Intrinsic disorder in spondins and some of their interacting partners

ABSTRACT

Spondins, which are proteins that inhibit and promote adherence of embryonic cells so as to aid axonal growth are part of the thrombospondin-1 family. Spondins function in several important biological processes, such as apoptosis, angiogenesis, etc. Spondins constitute a thrombospondin subfamily that includes F-spondin, a protein that interacts with Aβ precursor protein and inhibits its proteolytic processing; R-spondin, a 4-membered group of proteins that regulates Wnt pathway and have other functions, such as regulation of kidney proliferation, induction of epithelial proliferation, the tumor suppressant action; M-spondin that mediates mechanical linkage between the muscles and apodemes; and the SCO-spondin, a protein important for neuronal development. In this study, we investigated intrinsic disorder status of human spondins and their interacting partners, such as members of the LRP family, LGR family, Frizzled family, and several other binding partners in order to establish the existence and importance of disordered regions in spondins and their interacting partners by conducting a detailed analysis of their sequences, finding disordered regions, and establishing a correlation between their structure and biological functions.

Keywords:

• intrinsically disordered protein
• intrinsically disordered protein regions
• posttranslational modifications
• protein function
• protein structure
• spondin

Introduction

Until recently, it was believed that all proteins have well-defined 3 dimensional structures crucial for their unique biological functions. However, with recent advances in protein science, it has been proven that some genes in the eukaryotic genomes encode entire proteins or large segments of proteins that lack a well-structured 3-dimensional fold.^1-14 These intrinsically disordered proteins (IDPs) and intrinsically disordered protein regions (IDPRs) have important biological roles and are crucial for many cellular processes, such as signaling, transcriptional and translational activities.^{3,4,8,10-12,15-20} Some IDPRs function as linkers between ordered domains, many of them frequently serve as sites of posttranslational modifications, or as regions affected by disease-related mutations, gene truncations or translocations.^21-28

Thrombospondins (TSPs) are multimeric multidomain secreted glycoproteins with antiangiogenic functions found in the extracellular matrix (ECM).^29,30 This family of the matricellular glycoproteins includes 5 different members, namely, TSP1, TSP2, TSP3, TSP4, and TSP5, which is further subdivided to groups A (TSP1 and TSP2) and B (TSP3, TSP4, and TSP5). Just like many other ECM proteins, TSPs are large modular proteins (whose length in humans ranges from 757 to 1,170 residues) that contain series of repeated domains of different type. For example, members of the group A, TSP1 (UniProt ID P07996) TSP2 (UniProt ID P35442), being the longest members of the TSP family with 1,170 and 1,171 residues, respectively, contain an N-terminal laminin G-like domain, a VWFC domain, 3 TSP1 repeats, 3 epidermal growth factor-like repeats (EGF-like, also known as TSP2 repeats), 8 aspartic acid-rich TSP3 repeats, and a TSP C-terminal domain. The group B members, TSP3 (UniProt ID: P49746), TSP4 (UniProt ID: P35443), and TSP5 (also known as cartilage oligomeric matrix protein (COMP), UniProt ID: P49747), are noticeably shorter (there are 956, 961, and 757 residues in TSP3, TSP4, and COMP, respectively) and have domain structure that is different from the domain organization of the group A TSPs. Namely, they contain unique N-terminal laminin G-like domain, lack the VWFC domain and TSP1 repeats, contain 4 copies of EGF-like domains, and 8 copies of the TSP3 repeats. Furthermore, group A and B members have different oligomeric structures, with the group A TSPs being assembled as homo-trimer and the group B TSPs existing as homo-pentamers.³¹ These modular proteins act by bringing together cytokines, growth factors, other matrix components, membrane receptors, and extracellular proteases.^29-31

Spondins are a group of proteins from the thrombospondin superfamily. They are found in eukaryotic organisms and are grouped into different families, such as the R-spondin, the subcommissural organ (SCO)-spondin, the M-spondin (mindin), and the F-spondin. Spondins are engaged in various vital biological functions, such as regulators of Wnt signaling (R-spondins),^32-40 regulation of the developing skeleton, limb formation, and the maintenance of adult bone mass (R-spondins),^39,41 regulation of stem cells (R-spondin),^37,42,43 neuron/glia interaction and neuronal differentiation and development (SCO-spondin),^44,45 interaction with the β-amyloid precursor protein (APP) and its controlled proteolysis (F-spondin),^46,47 regulation of the accurate path-finding of embryonic axons (F-spondin),⁴⁸ and promotion of the neurite outgrowth and inhibition of the angiogenesis (F-spondin and mindin).⁴⁸ Being the members of the TSP family, these proteins have complex modular structures. Multifunctionality of these proteins, their ability to interact with numerous partners, their modular structure, and the presence of multiple aspartic acid-rich repeats suggest that spondins might belong to the family of hybrid proteins containing ordered domains and functional IDPRs.

The goal of this study was to establish the existence and importance of IDPRs in spondins and their interacting partners in human, and to conduct a detailed analysis of their sequences, find disordered regions, and establish a correlation between their structure and biological functions. Although most of the entities analyzed in this study are extracellular proteins possessing signal peptides that are removed following passage through the membrane, these secretion signals were included in the analysis. In other words, we analyzed entire protein sequences as reported in the corresponding UniProt entries.

R-spondin family

In human, there are 4 R-spondin proteins, which are secreted agonist of the canonical Wnt/β-catenin signaling pathway.^32-40 These proteins have molecular masses of approximately 35 kDa and are characterized by the presence of 2 N-terminal furin-like repeats, which are necessary for Wnt signaling. R-spondins can enhance responses to low-dose Wnt protein and also serves as activators of a canonical Wnt signaling pathway, where they act as ligands for the LGR4-6 receptors. Being potent stimulators of adult stem cells proliferation in vivo and in vitro, R-spondins have strong potential for therapeutic use in regenerative medicine.

R-spondin 1

R-spondin 1 is also referred to as Roof plate-specific spondin-1. This protein is encoded by RSPO1 gene located at the position 1p34.3 of the chromosome one, and is present as 3 isoforms in humans, a full-length canonical form (or isoform #1; UniProt ID: Q2MKA7-1) with sequence length of 263 residues, an isoform #2 (UniProt ID: Q2MKA7-2) that is characterized by MRLGLCVVALVLSWTHLTISSRGIKGKRQRRI → MIFRV substitution within the N-terminal region (residues 1–32), and an isoform #3 (UniProt ID: Q2MKA7-3) that misses residues 146–208. There are 5 functional domains in the canonical form of this protein, a signal peptide sequence at the N-terminus for secretion (residues 1–20), 2 cysteine-rich furin-like repeat domains (domains Fu1 and Fu2, residues 34–85 and 91–135, respectively), a TSP1 repeat domain (TSR, residues 147–207), and a basic amino acid-rich (BR) domain at the C-terminus (residues 208–263). Therefore, although alternative splicing does not affect the R-spondin 1 (Rspo1) N-terminal region with Fu1 and Fu2 domains, entire TSP type-1 domain is absent in its isoform #3, suggesting that this Rspo1 proteoform cannot interact with heparin sulfate proteoglycans (see below), and a signal peptide is removed in isoform #2, suggesting that this proteoform cannot be exported.

Rspo1 is known to strongly promote proliferation of the Wnt-dependent intestinal-crypt stem cell compartment,^49,50 with this activity being primarily attributed to the Fu1 and Fu2 domains of this protein,⁵¹ which are involved in direct physical interaction with the members of the leucine-rich repeat-containing G protein-coupled receptors 4–6 (LGR4–LGR6).^52-54 High affinity binding of R-spondins to LGR5 (as well as its homologs LGR4 and LGR6) mediates R-spondin contribution to the canonical Wnt pathway.^52-54 The TSR domain is responsible for interaction with heparin sulfate proteoglycans (HSPGs).⁵⁵

Besides interaction with the LGR4-6 receptors, Rspo1 regulates the canonical Wnt/β-catenin dependent pathway and non-canonical Wnt signaling by acting as an inhibitor of a transmembrane E3 ubiquitin ligase, zinc and ring finger 3 (ZnRF3), and the E3 ubiquitin-protein ligase RING finger protein 43 (RNF43), which are critical negative feedback regulators of the Wnt pathway, which function through promoting ubiquitination and degradation of Wnt receptors, and thereby playing a crucial role in the control of cancer development.³⁸ Therefore, Rspo1 serves as an important member of the R-spondin-ZnRF3/RNF43 signaling module that plays a crucial role in the control of Wnt signaling.³⁸ In fact, RSpo1 can simultaneously bind to the extracellular domains of ZnRF3/RNF43 and LGR4/5.⁵⁶ Formation of this complex induces auto-ubiquitination and membrane clearance of ZnRF3/RNF43, leading to the increased levels of frizzled (Fzd). This Rspo1-based regulation of the Fzd turnover allows R-spondin to control both Wnt/β-catenin and Wnt/PCP signaling pathways. The formation of the tripartite ZnRF3/RNF43-Rspo1-LGR4/5 complex is explained by the fact that Rspo1 utilizes its Fu2 domain for interaction with LGR4/5 and binds to ZnRF3/RNF43 via its Fu1 domain.³⁸ This tripartite complex represents a complex functional state of Rspo1, with LGR4/5 and ZnRF3/RNF43 acting as the engagement and the efficacy receptors for Rspo1, respectively.³⁸ Furthermore, Rspo1 can be engaged in the physical binary interactions with the extracellular domains of frizzled-8 (Fzd8)⁵⁷ and low-density lipoprotein receptor-related protein 6 (LRP6),⁵⁸ which serves as the Wnt co-receptor, without formation of a ternary complex with these 2 proteins.⁵⁷ Furthermore, Rspo1 partake in various biological processes including canonical male meiosis, positive regulation of protein phosphorylation, regulation of gene expression, regulation of male germ cell proliferation, regulation of receptor internalization.^36,39

