Wnt modulators in the biotech pipeline



This article is corrected by:

  1. Errata: Wnt modulators in the biotech pipeline Volume 239, Issue 3, 1034, Article first published online: 28 January 2010


The purpose of this review is to provide a better understanding for the LRP co-receptor-mediated Wnt pathway signaling. Using proteomics, we have also subdivided the LRP receptor family into six sub-families, encompassing the twelve family members. This review includes a discussion of proteins containing a cystine-knot protein motif (i.e., Sclerostin, Dan, Sostdc1, Vwf, Norrin, Pdgf, Mucin) and discusses how this motif plays a role in mediating Wnt signaling through interactions with LRP. Developmental Dynamics 239:102–114, 2010. © 2009 Wiley-Liss, Inc.


Since its discovery nearly 30 years ago, Wnt signaling has been extensively studied for its diverse roles ranging from embryology, i.e., neuronal development and plasticity, bone development and formation, to causing diseases such as cancer (Johnson and Rajamannan, 2006; Coombs et al., 2008; Hoeppner et al., 2009). The name “Wnt” arose from the combination of the two homologous genes that were first identified in drosophila and vertebrates, namely, “wingless” and “Int.” Wingless (wg) was identified in 1980 by Nusslein-Volhard and Wieschaus as a Drosophila segment polarity gene, and their vertebrate homologs, INT genes, were identified in 1982 by Nusse and Varmus (Nusslein-Volhard and Wieschaus, 1980; Nusse and Varmus, 1982).

Wnt proteins are a family of 19 conserved secreted glycoproteins, several of which also encode alternatively spliced isoforms, which play a critical role in all multicellular animals, including cell proliferation, migration, and morphology. Classically, Wnt proteins initiated signaling by interacting with Frizzled, a family of ten members each with seven transmembrane spanning domains. To add complexity to the pathway, Wnt has also been shown to interact with four alternate Wnt receptors: RYK (Derailed), ROR, Crypto, and LRP (Bafico et al., 2001; Forrester et al., 2004; Inoue et al., 2004; Lu et al., 2004). Derailed (RYK) and ROR are single transmembrane domain tyrosine kinase receptors. Crypto, also known as teratocarcinoma-derived growth factor 1 (TDGF-1), is a member of the EGF-CFC protein family (Salomon et al., 2000). The LRP receptors, which are widely expressed scavengers, belong to the larger Low Density Lipoprotein (LDL) receptor subfamily. All of the above receptors, LRP, ROR, Crypto, and RYK, are presumed to act as a tertiary complex with Wnt and Frizzled (Fzd) to influence the Wnt signaling pathway (http://www.stanford. edu/∼rnusse/wntwindow.html). Signal specificity may be achieved through cell-specific expression of different Fzd receptors, which are capable of forming homo- and hetero-oligomers or through association of Fzd receptors with different co-receptors (i.e., LRP) (Feldman and Pizzo, 1986; Gliemann et al., 1986; van Driel et al., 1987; Weisgraber and Shinto, 1991; Gong et al., 1996; Dann et al., 2001; Carron et al., 2003; Stiegler et al., 2009). Understanding how extracellular Wnt ligands interact with transmembrane receptors to modulate the intracellular signaling cascades is, therefore, of broad importance to biology and to human disease.


There are three branches to the Wnt signaling pathway: (1) the beta-Catenin pathway (canonical pathway), (2) the planar cell polarity pathway (PCP; non-canonical), and (3) the Wnt/Ca+2 pathways. The Canonical pathway is the signaling pathway involved resulting in cancer, pattern formation, and osteogenesis to name but a few. It becomes active when Wnt ligand binds to Fzd and LRP to activate Dishevelled (Dsh), which is responsible for changing the amount of nuclear beta-Catenin. Dsh works to regulate the stability of proteins Axin, Gsk-3, and APC, which promotes the degradation of beta-Catenin (the beta-Catenin destruction complex). When Wnt activates Dsh, the beta-Catenin destruction complex is destabilized and beta-Catenin accumulates. It is thought that GSK-3 also plays a role in the formation of the Wnt, Fzd, LRP, Dsh, and Axin membrane receptor complex (Farr et al., 2000). This membrane complex leads to beta-Catenin accumulation in the nucleus where it can interact with the TCF/LEF family of transcription factors to control gene transcription (Brannon et al., 1997; Brunner et al., 1997).

The PCP pathway is distinct from the canonical Wnt pathway due to the lack of beta-Catenin involvement (Axelrod et al., 1998). Wnt signals through Fzd receptors to mediate asymmetric cytoskelatal organization and the polarization of cells. Within this pathway, Dsh triggers two independent pathways that are mediated through small G-proteins (GTPase), Rho or Rac. Rho requires Daam-1, whereas Rac is independent of Daam-1. Instead, Rac activates Jun Kinase (JNK) activity (Huelsken and Behrens, 2002; Habas and Dawid, 2005).

The Ca+2 pathway, not surprisingly, leads to the release of intracellular calcium (Arrazola et al., 2009). Here, Wnt binds Fzd and ROR, and the genes activated by this pathway are not well understood, but NFAT, PLC, and PKC all appear to be involved. This pathway is important for cell adhesion and movements during gastrulation (Habas and Dawid, 2005; Kohn and Moon, 2005). Adipogenesis, calcium homeostasis, and apoptosis are examples of processes regulated by non-canonical Wnt signaling.


