Review Matrix Gla protein and osteocalcin: From gene duplication to neofunctionalization M. Leonor Cancela a,b,⇑,1 , Vincent Laizé a,1 , Natércia Conceição a,1 a Centre of Marine Sciences, University of Algarve, 8005-139 Faro, Portugal b Department of Biomedical Sciences and Medicine, University of Algarve, 8005-139 Faro, Portugal article info Article history: Received 8 April 2014 and in revised form 4 July 2014 Available online xxxx Keywords: Matrix Gla protein Osteocalcin Protein structure Gene architecture Splicing variants Gene duplication Molecular evolution abstract Osteocalcin (OC or bone Gla protein, BGP) and matrix Gla protein (MGP) are two members of the growing family of vitamin K-dependent (VKD) proteins. They were the first VKD proteins found not to be involved in coagulation and synthesized outside the liver. Both proteins were isolated from bone although it is now known that only OC is synthesized by bone cells under normal physiological conditions, but since both proteins can bind calcium and hydroxyapatite, they can also accumulate in bone. Both OC and MGP share similar structural features, both in terms of protein domains and gene organization. OC gene is likely to have appeared from MGP through a tandem gene duplication that occurred concomitantly with the appearance of the bony vertebrates. Despite their relatively close relationship and the fact that both can bind calcium and affect mineralization, their functions are not redundant and they also have other unrelated functions. Interestingly, these two proteins appear to have followed quite different evo- lutionary strategies in order to acquire novel functionalities, with OC following a gene duplication strat- egy while MGP variability was obtained mostly by the use of multiple promoters and alternative splicing, leading to proteins with additional functional characteristics and alternative gene regulatory pathways. Ó 2014 Elsevier Inc. All rights reserved. Introduction Osteocalcin 2 (OC or bone Gla protein, BGP) and matrix Gla pro- tein (MGP) are two members of the growing family of vitamin K- dependent (VKD) proteins, and the first found not to be involved in coagulation and being synthesized outside the liver [33,78,73,74]. Indeed, shortly after the discovery of Gla as a novel amino acid residue derived from the c-carboxylation of glutamate [88] and essential for the blood clotting capability of several coagu- lation factors, bone was found to contain high amounts of Gla sug- gesting the presence of Gla-containing proteins in this tissue. The follow up from this finding culminated with the purification of the bone Gla protein/osteocalcin from the mineralized matrix of bovine bone, which accounted for up to 2% of the total proteins of bone [73]. However, already in 1980 Price et al. predicted that OC was not the only Gla protein in bone since, when analyzing fetal bone, they could not extract its Gla content by demineralization, which suggested that it was not from osteocalcin and argued in favor of the presence of another Gla-containing protein in fetal bone but associated with the collagenous matrix [72]. This hypothesis proved to be correct and matrix Gla protein was later purified from the organic matrix of bone [78,77] and thought to account for the remaining Gla content of bone, since the presence of additional Gla proteins in bone was not anticipated at the time. OC and MGP share with the other mem- bers of this family their capability of binding calcium and calcified matrices through interaction with their Gla residues, which result from a c-carboxylation of glutamate residues, a post translation modification dependent of vitamin K and catalyzed by the enzyme c-glutamyl carboxylase, a ubiquitous protein found both in verte- brates and in invertebrates, and more recently also in bacteria [81] although its targets in the latter remain essentially unknown. War- farin, a vitamin K antagonist discovered in the late 1950s [51], is capable of inhibiting this process resulting in the appearance of undercarboxylated Gla proteins, a process that negatively affects their calcium binding capabilities and thus their established func- tions [17,69]. Both proteins were found to accumulate in bone, although osteocalcin was later identified as being secreted under normal, non-pathological conditions, only by osteoblasts, odontoblasts http://dx.doi.org/10.1016/j.abb.2014.07.020 0003-9861/Ó 2014 Elsevier Inc. All rights reserved. ⇑ Corresponding author at: Department of Biomedical Sciences and Medicine, University of Algarve, Campus de Gambelas, 8005-139 Faro, Portugal. Fax: +351 289 800971. E-mail address: [email protected](M.L. Cancela). 1 All authors contributed equally. 2 Abbreviations used: OC, osteocalcin; BGP, bone Gla protein; MGP, matrix Gla protein; VKD, vitamin K-dependent; Gla, c-carboxylated glutamate; OA, osteoarthri- tis; ucMGP, uncarboxylated MGP; cMGP, carboxylated MGP; AnaOC, Adriatic sturgeon osteocalcin. Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx Contents lists available at ScienceDirect Archives of Biochemistry and Biophysics journal homepage: www.elsevier.com/locate/yabbi Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys. (2014), http://dx.doi.org/10.1016/j.abb.2014.07.020
Transcript
Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
journal homepage: www.elsevier .com/ locate/yabbi
Review
Matrix Gla protein and osteocalcin: From gene duplicationto neofunctionalization
http://dx.doi.org/10.1016/j.abb.2014.07.0200003-9861/� 2014 Elsevier Inc. All rights reserved.
⇑ Corresponding author at: Department of Biomedical Sciences and Medicine,University of Algarve, Campus de Gambelas, 8005-139 Faro, Portugal. Fax: +351 289800971.
