Analysis of the role of Nidogen/entactin in basement membrane assembly and morphogenesis in Drosophila

Basement membranes (BMs) are thin sheet-like specialized extracellular matrices found at the basal surface of epithelia and endothelial tissues. They have been conserved across evolution and are required for proper tissue growth, organization, differentiation and maintenance. The major constituents of BMs are two independent networks of Laminin and Type IV Collagen interlinked by the proteoglycan Perlecan and the glycoprotein Nidogen/entactin (Ndg). The ability of Ndg to bind in vitro Collagen IV and Laminin, both with key functions during embryogenesis, anticipated an essential role for Ndg on morphogenesis linking the Laminin and Collagen IV networks. This was supported by results from in vitro and cultured embryonic tissues experiments. However, the fact that elimination of Ndg in C. elegans and mice did not affect survival, strongly questioned this proposed linking role. Here, we have isolated mutations in the only Ndg gene present in Drosophila. We find that while, similar to C.elegans and mice, Ndg is not essential for overall organogenesis or viability, it is required for appropriate fertility. We also find, alike in mice, tissue-specific requirements of Ndg for proper assembly and maintenance of certain BMs, namely those of the adipose tissue and flight muscles. In addition, we have performed a thorough functional analysis of the different Ndg domains in vivo. Our results support an essential requirement of the G3 domain for Ndg function and unravel a new key role for the Rod domain in regulating Ndg incorporation into BMs. Furthermore, uncoupling of the Laminin and Collagen IV networks is clearly observed in the larval adipose tissue in the absence of Ndg, indeed supporting a linking role. In light of our findings, we propose that BM assembly and/or maintenance is tissue-specific, which could explain the diverse requirements of a ubiquitous conserved BM component like Nidogen. Author Summary Basement membranes (BMs) are thin layers of specialized extracellular matrices present in every tissue of the human body. Its main constituents are two networks of Laminin and Type IV Collagen linked by Nidogen (Ndg) and proteoglycans. They form an organized scaffold that regulates organ morphogenesis and function. Mutations affecting BM components are associated with organ dysfunction and several congenital diseases. Thus, a better comprehension of BM assembly and maintenance will not only help to learn more about organogenesis but also to a better understanding and, hopefully, treatment of these diseases. Here, we have used Drosophila to analyse the role of Ndg in BM formation in vivo. Elimination of Ndg in worms and mice does not affect survival, strongly questioning its proposed linking role, derived from in vitro experiments. Here, we show that in the fly Ndg is dispensable for BM assembly and preservation in many tissues, but absolutely required in others. Furthermore, our functional study of the different Ndg domains challenges the significance of some interactions between BM components derived from in vitro experiments, while confirming others, and reveals a new key requirement for the Rod domain in Ndg function and incorporation into BMs.

. Finally, while Col IV deposition in wing imaginal discs and embryonic ventral nerve cord BMs is not required for localization of Laminins and Nidogens, it is essential for Perlecan incorporation (31,32). The Drosophila Laminin αβγ trimer family consists of two members comprised of two different α subunits encoded by LamininA and wing blister, one β and one γ subunit encoded by LamininB1 and LamininB2, respectively (33). Same as Col IV, Laminins can also self-assemble into a scaffold through N-terminal (LN) domains interaction in short arms (34). While, elimination of laminins in Drosophila, in contrast to mice, does not impair gastrulation, it affects the normal morphogenesis of most organs and tissues, including the gut, muscles, trachea and nervous system (35) (36). In addition, abnormal accumulation of Col IV and Perlecan was observed in laminins mutant tissues (35). Perlecan, encoded by the trol (terribly reduced optic lobes) gene, is subdivided into five distinct domains. Interactions with Laminins and Col IV occur through domains I and V (reviewed in (37). Mutations in trol affect postembryonic proliferation of the central nervous system, plasmatocytes and blood progenitors (38)(39)(40). Loss of trol also affects the ultrastructure and deposition of laminins and Col IV in the ECM around the lymph gland (40). All together, these results suggest that the BMs components laminin, Col IV and Perlecan are all essential for proper development. In addition, they also reveal that there is a hierarchy for their incorporation into BMs that seems to be tissue-specific and required for proper BM assembly and function. However, the role of Ndg in Drosophila morphogenesis and BM assembly has remained elusive. This may be in part due to the lack of mutations in this gene.
In this work, we have dissected the role of Ndg in Drosophila. Using a newly generated anti-Ndg antibody, we have shown Ndg accumulates in the BMs of embryonic, larval and adult tissues. By isolating several mutations in the single Drosophila Ndg gene, we find that while, similar to C.elegans and mice, Ndg is not required for overall organogenesis or viability, it is required for fertility. Also similar to the tissue-specific defects in mice and C. elegans, we find that the BM surrounding the larval fat body and flight muscles of the notum is disrupted in the absence of Ndg. Furthermore, we observed uncoupling of Laminin and Collagen IV in the fat body of Ndg mutants, indeed supporting a role of Ndg as a linker between the two networks. In addition, we have performed a thorough functional analysis of the different Ndg domains in vivo, which, on one hand, supports an essential requirement of the G3 domain for Ndg function and, on the other hand, uncovers a new key role for the Rod domain in regulating Ndg incorporation into BMs. Finally, we find that BM assembly is not universal but differs depending on the tissue and propose that this could explain the diverse requirements of a ubiquitous conserved BM component like Nidogen.

