Mineralization of synthetic polymer scaffolds: A bottom-up approach for the development of artificial bone

Article abstract of DOI:10.1021/ja043776z

The controlled integration of organic and inorganic components confers natural bone with superior mechanical properties. Bone biogenesis is thought to occur by templated mineralization of hard apatite crystals by an elastic protein scaffold, a process we sought to emulate with synthetic biomimetic hydrogel polymers. Cross-linked polymethacrylamide and polymethacrylate hydrogels were functionalized with mineral-binding ligands and used to template the formation of hydroxyapatite. Strong adhesion between the organic and inorganic materials was achieved for hydrogels functionalized with either carboxylate or hydroxy ligands. The mineral-nucleating potential of hydroxyl groups identified here broadens the design parameters for synthetic bonelike composites and suggests a potential role for hydroxylated collagen proteins in bone mineralization.

Full text of DOI:10.1021/ja043776z

Published on Web 02/17/2005
Mineralization of Synthetic Polymer Scaffolds: A Bottom-Up
Approach for the Development of Artificial Bone
Jie Song,*,†,‡ Viengkham Malathong,† and Carolyn R. Bertozzi*,†,‡,§
Contribution from the Materials Sciences DiVision, Lawrence Berkeley National Laboratory,
UniVersity of California, Berkeley, California 94720, Biological Nanostructures Facility,
Molecular Foundry, Lawrence Berkeley National Laboratory, UniVersity of California,
Berkeley, California 94720, and Departments of Chemistry and Molecular and Cell Biology,
and Howard Hughes Medical Institute, UniVersity of California, Berkeley, California 94720
Received October 13, 2004; E-mail: jsong@lbl.gov (J.S.); crb@berkeley.edu (C.R.B.)
Abstract: The controlled integration of organic and inorganic components confers natural bone with superior
mechanical properties. Bone biogenesis is thought to occur by templated mineralization of hard apatite
crystals by an elastic protein scaffold, a process we sought to emulate with synthetic biomimetic hydrogel
polymers. Cross-linked polymethacrylamide and polymethacrylate hydrogels were functionalized with
mineral-binding ligands and used to template the formation of hydroxyapatite. Strong adhesion between
the organic and inorganic materials was achieved for hydrogels functionalized with either carboxylate or
hydroxy ligands. The mineral-nucleating potential of hydroxyl groups identified here broadens the design
parameters for synthetic bonelike composites and suggests a potential role for hydroxylated collagen proteins
in bone mineralization.
scales,7,10,11 existing composite implants lack a well-defined
interface between their constituents.
Introduction
The continued increase in the age of the population has
generated higher demands for bone grafting.1 Existing ortho-
pedic implants typically consist of a single bioinert material,
such as metals, ceramics, or polymers, or of a relatively coarse
combination of two or three components.2-4 Although these
synthetic materials provide an immediate solution for many
patients, their long-term performance is generally not satisfac-
tory. This is often due to a mechanical property mismatch
between the implant and its surrounding native tissue, which
ultimately leads to implant failure and tissue damage.5-7 In
addition, these traditional implants rarely bear functionalities
that encourage communication with their cellular environment,
limiting the potential for self-repair, adaptation to physiological
conditions,8 and tissue attachment and ingrowth.9 Finally, unlike
natural skeletal tissue, where organic and inorganic components
are integrated into well-defined architectures at all length
The development of bonelike composites with improved
mechanical properties and enhanced biocompatibility calls for
a biomimetic approach using natural bone as a guide. Natural
bone is a composite of collagen, a protein-based hydrogel
template, and carbonated apatite crystals with varying composi-
tions and microstructures (Figure 1A). Collagen provides a
structural framework for the growth of calcium apatite.12 The
unusual combination of a hard inorganic material and an
underlying elastic hydrogel network gives bone unique me-
chanical properties, such as low stiffness, resistance to tensile
and compressive forces, and high fracture toughness.7,13 Bone
biomineralization is thought to start with the formation of
transient amorphous calcium phosphates and poorly crystalline
apatites.14-16 These precursors then undergo several crystalline
phase transitions before the more stable crystalline hydroxy-
apatite (HA) finally forms.11 The stabilization of amorphous
calcium phosphate in the early stage of bone mineralization and
the subsequent formation of nanometer-sized particles are
believed to be mediated by anionic proteins attached to collagen
† Materials Sciences Division, Lawrence Berkeley National Laboratory.
‡ Biological Nanostructures Facility, Lawrence Berkeley National Labo-
ratory.
§ Departments of Chemistry and Molecular Cell Biology, and Howard
Hughes Medical Institute.
(1) Hall, M. J.; Owings, M. F. 2000 National Hospital Discharge SurVey;
National Center for Health Statistics: Hyattsville, MD, 2002; Vol. 329.
