Biophysical characterization ofrecombinant human ameloblastin

Tom�Ð Wald1,2, Lucie Bedn�rov�3,Radim Osicka1, Petr Pachl3, Miroslav�ulc1, Stale Petter Lyngstadaas4,Ivan Slaby5, Jir� Vondr�Ðek3,5

1Institute of Microbiology v.v.i, Academy ofSciences of the Czech Republic, V�densk�,1083, 142 20 Prague 4 Czech Republic;2Institute of Chemical Technology, Prague 6,Czech Republic; 3Institute of OrganicChemistry and Biochemistry v.v.i., Academy ofSciences of the Czech Republic, Flemingovonam. 2, 166 10 Prague 6, Czech Republic;4Department of Biomaterials, University ofOslo, Oslo, Norway; 5Institute of Biotechnologyv.v.i., Academy of Sciences of the CzechRepublic, V�densk� 1083, 142 00 Prague 4,Czech Republic

Ameloblastin (AMBN, also known as amelin orsheathlin) is an interesting protein to study in terms of itsstructural, functional, and medical properties. The pro-tein is well conserved among species and has been foundto be constantly present throughout the evolution andontogenesis of mineralized tissues (1, 2). So far, no otherprotein family with structural or sequence similarity toAMBN has been characterized. The Ambn mRNA wasfirst cloned and sequenced from developing rat teeth in1995 (EMBL database no.: Z50083). As derived from thecDNA sequences, the AMBN molecule in most speciesexists as two variants generated by alternative splicing,creating a deletion of 15 amino-acid residues at theN-terminus (3, 4). The rodent AMBN contains a DGEAintegrin-binding sequence as well as a VTXG thrombo-spondin cell-adhesion sequence, suggesting that AMBNcould be involved in cell–matrix interactions. However,neither of these sequences is conserved in porcine andhuman molecules. The human homologue (434 amino-acid residues) has a unique 26 amino-acid residue insertin the middle part of the molecule that appears to be aduplication of short exon 7 (5). Extensive studies ofporcine AMBN have identified 13, 15, 27, and 29 kDacalcium-binding peptides, all derived from the C-termi-nal part of the molecule (6, 7). Furthermore, severalpost-translational modifications have been identified,including sulphated O-linked glycosylations (Ser112 andThr387), hydroxylated prolines, and phosphorylations(8). Interestingly, two phosphorylated residues (Ser43and Thr277) and three kinase-binding sites are absolutely

conserved in AMBN isolated, to date, from all knownspecies. The sequence of AMBN contains stretches ofproline-rich motifs (PMMs) mostly at the N-terminus,which could be important for signalling activity.In a previous study, we performed a bioinformatic

analysis of AMBN and a Robetta ab initio prediction ofthe three-dimensional structure of the molecule (9). Theresults suggested that AMBN is a two-domain, intrinsi-cally disordered protein (IDP) (10). All analyses andmodelling data were in agreement with earlier experi-mental observations that tooth-specific proteases(enamelysin and kallikrein) separate the basic N-termi-nus from the acidic C-terminal region of the AMBNmolecule (11). These separate domains were suggested tohave different localizations and functions during odon-togenesis (12). As reviewed by Tompa (13) and Uversky

& Dunker (14), the IDP structures are mostly dependenton the environment of the solution and on the nature ofthe interactions with other proteins that could determinethe functional structure. Some parts of the interactingproteins may also remain disordered. Therefore, IDPscan have several transient, function-related structures. Anumber of theoretical, as well as experimental, tech-niques are able to identify and describe IDPs, includingcircular dichroism (CD) spectroscopy, nuclear magneticresonance (NMR), dynamic light scattering (DLS), andsmall-angle X-ray scattering (SAXS).Based on the localization of AMBN expression and

the results of knockout experiments, AMBN is suggestedto have both a structural function (1, 3, 4, 15) and a role

Wald T, Bednarova L, Osicka R, Pachl P, Sulc M, Lyngstadaas SP, Slaby I, VondrasekJ. Biophysical characterization of recombinant human ameloblastin.Eur J Oral Sci 2011; 119 (Suppl. 1): 261–269. � 2011 Eur J Oral Sci

Ameloblastin (AMBN) is a protein expressed mainly during dental hard tissuedevelopment. Biochemically, it is classified as an intrinsically disordered protein (IDP).Its biological role remains largely unknown; however, the question of AMBN functionwill undoubtedly be connected to its structural properties and its potential for protein–protein and protein–cell interactions. A basic biophysical characterization of humanrecombinant ameloblastin (hrAMBN) and its N- and C-terminal domains by means ofcircular dichroism spectroscopy and dynamic light scattering showed that underphysiological conditions ameloblastin is an IDP with a prevalent polyproline-II (PPII)conformation. Both the N- and C-terminal polypeptides, when expressed indepen-dently, showed different structural preferences upon heating as well as differentbehaviour in the presence of trifluoroethanol and CaCl2 salt. The N-terminal peptideshowed a more ordered structure with a strong tendency to adopt a helical confor-mation upon the addition of trifluorethanol, whereas the C-terminal domain seemed tobe primarily responsible for the structural disorder of the entire AMBN molecule.

