Replacing the PDZ-interacting C-termini of DSCAM and DSCAML1 with epitope tags causes different phenotypic severity in different cell populations
Figures
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig1-v1.tif/full/617,/0/default.jpg)
The C-terminus of DSCAM is not required for protein stability or localization.
(A) In the Dscam∆C allele, the sequence encoding the final ten amino acids was replaced with a Myc tag by homologous recombination. (B) Western blots of protein immunoprecipitated from HEK293T cells co-transfected with MAGI-3 and V5-tagged DSCAM intracellular domain (ICD) or V5-tagged DSCAM-∆C ICD (∆C) demonstrates that the ∆C mutation disrupts the PDZ-binding of DSCAM’s C-terminus. (C) Western blots of DSCAM protein immunoprecipitated from neonatal brains showed no change in the size or amount of DSCAM in Dscam∆C/∆C mutants. The antibody specificity is confirmed by the lack of signal from Dscam-/- brains. (D–F) Immunofluorescent labeling for DSCAM in vertically sectioned retinas from 3-week old mice demonstrated that the protein is found in a normal, punctate localization in the synaptic plexiform layers in Dscam+/+ (D) and Dscam∆C/∆C (E) mice, consistent with earlier reports for wild type DSCAM (de Andrade et al., 2014). No staining above background was found in Dscam-/- retinas (F), demonstrating the specificity of the DSCAM antibody. (G–I) Hematoxylin and eosin staining of adult retinas shows that, compared to controls (G), Dscam-/- retinas (I) are severely expanded and disorganized. Dscam∆C/∆C retinas (H) have modest expansion, but not the extensive disorganization found in the null mutant. Scale bar is 100 µm. See also Figure 1—figure supplement 1 and Figure 1—figure supplement 2.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig1-figsupp1-v1.tif/full/617,/0/default.jpg)
DSCAM’s C-terminus interacts with PDZ domains.
(A,B) Both MAGI-2 and MAGI-3 were found to interact with DSCAM by yeast two-hybrid. The C-terminal 20 amino acids of DSCAM (PDZ) were used as bait and the Gal4 binding domain alone (GBD) was used as a negative control. Successful interaction is revealed by the expression of β-galactosidase on LacZ (A, note colony color) and expression of HIS3 promoting survival on plates without histidine (B, -His). When transfected into HEK293T cells, both DSCAM (C) and DSCAM-∆C (D) protein localized to the cell surface as revealed by live cell staining with an antibody raised against the entire extracellular domain.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig1-figsupp2-v1.tif/full/617,/0/default.jpg)
DSCAM protein localization is grossly unchanged in Dscam∆C/∆C retinas.
(A–F) Confocal images of P9 retinas immunolabeled for DSCAM and melanopsin show normal colocalization between the proteins in control (A–C) Dscam∆C/∆C animals (D–F). The colocalization is quantified in G. N = 4 retinas per genotype. *p<0.05. (H–L) Confocal images of P7 retinas immunolabeled for DSCAM collected under standardized microscope settings from Dscam+/+ (H), Dscam-/+ (I), Dscam∆C/∆C (J), and Dscam-/- (K) retinas show the relative staining intensities in each mouse. (L) Intensities were measured along a 10 µm line adjacent and perpendicular to the INL, a region that includes S1. Fluorescence intensity in Dscam∆C/∆C retinas was not reduced. Box plots represent the median, first and third quartile, range, and outliers. N = 6 retinas each for Dscam+/+, Dscam-/+, and Dscam∆C/∆C genotypes and 2 retinas for Dscam-/-. *p<0.05.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig2-v1.tif/full/617,/0/default.jpg)
DSCAM-mediated self-avoidance requires C-terminal interactions in only some amacrine cell types.
