Programmed cell senescence is required for sensory organ development in Drosophila

Programmed cell senescence occurs in developing Drosophila wing disc

To investigate whether programmed cell senescence occurs during Drosophila development, we performed SA-β-gal staining in a variety of tissues in 3rd instar larvae. An intense SA-β-gal signal at pH6.0 was reproducibly observed around the center of the wing imaginal discs (Figure S1A, yellow arrowheads). A cluster of cells with elevated SA-β-gal activity began to appear at the middle L3 (ML3) stage, and the signal intensity increased as larvae grew into late L3 (LL3) (Figures 1A and S1A). Notably, no β-gal activity was observed at pH7.5 (Figure S1A), indicating that the elevated SA-β-gal signal was not simply due to increased β-gal expression but specifically due to senescence-associated β-gal activation. We further analyzed SA-β-gal activity in the wing discs at pupal stage using a more sensitive fluoresce probe SPiDER-βGal at pH6.0.¹⁶ The morphology of the wing disc changed dramatically after puparium formation (APF), but the SA-β-gal activity still existed even at 5-h APF (Figure 1A, yellow arrowheads). As later stage, wing discs recruit many hemocytes, which are strongly stained with SPiDER-βGal or SA-β-gal and thus hampers our analysis. We performed ex vivo culture of L3 wing discs in a hemocyte-free medium for 24 h and found that the cluster of cells with elevated SA-β-gal activity still existed in the wing disc at pre-pupal phase 3 (PP3) (∼9 h APF) (Figure 1A).^17,18 We also noticed that there existed a smaller cluster of cells with elevated SA-β-gal activity at the ventral hinge of the wing disc (Figures 1A and 1A′, white arrowheads). Hereafter, we call these reproducibly observed large and small clusters of cells with elevated SA-β-gal activity as dorsal senescent cells (DSCs; yellow arrowheads) and ventral senescent cells (VSCs; white arrowheads), respectively. Other tissues such as leg discs, haltere discs, and the brain, but not salivary gland, fat bodies, and eye discs (except for the hemocytes attached to the discs), showed some SA-β-gal activities (Figure S1B). We focused our analyses on the SA-β-gal-positive clusters in wing discs, as they exhibit highly reproducible, intense SA-β-gal staining with low background.

One important aspect of cellular senescence is cell-cycle arrest. Consequently, we first analyzed cell-cycle phases of DSCs and VSCs using the cell-cycle monitoring probe fly FUCCI.¹⁹ We found that 85.8% of DSCs (yellow arrowheads) were GFP (G2/M/G1)-positive and 88.0% of VSCs (white arrowheads) were GFP (G2/M/G1)-positive (Figures 1B and 1C). Both DSCs and VSCs exhibited GFP (G2/M/G1) and RFP (S/G2/M) signals (Figures 1D and 1E), suggesting that they were at G2 or M phase. In addition, the lack of EdU staining, a widely used marker for S phase, in DSCs and VSCs indicated that those cells are not in S phase (Figure 1D). Moreover, phosphohistone H3 (pH3) staining, which labels cells in M phase, failed to stain DSCs and VSCs, indicating that they are arrested in G2 phase (Figure 1E, yellow and white arrowheads). These data indicate that a major population of SA-β-gal-positive cell clusters in the wing discs undergo cell-cycle arrest in G2 phase.

To verify whether DSCs and VSCs are indeed undergoing cellular senescence, we further examined other hallmarks of senescence. First, upregulation of a CDK inhibitor Dacapo (Dap, a p21 homologue), an important feature of cellular senescence, was observed in these cell clusters as visualized by anti-Dap immunostaining (Figure 2A, quantified in Figure 2B). This suggests that besides G2 arrest, at least small populations of SA-β-gal-positive cell clusters undergo G1 arrest. Second, trimethylation of histone H3-K9, which is associated with the senescence-associated heterochromatin foci, was elevated in DSCs and VSCs (Figure 2C, quantified in Figures 2D and 2E). Third, oxidative stress, which is often associated with senescent cells, was increased in DSCs and VSCs, as visualized by the gstD-GFP probe (Figure S2A). Fourth, both DSCs and VSCs underwent cellular hypertrophy, another hallmark of senescent cells (Figures 2F and 2G). These data suggest that two clusters of cells, DSCs and VSCs, in Drosophila wing discs undergo developmentally programmed cell senescence.

We further explored whether senescence-associated secretory phenotype (SASP) is induced in the senescent clusters. We did not detect expression of unpaired (upd, an interleukin-6 [IL-6] homologue) or matrix metalloproteinase 1 (mmp1), typical SASP factors in mammals, in the wing disc (Figures S2C–S2E). However, intriguingly, an IL-6 homologue unpaired3 (upd3), which activates Drosophila JAK/STAT signaling, was significantly upregulated in DSCs and VSCs as visualized by the upd3-gal4 reporter (Figure 2H). These data suggest that SASP may also be induced in programmed senescent cells in the wing discs.

