NF-κB signaling driven by oncogenic Ras contributes to tumorigenesis in a Drosophila carcinoma model

Abstract

Cancer-driving mutations synergize with inflammatory stress signaling pathways during carcinogenesis. Drosophila melanogaster tumor models are increasingly recognized as models to inform conserved molecular mechanisms of tumorigenesis with both local and systemic effects of cancer. Although initial discoveries of the Toll-NFκB signaling pathway in development and immunity were pioneered in Drosophila, limited information is available for its role in cancer progression. Using a well-studied cooperative RasV12-driven epithelial-derived tumor model, we here describe functions of Toll-NF-κB signaling in malignant RasV12, scrib- tumors. The extracellular Toll pathway components ModSP and PGRP-SA and intracellular signaling Kinase, Pelle/IRAK, are rate-limiting for tumor growth. The Toll pathway NFκB protein Dorsal as well as cactus/IκΒ show elevated expression in tumors with highest expression in invasive cell populations. Oncogenic RasV12, and not loss of scribble, confers increased expression and heterogenous distribution of two Dorsal isoforms, DorsalA and DorsalB, in different tumor cell populations. Mechanistic analyses demonstrates that Dorsal, in concert with the BTB-transcription factor Chinmo, drives growth and malignancy by suppressing differentiation, counteracting apoptosis, and promoting invasion of RasV12, scrib- tumors.

Fig 1. Extracellular and intracellular Toll signaling components contribute to Ras^V12; scrib^-/- tumor growth.

(A) A transcriptomics analysis comparing the profile of Ras^V12; scrib^-/- tumor cells to GMR>GFP control cells in the eye-antennal disc (EAD) identified 364 genes that are upregulated in the tumors. Among them, PGRP-SA, GNBP2, easter, spz3, and spz5 are genes taking part in the extracellular machinery leading to Toll pathway activation. PGRP-SA is highlighted in green because it is also found upregulated in two other tumor models. (B) Simplified version of Toll pathway activation upon bacterial infection. PGRP-SA (Peptidoglycan recognition protein SA) is a carboxypeptidase which recognizes and binds to peptidoglycans upon bacterial infection. Follows the formation of a trimeric complex together with a member of the GNBP family (Gram Negative Binding Protein) and the serine protease (ModSP). This complex initiates a multi-step proteolytic cascade that ultimately leads to the cleavage and activation of the ligand Spz. After the binding of Spz to the Toll receptor, the pathway is activated intracellularly through the activation and formation of a complex composed of the adaptor proteins Myd88 and Tube as well as the serine-threonine protein kinase Pelle. Pelle subsequently phosphorylates the inhibitor cactus, the homolog of mammalian IkappaB, which is targeted to proteasome degradation. As Cactus sequesters the NF-kB Dorsal into the cytoplasm, its degradation triggers the translocation of Dorsal into the nucleus and the transcription of its target genes. (C) Cartoon of the genetic setting used in this study for manipulating Ras^V12; scrib^-/- tumors. GFP-labeled tumors are generated from randomly selected single epithelial cells from the EAD. They grow and fuse to form large tumors that start invading the central nervous system through the optic lobe first and the ventral nerve cord later. Tumor cells also sometimes reach the leg discs. (D) Representative confocal pictures and quantifications of the mean tumor volumes of GFP-labeled control tumors (n = 21, m = 3,71 × 10⁷ µm³, SD = ±0,47 × 10⁷ µm³) and PGRP-SA Ras^V12; scrib^-/- tumors (n = 17, m = 2,11 × 10⁷ µm³, SD = ±0,19 × 10⁷ µm³) at Day 8 after egg laying, statistical significance was determined with an unpaired T test with Welch’s correction. (E) Representative confocal pictures and quantifications of the mean tumor volumes of GFP-labeled control (n = 20, m = 8,46 × 10⁷ µm³, SD = ±1,77,53 × 10⁷ µm³) and ModSP^IR Ras^V12; scrib^-/- Dcr2 tumors (n = 30, m = 5,07 × 10⁷ µm³, SD = ±1,00 × 10⁷ µm³) at Day 8 (29°C), statistical significance was determined with an unpaired T test with Welch’s correction. (F) Representative confocal pictures and quantifications of the mean tumor volumes of GFP-labeled control tumors (n = 10, m = 2,64 × 10⁷ µm³, SD = ±0,76 × 10⁷ µm³) and pelle^IR Ras^V12; scrib^-/- tumors (n = 10, m = 0,78 × 10⁷ µm³, SD = ±0,22 × 10⁷ µm³) at Day 6 (29°C), statistical significance was determined with a Mann–Whitney test. Scale bars = 50 µm.

Fig 2. Two splicing variants of the NF-kB Dorsal are aberrantly expressed in Ras^V12-driven tumors.

