Biomimetic peptides that engage specific integrin-dependent signaling pathways and bind to calciumphosphate surfaces

Michele Gilbert, Cecilia M. Giachelli, Patrick S. StaytonDepartment of Bioengineering, University of Washington, Seattle, Washington 98195

Received 21 October 2002; revised 18 December 2002; accepted 24 January 2003

Abstract: Many important matrix proteins involved inbone remodeling contain separate domains that orient theprotein on hydroxyapatite and interact with target cell re-ceptors, respectively. We have designed two synthetic pep-tides that mimic the dual activities of these large, complexproteins by binding to calcium phosphate minerals and byengaging integrin-dependent signaling pathways in osteo-blasts. The addition of either PGRGDS from osteopontin orPDGEA from collagen type I to the HAP-binding domain ofstatherin (N15 domain) did not alter its �-helical structure ordiminish its affinity for hydroxyapatite. Immobilized N15-PGRGDS bound MC3T3-E1 osteoblasts predominantly viathe �v�3 integrin and induced focal adhesion kinase (FAK)

phosphorylation at comparable levels to immobilized os-teopontin. Immobilized N15-PDGEA bound MC3T3-E1 os-teoblasts predominantly through the �2�1 integrin and in-duced similar levels of FAK phosphorylation. Althoughboth peptides induced FAK phosphorylation with similartime courses, only the N15-PDGEA peptide inducedERK1/2 phosphorylation, showing that these peptides arealso capable of engaging integrin-specific signaling path-ways. This peptide system can be used to study adhesion-dependent control of signaling in the context of the relevantbiomineral surface and may also be useful in biomaterialand tissue engineering applications. © 2003 Wiley Periodi-cals, Inc. J Biomed Mater Res 67A: 69-77, 2003

INTRODUCTION

Nature has evolved sophisticated strategies for en-gineering hard tissues through the interaction of pro-teins, and ultimately cells, with inorganic mineralphases. The extracellular matrix and matricellular pro-teins that function directly at the biomineral- biologi-cal interface have the interesting property of bothrecognizing the inorganic mineral and displaying li-gand sequences that engage specific target cell recep-tors. Proteins such as osteopontin, bone sialoprotein,osteocalcin, statherin, and phosphoryn all containhighly acidic domains that direct biomineral recogni-tion. Many of the matrix proteins also contain se-quences of amino acids that are known to bind to theextracellular domains of membrane-bound adhesionreceptors. For example, collagen I contains an RGD

sequence, although the RGD is generally inactive un-less collagen is denatured or degraded.1 The Asp-Gly-Glu-Ala (DGEA) sequence of collagen has been shownto be specific for the �2�1 integrin on osteoblast-likecells.2 These adhesive domains displayed on biomin-erals can activate many different intracellular signal-ing pathways, including the mitogen-activated proteinkinase (MAPK) pathway, the focal adhesion kinase(FAK) pathway, and G-protein-coupled receptor path-ways. Adhesive interactions may also connect to reg-ulators of intracellular calcium levels3 and down-stream to transcriptional activation.4 The MAPKpathway has been linked to the activation of Cbfa1,5

and collagen-based peptides might thus direct osteo-blast differentiation.

Here we describe the design of peptides that com-bine a domain for recognition of the biomineral sur-face with a domain that can engage specific integrin-dependent signaling pathways. For HAP recognition,a well-characterized hydroxyapatite (HAP)-bindingdomain has been used from statherin, a salivary pro-tein that controls HAP nucleation and growth. High-resolution structure and dynamic studies have de-tailed the HAP recognition mechanisms of theN-terminal 15 amino acid (N15) domain.6,7 Two pep-tides containing this N15 domain and the integrin-

Correspondence to: P. Stayton; e-mail: stayton@u.washington.edu

Contract grant sponsor: National Institute of Dental Re-search; contract grant number: DE 12554

Contract grant sponsor: National Science Foundation;contract grant number: EEC-9529161

Contract grant sponsor: The Whitaker Foundation

© 2003 Wiley Periodicals, Inc.

binding domain of osteopontin (GRGDS) or collagen I(DGEA) were synthesized. The N15-PGRGDS peptidewas previously shown to bind to the �v�3 integrin onmelanoma cells.8 Here we have studied the integrinspecificity of the N15-PGRGDS peptide and comparedit with the N15-PDGEA peptide, which was designedto bind the �2�1 integrin in a comparable manner tocollagen I. These integrin specificities have subse-quently been connected to FAK and extracellular sig-nals-regulated kinase (ERK) activation, to characterizewhether the peptides can induce these importantmarkers for early signaling pathway activation.