Structural information is available for the Fu1/Fu2-containing region of Rspo1 (residues 40–127; e.g., see PDB ID: 4BSP).⁵⁹ Although longer regions of Rspo1 are typically used in crystallization studies (residues 31–146), no coordinates have been determined for their N- (residues 31–39) and C-terminal regions (residues 128–146). Recent structural analysis revealed that in its unbound form, the Fu1/Fu2 region of Rspo1 is characterized by the non-equivalent Fu domains, where the Fu1 domain has 3 cysteine-knotted β-hairpins, whereas there are 2 such hairpins in the Fu2 domain, with the C-terminal hairpin being missing from the crystal structure, likely due to the high conformational flexibility.⁵⁹ Unbound Fu1/Fu2 region is organized in such a way that the sets of Fu1 and Fu2 β-hairpins are oriented at ∼90° to each other, with each Fu domain forming a leaflet consisting of 3 (Fu1) or 2 (Fu2) disulfide bond-connected β-hairpins.⁵⁹ However, binding to the LGR5 induced noticeable changes in the Fu1/Fu2 region, especially in its Ru2 part. Fig. 1 shows that the overall shape of the LGR5-bound TSR was flattened because the 2 sets of β-hairpins from the Ru1 and Ru2 domains were aligned due to a twist caused by the rotation around the TSR longitudinal axis. Furthermore, the C-terminal part of the Fu1/Fu2 region became visible upon its binding to LGR-5, indicating dramatic binding-induced reduction in the conformational flexibility of this region.⁵⁹

Figure 1. Structural characterization of the Fu1/Fu2-containing region of Rspo1 in its unbound (PDB ID: 4BSP-A, red ribbon and 4BSO-A, blue ribbon) and bound forms (PDB ID: 4BSR-C, yellow ribbon and 4BSR-D, cyan ribbon).⁵⁹ Results of multiple structure alignment are shown in the middle of the plot, whereas individual chains are located around. To show location of each individual chain within the aligned structural ensemble, the remaining structures are made transparent. Multiple structural alignment was conducted using the MultiProt tool (http://bioinfo3d.cs.tau.ac.il/MultiProt/).¹⁹⁸ Resulting structures are drawn using the visual molecular dynamic tool VMD.¹⁹⁹

Figure 2 represents the results of intrinsic disorder analysis in human Rspo1 (UniProt ID: Q2MKA7) and shows that this protein is expected to have several functional disordered regions. For example, a linker connecting signaling peptide to the rest of Rspo1 is disordered, likely to facilitate the accessibility of site of a proteolytic attack leading to the release of the signaling peptide and generating mature Rspo1 (residues 21–263). Although both Fu1 and Fu2 domains are mostly ordered, both a TSR and a BR domains are predicted to be mostly disordered. BR domain is predicted to have 2 potential disorder-based protein binding sites identified by the ANCHOR algorithm, that acts on the hypothesis that long regions of disorder include localized potential binding sites which are not capable of folding on their own due to not being able to form enough favorable intrachain interactions, but can obtain the energy to stabilize ordered conformation via interaction with a globular protein partner.^60,61 Furthermore, it is known that human Rspo1 can be subjected to several posttranslational modifications (PTMs), with sites of these PTMs being located within the IDPRs of this protein. For example, it was shown that the N-glycosylation of human Rspo1 at Asn137 (which is a residue located at the C-terminus of the Fu2 domain of this protein and is predicted to be disordered, being characterized by the mean disorder score of 0.53) is required for the efficient secretion and stability of this protein, but does not play a role in interaction of Rspo1 with heparin sulfate proteoglycans.⁶² A single phosphorylation site (Thr253) is located within the disordered C-tail of human Rspo1. Since this site is positioned within one of the MoRFs, it is likely that it plays a role in regulation of the disorder-based interactivity of this protein.

Figure 2. Multifactorial computational disorder analysis of in human Rspo1. (A) Intrinsic disorder profile of Rspo1 (UniProt ID: Q2MKA7) generated by the superposition of the outputs of PONDR® VLXT, PONDR® FIT, PONDR® VL3, PONDR® VSL2, IUPred_short and IUPred_long and a consensus disorder profile calculated by averaging disorder profiles of individual predictors. (B) Intrinsic disorder propensity and some important disorder-related functional information generated for human Rspo1 by the D²P² database (http://d2p2.pro/).²⁰⁰ Here, complementary disorder evaluations together with some disorder-related functional information are shown. To this end, the D²P² database uses outputs of IUPred,¹⁷⁹ PONDR® VLXT,²⁰¹ PrDOS,²⁰² PONDR® VSL2B,^183,184 PV2,²⁰⁰ and ESpritz.¹⁸⁰ Positions of disorder-based interactions sites (MoRFs) and sites of curated posttranslational modifications are also shown.

Figure 3A shows the Rspo1 interactome evaluated by Search Tool for the Retrieval of Interacting Genes; STRING, http://string-db.org/, which generates a network of predicted associations based on predicted and experimentally-validated information on the interaction partners of a protein of interest.⁶³ In the corresponding network, the nodes correspond to proteins, whereas the edges show predicted or known functional associations. Seven types of evidence are used to build the corresponding network, where they are indicated by the differently colored lines: a green line represents neighborhood evidence; a red line - the presence of fusion evidence; a purple line - experimental evidence; a blue line – co-occurrence evidence; a light blue line - database evidence; a yellow line – text mining evidence; and a black line – co-expression evidence.⁶³ In our analysis, the most stringent criteria were used for selection of interacting proteins by choosing the highest cut-off of 0.9 as the minimal required confidence level. According to these criteria, human Rspo1 interacts with other members of the R-spondin family (Rspo2, Rspo3, and Rspo4), leucine-rich repeat containing G protein-coupled receptors 4, 5, and 6 (LGR4, LGR5, and LGR6), low density lipoprotein receptor-related protein 6 (LRP6), zinc and ring finger 3 (ZnRF3), ring finger protein 43 (RNF43), ribosomal protein S27a (RPS27A), ubiquitin A-52 residue ribosomal protein fusion product 1 (UBA52), ubiquitin B (UBB), ubiquitin C (UBC), dickkopf 1 homolog (DKK1), wingless-type MMTV integration site family, member 4 (WNT4), SRY (sex determining region Y)-box 9 (SOX9), and forkhead box L2 (FOXL2) (see Fig. 3A).

Figure 3. Interactome generated by Search Tool for the Retrieval of Interacting Genes; STRING, http://string-db.org/ for: (A) Human Rspo1 (UniProt ID: Q2MKA7); (B) Human Rspo2 (UniProt ID: Q6UXX9); (C) Human Rspo3 (UniProt ID: Q9BXY4); and (D) Human Rspo4 (UniProt ID: Q2I0M5).

R-spondins 2, 3, and 4

Rspo2. Protein R-spondin 2 (Rspo2) is encoded by the RSPO2 gene located at the 8q23.1 position of the chromosome eight. Rspo2 is also known as roof plate-specific spondin-2. Similar to Rspo1, this protein is involved in a wide range of biological processes, such as bone mineralization, dopaminergic neuron differentiation, embryonic forelimb morphogenesis, embryonic hind-limb morphogenesis, epithelial tube branching involved in lung morphogenesis, lung growth, negative regulation of odontogenesis of dentin-containing tooth, osteoblast differentiation, and trachea cartilage morphogenesis.³⁶ Rspo2 also plays key role in regulation of the canonical Wnt pathway by passing signal into a cell through cell surface receptors, causing an accumulation of β-catenin in the cytoplasm and its eventual translocation into the nucleus.³⁵

Rspo2 has a signal peptide (residues 1–21), 2 Fu domains (residues 39–89 and 90–134, respectively), a TSR1 domain (residues 144–204), and a C-terminal basic amino acid rich domain (residues 205–243). Rspo2 (UniProt ID: Q6UXX9) has 3 proteoforms generated by alternative splicing. The full-length canonical isoform #1 (UniProt ID: Q6UXX9-1) has 243 residues. Isoform #2 (UniProt ID: Q6UXX9-2) is characterized by missing entire N-terminal domain (residues 1–67), whereas in the isoform #3 (UniProt ID: Q6UXX9-3) the 32–95 region is shrunk to one glycine residue and a residue 143 is missing.