The way in which a ligand is presented to its receptor has critical consequences to a ligand's ability to transduce a signal. Heparan sulfate proteoglycans (HSPGs) are major constituents of the extracellular matrix, which are implicated in the pathophysiology of diseases, including cancer, in which signals and tissue interactions malfunction (Selva and Perrimon, 2001; Nybakken and Perrimon, 2002). HSPGs, such as Glypicans and Syndicans, are soluble and membrane-intercalated proteins, which are composed of a core protein and decorated with covalently linked heparan sulfate (HS) chains (Bernfield et al., 1999). In Drosophila and mammalian systems, mutants that are completely deficient in HS sulfation have disrupted Wnt signaling (Reichsman et al., 1996; Lin et al., 1999; Toyoda et al., 2000; Dhoot et al., 2001). Since Wnt signaling is controlled by HS sulfation, the sheer presence of Wnt ligand is not sufficient to assume Wnt pathway function. The ability of Wnt to transduce a signal is dependent on its proper presentation by HSPG to either or both Fzd and its coreceptors, i.e., LRP.


In 1999, the LDL-receptor family was composed of five mammalian proteins and was thought to mainly mediate endocytosis. Today there are a total of 12 members of the LDL-receptor family and their functional roles are quite diverse and have yet to be fully understood (Table 1, Fig. 1). LRP4 and LRP11 were identified during our research for this review, on the basis of their protein sequence homology. All LRP receptors are type I transmembrane protein receptors (C-terminus in cytosol), with the exception of LRP4, which appears to have its amino terminus in the cytosol and thus a Type II transmembrane protein receptor (Fig. 1). An additional feature that the LDLR family shares is the highly conserved extracellular LDLR ligand binding repeat (LA, or complement-type cysteine rich repeat; shown as a green oval in Fig. 1). Most members of the family also share several structural features such as (1) epidermal growth factor (EGF)-precursor homology domains (blue ovals in Fig. 1), which themselves are composed of EGF repeats (cysteine-rich class B repeats) and spacer domains with YWTD propeller repeats (red propellers in Fig. 1), and mediate the acid-dependent dissociation of the ligands from the LDL receptor; (2) an O-linked sugar domain; and (3) an intracellular NPXY (or FXNPXY) sequence(s). LRP receptors also contain a putative cytoplasmic dishevelled (DSH) protein domain, which is specific to the signaling protein dishevelled (MotifFinder IPB003351A). This domain is found adjacent to the PDZ domain (PF00595) and often in conjunction with both DEP (PF00610) and DIX (PF0778) (Theisen et al., 1994). The presence of the putative DSH domain located in the cytoplasmic tail of LRP makes one question whether DSH also interacts with LRP. If this is so, there may be more LRP receptors acting as Wnt co-receptors than just LRP5 and 6.

Figure 1.

LRP family schematic. Note LRP4 is inverted for the simplicity of this diagram.

Table 1. LDLR Protein Family Membersa
1999 Nomenclature2009 Nomenclature
  • a

    Five LDLR receptors were identified before 1999, and today there is a total of 12 family members (new members are shown shaded in grey). All but LRP4 are Type I transmembrane protein receptors. LRP4 is a Type II transmembrane protein receptor. LRP4 and LRP11 were identified on the basis of protein sequence homology.

LRP (A2MR; alpha 2 macroglobulin receptor)LRP1
 LRP9 (LRP10, SORL1)

Of the several unique structural features that distinguish the LRP family, six subfamilies are discernable (I–VI; Table 2). The first (I) consists of the LDLR, VLDLR, and ApoER2 (LRP8) with seven LA domains, three EGF-like domains, and one YWTD propeller. The second (II) subfamily consists of LRP1 and LRP2, and their soluble isoforms (Quinn et al., 1997). These two receptors have the largest extracellular domains each with eight YWTD propellers spaced by EGF and LA domains. LRP4, LRP5, and LRP6 (third family, III) carry high-sequence homology to a region within LRP1, YWTD repeat three through six. This suggests that LRP4, 5, and 6 may have evolved from LRP subfamily II, with LRP4 undergoing a genomic inversion event. In addition to the sequence homology evidence suggesting LRP5 and 6 have evolved from LRP subfamily I, these receptors also share some of the same ligands (i.e., CTGF). LRP12 and LRP3 are the only receptors with CUB domains (CUB is from “complement C1r/C1s, Uegf, Bmp1” functions during embryogenesis) and compose the fourth (IV) subfamily. The unique LRP9 with six fibronectin domains (Fn3) makes up the fifth (V) subfamily, and the sixth (VI) subfamily consists of LRP11 possessing the only MANEC domain (a domain with eight conserved cysteines). The MANEC domain was previously named MANSC and is found in the N terminus of higher multicellular animal membrane and extracellular proteins. It is postulated that this domain may play a role in the formation of protein complexes involving various protease activators and inhibitors.

