ARTICLE pubs.acs.org/Biomac
Biomimetic Calcium Phosphate Mineralization with Multifunctional Elastin-Like Recombinamers Susana Prieto,†,‡ Andriy Shkilnyy,§,|| Claudia Rumplasch,§ Artur Ribeiro,†,‡ F. Javier Arias,†,‡ J. Carlos Rodríguez-Cabello,*,†,‡ and Andreas Taubert*,§,|| †
GIR Bioforge, University of Valladolid, E-47011 Valladolid, Spain Networking Research Center on Bioengineering, Biomaterials and Nanomedicine (CIBER-BBN), Valladolid, Spain § Institute of Chemistry, University of Potsdam, D-14476 Golm, Germany Max-Planck-Institute of Colloids and Interfaces, D-14476 Golm, Germany
)
‡
bS Supporting Information ABSTRACT: Biomimetic hybrid materials based on a polymeric and an inorganic component such as calcium phosphate are potentially useful for bone repair. The current study reports on a new approach toward biomimetic hybrid materials using a set of recombinamers (recombinant protein materials obtained from a synthetic gene) as crystallization additive for calcium phosphate. The recombinamers contain elements from elastin, an elastic structural protein, and statherin, a salivary protein. Via genetic engineering, the basic elastin sequence was modified with the SNA15 domain of statherin, whose interaction with calcium phosphate is well-established. These new materials retain the biocompatibility, “smart” nature, and desired mechanical behavior of the elastin-like recombinamer (ELR) family. Mineralization in simulated body fluid (SBF) in the presence of these recombinamers reveals surprising differences. Two of the polymers inhibit calcium phosphate deposition (although they contain the statherin segment). In contrast, the third polymer, which has a triblock structure, efficiently controls the calcium phosphate formation, yielding spherical hydroxyapatite (HAP) nanoparticles with diameters from 1 to 3 nm after 1 week in SBF at 37 °C. However, at lower temperatures, no precipitation is observed with any of the polymers. The data thus suggest that the molecular design of ELRs containing statherin segments and the selection of an appropriate polymer structure are key parameters to obtain functional materials for the development of intelligent systems for hard tissue engineering and subsequent in vivo applications.
’ INTRODUCTION Biomimetic calcium phosphate/organic hybrid materials have attracted an ever growing interest over the years.1,2 Polymers are popular organic components for those purposes because polymeric growth modifiers are well-known to enable the synthesis of calcium phosphate hybrids, some of which resemble natural bone.13 As a result, calcium phosphate/polymer hybrids have, among others, been promoted as near-natural materials for use in implantology and traumatology. Responsive polymers are of particular interest for several reasons. For example, a polymer, which is water-soluble at room temperature but precipitates at body temperature and forms a gel in a bone defect, could be a potential candidate for bone repair. This is particularly true if the polymer not only precipitates in the bone defect but also activates bone cell growth into the defect, supports the deposition of biological bone material (that is, nucleates calcium phosphate), and, finally, degrades after its task in the bone defect is done. Some reports describe polymer hydrogels bearing peptide side chains.4 These gels have been suggested as candidates for applications such as those mentioned above. However, there r 2011 American Chemical Society
have been no studies that fully exploit the potential of such polymeric materials, even though the incorporation of biological or bioinspired functionalities for regulating polymercell59 and polymermineral10 interactions has been described. Recombinant protein technology, that is, the synthesis of tailored proteins in living organisms by modification of the genetic code of the organism, is an attractive alternative to conventional polymer synthesis, in particular for polymers with complex behavior and composition.1116 Interestingly, however, there are so far no examples of functional hybrid materials based on recombinant proteins and calcium phosphate for biomedical applications. The current publication describes the mineralization of elastinlike recombinamers (ELRs)17 with calcium phosphate in simulated body fluid (SBF). ELRs are recombinant polypeptides with physical properties that are remarkably similar to those of native elastin.1820 ELRs were chosen because their composition Received: November 29, 2010 Revised: March 24, 2011 Published: March 25, 2011 1480
