Collagen, stiffness, and adhesion: the evolutionary basis of vertebrate mechanobiology.

The emergence of collagen I in vertebrates resulted in a dramatic increase in the stiffness of the extracellular environment, supporting long-range force propagation and the development of low-compliant tissues necessary for the development of vertebrate traits including pressurized circulation and renal filtration. Vertebrates have also evolved integrins that can bind to collagens, resulting in the generation of higher tension and more efficient force transmission in the extracellular matrix. The stiffer environment provides an opportunity for the vertebrates to create new structures such as the stress fibers, new cell types such as endothelial cells, new developmental processes such as neural crest delamination, and new tissue organizations such as the blood-brain barrier. Molecular players found only in vertebrates allow the modification of conserved mechanisms as well as the design of novel strategies that can better serve the physiological needs of the vertebrates. These innovations collectively contribute to novel morphogenetic behaviors and unprecedented increases in the complexities of tissue mechanics and functions.

INTRODUCTION

Three major driving forces in the evolution of vertebrates are the development of a stiffer extracellular environment, the emergence of collagen-binding integrins, and the significant increases in force generation in nonmuscle cells. These conditions, together with two rounds of whole genome duplications at the base of vertebrate evolution, give rise to new structures and cell types. The purpose of this Perspective is to speculate on how collagen I, stiffness, and adhesion impacted the evolution of mechanobiology, resulting in divergent mechanisms in vertebrates and nonvertebrates (Nakatani ; Hufton ; Hoffmann ).

Multiple biological and animal models are used to study cellular behaviors such as tissue morphogenesis. However, in many instances, different systems produce conflicting observations and, in the field of mechanobiology, very different conclusions. These discrepancies are seen at many levels of investigations. At the tissue level, cell movement appears to result from different mechanisms during the development of Drosophila and Xenopus embryos (Keller, 1986; Keller ; Keller and Winklbauer, 1992; Shih and Keller, 1992; Irvine and Wieschaus, 1994; Bertet ; Zallen and Wieschaus, 2004; Shindo, 2018). At the signaling level, collective cell migration appears to be activated by distinct ERK-dependent mechanisms in Drosophila and mammalian cells (Aoki ; Ogura ). At the molecular level, different amounts of molecular force appear to be exerted at cell–cell adhesions in the developing Drosophila and zebrafish embryos (Yamashita ; Eder ; Lagendijk ). Furthermore, the physiological roles of mechanosensitive molecules such as vinculin also appear to be different. For example, knockout of vinculin has little effect in Drosophila but is detrimental to zebrafish and mice (Alatortsev ; Xu ; Zemljic-Harpf ; Cheng ; Lausecker ). In Drosophila, hyperactive vinculin inhibits the formation of integrin adhesion complexes, whereas the opposite is true for the mouse (Marg ; Maartens ). At the cellular level, novel cellular and molecular structures are found only in vertebrates. One novel structure found in nonmuscle cells is the stress fiber, which has an alternating actin and myosin II arrangement, analogous to the sarcomeres in muscle cells. The “sarcomere-like” actomyosin organization has not been described in nonvertebrate cells, although basal actin arrays and oriented actin bundles are found in Drosophila (Byers and Fujiwara, 1982; Drenckhahn and Wagner, 1986; Nehls and Drenckhahn, 1991; Delon and Brown, 2009; Cetera ; Xie ; Goodwin ; Qin ; Cerqueira Campos ). Sarcomeric stress fibers are present at the apical junction of vertebrate epithelial cells but are not found at the Drosophila junction despite the significance of junctional actomyosin networks in intercellular movements (Fernandez-Gonzalez ; Fernandez-Gonzalez and Zallen, 2011, 2013; Martin ; Sawyer ; Curran ; Heuze ). While such differences might reflect simple interspecies variability, it is also possible that much of mechanoregulation is not well-conserved between vertebrates and nonvertebrates (Stephenson ; Soslau, 2020).

