Phenotypic analysis of Myo10 knockout (Myo10tm2/tm2) mice lacking full-length (motorized) but not brain-specific headless myosin X

We investigated the physiological functions of Myo10 (myosin X) using Myo10 reporter knockout (Myo10tm2) mice. Full-length (motorized) Myo10 protein was deleted, but the brain-specific headless (Hdl) isoform (Hdl-Myo10) was still expressed in homozygous mutants. In vitro, we confirmed that Hdl-Myo10 does not induce filopodia, but it strongly localized to the plasma membrane independent of the MyTH4-FERM domain. Filopodia-inducing Myo10 is implicated in axon guidance and mice lacking the Myo10 cargo protein DCC (deleted in colorectal cancer) have severe commissural defects, whereas MRI (magnetic resonance imaging) of isolated brains revealed intact commissures in Myo10tm2/tm2 mice. However, reminiscent of Waardenburg syndrome, a neural crest disorder, Myo10tm2/tm2 mice exhibited pigmentation defects (white belly spots) and simple syndactyly with high penetrance (>95%), and 24% of mutant embryos developed exencephalus, a neural tube closure defect. Furthermore, Myo10tm2/tm2 mice consistently displayed bilateral persistence of the hyaloid vasculature, revealed by MRI and retinal whole-mount preparations. In principle, impaired tissue clearance could contribute to persistence of hyaloid vasculature and syndactyly. However, Myo10-deficient macrophages exhibited no defects in the phagocytosis of apoptotic or IgG-opsonized cells. RNA sequence analysis showed that Myo10 was the most strongly expressed unconventional myosin in retinal vascular endothelial cells and expression levels increased 4-fold between P6 and P15, when vertical sprouting angiogenesis gives rise to deeper layers. Nevertheless, imaging of isolated adult mutant retinas did not reveal vascularization defects. In summary, Myo10 is important for both prenatal (neural tube closure and digit formation) and postnatal development (hyaloid regression, but not retinal vascularization).

. Myo10 reporter knockout (Myo10 tm2/tm2 ) mice lack full-length (motorized) Myo10, but express the brain-specific headless isoform. (A) Schematic diagram showing the reporter knockout (tm2) targeting strategy. Insertion of the targeting sequence by homologous recombination causes loss of 9594 bp, including exon 19 and part of intron 19. Notably, the headless Myo10 isoform begins at exon 20. The gene trap, polyadenylation (pA) signal, is harbored in the IRES:lacZ cassette (IRES stands for internal ribosome entry site). (B) Southern blot analysis. DNA was fragmented using the restriction enzyme EcoRI or BamHI. The position of the radiolabeled hybridization probe at the 3′-end is indicated. Labeled DNA fragments were detected using X-ray film. (C) Western blot analysis. Lysates of HEK293T cells overexpressing full-length mouse (m) Myo10 (plasmid, pCMV-Tag2B-mMyo10) or headless (Hdl) mouse Myo10 (plasmid, pCMV-Tag2B-Hdl-mMyo10) were used as positive controls (blot on the left). Whole brain lysates obtained from P10 mice were used to screen for expression of full-length and Hdl-Myo10 (blot on the right). (D) Level of Hdl-Myo10 protein expressed in Myo10 tm2/tm2 mouse brain (n = 3) relative to wild-type (WT) brains (n = 3), and mouse Hdl-Myo10 transcripts obtained from the National Center for Biotechnology Information (NCBI), National Institutes of Health (NIH). The accession prefix NM_ denotes confirmed protein-coding transcripts, whereas the prefix XM_ indicates predicted protein

Results
Myo10 reporter knockout mice. The reporter knockout (tm2) targeting strategy for Myo10 is shown in Fig. 1A. Insertion of the targeting cassette causes deletion of exon 19 and part of intron 19, and introduces both a reporter (lacZ gene) of endogenous gene expression and a gene trap (SV40 (simian virus 40) polyadenylation (pA) signal). Notably, the mutant (Myo10 tm2 ) allele cannot be converted to a conditional allele. Instead, Cre recombination would delete the loxP-flanked selection cassette, whereas Flp recombination would produce an exon 19 deletion allele without reading frame shift. Southern blot analysis using the hybridization probe shown in Fig. 1A confirmed correct targeting (Fig. 1B). Western blot analysis confirmed that full-length Myo10 was deleted in mouse postnatal (P10) brain (Fig. 1C). However, the headless (Hdl) isoform of Myo10 (Hdl-Myo10) could be clearly detected in homozygous mutants. Notably, lysates from HEK293T cells overexpressing fulllength mouse Myo10 (mMyo10) or mouse headless Myo10 (Hdl-mMyo10) were used as positive controls for the anti-Myo10 antibody (Fig. 1C). On the one hand, lack of Hdl-Myo10 deletion in mutant mice is not completely surprising since the transcript for Hdl-Myo10 is downstream from exon 19 29 . On the other hand, as alluded to previously 27 , the 5′ untranslated region (5′-UTR) of one of the two confirmed transcripts (NM_001353141.1 and NM_001353142.1) corresponding to Hdl-Myo10 is predicted to be disrupted in the Myo10 tm2 allele (Fig. 1D). The affected transcript (NM_001353142.1) lacks exons 20 and 21, suggesting that it may encode a minor isoform of Hdl-Myo10 30 (Fig. 1D). In any case, Western blot analysis revealed no significant difference in relative mouse brain Hdl-Myo10 levels: blot densities (Fig. 1D). Thus, the Hdl-Myo10 isoform encoded by NM_001353141.1 may be the major Hdl-Myo10 isoform in mouse brain or it may be upregulated in response to loss of the transcript NM_001353142.1 in the Myo10 tm2 allele.