EXPERIMENTAL PROCEDURES

Materials

Hamster anti-mouse �2 HM�2, hamster anti-rat �2 HA1/29, hamster anti-mouse �3 chain antibody (2C9.G2), hamsteranti-mouse �5 chain (HM�5), hamster IgG, group1, � (anti-TNP) control (A19-3), and mouse anti-hamster cocktail flu-orescein isothiocyanate (FITC)-conjugated immunoglobulinG (IgG) antibody were all purchased from Pharmingen (SanDiego, CA). The Tween-20 (EIA grade), PVDF membranes,and Tris-HCl gels were obtained from Bio-Rad (Hercules,CA). The micro-BCA protein assay kit, DuraWest chemilu-minescence kit, and secondary antibodies of goat anti-mouseperoxidase and goat anti-rabbit peroxidase were purchasedfrom Pierce (Rockford, IL). Mouse monoclonal antibody(mAb) against phosphotyrosine (P11120) and mouse mAbagainst FAK (F15020) were obtained from BD TransductionLaboratories (Lexington, KY). Rabbit polyclonal antibodyagainst ERK1/2 (AB3053) was obtained from Chemicon In-ternational (Temecula, CA). The rabbit polyclonal antibodyagainst active MAPK (pTEpY) was purchased from Promega(Madison, WI). The poly-D-lysine-coated 100-mm plateswere purchased from Beckton-Dickenson. Fmoc-protectedamino acids, including protected phosphoserine (Fmoc-Ser-(PO(OBzl)OH)-OH) and preloaded resins were purchasedfrom Novabiochem (San Diego, CA).

Peptide synthesis

The N15-PGRGDS, N15-PGRGES, N15-PDGEA, N15-PDGAA, and N15 peptides were synthesized by using stan-dard Fmoc solid phase synthesis. For N15-PDGEA and N15-PDGAA, the alanine-coupled resin used was Fmoc-Alanine-NovaSyn� TGA (Novabiochem). The soluble controlpeptides KDGEAG and KDGAAG were synthesized byUnited Biochemical Research, Inc. (Seattle, WA). Matrix as-sisted laser desorption/ionization mass spectrometry wasused to verify molecular weight and purity of the peptides.N15-PDGEA, N15-PDGAA, KDGEAG, and KDGAAG wereanalyzed by using electrospray mass spectrometry.

Circular dichroism and adsorption isotherms

Solutions (100 �M) of N15-PGRGDS, N15-PGRGES, N15,N15-PDGEA, and N15-PDGAA were made in 1X phos-phate-buffered saline (PBS; 140 mM NaCl, 2.7 mM KCl, 10mM Na2HPO4, 1.8 mM KH2PO4, pH 7.3). The circular di-chroism spectra were recorded for each sample on a JascoJ720 Spectropolarimeter at 0.2-nm intervals, at a rate of 100nm/min using a quartz cell with a 0.1-cm path length in thewavelength range of 195–250 nm. All spectra were correctedby using the solvent spectrum and they represent the aver-age of 10 scans.

Concentrations of stock solutions of N15, N15-PGRGDS,N15-PGRGES, N15-PDGEA, and N15-PDGAA in ddH2Owere determined by using amino acid analysis. A concen-tration versus fluorescence plot was determined by using100 �L of serially diluted peptides in 1 mL of fluoraldehyde(Pierce), which reacts with primary amines. Triplicate sam-ples of peptides were bound to 1 mg ceramic HAP for 4 h at37°C. The fluoraldehyde fluorescence was determined withan excitation at 360 nm and emission at 445 nm on a HitachiModel F-4500 Fluorescence Spectrophotometer. Langmuirisotherms were plotted with the C values calculated fromthe depleted supernatant and the Q values determined bydepletion using a BET measured surface area of 55 m2/g(performed by Dr. Allison Campbell, Batelle Pacific North-west Laboratory) for the autoclaved Bio-Rad HAP.

Cell culture

The MC3T3-E1 cells9 were obtained from the GermanCollection of Microorganisms and Cell Cultures (DSMZ)(DSMZ ACC 210). The cells were maintained in Eagle’sminimum essential medium (MEM), �-modified (Gibco),which contained 10% fetal calf serum (FCS) and 1% v/vpenicillin/streptomycin. The ability of the MC3T3-E1 cells toreduce alamarBlue was tested by using the same methodol-ogy as that used in previous studies for human melanomacells.8 At each measured time point, the fluorescence of thesupernatant was measured by combining 150 �L of super-natant with 1 mL of 1X PBS and exciting the sample at 530nm and measuring the emission at 584 nm.

Flow cytometry analysis of MC3T3-E1integrin complement

MC3T3-E1 cells were split and plated at a density of40,000 cells/cm2 on 75 cm2 TCPS flasks. At days 3, 6, 12, and18, analyses were performed. All cells were passage 3 exceptfor day 18 cells that were passage 4. The cells were plated in�MEM/10% FCS/1% penicillin/streptomycin. At the desig-nated day, 500,000 cells were aliquoted into sterile 15-mLtubes and pelleted by centrifugation at 1000 rpm for 3 min.The cells were resuspended in 1 mL of ice-cold 1X PBS/0.2%bovine serum albumin (BSA) and pelleted again, and thePBS/BSA was aspirated off. The cells were resuspended in50 �L of 1X PBS/0.2% BSA containing 1 �g of one of the

