Site-Directed Immobilization of BMP-2: Two Approaches for the

Article pubs.acs.org/Biomac

Site-Directed Immobilization of BMP-2: Two Approaches for the Production of Innovative Osteoinductive Scaffolds Barbara Tabisz,† Werner Schmitz,‡ Michael Schmitz,§ Tessa Luehmann,∥ Eva Heusler,∥ Jens-Christoph Rybak,∥ Lorenz Meinel,∥ Juliane E. Fiebig,⊥ Thomas D. Mueller,*,⊥ and Joachim Nickel*,†,# †

Lehrstuhl für Tissue Engineering und Regenerative Medizin, Universitätsklinikum Würzburg, Röntgenring 11, D-97070 Würzburg, Germany ‡ Lehrstuhl für Biochemie und Molekularbiologie, Theodor-Boveri-Institut für Biowissenschaften, and ∥Lehrstuhl für Pharmazeutische Technologie und Biopharmazie, Universität Würzburg, Am Hubland, D-97074 Würzburg, Germany § Lehrstuhl für Funktionswerkstoffe der Medizin und der Zahnheilkunde, Universitätsklinikum Würzburg, Pleicherwall 2, D-97070 Würzburg, Germany ⊥ Lehrstuhl für molekulare Pflanzenphysiologie und Biophysik, Julius-von-Sachs Institut für Biowissenschaften, Universität Würzburg, Julius-von-Sachs-Platz 2, D-97082 Würzburg, Germany # Translationszentrum Würzburg “Regenerative Therapien für Krebs- und Muskuloskelettale Erkrankungen”, Institutsteil Würzburg, Fraunhofer-Institut für Grenzflächen- und Bioverfahrenstechnik (IGB), Röntgenring 11, D-97070 Würzburg, Germany S Supporting Information *

ABSTRACT: The regenerative potential of bone is strongly impaired in pathological conditions, such as nonunion fractures. To support bone regeneration various scaffolds have been developed in the past, which have been functionalized with osteogenic growth factors such as bone morphogenetic proteins (BMPs). However, most of them required supra-physiological levels of these proteins leading to burst releases, thereby causing severe side effects. Site-specific, covalent coupling of BMP2 to implant materials might be an optimal strategy in order to overcome these problems. Therefore, we created a BMP-2 variant (BMP2-K3Plk) containing a noncanonical amino acid (propargyl-L-lysine) substitution introduced by genetic code expansion that allows for site-specific and covalent immobilization onto polymeric scaffold materials. To directly compare different coupling strategies, we also produced a BMP2 variant containing an additional cysteine residue (BMP2-A2C) allowing covalent coupling by thioether formation. The BMP2-K3Plk mutant was coupled to functionalized beads by a copper-catalyzed azide−alkyne cycloaddition (CuAAC) either directly or via a short biotin-PEG linker both with high specificity. After exposing the BMP-coated beads to C2C12 cells, ALP expression appeared locally restricted in close proximity to these beads, showing that both coupled BMP2 variants trigger cell differentiation. The advantage of our approach over non-sitedirected immobilization techniques is the ability to produce fully defined osteogenic surfaces, allowing for lower BMP2 loads and concomitant higher bioactivities, for example, due to controlled orientation toward BMP2 receptors. Such products might provide superior bone healing capabilities with potential safety advantages as of homogeneous product outcome.

■

INTRODUCTION Long bone fractures are reported to occur in the western world with an incidence rate of 300−400 cases per 100000 individuals per year.1 Most fractures spontaneously heal without surgical intervention within 20 weeks, but an incidence of nonunion fractures has been reported to range from 4−10%2 sustainably challenging the patient’s life quality. Hence, there is a strong demand to develop adequate treatment options for such nonunion fractures. Even though bone is the second most transplanted tissue, only exceeded by blood transfer,3 current treatment options for orthopedic deficiencies are rather limited. They predominantly involve either auto- or allo-transplantation of bone tissue, which both seems a suboptimal choice. An © XXXX American Chemical Society

autograft replacement is considered the golden standard, however, time-consuming surgeries and a second trauma site with potential complications such as residual pain as well as requirement for additional cosmetic procedures are disadvantages of this approach. Furthermore, as most osteogenic elements in the graft do not survive transplantation the reapplication of autograft transplants is limited. On the other hand using allograft replacement, regarded as the surgeon’s second choice, also bears risks such as transmission of viral Received: September 20, 2016 Revised: January 26, 2017

A

DOI: 10.1021/acs.biomac.6b01407 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

introduced through genetic codon expansion (BMP2K3Plk)13 for site-directed coupling via a highly specific bioorthogonal chemistry. The advantages of our approach compared to previous strategies are a highly controllable loading of the scaffold along with an inhibition of a release of BMP2 from the site of action. In this manuscript, we report the recombinant production of the two BMP-2 variants and present different coupling techniques to scaffolds while maintaining the ligand’s biological activity.

diseases or tumors. In addition, to limit immune responses allografts have to be processed, which might weaken their biological and mechanical properties. All these limitations are arguing for the development and application of innovative scaffolds with strong osteogenic properties. In the past, large efforts were taken to improve the osteogenic potential of these scaffolds, for example, by incorporation of osteogenic growth factors such as bone morphogenetic protein (BMP)2 or osteogenic protein (OP)1 (also known as BMP7), which received FDA approval for clinical use for nonunion tibia fractures and spinal fusions. An analysis of a large number of studies suggested that the osteogenic factor must be applied at rather high doses to or within the scaffold.4 As the osteogenic factors are continuously drained from their site of action, the currently available approaches of immobilization, such as encapsulation or adsorption to the scaffold, require supra-physiological amounts of the applied growth factor, resulting in unwanted initial burst release phenomena.5 In several clinical trials, up to 40 mg of BMP2 were used per implantation.4 Given the fact that physiological systemic concentrations of those osteoinductive factors are very low,6 the release of BMP2 from such scaffolds into extracellular fluids has severe disadvantages as it affects surrounding tissues, possibly leading to bone overgrowth, osteolysis, swelling, and inflammation.7 To circumvent this dilemma it is therefore necessary to implement either a method for a controlled slow release or, if possible, a complete immobilization, creating a functional activating surface.8,9 The logical best option would be, therefore, a covalent coupling of the growth factor to a biodegradable scaffold, which then ideally resists biodegradation as long as the bioactivity is needed. In most studies so far, a coupling chemistry has been used that relies on activated carboxylic acid groups on the scaffold that are subsequently reacted with primary amines, that is, those of lysine side chains of the protein.10,11 However, selection of the coupling lysine is a random process thus it is highly unlikely that only one particular lysine in the BMP2 dimer is solely used for coupling. Consequently, such an approach results in a scaffold in which BMP2 is presented in variable arrangements hence leading to an inhomogeneous product outcome. As some of these might inactivate BMP2 this not only leads to a scaffold with unknown activity but might also severely limit the reproducibility of a scaffold with defined activity. Furthermore, already coupling by only one of the 18 lysine residues could block one of the four receptor binding epitopes resulting in BMP2 molecules with significant lower bioactivities. Accordingly, studies using BMP2 heterodimers in which one receptor epitope is blunted have shown that all four receptors must bind to BMP2 for full activation of all signaling cascades.12 Furthermore, partial unfolding as a result of inhomogeneity may increase the safety risk as of anti-BMP2 antibody formation, thereby substantially challenging the efficacy of endogenously produced BMP2 and leading to substantial, chronic sequelae. To overcome the partial inactivation and the potentially increased risk of anti-BMP2 antibody formation due to random coupling, we generated BMP2 variants that allow for sitedirected coupling to the scaffold. In a first approach we introduced an additional unpaired cysteine residue at the Nterminus of BMP2 (BMP2-A2C), allowing targeted coupling to scaffolds using classical sulfhydryl-iodoacetamide coupling chemistry. As an improved strategy, we constructed a BMP2 variant that harbors a unique non-natural amino acid

