Mutual repression between JNK/AP-1 and JAK/STAT stratifies cell behaviors during tissue regeneration

Epithelial repair relies on the activation of stress signaling pathways to coordinate cellular repair behaviors. Their deregulation is implicated in chronic wound and cancer pathologies. Despite such translational importance, an understanding of how spatial patterns of signaling pathways and repair behaviors arise in damaged tissues remains elusive. Using TNF-α/Eiger-mediated inflammatory damage to Drosophila imaginal discs, we uncover that JNK/AP-1 signaling cells act as paracrine organizers and initiate a mutual repression network that spatially segregates JNK/AP-1 and JAK/STAT signaling cells into distinct populations. While JNK/AP-1 signaling cells produce JAK/STAT-activating Upd ligands, these signal-sending cells suppress activation of JAK/STAT via Ptp61F. Conversely, responding cells with activated JAK/STAT suppress JNK activation via Zfh2. The resulting bistable segregation of signaling domains is associated with distinct cellular tasks and regenerative potential. While JNK/AP-1 signaling cells at the wound center act as paracrine organizers, their cell cycle is senescently arrested. Thus, compensatory proliferation occurs exclusively in JAK/STAT signaling cells at the wound periphery. This spatial stratification is essential for proper tissue repair, as co-activation of JNK/AP-1 and JAK/STAT in the same cells creates conflicting inputs on cell cycle progression, leading to excess apoptosis of senescently arrested organizer cells. Finally, we demonstrate that bistable spatial segregation of JNK/AP-1 and JAK/STAT drives senescent and proliferative behaviors in transient as well as chronic tissue damage models, and importantly, in RasV12, scrib tumors under the influence of JNK/AP-1 activity. Revealing this previously uncharacterized regulatory network between JNK/AP-1, JAK/STAT and associated cell behaviors have important implications for our conceptual understanding of tissue repair, chronic wound pathologies and tumor microenvironments, where both pathways are strongly implicated.