70 GILBERT ET AL.

primary antibodies or PBS/BSA and allowed to incubate onice for 30 min. After washing, 1 �g of the anti-hamster IgGFITC conjugated secondary antibody in 50 �L of BSA/0.2%BSA was added to each of the samples. Samples were incu-bated on ice for 30 min, and after washing were resuspendedin 300 �L of 1X PBS for flow cytometric analysis. Flowcytometric analysis was performed on a nonsorting BectonDickenson FACScan analyzer set at the high flow rate. Tenthousand events were counted per sample. Gating elimi-nated counts that had both low forward and side scatter(cellular debris). The geometric mean value for each anti-body tested was baseline corrected for cell autofluorescenceby subtracting the geometric mean of the cells treated withPBS/BSA but no antibodies. To determine relative levels ofintegrin expression, the baseline-corrected geometric meansof the cells that were incubated with anti-integrin subunitantibodies were divided by the baseline-corrected value forcells incubated with the hamster IgG isotype control.

Adhesion assays

The adhesion assays used previously reported methodol-ogies with minor modifications.8 The MC3T3-E1 cells weredetached with 0.25% % trypsin/1 mM EDTA � 4 Na solutionand resuspended in �-MEM. The cells were pelleted andthen resuspended in �-MEM/BSA/HEPES to a final densityof 100,000 cells/mL. These cells (200 �L) were then added tothe peptide-coated HAP. The protocol for the competitionexperiments is similar to the cell adhesion assays with thefollowing exceptions. First, the HAP was coated with aconstant concentration of either 1 mg/mL N15-PDGEA or 1mg/mL N15-PGRGDS. Second, before addition of theMC3T3-E1 cells to the peptide-coated HAP, the cells wereincubated with varying concentrations (0–1 mM) ofKDGEAG, KDGAAG, GRGDSP, or GRGESP for 15 min at37°C/5% CO2. For the integrin-blocking experiments, ratherthan preincubating the cells with a soluble peptide, the cellswere incubated with a 1:200 dilution of one of the threeantibodies (anti-�2, anti-�3, or anti-hamster IgG).

Preparation of peptide-coated plates

The 100-mm poly-D-lysine-coated plates were washedtwo times with 1X PBS. After washing, 5 mL of 1 mg/mLsolution of peptide in 1X PBS (N15, N15-PGRGDS, N15-PGRGES, N15-PDGEA, or N15-PDGAA) or just 5 mL of 1XPBS was added to each plate, and the plates were allowed toincubate at 4°C overnight. The peptide solution (or PBSalone) was then removed, and the plates were UV irradiatedto sterilize them and then washed two times with 5 mL of 1XPBS to rinse off any loosely bound peptide before addition ofcells.

Cell culture and attachment topeptide-coated plates

The MC3T3-E1 cells were maintained in culture in �-MEMmedia supplemented with penicillin and streptomycin and

10% FBS. The cells were allowed to proliferate for 3 days andthen were washed with 1X PBS and detached by usingtrypsin/EDTA. The cells were then washed twice in�-MEM/0.5% BSA, counted and suspended at a density of4.4 � 105 cells/mL. After 40 min, 5 mL of cells was added toeach of the peptide-coated plates and allowed to incubate fora set time between 20 and 120 min at 37°C/5% CO2.

Cell lysis and western blot analysis

After incubation, the media were aspirated, and the cellswere washed with 5 mL of ice-cold 1X PBS. The PBS wasthen aspirated and 100 �L of lysis buffer (10 mM Tris-HCl,1% SDS, 1 mM sodium orthovanadate, pH 7.4) was added toeach plate. The cell lysates were scraped off the plates anddrawn through a 251/2-gauge needle. The lysates wereboiled at 95°C for 5 min, and any precipitate was spun downat 15,000 rpm for 15 min. The total protein concentration ofthe cell lysates was measured by using a micro-BCA kit. ForWestern blot analysis, equal protein amounts were loadedinto Tris-HCl gels (either 10% or 4–15% gradient gels) andrun under denaturing conditions. The blot chemilumines-cence was detected by using autoradiography, and the im-ages were analyzed after scanning with NIH Image.

RESULTS

Characterization of peptide solution structure andHAP adsorption isotherms

The circular dichroism spectra of N15, N15-PGRGDS, and N15-PGRGES were previously charac-terized, and the peptides displayed comparable �-hel-ical content in solution.8 The N15-PDGEA and N15-PDGAA peptides were found to have the same looselyhelical structure (data not shown). Adsorption iso-therm analysis was also performed for N15-PDGEAand N15-PDGAA to compare to the previously deter-mined properties of N15-PGRGDS, N15-PGRGES, andN15.8 The N15-PDGEA and N15-PDGAA peptides dis-played similar binding isotherms as N15-PGRGDSand N15-PGRGES. The addition of the C-terminal se-quences slightly increased the area per molecule forthe fusion peptides relative to N15, but the recognitionof HAP was clearly being directed by the N-terminalregion of the statherin sequence (Table I). This wasconsistent with solid-state NMR experiments thatshowed that the cell-binding domains of N15-PGRG(D/E)S were dynamic while the peptides wereadsorbed to HAP8 and that the N-terminal region ofthe N15 peptide directs HAP binding.6,7