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys. (2014), http://dx.doi.org/10.1016/j.abb.2014.07.020
M. Leonor Cancela a,b,⇑,1, Vincent Laizé a,1, Natércia Conceição a,1
a Centre of Marine Sciences, University of Algarve, 8005-139 Faro, Portugalb Department of Biomedical Sciences and Medicine, University of Algarve, 8005-139 Faro, Portugal
a r t i c l e i n f o a b s t r a c t
Article history:Received 8 April 2014and in revised form 4 July 2014Available online xxxx
Osteocalcin (OC or bone Gla protein, BGP) and matrix Gla protein (MGP) are two members of the growingfamily of vitamin K-dependent (VKD) proteins. They were the first VKD proteins found not to be involvedin coagulation and synthesized outside the liver. Both proteins were isolated from bone although it isnow known that only OC is synthesized by bone cells under normal physiological conditions, but sinceboth proteins can bind calcium and hydroxyapatite, they can also accumulate in bone. Both OC andMGP share similar structural features, both in terms of protein domains and gene organization. OC geneis likely to have appeared from MGP through a tandem gene duplication that occurred concomitantlywith the appearance of the bony vertebrates. Despite their relatively close relationship and the fact thatboth can bind calcium and affect mineralization, their functions are not redundant and they also haveother unrelated functions. Interestingly, these two proteins appear to have followed quite different evo-lutionary strategies in order to acquire novel functionalities, with OC following a gene duplication strat-egy while MGP variability was obtained mostly by the use of multiple promoters and alternative splicing,leading to proteins with additional functional characteristics and alternative gene regulatory pathways.
� 2014 Elsevier Inc. All rights reserved.
Introduction
Osteocalcin2 (OC or bone Gla protein, BGP) and matrix Gla pro-tein (MGP) are two members of the growing family of vitamin K-dependent (VKD) proteins, and the first found not to be involvedin coagulation and being synthesized outside the liver[33,78,73,74]. Indeed, shortly after the discovery of Gla as a novelamino acid residue derived from the c-carboxylation of glutamate[88] and essential for the blood clotting capability of several coagu-lation factors, bone was found to contain high amounts of Gla sug-gesting the presence of Gla-containing proteins in this tissue. Thefollow up from this finding culminated with the purification of thebone Gla protein/osteocalcin from the mineralized matrix of bovinebone, which accounted for up to 2% of the total proteins of bone [73].However, already in 1980 Price et al. predicted that OC was not the
only Gla protein in bone since, when analyzing fetal bone, they couldnot extract its Gla content by demineralization, which suggestedthat it was not from osteocalcin and argued in favor of the presenceof another Gla-containing protein in fetal bone but associated withthe collagenous matrix [72]. This hypothesis proved to be correctand matrix Gla protein was later purified from the organic matrixof bone [78,77] and thought to account for the remaining Gla contentof bone, since the presence of additional Gla proteins in bone wasnot anticipated at the time. OC and MGP share with the other mem-bers of this family their capability of binding calcium and calcifiedmatrices through interaction with their Gla residues, which resultfrom a c-carboxylation of glutamate residues, a post translationmodification dependent of vitamin K and catalyzed by the enzymec-glutamyl carboxylase, a ubiquitous protein found both in verte-brates and in invertebrates, and more recently also in bacteria [81]although its targets in the latter remain essentially unknown. War-farin, a vitamin K antagonist discovered in the late 1950s [51], iscapable of inhibiting this process resulting in the appearance ofundercarboxylated Gla proteins, a process that negatively affectstheir calcium binding capabilities and thus their established func-tions [17,69].
Both proteins were found to accumulate in bone, althoughosteocalcin was later identified as being secreted under normal,non-pathological conditions, only by osteoblasts, odontoblasts
2 M.L. Cancela et al. / Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx
and cementoblasts [8,35,61], while MGP was found to be expressedmainly by chondrocytes and fibroblasts [11,28], smooth musclecells [85] and more recently by tooth cementum [30] and trabecu-lar meshwork cells from the eye [25,90].
Both OC and MGP share similar structural features, both interms of protein domains and gene organization. However, previ-ous work has demonstrated that the evolutionary appearance ofMGP is likely to have preceded osteocalcin, both proteins sharinga common ancestor [12,47,80]. Furthermore, the earlier appear-ance of MGP is concomitant with the development of cartilage-liketissues in ancestral vertebrates, being still debatable if it is presentin the jawless fish (Agnatha). In contrast, OC appears to haveevolved together with the appearance of bone tissue in bony verte-brates (Osteichthyes; [47] and references therein). This articlereviews the knowledge associated with the discovery of OC andMGP and their structural domains, and the landmark discoveriesthat culminated with the present understanding of their functions.Although both proteins have a common ancestor, our data suggeststhat each protein followed distinct evolutionary strategies toachieve the present diversification of proposed functions.
MGP and OC: protein structure and functional domains
MGP and osteocalcin share some common protein features butthey also have their own individuality (Fig. 1). Both are smallsecreted proteins that localize in the extracellular matrix and thusboth of their primary structures contain a signal peptide. In addi-tion, the mature MGP contains in the N-terminal moiety a domainof phosphorylation (SxxSxxS) and a cleavage site identified by theconsensus sequence ANxF, which appear to be conserved in all spe-cies analyzed [10,47,80]. The conserved c-carboxylase recognitionsite is also included in this N-terminal moiety, a characteristicwhich is not found in all other known secreted VKD proteins inwhich that domain is included in the propeptide. In addition,MGP contains another cleavage site between either two conservedarginine residues (RR) or an arginine and a glycine (RG) locatedclose to its carboxy-terminal end and previously suggested to beinvolved in its function [29]. Four (out of five in human) of thec-carboxyglutamate residues and the two conserved cysteines,which form MGP characteristic disulfide bond, are located betweenANxF and RR/RG cleavage sites. To date, and due to the low solubil-ity of the protein in aqueous solutions [71], no tridimensional (3D)structure of MGP is available.
