Apatite-Polymer Composites for the Controlled Delivery of
Bone Morphogenetic Proteins
by
Tseh-Hwan Yong
S.B. Chemical Engineering
Massachusetts Institute of Technology, 1998
Submitted to the Department of Materials Science and Engineering
in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy in Materials Science and Engineering
M
INSIIIIJ-
H
__
MASSACHUSETTSNSTTTE
OF TECHNOLOGY
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUL 2 2 2005
June 2005
LIBRARIES
_
_
_~~~~~~~~~~
©Massachusetts Institute of Technology. All rights reserved.
.
Author:
II I J-
Department of Materials*Science and Engineering
March 16, 2005
Certified by:
Professor Ja'cki-Y. Ying
-Adjunct Professor of Chemical Engineering
Thesis Supervisor
Certified by:
_
- - ---
-- -
-
- -
I
Professor Michael Rubner
Professor of Materials Science and Engineering
Thesis Reader
Accepted by:
~- ---- Fecssor GeibtaftCeder
Professor of Materials Science and Engineering
Chairman, Departmental Committee on Graduate Students
uAIRCHIVE
Apatite-Polymer Composites for the Controlled Delivery of
Bone Morphogenetic Proteins
by
Tseh-Hwan Yong
S.B. Chemical Engineering
Massachusetts Institute of Technology, 1998
Submitted to the Department of Materials Science and Engineering
on March 16, 2005 in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy in Materials Science and Engineering
Abstract
Current treatment of bone defects due to trauma, cancer, or degenerative spine
diseases involves the implantation of a bone graft. Autografts, which are harvested from the
patient's own body, are associated with problems of limited availability and surgical
morbidity. The use of allografts obtained from donors is also not desirable due to the risks of
disease transmission and the costs of maintaining bone banks. The ideal solution would be to
regenerate native bone to fill the defects. A group of potent growth factors known as bone
morphogenetic proteins (BMPs) have been hailed as alternatives to bone grafts due to their
ability to elicit new bone formation. Clinical use of BMPs involves loading the protein
solution onto collagen sponges and subsequent implantation. However, these conventional
collagen carriers show rapid clearance of BMPs within - 2 weeks, whereas bone healing is a
longer process, especially in higher mammals. The poor BMP retention in collagen sponges
may explain the greater response variability in higher mammals, ranging from full bone
bridging within weeks to no bone union. These sponges are also not capable of tunable or
multifactor release that could benefit healing in certain anatomic sites, e.g. avascular sites and
prolonged non-unions. Hence, the motivation of this thesis is to develop new carriers that
allow more efficacious and flexible delivery of BMPs to achieve bone healing.
The carrier should ideally exhibit (i) sustained release to maintain the response and
activity of bone-forming cells, (ii) low initial burst to avoid adverse effects of a bolus
administration anmdto conserve the expensive growth factor, and (iii) tunable release to meter
out BMPs at the desired rate. In particular, tunable release and low burst release have long
been challenges in controlled delivery systems. A carrier that can offer such temporal control
will be highly valuable to the delivery of other therapeutic proteins, drugs and genes as well.
To this end, we have devised a novel composite of two biomaterials with proven track
records: poly(lactic-acid-co-glycolic
acid) (PLGA) and apatite.
The controlled
release
strategy was based on the use of a biodegradable polymer with acidic degradation products to
manipulate the dissolution of the basic apatitic component. Proteins were pre-adsorbed onto
the apatitic component such that as the apatite dissolved, proteins were released. Apatite-
2
PLGA composites were formed into microparticles by a solid-in-oil-in-water emulsion
process. In contrast to polymeric microparticles prepared by the conventional water-in-oil-inwater emulsion process, these composite microparticles exhibited zero-order, low burst
release. Low burst release was attributed to the affinity of the apatite for the protein; until the
apatite was dissolved, the protein was sequestered and prevented from premature release.
Accordingly, the use of apatite singly as a carrier would have led to extremely slow release.
A model protein, bovine serum albumin (BSA), and a therapeutic protein, recombinant
human BMP-2 (rhBMP-2), were encapsulated in these apatite-PLGA composite particles.
The release profile was modified systematically by changing variables that affected polymer
degradation and apatite dissolution, such as polymer molecular weight, polymer
hydrophobicity, apatite loading, and apatite particle size. An increase in polymer molecular
weight, apatite loading or apatite particle size reduced the release rate of both BSA and
rhBMP-2. Interestingly, increasing polymer hydrophobicity diminished the release of BSA,
but enhanced the release of rhBMP-2. Slower polymer degradation associated with greater
polymer hydrophobicity might have decreased the total amount of protein released, but
preserved a larger bioactive fraction due to milder pH conditions within the particles. A
suitable particle formulation for sustained rhBMP-2 delivery was identified as protein-sCAP59 kD PLGA.
When rhBMP-2 was encapsulated in these composite microparticles, it was released in
a sustained fashion over 100 days. More importantly, the bioactivity of the protein was
retained, as evaluated by its ability to induce the differentiation of mesenchymal stem cells
toward the osteoblast lineage. Specifically, the levels of osteoblastic phenotype markers such
as alkaline phosphatase
(ALP) and osteocalcin were found to be significantly
elevated
compared to the controls. In contrast, rhBMP-2 released from conventional collagen sponges
after 2 weeks did not increase the ALP expression over the controls.
Protein-loaded composite microparticles were dispersed in secondary matrices, either
gelatin or collagen sponges, for bone tissue engineering. Multifactor release from these
scaffolds was possible through the incorporation of different sets of composite microparticles
containing different proteins and exhibiting distinct release profiles. Collagen sponges
injected with rhBMP-2-loaded composite microparticles were implanted in subcutaneous sites
in rats. These composite collagen sponges stimulated a much higher degree of cellularity and
vascularity than the controls without BMPs. The increased vascularity might be evidence of
the angiogenic activity of rhBMP-2 at low concentrations in vivo.
Thesis Supervisor:
Jackie Y. Ying
Adjunct Professor of Chemical Engineering
3
Acknowledgments
This thesis would not have been possible without the kind support of my thesis advisor,
Professor Jackie Ying. I would like to thank her for her guidance, mentorship and for
providing the opportunity to pursue a new research direction. I gratefully acknowledge my
thesis committee members, Professors Anne Mayes, Michael Rubner and Myron Spector for
their support, informative discussions, and valuable suggestions. I also thank Professor Paul
Laibinis for his guidance and for providing me with a good grounding in research.
I have truly enjoyed the camaraderie and stimulating environment of the
Nanostructured Materials Research Laboratory. I thank Pemakorn Pitukmanorom, Yee San
Su, Noreen Zaman, and Cindy Ren for their wonderful friendship and support both inside and
outside of the laboratory. I also thank Dr. Suniti Moudgil, Dr. Todd Zion, Dr. Edward Ahn,
Dr. Su Seong Lee, Dr. Jason Sweeney, Dr. Justin McCue, Dr. Neeraj Sangar, Dr. Thomas
Lancaster, Steven Weiss, Jianyi Cui, Dr. Xiaohua Huang, Hong He, and Linda Mousseau for
helpful discussions and many instances of kind assistance. I also acknowledge Pemakorn
Pitukmanorom for help with the nitrogen adsorption analysis and mercury porosimetry.
I was very fortunate to have two bright, capable and enthusiastic UROP students on
the project, Elizabeth (Lizzie) Hager and Anita Kris. Lizzie was involved in the synthesis and
testing of BSA-loaded composite particles. Much of our understanding of gelatin scaffolds
was established through her senior thesis research. Anita was involved in the in vitro and in
vivo testing of BMP-loaded composite collagen sponges.
I thank Professor Clark Colton and Daryl Powers for the use of the cell culture
facilities, Professor Hammond, Kris Stokes and LaRuth McAfee for the use of the GPC, Dr.
Anthony Garratt-Reed and Patrick Boisvert for assistance with electron microscopy at the
CMSE, and the Institute for Soldier Nanotechnologies for the use of the mercury porosimeter
and the Zwick/Roell machine.
I would like to thank the staff at the Institute of Bioengineering and Nanotechnology
in Singapore who showed me much generosity with their time and expertise during my stay in
summer 2004. In particular, I would like to thank Shujun Gao for his guidance and help with
the animal studies.
My heartfelt gratitude goes to my father, mother and sister for their unwavering love,
support and understanding. I also thank my father-, mother-, brother-, and sister-in-law for
their encouragement and constant support. Finally, I thank my wonderful husband, Steve, for
his immense patience, love, emotional (and financial) support, and belief in me. Thank you,
Steve. This thesis is as much your accomplishment as it is mine.
This research was supported by the Singapore-MIT Alliance, and the Institute of
]3ioengineering
and Nanotechnology
(Agency
Singapore).
4
for Science,
Technology
and Research,
Table of Contents
Chapter 1 - Background and Motivation
15
1.1
Bone Regeneration
15
1.2
Bone Morphogenetic Proteins as Alternatives to Bone Grafts
15
1.3
Delivery Systems for Bone Morphogenetic Proteins
17
1.4
Research Objective
19
1.5
Apatite-Polymer Composites for Delivery of BMPs
20
1.6
References
22
Chapter 2 - Synthesis, Characterization, and In Vitro Release Profiles of ApatitePolymer Composite Microparticles for the Controlled Delivery of a Model Protein
31
2.1
31
2.2
2.3
Introduction
2.1.1
Conventional Double Emulsion Processing
31
2.1.2
Solid-in-Oil-in-Water Processing
32
2.1.3
Adsorption of Proteins onto Apatite
32
Experimental
33
2.2.1
Synthesis of Hydroxyapatite and Carbonated Apatite
33
2.2.2
Characterization of Apatite
34
2.2.3
Adsorption of Proteins onto Apatite
34
2.2.4
Effect of pH on the Dissolution of Apatite and the Release of Adsorbed
Proteins
35
2.2.5
Preparation of Composite Microparticles
36
2.2.6
Preparation of PLGA Microspheres by Double Emulsion Processing
37
2.2.7
Characterization of Microparticle Morphology and Size
38
2.2.8
Evaluation of Encapsulation Efficiency of Composite Microparticles
38
2.2.9
Evaluation of In Vitro Protein Release
39
2.2.10 Evaluation of Polymer Degradation
39
2.2.11 Evaluation of Calcium Release from Composite Microparticles
39
Results and Discussion
2.3.1
40
40
Characterization of Apatite
5
2.3.2
Protein Adsorption onto Apatite
42
2.3.3
Effect of pH on Protein Release from Protein-Apatite Complexes
45
2.3.4
Effect of Processing Parameters on Composite Particle Size
47
2.3.5
Encapsulation Efficiency of Protein in Composite Microparticles
49
2.3.6
Effect of Material and Processing Parameters on the In Vitro Release from
Composite Microparticles
50
2.3.6.. 1
Polymer Molecular Weight
50
2.3.6.2
Polymer Hydrophobicity
51
2.3.6.3
Apatite Particle Size
53
2.3.6.4
Addition of Buffering Apatite
54
2.3.6.5
Protein Loading on Apatite
55
2.3.6.6
BSA-Apatite Complex Loading in Particles
56
Comparison of Protein Release from Polymeric Microspheres Prepared by
Double Emulsion Method
57
2.3.8
Degradation of the Polymeric Phase of Composite Microparticles
58
2.3.9
Calcium Release from Composite Microparticles
59
2.3.7
2.4
Summary
61
2.5
References
63
Chapter 3 - Application of Apatite-Polymer Composite Microparticles to the Controlled
Delivery of BMPs
68
3.1
68
Introduction
3.1.1
3.2
Osteoinductive Effect of BMPs
68
69
Experimental
3.2.1
Adsorption of BMPs onto Apatite
69
3.2.2
Aseptic Preparation of rhBMP-2-Loaded Composite Microparticles
70
3.2.3
Estimation of the Magnitude of Initial Burst from Composite Microparticles
71
3.2.4
Evaluation of In Vitro Release
71
3.2.5
Cellular Response
72
3.2.5.1
Effect of Release Medium Collected at Each Time Point
72
3.2.5.2
Effect of Prolonged Exposure to Release Medium
73
6
3.2.5.3
3.3
Statistical Analysis
74
Results; and Discussion
74
3.3.1
Adsorption of rhBMPs onto Apatite
74
3.3.2
Estimation of Proportion of Unbound rhBMPs in Composite Microparticles
75
3.3.3
Effect of Material and Processing Parameters on the In Vitro Release of
rhBMP-2 from Composite Microparticles
75
3.3.3.1
Polymer Molecular Weight
76
3.3.3.2
Apatite Particle Size
76
3.3.3.3
Apatite Loading
77
3.3.3.4
Polymer Hydrophobicity
78
3.3.3.5
Release Medium
79
3.3.4
Effect of Conditioned Medium Collected at Each Time Point
81
3.3.5
Effect of Prolonged Exposure to Conditioned Medium
85
3.3.5.1
Measurement of Induced Alkaline Phosphatase Activity
85
3.3.5.2
Measurement of Osteocalcin Expression
87
3.3.5.3
Evaluation of Cell Proliferation
88
3.4
Summary
90
3.5
References
92
Chapter 4 - Incorporation of Apatite-Polymer Composite Microparticles into Bulk
Matrices
95
4.1
95
Introduction
4.1.1
4.2
Tissue Engineering Scaffolds
96
4.1.1.1
Gelatin Scaffolds
97
4.1.1.2
Collagen Sponges
97
Experimental
97
4.2.1
Preparation of Composite Gelatin Scaffolds
97
4.2.2
Preparation of Composite Collagen Sponges
99
4.2.3
Characterization of Scaffolds and Sponges
99
4.2.3.11 Mechanical Testing
99
7
4.2.3.2
4.2.4
Evaluation of Release Kinetics
100
4.2.4.1
Release of BSA
100
4.2.4.2
Release of rhBMP-2
100
4.2.5
100
Cellular Response
4.2.5..1 Evaluation of Cell Viability by Microscopy
101
4.2.5.2
Evaluation of Cell Proliferation
102
4.2.5.3
Assessment of Osteoblastic Markers
102
4.2.5.4
Statistical Analysis
103
4.2.6
4.3
100
Porosity Measurements
Preliminary In Vivo Testing in a Rat Ectopic Model
Results and Discussion
103
104
4.3.1
Mechanical Properties of Scaffolds
104
4.3.2
Porosity and Pore Size Distribution of Scaffolds
106
4.3.3
In Vitro Release of Proteins from Scaffolds
110
4.3.3.1
Composite Gelatin Scaffolds
110
4.3.3.2
Composite Collagen Sponges
112
4.3.4
114
Cellular Response
4.3.4.1
Evaluation of Cell Viability by Microscopy
114
4.3.4.2
Evaluation of Cell Proliferation
115
4.3.4.3 Measurement of Induced Alkaline Phosphatase Activity
116
4.3.4.4 Measurement of Osteocalcin Expression
118
119
Preliminary In Vivo Testing
4.3.5
4.3.5.1
Evaluation of Dosage Required for Ectopic Bone Formation
124
4.4
Summary
127
4.5
References
129
Chapter 5 - Recommendations for Future Work
134
5.1
Further Investigation of Release Mechanism
134
5.2
Exploration of Additional Material Parameters
135
8
5.3
Incorporation of Composite Microparticles into Tissue Engineering Scaffolds
135
5.4
In Vivo Studies
136
5.5
Use of Composites in Other Shapes and Sizes
136
5.6
Application to Other Therapeutic Agents
136
5.7
References
137
Chapter 6 - Conclusions
139
9
List of Figures
1.1.
A local pharmacokinetic curve showing release rate of radiolabeled BMP-2 from the
rabbit ulna osteotomy site
18
1.2.
Protein is released from the composite microparticle as a result of polymer hydrolysis
leading to dissolution of the apatite substrate
21
2.1.
Schematic of the S/O/W process for preparing composite microparticles.
36
2.2.
Schematic of the W/O/W process for preparing polymeric microparticles.
37
2.3.
XRD patterns of apatites prepared with surfactant (sCAP, sHAP) and without
surfactant (CAP, HAP), and commercial BABI-HAP-SP and BABI-HAP-20N
powders
41
2.4.
FITC-BSA adsorption isotherms of (a) HAP and (b) CAP at room temperature, and (c)
HAP and (b) CAP at 4 0 C
43
2.5.
Adsorption of FITC-BSA onto HAP with time at room temperature
2.6.
Release: of BSA from BSA-HAP and BSA-CAP complexes in buffers of pH 3, 4, 5
and 7.4
47
2.7.
ESEM micrographs of composite microparticles prepared with 59 kD PLGA and (a)
submicron-sized sCAP particles and (b) micron-sized CAP particles
48
2.8.
ESEM micrographs of composite microparticles prepared with CAP and (a) 13 kD
PLGA and (b) 50/50 blend of 13 kD and 24 kD PLGA.
2.9.
44
48
Protein release from composite particles loaded with 15 itgof FITC-BSA per mg of
carrier
51
2.10. Effect of PLA incorporation on protein release
52
2.11.
Effect of PEG incorporation on protein release
53
2.12.
Effect of carbonated apatite particle size on protein release
54
2.13.
Effect of buffering apatite on protein release, expressed as (a) cumulative mass of
BSA released per mg of carrier and (b) cumulative percentage of BSA released
55
2.14.
Effect of protein loading in BSA-CAP complex on protein release
10
56
2.15.
Effect of BSA-CAP complex loading on protein release, expressed as (a) cumulative
mass of BSA released per mg of carrier and (b) cumulative percentage of BSA
released
2.16.
57
Release of (a) BSA from PLGA and PLGA-PLA (2:3) microparticles prepared inhouse by the double emulsion method, and (b) FITC-IgG from PLGA microparticles
blended with 0, 1, 2 and 5 w/w% of PEG
58
2.17.
Release of acid from the degradation of PLGA of different molecular weights in blank
CAP-PLGA composite particles
59
2.18.
Calcium release from microparticles prepared with CAP-13 kD PLGA, CAP-59 kD
PLGA, BSA-CAP-59 kD PLGA, and bare CAP
60
3.1.
Effect of PLGA molecular weight on rhBMP-2 release from composite microparticles
3.2.
Release of rhBMP-2 from composite particles fabricated from 59 kD PLGA, and 1-
76
jtm sCAP and 7-tm CAP
77
3.3.
Effect of sCAP loading on rhBMP-2 release
78
3.4.
Effect of polymer hydrophobicity on rhBMP-2 release
79
3.5.
