Basement Membrane Structure and Function: Relating Biology to Mechanics

Abstract

Basement membranes are key mediators of many biological processes such as epithelial morphogenesis, kidney filtration, and muscle function among others. Basement membranes provide structural support to tissues so understanding their mechanical properties is important for determining how they contribute to tissue form and function. Further, basement membranes are altered in many diseases including cancer, diabetes, and fibrosis, and these changes may contribute to disease pathogenesis and progression. Understanding how basement membrane mechanics integrate with tissue function is the work of both biologists and engineers/material scientists, yet these disciplines have very different foundations. This review discusses basement membrane macromolecular structure with a view to illuminate how this structure confers basement membranes with unique mechanical properties adapted to resisting physiological stresses. The pathological implications of altered basement membrane mechanics are discussed in the context of different diseases. Additionally, we survey methods used to measure basement membrane mechanical properties, including atomic force microscopy, tensile stiffness assays, and non-quantitative assays such as cell bursting, assessing their strengths and limitations and their accessibility for different types of in vivo studies. We focus on explaining and illuminating the complexities of basement membrane material properties for biologists, and explaining the biological aspects for engineers, with the goal of making interdisciplinary science more accessible to experimentalists and readers.

Introduction

Basement membranes are ubiquitous sheet-like structures that lie beneath all epithelial cell sheets and wrap around all epithelial and endothelial tubes. Basement membranes are found in multiple organs and tissues including the skin, digestive system, respiratory tract, circulatory and lymphatic systems, as well as muscle, fat, Schwann cells, and more. Some specific examples in mammals include the glomerular basement membrane (GBM) which is a critical component of the kidney filtration barrier, the tubular basement membrane that support the kidney tubular epithelium, the lens basement membrane (LBM) that surround the eye lens and supports the lens epithelium, and Reichert’s basement membrane that surrounds the embryo during development. Often considered as stable structural supports, basement membranes are actually dynamic and multifunctional. Their many functions include separating different compartments within and between tissues, regulating cell polarity and migration, driving tissue morphology during development and repair, and serving as biochemical signaling hubs [1-3]. Though diverse, these biological functions are related to their mechanical function of providing structural support for adherent cells.

Given their structural role in maintaining tissue form and resisting normal and pathological stresses, the mechanical properties of basement membranes are important to understand. Extracellular matrices including basement membranes do not behave according to traditional principles of solid mechanics [4, 5]. Rather, they exhibit a number of biophysical properties that are mediated by their unique nanoscale macromolecular architecture and allow them to respond to physiological stress and maintain tissue form and function. In the setting of disease, these properties may be compromised and directly contribute to disease pathogenesis or progression. Several reviews have provided a comprehensive overview of basement membrane composition in different organs and tissues in health and disease [3, 6, 7]. The goal of this review is to interface between the fields of engineering and biology: to provide the basement membrane biologist with an overview of the nuanced biomechanical behavior of basement membranes, and to provide engineers and biophysicists with a primer on the underlying biology that is relevant to basement membrane mechanics.

I. Basement membrane assembly and organization.

Basement membranes were first observed by light microscopy and then by electron microscopy (EM), where they appear as thin dense sheets underlying epithelia, surrounding muscle, and separating tissue layers [8, 9]. They are made chiefly of collagen IV, laminin, nidogen, and heparan sulfate proteoglycans (HSPGs) such as perlecan or agrin. In the 1980s biochemists made important discoveries about their structure and assembly: in vitro, collagen IV and laminin each self-assemble into sheet-like structures [10, 11] (Fig. 1A). This suggested a simple model for the extracellular assembly of basement membranes, namely, that cells secrete the components, and in the extracellular environment these self-assemble into sheets. In embryos of mice and flies, laminin assembles first, providing a platform for subsequent collagen IV assembly. This model is supported by genetic experiments: when laminin was knocked out in embryos, collagen IV was not able to assemble appropriately in basement membranes [12, 13]; however, when collagen IV was knocked out, laminin was still able to assemble in the right place and time on tissue surfaces [14, 15] (Fig. 1B and Table 1). More recently, imaging confirms that, in vivo, basement membranes contain distinct layers of collagen IV and laminin [16, 17] (Fig. 1C). Interestingly, in older animals and in some human pathologies, basement membranes may gain additional layers visible by EM [18-20]. The thickness of basement membrane varies significantly depending on the specific tissue. Capillary basement membranes are approximately 100-200 nm [21, 22], tubular basement membranes are 200-300 nm [23, 24], glomerular basement membrane in healthy human kidney is 300-400 nm [25, 26]. Examples of thicker basement membranes can be found in the eye, with Descemet’s (cornea) and lens basement membranes (10-20 μm) [27, 28]. However, thickness can vary with age and disease, and can depend on preparation and characterization methods.

Figure 1. Basement membrane assembly.

(A) In vitro, laminin and collagen IV proteins can each self-assemble. Based on [10, 11].

(B) Analysis of collagen IV and laminin mutant mouse embryos demonstrates that laminin incorporation is required for collagen IV to assemble into basement membranes. Based on [12, 14].

(C) Laminin and collagen IV can be visualized in distinct layers in adult basement membranes. Based on [16, 17].

Table 1:

Summary of Basement Membrane Mouse Knockout Mutants Discussed

Gene(s) Mutated	Phenotype	Reference
LamC1	embryonic lethal, basement membranes absent	[12]
Col4a1 and Col4a2	embryonic lethal, basement membranes assemble but are structurally deficient	[14]
Nidogen1 and Nidogen 2	perinatal lethal, basement membrane defects	[29]
Netrin4	viable, basement membranes more sFff	[30]
Peroxidasin	semi-lethal, basement membranes less sFff	[23, 31]

Surprising work in Drosophila in the last two decades has demonstrated that the components of basement membranes can be supplied to a tissue non-autonomously, that is, the component proteins can be secreted from distant tissues to be assembled locally [32-35] dependent on integrins in the adjacent cells [32]. Again, these findings support the idea that basement membranes can self-assemble given the right conditions, although a great deal more remains to be determined about how assembly is regulated in a common extracellular bath of protein components. In contrast, the first basement membranes assembled in fly embryos do not extract components from this extracellular bath, but rather have components secreted locally by migratory blood cells that deliver them to the developing epithelium [15, 36]. This result confirms that initial assembly in vivo is more complex than self-assembly in vitro.

II. Basement membrane core components.

Collagen IV.

The mechanical properties of basement membranes are determined by many factors, but chief among them is the structural protein collagen IV. Several genes encode collagen IV proteins, which are translated and then assembled in the cell into heterotrimeric “protomers”, the building blocks of the collagen IV lattice. Each protomer contains an extended triple-helical collagenous domain, a 7S domain at the N-terminus, and a non-collagenous (NC1) domain at the C-terminus. Unlike the fibrillar collagens, the collagen IV triple helical domain has over a dozen breaks in the Gly-X-Y triple helix, conferring flexibility to the triple helix, which likely gives basement membranes their high elasticity and resistance to tensile stress and hoop stress, which is also referred to as circumferential stress [37, 38] (defined in Box 1 ). The larger collagen IV sheet-like network assembles extracellularly in the basement membrane through non-covalent interactions of four 7S domains at the N-termini [39] and two NC1 domains at the C-termini (Fig. 1A) [40]. After assembly, collagen IV becomes variably crosslinked, 7S to 7S and NC1 to NC1, to form a covalent network in which the degree of crosslinking partially determines the mechanical properties of the basement membrane. Consistent with its crosslinking, collagen IV in adult tissues can have a half-life of months, as determined by mass spectrometry studies [41-43], and it has little mobility within the basement membrane, as assayed by FRAP in C. elegans larvae [44]. In contrast, collagen IV is dramatically more dynamic during development, with half-life measured in hours [45].