Table 2. The Different Members of Each LRP Subfamily
LRP subfamilyMembers

LDLR has a well-characterized role in the regulation of cholesterol metabolism. LRP1 and LRP2 (Megalin) are multifunctional and bind a diverse group of ligands, including ApoE. LRP1 has also been shown to interact with Fzd1 to down-regulate Wnt signaling (Zilberberg et al., 2004a). VLDLR, LRP8, and LRP9 can bind ApoE and LRP8 and VLDLR can bind Reelin. Reelin activates the tyrosine kinases and subsequent phosphorylation of the PTB protein (phosphotyrosine binding) domain containing adaptor protein (Dab1) in migrating neurons (Herrick and Cooper, 2004). LRP3 and LRP12 bind integrins and Golgi-associated proteins. The second best studied subfamily of LDL receptors is LRP5 and LRP6, which bind Wnt, Sclerostin, Wise, Dkk, and CTGF to name a few. Importantly, at the 2009 ASBMR meeting this past September, Leupin and coworkers (Novartis) presented results demonstrating that LRP4 is indeed capable of mediating WNT inhibition during bone formation (ASBMR 2009). The subfamilies have been shown to dimerize with, as homo or heterodimers, other LRP receptors within the family. For example, LRP6 forms inactive homodimers at the cell surface that are mediated through the extracellular EGF-like repeats (Liu et al., 2003); upon Wnt binding, this allosteric inhibition is relieved and an intracellular conformational switch leads to LRP6 activity.

Interestingly, WNT is not the only ligand for the Frz-LRP6 complex. Spondin (structurally similar to Mindin), which has a Reelin domain, also interacts with FZ and LRP6 (Nam et al., 2006). This interaction with R-Spondin takes place in the second YWTD propeller of LRP6, similar to Wnt, Wise, and Sclerostin. However, R-Spondin synergizes with Wnt to stimulate activity, and this activity can also be inhibited by Dkk1 (presumably Sclerostin and Wise also) (Kim et al., 2008). LRP subfamily 1 (ApoER2, VLDLR, LDLR) has also been shown to bind Reelin, but there have been no reports on its participation in Wnt signaling (May et al., 2005). In addition, one has to wonder if other LRP6 subfamily members (LRP5 and LRP4) bind Reelin and/or Spondin. Reelin is a large extracellular glycoprotein (Balthazart et al., 2008) similar to the Mucin family that contains a cystine knot protein motif (CK domain), and may function to stabilize receptor dimers (Tan et al., 2008).


The cystine-knot (CK) is a highly conserved protein motif found in many growth factors and other proteins as discussed above. The cysteine residues that form disulphide bridges are important in directing and maintaining the secondary and tertiary protein structures (Fig. 2D; grey boxes). A mutation in any of the cysteine residues or change in amino acid spacing would result in an incorrectly folded protein. This incorrectly folded protein would then result in the proteins' inability to bind to its target (LRP) receptor and modulate signaling, in turn, altering downstream gene expression required for normal cellular function. Changes to this “normal” cellular function may result in an altered or possiblly diseased cellular and tissue function.

Figure 2.

Published structure of Sclerostin protein in solution. A: Schematic representation of the structure. B: Contact surface view. C: Backbone amide chemical shift observed for the Sclerostin blocking monoclonal antibody binding to Sclerostin (amino acid sequence with high binding is displayed). D: Amino acid sequence of Sclerostin. Loop 2 displays the highest levels of binding activity. The amino acids displaying high binding affinity are shown in red, and cystein residues are bolded in grey boxes. Modified from Veverka et al. (2009).


Studies have now conclusively shown that many members of the above subfamilies do interact directly with LRP5 or LRP6 to modulate Wnt signaling. The best-characterized interactions are those from Sclerostin, Wise, and the CCN family member, CTGF (Fig. 3). The one common characteristic these proteins share is their cystine-knot domain (CK domain). Deletion of this CK domain or mutation of any one of its cysteine residues results in its inability to bind LRP resulting in altered Wnt signaling (Itasaki et al., 2003; Mercurio et al., 2004). Mercurio et al. (2004) used CTGF protein module deletions to report that it was the CK domain that was necessary for its interaction with LRP6 to inhibit Wnt signaling. The ability of Mucin, CCN, DAN, Norrin, and PDGF cystine-knot subfamilies to modulate Wnt and their possession of a CK domain begs the question of whether they also are able to directly interact with LRP5 or LRP6. Alternatively, since CTGF has been shown to bind to the LRP6 and LRP1 subfamily, can Sclerostin, Wise, PDGF, Norrin, CCN, or other cystine-knot proteins also bind to the LRP1 and other LRP subfamilies?

Figure 3.

Schematic showing the interaction domains between LRP/Fzd, Wnt, and WISE/Sclerostin Cystine-knot protein family. Dkk cystein-rich domains do not share high enough homology to the Wise/Sclerostin family. CTGF/PDGF/VEGF all bind to the sequence region on LRP1, which is homologous to the Sclerostin/Wise/CTGF region in the LRP III subfamily.