RhBMP-2 release in BES buffer, complete BME, and serum-free BME, all at pH 7.4
80
3.6.
Change in PLGA molecular weight with time in BES buffer and complete BME, both
at pH 7.4
80
3.7.
Effect of rhBMP-2 concentration on induced ALP activity in C3H1OT1/2cells
81
3.8.
ALP activity induced by release medium collected at each time point from rhBMP-2loaded composite microparticles fabricated from 59 kD PLGA with a sCAP loading of
0.08 mg/mg of PLGA
82
13.9.
ALP activity induced by release medium collected at each time point from rhBMP-2loaded composite microparticles fabricated from 59 kD PLGA with a sCAP loading of
0.1 8 mg/mg PLGA
83
13.10. ALP activity induced by release medium collected at each time point from rhBMP-2loaded composite microparticles fabricated from a 3:2 blend of 59 kD PLGA and 25
kD PLA
3.11.
83
ALP activity induced by release medium collected at each time point from rhBMP-2loaded Hlelistat sponges
84
11
3.12.
Effect of length of exposure to rhBMP-2 on ALP activity
86
3.13.
Effect on ALP activity of prolonged exposure to release media collected from rhBMP86
2-loaded composite microparticles fabricated from 59 kD PLGA
3.14.
Effect of length of exposure to rhBMP-2 on osteocalcin levels
3.15.
Effect on osteocalcin levels of prolonged exposure to release media collected from
88
rhBMP.-2-loaded composite microparticles fabricated from 59 kD PLGA
3.16.
Effect of length of exposure to rhBMP-2 on cell number
3.17.
Effect on cell number of prolonged exposure to release media collected from rhBMP89
2-loaded composite microparticles fabricated from 59 kD PLGA
4.1.
Typical stress-strain curve for composite gelatin scaffolds
4.2.
Compressive moduli of composite gelatin scaffolds with different composite particle
87
89
105
106
loadings
4.3.
Pore size distributions of composite gelatin scaffolds loaded with (a) 0 w/w%, (b) 10
w/w%, (c) 33 w/w% and (d) 43 w/w% of composite microparticles
107
4.4.
ESEM mnicrographsof a composite gelatin scaffold (10 w/w% particles) showing (a) a
pore greater than 200 tm, and (b) composite microparticles dispersed in the scaffold.
108
4.5.
Pore size distributions of composite Helistat® sponges loaded with (a) 0 w/w%, (b) 10
w/w% and (c) 33 w/w% of composite microparticles
109
4.6.
ESEM micrographs of the cross-section of a composite collagen sponge showing (a)
pores greater than 200 pm and (b) composite microparticles held in the sponge
110
4.7.
BSA release from composite particles and composite gelatin scaffolds
111
4.8.
BSA release from composite particles and composite collagen sponge
113
4.9.
RhBMP-2 release from composite particles and composite collagen sponge
113
4.10. Fluorescent cell viability staining of cell-seeded composite collagen sponges at (a) 4
114
days and (b) 7 days
4.11. Number of C3H1OT1/2 cells growing on composite collagen sponges or on the bottom
of wells containing composite collagen sponges
12
115
4.12. ALP activity induced by prolonged exposure to composite collagen sponges containing
composite particles loaded with rhBMP-2, rhBMP-6, rhBMP-2 and rhBMP-6 and no
117
BMP (control)
4.13. Effects of rhBMP-2 and rhBMP-6 concentrations on induced ALP activity in
C3H10['1/2 cells
118
4.14. Osteocalcin levels induced by prolonged exposure to composite collagen sponges
containing composite particles loaded with rhBMP-2, rhBMP-6, rhBMP-2 and rhBMP119
6 and no BMP (control)
4.15. H&E staining of excised composite collagen sponges at 1 week and 2 weeks
122
4.16. H&E staining of excised collagen sponges at 1 week and 2 weeks
123
4.17. Radiographs at 2 weeks and 4 weeks of rats implanted with collagen sponges loaded
with rhBMP-2
of 2 fpg, 5 jtg, 10 pig, 20 pig, and 50 Etg
125
4.18. Radiographs at 2 weeks and 4 weeks of rats implanted with collagen sponges loaded
with rhBMP-6 of 2 pig, 5 ig, 10 pg, 20 tg, and 50 pg
13
126
List of Tables
2.1.
Grain sizes and particle sizes of apatite powders
41
2.2.
BET surface areas, pore volumes and mean pore radii of apatites
42
2.3.
Protein adsorption capacities of apatites
42
2.4.
Adsorption of proteins of different IEP and size onto CAP
44
2.5.
Zeta potential and particle size of apatites and BSA-apatite complexes
45
2.6.
Effects of processing parameters on composite particle size
49
2.7.
Encapsulation efficiency of selected BSA-loaded composite microparticles
50
3.1.
Adsorption of rhBMPs onto 5 mg of CAP
75
4.1.
Porosity of composite gelatin scaffolds and composite collagen sponges
14
107
Chapter 1 - Background and Motivation
1.1
Bone Regeneration
Bone is a naturally regenerative tissue; it is able to heal from fractures and breaks by
recapitulating the embryonic skeletal developmental process. However, an estimated 5-10%
of fractures fail to recover properly and proceed to delayed union or nonunion [1]. The repair
of bone loss associated with trauma and cancer is also typically not observed.
Current
treatment involves the implantation of autogenic or allogenic bone grafts, a procedure that an
estimated 1.5 million patients undergo in the United States each year [2]. Autografts, long
considered the gold standard in bone grafting, are plagued with problems of limited supply,
morbidity associated with graft harvest, and variability in fusion success rate. The use of
allografts is dampened by the risks of disease transmission and the costs of maintaining bone
banks. Synthetic grafts constructed of metals and ceramics are also used, but their mechanical
and chemical incompatibility with bone tissue often leads to implant failure [3]. The ideal
solution would be to regenerate native bone to fill the defects.
A group of potent growth
factors known as bone morphogenetic proteins (BMPs) has the ability to elicit new bone
formation. These proteins provide a promising alternative to bone grafts, and have garnered
much excitement and interest.
1.2
Bone Morphogenetic Proteins as Alternatives to Bone Grafts
BMPs are a class of fourteen cytokines belonging to the transforming growth factor-3
(TGF- [3) superfamily.
They were discovered by Urist in 1965 as the components
in
demineralized bone matrix responsible for inducing ectopic bone and cartilage formation in
muscle and subcutaneous sites of rodents [4, 5]. BMPs induce bone morphogenesis in a
multistep cascade reminiscent of embryonic skeletal formation, beginning with the migration
of mesenchymal cells, their differentiation into cartilage-forming cells (chondrocytes) or
bone-forming cells (osteoblasts), deposition of bone matrix, establishment of functional
marrow within the bone, and culminating in remodeling of the bone consistent with the
anatomic site and biomechanic
environment
[1, 4, 6, 7].
In addition to bone induction
(osteoinduction), BMPs also play a role in the development of other tissues and organs such
as kidney, gut, lung, teeth, heart, limb, and brain [6, 8].
15
Among the osteoinductive BMPs, BMP-2, BMP-4 and BMP-7 appear to have greater
potency [4, 8], and are being produced with high bioactivity and purity via recombinant DNA
technology. In vitro administration of BMP-2 and BMP-7 to embryonic rat calvarial cells, rat
osteosarcoma cells or mouse fibroblasts resulted in enhanced osteogenic activity, as
evidenced by (differentiationinto osteoblasts and elevated expression of bone mineralization
proteins.
In vivo treatment with BMP-2 or BMP-7 augmented the healing of defects in
rodents [9, 10], rabbits [11-15], dogs [16-18], sheep [19], and non-human primates [13, 20,
21]. Critical-size defects, which do not heal spontaneously, were bridged within 3 months in
primates [20, 21]. These animal studies validate the safety and efficacy of BMP-2 and BMP7 in promoting orthopedic repair. However, the results from human trials have shown more
variation. Geesink et al. reported that amongst six patients receiving BMP-7 with a collagen
carrier for the treatment of high tibial osteotomy, four showed bridging by 6 weeks, one by 10
weeks, and one showed no bone formation during the course of the study [22]. When similar
BMP-7 carriers were implanted in the maxillary sinuses of three patients with maxillary
atrophy, only one exhibited satisfactory bone formation, whereas the remaining two showed
little bone formation after 6 months [4]. More recently, 450 patients with open tibial fractures
were randomly assigned to 3 groups receiving (i) the standard of care (tissue irrigation,
d6bridement, followed by intramedullary nail fixation), (ii) standard of care and 6 mg of
rhBMP-2 on a collagen sponge, or (iii) standard of care and 12 mg of rhBMP-2 on a collagen
sponge [23]. One year after treatment, 52% of the patients who received the standard of care
had healed, compared to 59% and 72% of patients who were further treated with 6 mg and 12
-mgof rhBMP-'2, respectively. While the value of BMP treatment was undeniable, the results
from clinical studies were not as impressive as those observed in animal studies where healing
occurred more quickly and more completely. The greater variability and slower response in
humans may be attributed to a smaller population of multipotent cells, which are also less
responsive than,those in smaller animals. It has been proposed that the therapeutic outcomes
may be enhanced by carriers capable of (i) delivering BMPs at a rate that matches the
responsiveness of the cells or (ii) delivering a suitable cocktail of growth factors to stimulate
the cells [24, 25;].
16
1.3
Delivery Systems for Bone Morphogenetic Proteins
In order for BMPs to exert an effect on bone healing, they must be delivered to the
defect site. Bolus injections have limited effect because of the rapid clearance of exogenous
proteins from the body. To increase retention of BMPs at the defect site, a carrier is needed
that has desirable release kinetics. Furthermore, if the volume of bone to be regenerated is
large, the carrier has to be combined with a scaffold or serve as a scaffold itself to allow for
the cell migration and growth, and the deposition of extracellular matrix [25].
Delivery systems for BMPs have been constructed from a variety of materials, which
can be categorized as natural polymers, synthetic polymers, inorganic materials, and
composites of the above.
Of the natural polymers, collagen is the most widely used [9, 21, 26, 27] and was the
first to be employed commercially for BMP delivery. Demineralized bone matrix (DBM), the
material from which Urist originally extracted BMPs, is also commonly used [28, 29] as the
residual levels of BMPs, matrix proteins and calcium in DBM could enhance osteoinductivity.
Other natural polymers such as gelatin [30-33], hyaluronic acid [34], chitosan [35] and
alginate [36, 37] have also been tested. Potential issues with naturally derived materials
include batch variations in purity and quality, as well as the risk of disease transmission.
Biodegradable and/or biocompatible synthetic polymers, such as poly(oc-hydroxy
acids) [38-41], poly(ortho esters) [42], polypropylene fumarate [43] and polyethylene glycol
[44], offer benefits of reproducible manufacture and readily tailored functionality.
In
particular, poly(cc-hydroxy acids) comprising lactic acid and glycolic acid are approved by the
Federal Drug Administration (FDA), and have been used as suture material since the 1970s.
Biodegradable polymers are designed to decompose to non-toxic products in physiological
environments. However, some degradation products, such as those from poly(u-hydroxy
acids) and polyanhydrides, are acidic and may cause tissue inflammation.
Hydroxyapatite
(HAP) [19, 45, 46], calcium phosphate
[45, 47, 48], silica [49],
bioglass [50, 51] and titanium [52, 53] are some of the inorganic materials that have been
used as BMP carriers. The most commonly used is HAP because of its similarity to the
mineral component of bone and its osteoconductivity. The osteoconductivity of HAP and
calcium phosphate is sometimes attributed to their ability to concentrate growth factors,
including BMPs, in the body [54]. The release of proteins from ceramics can be very slow
17
and sustained because of the high affinity of certain ceramics for proteins [54-58].
Drawbacks of ceramics include their brittleness and difficulty in processing.
A fruitful combination of the above three classes of materials as composites would
allow the harnessing of the benefits of each component. Furthermore, the drawbacks of one
material may be countered by another. For example, HAP and calcium phosphate can act as a
buffer against the acidity of the degradation products from poly(oc-hydroxy acids) [59, 60].
Composites have been produced with enhanced mechanical and handling properties [61-66],
improved bioactivity [67-69], optimized biodegradation [59, 60, 70], and microstructures that
more closely resemble natural tissues [71, 72]. Some composites that have been explored for
BMP delivery include HAP-collagen [17, 26, 73, 74], poly(lactic acid-co-glycolic acid)
(PLGA)-calcium phosphate [75], PLGA-cellulose [76, 77], and PLGA-gelatin [14].
1U
Ll ,nn
r 4 Thu -
. Q(C0
0
4:
+
50 -
4020-
4..,
r
-.
a~~~~~~~<
-
10
2
4
!3
8
'10
t2
14
'16
Days post-treatrent
Fig. 1.1. A local pharmacokinetic curve showing release rate of radio-labeled BMP-2 from
the rabbit ulna osteotomy site [25]. The various carriers shown were either implanted (*
hyaluronic acid pad and A collagen sponge) or injected ( Gelfoam paste and * buffer) into
the defect. Reprinted from [25] with permission from Elsevier.
Many of these materials are transformed into BMP carriers by simply mixing in the
proteins during processing or by soaking pre-fabricated carriers in the protein solution.
Variations in release kinetics amongst the carriers are observed due to inherent differences in
the affinity of the materials for BMPs and in the carrier dimensions (Fig. 1.1). However,
adjustment of BMP release kinetics with such carriers in order to attain the optimal profile is
difficult and has not been accomplished. The optimal BMP release profile may vary with
18
animal species, age of host, anatomic site, wound history and other factors [25]. For example,
slower release rates may be required in more fluid environments where BMP clearance may
be faster, and in more compromised sites where the healing response is diminished.
Tunable
carriers would offer greater potential and flexibility in realizing the desired release profile.
Two products containing BMPs have recently been approved by FDA for spinal fusion:
Stryker's OP-I comprising BMP-7 and Medtronic Sofamor Danek's INFUSE containing
BMP-2. Both products use collagen sponges as the carrier for BMPs; the sponges are loaded
by soaking with a BMP solution for 10-20 min. The BMP loading in these sponges is on the
order of milligrams per implant, which is several orders of magnitude above the natural
occurrence in bone (- 0.002 mg of BMP-2 can be extracted per kg of powdered bone [78]).
The release of BMPs from collagen tends to be rapid: 70-90% of the load is depleted by the
first week [29., 34, 79]. However, bone healing is often a much longer process requiring
weeks or months, especially in higher mammals with less responsive cells. A possible reason
for the use of supra-physiological
BMP doses in collagen carriers is the need to overcome the
low availability of BMPs at later stages of healing [25] since cellular response to BMPs has
been found to increase with dosage as well as time of exposure [77, 80]. Therefore, carriers
that are capable of sustained release of lower but still therapeutic levels of BMPs would allow
greater efficacy and cost-savings by optimizing the use of these expensive proteins. The
presence of BMPs over the entire duration of healing in higher mammals may also reduce the
variation in response (see Section 1.2), and enhance the therapeutic outcomes. Furthermore,
tunable release or multifactor release at different rates, which current collagen sponge carriers
cannot accomplish, may also augment the bone healing response.
1.4
Research Objective
Certain anatomic sites and certain indications, e.g. prolonged non-unions, may require
BMP release kinetics that current collagen sponge carriers cannot provide. The objective of
our research is to improve the efficacy of BMP delivery for bone healing by developing
carriers that can retain and meter out BMPs at the appropriate dose and for a sufficient
duration to achieve the desired host response. Ideal release characteristics of such carriers
include:
19
1. Sustained release - to make BMPs available over the entire period of healing, possibly
lasting for several weeks to several months.
2. Low burst release - to avoid the adverse effects of an overdose, and to conserve the
expensive therapeutic proteins, making the formulation more cost-effective.
3. Tunable release - to allow flexibility in designing release rates for different therapeutic
proteins, host species, and anatomic sites.
Other desirable properties of the carrier include biocompatibility, biodegradability, bioactivity,
ease of manufacture, and surgical practicability. Such a carrier would also have broader
applications in the delivery of other therapeutic proteins and drugs.
To achieve the research objective, the following approach was taken:
* Development of a biocompatible, biodegradable carrier capable of tunable, sustained,
and low-burst release (Chapter 2)
* Application of the carrier to the delivery of BMPs (Chapter 3)
·
In vitro testing to establish bioactivity of released BMPs (Chapter 3)
* Incorporation of BMP carriers into scaffolds capable of supporting in vivo cell migration
and growth (Chapter 4)
·
1.5
Ascertaining the safety and efficacy of these scaffolds in vitro and in vivo (Chapter 4)
Apatite-Polymer Composites for Delivery of BMPs
Our strategy for a tunable controlled delivery platform is based on the use of a
polymer with acidic degradation products to control the dissolution of a basic inorganic
substrate on which BMPs have been adsorbed. The release of acid from the hydrolysis of the
polymer leads to the dissolution of the inorganic substrate and consequently, the desorption
and release of the protein. Hence, protein release is anticipated to be accelerated over passive
leaching from the inorganic substrate, yet, more controlled than the release from polymeric
microspheres due to the affinity of the substrate for the protein. The release mechanism is
depicted in Fig. 1.2.
Systematic modulation of the release profile may be possible by changing variables
that affect polymer degradation, subsequent acid generation, and/or inorganic substrate
dissolution.
These variables include polymer type, polymer molecular weight, polymer
composition (including copolymers and blends), inorganic substrate type, inorganic substrate
20
loading, inorganic substrate particle size, relative proportion of polymer and inorganic
substrate, and protein loading on the inorganic substrate. This delivery platform can be
applied to any therapeutic agent that can be adsorbed and sequestered on the surface of a basic
inorganic material.
-
O
e'
°e
water
O.
composite microparticle
Fig. 1.2.
o
protein
O
protein adsorbed on apatite
Protein is released from the composite microparticle as a result of polymer
hydrolysis that leads to dissolution of the apatite substrate.
Candidates for the polymeric component include polyanhydrides, poly(oc-hydroxy
acids) and poly(ortho esters), all of which degrade to produce acids. Poly(c-hydroxy
acids)
comprising lactic acid (LA) and glycolic acid (GA) - PLGA, PLA, and PGA - are attractive
candidates because they are FDA-approved, and are commercially available in a wide range
of molecular weights. Furthermore, poly(oc-hydroxyacids) are bulk eroding polymers, which
may be more suitable for this controlled release mechanism as they allow the buildup of acid
within composite particles necessary for dissolving the apatitic phase. On the other hand,
acidic degradation products from surface eroding polymers, e.g. polyanhydrides, may diffuse
away too quickly
to encounter
the apatitic phase.
