Drosophila genetics

Drosophila stocks were maintained at 25°C on standard molasses-cornmeal agar. The following strains were obtained from Bloomington Drosophila Stock Center: UAS-RNAi lines (cut^HMS00924 [23]; Atg1^JF02273; Atg1^GL00047/TM3, Sb¹; Atg2^HMS01198; Atg2^JF02786; Atg5^HMS01244; Atg5^JF02703; Atg8a^JF02895/TM3,Sb¹; Atg8a^HMS01328; Atg18^JF02898; Atg18^HMS01193); GAL4⁴⁷⁷,UAS-mCD8::GFP; UAS-Atg1^6A; UAS-Atg1^6B; UAS-GFP-hiw^A; UAS-MJD-78Q; UAS-eGFP-Atg5; UAS-Atg8a.GFP; UAS-wndK188A. Additional strains from other sources included the pan-md reporter strain GAL4^21-7,UAS-mCD8::GFP [24]; class I md reporter strain GAL4²²¹,UAS-mCD8::GFP [25]; class III md neuron reporter strains GAL4^19-12,UAS-hCD4::tdGFP [26]; nompC-GAL4,UAS-mCD8::GFP [27]; and ppk-GAL4,UAS-mCD8::GFP, ppk-GAL80 [23]; class IV md neuron reporter strains GAL4⁴⁷⁷,UAS-mCD8::GFP; ppk1.9-GAL4,UAS-mCD8::GFP [28–30]; and ppk::hCD4::tdTomato (donated by Dr. Yuh-Nung Jan) [31]; UAS-cut; UAS-Atg1^K38Q (donated by Dr. Thomas Neufeld, University of Minnesota) [32], UAS-Hiw^ΔRING and UAS-wnd.C (donated by Dr. Catherine Collins, University of Michigan) [33,34]. A minimum of two independent gene-specific UAS-RNAi (IR) lines were used for each Atg gene to control for any potential off-target effects and a representative IR line for each Atg gene is depicted in the figures. Detailed genotypes for each figure are reported in S1 Table.

Cell isolation and microarray expression profiling

Class-specific isolation and purification of md neurons was performed as previously described [30,35,36]. For C-I, III, and IV profiling, neurons were extracted from GAL4^2-21,UAS-mCD8::GFP (C-I) [36], ppk-GAL4,UAS-mCD8::GFP, ppk-GAL80 (C-III) [37], and ppk1.9-GAL4,UAS-mCD8::GFP (C-IV) [30] age-matched third instar larvae, respectively. For comparative expression profiling between WT C-I neurons and those ectopically overexpressing Cut, neuronal isolations were taken from third instar larvae expressing UAS-mCD8::GFP under the control of GAL4^2-21 in the presence or absence of UAS-cut [36]. mRNA isolation, amplification, labelling, hybridization, and microarray analyses was then performed by Miltenyi Biotec from the class-specific neuronal isolations as previously described [30,35–37]. Microarray data, including metadata, raw data and quantile normalized datasets have been deposited into the Gene Expression Omnibus (GEO) under the following accession numbers: GSE46154 (WT C-I and C-IV neurons) [30]; GSE69353 (WT C-III neurons) [37]; and GSE83938 (WT C-I and C-I neurons ectopically overexpressing Cut) [36]. All microarrays were performed on identical Agilent whole D. melanogaster genome oligo microarrays (4x44K) and statistical analyses of microarray data were performed as previously described [38]. Raw microarray data files obtained from the Agilent microarrays for each md neuron class were read into GeneSpring GX software in which the data was quantile normalized and only those gene probes which were flagged positive and significantly expressed above background were selected for further analyses. GeneSpring software was used to calculate mean fold change gene expression where a given gene/isoform was represented by multiple probe IDs on the microarray.

Immunofluorescent labeling

Dissection and immunofluorescent labeling of third instar larval filets was performed as previously described [39]. Primary antibodies used in this study include: rabbit anti-GABARAP/Atg8a (1:200) (Abcam), rat anti-Atg8a (1:320) (donated by Dr. Gábor Juhász, Eötvös Loránd University) [40], rabbit anti-WndA1 (1:300) (donated by Dr. Catherine Collins, University of Michigan) [34], Dylight AffiniPure Goat anti-horseradish peroxidase (HRP) 488 and 549 conjugated (1:200), AlexaFluor Goat anti-HRP 647 conjugated (1:200). Secondary antibodies used include: donkey anti-rat (1:1600) (Jackson Immunoresearch) and donkey anti-rabbit (1:200) (Life Technologies). Filets were imaged on either a Nikon C1 Plus confocal microscope or a Zeiss LSM780 confocal microscope and fluorescence intensities quantified using the Measure–mean gray value function in ImageJ [41] and were normalized to area to control for differences in md neuron subclass cell body size. Identical confocal settings for laser intensity and other image capture parameters were applied for comparisons of control vs. experimental samples.