Flow cytometric analysis of MC3T3-E1integrin profiles

The integrin profiles of osteoblasts have been foundto vary, depending on their source (e.g., primary vs

PEPTIDES THAT ENGAGE SIGNALING PATHWAYS 71

cell line) and species. The MC3T3-E1 cells were testedfor the presence of �2, �5, and �3 subunits, which werefound at all time points and at the same levels fromdays 3 to 18 in culture (Fig. 1). From the knownheterodimer repertoire, this finding implicates the�2�1 integrin, the �v�3 integrin, and the �5�1 integrinon the cell surface. The �3 subunit was found at thehighest level, followed by the �5 subunit and finallythe �2 subunit at a relative ratio of �8:4:1. These dataagree with other integrin protein level studies thatshowed little time-dependent regulation of �5�1 and�v�3 integrin expression. Time-dependent expressionwas measured by immunoprecipitation of differentintegrin subunits from rat primary osteoblast culturesfollowed by Western blotting, but the relative levels ofthe different integrin subunits were not quantified.10

Determination of integrin specificityin adhesion assays

The MC3T3-E1 cells displayed dose-dependentbinding to the N15-PDGEA and N15-PGRGDS pep-tides adsorbed to HAP (Fig. 2). The negative control

peptides (N15-PDGAA and N15-PGRGES, respec-tively) did not exhibit dose-dependent binding. TheN15 peptide inhibited nonspecific cell binding to themineral surface with increasing peptide coverage ofthe HAP surface, which was similar to the inhibitionof nonspecific cell binding to acidic urinary protein-coated calcium oxalate crystals.11–14 To further con-firm the binding sequence specificity of these pep-tides, the MC3T3-E1 cells were preblocked incompetition assays with either the soluble inhibitorsGRGDSP or KDGEAG or their control GRGESP orKDGAAG sequences, respectively. The GRGDSPpeptide completely abrogated cell binding to theimmobilized N15-PGRGDS peptide at a dose of 0.5mM, with an EC50 of �0.1 mM, whereas it required�1 mM KDGEAG to block binding of the cells toN15-PDGEA-coated HAP (Fig. 3). The control pep-tides KDGAAG and GRGESP did not block cellbinding to the N15-PDGEA or N15-PGRGDS-coatedHAP, respectively. Cross-talk between RGD andDGEA-binding integrins could also be measured byusing the soluble competitors (i.e., the effects of the

TABLE ILangmuir Adsorption Isotherm Parameters Calculated for N15, N15-PGRGDS, N15-PGRGES, N15-PDGEA, and

N15-PDGAA on 80-�m Ceramic Bio-Rad HAPa

N (mol/m2) Error in N K (liter/mol) Error in K

N15 4.65e-7** 7.75e-8 3.3e4 8.03e3N15-PGRGDS 3.62e-7 9.27e-9 3.0e4 3.15e3N15-PGRGES 3.71e-7 3.42e-8 3.45e4 4.82e3N15-PDGEA 3.47e-7 8.5e-9 1.7e4 5.37e3N15-PDGAA 3.48e-7 4.04e-8 1.9e4 4.73e3

aThe only value that was significantly different as calculated using the SNK analysis was the N value for N15 which wassignificantly higher than any of the other peptides. All K values were deemed the same for �T � 0.05 using SNK.

**�T � 0.05.

Figure 1. Relative levels of integrin subunits as measuredby flow cytometry. The �2 subunit is most likely part of theheterodimer �2�1, and the �3 subunit is most likely part ofthe �v�3 integrin. The �5 subunit should be part of thefibronectin integrin �5�1. The relative levels of the ex-pressed subunits were 8:4:1 for �3:�5:�2.

Figure 2. Ligand specific binding of MC3T3-E1 cells topeptide-coated HAP. The MC3T3-E1 cells bound in a dose-dependent manner to the HAP that was coated with pep-tides that contained integrin-binding sequences RGD andDGEA but not to the controls that contained the sequencesRGE or DGAA. The specific cell binding was also not due tothe N15 domain. In fact, this acidic domain decreased thelevel of nonspecific cell binding to the HAP. BSA, whichcontains no integrin-binding domains, did not show anydose-dependent binding of cells.

72 GILBERT ET AL.

soluble KDGEAG on cell binding to N15-PGRGDSor the effect of soluble GRGDSP on cell binding toN15-PDGEA). At concentrations up to 200 �M(highest point measured), the KDGEAG peptide didnot alter binding to N15-PGRGDS-coated HAP, andGRGDSP did not interfere with cell binding to N15-PDGEA-coated HAP.

