (C) or head cuticle (D). (E) Adults with sqaf01512 mutant clones showed antenna-like tissue at the borders of compound eyes. (F) Necrotic-scar like cells were observed in zip2 mutant adults (highlighted in blue). (G-G’) hth-GAL4 expression (RFP, green) completely covers the Lim1 expression field (red) and partially overlaps with Dll (blue). (H-I) Knockdown of zip or sqh by hth-GAL4 from L2 (hthL2: hth-GAL4+tub-GAL80ts, shifted to non-permissive temperature from L2) showed high penetrance of Lim1 and Dll cell mixing (yellow arrow, zip: 24/24; sqh: 21/22). Scale bars: 50μm, except C-F: 100μm.
Fig 5.
Disruption of myosin activity using CALI permits boundary crossing.
(A) Inactivation of actomyosin by Sqh-GFP-mediated CALI. Ex vivo-cultured e-L3 EAD were fixed immediately after CALI treatment (see Materials and Methods). EAD morphology is revealed by Sqh-GFP (green), aPKC (magenta) and Coracle (white) staining. The fold in the control region (non-CALI, white box) and CALI treated region (yellow box) were compared. (A’) In contrast to the control (non-CALI, white arrow), the CALI-treated region (yellow arrow) exhibits a significant reduction in the extent of tissue fold. (B) The actin-binding ERM protein Moe was used as a control for CALI specificity. Expression of Moe::GFP (green) only decorates actin filaments without notable effects. (B’) CALI on Moe::GFP does not affect the A1 fold (compare the non-CALI, white box/arrow, and the CALI-treated region, yellow box/arrow). (C-D) CALI in combination with a clonal tracing experiment. Cells expressing RFP (magenta) were induced upon heat-shock in L2. After 24h, the EAD was dissected, treated with CALI, and monitored for 14-16h. The images were shown in 3D projection. (C-C”) In cultured EAD, inactivation of Sqh-GFP (green) via CALI (yellow boxed region) at the A1 fold (white dashed line) showed that the RFP cells originated from the Dll field across the A1 boundary (arrows in C”, see S2 Movie) and maintained Dll expression (D, blue). (D) Dll expression (blue) after live imaging for 14h. N = 3. (E) Trajectory of the two RFP cells that cross the A1 fold (as cells pointed by yellow arrows in C”). Red squares indicate the RFP positions at post CALI time 6:01:48, with overall trajectory shown in color-coded time map (dashed line: A1 fold). Star and diamond indicate position of RFP cells at T0 and Tfinal post CALI treatment, respectively. (F-G) Cross sections to show that the two RFP clones in C”(marked by white arrowheads) that seemed to have crossed the A1 boundary are actually located in the peripodial epithelium. White and yellow lines outline peripodial and disc proper, respectively. Scale bars: 50μm, except in E: 10μm. Time indicated as hh:mm:ss.
Fig 6.
Notch activation and differential expressions of Delta, Serrate and fng-lacZ precede epithelial fold.
(A-E) N activation is indicated by Nintra, (A’, B’, green, shown in 3D projection) and the N transcriptional reporters Su(H)Gbe-lacZ (C, D, white) and E(spl)mβ -lacZ (E, white) in an antennal disc from l-L2 (A,C, E) to e-L3 (B,D). Nintra is low and relatively uniform in l-L2 (A) and highly enriched at the Lim1/Dll expression border in e-L3 (B). (C, D) Su(H)Gbe-lacZ shows a circular pattern before (C, l-L2) and after (D, e-L3) A1 fold. In the cross section, as shown on the right for C and D, the reporter intensity is strongest at the center of the putative fold and gradually declines in surrounding cells (marked by brackets). (E) E(spl)mβ-lacZ expression also appears in a circular pattern, corresponding to the future A1 fold at l-L2. (F) Expression of the selector genes Lim1 (red) and Dll (blue), the N ligands Dl (green) and Ser (magenta), as well as fng-lacZ (orange) were analyzed prior to tissue fold in three groups of discs of increasing size putatively representing increasing developmental time. Pixel profiling shows clear segregation of Dll/Lim1 expression in groups 1, 2 and 3. Segregation of Dl/Ser expression only begins in group 2 and becomes clear in group 3. Dl is high in the central (Dll-expressing) region, and Ser is high in the peripheral (Lim1-expressing) region. fng-lacZ exhibits only a slightly elevated level in the peripheral region of group 1, but becomes higher in the peripheral compared to the central region of group 3 (see S7 Fig for raw images). (G) Ratios of Dll/Lim1 and Delta/Serrate from single cells are plotted for group 1 (green, N = 49), group 2 (orange, N = 50), and group 3 (black, N = 51). Regression lines from the three stages show a positive correlation, with increasing correlations over development time (compare R2 indicated next to the regression lines). Scale bars: 50μm.
