ABSTRACT
Basement membranes (BMs) are specialized layers of extracellular matrix (ECM) mainly composed of Laminin, type IV Collagen, Perlecan and Nidogen/entactin (NDG). Recent in vivo studies challenged the initially proposed role of NDG as a major ECM linker molecule by revealing dispensability for viability and BM formation. Here, we report the characterization of the single Ndg gene in Drosophila. Embryonic Ndg expression was primarily observed in mesodermal tissues and the chordotonal organs, whereas NDG protein localized to all BMs. Although loss of Laminin strongly affected BM localization of NDG, Ndg-null mutants exhibited no overt changes in the distribution of BM components. Although Drosophila Ndg mutants were viable, loss of NDG led to ultrastructural BM defects that compromised barrier function and stability in vivo. Moreover, loss of NDG impaired larval crawling behavior and reduced responses to vibrational stimuli. Further morphological analysis revealed accompanying defects in the larval peripheral nervous system, especially in the chordotonal organs and the neuromuscular junction (NMJ). Taken together, our analysis suggests that NDG is not essential for BM assembly but mediates BM stability and ECM-dependent neural plasticity during Drosophila development.
INTRODUCTION
Proteome analysis of isolated basement membranes (BMs) has identified over 100 associated proteins, highlighting the complex nature of these specialized extracellular matrix (ECM) sheets (Uechi et al., 2014). However, BMs mainly assemble from a small subset of ECM proteins referred to as the ‘BM toolkit’, which is a highly conserved feature of most metazoan species (Hynes, 2012). BMs consist of two different mesh-like networks formed by self-assembly of either heterotrimeric Laminin molecules or type IV Collagen, which are then linked by Nidogen (NDG), Perlecan and Collagen XV/XVIII homologs (Jayadev and Sherwood, 2017; Yurchenco, 2011). The Drosophila genome harbors a minimal set of nine genes encoding the ‘BM toolkit’ (Hynes, 2012). Four genes encode two Laminin α-, one Laminin β- and one Laminin γ-subunit, which form the only two Laminin heterotrimers in the fly. Two adjacent loci encode for type IV collagens (Col4a1 and viking): one for the ColXV/XVIII homolog Multiplexin and one Nidogen/entactin gene. The terribly reduced optic lobes (trol) locus encodes Perlecan (Borchiellini et al., 1996; Dai et al., 2018; Datta and Kankel, 1992; Friedrich et al., 2000; Henchcliffe et al., 1993; Lindsley and Zimm, 1992; Martin et al., 1999; Meyer and Moussian, 2009; Urbano et al., 2009; Voigt et al., 2002; Wolfstetter and Holz, 2012; Yarnitzky and Volk, 1995; Yasothornsrikul et al., 1997).
In 1983, Timpl et al. (1983) reported the identification of an 80 kDa protein from mouse Engelbreth-Holm-Swarm (EHS) sarcoma cells that was named Nidogen due to its ability to self-aggregate into ‘nest-like structures’. Although this protein was later identified as a proteolytic fragment of Entactin, a 150 kDa sulfated glycoprotein described by Carlin et al. (Carlin et al., 1981), the terms Nidogen or Nidogen/Entactin are commonly used to refer to the non-cleaved protein (Martin and Timpl, 1987). The Nidogen protein consists of three globular domains (G1 to G3), a rod-like segment between G1 and G2, as well as an EGF-like domain-containing segment that connects G2 and G3 (Durkin et al., 1988; Fox et al., 1991). Nidogen forms stable complexes with Laminin through binding of its globular G3 domain to the Laminin γ-subunit, whereas the G2 domain mediates binding to Collagen IV and Perlecan (Fox et al., 1991; Hopf et al., 2001; Mann et al., 1989; Reinhardt et al., 1993). Therefore, it has been suggested that Nidogen supports the formation of ternary complexes within the BM and functions as an ‘ECM linker molecule’ connecting the Laminin and Collagen IV networks (Aumailley et al., 1993; Ho et al., 2008; Mayer et al., 1993, 1995).
Functional analysis of the two Nidogen family members in mammals reveals that the single knockouts of nidogen 1 (Nid1) and nidogen 2 (Nid2) in mice neither result in lethality nor affect BM formation and morphology grossly. However, Nid1−/− animals exhibit neurological phenotypes such as spontaneous seizures and hind limb ataxia (Dong et al., 2002; Murshed et al., 2000; Schymeinsky et al., 2002). Analysis of double mutant Nid1−/− Nid2−/− mice does not reveal an essential function during embryogenesis and embryonic BM formation, but complete absence of Nidogen causes perinatal lethality due to impaired lung and heart development accompanied by defects in the organ-associated BMs. Moreover, a varying degree of syndactyly and occasional twisting of forelimbs is observed (Bader et al., 2005; Böse et al., 2006). In accordance, a non-essential function of Nidogen for invertebrate development is revealed by the analysis of C. elegans nidogen-1 (nid-1) mutants, which are viable and fertile, and do not display abnormalities during BM assembly. In addition to a reduction in fecundity, nid-1 mutant animals exhibit aberrant guidance and positioning of longitudinal nerves, movement defects in a body-bending assay and altered neuromuscular junction (NMJ) organization, which suggests a role in nervous system patterning (Ackley et al., 2003; Hobert and Bülow, 2003; Kang and Kramer, 2000; Kim and Wadsworth, 2000; Kramer, 2005). A recent study by Zhu et al. (2017) reports body length reduction in Danio rerio upon depletion of nid1a, one of the four predicted Nidogen genes in zebrafish.
Here, we report the characterization of the single Nidogen/entactin (Ndg) gene in Drosophila. We found that Ndg was highly expressed in the Drosophila embryo and NDG protein was abundant in all basement membranes (BMs), suggesting an essential function during development. In contrast, analysis of Ndg mutants revealed that loss of NDG did not affect viability and fertility in general, arguing for a negligible role during development. Moreover, examination of BM assembly in Ndg mutant embryos revealed no overt changes in overall BM protein distribution. Interestingly, ultrastructural SEM analysis, and further functional assays revealed a porous BM surface and severe defects in BM barrier function. Ndg mutant larvae displayed a range of behavioral phenotypes, such as reduced responses to external stimuli, impaired motility and climbing performance, as well as altered crawling behavior and gravitaxis. Morphological analysis of the larval nervous system indicated improper positioning of dda neurons, impaired targeting by the transversal nerve as well as defects in chordotonal organ and neuromuscular junction (NMJ) organization. Taken together, our analysis did not reveal an essential role for NDG in BM assembly but suggested that NDG contributes to the robustness of the BM barrier.
RESULTS
Ndg transcript expression and protein localization
Northern blot analysis revealed high levels of Ndg expression in Drosophila embryos after gastrulation (Fig. S1). We therefore applied fluorescence in situ hybridization on white1118 (w1118) embryos to analyze Ndg mRNA distribution in comparison with NDG protein localization revealed by antibody staining (Fig. 1). In line with our Northern blot analysis, Ndg expression was apparent at stage 11/12 and could be detected in single cells of the head, especially in the gnathal segments (Fig. 1A, asterisk) as well as in segmentally located patches of cells in the dorsal mesoderm (arrowhead, Fig. 1A). Moreover, the midline-associated, mesodermal dorsal median cells (DMC, Fig. 1B,C), as well as four surrounding somatic myoblasts (sMB, Fig. 1B), exhibited strong Ndg expression (see also Discussion), whereas we detected weak signals in the amnioserosa (AS, Fig. 1C). NDG protein, embedding DMCs and sMBs, formed a thin sheet between ecto- and mesoderm, and around the prospective anal plate (AP, Fig. 1C). After germ band retraction (Fig. 1D) mRNA expression was visible in the forming dorsal and ventral muscles, the amnioserosa (AS) and the segmentally located chordotonal organs (Cho). This was in agreement with our Northern blot analysis, which revealed a strong Ndg mRNA increase in older embryos (Fig. S1). Internal views indicated a decrease of expression in the DMCs (arrow), whereas Ndg was expressed in the esophageal visceral muscle primordium (eVM, arrowhead) and in the joint region between hind- and midgut (Fig. 1E). At this time, accumulation of NDG protein in the forming BMs around the developing brain and ventral nerve cord (VNC), in the differentiating tracheal system, in the future digestive tract as well as in the forming somatic muscles was observed (Fig. 1D,E). Embryos at stage 16 displayed strong mRNA expression in somatic and visceral muscles, and in the cap cells (CCs, Fig. 1H) of the chordotonal organs (see also Discussion), whereas the protein localized to all embryonic BMs (Fig. 1F-I). Interestingly, BMs surrounding the chordotonal neurons (CNs), the VNC and the brain were highly NDG positive, although these tissues did not express Ndg mRNA (Fig. 1H,I). In summary, Ndg expression during embryogenesis followed a dynamic and distinct pattern involving mostly mesodermal cells and tissues, but also chordotonal cap cells and the amnioserosa, whereas the protein was highly abundant in all embryonic BMs. No transcript expression was observed in embryonic hemocytes or fat body cells, although these tissues strongly express Laminin and type IV Collagen (Le Parco et al., 1986; Pastor-Pareja and Xu, 2011; Rodriguez et al., 1996; Wolfstetter and Holz, 2012).