Live imaging confocal microscopy, neuronal reconstruction, and morphometric data analyses

Live neuronal imaging was performed as previously described [23,30]. We focused on the dorsal cluster of md neurons including C-I ddaE neurons; C-III ddaF and ddaA neurons; and C-IV ddaC neurons as morphological representatives of these md neuron subclasses. Dendritic morphology was quantified as previously described [30]. Briefly, maximum intensity projections of confocal Z-stacks were exported as a jpeg or TIFF. Once exported, images were manually curated to eliminate non-specific auto-fluorescent spots (such as the larval denticle belts) using a custom designed program, Flyboys (freely available upon request). For total dendritic length measurements, images were processed and skeletonized in ImageJ [30,41]. Quantitative neuromorphometric information was extracted and compiled using custom Python algorithms. The custom Python scripts were used to compile the output data from the Analyze Skeleton ImageJ plugin and the compiled output data was imported into Excel (Microsoft). For total dendritic branches and number of terminal branches, images were reconstructed using NeuronStudio [42]. Branch number and order were then extracted using the centripetal branch labeling function and output data was compiled in Excel.

qRT-PCR

qRT-PCR analysis of WT and cut-IR expressing neurons was done in quadruplicates as previously described [23]. Briefly, UAS-mCD8::GFP expressing C-IV md neurons were isolated using superparamagnetic beads (Dynabeads MyOne Streptavidin T1, Invitrogen) that were coupled to biotinylated anti-CD8a antibody (eBioscience). RNA was then isolated from these cells using the miRCURY RNA Isolation Kit (Exiqon) and qRT-PCR was performed using the following pre-validated QuantiTect Primer Assays: Atg1 (QT00963536), Atg2 (QT00956963), Atg5 (QT00499723), Atg8a (QT00919695) and Atg18 (QT00960337). Expression data was normalized to RpL32 (QT00980007).

Statistical analysis and data availability

Statistical analyses of neuromorphometric data and data plotting were performed using GraphPad Prism 7. Error bars reported in the study represent SEM. Statistical analyses were performed using either two-tailed unpaired t-test with Welch’s correction or one-way ANOVA using Dunnett’s multiple comparisons test when data sets were normally distributed as determined by the Shapiro-Wilk normality test. When data was not normally distributed, appropriate non-parametric tests were used (see figure legends for specific tests used in each case). Significance scores indicated on graphs are (* = p≤0.05, ** = p≤0.01, *** = p≤0.001). Detailed information on statistical analyses for each figure is reported in S2 Table. All newly reported genotypes presented here are available upon request. The microarray data is publicly available via the following GEO accession numbers: GSE46154, GSE69353, and GSE83938.

Cut positively regulates the expression of Atg genes

The homeodomain transcription factor Cut has been shown to promote dendritic diversification in morphologically complex C-III and C-IV md neurons via regulation of a variety of cellular processes including the secretory pathway and the cytoskeleton [3,23,36,43]. Previous studies have demonstrated that Cut protein is differentially expressed in md neuron subclasses with the highest levels in C-III neurons, followed by moderate and low levels in C-IV and C-II md neurons, respectively [25]. The levels of Cut expression are diversified in md neuron subclasses via the transcriptional regulators Scalloped and Vestigial [44]. In a recent study, we reported on comparative neurogenomic profiling of Cut-mediated transcriptional targets implicating a large number of differentially expressed genes and cellular processes that are positively regulated by Cut [36]. For these analyses, we capitalized on the observation that Cut is not normally detectable in C-I md neurons and therefore ectopic expression of Cut in these neurons can be used to examine Cut-mediated gene expression relative to control C-I neurons. This neurogenomic strategy also eliminates potential experimental confounds that may result from Cut overexpression in md neuron subclasses that normally express Cut [36].