OC is synthesized as a pre-pro-protein. As for MGP, the signalpeptide will target OC to the extracellular matrix, while theprodomain, which is cleaved by a furin-like proteolytic enzyme atRxxR site, only contains the c-glutamyl carboxylase recognition
Fig. 1. Schematic representation of matrix Gla protein (MGP) and osteocalcin (OC)archetype protein structures. Gla residues are indicated by blue dots and Gladomain (Gla) is represented by a blue box; Phosphoserine residues are indicated bygreen dots and phosphorylated domain (P) is represented by a green box; redtriangles indicate proteolytic cleavage sites AXXF and RR (MGP) and RXXR (OC); cindicates the docking site for c-glutamyl carboxylase; Circled C indicate conservedcysteine residues involved in intramolecular disulfide bond (dashed line); SP, signalpeptide.
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
site. The mature protein is small (49 amino acids in human) andcontains the three glutamate residues, which once c-carboxylatedare responsible for its binding to calcium and hydroxyapatite, andthe two conserved cysteine residues forming the intramoleculardisulfide bond, which contributes to stabilize its 3D structure.NMR studies have previously suggested that the configuration ofOC apoprotein form was in a disorganized state in the absence ofcalcium, likely acquiring its 3D structure upon binding of its c-car-boxylated glutamate (Gla) residues to calcium [19,31]. Analysis ofthe crystal structures obtained for OC from pig [38] and fish [22]confirmed that Gla residues are clustered at the surface of the pro-tein, coordinating the Ca2+ ions present in each of those structuresand leaving its carboxy-terminus accessible. This was indicative ofthe existence of a mechanism for attachment to the surface ofthe hydroxyapatite crystals and for promoting the adhesion toosteoblasts and osteoclasts during bone remodeling. Recently, theX-ray crystal structure of bovine 3Glu-osteocalcin (i.e. non-carbox-ylated OC) was described, and surprisingly its structure was foundto be very similar to that of the 3Gla-osteocalcin (i.e. the fully car-boxylated OC), contradicting the previous idea that, in the absenceof Gla, osteocalcin would have a disorganized structure and sug-gesting that the helical structure of the uncarboxylated formadopted a structure similar to that of the carboxylated osteocalcinand thus folded in a calcium-independent way [54]. In light ofthese results, authors suggested that the 3Glu-OC, which doesnot appear to bind calcium, could be involved in interactions withits recently identified receptor [67] in a calcium poor environment,while the 3Gla-OC would bind the crystal structure and throughthese interactions affect bone. The binding of osteocalcin toGPRC6A receptor is however controversial after the recent publica-tion of the study by Jacobsen et al. where agonistic activity ofosteocalcin could not be detected [41].
MGP and OC: established functions
The development of knockout mice [20,53] clearly showed therole of MGP and OC in the control of tissue mineralization,although acting through quite different mechanisms. Accordingly,expression of OC at sites of ectopic mineralization in MGP nullmice cannot reverse the abnormal phenotype, in contrast withre-expression of MGP at those same sites. This indicates thatdespite their evolutionary proximity, OC and MGP functions arenot redundant [58]. Previous works had already suggested theinvolvement of VKD proteins in the mineralization process. Whilecharacterizing the newly identified Gla protein of bone, Price et al.also identified Gla in calcified arteries and suggested that a Glaprotein could be involved in soft tissue calcification [73]. Later,Hale et al. hypothesized that MGP inactivation in cartilage couldbe responsible for both (i) the appearance of foci of mineralizationin the growth plate of infants affected by fetal warfarin syndrome,occurring when mothers received this anticoagulant drug and vita-min K antagonist during the first trimester of their pregnancy, and(ii) the abnormal growth plate mineralization observed in ratstreated with warfarin. The authors hypothesized that these pheno-types could be caused by the functional inhibition of a proteinresponsible for actively preventing ectopic mineralization in carti-lage, which they suggested could be MGP [28]. Indeed, this provedto be true and although its mode of action at the molecular level isstill not completely elucidated, all available evidence points to arole for MGP in the control of tissue mineralization. In 1997 thepublication of the phenotype developed by the mutant MGP�/�
mouse, which included, in addition to cartilage calcifications, deathby artery rupture within 8 weeks after birth due to calcifications ofvessel walls, clearly showed that MGP was indeed a physiologicalinhibitor of calcification. The identification of a human pathology
OsteocalcinSignal peptide Secretion into the extracellular matrix [68]Gla domain Binding of calcium and hydroxyapatite [31,34,73]
Interacts with the inter-calcium spacing in the hydroxyapatite latticeC-terminal region Promotes the adhesion of osteoblasts and osteoclasts in bone replacement [38]
Fig. 2. Occurrence of matrix Gla protein (MGP) and osteocalcin (OC) formsthroughout vertebrate taxonomy. Presence/absence of MGP, OC1, OC2 and OC3were inferred from sequence data collected from NCBI sequence database andrelated to vertebrate species listed in Supplementary Fig. 1. Cardinals indicate thenumber of sequences identified for each protein form in each taxonomic group.Question marks indicate unknown occurrences. V, Vertebrata; G, Gnathostomata; O,Osteichthyes; S, Sarcopterygii; T, Tetrapoda; A, amniota.
