The triple helix of collagens – an ancient protein structure that enabled animal multicellularity and tissue evolution

The extracellular matrix (ECM) played an essential role during the transition from unicellular organisms to multicellular animals (metazoans). The ECM comprises a basement membrane (BM) that underlies epithelia cells, and an interstitial matrix (IM) that is positioned between cells in the intercellular spaces and undergoes continuous controlled remodeling (Hynes, 2012; Bonnans et al., 2014; Nelson and Bissell, 2006; Inman et al., 2015). Yet, a major gap in cell biology is to understand how cells generate and interact with the ECM (Sherwood, 2015; Jayadev and Sherwood, 2017).

The collagen superfamily of proteins is a major component of ECMs, which – in vertebrates – comprises 28 types (I–XXVIII) that are derived from a total of 46 α-chains across the superfamily (Fig. 1) (Ricard-Blum, 2011; Kadler et al., 2007; Ricard-Blum and Ruggiero, 2005). Invertebrates generally contain collagen IV, XV or XVIII, some fibrillar collagens, as well as some fibril-associated collagens with interrupted triple helices (FACITs) (Fidler et al., 2014, 2017; Fahey and Degnan, 2010; Meyer and Moussian, 2009; Boot-Handford and Tuckwell, 2003; Whittaker et al., 2006; Kadler et al., 2007). Among these collagens, type IV is the evolutionarily most ancient, based on recent studies of non-bilaterian animals (sponges, ctenophores, placozoans and cnidarians) and unicellular groups (Fidler et al., 2017; Grau-Bove et al., 2017) (Fig. 1).

Collagens are the most abundant protein in the human body (Kadler et al., 2007; Shoulders and Raines, 2009). They occur as diverse supramolecular assemblies, ranging from networks to fibrils, and broadly function in structural, mechanical and organizational roles that define tissue architecture and influence cellular behavior (Shoulders and Raines, 2009; Ricard-Blum, 2011; Ricard-Blum and Ruggiero, 2005). Defects in collagens underlie the cause of almost 40 human genetic diseases, affecting numerous organs and tissues in millions of people worldwide (summarized in Table 1).

Disease pathogenesis typically involves genetic alterations of the triple helix, a unique structure that is a hallmark feature common to all collagens. The triple helix bestows exceptional mechanical resistance to tensile forces and a capacity to bind a plethora of macromolecules. Yet, there is a gap in our current knowledge in understanding the mechanisms of how a triple helix encodes and utilizes information in building supramolecular assemblies on the outside of cells. Here, we present collagen IV, the most ancient of the collagen superfamily, and argue that it is ideally suited to serve as an archetype for investigating and describing core functions of a triple helix.

The chemical structure of the triple helix was determined through the seminal work of structural biologists and chemists over the last century (see Box 1 in supplementary material). Its unique structure bestows upon collagens an exceptional mechanical resistance to tensile forces and a plethora of organizing information for building an ECM (Fig. 2). The triple helix presents all residues, except glycine (Gly), on its surface, which is the most economical and robust way to encode binding motifs of any protein structure. Moreover, the triple helix exhibits extensive post-translational modifications (PTMs), such as hydroxylation, glycosylation and phosphorylation, adding – in tandem – a secondary layer of information in addition to its amino acid (aa) code (Yamauchi and Shiiba, 2008). These PTMs confer even more diversity with tissue-specific and disease-specific variations, even amongst identical types of collagen (Pokidysheva et al., 2013). Furthermore, additional collagen modifications are mediated by specific extracellular enzymes, such as peroxidasin and lysyl oxidases-like proteins (LOXLs) for crosslinking, and Goodpasture antigen-binding protein (COL4A3BP, hereafter referred to as GPBP) and other extracellular kinases for phosphorylation (Bhave et al., 2012; Añazco et al., 2016; Revert et al. 1995, Raya et al. 1999; Yalak and Olsen, 2015). Non-enzymatic modifications, such as glycation, oxidation or chlorination, add even more complexity (Brown et al., 2015). Together, these modifications may serve as regulatory mechanisms on the outside of cells that may instruct cell behavior and influence tissue architecture and stability (Yalak and Olsen, 2015; Pedchenko et al., 2010).