Bioinformatic analyses of Cut-mediated differential gene expression identified a variety of intriguing cellular pathways including coordinated upregulation of numerous Atg genes. Relative to wild-type (WT) control C-I md neurons, Cut misexpressing C-I neurons exhibited significantly increased expression of the following Atg genes (Fig 1A): Atg1, encoding an autophagy-specific serine/threonine protein kinase involved in the induction of autophagosome biogenesis [32]; Atg2, encoding a vacuolar protein sorting (VPS)-associated 13 domain containing protein involved in Atg protein cycling between peripheral sites and the phagophore assembly site (PAS); Atg5, encoding an E3-like ubiquitin ligase component involved in autophagosome vesicle completion via lipidation of Atg8; Atg8a, encoding a ubiquitin-like protein involved in autophagosome vesicle completion; and Atg18, encoding a WD40 repeat domain phosphoinositide-interacting protein involved in protein cycling and autophagosome formation [12].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Cut positively regulates the expression of Atg genes.

(A) Microarray analyses reveal that ectopic expression of Cut in C-I md neurons results in upregulation of select Atg genes. Data is presented as average fold change in expression from wild-type (WT) control C-I md neurons (red dashed line). (B) C-I md neurons ectopically expressing Cut (CI>cut) show a significant (p<0.0001) increase in Atg8a protein expression as compared to WT. (C) Representative confocal images of C-I md neuron cell bodies from third instar larval WT control and C-I driven cut overexpression (CI>cut) genetic backgrounds. Third instar larval filets were stained with anti-Atg8a antibodies to mark autophagosomes and fluorescently-conjugated HRP to label md sensory neurons. (D) Normalized average expression values for Atg genes based on WT class-specific (C-I, C-III, C-IV) microarray analyses. Key includes graphic depicting increasing Cut protein expression levels by md neuron subclass. (E) qRT-PCR analyses reveal that knockdown of cut in C-IV neurons results in substantial reduction of mRNA levels for the selected Atg genes. (F,G) Representative confocal images of C-I (G), C-III, and C-IV (F) md neuron cell bodies from third instar larval WT control and pan-md driven cut RNAi knockdown (md>cut-IR) genetic backgrounds. Third instar larval filets were stained with anti-Atg8a antibodies to mark autophagosomes and fluorescently-conjugated HRP to label md sensory neurons. (H) Quantitative analyses of Atg8a fluorescence intensity normalized to area in C-I, C-III, and C-IV md neurons reveals significantly reduced expression in both C-IV and C-III neurons in cut-IR relative to WT controls. There is no significant change in Atg8a levels in C-I md neurons between cut-IR and WT controls. Statistics: two-tailed unpaired t-test with Welch’s correction (B,H) or one-way ANOVA using Dunnett’s multiple comparisons test (E) (** = p≤0.01, *** = p≤0.001). For detailed genotypes see S1 Table. For detailed statistics see S2 Table. a.u. = arbitrary units. Quantitative data is reported as mean ± SEM in all figures unless indicated otherwise.

https://doi.org/10.1371/journal.pone.0206743.g001

To validate the putative regulatory relationship between Cut and Atg proteins independent of the neurogenomic analyses, we performed immunofluorescence analyses to determine whether Cut can directly increase Atg protein expression. For these analyses we examined the expression of Atg8a based upon characterized and available antibodies. Relative to WT control C-I md neurons, we observed a significant increase in Atg8a expression levels in C-I neurons ectopically expressing Cut (p<0.0001) (Fig 1B and 1C).

As Cut protein is not normally detectable in C-I neurons, we next investigated whether patterns of Atg gene expression in C-III and C-IV md neurons displayed any relationship to Cut protein expression levels in these neurons. Based upon previously published WT C-I, C-III, and C-IV md neuron transcriptome expression profiling data [30,36,37], we discovered that the normalized mRNA expression values of the Atg genes identified above were largely correlated with differential Cut protein expression levels. With the exception of Atg5, we observed the highest levels of Atg gene expression in C-III neurons followed by C-IV neurons, which corresponds with previously reported Cut differential expression levels (Fig 1D) [25]. Between md neuron subclasses, Atg genes were relatively enriched in C-III and C-IV neurons compared to C-I neurons (Fig 1D). To further confirm the regulatory relationship between Cut and Atg genes we demonstrate that cut knockdown in C-IV neurons results in significant reductions in the mRNA expression levels of all five of these genes as measured by qRT-PCR (p<0.0001 for all five genes) (Fig 1E).