Jiri Vondrasek, Institute of Biotechnology v.v.i.,Academy of Sciences of the Czech Republic,V�densk� 1083, 142 20 Prague 4, CzechRepublic

Telefax: +420–244–472282E-mail: jiri.vondrasek@img.cas.cz

Key words: circular dichroism spectroscopy;dynamic light scattering; enamel matrix protein;intrinsically disordered proteins; protein struc-ture

Accepted for publication October 2011

Eur J Oral Sci 2011; 119 (Suppl. 1): 261–269DOI: 10.1111/j.1600-0722.2011.00913.xPrinted in Singapore. All rights reserved

� 2011 Eur J Oral Sci

European Journal ofOral Sciences

as a signalling molecule (16). Ameloblastin was originallydescribed as a tooth-enamel-specific matrix proteinsecreted only by ameloblast cells (3, 4). In enamel,AMBN is the most abundant non-amelogenin proteinand essential for normal enamel formation. Amelobla-stin-null mice show severe enamel hypoplasia anduncontrolled differentiation of the adjacent epithelialcells (17). In later studies, AMBN expression was alsodetected during the development of mesenchymal dentalhard tissues (18, 19), during trauma-induced reparativedentin formation (20), and in the early stages of bonedevelopment (21, 22), suggesting that it is involved inepithelial–mesenchymal interactions. Rat recombinantfusion AMBN has also been used to induce reparativedentin formation in pulpotomized teeth (23) and topromote bone formation in bone defects (24).To understand the proper function of AMBN as an

IDP, a reasonable knowledge of the structural features ofthe molecule must be obtained. So far, the main obstaclein studying the structure–function relationships ofAMBN has been an insufficient amount of purifiedprotein. In the study presented here, recombinantAMBN protein and its N- and C-terminal domains wereprepared and purified for CD spectroscopy and DLSexperiments.

Material and methods

Sequence analysis

For basic physical and chemical analysis of the human re-combinant AMBN (hrAMBN) and its N-terminal(hrAMBN-Nter) and C-terminal (hrAMBN-Cter) domains,we first used the analytical tool ProtParam, freely availableat (http://web.expasy.org/cgi-bin/protparam/protparam)(25). We also repeated the secondary structure predictionpublished in our previous paper on the bioinformaticsanalysis of AMBN (9) and its corresponding N- andC-terminal domains. The particular protein sequences andtheir lengths are described in the following paragraph. Weused Jpred (http://www.compbio.dundee.ac.uk/www-jpred/index.html) – the Protein Secondary Structure Predictionserver which incorporates the Jnet (26) algorithm. Jpred isalso able to predict solvent accessibility and coiled-coil re-gions in proteins. The current version (v3) (27) of Jpredbuilds on the previous versions of Jpred, developed andmaintained by James Cuff and Jonathan Barber. Thepractical aspects of this analysis were to identify differencesin both termini to form secondary structure elements.

Cloning, expression, and purification of hrAMBN

To construct plasmids for the expression of hrAMBN(amino-acid residues 27-447), hrAMBN-Nter (residues 27–222), and hrAMBN-Cter (residues 223–447) in prokaryoticcells, the sequences for AMBN and its subdomains wereamplified from the cDNA (Image clone 40033703, SourceBioscience UK Limited, Nottingham, UK) using PCR. ThePCR products were cloned into a modified pET28b vector(Novagen, Darmstadt, Germany) under the control of thetranscription and translation initiation signals of gene 10from bacteriophage T7. The resulting plasmids were usedfor the production of hrAMBN, hrAMBN-Nter, andhrAMBN-Cter in Eshcerichia coli BL21(kDE3) cells upon

induction with isopropyl thio-b-d-galactoside (IPTG). Thepoly-His tagged proteins were purified by affinity chroma-tography from crude cell extracts on a nickel-nitrilotriaceticacid (Ni-NTA) agarose column (Qiagen, Hilden, Germany).The homogeneity of the final protein preparations was tes-ted by SDS-PAGE and western blot, and the identity of theproteins was confirmed by mass spectrometry (data notshown).

CD spectroscopy

The CD measurements were performed on a Jasco-815spectropolarimeter equipped with the Peltier type temper-ature control system PTC-423S/L (JASCO Inc. Easton,MD, USA). The spectra were measured from 195 to 300 nmwith standard instrument sensitivity and with a scanningspeed of 20 nm min)1, a response time of 8 s, and twospectra accumulations for hrAMBN-Cter, and with highinstrument sensitivity and with a scanning speed of5 nm min)1, a response time of 16 s, and two spectraaccumulations for full-length hrAMBN and hrAMBN-Nter;all spectra were measured in a 1-mm quartz cell. For allmeasurements, the temperature of the sample was keptconstant at physiological temperature (37�C), except for theexperiment studying the thermal dependency of the CDspectra. In this experimental set up, the start and finishtemperatures were 4�C and 90�C, respectively, with anincrement of 20�C. The reversibility was checked by mea-surement of the chilled solution back to 4�C. The proteinwas dissolved in a 5 mM Tris/40 mM NaCl buffer and forthe entire experimental set the concentration of protein waskept constant (0.05 mg ml)1). The spectral variations ofhrAMBN, hrAMBN-Cter, and hrAMBN-Nter were alsomeasured in the presence of trifluorethanol [TFE; at finalconcontrations of 10%, 20%, 30%, 40%, 50%, and 80%(volume by volume)] NaCl (at final concentrations of0.5 M, 1 M, and 3.6 M), and calcium chloride (Cacl2; atfinal concentrations of 20 mM, 100 mM and 1,000 mM).Numerical analysis of the secondary structure and second-ary structure assignment were performed using an onlineCD analysis program (28, 29) Dichroweb (http://dichro-web.cryst.bbk.ac.uk).

DLS measurement

The DLS measurements were performed using the RiNALaser Spectroscater 201 (RiNA, Berlin, Germany). Theprotein samples were kept at 20�C and the scattered lightsignal was collected at 90� to the direct beam. For eachsample, 60 measurements were taken, where the time dif-fered depending on the quality of each sample. For the DLSmeasurement the same samples as for CD spectroscopy wereused without any additional dilution. The solutions werecentrifuged before each experiment to minimize the addi-tional parasite scattering.