THICKER AND STIFFER FIBRILS INCREASE THE STIFFNESS OF EXTRACELLULAR MATRIX

Collagen I is the primary component of the vertebrate extracellular matrix (ECM), making up ∼30% of the total protein mass in humans (Huxley-Jones ; Kadler ; Silver, 2009; Exposito ). Collagen I forms thick, stiff, and long fibrils that are responsible for the dramatic decrease in tissue compliance in vertebrates (Silver , 2002; Ushiki, 2002; Boot-Handford and Tuckwell, 2003; Yang ). Depending on the density and packing order of collagen I fibrils, vertebrates can construct ECM up to hundreds of megapascals in stiffness (van der Rijt ; Candiello ; Shen , 2011; Yang ; Svensson ; Aifantis ; Kohn ); Mineralization of collagen I fibers further increases the stiffness of the matrix to several gigapascals, a magnitude that is orders of magnitude higher than the ECMs than most nonvertebrates would have ever experienced (Figure 1A) (Kawasaki ; Landis and Silver, 2009; Chlasta ).

FIGURE 1:

A reductionist’s view of the extracellular experience in vertebrates and nonvertebrates. (A) The evolutionarily conserved fibronectin, collagen IV, and laminin form meshwork-type matrixes with stiffnesses up to hundreds of kilopascals (kPa). In vertebrates, collagen I forms matrixes with stiffnesses ranging from megapascals (MPa) to gigapascals (GPa). Nonvertebrates and vertebrates can bind to laminin and fibronectin, whereas only vertebrates interact directly with collagen (see the text for details). (B) Two cells are interacting with the ECM and with each other. Force generated by stress fibers inside the cell can be transmitted to the extracellular collagen matrix via collagen-binding integrins. A basement membrane containing a mixture of matrix proteins is present in some specialized tissues.

CONNECTING ECM TO CELLS

The major force-transmission pathway connecting the force-generating actomyosin structures inside the cells to the ECM occurs via cell surface integrins (Figure 1B) (Niland ; Livne and Geiger, 2016; Carey ). Vertebrate and nonvertebrate cells can interact with the evolutionarily more conserved fibronectin and laminin through fibronectin-binding and laminin-binding integrins. However, only vertebrates have evolved collagen-binding integrins to interact with collagens. Thus, despite the presence of collagen IV in all metazoa, nonvertebrate cells do not interact directly with collagen IV because they do not have the collagen-binding integrins (Ewan ; Chouhan ; Johnson and Chouhan, 2014).

In vertebrates, fibronectin can coassemble with collagen I to form fibers that potentially can interact with both fibronectin-binding and collagen-binding integrins concomitantly (Kadler ; Singh ; Kubow ; Lemmon and Weinberg, 2017; Mezzenga and Mitsi, 2019). Thus, the emergence of collagen I not only increases the stiffness of the overall extracellular environment, it also alters the rigidity and organization of the fibronectin network, allowing a softer meshwork-type matrix to behave as a stiffer fibrillar-type substrate. This has far-reaching implications and consequences because fibronectin-binding integrins have mechanotransduction roles and can sense the stiffness of the fibronectin matrix (Tee ; Trichet ; Carraher and Schwarzbauer, 2013; Ribeiro ; van Geemen ; Zhou ; Janmey ). By dramatically increasing the stiffness and organization of the fibronectin matrix, collagen I supports a wider range of mechanotransduction responses elicited by fibronectin-binding integrins.

COLLAGEN-BINDING INTEGRINS FORM THE STRONGEST CELL–ECM BONDS

The emergence of collagen-binding integrins is one of the most important events in the evolution of vertebrate mechanobiology (Calderwood ; Tuckwell ; Tulla ; Jokinen ; Ewan ; Chouhan ; Johnson and Chouhan, 2014). Collagen-binding integrins form bonds with collagen that are stronger than any integrin–fibronectin or integrin–laminin bonds, and thus are the load-bearing structures for the vertebrate cell-matrix adhesion (Tiger ; Chen ; Huhtala ; Zhang ; Louis ; Kong ; Roca-Cusachs , 2012; Carracedo ; Hu ; Niland ; Popov ; Zeltz and Gullberg, 2016; Ciuba ). The direct coupling between integrin and collagen plays a significant role in the generation of higher tension and more efficient force transmission in the collagen matrix (Hu ; Mohammadi ). Furthermore, the ability to exert force directly on collagen fibers allows the acquisition of novel cellular behaviors including multicellular streaming and contact guidance (Tiger ; Tamariz and Grinnell, 2002; Wolf , 2009; Miron-Mendoza ; Carey ; Grossman ; Han ; Nuhn ; Sarker ). Collagen-binding integrins and fibronectin-binding integrins form discrete molecular complexes and have distinct mechanotransduction functions, contributing to cell-matrix biology in complementary ways in vertebrates (Ylanne ; Hocking ; Teravainen ; Ribeiro ; Burridge and Guilluy, 2016; Roca-Cusachs ).