To complement the mRNA expression analyses, we performed RNAi-mediated knockdown of cut (UAS-cut-IR) using a pan-md GAL4 driver and labeled third instar larval filets with anti-Atg8a. We focused our analyses on C-I, C-III, and C-IV md neurons based upon differential Cut protein expression levels and our neurogenomic analyses. In WT control neurons, Atg8a protein expression appeared in punctate cytosolic vesicular structures representing presumptive autophagosomes throughout the md neuron cell bodies (Fig 1F and 1G). Quantitative analysis of C-I, C-III, and C-IV neurons in the dorsal cluster of md neurons revealed significant reductions in Atg8a protein expression levels with cut knockdown in the C-III (p<0.0001) and C-IV (p = 0.0093) md neurons compared to WT controls, supporting a positive regulatory relationship between Cut and Atg8a (Fig 1F and 1H). In contrast, we observed no significant change in Atg8a protein expression upon cut knockdown in the C-I md neurons (p = 0.1203), which was predicted based upon the lack of detectable Cut protein expression in C-I neurons [25] (Fig 1G and 1H). Collectively, these findings indicate that Cut positively regulates components of the autophagic machinery.

Basal autophagy functions as a downstream effector of Cut-mediated dendritic terminal branching

The regulatory relationship identified between Cut and Atg genes suggests that Cut may utilize the autophagy pathway to drive md neuron dendritic development and perhaps to regulate specific aspects of dendritic morphogenesis that contribute to md neuron subtype-specific dendritic diversification. If Cut requires these Atg genes, then we would predict that Atg gene disruption should result in a suppression of Cut-mediated dendritogenesis resulting from Cut ectopic expression in C-I md neurons. Relative to WT controls, ectopic expression of Cut in C-I md neurons (CI>cut) causes a dramatic increase in dendritic branching complexity characterized by extensive de novo formation of dendritic terminal filopodia-like branches emanating from the primary arbors (Fig 2A). Cut ectopic expression in C-I neurons results in significant increases in the total number of dendritic branches (Fig 2B), total dendritic length (Fig 2C), and number of terminal dendrites (Fig 2D). To test the hypothesis that Atg genes act as downstream effectors of Cut-mediated dendritogenesis in these neurons, C-I neurons ectopically expressing Cut were phenotypically compared to C-I neurons in which Cut was ectopically expressed with simultaneous expression of gene-specific RNAi for Atg genes identified as putative Cut targets. We have previously established that suppression of the Cut ectopic expression phenotype is not the result of GAL4 titration via co-expression of a neutral UAS-mCD8::RFP transgene [23], thus any morphological suppression effects are attributable to Atg-IR expression rather than GAL4 titration. Compared to C-I neurons ectopically expressing Cut alone, C-I neurons co-expressing Cut and Atg gene-specific RNAi transgenes (Atg-IR) displayed suppression of Cut-mediated de novo dendritic arborization, which was particularly notable with respect to Cut-induced short terminal dendritic filopodial branching (Fig 2A, compare insets). Quantitative neuromorphometric analyses support this observation as knockdown of Atg genes largely resulted in significant reductions in the total number of dendritic branches, total dendritic length, and number of terminal dendrites with a few exceptions (Fig 2B–2D, S2 Table). In most cases, there were significant concomitant reductions in the total number of branches and number of terminal dendrites with Atg gene knockdown (Fig 2B–2D) indicative of a requirement for autophagy in promoting the formation of Cut-induced de novo dendritic filopodial terminals. In addition to knockdown analyses, we drove expression of UAS-Atg1^K38Q, a kinase dead transgene of Atg1 [32], in C-I neurons ectopically expressing Cut. Expression of the kinase dead transgene competitively inhibits native Atg1 thereby impairing autophagic function through a reduction in kinase activity and autophagosome biogenesis [32]. Quantitative analyses of C-I neurons co-expressing Cut and UAS-Atg1^K38Q revealed significant reductions for all neuromorphometric parameters examined (Fig 2B–2D). These findings suggest that Cut-mediated dendritic terminal branching requires the basal autophagy pathway.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Autophagy is required for Cut-mediated dendritic terminal branching.

(A) Representative confocal images of third instar larval C-I ddaE md neurons. Compared to WT, ectopic expression of Cut (CI>cut) dramatically increases dendritic branching complexity. Simultaneous misexpression of Cut and gene-specific UAS-RNAi transgenes for select Atg genes (e.g. +Atg1-IR) reveals suppression of Cut-mediated increases in dendritic growth and terminal branching compared to Cut misexpression alone. Insets present magnified view of corresponding dendritic regions between genotypes. (B-D) Quantitative neuromorphometric analyses of the total number of dendritic branches (B), total dendritic length in voxels (C) and number of terminal dendrites (D). Statistics: Kruskal-Wallis test using Dunn’s multiple comparisons test (* = p≤0.05, ** = p≤0.01, *** = p≤0.001. For detailed genotypes see S1 Table. For detailed statistics see S2 Table.