Nidogen localizes to the BM of embryonic, larval and adult tissues.
Previous analysis has shown that, during embryogenesis, Ndg is expressed in multiple mesodermal cell types, such as visceral mesoderm, somatic muscle founder cells, a subset of pericardial and cardial cells and at the edges of the visceral mesoderm (41) (42,43). Here, we decided to further analyse Ndg expression in embryonic, larval and adult tissues. In order to do this, an antibody against a peptide encoded by exon 7 was developed (see Materials and Methods). We found that in addition to the pattern described previously, similar to laminins (35), Ndg was also detected in the BM surrounding most embryonic tissues, including muscles, gut and ventral nerve cord ( Fig 1A, A', B, B'), in stage 16 embryos. However, in contrast to laminins, Ndg was not enriched at muscle attachment sites (Fig 1A). In addition, a careful analysis of Ndg expression in stage 13 embryos revealed a dotted pattern along the visceral mesoderm, which differs from the continuous line observed around the muscles or the ventral nerve cord (Fig 1C, C'). At this stage, caudal visceral mesodermal cells migrate over the visceral mesoderm. In fact, using a marker for these cells, croc-lacZ (44), we found that Ndg accumulated around them as they migrate (Fig 1C, C'). In this case, Ndg seem to be organized in track-like arrays, similar to the distribution of laminins around migrating hemocytes. Ndg was also found in migrating hemocytes, as visualized using a version of Nidogen tagged with superfolder GFP (sGFP), expressed from a duplication of the Ndg genomic region (Fig 1D, (45).
Finally, Ndg was also found at high levels in chordotonal organs ( Fig 1A, A', asterisk). These results suggest that as it is the case for laminins, Ndg can be deposited and/or assembled in different patterns throughout embryogenesis.
In addition, and similar to the other BM components, Ndg was found in the BMs that surround most larval tissues, including fat body, imaginal discs, tracheae, salivary glands, midgut, mature muscles and heart ( Fig 1E and Fig 2), as well as in the follicular epithelium of the adult ovary ( Fig 1F).

