Fig 1.
Structures, binding partners and PY motifs of Comm, Rcr, PRRG and Shisa proteins.
A. Schematic representations of potential Comm homologues and their interacting partners. Drosophila Comm1 physically associates with Robo receptors [1], but it is not known if this interaction is direct or mediated by additional proteins. Comm1 regulation of fly Robo1 requires the transmembrane domain of Robo1 [62]. An interaction between PRRG4 and Robo receptors is predicted by the work in this study. The PY motifs of Comm1 interact with the WW domain of the Nedd4 E3 ubiquitin-protein ligase [9]. The PY motifs of the human PRRG4 protein also interact with the WW domain of Nedd4 [18]. The interaction with Nedd4 is only implicated in endocytosis of Robo from the cell surface and not in regulation of Robo exocytosis [10]. In yeast the Rcr1 and Rcr2 proteins regulate the cell surface expression of amino acid permease [38], and Rcr1 physically interacts with the Rsp5 ubiquitin ligase via cytoplasmic PY motifs [37]. Shisa proteins physically interact with Fibroblast Growth Factor Receptors (FGFR) and Frizzled Wnt receptors and prevent their trafficking to the cell surface [41]. Shisa proteins are distinguished by Cys-rich N-terminal domains, a feature shared by CYYR1 [40]. B. Amino acid alignments of the proline rich (PY) motifs in the cytoplasmic domains of candidate Comm homologues. These motifs are generally of the form PPxY or LPxY, where P is proline, Y is tyrosine and L is leucine and x is any amino acid. Comm1 has an extended PY motif, GLPSYDEAL, that is critical for Comm1 function. The core LPxY motif together with conserved acidic and hydrophobic residues is shared with the PRRG proteins. In PRRG proteins, a PPxY motif occurs after this extended motif, whereas in invertebrate Comm proteins, additional PY motifs occur before the LPSY sequence. Outside of the motifs, conservation is remarkably low even in other insect species. The species used for the alignment are shown underneath.
Fig 2.
Comparison of the gamma-carboxylation domains of PRRG proteins with Comm family members.
A. Schematic comparing the domain structures of Comm1 and PRRG4. PRRG4 has a signal sequence (SIG) that is lacking in Comm1 (and also PRRG1 and PRRG3). The enzyme γ-glutamyl carboxylase (GGCX) binds to the propeptide (PRO) and proceeds to modify glutamic acid residues in a processive fashion, frequently starting in the adjacent or ωloop or keel domain (ω) before proceeding to the conserved Gla domain (Gla) that contains multiple glutamic acid residues. In our alignment of PRRG4 and Comm1, the ω and Gla domains are separated in Comm1. The transmembrane domains (TM) and PY motifs are also indicated. B. Amino acid alignment of the extracellular/lumenal domains of PRRG and Comm proteins. The critical residues in propeptides are phenylalanine (F), alanine (A) and leucine (L) at positions -16, -10 and -6 relative to the propeptide proteolytic cleavage site (vertical arrow). These residues are conserved in the Drosophila Comm1 protein, and appears partially conserved in other insect Comm proteins. The ω-loop contains glutamic acid (E) residues flanked by conserved phenylalanine and leucine residues, and the latter are conserved throughout Comm proteins. C. Gla domains are characterized by a high frequency of glutamic acid residues, and conserved Cys and phenylalanine (Phe) residues. Comm1 displays weak homology to the Gla domain as a phenylalanine and a pair of glutamic acid residues are in conserved positions. In Comm1, the region that aligns with the Gla domain corresponds to the juxtamembrane lumenal peptide identified by Keleman et al. as critical to Comm1 localization and function [10]. The phenylalanine and a glutamic acid residue are present in some but not all insect Comm proteins. The yeast Rcr proteins lack Gla domains but display some extremely limited homology to Comm1. A conserved Cys is present near the cytoplasmic end of the transmembrane domain in all Comm proteins and in the PRRG3 and PRRG4, both of which can regulate Robo in cell culture (albeit partially in the case of PRRG3). The species used to supply the sequences for the alignment are shown at the bottom.
Fig 3.
Axonal localization of yeast Rcr proteins.