To fully appreciate the capacity and versatility of the triple helix, one should consider the building blocks (polypeptide chains) of this structure. These chains are simple ‘rope-like’ structures built from proline residues. In aqueous solutions, such a polyproline chain adopts a polyproline type II (PPII) helix in a left-handed conformation (Adzhubei et al., 2013) (Fig. 2A). A unique feature of this structure is that it is non-extensible (Okuyama et al., 1981). Such simple ‘ropes’ occur in the extracellular space of plants (Lamport, 1974) and have been identified as extensions of triple helices within mini-collagens of cnidarians (e.g. Hydra) (Holstein et al., 1994), where they perform structural and mechanical roles. To allow for the formation of a triple helix from this simple rope, every third proline residue of the PPII helix is replaced with a glycine residue – the smallest aa residue as it lacks a side chain. This endows the structure with flexibility and sufficient space to tightly pack three chains (Fig. 2A) that are each a left-handed helix, wind around each other with a shift in one residue, so that the smaller glycine residues are buried inside the triple helix, whereas only proline residues are exposed on the surface (Fig. 2A) (Ramachandran and Kartha, 1954; Rich and Crick, 1955; Okuyama et al., 1981). Moreover, as these three chains wind together forming a triple helix, each individual left-handed chain adopts a right-handed superhelix. The unique role of the glycine residues in the packing of the collagen triple helix explains the adverse effects of mutations at these positions, as any other residue in place of glycine will distort its tight packing (Bella et al., 1994).

Three main changes are achieved upon the transition from a single polyproline helix to a triple helix: (i) higher bending rigidity, (ii) a less accessible chain backbone that is, thus, less prone to proteolysis and, (iii) the ability to essentially accept any aa in place of the proline residues at position X and Y without any significant destabilization to the triple helix structure (Fig. 2B). Moreover, proline and other aa within the triple helix remain accessible to solvent and, thus, their numerous known PTMs are possible without disturbing the native helix structure. For example, probably the most important PTM is the 4(R)-hydroxylation of proline residues position Y, because it substantially increases the thermal stability of the collagen triple helix (Sakakibara et al., 1973). Although all proline and hydroxyproline residues are essential for the stability of a PPII helix, ∼65% of aa at these positions can vary in the native triple helix, while still maintaining its stability. Thus, the triple helix, in contrast to a single PPII helix, confers additional capacity to specify binding partners.

Once assembled from three superhelical α-chains, the collagen triple helix provides several distinct sites to tether macromolecules through three modes of binding motifs (single, double and triple α-chain) with further three levels of variations to these modes (mode 1, 2 and 3) (Fig. 2B). The three binding modes operate by utilizing one, two or all three distinct chains of the triple helix in different combinations to directly bind a molecule (Fig. 2B; modes 1–3). The three levels of variation to these modes add additional diversity to binding specificity on top of involving either one, two or three chains. The first level stems from the variation that is provided by the ≤20 possible aa residues that can occupy the variable positions X and Y of each tripeptide; this results in extensive structural variability across the three chains, as well as within each Gly-X-Y tripeptide of the same chain (Fig. 2B; level 1). The second level confers even more diversity because of the proclivity of collagens to become post-translationally modified and change their structure (Fig. 2B; level 2). The third level utilizes chain staggering that occurs during triple helix assembly between either two or three chains and enables new combinations of chains that can give rise to additional binding motifs (Fig. 2B; level 3).

In addition to these binding specificities, the triple helix also possesses other structural features that underlie its biological function. It is non-stretchable along its longitudinal axis, providing great tensile strength to withstand all physiological mechanical loads and stresses in our body. It also confers resistance to proteases – making collagens some of the most long-lived proteins – as well as to a wide range of pH values in order to withstand adverse conditions (Steven, 1965; Hafter and Höermann, 1963; Grant and Alburn, 1960; Drake et al., 1966; Uzawa et al., 1998; Pokidysheva et al., 2013; Hudson et al., 2017; Eyre et al., 2011). Furthermore, the triple helix exhibits variable lengths among collagen suprastructures, such as networks, fibrils and filaments (Ricard-Blum, 2011). For example, the triple helix length of fibrillar collagens results from multiple duplications of exons that are either 54 or 45 base pairs in length and encode Gly-X-Y (Yamada et al., 1980; Exposito et al., 2010). Moreover, the terminal and internal incorporation of non-triple helical sequences (e.g. interruptions) in all collagen types extends their functional capacity (Ricard-Blum, 2011; Khoshnoodi et al., 2006). Collectively, the binding motifs (see above) together with these features of a collagen triple helix underlie the diversity of biological activities of the diverse supramolecular assemblies of collagens.