Fat body adipocytes and blood cells are the main source of Nidogen in larval BMs
A recent study has shown that macrophages are the major producers of BM components in the Drosophila embryo (32). To investigate the cellular origin of Nidogen in the developing fly, we designed a GAL4-driven UAS-controlled short hairpin against super-folder GFP (sGFP) to eliminate sGFP tagged Nidogen without disrupting normal function of endogenous untagged Nidogen (isGFPi, Fig 2A). This approach is similar to iGFPi (in vivo GFP interference, Pastor-Pareja and Xu, 2011) and iYFPi (Zang et al., 2015), which we previously used to show that fat body adipocytes are the major source of Collagen IV and Perlecan in the larva. We found here that isGFPi knock down of Ndg.sGFP driven by Cg-GAL4, which drives expression in fat body and blood cells (Cg>isGFPi), reduced the presence of Ndg.sGFP in the whole animal. Ndg.sGFP signal was largely reduced or undetectable in most tissues, including fat body itself, imaginal discs, tracheae, midgut and heart ( Fig 2B). Deposition of Ndg.sGFP was only partially reduced in the ventral nerve cord, the imaginal ring of the salivary gland and body wall muscles, and was not affected in myoblasts, suggesting that these tissues could produce their own Ndg ( Fig 2B).
These results strongly suggest that, as it is the case for Collagen IV, fat body and blood cells are the main source of Ndg in the larvae.
We next decided to assess the origin of other components by performing the same essay for sGFP tagged Laminin, GFP tagged Col IV and YFP-tagged Perlecan (S1 Fig). We found that, similar to Nidogen, fat body and blood cells are the main source of the laminins, Col IV and Perlecan found in the BM of all tissues, except myoblasts and partially the VNC (S1 Fig).
Tracheal cells seem also able to produce some Perlecan but not laminins or Col IV (S1 Fig). Finally, we found that wing imaginal disc cells and muscles can also produce their own laminins (S1 Fig). In all, our results show that although the four core BM components are largely produced by fat body adipocytes and blood cells, there are tissue-specific and componentspecific exceptions to this rule.

Generation of Nidogen mutant alleles
The Drosophila genome contains a single Ndg gene. To analyse Ndg requirements during development, we aimed to isolate a deficiency uncovering the gene Ndg As these deficiencies removed other genes, we took advantage of the CRISPR-Cas9 technology to isolate a series of specific Ndg alleles (Fig 3A; see Materials and methods). To generate Ndg null alleles, embryos were injected with Cas9 mRNA and a combination of four sgRNAs designed against the 5'UTR exon, exon 3 (two sg RNAs) and exon 8. Two mutant lines in which the intervening Ndg sequence had been deleted partially (Ndg 1 ) or completely (Ndg 2 ) ( Fig 3A) were isolated. Gene CG3422, contained between exons 9 and 10 of the Ndg gene was not perturbed. Both mutations are predicted to be Ndg null alleles because of absence transcription start site. In fact, qRT-PCR, using different primers along the Ndg gene, showed no mRNA expression in Ndg 1 homozygous mutant larvae compared to wild type controls ( Fig 3B).
Furthermore, consistent with Ndg 1 and Ndg 2 being null alleles staining with our Ndg antibody could not detect presence of the protein in larval or embryonic tissues (Fig 3C and D).
As domain G3 has been postulated to be critical for binding of Ndg to laminins, we also isolated Ndg mutant alleles in which this domain and the adjacent rod domain were eliminated (Rod-G3 alleles) in order to analyze its function in the context of the whole organism. In this case, transgenic lines stably expressing an sgRNAs against exon 5 were generated and crossed to flies expressing Cas9 (see Materials and Methods). Three mutants alleles, Ndg Rod-G3.1 , Ndg Rod-G3. 2 and Ndg Rod-G3.3 , were selected ( Fig 3A). Two of them, Ndg Rod-G3.1 , Ndg Rod-G3.2 , were deletions of five and eight base pairs that resulted in frame-shifts generating stop codons eight and seven amino acids after the shift, respectively. In the other one, Ndg Rod-G3.3 , six base pairs were replaced by seven different ones, generating a frame-shift and a stop codon right after the shift. As expected, no staining using the antibody generated in this study was detected in Ndg Rod-G3.1 homozygous embryos (Fig 3D and S3 Fig).
All CRISPR/Cas9 Ndg mutant alleles we generated were homozygous viable with no obvious morphological abnormalities (Fig 3E and data not shown). These data shows that, in

