Distinct contributions of ECM proteins to basement membrane mechanical properties in Drosophila

The shaping of epithelial tissues during animal development depends on the generation of mechanical force and compliance of cells towards mechanical stress. Compliance towards stress is influenced by cell mechanical properties determined by the actomyosin network structure, Myosin motor protein activity and the underlying basement membrane (Khalilgharibi and Mao, 2021; Molnar and Labouesse, 2021). Basement membranes, specialized extracellular matrices (ECMs) that provide mechanical support to epithelia, promote cell survival, cell polarization and cell differentiation, serve as a substrate for cell migration, and are important for shaping epithelia during animal development (Jayadev and Sherwood, 2017; Sekiguchi and Yamada, 2018; Walma and Yamada, 2020). Basement membranes influence epithelial shape based on their mechanical properties (Khalilgharibi and Mao, 2021). Basement membranes are typically composed of four main components: Laminin, Collagen IV, Nidogen and Perlecan (Pozzi et al., 2017; Sekiguchi and Yamada, 2018). Laminins, a large family of heterotrimeric (α,β,γ) glycoproteins, self-assemble into networks linked via integrins and dystroglycans to the cell plasma membrane (Yamada and Sekiguchi, 2015). Collagen IV has a triple-stranded helical structure forming networks by covalent interactions strengthened through their terminal domains (Wu and Ge, 2019). Laminin and Collagen IV interact with the glycoprotein Nidogen and the proteoglycan Perlecan to form a highly interconnected molecular meshwork (Pozzi et al., 2017; Sekiguchi and Yamada, 2018). The de novo basement membrane assembly is hierarchical, beginning with Laminin, followed by the addition of Collagen IV (Matsubayashi et al., 2017; Pöschl et al., 2004). To what extent the presence of these main components is interdependent during the maintenance of basement membranes and to what extent individual components contribute to mechanical properties of basement membranes is not well understood.

The Drosophila egg chamber is a useful model system to investigate how individual ECM components contribute to mechanical properties of basement membranes (Isabella and Horne-Badovinac, 2015a; Khalilgharibi and Mao, 2021). In contrast to vertebrates, the Drosophila genome comprises a simple set of genes encoding the main basement membrane components: four genes (LanA, wing blister, LanB1, LanB2) encode Laminin subunits, the two α1-chains and one α2-chain forming the Collagen IV triple helical structure are encoded by ColIVα1 (also known as Col4a1) and ColIVα2 (viking), and Nidogen and Perlecan are encoded by single genes, Ndg and Pcan (terribly reduced optic lobes), respectively (Fig. 1A) (Hynes and Zhao, 2000). Egg chambers are composed of germline cells surrounded by a single-cell-layered somatic follicle epithelium, assemble in the germarium and mature through 14 stages to give rise to eggs (Fig. 1B) (Spradling, 1993). During maturation, egg chambers increase volume ∼1000-fold and change shape from spherical to ellipsoid (elongated). Follicle cells migrate as a collective on their underlying basement membrane in a direction perpendicular to the anteroposterior axis of egg chambers. Collective follicle cell migration results in egg chamber rotation and the orientation of basement membrane fibers parallel to the migration direction (Haigo and Bilder, 2011). Oriented basement membrane fibers, in conjunction with parallel oriented intracellular actin stress fibers, build a ‘molecular corset’ that contributes to egg chamber elongation (Bateman et al., 2001; Conder et al., 2007; Frydman and Spradling, 2001; Gutzeit et al., 1991; Haigo and Bilder, 2011; Horne-Badovinac et al., 2012; Lerner et al., 2013; Lewellyn et al., 2013; Viktorinová and Dahmann, 2013; Viktorinová et al., 2009). Atomic force microscopy (AFM) measurements have shown that the stiffness of the basement membrane increases during stages 3-8 and that basement membrane is stiffer in the central region of egg chambers compared with the poles at stage 8 (Chlasta et al., 2017; Crest et al., 2017; Díaz de la Loza et al., 2017). Basement membrane stiffness depends on Collagen IV (Chlasta et al., 2017; Crest et al., 2017). Mutants of Collagen IV fail in egg chamber elongation (Haigo and Bilder, 2011), indicating that basement membrane stiffness is important for reshaping egg chambers.