The evidence that collagen IV is critical for basement membrane mechanics comes from the mouse mutant, which can assemble basement membranes without collagen IV during embryogenesis, but these have structural deficiencies and fail when they are subject to mechanical demands [14]. Further, experiments in Drosophila egg chambers demonstrate that collagen IV is critical for generating basement membrane stiffness, as directly measured by AFM, and this stiffness is important for determining egg chamber shape and resistance to swelling-induced failure [46-48]. Basement membrane mechanics were altered more by the loss of collagen IV than by the loss other structural components such as laminin, nidogen, and perlecan [48]. Changes in collagen IV structure and mechanics have been implicated in the pathologies of cancer and dementia [49-51].

In vertebrates, there are three distinct protomers of collagen IV, made from combinations of six different genetically encoded proteins, collagen IV α1- α6. These form protomers of α1/α1/α2, α3/α4/α5, and α5/α5/α6, which assemble into distinct collagen IV scaffolds found in different basement membranes at different developmental stages (Fig. 2A). Collagen IV α1/α1/α2 is the most ubiquitous of the collagen IV scaffolds, found widely in basement membranes throughout the body and across evolution [52]. The collagen IV α3/α4/α5 scaffold is found in specific vertebrate tissues including the glomerular basement membrane (GBM; the filtration unit of the kidney) [53-55], where the α3/α4/α5 scaffold is estimated to comprise 70% of the total protein [56]. α3/α4/α5 forms a layer distinct from the α1/α1/α2 layer, visible with near-super resolution microscopy (120 nm) [57] (Fig. 1C). The α3/α4/α5 scaffold is more compact than the other collagen IV scaffolds, with looping and supercoiling of the lattice stemming from increased interchain disulfide bonding in the α3/α4/α5 scaffold, as α3 and α4 have many more cysteine residues [56, 58]. The restrictions on scaffold mobility imposed by the increased disulfide covalent bonding likely explain its compact nature and may provide both enhanced molecular selectivity of the GBM and increased mechanical stability to resist high filtration pressure associated with flow. However, direct comparison of the mechanical properties of the collagen IV α3/α4/α5-rich basement membranes to those enriched for other collagen IV isoforms are lacking.

Figure 2. Structure of the collagen IV network.

(A) Three different types of collagen IV protomers are found in vertebrates, consisting of α1/α1/α2, α3/α4/α5, and α5/α5/α6. Protomers assemble into networks via head-to-head interactions of two C-terminal NC1 domains, and the assembly of four N-terminal 7S domains to form a 7S tetramer. The NC1 domains are crosslinked through formation of sulfilimine bonds (orange) catalyzed by peroxidasin, while 7S domains are crosslinked (gray) by disulfide bonds and lysine-lysine bonds catalyzed by lysyl oxidase 2.

(B) The interruptions in the Gly-X-Y repeat of collagen IV impart flexibility to the network when no load is applied.

(C) Under increasing tensile load, the slack is removed from the structure, allowing the network to stiffen with increasing stress.

Like other collagens, the collagen IV protomer has an extended triple helical domain that confers tensile strength to the network. In fibers made from fibrillar collagens (collagen I), many triple helices are bound together and aligned to confer additional strength to load-bearing tissues such as tendons and can be mineralized in bone to provide compressive strength. However, collagen IV does not form fibers, but rather branched scaffolds that have significant flexibility introduced by the large number of Gly-X-Y interruptions in the triple helix [38, 56, 58]. This is illustrated in Fig. 2B where the unstressed basement membrane shows flexibility and bending of the collagen IV network. Upon application of an external tensile stress as indicated by arrows in Fig. 2C, the slack in the network has to be removed before the collagen IV network can maximally resist deformation in response to additional applied stress. This, in part, provides basement membranes with their non-linear elasticity, discussed in detail below. This schematic representation is somewhat oversimplified as this illustration does not consider intermolecular interactions between different collagen IV chains or intermolecular disulfide crosslinks that would provide additional resistance to tensile or hoop stress [59].

Collagen IV crosslinking proceeds by different mechanisms at the N- and C-termini. At the N-terminus, a protomer’s triple-helical 7S domain is assembled non-covalently into a tight rod with the 7S domains from three other protomers. This 7S tetramer, comprising the 7S domains of 12 polypeptides, is crosslinked at cysteine residues by interchain disulfide bonds and at lysine residues by Lysyl Oxidase 2 (Loxl2) [60, 61]. At the C-terminal end of each protomer, the trimeric globular NC1 domain can be crosslinked to the NC1 domain of another protomer in a biochemical process that is well understood. After non-covalent assembly of the collagen IV network mediated by extracellular chloride, the two NC1 domains abut each other, flat face to flat face [40, 62]. Within each constituent monomer of the NC1 domain, an exposed lysine on one face can be covalently bound to a methionine on the opposing face by a sulfilamine bond (S=N) [63], a bond that appears to be unique for NC1 crosslinking but is highly conserved throughout animal collagen IV [64]. The sulfilimine bond is catalyzed extracellularly by the enzyme peroxidasin, which resides in the basement membrane [65] and is also highly conserved across metazoans [66]. Each of the three pairs of opposing peptide chains can have zero, one, or two crosslinks, distinguished experimentally by their mobility during gel electrophoresis, implying that each NC1-NC1 interaction surface can in principle have from zero to six crosslinks. The variable number of crosslinks provides a potential dynamic range for the resulting mechanical properties of basement membranes. Actual crosslink occupancy appears to be tissue specific, with average values between 2-4 per hexamer reported [67]. Peroxidasin-based crosslinking is an important contributor to basement membrane stiffness, as measured in a tensile strain assay of either mouse or Drosophila tissues lacking peroxidasin [23, 31]. Although it is intriguing to speculate that the crosslink occupancy may respond to environmental stresses on the basement membrane, there is as of yet no data to support this hypothesis. For both NC1 and 7S crosslinking, the extent of crosslinking and its regulation are not well understood. Nevertheless, collagen IV crosslinking is an important determinant of basement membrane stiffness [23]. Treatment of endothelial cells with high glucose or methylglyoxal to model diabetes leads to increased expression of lysyl oxidases in endothelial cells and increased subendothelial matrix stiffness. This process can be mitigated by treatment with β-aminopropionitrile (BAPN), a pan-LOX inhibitor [68, 69]. These studies highlight the importance of collagen IV crosslinking in determining the mechanical properties of basement membranes and show that multiple physiological and pathophysiological crosslinking mechanisms can play a role in determining the properties of a given basement membrane.

Laminin.