COLLAGEN I SUPPORTS LONG-RANGE FORCE TRANSMISSION

Vertebrate collagen I has evolved new biochemical strategies to form compact, staggered, and covalently cross-linked fibrils that are thicker, stiffer, and longer than any ancestral fibrillar or meshwork collagens (Pins ; Eyre and Wu, 2005; Ricard-Blum, 2011; Kwansa ). Collagen I is more efficient for directional force propagation due to the increased fiber length, the higher rigidity, and the precise alignment of collagen I molecules in the fibers. The amount of force that can be generated in collagen I matrixes is substantially higher, up to several orders of magnitude, than that that can be supported by fibronectin, laminin, or collagen IV meshwork-type matrixes (Hocking ; Boot-Handford and Tuckwell, 2003; Candiello ; Dominguez-Gimenez ; Araki ; Silver, 2009; Exposito ; Maruthamuthu ; Sztal ; Tee ; Fahey and Degnan, 2012; Trichet ; Roca-Cusachs ; Teravainen ; Cetera ; Xie ; Adams ; Kubow ; Burridge and Guilluy, 2016; Goodwin ; Chlasta ; Fidler , 2018; Filla ; Qin ; Sancho ; Shook ; Zollinger ; Draper ; Cerqueira Campos ; Chronopoulos ).

How exactly might collagen I bring new mechanobiology to vertebrates? At least four factors are important (Figure 2):

FIGURE 2:

Collagen I forms stiffer, thicker, and longer fibers to support the generation of tension and long-range force propagation. (A) In nonvertebrates, cells can interact with laminin and fibronectin (FN) via laminin-binding and fibronectin-binding integrins. However, nonvertebrates do not have collagen-binding integrins and cannot exert force on the collagen (COL) IV matrix. Vertebrate cells have both fibronectin-binding and collagen-binding integrins and can bind directly to fibronectin and collagens (see the text for details). (B) Pulling on fibronectin-binding integrins that are bound to fibronectin dimers results in movement of the integrins with no change in tension in the matrix. (C) Pulling on collagen-binding integrins will generate tension in collagen IV meshworks (see the text for details). Pulling on a distensible fibronectin network may also result in the generation of tension in the matrix (see the text for details). (D) Long-distance force transmission and tension generation is supported by collagen I fiber attached to two cells via collagen-binding integrins. Pulling a collagen fiber by a cell can translocate the fiber.

(1) The number of binding sites matters. The number of integrin-binding sites on a matrix molecule determines whether it can cluster the integrins and how much force it can transmit (Figure 2A). A laminin trimer interacts with only one integrin, whereas a fibronectin dimer has two integrin-binding sites. A collagen fibril contains tens to hundreds of covalently cross-linked molecules, providing many integrin-binding sites (Xu ; Pichard ; Turner ). In vivo, fibronectin can form larger assemblies via noncovalent interactions, increasing the apparent binding sites. Collagen IV can also form large assemblies via noncovalent interactions and disulfide bonds. However, these networks are deformable and can buckle, collapse, or rupture when stretched. On the other hand, the extensively cross-linked collagen fiber behaves as one unit under force. Thus, the total amount of force that can be applied to a collagen fiber is substantially higher than the forces that can be applied to meshwork-type matrixes. Multiple interactions between integrins and the collagen fiber would allow force integration necessary for stiffness sensing by the cells (Jiang ; Janmey ).

(2) Size and length matters. The nonmuscle myosin II minifilament, which is ∼300 nm long, forms the force-generating unit of the stress fiber (Pellegrin and Mellor, 2007; Billington ; Dasbiswas ). To generate tension at integrin adhesions, two integrins must be positioned at least as far apart as the length of a myosin II minifilament. A fibronectin dimer has a length of ∼100 nm, and while it can cluster two integrins it cannot support tension generation. Thus, contraction of the stress fiber will pull on both integrins, dragging the fibronectin dimers and the integrins in the same direction. If the fibronectin dimers are isolated, there will be no resistance to the movement and no tension will be generated within the matrix (Figure 2B). In a fibronectin network, pulling on fibronectin would lead to stretching of the network. However, if the pulling force exceeds the strength of the noncovalent bonds, the fibronectin network could break, interrupting the force-transmission pathway and tension generation. Collagen IV meshworks are held together by disulfide bonds and noncovalent interactions and thus can withstand pulling force when stretched, possibly permitting tension to build up in the distorted matrix (Figure 2C). In vertebrates, a collagen I fiber can be up to tens of microns in length and thus can integrate many “sarcomere-like” contractile units, increasing the force exerted at individual cell-matrix adhesions and resulting in greater tension to be generated within the ECM. When two noncontacting cells are bound to the same long collagen I fiber, pulling force from one cell can be sensed by a neighboring cell (Figure 2D). Traction force generated at collagen-binding integrins can realign or translocate the collagen I fiber, resulting in the remodeling of the ECM (Ramos and DeSimone, 1996; Davidson , 2004, 2008; Marsden and DeSimone, 2003; Miron-Mendoza ; Wang ; Han ; Ban ; Shiflett ).

