Transcriptome Profiling Identifies Multiplexin as a Target of SAGA Deubiquitinase Activity in Glia Required for Precise Axon Guidance During Drosophila Visual Development

Abstract

The Spt-Ada-Gcn5 Acetyltransferase (SAGA) complex is a transcriptional coactivator with histone acetylase and deubiquitinase activities that plays an important role in visual development and function. In Drosophila melanogaster, four SAGA subunits are required for the deubiquitination of monoubiquitinated histone H2B (ubH2B): Nonstop, Sgf11, E(y)2, and Ataxin 7. Mutations that disrupt SAGA deubiquitinase activity cause defects in neuronal connectivity in the developing Drosophila visual system. In addition, mutations in SAGA result in the human progressive visual disorder spinocerebellar ataxia type 7 (SCA7). Glial cells play a crucial role in both the neuronal connectivity defect in nonstop and sgf11 flies, and in the retinal degeneration observed in SCA7 patients. Thus, we sought to identify the gene targets of SAGA deubiquitinase activity in glia in the Drosophila larval central nervous system. To do this, we enriched glia from wild-type, nonstop, and sgf11 larval optic lobes using affinity-purification of KASH-GFP tagged nuclei, and then examined each transcriptome using RNA-seq. Our analysis showed that SAGA deubiquitinase activity is required for proper expression of 16% of actively transcribed genes in glia, especially genes involved in proteasome function, protein folding and axon guidance. We further show that the SAGA deubiquitinase-activated gene Multiplexin (Mp) is required in glia for proper photoreceptor axon targeting. Mutations in the human ortholog of Mp, COL18A1, have been identified in a family with a SCA7-like progressive visual disorder, suggesting that defects in the expression of this gene in SCA7 patients could play a role in the retinal degeneration that is unique to this ataxia.

Keywords: SAGA; SCA7; axon guidance; glia; histone ubiquitination.

Figure 1

Glial nuclei are labeled with KASH-GFP in wild-type, nonstop and sgf11 optic lobes. (A) Genetic scheme for positive labeling of glial nuclei in nonstop or sgf11 homozygous mutant larvae. The expected percentage of progeny with each genotype at the larval stage is indicated. (B) Glial nuclei in a wild-type optic lobe were positively labeled with KASH-EGFP (GFP, green) and stained with DAPI (blue). Arrowheads mark the edges of the lamina (la), and the medulla (me) is indicated. A single KASH-GFP labeled glial nuclei is indicated by the arrow. Scale bars, 20 μm. (C–E) Glial nuclei in wild-type (C), nonstop (D), and sgf11 (E) optic lobes were positively labeled with KASH-EGFP (GFP, green). Mutant genotypes (panels D and E) were labeled using the genetic approach described in panel A. In third instar larvae, R1–R8 photoreceptor axons (anti-chaoptin, mAB24B10, magenta) extend axons from the eye imaginal disc into the optic lobe where these terminate in either the lamina (R1–R6, la) or medulla (R7–R8, me). In the wild-type optic lobe, R1–R6 photoreceptor axons form a densely packed growth cone, whereas in nonstop and sgf11 optic lobes, a subset of photoreceptor axons project through the lamina into the medulla. Arrowheads mark the edge of lamina. Scale bars, 20 μm. (F) Bar graph indicating the transcript level (FPKM) for glial markers in affinity-enriched glia (glia: this study) relative to the larval central nervous system (CNS: ModEncode). (G) Bar graph indicating the transcript level (FPKM) for neuronal markers in affinity-enriched glia relative to the larval central nervous system. CNS, central nervous system; DAPI, 4′,6-diamidino-2-phenylindole; EGFP, enhanced green fluorescent protein; FPKM, fragments per kilobase of transcript per million mapped reads; GFP, green fluorescent protein; KASH, Klarsicht, Anc-1, and Syn3-1 homology; WT, wild-type.