Headless Myo10 localizes to the plasma membrane independent of the MyTH4-FERM domain. The domain structures of the mouse Myo10 (mMyo10) and EGFP-tagged truncation constructs used to explore the subcellular localization of headless Myo10 (Hdl-mMyo10) are shown in Fig. 3A. Cells were fixed, stained with Alexa Fluor 594-conjugated phalloidin (an F-actin probe) and imaged by superresolution structured illumination microscopy. As expected from earlier work 29,30 , transfection of HEK293T cells with full-length EGFP-tagged mouse Myo10 (EGFP-mMyo10) induced filopodia formation, whereas transfection with EGFP-Hdl-mMyo10 failed to induce filopodia (Fig. 3B). However, EGFP-Hdl-mMyo10 impressively localized to the plasma membrane suggesting that the tail PH domains readily recruits the protein to membrane transcripts. Green (vertical) bars are exons (ranging from 1 to 41; indicated above) and red bars are 5′-UTRs (5′ untranslated regions), preceding the coding sequence. Notably, the 5′-UTR of Hdl-Myo10 transcript NM_001353142.1, which lacks exon 20 (labeled blue) and exon 21, is disrupted by insertion of the cassette (L1L2_Bact_P) used to generate the Myo10 tm2 allele.
Scientific REPORTS | (2019) 9:597 | DOI:10.1038/s41598-018-37160-y phosphoinositides, possibly due to loss of head-tail autoinhibition. Consistent with this notion, deletion of the MyTH4-FERM domain had no effect, whereas deletion of the PH domains completely blocked membrane localization (Fig. 3B). In living cells stained with the fluorescent plasma membrane probe CellMask Orange and transfected with various deletion constructs, we confirmed using quantified linear profile plots that EGFP-Hdl-mMyo10 and Myo10 lacking the MyTH4-FERM domain (EGFP-mMyo10-ΔMF) strongly localized to the plasma membrane, whereas Myo10 lacking PH domains (EGFP-Hdl-mMyo10-ΔMF-ΔPH1-3) did not localize to the membrane (Fig. 4A,B). Furthermore, deletion of one of the PH domains (EGFP-Hdl-mMyo10-ΔMF-ΔPH3) reduced membrane localization (Fig. 4A,B). Thus, Hdl-Myo10 is probably strongly recruited to the membrane due to the absence of head-tail inhibition. This mechanism would complement the head-tail interaction model proposed by Umeki et al. 32 in which phospholipid binding to Myo10 disrupts head-tail interactions and promotes dimerization, converting the myosin into a filopodial cargo transporter. Head-tail interactions, otherwise, maintain the inactive folded conformation.   Fig. 5A. We confirmed that DCC localized to the tips of Myo10-induced filopodia in HEK293T cells transfected with both human DCC and EGFP-tagged human Myo10 (Fig. 5B). DCC was labeled with mouse monoclonal antibodies which recognize the extracellular domain of human DCC. Using high-resolution MRI (magnetic resonance imaging), we investigated whether Myo10 tm2/tm2 mice had defects in the commissures of the brain (Fig. 5C). MRI of fixed brains isolated from WT and Myo10 tm2/tm2 mice revealed that full-length Myo10 is not critical for the formation of the corpus callosum, anterior commissure and hippocampal commissure (Fig. 5C), in contrast to Dcc (DCC) and Ntn1 (Netrin-1) knockout mice 34,35 , as    Fig. 5D. In coronal views, for example, the corpus callosum had a mean midline thickness of 0.28 ± 0.02 mm in both WT and Myo10 tm2/tm2 brains (n = 3 for each group).
White belly spots are devoid of melanocytes. Myo10 has been implicated in the transfer of melanosomes from melanocytes into epidermal keratinocytes 22 . If Myo10 was important for melanosome transfer, deletion of full-length Myo10 would be expected to produce a coat color phenotype similar to Myo5a-deficient mice 37 , which have widespread hypopigmentation. Instead, Myo10 tm2/tm2 mice have white belly patches, useful as an indicator of the homozygous genotype ( Fig. 2A), and infrequently a dorsal white patch (Fig. 6A). We confirmed using histological skin sections and antibodies against DCT (dopachrome tautomerase), a melanocyte marker, that the hair follicles in white patches were not populated with melanocytes (Fig. 6B).