(A) Immunostainings against Dorsal-A (DlA in green) and Dorsal-B (DlB in magenta) in BFP-labeled wt clones (n = 20 EADs), scrib^IR tumors (n = 23 EADs), Ras^V12 tumors (n = 20 EADs), and Ras^V12; scrib^IR tumors (n = 24 EADs) at Day 6. DlA and DlB accumulation is observed only in Ras^V12-driven tumors and appears mainly mutually exclusive. (B) Immunostaining against DlA (green) and DlB (magenta) in BFP-labeled Ras^V12; scrib^IR tumors (white) (n = 24 EADs) and close-ups showing that DlA and DlB are not expressed in the same subset of cells. Colocalization analysis assessed with the Pearson’s correlation coefficient reveals a poor colocalization between DlA and DlB in the tumors (n = 24, m = 0,11, SD = ±0,08). (C) Heat map of DlA intensity within a representative BFP-labeled Ras^V12; scrib^IR tumor-bearing EAD at Day 6. The tumor is outlined with the straight white line. An arbitrary intensity threshold set at 250 (gray line on the graph) allows the clear segregation of a DlAhigh (“tumor High”) and a DlAlow (“tumor Low”) cell population within the tumor. Quantifications of the mean DlA intensity of the “tumor High” (n = 23, m = 458,7, SD = ±28,4), “tumor Low” (n = 23, m = 135,2, SD = ±7,0), and wt tumor microenvironment (TME) (n = 23, m = 82,2, SD = ±6,7) demonstrate a general elevation of DlA levels within the tumor compared with wt cells. Statistical significance was determined with a Tukey’s multiple comparison test. (D) Heat map of DlB intensity within a representative RasV12; scribIR tumor-bearing EAD at Day 6. The tumor is outlined with the straight white line. An arbitrary intensity threshold set at 10,000 (gray line on the graph) allows the clear segregation of a DlBhigh (“tumor High”) and a DlBlow (“tumor Low”) cell population within the tumor. Quantifications of the mean DlB intensity of the “tumor High” (n = 24, m = 20,331,9, SD = ±3,108,7), “tumor Low” (n = 24, m = 3,955,8, SD = ±584,5), and wt TME (n = 23, m = 4,660,9, SD = ±1,187,3). Normalization was done by subtracting the background intensity. Statistical significance was determined with a Dunn’s multiple comparison test. Scale bars = 20 µm.

Fig 3. Cell type-dependent Dorsal isoforms accumulation correlates with tumor neurogenesis and apoptosis.

(A) Representative confocal pictures of BFP-labeled Ras^V12; scrib^IR tumors, expressing a DlA::GFP fusion protein at endogenous levels (green), overtime (Day 4 n = 18 EADs, Day 5 n = 14 EADs, Day 6 n = 18 EADs, Day 7 n = 17 EADs, Day 8 n = 20 EADs, and Day 9 n = 11 EADs) stained for the pan-neuronal marker Elav (magenta). The cephalic complex is outlined with the straight gray line and the tumor is outlined with the straight white line on the close-ups. Yellow arrows point at neurons that are negative for DlA. Scale bar = 50 µm. (B) Representative confocal pictures of GFP-labeled cherry^IR Ras^V12; scrib^IR (n = 8 EADs) at Day 6 (29°C) stained for Elav (green) and DlA (magenta). The tumor is outlined with the straight white line. DlA and Elav appears mutually exclusive. Scale bar = 20 µm. (C) Representative confocal pictures of GFP-labeled cherry^IR Ras^V12; scrib^IR (n = 8 EADs) at Day 6 (29°C) stained for Elav (green) and DlB (magenta). The tumor is outlined with the straight white line. DlB and Elav are expressed in the same tumor cells. Scale bar = 20 µm. (D) Representative confocal pictures of GFP-labeled cherry^IR Ras^V12; scrib^IR control tumors (n = 62 EADs) at Day 6 (29°C) stained for Elav (green) and cDcp1 (magenta). The tumor is outlined with the straight white line. cDcp1⁺ cells are mostly found in areas of the tumor where neurons are present where we often observe neuron apoptosis (yellow arrows). Scale bar = 20 µm. (E) Representative confocal pictures of GFP-labeled Ras^V12; scrib^IR control tumors (n = 7 EADs) and Diap1^OE Ras^V12; scrib^IR (n = 9 EADs) stained for cDcp1 (magenta) at Day 6. Scale bar = 50 µm. (F) Representative confocal pictures of GFP-labeled Ras^V12; scrib^IR control tumors (n = 13 EADs) and Diap1^OE Ras^V12; scrib^IR (n = 7 EADs) stained for DlA (green) and DlB (magenta) at Day 6. Scale bar = 20 µm.

Fig 4. Dorsal represses differentiation and apoptosis in Ras^V12 scrib^RNAi tumors while promoting invasive behavior.

