Fig 1.
Analysis pipeline and expression dynamics at stage transitions.
(A) The mRNA of eye-antennal imaginal discs was sequenced at three developmental stages: late L2, mid L3 and late L3. Upwards arrows illustrate the applied analysis pipeline. The total number of expressed genes for each stage transition is shown below the imaginal discs (B: late L2 to mid L3; C: mid L3 to late L3). The numbers next to the green arrows represent upregulated genes in each transition and the numbers next to the red arrow represent downregulated genes. (B) GO term enriched in the upregulated (green, left) and downregulated (red, right) genes in the transition from late L2 to mid L3 stage. (C) GO terms enriched in the upregulated (green, left) and downregulated (red, right) genes in the transition from mid L3 to late L3 stage.
Fig 2.
All 13 profiles of co-expressed genes predicted by HTSCluster (see Materials and Methods). The number of genes assigned to a particular cluster are indicated below the cluster name. Colored dots represent relative expression levels (lambda value) of the genes of that cluster (y-axis on the right) at each stage. Background grey lines represent the normalized mean count of all genes belonging to a cluster (y-axis on the left). Below each cluster plot, the first four non-redundant GO terms enriched in the genes of that cluster are listed (see S1 Table) and the significantly enriched transcription factors (NES > 3, see S2 Table) are shown (the NES value is given in brackets).
Fig 3.
Assignment of key retinal genes to identified co-expression clusters.
Directed gene interaction network (FlyOde,[50]) of the genes known to be involved in retinal development in D. melanogaster. Only genetic and protein-DNA interactions are included. The horizontal axis represents developmental time, with early L3 at the most left, and the vertical axis displays network hierarchy (please see [50] for details). Each node represents a gene, and these are color-coded according to the assigned co-expression cluster (see also Fig 2). In grey are genes that have not been allocated to any cluster, either because they did not pass the expression level threshold or the clustering confidence threshold (see Materials and Methods for details).
Fig 4.
hb is expressed in migratory subperineurial glia cells.
(A-C) hb expression (VT038544-Gal4 driving histone-bound RFP (UAS-H2B::RFP) (A-C), red in A”-C”) co-localizes (white arrows) with the pan-glial marker Repo (detected with rabbit α-Repo antibody (A’-C’), green in A”-C”) in two large cells. These cells migrate from the optic stalk during L2 stage (A”) into the posterior end of the eye-antennal disc at mid L3 stage (B”) and are located in the posterior region of the retinal field, at each side of the optic stalk, at late L3 stage (C”). (VT038544-Gal4 driver line obtained from the Vienna Tile collection, see S4 Fig for details) (D) Hb (detected with rabbit α-Hb antibody, red in D”) is present in the same cells (white arrows) that express the subperineurial glia cell marker moody (D’) (moody-Gal4 driving UAS-GFP, green in D”). In all pictures, anterior is to the right. Eye disc (ed), optic stalk (os). Scale bar = 20 μm.
Fig 5.
The number of polyploid glia cell nuclei is reduced after loss of Hb function.
(A-C) Staining with rabbit α-Repo antibody (red) and Phalloidin (green) of late L3 eye-antennal discs in wild type (A), repo driven hb RNAi (B) and Hb temperature sensitive mutant (C). This figure represents the phenotypes that were analyzed in (D), where the number of polyploid nuclei (white arrows) have been quantified. In all pictures, anterior is to the right. Scale bar = 20 μm. (D) Box plot of glia cell nuclei size. When two carpet cells were present, these are significantly larger than other surrounding glia cells. When only one carpet cell was present in the retinal field, its nucleus was significantly larger than regular carpet cells. Significance calculated by t-test (“Two CCs” vs. “One CC”, homogeneous distribution of variances) and t-Welch-test (“Other glia” vs. “Two CCs”, not homogeneous distribution of variances); “***”: p < 0.0005. (E) Quantification of the number of polyploid nuclei observed in wild type (WT), repo-Gal4 and moody-Gal4 driven UAS-hb RNAi (hbdsRNA) and Hb temperature sensitive mutant (hbTS). (F) Quantification of the number of polyploid nuclei observed in late L3 eye-antennal discs of flies that were raised at 18°C until the indicated time points (24h AEL, 48h AEL, 72h AEL and 96h AEL), when they were transferred to the restrictive temperature of 28°C. In D and E, the black bar indicates percentage of discs with two polyploid glia cell nuclei, grey indicates discs with one polyploid glia cell nucleus and white indicates discs without polyploid glia cell nuclei. Pearson’s Chi-squared test was performed to determine if the distribution of the different number of cells (0, 1 or 2) was equal between wild type and RNAi. *: p-val < 0.05, ***: p-val < 0.0005.