(3) Stiffness matters. Matrixes composed of fibronectin, laminin, and collagen IV form meshworks that are stretchable by cellular force, whereas collagen I is long and compact fibers that are stiff and relatively unbendable by cellular force. The propagation of force within fibronectin and collagen IV meshworks primarily goes through noncovalent protein–protein interfaces. In the presence of pulling force that can break noncovalent bonds, meshwork-type matrixes will fail to transmit force while the covalently cross-linked collagen I fiber will continue to support force propagation. In vertebrates, the interaction of fibronectin with fibrillar collagen I via noncovalent bonds could alter the apparent biophysical properties and stiffness of the fibronectin network. However, fibronectin is extremely distensible and unfolds almost completely under physiological force, reaching a tensile stiffness of only 1–2 MPa when maximally stretched, which is at least two orders of magnitude lower than the tensile stiffness of collagen I fibrils (van der Rijt ; Shen ; Yang ; Klotzsch ; Svensson ; Bradshaw and Smith, 2014). Moreover, fibronectin–integrin bonds are not as strong as collagen–integrin bonds. Therefore, force propagation in collagen I matrix is much more efficient due to the covalent intermolecular bonds, the higher tensile stiffness of the collagen I fibers, and the stronger bonds between collagen and collagen-binding integrins. In this regime, higher levels of tension can be accumulated within the ECM. The tensile energy in the stiffer matrix, in turn, would increase the stiffness of the cell and promote greater force generation that can be used to assemble stress fibers, remodel focal adhesions, promote cell–cell adhesions, and ultimately increase the overall potential energy of the tissue (Walcott and Sun, 2010; Maruthamuthu ; Wolfenson ; DeMali ; Wang ; Wong ; Ye ; Broaders ; Ronan ; Hall ; Hu ; Le ; Dasbiswas ; Han ; Kuragano ; Chang ).

(4) Geometry matters. Collagen IV and fibronectin form meshworks that are isotropic and would dissipate force in all directions, thus cannot faithfully reproduce the vector quality of force (Figure 2A). In contrast, collagen I fibrils behave like linear cables that can support the propagation of directional information, alignment of stress fibers, and the asymmetric growth of focal adhesions (Figure 2D). Importantly, linear collagen I fibers provide physical cues for directed protrusions and cell migration (Kaunas ; Besser and Safran, 2006; Miron-Mendoza ; Foolen ; Tondon and Kaunas, 2014; Mohammadi ; Han ; Xie ; Brauer ; Sarker ).

THICKER, STIFFER, LONGER FIBRILS SUPPORT HIGHER FORCE PRODUCTION

Vertebrate cells stiffen when attached to collagen I, exerting forces up to hundreds of nanonewtons which is nearly an order of magnitude higher than that can be generated by fibronectin-attached cells (Hocking ; Araki ; Maruthamuthu ; Tee ; Trichet et al., 2012; Roca-Cusachs ; Teravainen ; Kubow ; Burridge and Guilluy, 2016; Sancho ; Zollinger ; Chronopoulos ). Experimental evidence indicates that the maturation process of fibronectin-based adhesions is independent of force (Oakes ; Stricker ), raising the possibility that fibronectin-based adhesions are fundamentally different from collagen-based adhesions.

There are many possible ways to increase force production at collagen-based adhesions in vertebrates:

(1) Turn on myosin II ATPase. One way to increase force production is to activate pathways that phosphorylate and activate myosin II and turn off pathways that dephosphorylate and inactivate myosin II (Tojkander ; Ciuba ). The molecular motor myosin II is responsible for force generation in cells and is regulated by the phosphorylation status of the myosin light chain. In cells, myosin light chain is phosphorylated by myosin kinase. Thus, activating signals such as Rho, Rock, and calcium-calmodulin that increase the activities of myosin kinase would stimulate myosin contraction. Myosin phosphatase, in turn, dephosphorylates the myosin regulatory light chain and ends contraction. Activating signaling pathways that inhibit myosin phosphatase will promote myosin light chain phosphorylation and sustained contractility.