Sclerostin (Sost)/Wise (Ectodin; SOSTDC1; Usag1) Subfamily

The discovery of the human SOST (Sclerosteosis) gene is a triumph for the study of human genetics. Sclerosteosis and Van Buchem diseases are rare sclerosing bone dysplasias, discovered more than 40 years ago by Hansen et al. (1967), and are inherited in an autosomal recessive fashion. Each disease is manifested by a greatly increased amount of bone mass, with Sclerosteosis being the most severe (Beighton, 1976; Beighton et al., 1976a, b, 1977a, b; Balemans et al., 2001; Brunkow et al., 2001). Positional cloning revealed that both diseases are caused by a loss of function of the human SOST gene (Brunkow et al., 2001). Sclerosteosis is the result of point mutations causing either a termination or frame shift during protein translation. Alternatively, Van Buchem results from a 52K genomic deletion, 35 Kb downstream of SOST, which contains certain regulatory sequences responsible for regulation of Sost gene expression (Loots et al., 2005).

WISE (Ectodin, SOSTDC1, and USAG-1) was originally identified as a modulator of avian Hox gene expression. Hox genes are key regulators in anterio-posterior patterning (Itasaki et al., 2003). WISE maps to human chromosome 7p21.1, 10.6 Mb downstream of the HOXA cluster (Fig. 4). In a search for Wise family members, Ellies et al. (2006) found that SOST maps to chromosome 17q21.31, 5 Mb downstream of the HOXB complex (Ellies et al., 2006) (Fig. 4). Both loci have a similar structure and, in combination with their linkage to HOX complexes, seem to have arisen by duplication and divergence from a common ancestral chromosome region. Using Ensembl to search for genomic similarity to SOST or Wise, other putative family members, linked to the HOXC and HOXD complex, are yet to be identified. Human contig sequence AC011316 found at 12q13.11 is linked to HOXC and is 63% homologous to exon 2 of SOST (Fig. 4). At this time, it is unknown if this sequence encodes a functional protein. An important question is whether other sequences exist that are homologous to SOST or WISE.

Figure 4.

Schematic showing the chromosomal locations of Wise and SOST, alongside their HOX clusters.

Based on their weak protein sequence similarity to the DAN and CCN family of cystine-knot proteins, which themselves bind BMPs, Sclerostin (protein product of the SOST gene) was initially postulated to exert its function via its biochemical ability to bind and inhibit BMP signaling (Brunkow et al., 2001; Kusu et al., 2003; Winkler et al., 2005; Ellies et al., 2006). In cell culture models, using alkaline phosphatase activity as a late marker for BMP-mediated osteoblast differentiation, Sclerostin was found to inhibit BMP6 but not BMP4. In an ATDC-5 cell assay, Wise did not inhibit BMP6 and only weakly influenced BMP4-dependent activity at a very high concentration (Ellies et al., 2006). Therefore, Wise has a preference for inhibiting BMP7, BMP2, and BMP4, and, more weakly, BMP6. Despite the significant homology between Wise and Sclerostin, they are not identical in their functional ability to inhibit BMP signaling.

Both Wise and Sclerostin function to inhibit Wnt signaling in in-vitro assays by binding to YWTD propeller 1 and 2 of LRP5 or LRP6 (Ellies and Krumlauf, 2002; Itasaki et al., 2003). Wise, but not Sclerostin, also acts as a stimulator or, conversely, a mild repressor of the Wnt pathway (Ellies and Krumlauf, 2002; Itasaki et al., 2003). Biochemical competitive studies have revealed that Wise has a higher affinity, than Wnt, for LRP6 (Itasaki et al., 2003). It remains unclear if (1) Sclerostin and Wise can bind simultaneously to LRP and (2) which of the ligands (CTGF, Sclerostin, Wise, DKK, Wnt) contain the highest affinity for LRP5 and LRP6, the LRP3-subfamily. A recent study has very nicely shown that BMP/BMPR1A can effectively repress 90% of SOST gene expression, which ultimately results in the activation of the canonical Wnt pathway (Kamiya et al., 2008).

Like its closest family member, Sclerostin, Wise encodes a 200–amino acid, 28-kDa, secreted protein containing a leader sequence and a cysteine-rich protein domain (CK domain; Fig. 5; SMART domain SM00041). Proteins containing this CK domain are structurally related and share a common overall topology. These proteins have very little overall sequence homology, but they all have an unusual arrangement of six cysteines linked to form a “cystine-knot” conformation (Ellies et al., 2006). The active forms of these proteins are thought to be dimers, either homo- or heterodimers. Because of their shape, there appears to be an intrinsic requirement for the cystine-knot growth factors to form dimers. This extra level of organization increases the variety of structures built around this simple structural motif (Fig. 2A, B).

Figure 5.

Schematic showing an alignment of highly conserved protein domains from cystine-knot proteins most homologous to Sclerostin and Wise.

Significant similarity exists within the cystine-knot motifs from DAN (Cerberus, DAN, Gremlin, Caronte), CCN (NOV, CTGF, Cyr61), Slit, and Mucin family members (CK domain, Fig. 5). There is also homology to the cysteine motifs in individual genes, such as Von Willderbrand Factor (VFW), PDGF, and Norrie Disease Protein (NDP) (Fig. 5). They all contain a consensus organization of eight core cysteine residues and one glycine residue (between cys-3 and cys-4) (Ellies et al., 2006). In contrast to Wise and Sclerostin, the other subfamilies contain one to two additional cysteine (cys) residues between cys-4 and cys-5 of the core motif (Ellies et al., 2006). The cystine-knot domain is the only recognizable motif found in the Wise/Sclerostin, Norrie, and DAN subfamilies, whereas members of the other groups contain multiple motifs, such as insulin binding, Von Willderbrand, and TSP1 domains.