M.L. Cancela et al. / Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx 3
associated to MGP loss of function mutations, the Keutel syn-drome, an autosomal recessive disease linked to chromosome12p [9,57], raised the level of interest in elucidating the mode ofaction of this protein, hoping to identify possible therapeutic tar-gets. In the following years and until today, additional works fromdifferent laboratories have been providing further evidence sup-porting this function for MGP and extending its protective functionto other target tissues such as trachea, lungs, kidney, brain, skinand eye [40,55,59,101]. Interestingly, recent works confirm thatskin may be a target for MGP function and abnormal protein levelsin skin have been associated with abnormal phenotypes (e.g. laxand doughy skin) in pseudoxanthoma elasticum [4,36] and in pso-riasis [23].
MGP appears early in development, both in rodents [64], inamphibians such as Xenopus [15], in birds [102] and in fish ([86]and our unpublished data), suggesting a developmentally regu-lated expression which could be associated not only to an inhibi-tion of tissue mineralization but also to cell differentiation andproliferation. This effect could be mediated, at least in part, byinteracting with BMPs, to which MGP has been proven to bind,and consequently inhibiting BMP mediated signaling cascade ofevents [6,97,105].
Unlike MGP, OC was shown to be nearly absent in the mineral-ized fraction of fetal bone [72] and to be specific for vertebrate cal-cified tissues. Originally identified as a marker for bone formationand a strong inhibitor of calcium salt precipitation from supersat-urated solutions ([42,50,92] and references therein), OC was laterconfirmed to be involved in the correct maturation of the hydroxy-apatite crystal in mammalian bone [5] following the developmentof the OC�/�mouse model [20], more than 20 years after its discov-ery. However, another 10 years passed before its putative role as anegative regulator of bone growth could be clarified by the hypoth-esis of mineralization by inhibitor exclusion [76]. Osteocalcin isreleased into the circulation when new bone is formed (e.g. afterbone resorption when bone is remodeled) and, therefore, is consid-ered a marker of bone formation [18,45]. The knowledge of its 3Dstructure, which revealed a negatively charged protein surfacewith five calcium ions in positions complementary to those inhydroxyapatite, provided additional justification and contributedto better understand how OC might modulate morphology andgrowth of hydroxyapatite crystals. Table 1 summarizes thefunctional motifs of MGP and OC and how they contribute to theiridentified functions.
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
MGP and OC: new functions for old proteins and associatedpathologies
In the last decade, a new paradigm has arisen concerning thefunctions of OC and MGP. Indeed, new data has become progres-sively available pointing to the fact that both proteins are likelyto have a more diverse activity, depending on their status of c-car-boxylation, spatial/temporal expression and transcription variants.
Both MGP and OC were originally thought to be functional onlywhen c-carboxylated since this post-translation modification isrequired for the proteins to bind calcium and thus presumablyexert their functions [32,70]. Additionally, many works were cen-tered on the hypothesis that their incomplete status of c-carboxyl-ation indicated some degree of vitamin K deficiency.
4 M.L. Cancela et al. / Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx
For MGP, its undercarboxylated or non-carboxylated status,associated or not to a partial phosphorylation of its N-terminalmoiety, has been extensively associated by many authors to ahigher increase in risk of incidence of cardiovascular events includ-ing vascular calcification, calcification of other soft tissues([16,52,82,91,94,100] among others), and calcification of synovialmembranes in osteoarthritis (OA) patients [87]. In the latter case,this appears to be a local impairment since these authors alsofound that circulating uncarboxylated MGP (ucMGP) levels didnot correlate with levels of ucMGP in synovial fluid from OApatients. Previous in vitro studies have shown that human osteoar-thritic chondrocytes produced mainly ucMGP, while normal chon-drocytes secreted carboxylated MGP (cMGP) [98]. Although the
Fig. 3. Schematic representation of matrix Gla protein (A) and osteocalcin (B) gene ssequences are indicated in black and red. Non-coding exonic sequences are indicated inand alternative transcript structures are indicated on top and bottom of each scheme, respbuild each gene representation.
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
specific mechanism responsible for these findings is not clear,these authors also found a five-fold decrease in c-glutamyl carbox-ylase activity in OA chondrocytes cultured under similar mediumconditions as normal chondrocytes. These results could imply thatinadequate vitamin K concentrations may not be the only or themajor reason behind this difference since in the in vitro cultureboth cells received the same medium and additives.
Furthermore, it remains controversial whether in humans thenormal vitamin K condition would imply that circulating levelsof VKD proteins, in particular MGP and osteocalcin, should bealways fully c-carboxylated in all tissues. In particular, publishedwork is not always consensual about the presence of circulatinguncarboxylated OC being indicative of a vitamin K deficiency, or
tructures and transcript variants throughout vertebrate taxonomy. Coding exonicgrey and pink. Circled roman numbers indicate phase of intron insertion. Archetypeectively. Figures indicated in the last exon indicate the number of sequences used to
M.L. Cancela et al. / Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx 5
if it can be or should be prevented by vitamin K supplementation([3,26] and references therein). This raises the question of whetherthe levels of undercarboxylated VKDs can have a physiological rel-evance. Accordingly, recent studies have shown that osteocalcinappears to have two functional structures, carboxylated for itsinteraction with hydroxyapatite crystals and modulation of crystalgrowth, and the undercarboxylated form, which would have ahormone-like function in energy metabolism, fertility and braindevelopment [37,43,44,66,67,99]. Since most of this data wasobtained in mice, it is still controversial whether they accuratelyreproduce what happens in humans. However, Oury et al. haverecently demonstrated that disrupting osteocalcin signaling leadsto glucose intolerance in both humans and mice [65] providing evi-dence that there is some similarity between mouse and human OC-mediated pathways.