Phenotype of syndactyly in Myo10 tm2/tm2 mice. Mice lacking full-length Myo10 exhibited simple (soft tissue) syndactyly with high penetrance (Fig. 7A). Typically, digits 2 and 3 and/or digits 3 and 4 were completely fused (frequency of 31: 7: 1 for fusion of digits 2/3, 3/4 and 2/3/4, respectively; n = 14 mice). The rare combination of syndactyly and pigmentation defects in Myo10 tm2/tm2 mice is reminiscent of variations of Waardenburg syndrome (type 3), a neural crest cell disorder which is usually associated with hearing loss 38,39 . We confirmed using µCT (micro-computed tomography) that the syndactyly phenotype of Myo10 tm2/tm2 mice did not involve osseous fusion ( Fig. 7B; the µCT scans shown in Fig. 7B correspond to the photographed paws shown in Fig. 7A and can be matched by the Roman numerals in the lower left corner of each image). X-gal staining of E10.5 -E13.5 embryos revealed Myo10 expression in the developing limb bud and digit primordia (Fig. 7C). At E12.5 and E13.5, there was clear X-gal staining (Myo10 expression) of autopod condensations (Fig. 7C). Histological sections of the footplate at E14.5, when digit separation was nearly complete, revealed X-gal staining in the developing perichondrium and joints (Fig. 7D).
Persistence of hyaloid vasculature. The hyaloid vasculature extends from the optic disc and spans the vitreous to supply blood to the growing lens. In mouse, this vasculature is extensive in the first few postnatal days, but almost completely regresses in the following 2-3 weeks, with marked regression already apparent at postnatal day 8 (P8) 40 . High-resolution MRI of fixed enucleated eyes revealed that the hyaloid vasculature persists in adult Myo10 tm2/tm2 mice (Fig. 8A). Persistent hyaloid vessels could also be seen in retinal whole-mount preparations labeled with anti-collagen, type IV antibodies (Fig. 8B). Lobov et al. 41 . elegantly deduced that regression of the hyaloid vasculature is initiated by macrophages and requires Wnt7b signaling from macrophages to target vascular endothelial cells, which express the Wnt receptor Fzd4 (frizzled class receptor 4) and its co-receptor Lrp5 (low density lipoprotein receptor-related protein 5). Notably, mice lacking the receptor Fzd4, the co-receptor Lrp5 or the ligands Wnt7b or Norrin exhibit persistence of the hyaloid vasculature [41][42][43][44] , schematically illustrated in Fig. 8C. To test whether overexpression of Myo10 enriches Fzd4 at the tips of filopodia, we cotransfected HEK293T cells with EGFP-tagged mouse Fzd4 (Fzd4-EGFP) and mCherry-tagged bovine Myo10 (mCherry-bMyo10). Fzd4-EGFP localized to the plasma membrane and filopodia, but it was not enriched at filopodial tips (not shown). In contrast, HEK293T cells transfected with TurboGFP-tagged human Wnt7b (Wnt7b-tGFP) did not exhibit plasma membrane labeling (not shown).
In addition to persistence of the hyaloid vasculature, mice lacking the Norrie disease protein (Ndp) Norrin, encoded by Ndp, develop retinal hypovascularization, which is phenocopied in mice lacking Fzd4 or Lrp5 (Fig. 8C). We therefore investigated whether mice lacking full-length Myo10 develop retinal hypovascularization in addition to persistence of the hyaloid vessels ( Fig. 9). RNA sequence analysis (next generation sequencing) of ribosome-bound RNA isolated from postnatal retinal endothelial cells at P6, P10, P15, P21 and P50, respectively, revealed that Myo10 (full-length isoform) is the most abundantly expressed unconventional myosin (Fig. 9A). Between P6 and P15, Myo10 expression increased more than 4-fold ( Fig. 9A), when vertical sprouting angiogenesis drives the formation of a deep vascular plexus (P7-P12), followed by intermediate vascular plexus formation (P12-P15), as schematically illustrated in Fig. 9B. We speculated that Myo10 may be important for the formation of filopodia at tip cells, the cells at the tips of vascular sprouts 45 . However, imaging of whole-mount retinas by fluorescence stereomicroscopy and spinning disk confocal microscopy did not reveal impaired vascularization in adult retinas (Fig. 9C,D). Deeper vascular layers could be clearly detected in retinas from Myo10 tm2/tm2 mice (Fig. 9E,F). Thus, Myo10 tm2/tm2 mice show persistence of the hyaloid vasculature without retinal hypovascularization.

Macrophages from Myo10 tm2/tm2 mice do not have impaired Fcγ receptor-mediated phagocytosis.