dx.doi.org/10.1021/bm200287c | Biomacromolecules 2011, 12, 1480–1486
Biomacromolecules and hence their physical properties can be tuned via genetic engineering.21 Moreover, ELRs exhibit virtually no endotoxicity, cytotoxicity, and no antigenic properties upon implantation.22,23 Finally, depending on the exact sequence, ELRs have a lower critical solution temperature (LCST) below body temperature. As a consequence, ELRs can be designed such that they are water-soluble at room temperature and precipitate at body temperature. They can thus be expected to form a “precipitated plug” in a bone defect, which will remain in the defect over an extended period of time. To be useful as a biomaterial, ELRs not only need to remain in the bone defect upon implantation or injection but also need to trigger cell ingrowth and activate biological bone formation. A control of calcium phosphate nucleation and growth by the polymeric matrix could be favorable for bone formation, although this is still a matter of debate.2428 It nevertheless appears clear that the enrichment of calcium and phosphate in the defect would be beneficial for bone formation. Salivary statherin has a high affinity for calcium phosphate.2931 The amino-terminal 15-residue fragment (SN15) and its analog (SNA15) exhibit a higher affinity for HAP than the entire statherin molecule due to the negative charge density and helical conformation present in the fragment. Previous studies also demonstrated the ability of these fragments to adsorb at hydroxyapatite (HAP) surfaces and to inhibit its mineralization in supersaturated solutions.32 The SN15 and SNA15 fragments are thus attractive candidates for incorporation into ELRs for the synthesis of biocompatible and biofunctional calcium phosphate hybrid materials. The current work describes in vitro model mineralization experiments, where different ELRs with the SNA15 statherin analog domain have been used as mineralization templates at 25 and 37 °C in SBF. The goal of this study was to evaluate the potential of SNA15 modified ELRs for bone repair and to determine if there are differences in the effects of different ELR compositions. Indeed, as will be shown in the remainder of the text, the polymer composition is a key parameter for mineralization control.
’ EXPERIMENTAL SECTION Polymer Synthesis. Cloning and molecular biology experiments for gene construction were performed using standard methods.33 Expression, purification, and physicochemical characterization of the ELRs were carried out according to previously published procedures.34 Three ELRs were synthesized (sequences shown below). The simplest recombinamer is a homorecombinamer containing a modified motif from the statherin sequence denoted HSS (human salivary statherin). This modified motif, where the phosphoserines at positions 2 and 3 were replaced by two aspartic acid residues, was chosen instead of the original sequence because it shows the same net negative charge, comparable adsorption and affinity to HAP, and a helical structure essential for the function displayed by the protein32 but avoids problems associated with post-translational modification such as serine phosphorylation. The other two ELRs are block corecombinamers: an AB diblock corecombinamer, where the first block is identical to the homorecombinamer described above and a hydrophobic (VPAVG)20 block. The last ELR is an ABA triblock corecombinamer, where the A blocks are identical to the homorecombinamer, and the B block is the (VPAVG)20 block also used in the diblock corecombinamer. To assess further the amino acid composition of the ELR, HCl hydrolysis, derivatization by the AccQ-Tag Waters method, and
ARTICLE