(A) Representative confocal pictures and quantifications of the mean tumor volumes of GFP-labeled Cherry^IR Ras^V12; scrib^IR control tumors (n = 36, m = 8,32 × 10⁷ µm³, SD = ±1,31 × 10⁷ µm³), dorsal^IR-1 Ras^V12; scrib^IR tumors (n = 25, m = 1,97 × 10⁷ µm³, SD = ±1,00 × 10⁷ µm³), Cherry^IR Ras^V12; scrib^IR Dcr2 control tumors (n = 43, m = 8,52 × 10⁷ µm³, SD = ±1,64 × 10⁷ µm³), and dorsal^IR-2 Ras^V12; scrib^IR Dcr2 tumors (n = 37, m = 4,71 × 10⁷ µm³, SD = ±1,85 × 10⁷ µm³) at Day 12 (29°C), statistical significance was determined with an unpaired T test with Welch’s correction. Scale bar = 50 µm. (B) Representative confocal pictures and quantifications of the proliferation ratio of GFP-labeled Cherry^IR Ras^V12; scrib^IR control tumors (n = 36, m = 1,08%, SD = ±0,41%) and dorsal^IR-1 Ras^V12; scrib^IR tumors (n = 35, m = 1,39%, SD = ±0,83%) at Day 8 (29°C). Proliferation was detected through immunostainings against the mitotic marker pH3 (phospho-Histone 3, in magenta). Here, the proliferation ratio is defined as the ratio between the pH3 area within the tumor and the area of the tumor itself. Scale bar = 20 µm. (C) Representative confocal pictures and quantifications of the differentiation ratio of GFP-labeled Cherry^IR Ras^V12; scrib^IR control tumors (n = 63, m = 8,29%, SD = ±2,87%) and dorsal^IR-1 Ras^V12; scrib^IR tumors (n = 57, m = 24,40%, SD = ±7,09%) at Day 8 (29°C). Differentiation was detected through immunostainings against the pan-neuronal marker Elav (magenta). Here, the differentiation ratio is defined as the ratio between the Elav area within the tumor and the area of the tumor itself. Scale bar = 20 µm. (D) Representative confocal pictures and quantifications of the apoptosis ratio of GFP-labeled Cherry^IR Ras^V12; scrib^IR control tumors (n = 63, m = 0,71%, SD = ±0,29%) and dorsal^IR-1 Ras^V12; scrib^IR tumors (n = 57, m = 1,45%, SD = ±0,46%) at Day 8 (29°C). Apoptosis was detected through immunostainings against the cleaved effector caspase, cDcp1 (cleaved-Death Caspase 1, in magenta). Here, the apoptosis ratio is defined as the ratio between the cDcp1 area within the tumor and the area of the tumor itself. Statistical significance was determined with an unpaired T test with Welch’s correction. Scale bar = 20 µm. (E) Z-projections of representative confocal pictures of size-matched GFP-labeled Cherry^IR Ras^V12; scrib^IR tumors (n = 3 EADs) and dorsal^IR-1 Ras^V12; scrib^IR tumors (n = 8 EADs) at Day 12 (29°C), stained for Hoechst (magenta). Scale bars = 50 µm. The cephalic complexes (and leg discs [LD] on the small inserts) are outlined with the straight gray line. A cartoon represents the three invasion parameters assessed in this study: the VNC (Ventral Nerve Cord) invasion score, ranged between c and 1, is greater the further tumor cells migrate toward the tip of the VNC (green arrow), indicated as “VNC=” on the Z-projection images; the frequency of LD invasion (small insert with LD confocal pictures, Scale bars = 20 µm) and the fusion of distinct Eye-Antennal Disc (EAD) tumors (white arrow, dashed line indicates the visible separation of both neighbor EADs). For the VNC score, statistical significance was determined with Dunnett’s T3 multiple comparisons test. (F) Z-projections of representative confocal pictures of GFP-labeled Ras^V12; scrib^IR (n = 15 EADs) and dorsal^OE Ras^V12; scrib^IR tumors (n = 24 EADs) at Day 6, stained for Hoechst (magenta). The cephalic complexes (and LD on the small inserts) are outlined with the straight gray line. Quantification of the mean tumor volumes of GFP-labeled Ras^V12; scrib^IR (n = 9, m = 9,71 × 10⁶ µm³, SD = ±1,53 × 10⁶ µm³) and dorsal^OE Ras^V12; scrib^IR tumors (n = 11, m = 6,52 × 10⁶ µm³, SD = ±1,12 × 10⁶ µm³) at Day 6. OL: Optic Lobe. Statistical significance was determined with an unpaired T test with Welch’s correction. Scale bars = 50 µm.

Fig 5. Dorsal amplifies JNK signaling.