Cross-linking reactions

The cross-linking reaction using different concentrations (0,20, 40, 100, 200, or 400 lg ml)1) of 1-ethyl-3-[3-dimethyl(aminopropyl)] carbodiimide hydrochloride (EDC) wascarried out at a protein concentration of 2 lM in a systemcontaining 50 mM pyridine:HCl buffer (pH 6.0) and 0 or 20mM CaCl2, at room temperature for 2 h. Protein–proteininteractions were tested in several systems: (i) homodimer-ization of the N-terminal domain of hrAMBN, (ii)

262 Wald et al.

homodimerization of the C-terminal domain of hrAMBN,(iii) heterodimerization of the N- and C-terminal domains ofhrAMBN, and (iv) homodimerization of 14-3-3f (positivecontrol). The cross-linking reaction was quenched by theaddition of 2-mercaptoethanol in electrophoretic samplebuffer, after which the reaction mix was separated on a 10%SDS–polyacrylamide gel and the protein bands were visu-alized by silver staining of the gel.

Chemicals

Tris, NaCl, SDS, CaCl2, and TFE (of NMR purity) werepurchased from Sigma-Aldrich (Sigma-Aldrich, St Louis,MO, USA). Pyridine and EDC were from Fluka (St Louis,MO, USA).

Results and Discussion

Secondary structure prediction and analysis of thefull-length hrAMBN and its N- and C-terminal domains

The results of the ProtParam (http://web.expasy.org/cgi-bin/protparam/protparam) analysis of hrAMBN,hrAMBN-Nter, and hrAMBN-Cter are summarized inTable 1. It is apparent that the physical-chemical beha-viour of the full-length hrAMBN is influenced to a majorextent by the acidic and the least structurally stableC-terminal domain. The less charged and more compactN-terminal domain seems to influence the physical-chemical behaviour of the full-length hrAMBN onlymarginally. The high instability and low aliphatic indexesof the C-terminal domain also suggests its predominantrole in influencing the IDP structural characteristics ofthe full-length molecule.

Only one profiled a-helix was predicted in the N-ter-minal part of the full-length AMBN (Fig. 1) About sevenb-sheet stretches of different lengths were detected alongthe sequence, mainly located at the C-terminal region.The secondary structure prediction for the separate N-terminal domain showed, in contrast to the full-lengthmolecule, three a-helices in this region. It is difficult todecide if the prediction reveals the different propensity ofthe same sequence in a different context or if it can beaccounted for by the inaccuracy of the predictionmethod itself. Strangely enough, the NSLW sequence atthe N-terminal region is assigned as a b sheet in thefull-length AMBN but is predicted to be helical in theN-terminus of AMBN. All predictions show a zeroprobability of coiled-coil regions (motifs in which ahelices are coiled together) as well as less than 5%exposure for the involved amino acid residues. The dif-ference in secondary structure prediction between thefull-length AMBN and the isolated N-terminal domainsuggests that the context of the full-length moleculecould influence the secondary structure content of itsN-terminus. Thus, it seems likely that the isolated

Table 1Physical-chemical characteristics of human recombinant

ameloblastin (hrAMBN), the N-terminal domain of humanrecombinant ameloblastin (hrAMBN-Nter), and the C-terminaldomain of human recombinant ameloblastin (hrAMBN-Cter)

Molecule pIEstimated

charge at pH 7Instability

indexAliphaticindex

hrAMBN 4.72 )19.4 64.54 60.74hrAMBN-Nter 6.06 )0.6 59.90 69.60hrAMBN-Cter 4.32 )18.8 68.54 53.07

>sp|Q9NP70|AMBN_HUMAN Ameloblastin OS=Homo sapiens GN=AMBNPFFPQQSGTPGMASLSLETMRQLGSLQRLNTLSQYSRYGFGKSFNSLWMHGLLPPHSSLPWMRPREHETQQYEYSLP-------------HHHHHHHHHHHH-------------------EEEE-----------------------------VHPPPLPSQPSLKPQQPGLKPFLQSAAATTNQATALKEALQPPIHLGHLPLQEGELPLVQQQVAPSDKPPKPELPGV-----------------------------------------------------------------------------DFADPQGPSLPGMDFPDPQGPSLPGLDFADPQGSTIFQIARLISHGPMPQNKQSPLYPGMLYVPFGANQLNAPAR ---------------------------------------EEEE----------------EEEEE-----------LGIMSSEEVAGGREDPMAYGAMFPGFGGMRPGFEGMPHNPAMGGDFTLEFDSPVAATKGPENEEGGAQGSPMPEANPEEEEE--------------HH------------------------EEEEE---------------------------DNLENPAFLTELEPAPHAGLLALPKDDIPGLPRSPSGKMKGLPSVTPAAADPLMTPELADVYRTYDADMTTSVDFQE-------------------E---------------------------------------------------------EATMDTTMAPNSLQTSMPGNKAQEPEMMHDAWHFQEP ---EE-------------------EEE----------

>N terminal AMBN_HUMAN Ameloblastin PFFPQQSGTPGMASLSLETMRQLGSLQRLNTLSQYSRYGFGKSFNSLWMHGLLPPHSSLPWMRPREHETQQYEYSLP------------HHHHHHHHHHHHH------HHHH---------HHH------------------------------VHPPPLPSQPSLKPQQPGLKPFLQSAAATTNQATALKEALQPPIHLGHLPLQEGELPLVQQQVAPSDKPPKPELP ---------------------------------------------------------------------------GVDFADPQGPSLPGMDFPDPQGPSLPGLDFADPQGSTIFQIAR EEE----------------------------------------

>C terminal AMBN_HUMAN Ameloblastin LISHGPMPQNKQSPLYPGMLYVPFGANQLNAPARLGIMSSEEVAGGREDPMAYGAMFPGFGGMRPGFEGMPHNPAMG------------------EEEEE-----------EEEEE---------------H----------------------GDFTLEFDSPVAATKGPENEEGGAQGSPMPEANPDNLENPAFLTELEPAPHAGLLALPKDDIPGLPRSPSGKMKGLP--EEEE-----E-----------------------------------------E-----------------------SVTPAAADPLMTPELADVYRTYDADMTTSVDFQEEATMDTTMAPNSLQTSMPGNKAQEPEMMHDAWHFQEP ---------------------------EE--------EE--------------------------------

Fig. 1. The Jpred secondary structure prediction for human recombinant ameloblastin (hrAMBN), the N-terminal domain of humanrecombinant ameloblastin (hrAMBN-Nter), and the C-terminal domain of human recombinant ameloblastin (hrAMBN-Cter). E, betastrands; H, helices.