Nidogen is required for integrity of the BM of the larval fat body and adult flight muscles
Once shown that Ndg mutant flies are viable, we decided to analyze the effects of Ndg loss in the BMs of the fly. We could not detect defects in most of the BMs we analyzed, including those present in the embryo, larval epidermis, imaginal discs salivary glands, gut, muscles, ventral nerve cord and the follicular epithelium in the ovary. However, we did observe a clear defect in the BM surrounding the larval fat body (Fig 4A). The larval fat body is an organ formed by large polyploid cells (adipocytes) covered by a BM that separates it from the hemolymph (47). This BM contains, besides Ndg, the other three major components of BMs, Col IV, Laminins and Perlecan. Using tagged versions of these proteins, we found that the BM surrounding the fat body adipose tissue of Ndg 1 mutant larvae showed many holes, in contrast to the continuous appearance of the BM in wild type controls ( Fig 4A). This phenotype was also observed when knocking down Ndg expression using the Cg-GAL4 (Ndg i , Fig 4A). BM rupture upon Ndg loss was also observed in transheterozygotes Ndg 1 /Ndg 2 and Ndg 1 /Df(2R)BSC281 (Fig 4B), confirming a requirement for Ndg to preserve fat body BM integrity. Furthermore, this phenotype was also observed in transheterozygotes Ndg 1 / Ndg Rod-G3.1 larvae and in homozygous Ndg Rod-G3.1 (Fig 4B and data not shown), indicating a strong requirement of the Rod and G3 domains for this Ndg function. Confirming that these phenotypes reflected a loss of Ndg function, the Ndg.sGFP transgene rescued the rupture of fat body BMs in Ndg 1 mutants ( Fig 4C).
We next investigated whether adipose tissue physiology was affected in Ndg mutants. To do this, we stained fat body adipocytes with neutral lipid dye BODIPY and found that the lipid content in Ndg 1 and Ndg 2 mutant adipocytes was reduced, with some cells presenting fewer and smaller lipid droplets than controls ( Fig 4C).
In addition to fat body BM defects observed in Ndg 1 mutant larvae, we further discovered BM integrity defects in the flight muscles of the notum in Ndg mutant flies (S4C Fig). In addition, while flies appeared to fly normally and negative geotaxis climbing assays did not show differences with the wild type (not shown), Chill-Coma Recovery Time (CCRT) assays (48) showed increased recovery times after cold exposure in flies lacking Ndg, suggesting mild behavioral or motor defects (S4D Fig). In summary, our results show that although Ndg is not critically essential for fly development and assembly of most BMs, it is necessary for the integrity of BMs around specific tissues, such as the larval fat body and the adult flight muscles, and for appropriate fertility and fitness of the fly.

Functional analysis of the different Nidogen domains.
In order to understand the function of Ndg and the basis for its specific requirement in the fat body, we performed a functional analysis of the different Ndg domains. To do that, we generated transgenic flies capable of expressing GFP tagged versions of the wild type Ndg protein as well as a series of mutant variants lacking one or several of these domains ( Fig 5A).
First, we addressed whether these proteins localized normally to BMs. When expressed in the fat body and blood cells under control of the Cg-GAL4 driver, full length Ndg (Ndg FL .GFP) was able to localize to the BMs of imaginal discs, as expected ( Fig 5B). A similar analysis of the localization properties of the deletion constructs showed that no single domain of the protein was capable by itself to confer localization to BMs, suggesting cooperative interactions among domains are required for BM localization (Fig 5B). In addition, analysis of the localization of proteins in which a single domain was deleted (Ndg G1 , Ndg G2 , Ndg G3 and Ndg Rod ) showed that the only domain absolutely required for BM localization was the Rod domain ( Fig 5B).
However, the Rod domain was insufficient to drive protein localization on its own (Ndg Rod ), but required the presence of the G2 or the G3 domains (Ndg G2Rod and Ndg RodG3 , respectively, Fig 5B). We found that knock-down of LanA or Cg25C, encoding a Laminin  chain and Col IV 1, caused a marked reduction in the incorporation of full length Ndg FL .GFP (Fig 6A-C). In addition, while knocking down LanA had not effect on Ndg G3 localization, LanA loss caused a marked reduction in the localization of both Ndg G1 and Ndg G2 (Fig 6A and B). Conversely, knocking down Cg25C resulted in a strong reduction in Ndg G3 localization and significant but not as drastic effects on the localization of Ndg G1 and Ndg G2 (Fig 6A and C). Altogether, these results show that localization directed by the G3 domain depends on laminins, whereas localization by the G1 and G2 domains depends on Col IV.
Next we tested the ability of the Ndg mutant proteins lacking the G1, G2 or G3 domains, all three capable of localizing to BMs, to rescue the fat body BM defects observed in Ndg 1 mutant larvae ( Fig 6D). Overexpression of the mutant variants Ndg G1 or Ndg G2 was able to rescue integrity of the fat body BM (Fig 6D), as imaged with Cg25C.RFP (50) (50). In contrast, expression of the mutant form Ndg G3 failed to rescue BM rupture, indicating that G3 is a key domain for Ndg function, while the G1 and G2 domains may function in a partially redundant way. This is supported by our results showing that Ndg 1 /Ndg Rod-G3.1 transheterozygous mutant larvae show fat body BM defects indistinguishable from those found in Ndg 1 homozygotes ( Fig   4A and B).
The localization and rescue properties of the different domains of Ndg suggest that Ndg may indeed act as a linker between Laminin and Collagen IV, as originally proposed.
Confirming this, simultaneous imaging of Collagen IV and Laminin in fat body BMs shows that in the Ndg 1 mutant Laminin and Collagen IV appear separate from each other when the damaged BM is observed at high magnification ( Fig 6E). In all, these results are consistent with a function of Ndg as a linker of the Col IV and Laminin networks (Fig 6F). This linker function would depend on binding to Laminin through G3 and to Col IV through either G1 or G2.