(A) Representative confocal pictures of BFP-labeled Ras^V12; scrib^IR tumors, expressing the JNK activity reporter, TRE-eGFP (green) and stained for DlA (magenta). The tumor is outlined with the straight white line. Colocalization analysis assessed with the Pearson’s correlation coefficient reveals a weak colocalization between DlA and JNK activity in the tumors (n = 31, m = 0,23, SD = ±0,08). Scale bar = 20 µm. (B) Representative confocal pictures of GFP-labeled cherry^IR Ras^V12; scrib^IR tumors stained for pJNK (green) and DlB (magenta) at Day 6 (29°C). The tumor is outlined with the straight white line. Colocalization analysis assessed with the Pearson’s correlation coefficient reveals a good colocalization between DlB and pJNK in the tumors (n = 25, m = 0,41, SD = ±0,06). Scale bar = 20 µm. (C) Representative confocal pictures and quantifications of the mean MMP1 intensity of GFP-labeled Ras^V12; scrib^IR tumors (n = 11, m = 1,00, SD = ±0,06), dorsal^OE Ras^V12; scrib^IR tumors (n = 11, m = 2,85, SD = 0,57), and modSP^OE Ras^V12; scrib^IR tumors (n = 10, m = 1,49, SD = ±0,32) at Day 6. Intensity normalization was done by subtracting the background intensity on each picture and finally dividing each data point by the mean intensity value for the control tumors for each replicate. Statistical significance was determined with Dunnett’s T3 multiple comparisons test. Scale bars = 50 µm.

Fig 6. Dorsal and chinmo mutually influence each other’s expression, driving tumor growth while inhibiting differentiation.

(A) Representative confocal pictures and quantifications of control clone chinmo levels (mean gray value) of GFP-labeled wt clones (n = 18, m = 28,39, SD = ±4,129) and dorsal^OE clones (n = 19, m = 36,66, SD = ±6,701) at Day 6. Chinmo (magenta) upregulation upon Dorsal overexpression in clones is visible in the insets bellow each image. Statistical significance was determined with an Unpaired t test with Welch’s correction. Scale bar = 50 µm. (B) Representative confocal images and quantifications of clone dorsal levels (mean gray value) of GFP-labeled Ras^V12; scrib^IR tumors (n = 34, m = 36,09, SD = ± 4,009) and chinmo^OE Ras^V12; scrib^IR tumors at Day 6 (n = 45, m = 53,93, SD = ± 7,683). Single channel image below shows a clear upregulation of Dorsal (magenta) upon chinmo overexpression in Ras^V12; scrib^IR tumors. Statistical significance was identified with an Unpaired t test with Welch’s correction. Scale bar = 50 µm. (C) Representative confocal pictures and quantifications of tumor volumetrics for GFP-labeled Cherry^IR Ras^V12; scrib^IR tumors (n = 17, m = 3,429,819 µm³, SD = ±942,847 µm³) and chinmo^IR Ras^V12; scrib^IR tumors (n = 11, m = 1,502,713 µm³, SD = ±797,983 µm³) at Day 9. The cephalic complex is counterstained with Hoechst (magenta). Statistical significance was assessed with a Mann–Whitney test. Scale bar = 100 µm. (D) Representative confocal images and quantifications of tumor volumetrics and clone elav+ volume coverage for GFP-labeled (left) Cherry^IR Ras^V12; scrib^IR tumors (Volumetrics: (n = 30, m = 376,663 µm³, SD = ±142,575 µm³); Elav+ coverage: (n = 30, m = 27,28, SD = ±12,19)) and chinmo^IR Ras^V12; scrib^IR tumors (Volumetrics: (n = 25, m = 247,329 µm³), SD = ±69,250 µm³; Elav+ coverage: (n = 25, m = 44,25, SD = ±10,19)). Statistical significance for volumetrics was identified with an Unpaired t test with Welch’s correction and a Mann–Whitney test was used for clone elav+ coverage quantifications. Insets reveal an increase of elav+ (magenta) cells within the tumor upon chinmo^IR; (right) Ras^V12; scrib^IR tumors (Volumetrics: (n = 34, m = 439,561 µm³, SD = ±133,270 µm³); Elav+ coverage: (n = 44, m = 16,70, SD = ± 4,944)) and chinmo^OE Ras^V12; scrib^IR tumors (Volumetrics: (n = 45, m = 1,118,657 µm³, SD = ± 247,752 µm³); Elav+ coverage: (n = 51, m = 0,4,118, SD = ± 0,6,376)). Statistical significance for volumetrics was identified with an Unpaired t test with Welch’s correction and a Mann–Whitney test was used for clone elav+ coverage quantifications. Insets show a decrease of elav+ (magenta) cells within the tumor upon chinmo^OE. EAD nuclei were counterstained with Hoechst (blue). Scale bars = 50 µm.

Fig 7. Working model.