Dissected Drosophila embryonic nerve cords imaged with Nomarksi or differential interference (DIC) microscopy that allows unstained axons to be visualized. The CNS axon scaffold forms a characteristic ladder like pattern and lies on top of the cell bodies that make up the nerve cord. For mis-expression experiments, a single copy of the sca-GAL4 pan-neural driver was combined with a single copy of the indicated UAS transgene (B-D). A. dRobo1protein is primarily localized to the longitudinal axon tracts (arrows) with much less staining in the commissures that cross the CNS midline (arrowheads). The axons of the motor nerve roots are also labeled (asterisks). B. Expression of epitope tagged yeast Rcr1 in CNS axons. There is some staining in the underlying cell bodies, but expression clearly is stronger in the longitudinal axon tracts (arrows) compared to the commissures (arrowheads). The motor nerve roots display strong axonal localization (asterisks). C. Very light axonal staining can be seen for yeast Rcr2 with much stronger expression in the underlying cell bodies (arrows, arrowheads). The nerve roots have faint staining (asterisks). D. PRRG4 protein expression is restricted to cell bodies and is absent from axons. Trace amounts of PRRG4 protein may be present in the axons but the primary reason the axons are visible is due to DIC microscopy. Cell body expression is most evident underneath the commissures (arrowheads), but also lateral to the longitudinal tracts (arrows). The nerve roots lack detectable PRRG4 expression (asterisks).
Fig 4.
PRRG4 induces axon guidance errors when co-expressed with human Robo1 in the Drosophila ventral nerve cord.
Dissected Drosophila embryonic ventral nerve cords stained with the monoclonal antibodies BP102, which stains the CNS axon scaffold (A-E), or anti-dRobo1 13C9 (F-I), or anti-vertebrate Robo1 (J). Both stains are brown. For over-expression experiments, a single copy of the sca-GAL4 pan-neural driver and the indicated UAS transgene is present (B-D, G, I). A. The wild type axon scaffold of a stage 16 embryo exhibits a regular arrangement of commissures that cross the CNS midline (arrowheads) and longitudinal tracts that project along the anterior-posterior body axis (arrow). The anterior and posterior commissures in each segment are separated by the cell bodies of midline glia and motor neurons (asterisk). B. Pan-neuronal expression of PRRG4 results in a failure to fully separate the commissures leading to a fuzzy appearance of commissures (arrowheads). C. Pan-neuronal expression of human Robo1 causes the axon scaffold to partially collapse on the midline in some segments (arrow). In other segments the commissures appear thicker (arrowhead) and in several segments the width of the axon scaffold is reduced even though the commissures remain separated (asterisk). D. Co-expression of PRRG4 and hRobo1 results in strong axon phenotypes, including collapse of axons onto the midline in a manner resembling slit mutants (arrowhead), fuzzy and unseparated commissures with disrupted longitudinals (arrow) or fuzzy and only partially separated commissures (asterisk). E. Stage 16 embryo homozygous for a null mutation in the γ-glutamyl carboxylase gene (GC). No defects in the axon scaffold were observed. F. Stage 13 wild type embryo stained for fly Robo1 protein. The longitudinal portion of the axon scaffold stains brown (arrow), but the commissures lack Robo1 protein (arrowhead). The commissures are just beginning to be separated by the migration of midline glia. G. Pan-neural expression of PRRG4 and hRobo1 results in Robo1 protein entering the commissures (arrowheads), a phenotype seen when comm is over-expressed. H. Stage 14 wild type embryo in which the commissures are separated but lack visible Robo1 staining (arrowhead). The longitudinal tracts have Robo1 staining (arrow). I. An embryo expressing both PRRG4 and hRobo1 displays Robo1 protein in the commissures (arrowheads). The phenotypic effects of expressing both PRRG4 and hRobo1 can be witnessed in the asymmetry of the staining and lateral reduction in the width of the scaffold (asterisk). J. An embryo with pan-neuron expression of UAS-hRobo1 using the scratch-GAL4 driver was stained with anti-Robo1 (Abcam ab7279). Human Robo1 protein is visible in commissural axons (arrowheads), indicating that hRobo1 is not subject to regulation by fly Comm. The data is summarized in Table 1 and the underlying data are shown in S1 Data.
Table 1.
Quantification of axon guidance defects in PRRG4 overexpression experiments.
Fig 5.
Co-localization of Comm, PRRG and Robo proteins.