Introduction
mechanisms and molecular effectors that resolve these conflicting cell behaviors downstream 84 of JNK/AP-1 signaling in the same tissues remains elusive. 85 JAK/STAT signals strongly correlate with spatial segregation of regenerative cell behaviors, 157 specifically G2 stalling and proliferation, into spatially distinct signaling domains. We wanted to first understand whether JNK/AP-1 directly represses JAK/STAT activity and 163 thereby restricts it to the wound periphery. We therefore tested if clonal expression of JNK/AP-164 1 directly represses JAK/STAT activity, by co-expressing a constitutively activated form of the 165 JNKK Hep along with p35 to prevent cell death [75,76]. Indeed, hep act clones cell-166 autonomously repressed JAK/STAT activity ( Fig interact [83][84][85] or are integrated to define a cell's signaling state and establish distinct cellular 217 outcomes [86,87]. How regulatory networks can also organize cellular behaviors to influence 218 tissue-level outcomes has been extensively described using mathematical modeling [88][89][90]. 219 220 To characterize the rules of the regulatory network between JNK/AP-1 and JAK/STAT that 221 ultimately drives emergence of spatial segregation to organize tissue stress responses, we 222 developed a mathematical model using partial differential equations. These described the 223 spatio-temporal evolution of JNK/AP-1 and JAK/STAT activation depending on relative 224 concentrations of components in the system ( Fig. 3A-C) [86,91]. We defined a set of 225 experimentally determined rules, namely: (1) JNK/AP-1 effectors propagate JNK/AP-1 via 226 production of paracrine factors, such as egr or diffusible ROS [28,54,92]; (2) JNK/AP-1 227 effectors activate JAK/STAT via production of paracrine factors, such as upd's [20,28,49,51]; 228 (3) Both pathways use positive feedback loops to enhance and stabilize their own activation 229 [42,55,56,93] and (4) JAK/STAT represses JNK/AP-1 cell-autonomously, via the ZEB-1 homologue Zfh2 (Fig. S3A-C) [51]. To simplify the model (Fig. 3B), we excluded negative 231 segregation of these two pathways. The spatial segregation of JNK/AP-1 and JAK/STAT via 268 mutual repression allows the tissue to maintain a JNK-dependent G2 stall in the presence of 269 pro-mitogenic signals, and restricts proliferation to a separate cell population. Thus, induction 270 of JNK/AP-1 signaling upon tissue damage and mutual repression with JAK/STAT is sufficient 271 to set up temporal and spatial signaling patterns, and functionally distinct cell populations. We then turned to the tyrosine phosphatase Ptp61F, a known repressor of Stat92E activation 291 [113,114] which is transcriptionally induced in JNK-dependent tumors and other stress 292 conditions [23,115]. Repression of JAK/STAT signaling by Ptp61F may occur at the level of 293 receptor complex signaling or Stat92E activity [113,114,116]. We found that knock-down of 294 ptp61F caused upregulation of the Stat92E reporter activity in high JNK-signaling cells of egr-295 expressing discs ( Fig. 4A-E, Fig. S4. 3A-C). This suggests that Ptp61F is specifically required 296 in JNK-signaling cells to suppress Stat92E activity. Hence, to test if dephosphorylation of 297 Stat92E in JNK-signaling cells is rate-limiting for JAK/STAT signaling, we overexpressed an 298 HA-tagged Stat92E [117] in egr-expressing cells to saturate the dephosphorylation capacity of 299 We thus hypothesized that JAK/STAT interfered with the protective G2-stall of high JNK-320 signaling cells, and thereby increased apoptosis [31,37], While Stat92E expression altered 321 the proportion of S-phase cells in undamaged control discs, we observed no changes in the 322 proportion of cells in G1 or G2 ( Fig. S5E-H). In contrast, a cell cycle analysis of egr,Stat92E-323 HA co-expressing domains revealed a substantial increase in G1 and S-phase cells, indicating 324 that JNK-signaling cells failed to stall in G2 ( Fig. 5E-J). Thus, forced Stat92E activation in high 325 JNK-signaling cells overrides the protective JNK-induced G2 stall and consequently increases 326 apoptosis. Supporting these results, we found that knock-down of Ptp61F in egr-expressing 327 discs also led to an increase in cycling cells in the tissue (Fig. 5O,P) correlating with increased 328 levels of apoptosis . This highlights the necessity of JNK/AP-1 and JAK/STAT 329 signaling segregation within the tissue, as it maintains the G2 stalled cell population. Thus, the 330 JNK-induced mutual repression network ensures the establishment of two distinct and 331 indispensable cell populations -one that stalls in G2 to prevent apoptosis and secrete 332 necessary pro-mitogenic factors like Upd and other paracrine effectors like Dilp8 and ImpL2 333 [20,74], and a second that is able to respond to mitogenic signals and undergoes proliferation 334 necessary for regeneration. The strong similarities with paracrine signaling centers regulating 335 spatial patterning networks linked to specific cell behaviors and fate during development 336 encourages us to propose that the JNK-induced mutual repression network with JAK/STAT 337 functions as a central wound organizer network [118,119]. 338