Role of Laminins, Collagen IV and Perlecan in Nidogen incorporation into BMs.
We have previously shown that Drosophila laminins are critical for proper assembly of other ECM components in the BM of embryonic tissues (35). Furthermore, recent studies have shown that there is a temporal hierarchy of BM components expression in the Drosophila embryo, with laminins being expressed first, followed by Col IV and then Perlecan (32). This seems to be critical for proper formation of the BM around the embryonic VNC. Thus, while elimination of laminins affects both Col IV and Perlecan deposition, laminin incorporates in the absence of any of these two components and Perlecan requires Col IV (32). The requirements of these BM proteins for Ndg incorporation into embryonic BMs are still unknown. Here, we decided to investigate this by analysing Ndg expression in embryos devoid of the other BM components. We found that depletion of LanB1 results in a strong reduction of Ndg accumulation in the gut, muscles and VNC (Fig 7A and B). However, elimination of SPARC, a carrier for Col IV (31) (51), or Perlecan did not prevent Ndg deposition into embryonic BMs ( Fig   7C and D). This is in agreement to what we have previously found in wing imaginal discs where Col IV is not required for Ndg localization (31).
Next, we tested the requirements of laminins, Col IV and Perlecan for Ndg incorporation into the BM of the larval fat body. To this end, we analysed the expression of the transgene Ndg.sGFP in the fat body of larvae where we had knocked down expression of BM components under the control of the Cg-Gal4 driver. We found that the knock down of laminins or Col IV, but not of Perlecan, caused a reduction in the amount of Ndg in fat body BMs (Fig 7 and S5 Fig), consistent with our functional analysis of the different Ndg domains (Fig. 6).
We aditionally decided to analyze the mutual requirements of the remaining components of the adipose tissue BM. We found that loss of Col IV result in a strong reduction in Laminins levels and in a depletion of Perlecan (Fig 7E and S5 Fig). This is in agreement with previous results showing that knocking down vkg with hsp70-gal4, which is a heat shock inducible promoter, reduced the presence of Nidogen and Laminin in fat body BM (52). In contrast, similar to the loss of Ndg (Fig 4), absence of laminins led to rupture of the BM without apparent reduction in collagen or Perlecan levels ( Fig 7B and S5 Fig). Finally, knock down of Perlecan did not affect the presence of any of the other components, consistent with the notion that it is a terminal BM component (Fig 7E) (31).
In summary, these results show that Ndg incorporation into embryonic and fat body BMs depends on both Laminins and Collagen IV. They also suggest a model for the assembly and maintenance of the adipose tissue BM in which Ndg is not not essential for the incorporation of other components, but reinforces the connection between laminins and collagens, thus preventing the rupture of the BM (Fig 7G).