(2) Thomson, R. C.; Yaszemski, M. J.; Powers, J. M.; Mikos, A. G.
Biomaterials 1998, 19, 1935-1943.
(10) Vincent, J. F. V. Structural Biomaterials; Princeton University Press:
Princeton, NJ, 1990.
(11) Mann, S. Biomineralization: Principles and Concepts in Bioinorganic
Materials Chemistry; Oxford University Press: New York, 2001.
(12) Landis, W. J.; Hodgens, K. J.; Arena, J.; Song, M. J.; McEwen, B. F.
Microsc. Res. Tech. 1996, 33, 192-202.
(3) Marra, K. G.; Szem, J. W.; Kumta, P. N.; DiMilla, P. A.; Weiss, L. E. J.
Biomed. Mater. Res. 1999, 47, 324-335.
(4) Ma, P. X.; Zhang, R. Y.; Xiao, G. Z.; Franceschi, R. J. Biomed. Mater.
Res. 2001, 54, 284-293.
(13) Baron, R. In Handbook of Experimental Pharmacology; Abou-Samra, A.
B., Mundy, G. R., Martin, T. J., Eds.; Springer: Berlin, New York, 1993;
Vol. 107, pp 111-147.
(5) Black, J. Biological Performance of Materials: Fundamentals of Biocom-
patibility, 3rd revision and expanded ed.; Marcel Dekker: New York, 1999.
(6) Hillsley, M. V.; Frangos, J. A. Biotechnol. Bioeng. 1994, 43, 573-581.
(7) Weiner, S.; Wagner, H. D. Annu. ReV. Mater. Sci. 1998, 28, 271-298.
(8) Piehler, H. R. MRS Bull. 2000, 25, 67-70.
(14) Roberts, J. E.; Heughebaert, M.; Heughebaert, J. C.; Bonar, L. C.; Glimcher,
M. J.; Griffin, R. G. Calcif. Tissue Int. 1991, 49, 378-382.
(15) Roberts, J. E.; Bonar, L. C.; Griffin, R. G.; Glimcher, M. J. Calcif. Tissue
Int. 1992, 50, 42-48.
(9) Willmann, G. AdV. Eng. Mater. 1999, 1, 95-105.
(16) Kim, H.-M.; Rey, C.; Glimcher, M. J. Calcif. Tissue Int. 1996, 59, 58-63.
9
3366
J. AM. CHEM. SOC. 2005, 127, 3366-3372
10.1021/ja043776z CCC: $30.25 © 2005 American Chemical Society

Mineralization of Synthetic Polymer Scaffolds
A R T I C L E S
Figure 1. Recapitulating natural bone synthesis with synthetic polymers functionalized to mimic the mineral-nucleating proteins in bone. Natural bone is
a biocomposite of collagen, a protein-based hydrogel template, and inorganic apatite crystals (A). The anionic noncollagenous proteins attached to the
collagen matrix are thought to play important templating and regulatory roles in the bone formation and remodeling process (B). Inspired by the natural
bone, polymethacrylamide/polymethacrylate-based hydrogel copolymers containing biomimetic functional domains were designed (C). Anionic residues
mimicking acidic noncollagenous proteins that template the nucleation and growth of biominerals in bone are copolymerized with other monomers in
aqueous media (D).
(Figure 1A,B).17-19 The exact role of individual acidic matrix
proteins in mineralization, however, is still far from understood.
We are interested in a “bottom-up” approach to the design
and synthesis of artificial bone. This entails the design of simple
model systems with well-defined chemical, physical, and bio-
logical properties, followed by an iterative increase in the com-
plexity of the system to realize a higher-order approximation
of natural bone. Impressive progress has been made in this direc-
tion, with some recent examples demonstrating that bonelike
properties can be engendered in wholly synthetic systems.20,21
Here, we describe an approach for recapitulating bone biogenesis
by using hydrogel polymers functionalized to mimic the mineral-
nucleating proteins of bone (Figure 1B). We synthesized cross-
linked polymethacrylamides and polymethacrylates containing
biomimetic mineral-nucleating ligands and investigated their
integration with calcium phosphates (Figure 1C,D) using a
recently developed22 mineralization approach. We discovered
that the morphology and crystallinity of the mineral, as well as
the binding strength at the polymer-mineral interface, were gov-
erned by the structure and density of the templating ligands.
These results provide a framework for generating synthetic com-
posites with defined organic/inorganic interfaces similar to
natural bone.
groups similar to glutamate-, aspartate-, and phosphoserine-rich
bone proteins.17-19 We chose poly(2-hydroxyethyl methacrylate)
(pHEMA)-based hydrogels for this purpose due to their well-
established biocompatibility and ease of functionalization.23-25
As shown in Figure 2, we synthesized a library of anionic
methacrylamides. Copolymerization with either 2-hydroxy-
ethyl methacrylate (HEMA) or 2-hydroxyethyl methacrylamide
(HEMAm)22 formed three-dimensional hydrogel copolymers.