Figure 6 340
The wound organizer network segregates JNK/AP-1 and JAK/STAT signaling to drive 341 oncogenic growth 342 343 Our observation that tissue repair processes after chronic damage support the establishment 344 of a signaling-sending and a responding cell population, which are respectively G2-stalled or 345 proliferative, led us to speculate that such a wound organizer network could well-support a 346 tumor microenvironment. Interestingly, the coexistence of JNK/AP-1 and JAK/STAT pathways 347 in tumors has been extensively described [18][19][20][21][22][23]  not co-opt either pathway (Fig. 6A,B, Fig. S6A,B). RNAi-driven knock-down of scrib induced 358 barrier dysfunction (Fig. S6C,D) and, consistent with previous reports [123,124], moderately 359 activated both JNK/AP-1 and JAK/STAT in the pouch (Fig. S6E-H). A closer examination, 360 however, revealed that the scrib-RNAi expressing pouch predominantly segregated JNK/AP-361 1 and JAK/STAT signaling at the level of small cell clusters, while only few cell clusters 362 displayed co-activation of both pathways (Fig. 6C). This pattern mirrored our observations of 363 short-term egr-expression (Fig. S2L,M), thus suggesting that different contexts of mild JNK-364 activation cause segregation at short distances. These observations indicate that JNK/AP-1 365 activation upon barrier dysfunction via disruption of cell polarity also induces the wound 366 organizer network, which subsequently sets up a bistable segregation of both pathways even 367 at a short range between neighboring cells. 368 369 Curiously, if Ras V12 and scrib mutations occur in the same cell, they cooperate to cause 370 dramatic overproliferation of imaginal discs, which is not observed in single Ras V12 or scrib 371 mutant tissues. This cooperativity is thought to be driven by activation of JNK/AP-1, as a 372 consequence of scrib disrupting cell polarity. JNK/AP-1 signaling then induces upd's and 373 JAK/STAT activation, which can be utilized by a Ras V12 controlled network to drive overgrowth 374 [18,122]. We thus wondered if Ras V12 may disable the mutual repression motif between 375 activation of JNK/AP-1 may then be protected by anti-apoptotic functions of Ras V12 . We wanted 378 to test this hypothesis and closely monitored JNK/AP-1 and JAK/STAT signaling as well as 379 cell proliferation in Ras V12 and scrib RNAi-coexpressing discs. Surprisingly, we found that 380 JNK/AP-1 and JAK/STAT activation were still clearly segregated into exclusive signaling 381 domains ( Fig. 6D-F, Fig. S6I) with proliferation still predominantly associated with JAK/STAT 382 ( Fig. 6G-I). This suggests that even in a Ras V12 and scrib cooperativity model, where all cells 383 are genetically identical, bistability driven by the mutual repression network organizes the 384 pouch into distinct JNK/AP-1 or JAK/STAT-signaling domains. This observation aligns with 385 previous observations that cooperativity between Ras and scrib can rely, in principle, on non-386 autonomous interactions [18,125]. Crucially however, our observations demonstrate that non-387 autonomous cooperation between JNK/AP-1-activating and Ras V12 mutation arises as a self-388 organizing principle from a wound organizer network creating segregating signaling domains 389 in a genetically homogenous tumorigenic tissue. Moreover, it suggests that activation of this 390 wound organizer network and segregation of signaling and proliferative tasks may provide an 391 advantage to tumors over autocrine integration of JNK/AP-1 and JAK/STAT signaling in the 392 same cell. 393 394

396
Previous studies established a role for JNK/AP-1 in promoting regeneration upon injury [35, 397 59]. This role is, in part, mediated by the expression of upd's which are essential to promote 398 proliferation and survival via JAK/STAT signaling in a non-autonomous manner [28, 35, 48, 399 51, 126-129]. By identifying a mutual repression network between JNK/AP-1 and JAK/STAT, 400 we conceptually establish JNK as a core organizer of tissue repair with parallels to organizers 401 of cell fate patterning in developing tissues. The rules of the wound regulatory network we 402 describe have important implications. Previous reports suggest that JNK/AP-1 activates itself 403 non-autonomously, for example via activation of paracrine egr or ROS [28,54,92]. This implies 404 that JNK/AP-1 activation could result in unchecked spatial expansion of JNK/AP-1-signaling, 405 and accordingly of JAK/STAT signaling. Their unrestrained co-expansion would likely result in 406 chronic inflammation. However, the mutual repression motif restrains the expansion of both 407 JNK/AP-1 and JAK/STAT signaling domains, thus defining stable regions of inflammatory 408 (mitogenic signals) and regenerative (pro-proliferative)  By reducing Stat92E activation upon damage, JNK/AP-1 creates a permissive state for the G2 420 stall by preventing cycling, thereby protecting cells from apoptosis in a cytotoxic, inflammatory 421 environment ( Fig. 7C; Fig. S1A). When high levels of both pathways are co-activated in the 422 same cell, the anti-apoptotic state of the G2 stall is in conflict with proliferative outcomes and 423 therefore active cycling which causes an increase in apoptosis. Consequently, apoptosis of 424 these wound center signaling cells could jeopardize wound organizer function and ultimately 425 tissue repair behaviors (Fig.7D). Therefore, the JNK-activated G2-stall and JAK/STAT-426 activated cell cycling need to be spatially segregated. Importantly, our observations indicate 427 that, with exceptions [132], even successful tumors avoid coexistence of JNK and JAK/STAT 428 signaling in the same cell as they do not profit from the conflicting cell-autonomous inputs on 429 proliferation and survival decisions. Instead, tumors may highjack a wound organizer network 430 to chronically activate paracrine mitogenic signals but outsource proliferation to other cells as 431 a consequence of the senescent properties of JNK-signaling, G2-stalled cells [31]. is evidence of chronic mitogenic signals emanating from the wound [3,9,10,133]. These 453 clinical patterns resemble patterning in egr-expressing discs, thereby suggesting that 454 Drosophila imaginal discs may represent a suitable model to study wound pathologies driven 455 by prolonged activation of inflammatory signaling.