Fig 7.
Notch activation drives apical constriction and epithelial fold.
(A) Dominant-negative Notch (NDN) expression by hth-GAL4 disrupts the A1 fold (arrow, Coracle, white; RFP, green) and mixing of Lim1 (red) and Dll (blue) cells (A’: enlargement of boxed region in A, 25/29). (B) NDN clones within Dll or Lim1 single fields do not alter cell fate (14/14, compare Dll and Lim1 intensity in NDN and in control cells). (C) Ex vivo observation of N RNAi driven by hth-GAL4 (hth>NKD) in the EAD. Changes in cell apical size were monitored by Sqh-GFP. The three types of cells (decreasing, fluctuating and constant, see Fig 3B and 3C) are color-coded. Still frames from two samples at Tfinal are shown. The A1 fold fails to form in the absence of N signaling. (D) Proportions of the three cell populations in hth>NDN EAD. There is a reduction of cells with a decreasing apical area (red), and an increase of cells with a fluctuating apical area (blue). Cell numbers of constant, fluctuating, and decreasing groups are 17, 31, and 17 in the control, and 15, 29, and 5 in NKD. ((E-E’) DlRevF10, SerRX82 MARCM clones (mutant cells marked by mCD8-GFP, green) show reduced A1 fold (E’: compare white and yellow arrows in the control and mutant cells, respectively, in the optical section, 14/18). White line indicates a clone border. (F-F’) Clonal expression of Nact (RFP, magenta) causes ectopic fold (arrow) within the clone, even when away from the fold (37/49). (F’) Optical section along the yellow line indicated in (F). (G) The effect of N activity on cellular apical surface area. Cells with constitutively-active (Nact, N = 23) or dominant-negative (NDN, N = 28) N activity have apical areas similar to cells at the A1 fold (fold cells, N = 11) or cells outside of the fold (non-fold cells, N = 21), respectively. (H) The volumes of the apical and basolateral domains (defined by aPKC and FasIII, respectively) in cells expressing Nact or NDN (N = 12) were quantified. Nact cells were assayed at 48h (N = 16) and 72h (N = 14) post-clonal inductions. (I-K) Clonal expression of bantam (marked by RFP, red in I; green in J-K). The clone border is marked by a yellow line (I”) or a white line (J-K). (I) bantam blocks epithelial fold (arrow) via reduced Ena (I’, white) within the clone (16/22). (J) bantam overexpression results in mixing of Lim1 and Dll cells within the clone (9/14). (L) Coexpression of NDN and bantamsponge by hth-GAL4 shows normal A1 fold and no mixing of Lim1/Dll (L’, 10/13). Scale bars: 50μm, except in C: 5μm. ** P ≤ 0.01 *** P ≤ 0.001 (ANOVA-Tukey’s multiple comparisons).
Fig 8.
The folded epithelial structure reinforces N signaling.
N activity is indicated by the reporter Su(H)Gbe-lacZ (in heat map). The epithelial fold is marked by F-actin (white). (A-C) EAD. (D-F) Wing disc. (A, D) The dpp expression domain in the EAD (A) and wing disc (D) is indicated by GFP expression (dppL2>GFP, green). zip (B,E) and sqh (C, F) were individually knocked down from L2. Knockdown of actomyosin specifically disrupted A1 fold in the dpp region (B’ and C’, arrow) and reduced Su(H)Gbe-lacZ levels in the EAD (B”, C”) but not in wing disc (E’, F’). (G) The level of Su(H)Gbe-lacZ in dpp-GAL4 was normalized with a non-dpp region in the respective EAD or wing disc. Total analyzed disc numbers in control, sqh KD, and zip KD were 7, 8, and 9 (for EAD) and 6, 8, and 10 (for wing disc) respectively. Scale bars: 50μm. *** P ≤ 0.001 **** P ≤ 0.0001(ANOVA-Dunnett’s multiple comparisons).
Fig 9.
The process of A1 boundary formation.
At m-L2, expression of the Dll and Lim1 selector genes in cells is initially low, with fuzzy coexpression. From l-L2 to e-L3, expression of both Dll and Lim1 becomes elevated and sharply segregated. This is followed by differential expression of Dl in the Dll domain, and Ser and Fng in the Lim1 domain, and then subsequently by N activation at the Dll/Lim1 interface. N activity then represses bantam, resulting in the de-repression of Ena, which triggers non-cable actomyosin-dependent cytoskeleton reorganization to drive apical constriction and epithelial fold. The epithelial fold then serves as a physical barrier to prevent mixing of cells from the Dll and Lim1 field.