Biophysical characterization of recombinant human ameloblastin 263

N-terminal region of the AMBN molecule is moreordered than the C-terminal region. However, this pre-diction needs to be confirmed by additional experiments.

Thermal dependency of the structure contents ofhrAMBN, hrAMBN-Cter, and hrAMBN-Nter determinedusing CD spectroscopy

The temperature-dependent CD spectra of hrAMBN,hrAMBN-Cter, and hrAMBN-Nter are shown in Fig. 2.Obviously, the spectra look different but all share asimilar character, indicating an equilibrium betweenprevalent polyproline-II (PPII) and b-strand-type struc-tures presented in the molecules. Generally, the CDspectra of b-rich proteins can be divided into two cate-gories – bI and bII – where the bII CD spectra resemblethose of unfolded proteins (30). The bII structure isusually created by either distorted or short stretches of

the polypeptide chain, which seems to be in agreementwith the secondary structure prediction at the C-terminaldomain discussed in the previous paragraph. The spec-trum of the full-length hrAMBN at 4�C, with a negativeminimum around 203 nm and a small shoulder profiledaround 220 nm, is characteristic for bII proteins. Withincreasing temperature, an isoelliptic point appears ataround 213 nm. This is generally considered as evidenceof the PPII structure population changes with tempera-ture (31). The intensity of the shoulder in the 215–240 nm region becomes more negative upon thetemperature increase, whereas the intensity of the spec-tral band around 200 nm is positively correlated withincreasing temperature. The differential spectrum of 4�Cand 50�C has a negative band at around 220–225 nm anda positive band at around 200 nm. In analogy with thework of Lakshminaraynan et al. (32) this indicates that,upon heating, the b-sheet-like conformations areincreasingly populated at the expense of the PPII con-formation. Additionally, the spectra deconvolution (seeTable 2) shows a linear correlation of the helicity contentand the temperature as well as a decrease of the bstructure and random coil content.The CD spectrum of the hrAMBN-Nter has a distinct

negative minimum shifted towards 207 nm and theisoelliptic point is not well profiled at 215 nm, in com-parison with the CD spectrum of the full-lengthhrAMBN. There is an almost negligible decrease in thenegative part of the shoulder beyond the isoelliptic point.According to the work of Sreerama & Woody (30), wecan assign the structural state of the molecule to bI, butit does not provide information about the real content ofbI in the N-terminus. In contrast to the full-lengthhrAMBN, the spectrum deconvolution shows that thereis a significant decrease of helicity upon increasing

Fig. 2. Thermal dependency of the circular dichroism (CD)spectra of (A) full-length human recombinant ameloblastin(hrAMBN), (B) the N-terminal domain of human recombinantameloblastin (hrAMBN-Nter), and (C) the C-terminal domainof human recombinant ameloblastin (hrAMBN-Cter).

Table 2Structure content deconvolution of thermal dependent circulardichroism (CD) spectra of human recombinant ameloblastin(hrAMBN), the N-terminal domain of human recombinantameloblastin (hrAMBN-Nter), and the C-terminal domain of

human recombinant ameloblastin (hrAMBN-Cter)

Temperature

a-Helix(%)

Antiparallel(%)

Parallel(%)

b-Turn (%)

Randomcoil(%)

Totalsum(%)

hrAMBN4�C 18 24 10 19 30 10020�C 14 32 10 19 25 10037�C 15 30 10 19 26 10150�C 16 29 10 19 27 10090�C 26 18 9 19 28 100

hrAMBN-Cter

4�C 11 37 10 19 24 10020�C 14 32 10 19 25 10037�C 15 30 10 19 26 10150�C 16 29 10 19 27 10090�C 26 18 9 19 28 100

hrAMBN-Nter

4�C 30 11 9 19 31 10120�C 32 12 10 20 32 10537�C 30 10 9 18 32 10050�C 24 11 11 18 37 100

264 Wald et al.

temperature although it should be noted that the helicityvalue is still double that of the full-length hrAMBNaccompanied by an increase of the random coil type ofstructure.The best profiled among the reported spectra is the

hrAMBN-Cter CD spectrum. The well-defined isoellipticpoint at 210 nm, accompanied by little red shift ofobserved shorter negative minima (�196 nm at 4�C and�203 nm at 50�C), as well as a decrease of intensity inthe 215–240 nm region, ultimately indicates that theC-terminal domain changes its conformation uponheating dramatically. This dramatic increase in tem-perature induces a change in the conformation of thec-terminal domain, and this alters the equilibrium amongb structures, random coil, and the PPII type of con-formation. As shown in Table 2, we also observed anincrease of the helical content with the rise in tempera-ture. From this point of view, the CD spectrum ofhrAMBN-Cter is more similar to the full-lengthhrAMBN than to hrAMBN-Nter.We have to point out that there is a significant dif-

ference in the ellipticity signals in the CD spectra of themolecules: a difference of almost twofold exists betweenhrAMBN-Cter and hrAMBN or hrAMBN-Nter. Thisseems to be a common problem for IDPs and linked tothe concentration of a protein. Despite the fact that wedetermined the protein concentration using the samemethod and treated all samples equally, we cannoteliminate an effect of unobservable aggregation in theexperimental sample and a decrease in the concentrationof protein in the sample. This is a known problem forCD spectra measurement of IDPs. Nevertheless, animprovement of the reliability of the CD spectroscopyanalysis should be achieved by combining ECD spec-troscopy with other conformationally sensitive spectro-scopic techniques (e.g. intra red (IR) or vibrationalcircular dichroism (VCD).