■

MATERIALS AND METHODS

Cloning of BMP-2 Variants. The BMP-2 variant BMP2-A2C was generated by site-directed mutagenesis using the Rapid-PCR methodology from Costa and Weiner.14 For the variant BMP2-K3Plk, the codon (AAA) coding for Lys3 (amino acid at position 3 of the BMP2’s mature part) in the wild-type sequence was replaced by TAG (amber-, stop-codon) by PCR. The TAG stop codon terminating the mature part of the wild-type BMP2 sequenced was replaced by a TAA stop codon. The resulting sequence was subcloned into the NdeI/BamHI site of the bacterial expression vector pET11a-pyltRNA (kind gift of M. Rubini; Konstanz, Germany) using standard techniques. This plasmid also encodes the pyrrolysyl-tRNA with the anticodon being complementary to TAG. The resulting plasmid was cotransfected together with pRSFduet-pyrtRNAsynth encoding the corresponding aminoacyl-tRNA synthetase (kind gift of M. Rubini; Konstanz, Germany) into BL21(DE3) bacteria. All inserts encoding for the genes of interest were verified by DNA-sequencing. (Depiction of the location of the amino acid substitution, see Supporting Information, Figure 1A,B.) Expression and Purification of BMP2 Variants and BMP2 Receptor Ectodomains. Expression and purification of wild-type BMP2 was performed according to protocols published by Kirsch et al.15 BMP2-A2C was expressed and purified as published.16 From 7 g of inclusion bodies, approximately 9 mg of purified BMP2-A2C was recovered. For BMP2-K3Plk expression, BL21(DE3) bacteria were cotransfected with pET11a-pyrtRNA/BMP2-K3Plkplk and a plasmid encoding for the corresponding pyrrolysyl-tRNA synthetase (pRSFduet-pyltRNAsynth). An isolated single colony was propagated overnight in LB medium supplemented with 50 μg/mL of kanamycin and 100 μg/mL of ampicillin at 37 °C. Overnight LB cultures were transferred into TB medium (1:20) supplemented with antibiotics and grown until OD600 reached 0.7. Propargyl-L-lysine was synthesized according to published procedures.17 To establish the best concentration of propargyl-L-lysine, final concentrations of 1, 2, 3, 4, 5, 7, and 10 mM of propargyl-L-lysine were added. Cultures were induced by adding IPTG to a final concentration of 1 mM. Expression was carried out overnight at 37 °C. For the final large scale expression 10 mM Plk was added at OD 1.0 and cultures were induced with IPTG at OD 1.5. After 16 h of incubation, cultures were centrifuged, and approximately 40 g of the pellet was obtained. The pellet was subjected to four cycles of sonication (10 min with a 40 s pulse, 20 s break, and 30% amplitude) in STE buffer (10 mM Tris pH 8.0; 150 mM NaCl; 1 mM EDTA; 375 mM sucrose; freshly added 1/1000 vol/ vol of β-mercaptoethanol) and centrifugation in order to extract inclusion bodies. IBs were recovered as described.15 The final yield of refolded BMP2-K3Plk was approximately 22 mg. Expression and purification of the extracellular domain (EC) of bone morphogenetic protein receptor IA (BMPR-IAEC) and activin receptor IIB ActR-IIBEC was carried out as previously described.15,18 Biosensor Measurements. A ProteOn XPR36 biosensor system (Bio-Rad) was used for all surface plasmon resonance experiments. Measurements were performed at 25 °C using 10 mM HEPES pH 7.4, 500 mM NaCl, 3.4 mM EDTA, 0.005% (v/v) Tween-20 as running buffer, the flow rate for interaction data acquisition was set to 100 μL min−1. For the interaction analysis of BMP2 proteins with their B


Article

Biomacromolecules

Quantification of Immobilized BMP2-K3Plk and BMP2-A2C. The amount of BMP2-K3Plk immobilized to azide- and neutravidinactivated agarose beads and BMP2-A2C immobilized to iodoacetamide-activated acrylamide beads was determined in triplicates by the use of a BCA Protein Assay Reagent (Thermo Fisher Scientific) according to the manufacturer’s protocol. Neutravidin beads coupled to BMP2-K3Plk were boiled in a 2% SDS solution prior to measurement in order to neutralize biotin-neutravidin interaction. The amount of the released protein was analyzed as described above. Visualization of Immobilized BMP2-K3Plk and BMP2-A2C. Both immobilized BMP variants were visualized using an anti-BMP2 antibody (Abcam). BMP2-coupled beads were blocked with 5% milk in TBST for 1 h at 4 °C and incubated with the primary anti BMP2 antibody (ab 17885, Abcam) overnight at 4 °C. Beads were washed three times with TBST (0.1% Tween) and incubated with an Alexa Fluor 488-coupled secondary antibody (ab150061, Abcam). Fluorescence was analyzed by microscopy. To test the bioactivity of the immobilized BMP2 variants, the beads were alternatively exposed to the Texas Red labeled BMPR-IAEC (1 μM in HBS500 buffer). The beads were washed three times washed in the same buffer prior to use. Stability of BMP2-K3Plk Coupled to Azide-Functionalized Agarose Beads. The 2 × 105 beads were incubated over a 1 week period in 1 mL of medium (DMEM) at 37 °C. The beads were centrifuged (10 min, 1200g, RT), the supernatant recovered and the beads resuspended in 1 mL of fresh medium at days 1, 3, and 7. Soluble BMP2-K3Plk in the supernatant was analyzed by ELISA using the human BMP2 ELISA Kit from Peprotech (Peprotech, Rocky Hill, NJ, U.S.A.) according to the supplierś protocol. Beads were incubated for one further week and the immobilized BMP2-K3Plk visualized. Alkaline Phosphatase (ALP) Assay. The promyoblastic C2C12 cell line (ATCC CRL-172) was grown in DMEM containing 10% FCS, 100 U/mL penicillin G and 100 μg/mL streptomycin. The ALP assays were carried out in 96-well microplates as described previously.19 Alternatively, ALP expression was induced by the different BMP2 coupled beads. Which were added to the cells and subsequently overlaid with 0.4% low-melting-point agarose. After 3 days of incubation in differentiation medium (DMEM, 2% FCS) the medium was removed and ALP activity determined by addition of one-step NBT/BCIP substrate solution (Thermo Fisher Scientific). Protein Analyses. For qualitative and quantitative analyses proteins were subjected to SDS-PAGE. The proteins were visualized either by Coomassie staining or by Western blotting using a BMP-2 specific antibody (ABCAM, Cambridge, U.K.). In terms of BMP2-K3Plk which was coupled via CuAAC to a Biotin-functionalized linker a Streptavidin-Alexa Fluor 488 conjugate (Life Technologies) was used as detection reagent.