Programmed cell senescence is required for proper development of sensory organ campaniform sensilla

We next investigated the upstream trigger of programmed cell senescence in the wing discs. Epidermal growth factor receptor [EGFR]-Ras signaling, a known trigger of cellular senescence in mammals, was upregulated in DSCs and VSCs, as visualized by the Ras signaling effector pERK staining, which was co-localized with the SA-β-gal staining (Figure 3A, quantified in Figure 3B). Elevation of EGFR signaling was also confirmed by downregulation of a transcription factor Capicua²⁰ (Figure S3A). In addition, inhibition of Ras signaling by Ras^N17, a dominant-negative form of Ras, using the ci-Gal4 driver suppressed endogenous SA-β-gal activity (Figure 3C, quantified in Figure 3D). We further verified this by using the ap-GAL4 driver and found that SA-β-gal activity is indeed suppressed by Ras^N17 (Figure S3B, quantified in Figure S3C). These data suggest that Ras signaling acts as an upstream trigger of programmed cell senescence.

We next sought to identify the cell type of the clusters undergoing programmed cell senescence in the wing discs. The clusters of DSCs and VSCs are located at the anterior part of the wing disc near the anterior/posterior (A/P) boundary (Figure S3D, arrowheads), as assessed by ci-Gal4 that is expressed in the whole anterior compartment. Analysis with nub-GAL4, which drives gene expression in the wing pouch and a part of hinge, revealed that both DSCs and VSCs are located at the hinge region right next to the wing pouch (Figure S3E, arrowheads). Further analysis with DAPI staining, which visualizes the folding of the wing disc, showed that both DSCs and VSCs are located on the distal hinge (DH) (Figure 3E). Interestingly, these two locations, the regions of DSCs and VSCs, develop into the dorsal and ventral compartments of the adult wing hinge close to the wing blade, respectively: the regions called dorsal and ventral radiuses (Figure 3F). The dorsal and ventral radiuses possess sensory organs called campaniform sensilla, mechanoreceptors derived from proneural clusters in the wing disc (Figure 3G, arrows). Notably, although there are different types of campaniform sensilla, the campaniform sensilla on DSCs and VSCs commonly exhibit a specific feature denoted as circular, low-profile sensilla with a socket.²¹ We further analyzed DSCs and VSCs in the wing disc and found that they were indeed co-localized with the proneural markers achaete (Ac) and scabrous (Sca) (Figures S3F and S3G). Strikingly, suppression of programmed cell senescence by blocking Ras signaling (Ras^N17) using the ci-Gal4 driver resulted in a significantly reduced number of campaniform sensilla on both dorsal and ventral radiuses (Figure 3H, quantified in Figure 3I). These data suggest that programmed cell senescence is required for proper development of the special type of campaniform sensilla at dorsal and ventral radiuses.

Programmed cell senescence is caused by Ras-induced pnt expression

We next examined how programmed cell senescence is induced in the wing disc. In mammals, Ras-induced cellular senescence is mediated by the ETS1/2 transcription factor.²² Drosophila has an ETS1/2 homologue pointed (pnt), which encodes two protein isoforms PntP1 and PntP2.^23,24 PntP1 is transcriptionally regulated by EGFR-Ras signaling and induces downstream genes that are required for development of photoreceptor neurons in eye imaginal discs.²⁵ We have found that PntP1 is a necessary and sufficient downstream effector of Ras-induced cellular senescence in the imaginal discs.²⁶ Notably, we found that PntP1 expression was induced in both DSCs and VSCs (Figure 4A). This upregulation of PntP1 was abolished when Ras-MAPK signaling was blocked by overexpression of Ras^N17 or knockdown of MAPK rolled (rl) (Figures 4B and 4C), suggesting that PntP1 acts as a downstream effector of Ras-MAPK signaling in DSCs and VSCs. Crucially, knockdown of Pnt using the ci-Gal4 driver suppressed SA-β-Gal activity in these senescent cells (Figure 4D, quantified in Figure 4E). We further confirmed this result by using ap-GAL4 and pnt-RNAi, which significantly suppressed SA-β-Gal activity in DSCs (Figure S4A, quantified in S4B). Furthermore, blocking Ras-MAPK signaling caused the emergence of cells at S and M phase in DSCs and VSCs, as visualized by EdU labeling and pH3 staining (Figures 4F, 4G, and 4J–4K; asterisks, quantified in 4H, 4I, 4L, and 4M). We also found by pH3 and EdU staining that cell-cycle arrest in DSCs and VSCs was significantly suppressed by overexpression of String (Stg), a downstream cell-cycle regulator of Ras signaling (Figures S4C and S4F, quantified in Figures S4D, S4E, S4G, and S4H). These data indicate that programmed cell senescence in the wing discs is caused by Ras-induced PntP1 expression.