https://doi.org/10.1371/journal.pone.0206743.g002

Previous studies have demonstrated that cut mutant C-III md neurons exhibit defects in dendritic growth and loss of terminal dendritic filopodia-like branches, which are a characteristic feature of C-III neurons [25]. Consistent with these previous findings, we find that C-III-specific RNAi knockdown of cut leads to reductions in growth and a dramatic loss of dendritic terminal branches relative to WT controls (Fig 3A). If Cut utilizes the basal autophagy machinery to promote growth and dendritic terminal branching, then we would predict that upregulation of autophagic function may be sufficient to rescue aspects of defects in dendritic arborization associated with cut disruption. To test this hypothesis, we simultaneously knocked down cut and overexpressed the autophagy initiator kinase Atg1 in C-III neurons. Consistent with our prediction, Atg1 overexpression in a cut-IR loss-of-function genetic background resulted in a partial rescue of dendritic morphology defects relative to cut-IR alone (Fig 3A). Atg1 overexpression leads to significant increases in both the total number of dendritic branches (p<0.0001) and number of terminal dendrites (p = 0.0010), although total dendritic length is not rescued (p = 0.1882) (Fig 3B–3D). To further validate this rescue effect, we also performed comparative analyses with overexpression of Atg5 and Atg8a in a cut-IR background. Expression of these Atg genes resulted in partial rescue of the cut-IR phenotype (Fig 3A). Interestingly, these two genes, which function later in the autophagic process than Atg1, demonstrated significant rescue of dendritic length (Atg5: p = 0.0004, Atg8a: p = 0.0343) but less significant rescue of total dendritic branches (Atg5: p = 0.0124, Atg8a: p = 0.0172) and terminal dendrites (Atg5: not significant, Atg8a: p = 0.0396) relative to Atg1 (Fig 3B–3D). Collectively, these results support a role for the basal autophagy pathway as a downstream effector of Cut-mediated dendrite morphogenesis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Autophagy partially rescues dendritic complexity deficits caused by cut knockdown in C-III md neurons.

(A) Representative confocal images of third instar larval C-III ddaA md neurons. Compared to WT, expression of cut-IR results in severe loss of terminal dendritic filopodia-like branching. Simultaneous expression of cut-IR and overexpression of Atg1, Atg5, or Atg8a results in partial rescue of dendritic growth and branching deficits. (B-D) Quantitative analyses reveal that C-III Atg1 overexpression in a cut-IR loss-of-function background (CIII>Atg1;cut-IR) partially rescues reductions in the total number of dendritic branches (B) and number of terminal dendrites (D), but fails to rescue deficits in total dendritic length (C) relative to cut-IR loss-of-function C-III neurons (CIII>cut-IR), while Atg5 overexpression in a cut-IR loss-of-function background partially rescues reductions in the total number of dendritic branches (B) and total dendritic length (C) but fails to rescue reductions in the number of terminal dendrites (D), and Atg8a overexpression in a cut-IR background partially rescues all three morphological aspects. Statistics: one-way ANOVA with Dunnett’s multiple comparisons test (* = p≤0.05, *** = p≤0.001). For detailed genotypes see S1 Table. For detailed statistics see S2 Table.

https://doi.org/10.1371/journal.pone.0206743.g003

Homeostatic regulation of basal autophagy is required for class-specific dendritic growth and branching

We next conducted phenotypic analyses to investigate the putative roles of autophagy in regulating cell type-specific dendritic development in C-III and C-IV md neurons, which natively express Cut at high and moderate levels, respectively. Relative to controls, the most notable effect of C-III specific knockdown of Atg genes was a reduction in the occurrence of terminal dendritic filopodia-like branches emanating from the primary arbors (Fig 4A). Quantitative morphometric analyses identified significant reductions in the total number of dendritic branches and number of terminal dendrites upon knockdown of all Atg genes analyzed, as well as significant reductions in all but Atg1-IR for total dendritic length. These results suggest that the major effect on branching was localized to C-III dendritic terminals. These morphological defects were likewise observed with C-III-specific expression of the Atg1^K38Q kinase dead transgene (Fig 4B–4D, see S2 Table for p-values).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Atg genes are required to promote C-III md neuron dendritic growth and terminal branching.