Here, we report a systematic analysis of interdependencies of the four ECM components Laminin, Nidogen, Perlecan and Collagen IV and their individual contributions to basement membrane mechanical properties. We show that Laminin and Collagen IV networks can persist partially independently of each other. Moreover, Laminin has a minor contribution and Collagen IV has a major contribution to basement membrane mechanical properties, thus revealing distinct contributions of ECM components to basement membrane mechanical properties.

The basement membrane is a highly complex network in which single components interact with multiple binding partners (Naba et al., 2012; Randles et al., 2017). To test whether level and subcellular localization of different basement membrane proteins is interdependent, we used RNA interference to knock down expression of either LanA, LanB1, Ndg, Pcan, ColIVα1 or ColIVα2 in Drosophila follicle cells from stage 3 onwards using the GR1-Gal4 driver line (Tran and Berg, 2003) (Fig. S1A,A′) (Laminin A is the only α-subunit expressed in these cells; Schneider et al., 2006). Protein expression and localization were quantified by staining with antibodies against Laminin, Nidogen, Perlecan or a GFP antibody against ColIVα2::GFP expressed from an exon trap line (Buszczak et al., 2007). Compared with controls (Fig. 1C-F), levels of the targeted proteins were significantly decreased from stage 6 through at least stage 10 (Fig. 1C-G,L,Q,V-Z; Fig. S2A-Z), indicating that the knockdowns were efficient. Protein expression levels and localizations were analyzed in detail at stage 8. Laminin protein level decreased (reduction of staining intensity by ∼40%) in Ndg knockdown (Fig. 1K,W), was unaffected in Pcan knockdown (Fig. 1O,W) and was slightly increased (by ∼20-30%) in ColIVα1 or ColIVα2 knockdowns (Fig. 1S,W), the latter indicating a compensatory increase in Laminin when Collagen IV levels are reduced. Furthermore, Nidogen protein level strongly decreased in Laminin or Collagen IV knockdowns (by ∼91% and ∼75%, respectively) (Fig. 1H,T,X) and only slightly decreased (by ∼26%) in Pcan knockdown (Fig. 1P,X). Moreover, Perlecan protein level was only affected in Collagen IV knockdowns (Fig. 1I,M,U,Y), where it decreased (by ∼47% and ∼61%), consistent with observations in Drosophila embryo (Hollfelder et al., 2014), wing imaginal disc (Pastor-Pareja and Xu, 2011), fat body (Dai et al., 2018) and egg chamber (Haigo and Bilder, 2011; Isabella and Horne-Badovinac, 2015b) basement membranes. Finally, Collagen IVα2-GFP-protein level slightly decreased (by ∼21-31%) in each of the knockdowns (Fig. 1J,N,R,Z). Subcellular localizations of Collagen IV and Nidogen were not detectably altered in knockdowns of other basement membrane proteins (Fig. S3A,C-C″,E-E″,G-G″,I-I″,K-K″,M-M″,O-O″,Q-Q″,S-S″,U-U″). Subcellular localization of Perlecan was also unchanged in Laminin or Ndg knockdowns (Fig. S3D-D″,H-H″,L-L″,P-P″). Interestingly, Perlecan mislocalized to lateral cell membranes, in particular at tricellular vertices, in Collagen IV knockdowns (Fig. S3T-T″). Similarly, Laminin mislocalized to lateral cell membranes in Collagen IV knockdowns (Fig. S3B-B″,F-F″,R-R″’), but also in Pcan knockdown (Fig. S3N-N″). Laminin localization was unaffected in the Ndg knockdown (Fig. S3J-J″).