Influence of TFE on the AMBN CD spectrum

The CD spectra of all three AMBN molecules at physi-ological temperature did not provide a clear signal of adominant and well-profiled structure. To mimic condi-tions relevant to different protein environments, we tes-ted the AMBN molecules and their properties in twomodel environments: in the presence of different salts(NaCl and CaCl2) and in the presence of TFE.Trifluorethanol is a compound used to promote the

helical content in proteins and for mimicking conditionssuitable for protein–protein interactions (33). Increasingamounts of TFE (from 0 to 50%) were investigated fortheir ability to induce structural changes (Fig. 3) ofhrAMBN, hrAMBN-Nter, and hrAMBN-Cter.The CD spectrum of the full-length hrAMBN showed

significant structural changes upon the addition of TFE,showing that the ability to adopt a more ordered a-typestructure is inherent in the sequence of AMBN. At 50%TFE, the hrAMBN shows a CD spectrum typical of ahelical structure, with two minima at 208 and 222 nm.The helical content of hrAMBN at 50% TFE was cal-culated to be �45% (See Table 3). We also observed an

isoelliptic point at 213 nm, which is evidence of astructural transition between two states: disordered tostructured (34). Comparison of the CD spectra ofhrAMBN-Nter with that of hrAMBN-Cter shows that theregion dominantly responsible for the increasing struc-ture content of the full-length hrAMBN is the N-termi-nal region of the molecule (hrAMBN-Nter). Apparently,the hrAMBN-Nter helical content at 50% TFE is evenhigher than for the full-length hrAMBN (62%) and theincrease is about 30% in comparison with a TFE-freeenvironment. The TFE stimulates an increase of thehelical content also in hrAMBN-Cter but this is muchweaker for hrAMBN and hrAMBN-Nter. There is aninteresting shift of the minimum – from 203 nm to208 nm – and a decrease of the elipticity signal at222 nm, pointing out that even the hrAMBN-Cter

undergoes some structure transition.

Fig. 3. Trifluorethanol (TFE) dependency of the circulardichroism (CD) spectra of (A) full-length human recombinantameloblastin (hrAMBN), (B) the N-terminal domain of humanrecombinant ameloblastin (hrAMBN-Nter), and (C) the C-ter-minal domain of human recombinant ameloblastin (hrAMBN-Cter). The spectra were measured at 37�C.

Biophysical characterization of recombinant human ameloblastin 265

Titration by NaCl and CaCl2

There is a question of how the molecule adjusts itsstructural properties at different ionic strengths and inthe presence of different ions. During the process ofenamel formation, AMBN is present in an extracellularcompartment supersaturated with calcium ions. Ananalysis of the physical-chemical properties derived fromthe sequence analysis showed that the highly acidicC-terminus of AMBN can be generally influenced bycations – namely Na+, K+, and Ca2+. Calcium ions areknown to promote structuralization (35) of otherwiseunstructured regions. This interaction is believed to be animportant factor also in the formation of Ca2+-bindingmotifs in IDPs. We therefore performed a measurementof the CD spectra in the presence of 1 M CaCl2 and 0.5and 1 M NaCl.

Both NaCl and CaCl2 salts showed similar effects onthe CD spectra of full-length hrAMBN, as shown inFig. 4A. The differences were almost negligible. Anincrease of ellipticity in comparison with the full-lengthhrAMBN without ions around their minima suggests achange in equilibrium between PPII and random coil,similar to the effect of the temperature on the CD spectraof the full-length hrAMBN.We observed a very similar effect of NaCl and CaCl2

salts also on the hrAMBN-Cter molecule. The increase ofellipticity around the minima and almost no differencebetween the effect of the Na+ and Ca2+ cations suggestsagain that the response of the full-length protein to theionic environment is mainly driven by the C-terminusand its physical-chemical properties, and reflected bychanges in the equilibrium between PPII and disorderedcontent.The strong effect of salts and a remarkable difference

between the effects of Na+ and Ca2+ cations wereobserved in the CD spectrum of hrAMBN-Nter. Theincrease of the ellipticity signal for NaCl seems to beconcentration independent and half the size of the re-sponse of the molecule on the content of CaCl2 salt. Thisis rather surprising regarding the strong acidic characterof the hrAMBN-Cter where binding of cations andinduced structural changes were expected primarily.Another explanation would be a stronger non-specificsalting-out effect of CaCl2 than of NaCl, which corre-sponds to the change in protein concentration.

DLS results

Figure 5 shows the DLS spectra of the full-lengthhrAMBN, hrAMBN-Nter, and hrAMBN-Cter, and alsoof BSA, which was used as a standard representative of aglobular, well-structured protein of similar size toAMBN. The broadest peak (with a half-width spanninginterval of approximately 90 nm), corresponding tohrAMBN-Cter, illustrates clearly the disordered nature ofthe C-terminus and the huge spectrum of its structuralforms. It is apparent from the spectrum of hrAMBN-Nter (red curve) that there are two forms in solution: the

Fig. 4. Circular dichroism (CD) spectra of (A) full-length humanrecombinant ameloblastin (hrAMBN), (B) the N-terminaldomain of human recombinant ameloblastin (hrAMBN-Nter),and (C) the C-terminal domain of human recombinant amelob-lastin (hrAMBN-Cter) in the presence of NaCl and CaCl2 salts at37�C.

Fig. 5. Dynamic light scattering (DLS) spectra of humanrecombinant ameloblastin (hrAMBN) (green), the C-terminaldomain of human recombinant ameloblastin (hrAMBN-Cter)(blue), the N-terminal domain of human recombinant amel-oblastin (hrAMBN-Nter) (red), and BSA (yellow). BSA is givenas a representative example of a globular protein of comparablesize to AMBN that has a stable three-dimensional structure. Itshydrodynamic radius is significantly lower than intrinsicallydisordered proteins (IDPs).