Ndg transcript expression and protein localization during embryonic development. Confocal stacks of Drosophila white (w1118) embryos at developmental stages 12 (A-C), 14 (D,E) and 16 (F-I). Ndg mRNA (green) was visualized by fluorescence in situ hybridization (FISH), and antibody staining against NDG (magenta) reveals the secreted protein. Embryos are orientated laterally, except for B, G and I (ventral views). AS, amnioserosa; AP, anal plate; CC, chordotonal cap cell; Cho, chordotonal organ; CN, chordotonal neuron; DMC, dorsal median cell; eVM, esophageal visceral mesoderm; sMB, somatic myoblast. Asterisk and arrowhead in A depict Ndg expression in the head region and the dorsal mesoderm, respectively. Scale bars: 50 µm.
Ndg transcript expression and protein localization during embryonic development. Confocal stacks of Drosophila white (w1118) embryos at developmental stages 12 (A-C), 14 (D,E) and 16 (F-I). Ndg mRNA (green) was visualized by fluorescence in situ hybridization (FISH), and antibody staining against NDG (magenta) reveals the secreted protein. Embryos are orientated laterally, except for B, G and I (ventral views). AS, amnioserosa; AP, anal plate; CC, chordotonal cap cell; Cho, chordotonal organ; CN, chordotonal neuron; DMC, dorsal median cell; eVM, esophageal visceral mesoderm; sMB, somatic myoblast. Asterisk and arrowhead in A depict Ndg expression in the head region and the dorsal mesoderm, respectively. Scale bars: 50 µm.
Localization of NDG to BMs depends on Laminin
We wanted to determine whether the absence of other major BM components could affect NDG localization to BMs. Therefore, we performed NDG antibody staining on late stage 16 embryos lacking either Laminin, type IV Collagen or Perlecan (Fig. 2). Control siblings exhibited strong NDG staining of all embryonic BMs and around the chordotonal organs (Fig. 2A). To differentiate between the NDG localization properties of both Laminin trimers, we employed either LanA or wing blister (wb) loss-of-function mutations that impair formation and secretion of only one Laminin trimer (Urbano et al., 2009). Interestingly, we found a punctuated NDG pattern in transheterozygous LanA9-32/Df LanA embryos lacking the Laminin A-yielding trimer (Fig. 2B), whereas we did not detect changes in wbHG10/Df wb embryos in which the Wb-containing trimer is absent (Fig. 2C). The loss of all secreted Laminin trimers in embryos lacking the only Laminin γ-subunit (LanB2knod/Df LanB2) resulted in a severely disrupted NDG pattern (Fig. 2D), indicating a redundant function of both Laminin trimers in this process. On the other hand, embryos deficient for trol (the Perlecan encoding locus) or the two Drosophila type VI Collagen genes (Col4a1 and viking) displayed no changes in NDG localization and distribution (Fig. 2E,F), suggesting that Laminin is the critical component for NDG distribution and localization to BMs.
NDG localization in the absence of core BM proteins. Stage 16 embryos in lateral orientation were stained for NDG protein (white). (A) Control sibling embryo with balancer-associated YFP expression (magenta) displays proper NDG localization to BMs with strong accumulation at the chordotonal organs. (B) Punctate NDG patterning in a LanA9-32/Df LanA transheterozygous embryo. (C) NDG protein distribution appears unaffected in wbHG10/Df wb embryos. (D) Disruption of the NDG pattern in LanB2knod/Df LanB2 embryos. (E) Df trol (loss of Perlecan) embryos display a segmentation phenotype due to the additional loss of the adjacent giant locus but no changes in NDG localization. (F) Regular NDG distribution in an embryo deficient for the two Drosophila type VI Collagen-encoding loci Col4a1 and viking (vkg). Scale bar: 50 µm.
NDG localization in the absence of core BM proteins. Stage 16 embryos in lateral orientation were stained for NDG protein (white). (A) Control sibling embryo with balancer-associated YFP expression (magenta) displays proper NDG localization to BMs with strong accumulation at the chordotonal organs. (B) Punctate NDG patterning in a LanA9-32/Df LanA transheterozygous embryo. (C) NDG protein distribution appears unaffected in wbHG10/Df wb embryos. (D) Disruption of the NDG pattern in LanB2knod/Df LanB2 embryos. (E) Df trol (loss of Perlecan) embryos display a segmentation phenotype due to the additional loss of the adjacent giant locus but no changes in NDG localization. (F) Regular NDG distribution in an embryo deficient for the two Drosophila type VI Collagen-encoding loci Col4a1 and viking (vkg). Scale bar: 50 µm.
Generation of Ndg deletion mutants
In order to investigate the function of Ndg in Drosophila, we employed imprecise excision of the Minos transposon insertion Mi{ET1}MB04184 to generate deletions in the Ndg locus (Fig. S2A). Two excision lines were obtained in which 0.4 kb or 1.4 kb of the Ndg locus, including parts of the 5′ UTR, were deleted (henceforth referred to as NdgΔ0.4 and NdgΔ1.4, respectively; see also Materials and Methods). Further analysis employing NDG-specific antibodies revealed that the strong NDG staining observed in control siblings (Fig. S2B) was completely absent in homozygous NdgΔ1.4 (Fig. S2C) and Ndg-deficient embryos (Df Ndg, Fig. S2E), whereas faint expression was detected in the NdgΔ0.4 mutant (Fig. S2D). These results were further confirmed by detecting Ndg transcripts in stage 16 mutant embryos employing in situ hybridization (Fig. S2F-I). Therefore, we concluded that the deletion in NdgΔ1.4 resulted in a protein null allele, whereas the smaller molecular lesion in NdgΔ0.4 caused a hypomorphic Ndg allele characterized by reduced NDG expression.
Phenotypic analysis of Ndg mutants
Nidogen mutants generated in C. elegans and mouse are viable and display only mild phenotypes (Bader et al., 2005; Böse et al., 2006; Dong et al., 2002; Kang and Kramer, 2000; Murshed et al., 2000; Schymeinsky et al., 2002), a surprising finding that is in strong contrast to the essential developmental roles demonstrated for the other BM molecules (Arikawa-Hirasawa et al., 1999; Clay and Sherwood, 2015; Poschl et al., 2004; Yao, 2017). In agreement with these former observations, Drosophila NdgΔ mutants were viable and fertile, and could be maintained as homozygous stocks. Despite the lack of any obvious developmental function, we noticed several peculiarities in NdgΔ homozygous stocks. Under non-crowded, standard culture conditions, NdgΔ pupal cases were preferentially formed in the lower half of the vial or directly on the food, indicating impaired larval climbing abilities (Fig. S3A, quantified in Fig. S3B). Moreover, NdgΔ mutant pupae were significantly smaller than wild-type controls (Fig. S3B). We also noticed a difference in orientation of the pupal cases that was shifted towards the horizontal axis in NdgΔ vials (Fig. S3B). Moreover, we observed a decrease in fecundity in homozygous NdgΔ animals compared with the w1118 control (Fig. S3C) but we did not notice obvious alterations in overall egg morphology. Additionally, NdgΔ1.4 flies exhibited incompletely inflated wing blades at low (∼5%) penetrance rates (Fig. S3D,E).
Distribution of major BM components is not altered in Ndg mutant embryos
To analyze the proposed function of NDG as an ECM cross-linker that connects Laminin and Collagen layers of BMs (Timpl and Brown, 1996), we studied the distribution of BM components in NdgΔ1.4/Df Ndg embryos. Therefore, we applied antibodies raised against Laminin subunits, type IV Collagen and Perlecan (Fig. 3).