Our analyses of protein levels and subcellular localizations after knockdown lead us to the following conclusions. First, the presence of Nidogen depends on both Laminin and Collagen IV, consistent with its role as a protein linking the Laminin and Collagen IV networks (Dai et al., 2018, and references therein). A dependency of Nidogen on Laminin has been previously shown in the Drosophila embryo (Hollfelder et al., 2014; Wolfstetter and Holz, 2012). Second, the presence of Laminin, and to a lesser extent of Collagen IV, requires Nidogen, showing a mutual interdependency between Nidogen and Laminin proteins. Third, the Laminin network is maintained largely independently from Collagen IV, consistent with the Laminin network being directly bound by integrins to the basal cell surface (Yamada and Sekiguchi, 2015). However, partial mislocalization of Laminin in Collagen IV or Pcan knockdowns indicates a role for Collagen IV and Perlecan in Laminin trafficking or retention. Fourth, localization and expression of Perlecan solely depend on Collagen IV, but not on Laminin or Nidogen. Fifth, although proper Collagen IV recruitment during the establishment of the basement membrane requires the Laminin network (Huang et al., 2003; Matsubayashi et al., 2017; Smyth et al., 1999; Urbano et al., 2009; Wolfstetter and Holz, 2012), our findings suggest that the maintenance of Collagen IV in the basement membrane does not strongly depend on the Laminin network.

We next sought to analyze individual contributions of Laminin, Nidogen, Perlecan and Collagen IV with respect to mechanical properties of the basement membrane. Egg chamber elongation depends on basement membrane mechanical properties (Chen et al., 2019; Chlasta et al., 2017; Crest et al., 2017; Díaz de la Loza et al., 2017; Haigo and Bilder, 2011), providing a functional assay. We knocked down LanA, Ndg, Pcan or ColIVα2 expression in follicle cells and quantified egg chamber elongation by calculating the length ratio of the long-to-short axis of stage 2-14 egg chambers (aspect ratio, Fig. 2C). Compared with controls (Fig. 2A,B), knockdowns of LanA or Ndg had no obvious effect on egg chamber elongation except for at stage 8 (Fig. 2D-I). At this stage, LanA knockdown resulted in an increased aspect ratio that correlated with an unusually pointed anterior end of the egg chamber (Fig. 2D,F). Decreased LanA activity already induced in the germarium results in hypo-elongated stage 14 egg chambers (Andersen and Horne-Badovinac, 2016; Frydman and Spradling, 2001), Ndg knockdowns had a decreased aspect ratio at stage 8 (Fig. 2I). Knockdown of Pcan, using two independent RNAi lines, resulted in a moderate decrease of egg chamber elongation from stage 8 onwards (Fig. 2J-L) (Isabella and Horne-Badovinac, 2015b). Egg chamber elongation was severely diminished from stage 8 onwards in ColIVα2 knockdown (Fig. 2M-O) (Crest et al., 2017; Haigo and Bilder, 2011; Isabella and Horne-Badovinac, 2015b). Thus, Laminin and Nidogen have only a minor contribution to egg chamber elongation, whereas Collagen IV, and to a lesser extent Perlecan, are the major components of the basement membrane required for egg chamber elongation.

Egg chamber elongation, in part, depends on aligned basement membrane fibers, the orientation of which require egg chamber rotation (Haigo and Bilder, 2011). The velocity of egg chamber rotation at stage 8 was indistinguishable between controls and knockdowns of LanA, Ndg or Pcan (Fig. S4A-D,F), suggesting that the changes in basement membrane composition and mechanical properties observed in these knockdowns cannot be accounted for by a lack of proper egg chamber rotation. Velocity was decreased in ColIVα2 knockdown (Fig. S4E,F) (Haigo and Bilder, 2011). The greatly different contributions of Laminin and Nidogen versus Collagen IV to egg chamber elongation and rotation is consistent with our observation that Collagen IV protein levels are partially independent of Laminin and Nidogen (Fig. 1). Taken together, these results indicate that Laminin, Nidogen, Perlecan and Collagen IV make distinct contributions to basement membrane mechanical properties.

To further test the contribution of Laminin, Nidogen, Perlecan and Collagen IV to basement membrane mechanical properties, we knocked down their expression and analyzed the mechanical resistance of egg chambers to osmotic stress using an established assay (‘organ-swelling assay’) (Crest et al., 2017; Pastor-Pareja and Xu, 2011). Dissected egg chambers were placed in distilled water for 1 h, which led to their swelling and subsequent bursting (Fig. 3A). In general, the bursting of collagen networks is determined by the failure strain of the material, i.e. the amount of deformation at which the network breaks (Fig. 3B) (Roeder et al., 2002; Wang et al., 2002). In controls, ∼56% of egg chambers burst, with the egg chambers bursting on average after ∼12 min (Fig. 3C-E). Knockdown of LanA increased the fraction of burst egg chambers and slightly reduced the time span until the burst (Fig. 3C,D,F). Ndg knockdown behaved indistinguishably from controls (Fig. 3C,D,G). By contrast, all analyzed Pcan or ColIVα2 knockdown egg chambers burst rapidly (Fig. 3C,D,H,I), as previously shown for Collagen IV knockdowns (Crest et al., 2017). Similarly, knockdown of Pcan leads to a rapid bursting of Drosophila wing imaginal discs upon osmotic stress (Pastor-Pareja and Xu, 2011). Thus, Nidogen and Laminin make no or a minor contribution, and Perlecan and Collagen IV make a major contribution, to the mechanical resistance of egg chambers to osmotic stress.

We finally analyzed the contribution of Laminin, Nidogen, Perlecan and Collagen IV to apparent basement membrane stiffness, anticipating a linear relationship between stress and strain (Fig. 4A). Stiffness (apparent Young's modulus) was probed by indentation with the cantilever tip of an atomic force microscope in the central or pole region of dissected stage 8 egg chambers (Fig. 4B,C). In controls, basement membrane stiffness was ∼60 kPa in the central (Fig. 4D,E) and ∼45 kPa in the pole region (Fig. 4F), consistent with previous measurements using indentation depth similar to those used in this work (Chen et al., 2019; Crest et al., 2017). Collagenase treatment of egg chambers decreased basement membrane material (Fig. S5) and strongly reduced basement membrane stiffness (Fig. 4D), demonstrating that the measured stiffness depends on basement membrane (Chlasta et al., 2017; Crest et al., 2017). Knockdown of LanA increased basement membrane stiffness in the central region (Fig. 4E) and decreased it in the pole region (Fig. 4F). This increased stiffness anisotropy (Fig. 4G) correlated with hyper-elongation of LanA knockdown egg chambers at stage 8 (Fig. 2D,F), consistent with the notion that stiffness anisotropy drives egg chamber elongation (Crest et al., 2017). Ndg knockdown resulted in a slightly decreased basement membrane stiffness in the central region (Fig. 4E), but did not affect the pole region (Fig. 4F). Knockdown of Pcan did not significantly alter basement membrane stiffness in either region (Fig. 4E,F) [we note that Chlasta et al. (2017), using a different AFM setup, reported a decreased basement membrane stiffness in Pcan knockdowns]. Finally, knockdown of ColIVα2 strongly decreased basement membrane stiffness in both regions (Fig. 4E,F) (Chlasta et al., 2017; Crest et al., 2017). Thus, Collagen IV makes a major contribution to basement membrane stiffness in egg chambers.

In summary, we have used three different assays to investigate the contribution of individual ECM components to the mechanical properties of the basement membrane (Fig. 4H). Perlecan is important for egg chamber elongation and resistance to osmotic stress, but dispensable for basement membrane stiffness. Nidogen plays a minor role during egg chamber elongation and basement membrane stiffness, and is dispensable for the resistance to osmotic stress. Laminin plays a minor role during egg chamber elongation and resistance to osmotic stress and suppresses the stiffness anisotropy of the basement membrane. The differential requirements for Perlecan, Nidogen and Laminin indicate that the three assays reflect different functional aspects of the basement membrane in mechanically supporting the egg chamber. The organ-swelling assay interrogates in particular the mechanical properties of egg chambers at their poles, as suggested by frequent bursting at that location (Crest et al., 2017). Bursting frequency, however, does not correlate with basement membrane stiffness in pole regions. Pcan knockdown, for example, showed increased bursting frequency (Fig. 3C), yet basement membrane stiffness in the pole region was comparable with controls (Fig. 4F). The organ-swelling assay thus reflects the failure strain of the egg chamber material (see Fig. 3B) rather than basement membrane stiffness in the pole region. This is not surprising, as the failure strain and the stiffness often do not correlate positively (Bax et al., 2019; Haut, 1986; Leng et al., 2013). On the other hand, egg chamber elongation relies on the establishment of a stiffness gradient between the poles and the center of egg chambers (Chlasta et al., 2017; Crest et al., 2017). Finally, Collagen IV makes a major contribution to egg chamber elongation, resistance to osmotic stress and basement membrane stiffness. The greater contribution of the network-forming Collagen IV compared with Laminin towards conferring mechanical properties may reflect the ability of Collagen IV molecules to better resist mechanical stresses due to their covalent bonds, whereas laminins self-assemble only through weaker ionic interactions (Wu and Ge, 2019; Yamada and Sekiguchi, 2015).