Rspo3. R-spondin 3 protein (Rspo3) is encoded by the RSPO3 gene located at the 6q22.33 position of the chromosome six. RSPO3 has a 5-exon organization defining the presence of alternatively spliced isoforms and the domain structure of the protein. It is another member of the 4 vertebrate proteins that are secreted as agonists of the canonical Wnt/β-catenin signaling pathway (plays a role in cellular proliferation, differentiation and stem cell maintenance) and belongs to the superfamily of thrombospondin type 1 repeat (TSR1)-containing proteins.⁶⁴ It is approximately 35 KDa in length and characterized by the presence of 2 N-terminal furin-like repeats that are needed for Wnt signal potentiation and are present in important receptors for growth factors, such as EGF, HGF, and insulin; there are 41 proteins that contains TSR1 domains in the human genome.⁶⁵ As other members of the spondin family, this protein is characterized by the presence of the TSR1 domain that can be found in secreted proteins or extracellular portion of the transmembrane proteins. In Rspo3, TSR1 is located toward to the C-terminus and binds glycosaminoglycan and/or proteoglycans.⁶⁶ Rspo3 acts an upstream of Wnt pathway and in studying of the vasculogenesis and angiogenesis signals, the spondin-targeted disruption was shown to lead to severe vascular defects in the placenta.⁶⁷ Lspo3 may also serve as a negative regulator of the TGF-β pathway,^40,54,68 and as a key regulator of angiogenesis by activating the non-canonical Wnt signaling pathway in endothelial cells needed for the control of vascular stability and pruning.⁶⁷

The amino acid sequence of Rspo3 starts with a signal peptide (residues 1–21), followed by 2 Fu domains (residues 35–86 and 92–135, respectively), a TSR1 domain (residues 147–207), and a C-terminal basic amino acid rich domain (residues 208–272). Rspo3 (UniProt ID: Q9BXY4) is present in 2 alternatively splice isoforms. The full-length canonical isoform #1 (UniProt ID: Q9BXY4-1) has 272 residues. In isoform #2 (UniProt ID: Q9BXY4-1), the C-terminal region VSVSTVH (residues 266–272) is changed to GIEVTLAEGLTSVSQRTQPTPCRRRYL.

Rspo4. RSPO4 gene located at the position 20p13 on chromosome 20 encodes human R-spondin 4 protein (Rspo4, also called Roof plate-specific spondin-4). Rspo4 is a ligand for the LGR4, LGR5, and LGR6 receptors that activate the canonical Wnt signaling pathway. LGR4-6 interact with phosphorylated LRP6 and frizzled receptors that are activated by extracellular Wnt receptors leading to the activation of the canonical Wnt signaling pathway to increase expression of target genes. By acting as an inhibitor of ZnRF3, which is an important regulator of the Wnt signaling pathway, Rspo4 regulates both the canonical Wnt/β-catenin-dependent pathway and non-canonical Wnt signaling.^36,53,54

Domain organization of Rspo4 is similar to that of other members of the spondin family, and this protein has a signal peptide (residues 1–19), followed by 2 Fu domains (residues 37–85 and 85–128), a TSR1 domain (residues 138–197), and a positively charged C-tail (residues 198–234). There are 234 residues in the full-length canonical form of Rspo4 (UniProt ID: Q2I0M5-1), whereas the isoform #2 generated by the alternative splicing (UniProt ID: Q2I0M5-2) is characterized by missing the 137–198 region.

According to STRING analysis using the highest cut-off of 0.9 as the minimal required confidence level, Rspo2, Rspo3 and Rspo4 interact with each other, as well as bind to Rspo1 and interact with many binding partners of Rspo1, such as LGR4, LGR5 and LGR6, ZNFR3, UBA52, UBB, UBC, and RPS27A. In addition to this basic set, Rspo3 is known to interact with frizzled family receptor 8 (Fzd8) (see Fig. 3). The ability to interact with similar partners is likely to be due to some sequence and structural similarity of the R-spondin family members. Figure 4A shows that despite being encoded by different genes located at different chromosomes, these proteins share noticeable level of sequence similarity, with sequence identity ranging from 37.5% to 44.45%. These proteins also show rather similar disorder profiles (see Fig. 4B), indicating that the peculiarities of disorder distribution within their sequences is conserved, providing indirect support to the idea that intrinsic disorder is needed at least for some of their functions. Fig. 4C, 4D, and 4E further illustrate high levels of intrinsic disorder in the C-terminal halves of these proteins and also show that this disordered tail can be used for interactions with some binding partners of Rspo2, Rspo3, and Rspo4. Curiously, despite rather high sequence identity, close similarity of disorder profiles, highly overlapping interactomes, and important roles played in regulation and control of Wnt signaling, the biological activities of the members of the R-spondin family have very different outputs.

Figure 4. (A) Multiple sequence alignment of human Rspo1, Rspo2, Rspo3, and Rspo4 conducted by Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/). The corresponding similarity matrix is shown in the bottom of this figure legend. (B) Superposition of the consensus disorder profiles calculated for Rspo1 (black curve), Rspo2 (red curve), Rspo3 (green line), and Rspo4 (yellow line) illustrating remarkable similarity of the distribution of disorder propensities within sequences of these proteins. (C) Intrinsic disorder propensity and some important disorder-related functional information generated for human Rspo2 by the D²P² database. (D) Intrinsic disorder propensity and some important disorder-related functional information generated for human Rspo3 by the D²P² database. (E) Intrinsic disorder propensity and some important disorder-related functional information generated for human Rspo4 by the D²P² database. Percent Identity Matrix - created by Clustal2.1. Q2MKA7|RSPO1_HUMAN 100.00 40.93 44.79 38.96. Q6UXX9|RSPO2_HUMAN 40.93 100.00 45.45 41.05. Q9BXY4|RSPO3_HUMAN 44.79 45.45 100.00 37.50. Q2I0M5|RSPO4_HUMAN 38.96 41.05 37.50 100.00.

Finally, in order to illustrate evolutionary conservation of peculiarities of disorder distribution in R-spondins of different origin, Figure 5 represents disorder profiles of Rspo1 (see Fig. 5A), Rspo2 (see Fig. 5B), Rspo3 (see Fig. 5C), and Rspo4 (see Fig. 5D) from fish, frog, lizard, bird, and human and shows that specific disorder patterns are generally well preserved in these proteins, further inferring the functional importance of the predicted IDPRs. In fact, similar to Rspo1, human Rspo2 can undergo N-glycosylation, which is important for secretion and stability of this protein.⁶² Curiously, although N-glycosylation of Rspo2 was shown to occur at Asn160, a position near the N-terminus of TSR1 domain, it did not affect the heparin binding capability of this protein,⁶² a rather unexpected outcome since the TSR domain is known to be responsible for interaction with heparin sulfate proteoglycans (HSPGs).⁵⁵ Therefore, it seems that similar to the functional importance of long IDPRs on the intracellular side for mediating protein-protein interactions and/or PTMs, extracellular proteins might use their long IDPRs to carry PTM sites and sites of protein-protein interactions.

Figure 5. Analysis of the evolutionary conservation of intrinsic disorder propensity in Rspo1 (A), Rspo2 (B), Rspo3 (C), and Rspo4 (D). Disorder profiles were generated by PONDR-FIT for proteins from fish (black curves), frog (red curves), lizard (green curves), bird (yellow curves), and human (blue curves). These plots indicate the presence of rather close resemblance of the peculiarities of disorder distribution within orthologues, suggesting that intrinsic disorder might be of at least some functional importance.

Spondin family

F-spondin (spondin-1)

Protein spondin-1, also known as F-spondin, is encoded by the SPON1 gene located on chromosome 11. Although it was believed that F-spondin is preferentially expressed in the embryonic floor plate of vertebrates,^69-73 F-spondin transcripts were shown to be widely distributed in fetal and adult human tissues. For example, in normal adult tissue, F-spondin is expressed in various organs including the lung, ovary, small intestine, and kidney. An especially high expression of F-spondin is found in the adult human ovary, where it serves as a major factor for vascular smooth muscle cell (SMC) proliferation.⁷⁴

F-spondin is an extracellular matrix glycoprotein playing major role in the central nervous system (CNS), during development and later in life. In vitro, it participates in cell adhesion and can stimulate the sensory neuron and spinal cord cells attachment and outgrowth of neurites.⁷⁵ During early development, F-spondin also plays a role in the growth of axons in both the spinal cord and the peripheral nervous system (PNS). It is believed that F-spondin defines the axonal trajectory in the spinal cord by promoting the outgrowth of commissural axons⁷³ and inhibiting the outgrowth of motor axons.⁷⁶ In adult vertebrate, F-spondin displays broad distribution.⁷⁷ In the periphery, the expression of F-spondin is abundant in enteric neurons and in tissues regenerating following trauma. In the brain, it is associated with structures forming long neuronal tracts, such as the retina, the olfactory bulb, the habenula, and the nucleus of the medial longitudinal fasciculus (nMLF). F-spondin is expressed in the neurogenic niches of adult brain and abundant in CSF-contacting secretory neurons, especially those in the hypothalamus. Importantly, this protein can also play a role in neuronal regeneration,⁷⁸ and was shown to bind to the Aβ precursor protein (APP) and prevent its cleavage.⁷⁹

The total length of the human F-spondin (UniProt ID: Q9HCB6) is 807 amino acids. Like many large ECM proteins, F-spondin has a complex multidomain structure. This protein possesses N-terminally located reelin_N and F-spondin (FS) domains (residues 29–194 and 195–388, respectively), which are related to the corresponding domains found in the proteins Reelin and Mindin, and 6 C-terminal TSR1 domains (residues 442–495 (TSR1-1), 501–555 (TSR1-2), 558–611 (TSR1-3), 614–666 (TSR1-4), 668–721 (TSR1-5), and 754–807 (TSR1-6)). TSR1-2, TSR1-3, and TSR1-4 contain specific sequence motif (CSVTCG), which is responsible for the CD36 binding.⁸⁰ All these domains likely to play a role in the inhibition of the APP cleavage, where F-spondin is engaged in the formation of a heterotrimer on the membrane with APP and apolipoprotein E receptor 2 (apoEr2). This trimer is formed via the reelin_N domain and the FS domain interaction with APP and TSRs interaction with apoEr2.^46,81 Being a glycoprotein, F-spondin has several C-mannosylation and O-fucosylation sites,^82,83 with the first 5 TSR domains containing conserved tryptophan residues that have been identified as sites for C-mannosylation.⁸²