The pH within degrading
PLGA
microspheres has been found to be less than 4.7 [81] and as low as 1.5 [82]. Such a pH range
should be sufficient to cause the dissolution of a basic material.
A number of basic inorganic materials, such as HAP, carbonated apatite (CAP),
calcium phosphate and calcium carbonate, exhibit bioactivity and enhanced integration with
host bony tissue [83-88].
These materials are also employed as fillers to enhance the
mechanical properties of the softer natural and synthetic polymers [61, 62, 64]. In addition,
they have been used to alleviate the acidity created by the degradation of poly(oc-hydroxy
21
acids) [59, 60, 89, 90], which is relevant to our current application.
The high affinity of HAP
for proteins (see Section 1.3) suggests that HAP should be capable of holding BMPs on its
surface without premature release. Nanocrystalline HAP and CAP, the syntheses of which
were previously developed in our laboratory [91-93], were selected as the inorganic phase of
the composites.
CAP differs from HAP (Cal0(PO4) 6(OH) 2 ) in that it has carbonate ions
substituted in the hydroxyl or phosphate sites in the crystal lattice [94].
CAP is more
amorphous, more resorbable, and closer in composition to natural bone mineral.
The dimensions of the carrier would depend on the method of administration, defect
site to fill, and impact on release kinetics.
The apatite-polymer composite carriers may be
prepared as particles, which can be delivered by injection into the bone defect, compressed
into pellets for implantation, or dispersed in a secondary matrix that can be formed into tissue
engineering scaffolds. In addition, the composites may be fabricated as films or porous, bulk
scaffolds.
1.6
References
[1]
Wozney JM. Overview of bone morphogenetic proteins. Spine 2002;27:S2-S8.
[2]
Deutsche Bank Alex. Brown Report; 2001.
[3]
Lane JM, Tomin E, Bostrom MPG. Biosynthetic bone grafting. Clin Orthop Rel Res
1999;367:S1 07-S117.
[4]
Groeneveld EHJ, Burger EH. Bone morphogenetic proteins in human bone
regeneration. Eur J Endocrinol 2000; 142:9-21.
[5]
Urist MR. Bone: Formation by autoinduction. Science 1965;150:893-899.
[6]
Balemans W, Hul WV. Extracellular regulation of BMP signaling in vertebrates: a
cocktail of modulators. Dev Biol 2002;250:231-250.
1[7]
Kloen P', Doty SB, Gordon E, Rubel IF, Goumans M-J, Helfet DL. Expression and
activation of the BMP-signaling components in human fracture non-unions. J Bone Joint Surg
[Am] 2002;84A: 1909-1918.
1-8]
Hoffman A, Weich HA, Gross G, Hillmann G. Perspectives in the biological function,
the technical and therapeutic application of bone morphogenetic proteins. Appl Microbiol
Biotech 2001 ;57:294-308.
22
[9]
Lutolf MP, Weber FE, Schmoekel HG, Schense JC, Kohler T, Muller R, Hubbell JA.
Repair of bone defects using synthetic mimetics of collagenous extracellular matrices. Nat
Biotechnol 2003;21:513-518.
[10]
Saito N, Okada T, Horiuchi H, Narumichi M, Takahashi J, Nawata M, Ota H, Nozaki
K, Takaoka K. A biodegradable polymer as a cytokine delivery system for inducing bone
formation. Nat Biotechnol 2001;19:332-335.
[11]
Koempel
recombinant
JA, Patt BS, O'Grady K, Wozney
JM, Toriumi DM. The effect of
human bone morphogenetic protein-2
on the integration of porous
hydroxyapatite implants with bone. J Biomed Mater Res 1998;41:359-363.
[12]
Kokubo S, Fujimoto R, Yokota S, Fukushima S, Nozaki K, Takahashi K, Miyata K.
Bone regeneration by recombinant human bone morphogenetic protein-2 and a novel
biodegradable carrier in a rabbit ulnar defect model. Biomaterials 2003;24:1643-1651.
[13]
Suh DY, Boden SD, Louis-Ugbo J, Mayr M, Murakami H, Kim H-S, Minamide A,
Hutton WC. Delivery of recombinant human bone morphogenetic protein-2 using a
compression-resistant
matrix in posterolateral spine fusion in the rabbit and in the non-human
primate. Spine 2002;27:353-360.
[14]
Ueki K, Takazakura
D, Marukawa K, Shimada M, Nakagawa K, Takatsuka
S,
Yamamoto E. The use of polylactic acid/polyglycolic acid copolymer and gelatin sponge
complex containing
human
recombinant bone morphogenetic
protein-2
following
condylectomy in rabbits. J Crano-Maxillofac Surg 2003;31:107-114.
[15]
Zegzula HD, Buck DC, Brekke J, Wozney JM, Hollinger JO. Bone formation with use
of rhBMP-2. J Bone Joint Surg [Am] 1997;79A: 1778-1790.
[16]
Cullinane
DM, Lietman
SA, Inoue N, Deitz LW, Chao EYS. The effect of
recombinant human osteogenic protein-1 (bone morphogenetic protein-7) impregnation on
allografts in a canine intercalary bone defect. J Orthop Res 2002;20:1240-1245.
1-17]
Itoh S, Kikuchi M, Takakuda K, Nagaoka K, Koyama Y, Tanaka J, Shinomiya K.
Implantation study of a novel hydroxyapatite/collagen (HAp/Col) composite into weightbearing sites of dogs. J Biomed Mater Res 2002;63:507-515.
[-18]
Muraka:ni N, Saito N, Takahashi J, Ota H, Horiuchi H, Nawata M, Okada T, Nozaki
K, Takaoka K. Repair of a proximal femoral bone defect in dogs using a porous surfaced
23
prosthesis in combination with recombinant BMP-2 and a synthetic polymer carrier.
Biomaterials 2003;24:2153-2159.
[19]
den Boer FC, Wippermann BW, Blokhuis TJ, Patka P, Bakker FC, Haarman HJTM.
Healing of segmental bone defects with granular porous hydroxyapatite augmented with
recombinant human osteogenic protein-1 or autologous bone marrow. J Orthop Res
2003;21:521-528.
[20]
Ripamonti U, Ramoshebi LN, Matsaba T, Tasker J, Crooks J, Teare J. Bone induction
by BMPs/OPs and related family members in primates. J Bone Joint Surg [Am] 2001;83A:SI-116.
[21]
Cook SD, Wolfe MW, Salkfeld SL, Rueger DC. Effect of recombinant human
osteogenic protein-1 on healing of segmental defects in non-human primates. J Bone Joint
Surg [Am] 1995;77A:734-750.
[22]
Geesink RGT, Hoefnagels NHM, Bulstra SK. Osteogenic activity of OP-1 bone
morphogenetic protein (BMP-7) in a human fibular defect. J Bone Joint Surg [Br]
1999;81B:710-718.
[23]
Govender S, Csimma C, Genant HK, Valentin-Opran A. Recombinant human bone
morphogenetic protein-2 for treatment of open tibial fractures - A prospective, controlled,
randomized study of four hundred and fifty patients. J Bone Joint Surg [Am] 2002;84A:21232134.
[241
Einhorn TA. Clinical applications of recombinant human BMPs: Early experience and
future development. J Bone Joint Surg [Am] 2003;85A:82-88.
[25-1 Li RH, Wozney JM. Delivering on the promise of bone morphogenetic proteins.
'Trends Biotechnol 2001;19:255-265.
[26]
Takaoka K, Nakahara H, Yoshikawa H, Masuhara K, Tsuda T, Ono K. Ectopic bone
induction on and in porous hydroxyapatite combined with collagen and bone morphogenetic
protein. Clin Orthop Rel Res 1988;234:250-254.
[27]
Geiger M, Li RH, Friess W. Collagen sponges for bone regeneration with rhBMP-2.
Adv Drug Deliv,Rev 2003;55:1613-1629.
128]
Uludag H, D'Augusta D, Palmer R, Timony G, Wozney J. Characterization of rhBMP-
2 pharmacokinetics implanted with biomaterial carriers in the rat ectopic model. J Biomed
Mater Res 1999;46:193-202.
24
[29]
Uludag H, D'Augusta
D, Golden J, Li J, Timony G, Riedel R, Wozney
JM.
Implantation of recombinant human bone morphogenetic proteins with biomaterial carriers: A
correlation between protein pharmacokinetics and osteoinduction in the rat ectopic model. J
Biomed Mater Res 2000;50:227-238.
[30]
Holland TA, Tabata Y, Mikos AG. In vitro release of transforming growth factor-31
from gelatin microparticles encapsulated in biodegradable, injectable oligo(poly(ethylene
glycol) fumarate) hydrogels. J Control Release 2003;91:299-313.
[31]
Raiche AT, Puleo DA. In vitro effects of combined and sequential delivery of two
bone growth factors. Biomaterials 2004;25:677-685.
[32]
Tabata Y, Yamada K, Miyamoto S, Nagata I, Kikuchi H, Aoyama I, Tamura M, Ikada
Y. Bone regeneration by basic fibroblast growth factor complexed with biodegradable
hydrogels. Biomaterials 1998;19:807-815.
[33]
Yamamoto M, Takahashi Y, Tabata Y. Controlled release by biodegradable hydrogels
enhances the ectopic bone formation of bone morphogenetic protein. Biomaterials
2003;24:4375-4388.
[34]
Kim HI), Valentini RF. Retention and activity of BMP-2 in hyaluronic acid-based
scaffolds in vitro. J Biomed Mater Res 2002;59:573-584.
[35]
Lee J-Y, Nam S-H, Im S-Y, Park Y-J, Lee Y-M, Seol Y-J, Chung C-P, Lee S-J.
Enhanced bone formation by controlled growth factor delivery from chitosan-based
biomaterials. J Control Release 2002;78:187-197.
1-36]
Saito A, Suzuki Y, Ogata S, Ohtsuki C, Tanihara M. Prolonged ectopic calcification
induced by BMP-2-derived synthetic peptide. J Biomed Mater Res 2004;70A: 115-121.
[137] Simmons CA, Alsberg E, Hsiong S, Kim WJ, Mooney DJ. Dual growth factor
delivery and controlled scaffold degradation enhance in vivo bone formation by transplanted
bone marrow stromal cells. Bone 2004;35:562-569.
[38]
Andriano KP, Chandrashekar B, McEnery K, Dunn RL, Moyer K, Balliu CM, Holland
KM, Garrett S, Huffer WE. Preliminary in vivo studies on the osteogenic potential of bone
morphogenetic proteins delivered from an absorbable puttylike polymer matrix. J Biomed
Mater Res 2000.;53:36-43.
25
[39]
Bessho K, Cames DL, Cavin R, Ong JL. Experimental studies on bone induction
using low-molecular-weight poly(DL-lactide-co-glycolide) as a carrier for recombinant
human bone morphogenetic protein-2. J Biomed Mater Res 2002;61:62-65.
[40]
Duggirala SS, Mehta RC, DeLuca PP. Interaction of recombinant human bone
morphogenetic protein-2 with poly(d,l lactide-co-glycolide) microspheres. Pharm Dev
Technol 1996; 1:11-19.
[41]
Meikle MC, Mak W-Y, Papaioannou S, Davies EH, Mordan N, Reynolds JJ. Bone-
derived growth factor release from poly(ac-hydroxy acid) implants in vitro. Biomaterials
1993;14:177-183.
[42]
Rai B, Teoh SH, Ho KH, Hutmacher DW, Cao T, Chen F, Yacob K. The effect of
rhBMP-2 on canine osteoblasts seeded onto 3D bioactive polycaprolactone scaffolds.
Biomaterials 2004;25:5499-5506.
[43]
Vehof JWM, Fisher JP, Dean D, van der Waerden JCM, Spauwen PHM, Mikos AG,
Jansen JA. Bone formation in transforming growth factor 3-1-coated porous poly(propylene
fumarate) scaffolds. J Biomed Mater Res 2002;60:241-251.
[44]
Burdick. JA, Mason
MN, Hinman
AD,
Thorne
K, Anseth
KS. Delivery
of
osteoinductive growth factors from degradable PEG hydrogels influences osteoblast
differentiation anmd
mineralization. J Control Release 2002;83:53-63.
[45]
Alam MI, Asahina I, Ohmamiuda K, Takahashi K, Yokota S, Enomoto S. Evaluation
of ceramics composed of different hydroxyapatite to tricalcium phosphate ratios as carriers
for rhBMP-2. Biomaterials 2001 ;22:1643-1651.
[46]
Yuan fH, de Bruijn JD, Zhang X, van Blitterswijk
CA, de Groot K. Use of an
osteoinductive biomaterial as a bone morphogenetic protein carrier. J Mater Sci Mater Med
2001 ;12:761-766.
1[47] Niedhart C, Maus U, Redmann E, Schmidt-Rohlfing B, Niethard FU, Siebert CH.
Stimulation of' bone formation with an in situ setting tricalcium phosphate/rhBMP-2
composite in rats. J Biomed Mater Res 2003;65A:17-23.
1-48]
Ruhe PQ, Kroese-Deutman
HC, Wolke JGC, Spauwen PHM, Jansen JA. Bone
inductive properties of rhBMP-2 loaded porous calcium phosphate cement implants in cranial
defects in rabbits. Biomaterials 2004;25:2123-2132.
26
[49]
El-Ghannam A, Ning CQ, Mehta J. Cyclosilicate nanocomposite: A novel resorbable
bioactive tissue engineering scaffold for BMP and bone-marrow cell delivery. J Biomed
Mater Res 2004;71A:377-390.
[50]
Nicoll SB, Radin S, Santos EM, Tuan RS, Ducheyne P. In vitro release kinetics of
biologically active transforming growth factor-31 from a novel porous glass carrier.
Biomaterials 1997;18:853-859.
[51]
Mahmood J, Takita H, Ojima Y, Kobayashi M, Kohgo T, Kuboki Y. Geometric effect
of matrix upon cell differentiation: BMP-induced osteogenesis using a new bioglass with a
feasible structure. J Biochem 2001; 129:163-171.
[52]
Puleo IDA, Kissling RA, Sheu M-S. A technique to immobilize bioactive proteins,
including bone morphogenetic protein-4, on titanium alloy. Biomaterials 2002;23:2079-2087.
[53]
Schmidmaier G, Wildemann B, Cromme F, Kandziora F, Haas NP, Raschke M. Bone
morphogenetic protein-2 coating of titanium implants increase biomechanical strength and
accelerates bone remodeling in fracture treatment: A biomechanical and histological study in
rats. Bone 2002;30:816-822.
[54]
Yuan H, Zou P, Yang Z, Zhang X, De Bruijn JD, de Groot K. Bone morphogenetic
protein and ceramic-induced osteogenesis. J Mater Sci Mater Med 1998;9:717-721.
[55]
Barralet JE, Aldred S, Wright AJ, Coombes AGA. In vitro behavior of albumin-loaded
carbonate hydroxyapatite gel. J Biomed Mater Res 2002;60:360-367.
[561] Combes C, Rey C. Adsorption of proteins and calcium phosphate materials bioactivity.
Biomaterials 2002;23:2817-2823.
[57]
Gautier H, Guicheux J, Grimandi G, Faivre-Chauvet A, Daculsi G, Merle C. In vitro
influence of apatite-granule-specific area on human growth hormone loading and release. J
Biomed Mater Res 1998;40:606-613.
[58]
Ziegler J, Mayr-Wohlfart
U, Kessler S, Breitig D, Gunther K-P. Adsorption and
release properties of growth factors from biodegradable implants. J Biomed Mater Res
2002;59:422-428.
[159] Ara M, Watanabe M, Imai Y. Effect of blending calcium compounds on hydrolytic
degradation of poly(DL-lactic acid-co-glycolic acid). Biomaterials 2002;23:2479-2483.
27
[60]
Linhart W, Peters F, Lehmann W, Schwarz K, Schilling AF, Amling M, Rueger JM,
Epple M. Biologically and chemically optimized composites of carbonated apatite and
polyglycolide as bone substitution materials. J Biomed Mater Res 2001;54:162.
[61]
Balac
, Uskokovic PS, Aleksic R, Uskokovic D. Predictive modeling of the
mechanical properties of particulate hydroxyapatite reinforced polymer composites. J Biomed
Mater Res (Appl Biomater) 2002;63:793-799.
[62]
Durucan C, Brown PW. Calcium-deficient hydroxyapatite-PLGA composites:
Mechanical and microstructural investigation. J Biomed Mater Res 2000;51:726-734.
[63]
Kokubo T, Kim H-M, Kawashita M. Novel bioactive materials with different
mechanical properties. Biomaterials 2003;24:2161-2175.
[64]
Liu Q. de Wijn
JR, van Blitterswijk
CA. Nano-apatite/polymer
composites:
mechanical and physicochemical characteristics. Biomaterials 1997;18:1263-1270.
[65]
Liu Q, de Wijn JR, van Blitterswijk CA. Composite biomaterials with chemical
bonding between hydroxyapatite filler particles and PEG/PBT copolymer matrix. J Biomed
Mater Res 1998;40:490-497.
[661] Takagi S, Chow LC, Hirayama S, Eichmiller FC. Properties of elastomeric calcium
phosphate cement-chitosan composites. Dent Mater 2003;19:797-804.
[67]
Moursi AM, Winnard AV, Winnard PL, Lannutti JJ, Seghi RR. Enhanced osteoblast
response to a polymethylmethacrylate-hydroxyapatite composite. Biomaterials 2002;23:133144.
[68]
Ohtsuki C, Miyazaki T, Tanihara M. Development of bioactive organic-inorganic
hybrid for bone substitutes. Mater Sci Eng C 2002;22:27-34.
[69]
Rizzi SC, Heath DJ, Coombes AGA, Bock N, Textor M, Downes S. Biodegradable
polymer/hydroxyapatite composites: surface analysis and initial attachment of human
osteoblasts. J Biomed Mater Res 2001;55:475-486.
[70]
Matsumoto T, Okazaki M, Inoue M, Ode S, Chang-Chien C, Nakao H, Hamada Y,
Takahashi J. Biodegradation of carbonate apatite/collagen composite membrane and its
controlled release of carbonate apatite. J Biomed Mater Res 2002;60:651-656.
1-71] Estroff LA, Hamilton AD. At the interface of organic and inorganic chemistry:
Blioinspired synthesis of composite materials. Chem Mater 2001;13:3227-3235.