(2) Modify core machinery. To increase force production, myosin II assemblies with more motor heads can be formed by regulating the stability and the assembly of myosin II minifilaments. Increasing the duty ratio of the myosin motor would increase the duration of myosin interaction with actin, thus increasing the processivity of individual motors. In addition, altering the mechanoresponse of the myosin motor would increase the strength of the myosin–actin bond under mechanical stress. These properties are already implemented by the vertebrate-specific myosin IIB (Figure 3A), a paralogue of the ancestrally derived myosin IIA (Murakami , 2000; Chantler ; Stam ; Heissler and Sellers, 2016; Kuragano ; Melli ).

FIGURE 3:

Controlling force production by myosin II regulation. (A) Force generation by stress fibers can be modified by increasing the duty ratio of the myosin II motor, processivity of the myosin II minifilament, mechanosensitivity of the myosin molecule, and the size of the myosin minifilament. (B) Force production by myosin II minifilaments can be amplified by arranging them in series and in parallel.

(3) Build new structures using existing components. Each myosin II minifilament is capable of producing force of ∼30–50 piconewtons, but when the cells are attached to the collagen I matrix, they can generate forces up to hundreds of nanonewtons (Kaya and Higuchi, 2010; Sim ; Sancho ). This means that thousands of myosin II minifilaments must work synergistically to produce the amount of force that is measured at collagen I–based adhesions. To amplify force production, the cell organizes the myosin minifilaments and arranges them in parallel and in series, rather than randomly positioning them inside the cytoplasm (Figure 3B). Indeed, this organization is used in vertebrate nonmuscle cells to build actomyosin stress fibers capable of generating force that is an order of magnitude higher than the sum produced by individual myosin motors (Pellegrin and Mellor, 2007; Lohner ).

Forces generated by stress fibers are transmitted to the ECM through focal adhesions, which are sites of primary traction consisting of clustered integrins and strengthened linkage to actin (Ye ; Burridge and Guilluy, 2016; Livne and Geiger, 2016). Focal adhesions formed by collagen-binding integrins are essential for vertebrate mechanobiology by participating in stiffness sensing, cell migration, and remodeling of the ECM (Burridge and Fath, 1989; Fath ; Plotnikov ; Bays ; Hall ; Chang ; Puleo ).

TUNING FORCE IN A STIFFER ENVIRONMENT SUPPORTED BY WHOLE GENOME DUPLICATIONS

To manage the new biology that comes with a stiffer environment, vertebrates are provided with a new toolbox, brought in part by a major evolutionary event consisting of two rounds of whole genome duplications (Vandepoele ; Dehal and Boore, 2005; Holland and Ocampo Daza, 2018; Marletaz ). This major evolutionary event gives rise to paralogues at a genomic scale that marks the branching of vertebrates from the invertebrates ∼500–550 million years ago. Thus, a significant percentage of adhesion and cytoskeletal proteins in vertebrates have multiple copies, one from the ancestral gene and the others from duplicated genes, for example, filamin A-C, paxillin α-γ, myosin IIA-C, α-actinin-1-4, tropomyosin1-4, WAVE 1-3, RhoA-C, Rac1-3, and talin1-2 (Boureux ; Gehler ; Kurisu and Takenawa, 2009; Rahimzadeh ; Billington ; Austen ; Mohammadi ; Meacci ; Schiffhauer ; Kuragano ; Pathan-Chhatbar ; Puleo ; Sao ). While the ancestral copy often retains its original function, the duplicated copies frequently acquire new functions due to the lack of selection pressures on the copies. The expanded protein toolbox provides new options for the vertebrates to increase complexity and evolve novel molecular structures, mechanisms, and regulations (Brady ; Laurin ).

One way to build complexity into mechanoregulation is to adjust force production and tune mechanoresponses. Vertebrates have many options to handle this challenge:

(1) . For example, vinculin is expendable in Drosophila but, in vertebrates, it plays an important role in the strengthening of cell–cell and cell–matrix adhesions under mechanical stress (Alatortsev ; Cheng ; Bays and DeMali, 2017; Le ).