Interestingly, MGP has also been described as a critical regula-tor of endothelial cell function, since its expression is significantlyincreased during the angiogenesis process [1]. MGP has beenshown to regulate both physiological and tumor-related angiogen-esis [7,46] possibly through Notch and BMP signaling pathways[86]. Expression of abnormal levels of MGP have been observedin different tumors such as colon carcinoma [21], glioblastoma[46,93], breast [14,104], ovarian [39], urogenital [49], skin [56]and gastric [27] cancers which suggests, once again, the involve-ment of this protein in cell differentiation, as previously discussed.However, results are often contradictory; in some tumors MGP lev-els were found to be increased while in others they were decreasedwhen compared to normal tissue. These results suggest that addi-tional players must be involved and more work is required in orderto understand the involvement of MGP in tumorigenesis.
Fig. 4. Comparative analysis of MGP and OC protein structures, including ancestralOC form from Adriatic sturgeon (AnaOC). Gla residues are indicated by blue dotsand Gla domain (Gla) is represented by a blue box; Phosphoserine residues areindicated by green dots and phosphorylated domain (P) is represented by a greenbox; New domains in MGP variants 1 and 2 are represented by a red box; redtriangles indicate proteolytic cleavage sites; c indicates the docking site for c-glutamyl carboxylase; circled C indicate conserved cysteine residues involved inintramolecular disulfide bond (dashed line); SP, signal peptide; Question marksindicate domain either putative or with an unknown function. Archetype structuresare highlighted in a gray box.
MGP and OC genes: duplication and neofunctionalization
A better understanding of the molecular evolution of matrix andbone Gla proteins but also of their evolutionary relationship is notonly necessary to the better evaluation of the various hypothesesabout their role and function in tissue mineralization but shouldalso provide essential insights on how calcified tissues haveevolved. Genomic evidence collected within the scope of the sur-vey of public sequence database presented here (Fig. 2) indicatesthat MGP occurs in all jawed vertebrates (Gnathostomata) andOC in all bony vertebrates (Osteichthyes). While few sequence datawere available in 2005 from jawless fish (Agnatha) and cartilagi-nous fish (Chondrichthyes) at the time of the first survey [47], itis now possible to conclude with reasonable confidence, derivedfrom the large quantity of genomic data available, that MGP andOC are absent from the genome of agnathans and that OC is alsolacking in non-bony vertebrate genomes. Thus the hypothesis pre-viously presented [47] that MGP appearance may be concomitantwith the origin of the Chondrichthyes and cartilage-like skeletaltissue and that OC appearance may be concomitant with the originof the Osteichthyes and bony skeletal tissue remains valid and isnow reinforced by the additional available data.
The hypothesis that OC gene originated from MGP gene througha duplication event that would have occurred in the Osteichthyeslineage is supported by gene and protein structure data ([47] andFigs. 3 and 4). The presence of an OC gene in the vicinity of theMGP gene (intergenic region is approximately 4000 bp) in the gen-ome of ray-finned fish is an additional evidence that would favor atandem duplication event (Fig. 5A). Because MGP is present inChondrichthyes and not OC, MGP would be the parent copy andOC the duplicated copy. Ancestral function was retained by MGP,while OC would have diverged following the duplication eventthrough the accumulation of mutations and developed a new func-tion advantageous for Osteichthyes, probably related to the
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
appearance of bone tissue. Although this adaptive mutation pro-cess, known as neofunctionalization [106], is thought to occurrarely in evolution, we believe that two more events took placein vertebrate lineage to give rise to OC2 in teleost fish and OC3in tetrapods. As proposed previously by Laizé et al. [48] and morerecently by Cavaco et al. [13], teleost-specific OC2 gene, which issituated on a different chromosome than MGP and OC1 (Fig. 5A),most likely originated from OC1 gene through the well-establishedwhole genome duplication event that occurred in the Teleosteiafter branching from the Holostei about 230 million years agoand which has probably driven their appearance and diversity[83]. OC2 gene coexists with OC1 gene in the genome of almostall teleost fish surveyed here (Fig. 5A), suggesting that it was main-tained for 230 million years of evolution, certainly because itbrings some advantage/new function to teleost fishes. Although lit-tle information is available about the molecular function or thephysiological role of OC2, a recent report demonstrated its pres-ence in the mineral phase of teleost bone, suggesting that it shareswith OC1 the ability to bind the calcium mineral phase [13]. Inaddition, the time of developmental appearance is also different,with OC1 appearing within the first hours after fertilization in zeb-rafish, much before the development of a mineralized structure,while OC2 appearance is clearly associated with the mineral phaseof the forming bone [2]. The most relevant difference between thetwo teleostean isoforms resides in the prodomain: larger, moreacidic and polar, and potentially phosphorylated in OC2 ([48] andFig. 4). Although this hypothesis remains to be demonstrated,phosphate groups and acid residues could provide proOC2 (if
Fig. 5. (A) Schematic representation of the localization of matrix Gla protein andosteocalcin genes in vertebrate genomes. Ac, Actinopterygii. Length (bp) ofintergenic region between MGP and OC1/OC3 is indicated above red triangles.Solid line indicates chromosome or genomic scaffolds. (B) Molecular phylogeny ofosteocalcin and matrix Gla protein from Osteichthyes. Phylogenetic and molecularevolutionary analyses were conducted using MEGA version 6 [89] and multiplesequence alignment constructed using T-Coffee [62]. Alignment is presented inSupplementary Fig. 2 and original Neighbor-Joining tree is presented in Supple-mentary Fig. 3. (C) Evolutionary model describing the duplicative origin of MGP andOC genes and their evolutionary relationship.