Genetic interactions unmask a wider involvement of Nidogen in BM stability
To finally ascertain whether Nidogen incorporation had a wider stabilizing role on BMs despite limited phenotypic defects in the mutants, we tested genetic interactions with other genetic conditions compromising BM functionality. LanA 216 and LanA 160 are two embryonic lethal mutant alleles of LanA (53). While LanA 216 /LanA 160 3 rd instar larvae showed an elongated ventral nerve cord (VNC), no defects in VNC condensation were observed in LanA 216 /+, LanA 160 /+ or Ndg 1 . In contrast, we found that Ndg 1 mutants heterozygous for LanA 216 /+ or LanA 160 /+ showed VNCs that are significantly more elongated than those found in LanA 216 /+ or LanA 160 /+, respectively (Fig 8A and B). In addition, we found that Ndg interacted genetically with Perlecan.
Thus, while single knock down of either Ndg or Perlecan in the whole fly, using actin-GAL4, produced normal-looking pupae and viable adults, the double knock out of these genes caused a decrease in the size of pupae, which were unable to develop to adulthood (Fig 8). The combined interaction was exacerbated when knock down was driven at 30 o C, a temperature at which GAL4-driven transgene expression is higher (54) (55). In summary, these results prove that Nidogen interacts genetically with Laminins and Perlecan, suggesting a more general role of Nidogen in maintaining BM stability and consistent with its remarkable evolutionary conservation.

Discussion
BMs are thin extracellular matrices that play crucial roles in the development, function and maintenance of many organs and tissues (56). Critical for the assembly and function of BMs is the interaction between their major components, Col IV, laminins, proteoglycans and Ndg (57).
Both the ability of Ndg to bind laminin and Col IV IV networks and the crucial requirements for and epithelial morphogenesis in cultured embryonic lung, submandibular glands and kidney (22,23). However, elimination of Ndg in model organisms has shown that Ndg is not essential for BM formation per se but required for its maintenance in some tissues. Thus, while the early development of heart, lung and kidney, prior to E14, is not affected in nidogen-deficient mice, defects in deposition of ECM components and BM morphology were observed at E18.5 (11). In addition, whereas BM components localized normally underneath the apical ectodermal ridge (AER) in Nidogen-deficient mice at birth, this BM breaks down at later stages (12). In contrast, removal of Ndg does not impair assembly or maintenance of any BM in C.elegans (13). Here, we show that in Drosophila, as it is the case in mammals (11,12) (34) (61). In this context, Laminins and Collagens could self-assemble into networks in the embryo as they are produced, being this sufficient to assemble a BM capable to sustain embryonic development in the absence of the two subsequent components, Ndg and Perlecan. We also show here that, while fat body and blood cells are the source of the majority of the proteins in larval BMs, there are notable exceptions, a fact that highlights a diversity in the origins of BM components in different tissues. Thus, fat body produces entirely all its BM, the larval heart receives it all from the hemolymph, imaginal discs produce a portion of their laminins and similarly for tracheae with respect to Perlecan. These differences in the source of the components for the different tissues (incorporated vs. self-produced) may impose different assembly mechanisms, a possibility to study in more detail in the near future. In addition, although BM components are universally present in numerous tissues and organs, they are diverse depending on tissue and developmental stage (reviewed in (62). This heterogeneity arises from variations in protein subtypes, such as the two alternative Laminin  chains or the numerous Perlecan isoforms. Heterogeneity may also stem from differences in relative amounts of each component and posttranslational modifications thereof. In this respect, it is possible that BM assembly of the Drosophila fat body and adult flight muscles of the notum is such that is more dependent on Ndg function for its formation and stability than BMs found in other tissues.
Finally, dynamics of BMs can orchestrate organ shape changes. Reciprocally, the associated tissues can control properties of BMs by, for instance, expressing a specific repertoire of ECM receptors or remodeling factors. In this context, it is also possible that fat body or adult flight muscles sculpt BMs with properties demanding a high requirement of Ndg function.
We find here that Ndg mutant flies are less fertile and behave differently with respect to wild type in Chill-Coma Recovery Time assays. The physiological mechanisms underlying the response in insects to critical thermal limits remain largely unresolved. The onset and recovery of chill coma have been attributed to defects in neuromuscular function due to depolarization of muscles fibre membrane potential (63). Interestingly, flight muscle fibre membrane is particularly strongly depolarized upon exposure to low temperatures in Drosophila (63). In this context, the defects we observed in the BM of adult flight muscles in the absence of Ndg could be behind the defective response of Ndg mutant flies to chill coma recovery assays. Finally, we have found in a preliminary analysis that Ndg mutant larva show altered immune response to microbial infection (data not shown). All together, these results show that although not critical for survival, Ndg is required for overall fitness of the fly.
All Nidogen proteins consist of three globular domains, G1 to G3, and two connecting segments, one Rod domain separating G2 and G3 and a flexible link between G1 and G2.  (32), suggesting that Collagen IV is not affecting Laminin incorporation in these other tissues to the same degree or that it does not affect it at all. In addition, we found that BM assembly in Drosophila also differs from that in mammals and C. elegans. In this case, the divergences may arise during evolution, when different organisms might have incorporated novel ways to assemble ECM proteins to serve new specialized functions.
Nidogen has been proposed to play a key role on BM assembly based on results from in vitro experiments and on its ability to serve as a bridge between the two most abundant molecules in BMs, laminins and type IV collagens. However, phenotypic analysis of its knock out in mice and C. elegans have called into question a general role for nidogens in BM formation and maintenance. Here, we show that although Ndg is dispensable for BM assembly and preservation in many tissues, it is absolutely required in others. These differences on Ndg requirements stress the need to analyze its function in vivo and in a tissue-specific context. In fact, we believe this should also be the case when analyzing the requirements of the other ECM components for proper BM assembly, as we show here they also differ between species and tissues. One has to be cautious when simplifying functions to the different BM proteins or their domains based on experiments performed in vitro or in a tissue-specific setting. This might be especially relevant when trying to apply the conclusions derived from these studies to our understanding of the pathogenic mechanisms of BM-associated diseases or to the development of innovative therapeutic approaches.