Previous work has mainly analyzed mechanical properties of fibrillar collagens (including Collagen I) and other ECM components present, e.g. in connective tissues, ligaments or bones (reviewed by Sherman et al., 2015) and has shown that Collagen I, in contrast to laminins, bears tensile stresses. Similarly, our work suggests that in the basement membrane of Drosophila egg chambers Collagen IV mainly bears tensile stresses. The ability of Collagen I fibrils and higher order fibers to bear mechanical stress depends on multiple parameters, including fiber shape and orientation, mineralization and density of covalent cross-links (Sherman et al., 2015). Oriented fibril-like Collagen IV structures contribute to basement membrane stiffening of egg chambers (Chlasta et al., 2017), indicating that shape and orientation of fibril-like structures also determine mechanical properties of Collagen IV networks. Moreover, removal of bromide, an ion required for sulfilimine crosslinks of Collagen IV, from fly food results in hypo-elongated egg chambers (McCall et al., 2014), indicating that molecular cross-links further contribute to the ability of Collagen IV networks to bear mechanical stresses.

The fly stocks used were GR1-Gal4 [Bloomington Drosophila Stock Center (BDSC), 36287], UAS-mCD8::RFP (BDSC, 32219), UAS-LanA^dsRNA (BDSC, 28071), UAS-LanB1^dsRNA (BDSC, 42616), UAS-Ndg^dsRNA (BDSC, 62902), UAS-trol^dsRNA (BDSC, 42783), UAS-trol^dsRNA [Vienna Drosophila Resource Center (VDRC), 110494], UAS-Col4a1^dsRNA (VDRC, 28369), UAS-vkg^dsRNA (BDSC, 50895), vkg::GFP^G00454 [Drosophila Genetic Resource Center (DGRC), 110692] and LanA::GFP (VDRC, 318155). Flies were kept at 25°C on standard food. Crosses were raised at 18°C and virgin flies were transferred to 29°C for 2 days before dissection.

Ovaries from 2-day-old flies were dissected in PBS and fixed with 4% paraformaldehyde and 0.1% Triton X-100. The following primary antibodies were used: anti-Laminin (Gutzeit et al., 1991; 1:1000), anti-Nidogen (Wolfstetter et al., 2009; 1:500), anti-Perlecan (Friedrich et al., 2000; 1:1000), anti-GFP (Clontech, 632592, 1:2000) and anti-RFP (Chromotek, 5F8, 1:1000). The secondary antibodies used were goat anti-rabbit Alexa 488 (Molecular Probes, A27034, 1:200) and goat anti-rat Cy3 (Jackson ImmunoResearch, 112-165-167, 1:200). Rhodamin-Phalloidin (Thermo Fisher Scientific, R415, 1:200) was used to visualize F-actin. Images were acquired using a laser scanning microscope Zeiss LSM 880 or Zeiss LSM 980, or a microscope Zeiss Axio Observer Z1. Note that slight convolutions of the egg chamber surface (Fig. 1C-V) may be due to the long incubation time of fixation and permeabilization during the antibody staining.

Image stacks covering the entire apical-basal extent of follicle cells were acquired with the same laser intensity settings and were projected using the maximum projection method. Mean fluorescence intensity of z-projections was measured in corresponding regions of egg chambers using Fiji (Schindelin et al., 2012). To calculate the corrected mean fluorescence intensity, the mean fluorescence intensity value was reduced by the mean fluorescence intensity value of five egg chambers treated with secondary antibody only. For Fig. S3, image stacks were acquired in the central region of egg chambers and x-z images were then obtained by maximum projection of the orthogonal view using Fiji (Schindelin et al., 2012). For measurements of GFP fluorescence intensity during collagenase treatment, images were acquired every 2 min for 1 h.