6 M.L. Cancela et al. / Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx
secreted or working intracellularly) the ability to bind highamounts of calcium or calcium-containing crystals and play anactive role in calcium homeostasis and/or biomineralization.Because they live in a calcium-rich environment, fish have highserum levels of calcium and thus have the need for means to con-trol calcium homeostasis.
Interestingly, we recently isolated from sturgeon an ancestralform of OC [95] with structural characteristics which appear to
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
be hybrid between an OC and a MGP and thus much closer toMGP than all the other OC sequences identified so far (Fig. 4).And like OC2, the sturgeon OC contains a putative phosphorylatedpropeptide, which is this case is very similar to the phosphorylateddomain of MGP. Furthermore its pattern of tissue distribution isalso hybrid, being expressed in both mineralized and soft tissues,thus suggesting a close relationship between structure and func-tion. From an evolutionary point of view, sturgeon OC appears tohave retained structural features of the ancestral protein thatresulted, millions of years ago, from the duplication of the ancientMGP gene.
A gene duplication event related to a tandem duplication like inthe case of OC1 is probably at the origin of the appearance of OC3gene in the tetrapod lineage. The presence of the OC3 gene in thevicinity of the MGP gene (intergenic region is ranging from approx-imately 1700 to 6700 bp) in the genome of amphibian and saur-opsids (Fig. 5A), together with the similarity of OC3 with MGP atprotein level (Figs. 4 and 5B) are evidence towards similar mecha-nisms of duplication for OC1 and OC3 although separated by mil-lion years of evolution. Because OC3 would have resulted from alater MGP duplication event, this would justify why its molecularstructure is more similar to MGP than that of OC1. Our data alsoindicate that mammalian OC is orthologous to amphibian and sau-ropsidian OC1 and not OC3, a hypothesis sustained by syntenyevaluation of the genomic contexts of both genes, which clearlyshow a better conservation of the genomic environment betweenmammalian OC and amphibian/sauropsidian OC1 (results notshown). Absence of sequence data related to osteocalcin in coela-canth genome prevented us to localize this duplication event moreprecisely. Similarly, molecular function of OC3 and its role inamphibian and sauropsidian physiology is yet unknown. Theabsence of OC3 gene in mammalian genome suggest that OC3function/role did not present an advantage for mammals and thusthe gene was deleted within the course of mammalian evolution.Based on the data collected, we have enlarged and complementedour previous evolutionary model describing the duplicative originof MGP and OC genes and their evolutionary relationship [47],including now all the novel sequence information available in dat-abases. This refined model confirms the previous predictions and ispresented in Fig. 5C.
Interestingly, while a single MGP gene occurs in the genome ofall the gnathostomes surveyed in this study (Fig. 3A; occurrencesof two genes were found only in species with a known polyploidgenome i.e. Atlantic salmon, rainbow trout and the African clawedfrog), three distinct osteocalcin genes were identified (Fig. 3B).Similarly, while several splicing variants of MGP were identifiedin vertebrate species, a single transcript for each OC gene appearsto have been maintained throughout evolution (Fig. 3), suggestingdifferent evolutionary strategies for MGP and OC. In both casesnew functions and/or new regulatory pathways have been devel-oped by host organisms.
Conclusion
In this review we provide a summary of the established andnovel functions attributed to MGP and OC and have highlightedthe importance of the various strategies used throughout evolutionto accomplish this design. The two proteins have similar structures(Fig. 1) and although they have arisen from a common ancestor, afunctional MGP appeared concomitantly with cartilage while OCappeared with bone during evolution (Fig. 2). Functional diver-gence was achieved through a combination/modulation of posttranslation modifications of the proteins (carboxylation, phosphor-ylation, proteolytic cleavage) associated with either a strategy ofgene duplication for OC or the use of alternative gene promotersand splicing for MGP, which have produced similar proteins but
M.L. Cancela et al. / Archives of Biochemistry and Biophysics xxx (2014) xxx–xxx 7
exerting either different functions or presenting different spatial/temporal expression and regulation. Interestingly, the core struc-ture of each of these proteins never changed completely, as dem-onstrated in Figs. 3 and 4. Our hypothesis of the duplicativeorigin of MGP and OC genes and their evolutionary relationshipin presented in Fig. 5C.
Acknowledgments
This work was co-funded by the European Regional Develop-ment Fund (ERDF) through COMPETE Program and by NationalFund through the Portuguese Science and Technology Foundation(FCT) under PEst-C/MAR/LA0015/2011 project. NC was supportedby a post-doctoral grant (SFRH/BPD/48206/2008) from the FCT.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.abb.2014.07.020.
[2] A. Bensimon-Brito, J. Cardeira, M.L. Cancela, A. Huysseune, P.E. Witten, BMCDev. Biol. 12 (2012) 28.
[3] S.L. Booth, A. Centi, S.R. Smith, C. Gundberg, Nat. Rev. Endocrinol. 9 (2013) 43–55.
[4] F. Boraldi, M. Garcia-Fernandez, C. Paolinelli-Devincenzi, G. Annovi, L.Schurgers, C. Vermeer, P. Cianciulli, I. Ronchetti, D. Quaglino, Biochim.Biophys. Acta 1832 (2013) 2077–2084.
[5] A.L. Boskey, S. Gadaleta, C. Gundberg, S.B. Doty, P. Ducy, G. Karsenty, Bone 23(1998) 187–196.