(2) Evolve new proteins and structures that are absent in nonvertebrates. Many newly emerged vertebrate proteins and structures are designed to work with conserved molecules and core mechanisms to increase complexity (Mariani ). One example is the intermediate filament vimentin, which localizes to focal adhesions and regulates the assembly of integrin adhesion complexes under mechanical stress (Tsuruta and Jones, 2003; Sanghvi-Shah and Weber, 2017). Another example is the tight junction, which can modulate the mechanical input at the adherens junction in epithelial cells (Hatte ).

(3) Force generation can be adjusted by controlling stress fiber dynamics or focal adhesion stability. One example of subfunctionalization is the differential use of SORBS family members to control the amount of force applied to the ECM (Kuroda ). While the vertebrate paralogue SORBS2 interacts with α-actinin to regulate stress fiber contractility, the ancestral paralogue SORBS1 interacts with vinculin to control focal adhesion maturation (Ichikawa ). Another example is the differential usage of myosin IIA and IIB paralogues, resulting in the spatial regulation of stress fiber formation, front–back polarity, ECM remodeling, and intercellular junction dynamics (Even-Faitelson and Ravid, 2006; Vicente-Manzanares ; Sandquist and Means, 2008; Solinet and Vitale, 2008; Smutny ; Doyle ; Gutzman ; Stam ; Ridge ; Kim ; Kuragano , b; Heuze ).

A COMPLEX EXPERIENCE FOR THE VERTEBRATE CELL

The expanded toolbox, along with collagen I and collagen-binding integrins, created exceptional opportunities for the vertebrates to build complexities in mechanobiology (Figure 4, A and B). To illustrate, two cells were drawn in Figure 4A; the cell on the left has a myosin II minifilament that is composed of myosin IIA (red motor heads) and IIB (black motor heads). The hybrid myosin IIA/IIB minifilament is attached to two focal adhesions containing collagen-binding integrins; one focal adhesion is bound to collagen IV meshwork (red dotted circle on the left cell), and the other focal adhesion is bound to a collagen I fiber. Myosin IIB pulls harder than myosin IIA and moves the minifilament toward the collagen I fiber, exerting force on the collagen IV meshwork and ultimately breaking the noncovalent bonds within the collagen IV matrix (Figure 4B, red dotted oval on the left cell). Using the same argument, the cell on the right in Figure 4A has a hybrid myosin II minifilament composed of myosin IIA (red motor heads) and IIB (black motor heads). The stress fiber is attached to two focal adhesions; one focal adhesion is bound to the fibronectin meshwork via fibronectin-binding integrins (red dotted circle on the right cell), and the other focal adhesion is bound to a collagen I fiber via collagen-binding integrins. Myosin IIB wins the tug-of-war, and the asymmetric hybrid myosin IIA/IIB minifilament moves toward the collagen I fiber, pulling on the fibronectin-binding integrins and breaking their interaction with the fibronectin matrix (Figure 4B, red dotted oval on the right cell). Thus, the weakest link in the chain dictates the upper limit of force transmission (Burridge and Guilluy, 2016). This concept is central to mechanobiology and holds the biophysical basis for mechanoregulatory principles and models.

FIGURE 4:

The weakest-link-in-the-chain concept in force transmission (Burridge and Guilluy, 2016). (A) Example of a complex cellular experience involving two cells embedded in an extracellular environment composed of collagen IV, collagen I, and fibronectin. The two cells are expressing different combinations of fibronectin-binding and collagen-binding integrins on their cell surfaces. A collagen fiber is interacting with both cells. In the cell on the left, two focal adhesions with collagen-binding integrins are interacting with two different matrixes; one focal adhesion is bound to a collagen IV meshwork (red dotted circle on the left cell), and the other focal adhesion is bound to a collagen I fiber. A hybrid minifilament composed of myosin IIA (red motor heads) and IIB (black motor heads) is attached to the stress fiber linking the two focal adhesions. In the cell on the right, the hybrid myosin IIA/IIB minifilament is attached to two focal adhesions consisting of different integrins; one focal adhesion is bound to the fibronectin meshwork via fibronectin-binding integrins (red dotted circle on the right cell), and the other focal adhesion is bound to a collagen I fiber via collagen-binding integrins. (B) Contraction force from the stress fibers can differentially regulate cell-matrix adhesions and extracellular force transmission in the matrix. In the cell on the left, myosin IIB pulls harder than myosin IIA and moves the minifilament toward the collagen I fiber, exerting force on the collagen IV meshwork and ultimately breaking the noncovalent bonds within the collagen IV matrix (red dotted oval on the left cell). In the cell on the right, myosin IIB wins the tug-of-war and the myosin IIA/IIB minifilament moves toward the collagen I fiber, pulling on the fibronectin-binding integrins and breaking their interaction with the fibronectin matrix (red dotted oval on the right cell).