BM formation in the absence of Ndg. Distribution of LanA, LanWb, COLL IV and PCAN in (A-D) control and (E-H) transheterozygous NdgΔ1.4/Df Ndg embryos at stage 16. Embryos appear in dorsolateral (A,B,D,E,F,H) or ventrolateral (C,G) orientation. Scale bar: 50 µm.
The anti-Laminin antibodies employed stained embryonic BMs, indicating reactivity with secreted Laminin (Fig. 3A,B). Laminin A (LanA) outlined the BMs of all internal organs in control and Ndg mutant embryos (Fig. 3A,E), whereas antibody staining against the Laminin α1, 2 subunit Wing blister (LanWb) was found mainly in BMs around gut and apodemes of control and Ndg mutant embryos (Fig. 3B,F). Employing antibodies against either Collagen IV (COLL IV, Fig. 3C,G) or Perlecan (PCAN, Fig. 3D,H), we did not observe obvious differences between control and Ndg mutant embryos. Taken together, the complete loss of NDG did not influence the formation or overall assembly of embryonic BMs.
Loss of NDG affects BM barrier function and ultrastructure
The viability of NdgΔ mutant flies and the presence of major BM proteins suggested that NDG plays no essential role during BM assembly, thereby questioning its proposed role as an important ECM linker molecule. However, the loss of NDG might impair BM stability and function. To address this in more detail, we dissected w1118 and NdgΔ1.4 3rd instar larvae and performed a dextran-based permeability assay (Fig. 4). While Texas Red (TexRed)-coupled dextran was only weakly detected in somatic muscles of larval filets from w1118 control larvae (Fig. 4A), muscles of NdgΔ1.4 animals appeared significantly brighter (Fig. 4B, quantified in Fig. 4C) indicating enhanced diffusion into the muscle proper.
Ndg is required for permeability and mechanical stability of BMs. Confocal sections of somatic muscle preparations of control (A) and NdgΔ1.4 (B) larvae incubated with low-weight Texas Red-coupled dextran (TexRed-Dextran, rainbow RGB2 color LUT) to reveal BM leakage. Confocal images were acquired with identical channel settings. (C) Bee swarm box plot (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) of corrected fluorescence intensity measurements obtained from the permeability assay. Welch's t-test was applied to reveal significance (***P<0.001). (D) Quantification of osmotic stress applied to 3rd instar larval wing discs of the indicated genotypes. Data are mean±s.d. Compared with w1118 and w*; KrIf-1/CyO balancer controls, imaginal discs from NdgΔ animals resist osmotic stress for a significantly reduced mean time (NdgΔ0.4/CyO=472 s versus NdgΔ0.4=213 s; NdgΔ1.4/CyO=520 s versus NdgΔ1.4, 239 s; and Df Ndg/CyO=407 s versus NdgΔ1.4/Df Ndg, 241 s, n≥20 wing imaginal discs). Kruskal-Wallis ANOVA and Dunn's test of multiple comparisons for adjusted P-values was applied (w1118 versus NdgΔ1.4, ***P<0.001; w1118 versus NdgΔ0.4, **P=0.001; w1118 versus NdgΔ1.4/Df Ndg, **P=0.005). Scale bar: 50 µm.
Ndg is required for permeability and mechanical stability of BMs. Confocal sections of somatic muscle preparations of control (A) and NdgΔ1.4 (B) larvae incubated with low-weight Texas Red-coupled dextran (TexRed-Dextran, rainbow RGB2 color LUT) to reveal BM leakage. Confocal images were acquired with identical channel settings. (C) Bee swarm box plot (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) of corrected fluorescence intensity measurements obtained from the permeability assay. Welch's t-test was applied to reveal significance (***P<0.001). (D) Quantification of osmotic stress applied to 3rd instar larval wing discs of the indicated genotypes. Data are mean±s.d. Compared with w1118 and w*; KrIf-1/CyO balancer controls, imaginal discs from NdgΔ animals resist osmotic stress for a significantly reduced mean time (NdgΔ0.4/CyO=472 s versus NdgΔ0.4=213 s; NdgΔ1.4/CyO=520 s versus NdgΔ1.4, 239 s; and Df Ndg/CyO=407 s versus NdgΔ1.4/Df Ndg, 241 s, n≥20 wing imaginal discs). Kruskal-Wallis ANOVA and Dunn's test of multiple comparisons for adjusted P-values was applied (w1118 versus NdgΔ1.4, ***P<0.001; w1118 versus NdgΔ0.4, **P=0.001; w1118 versus NdgΔ1.4/Df Ndg, **P=0.005). Scale bar: 50 µm.
To further investigate whether absence of NDG comprises the mechanical stability of BMs, we exposed wing imaginal discs to osmotic shock by incubating them in deionized water. When water diffused into the hypertonic interior of the imaginal discs it created balloon-like swelling before eventual bursting of the disc. Wing discs of larvae resembling the genetic background of balanced and homozygous NdgΔ animals (w1118 and w*; KrIf−1/CyO) withstand the osmotic pressure for ∼430 s and ∼388 s, respectively. For wing discs derived from balanced NdgΔ siblings, we observed a slight but not significant increase (NdgΔ0.4/CyO=472 s, NdgΔ1.4/CyO=520 s and Df Ndg/CyO=407 s). In contrast to this, NdgΔ0.4 wing discs burst after ∼213 s and NdgΔ1.4 or NdgΔ1.4/Df Ndg discs after ∼240 s, indicating decreased BM stability (Fig. 4D).
Further ultrastructural analysis of w1118 control 3rd instar larvae revealed that BMs covering larval somatic muscles, the longitudinal and circular visceral muscles (LVM), and the larval fat body (Fig. 5A-C) appeared as smooth sheets with no detectable alterations (Fig. 5A′-C′). In contrast to this, NdgΔ1.4 mutant larvae displayed surface defects in the BMs surrounding these tissues. Small holes in the BM were observed in areas where tracheal (Tr) branches attached to the somatic muscles (Fig. 5D,D′) or in the close vicinity of the longitudinal visceral muscles (LVM) on the midgut surface (Fig. 5E,E′). Similar BM defects were observed in NdgΔ0.4 (Fig. S4C) mutant larvae and in transheterozygous NdgΔ1.4/Df Ndg animals (Fig. S4F) with some local variations in individual larvae (Fig. S4G). Interestingly, visceral BMs of heterozygous NdgΔ/CyO animals (Fig. S4B,D,E) already exhibited small surface holes that were not found in w1118 (Fig. 5B′) or in additional control larvae from the w*; KrIf-1/CyO balancer stock (Fig. S4A), indicating a dose-dependent effect of NDG on BM morphology. The strongest BM surface abnormalities were found around the larval fat body, where a continuous BM sheet (Fig. 5C,C′) was absent and the organ surface was instead covered with fibrillary structures (Fig. 5F,F′).
Loss of NDG affects BM ultrastructure. (A-F) Surfaces of somatic muscles, midgut and fat body from dissected w1118 control 3rd instar larvae (A-C) and homozygous Ndg mutants (D-F) revealed by scanning electron microscopy. (A′-F′) Areas highlighted in A-F are shown at higher magnification. LVM, longitudinal visceral muscle; Tr, tracheal branch. Scale bars: 100 µm in B; 50 μm in A,C,D,E,F; 5 µm in A′-F′.
Loss of NDG affects BM ultrastructure. (A-F) Surfaces of somatic muscles, midgut and fat body from dissected w1118 control 3rd instar larvae (A-C) and homozygous Ndg mutants (D-F) revealed by scanning electron microscopy. (A′-F′) Areas highlighted in A-F are shown at higher magnification. LVM, longitudinal visceral muscle; Tr, tracheal branch. Scale bars: 100 µm in B; 50 μm in A,C,D,E,F; 5 µm in A′-F′.
