dMyc is required in retinal progenitors to prevent JNK-mediated retinal glial activation

In the nervous system, glial cells provide crucial insulation and trophic support to neurons and are important for neuronal survival. In reaction to a wide variety of insults, glial cells respond with changes in cell morphology and metabolism to allow repair. Additionally, these cells can acquire migratory and proliferative potential. In particular, after axonal damage or pruning the clearance of axonal debris by glial cells is key for a healthy nervous system. Thus, bidirectional neuron-glial interactions are crucial in development, but little is known about the cellular sensors and signalling pathways involved. In here, we show that decreased cellular fitness in retinal progenitors caused by reduced Drosophila Myc expression triggers non cell-autonomous activation of retinal glia proliferation and overmigration. Glia migration occurs beyond its normal limit near the boundary between differentiated photoreceptors and precursor cells, extending into the progenitor domain. This overmigration is stimulated by JNK activation (and the function of its target Mmp1), while proliferative responses are mediated by Dpp/TGF-β signalling activation.

For a functional nervous system, neurons transmit information from cell to cell while glial cells provide crucial insulation and trophic support to neurons, which is important for neuronal survival. Glial cells are one of the most plastic cell types being able to adapt and respond to changing environmental stimuli. In this work we inhibit the function of the growth regulator dMyc in Drosophila retinal primordium, the eye imaginal discs. Glial cell numbers and migration pattern to the eye disc are tightly controlled but in dMycdepleted retinas the glial cells overcome their normal barriers and overmigrate into the eye progenitors domain. We show evidence that this process is mediated by JNK activation in the presence of metalloproteinases. We discuss the biological role of overmigrating glia in tissue regeneration and/or confinement of the damaged area. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
The nervous system is formed by neurons, which transmit information from cell to cell, and by glia, which supports and maintains a healthy and functional neuronal network (reviewed in [1]). A key function of glial cells is to provide insulation, trophic and survival support to neurons [2]. Furthermore, myelinating glia (or wrapping glia in invertebrates) can sculpt the structural and electrical properties of axons by controlling their diameter [3,4], or the spacing and clustering of ion channels at nodes and paranodes [1]. In addition, during development about 50% of neurons undergo programmed cell death (PCD) while others require axonal, dendritic, or synaptic pruning. Clearance of apoptotic corpses and engulfment of pruned parts is mediated by microglia and astrocytes in vertebrates [1,[5][6][7][8][9] and by various types of glia in Drosophila [10][11][12][13][14][15][16][17].
Neurons and glia interact in a bidirectional manner as distinct neuronal signals can regulate proliferation, survival, and differentiation of glial cells [18][19][20][21][22]. Despite the existence of some differences, similar mechanisms are involved in both invertebrates and vertebrates to match the axon surface area requiring wrapping and the number of wrapping glial cells [23]. This is well demonstrated by the sequential increase in retinal glia number to match the differentiating photoreceptors in the Drosophila eye [24] and the increase of glial cell size, through polyploidization, to match the increase of neuronal mass in the growing brain [25]. Furthermore, axonal neuregulin controls the proliferation and the survival of oligodendrocytes and Schwann cells in mammals [21,[26][27][28][29] and longitudinal glia in Drosophila [23,30]. Other factors such as PVF/PDGF and TGF-α have also been shown to exert both a gliatrophic and gliatropic function in glia [30][31][32]. For those functions to be accomplished, glial cells have to be extremely plastic in order to quickly respond to neuronal changes.
Despite the cumulative knowledge on the role of glia during axonal development and maintenance it is unknown if glia has the ability to recognise and respond to the microenvironment fitness changes in the context of development. Myc is a highly conserved helix-loop-helix leucine zipper transcription factor which mutations affect the ''cellular fitness" at the level of cell growth and proliferation [33][34][35][36][37][38]. The conserved MYC functions include control of cell growth, energy production (glycolysis, glutaminolysis and mitochondrial biogenesis), anabolic metabolism (synthesis of amino acids, nucleotides and lipids) and DNA replication [37,38]. At the mechanistic level, MYC binds and regulates a large subset of genes that control ribosome biogenesis [39,40], translation [41,42] and metabolism [39,40,43]. In fact, the ability of Myc to stimulate ribosome biogenesis is crucial both in development and oncogenesis [44]. This happens in part through the regulation of Polymerase I [45,46] and the RNA exonuclease Viriato (Nol12 in vertebrates) [47,48]. In this work, we have impaired retinal progenitors fitness in Drosophila eye imaginal disc by Drosophila Myc (dMyc) loss-of-function and found that it induces glial cell proliferation and overmigration. Myc acts in eye disc progenitors to prevent JNK activation, which is otherwise sufficient to induce matrix metalloproteinase 1 (Mmp1) expression and trigger glia overmigration.

