Two-tiered control of epithelial growth and autophagy by the insulin receptor and the ret-like receptor, stitcher

Abstract

Body size in Drosophila larvae, like in other animals, is controlled by nutrition. Nutrient restriction leads to catabolic responses in the majority of tissues, but the Drosophila mitotic imaginal discs continue growing. The nature of these differential control mechanisms that spare distinct tissues from starvation are poorly understood. Here, we reveal that the Ret-like receptor tyrosine kinase (RTK), Stitcher (Stit), is required for cell growth and proliferation through the PI3K-I/TORC1 pathway in the Drosophila wing disc. Both Stit and insulin receptor (InR) signaling activate PI3K-I and drive cellular proliferation and tissue growth. However, whereas optimal growth requires signaling from both InR and Stit, catabolic changes manifested by autophagy only occur when both signaling pathways are compromised. The combined activities of Stit and InR in ectodermal epithelial tissues provide an RTK-mediated, two-tiered reaction threshold to varying nutritional conditions that promote epithelial organ growth even at low levels of InR signaling.

Figure 1. Stit is required for optimal growth of the wing epithelium.

(A) Stit inactivation in the dorsal wing compartment using ap-GAL4-induced FRT stit mutant clones or expression of either Stit^KD or stit-IR caused an upwardly bent adult wing. (B) Wing cell area decreased in stit mutant clones compared to wild-type clones. (C) The expression of Stit^KD in the dorsal wing compartment led to a 33% reduction of the dorsal/ventral (D/V) cell number ratio (>15,000 cells from 10 Stit^KD pupal wings counted) relative to GFP expressing controls (8 animals, >15,000 cells counted). Student's t test, p<0.005. (D) Expression of stit-IR in the posterior compartment (en-GAL4) caused a backwardly bent wing. (E) The posterior compartment of en>stit-IR wings was reduced in size due to a reduction in both the total cell number and wing cell area relative to GFP-expressing control wings. (F) hs-flp;actin>CD2,stop>GAL4 (AFG4)-generated clones revealed a general reduction of proliferation and an increase in cell doubling time (CDT) when insulin (InR^DN) or amino acid signaling (RagA^DN) was reduced relative to clones expressing GFP alone (48 h after heatshock). A similar effect was observed in Stit^KD-expressing clones, whereas simultaneous reduction of Stit and InR signaling led to a further reduction in proliferation. The number of clones examined (n) for each genotype is indicated. (G) Analysis of Phospho-histone 3 (PH3)-positive mitotic profiles in third instar larval wing discs revealed a general (dorsal, D, and ventral, V, in graph) reduction in the number of mitotically active cells within ap>Stit^KD wing discs compared to control, Student's t test, p<0.005 when comparing either control compartment with either Stit^KD compartment. The number of discs examined (n) is indicated. Scale bar, 50 µm. (H) 10 min of EdU incorporation did not show a compartment-specific change in cells entering into S-phase in ap>Stit^KD discs versus ap>GFP control discs. However, the overall (both dorsal and ventral) labeling was more sparse in Stit^KD wings, although the D/V ratio was close to 1. The number of wing/discs examined (n) is indicated. All error bars indicate standard deviation. The nonautonomous compensatory growth effect is explored further in Figure S1.

Figure 2. Genetic interactions of stit with mutations affecting the PI3K-I/TORC1 pathway.

(A) InR receptor inactivation in the dorsal compartment (ap>InR^DN) reduced total wing size and caused a minor wing bending primarily around the margin. Co-reduction of Stit activity (by Stit^KD) further reduced wing size and gave an upwardly bent wing in addition to the bending at the margin. The stit-IR wing bending phenotypes could be suppressed by co-expression of a constitutively active form of the InR (InR^CA), Rheb, Akt, and constitutively active dS6K (dS6K^STQE), indicating that stit acts via or in parallel to the PI3K-I/TORC1 cassette. Dorsal expression of Rheb led to a downwardly bent wing, indicative of overgrowth. This phenotype was also observed upon mild expression of wild-type stit (see Figure S2). (B) Overexpression of stit by MS1096-GAL4 led to a severe crumpled wing phenotype. This could be further enhanced by co-expression of PI3K-I or suppressed by co-expression of PTEN. See also Table S1.

Figure 3. Stit activates PI3K-I to support growth and suppress starvation-induced autophagy.