Altered crawling behavior and NMJ morphology in Ndg mutant larvae
When placed on agar plates supplemented with yeast paste as food source, NdgΔ1.4 1st instar larvae were often observed outside the food (Fig. 6A,B). As we observed less directed and uncoordinated movements in these larvae, we wanted to monitor and compare their behavioral repertoire. Therefore, we recorded age-matched w1118 controls and NdgΔ1.4 animals that were placed on Agar plates in the absence of food (Fig. 6C,D). Crawling of control w1118 larvae was dominated by smooth forward movement with occasional turns and pauses that allowed the larvae to explore the experimental arena (Fig. 6C,E and Movie 1). In contrast, Ndg mutants generally moved within a smaller area and displayed a variety of sudden motion defects in their crawling pattern, such as head shaking, and spontaneous rolling and bending (Fig. 6D,F and Movie 1). Interestingly, the phenotypic strength exhibited age-dependent variation as 70% of 2nd instar larvae (n>40) and half of the analyzed 3rd instar larvae (n>40) displayed rather strong behavioral abnormalities, whereas the remaining animals exhibited fewer or weaker crawling defects. We further analyzed crawling of NdgΔ 3rd instar larvae in more detail, investigating the velocity and the stride frequency of the undisturbed crawling pattern (Fig. 6G,H). Comparison of the mean crawling velocity and the stride frequency showed comparable differences between the analyzed genotypes. The mean crawling velocity of NdgΔ0.4 mutant larvae was not significantly reduced, whereas NdgΔ1.4 larvae showed a significantly reduced mean velocity compared with w1118 larvae (Fig. 6G). The stride frequency was significantly lower in both NdgΔ strains when compared with w1118 larvae (Fig. 6H). However, NdgΔ1.4 mutants displayed the more severe phenotype in terms of reduced crawling speed and stride frequency (Fig. 6G,H).
Altered crawling behavior and neuromuscular junction phenotypes of Ndg mutant larvae. (A) Schematic example of the observed distribution of w1118 and NdgΔ1.4 1st instar larvae (black dots) on agar plates (depicted as circles) outside a central food source (red). (B) Quantification of the larval distribution assay. Mean number of larvae outside the food: w1118=1.44, NdgΔ0.4=2.31, NdgΔ0.4/NdgΔ1.4=8, NdgΔ1.4=19.1, NdgΔ1.4/Df Ndg=19.3. Data were analyzed using one-way ANOVA (P<0.001) followed by Tukey's multiple comparison test with adjusted P-value: w1118 versus NdgΔ0.4, P=0.947 (not significant, n.s.); w1118 versus NdgΔ0.4/NdgΔ1.4, *P<0.01; w1118 versus NdgΔ1.4, ***P<0.001; w1118 versus NdgΔ1.4/Df Ndg, ***P<0.001; n=400 larvae for each genotype. Boxes and whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles represent values from the individual measurements. (C,D) Snapshots taken from crawling recordings of w1118 and NdgΔ1.4 2nd instar larvae. (E,F) Representative tracking patterns of crawling w1118 and NdgΔ1.4 2nd instar larvae. Individually colored lines represent recordings of 25 s. Time points are indicated if the larva left the monitored area before the end of recording. The time interval in brackets indicates the absence from the recorded area. (G) Quantification of mean crawling velocity measured as 5 s of uniform crawling from recordings of w1118 and NdgΔ1.4 L3 larvae. Mean values: w1118=1.265, NdgΔ0.4=1.07, NdgΔ1.4=0.907 mm/s. Kruskal-Wallis ANOVA (P<0.0001) followed by Dunn's test of multiple comparisons for adjusted P-values: NdgΔ0.4 versus w1118, P=0.0597; NdgΔ1.4 versus w1118, ***P<0.0001. Data are mean±s.e.m. (H) Quantification of the stride frequency calculated from five consecutive strides from recordings of w1118 and NdgΔ1.4 3rd instar larvae. Mean values: w1118=73.32, NdgΔ0.4=62.25, NdgΔ1.4=47.75 strides/min; Kruskal-Wallis ANOVA (P<0.0001) followed by Dunn's test of multiple comparisons for adjusted P-values: NdgΔ0.4 versus w1118, *P=0.0207; NdgΔ1.4 versus w1118, ***P<0.0001. Data are mean±s.e.m. n≥23 larvae for each genotype. (I,J) Morphology of neuromuscular junctions (NMJs) innervating muscles 6/7 in the 2nd larval abdominal segment of w1118 and NdgΔ1.4 3rd instar larvae stained with anti-Bruchpilot (BRP, green) and anti-HRP (magenta). (K-O) Quantification of morphological NMJ features. Bee swarm box plots (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) are shown. (K,L) Welch's t-test was used to quantify NMJ area (n=number of NMJs analyzed, *P=0.022) (K) and NMJ size ratio (L) (n=number of animals analyzed, **P=0.006). (M,O) Unpaired, two-tailed t-tests were applied to reveal significant differences in (M) branch number (n=number of side branches analyzed, **P=0.008) and (O) bouton density (n=number of NMJs analyzed, **P<0.001). (N) A significant difference in branch length was revealed using the Mann-Whitney non-parametric test (n=number of NMJ side branches analyzed, ***P=0.001). Scale bars: 1 mm in C,E; 20 µm in I.
Altered crawling behavior and neuromuscular junction phenotypes of Ndg mutant larvae. (A) Schematic example of the observed distribution of w1118 and NdgΔ1.4 1st instar larvae (black dots) on agar plates (depicted as circles) outside a central food source (red). (B) Quantification of the larval distribution assay. Mean number of larvae outside the food: w1118=1.44, NdgΔ0.4=2.31, NdgΔ0.4/NdgΔ1.4=8, NdgΔ1.4=19.1, NdgΔ1.4/Df Ndg=19.3. Data were analyzed using one-way ANOVA (P<0.001) followed by Tukey's multiple comparison test with adjusted P-value: w1118 versus NdgΔ0.4, P=0.947 (not significant, n.s.); w1118 versus NdgΔ0.4/NdgΔ1.4, *P<0.01; w1118 versus NdgΔ1.4, ***P<0.001; w1118 versus NdgΔ1.4/Df Ndg, ***P<0.001; n=400 larvae for each genotype. Boxes and whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles represent values from the individual measurements. (C,D) Snapshots taken from crawling recordings of w1118 and NdgΔ1.4 2nd instar larvae. (E,F) Representative tracking patterns of crawling w1118 and NdgΔ1.4 2nd instar larvae. Individually colored lines represent recordings of 25 s. Time points are indicated if the larva left the monitored area before the end of recording. The time interval in brackets indicates the absence from the recorded area. (G) Quantification of mean crawling velocity measured as 5 s of uniform crawling from recordings of w1118 and NdgΔ1.4 L3 larvae. Mean values: w1118=1.265, NdgΔ0.4=1.07, NdgΔ1.4=0.907 mm/s. Kruskal-Wallis ANOVA (P<0.0001) followed by Dunn's test of multiple comparisons for adjusted P-values: NdgΔ0.4 versus w1118, P=0.0597; NdgΔ1.4 versus w1118, ***P<0.0001. Data are mean±s.e.m. (H) Quantification of the stride frequency calculated from five consecutive strides from recordings of w1118 and NdgΔ1.4 3rd instar larvae. Mean values: w1118=73.32, NdgΔ0.4=62.25, NdgΔ1.4=47.75 strides/min; Kruskal-Wallis ANOVA (P<0.0001) followed by Dunn's test of multiple comparisons for adjusted P-values: NdgΔ0.4 versus w1118, *P=0.0207; NdgΔ1.4 versus w1118, ***P<0.0001. Data are mean±s.e.m. n≥23 larvae for each genotype. (I,J) Morphology of neuromuscular junctions (NMJs) innervating muscles 6/7 in the 2nd larval abdominal segment of w1118 and NdgΔ1.4 3rd instar larvae stained with anti-Bruchpilot (BRP, green) and anti-HRP (magenta). (K-O) Quantification of morphological NMJ features. Bee swarm box plots (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) are shown. (K,L) Welch's t-test was used to quantify NMJ area (n=number of NMJs analyzed, *P=0.022) (K) and NMJ size ratio (L) (n=number of animals analyzed, **P=0.006). (M,O) Unpaired, two-tailed t-tests were applied to reveal significant differences in (M) branch number (n=number of side branches analyzed, **P=0.008) and (O) bouton density (n=number of NMJs analyzed, **P<0.001). (N) A significant difference in branch length was revealed using the Mann-Whitney non-parametric test (n=number of NMJ side branches analyzed, ***P=0.001). Scale bars: 1 mm in C,E; 20 µm in I.