receptors, the receptor ectodomains were biotinylated using SulfoNHS-LC-biotin (Pierce, Thermo Fisher Scientific, Rockford, IL, U.S.A.) following the manufacturer’s protocol, but using a 1:1 (biotin/BMP2 monomer) stoichiometric ratio. A GLC sensor chip was first activated using EDC/NHS according to manufacturer’s recommendation, then streptavidin was perfused over the activated sensor surface at a concentration of 40 μg mL−1 until resonance unit (RU) levels reached 2000 to 2200 RU. (RU). The biotinylated receptor ectodomains were subsequently immobilized onto this streptavidin sensor surface at a density of approximately 500 RU. For a single kinetics measurement, six different analyte concentrations starting at 25 nM (log 2 dilution series) were used. The association time was set to 245 s, dissociation data were obtained from perfusion with running buffer for 180 s for BMP-RIAEC and 130 s for ActRIIBEC. Interaction data were acquired employing the single-shot kinetic setup with all six analyte concentrations measured simultaneously. After each BMP2 perfusion, the sensor chip was regenerated with three subsequent 60 s pulses injecting 100 mM glycine pH 2.5, 10 mM glycine pH 1.5, and 4 M MgCl2 at a flow rate of 100 μL min−1. To remove bulk face effects (buffer jumps, etc.) and unspecific binding to the chip matrix the interaction of the analyte to the unmodified streptavidin surface at the so-called interspots was subtracted from all binding data. Binding affinities were calculated by fitting the association and dissociation phase of the sensorgrams using a grouped regression analysis of the rate constants and employing 1:1 Langmuirtype interaction model. Chi2 values for data fitting are less than 10% of the maximal signal amplitude, standard deviation of the kinetic rate and equilibrium binding constants were derived from three independent experiments using six different analyte concentrations. Coupling Techniques. Coupling of BMP2-K3Plk to AzideFunctionalized Linker. BMP2-K3Plk (20 μM) was incubated with PEG4-carboxamide-6-azidohexanyl-biotin (Life Technologies; 200 μM) in a reaction buffer (0.1 M HEPES pH 7.0, 3.7 M urea, 50 μM CuSO4 (Alfa Aesar), 250 μM THPTA (Baseclick), and 5 mM Na ascorbate) for 1, 2, 5, 10, or 15 min at RT. The reaction was stopped by addition of EDTA to a final concentration of 5 mM. As negative control BMP2-K3Plk was incubated for 15 min in the same buffer with Millipore water instead of CuSO4 (see Supporting Information, Figure 2A,E). Coupling of BMP2-K3Plk to Neutravidin-Agarose Beads. A total of 260 μg of BMP2-Plk (20 μM) were incubated with 200 μM PEG4carboxamide-6-azidohexanyl-biotin (Life Technologies) as described above. For the negative control BMP2-wt was used. Biotinylated BMP2-K3Plk and the negative control were dialyzed against 1 mM HCl, concentrated and incubated for 1.5 h at RT in 8 mL of HBS500 buffer (10 mM HEPES pH 7.4, 500 mM NaCl) with 50 μL of Neutravidin Agarose beads (Thermo Fisher Scientific) equilibrated to HBS500 buffer After incubation the beads were thoroughly washed with HBS500 buffer, 4 M MgCl2, with PBS, and finally stored in PBS at 4 °C (see Supporting Information, Figure 2B,E). Coupling of BMP2-K3Plk to Azide-Functionalized Agarose Beads. A total of 260 μg of BMP2-K3Plk (20 μM) was incubated with 20 μL of azide-activated agarose beads (Jena Biosciences) in a reaction buffer (0.1 M HEPES pH 7.0, 3.9 M urea, 50 μM CuSO4, 250 μM THPTA, 5 mM Na ascorbate) for 2 h at RT. The reaction was stopped with 5 mM EDTA. Beads were thoroughly washed with HBS500 buffer, 4 M MgCl2, PBS, and finally stored in PBS at 4 °C (see Supporting Information, Figure 2C,E). Coupling of BMP2-A2C to Iodoacetamide-Activated Acrylamide Beads. A total of 1 mM BMP2-A2C in water was reduced with 2-fold molar excess of TCEP for 1.5 h at RT. Upon reduction BMP2-A2C was transferred to the reaction buffer (50 mM Tris pH 8.5, 5 mM EDTA, 6 M urea), 10 μL of UltraLink Iodoacetyl Resin beads (Thermo Fisher Scientific) being equilibrated to reaction buffer were added, and the mixture was incubated for 1 h at RT. The beads were centrifuged, washed 3 times with reaction buffer, and blocked by incubation in a 50 mM cysteine solution for 30 min at RT. The supernatant was discarded and beads were thoroughly washed with HBS500, 4 M MgCl2, and PBS, and stored in PBS (see Supporting Information, Figure 2D,E).

■

RESULTS Production of BMP2 Variants. Bacteria expressing BMP2A2C or BMP2-K3Plk were harvested 3 h post-induction by centrifugation and lysed. Whole bacterial lysates analyzed by SDS-PAGE (Figure 1A,B) show strong protein expression for BMP2-A2C (Figure 1A, lane 2) in contrast to BMP2-K3Plk, which was seemingly not expressed under these conditions (data not shown). As stop codon suppression might influence the bacterial translation machinery, the kinetics of protein expression were analyzed by varying incubation time as well as propargyl-L-lysine concentrations. The results indicated that BMP2-K3Plk expression rates are slower as compared to wildtype BMP2 or the variant BMP2-A2C (data not shown). To our surprise, while a seemingly high concentration of propargylL-lysine was used to ensure the formation of sufficient propargyl-L-lysine-loaded tRNA, the initially chosen concentration of 1 mM was too low to allow production of BMP2K3Plk protein at high yields. Therefore, we increased the propargyl-L-lysine concentration to 10 mM, which finally resulted in sufficiently high yields (Figure 1B). The identity C


Article

Biomacromolecules

SDS-PAGE analysis using nonreducing conditions to monitor the formation of dimeric BMP2 protein (Figure 2A,C). After 5 days dimer formation reached peak levels and the BMP2 proteins were purified by cation-exchange (IEX) chromatography. For the BMP2-K3Plk variant harboring the non-natural amino acid propargyl-L-lysine an overall yield of approximately 7 mg of dimeric BMP2-K3Plk per liter of bacterial culture could be obtained. Quality analysis by mass spectrometry provided a mass of 26102.45 Da corresponding to the theoretical molecular weight of a homodimeric BMP2-K3Plk protein with two incorporated propargyl-L-lysine residues (one per chain, respectively; Figure 2D). The BMP2 variant BMP2-A2C was refolded and purified similarly. However, SDS-PAGE analysis of the purified BMP2 protein revealed several bands indicating heterogeneous protein compositions (Figure 2. A). Albeit the band in the SDS-PAGE migrating at an apparent molecular weight of 26 kDa was stained most intense other weaker bands were found at 13 kDa - most-likely representing BMP2 A2C monomer - as well as bands migrating slower thereby indicating BMP2 protein assemblies with higher stoichiometry (e.g., trimers, tetramers, etc.). Generation of multimeric species is possibly due to the unpaired additional cysteine at the N-terminus of this variant, predisposing for the formation of non-native intermolecular disulfide arrangements and assembly of multimers with higher stoichiometry. Unfortunately separation of such multimers from natively folded dimers proved to be rather difficult when using IEX chromatography or reversed-phase HPLC (Figure 2B), both of which were employed as final purification schemes. Thus, the only purification option to remove higher molecular weight species is gel filtration, which was however not employed here, as the use of reducing agents in the later coupling procedures potentially lead to the formation of new multimers again. The overall yield of produced dimeric BMP2A2C was thus estimated to 2 mg per 1 L of bacterial culture. Biological Activity of BMP2-K3Plk and BMP2-A2C. The biological activity of both BMP2 variants in their noncoupled form was assessed by cell-based assays. Upon BMP2 stimulation C2C12 cells, a murine myoblastic cell line (ATCC, CRL 1772), differentiate toward osteoblastic lineages,20 which is monitored by induction of alkaline phosphatase (ALP) gene expression, a marker gene of osteogenesis. Both BMP2-K3Plk as well as BMP2-A2C induced ALP expression in a dose-dependent manner. In case of BMP2-K3Plk, induction of ALP expression was identical to that of wild-type BMP2 (Figure 3B1). Only at high BMP2-K3Plk concentrations an unusual decrease in activity was observed that most likely is due to a limited solubility of the variant protein. BMP2 is badly soluble under physiological conditions. In the past, several BMP2 mutants were designed by our group, some of which showed better but others even worse solubility (unpublished observations). At high concentrations, these proteins tend to aggregate and to precipitate, resulting in a decrease in the effective BMP2 concentration. The calculated EC50 values of approximately 30 nM for wild-type BMP2 and approximately 20 nM for the BMP2-K3Plk variant (Figure 3B1) are in agreement with those already published.19 In contrast, for BMP2-A2C an EC50 value of approximately 110 nM was determined. Also, para-nitrophenylphoshate conversion was found significantly lower (0.24 nmol/min for BMP2-A2C vs 1.22 nmol/min for wild-type BMP2 (Figure 3B2)). Both observations, the reduced maximal ALP-level as well as the higher EC50 value, are related to the inactive or only partially