Antibody-blocking experiments were performedto determine which integrins the N15-PGRGDS andN15-PDGEA peptides recognize. The DGEA se-quence has been reported to be specific for the �2�1integrin,15 whereas the RGD sequence interacts with

both the �v�3 or �5�1 heterodimers (both expressedby the MC3T3-E1 cells). The anti-�2-antibodyblocked most cell binding to the immobilized N15-PDGEA peptide but did not alter cell binding to theN15-PGRGDS-coated HAP (Fig. 3). The anti-�3-an-tibody blocked �60% of the binding of theMC3T3-E1 cells to N15-PGRGDS-coated HAP buthad no effect on the binding of the cells to N15-PDGEA-coated HAP. The �3-antibody primarilyblocked the �v�3 integrin, and the incomplete block-age was probably due to the N15-PGRGDS engage-ment of the �5�1 integrin.

Figure 3. Blocking assays using small soluble peptides or antibodies to inhibit cells binding to N15-PGRGDS or N15-PDGEAimmobilized on HAP. The HAP was coated by using 1 mg/mL of N15-PGRGDS (a) or N15-PDGEA (b). MC3T3-E1 integrinswere preblocked with either soluble peptides or antibodies before being allowed to interact with peptide-coated HAP. Thesoluble GRGDSP could block binding of cells to N15-PGRGDS-coated HAP but not to N15-PDGEA-coated HAP. The solubleKDGEAG could selectively block MC3T3-E1 binding to the N15-PDGEA-coated HAP but not to the N15-PGRGDS-coatedHAP. The control peptides GRGESP and KDGAAG did not have any effect on cell binding. The anti-�3 antibody could blockcell binding to N15-PGRGDS-coated HAP but not N15-PDGEA-coated HAP. The anti-�2 antibody could block most ofMC3T3-E1 binding to N15-PDGEA-coated HAP but not to N15-PGRGDS-coated HAP. The control IgG antibody had no effecton cell binding to either peptide coated HAP.

PEPTIDES THAT ENGAGE SIGNALING PATHWAYS 73

Western blot analysis of FAK activation

The anchorage-dependent up-regulation of the FAKand MAPK pathways was measured as a function ofpeptide coating and time for N15-PGRGDS and N15-PDGEA. The Western blots showed that there wereequivalent steady-state quantities of ERK1/2 and FAKpresent on MC3T3-E1 cells on either of the peptide-coated surfaces (data not shown). Steady-state quan-tities of ERK1/2 and FAK were also found whenMC3T3-E1 cells were plated onto collagen type I16 andwhen F3 rat primary osteoblasts were plated ontofibronectin or collagen I.17 The relative level of specificadhesion to the different peptide-coated surfaces wasanalyzed by using Western blot analysis followed bydensitometric analysis of the bands using NIH Image.The same analysis methods were performed on twoindependent coating experiments that yielded two in-dependent Western blots. The Student-Newman-Keul’s test was used to examine the multiple compar-isons between the different coatings. Although theexact ranking order of the number of cells adhering tothe specific coatings was not identical in both experi-ments [suspended, N15-PDGAA, N15-PGRGES, poly-D-lysine, N15, N15-PGRGDS, N15-PDGEA for oneblot [Fig. 4(a)] and suspended, N15-PDGAA, N15-PGRGES, N15, poly-D-lysine, N15-PDGEA, N15-PGRGDS for the second blot [Fig. 4(b)], both showedthe same trend in significant differences when ana-lyzed via SNK]. A complete analysis of variance oneach data set showed that significant differences didexist between the different coatings at a p � 0.01. TheSNK analyses revealed that both the N15-PDGEA andN15-PGRGDS coatings were significantly differentfrom all the other coatings at �T � 0.01, whereas theother closer comparisons were not significant for bothexperiments.

When the MC3T3-E1 cells were plated onto im-mobilized N15-PGRGDS or N15-PDGEA for 1 h, thephosphorylated form of FAK was detected (Fig. 4).Cells plated onto the control peptides N15, N15-PGRGES, and N15-PDGAA showed a level of phos-phorylation comparable to that of the backgroundpoly-D-lysine coated plates. The adhesion of cells tothese surfaces was electrostatic and not receptormediated, but lower levels of tyrosine phosphory-lated FAK when osteoblasts or fibroblasts wereplated onto poly-L-lysine has been previously re-ported.18,19 Suspended cells showed no phosphory-lation. The time course of FAK phosphorylation wasdetermined on the N15-PGRGDS or N15-PDGEAsurfaces, and there was rapid up-regulation of ty-rosine phosphorylation (Fig. 5). At 20 min, cellsplated on either N15-PGRGDS or N15-PDGEAshowed a marked increase in tyrosine phosphoryla-tion of FAK. By 60 –90 min, a maximum amount of

phosphorylation was seen for FAK, with levels thendecreasing out to 120 min. The FAK phosphoryla-tion time courses on the peptide surfaces paralleledthe time-dependent tyrosine phosphorylation re-ported previously when MC3T3-E1 cells wereplated onto complete collagen type I.16