CCN Subfamily

The first family member for this subfamily was discovered in 1985 by Lau and Nathans. These proteins are 30–40 kDa and extremely cysteine rich (10% by mass), and multimodular (Brigstock, 1999; Perbal, 2001). The CCN family includes Cysteine-rich 61 (Cyr61/CCN1), Connective Tissue Growth Factor (CTGF/CCN2), Nephroblastoma Overexpressed (NOV/CCN3), and Wnt Induced Secreted Proteins-1 (Wisp-1/CCN4), -2 (Wisp-2/CCN5), and -3 (Wisp-3/CCN6). The function of the protein domain modules is not well understood (Fig. 5). Module 1 is an insulin-like growth factor–binding domain, module 2 is a Von Willebrand type-C domain, module 3 is a thrombospondin-1 domain, and module 4 is a C-terminal domain containing a putative cystine-knot. These proteins all can stimulate mitosis, adhesion, apoptosis, extracellular matrix production, growth arrest, and migration (hormone action, skeletal growth, placental angiogenesis, and diabetes-induced fibrosis). Many of these activities are born from their ability to bind and activate integrins. However, it has also been reported that CTGF/CCN2 can also bind to either LRP1 or LRP6 (Segarini et al., 2001; Gao and Brigstock, 2003; Mercurio et al., 2004). Mercurio and colleagues (2004) showed very eloquently that the C-terminal (CK domain), domain 4, of CTGF interacted with LRP6 to inhibit Wnt signaling. This demonstrates that domain 4 (CK domain) of the CCN family may have a role in regulating the Wnt pathway through its interaction with the LRP receptors. It is also interesting that module 3 contains a thrombospondin-1 domain. As we discussed above, Spondin interacts with the LRP 5/6 subfamily, and Spondin is a family member to thrombospondin. Could this thrombospondin-1 domain also interact with the LRP receptors?

DAN Subfamily

The DAN subfamily is an evolutionarily conserved group of proteins that function as transforming growth factor (TGF) Beta or BMP antagonists. Gremlin was the first member of this group that was identified in 1998 by Hsu et al. (1998). The subfamily is made up of the following genes: Dan (differential screening-selected gene aberrative in neuroblastoma), Cerberus, Gremlin, PRDC, and several other genes that have only been identified as expressed sequence tags. They have a very important role during development of the embryo; with that said, little is known about their biological role in the adult body plan. This protein family shares a cysteine-rich domain (the CK domain) and has much resemblance to the Mucins, a family of secreted factors found in mucus (Fig. 5). It is unclear whether the similarity to Mucin is functionally meaningful. One family member, Cerberus, finds itself apart from the rest of the Dan subfamily, due to its unique ability to inhibit Wnt signaling. It remains unclear if this ability to inhibit Wnt signaling may involve an interaction between its CK-domain and LRP receptor.

Mucin/Slits Subfamily

The Mucin subfamily contains 10 proteins: MUC1 (Mucin 1 transmembrane), MUC2 (Mucin 2), MUC4 (Mucin 4), MUC5AC (gastric), MUC6 (Mucin 6 gastric), MUC7 (Mucin 7 salivary), MUC13 (Mucin 13, epithelial transmembrane), MUC16 (Mucin 16), MUC19 (Mucin 19), and OVGP1 (Oviductal glycoprotein 1). Mucins are high molecular weight glycoproteins that are secreted by epithelial tissues. Indirect evidence suggests that these genes might encode the core protein of parasite Mucins, proposed to be involved in the interaction with, and invasion of, mammalian host cells. The subfamily harbors three repeated protein domains (Fig. 5). The first is a VWD domain and it is involved in multimerisation of proteins. The second is the CK domain containing 8 conserved cysteine residues. The third is the TIL domain, which consists of ten cysteine residues making 5 disulfide bonds. The N and C termini are conserved between all members of the family, whereas the central region is not well conserved but contains a large number of threonine residues that can be glycosylated. Recent evidence points to the ability of MUC to modulate the Wnt pathway though an interaction with the intracellular Wnt machinery, B-Catenin.

Members of the Slit protein family share a number of structural features with Mucins, such as the N-terminal leucine-rich repeats, the C-terminal epidermal growth factor–like motifs, and the cysteine-rich (CK) domain. The entire family of five rat and human Slit genes encode for a 1,523–1,530–amino acid protein. Signaling by these proteins requires the presence of heparan sulfate chains (Hu, 2001). Recently, Prasad et al. (2008) show for the first time that Slit has the capacity to modulate Wnt signaling via the modulation of Beta-Catenin.