These crawling phenotypes prompted us to examine neuromuscular junction (NMJ) morphology in wandering 3rd instar larvae. We therefore dissected larval filets and stained them with anti-Bruchpilot (BRP) antibodies to label T-zones in the synaptic boutons and employed anti-HRP staining to reveal the overall NMJ morphology. We focused on the NMJ at muscle 6/7 in the 2nd abdominal segment because it covers a considerably large area, innervates two muscles, and contains many synaptic boutons (Fig. 6I,J). Compared with NMJs in w1118 control larvae (Fig. 6I), NMJs of NdgΔ1.4 animals were altered in shape (Fig. 6J) displayed a slight but significant increase in NMJ size (Fig. 6K) and a significant size difference between corresponding NMJs on both sides in one segment (Fig. 6L). In addition, overall branching and average branch length were increased in NdgΔ1.4 animals (Fig. 6M,N). We also observed a significantly enhanced bouton density at NdgΔ1.4 junctions (Fig. 6O), suggesting that NDG actually suppresses NMJ maturation. Taken together, these results suggest that loss of NDG leads to defects in NMJ organization, presumably resulting in the climbing and crawling defects observed in NdgΔ mutant larvae.
Loss of NDG leads to reduced reaction to vibrational stimuli and aberrant chordotonal organ morphology
Our previous analyses revealed that developing chordotonal organs highly expressed Ndg in their cap cells and eventually became embedded in a NDG-positive BM. Notably, the altered larval crawling behavior and the pupal case orientation phenotype observed in the absence of NDG could also be interpreted as a result of impaired proprioception. Therefore, we wanted to characterize chordotonal function in NdgΔ mutants on a behavioral level and additionally analyze larval PNS morphology with respect to the chordotonal organs (Fig. 7). Stimulus-induced, relative body length reduction upon applied vibrational stimuli was compared between 3rd instar w1118 control and NdgΔ larvae (Fig. 7A,B). The body length reduction of control larvae in response to vibrational stimuli was on average ∼14%, whereas homozygous NdgΔ0.4 and NdgΔ1.4 larvae showed significantly reduced retraction values (Fig. 7B). NdgΔ1.4 mutants again displayed the most severe phenotype with a nearly abolished reaction (∼4% body length reduction) to vibrational stimuli (Fig. 7B).
Loss of Ndg affects response to vibrational stimuli and PNS morphology. (A) Body length recordings of w1118, NdgΔ0.4 and NdgΔ1.4 larvae exposed to pulsed vibrational stimuli (indicated by black bars). (B) Quantification of body length reduction induced by vibrational stimulation in w1118 and NdgΔ larvae. The relative body length reduction (Δ body length) is calculated from the mean larval length before and during vibrational stimulation. Data are mean±s.e.m., n=24, **P<0.01, ***P<0.001 versus w1118. Kruskal-Wallis ANOVA P<0.0001 and Dunn's test of multiple comparisons for adjusted P-values was applied for statistical analysis. NdgΔ0.4 versus w1118, P=0.0061; NdgΔ1.4 versus w1118, P<0.001. (C,D) Morphology of the lateral pentascolopidial chordotonal organ (lch5) of control (C) and NdgΔ1.4 (D) larvae revealed by Jupiter::GFP fusion protein (blue) expression, NDG (green) and HRP (magenta) antibody staining. Arrows highlight the tip of sensory cilia. (E,F) HRP labeling (black) was employed to reveal dda neurons in w1118 and NdgΔ1.4 larvae. (G) Quantification of the lch5 phenotype of NdgΔ1.4 larvae compared with a w1118 control. A bee swarm box plot (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) is shown. Mann–Whitney non-parametric analysis was applied to reveal significance (n=number of animals analyzed, ***P<0.001). (H) Quantification of the dda alignment phenotype of NdgΔ1.4 larvae compared with w1118 control animals. A bee swarm box plot (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) is shown. Unpaired, two-tailed t-test reveals significant difference (n=number of dda clusters analyzed, ***P<0.001). (I,J) Segments of w1118 (I) and NdgΔ1.4 (J) larvae were labeled with NDG (blue) and HRP (white) antibodies to reveal improper targeting by the dorsal projection of the transverse nerve (TN). (K) Quantification of TN targeting. Mean percentage of dorsal TN branches that target the alary muscles (w1118=84% and NdgΔ1.4=17%). Error bars indicate s.d. Unpaired, two-tailed t-test reveals significant difference (n=number of TNs analyzed in 17 w1118 and 16 NdgΔ1.4 larval filets, respectively, ***P<0.001). Scale bars: 20 µm.
Loss of Ndg affects response to vibrational stimuli and PNS morphology. (A) Body length recordings of w1118, NdgΔ0.4 and NdgΔ1.4 larvae exposed to pulsed vibrational stimuli (indicated by black bars). (B) Quantification of body length reduction induced by vibrational stimulation in w1118 and NdgΔ larvae. The relative body length reduction (Δ body length) is calculated from the mean larval length before and during vibrational stimulation. Data are mean±s.e.m., n=24, **P<0.01, ***P<0.001 versus w1118. Kruskal-Wallis ANOVA P<0.0001 and Dunn's test of multiple comparisons for adjusted P-values was applied for statistical analysis. NdgΔ0.4 versus w1118, P=0.0061; NdgΔ1.4 versus w1118, P<0.001. (C,D) Morphology of the lateral pentascolopidial chordotonal organ (lch5) of control (C) and NdgΔ1.4 (D) larvae revealed by Jupiter::GFP fusion protein (blue) expression, NDG (green) and HRP (magenta) antibody staining. Arrows highlight the tip of sensory cilia. (E,F) HRP labeling (black) was employed to reveal dda neurons in w1118 and NdgΔ1.4 larvae. (G) Quantification of the lch5 phenotype of NdgΔ1.4 larvae compared with a w1118 control. A bee swarm box plot (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) is shown. Mann–Whitney non-parametric analysis was applied to reveal significance (n=number of animals analyzed, ***P<0.001). (H) Quantification of the dda alignment phenotype of NdgΔ1.4 larvae compared with w1118 control animals. A bee swarm box plot (whiskers indicate minimum and maximum values, horizontal line indicates median, box hinges are 25th and 75th percentiles and circles are individual values) is shown. Unpaired, two-tailed t-test reveals significant difference (n=number of dda clusters analyzed, ***P<0.001). (I,J) Segments of w1118 (I) and NdgΔ1.4 (J) larvae were labeled with NDG (blue) and HRP (white) antibodies to reveal improper targeting by the dorsal projection of the transverse nerve (TN). (K) Quantification of TN targeting. Mean percentage of dorsal TN branches that target the alary muscles (w1118=84% and NdgΔ1.4=17%). Error bars indicate s.d. Unpaired, two-tailed t-test reveals significant difference (n=number of TNs analyzed in 17 w1118 and 16 NdgΔ1.4 larval filets, respectively, ***P<0.001). Scale bars: 20 µm.
To reveal potential morphological changes underlying the observed behavioral phenotypes, we examined the lateral chordotonal organs (lch5) of 3rd instar larvae. Therefore, we used the Jupiter::GFP fusion protein, which localizes to the microtubule network (Karpova et al., 2006). In addition, we employed anti-NDG and horseradish peroxidase (HRP) staining (Jan and Jan, 1982) to reveal the chordotonal BM and associated neurons, respectively (Fig. 7C,D). A NDG-positive BM surrounded the whole lch5 of Jupiter::GFP control larvae, including the aligned row of sensory cilia (Fig. 7C, arrow). We observed that the alignment of sensory cilia was frequently lost in NdgΔ1.4 mutant larvae (Fig. 7D, arrow). Although significantly increased in NdgΔ1.4 animals, sensory cilia alignment defects were also observed in control animals and the penetrance of this phenotype varied among the mutant larvae (Fig. 7G).
In addition to the lch5 phenotype, we noticed alterations in other parts of the larval peripheral nervous system. Cell bodies of the dorsal dendritic arborization sensory neurons (ddA neurons) failed to align in NdgΔ1.4 mutant larvae, resulting in a significantly increased dda cluster area (Fig. 7F,H). In addition, targeting of alary muscles by the dorsal projection of the transverse nerve (TN) was significantly reduced in NdgΔ1.4 mutant larvae compared with w1118 larvae (Fig. 7I-K).