The anionic monomers varied in overall polarity and in the num-
ber of negatively charged carboxylate groups. By varying the
percentage of the anionic monomers within the copolymers,
the average distance between potential nucleation sites was
modulated. Cross-linkers with ester (EGDMA) and amide
(EGDMAm) linkages were designed to generate HEMA- and
HEMAm-based hydrogels, respectively. The ester groups of
EGDMA and HEMA can be cleaved under basic conditions to
form the corresponding carboxylic acids, introducing additional
charge into the copolymer. As discussed later, we exploited this
feature for integration of calcium phosphates. By contrast,
EGDMAm and HEMAm are resistant to hydrolytic cleavage.
In addition to simple anionic monomers, we synthesized a
monomer bearing the peptide GRGD. This sequence includes
the RGD motif which is known to promote cell adhesion.26,27
All monomers and cross-linker EGDMAm were synthesized via
the direct coupling of the corresponding amino acid, oligopeptide
or diamine with methacryloyl chloride.
Results
Our first goal was to develop a three-dimensional scaffold
for HA mineralization that could be functionalized with anionic
(22) Song, J.; Saiz, E.; Bertozzi, C. R. J. Am. Chem. Soc. 2003, 125, 1236-
1243.
(17) Hunter, G. K.; Goldberg, H. A. Biochem. J. 1994, 302, 175-179.
(18) George, A.; Bannon, L.; Sabsay, B.; Dillon, J. W.; Malone, J.; Veis, A.;
Jenkins, N. A.; Gillbert, D. J.; Copeland, N. G. J. Biol. Chem. 1996, 271,
32869-32873.
(19) Boskey, A. L. Annu. ReV. N.Y. Acad. Sci. 1995, 760, 249-256.
(20) Hartgerink, J. D.; Beniash, E.; Stupp, S. I. Science 2001, 294, 1684-1688.
(21) Lutolf, M. P.; Weber, F. E.; Schmoekel, H. G.; Schense, J. C.; Kohler, T.;
Muller, R.; Hubbell, J. A. Nat. Biotechnol. 2003, 21, 513-518.
(23) Phillips, A. J.; Stone, J. Contact Lenses; Butterworth & Co.: London,
Boston, 1989.
(24) Oxley, H. R.; Corkhill, P. H.; Fitton, J. H.; Tighe, B. J. Biomaterials 1993,
14, 1064-1072.
(25) Flynn, L.; Dalton, P. D.; Shoichet, M. S. Biomaterials 2003, 24, 4265-
4272.
(26) Massia, S. P.; Hubbell, J. A. Ann. N.Y. Acad. Sci. 1990, 589, 261-270.
(27) Ratner, B. D. J. Mol. Recognit. 1996, 9, 617-625.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3367

A R T I C L E S
Song et al.
Figure 2. Monomers and cross-linkers designed for polymethacrylamide- and polymethacrylate-based hydrogels. Anionic monomers, GlyMA, SerMA,
AspMA, and GluMA, vary in the overall polarity and the number of negatively charged carboxylates. A methacrylamide-modified peptide, GRGDMA, that
possesses the RGD motif was designed to support cell adhesion. They were copolymerized with HEMA and HEMAm using the cross-linkers EGDMA and
EGDMAm, respectively, to present potential mineral nucleating sites on the polymer scaffold.
We next evaluated the ability of the functionalized hydrogels
to template HA mineralization. We have recently demonstrated
that high-affinity integration of calcium phosphates with
pHEMA can be realized by slowly heating the hydrogel in an
acidic solution of HA in the presence of urea.22 This method
takes advantage of the high and low solubilities of HA in acidic
and basic solutions, respectively,30 and utilizes the thermal
decomposition of urea to effect a gradual and homogeneous
pH change in the mineralization solution. In addition, the change
in both temperature and pH during this process promotes
Figure 3. Equilibrium water content (EWC) of pHEMA-based hydrogel
hydrolysis of the hydroxyethyl ester side chains of pHEMA to
the corresponding carboxylic acids, thereby generating abundant
Ca2+-binding sites.22
copolymers containing various percentages of anionic residues. The EWC
of 100% pHEMA gel was 40%.19
The hydrogels were formed by radical copolymerization of
HEMA or HEMAm and up to 10 wt % of an anionic monomer
with 2 wt % of a cross-linker. The incorporation of anionic
residues into the hydrogel copolymers was confirmed by
equilibrium water content (EWC) measurements.28,29 As shown
in Figure 3, the introduction of 10 wt % polar anionic ligands
increased the EWC of the pHEMA-based copolymers from 40%
(100% pHEMA) to 60-90%. We also tested the biocompat-
ibility of the functionalized hydrogels in cell culture. Human
osteosarcoma TE85 cells attached, spread, and proliferated on
all hydrogel copolymers with no apparent cytotoxicity (Sup-
porting Information, Figure S-1).