dMyc is required in eye progenitors to block glia overmigration
The eye imaginal disc of Drosophila is a good system to study neuron-glia crosstalk because these cell types have distinct and spatially separated origins. Photoreceptors originate from retinal progenitor cells in the eye disc, while glial cells derive from the optic stalk [24]. Glial cells migrate outwards from the brain along the optic stalk, towards the basal surface of the eye disc. This process is tightly coordinated with the ongoing photoreceptor differentiation [24] which takes place behind the morphogenetic furrow (MF), an epithelial indentation that progresses in a posterior to anterior direction (reviewed by [49]). Knocking-down dMyc function by dsRNA expression, specifically in eye progenitor cells (but not in glial cells) using the ey-Gal4 driver, leads to a reduced eye phenotype (Fig 1A-1C and S1A and S1B Fig), as expected given its role in cell growth regulation [34]. In control eye imaginal discs, perineurial glial cells (PG) migrate from the optic lobes into the eye disc and its migration terminate 3 to 4 ommatidial columns posterior to the MF (Fig 1D and 1F) [50][51][52]. Unexpectedly in dMyc RNAi retinas, glial cells can migrate pass the MF in approximately 90% of the discs (S1E Fig), moving beyond the atonal expression stripe anterior to the MF (Fig 1A-1Q). These cells migrate as multicellular chains or strands and importantly, detachment from other glial cells in the endogenous domain (posterior to the MF) was never observed. Glia overmigration was rescued in 96% of the ey>dMyc RNAi eye discs by dMyc co-expression (S1D and S1E Fig). During normal migration, PG cells stop just posterior to the MF when they contact a nascent photoreceptor axon (R-axon) and initiate differentiation into wrapping glia ( Fig 1H) (reviewed by [53]). In ey>dMyc RNAi retinas, glial cells still differentiate into wrapping glia as detected by the expression of sprouty (sty) [54] (Fig 1H and 1I) and by the presence of axonal wrapping ( Fig  1J, 1K, 1M and 1N). Furthermore, the majority of overmigrating cells do not express sty suggesting that these cells are PG and that differentiation into wrapping cells is still dependent on photoreceptor differentiation (only one disc out of the 10 analysed showed sty expression in overmigrating glia; Fig 1H and 1I). Glia overmigration is not a response towards patches of ectopic photoreceptor differentiation (Fig 1F and 1G) or axon misrouting in the anterior region, as those are not observed (S1F Fig). To assess whether excessive migration of PG cells is accompanied by changes in glial cell morphology we labelled these cells with LexA-driven expression of CD2-GFP independently of ey-Gal4-driven dMyc RNAi. In contrast with control glia (Fig 1L), cells on the leading edge of migration in ey>dMyc RNAi eye discs show very long projections ( Fig 1O) that extend anteriorly beyond differentiating photoreceptors (Fig 1J and 1M). dMyc is required for multiple functions in the cell, but a reduction of dMyc levels does not directly trigger significant autonomous apoptotic cell death (S1G Fig, left panel) [55,56]. Preventing residual apoptosis using a H99 deletion mutant which removes the pro-apoptotic genes reaper, hid, and grim [57] was not sufficient to interfere with excessive glia migration in ey>dMyc RNAi eye discs (S1G Fig,  left and middle panel). Furthermore, overexpressing dMyc in the eye disc (dMyc OE ) increased apoptotic cell death but did not induced glia overmigration (S1G Fig, right panel). These experiments suggest that in the context of Myc misregulation, apoptosis is not directly regulating the extension of glia migration. Furthermore, repo>dMyc OE caused no overmigration, thus it is not sufficient that glial cells express higher Myc levels in relation to progenitors to migrate excessively (S1H Fig). dMyc-dependent glia overmigration is observed already in the early 3 rd instar eye discs (Fig 1P and 1Q), before photoreceptor differentiation initiates. This strongly suggests that the overmigration phenotype results from dMyc depletion in photoreceptor progenitor cells. In fact, knocking-down dMyc in differentiated photoreceptors