Fly strains
Standard husbandry methods and genetic methodologies were used to evaluate segregation of mutations and transgenes in the progeny of crosses (71). The following stocks were used: The FTG, CTG and TTG balancer chromosomes, carrying twist-Gal4 UAS-2EGFP, were used to identify homozygous Ndg ΔRod-G3 mutants (Halfon et al., 2002). For Collagen IV knock down experiments (vkg i and Cg25C i ), thermosensitive GAL4 repressor GAL80 ts was used to prevent embryonic lethality. Cultures were grown at 18 o C for 6 days, followed by transfer of cultures to 30˚C (L2 stage) and dissection two days later (L3 stage).

Transgenic flies sGFP RNAi
Short hairpin oligoes to knock down sGFP were designed following instructions in DSIR website

Quantification
For quantification of egg laying (Fig. S4A), five 2-day old virgins were transferred to fresh vials daily for ten days and the eggs laid on each vial counted. Three such experiments were conducted per genotype.
For calculation of egg aspect ratio ( Fig. S4B; Frydman and Spradling, 2001), length and width of eggs were measured on images using the line tool in FIJI-ImageJ. Aspect ratio is defined as egg length divided by width).
In chill comma recovery time assays (Fig. S4D; Gilbert et al., 2001), 2-day old females were placed into 10 mL tubes. These tubes were submerged into an ice-water bath for 2 hours, resulting in paralyzed flies. The amount of time required for a fly at room temperature to stand after becoming paralyzed in this way was measured.
For quantification of fluorescence intensity of different Ndg.GFP constructs in fat body BM (Fig.   6B), GFP signal was measured on 4-6 confocal images per genotype using FIJI-ImageJ. Each measurement represents mean value intensity inside a 500 μm 2 square drawn on a flat portion of BM of an individual fat body cell, avoiding measuring intensity in cell contacts.
VNC length was measured on confocal images using the segmented line tool of FIJI-ImageJ.

Statistical analysis
Graphpad Prism software was used for graphic representation and statistical analysis. For statistical comparisons of fluorescence intensity in Fig. 6B Fig. 8A, we performed non-parametric Mann-Whitney tests. For egg production curves in Figure S4A, we conducted non-parametric Kolmogorov-Smirnov tests. For comparison of aspect ratio in Figure   S4B, we performed unpaired two-tailed Student's t tests. For comparison of chill comma recovery time in Figure S4C, Student's t-tests with Welch's correction were used. Significance of statistical tests is reported in graphs as follows: **** (p < 0.0001), *** (p < 0.001), ** (p < 0.01), * (p < 0.05), n.s. (p > 0.05).