We applied the urea-mediated mineralization method to the
pHEMA-based copolymers containing anionic ligands. The
hydrogel copolymers were immersed in the acidic HA-urea
solution at room temperature, and the solution was heated to
95 °C at 0.2 °C/min. We observed the mineral growth patterns
shown in Figure 4. Circular mineral domains that initiated from
individual nucleation sites (indicated by arrows) were templated
by pHEMA-based hydrogels containing 5 or 10 wt % GlyMA,
SerMA, GluMA, or AspMA. After prolonged mineralization
(10 additional hours at 95 °C), these circular mineral domains
eventually merged and covered the entire surface of the hydrogel
(Figure 4F) with a thickness >5 µm. A calibrated energy-
dispersive X-ray spectroscopy (EDS) analysis performed over
(28) Song, J.; Saiz, E.; Bertozzi, C. R. J. Eur. Ceram. Soc. 2003, 23, 2905-
2919.
(29) Kim, S. J.; Shin, S. R.; Lee, K. B.; Park, Y. D.; Kim, S. I. J. Appl. Polym.
Sci. 2004, 91, 2908-2913.
(30) Nancollas, G. H.; Zhang, J. In Hydroxyapatite and Related Materials;
Brown, P. W., Constantz, B., Eds.; CRC: Boca Raton, FL, 1994; p 73.
9
3368 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

Mineralization of Synthetic Polymer Scaffolds
A R T I C L E S
Figure 4. Urea-mediated integration of calcium phosphate with pHEMA-based hydrogel copolymers containing various anionic residues. SEM micrographs
shown are mineralized pHEMA-based hydrogels copolymerized with 5% GluMA (A and B), 5% GlyMA (C), 5% SerMA (D), 10% AspMA (E), or 10%
GluMA (F). The mineralization was performed by heating HA-urea solution from room temperature to 95 °C at a constant heating rate of 0.2 °C/min, and
then maintaining the sample at 95 °C for 0 h (A-E) or 10 h (F). Two-dimensional outward growth of circular calcium phosphate mineral domains from
multiple nucleation sites (indicated by arrows) was observed on all composite surfaces. When extended mineralization was applied, the circular mineral
domains eventually merged and covered the entire hydrogel surface with a final mineral layer several microns thick (F). A representative EDS analysis (inset
of A) performed over the circular mineral domains of the composite revealed a Ca/P ratio (1.6 ( 0.1) similar to that of HA. Note that the deliberate
fracturing of the composites (C, and the lower left corner of F) did not lead to delamination of the mineral layer, suggesting good mineral-gel interfacial
adhesion strength. All images were acquired at 15 kV.
the composite surface revealed a Ca/P ratio (1.6 ( 0.1) similar
to that of HA (Figure 4A inset). X-ray diffraction (XRD)
analysis (data not shown) suggested that the circular calcium
phosphate domains were either amorphous or nanocrystalline,
as no characteristic diffraction peaks matching those of crystal-
line apatites (e.g., HA) were detected. Deliberate fracturing of
the composites (Figure 4C,F) did not lead to delamination of
any mineral domains, suggesting good mineral-hydrogel
interfacial adhesion strength. This was further supported by
Vickers indentation performed on selected composites. After
>10 N load was applied to the surface of the composites for
30 s, the indented samples were analyzed by SEM. No
delamination of the mineral domains was observed on any
sample examined. The integrity of the strongly adhered circular
mineral domains remained unchanged over time.
It is worth noting that the carboxylates associated with the
incorporated amino acids and those generated by in situ
hydrolysis of pHEMA may have both contributed to the direct
and extensive mineral-hydrogel contacts responsible for the
strongly adhered mineral layer. To distinguish the mineral-
templating contribution of different anionic amino acid residues
from that of in situ generated surface carboxylates, a hydrogel
scaffold that is resistant to hydrolysis during urea-mediated
mineralization was investigated. The pHEMAm-based hydro-
gels were designed to serve this purpose. Mineralization of
pHEMAm-based hydrogels was achieved by incubating the
hydrogels in acidic HA-urea solution, starting at room tem-
perature and then heating to 95 °C (0.2 °C/min). The gels were
then held at 95 °C for 10 h. As shown in Figure 5, platelike
crystals formed on all surfaces examined. These crystals, often
appearing as spherical aggregates, covered the surfaces of the
pHEMAm-based hydrogels containing 10 wt % of GlyMA
(Figure 5A,B), SerMA (Figure 5C,D), AspMA (Figure 5E), and
GluMA (Figure 5F) residues. The morphology, XRD peaks
(Figure 5G), and Ca/P ratio (∼1.6, Figure 5H) were consistent
with those of crystalline HA. XRD analysis suggested that the
platelets were preferentially, although not perfectly, aligned
along the c axis direction, as indicated by the comparable inten-
sities of the (002), (004), and (211) peaks. For randomly oriented
synthetic HA samples, intensities of (002) and (004) are usually
less than 50% of that of the main reflection (211) (Supporting
Information, Figure S-2). When the c axis of the thin HA crystal
platelets formed in these composites is perpendicular to the
hydrogel surfaces, their thin edges appear as intense lines in
top view SEM micrographs (Figures 5A-F). The additional
broad peaks may be attributed to the embedded organic matrix
and amorphous calcium phosphate nodules underneath the
crystalline HA, which will be discussed later in more detail.