Laminins are heterotrimeric molecules assembled from an α, β, and γ chain. Before secretion from cells, trimers are assembled intracellularly such that the C-termini of the three chains form a coiled-coil rod, the long arm, which terminates in a globular domain composed of the α-chain C-terminal residues. In contrast, the three N-termini separate and reach outward like a “three-spoked umbrella” [70]. The globular domain at the end of the long arm binds to receptors of the cell surface, concentrating and positioning the spoke-like short arms to interact and polymerize with one another at the cell surface such that one α, one β, and one γ short-arm bind in a stereotyped fashion to form a triskelion-like node [71], whose structure was recently determined by cryoEM [72]. Because the triskelion is flat, the surface of the laminin sheet is expected to be flat and parallel to the cell surface [71]. Assembly of the α chain into the node is calcium dependent, and importantly, calcium is ~10,000 times more concentrated outside cells than inside cells, restricting laminin polymerization to the extracellular environment. The cell surface receptors that laminin binds to include proteins like integrins and syndecans, as well as HSPGs and sulfated lipids (sulfotides). In mammals, there are at least 15 laminin trimer combinations, composed from the five α, four β, and three γ chains encoded in the genome. In both Drosophila and C. elegans, only two trimer combinations exist, carrying one of two distinct α chains. Laminins seem to have widely varying turnover rates in basement membranes: in the adult mouse lung or glomerulus, the half-life of laminins can be months [41, 73] whereas in the adult mouse mammary gland – a tissue with constant growth and regression – the half-life of laminin B1 is estimated as 4-8 hours and even less in basement membranes around tumors [74]. This reduced stability is consistent with laminin in developing tissues of many model organisms [45]. Even in developing tissues, however, laminin does not exhibit mobility within the plane of the basement membrane [44]. Studies in Drosophila larval epidermis show that laminin does not need to be supplied to a basement membrane from local cells, but can instead be supplied from a distant organ [33]. Laminin networks may also contribute to basement membrane mechanics, although it is unclear if this is a direct effect or indirect effect through recruitment of the collagen IV network. Existing laminin networks can be disrupted by the laminin-like molecule Netrin-4 [75], and viable Netrin4 knockout mice have stiffer basement membranes than controls [30]. Similarly, reduction of laminin A in Drosophila egg chambers causes increased susceptibility to failure induced by osmotic swelling, but the loss of laminin has a less dramatic effect than does the loss of collagen IV or perlecan [48]. Loss of laminin had divergent effects on basement membrane stiffness in the central and pole regions of the Drosophila egg chamber measured by AFM: the center region showed an increase in basement membrane stiffness while the pole region had reduced stiffness [48].

Heparan Sulfate Proteoglycans (Perlecan/Agrin).

Heparan sulfate proteoglycans (HSPGs) have a protein core modified post-translationally with a substantial mass of sulfated carbohydrate. The sulfation endows HSPGs with extensive negative charges, causing them to become hydrated and act as a hydrogel, presumably resistant to compression. Assembly of the HSPG perlecan into basement membranes depends on prior assembly of collagen IV, as seen in Drosophila development and in repairing basement membranes [15, 32, 33]. Further, perlecan appears to augment collagen IV’s mechanical functions: tissues lacking perlecan are overly constricted in a collagenase reversible manner, whereas tissues overexpressing perlecan are overly expanded [32]. Like other basement membrane proteins, HSPGs can be supplied from distant sources in Drosophila and C. elegans [32, 44], but compared to collagen IV and laminin, HSPGs appear to have much shorter half-lives, measured in hours in the glomerulus and minutes in developing C. elegans [44, 76]. Interestingly, photobleaching experiments demonstrated that the HSPG agrin diffuses within the plane of the basement membrane [44]. HSPGs contribute to basement membrane mechanics by regulating the degree of hydration. Removal of highly charged proteoglycan side chains through enzymatic treatment with heparinase or chondroitinase reduced basement membrane thickness and increased stiffness due to removal of water [77, 78]. The role of hydration is further illustrated in measurements of basement membrane thickness by electron microscopy, which requires dehydration, compared to AFM, which can be performed in the hydrated state. Thickness measured by EM gives significantly smaller values than measurements by AFM, showing the importance of hydration state on morphological properties as well as response to compressive loading [77-79].

Nidogen.

Also known as entactin, nidogen is a monomeric glycoprotein that can bind both laminin and collagen IV, positioning it as a potential linker between these two distinct layers of the basement membrane. The first nidogen mutants, made in C. elegans, were viable with a mild phenotype, a result that challenged nidogen’s essential role in linking basement membrane layers [80]. Although just one nidogen gene exists in C. elegans and Drosophila, there are two nidogen genes in mammals: in mouse, the nidogen double mutant dies after birth with specific tissue defects in basement membranes [29]. Interestingly, the fly nidogen mutant is homozygous viable but shows basement membrane defects in specific tissues – flight muscles and adipose tissues – which may represent a requirement for nidogen in basement membranes around cells that are not apically-basally polarized [81]. In Drosophila embryos and larval epidermis, nidogen localization to the basement membrane depends on laminin, and nidogen can be supplied from muscles to local basement membranes or from a distant organ [33, 81]. Like other non-polymerized basement membrane components, nidogen has a short half-life and can diffuse within the plane of the basement membrane [44].

Other basement membrane components.

Identification and analysis of basement membrane components was facilitated in the 1980s by the use of mouse Engelbrecht Holm-Swarm (EHS) tumors. These tumor cells secrete large amounts of basement membrane components and are the source of Matrigel [82]; indeed, although Matrigel does not have the architecture, crosslinking, or mechanical properties of native basement membranes, its main components -- collagen IV, laminin, perlecan, nidogen – have come to define the core basement membrane proteins [83]. In addition to these four components, however, recent studies have identified hundreds more proteins in various basement membranes, confirming that basement membranes are extremely complex and consist of many additional structural, regulatory, and signaling molecules [6, 84]. Identification methods include in silico prediction, transcriptomics, and proteomics – but definitive identification depends on localizing a protein to an intact basement membrane. For example, unbiased transcriptomics analysis of over 60,000 mouse cDNAs identified 16 ECM proteins, 7 of which were localized to a basement membrane [85].

Recently, a near-complete set of 160 human basement membrane proteins was reported by Lennon, Sherwood, and colleagues. This comprehensive study mined existing literature for localization of predicted mammalian BM proteins, then augmented those studies with new protein localization in C. elegans complemented by cross-species homology [52]. This 160-member network was then used to probe proteomic studies of different tissues to determine which basement membrane proteins are common and which were tissue specific. Fortunately, the previously identified four core proteins -- collagen IV, laminin, perlecan, nidogen – were indeed abundant across tissues, but a fifth was added to their ranks, collagen VI (α1/α2/α3). Other basement membrane proteins were highly variable across tissues. Transcriptomics-based expression analysis found segregation of a distinct set of basement membrane proteins in tissues with mucosal linings, confirming the expectation that specific basement membrane functions are dictated by specific protein composition. To identify conserved regulators of basement membrane composition, they undertook genetic studies in C. elegans and zebrafish, identifying 19 genetic regulators of basement membrane composition, including the conserved protein families TGF-β, ADAMTS, and ROBO/SLIT. Thus, there is still a lot to learn about the diversity and complexity of basement membrane composition.

Other important basement membrane components include signaling molecules. For example, basement membranes serve as repositories for TGF-β and other growth factors. TGF-β specifically requires mechanical force to be liberated from the latent TGF-β complex into its signaling-active form, which has many functions including a significant role in fibrosis [86-88]. Interestingly, the absence of TGF-β signaling increases collagen levels as found by Lennon, Sherwood, and colleagues [52]. Further, application of mechanical force upregulates TGF-β production in multiple cell types [89-92]. Together, these studies suggest that TGF-β mediates a complicated relationship between mechanical force and collagen levels.

III. Basement membranes in disease.

Genetic diseases of the basement membrane.