We recently reported that resident peritoneal macrophages from Myo10 tm2/tm2 mice generate less nascent filopodia compared to wild-type cells, but have no defects in the phagocytosis of zymosan or immunoglobulin G (IgG)-coated polystyrene beads 46 . This was surprising since Myo10 had been previously implicated in Fcγ receptor-mediated phagocytosis 26 . Since macrophages can ingest unopsonized polystyrene beads, used in our previous study, we re-investigated Fcγ receptor-mediated phagocytosis using an alternative assay in which macrophages are presented with freshly isolated human red blood cells (hRBCs) coated (opsonized) with mouse anti-CD235a antibodies (mIgG; illustrated in Fig. 10A). Notably, we confirmed that mouse peritoneal macrophages do not ingest unopsonized hRBCs. In assays using macrophages isolated from NOTAM mice, which harbor a γ-chain mutation 47 , or Fcer1g-deficient mice, we have confirmed that mIgG-hRBCs are engulfed exclusively via Fcγ receptors (unpublished data). Using time-lapse spinning disk confocal microscopy, we found that  macrophages isolated from Myo10 tm2/tm2 mice had no defects in phagocytic cup formation or the ingestion of mIgG-hRBCs (Fig. 10A,B). Thus, Fcγ receptor-mediated phagocytosis is not impaired in macrophages lacking full-length Myo10. Next, we tested whether Myo10 is important for the engulfment of apoptotic cells. We induced externalization of phosphatidylserine, an "eat me" signal, in human red blood cells by incubation with 5 µM A23187, a Ca 2+ ionophore, for 50 min at 37 °C (Fig. 11A). After a wash step, the red blood cells were resuspended in annexin-binding buffer (Fig. 11A). This protocol consistently led to phosphatidylserine externalization, confirmed by incubating cells with Alexa Fluor 594-conjugated annexin V (Fig. 11A). We did not use N-ethylmaleimide to enhance phosphatidylserine externalization, as described by Closse et al. 48 , since we found that this compound is highly toxic to mouse macrophages. Human red blood cells with externalized phosphatidylserine adhered to macrophages, but only a small subset of particles were engulfed (Fig. 11B-D). This suggests that exposure of the phospholipid phosphatidylserine alone is probably not sufficient to trigger phagocytosis, a controversial issue 49 . Similar to wild-type cells, Myo10 tm2/tm2 macrophages sporadically ingested human red blood cells with phosphatidylserine exposure (Fig. 12A-D). The example in Fig. 12 nicely shows how the phagosomes of two ingested human red blood cells fuse (insets in Fig. 12C), such that the two blood cells become pressed together (Fig. 12D). There were no significant differences, wild-type versus Myo10 tm2/tm2 macrophages, in the number of adherent human red blood cells per macrophage and the rates of phagocytic cup formation, although phagocytic events were infrequent in both groups (Fig. 12E).
Exencephalus. Neurulation, the folding of the neural plate to form a neural tube, involves expansion of the mesoderm and elevation of the neural folds, followed by dorsolateral bending 50 . At E8.5, the folds close at the hindbrain-cervical boundary (closure 1) and zippering spreads rostrally and caudally. A second closure (closure 2) at the forebrain-midbrain boundary initiates a second site of zippering. Failure or disruption of closure 2-mediated zippering leads to exencephalus 50 . We speculate that susceptibility to exencephalus in Myo10 tm2/tm2 embryos may be due to impaired adhesion of opposing neural fold apices since cell-cell adhesion molecules implicated in tube closure, including cadherins, are known cargo proteins for Myo10 12,13,51 . Moreover, numerous filopodia, the hallmark of Myo10, have been shown to extend from the non-neural ectoderm and bridge opposing neural folds 52 ; see also review by Nikolopoulou et al. 53 . Genetic experiments in mice have strongly implicated N-cadherin (neural cadherin) in neural tube closure. N-cadherin null embryos rescued by cardiac N-cadherin expression, as well as neural crest-restricted N-cadherin knockout embryos, exhibit exencephalus due to failed closure of the anterior neuropore 54,55 . In this context, it would be interesting to test whether neural crest-restricted deletion of Myo10 predisposes to, firstly, exencephalus, due to loss of N-cadherin transport, and, secondly, pigmentation defects, due to decreased generation or migration of neural crest-derived melanoblasts. Interestingly, the incidence of exencephalus is higher (68% versus 24%) in Myo10 null (Myo10 tm1d/tm1d ) 27 versus Myo10 tm2/tm2 mice 28 . Thus, headless Myo10, which strongly localizes to the plasma membrane (see Figs 3 and 4), may partially compensate for loss of full-length Myo10 during neurulation. X-gal staining for Myo10 expression in Myo10 tm2 embryos appeared negative at the neural fold (see Fig. 2E). However, the expression of headless-Myo10 is probably not reported in Myo10 tm2 mutants since the lacZ reporter gene is located 5′ upstream of headless Myo10 transcripts (see Fig. 1), whereas both full-length and headless Myo10 are expected to be reported by X-gal staining in Myo10 tm1a embryos 27 .