[6] K. Boström, D. Tsao, S. Shen, Y. Wang, L.L. Demer, J. Biol. Chem. 276 (2001)14044–14052.
[7] K. Boström, A.F. Zebboudj, Y. Yao, T.S. Lin, A. Torres, J. Biol. Chem. 279 (2004)52904–52913.
[8] A.L.J.J. Bronckers, M.C. Farach-Carson, E. van Waveren, W.T. Butler, J. BoneMiner. Res. 9 (1994) 833–841.
[9] L. Cancela, C.-L. Hsieh, U. Francke, P.A. Price, J. Biol. Chem. 265 (1990) 15040–15048.
419–423.[13] S. Cavaco, M.K. Williamson, J. Rosa, V. Roberto, O. Cordeiro, P.A. Price, M.L.
Cancela, V. Laizé, D.C. Simes, Fish Physiol. Biochem. 40 (2014) 731–738.[14] L. Chen, J.P. O’Bryan, H.S. Smith, E. Liu, Oncogene 5 (1990) 1391–1395.[15] N. Conceição, A.C. Silva, J. Fidalgo, J.A. Belo, M.L. Cancela, FEBS J. 272 (2005)
1501–1510.[16] G.W. Dalmeijer, Y.T. van der Schouw, C. Vermeer, E.J. Magdeleyns, L.J.
Schurgers, J.W.J. Beulens, J. Nutr. Biochem. 24 (2013) 624–628.[17] J. Danziger, Clin. J. Am. Soc. Nephrol. 3 (2008) 1504–1510.[18] P.D. Delmas, R. Eastell, P. Garnero, M.J. Seibel, J. Stepan, Osteoporos. Int. 11
(2000) 2–17.[19] T. Dowd, J. Rosen, L. Li, C. Gundberg, Biochemistry 42 (2003) 7769–7779.[20] P. Ducy, C. Desbois, B. Boyce, G. Pinero, B. Story, C. Dunstan, E. Smith, J.
Bonadio, S. Goldstein, C. Gundberg, A. Bradley, G. Karsenty, Nature 382 (1996)448–452.
[22] C. Frazão, D.C. Simes, R. Coelho, D. Alves, M.K. Williamson, P.A. Price, M.L.Cancela, M.A. Carrondo, Biochemistry 44 (2005) 1234–1242.
[23] S. Gerdes, S. Osadtschy, N. Buhles, H. Baurecht, U. Mrowietz, Exp. Dermatol.23 (2014) 322–325.
[24] M. Goiko, J. Dierolf, J.S. Gleberzon, Y. Liao, B. Grohe, H.A. Goldberg, J.R. deBruyn, G.K. Hunter, PLoS ONE 8 (2013) e80344.
[25] P. Gonzalez, D.L. Epstein, T. Borrás, Invest. Ophthalmol. Vis. Sci. 41 (2000)3678–3693.
[26] C.M. Gundberg, J.B. Lian, S.L. Booth, Adv. Nutr. 3 (2012) 149–157.[27] L. Guo, X. Guo, J.L. Jiang, J.N. Zhang, J. Ji, B.Y. Liu, Z.G. Zhu, Y.Y. Yu, Zhonghua
Bing Li Xue Za Zhi 39 (2010) 436–441.[28] J.E. Hale, J.D. Fraser, P.A. Price, J. Biol. Chem. 263 (1988) 5820–5824.[29] J.E. Hale, M.K. Williamson, P.A. Price, J. Biol. Chem. 266 (1991) 21145–21149.[30] F. Hashimoto, Y. Kobayashi, E.T. Kobayashi, E. Sakai, K. Kobayashi, M. Shibata,
Y. Kato, H. Sakai, Arch. Oral Biol. 46 (2001) 585–592.[31] P.V. Hauschka, S.A. Carr, Biochemistry 21 (1982) 2538–2547.
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
(2010) 3793–3802.[47] V. Laizé, P. Martel, C.S.B. Viegas, P.A. Price, M.L. Cancela, J. Biol. Chem. 280
(2005) 26659–26668.[48] V. Laizé, C.S.B. Viegas, P.A. Price, M.L. Cancela, J. Biol. Chem. 281 (2006)
15037–15043.[49] E.N. Levedakou, T.G. Strohmeyer, P.J. Effert, E.T. Liu, Int. J. Cancer 52 (1992)
534–537.[50] J.B. Lian, C.M. Gundberg, Clin. Orthop. Relat. Res. 226 (1988) 267–291.[51] K.P. Link, Circulation 19 (1959) 97–107.[52] X. Lu, B. Gao, T. Yasui, Y. Li, T. Liu, X. Mao, M. Hirose, Y. Wu, D. Yu, Q. Zhu, K.
Kohri, C. Xiao, Kidney Blood Press. Res. 37 (2013) 15–23.[53] G. Luo, P. Ducy, M.D. McKee, G.J. Pinero, E. Loyer, R.R. Behringer, G. Karsenty,
Nature 386 (1997) 78–81.[54] V.N. Malashkevich, S.C. Almo, T.L. Dowd, Biochemistry 52 (2013) 8387–8392.[55] M. Meier, L.P. Weng, E. Alexandrakis, J. Ruschoff, G. Goeckenjan, Eur. Respir. J.
17 (2001) 566–569.[56] P. Micke, K. Kappert, M. Ohshima, C. Sundquist, S. Scheidl, P. Lindahl, C.-H.
Heldin, J. Botling, F. Ponten, A. Ostman, J. Invest. Dermatol. 127 (2007) 1516–1523.