Mutations in basement membrane genes cause a number of human diseases, the majority of which are associated with collagen IV and laminin [93-96]. Mutations in α1 or α2 collagen IV cause Gould Syndrome, which is clinically heterogenous and can include cerebrovascular hemorrhage, porencephaly, and ocular, renal and neuromuscular pathologies [97-101]. Genetic mutations in the α345 collagen IV network cause Alport Syndrome, characterized primarily by kidney disease as well as hearing and vision loss [102]. Laminin mutations result in various diseases including Pierson Syndrome (LAMB2), congenital muscular dystrophy (LAMA2) and Poretti-Boltshauser (LAMA1) syndrome [93, 95]. Given the ubiquity of basement membranes, these conditions affect many organs and tissues including the brain, muscle, kidneys, eyes, skin, and heart. The specific organs affected by basement membrane mutations reflect the tissue-specific localization, isoforms, or functions of various basement membrane components. Many of these conditions are spectrum disorders with varying severity depending on the nature of the mutation. How these mutations alter the mechanical properties of tissues is not fully understood. In Alport mice (Col4a3^−/−) glomerular stiffness decreases early in disease, possibly reflecting changes in both GBM and glomerular cell mechanics [103], and stiffness increases later in disease when fibrosis and glomerular sclerosis are present [104]. Fig. 3A,B show electron microscopy images of the normal and Alport (Col4a3^−/−) mouse GBM, with the mutant GBM showing low electron density and significant splitting (Fig. 3B).

Figure 3. Transmission electron microscopy images of normal and diseased glomerular basement membranes in Alport Syndrome and diabetes.

(A) High magnification image of the normal glomerular filtration barrier (scale bar=100 nm).

(B) Image of the filtration barrier in the Alport mouse showing splitting and delamination of the glomerular basement membrane, reprinted with permission from Funk et al, Alport syndrome and Pierson syndrome: diseases of the glomerular basement membrane, Matrix Biol. 71 (2018) [95].

(C) Low magnification image of the normal glomerular filtration barrier.

(D) Diabetic glomerular filtration barrier showing thickening of the glomerular basement membrane. C and D reprinted with permission from Marshall, Rethinking glomerular basement membrane thickening in diabetic nephropathy: adaptive or pathogenic? Am. J. Physiol. Renal Physiol. 311 (2016) [112].

Diabetes.

Basement membranes can be altered in response to disease, including chronic conditions such as diabetes. Diabetic basement membranes become thicker in both microvascular basement membranes such as the kidney glomerulus, eye, and skeletal muscle [22, 105-107] and in non-vascular basement membranes such as the kidney tubules, retina, and nerves [108, 109]. Fig. 3C,D show electron microscopy of the normal and thickened GBM in diabetic kidney disease. Indeed, basement membrane thickening is a common histological marker of diabetes and correlates with the degree of tissue damage [110, 111], but the biological mechanisms of thickening and the functional consequences of altered basement membrane structure are not well characterized [112]. In the human eye, diabetic basement membrane thickening is accompanied by increases in stiffness, as measured by AFM, and changes in basement membrane composition, with diabetes-induced increases in agrin, fibronectin and tenascin [113]. Changes in vascular permeability are known to accompany diabetic retinopathy and diabetic kidney disease among other pathologies, but the degree to which basement membrane mechanics regulate permeability either directly or indirectly through effects on adherent cells has not been elucidated.

In the kidney, the glomerular basement membrane (GBM) thickens early in diabetes progression, and the degree of thickening correlates with progression of chronic kidney disease and end stage kidney failure [114]. The increase in GBM thickness could be a compensatory mechanism to resist increases in capillary pressure and related increases in single nephron glomerular filtration rate in early disease. For a thin-walled capillary under pressure, hoop stresses (see Box 1) linearly increase with pressure and vessel diameter and linearly decrease with increased wall thickness. Therefore, basement membrane thickening would act to normalize hoop stress in the setting of increased capillary pressure. However, this is offset in part if there is glomerular hypertrophy and increased capillary diameter, which occurs following nephron loss and diabetes [115, 116]. Numerical modeling studies suggest that the increase in capillary wall thickness is not sufficient to offset the effects of glomerular hypertrophy with increased capillary diameter and increased glomerular capillary pressure, leading to an increase in hoop stresses [117]. A number of studies show that extracellular matrix crosslinking is upregulated in diabetic kidney disease [118, 119], and further that crosslinker inhibitors show beneficial effects on slowing progression of CKD in rodent models of diabetic kidney disease [120-122]. However, the actual changes in GBM mechanics in the setting of diabetes have not been well established, nor can the beneficial effects of crosslinker inhibitors be directly attributed to effects on the GBM. We have shown incubating decellularized glomeruli or kidney cortex in sugars induces an increase in stiffness that is dependent on advanced glycation end-product (AGE) crosslinks [123, 124]. However, diabetes-induced stiffness changes in the GBM in vivo have not been studied in detail, nor have the potential pathological consequences of altered GBM mechanics been definitively established.

Cancer metastasis.

In cancer, the ability of malignant cells to cross basement membranes is a critical component of metastasis, both for intravasation into the blood or lymph vessels and for extravasation out of those vessels to establish metastases. Breaching the basement membrane can be protease dependent or independent and may be related to the biophysical properties of the basement membrane [125-127]. The role of basement membranes in regulating metastasis, and the potential of targeting basement membranes to mitigate cancer progression, are only beginning to be explored. In the cancer setting, one could intuit that disruption of normal basement membrane assembly may reduce their mechanical integrity and promote metastasis by allowing cancer cells to more readily cross the diminished basement membrane. However, a recent study showed that individuals with high levels of Netrin-4 have softer basement membranes that protect against metastasis [30]. Several factors could contribute to how readily metastatic cells can cross the basement membrane. For example, cancer cell migration and proliferation are mediated by matrix mechanics [128, 129]. Higher tumor stromal stiffness is an important driver of tumor growth and metastatic potential, and similar effects may play a role in the ability of cancer cells to cross basement membranes. Another factor is protease-mediated degradation of the basement membrane, which plays a role in cancer cell migration [126]. Additional biophysical contributions to basement membrane breaching have recently been explored and are related to the biomechanics of both the basement membrane and the forces generated by the cancer cells that are then transmitted to the basement membrane [125]. These complex biochemical and biophysical mechanisms of metastasis may reveal novel therapeutic targets to minimize the metastatic potential of different cancers [130].

Fibrosis and degradation products.

Basement membranes play a role in disease progression in the setting of chronic fibrotic diseases of the kidney, lungs, liver, heart, cornea and other organs. In the cornea, the epithelial and endothelial basement membranes exclude TGF-β from entering the corneal stroma; when these basement membranes are damaged, TGF-β enters the stroma and promotes the differentiation of fibrosis-inducing myofibroblasts [88]. There are parallels with other fibrotic organ pathologies, and this may be a broadly applicable model of how basement membranes restrain fibrosis. Further, it has recently been appreciated that many fibrotic diseases are accompanied by an increase of basement membrane cleavage fragments in the blood, including pulmonary fibrosis [131], systemic sclerosis [132], fibrotic liver diseases [133-136], and acute heart failure [137]. These studies represent an exciting clinical development for diagnosing and treating a wide variety of diseases based on basement membrane biomarkers.

IV. Basics of basement membrane mechanics.

Basement membranes have several interesting mechanical behaviors arising from their nanoscale molecular organization and architecture. Because of the sheet-like meshwork of protein constituents and high degree of hydration, basement membranes will have a different response to forces in different orientations. The crosslinked collagen IV network imparts basement membranes with high resistance to tensile and circumferential (hoop) stresses (see Box 1). This is particularly important at high strains, where non-linear strain stiffening behavior imparts additional resistance to deformation. In many physiological settings such as the vascular or renal tubules these are the stresses that basement membranes must resist for normal function. As epithelial layers grow, basement membranes will become stretched along their planar axis unless remodeling intervenes [138, 139]. Basement membranes are well adapted to resist these circumferential and tensile stresses due to the high number of lateral interactions [59], the high degree of molecular flexibility, and heavy collagen IV crosslinking. In contrast, when under compression perpendicular to the plane of the basement membranes, basement membranes will tend to act as poroelastic compressible gels that partially collapse and change their pore structure under compressive loading. While this may appear to be a detrimental response, in certain biological contexts this may be physiologically important. For example, in the kidney, compression in the glomerular capillary wall due to the pressure drop across the filtration barrier reduces the porosity of the GBM and increases the size selectivity of the kidney filtration barrier [140, 141], and disease-induced loss of GBM compressibility may render the glomerular filtration barrier more permeable to protein thus contributing to development and progression of proteinuria [142-144]. The effect of compression on basement membrane porosity is discussed below, but here this example illustrates some of the complexities and nuances of the unique biomechanical properties of basement membranes that directly link to their physiological function. We will now discuss these unique properties in the context of the mechanical behavior of typical linear-elastic solids.