Pigmentation defects (white belly spots). During neural tube closure, neural crest cells delaminate from the apices of the neural folds and undergo epithelial-to-mesenchymal transition. Neural crest cells are highly migratory and give rise to diverse cell lineages, including melanocytes (pigment-producing cells), under the control of a network of regulatory transcription factors and downstream effector genes 56 . Neural crest-derived melanoblasts, the precursors of melanocytes, migrate dorsolaterally, populate the ectoderm and subsequently colonize hair follicles, where the cells produce the pigment melanin. Myo10 tm2/tm2 mice consistently showed white belly spots, and histology confirmed that the spots are devoid of melanocytes (see Fig. 6). Myo10 is expressed in reconstructions of retinal vessels, obtained by spinning disk confocal microscopy (via a 20x/0.45 objective lens). Anti-collagen, type IV, stained the superficial plexus (red), whereas all vessels were labeled with isolectin B4 (green). Scale bars: 100 µm.
Scientific REPORTS | (2019) 9:597 | DOI:10.1038/s41598-018-37160-y neural crest cells 57 and has been reported to be a signature gene of epidermal neural crest stem cells 58 . However, aside from white belly spots, Myo10 tm2/tm2 mice did not exhibit major neural crest-related disorders, such as craniofacial defects or megacolon (intestinal aganglionosis) 59 . We speculate that Myo10 may be an effector gene for the transcription factors specifying the melanocyte lineage, such as Sox10, Pax3 and Mitf, and may confer motility. In Xenopus laevis, Myo10 is a neural crest signature gene 60 and knockdown of the gene has been reported to decrease neural crest cell migration 24,25 . Furthermore, melanoblast-restricted deletion of the filopodia-inducing Rho GTPase Cdc42 impairs melanoblast motility and mice develop severe belly pigmentation defects, similar to Rac1 conditional knockout mice 61,62 . It would be useful to test whether neural crest-(as alluded to above)  Syndactyly. During embryonic development in the mouse, digit formation involves the initiation and progression of interdigital cell death. BMPs (bone morphogenetic proteins), secreted proteins, are thought to play an important role in initiating interdigital cell death, which involves inhibition of anti-apoptotic fibroblast growth factors and Wnt signaling. Consistent with this scheme, conditional deletion of Bmp2 (bone morphogenetic protein 2) in limb bud mesenchyme in mouse leads to soft tissue syndactyly between digits 3 and 4, with variable penetrance, whereas conditional deletion of both Bmp2 and Bmp4 produces complete syndactyly of fore-and hindlimbs 64   other defects. Similarly, digits 2 and 3 or 3 and 4 are fused in Myo10 tm2/tm2 mice with high penetrance (~95%); in contrast, Tokuo et al. 28 reported 72% penetrance for Myo10 tm2/tm2 mice, and Heimsath et al. 27 reported ~50% for Myo10 tm1d/tm1d mice. A different pattern of syndactyly is seen when the intrinsic pathway of apoptosis, mediated by cytochrome c release and caspase 9 activation, is inhibited in mice. Instead of fusions affecting digit pairs 2 and 3 or 3 and 4, persistent interdigital webbing is observed in mice lacking both (double knockout) Bax (encoded by Bax) and Bak (Bak1) 67 , required for cytochrome c release, or mice lacking the three (triple knockout) Bax/Bak activators Bid (Bid), Bim (Bcl2l11) and Puma (Bbc3) 68 . Thus, Myo10 probably regulates the regression of interdigital mesenchyme, presumably as a downstream effector of Bmp signaling 21 , by initiating interstitial cell death. Impaired phagocytic clearance in Myo10 tm2/tm2 mice is unlikely to explain the syndactyly phenotype, since we found that Myo10 tm2/tm2 macrophages have robust phagocytic cup formation and rapidly ingest large IgG-coated beads 46 or IgG-coated red blood cells, as well as apoptotic cells (see Figs 10 and 12). Moreover, impaired phagocytosis would probably give rise to persistent webbing rather than complete or near complete syndactyly of selected digits. Interestingly, syndactyly is not observed in embryos of PU.1 null mice, which lack macrophages, since mesenchyme cells can assume the function of phagocytes in the interdigital space 69 73 . How Myo10 regulates hyaloid regression is unclear, although we found that full-length Myo10 is expressed in both macrophages 46 , which trigger hyaloid regression 41 , and in retinal vascular endothelial cells. Interestingly, long filopodia and Myo10-mediated transport have been implicated in Wnt signaling 74 , implying that Myo10 may be involved in the cell death-inducing Wnt signaling between retinal microglia (macrophages) and hyaloid vessels 41 . Imaging of the pupillary membrane, a transient structure on the anterior surface of the lens, in postnatal mice has impressively shown that numerous long, thin filopodia extend from vascular endothelial cells and interact with resident macrophages 75 . However, whether Myo10-induced filopodia formation is important for endothelial cell-macrophage communication and programmed capillary regression remains to be clarified. In HEK293T cells, we found that Fzd4-EGFP localized to the plasma membrane, but, in contrast to DCC, it was not enriched at the tips of Myo10-induced filopodia (not shown).
