The temporal pattern of imaginal ring cell proliferation

Imaginal rings are organized and arranged into tube-like structures (Figure S1, G–I). Compared with their neighboring polyploid larval tissues, both the cell and nuclear sizes of imaginal ring cells are smaller (as depicted in Figure 1). The imaginal ring cells are epithelial cells that display typical apical–basal polarity, as revealed by staining with antibodies against an apical marker, aPKC (Rolls et al. 2003; Morais-de-Sá et al. 2010), and a basal marker, DG (Deng et al. 2003; Schneider et al. 2006). Similar to other tubes formed by epithelial cells, the apical surface of imaginal ring cells faces the lumen, whereas the basal side is away from the lumen and forms the outside surface (Figure S1, A–F).

Imaginal ring cells have been reported to undergo cell proliferation to increase the cell number during larval stages (Mandaravally Madhavan and Schneiderman 1977). To determine the precise temporal pattern of cell proliferation, we performed PH3 (an M-phase marker) antibody staining (Gurley et al. 1978) of imaginal rings from late first instar to early pupal stage. PH3 signal was hardly detectable during the first and early second larval instar, but became evident around the mid- to late second and early third instar. The peaks of cell proliferation in all imaginal ring cells were at 96 hr AEL. At this time, PH3 staining was present in 5.4, 6.2, and 6.1% of foregut, hindgut, and salivary gland imaginal ring cells, respectively (number of total cells observed = 5143, 4889, and 1557, respectively). The PH3 signal gradually declined when entering the pupal stage (Figure S2A). These results suggest that imaginal ring cells undergo rapid proliferation during the L3 stage. Consistent with this, we found an increase in the number of imaginal ring cells during different larval stages, which agrees with the previous study in the hindgut imaginal ring (Fox and Spradling 2009). The average numbers of foregut, hindgut, and salivary gland imaginal ring cells at the late first-instar larval stage were 51, 90, and 11 respectively. These numbers did not change substantially during the early- to midsecond instar. However, during the late third-instar larval stage the average number per imaginal ring increased to 504, 531, and 147 in foregut, hindgut, and salivary gland imaginal rings, respectively (Figure S2B). Taken together, these results suggest that imaginal ring cells are kept in a quiescent state until the midsecond-instar larval stage, then undergo rapid cell proliferation to reach an adequate number of progenitor cells for differentiation into adult tissues.

Notch signaling is active in imaginal rings

To determine how imaginal ring development is impacted by different signaling pathways, we studied the expression pattern of several signaling pathways by screening a series of reporters and Gal4 lines. From this screen, we found that the Notch signaling reporter, NRE-eGFP (Housden et al. 2014; Zacharioudaki and Bray 2014), was expressed in all three imaginal ring tissues from first instar to late third instar (Figure 2). Although the NRE-eGFP pattern in young larval stages presents relatively equally in all imaginal ring cells (Figure 2, A–C), the pattern does not appear to be uniform during the third instar (Figure 2, D–F). In salivary gland imaginal rings, the middle region has the highest Notch activity, whereas the two sides have lower Notch activity. In foregut imaginal rings, NRE-eGFP is expressed more strongly in the most posterior region than in the anterior part. In the hindgut, posterior imaginal ring cells have stronger NRE-eGFP expression as well. We confirmed that Notch is active in these imaginal rings with a different Notch activity reporter line, E(spl)-mβCD2 (Figure S3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Expression of Notch Responsive Element-enhanced GFP (NRE-eGFP) during larval stages. (A–C) NRE-eGFP expression in imaginal ring cells in midfirst-instar larvae. (D–F) NRE-eGFP presents in the foregut (FG), hindgut (HG), and salivary gland (SG) imaginal rings (ImRs) of the midthird instar. (A and D) FG ImR; (B and E) HG ImR; and (C and F) SG ImR. eGFP signal was enhanced by GFP antibody staining (green). Nuclei were labeled with DAPI (blue). Bar, 20 µm.