Figure 2

Mutations that disrupt SAGA deubiquitinase activity result in misregulated gene expression in optic lobe glia. (A) Hierarchical clustered heat map representing the Euclidean distance matrix of each genotype and biological replicate (n = 4) for RNA-seq analysis calculated based on regularized log₂ transformed counts of each gene. More closely related samples are shown in light green (smallest distance, 0), and more distantly related samples are shown in black (largest distance, 50). (B) Venn diagrams indicating the number of overlapping genes that are significantly upregulated or downregulated (FDR < 0.01) in nonstop and sgf11 optic lobe glia compared to wild-type optic lobe glia. (C) Scatter plots illustrating the average differential expression of each gene (dot) in mutant/wild-type glia relative to its average expression across all samples. Each gene is plotted based on the log₂ expression ratio in either the nonstop (upper panel) or sgf11 (lower panel) mutant/wild type (y-axis) relative to the log₂ of its average expression level in CPM across all genotypes (x-axis). Genes that were identified as being significantly differentially regulated in both mutant genotypes are shown in red, and genes identified as significantly differentially regulated only in a single mutant genotype are shown in blue. (D) qRT-PCR analysis of transcript levels in cDNA from wild-type, nonstop, and sgf11 optic lobe glia for a subset of the differentially regulated genes was compared with fold changes observed in the RNA-seq analysis. Mean transcript levels for each gene were normalized to RpL32 and plotted relative to the wild type, which was set to one (right panel). RNA-seq result with fold change of genes of interest were plotted using raw abundance values compared to the wild type, which was set to one (left panel). Error bars denote standard error of the mean for three biological replicates for qRT-PCR analysis and four biological replicates for RNA-seq. CPM, count per million; FDR, false discovery rate; qPCR, quantitative polymerase chain reaction; RNA-seq, RNA sequencing; SAGA, Spt-Ada-Gcn5 acetyltransferase; WT, wild-type.

Figure 3

SAGA deubiquitinase-regulated genes are enriched for Gene Ontology terms including proteasome function, protein folding, and axon targeting. Significantly enriched biological process GO terms (FDR < 0.001) of upregulated genes (black) and downregulated genes (red) in both nonstop and sgf11 optic lobe glia. Adjusted p-values for each GO term are shown to the right of each bar. Enrichment (x-axis) represents the fold increase of the number of genes in each GO term over the number expected by chance. FDR, false discovery rate; GO, Gene Ontology; SAGA, Spt-Ada-Gcn5 acetyltransferase.

Figure 4

SAGA deubiquitinase-regulated gene Multiplexin is required in glia for proper photoreceptor axon targeting. (A) Venn diagram showing the overlap between genes expressed in glia greater than one-and-a-half-fold increase relative to the central nervous system, genes that are significantly downregulated in both nonstop and sgf11 glial relative to the wild type (FDR < 0.01), and genes with GO terms that contain the term axon guidance. Genes selected for RNAi screening are highlighted in red. (B–D) Axon targeting defects were examined using X-gal staining of the R2–R5 axon marker ro-τlacZ and classified as either no defect (mistargeted axons ≤ 1, panel B), mild defect (mistargeted axons ≤ 4, panel C), or severe defect (mistargeted axons ≥ 5, panel D). Representative images for each category from glial-specific expression of RNAi against Luciferase (panel B), Mp (panel C), and sgf11 (panel D) are shown. (E) Stacked bar graph indicating the percentage of optic lobes exhibiting defects in R1–R6 photoreceptor axon targeting upon expression of RNAi against the indicated genes expressed in glia under repo-GAL4 control. n, number of optic lobes analyzed. Fisher’s exact test was performed between each Mp RNAi genotype (two independent lines) ± repo-Gal4 and the p-value is indicated for each comparison next to each bar. CNS, central nervous system; FDR, false discovery rate; GO, Gene Ontology; RNAi, RNA interference; SAGA, Spt-Ada-Gcn5 acetyltransferase.

Figure 5

Glial-specific expression of RNAi against Multiplexin modestly disrupts lamina glial organization, which correlates with photoreceptor axon mistargeting. (A) Lamina glia were visualized in WT and sgf11 optic lobes using the loco^rC56 marker (blue), and R1–R8 photoreceptor axons were labeled using anti-chaoptin (mAB24B10; red). In WT larvae, glia migrate appropriately to the lamina ganglia where they express the loco^rC56 marker. In sgf11 larvae, although some glia migrate appropriately and express loco^rC56 (boxed region R1), many glia are absent from the lamina (boxed region R2), correlating with mistargeting of photoreceptor axons. Merged images for glia and axons are shown in the upper panel, and single channel images for glia alone (loco^rC56) are shown in the lower panel. Optic stalk, os; Lamina, la; Medulla, me. Scale bars, 20 μm. (B–D) RNAi constructs against the indicated genes were expressed specifically in glia using the repo-GAL4 driver, and targeting of photoreceptor axons was examined using the R2–R5 photoreceptor axon marker ro-τlacZ (left and middle panels, green). Glial cells were labeled for comparison using anti-repo (middle and right panels, magenta). Maximum projection images of 0.5 μm z-stacks are shown for each knock-down. The positions of individual mistargeted photoreceptor axons are indicated by arrowheads in each panel. The expected position of lamina glial cells is indicated by dotted lines in the right panels for each genotype. Scale bars, 20 μm. RNAi, RNA interference; WT, wild-type.