(A–C) Dopaminergic amacrine cells (stained for tyrosine hydroxylase, TH) are non-randomly spaced in wild type retinas (A), but lose mosaic spacing and form neurite fascicles in two-week old Dscam∆C/∆C (B) and Dscam-/- (C) retinas. (D) In both mutants, there was a significant increase in cell density. Spacing was quantified by DRP analysis; relative cell densities normalized to the overall density at increasing distances from reference cells are plotted in (E). By Voronoi (F) and nearest neighbor (G) analyses spacing in Dscam∆C/∆C retinas was not significantly different than in Dscam-/- animals. (H–J) Conversely, bNOS-positive amacrine cells were not visibly different between controls (H) and Dscam∆C/∆C retinas (I) despite clear fasciculation and loss of mosaic spacing in Dscam-/- mice (J). Dscam∆C/∆C values were intermediate between control and Dscam∆C/∆C in cell density (K), DRP (L), Voronoi (M), and nearest neighbor (N) analyses, but differences from control were not statistically significant. Means ± s.e.m. are represented in D–E, K–L. Box plots represent the median, first and third quartile, range, and outliers. N = 4–8 retinas per genotype. *p<0.05; **p<0.01; ***p<0.001; n.s. is not significant by Tukey post-hoc test between indicated genotypes or compared to controls. Scale bar is 100 µm. Representative Voronoi domains are in Figure 2—figure supplement 1.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig2-figsupp1-v1.tif/full/617,/0/default.jpg)
Examples of Voronoi tessellation domains in Dscam mutants.
Representative Voronoi tessellation domains of TH-positive DA cells (A–C) and bNOS-positive amacrine cells (D–F) show the differential effect of the C-terminal deletion.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig3-v1.tif/full/617,/0/default.jpg)
RGCs also display differential dependence on DSCAM C-terminal interactions for self-avoidance.
(A–C) Melanopsin-positive intrinsically photoresponsive retinal ganglion cells are found in a mosaic pattern in wild type retinas (A), but at two weeks of age, ipRGC cell bodies in Dscam∆C/∆C (B) retinas are pulled into clusters similar to those in Dscam-/- (C) retinas. (D) Overall ipRGC density was not significantly increased in Dscam∆C/∆C retinas. (E) By DRP, cell body clustering was intermediate between Dscam+/+ and Dscam-/- retinas. Voronoi (F) and nearest neighbor (G) analyses also revealed a clear intermediate defect. Values from Dscam∆C/∆C retinas were significantly different from both control and Dscam-/- mutants. (H–J) Cdh3-GFP RGCs are mosaically spaced in control retinas (H), and this spacing is not perturbed in the Dscam∆C/∆C retinas (I), but these cells form clusters in Dscam-/- animals (J), indicating interactions mediated by DSCAM’s C-terminus are dispensable to prevent these cells from clustering. (K) Cdh3-GFP RGC overall cell density was significantly increased in Dscam-/- retinas, but not in Dscam∆C/∆C mutants. (L) A clear exclusion zone is detectable by DRP analysis in Dscam+/+ and Dscam∆C/∆C retinas. This exclusion zone is lost in Dscam-/- animals, where cell density is increased adjacent to any given cell, indicative of clustering. Similarly, Voronoi (M) and nearest neighbor (N) analyses describe a clear spacing defect in Dscam-/- but not in Dscam∆C/∆C retinas. Means ± s.e.m. are represented in D–E, K–L. Box plots represent the median, first and third quartile, range, and outliers. N = 6 retinas per genotype. *p<0.05; **p<0.01; ***p<0.001; n.s. is not significant by Tukey post-hoc test between indicated genotypes or compared to controls. Scale bar is 250 µm. Representative Voronoi domains are in Figure 3—figure supplement 1.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig3-figsupp1-v1.tif/full/617,/0/default.jpg)
Examples of Voronoi tessellation domains in Dscam mutants.
Representative Voronoi tessellation domains of melanopsin-positive ipRGCs (A–C) and Cdh3-GFP RGCs (D–F) show the differential effect of the C-terminal deletion.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig4-v1.tif/full/617,/0/default.jpg)
DSCAML1-mediated self-avoidance requires C-terminal interactions in only some cell types.