Crystal structures of the reelin-N (PDB ID: 3COO)^84,85 and spondin domains (PDB ID: 3Q13)⁸¹ of human F-spondin were solved. The reelin-N domain of human F-spondin has a sequence identity of 30% to the N-terminal domain of human reelin.⁸⁵ Structure of the reelin-N domain can be described as a variant immunoglobulin fold consisting of 2 β-sheets, where β-strands A, B, E, and D form one β-sheet, and β-strands A′ (or A0), G, F, C, D″ and D′ form the opposite β-sheet.⁸⁵ The protein is crystallized as a dimer, 2 chemically identical molecules of which are characterized by the substantial conformational differences. In fact, although the core structure of the domain of 2 monomers (Thr59-Glu100, A111-S162, total 92 Cα atoms from each monomer) can be superimposed with an RMSD value of 0.88 Å, the N-terminal region (up to Tyr58), CD loop and the C-terminal region (after S162) are totally out of the structural alignment with RMSD values of 4.97 Å, 2.03 Å, and 2.92 Å, respectively.⁸⁵

The F-spondin domain (FS domain) contains 2 distinctive conserved sequence repeats known as “FS1-domain” and “FS2-domain” that can also be found in other spondin family members, including mindin 1 and 2, and M-spondin.⁸⁵ Analysis of crystal structure of the FS domain (residues 191–434, PDB ID: 3Q13) revealed that it is characterized by the topology typical for the C2 domain β-sandwich folds,⁸⁶ where core of the FS domain has an 8-stranded β-sandwich fold, with β-strands β4, β1, β8, and β7 forming one β-sheet and β-strands β3, β2, β5 and β6 forming the other β-sheet.⁸¹

Although no structural information is available for any of the 6 TSR domains of human F-spondin, structure of TSR1-2 and TSR1-3 of the rat protein, along with the linker between the 2 domains, have been determined by X-ray crystallography,⁸⁷ and solution structures were solved for the TSR1-1 and TSR1-4 of rat F-spondin by the heteronuclear NMR spectroscopy combined with automated NOESY assignment and structure calculation,⁸⁸ all of which are characterized by the antiparallel 3-stranded β-sheet fold stabilized by 3 disulfide bonds. In solution structures, the first strand has a rippled conformation, whereas the 2 other strands form an antiparallel β-sheet (residues 462–467 and 484–489 in TSR1-1, and 634–640 and 657–663 in TSR1-4).⁸⁸

Results of the computational evaluation of the abundance of predicted intrinsic disorder in F-spondin is shown in Fig. 6B and 6C, which illustrate that this protein contains several IDPRs, albeit being not as disordered as R-spondins discussed in the preceding sections. Potential functional applications of disorder in F-spondin is outlined in the subsequent section dedicated to mindin.

Figure 6. (A) Aligned structures of FS domains of human F-spondin (235 residues, blue structure; PDB ID: 3Q13) and mindin (211 residues, red structure; PDB ID: 3D34). (B) Superposition of the consensus disorder profiles calculated for human F-spondin (blue curve; UniProt ID: Q9HCB6) and human mindin (red curve, UniProt ID: Q9BUD6). (C) Intrinsic disorder profile of human F-spondin (UniProt ID: Q9HCB6) generated by the superposition of the outputs of PONDR® VLXT, PONDR® FIT, PONDR® VL3, PONDR® VSL2, IUPred_short and IUPred_long and a consensus disorder profile calculated by averaging disorder profiles of individual predictors. (D) Intrinsic disorder profile of human mindin (UniProt ID: Q9BUD6) generated by the superposition of the outputs of PONDR® VLXT, PONDR® FIT, PONDR® VL3, PONDR® VSL2, IUPred_short and IUPred_long and a consensus disorder profile calculated by averaging disorder profiles of individual predictors.

Figure 7A shows remarkable evolutionary conservation of the disorder pattern in F-spondins of different origin. Obviously, high similarity of disorder profiles of F-spondins from fish, frog, lizard, bird, and human represents a strong indication that the peculiarities of distribution of the intrinsic disorder propensity within the amino acid sequences are of at least some functional importance for these proteins.

Figure 7. Analysis of the evolutionary conservation of intrinsic disorder propensity in F-spondin (A), mindin (B), and SCO-spondin (C). Disorder profiles were generated by PONDR-FIT for proteins from fish (black curves), frog (red curves), lizard (green curves), bird (yellow curves), and human (blue curves). These plots indicate the presence of rather close resemblance of the peculiarities of disorder distribution within orthologues, suggesting that intrinsic disorder might be of at least some functional importance.

Mindin (spondin-2 or DIL-1)

Spondin-2, also known as differentially expressed in cancerous and non-cancerous lung cells 1 (DIL-1), is a second member of the spondin family. It is encoded by SPON2 gene located on chromosome 4. The DIL-1 name was given to this protein when it was cloned from noncancerous lung cells and found to be downregulated in cancerous lung cells.⁸⁹ SPON2 expression is significantly associated with the stages of the colorectal cancer (CRC), and the upregulated spondin-2 can be used as a predictor of the poor survival prognosis of CRC patients.⁹⁰ High levels of SPON2 and spondin-2 protein expression are also observed in Barrett's adenocarcinoma,⁹¹ breast cancer,⁹² gastric cancer,⁹³ liver cancer,^94,95 ovarian cancer,^96,97 pancreatic cancer,⁹⁸ and prostate cancer.^99,100 It was also found that levels of this protein undergo a linear and significant increase in patients with type 2 diabetes (T2D) as the stage of their diabetic nephropathy (DN) increased.¹⁰¹

Spondin-2 is a cell adhesion protein that stimulates hippocampal embryonic neurons adhesion and growth.¹⁰² It functions as an opsonin for macrophage phagocytosis of bacteria by binding to bacteria and their components.¹⁰³ It plays a significant role in the initiation of the innate immune response and acts as a unique pattern-recognition molecule in the ECM for microbial pathogens providing defense responses against fungus and viruses.¹⁰³ It also binds bacterial lipopolysaccharides (LPSs).¹⁰³ Mindin participates in integrin-dependent trafficking of eosinophils.¹⁰⁴ In an experimental stroke model, mindin was shown to serve as a critical mediator of ischemic brain injury, most likely via mediation of/by Akt signaling.¹⁰⁵ It is specifically involved in the positive regulation of interleukin-6 production, macrophage cytokine production, and tumor necrosis factor production.¹⁰³

The total length of the human mindin (UniProt ID: Q9BUD6) is 331 amino acids. Mindin has a signal peptide (residues 1–26), a F-spondin domain (FS, residues 31–221), structure of which was solved (PDB ID: 3D34),¹⁰⁶ and a single C-terminally located TSP type-1 domain (residues 277–331). Similar to F-spondin, the FS of mindin is calcium-binding protein with Ca²⁺ being ligated by Asp¹⁶⁰, Asp¹⁸⁸, and Asp¹⁹².¹⁰⁶ The structure of the FS domain of mindin is characterized by the 8-stranded antiparallel β-sandwich topology containing 2 4-stranded β-sheets, β4β1β8β7 and β6β5β2β3 with 2 α-helices packed on one side of the β-sandwich.¹⁰⁶ Fig. 6A represents aligned structures of FS domains of human F-spondin (235 residues, blue structure) and mindin (211 residues, red structure) and show close structural similarity of the core regions of these 2 domains illustrated by the facts that the 200 structurally aligned residues are characterized by the RMSD value of 1.15 Å and that the structural dissimilarity is mostly found in loops and tail regions. This close structural similarity is only in part can be attributed to the sequence similarity between these 2 proteins which are characterized by sequence identity of 32.51% (see supplementary materials). Fig. 6B, and C and 6D show that F-spondin and mindin are predicted to have less intrinsic disorder in comparison with the members of R-spondin family. Since some similarity is present in disorder profiles of F-spondin and mindin (see Fig. 6B), one can assume that the peculiarities of disorder distribution within the sequences of FS domains could be of some functional importance. For example, it was pointed out that in mindin, residues Lys⁴², Glu¹²², and Glu¹⁴¹ can be engaged in interaction with the metal ion-dependent adhesion site (MIDAS) of the integrin,¹⁰⁶ and 2 of these residues, Glu¹²² and Glu¹⁴¹, are predicted to be highly flexible, possessing the mean disorder scores of 0.48 ± 0.07 and 0.48 ± 0.05, respectively.

Similar to F-spondins, intrinsic disorder profiles of mindins of different origin are characterized by high similarity levels, indicative of the important functional roles of intrinsic disorder in these proteins. In fact, Fig. 7B clearly shows that the distributions of the intrinsic disorder propensity within the amino acid sequences of mindins from fish, frog, lizard, bird, and human are very similar.