caused no glia overmigration phenotype (S1H Fig), whereas dMyc depletion in the anterior region of the disc (using hth-Gal4 driver) causes glia to overshoot beyond the MF (Fig 1S). In contrast, specific depletion of dMyc in glial cells (using the glia-specific repo-Gal4 driver) results in fewer glial cells in the disc (Fig 1T) [58]. In order to understand if the observed glia overmigration phenotype is a consequence of the defective tissue growth induced by Myc depletion, we next knocked down the growth regulator Pi3K92E [59]. Decreased levels of Pi3K92E also result in smaller discs, but glia did not overmigrate ( Fig 1R). Overall, these experiments suggest that Myc is required in retinal progenitors to control glia migration in a non-autonomous manner.  (Fig 2). The autonomous activation of the Dpp pathway, through activation of the Dpp/TGF-β receptor Thickveins (Tkv), induces an accumulation of glial cells in the eye disc ( Fig 2F) [60]. Relatively low levels of Mad activation (phospho-Mad; pMad), the transcription factor acting downstream of Dpp were detected in glial cells of control retinas (Fig 2A and S2 Fig). Interestingly, in ey>dMyc RNAi eye discs we observed an increased activation of Mad, but mainly in glial cells that migrate beyond the MF and overshoot anteriorly (Fig 2A and 2B). The localised pattern of pMad hyperactivation correlated well with increased EdU staining for cell proliferation in overmigrating cells, but not in the overall glia population where no increased EdU is observed ( Fig  2C, 2D and 2G). These experiments indicate that in eye discs with reduced dMyc expression, glial cells that pass the MF respond highly to Dpp (expressed normally at the MF) and proliferate. However, on its own, the accumulation of glial cells in the eye disc, when a constitutively active form of Tkv (Tkv QD ) (Fig 2E) or the constitutively active growth promoter Yorkie [58] are expressed in glia, is not sufficient to promote overmigration beyond the MF. This suggests that other mechanisms are at play in Myc-depleted discs.

dMyc inhibits JNK activation in the eye imaginal disc
The JNK signalling pathway has been widely implicated in morphogenetic and cell migration regulation [61,62]. Recently, JNK activity was shown to be important for neuron-glia crosstalk upon neuronal damage in the Drosophila embryonic CNS [63], Drosophila developmental axon pruning [64] but also in mammalian Schwann cells in response to axonal injury [65,66]. Thus, we tested if Myc function in limiting the extent of glia migration is associated to a role in modulating JNK activation during retinal development. We analysed JNK activation using two transcriptional reporters: an enhancer trap line, puc E69 -LacZ for the JNK target Puckered [67], and TRE-GFP that is under the control of AP-1 binding sites for JNK transcriptional effectors Jun/Fos [68]. There was a strong increase of puc-LacZ (Fig 3A and 3B) and TRE-GFP expression (Fig 3C and 3D) in the eye disc proper upon dMyc knockdown, especially at the anterior region of the disc (Fig 3A-3D arrows). Significantly, we observed a similar upregulation of JNK pathway activation in the anterior domain using an antibody against phosphorylated JNK (activated JNK-pJNK; Fig 3E-3H). We also detected weak ectopic activation of the JNK pathway in glia (using puc E69 -LacZ, TRE-GFP and pJNK; Fig 3A-3H asterisks), suggesting that Myc-depleted eye imaginal discs can induce non-autonomous glia JNK activation. Importantly, we also observed glia overmigration and pJNK activation in eye discs mutant for Myc (ey-dm P0 ) generated by ey-flp mediated removal of a rescuing dMyc transgene. In this experiment, glial cells are phenotypically wild-type (Fig 4A and 4B) [69]. Furthermore, in dMyc is required to prevent JNK-mediated retinal glial activation third-instar eye imaginal discs we observed pJNK autonomous upregulation in some of the dMyc RNAi mitotic flp-out clones, when induced at 48-72h AEL (Fig 4C-4F). The JNK upregulation in dMyc loss-of-function was surprising as it has previously been reported that JNK can regulate MYC through phosphorylation [70], ubiquitination and degradation [71], but never the opposite.