Platelet-Derived Growth Factor (PDGF)

PDGF is a growth factor that plays a significant role in controlling blood vessel formation (angiogenesis) from already existing blood vessel tissue. Uncontrolled angiogenesis due to uncontrolled PDGF is a trait of cancer. The PDGF subfamily was for more than 25 years assumed to consist of only PDGF-A and –B. However, with the recent discovery of members PDGF-C and PDGF-D, the subfamily is composed of four members. With little information on PDGF-C and -D protein structures to confirm or debate, PDGF-C protein appears to resemble VEGF-A structurally. The classical PDGF polypeptide chains, PDGF-A and PDGF-B, are well studied and they regulate a number of processes via two receptor tyrosine kinases, PDGF receptors α and β. In addition to their ability to bind to PDGF receptors, PDGF-B has also been reported to bind to LRP1, and Zilberberg et al. (2004b) reported that LRP1 is able to down-regulate Wnt signaling by interacting with Fzd. The interaction of PDGF with LRP has no apparent link, as of yet, to the Wnt signaling pathway. However, a recent study by Cohen et al. (2009) has reported a correlation between the expression of Wnt7a and the expression of PDGFR-A and -B. It is intriguing that PDGF proteins contain a CK domain similar to those from Sclerostin, Wise, and CTGF, and that PDGF-B has been shown, like CTGF, to interact directly with LRP1. However, it is yet to be determined whether PDGF has the ability to interact with LRP to modulate Wnt signaling (Fig. 5).


The Norrin protein is involved in Norrie Disease, an X-linked recessive disorder resulting in an incomplete development of retinal vasculature. Norrie Disease Protein (NDP) was originally discovered over a decade ago and named EVR2 (Toomes et al., 2004). Many retinopathies are classified as NDP-related: persistent fetal vasculature syndrome (PFVS), retinopathy of prematurity (ROP), Coats disease, and X-linked familial exudative vitreoretinopathy (X-linked FEVR). Norrie disease results from either point mutations or deletions, 70% of which affect a cysteine residue.

The Norrie gene encodes a small secreted 133–amino acid protein termed norrin. The norrin protein contains only one protein domain, the cystine-knot (CK domain), which is encoded from exon 2 and exon 3 (Fig. 5). The cystine-knot is involved in the formation of covalently-linked dimers, which ultimately leads to pathway activation. Norrin acts as an activator for the Wnt pathway through its interaction with Fzd4 and LRP5 or LRP6 (Xu et al., 2004). Yet the exact nature of this interaction with LRP5 or LRP6 is not clearly understood. The possibility that Norrin may interact directly with LRP has been tainted by studies showing that it was unable to bind LRP6. This does not omit the possibility that Norrin may bind to LRP5, as patients with mutations in LRP5 also present with vascular eye defects.

Von Willebrand Factor (VWF)

VWF is a glycoprotein involved in hemostasis. As a monomer, VWF is a 2,050–amino acid protein with five specific domains (Fig. 5). The first is the VWD domain, which binds to coagulation factor VIII. The second, the VWA domain, binds to the platelet GPI b-receptor, heparin, and possibly collagen. The third is the VWC domain that binds to platelet integrin αIIbβ3. The fourth is the “cystine knot” domain (CK). The last and fifth domain is the TIL domain, which consists of ten cysteine residues making 5 disulfide bonds. VWF monomer proteins are subsequently N-glycosylated and multimerize into functional proteins consisting of more than 80 subunits of 250 kDa each making the multimer >20,000 kDa. VWF protein contains domains that help in the making of blood clots. VWF is made in the endothelial cells that line the inside surface of blood vessels and bone marrow cells. VWF helps platelets stick together and adhere to the walls of blood vessels at the side of an injury. VWF also carries coagulation factor VIII to the area of the clot formation. No link to the Wnt pathway has been made as of today. Yet, coagulation factor VIII, which binds to the VWD domain, is known to interact with LRP to help in its clearance from the plasma (Lenting et al., 1999). No link to Wnt signaling has been made as of today.


Seminal discoveries have uncovered the importance of the Wnt signaling pathway during development, cell biology, and physiology. The importance of this pathway is reflected in the growing number of human diseases that are the consequence of abnormal Wnt signaling. Twenty-three human diseases have been associated with mutations in the Wnt pathway (Table 3). This number is small considering that there are 19 Wnt and 10 Fzd proteins.

Table 3. Diseases Associated With Mutations Within Genes of the Wnt Signaling Pathway
APC- Polyposis coli (Kinzler et al., 1991; Nishisho et al., 1991)
LRP5- Bone density (van Meurs et al., 2008)
- Vascular defects in the eye (Osteoperosis-pseudoglioma Syndrome, OPPG) (Gong et al., 2001; Boyden et al., 2002; Little et al., 2002)
LRP5- Familial Exudative Vitreoretinopathy (Toomes et al., 2004; Qin et al., 2005)
LRP6- Early coronary disease (Mani et al., 2007)
LRP6- Osteoporosis (Mani et al., 2007)
LRP6- Late onset Alzheimer (De Ferrari et al., 2007)
FZD4- Familial Exudative Vitreoretinopathy: retinal angiogenesis (Robitaille et al., 2002; Qin et al., 2005)
Norrin- Familial Exudative Vitreoretinopathy (Xu et al., 2004)
WNT3- Tetra-Amelia (Niemann et al., 2004)
WNT4- Mullerian-duct regression and virilization (Biason-Lauber et al., 2004)
WNT4- SERKAL syndrome (Mandel et al., 2008)
WNT5B- Type II diabetes (Kanazawa et al., 2004)
WNT7A- Fuhrmann syndrome (Woods et al., 2006)
WNT7A- Al-Awadi/Raas-Rothschild/Schinzel Syndrome (Woods et al., 2006)
WNT10A- Odonto-onycho-dermal dysplasia (Adaimy et al., 2007)
WNT10B- Obesity (Christodoulides et al., 2006)
WNT10B- Split-Hand/Foot Malformation (Ugur and Tolun, 2008)
AXIN1- Caudal duplication (Oates et al., 2006)
TCF7L2 (TCF4)- Type II diabetes (Florez et al., 2006; Grant et al., 2006; O'Rahilly and Wareham, 2006)
AXIN2- Tooth agenesis (Lammi et al., 2004)
DACT1- Adipogenesis (Lagathu et al., 2009)
WTX- Wilms tumor (Major et al., 2007; Rivera et al., 2007)
- Sclerosing skeletal dysplasia (Jenkins et al., 2009)
PORC1- Focal dermal hypoplasia (Grzeschik et al., 2007; Wang et al., 2007)
WIF- Osteosarcoma (Kansara et al., 2009)
RSPO4- Autosomal recessive anonychia (Bergmann et al., 2006; Blaydon et al., 2006)
VANGL1- Neural tube defects (Kibar et al., 2007)
SOST- Bone density (Brunkow et al., 2001)