(A) Representative confocal images of third instar larval C-III ddaF md neurons. Compared to WT, knockdown of autophagic activity via an Atg1 kinase dead transgene (K38Q) and UAS-Atg-IR transgenes results in reductions in dendritic growth and terminal branching. Insets present magnified view of corresponding dendritic regions between genotypes. (B-D) Quantitative analyses reveal reductions in the total number of dendritic branches (B), total dendritic length (C), and number of terminal dendrites (D) upon knockdown of Atg genes. Statistics: one-way ANOVA using Dunnett’s multiple comparisons test (* = p≤0.05, *** = p≤0.001). For detailed genotypes see S1 Table. For detailed statistics see S2 Table.

https://doi.org/10.1371/journal.pone.0206743.g004

In the case of C-IV neurons, knockdown of Atg genes resulted in variable qualitative phenotypic defects largely manifesting as reductions in dendritic arbor growth and terminal branching (Fig 5A). Quantitative analyses revealed significant reductions in the total number of dendritic branches (Fig 5B) and number of dendritic terminals (Fig 5D) for all Atg genes analyzed, whereas significant reductions total dendritic length (Fig 5C) were observed for all of the Atg genes except Atg8a (see S2 Table for p-values).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Atg genes function in regulating C-IV dendritic development and terminal branching.

(A) Representative confocal images of third instar larval C-IV ddaC md neurons. Compared to WT, knockdown of autophagic activity via an Atg1 kinase dead transgene (K38Q) and UAS-Atg-IR transgenes results in reductions in growth and terminal branching. (B-D) Quantitative analyses reveal reductions in the total number of dendritic branches (B) for all Atg genes investigated, decreases in total dendritic length (C) for all Atg genes investigated except Atg8a, and decreases in the number of terminal dendrites (D) without exception. Statistics: one-way ANOVA using Dunnett’s multiple comparisons test (*** = p≤0.001). For detailed genotypes see S1 Table. For detailed statistics see S2 Table.

https://doi.org/10.1371/journal.pone.0206743.g005

Previous studies have implicated both insufficient and excessive autophagic induction in neurodegeneration as well as degeneration of axo-dendritic processes in patients suffering from a variety of neurodegenerative disorders including Alzheimer’s (AD), Parkinson’s (PD) and Huntington’s (HD) diseases (4,12,18). These findings suggest that homeostatic regulation of autophagy is important for maintaining neuritic architecture and neuronal survival. To investigate how excessive autophagic induction may impact md neuron dendritic development, we conducted gain-of-function phenotypic analyses of the Atg1 initiator kinase in C-III and C-IV neurons. Relative to controls, Atg1 overexpression drastically decreased dendritic growth and higher order terminal branching in both C-III and C-IV md neurons (Fig 6A, 6B, 6G and 6H) resulting in severe reductions (p<0.0001 in all cases) in the number of dendritic branches (Fig 6D and 6J), total dendritic length (Fig 6E and 6K), and number of terminal dendrites (Fig 6F and 6L) in both C-III and C-IV md neurons. Interestingly, when we examine the effects of Cut overexpression in C-III and C-IV neurons, the phenotypic defects observed are strikingly similar to those observed with Atg1 overexpression in these neurons (Fig 6C and 6I). This observation is quantitatively supported in that Cut overexpression led to significant reductions in the total number of dendritic branches (Fig 6D and 6J), total dendritic length (Fig 6E and 6K), and number of terminal dendrites (Fig 6F and 6L) that were consistent with the Atg1 overexpression phenotypes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Atg1 or Cut overexpression restricts dendritic growth and higher order terminal branching.

(A-C) Representative confocal images of third instar larval dorsal C-III ddaF md neurons. (G-I) Representative confocal images of third instar larval C-IV ddaC md neurons. Compared to WT (A,G), overexpression of Atg1 (B,H) or Cut (C,I) results in reductions in dendritic growth and terminal branching. Quantitative analyses reveal significant decreases in the total number of dendritic branches (D,J), total dendritic length (E,K), and number of terminal branches (F,L). Statistics: one-way ANOVA using Dunnett’s multiple comparisons test (E, J-L) or Kruskal-Wallis test using Dunn’s multiple comparisons test (D,F) (* = p≤0.05, *** = p≤0.001). For detailed genotypes see S1 Table. For detailed statistics see S2 Table. OE = overexpression.

https://doi.org/10.1371/journal.pone.0206743.g006

Collectively, these analyses suggest that the autophagy pathway is required to promote terminal dendritic branching in both C-III and C-IV md neurons as well as to regulate cell type-specific dendritic arborization architectures. Moreover, these loss-of-function studies are in line with findings from the suppression analyses in C-I neurons where knockdown of Atg genes appears to largely suppress the formation of Cut-induced dendritic growth and terminal filopodial branching (Fig 2). Furthermore, gain-of-function analyses reveal that excessive autophagic induction leads to severe reductions in dendritic growth and branching. Combined with the loss-of-function studies, this suggests that homeostatic regulation of the basal autophagy pathway is required for class-specific dendritic arborization.