JNK is required downstream of dMyc to induce glia overmigration
Next, we analysed if the activation of the JNK pathway was important for glia overmigration. When dMyc expression is knocked down in the eye disc there is a significant Mmp1 upregulation, hallmark of JNK activation [72], close to the overshooting glial cells (  These experiments suggest that inhibition of dMyc in the proliferative non-differentiated region of the eye disc causes JNK pathway activation, primarily in the eye disc proper, and Mmp1 upregulation. This has a non-autonomous effect in glia allowing its migration to the retinal progenitors region (overcoming its normal boundary).

dMyc knockdown induces ECM remodelling
Glial cell migration in the CNS and PNS have been shown to require significant extracellular matrix (ECM) modulation [76][77][78][79]. In particular, we have previously shown that integrins are important for glial cell migration and Laminin A rearrangement [4]. Additionally it was previously reported that Laminin W is required for cell adhesion and migration during embryonic and imaginal development [80,81]. As we observed Mmp1 upregulation in ey>dMyc RNAi ( Fig  5A and 5B), we decided to assess if changes in ECM components or regulators are associated to glia migration in these eye discs. In Drosophila, two laminin trimers have been described: laminin A (α3,5; β1; γ1) and laminin W (α1,2; β1; γ1). Wing blister (wb) encodes a α-Laminin 1,2 protein (subunit of Laminin W) that in control eye discs accumulates in a column posterior to the MF, near the limit for glial cell migration. When dMyc RNAi was expressed in the eye disc, glial cells that overshoot the MF co-localise with regions of ectopic high WB levels (Fig 6A and  6B). LamininA (LanA; Drosophila α-Laminin 3,5 protein) subunit deposition is even more widespread accumulating in puncta both in the glia posterior domain, and in the overmigrating glia (Fig 6C-6F). Next, we investigated if tissue ECM remodelling mediated by Mmp1 upregulation was sufficient to explain glial cells overmigration in ey>dMyc RNAi. Clonal overexpression of Mmp1 in the eye disc (Fig 6G and 6H) or in differentiated photoreceptors using GMR-Gal4 (Fig 6I and 6J) was not sufficient to allow glia to overmigrate. Furthermore, inhibiting MMPs expression or activity in glial cells (by overexpressing Timp, and Mmp1 and Mmp2 RNAi) did not interfere with the normal glia migration pattern (S3E- S3I Fig). Thus, Mmp activity appears to have a limited role in allowing excessive glia migration in the eye disc, as it is necessary but not sufficient for this phenotype.  dMyc is required to prevent JNK-mediated retinal glial activation JNK activation in the eye disc is sufficient to induce glia overmigration Having shown an essential role for JNK activity in glia overmigration induced by dMyc RNAi, the next question we addressed was if JNK activity per se is sufficient to induce this phenotype. We overexpressed constitutively active hemipterous (hep CA ) in the eye disc driven by dppblk-Gal4, to avoid ey-Gal4-driven lethality (S5A and S5B Fig). Hep is homologous to Jun kinase kinase (JNKK) [82], and activates Bsk (Drosophila JNK) through phosphorylation. As expected, we observed increased upregulation of the JNK activation by phosphorylation (pJNK; Fig 7A and  7B). Interestingly, JNK activation in the eye disc was sufficient to induce glia overmigration ( Fig  7A-7E). The same was verified using the Optix-Gal4 driver (S5D Fig) and in the few larvae that survived from ey-Gal4 and A4-Gal4 driven hep CA (S5A-S5C Fig). ey>dMyc RNAi overmigration is accompanied of pMad activation and increased proliferation in overmigrating glia (Fig 2) so we analysed if this activation was mediated by JNK activation. Surprisingly, no Dpp/TGF-β signalling activation (Fig 7C and 7D) or increased proliferation (Fig 7E and 7F) was detected, for what we conclude that ectopic glia localization anterior to the MF is not sufficient to activate Dpp/TGF-β signalling and promote proliferation. pMad activation occurs downstream of Mycknockdown but JNK activation is not sufficient to promote it (Fig 8). Furthermore, JNK-mediated glia recruitment requires a non-autonomous mechanism, since modulation of the JNK pathway in glia does not appear to have a major role in glia migration (S5E to S5J Fig). Hep activation dMyc is required to prevent JNK-mediated retinal glial activation in the differentiated photoreceptors also does not affect the glia migration pattern (S5K Fig). We can conclude that JNK activation in eye progenitors is sufficient to non-autonomously induce overmigration of glial cells into the eye disc. dMyc is required to prevent JNK-mediated retinal glial activation