COS cells were transfected with epitope tagged constructs as indicated under the figure panels. Protein expression was detected by antibody labeling and fluorescence microscopy. Nuclei were detected by DAPI (blue) staining. A. Cells transfected with both Comm and rRobo1. Despite the presence of significant amounts of Comm protein (green), rRobo1 remains localized at the cell surface and throughout the cell (magenta), suggesting that these proteins do not interact in this assay. There is a very limited degree of overlap of Comm and rRobo1 expression within the cell (white; arrow in A*) but also a significant lack of overlap in most other areas. B. Co-expression of both PRRG4 and dRobo1. In this image, two cells are transfected with dRobo1 (magenta), but only one cell expresses high levels of PRRG4 (green). Comparison of the two cells reveals that the pattern of dRobo1 in the cell with a low level of PRRG4 (arrowhead) shows little difference with the cell expressing PRRG4 suggesting the two proteins do not interact. Almost no overlap (white) is seen in the magnified panel (B*). C. PRRG1 and rRobo1 display a slight degree of co-localization in the presumed ER/Golgi when co-expressed (white areas, arrows in C”‘ and C*). In areas not adjacent to the nucleus, strong separation of the two proteins is seen (arrowhead in C*) suggesting they are not interacting. D. Co-expression of PRRG2 and rRobo1 leads to little or no co-localization of the proteins. E. Expression of PRRG3 can lead to a reduction of rRobo1 on the cell surface (E”) and limited co-localization around the nucleus (arrow in E*). Nevertheless, the two proteins do not co-localize in many parts of the cell (arrowhead in E*). These results suggested that PRRG3 may have a limited capacity to interact. F. Co-expression of PRRG4 and rRobo1 results in a strong reduction of cell surface rRobo1 (F”) and co-localization of the two proteins throughout the cell, particularly in the presumed ER/Golgi adjacent to the nucleus (arrow in F*). An adjacent cell expresses a high level of PRRG4 and a low level of rRobo1 (arrowheads in F” and F”‘) suggesting that PRRG4 may be resulting in degradation of rRobo1. These results strongly suggest that PRRG4 and rRobo1 interact in cell culture.
Fig 6.
PRRG4 re-localizes rRobo1 from the plasma membrane.
COS cells were transfected with plasmids encoding the genes indicated in the panels and antibody stained with anti-Robo in magenta. Cells were counterstained with DAPI (blue) to reveal the nucleus. A. Drosophila Robo1 (dRobo1) and B. Co-expression of both dRobo1 and Comm leads to re-localization of dRobo1 from the cell surface to the ER/Golgi. C. Quantification of results for re-localization experiments. Cells transfected with the genes indicated were stained for dRobo1 or rRobo1. Subcellular localization at either the plasma membrane or predominantly in the ER/Golgi were scored by an experimenter blind to the plasmids present. At least 22 healthy cells as judged by nuclear staining were scored for each category. The percentage of cells with ER/Golgi localization is shown in the bar graph. Error bars are the 95% confidence interval to reflect sampling noise. Statistical significance relative to dRobo1 and rRobo1 controls is shown (*** p < 0.01, highly statistically significant) and was calculated using the Fisher exact test with two tails. For the PRRG and rRobo1, the Bonferroni correction was applied. Comparison of dRobo1 with and without Comm has a p value < 0.0001. Comparison of rRobo1 and PRRG4 has p < 0.0001. The PRRG3 and rRobo1 data are trending towards statistical significance, p = 0.0538 (cutoff value is p < 0.0125). D. rRobo1 in the presence of PRRG1 is localized predominantly to the cell surface. E. Co-expression of PRRG2 and rRobo1 results in cell surface localization of rRobo1. F. PRRG3 expression can result in a reduction in the level of rRobo1 on the cell surface. G. The majority of cells co-expressing PRRG4 and rRobo1 display an ER/Golgi localization for rRobo1. The underlying data are shown in S2 Data.
Fig 7.
PRRG4 lowers rRobo1 protein levels in COS cells.
COS cells were transfected with plasmids encoding either rRobo1 or hDscam and increasing amounts of PRRG4. A. Immunoblot analysis of hDscam and rRobo1 proteins levels in the presence of different amounts of PRRG4. Rat Robo1 protein levels fall with increasing amounts of PRRG4 plasmid, whereas hDscam levels remain relatively constant. B. Quantification of independent immunoblot experiments. The density of protein bands for hDscam and rRobo1 was quantified using ImageJ. The positive control was each plasmid transfected in the absence of PRRG4 plasmid and this densitometric value was set at one. All other values for each experiment were expressed as value relative to the control to normalize across different experiments. The values were analyzed in a one-way ANOVA with a Fisher LSD test. Asterisks represent p values for differences between hDscam and rRobo1, * p < 0.05, ** p < 0.01, *** p < 0.001. For 120ng of PRRG4, the difference was not statistically significant (p = 0.05118) but was clearly trending towards significance. The effect for 250ng PRRG4 was very strong and reproducible (p = 0.00002). Error bars represent standard error and are obscured by the 250ng PRRG4 data point for the dashed rRobo1 line. The underlying data are shown in S3 Data.