(A) Dscaml1∆C/∆Cmutant mice were generated by replacing the sequence encoding the final ten amino acids with an HA tag by homologous recombination. See also Figure 4—figure supplement 1. (B–D) Hematoxylin and eosin staining of adult retinas shows that, compared to controls (B), Dscaml1-/- retinas (D) are significantly expanded and disorganized. Dscaml1∆C/∆C retinas (C) have a more intermediate expansion without the extensive disorganization found in the null mutant. (E–G) AII amacrine cells (Dab1-positive) are organized in a mosaic pattern in two-week old control retinas (E). This pattern is disrupted in Dscaml1∆C/∆C retinas (F), but not as severely as in Dscaml1-/- retinas (G), where the cells form clusters. (H) There was a significant increase in cell density in Dscaml1-/- but not in Dscaml1∆C/∆C animals. I) DRP analysis revealed an intermediate effect in Dscaml1∆C/∆C retinas between the clear exclusion zone in control and clustering in Dscaml1-/-. AII amacrine spacing was slightly disrupted in Dscaml1∆C/∆C retinas by Voronoi analysis (J) but not by nearest neighbor analysis (K). L-N) Conversely, compared to controls (L) the disruption of VGLUT3-positive amacrine cell spacing in Dscaml1∆C/∆C (M) retinas was more similar to that in Dscaml1-/- (N) retinas. (O) VGLUT3-positive amacrine cell density was significantly increased in both Dscaml1∆C/∆C and Dscaml1-/- animals. (P) DRP analysis reveals the loss of exclusion zone in both mutants. (Q,R) Dscaml1∆C/∆C values were significantly different from controls in both Voronoi and nearest neighbor analyses. Means ± s.e.m. are represented in H–I, O–P. Box plots represent the median, first and third quartile, range, and outliers. N = 6–8 retinas per genotype. *p<0.05; ***p<0.001; n.s. is not significant by Tukey post-hoc test between indicated genotypes or compared to controls. Scale bars are 100 µm. Representative Voronoi domains are in Figure 4—figure supplement 2.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig4-figsupp1-v1.tif/full/617,/0/default.jpg)
DSCAML1-∆C has a normal membrane topology, but does not interact with MAGI-3.
Live staining of HEK293T cells transfected with full-length DSCAML1 with an N-terminal HA tag (A) and DSCAML1-∆C with a C-terminal HA tag (B) show that, for both proteins as expected, the N-terminus is presented to the extracellular space and the C-terminus to the cytoplasm. (C) Western blots of protein immunoprecipitated from HEK293T cells co-transfected with MAGI-3 and V5-tagged DSCAML1 intracellular domain (ICD) or V5-tagged DSCAML1-∆C ICD (∆C) demonstrates that the ∆C mutation disrupts the PDZ-binding of DSCAML1’s C-terminus.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig4-figsupp2-v1.tif/full/617,/0/default.jpg)
Examples of Voronoi tessellation domains in Dscaml1 mutants.
Representative Voronoi tessellation domains of AII amacrine cells (A–C, Dab1) and VGLUT3-positive amacrine cells (D–F) show the differential effect of the C-terminal deletion.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig5-v1.tif/full/617,/0/default.jpg)
Increased cell density is not sufficient to explain spacing defects.
(A) ipRGC cell density was similar in Dscam∆C/∆C and Bax-/- retinas. Despite this, clustering was more severe in Dscam∆C/∆C as measured by DRP (B) and Voronoi (C), but not by nearest neighbor analysis (D). (E) Likewise, DA cell density was similar in Dscam∆C/∆C and Bax-/- retinas. However, DA cell spacing was not significantly different between Dscam∆C/∆C and Bax-/- mutants (F–H). (I) VGLUT3-positive amacrine cell density was significantly higher in Bax-/- than in Dscaml1∆C/∆C retinas. Mosaic spacing was more disrupted in Dscaml1∆C/∆C as measured by Voronoi domain analysis (K) and nearest neighbor (L) but not by DRP (J). Means ± s.e.m. are represented in A–B, E–F, I–J. Box plots represent the median, first and third quartile, range, and outliers. *p<0.05; **p<0.01; ***p<0.001; n.s. is not significant by student’s t-test.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig6-v1.tif/full/617,/0/default.jpg)
Neurite fasciculation is separable from density-dependent cell body clustering.
(A) DA cell neurites evenly fill their receptive fields in control mice, but form fascicles in Dscam∆C/∆C (B) and Dscam-/- (C) animals. (D) DA fascicles are rarely observed in Bax-/- mutants. (E) M2 ipRGC dendrites imaged in the ON region of the IPL are evenly distributed in wild type mice and largely remain so in Dscam∆C/∆C (F) and Bax-/- (H) mutants, while severe fasciculation is evident in Dscam-/- retinas (G). I) In the OFF strata of the IPL, M1 ipRGC dendrites are diffusely organized. There is modest fasciculation in Dscam∆C/∆C (J) and Bax-/- (L) mice, while fasciculation in Dscam-/- retinas (K) is much more severe. (M–O) Elo ranking of fasciculation severity between genotypes demonstrates that DA neurites (M) are clearly fasciculated in Dscam∆C/∆C and Dscam-/- retinas, but not in Bax-/-, which had loss of mosaic cell body spacing. Conversely, ipRGCs in Dscam∆C/∆C did not have significantly more fasciculation than Bax-/- either in the ON (N) or OFF (O) layers, despite having a more severe cell body clustering. Box plots represent the median, first and third quartile, range, and outliers. *p<0.05; **p<0.01; ***p<0.001; n.s. is not significant by Wilcoxon rank sum test. Scale bar is 250 µm. See also Figure 6—figure supplement 1.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig6-figsupp1-v1.tif/full/617,/0/default.jpg)
Fasciculation of Cdh3-GFP RGCs and bNOS-positive amacrine cells.