DISCUSSION
Analysis of Ndg expression and NDG protein distribution
In this work, we have characterized the single Nidogen/entactin gene in Drosophila melanogaster. In the developing embryo, Ndg expression followed a highly dynamic pattern involving diverse mesodermal cell types such as somatic muscle founder cells, visceral and somatic muscles, as well as dorsal median cells (DMCs). Ndg expression in a subset of somatic muscle founder cells had been previously reported (Artero et al., 2003) and detailed analysis of an intronic ‘Ndg muscle enhancer’ revealed a complex interplay of Forkhead and Homeodomain transcription factors, and their binding sites, which are required to drive Ndg expression in the different mesodermal cell types (Busser et al., 2013, 2012; Philippakis et al., 2006; Zhu et al., 2012). Our analysis further revealed Ndg expression in the extraembryonic amnioserosa and in neuronal-derived chordotonal cap cells, indicating the presence of additional regulatory elements. Beside expression in cap cells, we did not detect Ndg expression in the embryonic CNS but in the closely associated DMCs, which express other BM components and ECM receptors such as Perlecan, Dystroglycan, Glutactin, Laminin and type IV Collagen (Friedrich et al., 2000; Mirre et al., 1992; Montell and Goodman, 1989; Olson et al., 1990). DMCs provide cues for proper axonal pathfinding during transverse nerve (TN) outgrowth, which we found impaired in Ndg mutant larvae, and regulate bifurcation of the median nerve in Drosophila (Chiang et al., 1994; Gorczyca et al., 1994). Therefore, studying a potential role of the secreted ECM in this context would be an interesting topic for future investigations. Interestingly, embryonic Ndg expression was absent from hemocytes and fat body cells, although these cells strongly secrete Laminin, type IV collagens and other ECM components (Le Parco et al., 1986; Pastor-Pareja and Xu, 2011; Rodriguez et al., 1996; Urbano et al., 2009; Wolfstetter and Holz, 2012), indicating that Ndg expression is regulated independently from other BM genes. Moreover, this raises the interesting question of how NDG is ubiquitously distributed to all BMs? This could be achieved by either diffusion of the secreted protein or a yet unknown transport mechanism.
CEL-Seq transcriptome analysis in C. elegans reveals high nid-1 mRNA expression in mesodermal tissues upon gastrulation (Hashimshony et al., 2015) and NDG proteins strongly associate with body wall muscles in Drosophila (Dai et al., 2018, this study), C. elegans, ascidians and mice (Ackley et al., 2003; Fox et al., 2008; Kang and Kramer, 2000; Nakae et al., 1993), suggesting an important, evolutionary conserved function in this particular tissue. Indeed, C. elegans NID-1 and NID2 in mice are enriched at neuromuscular junctions, and, in agreement with our analyses in Drosophila, are required for proper NMJ structure and function (Ackley et al., 2003; Fox et al., 2008).
NDG is not required for BM assembly
Our analysis of NDG distribution in the absence of major BM components revealed a dependency of NDG localization from Laminin but not from type IV Collagen or Perlecan. This is in agreement with studies on γ1III4-mice in which the NDG-binding site in the Laminin γ1 chain had been deleted. This analysis revealed severe reduction of NDG from BMs, as well as perinatal lethality and organ defects similar to those observed in Nid1−/− Nid2−/− double mutant mice, which further strengthens the assumption that Laminin plays an essential role in localizing NDG to forming BMs (Bader et al., 2005; Halfter et al., 2002; Willem et al., 2002). The two Laminin heterotrimers in the fly function in a redundant manner to localize NDG to embryonic BMs. In a single loss of function background, however, only absence of the LanA trimer (containing the Drosophila Laminin α3/5 homolog) affected NDG localization in our antibody staining, suggesting different binding and polymerization activities of the Drosophila laminins. In line with these findings, combined reduction of LanA and NDG around the developing gonad was also observed in βPS-integrin mutant embryos (Tanentzapf et al., 2007). No obvious changes in the distribution of BM core components (laminins, collagens and Perlecan) were found in our immunohistochemical analysis of Ndg mutant animals, reflecting its non-essential role during development. This finding, further supported by similar analyses in other species (Ackley et al., 2003; Bader et al., 2005; Kang and Kramer, 2000; Kim and Wadsworth, 2000; Rossi et al., 2015) contradicts the proposed role for NDG as linker between inner and outer ECM networks during BM formation, suggesting a negligible role for NDG in overall BM assembly (Ho et al., 2008; Hohenester and Yurchenco, 2013; Jayadev and Sherwood, 2017; Kramer, 2005; Timpl and Brown, 1996; Yurchenco, 2011).
Loss of NDG affects BM stability and function
Although BM changes are barely detectable in Ndg mutant animals by conventional immunohistochemistry, our ultrastructural analysis revealed that BM continuity is compromised if NDG is absent or reduced. Our findings that BM barrier function and stability were decreased in Ndg mutant larvae suggests that NDG, or NDG-mediated crosslinking of other ECM molecules, plays an important role in sealing the assembled BM. Intriguingly, Matsubayashi et al. observe holes in the forming BM around the embryonic Drosophila CNS that are progressively closed during hemocyte migration and concurrent Col IV deposition (Matsubayashi et al., 2017). In addition, studies on mice demonstrate that, although Laminin is sufficient for the assembly of early BM scaffolds, incorporation of type IV Collagen is essentially needed for BM integrity and stability (Poschl et al., 2004). Therefore, it is tempting to speculate that crosslinking of hemocyte-deposited Col IV might be impaired in Ndg mutants, leading to decreased BM stability and barrier function.
The increased number of BM holes in close proximity to tracheal branches and visceral longitudinal musculature in Ndg mutants and the reduced tolerance to osmotic shock further implies that NDG protects BM integrity from mechanical stress therefore separating and maintaining different compartments in the larval body. Although our experiments demonstrated reduced BM stability in Ndg mutant larvae, the BM phenotype observed exhibited some local variation. An explanation for this could be that, although the genetic background of the larvae determines overall BM stability, individual behavior, physical activity and motility levels could modulate Ndg loss-of-function phenotypes. A comparable finding was made by Tsai and colleagues (Tsai et al., 2012) who reported a process in which larval NMJ size changes in response to LanA levels at the NMJ, which are influenced by crawling activity as well as nervous responses to environmental cues. Given the NDG-localizing properties of LanA as well as the variations in NMJ size observed in NdgΔ1.4 animals, it is intriguing to think of a feedback mechanism that employs spatiotemporal reorganization of the larval BM by NDG to adapt NMJ morphology to altered environmental conditions.
Loss of NDG affects behavioral responses and PNS morphology
With the exception of Nid1−/− Nid2−/− double mutant mice that die at birth, loss of NDG is generally associated with a range of rather subtle morphological and behavioral phenotypes. Notably, abnormalities described in the absence of NDG are not always fully penetrant or they exhibit phenotypic variation (Bader et al., 2005; Böse et al., 2006; Dong et al., 2002; Hobert and Bülow, 2003; Kang and Kramer, 2000; Kim and Wadsworth, 2000; Murshed et al., 2000; Schymeinsky et al., 2002; Zhu et al., 2017). Moreover, subtle behavioral phenotypes can be enhanced by adjusting the experimental conditions (Ackley et al., 2003). In agreement with this, analysis of Drosophila Ndg mutants did not reveal an essential developmental function but a range of less overt, often variable phenotypes. After vibrational disturbance of a normal crawling phase, wild-type Drosophila larvae show a complex sequence of behavioral pattern. The larvae discontinue their forward movement and show a head retraction (‘hunch’) followed by a head turning phase (‘kink’) (Bharadwaj et al., 2013; Ohyama et al., 2013; Wu et al., 2011). This sequence of behavior is initiated upon mechanical stimuli delivered to the substrate, which are detected by segmental chordotonal organs, mediated by stimulus processing in the central nervous system and executed under neuromuscular control. Our analysis of the vibrational response behavior in Ndg mutant larvae shows a statistically significant reduction in body contraction. This finding points to a crucial role for NDG in vibrational response behavior. However, the precise level of NDG action cannot be inferred from these behavioral data. NDG expression in chordotonal cap cells, the morphological defects observed in larval lch5 as well as the altered geotaxis of Ndg mutant larvae suggest a contribution to mechanosensation. Notably, a similar lch5 phenotype was recently observed in mutants for the sensory ECM protein Artichoke, which further supports this assumption (Andres et al., 2014). Defective wing inflation and the aberrant NMJ morphology observed in Ndg mutant animals additionally indicate that loss of NDG affects the neuromuscular system. Indeed, previous research on C. elegans and mouse has uncovered a role of NDG for the structural development of neuromuscular junctions, as well as altered locomotor behavior upon loss of NDG (Ackley et al., 2003; Bader et al., 2005; Dong et al., 2002; Fox et al., 2008).