Axon guidance. Myo10 has been implicated in neuritogenesis and axon guidance 23,30,76 , and mice lacking the Myo10 cargo protein DCC fail to form commissures in the brain 34 . However, MRI of isolated, fixed brains revealed that the corpus callosum, anterior commissure and hippocampal commissure were intact in Myo10 tm2/tm2 mice (see Fig. 5C). These observations suggest that full-length Myo10 is redundant for Netrin-1-mediated axon pathfinding. We confirmed that headless Myo10, which is still expressed in Myo10 tm2/tm2 mice, does not induce filopodia, but we found that headless Myo10 robustly localizes to the plasma membrane independent of the MyTH4-FERM domain. This was somewhat surprising since GFP-tagged headless Myo10 has been reported to more diffusely localize in CAD (Cath.a-differentiated) cells 29 , a neuronal cell line. Thus, in addition to dimerization with full-length Myo10 and competition for cargo 29 , headless Myo10 may negatively regulate full-length Myo10 by masking membrane phosphoinositides.

Summary and Conclusions
We found that the expression of full-length Myo10, but not headless Myo10, is abolished in Myo10 tm2/tm2 mice. Homozygous mutant embryos developed exencephalus, a lethal phenotype, with low penetrance, whereas surviving Myo10 tm2/tm2 mice consistently exhibited white belly spots, simple syndactyly and persistence of the hyaloid vasculature without retinal hypovascularization. We confirmed in vitro that DCC localizes with Myo10 at the tips of filopodia, but in vivo we could not detect defects in the brain commissures of Myo10 tm2/tm2 mice, in contrast to Dcc knockout mice. We also showed in vitro that headless Myo10 strongly localizes to the plasma membrane in a PH domain-dependent fashion. The unusual combination of pigmentation defects and syndactyly observed in Myo10 tm2/tm2 mice is reminiscent of variants of Waardenburg syndrome, and we suspect that Myo10 tm2/tm2 mice develop progressive hearing loss. The phenotypes of hyaloid persistence and syndactyly suggest that Myo10 may be important for initiating apoptosis, which is also required for neural tube closure 77 . Intact phagocytosis by Myo10 tm2/ tm2 macrophages in vitro suggests that phagocytic clearance is not impaired in homozygous mutant mice. No experiments were performed on live vertebrates. All methods were carried out in accordance with the relevant guidelines and regulations, and all experimental protocols were approved by the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Germany.

Southern blot analysis. Mouse tail biopsies were lysed overnight at 55 °C in buffer containing 100 mM
Tris-HCl (pH 8.5), 5 mM EDTA, 0.2% sodium dodecyl sulfate, 200 mM NaCl, and 100 μg/ml proteinase K. After phenol/chloroform extraction, DNA samples were precipitated by isopropanol, washed in 80% ethanol, dried and dissolved in 50 μl TE Buffer (10 mM Tris (pH 7.9) and 0.2 mM EDTA). Approximately 5 μg genomic DNA was digested with BamHI or EcoRI restriction endonuclease, fractionated on 0.8% agarose gels, and transferred to GeneScreen nylon membranes (NEN-DuPont, Boston, MA). The membranes were hybridized with a 32 P-labeled 2.4 kb probe containing sequences 5′ to the targeted homology and washed with (final concentrations) 0.5x SSPE (1x SSPE contains 0.18 M NaCl, 10 mM NaH 2 PO 4 , and 1 mM EDTA; pH 7.7) and 0.5% sodium dodecyl sulfate at 65 °C. The hybridization probe was cloned as follows. A DNA PCR product was amplified from mouse genomic DNA using the oligonucleotide pair Myo10HR1d and Myo10HR1r and cloned into a custom vector using BsmBI restriction endonuclease sites, followed by sequencing for verification. The sequence of the Myo10HR1d oligonucleotide was 5′-GCTCTAGACGTCTCTGAGATGAGATGATCAGGTCCTGGTGTTA-3′, and Myo10HR1r was 5′-GTCTCAAGCGTCTCTTGGACATTCTAATATCCTGTATACCCCTCACA-3′. The sequences used for cloning of the PCR product are in italics.