dMyc non-autonomous effect in glia
Growth requires energy as well as protein synthesis and MYC activity plays an important role in controlling metabolic pathways such as glycolysis and glutaminolysis [43,83,84]. Hypomorphic dMyc mutant cells are characterized by a reduced growth rate and correspondingly smaller size, with proliferation rates being affected when dMyc levels are strongly reduced [38]. In here, we investigated if decreased cellular fitness in eye progenitors caused by reducing Drosophila Myc expression trigger non cell-autonomous responses in retinal glia development. Indeed, glial cells sense the decreased dMyc levels in retinal progenitors and respond with increased migration and proliferation (Fig 8). This phenomenon is non-autonomous, as when dMyc is depleted in glial cells it does not induce glia overmigration. Cell competition [55,85,86] was observed in mutant dMyc clones that were still functional but were culled by programmed cell death [55] if ''fitter" cells were present that could replace them. We found no evidence for dMyc is required to prevent JNK-mediated retinal glial activation neurons and glia to be comparing their Myc levels in the context of overmigration, as reducing dMyc expression in differentiated photoreceptors or overexpressing dMyc in glia did not change the normal glia migration pattern.
The Dpp/TGF-β pathway stimulates glia proliferation [58,60] and in fact we observed activation of Mad and increased proliferation in overmigrating glia (Fig 2). However, activating Dpp signalling in an autonomous fashion in all retinal glial cells (repo>tkv QD ) was not sufficient to stimulate migration of glia. Overmigration of retinal glial cells has been observed in a few genetic backgrounds, however the patterns of glia migration and proliferation in Mycdepleted eye discs are quite distinct. For example, when the Hh effector Ci is expressed in glial cells they overmigrate being followed by axonal misrouting [60], which was not observed in dMyc-depleted retinas (Fig 1J and 1M and S1F Fig). In dMyc-depleted eye discs, glial cells do not migrate towards patches of ectopic neurons nor are accompanied by axon misrouting as has been described for other contexts of alterations in glia migration [60,87]. If Spinster, a transmembrane protein localizing to the late endosome or lysosome, is downregulated in glia there is overmigration [88], but this is accompanied by increased glia proliferation that is restricted to the optic stalk, unlike the increased proliferation of overmigrating glia that overshot the MF in dMyc-knockdown eye discs. Pvr (PDGF receptor homologue)-activated glia dMyc is required to prevent JNK-mediated retinal glial activation also overmigrate, but distinctly from ey>dMyc RNAi, through both the basal and apical sides of the eye disc and without affecting glia proliferation [78]. Additionally, retinal glial cells overmigrating in response to dMyc loss-of-function in eye disc progenitors are still able to activate Sty-LacZ (a marker of wrapping glia) unlike overmigrating glial cells overexpressing Cut, a transcription factor activated by Dref and FGF [89].
Interestingly, we have shown that dMyc-RNAi eye discs show autonomous activation of JNK and ectopic Mmp1 expression, possibly associated to ECM remodelling. In hemocytes, JNK has been implicated in cell migration through Mmp secretion and Pi3K92E activation [90]. A different mechanism is in place in dMyc-dependent retinal glia overmigration, as dMyc RNAi does not induce ERK activation in migrating glia (S6A and S6B Fig) nor Pi3K92E activation in glia is sufficient to induce dMyc-like glia migration (S6C and S6D Fig). Autonomous JNK activation has been shown to lead to a propagation loop, mediated by upregulation of the Drosophila Eiger/TNF (Egr) that activates JNK in neighbouring cells [91]. We excluded a similar retinal progenitors-glia communication through JNK-mediated Egr expression, as manipulation of Egr levels did not change the glia overmigration phenotype (S6E-S6G Fig). Importantly, JNK activation is both necessary and sufficient to induce non-autonomous glia overmigration. However, we did not observe increased glia Mad activation when glia overmigration was induced by JNK activation in the eye disc and in this genotype we also did not observe increased proliferation of glial cells that overshoot the MF. Thus, we suggest that in ey>dMyc RNAi eye discs, glial cells that reach the anterior domain are highly responsive to Dpp signal but this response is not intrinsic to glia overmigration, requiring dMyc loss-offunction.