In conclusion, our initial characterization of Ndg mutants in Drosophila neither revealed an essential developmental function nor supported the proposed role for NDG as universal ECM-linker molecule. However, NDG appears to be important for BM sealing and stability, proper mechanosensation and neuromuscular function.
MATERIALS AND METHODS
Fluorescence antibody staining
Antibody staining of Drosophila embryos and larvae was essentially performed as described by Mueller in Dahmann (2008). For antibody staining of wandering 3rd instar larvae we adapted the protocol from Klein in Dahmann (2008) with the following modifications: animals were relaxed before dissection by briefly dipping them into 60°C water and a permeabilization step (10 min wash in PBS supplied with 1% Triton-X100) was added before blocking and primary antibody incubation. The following primary antibodies were used at their specified dilutions: mouse anti-Bruchpilot (Brp nc82, 1:100; Wagh et al., 2006; DSHB), guinea pig anti-Collagen IV (COLLIV, 1:500; Lunstrum et al., 1988), sheep anti-digoxigenin alkaline phosphatase-conjugated Fab fragments (DIG-AP, 1:4000, Roche Applied Science, 11093274910), mouse anti-green fluorescent protein (GFP, 1:500, Roche Diagnostics, 11814460001), rabbit anti-green fluorescent protein (GFP, 1:500, Abcam, ab290), chicken anti-green fluorescent protein (GFP, 1:500, Abcam, ab13970), goat anti-horseradish peroxidase, Cy3-conjugated (HRP, 1:200, Jackson ImmunoResearch, 123-165-021), guinea pig anti-Laminin A (LanA, 1:500; Harpaz and Volk, 2012), mouse anti-Laminin A (LanA, 1:500; Takagi et al., 1996), rabbit anti-Wing blister (LanWb, 1:100; Martin et al., 1999), rabbit anti-Nidogen (NDG, 1:1.000; Wolfstetter et al., 2009) and rabbit anti-Perlecan (PCAN, 1:1.000; Friedrich et al., 2000). Alexa Fluor-, Cy-, Biotin-SP- and HRP-coupled secondary antibodies were purchased from Dianova and Jackson ImmunoResearch; DAPI was from Sigma Aldrich. Embryos and larval tissues were embedded in Fluoromount-G (Southern Biotech) before visualization under Leica TCS SP2 or Zeiss LSM 800 confocal microscopes.
Developmental northern blot
Northern blot analyses were performed using standard procedures (Maniatis, 1982). RNA was extracted by the guanidine thiocyanate/phenol/chloroform extraction method (Chomczynski and Sacchi, 1987). Poly(A)-tailed RNA was isolated using a Pharmacia Kit (Pharmacia Biotech). A 2.5 kb fragment of the Ndg cDNA (cNdg5, S.B., unpublished) was radioactively labeled and hybridized to the Northern filter. Exposure time for northern blots was 1.5 days. The blot was re-probed with a Drosophila 19S probe to evaluate equimolar loading. The Ndg blot shown in this work was identically conducted as those for wing blister (LanWb) and LanA (Martin et al., 1999) allowing scarce comparison of the expression levels of these different ECM genes.
Whole-mount in situ hybridization
N- and C-terminal fragments from a full-length Nidogen cDNA clone (DGRC cDNA clone LP19846; GenBank: BT031149.1) were PCR amplified and sub-cloned into the pCRII-TOPO vector with the TOPO TA Dual Promoter Kit (Invitrogen). Primer sequences for a 571 bp N-terminal fragment were GGACCCATCCATATCCCGCCACAAT and GCAATCAGTGCCACCTGGAAGGTGT, whereas CGTGGCATTGCCGTGGATCCCT and GGTGCATCCTGTGGAGGCGCT were employed to amplify a 549 bp C-terminal fragment. Templates for probe synthesis were generated by PCR using M13 primers supplied with the TOPO TA Dual Promoter Kit. Digoxygenin (DIG)-labeled sense and antisense probes were made by SP6/T7 in vitro transcription with the DIG RNA Labeling Kit (Roche Applied Science). In situ hybridization on Drosophila embryos was performed according to Lécuyer et al. (2008) with modifications adapted from Pfeifer et al. (2012). Sheep anti-DIG-AP Fab fragments (1:4000, Roche Applied Science), biotinylated donkey anti-sheep IgG (1:400, Dianova), the Vectastain ABC Standard Kit (Vector Laboratories) and the TSA Amplification Renaissance Kit (PerkinElmer) were used for fluorescence in situ hybridization (FISH) detection.
Fly stocks and genetics
Flies were grown under standard conditions (Ashburner, 1989) and crosses were performed at room temperature or at 25°C. Staging of embryos was carried out according to Campos-Ortega and Hartenstein (1997). The following mutations and fly stocks were used in this study: as control or wild-type stocks we employed white1118 (w1118), w*; KrIf-1/CyO P{Dfd-EYFP}2, Oregon-R or balanced sibling embryos. We used Df(2R)BSC281 as a deficiency for Ndg (Df Ndg); Df(2L)Exel7022, which deletes the two adjacent Drosophila type IV Collagen genes viking (vkg) and Col4a1 (previously referred to as Cg25C); Df(3L)Exel8101 as deficiency for Laminin A; Df(2L)TE35B-2 as deficiency for Laminin wing blister (Gubb et al., 1985); Df(3L)Exel6114 as Laminin B2 deficiency (Df LanB2); and Df(1)Exel6230, which removes the trol locus but also the adjacent segmentation gene giant. Mi{ET1}MB04184, P{hsILMiT} and P{Δ2-3}99B were used to generate the Ndg deletion alleles NdgΔ0.4 and NdgΔ1.4 (Fig. S2, see below). The null alleles LanB2knod (Wolfstetter and Holz, 2012) and wbHG10 were used as well as Jupiter::GFP (Karpova et al., 2006) as protein trap line.
Generation of Nidogen deletions
The pMiET1 transposon insertion Mi{ET1}MB04184 (Metaxakis et al., 2005) was used in an imprecise excision screen. Lethality reported for this line was not associated with the Mi{ET1} insertion and homozygous viable, isogenic stocks were established after approximately five generations of free recombination over a wild-type chromosome. The position of the Mi{ET1} insertion, initially revealed by flanking sequence recovery (Bellen et al., 2011; GenBank accession number ET201740.1), was confirmed in the isogenic stocks by inverse PCR (Ochman et al., 1988). Therefore, we detected a 35 bp deletion 141 bp upstream of the insertion site that was present in the flanking sequence recovery data (ET201740.1) and in w1118 DNA samples but not in the 6th GenBank release of the D. melanogaster genome annotation (Hoskins et al., 2015) or the corresponding sequence derived from Oregon-R genomic DNA. Therefore, we considered this small deletion as a naturally occurring sequence variation in the Ndg locus. Before serving as Minos transposase source in the screen, the P{hsILMiT} insertion (Metaxakis et al., 2005) was remobilized from the SM6a balancer and inserted onto a w1118 X-chromosome employing P{Δ2-3}99B as transposase source. pMiET1 remobilization was induced in the germ line of 2nd and 3rd instar larvae by daily 1 h heat shocks at 37°C in the presence of the w1118, P{hsILMiT} helper chromosome. A total of 301 single excision events were identified due to the absence of the Mi{ET1}-associated Mmus/Pax6-GFP expression (Berghammer et al., 1999) and screened for deletions by single-fly PCR analysis with the following primer combination: GCCAAGGAATGGGAGTGCTCTGGAT and GGAGCCATCCTCGAACTCGTACAATT. Deletions were detected in two viable Mi{ET1}MB04184 excision lines (henceforth referred to as NdgΔ0.4 and NdgΔ1.4). Sequencing the molecular lesions uncovered a 1401 bp deletion (2R:10312780-10314180) and a P-Element remnant of 17 nucleotides (TGCCACGTAGCCGGAAT) in NdgΔ1.4. In the case of NdgΔ0.4 we detected a 417 bp (2R:10313549-10313965) deletion and an insertion of 19 nucleotides (CGAGCAAAATACAAAATAC) at the former P-Element insertion site (Fig. S2).