The last common ancestor to animals, the Urmetazoan, almost certainly reproduced by gametogenesis, underwent gastrulation during early development, had the ability for cells to differentiate both during development and as stem cells, and comprised an epithelial layer of cells forming the body of the animal – features that are still fundamental to extant animals (Richter and King, 2013; King and Rokas, 2017). Importantly, these cellular activities ultimately required the invention of an ECM to provide a substrate for attachment and signaling cues to regulate cell behavior and function in tissue genesis and homeostasis (Abedin and King, 2010). The appearance of a specialized form of ECM, the BM, coincided with the transition to multicellularity. The BM functions in several cellular activities, including migration, adhesion, delineation of apical–basal polarity and modulation of differentiation during development (Petersen et al., 1992; Lukashev and Werb, 1998; Daley and Yamada, 2013; Hynes, 2009, 2012; Yurchenco, 2011; Ozbek et al., 2010; Henry and Campbell, 1998).

Importantly, understanding the makeup of BMs between different animals sheds light on the functions of proteins in the evolution of animal multicellularity and tissues. BMs are composed of numerous proteins, vary between animals. The BM of bilaterian animals (e.g. human, fly, C. elegans, sea urchin) is composed of several proteins, including collagen IV, laminin, perlecan, nidogen, fibronectin, proteoglycans, peroxidasin, GPBP, and collagens XV and XVIII (Hynes, 2012; Yurchenco, 2011; Jayadev and Sherwood, 2017). Potentially, there are many more components of the BM (Chew and Lennon, 2018). Among these, collagen IV is a major component that is conserved among animal phyla (Fidler et al., 2014). In a recent study, we described collagen IV at the origins of animal multicellularity and in tissue evolution, as revealed by close examination of sponges, ctenophores and other non-bilaterally symmetrical animals (Fidler et al., 2017). Ctenophores and sponges have been established as the two most likely candidates to be the sister-groups to the rest of animals, based on phylogenetic analyses of genomic and transcriptomic data and cell-type evolution (Ryan et al., 2013; Moroz et al., 2014; Pisani et al., 2015; Whelan et al., 2015; Telford et al., 2016; King and Rokas, 2017; Feuda et al., 2017). Our genomic analysis of the extracellular matrix components within the ctenophores and Homoscleromorph sponges revealed a BM ‘toolkit’ consisting of just collagen IV and laminin (Fidler et al., 2017). However, the demosponges, a sponge class, lack both laminin and classic collagen IV but do contain spongins, which are short collagen IV variants (Exposito et al., 1991; Aouacheria et al., 2006; Fidler et al., 2017). The order in which these variants and collagen IV first appeared is still unknown. Despite containing fewer BM proteins than bilaterians, many sponges and ctenophores form classic BMs (Fidler et al., 2017; Boute et al., 1996; Leys et al., 2009).

Importantly, comparison of BM components of animals with those of unicellular lineages is key to determining their importance during the transition to multicellularity. Within choanoflagellates, the closest relatives of animals, no complete ECM proteins exist; yet, domains that are characteristic of laminins and short collagenous Gly-X-Y repeats are present (King et al., 2008; Fahey and Degnan, 2012; Fidler et al., 2017). Interestingly, a recent study reported the discovery of a collagen IV-like gene in the filasterean, Ministeria vibrans, a unicellular lineage that diverged prior to choanoflagellates and animals (Grau-Bove et al., 2017). This finding indicates that collagen IV has a premetazoan ancestry and a function for single cells. Collectively, these findings suggest that collagen IV played a role in the transition from unicellular organisms to multicellular animals (Grau-Bove et al., 2017; Fidler et al., 2017) (Fig. 1). Therefore, we consider collagen IV as an archetype of collagens to describe the fundamental features of a triple helix that underlie biological functions.