266 Wald et al.

first, more abundant, form is compact with a Stokesradius of around 7 nm (half width = 2.5 nm), and thesecond, less abundant, form has a Stokes radius of�20 nm (half width = �30 nm). This is in agreementwith reported structure properties of hrAMBN-Nter fromCD spectra, which points to a more ordered, morecompact, and partially helical structure of the indepen-dent N-terminal domain of the AMBN than its corre-sponding C-terminal part.The DLS spectrum of the full length hrAMBN is

characterized by one peak with a maximum at 35 nm(half width = 50 nm). This suggests that the influence ofhrAMBN-Nter on the hydrodynamic behaviour of thefull-length hrAMBN molecule is small, and contributesto reducing the half width value of the full-lengthhrAMBN, which is still large. The disordered characterof all the studied molecules becomes apparent whencompared with the DLS profile of BSA.

The N- and C-terminal domains of AMBN do notinteract with each other

No homodimer or heterodimer formation was observedusing zero-length cross-linker (soluble carbodiimideEDC) in protein systems containing the N-terminal and/or C-terminal domains of hrAMBN. Only the hrAMBN-Cter revealed non-specific intramolecular monomericcross-linking by EDC (the smear in lanes 5–8, Fig. 6).Similarly to NaCl, the presence of CaCl2 decreased thisnon-specific intramolecular cross-linking (data notshown) probably by compensation of its high negativecharge or decreasing its charge repulsion. In contrast, thechemical cross-linking reaction using the same cross-linker successfully formed a covalent homodimericproduct of human 14-3-3f to prove the system conditionsas a positive control (Fig. 6, lane 10).

CD spectrum of the equimolar mixture of thehrAMBN-Nter and hrAMBN-Cter

Additivity of the CD spectra of hrAMBN-Nter andhrAMBN-Cter was tested in their equimolar mixture at37�C. The results of the experiment were difficult tointerpret unambiguously (Fig. 7). Similarly to the workof Csizmok et al. (36) on CSD1 fragments, we obtained aresult showing that the CD spectra of the two halves ofAMBN are not additive. With respect to the results ofthe cross-linking experiment we can state that the C- andN-terminal domains do not mutually interact. Instead,they contribute individually to the specific �mixture�environment, which shifts the populations of individualconformational states of the C- and N-terminal domainsin their equimolar solution. This can be explained theo-retically by the increasing probability of producing var-ious formations of homomolecular or heteromolecularcomplexes that do not have stable and specific interac-tions but allow different conformational states to beadopted that are not possible in the full-length molecule.

Concluding remarks

The results presented here demonstrate that AMBN is anIDP with different biophysical features in the N- andC-terminal halves that do not interact with each other,suggesting that the molecule has at least two biologicalfunctions. Indeed, it has previously been shown that thecleavage of AMBN by matrix metalloproteinase 20(MMP20) upon secretion from ameloblasts separates thebasic N-terminal region from the acidic C-terminalregion (12). The relatively stable N-terminal cleavageproducts accumulate in the sheath space between enamelrods throughout the thickness of the enamel matrix, whilethe acidic C-terminal cleavage products are short livedand observed only in the more immature regions of thematrix, close to the secreting cells (1, 37). The N- andC-terminal halves of the AMBN molecule are thus sep-arated into different subcompartments in the developingenamel and probably have different biological functions

Fig. 6. Monitoring of protein–protein interactions using the1-ethyl-3-[3-dimethyl (aminopropyl)] carbodiimide hydrochlo-ride (EDC) cross-linker. Lanes 1, 3, and 4, the N-terminaldomain of human recombinant ameloblastin (hrAMBN-Nter);lanes 2, 5, and 6, the C-terminal domain of human recombinantameloblastin (hrAMBN-Cter); lanes 7 and 8, equimolar mixtureof the hrAMBN-Cter and hrAMBN-Nter; lanes 9 and 10, 14-3-3f. Calcium chloride (0 or 20 mM) was present in the reactionmixture with EDC (0 or 0.2 g l)1). The arrow indicates the 14-3-3f homodimeric product and the asterisk indicates theintramolecular 14-3-3f monomer. Proteins were separated byelectrophoresis on a 10% polyacrylamide gel and stained withsilver.

Fig. 7. Circular dichroism (CD) spectra of the equimolar mix-ture of the N-terminal domain of human recombinant amel-oblastin (hrAMBN-Nter) and the C-terminal domain of humanrecombinant ameloblastin (hrAMBN-Cter). For comparison,the CD spectra of isolated human recombinant ameloblastin(hrAMBN), hrAMBN-Nter, and hrAMBN-Cter are shown.

Biophysical characterization of recombinant human ameloblastin 267

during enamel formation, which may be related to theability of the molecule to undergo conformationalchanges when exposed to different environmental factors.From a functional aspect it is interesting that the major

enamel matrix protein, amelogenin, is also intrinsicallydisordered (32, 38). The structural disorder providesmanyfunctional advantages, especially during the formation ofhighly ordered mineralized structures like the tooth. TheIDPs can be involved in a variety of functions, fromscaffolding the crystal lattice during mineralization bybridging the large distances that initially separate thehydroxyapatite crystals, to more specific roles of adaptingand binding to specific molecules in the environment or incell membranes to anchor or stabilize the cells during en-amel formation. Also, the IDPs could simply provide away for the body to control extracellular biomineraliza-tion at the molecular level by providing a set of moleculesthat can interact and form assemblies at, or on, the crystalsurface to modulate crystal growth and control enzymaticactivity by substrate inhibition and topochemical func-tions. For instance, it could be that the IDPs interact withproteases, but are only cleaved when they are structuredcorrectly, possibly only when bound to or released from acrystal surface. Mutual interactions between amelogeninand AMBN (39), as well as between amelogenin and e-namelin (40, 41), have been suggested.