SCO-spondin

With length of 5,147 residues in its canonical form, SCO-spondin is the longest member of the group of human spondins considered in this article. It is encoded by the SSPO gene located at the position 7q36.1 on chromosome 7. As other spondins, SCO-spondin is a member of the trombospondin family that is involved in neuronal development by modulating cell aggregative mechanisms. It is an extracellular matrix glycoprotein that is secreted and expressed early in development by the subcommissural organ (SCO), an ependymal differentiation located in the roof of the Sylvian aqueduct.¹⁰⁷ SCO-spondin also makes part of the Reissner's fiber (RF), a thread-like structure present in the central canal of the spinal cord, and may also be involved in posterior commissure formation and spinal cord differentiation during ontogenesis of the central nervous system.⁴⁵ The role of this protein in adult brain, and that of the Reissner's fiber it forms, remains largely unknown.

The potent sites for protein-protein interaction include 26 thrombospondin type 1 repeats (TSR), 9 low-density lipoprotein receptor (LDLR) type A domains, 2 epidermal growth factor (EGF)-like domains, and N- and C-terminal von Willebrand factor (VWF) cysteine-rich domains.¹⁰⁸ Besides the full-length canonical form, SCO-spondin has an alternatively spliced isoform #2 (UniProt ID: A2VEC9-2) whose shortened sequence (1,314 residues) originates from removal of long N- and C-terminal regions (residues 1–1122 and 2315–5147, respectively), changes of Ala¹⁶³⁹ to ACVEAPAPPA…DCPQGEDELD, changes of Leu¹⁶⁷¹ to LVRVGVGGGGGSAMLPPSTRALTPLPPQ, and by substituting the 2179–2314 region LFPRNWDDLD…ETEHWPPGQE to VSPAQGRWGQ…RSGRNQSVLC.

The domain structure of human SCO-spondin is complex, which is not too surprising taking into account its length. The protein starts with the EMI domain (residues 18–802) followed by the von Willebrand factor type D domain-1 (VWFD-1, residues 194–408), TIL-1 domain (residues 469–924), VWFD-2 (residues 563–373), TIL-2 (residues 827–779), von Willebrand factor type C domain-1 (VWFC-1, residues 880–039), VWFD-3 (residues 1013–3219), TIL-3 (residues 1275–5331), LDL-receptor class A domain-1 (LDLRA-1, residues 1375–5412), LDLRA-2 (residues 1415–5450), LDLRA-3 (residues 1451–1487), LDLRA-4 (residues 1491–1529), LDLRA-5 (residues 1564–4600), LDLRA-6 (residues 1602–2641), LDLRA-7 (residues 1655–5693), TSR1-1 (residues 1694–4748), TSR1-2 (residues 1750–0808), EGF-like domain-1 (EGFL-1, residues 1824–4863), EGFL-2 (residues 1864–4901), TSR1-3 (residues 1909–9965), VWFC-2 (residues 1965–5025), F5/8 type C domain (residues 2065–5224), LDLRA-8 (residues 2233–3269), LDLRA-9 (residues 2390–0426), LDLRA-10 (residues 2463–3499), TSR1-4 (residues 2500–0553), TSR1-5 (residues 2555–5610), TIL-4 (2633–3675), TSR1-6 (residues 2715–5769), TSR1-7 (residues 2772–2828), TSR1-8 (residues 2830–0883), TSR1-9 (residues 2985–5040), TSR1-10 (residues 3041–1083), TSR1-11 (residues 3183–3250), TSR1-12 (residues 3252–2307), TIL-5 (residues 3311–1365), TSR1-13 (residues 3408–8470), TSR1-14 (residues 3472–2527), TSR1-15 (residues 3645–5693), TSR1-16 (residues 3811–1932), TSR1-17 (residues 3946–6002), TSR1-18 (residues 4004–4059), TSR1-19 (residues 4159–9212), TSR1-20 (residues 4253–3305), TSR1-21 (residues 4307–7363), TSR1-22 (residues 4365–5419), TSR1-23 (residues 4615–5665), TIL-6 (residues 4667–7723), TSR1-23 (residues 4763–3816), VWFC-3 (residues 4984–4042), and the C-terminal cystine knot (CTCK) domain (residues 5041–1140).

No structural information is available for any part of SCO-spondin. Figure 8 represents results of intrinsic disorder analysis in human SCO-spondin (UniProt ID: A2VEC9) and shows that this protein is expected to contain significant amount of IDPRs. In fact, according to different computational tools, the mean disorder score SCO-spondin ranges from 0.153 ± 0 .002 to 0.634 ± 0 .004 according to IUPred_long and PONDR® VSL2, respectively (the mean disorder score evaluated by averaging the results of 6 predictors shown in Fig. 8 is 0.358 ± 0 .002), whereas its overall disorder content (i.e., percentage of residues with disorder scores exceeding the threshold of 0.5) ranges from 71.2 % to 5.4 % as evaluated by the same predictors (the averaged disorder content is 20.3%). According to ANCHOR analysis, there are 6 disorder-based binding sites in this protein, residues 2051–2057, 2069–2076, 2269–2274, 2278–2309, 2328–2368, and 2429–2443. The protein is expected to have 101 disulfide bonds and 43 glycosylation sites.

Figure 8. Intrinsic disorder profile of human SCO-spondin (UniProt ID: A2VEC9) generated by the superposition of the outputs of PONDR® VLXT, PONDR® FIT, PONDR® VL3, PONDR® VSL2, IUPred_short and IUPred_long and a consensus disorder profile calculated by averaging disorder profiles of individual predictors. Due to large length of this protein, the corresponding disorder profile is split on 5 segments to better show the peculiarities of intrinsic disorder distribution within this protein.

Figure 7C shows that despite their very large length, SPO-spondins from different organisms (fish, frog, lizard, bird, and human) still share some levels of similarity in their disorder profiles. This is in line with an idea that similar to other members of spondin family, intrinsic disorder is of functional importance for SPO-spondins.

Spondin-interacting proteins

Frizzled-4

Frizzled 4 (Fzd4) is a protein in humans (UniProt ID: Q9ULV1) encoded by the FZD4 gene on chromosome 11q14-q21 that belongs to the frizzled family of genes which encodes 7 transmembrane domain proteins. Frizzled are receptors linked to the canonical β-catenin (CTNNB1) signaling pathway resulting in the activation of disheveled proteins (Dvl), inhibition of GSK-3 kinase, nuclear accumulation of CTNNB1, and activation of Wnt target genes. Fzd4 is the only member of the frizzled family that binds strongly to a secreted retinal growth factor with angiogenic and neuroprotective properties, norrin, also known as the Norrie disease protein or X-linked exudative vitreoretinopathy 2 protein.¹⁰⁹ Fzd4 plays a positive role in regulating Wingless type MMTV integration site signaling pathway and the signals of this protein induced by norrin regulates vascular development of vertebrate retina and controls important blood vessels in the ear.¹¹⁰ Wnt signaling pathway is crucial for the nervous system development and maintenance, and this cascade is initiated when Wnt binds with the frizzled transmembrane receptor family and results in activation of pathways like the β-catenin-dependent pathway or the planar cell polarity- (PCP-) dependent pathway, both of which are involved in early neural crest development.¹¹¹ Frizzled-4 also plays a role in regulating stem cell maintenance, cell migration and neuro protection in central nervous system.¹¹² R-spondins are considered as alternative Fzd ligands.¹¹³

Fzd4 contains a frizzled (FZ) cysteine-rich domain (residues 40–161) located within the N-terminal extracellular domain of this transmembrane protein and crystal structure of which is known (e.g., PDB ID: 5BPB),¹¹⁴ a Lys-Thr-X-X-X-Trp motif mediating the interaction with the PDZ domain of Dvl family members (residues 499–504), and a PDZ-binding motif (residues 535–537), both located within the C-terminal cytoplasmic domain, and 7 transmembrane helices (residues 223–243, 255–275, 303–323, 345–365, 390–410, 437–457, and 478–498). Although in agreement with its transmembrane nature, Fzd4 is predicted to be mostly ordered, it has 4 IDPRs (residues 1–44, 141–184, 419–429, and 513–537). Furthermore, the functional motifs needed for Fzd4 interaction with Dvl and PDZ are located within the C-terminal disordered tail (see Fig. 9A and Supplementary Materials Figure S1A). Results of the STRING-based analysis of the Fzd4 interactivity are shown in Figure S2A that clearly indicates high binding promiscuity of this protein.

Figure 9. Intrinsic disorder profiles of: (A) Human Fzd4 (UniProt ID: Q9ULV1); (B) Human Fzd8 (UniProt ID: Q9H461); (C) Human ZnRF3 (UniProt ID: Q9ULT6); and (D) Human RNF43 UniProt ID: Q68DV7). These disorder profiles were generated by the superposition of the outputs of PONDR® VLXT, PONDR® FIT, PONDR® VL3, PONDR® VSL2, IUPred_short and IUPred_long and a consensus disorder profile calculated by averaging disorder profiles of individual predictors.

Frizzled-8

The gene that codes for frizzled-8 (Fzd8) is FZD8 located on chromosome 10p11.2.¹¹⁵ The protein functions as receptor for Wnt proteins,¹¹⁶ and serves as a component of the Wnt-Fzd-LRP5-LRP6 complex that induces association of the receptor-ligand complexes into ribosome-sized signalosomes resulting in the initiation of the β-catenin canonical signaling pathway that leads to the inhibition of GSK-3 kinase, activation of disheveled proteins, accumulation of β-catenin inside the nucleus, and activation of Wnt target genes. Fzd8 may be involved in transduction and intercellular transmission of polarity information during tissue morphogenesis and/or in differentiated tissues. This protein serves as co-receptor of Wnt proteins, such as Wnt1.¹¹⁵ The extracellular domains of Fzd8 were shown to interact with Rspo1 and Rspo3.⁵⁷

Human Fzd8 (UniProt ID: Q9H461) is a 694 residue-long proteins that has a signaling peptide (residues 1–27) and N-terminally located FZ domain (residues 30–151), which is a part of the extracellular domain (residues 28–275). Similar to other members of the frizzled family, this protein has 7 transmembrane helices (276–296, 313–333, 397–417, 440–460, 484–504, 533–553, and 585–605) and a cytoplasmic C-terminal tail (residues 606–694). Regions 95–100 and 147–152 of Fzd8 are involved in Wnt binding, motif Lys-Thr-X-X-X-Trp located at 608–613 region mediates interaction with the PDZ domain of Dvl family members, and a PDZ-binding motif is located the very end of C-terminus (residues 692–694). Figure 9B shows that Fzd8 is predicted to have several IDPRs (residues 1–33, 156–249, 340–380, 516–526, 574–580, and 625–694) 4 disorder-based potential binding sites (residues 148–160, 196–210, 666–679, and 687–694), and several phosphorylation sites. Two functional motifs/regions of Fzd8 (one of the Dvl binding motifs (residues 147–152) and C-terminal PDZ-binding motif) are located within the disordered regions that are expected to undergo binding-induced disorder-to-order transitions, clearly indicating that intrinsic disorder is important for the functionality of this transmembrane protein (see Fig. 9B and Supplementary Materials Figure S1B). Figure S2B represents the results of the STRING-based analysis of the Fzd8 interactivity and shows that this protein is involved in a wide range of protein-protein interactions.

E3 ubiquitin-protein ligase ZnRF3

E3 ubiquitin-protein ligase is encoded by gene ZNRF3 located on chromosome 22. This proteins is also known as RING finger protein 203 and Zinc/RING finger protein 3 (ZnRF3). ZnRF3-driven ubiquitination and subsequent degradation of Wnt receptor complex components, Frizzled and LRP6, defines the involvement of this E3 ubiquitin-protein ligase in negative regulation of both canonical and non-canonical Wnt signaling pathways. It is also involved in the tumor suppressor process in the intestinal stem cell zone by inhibiting the Wnt signaling pathway which leads to size limitation of the intestinal stem cell zone.¹¹⁷ Overexpression of ZnRF3 was shown to negatively regulate both the Wnt and Hedgehog proliferative pathways (and thereby to negatively regulate cancer progression) via dramatic reduction of the levels of LGR5 and Gli1, which are component of the Wnt and Hedgehog signaling pathways, respectively.¹¹⁸

R-spondin proteins, such as Rspo1, are responsible for the negative regulation of ZNRF3, since indirect association between ZnRF3 and LGR4 mediated by Rspo1 promotes membrane clearance of ZnRF3.¹¹⁷ Interactions between the extracellular region of RNF43 and ZnRF3 provides a direct linkage between the extracellular recognition and E3 ligase activity needed for the modulation of cell surface signaling.¹¹⁹ This E3 ubiquitin ligase serves as an important component of the Rspo-LGR4/5-ZnRF3/RNF43 module that acts as a regulator of the Wnt/β-catenin-mediated metabolic liver zonation and controls hepatic growth and size during development, homeostasis, and regeneration.¹²⁰

Human ZnRF3 (UniProt ID: Q9ULT6) is a single-pass transmembrane protein containing N-terminal signal peptide (residues 1–55), extracellular domain (residues 56–219), transmembrane helix (residues 220–240), and a cytoplasmic domain (residues 241–936), where the zinc finger domain (RING-type, residues 293–334) is embedded. This protein exists in 2 isoforms, the full-length canonical form (936 residues) and alternatively spliced isoform #2 that differs from the canonical form by missing first 100 residues. Figure 9C show that, despite being a transmembrane protein, ZnRF3 is predicted to contain significant level of intrinsic disorder (>50 %), especially at its cytoplasmic domain, which seems to be mostly disordered. There are 16 disorder-based binding regions in this protein (residues 37–48, 59–66, 427–434, 507–512, 593–599, 658–670, 685–694, 696–702, 705–713, 728–741, 774–784, 801–808, 828–842, 877–892, 897–911, and 918–926) and several phosphorylation sites (see also Figure S1C). High levels of intrinsic disorder in ZnRF3 are in line with known high predisposition of protein ubiquitin E3 ligases for intrinsic disorder,¹²¹ and together with large number of disorder-based binding sites defines the ability of this protein to be engaged in numerous protein-protein interactions (see Figure S2C).

E3 ubiquitin-protein ligase RNF43

E3 ubiquitin-protein ligase RNF43 or RING finger protein 43 is encoded by the RNF43 gene located on the 17q23.2 chromosome. RNF43 is one of the interacting partners of spondins and functions as a negative regulator of both canonical and non-canonical Wnt signaling pathways by mediating the ubiquitination, endocytosis and subsequent degradation of Wnt receptor complex components, such as Frizzled.^38,117 Therefore, similar to ZnRF3, RNF43, which is considered as a stem-cell E3 ligase, reduces Wnt signals by selectively ubiquitinating frizzled receptors and targeting surface-expressed frizzled receptors to lysosomes for degradation.¹²² RNF43, cancer-associated RING finger protein, can be found in the endoplasmic reticulum (ER) and in the nuclear envelope.¹²³ It was suggested that RNF43 may be involved in cell growth control through the interaction with a nuclear protein HAP95.¹²³

Human RNF43 exists as 4 alternative splicing-generated proteoforms. Isoforms #2 and #3 (UniProt ID: Q68DV7-2 and Q68DV7-3) are different from the canonical form by missing region 85–125 and 1–125, respectively, whereas in the isoform #4 (UniProt ID: Q68DV7-4), the C-terminal tail region SEEELEELCEQAV (residues 771–783) is changed to EFSEGSGCGRERRLQ LNISGQVKSANKGLMEAEKDTAEMTTKILNHRDSVSCWLECRNTPPLPGATPLVGRSQGGPREVLVWLRHQKGTWKAGCDGSCL. Similar to ZnRF3, human RNF43 is a single-pass transmembrane protein that contains signal peptide (residues 1–23), extracellular domain (residues 24–197), transmembrane helix (residues 198–218), and a cytoplasmic domain (residues 219–783), with the zinc finger domain (RING-type, residues 272–313). It has 3 regions with the compositional bias, serine-rich (residues 443–503), histidine-rich (residues 547–557), and proline-rich (residues 569–760).

Crystal structure of the extracellular protease-associated (PA) domain (residues 44–198) of RNF43 in a complex with the CRD Rspo1 and the LGR5 ectodomain (ECD) was solved (PDB ID: 4KNG).⁵⁶ PA domains that function as ligand recognition motifs and play regulatory roles are commonly found in proteases and receptors.¹²⁴ Fig. 10 shows that the RNF43 PA domain consists of 7 β-strands that form a twisted β sheet, with the curved β3 defining a shallow binding groove used for binding of the β-hairpin protrusion of the Rspo1, and 3 peripheral α-helices.¹²⁴ Since there is no physical contact between LGR6 and RNF5, Fig. 10 clearly illustrates the scaffolding role of Rspo1 that assembles the tripartite LGR3/Rspo1/RNF43 complex. Cytoplasmic domain is predicted to be mostly disordered (see Fig. 9D) and contain 11 disorder-based binding motifs (residues 349–365, 386–396, 400–436, 451–460, 480–490, 497–512, 531–568, 597–650, 674–690, 723–730, and 777–783) and several phosphorylation sites (see Figure S1D). This high intrinsic disorder level and presence of multiple disorder-based binding sites makes RNF43 a promiscuous binder (see Figure S2D).

Figure 10. Evaluating disorder predisposition and some important disorder-related functional information evaluated for human LRP5 (UniProt ID: O75197, plot (A) and human LRP6 (UniProt ID: O75581, plot (B) by the D²P² database.

Low-density lipoprotein receptor-related proteins 5 and 6 (LRP5 and LRP6)

Low-density lipoprotein receptor-related proteins 5 and 6 are encoded by the LRP5 and LRP6 genes located on the 11q13.4 and 12p13.2 chromosome respectively. Being located on the cell surface, LRP5 and LRP6 serve as co-receptors of Wnt/β-catenin signaling and are involved in the formation of bones. These two proteins are engaged in the formation of the Wnt-Fzd-LRP5-LRP6 complex responsible for triggering the β-catenin signaling pathway via promotion of the assembly of receptor-ligand complexes into signalsomes.¹²⁵ After the Fzd/LRP6 co-receptor complex is induced by Wnt, it recruits Dvl1 (segment polarity protein dishevelled homolog-1) to the plasma membrane leading to the recruitment of the Axin1/GSK3β-complex to the cell surface. This results in the formation of signalsomes and leads to the inhibition of the Axin1/GSK3-mediated phosphorylation and destruction of β-catenin.¹²⁶ LRP5 is also involved in the norrin-mediated signal transduction.¹²⁷

Functionality of these proteins is regulated by various PTMs. For example, dual phosphorylation of cytoplasmic PPPSP motifs sequentially by GSK3 and CK1 is required for Axin1-binding, and subsequent stabilization and activation of β-catenin via preventing GSK3-mediated phosphorylation of this protein. In in vitro experiments, LRP6 can be phosphorylated by GRK5/6 within and outside the PPPSP motifs. Phosphorylation at Ser-1490 by CDK14 during G2/M phase leads to regulation of the Wnt signaling pathway during the cell cycle. Phosphorylation by GSK3β is induced by RPSO1 binding and inhibited by DKK1. Negative regulation of LRP6 function by casein kinase I epsilon phosphorylation may induce a negative feedback loop by phosphorylation of sites on LRP5/6 that modulate axin binding and hence β-catenin degradation.¹²⁸

LRP6 undergoes gamma-secretase-dependent regulated intramembrane proteolysis (RIP).¹²⁹ The extracellular domain is first released by shedding, and then, through the action of gamma-secretase, the intracellular domain (ICD) is released into the cytoplasm where it is free to bind to GSK3β and to activate canonical Wnt signaling.¹²⁹ Palmitoylation on the 2 sites near the transmembrane domain leads to release of LRP6 from the endoplasmic reticulum, whereas mono-ubiquitination retains LRP6 in the endoplasmic reticulum.¹³⁰

Both proteins can be ubiquitinated by ZnRF3, which leads to their degradation via the proteasome.¹³¹ As it was already pointed out, ZnRF3 (which is a cell-surface transmembrane E3 ubiquitin ligase) is associated with the Wnt receptor complex. The role of this protein in the inhibition of the Wnt signaling relies on its ability to promote the turnover of frizzled and LRP6. It was also shown that the Wnt/β-catenin signaling can be enhanced via the ZnRF3 inhibition which also disrupts the Wnt/planar cell polarity signaling in vivo.^38,117

Finally, N-glycosylation of LRP6 is required for its cell surface location. The maturation and plasma membrane localization of this protein can be blocked by the expression of Mest/Peg1 (mesoderm-specific transcript/paternally expressed gene 1) via the Mest/Peg1-controlled inhibition of the LRP6 glycosylation.¹³²

Although LRP5 and LRP6 have many overlapping activities and rather similar domain organization (see below), these proteins are not completely redundant and also have some unique functions during development and adult tissue homeostasis, with LRP6 playing a dominant role in embryogenesis.¹³³ This conclusion can be further illustrated by non-overlapping sets of diseases linked to abnormalities of these proteins. In fact, mutations in the human LRP5 gene are associated with several diseases, such as exudative vitreoretinopathy 1,¹²⁷ exudative vitreoretinopathy 4,¹³⁴ osteoporosis,^135,136 osteopetrosis, autosomal dominant 1,¹³⁷ osteoporosis-pseudoglioma syndrome,^138-142 high bone mass trait,^143-147 Worth type of endosteal hyperostosis,¹³⁷ and Van Buchem disease 2,¹³⁷ whereas mutations in the human LRP6 gene are known to cause autosomal dominant coronary artery disease 2,^148,149 and selective tooth agenesis 7.¹⁵⁰

Human LRP5 (UniProt ID: O75197) is a large (mature form has 1584 residues after removal of the N-terminally located 31 residue-long signal peptide), single-path transmembrane protein that have an N-terminal extracellular ectodomain (residues 32–1384), a transmembrane helix (residues 1385–1407) and a C-terminal cytoplasmic domain (residues 1408–1615). There are 4 β-propeller regions in this protein (residues 32–288, 341–602, 644–903, and 945–1212) connected by EGF-like domains (residues 295–337, 601–641, and 902–942). This cassette is connected via another EGF-like domain (residues 1213–1254) to a series of 3 LDL-receptor class A domains (residues 1258–1296, 1297–1333, and 1335–1371). It also has multiple repeats of different nature, such as 20 LDL-receptor class B repeats (residues 75–119, 120–162, 163–206, 207–247, 248–290, 385–427, 428–470, 471–514, 515–557, 558–600, 687–729, 730–772, 773–815, 816–855, 856–898, 989–1035, 1036–1078, 1079–1123, 1124–1164, and 1165–1207) and 11 YWTD repeats (residues 78–81, 123–126, 166–169, 251–254, 388–391, 431–434, 474–477, 559–562, 690–693, 819–822, and 859–862). Finally, C-terminal domain contains a proline-rich region (residues 1495–1610) that includes 5 PPPSP motifs (residues 1500–1506, 1538–1545, 1574–1581, 1591–1596, and 1605–1612). Although no structural information is available for human LRP5 protein, by prediction, it is expected to have rather disordered C-terminal tail (see Fig. 10A), where most PTM sites and disorder-based binding sites are concentrated.

After removal of the N-terminal signal peptide (residues 1–19), the amino acid sequence of human LRP6 (UniProt ID: O75581) consists of 1594 residues. This is another single-path transmembrane protein that has an N-terminal extracellular ectodomain (residues 20–1370), a transmembrane helix (residues 1371–1393) and a C-terminal cytoplasmic domain (residues 1394–1613). Similar to LRP5, LRP6 has 4 β-propeller regions (residues 20–275, 328–589, 631–890, and 933–1202) connected by EGF-like domains (residues 282–324, 588–628, and 889–930). The fourth EGF-like domain (residues 1203–1244) links β-propeller containing region to a set of 3 LDL-receptor class A domains (residues 1248–1286, 1287–1323, and 1325–1361). This protein also has 20 LDL-receptor class B repeats (residues 63–106, 107–149, 150–193, 194–236, 237–276, 372–414, 415–457, 458–501, 502–542, 543–584, 674–716, 717–759, 760–802, 803–842, 843–885, 977–1025, 1026–1068, 1069–1113, 1114–1156, and 1157–1198) and its C-terminal domain contains 5 PPPSP motifs (residues 1487–1493, 1527–1534, 1568–1575, 1588–1593, and 1603–1610).

Structural information is available for the β-propeller containing region, residues 20–635 containing first 2 β-propeller domains and first 2 EGF-like domains (PDB ID: 4DG6) and residues 629–1244 containing last 2 β-propeller domains and 2 EGF-like domains (PDB ID: 4A0P),¹⁵¹ as well as for the 2 phosphorylated PPPSP motifs, residues 1568–1575 (PDB ID: 4NM5) and 1603–1610 (PDB ID: 4NM7) bound to the GSK3/Axin complex.¹⁵² The structures of the individual β-propellers are typical of the YWTD class β-propeller domains (i.e., domains containing the N-terminal Tyr-Trp-Thr-Asp (YWTD) motif), with 6 blades of 4-stranded antiparallel β-sheets being symmetrically arranged around a central channel.¹⁵¹ Since the linker between β-propeller and EGF domain crosses the base of the β-propeller, the EGF domain is positioned to form a predominantly hydrophobic contact with the second and third blades.¹⁵¹ Obviously, proline-rich motifs are present in their complexes with GSK3/Axin in mostly irregular structure.¹⁵² In agreement with this structural characterization and similar to LRP5, LRP6 is predicted to have mostly ordered N-terminal ectodomain and rather disordered C-terminal cytoplasmic domain that includes all the disorder-based binding sites and which is heavily decorated with PTMs (see Fig. 10B).

It was pointed out that the majority of the Wnt- and antagonist-binding sites of LRP5 and LRP6 are located within the 4 tandem β-propeller–EGF-like domain (PE) pairs (P1E1–P4E4), suggesting that they serve as major functional recognition modules of the ectodomain.¹³³ In fact, functional analysis of LRP6 and LRP5 revealed that their 4 tandem PE pairs might represent 2 independent functional units, P1E1P2E2 and P3E3P4E4, where P1E1P2E2 serves as the primary binding domain for Wnt9b,¹⁵³ whereas P3E3P4E4 is engaged in interaction with Wnt3a¹⁵³ and Dkk1.¹⁵⁴ It was also noted that P1E1P2E2 can contribute to Dkk1 binding.^153,155,156

Complexity of both LRP5 and LRP6 is further increased by the fact that these 2 proteins operate as disulfide-linked homodimers. Combined with abundant intrinsic disorder in their C-terminal regions, presence of multiple PTMs and several disorder-based binding sites, this overall complexity defines the ability of these proteins to be engaged in multiple protein-protein interactions (see Fig. S3).