Calcium phosphates formed on pHEMA- (Figure 4) versus
pHEMAm-based (Figure 5) hydrogels were different with
respect to both morphology and crystallinity. Calcium phos-
phates formed on pHEMA-based gels create a continuous, two-
dimensional mineral layer with strong adhesive strength to the
underlying matrix. By contrast, HA formed on pHEMAm-based
gels nucleates on the gel surface but then propagates in three
dimensions without extensive surface interactions. We attribute
these differences to the additional anionic groups formed in
pHEMA-based gels during the mineralization process, which
presumably increase the number of strong calcium phosphate
binding sites on the gel surface.
For hydrogels based on pHEMAm, the morphology and
crystallinity of calcium apatites were independent of the structure
of the biomimetic anionic ligands. AspMA-, GluMA-, SerMA-,
and GlyMA-containing copolymers were equally capable of
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3369

A R T I C L E S
Song et al.
Figure 5. Urea-mediated calcium apatite growth on pHEMAm-based hydrogel copolymers containing various anionic residues. Mineralization was performed
by heating hydrogel copolymers in an acidic HA-urea solution from room temperature to 95 °C at a constant heating rate of 0.2 °C/min, and then holding
at 95 °C for 10 h. SEM micrographs shown are mineralized pHEMAm-based hydrogels copolymerized with 10% GlyMA (A, B, and J-L), 10% SerMA (C
and D), 10% GluMA (E), 10% AspMA (F), and pure pHEMAm (I). Note the platelike morphology of the crystals that aggregated on the surface of all
anionic hydrogel-mineral composites. A representative XRD (G) of the anionic hydrogel-mineral composites showed diffraction peaks characteristic of
crystalline HA. An EDS analysis (H) performed over the surface shown in (D) revealed a Ca/P ratio of 1.6 ( 0.1. The selected area EDS analysis (inset of
I) performed over a bulging nodule on the surface of 100% pHEMAm (I, indicated by the arrow) suggested the mineralized nature of the nodules. A 50°
tilt of the p(HEMAm-co-10%GlyMA)-mineral composite revealed similar nodules (indicated by arrows) beneath clusters of platelike HA crystals (J). X-ray
elemental mapping of Ca (K) and P (L) within the same sample area shown in (J) confirmed that both the bulging nodules and the platelike crystals
contained Ca and P. All images were acquired at 15 kV.
mediating mineralization. This observation is in agreement with
the interchangeable nucleating role of Asp and Glu residues,
for instance, in bone sialoproteins, as suggested by site-directed
mutagenesis studies.31
pHEMAm-based hydrogels containing various anionic ligands.
When the composites were tilted by 50°, the SEM indeed
revealed that similar nodules had formed beneath almost all
spherical clusters of platelike HA crystals. A representative
SEM micrograph of composites formed with p(HEMAm-co-
10%GlyMA) is shown in Figure 5J, with the nodules appearing
in darker color (indicated by arrows) and the spherical ag-
gregates of crystalline HA platelets appearing in white. X-ray
elemental mapping of Ca (Figure 5K) and P (Figure 5L) within
the same sample area confirmed that both the nodules and the
platelike crystals were composed of Ca and P.
As a control, we also examined whether 100% pHEMAm
gels lacking anionic ligands could be mineralized using the urea-
mediated method described above. Unlike the 100% pHEMA
gel,22 the 100% pHEMAm gel did not template the exten-
sive growth of a strongly adhered mineral layer. Instead, iso-
lated mineral nodules were formed across the surface of the
pHEMAm gel (Figure 5I). EDS analysis performed over the
nodules showed a Ca/P ratio of 1.5 ( 0.1 (inset of Figure 5I).
XRD of the composite did not reveal any characteristic reflec-
tions matching those of crystalline HA. Thus, our results suggest
that the hydroxyl groups of pHEMAm can template the
formation of calcium phosphate, yet in a way that is markedly
different from gels functionalized with anionic residues.