Relevance of force-displacement and stress-strain to basement membrane mechanics.

For clarity, we will first define some important terms. Let us use the simple example of a uniaxial tension test where force is applied along one axis to two samples of the same material, but Sample 1 is narrower, with a smaller rectangular cross section than Sample 2 (Fig. 4A). Force is the amount of pull applied to the samples and has units of Newtons (SI units). The tensile force will stretch or deform the samples by a certain displacement, with units of meters. Given the same force applied to each sample, Sample 1 will have a larger displacement owing to the smaller cross-sectional area, and the slope of the force-displacement curve will be less steep for Sample 1 than Sample 2 as shown in Fig. 4B. If we want to use the mechanical response to define a material property, it is necessary to take into account the difference in cross-sectional areas of the samples, where stress (σ) is defined as the amount of force (F) applied per unit of cross-sectional area (A) of the sample and has units of N/m² (Pascals). Similarly, we need to take into account any difference in length of the samples when we consider how far they stretch or deform, and strain (ε) is defined as the deformed sample length (l) normalized to the original length (l₀), and so it is unitless or may be expressed as a percent strain. For example, if a sample in uniaxial tension with an original length of 1 cm is deformed to a length of 1.01 cm, the strain is 0.01 or 1%. If the force is compressive, then the sample length decreases and the strain (ε) is negative. Standard materials (metals, ceramics, most polymers) behave according to well described mechanical responses: displacement increases linearly with applied force as shown in Fig. 4B. If the applied force is doubled, the sample deformation will double. This linear portion of the curve is the elastic regime and defines the elastic properties. If the force is removed in the elastic regime, the sample will return to its original shape. If the force is removed beyond the linear portion of the stress-strain response, the material will not return to the original shape. This linear response will persist until the sample begins to yield or permanently deform. This defines the plastic deformation and can be used to characterize the failure properties of the material.

Figure 4. Mechanical response of linear elastic materials in the elastic regime.

(A) Force (F) applied to two samples (red and blue) made of the same material but with different cross-sectional areas. A has a smaller cross-sectional area than B as shown below. The unloaded samples are shown with F=0.

(B) When the force-displacement response is analyzed, each sample shows a different slope, K.

(C) When the stress-strain response is analyzed, both samples show have the same slope, which is elastic modulus, E. Stress (σ) is the force (F) divided by the cross-sectional area (A) and strain (ε) is the change in length divided by the original length. Analyzing the stress-strain response demonstrates that these samples are made of material with the same elastic properties, which is true since they are made from the same material.

Stiffness is defined as the resistance to deformation in the elastic regime (i.e. before any permanent deformation). Stiffness can be determined in a manner that is either dependent or independent of the sample geometry. The geometrically dependent resistance to deformation is quantified by measuring the spring constant (K) as the slope of the force-displacement curve. For our example, the wider sample (Sample 2) will have a higher spring constant and act “stiffer” than Sample 1 because it better resists deformation under a given applied load (Figure 4B). Since both samples are made from the same material, their resistance to deformation is entirely dependent on the sample geometry. This stiffness difference has contextual importance depending on the mechanical loading conditions, but it does not define a material property. To analyze resistance to deformation independent of geometry, we evaluate the Elastic modulus (E), also called Young’s modulus, defined as the slope of the stress-strain curve (Fig. 4C). Like the spring constant, the elastic modulus is also a measure of resistance to deformation but is normalized to the sample’s cross-sectional area and therefore defines an intrinsic material property. In our example, the elastic modulus of the two samples is equal because they are made of the same material. When referring to a material property, engineers often use stiffness and elastic modulus interchangeably to refer to a material’s intrinsic ability to resist deformation. As biologists we are often concerned not only with the material properties which reflect composition, crosslinking, and structure of basement membranes, but also the physiological or pathological response to force, which reflects also the amount of material. A further complication is that although the terms stiffness and elastic modulus can be used to describe basement membranes, it is important to note that basement membranes do not respond to applied force with the same well-described force-displacement and stress-strain responses as standard materials. These atypical behaviors, discussed below, are largely mediated by the molecular architecture of basement membranes.

Non-linear elasticity and strain stiffening: Basement membranes become stiffer when stretched.

Material behaviors can be elastic but nonlinear, and many biological materials, including basement membranes, exhibit non-linear elastic behaviors where they stiffen with increased deformation [145]. This strain-stiffening could be a mechanism for biological tissues to maintain structural integrity and resist large deformations when subjected to high mechanical loads or possibly to provide mechanical stability during rapid tissue remodeling in development. The stress-strain response of basement membranes under tensile or circumferential stress can be described as hyperelastic [31, 123, 124, 146]. Hyperelastic materials are characterized by their ability to undergo large deformations (strains) while maintaining their original shape after the force is removed. Hyperelastic materials also exhibit non-linear stress-strain behavior. Some familiar hyperelastic materials include rubbers and silicone-based elastomers, so to understand this behavior, consider a rubber band. Rubber bands can be subjected to very high strains while maintaining their elasticity, and as the rubber band is stretched, more units of force are required to stretch it a given unit of displacement. This leads to a nonlinearity in both the force-displacement and stress-strain responses. This is illustrated in Fig. 5: a hyperelastic material subjected to a tensile load leads to a high tensile strain and a non-linear stress-strain response, where the slope of the stress-strain curve increases as strain increases. To compare two samples, it is important to measure the mechanical response at equivalent strains, to account for the fact that the elasticity is different depending on how much it is stretched. For example, tubular basement membrane stiffness in uniaxial tension gives a low strain modulus (i.e. slope of the stress-strain response at 0-10% strain) of approximately 400-500 kPa and a higher strain modulus (30-40% strain) of approximately 3-4 MPa [23]. Different formulations of hyperelastic models have been developed and generally describe the behavior of rubber and elastomeric materials very effectively over the entire strain range [147]. More recently, hyperelastic models have been adopted to describe the mechanical behavior of soft tissue and biological materials [147, 148]. We showed that non-linear hyperelastic models are well-suited for describing the mechanical behavior of basement membranes in uniaxial tension: decellularized kidney extracellular matrix consisting primarily of tubular basement membrane and isolated Drosophila Malpighian tubules exhibit non-linear strain stiffening that is well described by hyperelastic modeling [31, 123]. This strain-stiffening response effectively means that the material can better resist deformation with increasing applied stress. The mechanisms by which this behavior is imparted to basement membranes relates to their molecular organization, interactions, and the flexibility of the different components of the basement membrane as illustrated in Figure 2 for the collagen IV network.

Figure 5. Less intuitive material responses.

(A) Hyperelastic materials show non-linear stress-strain response and increased slope of the stress-strain curve (strain stiffening) with increased strain.

(B) Viscoelastic materials exhibit time-dependent mechanics including creep and stress relaxation.

(C) Poroelastic materials have alterations in pore structure that can cause stress-induced movement of water.

(D) Tension or compression anisotropy, where stiffness is dependent on the direction of the applied stress. Note that the compressive stain is shown as positive to more clearly illustrate the difference in the mechanical response in tension versus compression.