Figure 1. Expression of BMP2-A2C and BMP2-K3Plk variants: (A) BMP2-A2C, lane 1, before; land 2, after induction with IPTG; (B) BMP2-K3Plk, lane 1, BMP2-wt; lane 2, BMP2-Plk/+IPTG; lane 3, 1 mM Plk; lane 4, 2 mM Plk; lane 5, 3 mM Plk; lane 6, 4 mM Plk; lane 7, 5 mM Plk; lane 8, 7 mM Plk; lane 9, 10 mM Plk; lane 10, marker; (C) Western blot analyses, lane 1, BMP2-wt induced; lane 2, BMP2K3Plk-Plk/induced; lane 3, BMP2-K3Plk + Plk/induced; (D) BMP2K3Plk large scale expression, lane 1, BMP2-wt; lane 2, BMP2-K3Plk noninduced; lane 3, BMP2-K3Plk + Plk/noninduced; lane 4, BMP2K3Plk + Plk/induced; lane 5, BMP2-wt noninduced; lane 6, BMP2-wt induced; lane 7, marker; (E) MS spectrum of monomeric BMP2K3Plk (X-axis 900−2100). The m/z ratio of 9-fold charged ions equals 1451.844. This corresponds to the mass of 13058.596 Da (1451.844*9−8 = 13058.596), which is in agreement with the theoretically calculated mass for BMP2 containing one propargyl-Llysine substitution.

of the BMP2-K3Plk was confirmed by Western blot analyses using an anti-BMP2 antibody (Figure 1C). However, as these high concentrations of propargyl-L-lysine could lead to incorporation at other lysine positions within the BMP2 sequence from unspecific loading of native tRNAs by the coexpressed pyrrolysyl-tRNA synthetase, we extracted BMP2K3Plk protein from the inclusion bodies, in which BMP2 proteins typically accumulate upon expression. The extracted protein was dialyzed against water and soluble monomeric BMP2-K3Plk from the supernatant was subsequently analyzed by mass spectrometry (Figure 1E), yielding a single peak with a corresponding mass of 13058.596 Da (Figure 1E). This mass is consistent with the theoretical molecular weight of a BMP2K3Plk monomer containing a single propargyl-L-lysine residue. Refolding and Purification of BMP2-A2C and BMP2K3Plk. Both BMP2 variants were refolded to yield mature dimeric BMP2 proteins. Refolding progress was followed by D


Article

Biomacromolecules

Figure 2. Refolding of BMP2-K3Plk and BMP2-A2C: (A) BMP2-A2C, lane 1, BMP2-A2C, reducing conditions; lane 2, BMP2-A2C, nonreducing conditions; lane 3, marker; (B) BMP2-A2C purified byRP-HPLC, lane 1, BMP2-A2C purified with IEX; lanes 2−9, BMP2-A2C, RP-HPLC fractions; lane 10, marker; (C) BMP2-K3Plk after refolding, lane 1, purified BMP2-wt, reducing conditions; lane 2, purified BMP2-wt, nonreducing conditions; lane 3, refolded BMP2-K3Plk, nonreducing conditions; lane 4, marker; (D) Mass spectrometry, deconvoluted spectrum of dimeric BMP2-K3Plk. The mass of 26102.453 Da is in agreement with the theoretically calculated mass of the BMP2 dimer containing one propargyl-L-lysine substitution within each monomer; (E) Mass spectrometry, deconvoluted spectrum of dimeric BMP2-A2C. The mass of 26613.566 Da is in agreement with theoretically calculated mass for BMP2-A2C dimer with one glutathione group coupled to the terminal cysteine within each monomer.

noncoupled product (Figure 4B). Nevertheless, since only the biotinylated product will couple onto neutravidin-coated beads minor traces of nonbiotinylated BMP2-K3Plk could be neglected. The biological activity of the biotinylated BMP2 protein was subsequently tested by cell-based assays employing C2C12 cells (Figure 4.C). Similar activities (EC50 values) were observed for the nonreacted (approximately 24 nM) and the biotinylated BMP2-K3Plk (approximately 39 nM), respectively. Immobilization of BMP2-K3Plk and BMP2-A2C onto Polymer Beads. All BMP2 variants, BMP2-K3Plk, biotinylated BMP2-K3Plk, and BMP2-A2C, were coupled to commercially available polymer beads with nonmodified wildtype BMP2 serving as control. Neutravidin-coated beads were chosen to couple the biotinylated BMP2-K3Plk variant. After incubation, the beads were washed intensively with 4 M MgCl2 to remove BMP2 that was only physically adsorbed to the polymer surface. BMP coupling was then confirmed by immunofluorescence using an anti-BMP-2 antibody. Nontreated neutravidin beads and neutravidin beads incubated with wild-type BMP2 did not show any fluorescence staining above background (Figure 5B1,B2). In contrast, direct coupling of BMP2-K3Plk to azide-functionalized polymer beads via CuAAC chemistry resulted in a strong fluorescence (Figure 5, A3). While nontreated beads did not show any fluorescence above background (Figure 5A1), similar as found for the neutravidin-coated agarose beads, a weak fluorescence signal was however observed when azide-functionalized beads were incubated with wild-type BMP2 (Figure 5A2). In order to directly compare the coupling characteristics of both BMP2 variants, we next tested the direct coupling of BMP2-A2C via reactive thiol groups to iodoacetamide-functionalized poly-

active multimers being present in BMP2-A2C. To analyze the receptor binding capabilities of both BMP2 variants, we performed in vitro binding studies using surface plasmon resonance spectroscopy (SPR). Binding to the extracellular (EC) domains of the BMP type IA receptor (BMPRI-AEC) revealed apparent KD values of approximately 0.7, 0.6, and 0.8 nM for wild-type BMP2, BMP2-A2C, and BMP2-K3Plk, respectively. Binding of the ligands to the EC domains of activin type IIB receptor (ActR-IIBEC) showed apparent KD values of approximately 0.9, 1.7, and 0.8 nM for wild-type BMP2, BMP2-A2C, and BMP2-K3Plk, respectively (Figure 3A). Site-Specific Biotinylation of BMP2-K3Plk. We then tested different conditions for the immobilization to polymeric beads. However, as protein coupling to macroscopic structures is difficult to assess, we first analyzed the coupling of BMP2K3Plk to a short PEG4-carboxamide-6-azidohexanyl-biotin linker (Figure 4A). The reaction was rapid and the biotinylated product was detected already after 1 min of incubation. As control, the reaction was carried out in absence of copper ions (Figure 4A, lane 6). Maximal biotinylation was observed after 15 min. However, prolonged incubation times seemed to affect the architecture of BMP-K3Plk since SDS-PAGE analysis revealed a band with decreased electrophoretic mobility, which possibly indicated the formation of a tetramer as a result of increasing reaction times (Figure 4A, see lanes 4 and 5). For quantification we tried to separate the biotinylated from nonbiotinylated BMP2-K3Plk by RP-HPLC. However, nonreacted and biotinylated BMP2-K3Plk protein eluted with identical retention times when using a C4 column thereby impeding separation and quantification of the coupled and E


Article

Biomacromolecules

Figure 3. Receptor binding and activity of BMP2-K3Plk and BMP2-A2C. (A) Interaction of BMP2-wt and BMP2-K3Plk with BMPRIA and ActRIIB receptor ectodomains measured by surface plasmon resonance. Kinetic rate and equilibrium binding constants were derived from three independent experiments using at least five different analyte concentrations. The binding parameters kon, koff, and KD thus represent mean values and standard deviations of 18 individual measurements (N = 18); (B1) bioactivity of BMP2-K3Plk. Data represent mean values and standard deviations of four individual experiments (N = 4); (B2) bioactivity of BMP2-A2C in comparison to wild-type BMP2. Data represent mean values and standard deviations of two individual experiments (N = 2).

acrylamide beads. While nonreacted beads did not produce fluorescence signals (Figure 5C1), coupling of wild-type BMP2,

which had been activated with TCEP beforehand, resulted in signals even stronger when compared to beads reacted with F


Article

Biomacromolecules

Figure 4. Site-specific biotinylation of BMP2-K3Plk. (A) Site-specific biotinylation of BMP2-K3Plk: optimization of CuAAC reaction at RT, lane 1, 1 min; lane 2, 2 min; lane 3, 5 min; lane 4, 10 min; lane 5, 15 min; lane 6, negative control (without CuSO4); (B) HPLC separation of CuAACbiotinylated BMP2-K3Plk; (C) Bioactivity of CuAAC-biotinylated BMP2-K3Plk (ALP assay). Data represent mean values and standard deviations of four individual experiments (N = 4).

does not provide an adequate proof that the immobilized BMP2 is still able to bind to its cognate receptors or, even more, evoking a BMP2-specific response in cells. Therefore, to test the structural integrity of the immobilized BMP2 proteins we determined the receptor binding capacity employing fluorescently labeled receptor ectodomain protein of the type I receptor BMPR-IA. Beads coated with BMP2-K3Plk either directly via CuAAC chemistry or indirectly via a bivalent linker yielded fluorescence when incubated with the dye-labeled BMPR-IAEC (Figure 6A3,B3). Noncoated beads or beads that were incubated with wild-type BMP2 showed no fluorescence signal above background. Iodoacetamide-polyacrylamide beads that were reacted with BMP2-A2C also produced fluorescent signals (see Figure 6C3). However, beads that were reacted under the same conditions with wild-type BMP2 similarly showed fluorescence (see Figure 6C2) raising the question whether the coupling of BMP2-A2C would have solely occurred via the unpaired cysteine residue introduced in the

BMP2-A2C (compare Figure 5C2 and C3). Since after coupling the beads were washed intensively, thus, applying the same harsh conditions as mentioned before, a nonspecific, noncovalent adsorption of wild-type BMP2 or BMP2-A2C to the beads surface is unlikely. Instead, reactive thiol groups must have been produced, also within the wild-type BMP2 protein, which then allowed also covalent coupling of wild-type BMP2. The weaker fluorescence signal observed for BMP2-A2C coupled to beads might be explained with the multimeric protein (re)arrangements found in BMP2-A2C, which might impair the coupling to the beads or the interaction with the BMP2 antibody used for detection (compare Figure 5C2 and C3). Binding of Immobilized BMP2 Variants to BMPRI-AEC. As shown in Figure 5, coating of beads with both BMP variants was confirmed by immuno-fluorescence staining using an antiBMP2 antibody. However, as the antibody was raised to recognize only a small linear epitope within BMP2, the staining G


Article

Biomacromolecules

Figure 5. Immobilization of BMP2-wt, BMP2-K3Plk, and BMP2-A2C. Representative pictures from three individual experiments: (A1) noncoupled azide-agarose beads; (A2) azide-agarose beads incubated with BMP2-wt in the CuAAC reaction buffer; (A3) azide-agarose beads coupled to BMP2K3Plk; (B1) noncoupled neutravidin agarose beads; (B2) neutravidin agarose beads incubated with BMP2-wt in the CuAAC reaction buffer; (B3) neutravidin agarose beads coupled to BMP2-K3Plk; (C1) noncoupled iodoacetamide-acrylamide beads; (C2) iodoacetamide-acrylamide beads coupled to BMP2-wt; (C3) iodoacetamide-acrylamide beads coupled to BMP2-A2C.

comparison of the coupling targets with the negative controls. However, coupling of BMP2-K3Plk was highly specific regardless of whether immobilization was performed directly or an azide-functionalized linker moiety was used. The high specificity is evident when comparing coupling of BMP2-K3Plk to that of wild-type BMP2 serving as negative control. Direct coupling (without linker) showed very low levels of binding for wild-type BMP2 (approximately 35 pg/bead), whereas BMP2K3Plk was bound at approximately 378 pg/bead (Figure 7A). The stability of the ligand which was immobilized to these beads was tested by a standard release assay. From 2 × 105 beads 0.12 μg (0.6 pg per bead) immunoreactive BMP2 was released and found in the supernatant after 1 week as determined by ELISA (Supporting Information, Figure 3A) corresponding to approximately one per mill of the coupled molecules. Staining of the beads with a fluorescent BMP2 antibody after one further week of incubation in media still showed significant fluorescence (Supporting Information,

variant A2C. Thus, potentially also another cysteine residue, for example, one within the cystine-knot, might have been activated upon TCEP treatment to facilitate cross-linking outlining limitations not only about the homogeneity of the coupled BMP2-A2C protein, but also highlights the difficulty to assess the bioactivity of the decorated macromolecular scaffold material. Quantification of Immobilized BMP2 Variants. The use of protein-activated polymer scaffolds requires careful quantitative determination of the bioactive compound on or within the scaffold as well as knowledge about the bioactivity of the scaffold material employed when used in later therapeutic setups. A semiquantitative comparison based on the binding of an anti-BMP2 antibody or the fluorescently labeled BMPR-IAEC already showed that the diverse BMP2 proteins were coupled to the beads to similar extents despite using different coupling chemistries (Figure 7). However, the various chemistries resulted in different coupling specificities as evident from a H


Article

Biomacromolecules

Figure 6. Interaction of immobilized BMP2-wt, BMP2-K3Plk and BMP2-A2C with Texas-Red-labeled BMPR-IAEC. Representative pictures from three individual experiments.

(see Figure 6A3, B3, and C3). To finally test whether these biomaterials, that is, BMP proteins coupled to macroscopic structures, also exhibit bioactivity in a biological context, alkaline phosphatase expression in C2C12 cells was analyzed by a modified ALP assay. Since the macroscopic polymer beads did not release BMP2, we adapted the protocol by incubating the cells in 96-well plates for 3 days in close contact to the BMP2-coated beads. Movement of beads was impeded by embedding them in low-melting agarose. After gelling, differentiation medium was added, and the cells were incubated for 3 days. ALP expression was subsequently determined by staining with NBT (nitrotetrazoliumblue chloride) and BCIP (5-bromo-4-chloro-3′-indolyl phosphate p-toluidine salt). If the covalently coupled BMP2 is capable to form the active signaling ligand receptor complex ALP staining should occur in those cells that are in contact with the beads since the signaling initiation requires the physical interaction of the immobilized ligand with the receptors on the cell surface. We have observed similar patterns in analogous studies using fibroblast growth factor (FGF)2-K8Plk17 that together with interleukin (IL-4)21 represent the two growth factors which aside of BMP2-K3Plk

Figure 3B) An even slightly higher specificity was obtained when the protein was coupled after it had been attached first to a biotin-linker (wild-type BMP2:37pg/bead; BMP2-K3Plk: 502 pg/bead) (Figure 7B). In case of iodoacetamide-functionalized beads and coupling via thiol groups, cross-linking resulted in a higher coating density, but the specificity of coupling was much lower (approximately 1.23 ng/bead for wild-type BMP2 and 2.52 ng/bead for BMP2-A2C (Figure 7C). Although wild-type BMP2 is devoid of unpaired cysteine residues, as introduced in BMP2-A2C, it could nevertheless be coupled efficiently to the functionalized beads, most likely as the activation reaction provides for reducing conditions being strong enough to break up disulfide bonds within wild-type BMP2. Induction of ALP Expression Induced by BMP2-K3Plkor BMP2-A2C-Coupled Beads. The major goal of our study was the design and production of a scaffold with immobilized BMP2 that should exhibit strong osteogenic properties. So far we could not only show the successful coupling of two different BMP2 variants using different chemistries, but also highlight that both BMP2 variants maintain their receptor binding properties toward the BMP type I receptor BMPR-IA in vitro I


Article

Biomacromolecules

Figure 7. Quantification of immobilized BMP2-K3Plk, BMP2-wt, and BMP2-A2C: (A) azide-agarose beads; (B) neutravidin-agarose beads; (C) iodoacetamid-functionalized acrylamide beads. The values represent mean values of three independent measurements. Absolute values determined for uncoupled beads were subtracted. Data represent mean values and standard deviations of three individual experiments (N = 3).

In summary, the different coupling strategies employed in this study allowed for site-directed coupling of BMP2 to macromolecular scaffold structures while maintaining its biological activity. The covalent attachment to the scaffold not only clearly limits off-target site effects as the ligand is incapable to diffuse from the appointed site of action, it also “sustains” the bioactivity of the factor. The internalization of the ligand into the cells by endocytosis and its subsequent degradation is impeded, thus, keeping extracellular ligand concentrations over time. However, this leaves the question how long the bioactivity of the biomaterial is retained and whether (and how) this time can be controlled or manipulated to provide osteogenic activity for the desired time period.

have been coupled to solid surfaces using this technique so far. BMP2-A2C coupled to iodoactetamide-acrylamide beads revealed strong ALP staining (Figure 8A3). However, also beads coupled with wild-type BMP2-induced ALP expression in surrounding C2C12 cells (Figure 8A2). This is consistent with our observation that the conditions used for the activation also yielded free thiol groups in wild-type BMP2, which can then also be coupled to the iodoacetamide-functionalized beads (compare to Figure 5C2,C3 or Figure 6C2,C3). This is surprising at first sight, as the coupling of wild-type BMP2 via thiol groups requires that at least one native disulfide bond, most likely one of the cystine-knot motif, were opened. As these disulfide bonds are assumed to be structurally relevant, a complete or at least significant loss of bioactivity was expected. However, the experiments employing beads coated with wildtype BMP2 suggest that even BMP2 coupled under these conditions retains (at least partially) its bioactivity. Coupling of BMP2-K3Plk employing CuAAC chemistry, either directly to azide-functionalized beads or via bivalent, biotin-containing linkers to neutravidin-functionalized beads showed a clear coupling specificity. ALP staining was only observed in cells, which were in direct contact with BMP2K3Plk functionalized beads (Figure 9A3,B3), but not in those which were in contact to beads reacted with wild-type BMP2 (Figure 9A2,B2). In general, the lack of ALP-mediated staining in distances around all tested beads exceeding a single cell layer confirms, that the BMP2 proteins were covalently linked and not just adsorbed. As control, soluble BMP2 was added in one well to mimic a freely diffusible ligand protein as expected for BMP2 protein that would be released from the beads (see Figure 8A4 and Figure 9A4). In these cases, a broad uniform staining of the cells was observed.

■

DISCUSSION Despite continuous research, transplantation of autografts is still considered the golden standard for treating critical size bone defects. In the past, osteomimetic materials have been developed containing recombinant growth factors adsorbed to or embedded within a carrier. One such scaffold, that is, INFUSE Bone Graft, was approved for the treatment of nonunion fractures by the Food and Drug Administration (FDA) back in 2002. It is based on a collagen to which recombinant BMP2 was adsorbed before implantation, which due to its osteogenic properties confers superior bone forming properties to the scaffold. These grafts were believed to become the new standard in bone surgery, but serious complications after implanting INFUSE bone grafts into cervical spine, such as swelling of the neck resulting in difficulties in breathing, speaking, and dysphagia were reported.22 Although the causes of these dangerous side effects are still debated, they are potentially linked to the physical and biochemical properties of this growth factor class. J


Article

Biomacromolecules

Figure 8. Induction of ALP expression inC2C12 cells byBMP2-A2C coupled beads. Representative pictures from three individual experiments. Alkaline phosphatase (ALP) staining upon treatment with iodoacetamide-acrylamide beads coupled to BMP2-A2C or BMP2-wt: (A1) uncoupled beads; (A2) beads coupled to wild-type BMP2; (A3) beads coupled to BMP2-A2C; (A4) 25 nM of soluble BMP2-wt.

reduce the required growth factor doses, several strategies focused on the generation of systems providing a sustained release of the growth factor, for instance, by adsorption to scaffolds or by encapsulation into hydrogels.25−29 Other investigations utilized genetically engineered BMP2 variants with enhanced binding affinities to components of the extracellular matrix such as heparin or heparan sulfates to anchor the protein at the site of implantation.30 The results of all these experiments clearly showed that the osteogenic activity of the applied BMP2 can be increased by generating longer retention times of the ligand. Consequently, an ideal osteogenic scaffold should provide an efficient dose of the growth factor being covalently immobilized on its surface without compromising its bioactivity. Thereby, the activity lifetime would only depend on the biological half-life time of the underlying scaffold or the surface degradation of the decorated BMP2 used in the particular setup. Unfortunately, studies in that direction so far utilized immobilization chemistries for BMPs that relied on a random coupling for instance by using the side chain amine group of lysines, which were cross-linked to NHSactivated groups on the scaffold. While these scaffolds showed osteogenic activity, the BMP2 protein on these surfaces is heterogeneously linked and displays variable activities;

Like all members of the TGF-β superfamily BMPs signal via oligomerization of two types of transmembrane receptors termed type I and type II that, based on the dimeric nature of BMPs, provides two binding epitopes for each receptor subtype, consequently, leading to the formation of heterohexameric ligand−receptor assemblies. Despite high sensitivity of this receptor assembly, full signaling activity is achieved already at subnanomolar BMP concentrations (approximately 100−300 pM), and some cellular responses such as the differentiation of particular cell types to, for example, osteoblastic lineages require significant higher concentrations (5−40 nM) of the ligand over longer periods of time.23 Therefore, we assume that also in vivo the growth factor is required at the site of action at sustained high concentrations to initiate and support the bone remodeling process. Injected soluble BMPs are, however, rapidly cleared from the body with a half-lifetime of t1/2 = 16 min in the rat and t1/2 = 6.7 min in nonhuman primates.7,24 Thus, to avoid rapid dilution of the BMP protein provided with the scaffold mechanisms must be implemented to facilitate a slow release. In the past, soluble recombinant BMPs were used in several clinical trials that reported applied BMP2 doses in the milligram range in order to achieve efficient bone growth.4 In order to K


Article

Biomacromolecules

Figure 9. Induction of ALP expression in C2C12 cells by BMP2-K3Plk coupled beads. Representative pictures from three individual experiments. Alkaline phosphatase (ALP) staining upon treatment with beads coupled to BMP2-K3Plk or BMP2-wt: (A1) uncoupled neutravidin-agarose beads; (A2) neutravidin-agarose beads incubated with wild-type BMP2; (A3) neutravidin-agarose beads coupled to BMP2-K3Plk; (A4) 25 nM of soluble BMP2-wt; (B1) uncoupled azide-agarose beads; (B2) azide-agarose beads incubated with wild-type BMP2 in the CuAAC reaction buffer; (B3) azideagarose beads coupled to BMP2-K3Plk.

furthermore, reproducibility as well as producing defined BMP activities on the surface of the scaffold is difficult. In our work, we established two strategies to couple BMP2 to macroscopic structures in a site-specific and covalent manner. One approach utilized a BMP2 variant with an additional cysteine residue introduced at the ligand’s Nterminus, thereby allowing for coupling by thiol-chemistry.