Figure 4. NIH Image analysis of Western blot band in-tensities for the level of tyrosine phosphorylated focaladhesion kinase after 1 h of plating MC3T3-E1 cells ontopoly-D-lysine (PDL) surfaces treated with different N15fusion peptides. The results of two different independentexperiments are shown. Although there was some resid-ual tyrosine phosphorylation of FAK still present for cellsplated onto PDL, cells plated onto N15-PGRGDS andN15-PDGEA which can specifically engage the �v�3 or�2�1 integrins showed a 1.5–2-fold increase in FAK phos-phorylation over PDL. N15, N15-PGRGES, and N15-PDGAA all showed levels of FAK phosphorylation com-parable to the PDL control. Suspended cells showed noappreciable level of FAK phosphorylation. These inde-pendent experiments were statistically analyzed by usingSNK methods and the FAK phosphorylation on the N15-PDGEA and N15-PGRGDS were found to be statisticallysignificant at a level of �T � 0.01. Values were calculatedfrom band intensity using NIH Image from a Western blotdirected against phosphorylated tyrosine for MC3T3-E1cell lysates and the SEM is shown.

74 GILBERT ET AL.

Western blot analysis of ERK1/2 activation

To test whether the peptides could direct adhesion-specific MAPK activation, Western blot analysis ofERK1/2 activation was conducted at the 1-h timepoint (based on previous results with rat and mouseosteoblasts on collagen I that showed maximal phos-phorylation of ERK1 or ERK 2 after 60–120 min).16,17

The Western blot analysis of MC3T3-E1 cells showedthat only the cells plated onto N15-PDGEA, and notthe N15-PGRGDS peptide, had active ERK 1 and ERK2 present (Fig. 6). The binding of MC3T3-E1 cells toany of the control surfaces (N15, N15-PGRGES, N15-PDGAA, and poly-D-lysine) did not result in anyphosphorylation of ERK 1 or ERK 2. These resultsshowed that �v�3 engagement to N15-PGRGDS didnot activate either ERK 1 or ERK 2, whereas �2�1interaction with the N15-PDGEA peptide resulted inactivation of both ERK 1 and ERK 2.

DISCUSSION

The N15 fusion peptides were designed to use thestatherin domain for structured binding and orienta-tion on calcium phosphate or cationic polyelectrolytesurfaces and to present cell-accessible integrin-bind-ing domains that could trigger specific intracellularsignaling pathways. The N15 domain has been shownto recognize hydroxyapatite through the N-terminalacidic pentapeptide sequence, which contains twophosphorylated serines, and this acidic region alsoprovides electrostatic complementarity for interac-tions with cationic polyelectrolytes such as polylysine.

The remainder of the N15 peptide is �-helical and theC-terminus has been shown to be mobile and dynamicon the HAP surface. The mobility and dynamics areretained when cell-binding sequences are added to theC-terminus, and these properties are likely key to theexcellent functional activities of the fusion peptides.

Binding of osteoblasts to either N15 fusion peptideresulted in the rapid up-regulation of FAK tyrosinephosphorylation. FAK tyrosine phosphorylation hasbeen observed when osteoblasts are plated onto fi-bronectin, collagen I, or osteopontin.16,18,20 For fi-bronectin and collagen I, the �1 integrin subunit wascritical.16,17 Although osteoblast binding to osteopon-tin was determined to result in FAK tyrosine phos-phorylation, the integrin engaged by osteopontin wasnot determined.20 The �3-dependent tyrosine phos-phorylation of FAK had not been previously studiedfor osteoblasts. The MC3T3-E1 cells were preferen-tially bound to the N15-PGRGDS peptide with the�v�3 integrin compared to the �5�1 fibronectin inte-grin. The GRGDS linear peptide binds better to the�v�3 integrin than to �5�1 integrin.21,22 The results forthe N15-PGRGDS peptide are the first to show thatthere is a time-ependent and specific FAK tyrosinephosphorylation event on engagement of osteoblaststo the �v�3 integrin (although the broader finding inprimary osteoblasts is yet to be determined). The �3subunit has been linked to FAK phosphorylation incolon carcinoma cells23 and with chimeric receptorstudies in fibroblasts.24

Although both the N15-PGRDGS and N15-PDGEApeptides directed adhesion events leading to FAKphosphorylation, only the N15-PDGEA peptide led toMAPK pathway activation as monitored by ERK 1 and2 phosphorylation. Because �5�1 integrin engagementhas been shown to activate both ERK 1 and ERK 2 inosteoblasts,17 this result further confirms the selective�v�3 recognition of N15-PGRDGS. A recent study re-

Figure 6. Western blot analysis of ERK1/2 activation statewhen MC3T3-E1 cells are plated for 1 h onto N15 fusionpeptide-coated poly-D-lysine plates. ERK1/2 was found tobe at a constant level in all cells irrespective of whether theywere maintained in suspension or plated onto different sub-strates. However, only MC3T3-E1 cells plated onto N15-PDGEA showed the active form (tyrosine and threoninephosphorylated) of ERK 1 and ERK 2.