Genotyping. PCR for genotyping the Myo10 reporter knockout mice was performed in two steps. First, touchdown PCR was performed using the following thermocycling protocol: 94 °C for 5 min, then 6 cycles of 94 °C for 30 s, 61 °C (with subtraction of 1 °C per cycle) for 30 s, and 72 °C for 60 s. This was followed by 31 cycles of 94 °C for 30 s, 57.5 °C for 30 s, and 72 °C for 60 s. The final extension was 72 °C for 5 min, followed by a holding temperature of 12 °C. The following primer sequences (5′ → 3′) were used: Myo10_F, ATCTGTTTCCCCTTAAGCGAAAAT; Myo10_R, CTCTGTGGGGCCCAGAGCT; CAS_R1_Term, TCGTGGTATCGTTATGCGCC. The expected band (product) size for the primer pair Myo10_F and CAS_R1_Term was 295 bp (mutant allele), and the expected size for the pair Myo10_F and Myo10_R was 400 bp (wild-type allele). LacZ could also be detected using the primer pair LacZ_2_small_F (ATCACGACGCGCTGTATC) and LacZ_2_small_R (ACATCGGGCAAATAATATCG), which had an expected band size of 108 bp (LacZ positive).
Transfection and imaging. HEK293T   Fixed mouse eyes were measured in 1 mL PCR tubes, using a 0.7 T/m gradient system and a two element cryogenic surface coil (Bruker). The inner structures of the eye were visualized using a three-dimensional (3D) fast low-angle shot sequence in a total scan time of 5:23 hours with the following parameters: isotropic resolution, 23 µm; matrix size, 240 × 256 × 256; field of view, 5.5 × 6.0 × 6.0 mm; TR/TE, 37/4.5 ms; flip angle, 10 degrees; and averages, 8.
Fixed brains were imaged using a 1 T/m gradient system and a quadrature volume coil with an inner diameter of 35 mm (Rapid Biomedical, Rimpar, Germany). A 3D spin echo data set was acquired in 16:31 hours, using a turbo RARE sequence: isotropic resolution, 55 µm; matrix size, 296 × 232 × 232; field of view, 16 × 12.8 × 12.8 mm; TR/TE, 750/40 ms; RARE factor 12; and averages, 18. Histology. Harvested skin samples were placed (dermis side down) on Whatman filter paper, which preserves skin flatness during fixation, and then cut into strips. The strips were fixed in 10% neutral buffered formalin (Sigma-Aldrich), equivalent to about 4% formaldehyde, on ice. The next day, the tissue was dehydrated, embedded in paraffin wax and sectioned with a microtome. The formalin-fixed, paraffin-embedded sections were subsequently deparaffinized using the solvent xylene and graded washes with xylene and ethanol. Next, the sections were rehydrated with graded concentrations of ethanol in water. To optimize staining, antigen retrieval was performed by placing sections in target retrieval solution (Dako, Agilent Technologies, Santa Clara, CA), a modified citrate buffer (pH 6.1), and heated for 20 min in a pressure cooker. After washing and blocking with serum for 1 h, sections were incubated with primary antibody, goat anti-DCT (dopachrome tautomerase; also known as tyrosine-related protein 2) antibodies (sc-10451; Santa Cruz Biotechnology), for 2 h at room temperature. The sections were then incubated with biotinylated secondary antibody for 1 h at room temperature, washed and subsequently incubated with Vectastain ABC reagent (Vector Laboratories) before the addition of peroxidase substrate (alkaline phosphatase substrate kit, Vector Laboratories). Finally, the dark blue stain hematoxylin, which labels nucleic acid and other structures, was applied. Micro-computed tomography (µCT). Microtomography of ex vivo samples was performed using a SkyScan 1176 system (Bruker microCT, Kontich, Belgium). Tissue samples were fixed in 4% PFA in PBS for 24 h, followed by 3 × 30 min wash steps in 70% ethanol at room temperature. Subsequently, samples were stored in 70% ethanol at 4 °C. Images were acquired using the X-ray source voltage set at 40 kV (range, 20-90 kV) and the energy (transmission) was attenuated using a 0.2 mm aluminium filter. The exposure time of the X-ray detector, cooled digital X-ray CCD camera (pixel size, 8.52 µm), was set to 780 ms (1 × frame averaging) and scanning performed at 0.5° rotation steps. Volumetric reconstruction of data sets was processed using SkyScan NRecon software (v1.6.9.8; Bruker microCT). The following reconstruction parameters were used: smoothing = 1 (range, 1-10), ring artifact reduction = 14 (range, [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20], and beam-hardening correction = 36 (range, 1-100).
RNA sequence (RNA-Seq) analysis. Expression profiling of retinal endothelial cells at five different postnatal days (P6, P10, P15, P21, and P50) was performed as recently described 79 . Pdgfb-iCre/Rpl22 HA/HA transgenic mice, which express hemagglutinin (HA) tagged ribosomal protein L22 (Rpl22) under control of Cre recombinase specifically expressed in endothelial cells, were generated by crossing inducible endothelial cell-specific Cre (Pdgfb-iCre) mice with Rpl22 tm1.1Psam RiboTag knock-in mice. Ribosome-bound transcripts were immunoprecipitated from whole retina lysates using anti-HA antibodies coupled to magnetic beads, and gene expression was analyzed by RNA-Seq analysis after performing quality control for endothelial cell-specificity using quantitative RT-PCR.