Fly stocks 459
All fly stocks and experimental crosses were maintained on standard media and raised at 18 °C 460 or room temperature (22 °C) unless otherwise specified. For detailed genotypes, please refer 461 to     shock was induced on developmental day 5 or 6 at 37°C for 7-10 min . Larvae were dissected 498 at wandering 3rd instar stage or as indicated (28h or 48h after heat-shock, Fig. 2A . 500 501

Genetic cell ablation using Gal4/UAS/Gal80ts 502
To induce expression of egr, experiments were carried out as described in [31,38,51] with 503 few modifications. Briefly, larvae of genotype rn-GAL4, tub-GAL80 ts and carrying the desired 504 UAS-transgenes were staged with a 6h egg collection and raised at 18°C at a density of 50 505 larvae/vial. Overexpression of transgenes was induced by shifting the temperature to 30°C for 506 24h at D7 after egg deposition (AED), and larvae dissected at recovery time point R0 h ( Fig. S1.1C) unless noted otherwise. For time course and tumor experiments, transgenes were 508 induced by shifting the temperature to 30°C for 0h, 7h, 14h or 24h at D7 (Fig. S1.2, 2 and S2), 509 or 44h at D6 ( Fig. 6 and S6) respectively. Larvae were subsequently dissected for analysis or 510 allowed to recover at 22°C for the indicated time. All images represent R0 h unless noted 511 otherwise. Control genotypes were either rn ts >, or sibling animals (+/TM6B, tubGAL80 or 512 -Bolton et al. 2009). All experiments were performed with ≥ 2 biological 513 replicates. 514 515 Immunohistochemistry 516 Wing discs from third instar larvae were dissected and fixed for 15 min at room temperature in 517 4% paraformaldehyde in PBS. Washing steps were performed in PBS containing 0.1% 518 TritonX-100 (PBT). Discs were then incubated with primary antibodies (described in Table S1) 519 in PBT, gently mixing overnight at 4°C. Tissues were counterstained with DAPI (0.25 ng/µl, 520 Sigma, D9542), Phalloidin-Alexa Fluor 488/647 (1:100, Life Technologies) or Phalloidin-521 conjugated TRITC (1:400, Sigma) during incubation with cross-absorbed secondary antibodies 522 coupled to Alexa Fluorophores (Invitrogen or Abcam) at room temperature for 2h. Tissues were 523 mounted using SlowFade Gold Antifade (Invitrogen, S36936). Whenever possible, 524 experimental and control discs were processed in the same vial and mounted on the same 525 slides to ensure comparability in staining between different genotypes. Images were acquired 526 using the Leica TCS SP8 Microscope, using the same confocal settings and processed using 527 tools in Fiji. Of note, all wing discs expressing the 10xStat92E>dGFP reporter were boosted 528 with an anti-GFP antibody. 529 530

EdU Labelling 531
EdU incorporation was performed using the Click-iT Plus EdU Alexa Fluor 647 Imaging Kit, 532 (described in Table S1) prior to primary antibody incubation. Briefly, larval cuticles were 533 inverted in Schneider's medium and incubated with EdU (10µM final concentration) at RT for 534 15 minutes. Cuticles were then fixed in 4% PFA/PBS for 15 minutes, washed for 30 minutes 535 in PBT 0.5%. EdU-Click-iT labeling was performed according to manufacturer's guidelines. 536 Tissues were washed in PBT 0.1%, after which immunostainings, sample processing and 537 imaging were carried out as described above. 538 539

Western Blots 540
Cell lysates from third instar wing imaginal discs and brains were prepared in lysis buffer (50 541 mM Tris-HCl pH 7.5, 300mM NaCl, 0.1mM EDTA, 1% Triton X-100, 0.1% SDS, 5% Glycerol, 542 1mM PMSF, 1/10 tablet of Complete Mini Protease Inhibitor Cocktail) on ice. Protein samples 543 were loaded onto a 10% polyacrylamide gel, along with Chameleon Duo Pre-stained Protein Ladder (LI-COR, P/N 928-60000) as a molecular weight ladder and run at 150V. After SDS-545 PAGE, proteins were transferred onto a nitrocellulose membrane (Bio-Rad, 162-0115, 546 0.45µm) in transfer buffer (25 mM Tris, 192 mM glycine, 20% (v/v) methanol) using the wet-547 tank method with a current density of 300mA for 1h. Prior to antibody incubation, membranes 548 were blocked with 5% milk in PBS. Membrane was incubated with primary antibodies in PBS-549 T (1% Triton-X in 1xPBS) -rat anti-GFP (Chromotek, 3H9-100, 1:1000) and mouse anti-α- Maximum intensity projections of z-stacks for each disc were obtained in Fiji. Masks were 568 generated after applying a fixed threshold (10-255) to the DAPI channel and noise was reduced 569 using the 'Despeckle' function. Area measurement from the resulting masks were obtained 570 using the 'Measure' function. 571 572 10xStat92E>dGFP quantification 573 Maximum intensity projections of selected confocal sections were taken, excluding the 574 peripodium and carefully chosen to capture signaling activity within the disc proper. In the egr-575 expressing or egr,transgene-coexpressing discs, a mask of the pouch domain was generated. 576 Region Of Interest (ROI) outlines were either manually drawn on the DAPI channel to include 577 the pouch and hinge domains till the medial hinge fold or by thresholding Nubbin staining. 578 Centroid co-ordinates were identified using the 'Measure' function and marked with a point 579 tool. To measure the fluorescence intensity of the Stat92E>dGFP reporter, a square ROI 580 (25x25µm) was placed centrally over the centroid and mean intensity within the ROI was 581 obtained using the 'Measure' function in Fiji. (Fig. 4) 582 583

Dcp-1 area quantification 584
Maximum intensity projections of stacks for each disc were obtained in Fiji. Masks were 585 obtained by thresholding for signal of interest. Signal to noise ratio was enhanced by 'Remove 586 outliers' (bright, radius = 1.5) or 'Despeckle' functions. The '3D object counter' function was 587 used to obtain surface area of the generated mask. To control for differences arising from the 588 total size of the discs, ratios between the Dcp-1 area and total WID area were obtained for 589 each disc and reported as percentage values. (Fig. 5C For reporter patterns along disc center to disc periphery, JNK/AP-1 masks were made using 598 thresholding tools and ROI outlines were generated. Based on ROIs, the centroid value was 599 determined and n≥15 tracks were drawn from the centroid outwards towards the hinge for a 600 fixed distance -covering the JNK/AP-1 domain, the JNK/AP-1-JAK/STAT interface, and the 601 JAK/STAT signaling domain respectively. Using the 'Plot profile' function in Fiji, the 602 fluorescence intensity values of each reporter was obtained along these tracks and averaged. 603 Reporter patterns for each disc was graphed by scaling the averaged values from each track, 604 between 0 to 1 and plotted on the Y-axis. This was done independently for n≥3 discs, and a 605 representative image was selected for visualization of the spatial trends of the reporters. signaling pouch domain. After applying a 'Gaussian blur' filter (sigma=2), masks were generated by applying a low fixed threshold (value = 10) based on early, 7h egr-expressing 618 discs. Noise removal was done using the 'Despeckle' function, followed by 'Fill Hole' function. 619 ROI outlines were generated using 'Create selection'. All ROIs were placed on the source 620 TRE>RFP and Stat92E>dGFP reporter channels. 'Save X-Y co-ordinates' function was used 621 to obtain pixel fluorescence intensity values for both reporters within the ROI, and the 622 distribution of fluorescence intensities across the samples were graphed by generating a 623 binned (bins = 16) 2D histogram plot using R. After all NA values were set to 0, data density 624 per bin for each disc was calculated as a percentage. Values were averaged across 625 corresponding bins for all discs within a timepoint to generate the average fluorescence 626 intensity distribution profile. To distinguish regions with low, medium and high signaling for 627 each pathway, the graphs were subdivided into sections s1-s9 by defining a threshold to 628 distinguish low, medium and high JNK/AP-1 and JAK/STAT signal intensities (Fig. S2K). Crosses were setup as previously described and wing discs from larvae were dissected on D7 643 AED. Single sections at similar focal planes were carefully chosen during imaging. Using 644 drawing tools in Fiji, ROIs were generated to include JAK/STAT signaling cells in the hinge 645 within A and P compartments. The mean JAK/STAT reporter intensity within the ROI was 646 obtained from control and RNAi expressing discs using the 'Measure' function in Fiji. For each 647 dataset, paired t-tests were performed between A and P fluorescence intensity values for the 648 same genotype. (Fig. S4.3A-C) 649 650

Nuclear translocation of Stat92E-GFP 651
Images from the pouch (n=12) and hinge (n=9) domains of control discs and pouch (n=16) and 652 hinge (n=12) domains of p35+egr expressing discs were obtained to track the intra-cellular 653 localization of Stat92E-GFP within the nucleus and cytoplasm. Images were taken at high magnification (63x). After thresholding ('Huang'), the selection tool was used to generate a 655 nuclear ROI using the DAPI channel. The inverse selection tool was used to generate a 656 corresponding cytoplasmic ROI for each image (Fig S4.1L). These nuclear and cytoplamic 657 ROIs were then placed on the Stat92E-GFP channel and mean fluorescence intensity values 658 were obtained for each subcellular fraction and unpaired t-tests were performed. (Fig. S4.1M Area of each mask was obtained using the 'Measure' function. Summed area of these masks 678 were considered as the total area and each area was then represented as a percentage of 679 total. One-way paired ANOVA with Holm-Šídák's multiple comparison test was performed to 680 test for statistical significance. Mean intensity for TRE>RFP or 10xStat92E>dGFP reporters 681 were measured inside the JNK/AP-1 mask (JNK + ) and the exclusive JNKmasks. Paired t-682 tests were performed to test for statistical significance. (Fig. 6E,F and S6I blur' filter (sigma=1), 'Moments' threshold and 'Despeckle' function for noise correction was 692 applied. The DAPI channel was used to manually outline the pouch domain and generate 693 masks. Using the 'Image calculator' function, STAT + mask was 'Subtract'-ed from the pouch 694 mask to generate the STATmask. Areas of the STAT + and STATregions in the pouch were 695 obtained using the 'Measure' function. Then, the ph3 channel was duplicated and 'Li' threshold 696 was applied. 'Remove outlier' (radius=5 threshold=50) and 'Despeckle' function for noise 697 correction, followed by 'Watershed' function was applied to generate a ph3 mask. Using the 698 'AND' operator, the ph3 mask was overlayed with the STAT + and STATmasks. '3D object 699 counter' function was then used to obtain ph3 'Object counts' within the STAT + and STAT -700 regions and divided by area of each ROI to get the ph3 counts per mm 2 . Paired t-tests were 701 performed to test for statistical significance. (Fig. 6I) 702 703

Mathematical modelling 705
To test for the existence of a regulatory motif within our signaling network, a mathematical 706 model was derived to describe the temporal dynamics of the concentration of specific 707 molecules over a fixed 2D space. The system was modeled as a set of ordinary differential 708 equations (ODEs) (see also We performed a global analysis by sampling approach introduced in Rausenberger et. al [135]. 748 The models were simulated on a 1D spatial domain x = [0 1]. To represent the wound site, a 749 high basal production of Egr was fixed within a finite space at x = [0 0.05], while basal rate of 750 production of Egr for the rest of the region was set to zero. The rate of production of Upd was 751 defined as a function of JNK/AP-1 activity [18,20,39,51]. For all states, initial condition was 752 set as u (t=0,x) = 0.1. Solutions for each model (with and without repression) were simulated 753 using 10 6 parameter sets, drawn uniformly from an interval from 10 -3 to 10 +2 on a logarithmic        14h. Graphs display mean ± SEM for control 0h, n=2; 7h, n=2 and 14h, n=2 discs and egr 0h, n=2; 7h, n=4 and 14h, n=5 discs.

Maximum projections of multiple confocal sections are shown in B-D, H-J.
Discs were stained with DAPI to visualize nuclei.     14h, n=6 and 24h, n=6 discs. The same density color scale applies to all genotypes.