The formation of calcium phosphate nodules on the surface
of 100% pHEMAm prompted us to re-examine the mineralized
Discussion
We have developed a system for templated mineralization
of calcium phosphates on organic hydrogels functionalized to
mimic bone matrix proteins. The results highlight the importance
of anionic Ca2+-binding sites at a critical density for high-affinity
integration of the inorganic and organic materials. The genera-
tion of a continuous layer of amorphous or nanocrystalline
calcium phosphate that strongly adheres to the pHEMA-based
hydrogels provides additional possibilities for fine-tuning the
mechanical properties of bonelike composite materials. In natural
(31) Tye, C. E.; Rattray, K. R.; Warner, K. J.; Gordon, J. A. R.; Sodek, J.;
Hunter, G. K.; Goldberg, H. A. J. Biol. Chem. 2003, 278, 7949-7955.
9
3370 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

Mineralization of Synthetic Polymer Scaffolds
A R T I C L E S
at pH 8-9 throughout the reaction. The mixture was warmed to room
temperature and stirred for 4 h before it was quenched with Amberlite
IR-120 (H+ form) ion-exchange resin to a final pH of 5. After silica
gel chromatography (19:1 CHCl3/MeOH), EGDMAm was isolated in
71% yield: 1H NMR (500 MHz, CDCl3) δ 6.83 (2H, br), 5.75 (2H, d,
J ) 1.0 Hz), 5.35 (2H, d, J ) 1.0 Hz), 3.50 (4H, t, J ) 2.5 Hz), 1.96
(6H, s); 13C NMR (125 MHz, CDCl3) δ 169.62, 139.26, 120.35, 40.38,
18.53; HRMS FAB+ (NBA) C10O2N2H17 [M + H]+ calcd, 197.1290;
found, 197.1287.
calcified tissues, the isotropic and less brittle amorphous mineral
components,14,16,32,33 rather than their harder, stiffer, and less
soluble crystalline counterparts, are thought to contribute to
resistance to crack propagation.33 New strategies for integrating
an organic matrix with both amorphous and crystalline minerals,
as demonstrated in this work, should help bridge the gap
between synthetic and natural composite materials.
The distinctively different calcium phosphate nodules and
platelike HA crystals formed on the pHEMAm-based scaffolds
demonstrated that both hydroxyl and carboxylate groups may
be capable of templating the growth of calcium phosphates.
Conventional wisdom has focused overwhelmingly on acidic
noncollagenous ECM proteins17-19,34 as modulators of natural
bone biomineralization. The mineral-nucleating potential of
neutral hydroxyl groups identified here, coupled with evidence
from other synthetic systems,35,36 suggests a plausible involve-
ment of collagens in modulating mineral growth. The regularly
occurring hydroxyamino acids in the triplet repeats of col-
lagen37,38 may be involved in stabilizing transient amorphous
calcium phosphate and in guiding the growth of small crystallites
into larger apatite crystals that eventually fill the space between
collagen fibers.39,40
The model system described here provides a platform for
deriving basic rules for the design of synthetic bonelike
composites and for fundamental studies of the biomineralization
process. These synthetic models do not possess the level of
sophistication of natural bone.40 Nonetheless, by manipulating
the structure and density of mineral-binding ligands presented
on the hydrogels, we demonstrated that wholly synthetic organic
matrixes can be integrated with biominerals with varied affinity,
morphology, and crystallinity. To further bridge the gap between
synthetic bonelike materials and natural bone, more intelligent
use of the information obtained from both the “top-down” and
the “bottom-up” approaches will be necessary.