28
[72]
Itoh S, Kikuchi M, Koyama Y, Takakuda K, Shinomiya K, Tanaka J. Development of
an artificial vertebral body using a novel biomaterial, hydroxyapatite/collagen composite.
Biomaterials 21)02;23:3919-3926.
[73]
Louis-Ugbo J, Kim H-S, Boden SD, Mayr MT, Li RC, Seeherman H, D'Augusta D,
Blake C, Jiao A, Peckham S. Retention of
125I-labeled
recombinant human bone
morphogenetic protein-2 by biphasic calcium phosphate or a composite sponge in a rabbit
posterolateral spine arthrodesis model. J Orthop Res 2002;20:1050-1059.
[74]
Wang Y-J, Lin F-H, Sun J-S, Huang Y-C, Chueh S-C, Hsu F-Y. Collagen-
hydroxyapatite microspheres as carriers for bone morphogenetic protein-4. Artif Organs
2003;27:162-168.
[75]
Ruhe PQ, Hedberg EL, Padron NT, Spauwen PHM, Jansen JA, Mikos AG. rhBMP-2
release
from
injectable
poly(DL-lactic-co-glycolic
acid)/calcium-phosphate
cement
composites. J Bone Joint Surg [Am] 2003;85A:75-81.
[76]
Schrier JA, Fink BF, Rodgers JB, Vasconez HC, DeLuca PP. Effect of a freeze-dried
CMC/PLGA microsphere matrix of rh-BMP-2 on bone healing. AAPS Pharm Sci Tech
2001;2:article
[77]
118.
Woo B-H, Fink BF, Page R, Schrier JA, Jo YW, Jiang G, DeLuca M, Vasconez HC,
DeLuca PP. Enhancement of bone growth by sustained delivery of recombinant human bone
morphogenetic protein-2 in a polymer matrix. Pharm Res 2001;18:1747-1753.
[78]
Seeherman H, Wozney J, Li R. Bone morphogenetic protein delivery systems. Spine
2002;27:S 16-S23.
[79]
Uludag H, Gao T, Porter TJ, Friess W, Wozney JM. Delivery systems for BMPs:
factors contributing
to protein retention at an application site. J Bone Joint Surg [Am]
:2001;83A: S1-128.
[80]
Puleo DA. Dependence of mesenchymal cell responses on duration of exposure to
bone morphogenetic protein-2 in vitro. J Cell Physiol 1997;173:93-101.
1[81] Brunner A, Mader K, Gopferich A. pH and osmotic pressure inside biodegradable
microspheres during erosion. Pharm Res 1999;16:847-853.
182]
within
Fu K, Pack DW, Klibanov AM, Langer RS. Visual evidence of acidic environment
degrading
poly(lactic-co-glycolic
acid)
2000; 17: 100-106.
29
(PLGA) microspheres.
Pharm
Res
[83]
Ben-Nissan B. Natural bioceramics: from coral to bone and beyond. Curr Opin Solid
State Mat Sci 2003;7:283-288.
[84]
Damien E, Hing K, Saeed S, Revell PA. A preliminary study on the enhancement of
the osteointegration of a novel synthetic hydroxyapatite scaffold in vivo. J Biomed Mater Res
2003;66A:241.-.246.
[85]
Barralet JE, Grover L, Gaunt T, Wright AJ, Gibson IR. Preparation of macroporous
calcium phosphate cement tissue engineering scaffold. Biomaterials 2002;23:3063-3072.
[86]
Landi
, Celotti G, Logroscino G, Tampieri A. Carbonated hydroxyapatite as bone
substitute. J European Ceram Soc 2003;23:2931-2937.
[87]
Frankenburg EP, Goldstein SA, Bauer TW, Harris SA, Poser RD. Biomechanical and
histological evaluation of a calcium phosphate cement. J Bone Joint Surg [Am]
1998;80A: I I 12-1124.
[88]
Kim H-.M. Ceramic bioactivity and related biomimetic strategy. Curr Opin Solid State
Mat Sci 2003;7:289-299.
[89]
Schiller C, Epple M. Carbonated calcium phosphates are suitable pH-stabilising fillers
for biodegradable polyesters. Biomaterials 2003;24:2037-2043.
[90]
Tsunoda M. Degradation of poly (DL-lactic acid-co-glycolic acid) containing calcium
carbonate and hydroxyapatite fillers - Effect of size and shape of the fillers. Dent Mater J
2003;22:371-382.
[91]
Ahn
ES. Nanostructured
Apatites
as Orthopedic
Biomaterials
[Ph.D.
Thesis].
Cambridge, MA: Massachusetts Institute of Technology; 2001.
[92]
Ahn ES, Gleason NJ, Nakahira A, Ying JY. Nanostructure processing of
hydroxyapatite-based
[93]
bioceramics. Nano Lett 2001; 1:149-153.
Ying JY, Ahn ES, Nakahira A. Nanocrystalline apatites and composites, prostheses
incorporating them, and method for their production. U.S. Patent Number 6,013,591, 2000.
[94]
Fleet ME, Liu X. Carbonate apatite type A synthesized at high pressure: new space
group and orientation of channel carbonate ion. J Solid State Chem 2003;174:412-417.
30
Chapter 2 - Synthesis, Characterization, and In Vitro Release Profiles of ApatitePolymer Composite Microparticles for the Controlled Delivery of a Model Protein
2.1
Introduction
Proteins experience short half-lives in vivo on the order of minutes or hours due to
enzymatic degradation, evisceration through the reticulo-endothelial system, immunological
inactivation, and other pathways [1-3].
The encapsulation of proteins in microparticles
protects the proteins from their external environment and slows their release [4]. Methods of
preparing microparticles loaded with water-soluble drugs and proteins include water-in-oil-inwater emulsions [5-9], coacervation [10-12], and spray drying [10, 13-15].
The
microparticles are typically constructed of polymers; some examples are biodegradable
synthetic polymers such as poly(lactic-co-glycolic acid) (PLGA) and polycaprolactone, and
natural polymers such as chitosan and gelatin.
Formulating the protein delivery vehicle as microparticles allows direct injection of
the particles into the bloodstream
or a defect site.
Through compression
or sintering,
microparticles can also be formed into other shapes and sizes [16, 17]. A further application
of microparticles is in the release of multiple therapeutic proteins, such as a combination of
growth factors to encourage tissue regeneration.
Different sets of microparticles, each
containing a therapeutic protein and each designed with a distinct release profile, can be
dispersed in a bulk matrix or scaffold. For versatile multifactor release, it is important that the
release profiles of the microparticles be tunable.
2.1.1
Conventional Double Emulsion Processing
Water-in-oil-in-water (W/O/W) emulsion, also known as double emulsion, is arguably
the most frequently cited method of producing polymeric microparticles encapsulating water
soluble agents due to its relative processing ease and low equipment demand.
This process
involves adding an aqueous solution of the therapeutic agent to an organic solution of the
polymer. The mixture is agitated by vortexing, homogenization or sonication to form a waterin-oil (W/O) emulsion. This emulsion is then added to a large volume of water supplemented
with a surfactant, and agitated again to create a W/O/W emulsion. Solid microspheres are
obtained after the organic solvent is removed by extraction or evaporation [8].
31
While the microparticle yield and protein encapsulation efficiency can be relatively
high with this process, it has several limitations. These microparticles typically exhibit high
initial burst as a large proportion of the protein is only loosely associated with the
microparticles
on the surface or in the pores.
The release rate after the initial burst is also
difficult to predict and control. The effects of a wide range of material and processing
parameters on release have been studied, including polymer molecular weight [18], surfactant
type [19], ratio of inner and outer water phases [5, 20], addition of hydrophilic or hydrophobic
agents [21, 22], and addition of salt to the outer water phase to reduce protein leaching [6].
These studies have highlighted the difficulty in predicting a priori the effect of each
parameter on the release profile. In addition, the effect on release is often a change in
magnitude, and not a change in rate. A further drawback of the double emulsion process is
the exposure of the protein to denaturing organic solvent at the water-oil interface, which may
lead to protein unfolding, aggregation and deactivation [23].
2.1.2 Solid-in-Oil-in-WaterProcessing
To circumvent protein aggregation at the water-oil interface, a semi-aqueous technique
has been developed in which the protein is lyophilized and dispersed as a solid powder in the
organic polymer solution to create a solid-in-oil (S/O) suspension. This suspension is then
emulsified in an aqueous phase to form a solid-in-oil-in-water
(S/O/W) suspension [24]. The
rationale behind this technique is the observation that dehydrated proteins are less prone to
denaturation in anhydrous solvents because conformational changes are kinetically prohibited
[25]. Bovine serum albumin (BSA) encapsulated by this technique was found to show little
change in secondary conformation from its lyophilized state [24].
2.1.3 Adsorption of Proteins onto Apatite
The S/C)/W emulsion method overcomes the problem of protein denaturation at the
W/O interface, but does not address issues of high initial burst and poor tunability of release
from polymeric microspheres.
To reduce initial burst, the protein can be formulated into less
soluble forms such as by precipitation
substrate.
with divalent ions [26-28] or adsorption onto a
In the former case, the concentration of the divalent ion, such as zinc, needed to
precipitate each type of protein has to be carefully determined. The diffusion or displacement
32
of the divalent ions leads to solubilization and release of the protein. In the latter case, the
substrate used for adsorption should have a high affinity for the protein and a high surface
area for adsorption. Protein release is affected by desorption of the protein and/or dissolution
of the substrate. A suitable candidate for adsorbing bone morphogenetic proteins (BMPs) is
apatite, which has a natural affinity for BMPs. In fact, the osteoconductivity of apatite and
calcium phosphate is sometimes attributed to their ability to concentrate growth factors such
as BMPs in the body [29]. As a result of this high protein affinity, the release of proteins
from apatite and calcium phosphate is typically very slow and sustained [30].
Our strategy for a tunable, controlled delivery platform utilizes a polymer with acidic
degradation products, poly(lactic-co-glycolic acid) (PLGA), to affect the dissolution of a basic
apatitic substrate on which a protein has been pre-adsorbed.
Such a composite can be
formulated into microparticles by the S/O/W method. Protein release is anticipated to be
accelerated over passive leaching from apatite, yet, more controlled than release from
polymeric microspheres due to apatite's affinity for protein.
2.2
Experimental
2.2.1
Synthesis of Hydroxyapatite and Carbonated Apatite
Nanocrystalline hydroxyapatite (HAP) and carbonated apatite (CAP) were synthesized
according to the method developed by Ahn et al. [31-33]. For the synthesis of HAP, 900 ml
of 0.167 M Ca(NO 3) 2 (Fluka) and 900 ml of 0.100 M (NH4)2 HPO4 (Fluka) were prepared in
deionized (DI) water. The pH of the (NH4 )2 HPO4 solution was raised to 10.4 with ammonium
hydroxide. The Ca(NO3) 2 solution was added to the (NH4)2HPO4 solution at a rate of- 3
ml/min to precipitate HAP. The resulting suspension was stirred at room temperature for 72 h.
After this aging step, the white precipitate was washed with solutions of decreasing pH,
followed by two ethanol washes. The gel was air-dried overnight, then oven-dried at 120°C
for 24 h. The dried gel was ground in a heated mortar and calcined at 5500 C for 2 h (ramp
rate = 4C/min). After calcination, the HAP powder was sieved through a 45-tm mesh.
CAP was synthesized by the same method but with the following modifications. The
carbonate source, (NH 4 )HCO 3, was added to the (NH 4 )2 HP0
4
solution at a concentration of
0.100 M. After oven drying, the gel was ground and sieved. The powder was not calcined to
avoid driving off the carbonate groups at elevated temperatures.
33
For the synthesis of submicron-sized apatite particles, modifications were made to the
above protocol
to reduce
The concentrations
agglomeration.
of the Ca(NO 3)
2
and
(NH 4 )2HP0 4/(NH 4)HCO 3 solutions were reduced 10-fold. Tween 80 (Aldrich) was added as
a surfactant to constitute 11 v/v% of the (NH 4) 2HP0 4 /(NH 4)HCO 3 solution.
The particles
were collected and washed by ultrafiltration instead of centrifugation. Washes with ethanol
were not performed due to its incompatibility with the ultrafiltration device. Water was
removed by fieeze drying to obtain the final product. Calcination, which leads to grain
growth, was not performed on these apatite powders. The hydroxyapatite and carbonated
apatite thus prepared are referred to as sHAP and sCAP, respectively.
Two types of HAP powder were purchased from Berkeley Advanced Biomaterials, Inc.
The two products were BABI-HAP-SP
(BABI) for comparison with our materials.
and
BABI-HAP-N20, with advertised mean particle sizes of 5 ptmand 20 nm, respectively.
2.2.2 Characterizationof Apatite
Powder X-ray diffraction (XRD) patterns of the various apatite powders were obtained
with a Siemens D5000 0-0 diffractometer (45 kV, 40 mA, Cu K.). Grain size analyses were
performed on the <002> diffraction peaks using Scherrer's method. The BET surface area
was determined by nitrogen adsorption on a Micromeritics ASAP 2000/2010 Analyzer.
Particle size distribution was evaluated using a Horiba CAPA-300 Particle Size Analyzer.
2.2.3 Adsorption of Proteins onto Apatite
Fluorescein isothiocyanate bovine serum albumin (FITC-BSA, Sigma Aldrich) was
used as a model protein. The adsorption of FITC-BSA onto apatite was typically conducted
as follows. FITC-BSA was dissolved in DI water at a concentration of 0.25 mg/ml, and added
to 10 mg of apatite.
The suspension was stirred for 16 h at room temperature to allow the
adsorption of protein onto the apatite. The resulting BSA-apatite complex was collected by
centrifugation,
and the supernatant was filtered and saved.
Subsequently, the BSA-apatite
complex was washed with DI water and lyophilized. The amount of protein adsorbed was
determined as the difference between the protein concentration of the initial stock solution
and that of the supernatant after adsorption.
Coomassie Plus total protein assay (Pierce).
34
Protein concentration was analyzed by
Maximum protein adsorption was determined for HAP, CAP, sHAP, sCAP, BABI-
HAP-SP, and BABI-HAP-N20. For each type of apatite, the sample size tested was 3-5.
Adsorption isotherms for FITC-BSA onto CAP and HAP at 40 C and room temperature were
obtained by varying the concentration of the initial FITC-BSA solution from 0.25 to 6.00
mg/ml. Adsorption times ranging from 0.5 h to 7 d were used to determine the minimum
amount of time required for maximum protein adsorption at room temperature.
To explore the possibility of adsorbing other types of proteins, proteins of different
isoelectric point (IEP) and size were tested for adsorption onto CAP. These proteins (all from
Sigma) included lysozyme (IEP = 11, MW = 14 kD), cytochrome C (IEP = 10, MW = 12.4
kD), and alcohol dehydrogenase (IEP = 5.5, MW = 141 kD). Protein stock solutions of 100200 ytg/ml were prepared with DI water. Each protein solution was added to triplicates of 10
mg of CAP and stirred for 8 h at room temperature. The amount of protein adsorbed was
expressed as a percentage of the amount of protein in the original stock solution.
Zeta potential measurements (Brookhaven ZetaPALS Zeta Potential Analyzer) were
performed at piH 7 on CAP and HAP before and after the adsorption of FITC-BSA.
For each
apatite or BSA.-apatite complex, the number of specimens tested was 3.
2.2.4 Effect ofpH on the Dissolutionof Apatite and the Release of Adsorbed Proteins
To study the effect of pH on the dissolution of apatite and the release of adsorbed
proteins from BSA-apatite complexes, triplicates of 10 mg of either BSA-CAP or BSA-HAP
powder was added to microvials containing 1.5 ml of medium. The BSA loading on the
apatites was 7.5 w/w% (750 itgof BSA per 10 mg of BSA-apatite complex). The medium
As a comparison,
N,N-bis(2-hydroxyethyl)-2-
was citrate buffer
of pH 3, 4 or 5.
aminoethanesulfonic
acid (BES) buffer of physiological pH (pH 7.4) was also used. The vials
were incubated in a 37°C water bath for 8 weeks. At pre-determined time intervals of 1, 4, 7,
14, 21, 28, 42 and 56 days, the vials were centrifuged.
Half of the supernatant (0.75 ml) was
removed and filtered, and replaced with the same volume of fresh buffer. The vials were then
returned to the water bath. The supernatant was stored at 4C until analysis by Coomassie
Plus total protein assay for protein concentration. The results were used to construct a
cumulative protein release profile for each protein-apatite complex at a specified pH.
35
2.2.5 Preparationof CompositeMicroparticles
Composite microparticles were synthesized by a S/O/W emulsion process modified
from that described by Castellanos et al. for encapsulating proteins in the solid state [24]. A
typical synthesis involved dissolving 250 mg of PLGA (Alkermes) in 2 ml of
dichloromethane. To this polymer solution, 10 mg of protein-apatite complex were added and
vortexed to create a uniform suspension.
The S/O suspension was homogenized in 50 ml of
0.1 w/v% aqueous methyl cellulose solution at 8000 rpm for 2 min at room temperature.
The
resulting S/O/W suspension was heated at 300 C for 3 h to evaporate dichloromethane
and
solidify the particles. The particles were collected by centrifugation, washed with DI water,
and lyophilizecd. A schematic of this synthesis is presented in Fig. 2.1. This preparation was
varied systemically one parameter at a time to investigate the effect of different material and
processing parameters on particle size and morphology, as well as protein release.
Protein-Apatite Complex (s)
PLGA in CH 2C12 (o)
vortex
I,
Solid-in-Oil (S/O) Suspension
i
.
Methyl Cellulose in Water (w)
homogenize
.r
Solid-in-Oil-in-Water (S/O/W)Suspension I
evaporate CH2 C12
wash
freeze dry
Composite Particles Encapsulating Protein
Fig. 2.1. Schematic of the S/O/W process for preparing composite microparticles.
36
2.2.6 Preparationof PLGA Microspheresby Double Emulsion Processing
As a standard for comparison, the double emulsion technique of Blanco et al. [34] was
modified to prepare polymeric microspheres encapsulating proteins. An aqueous solution of
FITC-BSA was first created by dissolving 4 mg of the protein in 50 tl of DI water. This
aqueous phase was added to an organic phase containing 250 mg of PLGA in 2 ml of
dichloromethane.