[57] P.B. Munroe, R.O. Olgunturk, J.P. Fryns, L. Van Maldergem, F. Ziereisen, B.Yuksel, R.M. Gardiner, E. Chung, Nat. Genet. 21 (1999) 142–144.
[58] M. Murshed, T. Schinke, M.D. McKee, G. Karsenty, J. Cell Biol. 165 (2004) 625–630.
[59] A. Nanda, J.T. Anim, M. Al-gareeb, Q.A. Alsaleh, Am. J. Med. Genet. A 140(2006) 1487–1489.
[60] S.K. Nishimoto, M. Nishimoto, Matrix Biol. 24 (2005) 353–361.[61] S.K. Nishimoto, P.A. Price, J. Biol. Chem. 255 (1980) 6579–6583.[62] C. Notredame, D.G. Higgins, J. Heringa, J. Mol. Biol. 302 (2000) 205–217.[63] J. O’Young, Y. Liao, Y. Xiao, J. Jalkanen, G. Lajoie, M. Karttunen, H.A. Goldberg,
G.K. Hunter, J. Am. Chem. Soc. 133 (2011) 18406–18412.[64] Y. Otawara, P.A. Price, J. Biol. Chem. 261 (1986) 10828–10832.[65] F. Oury, M. Ferron, W. Huizhen, C. Confavreux, L. Xu, J. Lacombe, P. Srinivas, A.
Chamouni, F. Lugani, H. Lejeune, T.R. Kumar, I. Plotton, G. Karsenty, J. Clin.Invest. 123 (2013) 2421–2433.
[66] F. Oury, L. Khrimian, C.A. Denny, A. Gardin, A. Chamouni, N. Goeden, Y. Huang,H. Lee, P. Srinivas, X.-B. Gao, S. Suyama, T. Langer, J.J. Mann, T.L. Horvath, A.Bonnin, G. Karsenty, Cell 155 (2013) 228–241.
[67] F. Oury, G. Sumara, O. Sumara, M. Ferron, H. Chang, C.E. Smith, L. Hermo, S.Suarez, B.L. Roth, P. Ducy, G. Karsenty, Cell 144 (2011) 796–809.
125–128.[88] J. Stenflo, P. Fernlund, W. Egan, P. Roepstorff, Proc. Natl. Acad. Sci. U.S.A. 71
(1974) 2730–2733.[89] K. Tamura, G. Stecher, D. Peterson, A. Filipski, S. Kumar, Mol. Biol. Evol. 30
(2013) 2725–2729.[90] S.I. Tomarev, G. Wistow, V. Raymond, S. Dubois, I. Malyukova, Invest.
Ophthalmol. Vis. Sci. 44 (2003) 2588–2596.[91] T. Ueland, C.P. Dahl, L. Gullestad, S. Aakhus, K. Broch, R. Skårdal, C. Vermeer, P.
Aukrust, L.J. Schurgers, Clin. Sci. 121 (2011) 119–127.[92] P.G.F. Van de Loo, B.A.M. Soute, L.J.M. van Haarlem, C. Vermeer, Biochem.
Biophys. Res. Commun. 142 (1987) 113–119.[93] J. Van den Boom, M. Wolter, R. Kuick, D.E. Misek, A.S. Youkilis, D.S. Wechsler,
C. Sommer, G. Reifenberger, S.M. Hanash, Am. J. Pathol. 163 (2003) 1033–1043.
Please cite this article in press as: M.L. Cancela et al., Arch. Biochem. Biophys.
[94] E.G.H.M. Van den Heuvel, N.M. van Schoor, P. Lips, E.J.P. Magdeleyns, D.J.H.Deeg, C. Vermeer, M. den Heijer, Maturitas 77 (2014) 137–141.
[95] C.S.B. Viegas, D.C. Simes, M.K. Williamson, S. Cavaco, V. Laizé, P.A. Price, M.L.Cancela, J. Biol. Chem. 288 (2013) 27801–27811.
[96] N. Wajih, T. Borras, W. Xue, S.M. Hutson, R. Wallin, J. Biol. Chem. 279 (2004)43052–43060.
[97] R. Wallin, D. Cain, S.M. Hutson, D.C. Sane, R. Loeser, Thromb. Haemost. 84(2000) 1039–1044.
[98] R. Wallin, L.J. Schurgers, R.F. Loeser, Osteoarthr. Cartil. 18 (2010) 1096–1103.[99] J. Wei, M. Ferron, C.J. Clarke, Y.A. Hannun, H. Jiang, W.S. Blaner, G. Karsenty, J.
Food Res. 1 (2014) 1–16.[101] W. Xue, R. Wallin, E.A. Olmsted-Davis, T. Borrás, Invest. Ophthalmol. Vis. Sci.
47 (2006) 997–1007.[102] K. Yagami, J.-Y. Suh, M. Enomoto-Iwamoto, E. Koyama, W.R. Abrams, I.M.
Shapiro, M. Pacifici, M. Iwamoto, J. Cell Biol. 147 (1999) 1097–1108.[103] Y. Yao, A. Shahbazian, K.I. Boström, Circ. Res. 102 (2008) 1065–1074.[104] K. Yoshimura, K. Takeuchi, K. Nagasaki, S. Ogishima, H. Tanaka, T. Iwase, F.
Akiyama, Y. Kuroda, Y. Miki, Mol. Med. Rep. 2 (2009) 549–553.[105] A.F. Zebboudj, M. Imura, K. Boström, J. Biol. Chem. 277 (2002) 4388–4394.[106] J. Zhang, Trends Ecol. Evol. 18 (2003) 292–298.