Lineage analysis of imaginal ring cells

Although imaginal ring cells have been described as precursors for adult salivary gland, foregut, and hindgut cells, no direct evidence has been shown. To confirm the progenitor/descendent relationship between the imaginal rings and the adult organs, we decided to use the G-TRACE system for lineage analysis, in which RFP labels Gal4 real-time-expressing cells and GFP labels the lineage cells derived from Gal4-expressing cells (Evans et al. 2009). However, no Gal4 lines have been reported to show the imaginal ring-specific expression pattern needed to carry out the lineage tracing study. Since Notch shows specific activity patterns in all three imaginal rings, we generated NRE-Gal4 transgenic flies by replacing the eGFP sequence with Gal4. Crossing with a UAS-GFP line revealed that NRE-Gal4 is specifically expressed in all three types of imaginal rings (Figure S4, A–C). To prevent the possibility that expression of NRE-Gal4 in adult cells might confound this analysis, we also utilized Gal80^ts to control the timing of NRE-Gal4 activation. Flies were reared at 18° first, then transferred to 29° at the late second-instar stage to inactivate Gal80^ts and allow functional Gal4 to mediate the G-TRACE system during the third-instar larval stage. After pupation, the flies were transferred back to 18° and dissected for further analysis (Figure S12C). We found that adult foregut (esophagus, cardia, and crop) (Figure 3, A and B), hindgut (Figure 3C), and salivary gland (Figure 3D) cells were GFP positive but RFP negative, suggesting that the above-mentioned cells are the descendants of larval NRE-Gal4-expressing cells and NRE-Gal4 is indeed no longer real-time expressed because of the Gal80 function. We also noticed that a subset of adult foregut, hindgut, and salivary gland cells were not labeled with GFP. In addition, GFP-negative cells did not show a specific pattern but presented randomly, suggesting that this phenomenon might be caused by a threshold effect of the Gal4 driver (Evans et al. 2009), though the possibility that cells without GFP are derived from other precursor cells cannot be ruled out. These results confirm that imaginal ring cells are progenitor cells that develop and differentiate into adult salivary glands, foreguts, and hindguts.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Imaginal ring cells form adult foregut, hindgut, and salivary gland. G-TRACE on NRE^ts for lineage tracing study. eGFP-labeled progeny cells (green) in adult esophagus, cardia (A), crop (B), hindgut (C), and salivary gland (D). Real-time RFP expression is not detected. Nuclei were labeled with DAPI (blue). Bar, 20 µm. eGFP, enhanced GFP; G-TRACE, Gal4 technique for real-time and clonal expression; NRE^ts, temperature-sensitive Notch responsive element; RFP, red fluorescent protein.

Canonical Notch signaling is required for imaginal ring cell proliferation

During the third larval instar, the cell proliferation rate and Notch activity are the highest in all three imaginal rings (Figure S2A), (Figure 2, D–F). To determine the relationship between Notch signaling and imaginal ring cell proliferation, we knocked down Notch using the TARGET system (McGuire et al. 2004), which applied an Actin5C promoter-driven Gal4 and Gal80^ts to restrict the timing of Notch-RNAi expression only to the third-instar stage (Figure S12D; Act-Gal4, Tub-Gal80^ts abbreviated to Act^ts hereafter). Inactivation of Notch signaling during the third instar reduced tissue size as compared with wild-type and sibling controls (Figure 4, A–F). The ratio of PH3-positive cells to total cells counted (mitotic index hereafter) was significantly reduced in these loss-of-Notch-function imaginal rings as well (Figure 4G). To test the possibility that global dysregulation of the Notch pathway using Act-Gal4 might result in systemic effects that in turn disrupt cell proliferation in the imaginal rings, we used two other nonubiquitous drivers, NRE-Gal4 and retn-Gal4 (which are expressed in imaginal rings from late second instar; Figure S4, A–C and Figure S5, A–C), to knock down Notch signaling in imaginal ring cells, and we found that the mitotic index was also dramatically diminished (Figure S4D and Figure S5D). Moreover, use of a Notch temperature-sensitive (N^ts) mutant resulted in decreased mitotic activity and fewer imaginal ring cells when flies were reared at the restrictive temperature (29°) during the third-instar larval stage (Figure S6, A and B and Figure S12D). To further confirm that Notch mutation leads to a lower cell proliferation rate, the MARCM technique (Lee and Luo 1999, 2001) was used to generate Notch mutant clones (FRT19A N^55e11). We generated MARCM clones in late second instar and dissected L3 flies to observe the clone size. The average cell number per mutant clone was about half that of the control clones (FRT19A only) (Figure S6C). Taken together, these results suggest that Notch signaling in imaginal ring cells is critical for cell proliferation during the late larval stage.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Notch is required for imaginal ring (ImR) cell proliferation during the third-instar larval stage. (A–F) Cell proliferation was labeled by phospho-histone 3 (PH3, red) in controls and loss-of-Notch ImRs. (A and D) foregut (FG) ImR; (B and E) hindgut (HG) ImR; and (C and F) salivary gland (SG) ImR. Nuclei were labeled with DAPI (green). Bar, 20 µm. (G) Quantification of mitotic index. * P < 0.05 and ** P < 0.01. Error bars, mean ± SEM.