The two phases were homogenized at 8000 rpm for 2 min to form a W/O
emulsion. The emulsion was then transferred to 50 ml of aqueous 0.1 w/v% methyl cellulose
solution and homogenized at 8000 rpm for another 2 min to create a W/O/W emulsion.
This
double emulsion was stirred and heated at 300 C for 3 h to drive off dichloromethane.
The
resulting microparticles were collected by centrifugation, washed 3 times with DI water, and
freeze-dried to obtain the final product. A schematic of this synthesis is presented in Fig. 2.2.
The key difference between W/O/W and S/O/W processing lies in the protein formulation; the
protein is dissolved in an aqueous solution in the former but adsorbed onto solid apatite in the
latter.
Aqueous Protein Solution (w)
PLGA in CH 2Cl 2 (o)
vortex
I
I
Water-in-Oil (W/O) Emulsion
ho eI
I
Methyl Cellulose in Water (w)
homogenize
Water-in-Oil-in-Water (W/O/W) Emulsion
I
evaporate CH 2C12
wash
freeze dry
I Polymeric Particles Encapsulating Protein
Fig. 2.2. Schematic of the W/O/W process for preparing polymeric microparticles.
37
2.2.7 Characterizationof MicroparticleMorphologyand Size
The morphology of the composite microparticles was evaluated by environmental
scanning electron microscopy (ESEM, FEI/Philips XL30 FEG) at 6-7 kV. The particle size
distribution was determined by measuring the dimensions of a minimum of 100 particles by
ESEM.
2.2.8 Evaluation of EncapsulationEfficiency of CompositeMicroparticles
To evaluate the protein loading in the composite microparticles, the polymeric phase
was first dissolved with 0.5 ml of dichloromethane per 10 mg of composite particles. 3 ml of
citrate buffer of pH 2 were then added to the organic suspension to dissolve the apatitic phase.
Prior testing showed that 3 ml of pH 2 citrate buffer were sufficient to dissolve more than 8
mg of protein-apatite complex. Among all the sets of composite particles that were prepared,
the maximum loading of protein-apatite in 10 mg of composite particles was 2.65 mg. The
dichloromethane-citrate buffer mixture was vortexed periodically over 2 days to allow protein
extraction into, the aqueous phase.
The aqueous phase was then removed for protein
concentration analysis by Coomassie Plus total protein assay. The low pH of the citrate
buffer sometimes led to the precipitation of BSA. To dissolve the protein precipitate, 1 ml of
0.1 N NaOH was added. The resulting alkaline protein solution was also assessed for protein
concentration. The total protein content in the composite particles was the summation of the
quantities of protein measured in the citrate buffer and in the NaOH solution. Hence, the
protein loading of the composite particles was calculated as:
Protein Loading (%) = Protein Amt.citrate buffer + Protein Amt. NaOHsoin x 100
Mass of Composite Particles
'The encapsulation efficiency was calculated as follows:
Encapsulation Efficiency (%) =
Measured Protein Loading
x 100
Theoretical Protein Loading
Each set of composite particles was tested in triplicates.
An alternative method of loading these composite particles with protein was to allow
post-synthesis uptake of protein into blank apatite-PLGA composite particles. Both CAPPLGA and HAP-PLGA composite particles were tested in triplicates. 15 mg of blank apatitePLGA particles were mixed with 1 mg of FITC-BSA in 1 ml of DI water. The mixtures were
38
allowed to stir at room temperature for 24 h, and then centrifuged.
The protein loading for
each sample was derived from the difference between the protein concentration of the original
stock solution and that of the supematant.
2.2.9 Evaluation of In VitroProtein Release
A known amount of protein-loaded composite microparticles was resuspended in 1.5
ml of BES buffer (pH 7.4) supplemented with 0.02 w/v% sodium azide as a germicide, and
incubated at 37°C. Each set of particles was tested in duplicates. At pre-determined time
intervals (1, 4, 7, 14, 21, 28, 42 and 56 days, up to 20 weeks), the samples were centrifuged,
and 0.75 ml of the supernatant was withdrawn and replaced with 0.75 ml of fresh BES buffer.
For samples containing FITC-BSA, the collected supernatant was filtered and assayed for
fluorescence, and stored at 4C until further analysis of protein concentration with the
Coomassie Plus total protein assay. The concentration of the protein released at each time
point was used to construct cumulative release profiles.
2.2.10 Evaluation of PolymerDegradation
Blank composite microparticles fabricated from PLGA with molecular weight of 6, 13,
24 and 59 kD were tested. The apatite loading was 0.2 mg per mg of PLGA. For each set of
particles, 20-mg samples were suspended in 1.5 ml of pH 7.4 BES buffer supplemented with
0.02 w/v% sodium azide. One sample was prepared for each time point at 1, 4, 7, 14, 21 and
28 days. At each time point, the sample was centrifuged and the supernatant was removed.
The residue was freeze-dried, dissolved in tetrahydrofuran, filtered, and analyzed for
molecular weight by gel permeation chromatography using Waters 1525 HPLC system and
2414 refractive index detector, equipped with Styrogel® HT3, 4 and 5 columns.
2.2.11 Evaluation of CalciumReleasefrom CompositeMicroparticles
The calcium concentration of the release media incubated with composite
microparticles was analyzed using a fluorescent calcium indicator, Rhod-5N (Molecular
Probes). Rhocl-5N is a rhodamine-based fluorophore that exhibits enhanced fluorescence
(Ex/Em = 551/576) when bound to calcium.
It is sensitive in the range of 10 tM to 1 mM
calcium.
39
Calcium release from CAP-PLGA composite microparticles prepared from polymers
of different molecular weight was tracked. In addition, bare CAP and one set of BSA-CAPPLGA (59 kD) composite particles with a BSA loading of 1.67 w/w% were tested. The
apatite loading, for all sets of composite particles was 0.2 mg per mg of PLGA. For each set
of composite particles, duplicates were prepared consisting of 30 mg of particles incubated in
1.5 ml of BES buffer (pH 7.4) at 37°C. 5 mg were used for bare CAP, which were equivalent
in amount to the apatite loading in 30 mg of composite particles. The release medium was
collected, and renewed at 1, 4, 7, 14, 21, 28 and 42 days. To generate a calcium calibration
curve, standards ranging in concentration from 10 [tM to 1 mM were prepared using calcium
nitrate tetrahydrate (Fluka).
2.3
Results and Discussion
2.3.1
Characterization of Apatite
The preparation of apatite by chemical precipitation resulted in the formation of a pure
apatitic phase as shown in Fig. 2.3. The major peaks at 25.880, 31.77° , 32.20° , 32.90° and
39.82 ° corresponded to (002), (211), (112), (300) and (310) diffractions of HAP. Calcination
of HAP led to higher crystallinity, as indicated by the sharper and narrower peaks. HAP had a
much finer grain size than commercial BABI-HAP-SP and BABI-HAP-N20 powders (Table
2.1). CAP, sCAP, and sHAP, which were prepared without calcination, possessed smaller
grain sizes than HAP.
HAP and CAP have similar average particle sizes (Table 2.1). The use of surfactant
led to sHAP and sCAP with finer particle sizes. The measured particle size of BABI-HAP-SP
was close to its advertised particle size of 5 tpm,and was in the same range as HAP and CAP.
The measured particle size of BABI-HAP-N20 was 30 times larger than its advertised particle
size, probably due to agglomeration. Apatite particle size is important as it places a lower
limit on the size of the apatite-polymer composite microparticles that can be formed. Smaller
apatite particles would facilitate more uniform apatite dispersion in the polymeric matrix, as
well as the creation of smaller composite particles without phase separation.
Table 2.2 shows that CAP has a high BET surface area (302 m 2 /g) and a high pore
volume, suggesting that it might be able to sequester a larger amount of protein than HAP.
The commercial BABI powders have lower surface areas and pore volumes, and larger pores
40
than the apatites
we synthesized.
The sCAP and sHAP powders
gave inconsistent
measurements of BET surface areas, likely due to surfactants remaining in their pores.
Attempts to drive off the residual surfactants by gentle calcination to 3000 C resulted in the
discoloration of these powders and reduction in surface areas.
Hence, their nitrogen
adsorption results were not included in Table 2.2.
18000
16000
(3
C
14000
:D
12000
LL
(U
10000
.0
8000
6000
C:
4000
2000
0
20
25
30
35
40
45
20 (0)
Fig. 2.3. XRD) patterns of apatites prepared with surfactant (sCAP, sHAP) and without
surfactant (CAP, HAP), and commercial BABI-HAP-SP and BABI-HAP-20N powders. The
diffraction peak,positions of the HAP standard (std) are also included for comparison.
Table 2.1. Grain sizes and particle sizes of apatite powders.
Apatite Sample
Grain Size (nm)
Average Particle Size (pm)
HAP
27
5.3
CAIP
12
6.8
sHAP
15
0.7
sCAP
15
1.0
BABI-HAP-SP
47
4.5
BABI-HAP-N20
43
0.6
41
Table 2.2. BET surface areas, pore volumes and mean pore radii of apatites.
Property
CAP
HAP
BABI-HAP-SP
BABI-HAP-N20
BET Surface Area (m 2/g)
302
86
48
53
Pore Volume (cm 3 /g)
1.24
0.52
0.18
0.27
Mean Pore Radius (nm)
8.2
12.2
24
36
2.3.2 Protein Adsorption onto Apatite
The maximum FITC-BSA adsorption capacities of various apatites are summarized in
Table 2.3. BABI-HAP-SP and BABI-HAP-N20 were found to have lower protein adsorption
capacities due to their lower surface areas (Table 2.2). Compared to BABI-HAP-N20, sHAP
has approximately the same particle size, but adsorbed more than thrice the amount of protein.
Table 2.3. Protein adsorption capacities of apatites.
Apatite Sample
Maximum BSA Adsorption (w/w%)
HAP
15.9 ± 0.8
CAP
31.4+ 1.5
sHAP
25.3
1.6
sCAP
21.3
1.2
BABI-HAP-SP
7.1 ± 0.5
BABI-HAP-N20
8.0 ± 0.6
FITC-BSA adsorption isotherms of CAP and HAP are shown in Fig. 2.4. Adsorption
isotherms at 4°C show saturation at - 80% of the BSA levels at room temperature. The value
of the plateau was taken as the maximum amount of protein adsorbed for the given mass of
apatite. Table 2.3 shows that the maximum amount of FITC-BSA adsorbed onto CAP was
2.4 times higher than that adsorbed onto HAP, although the BET surface area derived from
nitrogen adsorption was 3.5 times higher for CAP (Table 2.2). Compared to HAP, CAP has a
finer mean pore size of 8.2 nm, which was equivalent to BSA's hydrodynamic diameter.
Thus, the smaller pores in the pore size distribution for CAP might not be accessible for BSA
adsorption.
42
Protein adsorption onto apatite was found to be a rapid process, and maximum
adsorption could be achieved within 6 h. A representative adsorption profile is shown in Fig.
2.5. The minimum time required for maximum protein adsorption should be used to reduce
processing time and exposure of sensitive proteins to elevated temperatures. However, in the
case of FITC-B3SA,which is a relatively stable protein, adsorption was typically conducted
overnight for 16 h. For the adsorption of BMPs [35], the adsorption time was shortened.
500
_
A_
on
E
(a)
a)
E
:L
500
400
400
U,
a
a)
'1
(U,
E
300
300
E)
E
a)
a)
Q0
_0
U)
ED
a
a)
200
.o
co
100
m
E
200
0
-0
100
co
0
E
-e
Fit-----C-----·-·-·-----·----------
100
0
200
300
0
200
400
Amt. BSA added per mg apatite (g/mg)
0)
400
600
800
Amt. BSA added per mg apatite (Vlg/mg)
500
E
500
(c)
400
400
12
(U
a
0)
E
E
300
300
a)
a)
a)
Q
a
200
200
-o
0o 100
m
co
E
<
aV
<
n1
'
u
v
0
100
200
100
m
,X
300
0
0
400
Amt. BSA added per mg apatite (g/mg)
200
400
600
800
Amt. BSA added per mg apatite (g/mg)
Fig. 2.4. FITC-BSA adsorption isotherms of (a) HAP and (b) CAP at room temperature, and
(c) HAP and (d) CAP at 4 0 C.
43
120
)
0)
E
.
110
100
90
U,
°
80
CU
C,,
70
E
<
60
70
0
10
20
30
40
50
60
Time (h)
Fig. 2.5. Adsorption of FITC-BSA onto HAP with time at room temperature.
The ability of apatite to adsorb other types of proteins of different isoelectric points
(IEP) and sizes was examined.
The amount of protein adsorbed is expressed as a percentage
of the total amount of protein in the initial stock solution. CAP was found to be capable of
adsorbing a high percentage of all the proteins tested, with the exception of lysozyme (see
Table 2.4). Only 24% of the lysozyme added to the apatite suspension was adsorbed.
This
lower affinity could be attributed to the rigid structure of lysozyme, which allowed it to only
partially unfold during adsorption and make fewer contacts with the substrate [36].
Table 2.4. Adsorption of proteins of different IEP and size onto CAP.
Protein
Isoelectric Point
Molecular Weight (kD)
Amt. Adsorbed (%)
FITC-BSA
4.7
66
99.2
Lysozyme
11
14
23.7
Cytochrome C
10
12
96.9
Alcohol
5.5
141
97.5
Dehydrogenase
The zeta potential of CAP at pH 7 was found to be slightly negative (Table 2.5). HAP,
which was also negatively charged, had a higher zeta potential. A higher zeta potential would
give rise to increased stability of the suspension since the charged particles would repel one
another and prevent agglomeration. Hence, a suspension of higher zeta potential could be
expected to have finer particle size (see Tables 2.2 and 2.5). Upon the adsorption of BSA
44
(IEP = 4.7), which was negatively charged at pH 7, the zeta potentials of both CAP and HAP
rose in magnitude.
The particle sizes of BSA-CAP and BSA-HAP were also found to
decrease.
Table 2.5. Zeta potential and particle size of apatites and BSA-apatite complexes.
Sample
Zeta Potential at pH 7 (mV)
Particle Size (m)
CAP
-2.82 ± 1.78
6.8
1.7
HAP
BSA-CAP
BSA-HAP
-11.19 ± 2.12 -18.59 ± 0.66 -22.16 ± 1.81
5.3 ±1.5
5.8
1.8
3.8
2.1
As BSA and apatite were both negatively charged at pH 7, electrostatic interaction
was not the driving force for protein adsorption in this case.
The surface of bare
hydroxyapatite is reported to be hydrophobic, so hydrophobic interactions between BSA and
hydroxyapatite could have induced protein adsorption [37, 38]. BSA, a "floppy" protein with
high conformational adaptability, could expose its hydrophobic groups for interactions with
the hydroxyapatite surface [39]. Other mechanisms of protein adsorption might include ionexchange and hydrogen bonding. Thus, adsorption onto apatite would not be limited to
proteins that were oppositely charged at the pH of interest. Both positively charged proteins
(such as cytochrome C and BMPs [35]) and negatively charged proteins (such as BSA and
alcohol dehydrogenase) were found to adsorb onto CAP (Table 2.4).
2.3.3 Effect of pH on ProteinReleasefrom Protein-ApatiteComplexes
The effect of pH on the BSA release from BSA-apatite complexes is shown in Fig. 2.6.
it was hypothesized that since apatite was basic, its dissolution in an acidic medium would
cause proteins sequestered on its surface or within its pores to be released. In line with our
hypothesis, we found that at the physiological pH of 7.4, protein release was very slow and
almost negligible over the 8-week test period. One additional data point was obtained at 436
days for pH 7.4, indicating that even after this extended period, only 2.6% and 1.5% of the
protein had been released from the BSA-HAP and BSA-CAP complexes, respectively. Hence,
under neutral pH conditions, the affinity of apatite for BSA remained high and apatite
dissolution was low over more than one year. The stability of ovine albumin in carbonated
hydroxyapatite gel over one year has also been reported by Barralet et al. [30].
45
Lowering of the pH to 5 led to a sharp increase in protein release from both BSA-HAP
and BSA-CAP complexes. The BSA-HAP complex released 45% of its protein in the first
week, but the release plateaued thereafter. Protein release from BSA-CAP complex was more
gradual and constant over the period tested.
The difference in release profile could be
attributed to the difference in the pore volume of HAP and CAP. HAP, being less porous,
held a larger proportion of BSA on its surface, whereas CAP has a more even distribution of
BSA on its surface and in its pores. Acid would first erode the surface of the apatite, resulting
in an initial burst of protein from BSA-HAP complex. For BSA-CAP complex, the protein
would be released more gradually since the acid would have to penetrate through the surface
layer of the CAP particle to release those proteins adsorbed deep within the pores.
Further reductions in pH to 4 and 3 were also investigated.
At pH 4, two observations
were noted: First, the release rate was approximately zero-order after a small initial burst,
suggesting that a constant protein release rate was possible in the constant presence or
addition of acid. Second, protein release decreased to - one-fifth the levels at pH 5 for BSACAP complex. A fall in the protein release with decreasing pH was further exemplified by
the profile at pH 3. A possible reason for the depressed protein release was a change in
electronic interactions between BSA and apatite as their points of zero change were crossed at
pH 4.7 and - 4 (preliminary experimental data not shown) [40], respectively.
both BSA and apatite were negatively charged.
At pH 7.4,
At pH 3 or 4, BSA was positively charged,
whereas apatite was negatively charged or mildly positively charged. Thus, at the lower pHs
tested, stronger interactions might have existed between BSA and apatite, leading to a
decrease protein desorption and release. Another possibility was that low pH might have
induced protein precipitation [41], which became more severe with increasing acidity.
Therefore, in the subsequent preparation of polymer-apatite composite microparticles, the pH
within the particles should be regulated by the selection of (i) an appropriate proportion of
apatite to neutralize the acid, and (ii) a polymer with a suitable degradation profile to avoid
overly rapid acid production.
46
60
50
a)
,o 40
a)
)
m
30
a)
co 20
E
10
0
0
10
20
30
40
50
60
70
Time (days)
Fig. 2.6. Release of BSA from BSA-HAP and BSA-CAP complexes in buffers of pH 3. 4, 5
and 7.4. Data shown are mean values for n = 3.
Matsumoto et al. have recently reported similarly enhanced protein release from
hydroxyapatite pre-adsorbed with cytochrome C when the pH was reduced from 7 to 4 [42].
However, the protein adsorption capacity of their HAP particles was very low; the maximum
cytochrome C adsorbed was only 1.5 w/w%.