Programmed cell senescence promotes campaniform sensilla development via pnt-AS-C-mediated proneural cluster formation

We next investigated how PntP1-mediated cellular senescence promotes development of campaniform sensilla. Consistent with our data that Pnt is required for the induction of programmed cell senescence, knockdown of pnt resulted in loss of campaniform sensilla in the adults (Figure 5A, quantified in Figure 5C). It is known that the basic helix-loop-helix transcriptional factor achaete-scute (ac-sc) complex (AS-C) plays an essential role in sensory organ development in Drosophila. However, whether Pnt regulates proneural clusters in campaniform sensilla has been unclear, as most studies have focused on sensory bristles and chordotonal organs.^27,28 Significantly, blocking programmed cell senescence by pnt-RNAi using the ci-Gal4 driver strongly suppressed Ac expression at the clusters of DSCs and VSCs (Figures 5C and 5D, quantified in 5G and 5H). Notably, Ac expression along the DV boundary, which become bristles in the adults, were unaffected by pnt knockdown (Figures 5C and 5D, asterisks). Furthermore, knockdown of ac and sc using the nub-Gal4 abolished campaniform sensilla development at dorsal and ventral radiuses of the adult hinge (Figure 5E, quantified in Figure 5I) without affecting SA-β-gal activity in the programmed senescent cells (Figure 5F, quantified in Figure 5J). These data indicate that AS-C acts downstream of Pnt, which has been suggested before²⁹. Given that AS-C is generally required for the induction of sensory organs, Pnt-induced cellular senescence may be required for the formation of a special type of campaniform sensilla at dorsal and ventral radiuses of the hinge. These results suggest that programmed cell senescence contributes to the development of a special type of sensory organ in the wing hinge through Pnt-AS-C-mediated formation of proneural clusters.

Ras-Pnt-dependent programmed cell senescence is mediated by zfh2

To further confirm the role of programmed cell senescence in campaniform sensilla development, we searched for other genes that regulate programmed cell senescence in the wing disc. By performing an RNAi screening of genes abundantly expressed in the hinge region of the wing disc,³⁰ we found that knockdown of a transcription factor Zn finger homeodomain 2 (zfh2) strongly blocked programmed cell senescence (Figure 6A, quantified in Figure 6C). We further confirmed this by using ap-GAL4, which showed a blockade of programmed cell senescence in dorsal cells (Figure S5A). Notably, zfh2 is an ortholog of AT-motif binding factor 1 (ATBF1), which is often mutated or deleted in various cancers and is decreased in Ras^ACT- and Notch^ACT-driven tumors in Drosophila,³¹ suggesting that zfh2 might act as a tumor suppressor by inducing cellular senescence. Indeed, loss of zfh2 suppressed not only programmed cell senescence but also oncogene-induced senescence, as Ras^V12-induced SA-β-gal activity¹⁵ was significantly reduced by zfh2 knockdown (Figure 6B, quantified in Figure 6D). This suggests that the inhibition of programmed cell senescence by zfh2 knockdown is not the result of abnormal hinge development but is due to blockage of senescence signaling. Indeed, the expression of pnt-lacZ in DSCs and VSCs was strongly downregulated by zfh2 knockdown (Figure 6E). Furthermore, the development of campaniform sensilla was severely impaired by loss of zfh2 (Figure 6F, quantified in Figure 6G). Together, these data suggest that pnt-dependent programmed cell senescence triggered by Ras signaling is mediated by zfh2 and that programmed cell senescence plays an essential role in campaniform sensilla development (Figure 6H).

Limitations of the study

Our observations in developing Drosophila wing discs revealed that programmed cell senescence is induced by the Ras-Pnt-Zfh2 pathway, which contributes to the formation of a specific sensory organ campaniform sensilla in the adults. In the future study, detailed investigation should be performed to track senescent cells from developing larval wing discs to sensory organ development in the adult, which is currently hampered by technical limitations. It should also be addressed how programmed cell senescence contributes to the formation of campaniform sensilla.

Lead contact

For additional information or requests for resources and reagents, please contact the lead contact, Tatsushi Igaki ([email protected]).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •

  No original code was generated in this study.

• •

  All data presented in this paper will be made available upon request to the lead contact.

• •

  Additional information necessary for data reanalysis can be obtained from the lead contact upon request.

Key resources table

Experimental model and subject details

Method details

Quantification and Statistical Analysis

The percentage of antibody staining/fluorescence marker-positive cells in SA-β-gal-elevated clusters was calculated in each disc by dividing the marker-positive area by SA-β-gal-elevated area.

The intensity of Tri-methylated H3K9 was quantified in individual nuclei, which was visualized by DAPI staining, and was normalized to the corresponding DAPI intensity.

Cell sizes were measured by ImageJ based on the outline of phalloidin staining.

The normalization of SA-β-gal intensity was conducted by dividing the intensity of senescence markers by the background intensity.

The number of campaniform sensilla was counted and was compared with the mean number of campaniform sensilla on wild-type wings.

The percentage of antibody staining/fluorescence marker-positive cells in senescent clusters was calculated in each disc by dividing the marker-positive area by FUCCI GFP and RFP-positive area.

Data were analyzed using Student’s t test for single comparisons. Bar graphs represent the mean ± SE. All n values correspond to biological replicates. Statistical significance is indicated as follows: ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, and n.s. (not significant) for p > 0.05.