Figure 5. NIH Image analysis of Western blot band inten-sities for the time course of FAK phosphorylation forMC3T3-E1 cells plated onto N15-PDGEA or N15-PGRGDS.Cells plated onto N15-PGRGDS showed an increase in FAKphosphorylation with a maximum level after 90 min ofplating, whereas cells plated onto N15-PDGEA reach a max-imum at 60 min. Values were calculated from band intensityusing NIH Image from a Western blot directed against phos-phorylated tyrosine for MC3T3-E1 cell lysates with the stan-dard deviation shown.

PEPTIDES THAT ENGAGE SIGNALING PATHWAYS 75

ported that phosphorylated ERK 1 and ERK 2 werepresent at basal levels in transfected MC3T3-E1 cellsoverexpressing the human �v�3 integrin,25 although itwas not determined whether both the tyrosine andthreonine residues of ERK1/2 were phosphorylated asnecessary for full enzymatic activity,26 and the cellswere not maintained in suspension before Westernblot analysis. There have been conflicting reports onwhich ERK species are activated on integrin engage-ment to collagen type I.16,17 These conflicting findingscould be due to the differences in antibody specifici-ties, cell source, or preparation. Another possible rea-son for the contradictory results could be differencesin collagen type I conformation on the plate surfaces.Full-length collagen type I can engage a variety ofintegrins, including �1�1,27 �2�1, and �3�1,28 and thusenable integrin cross-talk. The selectivity of the N15-PDGEA peptide for the �2�1 integrin may explain thedual phosphorylation of both ERK 1 and ERK 2, whichis similar to the response observed when either fibro-blasts or primary osteoblasts were bound to an anti-�1antibody.17,29

The selectivity of ERK activation by the N15-DGEApeptide versus N15-RGD, following the initial com-mon FAK phosphorylation event, is an interestingexample of how specific outside-in signaling eventscan be when induced by specific integrin engage-ments. FAK phosphorylation is not necessarily linkedto the MAPK pathway,4,30 and it is possible that thephosphorylation of FAK and ERK 1/2 on binding toN15-PDGEA were two distinct and separate pathwayactivation events specified by initial �2�1 engagement.The activity and specificity of these peptides mayprovide an interesting model system to further studyintegrin-mediated adhesion and subsequent signalingpathway activation, both with isolated peptides and incombination because they assemble on the basis of theN15 statherin sequence. These peptides may also findapplication as tissue engineering or biomaterial coat-ings, because the MAPK pathway has been linked tothe activation of Cbfa15 and the N15-PDGEA peptidemay thus trigger the further differentiation of osteo-blasts.

We gratefully acknowledge the support provided by theNational Institute of Dental Research, the National ScienceFoundation through the UWEB ERC, and The WhitakerFoundation (graduate fellowship to M.G.).

References

1. Davis GE. Affinity of integrins for damaged extracellular ma-trix: alpha v beta 3 binds to denatured collagen type I throughRGD sites. Biochem Biophys Res Commun 1992;182:1025–1031

2. Xiao G, Wang D, Benson MD, Karsenty G, Franceschi RT. Roleof the alpha2-integrin in osteoblast-specific gene expression

and activation of the Osf2 transcription factor. J Biol Chem1998;273:32988–32994.

3. Grzesik WJ. Integrins and bone—cell adhesion and beyond.Arch Immunol Ther Exp 1997;45:271–275.

4. Aplin AE, Howe A, Alahari SK, Juliano RL. Signal transduc-tion and signal modulation by cell adhesion receptors: the rollof integrins, cadherins, immunoglobulin-cell adhesion mole-cules, and selectins. Pharmacol Rev 1998;50:197–263.

5. Xiao G, Jiang D, Thomas P, Benson MD, Guan K, Karsenty G,Franceschi RT. MAPK pathways activate and phosphorylatethe osteoblast-specific transcription factor, Cbfa1. J Biol Chem2000;275:4453–4459.

6. Shaw WJ, Long JR, Dindot JL, Campbell AA, Stayton PS,Drobny GP. Determination of statherin N-terminal peptideconformation on hydroxyapatite crystals. J Am Chem Soc 2000;122:1709–1716.

7. Shaw WJ, Long JR, Campbell AA, Stayton PS, Drobny GP. Asolid state NMR study of dynamics in a hydrated salivarypeptide adsorbed to hydroxyapatite. J Am Chem Soc 2000;122:7118–7119.

8. Gilbert M, Shaw WJ, Long JR, Nelson K, Drobny GP, GiachelliCM, Stayton PS. Chimeric peptides of statherin and osteopon-tin that bind hydroxyapatite and mediate cell adhesion. J BiolChem 2000;275:16213–16218.

9. Sudo H, Kodama HA, Amagai Y, Yamamoto S, Kasai S. In vitrodifferentiation and calcification in a new clonal osteogenic cellline derived from newborn mouse calvaria. J Cell Biol 1983;96:191–198.

10. Moursi AM, Globus RK, Damsky CH. Interactions betweenintegrin receptors and fibronectin are required for calvarialosteoblast differentiation in vitro. J Cell Sci 1997;110:2187–2196.