In addition to Wnt and Fzd, Wnt co-receptors like LRP5 and LRP6 have genetic diseases associated with them. Human mutations in LRP5 cause bone density effects, eye vascular effects, and FEVR (Table 3). HLRP6 mutations cause early onset coronary disease, osteoporosis, and late onset Alzheimer's, and increased LRP5/LRP6 is associated with colon cancer (Table 3). Mutations in the SOST gene cause bone density effects. This list will increase once our understanding of the role of the other (ROR, RYK, and Crypto) Wnt co-receptors is advanced.

In addition to the genetic mutations described above, other diseases have been linked to the uncontrolled “abnormal” signaling of the Wnt pathway. Increased levels of Fzd are associated with gastric cancer, leukemia, kidney cancer, and liver cancer. The intracellular Wnt pathway components comprise beta-Catenin, APC, Axin, and GSK-3. Human genetic mutations in APC impair beta-Catenin degradation or in beta-Catenin itself, which stabilizes beta-Catenin, both result in constitutively active Wnt signaling and have been linked to familial adenomatous olyposis and cancer (colon, desmoid, endometrial, gastric-intestinal, hepatocellular, kidney, Wilms tumor, lung, medulloblastoma, melanoma, ovarian, pancreatic, pilomatricoma, prostate, esophageal, thyroid, and uterine) (Baehs et al., 2009; Barker et al., 2009; Bisson and Prowse, 2009; Bjorklund et al., 2009; Cheah, 2009; Chung et al., 2009; Han, 2009; Hirata et al., 2009; Jiang et al., 2009; Lu et al., 2009; Monga, 2009; Wang et al., 2009; Yamamoto et al., 2009; Yang et al., 2009). There has also been an association of hepatocelluar carcinoma with mutations in human Axin. Reduced Wnt signaling has been observed in the transition from hypertrophy to failing heart (Schumann et al., 2000).

With diseases comes a therapy. In Table 4, one can peek at what is being developed for commercial therapeutics around the Wnt pathway. Table 4 separates the therapeutics into 20 intracellular and 8 extracellular Wnt therapies. A therapeutic is categorized in one of two pools: (1) a biologic (protein or antibody) or (2) a small molecule. The Wnt extracellular therapeutics consists of 7 biologic and 3 small molecules, whereas the intracellular therapeutic area is vastly dominated by small-molecule therapeutics. The apparent focus is centered around the intracellular GSK3 target; 83% of the small-molecule technology surrounds inhibiting GSK-3. However, potential problems exist with the long-term use of GSK3 inhibitors. GSK3 inhibitors would be expected to mimic the overexpression of Wnt signaling and, therefore, may become oncogenic.