The second approach employed the non-natural amino acid propargyl-L-lysine, which was introduced in BMP2 instead of lysine 3 by bacterial protein expression using codon usage expansion.31 Both variants were analyzed for their biophysical and biochemical properties. We showed that the mutation introduced into variant BMP2-K3Plk did neither alter the receptor binding properties nor abrogated its bioactivity L


Article

Biomacromolecules

occurs in a 1:1 molar stoichiometry. We therefore assume that the density of the immobilized BMP2, which has been achieved in this work, might be considerably reduced while still remaining the osteogenic properties of the beads. Although the strategy to covalently immobilize BMPs onto a scaffold seems an appropriate answer to waive the application of growth factor at high doses ultimately leading to diffusion to off-target sites it bears an inherent risk of failure. More recent studies suggest that activation of receptors might often not occur at the cell surface but require endocytosis of the ligand/ ligand−receptor complex, which then starts signal transduction from endosomal compartments. It is known that BMP2 can indeed be internalized together with the receptors upon receptor assembly in a clathrin- or caveolin-mediated endocytosis33 and there has been an ongoing debate whether BMP2 must be internalized into the cell or whether the interaction with the assembly of its receptors at the cell surface is sufficient to initiate full BMP signaling.16 Some cell-based studies however demonstrated that BMP2 covalently immobilized on macroscopic, nonendocytosable surfaces is indeed still able to induce osteoblast differentiation.10 This strongly indicates that internalization of BMP2 is not strictly required for BMP2 signaling, but might rather serve as an additional regulatory step.33 Therefore, we tested also whether our macroscopic BMP2-coated scaffolds can induce BMP-specific signals in cell based assays which indeed resulted in the induction of ALP expression in the cells sharing direct contact with the BMP2-coated beads.

compared to those of wild-type BMP2. In contrast, the second variant BMP2-A2C showed decreased bioactivities, which most likely resulted from the presence of non-native protein structures (trimers, tetramers, etc.) formed from the additional cysteines located at the N-terminus of this variant. Coupling of the two BMP2 variants required different coupling chemistries. For the variant BMP2-A2C mass spectrometry analyses showed that the thiol group of the additional cysteine is protected by a glutathione moiety that is linked during the refolding process. Hence, the thiol group needs to be activated for coupling, which is achieved by a mild reduction of the glutathionylated BMP2-A2C. It has been previously reported that, after reduction and a subsequent coupling of a maleimide-activated fluorescent dye, only the Nterminal cysteine residue was activated resulting in a BMP2 dimer protein, in which both N-terminal cysteine residues (Cys2) were coupled to the fluorescent dye.16 Here, BMP2A2C was similarly activated prior to dye coupling, as in our study reported here, but protein multimers that were present from the refolding as well as unwanted byproducts that were formed due to unspecific reduction of cysteine residues other than the unpaired cysteine at position 2 could be removed in a subsequent purification step. Unfortunately, as already mentioned the coupling to solid surfaces as anticipated for the production of osteogenic scaffolds does not allow such subsequent purification and thus non-native protein species will potentially contaminate the biomaterial with BMP2 assemblies of unknown activity. This issue is solved with our second approach utilizing a nonnatural amino acid, which enables a site-directed, chemically highly specific coupling reaction eliminating or strongly reducing the formation of site products. For the coupling of BMP2-K3Plk to polymer beads, we tested two strategies. In the first approach, BMP2-K3Plk was first site-specifically linked to a bivalent azide-biotin functionalized linker, which was then coupled to neutravidin-conjugated agarose beads. Neutravidin binds to biotin with high affinity (KD = 10−15 M), which is for most biological applications equivalent to a covalent linkage.32 In an alternative approach, BMP2-K3Plk was directly coupled to azide-functionalized agarose beads by click chemistry. Both strategies showed similar coupling efficiencies (see Figure 7). In contrast to the thiol-coupling strategy, only very weak staining was observed when wild-type BMP2 protein was reacted with the beads using CuAAC, indicating that coupling occurs highly specific via the propargyl-L-lysin residue introduced into BMP2K3Plk. Thus, a high positional coupling specificity is a great improvement compared to the immobilization via classical NHS/EDC chemistry often used so far. To get an impression on the BMP2-K3Plk density on the bead surface the known bead size and the average BMP2-K3Plk amount per bead were used for a rough estimation. The average surface (average 100 μm in diameter) of the used beads is approximately 31500 μm2. The amount of coupled BMP2K3Plk was determined at approximately 0.4 ng (∼8.8 × 109 BMP2-K3Plk molecules) per bead, which results in a density of ∼280000 molecules/μm2. It has been demonstrated previously that ∼50000 to 200000 BMP-receptor complexes exist on the surface of C2C12 cells.23 Estimating that only 1% of the bead surface and 10% of the receptors on the cell membrane would come in contact with each other, ∼9 × 107 BMP2-K3Plk molecules would be able to contact and activate at least 5000 BMP-receptors, which approximately represents a 18000-fold molar excess assuming that the ligand receptor interaction

■

CONCLUSION The two BMP2 variants together with the coupling strategies presented in this study highlight a great improvement in the generation of novel osteogenic scaffolds over the currently known biomaterials that either release BMP2 in an uncompletely controllable manner or covalently, nonsitedirected immobilized BMP2. These new tools now allow coating of surfaces and scaffolds with BMP2 in a highly sitespecific manner leading to biomaterials with defined protein species attached and exhibiting a defined and selectable activity. Certainly, the functionalized scaffolds have to be tested in animal experiments in order to get a clear answer whether the covalently bound protein is also functional in vivo. However, since the introduced chemistry, especially that in BMP2-K3Plk, can serve as platform technology also coupling via linkers being cleavable, for example, by matrix metalloproteinases (MMPs) to newly designed scaffolds is feasible. The resulting product(s) can be analyzed and compared to the huge data set already obtained from the numerous clinical trials with “soluble” BMP2, which might lead to a fast return of this powerful agent into the clinic.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biomac.6b01407. Supplemental Figure 1: Depiction of the BMP2 variants with the localization of the introduced amino acid substitutions. Supplemental Figure 2: Coupling schemes. Supplemental Figure 3: Stability testing of BMP2-K3Plk coupled azide agarose beads (PDF). M