Collagen IV forms a network that functions as a scaffold within BMs (Fig. 3). These scaffolds provide tensile strength, connect adjacent cells and organize supramolecular protein assemblies that are able to influence cell behavior (Wang et al., 2008; Emsley et al., 2000; Parkin et al., 2011; Cummings et al., 2016; Vanacore et al., 2009). Network assembly begins with the intracellular formation of triple-helical protomers comprising three α-chains (Brown et al., 2017). Protomer assembly is directed and regulated by the non-collagenous (NC)1 recognition modules, which are located at the C-terminus of each α-chain; this is followed by the twisting together of the collagenous domains – the Gly-X-Y repeats – into a triple helix. In mammals, three distinct protomers (α112, α345 and α556) are formed from six genetically distinct α-chains (α1–α6), thereby forming three distinct networks (Khoshnoodi et al., 2008). Once secreted into the extracellular space, protomers adjoin via their NC1 domains and the N-terminal 7S domains (Fig. 3A). NC1-domain association is mediated by extracellular Cl⁻, which activates a molecular switch that enables adjacent protomers to adjoin two NC1 domain trimers (Fig. 3A) (Cummings et al., 2016). 7S domains assemble into a complex of four independent protomer 7S domains (Añazco et al., 2016). Upon association mediated through NC1- and 7S-domains, the collagen IV networks are reinforced through covalent crosslinks at both the NC1- and 7S-domain interfaces. NC1-domain hexamers are stabilized through sulfilimine (-S=N-) double bonds, crosslinks that are induced by peroxidasin (PXDN) – an animal heme peroxidase embedded in the BM – and by Br⁻ (Vanacore et al., 2009; Bhave et al., 2012; McCall et al., 2014). Concurrently, 7S dodecamers are crosslinked by lysyl oxidase-like protein 2 (LOXL2) (Añazco et al., 2016).

Collagen IV networks function as smart scaffolds, bestowing BMs with several capabilities (Fig. 3B). Via the triple helix, scaffolds tether different extracellular molecules, i.e. laminins, proteoglycans, perlecans, nidogens, growth factors and extracellular enzymes (such as peroxidasin and lysyl oxidase-like protein 2) (Parkin et al., 2011; Hynes, 2012; Bhave et al., 2012; Añazco et al., 2016). The information for the tethering of these molecules is encoded at sites within the triple helix, and depends on the 20 variable aa residues within a single chain, or a combination of one, two or three chains, chain stagger and post-translational modifications (see Fig. 2). The tethering of these molecules at specific sites along the triple helix spatially organizes binding partners (Fig. 3B), which – in turn – forms a diverse multi-protein complex that represents a BM. The distribution of binding partners within a BM is not a static arrangement, and can be dynamically regulated throughout early development and beyond (Inman et al., 2015; Jones-Paris et al., 2016). The resulting scaffold, populated with macromolecules bound to the triple helix, provides tensile strength to tissues, attaches to cells through cell-surface receptors and influences cell behavior in tissue development, function and regeneration (Eble et al., 1993; Emsley et al., 2000; Khoshnoodi et al., 2008; Valiathan et al., 2012; Fu et al., 2013).

The biological importance of the triple helix is displayed in several ways. It is an ancient structure that is conserved between animals and is expressed ubiquitously in their ECMs. The triple helix is a common protein structure of numerous and distinct collagen suprastructures with diverse biological activities, including the network-forming collagens (IV, VIII, X), the FACITs (IX, XII, XIV, XVI, XIX, XX, XXI, XXII), fibrils (I, II, III, V, XI, XXIV, XXVII), anchoring fibrils (VII) and beaded filaments (VI) (Fig. 1). There are almost 40 diseases wherein mutations of glycine residues affect multiple tissues and organs in millions of people (Fig. 4A,B, Table 1). Glycine residues are crucial for the structural integrity of the triple helix (see Figs 2 and 4A); therefore, mutated collagen molecules can assemble into faulty fibrils, networks, and other assemblies can cause tissue dysfunction.

As examples for such glycine mutations, osteogenesis imperfecta (OI; also known as brittle bone disease), and Alport syndrome are two collagen-dependent genetic disorders that are well-studied (Fig. 4C, Table 1). For OI, over 800 mutations in collagen I have been described (Marini et al., 2007; Forlino et al., 2011; Forlino and Marini, 2016). Approximately 80% are glycine mutations that occur in the triple helix (Forlino et al., 2011). However, such mutations, depending on the nature of substitution as well as its location, lead to different degrees of post-translational modifications and structural destabilization of the triple helix. For example, substitutions in the first 200 residues of collagen I are non-lethal, whereas there are two regions (helix positions 691–823 and 910–964), in which substitutions can cause lethality because they align with main ligand-binding sites for integrins, matrix metalloproteinases, fibronectin and cartilage oligomeric matrix protein (Marini et al., 2007). In Alport syndrome, the collagen IV scaffold is mutated, which leads to progressive organ failure in kidney, ear and eye (Williamson, 1961; Hudson, 2004; Hudson et al., 2003; Cosgrove et al., 2007; Cosgrove and Liu, 2017; Kashtan, 1993; Chew and Lennon, 2018). Over 1700 mutations have been found to occur in the collagen IV scaffold that is composed of α3, α4 and α5 chains. Of these, ∼85% are glycine substitutions that are located in the triple helix (Hertz et al., 2012) (Table 1).