Atg1 interacts with the DLK pathway in C-IV md neurons

To gain insight into putative mechanisms by which the autophagy pathway may regulate cell type-specific md neuron dendrite morphogenesis, we focused on potential interactions with the dual leucine zipper kinase (DLK) pathway. Previous studies demonstrated that the autophagy pathway regulates synaptic and axonal growth via degradation of the E3 ubiquitin ligase Highwire (Hiw) in CNS neurons [15], and that disruptions of the DLK pathway, which is comprised of Hiw and the MAPKKK Wallenda (Wnd), produce bimodal and opposing effects on axonal and dendritic development of C-IV md neurons [45]. More precisely, Hiw functions to promote C-IV dendritic arbor growth and branching, whereas Wnd suppresses it [45]. Consistent with previous findings, we independently confirmed these phenotypic requirements in C-IV neurons by loss-of-function and gain-of-function studies of hiw and wnd, respectively. C-IV specific expression of a hiw mutant transgene (UAS-Hiw^ΔRING) in which the Hiw E3 ubiquitin ligase RING domain is mutated generating a loss-of-function effect revealed defects in dendritic arbor growth and branching (Fig 7B) reminiscent of the defects observed with expression of the Atg1^K38Q kinase dead transgene (Fig 5A). In contrast, C-IV overexpression of Wnd dramatically reduced dendritic growth and higher order terminal branching (Fig 7D) in a manner that was phenotypically similar to the defects observed with Atg1 overexpression (Fig 7E). We therefore investigated whether the autophagy pathway may retain the same regulatory relationship with the DLK pathway in dendritic development as previously observed in axon growth and synapse development [15]. To assess this putative regulatory relationship we expressed a kinase dead Wnd transgene, UAS-wnd.K188A, in C-IV md neurons alone and in combination with Atg1 overexpression (Fig 7C and 7F). Consistent with previous findings [45], expression of the kinase dead Wnd transgene resulted in no notable morphological defects (Fig 7C), however co-expression with Atg1 resulted in dramatic and significant (p<0.0001 for all three measures) rescue of the morphological deficits induced by Atg1 overexpression (Fig 7F–7I), suggesting that Wnd functions downstream of Atg1 to regulate dendritic morphology in C-IV neurons. To assess the involvement of Hiw, we expressed a UAS-Hiw::GFP transgene [33] in C-IV md neurons with and without co-expression of UAS-Atg1. Live imaging of Hiw::GFP fluorescence revealed a significant (p<0.0001) decrease in the levels of GFP-tagged Hiw in neurons co-expressing UAS-Atg1 relative to controls, thereby suggesting that Atg1 overexpression negatively regulates Hiw expression in C-IV neurons (Fig 7J and 7L). To further examine this relationship we co-expressed UAS-Hiw::GFP and the Atg1^K38Q kinase dead transgene. If Atg1 negatively regulates Hiw then we would expect to see an increase in Hiw::GFP fluorescence as the kinase dead Atg1 outcompetes native Atg1, and consistent with this prediction, we observed a significant (p<0.0001) increase in levels of GFP-tagged Hiw (Fig 7J and 7L). Based upon the negative regulatory relationship between Hiw and Wnd, we predicted that Atg1-mediated downregulation of Hiw would lead to a concomitant increase in Wnd expression levels. To investigate this, we compared Wnd expression levels in control and Atg1 overexpressing C-IV neurons and discovered a notable increase in punctate Wnd staining in the cell body (Fig 7K). Quantitative analysis revealed significant (p = 0.0038) increases in Wnd expression upon Atg1 overexpression relative to WT controls (Fig 7M). These results suggest that Atg1 can interact with the DLK pathway by regulating both Hiw and Wnd expression levels and that autophagy may utilize the DLK pathway to direct C-IV dendrite morphogenesis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Atg1 negatively regulates Hiw and positively regulates Wnd in C-IV md neurons.