Furthermore, we used the MARCM and TARGET techniques to analyze whether other components of the canonical Notch pathway regulate the proliferation of imaginal ring cells. RNAi-mediated inactivation of the core transcription factor [Su(H)] and S3 Notch cleavage enzymes [Anterior pharynx defective 1(Aph-1), Nicastrin (Nct), and Presenilin (Psn)] caused lower mitotic activity in the imaginal rings (Figure S7A). In addition, the mutant MARCM clone sizes [Kuzbanian^ES24 (Kuz, S2 cleavage enzyme), aph-1^D35, Su (H)⁴⁷, and nct^R46) were all significantly decreased compared with their controls (Figure S7B), indicating that loss of canonical Notch signaling impedes cell proliferation in imaginal ring cells.

We next examined whether Notch activation is sufficient to drive cell proliferation. The active form of Notch (NICD) was misexpressed in imaginal ring cells by the Act-Gal4, NRE-Gal4, and retn-Gal4 drivers, during the third-instar larval stage (Figure S12D). Overexpression of NICD by either Act-Gal4 or retn-Gal4 drivers in salivary gland imaginal rings resulted in increased mitotic activity (13.5 and 7.5 mean number of cells per imaginal ring, respectively) as compared to control (3 and 3.1 mean number of cells per imaginal ring, respectively) (Figure 5G and Figure S5D) and oversized tissues (Figure 5, A–F). However, the number of mitotic cells was unaffected in hindgut and foregut imaginal rings. NRE-Gal4-driven NICD overexpression also caused the overproliferation phenotype in salivary gland imaginal rings [mitotic index was 8.5% compared to that of the control (5.7%)] (Figure S4D). Interestingly, NRE > NICD also increased proliferation in hindgut imaginal rings [mitotic index was 6.7% compared to that of the control (5.5%)] (Figure S4D). In addition, overexpression of Su(H) by Act^ts led to overproliferation in the salivary gland imaginal ring (Figure S7A). These results demonstrate that overexpressed Notch signaling induces excessive cell proliferation in the salivary gland imaginal ring, as well as in hindgut imaginal ring cells, depending on which Gal4 driver is used. On the other hand, cell proliferation in the foregut imaginal ring was not affected when NICD was misexpressed (Figure 5, A, D, and G, Figure S4D, and Figure S5D). Future research may determine why NICD overexpression has no effect on the foregut imaginal ring. Taken together, these observations indicate that overexpressed Notch signaling is sufficient to promote overproliferation in salivary gland and hindgut imaginal ring cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Notch is sufficient to induce overproliferation in salivary gland imaginal rings (ImRs) during the third instar. (A–F) Cell proliferation was labeled by phospho-histone 3 (PH3, red) in controls and overexpressed Notch intracellular domain (NICD) ImRs. (A and D) Foregut (FG) ImR; (B and E) hindgut (HG) ImR; and (C and F) salivary gland (SG) ImR. Nuclei were labeled with DAPI (green). Bar, 20 µm. (G) Quantification of mitotic index. *** P < 0.001. Error bars, mean ± SEM.

Notch signaling controls the initial size of imaginal rings during embryonic and young larval stages

In addition to its expression during the third instar, NRE-eGFP is also expressed in imaginal ring cells of young larvae (Figure 2, A–C). Previous studies showed that Notch and its ligands, Dl or Ser, are expressed in the developing larval foregut and salivary gland at middle embryonic stages (Haberman et al. 2003; Fuss et al. 2004). These indicate that Notch might play a role in the development of imaginal ring cells before they highly proliferate in late larval stages. To test this, we knocked down Notch during different developmental stages using the Act^ts system described above. Loss of Notch during early embryonic stages (stages 2–8) resulted in complete a loss of salivary glands and no imaginal ring cells were detected, confirming the critical role of Notch in the specification of ectodermal epithelia (Hartenstein et al. 1992).