(K)
Schematic showing different sections (s1-s9) from which mean pixel density values were obtained for 7h, 14h and 24h discs for statistical analysis. Black asterisks mark sections of interest. Graph comparing mean density of pixels in each section (s1-s9) for different durations of cell ablation Ordinary 2-way ANOVA with Dunnett's multiple comparisons test was used to test for statistical significance. Graphs represent mean ± SEM for 7h, n=5; 14h, n=6 and 24h, n=6 discs. Note the reduction in medium JAK/STAT activity within JNK/AP-1 signaling cells in S5 from 14h (15.37%) to 24h (5.25%) in egr-expressing discs.    (D) Quantification of total area of wing discs. Graphs display mean ± SEM for n=8, egr-expressing and n=10, egr,ptp61F-RNAi-co-expressing discs. T-tests were performed to test for statistical significance.
(E) Quantification of the Stat92E>dGFP reporter fluorescence intensity measured within the JNK/AP-1 signaling pouch domain. Black square in schematic shows measured region. Graphs display mean ± SEM for n=8, egrexpressing and n=10, egr,ptp61F-RNAi-co-expressing discs. U-tests were performed to test for statistical significance. (J) Quantification of the Stat92E>dGFP reporter fluorescence intensity measured within the JNK/AP-1 signaling pouch domain. Black square in schematic (E) shows measured region. Graphs display mean ± SEM for n=16, control discs (rn ts >), n=9, stat92E-expressing discs, n=11, egr-expressing and n=12, egr,stat92E-co-expressing discs. One-way ANOVA with multiple comparisons was performed to test for statistical significance.
Discs were stained with DAPI (magenta) to visualize nuclei. Scale bars: 50 µm independent replicates (B). Stat92E-GFP partially rescues larval lethality in stat 85c9 null mutants from 0% to 15% pupariation, at approximately half the expected rate of a full rescue (red dashed line), n=3 independent replicates (C). Thus, the Hop-GFP and Stat92E-GFP fosmid lines were used as proxies for further analysis of protein localization in egr-expressing discs.
(D) Western blots analyzed for Stat92E-GFP expression in imaginal discs. HP-1-GFP genotypes were included as positive control, along with a non-tagged negative control from wild type imaginal disc extracts. Increasing concentrations were loaded, and probed with anti-GFP and anti-Tubulin antibody. The GFP-tagged Stat92E protein is detected at the expected MW (125Kda and 110Kda), and likely representing overlapping isoforms running in two separate weight ranges (71.2-76.8 kD and 85.6-92.8 kD plus GFP-tag).
(E) A control w 118 disc used as a negative control to determine anti-GFP antibody background. expression is low in the medial hinge folds of control discs, the hinge shows elevated expression in egr discs (purple asterisk). This is consistent with JAK/STAT activity inducing expression of Stat92E to promote self-activation. In contrast, Stat92E-GFP protein levels remains low in the pouch (yellow asterisk).
(J-K) A control pouch (J) and a larger domain of p35,egr-expressing cells in the pouch (K) was used to test for changes in nuclear versus cytoplasmic localization of Stat92E-GFP within high JNK/AP-1-signaling cells.
(L) Schematic showing the centrally selected region (black square) in control and egr,p35-expressing discs, used to measure Stat92E-GFP fusion protein intensity in the JNK/AP-1 signaling pouch domain. Automated workflows were used to generate nuclear and cytoplasmic ROIs from selected regions, to quantify Stat92E-GFP protein levels within these area fractions.     p=0 .  Dcp-1 (red) to visualize apoptosis.
(C-D) Quantification of total wing disc area (C) and normalized area occupied by cleaved Dcp-1 (D). Graphs display mean ± SEM for n=10, egr-expressing and n=9, egr,stat92E-co-expressing discs. T-tests were performed to test for statistical significance.
Dashed white squares highlight the distinct shift from G2 (yellow cells) to G1 (green cells) in the pouch domain. (I-J) EdU incorporation assay to detect S-phase cells (gray) in egr-expressing (I), or egr,stat92E-co-expressing (J) disc. Discs were stained with DAPI (red) to visualize nuclei.
(M-N) Quantification of total wing disc area (M) and normalized area occupied by cleaved Dcp-1 (N). Graphs display mean ± SEM for n=11, egr-expressing and n=15, egr,ptp61F-RNAi-coexpressing discs. T-tests were performed to test for statistical significance.
(O-P) EdU incorporation assay to detect S-phase cells (gray) in an egr-expressing (O), and egr,ptp61F-RNAicoexpressing (P) disc. Discs were also stained with DAPI (red) to visualize nuclei.
Maximum projections of multiple confocal sections are shown in A-B, I-L, O-P.  which also express UAS-GFP (green), all under the control of rn-GAL4. Discs were stained for cleaved Dcp-1 (magenta) to visualize apoptosis. Note how almost all apoptotic debris is labelled by GFP (D), demonstrating that apoptotic cells truly originate from egr,stat92E-coexpressing cells.
TRE>RFP and Stat92E>dGFP reporter fluorescence intensities were adjusted to subsaturation in egr-expressing discs and all genotypes were imaged at comparable settings.