2.3.4 Effect of ProcessingParameterson CompositeParticleSize
ESEM micrographs of four representative sets of composite microparticles are shown
in Figs. 2.7 and 2.8. The first pair has an average diameter of - 40 ~pm(Fig. 2.7). Composite
particles prepared with sCAP have smoother surfaces, whereas those containing CAP were
rougher and more porous.
Fig. 2.8 illustrates the effect of blending in a higher molecular weight polymer.
Particles prepared with 13 kD PLGA were smaller (- 20
tm) and contained a larger
proportion of non-spherical debris (Fig. 2.8a). The addition of 24 kD PLGA led to slightly
larger (- 23 pm) and more spherical particles (Fig. 2.8b).
47
I
I,
Fig. 2.7. ESEM micrographs of composite microparticles prepared with 59 kD PLGA, and (a)
submicron-sizec sCAP particles and (b) micron-sized CAP particles.
IS
Fig. 2.8. ESENMmicrographs of composite microparticles prepared with CAP, and (a) 13 kD
PLGA and (b) 50/50 blend of 13 kD and 24 kD PLGA.
The effects of various processing parameters on the size of the composite particles are
summarized in Table 2.6. For each parameter, low and high values around a midpoint were
investigated to demonstrate the general trend. Variations that caused higher shear during
homogenization led to the creation of smaller particles.
For example, increase in
homogenization speed and time increased the amount of shear experienced by the suspension.
Decrease in polymer solution concentration and polymer molecular weight reduced the
viscosity of the organic phase and lowered the particle size concomitantly. Blends of PLGA
and poly(lactic acid) (PLA) or polyethylene glycol (PEG) were also examined.
48
Table 2.6. Effects of processing parameters on composite particle size.
Processing/Material Parameter
Particle Size (lm)
Standard processing conditions as described in Section 2.2.5
41 + 12
Increase homogenization speed to 12000 rpm
30 + 8
Increase homogenization speed to 16000 rpm
27 + 8
Decrease homogenization time to 1 min
55 ± 21
Increase homogenization time to 4 min
31 ± 10
Increase CH-C1lI volume to 4 1ml
27 + 8
Decrease CH-CL, volume to 1 ml
65 + 24
Reduce aqueous surfactant volume to 25 ml
35 + 12
Increase aqueous surfactant volume to 100 ml
51 + 15
Increase BSA-CAP complex loading to 30 mg
37 + 13
Increase BSA-CAP complex loading to 50 mg
37 + 12
Decrease apatite particle size to 1 um (with the use of sCAP)
44 + 13
Use of lower molecular weight PLGA (6 ikD)
23 + 7
Use of higher molecular weight PLGA (59 kD)
45 + 15
Blending in 20 w/w% PLA (25 kD)
42 + 13
Blending in 60 w/w% PLA (25 kD)
44 + 12
Blending in 10 wv/w%PEG (3.4 kD)
37 + 12
Blending in 30 w/w% PEG (3.4 kD)
23 + 7
2.3.5
Encapsulation Efficiency of Protein in Composite Microparticles
More than 60 formulations of BSA-loaded composite microparticles were prepared,
differing in polymer molecular weight, polymer blend, apatite type, apatite particle size, etc.
All 60 formulations were tested for encapsulation efficiency, and the average was found to be
69.8%
19.9%/. The multi-step nature of the extraction process used to detennine
encapsulation efficiency might have led to under-estimations due to protein loss at each step.
However, the average encapsulation
efficiency was high, and was comparable to that of
particles prepared by the double emulsion technique [18, 21, 43-46]. The encapsulation
efficiency of selected formulations is shown in Table 2.7.
49
Table 2.7. Encapsulation efficiency of selected BSA-loaded composite microparticles.
Formulation
Encapsulation
Theoretical BSA
Efficiency§
Polymer
PolymerMW
BSA-ApatiteComplex*
Loadingt
PLGA
6 kD
BSA-CAP (15 w/w%)
1.5 w/w%
77.1%
PLGA
24 kD
BSA-CAP (15 w/w%)
1.5 w/w%
52.1%
PLGA
24 kD
BSA-CAP (8 w/w%)
2.2 w/w%
68.7%
PLGA
24 kD
BSA-HAP (20 w/w%)
2.0 w/w%
73.8%
PLGA
59 kD
BSA-CAP (15 w/w%)
1.5 w/w%
50.9%
PLA
10-15 kD
BSA-CAP (15 w/w%)
1.5 w/w%
57.9%
PLGA +
PLGA +
24 kD, 70 kD
BSA-CAP (9 w/w%)
0.9 w/w%
94.0%
PEG (7:3)
* Values in parentheses reterred to the weight percent of the BSA-apatite complex in the
composite microparticles.
t Weight percent of BSA in the composite microparticles.
§ Mean values for n = 3.
The maximum amount of protein that could be loaded into the CAP-PLGA and HAPPLGA composite particles post-synthesis was 0.26 and 0.28 w/w%, respectively.
If the
apatite powders had been pre-adsorbed with protein, the maximum theoretical protein loading
would have been 5.3 and 2.7 w/w% for CAP-PLGA and HAP-PLGA particles, respectively.
PLGA is not known to have a particularly high affinity for proteins.
Thus, the less effective
protein uptake into the composite particles post-synthesis was expected, since the apatite
particles would then be embedded in the PLGA matrix and be less accessible for protein
adsorption. Hence, the pre-adsorption of protein onto apatite was a crucial step in ensuring
adequate and appropriate protein loading for the composite particles to be used in controlled
delivery applications.
2.3.6 Effect of Material and ProcessingParameterson the In VitroReleasefrom
CompositeMicroparticles
2.3.6.1
Polymer Molecular Weight
The molecular weight of a biodegradable polymer is expected to influence its
degradation rate and lifetime.
Apatite-polymer composite particles were prepared from
PLGA of different molecular weights ranging from 6 kD to 59 kD. The onset of rapid protein
50
release was observed to occur at different times depending on the molecular weight of the
polymer used (Fig. 2.9a). For particles prepared with short-chained, 6 kD and 13 kD PLGA,
the rapid release phase began immediately. An upturn in the release profile was observed
after - 3 weeks for particles synthesized with 24 kD PLGA, whereas protein release remained
gradual up to 8 weeks for particles prepared with 59 kD PLGA. Such control of release rate
by polymer molecular weight might be useful in sequential release, such as to direct the debut
of a protein in a wound healing process.
4
4 e
I .L
I .4
MI 1.0
M 1.0
E
E
m 0.8
m 0.8
0
.,
0.6
m 0.6
u,
m 0.4
m 0.4
0
' 0.2
E
O 0.2
E
0.0
0.0
0
20
40
60
Time (days)
80
100
0
10
20
30
40
Time (days)
50
60
Fig. 2.9. Protein release from composite particles loaded with 15 itgof FITC-BSA per mg of
carrier. Particles were prepared using CAP and (a) PLGA of different molecular weights and
(b) blends of PLGA of different molecular weights. Data shown are mean values for n = 2.
Polymers of different molecular weights could also be blended together to modulate
the acid production over time. For example, rapid protein release from PLGA of low
molecular weights (6 kD and 13 kD) was slowed by blending an equi-proportion of 24 kD
PLGA (Fig. 2.9b).
2.3. 6.2
Polymer Hydrophobicity
2.3.6.2.1
Blends of PLGA and PLA
Composite particles were synthesized from blends of PLGA and PLA.
As the
proportion of PLA in the composite particles was increased, the release rate decreased (Fig.
2.10).
Lactic acid contains a methyl group that augments its hydrophobicity relative to
51
glycolic acid, hence, PLA is more hydrophobic than PLGA. Since it is a homopolymer, PLA
is also more crystalline than PLGA, which is a random 50:50 copolymer of lactic acid and
glycolic acid.
An increase in polymer hydrophobicity would delay water penetration,
polymer degradation and acid production, and consequently, reduce protein release. This
parameter could be varied by using different types, blends, or compositions of polymers. For
example, the lactic acid to glycolic acid ratio in PLGA could be varied to obtain a range of
hydrophobicities and degradation rates.
tn ,-
U.b
, 0.5
0
E
-)
0.4
.
0.3
m 0.2
0,
E 0.1
E
0.0
0
20
40
60
Time (days)
80
100
Fig. 2.10. Effect of PLA incorporation on protein release. The CAP-polymer composite
particles contained 15 Vtgof FITC-BSA per mg of carrier. The molecular weights of PLGA
and PLA were 24 kD and 25 kD, respectively. Data shown are mean values for n = 2.
;2.3.6.2.2 Blends of PLGA and PEG
PEG, a hydrophilic, biocompatible and non-degradable polymer, was blended with
PLGA in the synthesis of composite particles. Increasing the PEG content led to an increase
in the protein release (Fig. 2.11). All release profiles showed an upturn at-
21 days, which
coincided with the degradation of 24 kD PLGA.
The hyclrophilicity of PEG was expected to draw water into the particles where the
PEG chains resided. In addition, PEG might have leached out of the polymer matrix, leaving
behind pores [47]. The aqueous channels thus formed would facilitate water penetration and
PLGA hydrolysis, as well as protein diffusion out of the particles.
incorporation of a hydrophilic polymer increased the protein release.
52
Therefore, the
0.6
'E
X 0.5
0
E
C 0.4
m 0.2
am
' 0.1
E
0
0.0
0
10
20
30
40
50
60
Time (days)
Fig. 2.11. Effect of PEG incorporation on protein release. The CAP-polymer composite
particles contained 9 fig of FITC-BSA per mg of carrier. Molecular weights of PLGA and
PEG were 24 kD and 3.4 kD, respectively.
2.3.6.3
Data shown are mean values for n = 2.
Apatite Particle Size
BSA-apatite complexes prepared with sCAP (1 jtm) and CAP (7
incorporated into the composite particles.
m) were
Reducing the apatite particle size resulted in a
protein release profile that more closely approximated zero-order for a longer period of time
(Fig. 2.12). In contrast, the release plateaued after the first week for composite particles
containing the larger, 7-jpm CAP particles.
For a given amount of acid produced by polymer degradation, a certain amount of
apatite would dissolve. Consequently, the protein adsorbed on that amount of apatite would
be released. Compared to CAP (which has a porosity of 1.24 cm3/g and a mean pore size of
8.2 nm), sCAP has a higher surface/volume ratio, and a lower porosity (0.27 cm 3/g) and mean
pore size (4.2 nm). Thus, most of the protein molecules would be expected to be adsorbed on
the surface of sCAP particles, whereas protein molecules would be adsorbed both on the
surface as well as within the pores of the CAP particles. To free the protein adsorbed within
the pores, a larger volume of apatite would have to be dissolved; this might explain the slower
rate of protein release from CAP after its initial protein release phase. By controlling the
apatite particle size and the available adsorption sites, the protein release profile could be
modulated to achieve extended zero-order release.
53
3.5
a,
'
3.0
E 2.5
8 2.0
a)
a0)
i 1.5
u,
m
a) 1.0
E 0.5
0.0
0
20
40
60
80
100
Time (days)
Fig. 2.12.
Effect of carbonated apatite particle size on protein release.
The composite
particles were prepared with 59 kD PLGA and contained 18 g of FITC-BSA per mg of
carrier. Data shown are mean values for n = 2.
2.3.6.4
Addition of Buffering Apatite
In addition to protein-apatite complexes, varying amounts of bare apatite with no
adsorbed protein could be loaded into the composite particles. As the proportion of this
"buffering apatite" was increased, the protein release rate decreased (Fig. 2.13). Protein
-release was dependent on apatite dissolution by acidic polymer degradation products. By
incorporating buffering apatite into the composite particles, a portion of the acidity was
neutralized by these bare apatite particles rather than by the protein-adsorbed apatite particles.
As a result, less protein was released for a given amount of acid produced.
54
14
0.6
12
g 0.5
E
10
° 0.4
us
a!
=o
(0
0
0.3
8
Co
io 0.2
= 0.1
E
D
6
'
E
4
2
0.0
0
0
10
20
30
40
Time (days)
50
60
0
10
20
30
40
Time (days)
50
60
Fig. 2.13. Effect of buffering apatite on protein release, expressed as (a) cumulative mass of
BSA released per mg of carrier and (b) cumulative percentage of BSA released. The
composite particles were fabricated from 24 kD PLGA (250 mg) and contained ()
9, ()
8.4
and () 7.6 tg of FITC-BSA per mg of carrier (theoretical loading). Results in (b) are based
on BSA content determined by extraction. Data shown are mean values for n = 2.
2.3.6.5
Protein Loading on Apatite
BSA-apatite complexes were prepared by adsorbing 9 and 20 w/w% of FITC-BSA
onto CAP. They were incorporated into composite particles (labeled "9 w/w% BSA in BSACAP" and "20 w/w% BSA in BSA-CAP" in Fig. 2.14) such that each set of particles
contained the same mass of BSA-CAP and essentially the same mass of apatite, but the BSA
content in the "20 w/w%" particles was twice that of the "9 w/w%" particles. Fig. 2.14 shows
that as the protein loading on apatite was increased, the amount released was increased.
However, the release profile of the "20 w/w%" particles was sharper initially and then leveled
off, which deviated from the typical release profile of composite particles prepared with 24
kD PLGA, where the upturn occurred after - 21 days. The higher initial burst might be
indicative of multilayer instead of monolayer adsorption of proteins on CAP. By applying the
Langmuir adsorption isotherm to the protein adsorption data (Section 2.3.2), it was estimated
that a protein loading of - 28 w/w% would constitute a monolayer coverage on CAP if all the
BET surface area were available for protein adsorption. However, a portion of the CAP's
13ET surface area would be occluded due to pores that were smaller than the size of BSA
molecules (Section 2.3.1). Hence, we postulated that a protein loading of 20 w/w% could
55
have exceeded monolayer coverage. Proteins that were adsorbed in multilayers would be
released more quickly and in larger quantity upon dissolution of the underlying apatite.
........
.
.
......................................................................................................................................
....
+-
(a)
20 w/w% BSA in BSA-CAP
L_.--
0.5
--
9 w/w% BSA in BSA-CAP
12
m0)
o
Er
010
0.4
o
0.2
EU
0.E
E
2
0.0
0
0
10
20
30
40
Time (days)
50
60
0
10
20
30
40
Time (days)
50
60
Fig. 2.14. Effect of protein loading in BSA-CAP complex on protein release. The composite
particles labeled "9 w/w% BSA in BSA-CAP" and "20 w/w% BSA in BSA-CAP" contained
9 and 18 glg of FITC-BSA per mg of carrier, respectively (theoretical loading). Results in (b)
are based on BSA content determined by extraction. PLGA molecular weight was 24 kD.
Data shown are, mean values for n = 2.
2.3.6.6
BSA-Apatite Complex Loading in Particles
Different amounts of BSA-CAP with the same BSA loading (9 w/w%) were
introduced to 24 kD PLGA to form composite particles. Increasing the BSA-CAP loading
diminished the amount of BSA released (Fig. 2.15a). The polymer-apatite composite particles
behaved very differently from the polymeric particles prepared by the W/O/W emulsification
method; the latter demonstrated faster and increased protein release in the absence of apatite
particles when the protein loading was increased [45, 48]. This was because for a given BSA
loading on apatite, increasing the BSA-CAP content in the composite particles would give
rise to increasing apatite/polymer content, so that less acid would be produced from polymer
degradation per unit mass of carrier to dissolve a larger amount of apatite. In addition, the
alleviation of acidity by larger amounts of apatite per unit mass of carrier would diminish the
autocatalytic effect of acid on polymer hydrolysis. Hence, the overall effect of increasing
I3SA-CAP content was a reduction in protein release from the composite particles.
56
.,
U.
nnI
4U
0.8
35
M 0.7
30
' 0.6
_D25
U) 0.5
m
U)20
at 0.4
X_
m 0.3
I
0.2
a
0.1
15
E 10
0
5
0
0.0
0
10
20
30
40
Time (days)
50
60
0
10
20
30
40
Time (days)
50
60
Fig. 2.15. Effect of BSA-CAP complex loading on protein release, expressed as (a)
cumulative mass of BSA released per mg of carrier and (b) cumulative percentage of BSA
released.
BSA-CAP loadings in the composite particles were 2.0, 3.5, 10.7 and 16.7 w/w%.
The composite particles contained a theoretical loading of 2, 3, 9 and 15 ptg of FITC-BSA per
mg of carrier, respectively. Results in (b) are based on BSA content determined by extraction.
PLGA molecular weight was 24 kD. Data shown are mean values for n = 2.
If the BSA-CAP loading was decreased to a very low level, the complex would be
quickly depleted by acid evolution from PLGA such that at later time points, even with the
continual degradation of PLGA, there would be no CAP to dissolve and no further protein
release. On the other hand, if the proportion of BSA-CAP complex was increased, some of
the complex might remain after complete PLGA degradation.
Plotting the percentage
cumulative release from such composite particles would lead to seemingly low release (Fig.
2.15b).
For our experiments, we have chosen to overload rather than underload protein-
apatite complexes so that protein was continually released and tracked over the course of the
study. In actual applications, the complex loading can be adjusted so that a high percentage
of the precious therapeutic protein gets released before complete polymer degradation.
2.3. 7 Comparison of Protein Release from Polymeric Microspheres Prepared by Double
Emulsion Method
The release profiles of polymeric microspheres prepared by the double emulsion
method (Fig. 2. 16a) show similarities to those obtained by other researchers [47] (Fig. 2.1 6b).
For the polymeric microspheres, protein release typically began with an initial burst,
57
sometimes as high as 55% [19, 47], due to the solubilization
of protein that was loosely
adsorbed on the surface or in the pores of the particles. Protein release would then reach a
plateau, as the physically entrapped proteins slowly diffused through tortuous channels in the
particles for release.
In the later stages, an upturn in the release profile might occur due to
polymer degradation, releasing proteins that were polymer-bound. Hence, the early stages of
protein release were diffusion-controlled, and the later stages were degradation-controlled.
Our 4-week release profiles did not reach the degradation-controlled phase. They showed that
the addition of a more hydrophobic polymer, PLA, reduced the initial burst. However, the
release rates for PLGA and PLGA-PLA particles were remarkably similar despite their
difference in composition. In addition, the similarity between our profiles and those of Cleek
et al., who used PEG to modify the profiles, further illustrated the limitation in using material
or processing parameters to alter the release rates of polymeric microparticles.
I.