11. Asplin JR, Arsenault D, Parks JH, Coe FL, Hoyer JR. Contri-bution of human uropontin to inhibition of calcium oxalatecrystallization. Kidney Int 1998;53:194–199.

12. Jiang XJ, Feng T, Chang LS, Kong XT, Wang G, Zhang ZW,Guo YL. Expression of osteopontin mRNA in normal andstone-forming rat kidney. Urol Res 1998;26:389–394.

13. Lieske JC, Toback FG. Interaction of urinary crystals with renalepithelial cells in the pathogenesis of nephrolithiasis. SeminNephrol 1996;16:458–473.

14. Wesson JA, Worcester EM, Wiessner JH, Mandel NS, KleinmanJG. Control of calcium oxalate crystal structure and cell adhe-sion by urinary macromolecules. Kidney Int 1996;53:952–957.

15. Staatz WD, Fok KF, Zutter MM, Adams SP, Rodriguez BA,Santoro SA. Identification of a tetrapeptide recognition se-quence for the alpha 2 beta 1 integrin in collagen. J Biol Chem1991;266:7363–7367.

16. Takeuchi Y, Suzawa M, Kikuchi T, Nishida E, Fujita T, Matsu-moto T. Differentiation and transforming growth factor-betareceptor down-regulation by collagen-alpha2beta1 integrin in-teraction is mediated by focal adhesion kinase and its down-stream signals in murine osteoblastic cells. J Biol Chem 1997;272:29309–29316.

17. Cowles EA, Brailey LL, Gronowicz GA. Integrin-mediated sig-naling regulates AP-1 transcription factors and proliferation inosteoblasts. J Biomed Mater Res 2000;52:725–737.

18. Krause A, Cowles EA, Gronowicz G. Integrin-mediated signal-ing in osteoblasts on titanium implant materials. J BiomedMater Res 2000;52:738–747.

19. Schlaepfer DD, Broome MA, Hunter T. Fibronectin-stimulatedsignaling from a focal adhesion kinase-c-Src complex: involve-ment of the Grb2, p130cas, and Nck adaptor proteins. Mol CellBiol 1997;17:1702–1713.

20. Liu YK, Uemura T, Nemoto A, Yabe T, Fujii N, Ushida T,Tateishi T. Osteopontin involvement in integrin-mediated cellsignaling and regulation of expression of alkaline phosphataseduring early differentiation of UMR cells. FEBS Lett 1997;420:112–116.

76 GILBERT ET AL.

21. Pytela R, Pierschbacher MD, Ruoslahti E. A 125/115-kDa cellsurface receptor specific for vitronectin interacts with the argi-nine-glycine-aspartic acid adhesion sequence derived from fi-bronectin. Proc Natl Acad Sci USA 1985;82:5766–5770.

22. Ruoslahti E. RGD and other recognition sequences for inte-grins. Annu Rev Cell Dev Biol 1996;12:697–715.

23. Yokosaki Y, Monis H, Chen J, Sheppard D. Differential effectsof the integrins alpha9beta1, alphavbeta3, and alphavbeta6 oncell proliferative responses to tenascin. Roles of the beta sub-unit extracellular and cytoplasmic domains. J Biol Chem 1996;271:24144–24150.

24. Akiyama SK, Yamada SS, Yamada KM, LaFlamme SE. Trans-membrane signal transduction by integrin cytoplasmic do-mains expressed in single-subunit chimeras. J Biol Chem 1994;269:15961–15964.

25. Cheng SL, Lai CF, Blystone SD, Avioli LV. Bone mineralizationand osteoblast differentiation are negatively modulated byintegrin alpha v beta 3. J Bone Miner Res 2001;16:277–288.

26. Payne DM, Rossomando AJ, Martino P, Erickson AK, Her JH,Shabanowitz J, Hunt DF, Weber MJ, Sturgill TW. Identificationof the regulatory phosphorylation sites in pp42/mitogen-acti-vated protein-kinase (MAP kinase). EMBO J 1991;10:885–892.

27. Roberts AI, Brolin RE, Ebert EC. Integrin alpha(1)beta(1)(VLA-1) mediates adhesion of activated intraepithelial lym-phocytes for collagen. Immunology 1999;97:679–685.

28. Lundstrom A, Holmbom J, Lindqvist C, Nordstrom T. The roleof alpha 2 beta 1 and alpha 3 beta 1 integrin receptors in theinitial anchoring of MDA-MB-231 human breast cancer cells tocortical bone matrix. Biochem Biophys Res Commun 1998;250:735–740.

29. Morino N, Mimura T, Hamasaki K, Tobe K, Ueki K, Kikuchi K,Takehara K, Kadowaki T, Yazaki Y, Nojima Y. Matrix/integrininteraction activates the mitogen-activated protein kinasep44erk-1 and p42erk-2. J Biol Chem 1995;270:269–273.

30. Boudreau NJ, Jones PL. Extracellular matrix and integrin sig-naling: the shape of things to come. Biochem J 1999;339:481–8.

PEPTIDES THAT ENGAGE SIGNALING PATHWAYS 77