subsequent analysis with UV detection for quantification was done by the Technical-Scientific Service at the University of Barcelona (Spain). Homorecombinamer (A block only), (HSS)3 (statherin sequence is italicized): [((VPGIG)2VPGKG(VPGIG)2)DDDEEKFLRRIGRFG((VPGIG)2VPGKG(VPGIG)2)]3 Diblock (A-B), statherin sequence is italicized: [((VPGIG)2VPGKG(VPGIG)2)DDDEEKFLRRIGRFG((VPGIG)2VPGKG(VPGIG)2)]3-[(VPAVG)]20 Triblock (A-B-A), statherin sequence is italicized: [((VPGIG)2VPGKG(VPGIG)2)DDDEEKFLRRIGRFG((VPGIG)2VPGKG(VPGIG)2)]3-[(VPAVG)]20-[((VPGIG)2VPGKG(VPGIG)2)DDDEEKFLRRIGRFG((VPGIG)2VPGKG(VPGIG)2)]3 Mineralization. All solutions were prepared just before use. Mineralization was carried out in SBF solutions at pH 7.4. SBF and doubly concentrated SBF (2 SBF) were prepared according to previously published procedures.35 Polymer concentrations were 4, 6, 8, and 10 mg/mL. Polymers were dissolved in 10 mL of SBF in 15 mL Falcon tubes. The resulting solutions were stored at 4 °C until the polymer was completely dissolved and a clear solution was obtained (ca. 30 min.). Then, solutions were incubated at 25 and 37 °C for 1 to 3 weeks. SBF was renewed after centrifugation every 48 h, and the pH was checked on samples at 37 °C. In samples incubated at 25 °C, no precipitate formation was observed. As a result, the SBF could not be exchanged because this would have led to loss of the recombinamers. Therefore, only pH values were checked. All pH values were maintained at 7.4 over the entire course of the mineralization to minimize problems associated with SBF preparation and stabilization.36,37 X-ray Diffraction. X-ray diffraction (XRD) was done on a Nonius PDS 120 with Cu KR radiation and position-sensitive detector (1 to 120° 2θ) and on a Nonius D8 with Cu KR radiation. Analysis of the XRD patterns was done with OriginLab Origin 6.1. FTIR. Transmission IR spectra were recorded on a Thermo Nicolet Nexus 670 Fourier transform infrared (FTIR) spectrometer. Samples were ground with KBr and pressed into pellets with a hand press. Spectra were recorded at room temperature from 400 to 4000 cm1 with a resolution of 4 cm1. Electron Microscopy and X-ray Microanalysis. Scanning electron microscopy (SEM) experiments were done on a Hitachi S-4800 operated at 5 kV. Prior to SEM imaging, samples were sputtered with Pt/Au using a Balzers SCD 050 sputter coater. TEM experiments were made on a Zeiss EM 902 operated at 80 kV. EDXS was done with an EDAX-4 energy-dispersive spectroscopy system. Thermal Analysis. Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were done on a Linseis L81 thermal analyzer (Linseis, Germany) working in vertical mode in a stationary air atmosphere (no purge) from 30 to 1000 °C. Heating rate was 10 °C/min and Al2O3 was used as reference. Control experiments at a heating rate of 5 °C/min showed ∼5% deviation from measurements performed at 10 °C/min. Differential Scanning Calorimetry. DSC experiments were done on a Mettler Toledo 822e DSC with liquid-nitrogen cooler and calibrated with indium. Solutions with polymer concentrations of 50 mg mL1 in SBF solutions at pH 7.4 were placed in a 40 μL sealed aluminum pan, and an equal volume of SBF was placed in the reference pan. Before the experiment, samples were held at 0 °C for 5 min. Measurements were run from 0 to 70 at 10 °C/min.
Matrix-Assisted Laser Desorption/Ionization Time-ofFlight Mass Spectroscopy. MALDI-TOF mass spectroscopy was performed in a Voyager STR (Applied Biosystems) in linear mode and with an external calibration using bovine serum albumin (BSA). 1481
dx.doi.org/10.1021/bm200287c |Biomacromolecules 2011, 12, 1480–1486
Biomacromolecules
ARTICLE
Table 1. Amino Acid Composition of Recombinamers homorecombinamer (HSS)3
diblock (HSS)3/(VPAVG)20
triblock (HSS)3/(VPAVG)20/(HSS)3
amino acid residues
theoretical value
experimental value
theoretical value
experimental value
theoretical value
experimental value
alanine arginine aspartic acid glycine glutamic acid isoleucine leucine lysine methionine phenylalanine proline serine valine
9 9 126 7 51 5 15 1 6 61 1 61
9.50 9.69 132 7 47.12 5.65 14.76 0.53 5.94 59.58 0.87 58.49
20 9 9 146 7 51 5 15 1 6 81 1 101
22.09 9.89 8.92 144.08 6.64 48.6 4.44 14.7 0.94 5.55 82.11 0.77 102.95
20 18 18 272 13 102 8 30 1 12 141 1 161
22.52 19.64 18.79 268.85 13 97.3 7.76 29.61 1.1 11.02 143.68 1.32 165.34
Table 2. Theoretical and Experimental Values of Recombinamersa
a
recombinamer
predicted Mw (Da)
experimental Mw ((SD) (Da)
(HSS)3 (HSS)3/VPAVG)20 (HSS)3/(VPAVG)20/(HSS)3
31 889 40 361 71 462
31 910.7 ( 6.1 40 420.9 ( 12.5 71 430.7 ( 12.9
Data are means ( SD (n = 9).
’ RESULTS Table 1 shows the theoretical and experimental amino acid composition of the ELRs. Both values are in agreement, indicating that the ELRs are of high purity and desired amino acid sequence. The presence of small amounts of serine and methionine is due to the addition of the tails MESLLP and V to the amino and carboxyl ends, respectively, in the last cloning step. Table 2 shows results from MALDI-TOF mass spectrometry experiments, which were performed to support and further assess the purity of the recombinamers and their molecular weight. Differences between predicted and experimental MW are always smaller than one single amino acid molecular weight. ELRs have an LCST which, via their composition, can be tuned at any temperature within the liquid water range. Below the LCST, the recombinamer is water-soluble but segregates from the solution at the LCST and forms nanometer- to micrometer-sized particles, yielding a turbid suspension.38 Figure 1 shows differential scanning calorimetry (DSC) data of the pure recombinamers in SBF. All thermograms show broad exothermic peaks that can be assigned to the phase segregation of the recombinamer at the LCST. In the case of the homorecombinamer, just one broad signal is visible. In the case of the diblock two (although broad) endotherms and in the case of the triblock a broad endothermic peak and a shoulder are visible. The appearance of two peaks is caused by the independent folding of the two different blocks, similar to a previous study on a related system.39 DSC thus suggests that the two blocks precipitate rather independently from one another. Moreover, the formation of polymeric, micelle-like, aggregates above the LCST is possible (via segregation of the acidic and hydrophilic HSS domains to the exterior and the hydrophobic and collapsed elastin to the interior of the aggregate), although we do not currently have direct evidence of a micellization or similar process. Mineralization experiments were conducted at 4, 20, and 37 °C in SBF. Macroscopic observation of the vials shows no
Figure 1. DSC thermograms of the three recombinamers in SBF. Recombinamer concentration is 50 mg/mL and vertical lines indicate reaction temperatures used for mineralization experiments; see Experimental Section and below.
precipitation at 4 and 20 °C. Even extensive centrifugation did not yield any product from these reactions, and dynamic light scattering (DLS) experiments yielded inconclusive results, but electron microscopy suggests that, if any, only very few nanoparticles precipitate. This holds for both the control samples (no recombinamer) and the samples containing ELRs, even after 3 weeks. At 37 °C, however, a large volume of precipitate is observed already after 1 week with all ELRs. The largest amounts of precipitate were collected from the samples grown with the triblock ELR (Figure S1 of the Supporting Information). Attempts at quantifying the inorganic fraction using TGA/DTA have been inconclusive (Figure S2 of the Supporting Information). This is partially due to the low overall inorganic contents (ca. 10%) and further complications related to heat-induced phase transitions in calcium phosphate during the TGA/DTA treatment.1,22 There are, however, significant macroscopic differences between the samples. Precipitates grown in the presence of the homorecombinamer and the diblock corecombinamer are, after isolation and drying, clear to slightly yellow, tough materials that are difficult to cut or otherwise manipulate. In contrast, the precipitate obtained with the triblock corecombinamer is a white powder, which easily moves under the pressure of a spatula. 1482
dx.doi.org/10.1021/bm200287c |Biomacromolecules 2011, 12, 1480–1486
Biomacromolecules
ARTICLE
Figure 3. TEM image of a sample precipitated in the presence of the triblock ELR at 37 °C after 1 week. The samples grown with the homorecombinamer and the diblock do not show any features. Inset: size distribution histogram of the particles shown in the main image. The average particle diameter is 1.8 ( 0.57 nm.
Table 3. EDXS Data of the Precipitates Isolated after 1 Week at 37 °C.a Error is below 1% homo recombinamer
diblock corecombinamer
triblock corecombinamer
element
weight %
atom %
weight %
atom %
weight %
atom %
C O Na Si P S Cl Ca
68.01 28.33 0.94 0.14 0.3 0.15 1.85 0.28
74.97 23.45 0.54 0.06 0.13 0.06 0.69 0.09
66.03 29.99 1.48 -
73.23 24.97 0.86 -
28.53 43.29 0.59 0.26 10.63 0.56 0.22 13.98
40.18 45.77 0.43 0.16 5.81 0.3 0.11 5.9
Figure 2. SEM images of the samples precipitated with ELRs at 37 °C after 1 week of incubation in SBF. Insets are higher magnification images of the same samples; scale bars in the insets are 1 μm.
a
Figure 2 shows representative SEM images of the samples. SEM clearly shows that the samples precipitated with the homorecombinamer and the diblock corecombinamer are morphologically quite different from the sample grown with the triblock corecombinamer. The former samples show large plateor step-like features. At higher magnification, tiny dots become visible, which could be calcium phosphate particles on the surface of the ELR. This is, however, difficult to conclude because of significant charging of the samples. In contrast, samples precipitated with the triblock corecombinamer exhibit a very uniform morphology of spherical particles with diameters of ca. 500 nm. Higher magnification SEM images suggest that the individual particles are composed of yet smaller particles with a diameter of a few nanometers. Figure 3 shows a representative transmission electron microscopy (TEM) image of a sample grown with the triblock ELR at 37 °C. As suggested by SEM, TEM finds numerous nanoparticles with diameters in a range from 1 to 3 nm. The nanoparticles are
spherical, have a reasonably narrow size distribution, and are mostly present as individual nanoparticles, that is, not as aggregated clusters. Table 3 shows the corresponding energy-dispersive X-ray spectroscopy (EDXS) data of the samples precipitated at 37 °C. Samples precipitated with the homorecombinamer and the diblock corecombinamer contain only traces of calcium and phosphorus. In contrast, the samples precipitated with the triblock corecombinamer show intense signals of both Ca and P along with C, N, and O. The oxygen signal is due to both the recombinamer and the phosphate anion, whereas C and N are due to the presence of a large amount of recombinamer. The Ca/P ratio is ca. 1.3, which is indicative of calcium deficient HAP, although this assignment is ambiguous because mixtures of two crystal phases, for example, dicalcium phosphate dihydrate (DCPD) and HAP, can also lead to such low Ca/P ratios.40 Figure 4 shows XRD patterns of the same samples. All samples exhibit a broad halo, which begins at ca. 10° 2θ and tapers off after 1483
Dashes indicate elements below the detection limit.
dx.doi.org/10.1021/bm200287c |Biomacromolecules 2011, 12, 1480–1486
Biomacromolecules
ARTICLE
Figure 4. XRD patterns of precipitates obtained after 1 week in SBF at 37 °C with concentration of recombinamer 10 mg/mL. Vertical lines are guides for the eye to identify matching reflections in the different samples. Label “x” refers to recombinamer reflections; numbers are the Miller indices of the respective HAP reflections. For full patterns, see Figure S3 of the Supporting Information.
ca. 20° 2θ. Moreover, all samples exhibit a number of peaks at higher angles. The sample precipitated with the homorecombinamer shows five reflections at 44.3, 64.3, 77.2, 78.8, and 81.4 and a broad hump at 56.1° 2θ. The sample precipitated with the diblock ELR shows the same reflections, but they are much weaker and broader, and the two last reflections are not observed anymore. We assign these reflections to the recombinamer, in particular, the (HSS)3 block. In addition, the sample precipitated with the diblock corecombinamer also shows a weak reflection at 31.8°, which can be assigned to the 211 reflection of HAP (JCPDS 09-0432). Along with a number of other reflections, this reflection is also observed in the samples grown with the triblock corecombinamer. In this last case, all reflections can be assigned to HAP (JCPDS 090432). Although the reflections appear rather sharp, a more detailed analysis reveals that the full width at half-maximum (fwhm) of the individual reflections is ca. 0.4 to 0.5° 2θ, which, although not very broad, cannot be considered very narrow either. XRD thus qualitatively confirms the presence of small particles. Figure 5 shows the IR spectra of the precipitates. The presence of the ELR in all precipitates is indicated by broad bands between 3500 and 2500 cm1. They can be attributed to hydrogen-bonded stretching modes of crystal water, NH, and CH vibrations of the protein backbone. The bands at 1651, 1532, 1436, and the shoulder at 1100 cm1 are due to amide I, amide II, and CH and CC stretching modes, respectively. Only the spectrum of the sample obtained in the presence of the triblock corecombinamer clearly indicates the presence phosphate. Here bands at 1642, 1030, and 580 cm1 can be assigned to the phosphate anion.4143 IR thus supports XRD and EDXS in that also here we find evidence of calcium phosphate deposition mainly with the triblock ELR.
’ DISCUSSION Somewhat surprisingly, there are significant differences among the three recombinamers used as crystallization additives for calcium phosphate mineralization. At 4 and 20 °C, no precipitation occurs in either sample, not even in the control sample (without recombinamer). At 37 °C, visual inspection (Figure S1 of the Supporting Information) and centrifugation experiments show that the samples grown with the homorecombinamer and
Figure 5. FTIR spectra of samples precipitated with 10 mg/mL of recombinamer.
the diblock ELRs are different from the samples grown with the triblock recombinamer: the latter is a white, powdery material, whereas the former two are tough, clear-to-yellowish solids. Electron microscopy (Figures 2 and 3) confirms these differences. The tough solids formed in the presence of the homorecombinamer and the diblock corecombinamer show micrometer-sized, terrace- or step-like features, but there is virtually no evidence of calcium phosphate deposition. EDXS (Table 3), XRD (Figure 4), and IR spectroscopy (Figure 5) confirm this finding. EDXS detects