1.3. Amino Acid Methacrylamides. To a solution of the amino
acid (25.0 mmol) in 60 mL of THF/H2O (pH 8-9, 0 °C) was added
methacryloyl chloride (27.5 mmol) dissolved in 20 mL of THF portion-
wise over 10 min. Potassium hydroxide (2 M, aqueous) was used to
maintain a pH of 8-9 throughout the course of the reaction. The
reaction was allowed to warm to room temperature and stirred for 3 h
before it was neutralized with aqueous HCl and stored at 4 °C
overnight.41,42 THF was removed in vacuo, and the pH was adjusted to
4-5 before the crude product was lyophilized. Salts were removed by
precipitation from ice-cold MeOH. The desalted product was concen-
trated in vacuo to afford a foamy solid which was further purified by
silica gel chromatography. Yields were 30-80%. Unwanted polym-
erization during either the coupling or the workup step contributed to
the lower yields. GlyMA: 1H NMR (500 MHz, D2O) δ 5.68 (1H, d, J
) 1.0 Hz), 5.39 (1H, d, J ) 0.5 Hz), 3.84 (2H, s), 1.84 (3H, s); 13C
NMR (125 MHz, D2O) δ 174.59, 171.58, 138.38, 121.86, 42.08, 17.55;
LRMS FAB+ C6O3NH10 [M + H]+ calcd, 144; found, 144. SerMA:
1H NMR (500 MHz, D2O) δ 5.65 (1H, s), 5.38 (1H, s), 4.37 (1H, dd,
J ) 5.0, 4.5 Hz), 3.83 (1H, dd, J ) 11.5, 5.5 Hz), 3.80 (1H, dd, J )
11.5, 4.5 Hz), 1.83 (3H, s); 13C NMR (125 MHz, D2O) δ 174.40,
171.70, 138.60, 121.68, 61.21, 55.74, 17.57; HRMS FAB+ C7O4NH12
[M + H]+ calcd, 174.0766; found, 174.0768. AspMA: 1H NMR (500
MHz, D2O) δ 5.59 (1H, s), 5.33 (1H, s), 4.48 (1H, dd, J ) 6.5, 5.5
Hz), 2.77 (1H, dd, J ) 16.5, 4.5 Hz), 2.62 (1H, dd, J ) 16.5, 7.5 Hz),
1.79 (3H, s); 13C NMR (125 MHz, D2O) δ 176.33, 175.83, 171.15,
138.71, 121.32, 51.06, 36.97, 17.48; HRMS FAB+ C8O5NH12 [M +
H]+ calcd, 202.0715; found, 202.0716. GluMA: 1H NMR (500 MHz,
D2O) δ 5.59 (1H, d, J ) 0.5 Hz), 5.33 (1H, d, J ) 0.5 Hz), 4.17 (1H,
dd, J ) 9.0, 4.5 Hz), 2.29 (2H, t, J ) 7.5 Hz), 2.06 (1H, m), 1.87 (1H,
m), 1.79 (3H, s); 13C NMR (125 MHz, D2O) δ 178.30, 177.42, 171.39,
138.75, 121.18, 54.18, 31.07, 26.64, 17.52; HRMS FAB+ (NBA)
C9O5NH14 [M + H]+ calcd, 216.0872; found, 216.0877.
Experimental Section
1. Synthesis. 1.1. General Techniques. Flash chromatography was
performed with 60 Å silica gel (Merck, 230-400 mesh). High-pressure
liquid chromatography (HPLC) was performed on a Varian ProStar
210 HPLC system using a preparative Dynamax C18 reversed-phase
(RP) column. NMR spectra were recorded on a Bruker DRX-500
spectrometer. Chemical shifts of 13C NMR in D2O are reported using
dioxane as a reference. Low-resolution electrospray ionization mass
spectrometry (ESI-MS) was performed on a Hewlett-Packard 1100 mass
spectrometer. High-resolution mass spectra (HRMS) were recorded at
the Mass Spectrometry Facility at the University of California at
Berkeley using either fast atom bombardment (FAB) or electrospray
ionization (ESI).
1.4. GRGDMA. Tetrapeptide GRGD was synthesized on Rink
amide-MBHA resin (0.1-0.2 mmol scale), using NR-Fmoc-protected
amino acids and DCC-mediated HOBt ester activation in NMP on a
Perkin-Elmer ABI 431A synthesizer. The peptide was purified by
preparative reversed-phase HPLC in 90% yield: HRMS ESI+ C14O6N8H27
[M + H]+ calcd, 403.2054; found, 403.2057. The N-terminus of the
peptide was then coupled with methacryloyl chloride in aqueous THF
(pH 7-8) at room temperature for 4 h. The product was purified with
reversed-phase HPLC in 89% yield: HRMS ESI+ C18O7N8H31 [M +
H]+ calcd, 471.2316; found, 471.2315.
2. Hydrogel Preparation. HEMA or HEMAm with various percent-
ages (1, 5, or 10 wt %) of functionalized methacrylamides (500 mg
total) were combined with 10 µL of cross-linker EGDMA (when HEMA
was the main monomer) or EGDMAm (when HEMAm was the main
monomer), 100 µL of Milli-Q water, and 150 µL of ethylene glycol.
To this mixture were added 50 µL of aqueous sodium metabisulfite
(150 mg/mL) and 50 µL of aqueous ammonium persulfate (400
mg/mL). The viscous solution was then allowed to polymerize in a
glass chamber and washed as previously described.22
1.2. EGDMAm. To 10 mL of an ice-cold methanolic solution of
ethylenediamine (0.40 mL, 6.0 mmol) was slowly added 1.24 mL (12.6
mmol) of methacryloyl chloride dissolved in 10.0 mL of THF.
Potassium hydroxide (1 M, aqueous) was added to maintain the solution
(32) Addadi, L.; Raz, S.; Weiner, S. AdV. Mater. 2003, 15, 959-970.
(33) Aizenberg, J.; Lambert, G.; Weiner, S.; Addadi, L. J. Am. Chem. Soc. 2002,
124, 32-39.
(34) Boskey, A. L. Connect. Tissue Res. 1996, 35, 357-363.