OU
hFe
50
Q -6
4)
a)
"I
CO 30
m
X
_
.r
-te
EL4
U~t4
20
E
0
10
Cz
0
0
R
15
20
2F
30
Time, Days
Time (days)
Fig. 2.16.
Release of (a) BSA from PLGA (24 kD) and 2:3 PLGA (24 kD)-PLA (25 kD)
microparticles prepared in-house by the double emulsion method, and (b) FITC-IgG from
PLGA microparticles blended with 0, 1, 2 and 5 w/w% of PEG [47]. Data shown in (a) are
mean values for n = 2. Reprinted from [47] with permission from Elsevier.
2.3.8 Degradationof the PolymericPhase of CompositeMicroparticles
Acid release from blank CAP-PLGA composite particles was estimated by tracking
the molecular weight of the remaining polymer by GPC, and converting the mass lost into the
amount of acid released (Fig. 2.17). The conversion was performed using the approximation
58
that a two-fold reduction in molecular weight (which would suggest the breakage of a
polymer chain into two) corresponded to the release of one acid molecule.
4.0
a)
E
0
a
3.5
2o
E
3.0
0o
E
0
a,
(D
2.5
2.0
_0
a.
1.5
1.0
0
0
E
0.5
0.0
0
5
10
15
20
25
30
Time (days)
Fig. 2.17. Release of acid from the degradation of PLGA of different molecular weights in
blank CAP-PLGA composite particles, n = 1.
6 kD and 13 kD PLGA degraded rapidly, as indicated by the steepness of the slope in
Fig. 2.17. Polymers of higher molecular weights degraded more gradually, giving rise to a
more constant rate of acid production. The increase in slope at later time points for PLGA of
different molecular weights might be due to an autocatalytic effect. Ester hydrolysis can be
catalyzed by acid or base, and hence, the accumulation of acidic degradation products in the
particles would accelerate the rate of hydrolysis [49].
These profiles bear relation to the profiles in Fig. 2.9 illustrating the effect of polymer
molecular weight on protein release. Lower molecular weight polymers that degraded more
quickly, as indicated by the steeper gradient of the profiles in Fig. 2.17, also showed faster
protein release. For particles fabricated from 24 kD and 59 kD PLGA, the change in the
gradient of the acid evolution profile appears to coincide with the upturn in the protein release
profile.
2.3.9
Calcium Release from Composite Microparticles
Subsequent to polymer degradation and acid release, apatite particles that were
embedded in the polymer matrix would dissolve to produce calcium and phosphate ions. The
59
release of calcium ions was detected with a calcium-sensitive fluorophore (Rhod-5N). A
caveat to the use of this fluorophore was its sensitivity to the pH and salt concentration of the
medium.
Hence, even with a calibration curve, only estimations of the calcium
concentrations could be obtained. Nevertheless, the fluorophore was useful in confirming the
presence of calcium and the relative change in its concentration with time. The results for
blank CAP-PLGA microparticles synthesized with 13 kD and 59 kD PLGA are shown in Fig.
2.18. Comparisons were made with 59 kD BSA-CAP-PLGA microparticles and with bare
CAP.
.
on
18
*
16
E
0)
12
4
+0 102
a 8
2
0
10
20
30
Time (days)
40
50
Fig. 2.18. Calcium release from microparticles prepared with (*) CAP-13 kD PLGA, (x)
CAP-59 kD PLGA, ()
BSA-CAP-59 kD PLGA, and ()
bare CAP. The CAP loading in
the composite particles was 16.67 w/w%; an equivalent amount of bare CAP was employed in
*. Data shown are mean values for n = 2.
For the 59 kD CAP-PLGA microparticles, the calcium release was gradual and bore
similarity to the protein release profile (Fig. 2.9a). The presence of BSA on CAP did not
appear to affect the calcium dissolution rate. However, it was unclear if the increased
magnitude of calcium release was due to a real enhancement in the presence of BSA or a
change in Rhod-5N fluorescence in the presence of BSA.
The 13 kD CAP-PLGA microparticles showed a sharp rise in calcium dissolution in
the first week, followed by a plateau. This resembled its protein release profile (Fig. 2.9a).
Although acid continued to be released past the first week (see Fig. 2.17), there appeared to be
60
little calcium or protein release. An unusual observation was the fall in the cumulative
calcium release profile after 7 days, which suggested that calcium was re-precipitated out of
the release medium, possibly due to interactions with the polymer's anionic groups or due to
pH shifts. The calcium indicator was able to confirm enhanced release of calcium from the
composite microparticles over the passive leaching of calcium ions from bare CAP.
Comparisons amongst the acid, calcium and protein release profiles of the composite
microparticles confirmed that as the apatite particles were dissolved by the acid from polymer
degradation, proteins were released into the medium.
2.4
Summary
Composite microparticles of a biodegradable polymer (PLGA) and a basic inorganic
substrate (apatite) were developed as protein delivery vehicles.
Controlled release was
accomplished through the interplay of polymer degradation to form acidic products and
dissolution of the basic apatitic phase. Consequently, protein that was pre-adsorbed onto
apatite was released.
The uniqueness of the controlled release strategy lies not only in the material
composition, but also in the way the individual components were combined. A key step was
the adsorption of protein onto apatite, thereby preventing its premature release until the
underlying apatitic substrate was dissolved or the protein-apatite interactions were weakened.
As discussed in Section 2.3.3, BSA-apatite interactions were sufficiently strong to frustrate
protein release for more than a year. Without such interactions to retain the proteins,
diffusion out of the carrier particles could be rapid and result in an initial burst, as was the
case for polymeric microparticles prepared by the W/O/W technique (Fig. 2.16). Therefore,
an essential criterion for the use of this release strategy was the adsorption of the therapeutic
agent onto apatite. The high affinity of apatite for proteins of different sizes and IEPs was
demonstrated in Section 2.3.2. An exception, however, was lysozyme. Hence, it is important
to verify the affinity of apatite for the therapeutic protein to be delivered.
The premise that acids from polymer degradation would dissolve the apatitic phase
and promote protein release was supported by the study of the effect of pH on BSA-apatite
complexes (Section 2.3.3). A reduction in pH to 5 caused a substantial increase in BSA
release. However, further pH reductions had the opposite effect, which could be attributed to
61
the stronger electronic interactions between BSA and apatite as BSA became progressively
more positively charged below its IEP (4.7), or to the precipitation of BSA.
The
augmentation in Ca release when CAP or BSA-CAP was encapsulated in PLGA was also
evidence that acid evolution from PLGA degradation accelerated apatite dissolution (Section
2.3.9).
Protein release from these apatite-polymer composite microparticles could be
modified by parameters that influence polymer degradation and/or apatite dissolution.
Amongst the factors studied, those that affect polymer degradation include polymer molecular
weight and polymer hydrophobicity, and those that affect apatite dissolution include apatite
loading and apatite particle size. Polymer molecular weight was found to determine the onset
of the accelerated phase of release; the use of a low molecular weight polymer led to an
immediate upturn in the release profile (burst release).
An increase in polymer
hydrophobicity through the addition of PLA was responsible for slowing aggregate polymer
degradation and consequently, reducing the protein release rate. When a non-degradable,
hydrophilic polymer, PEG, was blended into the particles, the magnitude of protein release
increased whereas the release rate remained approximately the same. PEG encouraged water
penetration throughout the composite particles, enabling degradation of a larger proportion of
PLGA without changing the polymer degradation rate significantly.
Apatite loading in the composite particles was altered by adding bare apatite in
addition to the protein-apatite complex or by adding different proportions of the proteinapatite complex. In both cases, increasing the apatitic content of the composite particles
meant that more acid had to be produced for apatite dissolution to provide for protein release.
The alleviation of acidity by apatite would also lessen the autocatalytic effect of acid on
polymer hydrolysis, further reducing the amount of acidic degradation products formed. The
result was a decrease in protein release rate.
Apatite particle size was observed to have a significant effect on protein release. The
use of sub-micron-sized apatite particles enhanced low-burst protein release that was zeroorder for the first 4 weeks. The enhancement of protein release could be attributed to a
surface effect: smaller apatite particles would sequester a larger proportion of proteins on their
surface than within the pores of the apatite; proteins on the surface would be more readily
released when apatite was eroded from the surface. Zero-order release would be encouraged
62
by a constant rate of acid production and apatite dissolution, as well as homogeneity of
protein distribution on apatite. In the experiment, the same polymer (59 kD PLGA) was used
to encapsulate protein-apatite complexes formed from either sCAP (1 ptm) or CAP (7 tm).
The degradation rate of 59 kD PLGA was found to be relatively constant over 4 weeks (Fig.
2.17). The dissolution rates of sCAP and CAP might differ, but it would be reasonable to
assume that each apatite would dissolve by a fixed amount in response to a given amount of
acid (preliminary experimental data not shown). Thus, protein distribution on apatite seemed
to be the factor driving zero-order release in this case. Protein concentration likely decreased
sharply from the surface to the interior of the particle for the protein-apatite complex prepared
from CAP. Uniform protein distribution might be promoted by using very small (nanometersized) and non-porous apatite particles, or large, highly porous apatite particles with large
pores.
With an understanding of the modulation of protein release by the above factors, we
can attempt to design protein release profiles by selecting specific parameters. For example,
to attain low burst, zero-order and sustained release of reasonable magnitude that might be
suitable for BMIP delivery, the following combination could be used: high molecular weight
polymer (e.g. 59 kD PLGA) of intermediate
hydrophobicity
(little or no PLA added)
encapsulating a protein-sCAP complex with no added bare apatite. An intermediate level of
apatite loading could be used, e.g. 0.08 mg of apatite per mg of polymer. To increase the
magnitude of release, PEG could be blended in, whereas to alter the rate of release, PLA or
bare apatite could be introduced in the appropriate amounts. The next chapter describes the
application of composite particles of selected formulations to the delivery of recombinant
human BMP-2..
2.5
References
[1]
Siegel RA, Langer R. Controlled release of polypeptides and other macromolecules.
Pharm Res 1984;1:2-10.
[2]
Sheffield W. Modification of clearance of therapeutic and potentially therapeutic
proteins. Curr i)rug Targets 2001;1:1-22.
[3]
Topp EM. Barriers to delivery of macromolecules. Med Chem Res 1997;7:493-508.
63
[4]
Crotts G, Park TG. Protein delivery from poly(lactic-co-glycolic acid) biodegradable
microspheres: release kinetics and stability issues. J Microencapsul 1998;15:699-713.
[5]
Meng FT, Ma GH, Qiu W, Su ZG. W/O/W double emulsion technique using ethyl
acetate as organic solvent: effects of its diffusion rate on the characteristics of microparticles.
J Control Release 2003 ;91:407-416.
[6]
Pistel K-F, Kissel T. Effects of salt addition on the microencapsulation of proteins
using W/O/W double emulsion technique. J Microencapsul 2000;17:467-483.
[7]
Yang YrY, Chung TS, Bai XL, Chan WK. Effect of preparation conditions on
morphology and release profiles of biodegradable polymeric microspheres containing protein
fabricated by double emulsion method. Chem Eng Sci 2000;55:2223-2236.
[8]
Jain R.
The manufacturing techniques of various drug loaded biodegradable
poly(lactide-co-glycolide) (PLGA) devices. Biomaterials 2000;21:2475-2490.
[9]
Pays K, Giermanska-Kahn
J, Pouligny
B, Bibette
J, Leal-Calderon
F. Double
emulsions: how does release occur? J Control Release 2002;79:193-205.
[10]
Blanco-Prieto
MJ,
Campanero
MA,
Besseghir
K, Heimgatner
F, Gander
B.
Importance of single or blended polymer types for controlled in vitro release and plasma
levels of a somatostatin analogue entrapped in PLA/PLGA microspheres. J Control Release
2004;96:437-448.
[11]
Jain RA, Rhodes CT, Railkar AM, Malick AW, Shah NH. Controlled delivery of
drugs from a novel injectable in situ formed biodegradable PLGA microsphere system. J
Microencapsul 2000; 17:343-362.
[12]
Lin WJ, Yu CC. The effect of solvent removal conditions on performance and release
property of protein-loaded microparticles. J Microencapsul 2002;19:767-773.
[13]
Bain DF, Munday DL, Smith A. Solvent influence on spray-dried biodegradable
microspheres. J Microencapsul 1999;16:453-474.
[14]
Weidenauer U, Bodmer D, Kissel T. Microencapsulation of hydrophilic drug
substances using biodegradable polyesters. Part II: Implants allowing controlled drug release a feasibility study using bisphosphonates. J Microencapsul 2004;21:137-149.
[15]
Fu YJ, Shyu SS, Su FH, Yu PC. Development
of biodegradable
co-poly(D,L-
lactic/glycolic acid) microspheres for the controlled release of 5-FU by the spray drying
method. Colloid Surf B-Biointerfaces 2002;25:269-279.
64
[16]
Borden MD, El-Amin SF, Attawia A, Laurencin CT. Structural and human cellular
assessment of a novel microsphere-based tissue engineered scaffold for bone repair.
Biomaterials 2003;24:597-609.
[17]
Murakami H, Kobayashi M, Takeuchi H, Kawashima Y. Utilization of poly(DL-
lactide-co-glycolide) nanoparticles for preparation of mini-depot tablets by direct compression.
J Control Release 2000;67:29.
[18]
Soppimath
KS, Aminabhavi
TM. Ethyl acetate as a dispersing
solvent in the
production of poly(DL-lactide-co-glycolide) microspheres: effect of process parameters and
polymer type. J Microencapsul 2002; 19:281-292.
[19]
Coombes AGA, Yeh M-K, Lavelle EC, Davis SS. The control of protein release from
poly(DL-lactide-co-glycolide) microparticles by variation of the external aqueous phase
surfactant in the water-in-oil-in water method. J Control Release 1998;52:311-320.
[20]
Chen X-Q, Yang Y-Y, Wang L, Chung T-S. Effects of inner water volume on the
peculiar surface morphology of microspheres fabricated by double emulsion technique. J
Microencapsul 2001; 18:637-649.
[21]
Cleek RL, Ting KC, Eskin SG, Mikos AG. Microparticles of poly(DL-lactic-co-
glycolic acid)/poly(ethylene glycol) blends for controlled drug delivery. J Control Release
1997;48:259.
[22]
Ruan G, Feng S-S, Li Q-T. Effects of material hydrophobicity on physical properties
of polymeric microspheres formed by double emulsion process. J Control Release
2002;84:151-160.
[23]
van de Weert M, Hennink WE, Jiskoot W. Protein instability in poly(lactic-co-glycolic
acid) microparticles. Pharm Res 2000;17:1159-1167.
[24]
Castellanos IJ, Carrasquillo KG, Lopez JDJ, Alvarez M, Griebenow K. Encapsulation
of bovine serum albumin in poly(lactide-co-glycolide) microspheres by the solid-in-oil-in
water technique. J Pharm Pharmacol 2001;53:167-178.
[25]
Griebenow K, Klibanov AM. On protein denaturation in aqueous-organic mixtures but
not in pure organic solvents. J Am Chem Soc 1996;118:11695-11700.
[26]
Gietz U, Arvinte T, Haner M, Aebi U, Merkle HP. Formulation of sustained release
aqueous Zn-hirudin suspensions. Eur J Pharm Sci 2000; 11:33-41.
65
[27]
Yang TH, Cleland JL, Lam X, Meyer JD, Jones LS, Randolph TW, Manning MC,
Carpenter JF. Effect of zinc binding and precipitation on structures of recombinant human
growth hormone and nerve growth factor. J Pharm Sci 2000;89:1480-1485.
[28]
Johnson OL, Jaworowicz W, Cleland JL, Bailey L, Charnis M, Duenas ET, Wu C,
Shepard D, Magil S, Last T, Jones AJS, Putney S. The stabilization and encapsulation of
human growth hormone into biodegradable microspheres. Pharm Res 1997;14:730-735.
[29]
Yuan Hi, Zou P, Yang Z, Zhang X, De Bruijn JD, de Groot K. Bone morphogenetic
protein and ceramic-induced osteogenesis. J Mater Sci Mater Med 1998;9:717-721.
[30]
Barralet JE, Aldred S, Wright AJ, Coombes AGA. In vitro behavior of albumin-loaded
carbonate hydroxyapatite gel. J Biomed Mater Res 2002;60:360-367.
[31]
Ahn ES. Nanostructured Apatites as Orthopedic Biomaterials [Ph.D. Thesis].
Cambridge, MA: Massachusetts Institute of Technology; 2001.
[32]
Ahn ES, Gleason NJ, Nakahira A, Ying JY. Nanostructure processing of
hydroxyapatite-based
[33]
bioceramics. Nano Lett 2001; 1:149-153.
Ying JY, Ahn ES, Nakahira A. Nanocrystalline apatites and composites, prostheses
incorporating them, and method for their production. U.S. Patent Number 6,013,591, 2000.
[34]
Blanco MD, Alonso MJ. Development and characterization of protein-loaded
poly(lactide-co-glycolide) nanospheres. Eur J Pharm Biopharm 1997;43:287-294.
[35]
Yong TH, Ying JY. Apatite-polymer composite microparticles for the controlled
delivery of bone morphogenetic proteins. To be submitted to Biomaterials.
[36]
Green RJ, Hopkinson
I, Jones RAL. Unfolding and intermolecular
association
in
globular proteins adsorbed at interfaces. Langmuir 1999;15:5102-5110.
[37]
Matsumura H, Kawasaki K, Okumura N, Kambara M, Norde W. Characterization of
the surface of protein-adsorbed dental materials by wetting and streaming potential
measurements. Colloid Surf B-Biointerfaces 2003;32:97-103.
[38]
Misra DN. Adsorption from solutions on synthetic hydroxyapatite: nonaqueous vs.
aqueous solvents. J Biomed Mater Res 1999;48:848-855.
[39]
Kawasaki K, Kambara M, Matsumura H, Norde W. A comparison of the adsorption of
saliva proteins and some typical proteins onto the surface of hydroxyapatite. Colloid Surf BBiointerfaces 2003;32:321-334.
66
[401
Smiciklas ID, Milonjic SK, Pfendt P, Raicevic S. The point of zero charge and
sorption of cadmium (II) and strontium (II) ions on synthetic hydroxyapatite. Separation
Purification Technol 2000;18:185-194.
[41]
Cleland JL, Powell M, Shire S. The development of stable protein formulations: a
close look at protein aggregation, deamidation, and oxidation. Crit Rev Therapeutic Drug
Carrier Systems 1993;10:307-377.
