Fig 1.
Drosophila Canoe and human Afadin shared conserved folded domains and a long C-terminal intrinsically disordered region.
A. Diagram to scale of Drosophila Canoe and Human Afadin. B. Output of D2P2, the database of disordered protein predictions. Predicted folded domains are numbered boxes along the line. Above the line are disorder predictions from different programs, and below the line are regions of Predicted Disorder Agreement and sequences defined by the software as molecular recognition features (MoRFs).
Fig 2.
Amino acid conservation of the folded domains and IDRs of Cno and Afadin.
For each folded domain and for the IDR and FAB we calculated amino acid identity. Comparisons were made relative to Drosophila melanogaster Cno (relative to other insects—all except Tribolium and Heliconius are in the Order Diptera) or human Afadin (relative to other vertebrates and one non-vertebrate chordate, Amphioxus). Color coding indicates ranges of sequence identity. Green = 85–100% identical. Blue = 70–84% identical. Yellow = 55–69% identical. Red = <55% identical. Small insertion/deletion polymorphisms (indels) are noted. “No significant similarity” means the program Blast found no match.
Fig 3.
Clustal sequence alignments of the N-terminal folded domains of Cno and its orthologs from selected Dipteran insects reveals strong conservation.
Asterisks indicate identical amino acids, colons and periods conservation between groups of strongly similar or weakly similar properties, respectively, and dashes gaps in the alignment. Predicted folded domains are indicated and highlighted in different colors.
Fig 4.
Clustal sequence alignments of the N-terminal folded domains of human Afadin and its orthologs from selected vertebrates reveals strong conservation.
Asterisks indicate identical amino acids, colons and periods conservation between groups of strongly similar or weakly similar properties, respectively, and dashes gaps in the alignment. Predicted folded domains are indicated and highlighted in different colors.
Fig 5.
Conserved motifs in the Afadin IDR that are predicted to be alpha-helical by AlphaFold correspond with the mapped binding sites of alpha-catenin and actin.
Clustal sequence alignments of the IDRs of vertebrate Afadins. Motifs of 14 amino acids or more that were from 48–88% identical in sequence are highlighted in yellow, or, if the motif was predicted to be alpha-helical by AlphaFold, highlighted in green. Degree of sequence identity in each motif is indicated below the motif. The predicted binding site of alpha-catenin is indicated by blue overlining and overlaps the most N-terminal predicted alpha-helix. The F-actin-binding region identified by Carminati et al. is indicated by red overlining and corresponds to the next three predicted alpha-helices. The fragment found by Mandai et al to bind F-actin is indicated by green overlining, the end of the s-Afadin isoform, which does not bind F-actin, is indicated by black overlining, and the “FAB” as referenced in Sakakibara et al 2020 is indicated by cyan overlining.
Fig 6.
Conserved motifs in the Cno IDR include several regions that are predicted to be alpha-helical by AlphaFold.
Within the IDR motifs of 14 amino acids or more that were from 48–88% identical in sequence are highlighted in yellow, or, if the motif was predicted to be alpha-helical by AlphaFold, highlighted in green. Degree of sequence identity in each motif is indicated below the motif. The region of Cno found to bind F-actin by Sawyer et al., 2009 is overlined in green. The region of sequence similarity to human Afadin is overlined in red.
Fig 7.
AlphaFold predicts alpha-helical regions in the IDRs of Cno and Afadin.
Images are from the predicted structures of Drosophila Cno and Human Afadin from the AlphaFold Protein Structure Database developed by DeepMind and EMBL-EBI. A-C. Predicted helices in Cno’s IDR. Amino acids at the ends of each helix are indicated. D-G.Predicted helices in human Afadin’s IDR. H. Blast alignment showing limited sequence identity in some of the predicted alpha-helices in Drosophila Cno and human Afadin. I. 3D model of the Afadin/α-catenin complex indicates a conserved dimer interface from human to zebrafish. AlphaFold-Multimer [56]in ColabFold [57] was used to predict heterodimeric structures of the α-catenin-binding region of rat l-Afadin (Uniprot accession: O35889-1, residues 1400–1460) and the M3 domain of mouse alpha-E-catenin (Uniprot accession: P26231, residues 507–631). Models with high confidence scores (> 0.87) consistently presented a dimer interface between a long α-helix formed by conserved Afadin residues 1403–1455 (blue) and α-catenin M3 (yellow)–the top-ranked model is shown.
Fig 8.
The zygotic mutant cuticle phenotype of the null allele cnoR2 is surprisingly less severe than the zygotic phenotype of the “classic” allele cno2.
A-D. Representative cuticles. In this and subsequent Figures, all cuticles are oriented anterior up. A. Wildtype. Head involution is complete, producing a wildtype head skeleton (arrow). Dorsal closure is complete and the epidermis is intact. B. Maternal-zygotic phenotype of the null allele cnoR2. Dorsal closure and head involution fail and in the stronger examples like this ventral epidermis is lost. C. Zygotic phenotype of the null allele cnoR2. Dorsal closure is completed, and the epidermis is intact. Head involution is disrupted, leading to a fragmented head skeleton (arrow). D. Zygotic phenotype of many embryos of the cno2 allele. Both head involution and dorsal closure (arrow) fail, giving the classic “canoe” phenotype. E-I. Immunofluorescence images of Stage 13/14 embryos. In this and all subsequent Figures, unless noted, embryos are oriented with anterior to the left and dorsal side up. E-E’. Use of a Balancer chromosome in which GFP is expressed in the mesoderm under control of the twist promotor allows us to identify cnoR2 homozygotes by lack of GFP expression. F-I. Representative cnoR2 homozygotes and wildtype or heterozygous siblings, stained to visualize Arm or Cno. Cno accumulates at lower but still detectable levels at cell-cell junctions in cnoR2 homozygotes. J. Quantification of Cno levels.