Fig 6.
Carpet cell membranes after loss of carpet cells.
(A-C) All glia cells are stained with rabbit α-Repo antibody (red in A”-C”). In control discs, two carpet cells can be identified by their large size (white arrows). After moody driven knock-down of hb expression (B and C) some discs present only one carpet cell nucleus (B, white arrow) and others no carpet cell nucleus (C) (refer to Fig 5 for percentages). (A’-C’) Membranes of carpet cells in late L3 eye-antennal discs are labelled with moody-Gal4 driven UAS-mCD8::GFP expression (green in A”-C”). The presence of these membranes was analyzed in the optic stalk (orange arrowheads) and in the retinal field (determined by the presence of these membranes in the posterior margins of the disc and their extension to the MF (blue arrowheads)). In all pictures, anterior is to the right. Eye disc (ed), optic stalk (os). Scale bar = 20 μm. (D) Quantification of the presence of carpet cell membrane as described in A’-C’. Pearson’s Chi-squared test was performed to determine significance between each pair of condition. *: p-val < 0.05, ***: p-val < 0.0005.
Fig 7.
Blood-brain barrier function is impaired after loss of hb expression in carpet cells.
(A) After injection of fluorescently labelled dextran into the abdomen of adult flies, animals with correctly formed blood-brain barrier show no fluorescence in the compound eye. (A’) In flies with incomplete blood-brain barrier, fluorescent dye can be observed in the compound eye as well as in the body. (B) Quantification of eyes with (green) or without (red) dye penetration. hb knock-down flies have a significant increase in the penetrance of dye into the eye, indicating a defective blood-eye barrier. Pearson’s Chi-squared test was performed to determine significance between the wild type results and the RNAi. ***: p-val < 0.0005.
Fig 8.
Differentially expressed putative Hb target genes.
Green and red shaded circles are up- and down-regulated genes, respectively. Larger circle size indicates higher log2-fold change. (A) 267 genes from the high confidence list of Hb targets are differentially expressed in the eye-antennal discs during the transition from late L2 (72h AEL) to mid L3 (96h AEL) stages. 33 genes are down-regulated and 234 are up-regulated (see S4 Table). (B) 52 genes from the high confidence list of Hb targets are differentially expressed in the eye-antennal discs during the transition from mid L3 (96h AEL) to late L3 (120h AEL) stages. 10 genes are down-regulated and 42 are up-regulated (see S4 Table).
Fig 9.
Expression of Hb target genes in the eye-antennal discs.
Four of the tested target genes show expression in carpet cells. Eye disc (ed), optic stalk (os). Scale bar = 20 μm. (A) CadN-Gal4 drives UAS-GFP expression (green in A”) in one of the two carpet cells (white arrow), as well as other cells in the disc, possibly glia cells. (A’ and A”). Glia cells are marked with an antibody against the pan-glial marker Repo (rabbit α-Repo, red) and Repo-positive carpet cell nuclei are recognized by their large size (see also Fig 5D). (B) mouse α-Cut (red in B”) showed a clear signal in the two carpet cells (white arrows). (B’ and B”). Carpet cell nuclei were recognized by the expression of the subperineurial glia marker moody (moody-Gal4 driving UAS-GFP expression, green). DAPI shows the eye-antennal disc surface. (C) Fas2-Gal4 drives UAS-H2B::RFP (red in C”) expression in the two carpet cells (white arrows). (C’ and C”) Carpet cell nuclei are recognized by their large size with DAPI and their location on the posterior edge of the retinal field between the outgoing axons visualized with Phalloidin staining (green). (D) Sty-Gal4 drives UAS-H2B::RFP (red in D”) expression in the two carpet cells (white arrows), as well as in other cells in the disc. Due to folding of the imaged disc, the right (D-I) and left (D-II) carpet cells where not found in the same focal plane. (D’ and D”) Carpet cell nuclei are recognized by their large size with DAPI and their location on the posterior edge of the retinal field between the outgoing axons visualized with Phalloidin staining (green).