(A) Clonal overexpression in the fat body (marked by RFP, asterisk) of PI3K-CaaX led to increased recruitment of GFP-tagged PH probe (tGPH) to the fat body cell plasma membrane and cell enlargement. This was more evident under starvation conditions, where membrane-bound levels of GFP-PH declined in neighboring cells. (B) Clones expressing stit were slightly enlarged, more rounded, and had higher membrane-bound GFP-PH levels than neighbor cells. This persisted upon starvation. (C) Clonal overexpression of PI3K-CaaX or (D) stit (marked by GFP, asterisk) in the fat body of larvae expressing a Cherry-tagged Atg8a reporter expressed under the control of a fat body promoter showed that both PI3K-CaaX and Stit can block the starvation-induced punctate accumulation of Ch::Atg8a. Quantified in (G). (E) Clonal overexpression of either PI3K-CaaX (E) or stit (F) in larvae during programmed autophagy (P.A.) demonstrated that both can block Ch::Atg8a accumulation in the expressing cells. Quantified in (G). Feeding the TORC1 inhibitor rapamycin to larvae expressing PI3K-CaaX (E) or stit (F) in clones reverted the Stit-mediated block of Ch::Atg8a puncta accumulation. Quantified in (G). (G) The intensity of Ch::Atg8a in AFG4-positive cells was measured and compared to the nearest neighbor cells to calculate fold changes where a value of 1 (red hatched line) indicates no difference to the normal autophagy response to each condition (starved, fed, P.A./developmental, or rapamycin induced) observed in wild-type neighbor cells. Stit and PI3K-I could both suppress programmed and starvation-induced Ch::Atg8a accumulation/intensity increase. * indicates Student's t test scores of significance (p<0.005) between overexpressing cells and wild-type neighbor cells, while inset Student's t test scores indicate p values of the difference in response of stit-expressing cells/wild-type neighbor cells between conditions—that is, starved and fed. The number of transgene-expressing cells/wild-type neighbor cells where Ch::Atg8a intensity was measured (n) is indicated. Error bars indicate standard deviation. Scale bar, 50 µm. See also Figure S3.

Figure 4. The PI3K-I/TORC1signalling cassette is required for Stit-dependent protection against starvation.

Clonal expression (cells marked by GFP/hatched line) of stit and/or transgenes interfering with TORC1 signaling components in the fat body of starved (5 h) larvae expressing Ch::Atg8a under the control of a fat body promoter (A) PI3K-I-expressing cells have larger nuclei, are protected from starvation-induced Ch::Atg8a puncta formation, and maintain high p-4E-BP levels compared to wild-type neighbors. (B) stit-expressing cells behave like PI3K-I-expressing cells in (A). (C) PI3K-I^DN-expressing cells are smaller and show Ch::Atg8a autophagic puncta and low levels of p-4E-BP (indistinguishable from neighbor cells). (D) stit and PI3K-I^DN co-expression resulted in smaller cells with autophagic puncta and low levels of p-4E-BP following starvation, indicating PI3K-I signaling lies downstream of Stit. (E) Akt-IR, (G) RagA^DN, and (I) TOR^TED-expressing cells are smaller, readily form autophagic puncta, and had lower or similar p-4E-BP levels to neighbor cells. (F) Akt-IR, (H) RagA^DN, or (J) TOR^TED co-expression in cells expressing stit inhibited the increase in cell size, resistance to starvation-induced autophagy (Ch::Atg8a puncta), or maintenance of p-4E-BP levels, indicating that these members of the PI3K-I/TORC1 signaling cassette are required for Stit-dependent starvation resistance. Scale bars, 25 µm. (K) Plot of the ratios of p-4E-BP labeling intensities in cells expressing the transgene to wild-type neighbor cells. The number of overexpressing cells/wild-type nearest neighbor cells counted (n) is given. * indicates Student's t test p values <0.005 between transgene-expressing and wild-type neighbor cells. Co-expression of RagA^DN together with stit blocked stit-supported growth under starvation. This effect was so strong on p-4E-BP levels that it reduced levels far beyond the starved wild-type cell levels, giving significant differences. Inset p values compare differences in preservation of p-4E-BP signal between stit-expressing cells/wild-type neighbor cells and stit and transgene co-expressing cells/wild-type cells. Error bars indicate standard deviation.

Figure 5. Stit and the insulin receptor cooperate to activate the PI3K-I pathway.

(A) Clones of cells (marked with RFP) expressing PI3K-CaaX in the larval wing discs led to an increased recruitment of the tubulin::GFP-PH probe (tGPH) to the plasma membrane detectable following 24-h starvation. Inset X–Z section (3–4 µm) and line intensity graph is through/over the indicated region in merged panels. RFP (red) and GFP (green) intensities are represented as line intensity graphs. Peaks represent membrane-localized GFP, while troughs correspond to cytoplasmic signal. GFP-PH membrane-associated intensity in the PI3K-CaaX-expressing region is higher than the nonexpressing region. (B) MS1096 expression of stit (red) causes increased GFP-PH membrane recruitment in the expressing cells. (C) Clones of cells expressing Stit^KD or (D) InR^DN alone did not decrease membrane GFP-PH localization. (E) Clones expressing both Stit^KD and InR^DN had lower levels of membrane-localized GFP-PH than wild-type neighbors. Arrowheads denote clone boundaries. Insets show X–Z projections spanning the clones. All images are thin (8 µm) confocal projections. Scale bars, 10 µm.