Acknowledgements – This work was supported by the CzechScience Foundation (grant number P302/10/0427) and Institu-tional Research Concepts AV0Z505200701 (J.V.) andAV0Z50200510 (R.O.). Additionally, it was supported by grantno. LC512 from the Ministry of Education, Youth and Sports(MSMT) of the Czech Republic and was part of research pro-ject nos Z40550506 and MSM6198959216.

Conflicts of interest – The authors declare no conflicts ofinterest.

References1. Uchida T, Murakami C, Dohi N, Wakida K, Satoda T,

Takahashi O. Synthesis, secretion, degradation, and fate ofameloblastin during the matrix formation stage of the ratincisor as shown by immunocytochemistry and immunochem-istry using region-specific antibodies. J Histochem Cytochem1997; 45: 1329–1340.

2. Wang HJ, Tannukit S, Zhu DH, Snead ML, Paine ML.Enamel matrix protein interactions. J Bone Miner Res 2005; 20:1032–1040.

3. Cerny R, Slaby I, Hammarstrom L, Wurtz T. A novel geneexpressed in fiat ameloblasts codes for proteins with cell bind-ing domains. J Bone Miner Res 1996; 11: 883–891.

4. Krebsbach PH, Lee SK, Matsuki Y, Kozak CA, Yamada

RM, Yamada Y. Full-length sequence, localization, andchromosomal mapping of ameloblastin - A novel tooth-specificgene. J Biol Chem 1996; 271: 4431–4435.

5. MacDougall M, Simmons D, Gu TT, Forsman-Semb K,Mardh CK, Mesbah M, Forest N, Krebsbach PH, Yamada

Y, Bernal A. Cloning, characterization and immunolocaliza-tion of human ameloblastin. Eur J Oral Sci 2000; 108: 303–310.

6. Hu CC, Fukae M, Uchida T, Qian Q, Zhang CH, Ryu OH,Tanabe T, Yamakoshi Y, Murakami C, Dohi N, Shimizu M,Simmer JP. Sheathlin: cloning, cDNA/polypeptide sequences,and immunolocalization of porcine enamel sheath proteins.J Dent Res 1997; 76: 648–657.

7. Yamakoshi Y, Tanabe T, Oida S, Hu CC, Simmer JP, Fukae

M. Calcium binding of enamel proteins and their derivativeswith emphasis on the calcium-binding domain of porcinesheathlin. Arch Oral Biol 2001; 46: 1005–1014.

8. Kobayashi K, Yamakoshi Y, Hu JCC, Gomi K, Arai T, Fukae

M, Krebsbach PH, Simmer JP. Splicing determines the glyco-sylation state of ameloblastin. J Dent Res 2007; 86: 962–927.

9. Vymetal J, Slaby I, Spahr A, Vondrasek J, Lyngstadaas SP.Bioinformatic analysis and molecular modelling of human a-meloblastin suggest a two-domain intrinsically unstructuredcalcium-binding protein. Eur J Oral Sci 2008; 116: 124–134.

10. Dyson HJ, Wright PE. Intrinsically unstructured proteins andtheir functions. Nat Rev Mol Cell Biol 2005; 6: 197–208.

11. Chun Y-HP, Yamakoshi Y, Yamakoshi F, Fukae M, Hu JC-C, Bartlett JD, Simmer JP. Cleavage site specificity of MMP-20 for secretory-stage ameloblastin. J Dent Res 2010; 89: 785–790.

12. Iwata T, Yamakoshi Y, Hu JCC, Ishikawa I, Bartlett JD,Krebsbach PH, Simmer JP. Processing of ameloblastin byMMP-20. J Dent Res 2007; 86: 153–157.

13. Tompa P. Unstructural biology coming of age. Curr Opin StructBiol 2011; 21: 419–425.

14. Uversky VN, Dunker AK. Understanding protein non-fold-ing. Biochim Biophys Acta 2010; 1804: 1231–1264.

15. Paine ML, Wang HJ, Luo W, Krebsbach PH, Snead ML. Atransgenic animal model resembling amelogenesis imperfectarelated to ameloblastin overexpression. J Biol Chem 2003; 278:19447–19452.

16. Kitagawa M, Kitagawa S, Nagasaki A, Miyauchi M,Uchida T, Takata T. Synthetic ameloblastin peptide stimu-lates differentiation of human periodontal ligament cells. ArchOral Biol 2011; 56: 374–379.

17. Fukumoto S, Yamada A, Nonaka K, Yamada Y. Essentialroles of ameloblastin in maintaining ameloblast differentiationand enamel formation. Cells Tissues Organs 2005; 181: 189–195.

18. Fong CD, Cerny R, Hammarstrom L, Slaby I. Sequentialexpression of an amelin gene in mesenchymal and epithelialcells during odontogenesis in rats. Eur J Oral Sci 1998; 106:324–330.

19. Zeichner-David M, Chen LS, Hsu ZY, Reyna J, Caton J,Bringas P. Amelogenin and ameloblastin show growth-factorlike activity in periodontal ligament cells. Eur J Oral Sci 2006;114: 244–253.

20. Spahr A, Lyngstadaas SP, Slaby I, Haller B, Boeckh C,Tsoulfidou F, Hammarstrom L. Expression of amelin andtrauma-induced dentin formation. Clin Oral Investig 2002; 6:51–57.