CONCLUDING REMARKS

The emergence of collagen I drastically changed the compliance, porosity, stiffness, and organization of the ECM in vertebrate tissues. However, changing the biophysical properties of the extracellular environment alone is not sufficient for the evolution of new processes unless the stiffer matrix is linked to the cell. This is achieved by the coemergence of collagen-binding integrins at the base of the vertebrate evolution ∼550 million years ago, resulting in a powerful change in the energy landscape in cells and in tissues. Ultimately, the dawn of vertebrate mechanobiology is realized by two rounds of whole genome duplications, providing new tools for regulation, amplification, and integration.

The decrease in tissue compliance allows the development of blood pressures to support pressurized renal filtration, a vertebrate invention that is superior in design for the management of electrolytes and metabolites (Schulte ; Stephenson ; Soslau, 2020). The increase in hydrostatic pressures in the closed circulatory system of the vertebrates necessitates the development of a vascular lining to prevent transmural leakage of vascular contents. The emergence of endothelial cells that line the vascular wall is another significant vertebrate innovation that coemerged with collagen I and collagen-binding integrins. Endothelial cells are equipped with specialized cell–cell junctions to act as permeability barriers and have strengthened cell–cell adhesions to resist hydraulic mechanical stress. The plasticity of the endothelial cells allows functional specialization of the endothelium, including a system of low-pressure capillary beds for nutrient delivery and the formation of the blood–brain barrier (Bundgaard and Abbott, 2008; O’Brown ). The morphogenesis and the development of the endothelial vasculature, a process known as angiogenesis, is dependent on collagen-binding integrins on the endothelial cells and is supported by the collagen I ECM (Senger ; Sweeney ; Whelan and Senger, 2003; San Antonio ; Kick ; Post ; Turner ).

The endothelium ushers the appearance of multiple vertebrate-specific organs that are the footprints and signatures of vertebrate evolution (Shigei ; Vize and Smith, 2004; Munoz-Chapuli ; Monahan-Earley ). Endothelial cells are essential for the development of the vertebrate liver, pancreas, lungs, and kidneys during embryo morphogenesis. In adult animals, endothelial cells play significant roles in the physiological functions of the kidneys to concentrate urine and the lungs to perform gaseous exchange (Lammert ; Cleaver and Dor, 2012; Ramasamy ; Rafii ; Bastidas-Ponce ; Daniel and Cleaver, 2019). The morphogenesis of these vertebrate-specific organs is dependent on the proper deposition and organization of collagen I (Aycock and Seyer, 1989; Goldstein, 1991; Saelman ; Shimeld and Holland, 2000; Chen ; Riopel and Wang, 2014; Buchtler ; Stephens ).

Endothelial cells also play a significant role in the development of neural crest derivatives including the vertebrate skull, the brachial skeleton, and the sensory ganglia. Endothelial cells interact with neural crest cells, another key innovation of the vertebrates, and influence the distribution, the migratory pathways, and the patterned formation of neural crest cell derivatives. Neural crest migration from their origin of formation between the neural plate and the surface ectoderm is dependent on collagen I and requires interaction with the matrix via collagen-binding integrins (Duband and Thiery, 1987; Perris ; Nagy ; Diogo ; George ).

Further along the evolutionary timeline, mammals continue to utilize fibrillar collagens and collagen-binding integrins to invent new mechanosensitive processes, structures, and physiology, including the development of the mammalian placenta and invasion of the uterine wall by trophoblasts in a stiffness-dependent manner during embryo implantation (Damsky ; Braasch ; Yoshida ; Griffith and Wagner, 2017; Park ; Abbas ; Zambuto ). In conclusion, vertebrate mechanobiology is created, in part, by three major evolutionary forces including collagen I, collagen-binding integrins, and vertebrate-specific proteins to support the development of novel molecular mechanisms, adding to the repertoire of mechanisms that is already in place and shared by nonvertebrates and vertebrates (Rozario and DeSimone, 2010).