Even for hyper-elastic behavior, the dependency on cross-sectional area is still contextually important. Consider the deformation of two rubber bands of different cross-sectional areas with the same applied force: the narrower one will stretch more than the wider one. Ignoring their cross-sectional area would result in a misleading assertion that the intrinsic mechanical properties of the rubber bands were different, despite being the same material. Instead, if the elasticity is measured at the same strain with normalization to the cross-sectional area, one would correctly determine that the elastic properties of the rubber bands are equal. For a relevant basement membrane example, there are reports that basement membranes become thicker with age [18, 78]; even if the material composition of older basement membranes is the same, the increased thickness would change the response of older basement membranes to forces. Thus, when we compare basement membranes it is important to consider both their intrinsic properties and thicknesses. It is also important to consider if one is interested in how a basement membrane responds to forces in a given physiological context (e.g. spring constant) or if one is interested in defining the materials properties (e.g. elastic modulus). It should be noted that this discussion is oversimplified because, as the strain increases, cross-sectional area decreases, a phenomenon known in the engineering field as true stress versus engineering stress [149]. For materials that remain elastic at large strains, the change in cross-sectional area may be more relevant.

Viscoelasticity: Basement membranes may respond differently to forces over longer periods of time.

Many biological materials, including basement membranes, exhibit viscoelastic behavior, referring to the combination of fluid (viscous) and solid (elastic) properties, and viscoelasticity is characterized by differences in mechanical responses applied over different times or strain rates. Different materials can be on a spectrum between more elastic and more viscous. Viscoelastic responses are generally characterized using stress relaxation and creep testing (Fig. 5B). Creep is a time-dependent change in displacement or strain under a fixed applied force or stress. An example of creep in vivo is observed in cartilage where the knee joint displacement will continue to increase over time under a constant applied force [150]. Stress relaxation is a reduction in force or stress required to impart a fixed displacement or strain over time. An intuitive example of biological viscoelasticity is removing an article of tight clothing, like a pair of socks -- initially, the deformation is maintained in the skin, but after the load is removed, the skin returns to its original shape after a certain amount of time. Interestingly, the different components of basement membranes may have different viscoelastic properties. In C. elegans larval basement membranes, FRAP (fluorescence recovery after photobleaching) experiments have documented that although laminin and collagen IV are stationary, as expected in an elastic solid, several other basement membrane components flow laterally within a basement membrane, including fibulin, nidogen, agrin, spondin, and peroxidasin-1 [44], exhibiting viscous properties. Viscoelastic properties are important for how basement membranes respond to stresses over short and long durations, and they impact the behavior of adherent cells [4, 151-153].

Poroelasticity: Basement membrane porosity decreases when compressed.

Extracellular matrices and basement membranes behave poroelastically. This refers to movement of fluid within the material in response to an applied force (Fig. 5C). An intuitive example of this response is a sponge. When a hydrated sponge is squeezed, water is forced out of the material. Here water was stored in the pores of the sponge, and the pores became smaller in response to the applied compressive force. Given that many basement membrane have porosity or void volumes of >90% [154], a large proportion of the volume is a discrete fluid component that can move through the matrix. The matrix provides resistance to movement of fluids and obstructs the movement of larger solutes. A number of studies in both isolated basement membrane and reconstituted Matrigel have shown that application of pressure to basement membranes reduces both fluid and macromolecular permeability due to a strain-induced reduction in effective pore size with increasing pressure [140, 141, 155, 156]. In the setting of kidney disease, when the GBM filter is unable to compress either due to disruption in podocytes or stiffening of the GBM, the molecular permeability increases, allowing larger molecules to pass through to the filtrate. This may represent a biomechanical mechanism by which changes in the poroelasticity of the GBM contributes to proteinuria in chronic kidney disease [142]. We have shown that enzymatic crosslinking and stiffening of the GBM with the crosslinking enzyme transglutaminase reduces compression and increases molecular permeability of the GBM under applied pressure [143]. Specific application of poroelastic modeling to the GBM suggests that reduction in podocyte buttressing forces would reduce compression, increase porosity, and result in a more permeable GBM under applied pressure [157].

Tension-Compression Anisotropy: Basement membranes respond differently to tension and compression.

Because of the overall sheet-like organization of basement membranes, they are more resistant to displacement in the tensile or circumferential direction as compared to compression (Fig. 5D). Tension in the sheet is resisted by the crosslinked collagen IV network, whereas compression in the sheet would cause buckling; compression orthogonal to the sheet would compress the layers of basement membrane and be opposed not primarily by collagen IV but by hydrated proteoglycans. Their different responses to tension and compression can complicate comparison of measured basement membrane properties in different assays. When responses to tensile or hoop stresses are reported, basement membranes have similar properties. For example, the elastic moduli of basement membranes from mouse renal tubules and Drosophila Malpighian tubules are similar at equivalent strain (10-20%): renal tubules around 1 MPa and Malpighian tubules around 1.5 MPa [23, 31, 158]. This is the same order of magnitude as renal tubule basement membranes measured under circumferential strain (inflation), around 7-10 MPa at high strain [24]. Differences could be attributed to the measurement being taken at different strains or anisotropy in the mechanical response to circumferential versus axial directions. These values are also in line with measurements in lens capsule basement membrane with low strain modulus of approximately 800 kPa and high strain modulus of approximately 7 MPa [159]. By contrast, basement membrane mechanical properties measured using AFM or other compression techniques give values on the order of around 1-100 kPa depending on the origin of the basement membrane. The AFM (compression) measurements report stiffnesses that are not comparable to tensile strain measurements. This is not unexpected given the nature of the collagen IV network, which is well suited to resist tensile and circumferential stresses, but not compressive forces.

Failure Mechanics: Basement membranes beyond the elastic regime.

There are relatively few studies on the failure mechanics of basement membranes, likely owing to the high stresses and strains needed to induce failure. Failure mechanics refers to any mechanical behavior inducing permanent deformation or rupture of basement membranes. Using our tension assay, we were not able to reliably induce failure in either intact mouse renal tubules or Drosophila Malpighian tubules. In the few instances where failure has been reported, the authors noted a “very high value for biological structures,” with failure stresses of approximately 20 MPa in renal tubules [160]. It should be noted that only a small number of the tubules that were tested failed (2 of 11) even at these high strains, suggesting even higher failure strains on average. We observed (unpublished data) that basement membranes can be made to fail at very low strain with limited collagenase digestion, reinforcing the importance of collagen IV in providing the resistance to tensile and circumferential stress. This was investigated by Töpfer et al. who showed that collagenase treatment of Drosophila egg chambers removed basement membrane material, and so not surprisingly stiffness decreased to near zero following collagenase treatment [48], explaining the failure.

Failure is a relevant mechanical property for biological materials. We speculate that basement membranes have high failure stresses to maintain tissue integrity even in the presence of supraphysiological stresses. A potential exception to this is noted by West and Matheiu-Costello [161] who point out an interesting case of thoroughbred racehorses, which require a very high vO₂ max that necessitates a very thin alveolar membrane to provide sufficient oxygen exchange, but also very high capillary pressures. They note that “essentially all thoroughbreds bleed into their lungs during racing,” ostensibly due to mechanical failure of the alveolar basement membrane during this extreme stress state. One could also speculate on the role of basement membrane failure in the setting of pathological conditions such as Alport Syndrome, which is characterized by early onset hematuria. The presence of red blood cells in the urine would be indicative of catastrophic failure in the glomerular filtration barrier, likely due to mechanical failure resulting from a basement membrane defect. A basement membrane’s failure point may not be distributed evenly across the sheet but instead represent a single weak point, which like a snag in a stocking, causes a much larger area to break. For example, in early mouse embryonic development, failure points are introduced into its basement membrane to allow embryo growth [162].