In summary, the many collagen diseases that involve glycine mutations directly demonstrate the essentiality of the triple helix for tissue architecture and function. Its essentiality is further supported experimentally by collagen knockout studies in mice (Table 1). The respective mouse phenotypes include developmental lethality upon knockout of collagen types I, II, III, IV, V, VII, XII, XVII, XIX, XXVI and XXVII, muscle deformities in that of types VI, XII, XIII, XV and XXV, and bone and cartilage deformities in that of types IX and XI (see references in Table 1).

The triple helix is unique among all other protein structures – globular or fibrous – in its capacity to encode vast amounts of information that is available on its surface for utilization on the outside of cells (Fig. 5A). The triple helix, arranged in various patterns forming diverse supramolecular scaffolds, tethers and spatially organizes macromolecules, thus providing tensile strength to tissues and influencing cell behavior. This unique structure of the triple helix with its encoded degree of information evokes an analogy to the DNA double helix (Fig. 5A).

The biological importance of the triple helix is also evident from the almost 40 genetic diseases and its ubiquitous presence in animals. The triple helix was co-opted in the form of collagen IV to enable the evolutionary transition from unicellular organisms to multicellular animals, and the triple helix was also adapted to give rise to all the other members of the diverse collagen superfamily, thereby enabling the evolution of tissues and organs (Fig. 5B). Thus, the triple helix represents a fundamental protein structure that nature adapted for building an extracellular matrix.

There are many provocative and unanswered questions regarding the function of triple helices in suprastructures and their dysfunction in diseases. These include: (i) What are the unknown sites encoded in the triple helix for binding partners as exemplified by those in collagen I fibrils and collagen IV networks (Di Lullo et al., 2002; Parkin et al., 2011)? (ii) How is information in the triple helix used to assemble suprastructures (Orgel et al., 2011)? (iii) What are the mechanisms of how triple helices within suprastructures influence cell function? (iv) What is the impact of PTMs on the structure and function of the triple helix? (v) How do genetic mutations in the triple helix cause tissue dysfunction? (vi) How do genetic backgrounds affect phenotype variations? (vii) What are the mechanisms for the function of the triple helix in the transition from unicellular organisms to multicellular animals.

To answer such fundamental questions, collagen IV is an ideal archetype because it is the most ancient of the collagens, and it is present in unicellular organisms and non-bilaterian animals (Fig. 5B). Furthermore, recent studies have revealed that, in some organisms, collagen IV also occurs in the absence of a BM, such as in certain ctenophores and sponges, placozoans and the unicellular filasterean Ministeria vibrans (Fidler et al., 2017; Grau-Bove et al., 2017; Schierwater et al., 2009). Moreover, a recent study of Drosophila development provided evidence that, in the absence of BM, collagen IV has a role in intercellular adhesion and pro-growth signaling (Dai et al., 2017; Zajac and Horne-Badovinac, 2017). Together, these recent studies clearly indicate that collagen IV can have a direct role in influencing cell behavior, outside of the BM; thus, the core functions of the collagen triple helix, as well as its role in the transition from unicellular organisms to multicellular animals, can be addressed by comparative studies in these organisms. Such knowledge may provide insights into unknown roles of collagens in cell biology, disease pathogenesis and evolution of animals.

This article is the culmination of a ten-year Aspirnaut expedition to the dawn of the animal kingdom in search of the evolutionary origin of collagen IV. The expedition involved over 100 middle school, high school, undergraduate and graduate students from disadvantaged backgrounds, and was championed by Aaron Fidler, Julie Hudson and Billy Hudson. We are grateful for the contributions of all of these students in the discovery that collagen IV occurred in all animals, and the fundamental importance of the triple helical protein structure in animal evolution as presented in this capstone article.