Fasciculation of Cdh3-GFP RGC dendrites (A–C) and bNOS-positive neurites (D–G) were compared between genotypes at two weeks of age. (H) Elo ranking of fasciculation severity demonstrates that Cdh3-GFP RGC dendrites are clearly fasciculated in Dscam-/- retinas, but not in Dscam∆C/∆C or control retinas. (I) Similarly, bNOS-positive neurites were fasciculated in Dscam-/- retinas, but not in Dscam∆C/∆C or controls, and there was only mild fasciculation in Bax-/- animals. Box plots represent the median, first and third quartile, range, and outliers. *p<0.05; **p<0.01; n.s. is not significant by Wilcoxon rank sum test. Scale bar is 250 µm.
![](https://iiif.elifesciences.org/lax/16144%2Felife-16144-fig7-v1.tif/full/617,/0/default.jpg)
Laminar specificity in ∆C mutants.
Neurite stratification in the IPL was analyzed in immunolabeled cryosections. (A,B) bNOS-positive amacrine cells stratified normally in Dscam∆C/∆C retinas, as did ipRGCs and DA cells (C,D), which co-stratify in the OFF region adjacent to the INL. (E) Imaged en face, DA neurites co-fasciculated with ipRGC dendrites in Dscam∆C/∆C mutants (arrowheads), as we have previously found in Dscam-/- retinas (Fuerst et al., 2009). In Dscaml1∆C/∆C mutants, AII amacrine cells (F,G) and rod bipolar cells (H,I) terminate their processes normally. (J–M) TEM analysis revealed that Dscaml1∆C/∆C retinas contained structurally normal dyad synapses between rod bipolar cells and AII/A17 amacrine cells with distinct ribbons (arrows). (K) Dscaml1-/- RBC dyad synapses are characterized by excess in synaptic vesicle number and indistinct ribbons. Four retinas analyzed by TEM per genotype, > 10 synapses inspected per retina. (N,O) VGLUT3-positive amacrine cells misprojected beyond the ON ChAT layer in Dscaml1∆C/∆C mutants. (P,Q) These ectopic neurites became associated with AII amacrine terminals adjacent to the retinal ganglion cell layer (arrowheads). This association was observable at 3 weeks of age (P) and persisted through adulthood (Q, 18 months of age). (R) These misprojections were quantified by imaging whole-mount retinas stained for VGLUT3 and Dab1 en face and calculating the percent of area occupied by VGLUT3 in projections through Dab1-positive AII amacrine terminals. Means ± s.e.m. at three threshold levels are represented in R. n = 6–8 retinas per genotype. Scale bar is 20 µm in A–O, 110 µm in E, 10 µm in P,Q, and 500 nm in J–M.
Tables
Phenotypes by cell type in ∆C mutants. The phenotypes assessed in Dscam∆C/∆C and Dscaml1∆C/∆C are summarized. n.a. is not applicable, n.s. is not shown.
Cell type | Cell spacing | Cell density | Fasciculation | Neurite stratification |
---|---|---|---|---|
DA | nearly null | nearly null | nearly null | normal |
bNOS | nearly wild type | nearly wild type | nearly wild type | normal |
Cdh3-GFP | nearly wild type | nearly wild type | nearly wild type | normal (n.s.) |
ipRGC | Intermediate | nearly wild type | Intermediate | normal |
VGLUT3 | nearly null | nearly null | n.a. | misprojection |
AII | Intermediate | nearly wild type | n.a. | normal |
Additional files
-
Supplementary file 1
p-values from all pairwise comparisons.
p-values were calculated from each pairwise comparison across genotypes using the indicated measurement and test. NN = nearest neighbor analysis; PF = packing factor derived from DRP (density recovery profiling).
- https://doi.org/10.7554/eLife.16144.018