(A-F) Representative confocal images of third instar larval C-IV ddaC md neurons. Compared to WT controls (A), disruption of the E3 ubiquitin ligase activity of Hiw (Hiw^ΔRING) (B) and overexpression of the MAPKKK Wnd (D) result in reductions to C-IV dendritic growth and higher order terminal branching while overexpression of a kinase dead Wnd (C) has no notable effect on morphology, consistent with previous studies [45]. (E) Atg1 overexpressing C-IV neurons likewise exhibit defects in dendritic growth and higher order terminal branching consistent with defects observed with Wnd overexpression (D), and these defects are rescued when Atg1 overexpression is coupled with expression of kinase dead Wnd (F). (G-I) Quantitative analyses reveal significant decreases in the total number of dendritic branches, total dendritic length, and number of terminal branches for Atg1 overexpression as ^compared to WT. Co-expression of Atg1 and kinase dead Wnd (wnd.K188A), however, demonstrates significant rescue of all measured aspects of the Atg1 overexpression phenotype. (J) Representative confocal images of third instar larval C-IV md neurons expressing UAS-Hiw::GFP alone compared to C-IV md neurons expressing UAS-Hiw::GFP along withUAS-Atg1 or UAS-Atg1^K38Q. ppk-hCD4::tdTomato was used for C-IV md neuron visualization. (L) Quantification of GFP signal (fluorescence intensity) reveals a significant (p<0.0001) decrease in GFP-tagged Hiw expression upon Atg1 overexpression, whereas increased (p<0.0001) Hiw::GFP expression was observed upon expression of the kinase dead Atg1^K38Q. (K) Representative confocal images of third instar larval WT C-IV md neurons and neurons overexpressing UAS-Atg1 labeled with anti-Wnd antibodies. HRP labeling was used for neuron visualization. (M) Quantitative analysis reveals a significant (p = .0038) increase in Wnd expression (fluorescence intensity) with Atg1 overexpression. Statistics: one-way ANOVA using Dunnett’s multiple comparisons test (G-I, L) or two-tailed unpaired t-test with Welch’s correction (M) (** = p≤0.01, *** = p≤0.001). For detailed genotypes see S1 Table. For detailed statistics see S2 Table. OE = overexpression; a.u. = arbitrary units.

https://doi.org/10.1371/journal.pone.0206743.g007

Autophagic induction partially rescues dendritic defects under conditions of cellular stress induced by polyglutamine toxicity

Autophagy has been demonstrated to play an essential neuroprotective role with respect to inhibiting neurodegeneration and loss of neuritic processes [4]. This led us to question whether autophagic induction may be capable of ameliorating loss of dendritic complexity associated with conditions of cellular stress. Recent studies in md neurons have modeled the neurodegenerative-like defects that occur in the polyglutamine (polyQ) disease spinocerebellar ataxia type 3 (SCA3), also known as Machado-Joseph disease (MJD). These studies revealed that expression of pathogenic SCA1 and SCA3 proteins results in severe dendritic atrophy [46]. Consistent with previous analyses, we found that C-IV overexpression of a transgene encoding a C-terminal truncated form of SCA3 (UAS-MJD-78Q) that retains a long polyQ repeat leads to severe defects in dendritic growth and higher order terminal branching presumptively linked to polyQ toxicity (Fig 8A). To determine whether the severe dendritic growth and branching deficits associated with SCA3 pathology could be rescued by autophagic induction, we overexpressed the Atg1 initiator kinase in C-IV neurons expressing the MJD-78Q transgene. Relative to MJD-78Q alone, Atg1 overexpression led to a partial rescue of the dendrite growth and higher order branching defects resulting from expression of SCA3 aggregates (Fig 8A and 8B). Quantitative morphometric analyses reveal significant rescue of the number of dendritic branches (Fig 8C, p = 0.0067), total dendritic length (Fig 8D, p = 0.0053), and number of terminal dendrites (Fig 8E, p = 0.0056). These results suggest that Atg1-mediated autophagic induction is capable of partially rescuing neurodegenerative-like dendritic arbor atrophy in the SCA3 aggregate toxicity model.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Atg1 overexpression partially rescues dendritic atrophy in a neuronal model of SCA3 polyQ toxicity.

(A,B) Representative confocal images of third instar larval C-IV ddaC md neurons expressing MJD-78Q with and without co-expression of Atg1. Expression of MJD-78Q alone (A) results in severe dendritic atrophy. Simultaneous expression of MJD-78Q and Atg1 (B) results in partial rescue of dendritic atrophy leading to increased growth and higher order branching. (C-E) Quantitative analyses reveal partial rescue of total number of dendritic branches (p = 0.0067) (C), total dendritic length (p = 0.0053) (D), and number of terminal dendrites (p = 0.0056) (E). Statistics: two-tailed unpaired t-test with Welch’s correction (** = p≤0.01). For detailed genotypes see S1 Table. For detailed statistics see S2 Table.

https://doi.org/10.1371/journal.pone.0206743.g008