Interestingly, in 1995 Welling et al. noted the similarities in mechanical properties of basement membranes isolated from different tissues and noted that “it is tempting to conclude at this point that all basement membranes are alike and that, because similar results were obtained from several tissues regardless of the presence or absence of an epithelium or endothelium, it is always the basement membrane alone that determine the elastic properties of such tissues. As we will see, however, such a conclusion probably is much too broad and certainly is premature [160].” This was a prescient concept; even while more recent studies have shown extraordinary diversity of basement membrane composition and tissue specific function, basement membranes do exhibit high strength, extraordinary elasticity, and resistance to failure, and the conclusion that basement membranes provide the primary resistance to deformation still holds true.

Physiological changes in basement membranes affect mechanical behavior.

In summary, basement membranes behave as non-linear, strain-stiffening, poro-visco-elastic materials, with these properties depending on whether the basement membrane is under tension or compression. This requires some care to be taken in the interpretation of mechanical testing since multiple variables can affect the measured output. One cannot necessarily define a single elastic modulus for a basement membrane, as the modulus depends on whether measurements were made by compressing or stretching the basement membrane, and further the modulus depends on the strain at which the slope of the stress-strain response is measured. In a physiological or pathological setting, multiple structural changes can be imparted to basement membrane simultaneously, illustrated schematically in Figure 6. With the normal basement membrane as the control (Fig. 6A), a simple increase in the basement membrane thickness with more of the same material but no change in composition (Fig. 6B) would result in an upward shift in the force displacement response that could be misinterpreted as an increase in the elastic modulus of the material. If a proper analysis of the basement membrane thickness is conducted and the stress-strain response plotted, these plots would be identical (Fig. 6C), and one would correctly determine that the material properties are unchanged between these conditions. Even so, the thicker basement membrane is more resistant to deformation, and thus the increase in basement membrane thickness may be physiologically relevant as a compensatory mechanism to resist physiological or pathological mechanical loading. In a different scenario where the thickness of the basement membrane is unchanged, but there is an increase in either density or crosslinking of the basement membrane (Fig. 6B), imaging alone might lead to the false assertion that the basement membrane properties are unchanged. However, a mechanical analysis would reveal an upward shift in both the force-displacement and stress-strain response (Fig. 6C), indicating an increase in the elastic response across the entire spectrum of strain. Indeed, we measured the mechanical properties of basement membranes excessively crosslinked, either by the enzymatic crosslinker transglutaminase or by non-enzymatic glycation with a reducing sugar, and the elastic modulus of the basement membrane increases. We have observed that this effect is more pronounced under tension than compression, pointing to the tension-compression anisotropy in the basement membrane properties [123, 124, 143].

Figure 6. Effects of changes in basement membrane architecture on force-displacement and stress-strain response.

(A) Normal basement membrane with collagen IV lattice shown in the inset.

(B) Changes in basement membrane thickness, crosslinking, and damage.

(C) Effects of these BM changes on force-displacement and stress-strain behavior under tension. Increased basement membrane thickness, in the absence of changes in crosslinking or composition, will result in an upward shift in the force-displacement response but no change in the stress-stain response indicating no change in BM stiffness. Increased crosslinking or increased density results in an upward shift in both the force-displacement and stress-strain response indicating increased stiffness. BM damage or reduced density results in a downward shift in both the force-displacement response and stress-strain response indicating decreased stiffness.

The counterpoint to this example is a damaged basement membrane or one with reduced crosslinking (Fig. 6B): these basement membranes may appear morphologically similar by imaging, but a mechanical analysis would reveal a downward shift in the force-displacement and stress-strain curves, indicating a reduced elastic modulus (Fig. 6C). We measured the mechanical properties of basement membranes disrupted genetically by knockout of the crosslinking enzyme peroxidasin, chemically by the irritant dextran sodium sulfate, pharmacologically by inhibition of peroxidasin, or structurally by enzymatic digestion of the basement membrane, and all of these interventions result in a downward shift in the stress-strain response [23, 31, 143, 158]. Indeed, this reduction in the stress-strain mechanics appears to be an important trigger for basement membrane repair [163]. It remains to be determined how specific physiological or pathological changes result in differences in the mechanical properties of basement membranes. For example, the GBM thickens relatively early in diabetes, possibly as a compensatory response to hyperfiltration, but the actual changes in mechanics of the basement membrane remain to be determined. Moreover, the importance of altered basement membrane mechanics to initiation and progression of pathological conditions such as kidney disease, vascular permeability, cancer metastasis, and organ and tissue fibrosis are now beginning to be elucidated.

V. Measuring basement membrane mechanical properties

There are several techniques for measuring basement membrane mechanical properties. Some of the earliest studies from the early 1970’s measured basement membrane stiffness in isolated renal tubules using a single tubule perfusion system to apply circumferential stress (Fig. 7A). Micropipettes were used to clamp one end of the tubule while a known pressure was applied to the other end to inflate the tubule. By analyzing the change in tubule diameter as a function of increasing pressure, they were able to calculate the strain dependent mechanical properties of the tubular basement membrane [24]. Grantham et al. further measured the viscoelastic properties of the tubular basement membrane from normal and cystic kidney disease models using a micropipette aspiration technique [164]. Another early study characterized basement membrane mechanics using volumetric strain measurements in which a closed system was used to apply pressure-induced volumetric strain on the lens capsule basement membrane, and low and high strain elastic properties were determined [159]. Lens capsule has been a widely utilized model for understanding both collagen IV biochemistry and basement membrane mechanics because the anterior lens capsule basement membrane is much thicker than most other basement membranes and can be easily dissected and manipulated [159, 165, 166]. Even these early measurements recognized the extraordinary mechanical properties of basement membranes and characterized the highly non-linear response to applied stress. More recently, a number of additional technical approaches have been developed and employed to characterize basement membrane mechanical properties. In particular, the advent of atomic force microscopy (AFM) has enabled analysis of mechanical properties of many biological tissues including basement membranes. We discuss the conceptual basics of each technique and their advantages and disadvantages for measuring basement membranes stiffness.

Figure 7. Methods for measuring basement membrane mechanics.

(A) Tubule dilation used to measure the change in tubule diameter (d₁-d₀) with increased applied pressure (ΔP). The outer diameter of the renal tubule is approximately 50-60 μm with some variation based on the tubule segment.

(B) Atomic force microscopy (AFM) used to induce a displacement (δ) with force (F) determined from the bending of the cantilever with a known spring constant.

(C) Tubule tensile test used to apply tensile stress to a tubule structure. This technique has been used to measure basement membrane stiffness in kidney tubules and Drosophila Malpighian tubules.

(D) Drosophila egg chamber burst assay used to assess strength. Eggs are incubated in hypotonic solution driving fluid into the egg chamber. Swelling-induced failure can be used to infer basement membrane strength. Egg chambers used in this assay range from 40-150 μm in length.

(E) Glomerular compression testing used to determine glomerular basement membrane stiffness in compression. Scale bar is C is 200 μm and E is 100 μm (approximate).

Atomic Force Microscopy:

The most common method for measuring extracellular matrix, tissue, and single cell mechanics at the micro/nanoscale is atomic force microscopy (AFM) (Fig. 7B). For AFM, a cantilever is used to indent the surface of a biological sample, and contact-mechanics models are used to calculate the sample stiffness. Stiffness in this case refers to the material property, as the modeling of the tip indentation accounts for the contact area. Indentation can be performed with pyramidal or spherical AFM tips and specific models can be applied depending on the tip geometry. Technical details of instrumentation, contact mechanics, and data analysis for biological samples and extracellular matrices has been reviewed previously [167]. Use of AFM specifically for measuring basement membrane mechanics has been applied to the eye lens, retinal, and corneal basement membranes of the eye [166, 168, 169], Drosophila egg chamber [46-48], and lung cancer basement membrane [30, 170]. Reported stiffness values are typically in the tens to hundreds of kPa for ocular basement membranes with some variability depending on the specific basement membrane and the technical details of the measurements.

Advantages of AFM for measuring basement membrane mechanics are the well-established use of the technique for mechanical analysis, the micro/nanoscale spatial resolution of AFM, the ability to measure samples in their hydrated state, ability to perform mechanical mapping of tissue samples with heterogeneous mechanical properties, and the potential to combine mechanical analysis with optical/fluorescence imaging and surface topographical mapping. Disadvantages of AFM are a relatively high bar for technical proficiency, high cost of the instrument, need for careful instrument calibration, and in most cases, AFM is confined to compression testing. As described above, the differences in the mechanical properties of basement membranes can be significant when analyzed in tension versus compression, and disease-mediated changes in basement membrane structure such as elevated crosslinking are likely to manifest more dramatically under tensile loads.

Microscale Tensile and Compression Testing:

We developed a microscale tensile testing assay for measuring the tensile stiffness of individual basement membranes [23, 31, 124, 158] (Fig. 7C). This technique utilizes pulled glass capillary tubes as cantilever force sensors. Cantilevers were fabricated and calibrated using similar methods to those described by Shimamoto and Kapoor [171]. The basement membrane of interest is held between the measurement cantilever and a rigid holding pipette with vacuum. The holding pipette is attached to a motorized micromanipulator to apply a fixed displacement (d_h). The force is calculated from the displacement of the cantilever (d_m) using the cantilever spring constant. Stress is calculated as the force divided by the cross-sectional area of the basement membrane, measured by EM in separate samples. Strain is calculated from the dimensional changes in the basement membrane length (Δl) normalized to the original tubule length (l₀). We have applied this technique to measure the stress-strain response in peroxidasin knockout renal tubule basement membranes [23], dextran sodium sulfate (DSS) and peroxidasin-inhibitor treated Drosophila Malpighian tubules [31, 158], and sugar modified tubular basement membranes [124]. Similar to Welling and Grantham [24], in both proximal tubules and Malpighian tubules, cells had a minimal impact on the response to tensile stress further reinforcing the concept that basement membranes are responsible for resisting these forces. We found that disruption of the basement membrane -- by either reducing crosslinking or directly inducing chemical damage -- reduced basement membrane stiffness, whereas excessive crosslinking through non-enzymatic glycation increased basement membrane stiffness.

We have modified the above technique to measure properties of decellularized glomerular extracellular matrix in compression (Fig. 7E). The glomerular ECM is composed of glomerular basement membrane and mesangial matrix, so measurements are an aggregate of the mechanics of these matrices. However, the bulk properties of the glomerular matrix reflect changes in the biomechanics of the glomerular basement membrane, particularly at low strains since the GBM is on the outer regions of the glomerulus. We showed that basement membrane damage induced by hypochlorous acid resulted in reduced stiffness as well as increased molecular permeability [172]. We used the same technique to show that increased crosslinking of the GBM either through AGE-mediated glycation or enzymatic crosslinking increased glomerular ECM stiffness [124, 143]. Advantages of these techniques include ability to perform measurements on in vivo derived basement membranes with minimal sample processing, relatively simple image-based data analysis, and the fact that there is no specialized or expensive equipment. However, these techniques are only suited to specific structures, like cylindrical tubes or spheres and are not well suited to sheets or planar structures. These techniques also have limited spatial resolution compared to AFM. Therefore, any mechanical or compositional heterogeneity in the material is not captured in the bulk measurement.

Inflation, swelling, and micropipette aspiration:

For spherical or semi-spherical structures surrounded by a basement membrane, inflation and burst testing can be used to evaluate basement membrane mechanics and failure. Li et al. measured the elasticity of the basement membrane surrounding cancer spheroids by injecting fluid to rapidly inflate and deflate the spheroid [146]. They calculated the non-linear stress-strain response, which was fit to the Fung model of hyperelasticity. The tangent modulus, which can be described by taking the derivative of the stress-strain curve (K_t = dσ/dε), increased with increasing strain. Advantages of this technique include the ability to simultaneously measure mechanical properties and water permeability, minimal processing of the sample prior to measurements (e.g. no decellularization process is needed), and the ability to measure properties at high strains. Limitations of this method are that it can only be applied to closed spherical structures such as cancer spheroids with external basement membranes that are amenable to inflation and deflation. Therefore, it cannot be broadly applied to in vivo basement membranes.

A related but non-quantitative assay was developed to measure failure in Drosophila egg chambers: incubating the tissue in a hypotonic media results in movement of water into the egg chamber that will eventually cause the structure to fail by bursting, often explosively (Fig. 7D). While not a direct measure of basement membrane elasticity, the likelihood of chamber failure can be used as a measure of the mechanical integrity of the basement membrane. This assay provides a surrogate measure of the failure strain of the basement membrane. This has been applied to egg chambers following genetic knockout of various components of basement membrane including collagen IV, laminin, perlecan, and nidogen [48]. This analysis showed a significant increase in likelihood of failure with loss of perlecan and collagen IV, a lesser effect in laminin knockdown, and no effect of loss of nidogen. The advantage of this technique is that it requires no specialized equipment or complex data analysis; the disadvantage is its lack of quantitative data and inability to directly measure basement membrane stresses which precludes direct analysis of material properties and limits comparisons with other assays.

In summary, each of these techniques has advantages and disadvantages. The choice of characterization method for determining the mechanics of the basement membrane should be chosen based on the specific questions to be answered with consideration of the temporal, spatial, and cost constraints of each technique. Importantly, the data should be carefully analyzed with consideration of the unique mechanical properties of basement membranes described here to avoid misinterpretation of the results.

Conclusions

We hope we have illustrated that basement membranes have unique mechanical properties that stem from their macromolecular structure. The collagen IV network, with its cross-linked sheet-like structure, is well-designed to resist tensile stresses. Like other elastic materials, the amount of deformation in response to a force will be determined not only by an individual basement membrane's mechanical properties, quantified as the elastic modulus, but also by its thickness, which is not accounted for in the elastic modulus. Unlike traditional linear elastic materials, basement membranes exhibit strain-stiffening, that is, they get stiffer the more they are stretched. Some of its strain-stiffening is explained by the kinks in the collagen IV triple helix, which provide the network with slack; as the network is stretched the slack is removed until the stresses are resisted by the crosslinked network architecture itself. Basement membranes also exhibit other behaviors that differ from linear elastic materials, including viscoelasticity, characterized by stress relaxation and creep, and poroelasticity. As sheet-like structures, basement membranes respond differently to tensile stresses in the plane of the sheet than to compressive stresses normal to the plane. Tensile stresses are largely resisted by the crosslinked collagen IV network, whereas compressive stresses are significantly resisted by hydrated proteoglycans. For all these reasons, when comparing measurements of basement membrane properties, it is important to compare like with like: stiffness can only be compared at a given strain magnitude and strain rate, and AFM measurements under compression cannot be directly compared with tensile strain measurements. The number of assays for measuring basement membrane mechanical properties is continuing to grow, providing the field with an important foundation for understanding the relationship between basement membrane mechanics and biological function.

Box 1.

Schematic illustrating tensile, compressive and hoop (aka circumferential) stresses.