Table 3Structure content deconvolution of trifluorethanol (TFE)-dependent circular dichroism (CD) spectra of human

recombinant ameloblastin (hrAMBN), the N-terminal domain ofhuman recombinant ameloblastin (hrAMBN-Nter), and theC-terminal domain of human recombinant ameloblastin

(hrAMBN-Cter)

TFE content

a-Helix(%)

Antiparallel(%)

Parallel(%)

b-Turn(%)

Randomcoil(%)

TotalSum(%)

hrAMBN 19 22 10 18 30 10010% TFE 20 18 11 18 34 10020% TFE 25 13 10 17 35 10030% TFE 23 13 10 17 35 10040% TFE 33 11 8 17 30 10050% TFE 45 8 6 17 22 10080% TFE 52 6 6 17 20 100

hrAMBN-Cter 14 30 10 19 26 10010% TFE 18 29 9 19 24 10050% TFE 24 21 9 19 27 100

hrAMBN-Nter 31 11 9 19 30 10010% TFE 34 8 9 17 33 10130% TFE 38 7 8 16 31 9950% TFE 62 4 4 13 16 99

268 Wald et al.

21. Spahr A, Lyngstadaas SP, Slaby I, Pezeshki G. Ameloblastinexpression during craniofacial bone formation in rats. Eur JOral Sci 2006; 114: 504–511.

22. Tamburstuen MV, Reseland JE, Spahr A, Brookes SJ,Kvalheim G, Slaby I, Snead ML, Lyngstadaas SP. Amel-oblastin expression and putative autoregulation in mesenchy-mal cells suggest a role in early bone formation and repair. Bone2011; 48: 406–413.

23. Nakamura Y, Slaby I, Spahr A, Pezeshki G, Matsumoto K,Lyngstadaas SP. Ameloblastin fusion protein enhances pulpalhealing and dentin formation in porcine teeth. Calcif Tissue Int2006; 78: 278–284.

24. Tamburstuen MV, Reppe S, Spahr A, Sabetrasekh R,Kvalheim G, Slaby I, Syversen U, Lyngstadaas SP, Rese-

land JE. Ameloblastin promotes bone growth by enhancingproliferation of progenitor cells and by stimulating immun-oregulators. Eur J Oral Sci 2010; 118: 451–459.

25. Gasteiger E, Gattiker A, Hoogland C, Ivanyi I, Appel RD,Bairoch A. ExPASy: the proteomics server for in-depthprotein knowledge and analysis. Nucleic Acids Res 2003; 31:3784–3788.

26. Cuff JA, Barton GJ. Application of multiple sequencealignment profiles to improve protein secondary structureprediction. Proteins 2000; 40: 502–511.

27. Cole C, Barber JD, Barton GJ. The Jpred 3 secondarystructure prediction server. Nucleic Acids Res 2008; 36: W197–W201.

28. Lobley A, Whitmore L, Wallace BA. DICHROWEB: aninteractive website for the analysis of protein secondary struc-ture from circular dichroism spectra. Bioinformatics 2002; 18:211–212.

29. Whitmore L, Janes RW, Wallace BA. Protein CircularDichroism Data Bank (PCDDB): data bank and website de-sign. Chirality 2006; 18: 426–429.

30. Sreerama N, Woody RW. Structural composition of beta(1)and beta(11) proteins. Biophys J 2003; 84: 171A–172A.

31. Bochicchio B, Tamburro AM. Polyproline II structure inproteins: identification by chiroptical spectroscopies, stability,and functions. Chirality 2002; 14: 782–792.

32. Lakshminarayanan R, Yoon I, Hegde BG, Fan DM, Du C,Moradian-Oldak J. Analysis of secondary structure and self-assembly of amelogenin by variable temperature circular

dichroism and isothermal titration calorimetry. Proteins 2009;76: 560–569.

33. Moncoq K, Broutin I, Craescu CT, Vachette P, Ducruix A,Durand D. SAXS study of the PIR domain from the Grb14molecular adaptor: a natively unfolded protein with a transientstructure primer? Biophys J 2004; 87: 4056–4064.

34. Dettin M, Scarinci C, Zanotto C, Roncon R, De Rossi A,Di Bello C. Biological and conformational studies on ana-logues of a synthetic peptide enhancing HIV-1 infection. J PeptSci 1998; 4: 436–448.

35. He ZP, Dunker AK, Wesson CR, Trumble WR. Ca2 + -Induced Folding and Aggregation of Skeletal-Muscle Sarco-plasmic-Reticulum Calsequestrin - the Involvement of theTrifluoperazine-Binding Site. J Biol Chem 1993; 268: 24635–24641.

36. Csizmok V, Bokor M, Banki P, Klement T, Medzihradszky

KF, Friedrich P, Tompa KA, Tompa P. Primary contact sitesin intrinsically unstructured proteins: the case of calpastatinand microtubule-associated protein 2. Biochemistry (Mosc)2005; 44: 3955–3964.

37. Murakami C, Dohi N, Fukae M, Tanabe T, Yamakoshi Y,Wakida K, Satoda T, Takahashi O, Shimizu M, Ryo OH,Simmer JP, Uchida T. Immunochemical and immunohisto-chemical study of the 27- and 29-kDa calcium-binding proteinsand related proteins in the porcine tooth germ. Histochem CellBiol 1997; 107: 485–494.

38. Delak K, Harcup C, Lakshminarayanan R, Sun Z, Fan

YW, Moradian-Oldak J, Evans JS. The Tooth Enamel Pro-tein, Porcine Amelogenin, Is an Intrinsically Disordered Proteinwith an Extended Molecular Configuration in the MonomericForm. Biochemistry (Mosc) 2009; 48: 2272–2281.

39. Hatakeyama J, Fukumoto S, Nakamura T, Haruyama N,Suzuki S, Hatakeyama Y, Shum L, Gibson CW, Yamada Y,Kulkarni AB. Synergistic Roles of Amelogenin and Amelob-lastin. J Dent Res 2009; 88: 318–322.

40. Fan D, Iijima M, Bromley KM, Yang X, Mathew S,Moradian-Oldak J. The cooperation of enamelin and ame-logenin in controlling octacalcium phosphate crystal morphol-ogy. Cells Tissues Organs 2011; 194: 194–198.

41. Fan DM, Du C, Sun Z, Lakshminarayanan R, Moradian-Oldak J. In vitro study on the interaction between the 32 kDaenamelin and amelogenin. J Struct Biol 2009; 166: 88–94.

Biophysical characterization of recombinant human ameloblastin 269