Whole-mount retinal staining. The retina was dissected and whole-mount immunostaining performed as previously described by Pitulescu et al. 80 . In brief, eyeballs were enucleated and placed in 2.0 ml Eppendorf microcentrifugation tubes containing 4% PFA in PBS, and incubated for 2 h at room temperature in a tube rotator. After fixation, the eyes were washed with PBS, placed in a Petri dish and dissected under a stereomicroscope. Spring scissors with 8 mm blades (15003-08; Fine Science Tools) were used to both make an initial cut and excise the cornea. Next, two Dumont #5 forceps were used to remove the sclera and underlying choroid. Susequently, the lens was removed and four radial incisions were made to divide the retina into quandrants. The dissected retinas were placed in 2.0 ml tubes containing blocking and permeabilization buffer (1% bovine serum albumin and 0.3% Triton X-100 in PBS) and incubated at 4 °C overnight in a tube rotator. After washing ( Note that GSL I-B 4 isolectin is a marker for mouse endothelial cells, useful for labeling deeper vascular plexuses, and the extracellular matrix protein collagen IV is a component of the basal lamina of blood vessels. Alexa Fluor 488-conjugated streptavidin (S11223; Thermo Fisher Scientific, Darmstadt, Germany), diluted 1:100, and secondary antibody (Alexa Fluor 546-conjugated donkey anti-rabbit IgG (H + L)), diluted 1:500, were introduced for 1.5 h at room temperature after washing 1 × 15 min with washing buffer (blocking and permeabilization buffer diluted 1:1 with PBS) and 3 × 10 min with PBS. The washing steps were repeated to remove unbound secondary antibodies. Stained retinas were transferred onto standard (25 mm × 75 mm) glass microscope slides via a plastic transfer pipette. For each retinal preparation, excess medium was aspirated and a glass coverslip (24 mm × 32 mm) containing a hanging drop of Fluoromount-G mounting medium (SouthernBiotech, Birmingham, AL) was gently applied. Whole-mount, stained retinas were imaged at low magnification (x2) using a Leica MZ16 F fluorescence stereomicroscope. Higher magnification images were obtained via Nikon Plan Fluor ELWD 20x/0.45 (dry) and Apochromat TIRF 60x/1.49 (oil-immersion) objective lenses of a spinning disk confocal microscope (UltraVIEW Vox 3D live cell imaging system). Quantitative analysis of retinal vasculature was performed using AngioTool 81 .
Live-cell phagocytosis assays. Mouse resident peritoneal macrophages were isolated and seeded into fibronectin-coated Ibidi µ-Slide I chambers as previously described 46 . After overnight incubation of the mouse macrophages, freshly isolated hRBCs (human red blood cells) were incubated with CellMask Orange (C10045; Thermo Fisher Scientific) plasma membrane stain (1:1000 dilution and 5 min incubation at 37 °C) and washed twice with bicarbonate-free RPMI 1640 medium containing 20 mM Hepes. The hRBCs were subsequently opsonized with mouse IgG by incubation for at least 8 min at 37 °C with (mouse) anti-CD235a monoclonal IgG (IgG2b; clone HIR2) antibodies (MA1-20893; Thermo Fisher Scientific), diluted 1:400. The opsonized hRBCs were not washed to avoid agglutination and directly pipetted into a µ-Slide I chamber seeded with macrophages freshly labeled (20 min pre-incubation at 37 °C, followed by wash) with rat anti-mouse F4/80 antibodies conjugated to Alexa Fluor 488 (1:40 dilution; MF48020, Thermo Fisher Scientific).
Phagocytosis of IgG-opsonized (red fluorescent) hRBCs by (green fluorescent) mouse macrophages was imaged by time-lapse spinning disk confocal microscopy. Z-stacks (22 slices at 0.8 µm steps) for each channel (488 nm (green channel) and 561 nm (red channel) laser excitation, respectively) were obtained every 15 s for 16 min. Notably, the bicarbonate-free RPMI 1640 medium (containing 20 mM Hepes) was supplemented with 1 mM MPG (N-(2-mercaptopropionyl)glycine), a free radical scavenger, to reduce phototoxicity. Focal drift was circumvented using the Nikon Perfect Focus Sytem.
Statistics. Normality and homoscedasticity were tested using the Shapiro-Wilk and Levene tests, respectively.
A one-way ANOVA (analysis of variance) was used to test for statistical differences at the 0.05 level of significance. When the assumed conditions of normality and homogeneity of variance were not fulfilled, as in most cases, we used the non-parametric Mann-Whitney U test or Kruskal-Wallis one way analysis of variance on ranks (at the 0.05 level of significance). Statistical analyses were performed using Origin 2016 (OriginLab), and data are presented as box plots or mean ± standard error (s.e.m.).