Fig 9.
Our set of cno alleles include mutations leading to predicted early stop codons and potential truncated proteins across the span of the Cno protein.
A. Examples of Sanger sequencing confirming the mutations found in whole genome sequencing, as indicated by double peaks in the chromatogram. B. Diagram illustrating the positions and nature of the 18 mutants with a clear lesion in the coding sequence.
Fig 10.
The cuticle phenotypes of zygotic mutants fall into two broad classes, with the null allele cnoR2 in the less severe category, and genetic background is not the sole cause.
A-F. Representative cuticles of the six categories used as criteria in scoring, and their numerical equivalents. Anterior up. G, H. Cuticles of the null allele cnoR2 are on average quite a bit less severe than those of the “classic” allele cno2. I,J. Genetic background is not the sole cause of this difference, as cnoR3, which has a strong phenotype, is on the same background as cnoR2, while cno3, which has a weaker phenotype, is on a third genetic background. K. Comparisons of the phenotypes of the four cno alleles displayed as a 100% cumulative bar chart.
Table 1.
Lesions and phenotypes of cno alleles under study.
Fig 11.
Four of the five alleles with the strong phenotype result from stop codons near the beginning of the IDR.
A. Positions of the alleles with identified lesions and their “numerical cuticle score”. The five strong alleles are highlighted in red. B. Distribution of the cuticle phenotypes of the alleles not presented in Fig 10. C. Comparisons of the phenotypes of these cno alleles displayed as a 100% cumulative bar chart.
Fig 12.
The strong class of cno alleles have more severe effects on morphogenesis than the presumptive null alleles.
Embryos, anterior left, dorsal side up. Stages 13 (A, C-E) or 14/15 (B, F-J). A,B. GFP expression in the mesoderm from the Balancer chromosome allows us to distinguish Balancer or heterozygous wildtype siblings (A, B bottom left) from homozygous zygotic mutants (B). Zygotic mutants also have strongly reduced Cno staining at AJs. C-E. At stage 13 mutant phenotypes are relatively mild, with more persistent segmental grooves (red arrows) and somewhat more variable leading edge cell shapes. F. Wildtype sibling showing normal head involution (red arrow) and intact amnioserosa (green arrow). G, H. cnoR10 zygotic mutant exhibit fully penetrant defects in head involution (red arrows). I, J. cnoR17 zygotic mutant exhibit fully penetrant defects in head involution (J, red arrow), but also have occasional holes in the ventral cuticle (I, cyan arrow), as well as catastrophic defects in dorsal closure, leading to the gut protruding (I, green arrow).
Fig 13.
All the truncation alleles have a very strong maternal/zygotic loss-of-function.
A-G. Representative cuticles of the genotypes indicated, anterior end up. A. Wildtype cuticles have a well-developed head skeleton (arrow). B. Zygotic cnoR10 mutants have strong defects in head involution, disrupting the head skeleton (arrow), but are otherwise normal. C. Maternal/zygotic cnoR10 mutants exhibit complete failure of head involution and dorsal closure, and many have holes in the ventral cuticle. D-F. Maternal/zygotic cnoR5, cnoR3, and cnoR24 mutants have similar cuticle phenotypes. G. In contrast, while maternal/zygotic cnoR8 mutants die as embryos, their cuticles are wildtype or have mild defects in the head skeleton (arrow). H. Diagram showing location of the early stop codons or missense mutation in the alleles examined here and their average cuticle scores. I-N. Distribution of the cuticle phenotypes of the alleles presented. O. Comparisons of the phenotypes of these cno alleles displayed as a 100% cumulative bar chart.
Fig 14.
The three truncation alleles analyzed all disrupt dorsal closure, ventral epidermal integrity and segmental groove retraction.
Embryos, stages 14–15. Anterior to the left. A. Side-by-side maternal/zygotic cnoR3 mutant and zygotically-rescued sibling. They are easily distinguished by the restoration of junctional Cno staining. B,C. Closeups of the embryos in A. In maternal/zygotic mutants the junctional Cno signal is lost. D-J. Representative stage 14 or stage 15 embryos from the indicated genotypes. For all four genotypes, maternal/zygotic mutants exhibit dorsal closure failure (magenta arrows), defects in ventral epidermal integrity (cyan arrows) and persistent deep segmental grooves (yellow arrows).
Fig 15.
Loss of Sdk enhances the zygotic phenotype of the null allele cnoR2.
A,B. Cuticle phenotypes of the null allele cnoR2 versus those of sdk; cnoR2mutants. C. Comparisons of the phenotypes of these genotypes displayed as a 100% cumulative bar chart. D-J. Stage 14 embryos stained for Arm, lateral view unless noted. D, F. Wildtype. Dorsal closure and head involution are proceeding. Segmental grooves are no longer deep (D, arrows), and have receded from the leading edge (F, arrows), where cell shapes are relatively uniform. E,G. cnoR2 zygotic mutant. Segmental grooves remain deep (E, arrows) and leading-edge cell shapes are less uniform (G, arrows). H,I. sdk; cnoR2mutant. Segmental grooves are very deep (white arrows). Dorsal closure has failed, exposing internal tissues (H, red arrows), and cell shapes at the leading edge are less elongated along the dorsal-ventral axis (I, red arrows). J. Ventral view of a sdk; cnoR2mutant revealing disruption of the ventral epidermis. K. Stage 11 sdk; cnoR2mutant with disruptions of the ventral epidermis.
Table 2.
Tests of heterozygous viability.
Table 3.
Antibodies and probes used in this study.