Table 4. Wnt Modulators in the Biotech Pipeline
 CompanyTherapeuticPathway TargetTechnologyDevelopment Stage
Extracellular Modulators
 BoneNuveloLRP5 MabLRP5BiologicDiscovery
 BoneNuveloDkk1 MabDkk1BiologicDiscovery
 CancerFibrogenCTGF MabCTGFBiologicPreclinical
 BoneAmgenSclerostin MabSOSTBiologicPhase II
 BoneNovartisSclerostin MabSOSTBiologicPreclinical
 BoneEli LillySclerostin MabSOSTBiologicPreclinical
 CancerWyethWAY-316606SFRPSmall MoleculePreclinical
 UTSWNiclosamideFrizzledSmall MoleculeDiscovery
 BoneEnzo BiochemSmall Mole NCIDkkSmall moleculePreclinical
 BoneOsteoGeneXAnti-SclerostinSOST/LRPSmall moleculelead opt, preclinical
 BoneGalapagoswnt pathwayLRP5Small moleculeDiscovery
Intracellular Modulators
 CancerNovartisXAV939tankyrase1/AxinSmall moleculeDiscovery
 CancerUTSWIWRAxinSmall moleculeDiscovery
 Asahi Kasei CorporationIQ1PP2ASmall moleculeDiscovery
 CancerScrippsQS11ARFGAP1Small moleculeDiscovery
 CancerSt. Jude Children'sNSC668036DshSmall moleculeDiscovery
 DiabetesSmithKline BeechamSB-216763GSK3Small moleculeAbandoned
 Dia betesSmithKline BeechamSB-216763GSK3Small moleculeAbandoned
 Stem cell renewalRockefellerBIO(6-bromoin dirubin-3′ -oxime)GSK3Small moleculeDiscovery
 CancerDepartment of Veterans Affairs Medical CenterDCAbeta-cateninSmall moleculeDiscovery
 Scripps/Novartis(methyl enedioxy) benzyl-amino]- 6-(3-unknownSmall moleculeAbandoned
 CancerWyeth2,4-diamino- quinazolineTCF/ beta-cateninSmall moleculeLead Optimization
 CancerSeoul National UniversityQuercetinTCFSmall moleculeAbandoned
 CancerInstitute for Chemical GenomicsICG-001CREB-binding proteinSmall moleculeAbandoned
 CancerDana Farberseveral otherTCF/ beta-cateninSmall moleculeAbandoned
 Cancer, Bone, ObesityOsteoGeneXwnt pathwayvarious wnt pathwaySmall moleculelead opt, preclinical
 CancerAval onAVN316BcateninSmall moleculelead candidate selected
 CancerCuriswnt pathwayWnt pathwaySmall moleculeDiscovery
 CancerCelon WNTRNA 
 CancerEthical Oncologywnt pathwayBCL-beta-catenin lead optimization
 AlzheimersNeuropharma SAwnt pathwaygsk3 inhibitSmall moleculephase I
 DiabetesDeveloGen AGwnt pathwaygsk3 inhibitSmall moleculepreclinical
 DiabetesEli-Lillywnt pathwaygsk3 inhibitSmall moleculepreclinical
 BoneRochewnt pathwaygsk3 inhibitSmall moleculepreclinical
 CancerDeci pherawnt pathwaygsk3 inhibitSmall moleculepreclinical
 DiabetesCrystal Genomicswnt pathwaygsk3 inhibitSmall moleculepreclinical
 DiabetesCyclacelwnt pathwaygsk3 inhibitSmall moleculepreclinical
 Neuro, Metabolic, PainAmphorawnt pathwaygsk3 inhibitSmall moleculepreclinical
 DiabetesKemiawnt pathwaygsk3 inhibitSmall moleculepreclinical
 DiabetesChironwnt pathwaygsk3 inhibitSmall moleculepreclinical
 AlzheimersAstraZenicawnt pathwaygsk3 inhibitSmall moleculephase I- Discontinued
 Alzheimers, DiabetesMitsubishiwnt pathwaygsk3 inhibitSmall moleculepreclinical - no dev since 2007
 All IndicationsXcellsyzwnt pathwaygsk3 inhibitSmall moleculepreclinical-no dev since 2005
 Diabetes, CancerKinetek Pharmawnt pathwaygsk3 inhibitSmall moleculepreclinical - no dev since 2005
 DiabetesNOVO Nordiskwnt pathwaygsk3 inhibitSmall moleculePreclinical - no dev since 2004
 CancerUTSWIWPPorcupine Discovery

A more favorable approach to the modulation of the Wnt pathway has been to focus on extracellular mediators of the pathway. Currently, Amgen is the first in class to develop a biologic therapeutic against Sclerostin (Human Clinical Phase II). Nuvelo is following Amgen with biologics against LRP5, Dkk1, and R-Spondin (Discovery). Second in class for Sclerostin-blocking antibodies will be Novartis and Eli Lilly (Preclinical). Fibrogen, which is taking a different approach, has developed a biologic against CCN family member CTGF (Preclinical). As for small molecules, OsteoGeneX is first in class to develop a Sclerostin small-molecule inhibitor, currently in preclinical and lead optimization. Alternatively to Sclerostin, Galapagos is developing small-molecule leads against LRP5 in a partnership with Eli-Lilly (Discovery).


In summary, one important point that must be remembered when thinking of the Wnt signaling pathway is that it is not an off-on situation, and that it is cell specific. Wnt signaling is not binary; it requires fine tuning in order to be properly controlled during normal physiology (Fig. 6). Imagine a scale from 0–150%, where 0% represents no Wnt activity and 150% represents 50% over normal Wnt activity. At 150%, Wnt is more than constitutively activated and this leads to cancer. One hundred percent Wnt activity leads to high bone mass, and 60% Wnt activity leads to normal bone physiology. Heart failure and osteoporosis would be found at 0% Wnt activity. This modulation is critical and developing therapeutics against the Wnt pathway is tricky. Finding the right target and making sure it acts on the target and nowhere else within the Wnt pathway is vital to making sure the therapeutic will not be triggering off-target side effects especially as reflected in the unwanted scale above.

Figure 6.

Universal schematic of Wnt Thermostat. High reading leads to cancer, whereas a low reading leads to heart failure or osteoporosis depending on the cell type. Modulation of the Wnt pathway is not binary; one needs to fine-tune the signal to receive the desired outcome.


The authors are supported by grants to D.L.E. from the Kansas Bioscience Authority, US Department of Health and Human Services, the National Institute of Health, and the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS).