Protein phosphatase 4 coordinates glial membrane recruitment and phagocytic clearance of degenerating axons in Drosophila

Neuronal damage induced by injury, stroke, or neurodegenerative disease elicits swift immune responses from glial cells, including altered gene expression, directed migration to injury sites, and glial clearance of damaged neurons through phagocytic engulfment. Collectively, these responses hinder further cellular damage, but the mechanisms that underlie these important protective glial reactions are still unclear. Here, we show that the evolutionarily conserved trimeric protein phosphatase 4 (PP4) serine/threonine phosphatase complex is a novel set of factors required for proper glial responses to nerve injury in the adult Drosophila brain. Glial-specific knockdown of PP4 results in reduced recruitment of glia to severed axons and delayed glial clearance of degenerating axonal debris. We show that PP4 functions downstream of the the glial engulfment receptor Draper to drive glial morphogenesis through the guanine nucleotide exchange factor SOS and the Rho GTPase Rac1, revealing that PP4 molecularly couples Draper to Rac1-mediated cytoskeletal remodeling to ensure glial infiltration of injury sites and timely removal of damaged neurons from the CNS.

Glial cells continuously survey the brain and respond swiftly to any form of stress or damage. 1 Acute insults, as well as chronic neurodegenerative conditions, trigger robust immune responses from glia. 2,3 Reactive glia undergo striking morphological changes and infiltrate injury sites to rapidly phagocytose cellular debris. 4,5 Glial changes in cell shape, size, and migration in response to neurodegeneration are highly conserved hallmark reactions to trauma in species ranging from Drosophila to humans. Following injury, glial cells either migrate to injury sites or, in instances where the cell soma remains in a fixed location, reactive glia send dynamic membrane projections into regions that house damaged neurons. [5][6][7] Importantly, inhibiting these glial morphogenic responses delays phagocytic clearance of neurotoxic cellular debris, which can attenuate postinjury neuronal plasticity, trigger secondary inflammatory reactions, and exacerbate damage. 8,9 Despite the fact that glial cells undergo significant changes in size and shape to access trauma sites and clear damaged cells, the molecular mechanisms responsible for glial migration and directed extension of processes are not entirely understood.
Acute axotomy of the olfactory nerve in adult Drosophila melanogaster is a well-established injury paradigm to investigate the molecular mechanisms that govern glial morphogenesis and phagocytic function in response to axon degeneration. 4,5,[10][11][12][13][14][15][16] Drosophila glia are morphologically and functionally similar to mammalian glia, and fly glial responses to axon injury mirror those that occur in vertebrate models. 4,[17][18][19] After severing adult maxillary palp or antennal olfactory nerves, local ensheathing glia extend membrane projections to infiltrate antennal lobe neuropil and phagocytose degenerating axonal debris; glial invasion of the antennal lobes requires the highly conserved glial immune receptor Draper/MEGF10. 5 Activated Draper signals via Src family kinases, which leads to activation of Rac1 and cytoskeletal remodeling. [13][14][15][16]20 Recent work has identified two guanine nucleotide exchange factors (GEFs), Crk/Mbc/Ced-12 and DRK/DOS/SOS, that activate Rac1 in this context, directly associating with Rac1 to catalyze the exchange of GDP for GTP, 13,16 but we still have a poor understanding of the molecular effectors that couple the transmembrane receptor Draper to Rac1-mediated cytoskeletal changes.
The evolutionarily conserved serine/threonine protein phosphatase 4 (PP4) complex is involved in diverse cellular functions, including cell proliferation and apoptosis, and is required for embryonic development across species. [21][22][23][24][25][26][27][28] The PP4 complex consists of three subunits: one catalytic subunit (PP4c), which is required for dephosphorylation of target proteins, and two regulatory subunits (PP4r2 and Falafel/ PP4r3), which control subcellular localization of the complex and specificity of phosphatase target association. 21,29 Interestingly, PP4 is also linked to cell motility and tumor invasiveness in several organisms and cell types. For example, in the slime mold Dictyostelium discoideum, PP4 phosphatase activity is necessary for chemotaxis, and in human colorectal carcinoma cells, active PP4 promotes cell migration. 25,30 PP4 is also a positive regulator of Rac1dependent cell movement in cultured HEK293 cells. 30,31 The role of the PP4 complex in glial responses to neural injury, however, has never been explored. Here, we demonstrate a novel role for PP4 in glia as they respond to severed axons. We propose that PP4 is a downstream effector of the glial receptor Draper and that it signals through the SOS GEF complex and the GTPase Rac1 to promote proper glial membrane infiltration of injury sites and clearance of degenerating neuronal material.

Results
PP4 is required for proper glial clearance of severed axons. To inhibit PP4 function in adult Drosophila glia, we expressed UAS-falafel RNAi with the pan-glial driver repo-Gal4 and assayed clearance of degenerating olfactory receptor neuron (ORN) axons. Importantly, because PP4 phosphatase activity is critical for proper development, including asymmetric localization of cell fate determinants in larval neuroblasts, 32 these flies also carried the tubulin-Gal80 ts transgene, which allowed us to temporally regulate GAL4 activity and express falafel RNAi specifically in adult glia. 33 To monitor clearance of axonal debris, a subset of maxillary palp ORNs were labeled with membrane-tethered GFP (OR85e-mCD8::GFP). We severed the maxillary nerves that project into the antennal lobes and then quantified axonal GFP+ fluorescence in each OR85e glomerulus one day after axotomy using previously published methods. 4,11,12 In controls, most GFP+ axonal debris was cleared within 1 day (Figures 1a and b). In falafel RNAi flies, significantly more GFP+ axonal material was present 1 day postaxotomy (Figures 1c-e, Po0.001), suggesting that Falafel is necessary for proper glial engulfment of severed axons. To confirm efficacy of falafel RNAi , we performed immunostaining against Falafel and the glial-specific transcription factor Repo on adult brains. In control animals, Falafel appeared to be localized, or enriched, in glial nuclei (Figure 1m), and we detected a 70% reduction of glial nuclear Falafel fluorescence in falafel RNAi flies (Figures 1m and n).
In addition to Falafel, the PP4 complex contains a second regulatory subunit, PP4r2, and a catalytic phosphatase subunit, PP4c. To determine if a complete PP4 complex is necessary for proper glial engulfment of debris, we again used the Gal4/Gal80 ts system to knockdown PP4c (UAS-PP4c RNAi ) and PP4r2 (UAS-PP4r2 RNAi ) independently in adult glia. One day after severing maxillary nerves, we observed significantly more OR85e GFP+ axonal debris lingering in the antennal lobes in PP4c-and PP4r2-depleted flies (Figures 1f-l, Po0.01). To confirm efficacy of our Gal4/Gal80 ts experiments, we repeated these clearance assays while maintaining flies at the permissive temperature of 22°C, and observed normal clearance ( Supplementary Figures 1a-i). Finally, we performed a short time course to assess clearance in PP4cdepleted flies and found that significantly more axonal material persisted in the brain for at least a week after axotomy (Supplementary Figure 1j). Taken together, these results indicate that the PP4 serine/threonine phosphatase complex is essential for efficient glial engulfment of axonal debris in the adult brain.
PP4 is essential for proper recruitment of Draper and glial membranes to severed axons. Following ORN axotomy, local ensheathing glial cells robustly upregulate the Draper receptor and extend their membranes into the antennal neuropil regions to phagocytose axonal debris, 5 and, in fact, Draper is essential for these cells to invade the neuropil and access degenerating nerves. 5,16 To determine if recruitment/accumulation of Draper and glial membranes was altered in PP4-depleted flies, we first expressed RNAi against Falafel (UAS-falafel RNAi ), PP4c (UAS-PP4c RNAi ), or PP4r2 (UAS-PP4r2 RNAi ) in adult glia using the Gal4/Gal80 ts system, severed maxillary nerves, and then immunostained brains for Draper. In controls, one day after maxillary nerve axotomy, we observed a significant increase in Draper on maxillary ORN-innervated glomeruli (Figures 2b and i), but this response was significantly attenuated in falafel RNAi , PP4c RNAi , and PP4r2 RNAi animals (Figures 2d, f, h, and i Po0.0001), suggesting that the PP4 complex is essential for proper recruitment of Draper to severed nerves.
Next, because Draper accumulation on injured nerves is tightly coupled to the process of glial infiltration of neuropil, we assessed glial membrane responses in control and PP4c knockdown flies. To visualize glial membranes, we used a fly line expressing membrane-tethered red fluorescent protein (RFP) (UAS-mCD4::RFP) under the control of repo-Gal4, and again used the tubulin-Gal80 ts system to express PP4c RNAi specifically in adult glia. One day after antennal nerve axotomy, we observed a striking increase in ensheathing glial membrane RFP+ fluorescence around the antennal lobes (RFP in gray scale, Figure 3b and b′, Po0.01), which represents expansion of responding glial membranes. 5 Importantly, in flies expressing glial PP4c RNAi , there was no detectable increase of glial membrane RFP after injury (Figures 3d and d′). Similarly, while maxillary palp ablation resulted in accumulation of glial membranes on the maxillary nerves of control animals (Figures 3f-g′, Po0.0001), we did not detect an increase in glial membrane expansion along injured maxillary nerves in PP4c RNAi flies (Figures 3h-j). Finally, in uninjured flies, we did not observe any obvious changes in gross glial morphology following PP4 knockdown, nor did we detect a decrease in glial cell numbers (Supplementary Figure 2), further indicating that glial cell development is not overtly affected in these animals. Taken together, these results indicate that the PP4 complex is required in adult glia to activate a program that drives dynamic glial membrane responses to axotomy. Because PP4 inhibition did not lower Draper levels in the adult brain, we reasoned that the PP4 complex may function downstream of Draper to promote glial infiltration of antennal lobes and clearance of severed olfactory axons. To further explore this, we overexpressed PP4c (UAS-PP4cHA) and draper RNAi in adult glia and assayed OR85e axonal clearance. Glial depletion of draper significantly inhibits clearance of severed axons as compared with controls 5 (Figures 4a, b, e, f, and i, Po0.0001). Glial expression of PP4c partially, but significantly, reversed this clearance defect in draper RNAi flies (Figures 4d, h, and i, Po0.05), suggesting that boosting PP4c can partially bypass the requirement for Draper in this injury paradigm. Unpaired t-test. Po0.0001. Scale bars = 20 μm. Genotypes: Control = OR85e-mCD8::GFP,tub-Gal80 ts /+; repo-Gal4/+. Falafel RNAi = OR85e-mCD8::GFP,tub-Gal80 ts /+; repo-Gal4/UAS-falafel RNAi . RNAi = OR85e-mCD8::GFP,tub-Gal80 ts /UAS-PP4c RNA i; repo-Gal4/+. PP4r2 RNAi = OR85e-mCD8::GFP,tub-Gal80 ts /UAS-PP4r2 RNAi ; repo-Gal4/+ PP4 is dispensable for injury-induced activation of STAT92E in ensheathing glia. Although Draper is basally expressed in glia in the healthy adult brain, axon injury triggers transcriptional upregulation of draper, which ensures that adequate levels of the receptor are available to drive dramatic morphogenic changes in glial cell morphology and phagocytic function in the days after axotomy. Upregulation of draper after nerve injury requires the transcription factor STAT92E, and requisite STAT92E binding elements have been defined in the draper promoter. 20 Activation of STAT92E in glia can be easily tracked in adult brains by monitoring the activation of a 10XSTAT92E-dGFP reporter, 20 which contains 10 tandem canonical STAT92E binding sites that control expression of a destabilized form of cytosolic GFP. 35 To determine if STAT92E signaling requires the PP4 complex, we severed the antennal nerves of flies expressing glial PP4c RNAi as well as the 10XSTAT92E-dGFP transgene and quantified GFP levels one day postinjury. Notably, activation of 10XSTAT92E-dGFP was indistinguishable from control animals (Figures 5a-d and f), although the robust increase in Draper protein typically observed in controls (Figures 5a, b, and e, Po0.0001) was inhibited in glial PP4c-depleted animals (Figures 5c, d, and e). This finding that STAT92Edependent transcription appears unchanged in PP4c RNAi animals, combined with our PP4cHA rescue experiment (Figure 4), further supports a model in which the PP4 complex is acting downstream of Draper to drive glial membrane infiltration of neuropil regions to access severed axons.
Two independent GEF complexes (DRK/DOS/SOS and Crk/Mbc/Ced-12) reportedly activate Rac1 in ensheathing glia postaxotomy. 13,16 To determine if PP4 is coupled to activation of the SOS GEF complex, we used repo-Gal4 to overexpress SOS (UAS-SOS-Myc) and PP4c RNAi flies in adult glia, severed maxillary palp nerves, and then assayed Draper accumulation on maxillary palp glomeruli that house severed axons. Interestingly, SOS overexpression significantly reversed Draper recruitment phenotypes in PP4c RNAi animals (Figures 7a-i, Po0.01). We performed comparable experiments in an attempt to manipulate the PP4 complex and Crk/ Mbc/Ced-12 in glia by coexpressing Ced-12 RNAi and PP4c:: HA or coexpressing PP4c RNAi and mbc, but these experiments resulted in lethality. Therefore, although the mechanistic connection between PP4, the Mbc GEF complex, and Rac1 in reactive glia is still unclear, our findings do highlight DRK/SOS/DOS as one key GEF complex that promotes PP4-mediated dynamics in reactive glia required for proper Draper accumulation at injury sites.
Rac1 localization is often coupled to its activity within a cell. We performed Rac1 immunostaining on PP4c glial knockdown and control brains before and one day after maxillary nerve axotomy. Significant Rac1 accumulation was visible on injured axons in control animals (Figures 7j, l, n, p, and s, Po0.01), but not in PP4c RNAi brains (Figures 7k, m, o, q,  and s). We also confirmed that PP4c depletion did not alter basal Rac1 levels in the central brain by quantifying Rac1 fluorescence in regions immediately adjacent to the antennal lobes (Figure 7r). To further explore the connection between PP4c and glial cytoskeletal remodeling after antennal nerve injury, we performed phalloidin stains to visualize filamentous actin (F-actin). One day after antennal nerve axotomy, phalloidin levels were markedly increased in the antennal lobe neuropil regions of controls (Figures 7t, u, and x, Po0.0001), but almost undetectable in PP4c RNAi flies (Figures 7u,w, and x). Collectively, these experiments indicate that PP4 does not influence basal expression of Rac1 in adult glia but instead bolster the notion that PP4 activates Rac1mediated cytoskeletal remodeling via DOS/SOS/DRK to promote glial responses to nerve injury.
Axotomy results in reduced nuclear Falafel expression in responding glia. The regulatory subunits Falafel and PP4r2 regulate PP4 phosphatase complex activity by influencing subcellular localization and substrate recognition. 21,23,29,32 Translocation of Falafel between the nucleus and cytoplasm to access targets for dephosphorylation has been reported in various species and cell types. [23][24][25]33 Thus, we wondered if Falafel might exit the nucleus in glia responding to axotomy to facilitate PP4 complex activity. We expressed nuclear β-galactosidase (β-gal) (UAS-LacZ::NLS) under the control of the ensheathing glial driver TIFR-Gal4 to label ensheathing glial nuclei, performed antennal nerve axotomy, and then immunostained brains with anti-β-gal and anti-Falafel. Comparing uninjured and injured animals, we quantified nuclear Falafel levels by computationally segmenting to β-gal, and found that Falafel fluorescence was significantly decreased at 3 and 6 h postinjury (Figures 8a-d, Po0.05). Notably, we repeated this experiment labeling the nuclei of cortex glia (NP2222-Gal4,  (Figures 8e-i), indicating that Falafel location and/or levels are specifically influenced in the ensheathing glia responding to axotomy. Antennal lobe astrocytes did not express any detectable Falafel (Figure 8j and k).

Discussion
The Draper receptor is essential for proper initiation of dynamic glial responses to axotomy in the adult Drosophila olfactory system. 5 Ensheathing glia fail to infiltrate neuropil regions after olfactory nerve axotomy in draper mutant animals due to inadequate Rac1 activity. 16 The mechanisms that couple activation of Draper to Rac1-mediated cytoskeletal remodeling, glial recruitment to injury sites, and phagocytic clearance of severed axons are poorly understood. Our results now implicate the PP4 phosphatase complex as a critical molecular effector that functions downstream of Draper to activate the DOS/SOS/DRK GEF complex and Rac1 to promote dynamic cytoskeletal rearrangements in glia responding to axotomy. The PP4 phosphatase complex is implicated in diverse cellular functions, including mitosis, DNA strand break repair, and differentiation. [21][22][23][24][25][26][27][28][29][30][31][32][36][37][38][39] Our work now highlights a previously unexplored role for PP4 in governing innate glial immune responses to neurodegeneration and poses interesting questions for future efforts aimed at understanding precisely how PP4 activity promotes cell migration. We show that forced SOS GEF expression rescues loss of PP4c (Figures 7a-i), implicating the SOS GEF complex as one key effector downstream of PP4 required for proper Rac1 activity in responding glia. To the best of our knowledge, direct biochemical interactions between PP4 and GEF complexes have not been reported. Thus, it is unlikely that the SOS/DOS/ DRK complex is directly targeted by PP4 in glia; future screening efforts will be required to delineate the complete signaling pathway that couples PP4 to the SOS complex. We also cannot rule out the possibility that additional GEF complexes (e.g. Crk/Mbc/Ced-12) converge on glial Rac1 to coordinate the assorted dynamic reactions required for glia to access and dispose of degenerating axonal debris. Finally, because glial activation (e.g. recruitment of glial membranes) In Drosophila, Falafel and PP4r2 are the exclusive regulatory subunits that associate with the catalytic subunit PP4c to form a functional trimeric complex. Mammalian genomes contain six or more genes that encode regulatory PP4 subunits, which enhances the capacity for combinatorial control over PP4 activity across cell types and biological states. 21 PP4 complex activity can be regulated, in part, by subcellular localization of the regulatory subunits. For example, in starving Dictyostelium, the Falafel homolog SMEK translocates from the cytoplasm into the nucleus where it activates PP4c to facilitate cell stress responses. 25 Drosophila PP4 complex components also cycle between the nucleus and cytoplasm of proliferating neural precursors, which is essential for PP4 to associate selectively with key targets in  23,32 Our observation that nuclear levels of Falafel decrease significantly in ensheathing glial cells surrounding the antennal lobes within hours after olfactory nerve injury suggests that expression and/or function of PP4 is modified in the glia as innate glial immune responses are elicited. We favor the model that Falafel is translocated out of the nucleus, but we cannot exclude the possibility that it becomes incorporated into a complex that hinders antibody accessibility or becomes degraded. We did not detect a significant increase in Falafel levels in glial cytoplasm postinjury, but this may reflect in vivo imaging limitations while attempting to visualize low concentrations of Falafel distributed throughout the cell.

Figure 6
The serine/threonine PP4 complex was recently identified as a requisite factor for proper immune responses in T cells, B cells, and macrophages, [40][41][42] including proliferation and immune gene induction. Although the specific role of PP4 in glial cell immunity has not been investigated, increased PP4 expression has been reported in glial tumors, suggesting a connection between PP4 function and glial cell invasiveness. 43 Draper is a highly conserved glial receptor essential for glial clearance of damaged and dying neurons across species. The mammalian homolog, MEGF10, is required for glial clearance of apoptotic neurons, as well as developmental axonal/synaptic pruning. 4,5,[10][11][12][13][14][15][16]20,[44][45][46][47][48][49][50] The high conservation of GEF/Rac1-mediated control of cell migration is also well documented. [51][52][53][54] Our work now reveals that the PP4 phosphatase complex unifies these two conserved molecular signaling pathways in the context of glial immunity and may also provide new molecular insight into glial tumor cell migration.
Adult fly brain injury, dissection, and immunostaining. Maxillary palp and antennal ablations were performed on adult flies as described previously. 5 Maxillary or antennal nerves were severed by removing maxillary palps or third antennal segments, respectively, with forceps. One to five days postinjury, heads were removed and fixed in 4% PFA+0.01% Triton-X for 15 min, followed by washing with 1 × PBS+0.01% Triton-X for 3 × 2 min. Brains were dissected in 1 × PBS+0.01% Triton-X in glass well plates, and then fixed in 4% PFA+0.1% Triton-X for 15 min. Brains were washed in 1 × PBS+0.1% Triton-X for 3 × 2 min, and then placed in primary antibody diluted in 1 × PBS+0.1% Triton-X overnight at 4°C. Brains were then washed in 1 × PBS+0.1% Triton-X for 3 × 2 min, and then placed in secondary antibody diluted in 1 × PBS+0.1% Triton-X for 2 h at room temperature. Brains were washed in 1 × PBS+0.1% Triton-X for 3 × 2 min, and then placed in CitiFluor CFM-I mounting media (Electron Microscopy Sciences, Hatfield, PA, USA) for 30 min before being mounted on glass slides. Flies with tub-Gal80 ts were raised at 22°C, shifted to the restrictive temperature 30°C posteclosion for 3-7 days, injured, and then returned to 30°C until dissection. Each genotype had equal amounts of male and female flies. The following antibodies were used: mouse anti-Draper (1:400, Developmental Studies Hybridoma Bank, Iowa City, IA, USA); rat anti-Falafel Western blot analysis. Central brains (optic lobes manually removed) were homogenized in 4 ml 1xLB (loading buffer) per brain. Lysates were loaded into 4-20% Tris-glycine gels (Lonza, Allendale, NJ, USA) and transferred to Immobilon-FL (Millipore, Billerica, MA, USA). Blots were probed with rabbit anti-Draper (1 : 1000, kind gift from Marc Freeman) and sheep anti-tubulin (Cytoskeleton, Denver, CO, USA; no. ATN02). Blots were incubated with primary antibodies overnight at 4°C, washed 3 × with 1xPBS+0.01%Tween-20, and then incubated with secondary antibodies (713-625-147 and 711-655-152, from Jackson Immunoresearch, West Grove, PA, USA) for 2 h at room temperature. Blots were then washed 3 × with 1xPBS+0.01% Tween-20 and 1 × with 1xPBS. Blots were imaged on LI-COR Odyssey CLx Quantitative Western Blot Imaging System, and data were quantified with Li-COR Image Studio software (Li-COR, Lincoln, NE, USA).
Confocal microscopy and image analysis. Brains were mounted in CFM-I mounting medium and imaged using a Zeiss LSM 710 confocal microscope (Zeiss, Thornwood, NY, USA). Brains were imaged in 1 μm steps with a x40 1.4 NA oil immersion plan apochromatic lens. Brains in a single experiment were imaged in the same day on the same slide with the same confocal settings.
Draper recruitment. Draper pixel intensity was quantified in 3D regions of interest in the antennal lobe of 15 μm z-stacks (see dotted outline in Figure 2). These dotted regions were selected because they correspond to OR85e − glomeruli, which was visualized by the introduction of a OR85e-mCD8::GFP transgene.
Membrane expansion. Glial membrane expansion after antennal ablation was measured as RFP+ intensity in 3D regions of interest in the antennal lobe of 15 μm z-stacks.
Falafel translocation. Falafel translocation experiments were quantified by segmenting to the β-gal-positive nuclei (either ensheathing glia or cortex glia) and then measuring mean Falafel fluorescence in these glial nuclei only using Volocity. All image analysis was performed using Volocity image analysis software (Perkin-Elmer, Hopkinton, MA, USA).
Statistics. All statistics were performed in GraphPad Prism 6 (GraphPad, La Jolla, CA, USA). T-tests and one-way ANOVAs were performed as appropriate (see figure legends). All experiments were repeated in full at least three times and post hoc power tests were run to ensure sample size adequacy. Experiments were not blinded. N for each genotype for each experiment: Figures 1a-d)