Figure 6. Stit and InR cooperatively govern TORC1 activity.

Anti-p-dS6K labeling of third instar larval wing discs revealed discrete puncta lying basally within cells. See Figure S5 for control experiments addressing labeling specificity. Transgenes were expressed by ap-GAL4 and expressing cells were identified by GFP (A–H and J) or by Stit detection (I, K). Arrows or arrowheads mark the D/V compartment boundary. D/V ratios of p-dS6K intensities are quantified in (L). (A) Expression of RagA^DN led to a reduction in basal p-dS6K puncta in the dorsal compartment. X–Z sections of the boxed area are shown in (B). (B–F) X–Z sections of discs expressing RagA^DN, raptor-IR, InR^DN, stit-IR, or Stit^KD in the dorsal compartment. The basal p-dS6K signal of interest is boxed in (B) and the dorsal (D) and ventral (V) compartments indicated. Dividing cells at the top of the panel also label strongly (see Figure S6). The dorsal compartment lies to the right in all X–Z sections and arrowheads mark the compartment borders. (G) Expression of stit caused overgrowth of the dorsal region of the disc accompanied by a change in disc morphology, preventing analysis of basal p-dS6K levels. (H) Co-expression of Stit^KD and InR^DN reduced the p-dS6K signal, to a similar extent as either inactivation alone; see (L) for quantification. (I) Co-expression of stit with Stit^KD leads to a reduction of the stit overgrowth phenotype but gave excessively high p-dS6K levels precluding quantification. (J) Rheb expression increased p-dS6K levels within the dorsal region. (K) Co-expression of Rheb with Stit^KD reverted the decrease in p-dS6K levels resulting from Stit inactivation (see L). Scale bar, 50 µm. (L) The intensity of basal p-dS6K within both dorsal and ventral compartments was measured and the D/V ratio for each genotype calculated. Student's t test showed p<0,001 when wild-type D/V ratios were compared with all genotypes except when compared to ap>Stit^KD>Rheb (p<0.01). ap>Stit^KD>Rheb was significantly different from ap>Stit^KD (p = 0.0015). (M) Immunoblots of total lysates prepared from third instar larval wing discs of wild-type, wild-type starved, and stit mutant larvae. The ratio of p-dS6k/dS6K was reproducibly lower (40% average reduction (45% in M), 15% standard error, 5 independent experiments, 7 samples) in stit mutant discs compared to wild-type fed animals. Arrowhead indicates the band recognized by the anti p-dS6K antibody.

Figure 7. Stit and InR are interchangeably sufficient to block autophagy.

ptc-GAL4 was used to drive transgenes and GFP in the wing discs of third-instar larvae expressing a Cherry-tagged Atg8a autophagy reporter under the control of its own promoter (Ch::Atg8a). GFP expression did not affect the formation of Ch::Atg8a puncta (see Figures S6A and 7G). (A) Expression of a dominant negative form of TOR (TOR^TED) led to a robust induction of autophagic puncta compared to wild-type neighboring tissue. Quantified in (G). (B) Expression of PTEN induced the formation of autophagic puncta, indicating PI3K-I signaling normally holds the autophagic machinery dormant. Expression of either (C) InR-IR or (D) Stit^KD in the ptc domain did not trigger an increase in punctate Ch::Atg8a accumulation. (E) Co-expression of InR-IR together with Stit^KD led to a robust increase in the formation of autophagic Atg8a puncta. Comparable levels of autophagic puncta were observed upon co-reduction of InR/Stit signaling via double RNAi or double dominant negative approaches (see G and Figure S6). (F) Rearing Stit^KD-expressing animals on nutrient-restricted (N.R.) low-energy food led to a notable induction of autophagic puncta, quantified in (G). Scale bar, 50 µm. (G) Graph displaying the ratios obtained from comparison of Ch::Atg8a intensities within the ptc-GFP stripe with the flanking tissue. Student's t test scores for significant differences to control are indicated.

Figure 8. A model for the different modes of autophagy regulation in tissues that express stit.

(A) A model depicting Stit and its intersection with the InR/PI3K-I/TORC1 pathway. As Stit is required and sufficient for PI3K-I activation, we place it in parallel to the InR. While InR is known to signal through Chico (IRS) and Lnk, the RTK adaptor that couples Stit to PI3K-I signaling is unknown (X). (B) The presence of InR or stit is sufficient to block autophagy in the wing, while both are required for optimal wing growth. When reared on low-energy food, Stit is required for repressing autophagy. These findings predict a two-tiered model for the switch between anabolism to catabolism in epithelial tissues.