The Act^ts system allowed us to bypass this earlier role of Notch signaling by inducing Notch-RNAi during embryonic stages 8–11 (Figure S12A). We found that the average numbers of foregut, hindgut, and salivary gland imaginal ring cells were dramatically decreased as compared with wild-type controls reared in the same conditions (Figure 6, A–F and M). For example, normally, the salivary gland imaginal ring has ∼10–11 cells in early L2, but Notch knockdown in embryonic stages 8–11 resulted in only three to four salivary gland imaginal ring cells. Consistently, when NRE^ts was used to induce Notch-RNAi expression specifically in Notch-active cells during embryonic stages 8–11, the number of imaginal ring cells was also significantly reduced (Figure S4E). Furthermore, using the Act^ts or NRE^ts driver to knock down Notch during the first-instar larval stage (Figure S12B), the number of cells in all three imaginal rings was lower; for instance, there were only six to seven salivary gland imaginal ring cells when Notch was knocked down as compared to 11 cells in the control condition (Figure 6, G–M and Figure S4E). Since cell proliferation in imaginal rings is hardly detected before the early second instar (Figure S2A), these results suggest that Notch signaling during the embryonic to first-instar larval stages is probably involved in maintaining the size of the imaginal ring cell pool during this period of development.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Notch maintains the number of imaginal rings (ImRs) during young larval stages. Notch-RNAi (RNA interference) was induced by Act^ts (Act-Gal4, Tub-Gal80^ts) during the midembryonic stage (A–F) or early first-instar larval stage (G–L). Nuclei were labeled with DAPI (white). (A, D, G, and J) foregut (FG) ImR; (B, E, H, and K) hindgut (HG) ImR; and (C, F, I, and L) salivary gland (SG) ImR. Dashed lines indicate ImR cells. Bar, 10 µm. (M and N) Quantification for cell numbers. E-Ctrl, embryonic control; E-loss 1, embryonic loss of Notch by Notch-RNAi (BL27988); E-loss 2, embryonic loss of Notch by Notch-RNAi (BL33611); L-Ctrl, larval control; L-loss 1, larval loss of Notch by Notch-RNAi (BL27988); and L-loss 2, larval loss of Notch by Notch-RNAi (BL33611). Embryonic experimental design was based on Figure S12A; Larval experimental design was based on Figure S12B. * P < 0.05, ** P < 0.01, and *** P < 0.001. Error bars, mean ± SEM.

To determine the cause for the reduced early imaginal ring cells upon Notch knockdown, we first tested whether cell death occurred. The expression of DCP1, an effector caspase that induces programmed cell death (Song et al. 1997), was assessed, but no signal was detected in imaginal rings or nearby regions when Notch was knocked down (Figure S8), suggesting that the cells lacking Notch activation did not undergo apoptosis. Another possibility for the reduced cell number is that the imaginal ring cells become polyploid cells, as in a previous study in which salivary gland imaginal ring cells switched from diploidy to polyploidy when Ser was disrupted during embryogenesis (Haberman et al. 2003). Indeed, loss of Notch during late embryonic stages increased the numbers of polyploid cells slightly but significantly in the regions of the proventriculus, hindgut, and salivary gland (Figure 6N and Figure S4F), whereas some groups of neighboring polyploid cells increased slightly, but with statistical significance, when Notch signaling was reduced in L1 (Figure 6N and Figure S4F). These findings suggest that a small number of imaginal ring cells close to the boundaries between the imaginal rings and neighboring polyploid tissues become polyploid cells when Notch signaling is not maintained during embryonic and early larval stages in imaginal rings.

Ser is the ligand that promotes Notch activation in imaginal ring cells

Dl and Ser are two major ligands for Notch signaling in Drosophila (Zeng et al. 1998). To determine the Notch ligand in the imaginal ring, we first used Dl and Ser antibodies to examine their expressions in imaginal rings, but they were undetectable. Alternatively, we used Dl-lacZ (Spradling et al. 1999; Zeng et al. 2010) and Ser-LacZ (Yamamoto et al. 2012) reporters to investigate their activation the in imaginal rings, and found that both were expressed in imaginal ring cells (Figure S9). To determine which ligand promotes Notch-induced cell proliferation in imaginal ring cells, Ser or Dl was knocked down using Act^ts individually during the third larval instar. Ser but not Dl knockdown resulted in reduced tissue sizes and lower mitotic activity in all three imaginal rings (Figure 7A), suggesting that Ser is the ligand for Notch signaling in imaginal rings. To confirm this, we knocked down fringe (fng), which encodes a glycosyltransferase that promotes Dl–Notch interaction (Brückner et al. 2000; Sasamura et al. 2003), and found that fng knockdown did not reduce cell proliferation in the imaginal rings (Figure 7A), further suggesting that Dl does not facilitate Notch activation in imaginal rings during the third-instar larval stage.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