Dextran permeability assay
Wandering 3rd instar larvae were picked and relaxed by briefly dipping them into 60°C water. Larval filets were dissected in PBS, fixed for 10 min in a drop of 4% formaldehyde in PBS, rinsed with 500 µl PBS and incubated in 50 µl of 25 µg/ml anionic, lysine-fixable Texas Red-coupled dextran (Thermo Fisher Scientific, D3328) in PBS for 10 min. The dextran solution was removed and filets were re-fixed for 10 min in 4% formaldehyde followed by 10 min washing in PBS containing 0.3% NP-40. Filets were mounted in Fluoromount-G and analyzed under a Zeiss LSM 800 confocal microscope using the same setting for wild-type and Ndg mutants. To quantify fluorescence intensity levels, three measurements from somatic muscle 7 in the 2nd abdominal segment (approximately 50, 100 and 200 µm away from the muscle's attachment site) were taken from equally sized image stacks (n=24 measurements from 4 experimental replicates) using the Fiji distribution of ImageJ (Schindelin et al., 2012). The mean fluorescence background (calculated from three independent measurements of areas containing no larval tissue) was subtracted from every measurement to obtain the corrected fluorescence intensity applying the formula: CTF=integrated density ROI-(area ROI×mean fluorescence background).
Larval climbing and gravity assay
Matching numbers of flies were placed on food-filled vials and allowed to lay eggs for 3 days. Pupae were examined at the pharate adult stage (7-9 days after egg laying at 25°C). To assay larval climbing performance, the distance between the food surface and the cotton plug was divided into four zones. Pupae formed in each zone were counted. Pupae positioned on the border between two zones were assigned to the next higher zone. Pupae were further assigned into three categories: (1) upright=0° ‘head up’ or 180° ‘head down’ ±22.5°; (2) tilted=45° or 225°±22.5°; and (3) flat=90° or 270°±22.5°, according to their orientation along the gravity axis.
Scanning electron microscopy
Wandering 3rd instar larvae were dissected in PBS. For somatic muscle preparations larvae were heat-relaxed in 60°C water prior to dissection. After 3 h fixation in 2.5% glutaraldehyde in PBS, samples were rinsed several times in PBS and dehydrated in an ascending ethanol series (50%, 70%, 80%, 90% and 2×100% ethanol for 10 min each). After critical-point drying in a Balzers CPD 030 at 40°C, samples were mounted on stubs using double-sided adhesive tape and sputter-coated with a thin layer of gold (Balzers SCD 004). Samples were analyzed under a Zeiss DSM982 scanning electron microscope. Acceleration voltage was set to 3 kV. Settings for tilt angle, spot size, scanning mode and magnifications were kept constant throughout image acquisition.
Osmotic stress assay
Wandering 3rd instar larvae of indicated genotypes were picked and rinsed with PBS. The larvae were cut in half and the anterior part was inverted in order to expose the wing imaginal discs. After removing the anterior part of the gut and fat body, the samples were transferred to distilled water and the time until bursting of the first wing imaginal disc was measured. Measurements were stopped after 10 min and wing imaginal discs that were found intact after this period were assigned to the 10 min group.
Lethality tests
Parental flies (3-7 days old) of the indicated genotype were shifted on grape juice agar plates for 8 days. Agar plates were changed every 2nd day and defined numbers of eggs were transferred to fresh agar plates supplemented with dried yeast for further comparative analysis. Undeveloped eggs, hatched larvae, pupal cases and eclosure were counted daily. All plates were incubated at 25°C and constantly humidified.
Larval feeding assay
Fifty freshly hatched 1st instar larvae were transferred to apple juice agar plates with a central spot of food dye-supplemented yeast paste. Larvae moving outside the food were counted 1 and 2 h after the transfer. Fresh yeast paste was added every day and individual larvae were checked for the uptake of dyed food after 72 h. The assay was performed at 25°C and 60% humidity conditions.
Larval locomotion assay
Flies were allowed to lay eggs for 2 h on apple juice agar plates supplemented with yeast paste. Fresh yeast paste was added every day and individual 2nd instar larvae were picked after 72 h (at 25°C and 60% humidity) and rinsed with distilled water. A single larva was placed in the middle of a ø52.5 mm Petri dish prepared with a ø35 mm central arena (1% agar in PBS) surrounded by a ring of high-salt agar (3 M NaCl in PBS) as a locomotion barrier. For every recording, a fresh agar plate was used. Larvae were allowed to move freely in the central arena for 2-5 min, respectively, and locomotion was recorded in a square of 12.5×9.5 mm2 using a Zeiss AxioZoom.V16 stereo zoom microscope equipped with LED ring light and an Axiocam 503 color camera. Time series were recorded with four frames/second employing ZEN Blue software and 25 s subsets were further analyzed and edited with the manual tracking plugin MTrackJ in ImageJ.
Larval crawling assay
Larval crawling analyses were performed as described by Eschbach et al. (2011) for responses to vibrational stimulation and measurement of body length and stride frequency. For the trials, 3rd instar larvae were placed on 1% agarose covered Petri dishes (60 mm diameter) as a crawling stage. This stage was placed in a silencing foam-covered box (57×57×47 cm3) and illuminated by a red darkroom lamp (Pf712em, Philips, 15 W and 7 lm). Video recordings of the trials were performed via an infrared CMOS-Color-Camera (CCD-651, Conrad Electronics). The analog video signals were grabbed by using an analog/digital video converter (700-USB, Pinnacle Systems) and recorded via a computer for further analysis (25 fps). Measurements of crawling speed, stride frequency and body length of the larvae were performed by using the video tracking software Cabrillo Tracker (v4.91, Douglas Brown, Open Source Physics, 2015). To quantify mean velocity and stride frequency for five larvae per trial, the terminal ends were video-tracked for 10-20 s. Only periods of unrestricted and constant crawling behavior were used for analysis. The mean velocity was calculated from 5 s of crawling and the stride frequency from the duration of five complete strides in a row. The maximum crawling speed of a single stride was defined as the starting point for each measurement. To determine body length reduction due to vibrational stimuli, apical and terminal ends of four larvae per trial were video tracked. The vibrational stimuli were generated via the audio edition software Audacity (v2.05, Audacity-Team, GNU GPLv2+, 2015) at a frequency of 100 Hz and delivered via a full-range speaker (BPSL 100/7, Isophon/Gauder Akustik, 60-20.000 Hz and 7 W) to the crawling stage. For each trial, three stimuli with the duration of 1 s and an acceleration of 12 m/s2 were applied (interpulse-interval 9 s).
Statistical analyses
Data for quantification of PNS phenotypes was obtained by employing measurement tools in Zeiss ZEN lite blue (version 1.1.2) and black (version 8.1 with Sp1) edition software as well as the measurement tool in the Fiji distribution of ImageJ (Schindelin et al., 2012). Prism v6.0 and v7.02 (GraphPad Software) was used for all statistical analyses.
Acknowledgements
We thank Bloomington Drosophila Stock Center (NIH P40OD018537), Lynn Cooley, the Drosophila Genomics Resource Center (DGRC, NIH 2P40OD010949-10A1), the Developmental Studies Hybridoma Bank (created by the NICHD of the NIH and maintained at the University of Iowa), the imaging unit at the Justus-Liebig-Universität Giessen, Liselotte Fessler, the KYOTO Stock Center at the Kyoto Institute of Technology, Maurice Ringuette, John Roote, Talila Volk and Gerd Vorbrüggen for sending materials and fly stocks.
Footnotes
Author contributions
Conceptualization: G.W., A.H.; Methodology: G.W., A.H.; Software: G.W.; Validation: G.W., I.D., K.P., U.T., J.A.A., A.H.; Formal analysis: G.W., I.D., K.P., U.T., J.A.A.; Investigation: G.W., I.D., K.P., U.T., J.A.A., D.C.P., S.B., A.H.; Resources: G.W.; Data curation: G.W.; Writing - original draft: G.W., I.D., K.P., U.T., J.A.A., D.C.P., S.B., A.H.; Writing - review & editing: G.W., S.B., R.H.P., A.H.; Visualization: G.W., I.D., K.P., U.T., J.A.A., D.C.P., A.H.; Supervision: G.W., R.L.-H., R.H.P., A.H.; Project administration: A.H.; Funding acquisition: S.B., R.H.P., A.H.
Funding
This work was supported by Vetenskapsrådet (621-2003-3408 to S.B. and 621-2015-04466 to R.H.P.), Cancerfonden (4714-B03-02XBB to S.B. and 2015/391 to R.H.P.), Barncancerfonden (2015/0096 to R.H.P.), Göran Gustafsson Stiftelser and Deutsche Forschungsgemeinschaft (Ho-2559/3-3 and Ho-2559/5-1 to A.H.).
References
Competing interests
The authors declare no competing or financial interests.