Leucine-rich repeat-containing G-protein-coupled receptors 4, 5, and 6 (LGR4, LGR5, and LGR6)

Leucine-rich repeat-containing G-protein-coupled receptors (LGRs) belong the superfamily of G protein-coupled receptors (GPCRs) includes at least 800 7 transmembrane receptors participating in a multitude of physiological and pathological functions.¹⁵⁷ Various physiological and pathological roles of GPCRs rely on the ability of these receptors to transduce extracellular signals to various intracellular pathways via their activation mediated by binding to a broad range of ligands, such as eicosanoids, organic compounds, peptides, and proteins.¹⁵⁷ Until quite recently, LGRs were considered as orphan GPCRs, since their endogenous ligands remained unidentified. However, it is known now that LGRs belong to the Rhodopsin subfamily of GPCRs, being typically considered as classical GPCRs in terms of their structure and signal transduction.¹⁵⁷ Based on their natural ligands, LGRs 4–8 can be grouped into 2 classes, with R-spondins being the ligands for the LGRs 4, 5, and 6,⁵² and with the relaxin family peptide (RXFP) serving as a ligand for the LGRs 7 and 8.¹⁵⁸ The LGR 4–6 proteins are known to interact with Wnt receptors, mediate R-spondin signaling, and lead directly to the activation of Wnt canonical pathway through the Frizzled and LRP proteins.¹⁵⁹

Despite their sequence and structural similarities (see below), these 3 LGRs have rather different functions. For example, LGR4 is expressed in proliferating cells of diverse tissues, including adult stem cells and progenitor cell.¹⁶⁰ Its expression was reported in adrenal gland, cartilage, eyes, kidney, nervous system cells, and reproductive tracts, where it engaged in a broad range of important physiological functions.^161,162 The loss of this protein causes abnormal renal development, defective development of the gall bladder and cystic ducts,¹⁶³ defective postnatal development of the male reproductive tract,¹⁶⁴ developmental defects in bone formation and remodeling dysfunction,¹⁶⁵ impaired hair placode formation,¹⁶⁶ intrauterine growth retardation associated with embryonic and perinatal lethality,¹⁶⁷ and ocular anterior segment dysgenesis.¹⁶⁰ On the other hand, LGR5 serves as a marker of stem cells of hair follicle¹⁶⁸ and gastrointestinal tract.¹⁶⁹ In mice, total neonatal lethality accompanied by ankyloglossia and gastrointestinal distension was reported when LGR5 gene was knockout.¹⁷⁰ Although LGR6 also marks stem cells in hair follicles, all the cell lineages of the skin (including those of the hair follicle, sebaceous gland, and interfollicular dermis) were shown to be generated by the LGR6-positive stem cells.¹⁷¹ Increased phosphorylation of LRP6 allows this protein to interact with R-spondin 1–3 with high affinity leading to the enhancement of the Wnt signaling.⁶⁸

Finally, high level of LGR4 and LGR5 expression were reported in several types of cancers. For example, cervical and colon cancer cell invasiveness and metastasis are enhanced by the LGR4 overexpression,¹⁷² whereas human colon and ovarian tumors are characterized by the LGR5 up-regulation which promotes cell proliferation and tumor formation in basal cell carcinoma.^173,174 Although the LGR6 has a positive effect on the Wnt signaling pathway, it can also function as a tumor suppressor,⁶⁸ functioning via protein binding, transmembrane receptor activation and regulation of the Fzd proteins through ubiquitination.¹⁷⁵

With their almost 50% sequence identities, LGR4 (UniProt ID: Q9BXB1, 951 residues, also known as G-protein coupled receptor 48, GPR48), LGR5 (UniProt ID: O75473, 907 residues, also known as G-protein coupled receptor 49, GPR49), and LGR6 (UniProt ID: Q9HBX8, 967 residues) are closely related to each other and are characterized by similar domain organization. In fact, the N-terminal extracellular ectodomains domains (ECDs) of these 7 transmembrane receptors are composed of 17 leucine rich repeats (LRR) flanked by cysteine-rich sequences at both the N- and C-termini.¹⁷⁶ Their ECDs are substantially longer than the cytoplasmic C-terminal domains.

All three LGRs have several isoforms produced by alternative splicing. For example, the 62–85 region is missing in the isoform #2 of the LGR4, whereas isoforms #2 and #3 of the LGR5 are different from the canonical form of this protein by missing regions 263–286 and 143–214, respectively. In the isoform #2 of LGR6, the 1–71 region MPSPPGLRAL…GDLDPLTAYL is substituted to a shorter N-tail MRLEGEGRSARAGQNLSRAGSARRGAPR and the 144–239 region is missing, whereas isoform #3 of this protein differs from the canonical form possessing even shorter N-tail, MGRPRLTLVCQVSIIISAR.

Structural information is available for the ECDs of human LGR4 (residues 26–502; PDB ID: 4KT1),¹⁷⁷ and LGR5 (residues 22–543, PDB ID: 4BST).⁵⁹ In these structures, ectodomains of LGR4 and LGR5 were shown to possess a typical horseshoe-shaped structure consisting of the 17 leucine-rich repeat (LRR) units. Figure 11 illustrates that these 3 LGRs are characterized by rather similar disorder propensities, with their cytoplasmic C-terminal domains being mostly disordered. Interactomes of LGRs 4–6 are shown in Figure S4 that clearly illustrates their high binding promiscuity.

Figure 11. Evaluating disorder predisposition and some important disorder-related functional information evaluated for human LGR4 (UniProt ID: Q9BXB1, plot A), LGR5 (UniProt ID: O75473, plot B), and LGR6 (UniProt ID: Q9HBX8, plot C) by the D²P² database.

Concluding remarks

Our study shows that all human spondins and their major interacting partners are predicted to have significant levels of intrinsic disorder and possess a number of functionally important IDPRs. In fact, according to PONDR® VSL2 (which is a rather accurate stand-alone disorder predictor) all proteins considered in this article were predicted to have noticeable levels of disordered residues: 74.8% (ZnRF3), 71.2% (SCO-spondin), 69.2% (RNF43), 63.5% (Rspo1), 50.10% (F-spondin), 50.0% (Rspo4), 49.6% (Rspo3), 46.0% (mindin), 44.4% (Rspo2), 39.2% (Fzd8), 26.2% (LRP6), 25.3% (LRP5), 23.2% (LGR6), 19.9% (LGR5), 19.2% (Fzd4), and 16.0% (LGR4). To gain information on the average disorder propensity of these proteins, the MobiDB database (http://mobidb.bio.unipd.it/) was utilized.^178,179 Since MobiDB generates consensus disorder scores by aggregating the output from 10 predictors, such as 2 versions of IUPred,¹⁸⁰ 2 versions of ESpritz,¹⁸¹ 2 versions of DisEMBL,¹⁸² JRONN,¹⁸³ PONDR® VSL2B,^184,185 and GlobPlot,¹⁸⁶ it is likely that this database provides the most conservative estimates of intrinsic disorder in a query protein. However, even according to their MobiDB disorder consensus scores, spondins and their major interactors ranges from 1.3% to 48.8% and were arranged as follows: 48.8% (RNF43), 45.6% (ZnRF3), 40.1% (Rspo3), 34.4% (Rspo1), 29.5% (Fzd8), 25.2% (Rspo4), 16.9% (Rspo2), 12.9% (F-spondin), 12.7% (mindin), 12.1% (Fzd4), 11.3% (LRP6), 9.8% (SCO-spondin), 9.0% (LGR6), 8.6% (LRP5), 5.2% (LGR4), and 1.3% (LGR5). In other words, on average, spondins and their binding partners contain 20.3 ± 3 .8% (MobiDB) or 43.7 ± 5 .1% (PONDR® VSL2) disordered residues. Furthermore, these proteins are characterized by relatively low structural coverage of 24.6 ± 7 .1%, which is the percent of residues with known structure, with the structure being unknown for 8 of 16 proteins analyzed in this study. In fact, according to their structural coverage, spondins and their partners forms the following series: 75.9% (LRP6), 67.3% (mindin), 57.6% (LGR5), 52.8% (LGR4), 50.3% (F-spondin), 44.0% (Rspo1), 25.7% (Fzd4), 19.8% (RNF43), 0% (Rspo2), 0% (Rspo3), 0% (Rspo4), 0% (SCO-spondin), 0% (Fzd8), 0% (ZnRF3), 0% (LRP5), and 0% (LGR6)). Even if proteins with unknown structure would be excluded from consideration, the structural coverage of the remaining proteins is of 49.2 ± 6 .8%.

One can imagine multiple reasons for the lack of structural information about a query protein, ranging from the absence of attempts to solve its structure that reflects the absence of the overall interest in this protein, to the presence of some challenges associated with its crystallization. However, the “lack of interest” argument can be excluded for a protein with a partial structural coverage, since structure was determined for a part of protein sequence. It is known that the presence of intrinsic disorder can serve as one of the bottlenecks in structural characterization of a target protein that can affect several processes of the structure determination pipeline, such as expression and stability,^187,188 as well as solubility,^189,190 and predisposition for crystallization.^189,191-196 In application to spondins and their partners, this hypothesis is supported by the observation that proteins with more predicted disorder typically have lower structure coverage.

Data presented in our study revealed that spondins and their binding partners contain considerable levels of disorder and intrinsic disorder might be related to functionality of these proteins. Although presence of high levels of intrinsic disorder was previously reported for many proteins from the extracellular matrix, including collagens,^197,198 our analysis of spondins and their interactors represents the first detailed analysis of functional intrinsic disorder in these proteins. Furthermore, the abundance and functional roles of intrinsic disorder are not too often compared for intracellular vs. extracellular proteins involved in a particular biological process, and our study represents one of the first attempts to cover this gap. Structural disorder is typically associated with intracellular proteins or with intracellular regions of transmembrane proteins. We show here that similar to other signaling systems and regulatory pathways (which are mostly intracellular), intrinsic disorder is crucial for various functional aspects of extracellular spondins. In fact, their IDPRs are utilized in protein-protein interactions and serve as sites of various posttranslational modifications. It is tempting to hypothesize that the proteins covered in this study can serve as representatives of all extracellular proteins, meaning that structural disorder in the extracellular space is as general as observed here.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported in part by Chaikin-Wile foundation (IVZh).