Article

Biomacromolecules

■

(16) Paarmann, P.; Dorpholz, G.; Fiebig, J.; Amsalem, A. R.; Ehrlich, M.; Henis, Y. I.; Muller, T.; Knaus, P. Dynamin-dependent endocytosis of Bone Morphogenetic Protein2 (BMP2) and its receptors is dispensable for the initiation of Smad signaling. Int. J. Biochem. Cell Biol. 2016, 76, 51−63. (17) Lühmann, T.; Jones, G.; Gutmann, M.; Rybak, J.-C.; Nickel, J.; Rubini, M.; Meinel, L. Bio-orthogonal Immobilization of Fibroblast Growth Factor 2 for Spatial Controlled Cell Proliferation. ACS Biomater. Sci. Eng. 2015, 1 (9), 740−6. (18) Weber, D.; Kotzsch, A.; Nickel, J.; Harth, S.; Seher, A.; Mueller, U.; Sebald, W.; Mueller, T. D. A silent H-bond can be mutationally activated for high-affinity interaction of BMP-2 and activin type IIB receptor. BMC Struct. Biol. 2007, 7, 6. (19) Kirsch, T.; Nickel, J.; Sebald, W. BMP-2 antagonists emerge from alterations in the low-affinity binding epitope for receptor BMPR-II. EMBO J. 2000, 19 (13), 3314−24. (20) Katagiri, T.; Yamaguchi, A.; Komaki, M.; Abe, E.; Takahashi, N.; Ikeda, T.; Rosen, V.; Wozney, J. M.; Fujisawa-Sehara, A.; Suda, T. Bone morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into the osteoblast lineage. J. Cell Biol. 1994, 127 (6 Pt 1), 1755−66. (21) Luhmann, T.; Spieler, V.; Werner, V.; Ludwig, M. G.; Fiebig, J.; Mueller, T. D.; Meinel, L. Interleukin-4-Clicked Surfaces Drive M2Macrophage Polarization. ChemBioChem 2016, 17 (22), 2123− 2128. (22) FDA Letter to Healthcare Practitioners, 2008; Center for Devices and Radiological Health Food and Drug Administration, http://www.fda.gov/medicaldevices/safety/alertsandnotices/ publichealthnotifications/ucm062000.htm. (23) Heinecke, K.; Seher, A.; Schmitz, W.; Mueller, T. D.; Sebald, W.; Nickel, J. Receptor oligomerization and beyond: a case study in bone morphogenetic proteins. BMC Biol. 2009, 7, 59. (24) U.S. Food and Drug Administration (FDA), Summary of Safety and Effectiveness Data, INFUSE Bone Graft, 2004, http://www. accessdata.fda.gov/cdrh_docs/pdf/P000058b.pdf. (25) Kirker-Head, C.; Karageorgiou, V.; Hofmann, S.; Fajardo, R.; Betz, O.; Merkle, H. P.; Hilbe, M.; von Rechenberg, B.; McCool, J.; Abrahamsen, L.; Nazarian, A.; Cory, E.; Curtis, M.; Kaplan, D.; Meinel, L. BMP-silk composite matrices heal critically sized femoral defects. Bone 2007, 41 (2), 247−55. (26) Kisiel, M.; Martino, M. M.; Ventura, M.; Hubbell, J. A.; Hilborn, J.; Ossipov, D. A. Improving the osteogenic potential of BMP-2 with hyaluronic acid hydrogel modified with integrin-specific fibronectin fragment. Biomaterials 2013, 34 (3), 704−12. (27) Minier, K.; Toure, A.; Fusellier, M.; Fellah, B.; Bouvy, B.; Weiss, P.; Gauthier, O. BMP-2 delivered from a self-crosslinkable CaP/ hydrogel construct promotes bone regeneration in a critical-size segmental defect model of non-union in dogs. Vet Comp Orthop Traumatol 2014, 27 (6), 411−21. (28) Poldervaart, M. T.; Wang, H.; van der Stok, J.; Weinans, H.; Leeuwenburgh, S. C.; Oner, F. C.; Dhert, W. J.; Alblas, J. Sustained release of BMP-2 in bioprinted alginate for osteogenicity in mice and rats. PLoS One 2013, 8 (8), e72610. (29) Priddy, L. B.; Chaudhuri, O.; Stevens, H. Y.; Krishnan, L.; Uhrig, B. A.; Willett, N. J.; Guldberg, R. E. Oxidized alginate hydrogels for bone morphogenetic protein-2 delivery in long bone defects. Acta Biomater. 2014, 10 (10), 4390−9. (30) Wuerzler, K. K.; Emmert, J.; Eichelsbacher, F.; Kuebler, N. R.; Sebald, W.; Reuther, J. F. Evaluation der osteoinduktiven Potenz von gentechnisch modifizierten BMP-2-Varianten. Oral Maxillofacial Surg. 2004, 8 (2), 83−92. (31) Eger, S.; Scheffner, M.; Marx, A.; Rubini, M. Formation of ubiquitin dimers via azide-alkyne click reaction. Methods Mol. Biol. 2012, 832, 589−96. (32) Green, N. M. Avidin. Adv. Protein Chem. 1975, 29, 85−133. (33) Alborzinia, H.; Schmidt-Glenewinkel, H.; Ilkavets, I.; BreitkopfHeinlein, K.; Cheng, X.; Hortschansky, P.; Dooley, S.; Wolfl, S. Quantitative kinetics analysis of BMP2 uptake into cells and its modulation by BMP antagonists. J. Cell Sci. 2013, 126 (Pt 1), 117−27.

AUTHOR INFORMATION

Corresponding Authors

*Tel.: # 49 (0) 931 31 84122. Fax: #49 (0) 931 31 81068. Email: [email protected]. *Tel.: # 49 (0) 931 31 89207. Fax: #49 (0) 931 31 86158. Email: [email protected]. ORCID

Lorenz Meinel: 0000-0002-7549-7627 Joachim Nickel: 0000-0003-0154-1370 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Dr. M. Rubini (Konstanz, Germany) for providing the plasmid encoding pyrrolysyl-tRNA and for providing pRSFduet-pyrtRNAsynth encoding the corresponding aminoacyl-tRNA synthetase.

■

REFERENCES

(1) Meling, T.; Harboe, K.; Soreide, K. Incidence of traumatic longbone fractures requiring in-hospital management: a prospective ageand gender-specific analysis of 4890 fractures. Injury 2009, 40 (11), 1212−9. (2) Brighton, C. T.; Shaman, P.; Heppenstall, R. B.; Esterhai, J. L., Jr.; Pollack, S. R.; Friedenberg, Z. B. Tibial nonunion treated with direct current, capacitive coupling, or bone graft. Clin. Orthop. Relat. Res. 1995, 321, 223−34. (3) Giannoudis, P. V.; Dinopoulos, H.; Tsiridis, E. Bone substitutes: an update. Injury 2005, 36 (Suppl 3), S20−7. (4) Oryan, A.; Alidadi, S.; Moshiri, A.; Bigham-Sadegh, A. Bone morphogenetic proteins: a powerful osteoinductive compound with non-negligible side effects and limitations. Biofactors 2014, 40 (5), 459−81. (5) Luginbuehl, V.; Meinel, L.; Merkle, H. P.; Gander, B. Localized delivery of growth factors for bone repair. Eur. J. Pharm. Biopharm. 2004, 58 (2), 197−208. (6) Utturkar, A.; Paul, B.; Akkiraju, H.; Bonor, J.; Dhurjati, P.; Nohe, A. Development of Physiologically Based Pharmacokinetic Model (PBPK) of BMP2 in Mice. Biol. Syst.: Open Access 2013, 2, n/a. (7) Haidar, Z. S.; Hamdy, R. C.; Tabrizian, M. Delivery of recombinant bone morphogenetic proteins for bone regeneration and repair. Part A: Current challenges in BMP delivery. Biotechnol. Lett. 2009, 31 (12), 1817−24. (8) Meinel, L.; Kaplan, D. L. Silk constructs for delivery of musculoskeletal therapeutics. Adv. Drug Delivery Rev. 2012, 64 (12), 1111−22. (9) Luhmann, T.; Germershaus, O.; Groll, J.; Meinel, L. Bone targeting for the treatment of osteoporosis. J. Controlled Release 2012, 161 (2), 198−213. (10) Pohl, T. L.; Boergermann, J. H.; Schwaerzer, G. K.; Knaus, P.; Cavalcanti-Adam, E. A. Surface immobilization of bone morphogenetic protein 2 via a self-assembled monolayer formation induces cell differentiation. Acta Biomater. 2012, 8 (2), 772−80. (11) Luhmann, T.; Meinel, L. Nanotransporters for drug delivery. Curr. Opin. Biotechnol. 2016, 39, 35−40. (12) Knaus, P.; Sebald, W. Cooperativity of binding epitopes and receptor chains in the BMP/TGFbeta superfamily. Biol. Chem. 2001, 382 (8), 1189−95. (13) Wang, L.; Xie, J.; Schultz, P. G. Expanding the genetic code. Annu. Rev. Biophys. Biomol. Struct. 2006, 35, 225−49. (14) Costa, G. L.; Weiner, M. P.; Rapid, P. C. R. Site-Directed Mutagenesis. CSH Protoc 2006, 2006 (1), n/a. (15) Kirsch, T.; Nickel, J.; Sebald, W. Isolation of recombinant BMP receptor IA ectodomain and its 2:1 complex with BMP-2. FEBS Lett. 2000, 468 (2−3), 215−9. N


Site-Directed Immobilization of BMP-2: Two Approaches for the

Recommend Documents