Egg chamber stages were determined as previously described (Jia et al., 2016; Spradling, 1993). Length and width of egg chambers were measured using Fiji (Schindelin et al., 2012).

The organ-swelling assay was performed as previously described (Crest et al., 2017) with the following exception: distilled water was set to a conductivity of 100 µS/cm³.

Dissection of ovaries and measurements were performed in Schneider medium with supplements (Prasad et al., 2007). Basement membrane stiffness was measured using an atomic force microscope (Nanowizard I, JPK Instruments) mounted on a Zeiss Axiovert 200M wide-field microscope with a 20× objective (Zeiss, Plan Apochromat, NA=0.8) and a CCD camera (DMK 23U445, The Imaging Source). Standard pyramidal tips (MLCT-C, Bruker AFM probes) were used in all experiments. The nominal tip radius given by the manufacturer is 20 nm with a nominal spring constant of 0.01 N/m. Before each experiment, the actual cantilever spring constant was measured by means of thermal noise analysis (built-in software, JPK Instruments). Actual spring constant values ranged from 0.015-0.019 N/m. All measurements were performed in culture medium at room temperature.

For measurements in the central region, dissected egg chambers were placed into cell culture dishes (FluoroDish FD35-100) coated with poly-D-lysine. For measurements in the pole region, polydimethylsiloxane (PDMS) blocks with molds of diameter 120 µm and 140 µm and a depth of 100 µm were made [Center for Molecular and Cellular Bioengineering (CMCB) Microstructure Facility of the Technische Universität Dresden]. The PDMS block was mounted inside a cell culture dish and adhered to the glass bottom through gentle mechanical pressure. Then, the mounted PDMS block was plasma activated and coated with poly-D-lysine. Stage 8 egg chambers were manually isolated and transferred into the coated PDMS block (posterior to the top). Quick removal and addition of culture medium allowed adhesion of the egg chamber to the coated PDMS block.

AFM measurements were carried out in contact mode with the cantilever oriented approximately orthogonal to the egg chamber long axis (Fig. 4B,C). Each measurement consisted of 64 force curves acquired on a regular square lattice of 10×10 μm² placed in the central region or in the pole region of the egg chamber surface (Fig. 4B,C). Indentations were performed at an extension speed of 0.5 μm/s up to a maximal force of 0.4 nN. For each condition, at least five egg chambers were measured.

In order to extract basement membrane Young's moduli, force-indentation curves were fitted to the Hertz-Sneddon model (Sneddon, 1965) for a tip with a quadratic pyramid geometry and a half-angle to edge of 17.5°, assuming a Poisson ratio of 0.5. Fitting was performed using only the first 50 nm of indentation as reported earlier (Crest et al., 2017).

Egg chambers were dissected in egg chamber culture medium (Prasad et al., 2007) and adhered to glass bottom microwell dishes (MatTek) coated with poly-D-lysine. Egg chambers were imaged every minute for 20 min. GR1-Gal4 driven expression of mCD8::RFP was used to visualize cell membranes. The distance of tricellular junction displacement after 20 min was measured by Fiji (Schindelin et al., 2012) and used to calculate the velocity of egg chamber rotation.

Egg chambers were dissected in Schneider medium with supplements (Prasad et al., 2007). Dissected egg chambers were adhered to glass bottom microwell dishes (MatTek) coated with poly-D-lysine and were either treated with or without (control) 0.05% Collagenase I (Merck).

Statistical significance was calculated using the Welsh two-sided t-test or Chi-square test using R. To determine error propagation of the index of anisotropy, we used the standard formula for a ratio, i.e., if ⁠, the propagation of the standard error of the mean is ⁠. Where and s_y are the standard error of the mean of the function, variable x and variable y, respectively.

We thank Stefan Baumgartner, Anne Holz, and Christoph von Bredow for providing antibodies, Klaus Reinhardt for access to the Zeiss LSM 880 microscope as well as the CMCB Light Microscopy Facility and the CMCB Microstructure Facility of the Technische Universität Dresden for their support. Stocks obtained from the BDSC (NIH P40OD018537), the VDRC and the DGRC were used in this study.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200456.