(35) Merten, H. L.; Bachman, G. L.: U.S. Patent 4,237, 147, 1980.
(36) Han, Y.-J.; Aizenberg, J. Mater. Res. Soc. Symp. Proc. 2004, 823,
W10.15.11-W10.15.15.
3. Equilibrium Water Content (EWC) Measurements. The EWC
at room temperature is defined as the ratio of the weight of water
(37) Grant, M. E.; Prockop, D. J. N. Engl. J. Med. 1972, 286, 194-199.
(38) Engel, J.; Chen, H. T.; Prockop, D. J.; Klump, H. Biopolymers 1977, 16,
601-622.
(39) Glimcher, M. J.; Krane, S. M. In Biology of Collagen; Gould, B. S., Ed.;
Academic Press: London, 1968; Vol. 2B, pp 68-251.
(40) Landis, W. J.; Song, M. J.; Leith, A.; McEwen, L.; McEwen, B. F. J. Struct.
Biol. 1993, 110, 39-54.
(41) Menon, S. K.; Bote, A. N.; Dhoble, D. A.; Rajamohanan, P. R.; Kulkarni,
R. A.; Gundiah, S. Polymer 1992, 33, 2456-2458.
(42) Camail, M.; Margaillan, A.; Thuret, S.; Vernet, J. L. Eur. Polym. J. 1998,
34, 1683-1688.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3371

A R T I C L E S
Song et al.
absorbed by a dry hydrogel to the weight of the fully hydrated hydrogel.
The amount of water absorbed by the hydrogel is determined from the
weight of a freeze-dried gel (Wd) and the weight of the corresponding
hydrated gel (Wh) according to the following equation:
determination of Ca/P ratios of all composite materials was based on
calibration using a standard synthetic HA sample.
5.2. XRD. The presence and overall orientation of crystalline phases
in the precipitated mineral layers were evaluated by XRD with a
Siemens D500 instrument using Cu KR radiation. Phases were identified
by matching the diffraction peaks to the JCPDS files.
EWC (%) ) [(Wh - Wd)/Wh] × 100
5.3. Indentation of Hydrogel-Mineral Composites. The adherence
of the mineral layers attached to the pHEMA-based hydrogel copoly-
mers was qualitatively evaluated by indenting freeze-dried hydrogel-
mineral composites. Loads from 5 to 15 N were applied by a Vickers
indenter for 30 s for each measurement. After indentation, the samples
were analyzed by SEM to check for delamination of mineral layers.
Lack of delamination was interpreted as an indication of strong adhesion
between the mineral and the hydrogel substrate.
4. Formation of Hydrogel-Calcium Phosphate Composites. A
urea-mediated mineralization method22,28 was applied to pHEMA- and
pHEMAm-based copolymers containing various percentages of anionic
ligands. In a typical procedure, a strip of hydrogel copolymer (∼11
mm × 20 mm × 1 mm) was immersed in 15 mL of an acidic aqueous
solution of HA (15 mg/mL, pH 2.5) containing 2 M urea. The solution
was heated without stirring from room temperature to 95 °C at a
constant heating rate of 0.2 °C/min and was maintained at 95 °C for 0
or 10 h. A detailed experimental setup was described previously.28
Acknowledgment. The authors would like to thank Dr.
Eduardo Saiz for technical advice and helpful discussions
regarding SEM and XRD analyses. This work was supported
by the Laboratory Directed Research and Development Program
of Lawrence Berkeley National Laboratory under the Depart-
ment of Energy Contract No. DE-AC03-76SF00098 and the
National Institute of Health Grant No. R01 DE015633-01.
5. Hydrogel-Mineral Composite Characterization. Mineralized
hydrogel strips were repeatedly washed in water to remove loosely
attached minerals and soluble ions before they were freeze-dried for
further structural analyses and mechanical characterization. The surface
microstructures and crystallinity of the minerals grown on the surface
of the hydrogel were analyzed by scanning electron microscopy (SEM)
with associated energy-dispersive spectroscopy (EDS) and X-ray
powder diffraction (XRD).
Supporting Information Available: Evaluation of the bio-
compatibility of pHEMA-based hydrogel copolymers via in vitro
cell culture and X-ray powder diffraction (XRD) of randomly
oriented polycrystalline hydroxyapatite. This material is avail-
able free of charge via the Internet at http://pubs.acs.org.
5.1. SEM-EDS. All SEM micrographs of hydrogel-mineral
composites were obtained with an ISI-DS 13OC dual-stage SEM with
associated EDS. Samples were coated with either Au on a BAL-TEC,
SCD 050 sputter coater to achieve optimal imaging results, or with
carbon for EDS analysis. All images were acquired at 15 kV. EDS
analyses were performed with a 15° tilt unless otherwise specified. The
JA043776Z
9
3372 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005