The interaction of Notch and its ligands. Cell proliferation of imaginal rings (ImRs) during the third instar were labeled with phosphor-histone 3 (PH3). (A) The quantification of mitotic index for knockdown of Ser, Dl, or fng by Act^ts (Act-Gal4, Tub-Gal80^ts). (B) The quantification of mitotic index for overexpression of Ser and Dl^myc driven by Act^ts. * P < 0.05, ** P < 0.01, and *** P < 0.001. Error bars, mean ± SEM. FG, foregut; HG, hindgut; SG, salivary gland.

We further examined the Ser source for Notch activation in imaginal rings. Direct cell-to-cell contact between ligand- and receptor-expressing cells is essential for the activation of the Notch pathway (Lai 2004; Kopan and Ilagan 2009). Although foregut imaginal rings contact with outer proventricular cells and inner esophagus cells, the tissue structures of hindgut and salivary gland imaginal rings are only single layers. Therefore, we proposed that the ligands for Notch activation in imaginal ring cells are provided from the neighboring imaginal ring cells. To test this hypothesis, we used the Flip-out Gal4 driver (Pignoni and Zipursky 1997) to generate Ser or Dl knockdown mosaics and monitor NRE-eGFP expression levels. We found that the NRE-eGFP expression level in wild-type cells surrounded by Ser knockdown cells was significantly reduced as compared to wild-type NRE-eGFP expression (Figure 8, A–C and G), indicating that Ser knockdown clone cells were unable to induce Notch activation in their neighboring wild-type cells. These results suggest that the receptor–ligand interaction between two neighboring imaginal ring cells underlies Notch activation. However, in contrast, Dl knockdown clones did not reduce NRE-eGFP expression in their neighboring cells (Figure 8, D–G), which confirms that Dl is not the ligand to activate Notch signaling in imaginal ring cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

Ser is provided from neighboring cells. Flip-out Gal4-generated mosaic ImR with clones expressing Ser-RNAi (A–C) or Dl-RNAi (D–F). Third-instar larvae were observed. (A and D) RG ImR; (B and E) HG ImR; and (C and F) SG ImR. RNAi-expressing cells were labeled with RFP expression (red). NRE-eGFP indicated Notch activation (green). Nuclei were labeled with DAPI (blue). Dashed lines indicate nonclone cells. Bar, 20 µm. (G) The quantified ratio of eGFP intensity. The NRE-eGFP intensities in the WT cells that are completely surrounded by Ser or Dl loss-of-function clones and WT clones was measured. The intensities of eGFP in above nonclone cells were further divided by the intensities of eGFP from WT NRE-eGFP flies. *** P < 0.001. Error bars, mean ± SEM. eGFP, enhanced GFP; FG, foregut; HG, hindgut; ImR, imaginal ring; NRE, temperature-sensitive Notch responsive element; RFP, red fluorescent protein; RNAi, RNA interference; SG, salivary gland; UAS, upstream activating sequence; WT, wild-type.

Cis-inhibition modulates Notch activity in imaginal ring cells

From the above Ser and Dl knockdown mosaic studies, we found that NRE-eGFP expression was increased in the clones of a single cell layer (Figure S10), indicating that loss of ligands in the same cell results in higher Notch activation. When the ligands of Notch are expressed in the same cells as the Notch receptor, they can induce cis-inhibition, which reduces Notch activation (Sprinzak et al. 2010; del Álamo et al. 2011). This result prompted us to further examine whether Notch signaling is also regulated by cis-inhibition in the imaginal rings. To determine this, we applied the Flip-out Gal4 system to generate clones with overexpressed Dl or Ser ligand and examined NRE-eGFP expression levels in imaginal ring cells, and found that the expression of NRE-eGFP in clone cells was dramatically decreased compared to wild-type NRE-eGFP expression (Figure S11), suggesting that overexpression of both Dl and Ser ligands is sufficient to induce cis-inhibition to reduce Notch activation. We next overexpressed Dl and Ser during the third-instar larval stage and monitored cell proliferation with PH3 staining. In both cases of Dl and Ser overexpression driven by Act^ts, the mitotic index was significantly reduced (Figure 7B). Taken together, these results suggest that cis-inhibition of the Notch pathway results in the reduction of Notch activation, which further causes a lower cell proliferation rate in imaginal ring cells.