Our results show that dMyc has a role in limiting JNK activation in undifferentiated eye disc cells. Thus, it will be interesting to address if dMyc has a general role in JNK repression through direct transcriptional regulation of JNK pathway members. Available ChIP-sequencing data shows that puc and Dusp10 (the mammalian orthologue of puc) are directly bound by MYC [92][93][94][95], suggesting that MYC could regulate JNK signalling via direct control of puc/ Dusp10 transcription.
In the absence of dMyc in the eye, glial cells overmigrate as a chain (distinctly from the single cell migration mode [96]) suggesting that the leading tip cell could be competing for external signals such as FGF [54,97]. This could be the case as FGF is important for normal glia migration and differentiation [54,98]. Importantly, we confirm that depletion of bnl function in the eye disc causes non-autonomous stimulation of glia migration [98] and that this is accompanied by activation of Mmp1 and pMad in the anterior domain, and glial cells proliferation, in a similar manner to dMyc knockdown (S8 Fig). However, we failed to observe a strong genetic interaction between bnl and dMyc, suggesting that distinct changes in the context of growth potential of retinal progenitors could converge on JNK activation.

JNK activation in eye disc progenitors regulates glia overmigration
dMyc-depleted eye discs induce glia proliferation and overmigration, characteristics of many types of tumors, including glioblastomas. In recent years it has become more evident that tumour formation involves interactions between the tumour-initiating cells and the tumour microenvironment niche, all of which contribute to the tumour proliferative and invasive capacity [99]. In this work we show that the dMyc-depleted microenvironment (retinal progenitor cells) activate a stress pathway triggering epithelial/ECM changes that are actively interpreted by glia. This stress response is triggered by JNK activation and mediated by Mmp1 expression. JNK is indispensable for both cell proliferation and apoptosis depending on the cell type, the nature of the upstream stimulus, the duration of its activation and the activities of other signalling pathways [100,101]. We show that JNK activation induced by dMyc loss-offunction in Drosophila eye was not caused by or result in apoptosis, demonstrating a non-apoptotic role for Drosophila JNK in this context. Activation of JNK in eye progenitors was sufficient to activate Jra/Kayak (c-Jun/c-fos homologues) and trigger non-autonomous glia migration. Interestingly, JNK function in developmental axon pruning and injured axons has been described either as deleterious, through the induction of axonal degradation [102][103][104][105][106] or beneficial, through the activation of axonal regeneration [63,64,[107][108][109]. JNK activation is essential for phagocytosis and autophagy both in Drosophila glia [63,110], mammalian astrocytes [111], Schwann cells [66,112,113], microglia [114] as well as in non-glial cells types [115,116]. We envisage that overmigrating glia in Drosophila eye discs mimics reactive gliosis in mammals (important mechanism of rescue/confinement of a brain injured area) [117,118]. Thus overmigrating glia might play roles in promoting tissue regeneration through its phagocytic activity and possible secretion of growth and/or proliferation factors. Undoubtedly, future studies are required to characterise the roles of retinal dMyc in modulating glia responses. Overall, we describe here a useful system to further understand the mechanisms and functional roles of glia activation in response to perturbations of photoreceptor neuronal development.
Images were obtained with the Leica SP5 confocal system and processed with Adobe Photoshop.
Glial cells were counted and eye discs measured in Fiji. Mean and standard deviation were calculated for each case. Results from glial cell counting were analysed by unpaired Student's t test. Results from overmigration discs were analysed by Fisher's exact test in contingency tables. p-values are shown with numbers or with the following asterisk code: Ã = p<0.05; ÃÃ = p<0.01; ÃÃÃ = p<0.001; ÃÃÃÃ = p<0.0001. Error bars present in all graphs represent the standard deviation.