[42]
Matsurnoto T, Okazaki M, Inoue M, Yamaguchi S, Kusunose T, Toyonaga T, Hamada
Y, Takahashi J. Hydroxyapatite particles as a controlled release carrier of protein.
Biomaterials 2004;25:3807-3812.
[43]
Lu L, Stamatas GN, Mikos AG. Controlled release of transforming growth factor 31
from biodegradable polymer microparticles. J Biomed Mater Res 2000;50:440.
[44]
Perez C, Griebenow K. Effect of salts on lysozyme stability at the water-oil interface
and upon encapsulation in poly(lactic-co-glycolic) acid microspheres. Biotechnol Bioeng
2003;82:825-832.
[45]
Yang YY, Chung TS, Ng NP. Morphology,
drug distribution, and in vitro release
profiles of biodegradable polymeric microspheres containing protein fabricated by doubleemulsion solvent extraction/evaporation method. Biomaterials 2001;22:231-241.
[46]
Cleland JL, Lim A, Barron Lo, Duenas ET, Powell MF. Development of a single-shot
subunit vaccine for HIV-1: Part 4. optimizing microencapsulation and pulsatile release of
MN rgpl20 from biodegradable microspheres. J Control Release 1997;47:135-150.
[47]
Cleek RL, Ting KC, Eskin SG, Mikos AG. Microparticles of poly(DL-lactic-co-
glycolic acid)/poly(ethylene glycol) blends for controlled drug delivery. J Control Release
1997;48:259-268.
[48]
Sun SW, Jeong
YI, Jung
SW, Kim
SH. Characterization
of FITC-albumin
encapsulated poly(DL-lactide-co-glycolide) microspheres and its release characteristics. J
Microencapsul 2003;20:479-488.
[149]
within
Fu K, Pack DW, Klibanov AM, Langer RS. Visual evidence of acidic environment
degrading
poly(lactic-co-glycolic
acid)
2000;17:100-106.
67
(PLGA) microspheres.
Pharm
Res
Chapter 3 - Application of Apatite-Polymer Composite Microparticles to the Controlled
Delivery of BMPs
3.1
Introduction
In the previous chapter, the potential of apatite-PLGA composite microparticles for
protein delivery was established. The goal of this chapter is to explore the application of
these composite microparticles to the delivery of bone morphogenetic proteins (BMPs),
specifically, BMP-2. Primary criteria in the delivery of therapeutic proteins include (1) safety
and biocompatibility of the carriers, and (2) preservation of biological activity (bioactivity) of
the therapeutic protein. When these criteria are met, tunability of release from the carriers
would provide added versatility and efficacy over conventional carriers, such as collagen
sponges.
3.1.1
Osteoinductive Effect of BMPs
As their name implies, BMPs promote bone formation by inducing other cell types to
differentiate into bone cells (osteoblasts) that deposit the matrix material constituting bony
tissue [1-7].
BMPs begin their signaling cascade by binding to type I and type II
transmembrane serine/threonine kinase receptors.
The resulting ligand/receptor complex
phosphorylates intracellular signaling proteins, Smad 1, 5 and 8. The activated Smad proteins
heterodimerize with Smad 4, and translocate into the nucleus where they regulate gene
expression associated with osteogenesis [1, 7-11]. Genes that are increased in expression
include those that code for osteocalcin, type I collagen, bone sialoprotein and alkaline
phosphatase (ALP). These proteins are typically expressed by osteoblasts during matrix
deposition and mineralization, and are considered markers of the osteoblastic phenotype. In
particular, ALP and osteocalcin were chosen as markers in our studies. ALP is a glycosylated
enzyme attached on the surfaces of osteoblasts.
It is able to cleave phosphates off proteins,
and is involved in the regulation of extracellular phosphate concentration. An increase in
ALP activity is associated with the onset of mineralization [11, 12]. Osteocalcin is a small
gamma-carboxylated protein that binds to apatite with high affinity.
It is produced by
osteoblasts during mineralization, and hence, appears at a later stage than ALP. Osteocalcin
is believed to be a regulator of bone formation and mineral maturation; its absence in
knockout mice has been linked to excessive calcification [12].
68
BMP-induced osteoblastic differentiation of uncommitted mesenchymal stem cells
and osteoprogenitor cells has been widely demonstrated [6, 13-21]. To assess the bioactivity
of BMPs, we selected a pluripotent murine embryonic fibroblast cell line, C3H10T1/2, which
can be induced to differentiate along osteoblast, chondrocyte, adipocyte or myoblast lineages,
depending on the signals provided [16, 18, 20].
BMPs promote the development
of the
osteoblastic phenotype in these cells, which can be assessed by an increased expression of
ALP and osteocalcin.
3.2
Experimental
.3.2.1 Adsorption of BMPs onto Apatite
Recombinant human bone morphogenetic proteins (rhBMPs) were purchased from
R&D Systems. The capacity of apatite for adsorbing BMPs was tested using three members
of the BMP family, rhBMP-2, rhBMP-4 and rhBMP-6. Each rhBMP was added to 5 mg of
carbonated apatite (CAP) in the amounts of 0, 2.5, 5.0 or 10.0 jpg in 1 ml of deionized (DI)
water. Duplicates of each combination were tested. The suspensions were stirred at room
temperature for 8 h, and then centrifuged.
The supernatant was collected and filtered.
The
amount of rhBMP adsorbed was determined by measuring the difference in concentration
between the original stock solution and the supernatant. RhBMP concentration was evaluated
with a sandwich enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems) specific to
each rhBMP. In general, the assay consisted of adding 50 tplof standards and test samples in
duplicates to a 96-well plate coated with anti-BMP monoclonal antibody. The plate was
incubated at room temperature for 2 h and washed. Biotinylated anti-BMP antibody was then
added, and the plate was further incubated for 2 h. Unbound antibody was removed by
washing, following which streptavidin-horse radish peroxidase was introduced. After 30 min,
color was developed by the addition of hydrogen peroxide and tetramethylbenzidine.
Absorbance was read at 450 nm using a UV-Vis microplate reader (VersaMax, Molecular
Devices) with vvavelength correction at 570 nm.
Only srnall quantities of rhBMPs were tested for adsorption onto apatite as their
potency suggested that only tiny amounts were needed to achieve efficacy in subsequent
experiments. For example, the ED50 of rhBMP-2 for the induction of ALP activity in ATDC5
chondrogenic cells is 100-300 ng/ml according to the manufacturer, R&D Systems. (ED50 or
69
effective dose 50 is the amount required to achieve a specific effect in 50% of the cell or
animal population.) The high cost of these cytokines (- US$40/tg) was also taken into
consideration in the planning of the experiments so as to maximize utility and conservation.
For the preparation of rhBMP-apatite complexes for encapsulation in composite
microparticles, a protein mixture of 1.4 w/w% rhBMP and 98.6 w/w% BSA was used. As the
amount of rhBMPs required was small, the intention was for BSA to cap some of the
adsorption sites, especially in the pores of the apatite, so that rhBMPs, which are less than
one-third the size of BSA, would not tunnel too deep into the apatite particle and result in
delayed release. Adsorption was conducted under aseptic conditions in a biosafety cabinet.
Sterile, cell culture grade water (Sigma) was used to prepare the protein solutions, which were
subsequently sterile-filtered.
The apatite was either sterilized with 70% ethanol and
lyophilized, or sterilized by UV irradiation. A typical adsorption consisted of adding 40 ptgof
rhBMP-2 and 2800 jtg of BSA to 20 mg of CAP in 10 ml of water. After adsorption for 8 h,
the suspension was lyophilized. Washing of the rhBMP-apatite complex to remove any free
protein was not performed in order to minimize loss of these expensive proteins. Although
the unadsorbed proteins may lead to an initial burst during release, some studies have shown
that an initial burst of BMP may be desirable in jump-starting osteoinduction by
chemotactically attracting cells to the wound site. A sustained release of BMPs is then
required to differentiate the cells into osteoblasts, and to maintain their phenotype for bone
deposition [22].
3.2.2
Aseptic Preparation of rhBMP-2-Loaded Composite Microparticles
Of the rhBMPs tested, rhBMP-2 has been more extensively studied in the literature,
and has been reported to have higher potency [1, 6, 8]. Hence, it was used in the majority of
the formulations. Composite microparticles encapsulating rhBMP-2 were synthesized under
aseptic conditions by a solid-in-oil-in-water (S/O/W) emulsion process similar to that
described in Section 2.2.5 [23]. A typical synthesis involved dissolving 250 mg of PLGA
(Alkermes) in 2 ml of dichloromethane. The polymer solution was sterile-filtered through a
0.45-tm Teflon membrane into a vial containing the rhBMP-2-apatite complex. The mixture
was vortexed to create a solid-in-oil suspension, which was then transferred to 50 ml of 0.1
w/v% methyl cellulose solution that had been autoclaved. A S/O/W suspension was formed
70
by homogenizing
at 8000 rpm for 2 min at room temperature.
To solidify the composite
particles, dichloromethane was evaporated by heating the suspension with stirring at 300 C for
3 h. The particles were collected by centrifugation, washed with sterile water, and freeze
dried.
3.2.3 Estimation of the Magnitude of Initial Burstfrom CompositeMicroparticles
BMPs that were not adsorbed onto apatite were expected to contribute to burst release
when the composite microparticles were introduced to an aqueous medium. To estimate the
amount of free rhBMP-2 in the microparticles, we used a similar procedure as that described
in Section 2.2.8 [23] for determining the protein encapsulation efficiency of composite
microparticles. The polymeric phase of the composite particles was dissolved with 0.5 ml of
dichloromethane per 10 mg of particles.
suspension to extract free protein.
2 ml of DI water were added to the organic
The mixture was vortexed periodically over 6 h to
facilitate extraction, and then centrifuged. Protein associated with apatite settled as a residue,
whereas free protein was dissolved in the supernatant. The rhBMP-2 concentration of the
filtered supernatant was evaluated by ELISA, and was used to calculate the amount of free
rhBMP-2 per rng of composite particles. Each set of particles was tested in duplicates.
RhBMP-2 encapsulation efficiency of the composite particles was not determined as
preliminary experiments suggested that the low pH (pH 2) of the citrate buffer used to
dissolve the apatitic phase led to denaturation/deactivation of the rhBMP-2 molecules and
very poor detection by ELISA.
.3.2.4 Evaluation of In VitroRelease
Release studies were conducted in Eagle's basal medium (BME; Sigma) supplemented
with 10 v/v% heat-inactivated fetal bovine serum (Invitrogen) and 1 v/v% antibiotics (100
IU/ml penicillin and 100 tg/ml streptomycin; ATCC).
This medium will be referred to as
'complete BME'. Each set of rhBMP-2-loaded composite microparticles was suspended at a
concentration of 20 mg/ml in complete BME in 2 sample vials, and incubated at 37°C. At
pre-determined
time intervals (1, 4, 7, 10, 14, 18, 22, 26, 30 days, up to 100 days), the vials
were centrifuged, and the supernatant was withdrawn completely and replaced with the same
volume of fresh medium. The collected supernatant was filtered and stored at -20°C until
71
evaluation by an rhBMP-2 ELISA kit. The concentration of the protein released at each time
point was used to construct cumulative release profiles.
In addition to complete BME, serum-free BME and BES buffer, both of pH 7.4, were
tested as release media.
3.2.5 CellularResponse
A murine pluripotent embryonic fibroblast cell line, C3H10T1/2, was obtained from
ATCC. The cell line was maintained in complete BME. Cells were expanded in T-75 flasks
at a density of 2000 cells/cm2 with two medium renewals per week.
Subculturing was
performed before cells reached confluence due to the sensitivity of these cells to contact
inhibition of cell division. 0.25% trypsin/0.53 mM EDTA (ATCC) was used to dissociate
cells, which were then plated onto 24-well plates for in vitro experiments. Cells from the 11th
to 15th passages were used.
.3.2.5.1
Effect of Release Medium Collected at Each Time Point
Different sets of rhBMP-2-loaded composite microparticles were incubated at 37°C in
complete BME at a concentration of 20 mg/ml. Blank composite particles prepared under the
same processing and material parameters but without rhBMP-2 served as the controls. The
medium was renewed every 3-4 days as described in Section 3.2.4. The release medium
collected at each time point was stored at -20 0 C until incubation with cells.
As a positive
reference, Helistat® collagen sponges (Integra Life Sciences) were soaked with rhBMP-2
solution for 10---15min based on protocols in clinical use [24, 25]. Sponges weighing - 7 mg
apiece were cult and were each loaded with 2.0 ptgof rhBMP-2. Loaded and blank (control)
sponges were then transferred into 2 ml of complete BME per sponge for release studies.
ALP activity was determined by a modification of the protocol of Lowry et al. [26].
C3H1OT1/2 cells were seeded on 24-well plates at a density of 6000 cells/cm 2 (- 12,000
cells/well). The cells were allowed to adhere overnight for 16 h. The medium was aspirated
and replaced with 0.5 ml per well of release medium from a specific time point. For each
time point, 5 wells were prepared. In addition, complete BME enriched with 0-2 Vtg/mlof
rhBMP-2 was incubated with cells. For each rhBMP-2 concentration, 5 wells of cells were
72
tested, each containing 0.5 ml of enriched BME. This experiment was intended to establish
the bioactivity of rhBMP-2, as well as to evaluate dose-dependent cellular response.
After 4 days of culture, the cells were rinsed twice with phosphate buffered saline
(Gibco), and the plates were blotted dry. The cells were lysed according to the protocol
provided by R&D Systems. 100 pl of 0.1% Triton X-100 solution (Sigma) containing 150
mM of NaCl and 3 mM of NaHCO3 (pH 9.3) were added to each well. The plates were
incubated at 37°C for 30 min.
25 pl of the cell lysate were assayed in duplicate for ALP
activity using 100 ptlof p-nitrophenyl phosphate substrate solution (pNPP; Sigma) in 96-well
plates. The absorbance at 405 nm was read every 5 min for 30 min at 37°C. ALP cleaves the
phosphate off pNPP to form p-nitrophenol. Using Beer's Law and an extinction coefficient of
18.45 cmL/mmol for p-nitrophenol, the amount ofp-nitrophenol (in nmol) formed per minute
was determined. ALP activity was normalized by the protein concentration of the lysate as
determined by total protein assay (Pierce) and expressed as nmol/minmg protein.
ALP
activity induced by the medium collected at each time point was plotted as a function of time.
.3.2.5.2
Effect of Prolonged Exposure to Release Medium
Based on the results from in vitro release and induced ALP activity at each time point,
a set of composite microparticles with a sustained release profile of bioactive rhBMP-2 was
chosen. This set of composite particles was constructed of 59 kD PLGA and 0.08 mg of
sCAP per mg of PLGA. RhBMP-2 loading in the particles was 145 ng per mg carrier.
The
effect of prolonged exposure of C3H10T1/2 cells to release medium from this set of particles
was studied. As before, the composite microparticles were incubated at 37°C in complete
BME at a concentration of 20 mg/ml.
C3H10OT1/2 cells were seeded in 24-well plates at a
density of 6000 cells/cm2 . Twice per week, the medium in the wells was aspirated and
replaced with 0.5 ml of filtered release medium collected from the particles. Release medium
from blank composite particles served as the control.
For comparison, complete BME
systems enriched with fixed concentrations of rhBMP-2 (10 and 100 ng/ml; 0.5 ml per well)
were used. Cell culture was conducted for 8 weeks in all groups.
At 1, 2, 3, 4, 6 and 8 weeks, cells in 4 wells for each group were lysed, and the ALP
activity of the cell lysate was measured. Each week, the osteocalcin level of the conditioned
73
medium of 4 wells in each group was assayed by sandwich ELISA (Biomedical Technologies,
Inc.). The procedure for this ELISA was similar to that of the rhBMP-2 assay.
Cell proliferation was evaluated for C3H10T1/2 cells exposed to fixed concentrations
of rhBMP-2 (10)or 100 ng/ml), or to release medium collected from blank or rhBMP-2-loaded
composite microparticles.
Cell counting was performed using the CyQUANT® assay
(Molecular Probes). At one week intervals and for each group, 4 wells of cells were detached
with 0.25% trypsin and centrifuged. The cell pellets were lysed by freeze-thaw cycles in a
lysis buffer, and digested with RNase (1.35 Kunitz units/ml; Sigma) for 1 h at room
temperature to reduce interference from RNA. Green fluorescent CyQUANT®DNA binding
dye was then added, and the fluorescence was read at 485/538 nm (ex/em) using a microplate
reader (fMax, Molecular Devices). DNA and cell number standard curves were generated to
quantify the amount of DNA (and associated fluorescence) per cell. This value (- 6.5 pg of
DNA per cell) subsequently allowed for the determination of the number of cells in each test
sample.
3.2.5.3
Statistical Analysis
Statistical analysis was conducted using analysis of variance (ANOVA) with respect
to treatment or time, followed by Tukey-Kramer HSD post hoc test for multiple comparisons
(JMP v.5.0.1). Data were plotted as mean ± standard deviation. A significance level of p <
0.05 was used.
3.3
Results and Discussion
3.3.1 Adsorption of rhBMPs onto Apatite
The affinity of CAP for rhBMP-2, rhBMP-4 and rhBMP-6 was found to be high. For
all BMP loadings tested (2.5, 5.0 and 10.0 Ctgof BMP per 5 mg of CAP), the amount of BMP
remaining in the supematant after adsorption was negligible (Table 3.1). The co-adsorption
of 10 tg of BMP and 700 tg of FITC-BSA onto 5 mg of CAP, which constituted a total
protein loading of 12.4 w/w% on CAP, also led to the complete adsorption of both BMP and
FITC-BSA.
74
Table 3.1. Adsorption of rhBMPs onto 5 mg of CAP.
Amount of Protein
Amount of BMP in
Amount of BMP
Added (tg)
Supernatant (ng)
Adsorbed (g)
rhBMP-2
2.5
0.0
5.0
0.0
5.0
rhBMP-6
10.0
2.5
0.0
0.1
10.0
2.5
5.0
0.0
5.0
rhBMP-6 (+ BSA)
2.5
10.0
0.9
10.0
10.0 (+ 700.0)
25.4
9.97
3.3.2 Estimation of Proportionof Unbound rhBMPs in CompositeMicroparticles
The percentage of rhBMP-2 that was not bound to apatite, and hence, extractable by
water, was found to be 0.06-0.58 w/w% of the theoretical rhBMP-2 loading in the composite
microparticles.
(The actual rhBMP-2 loading was not determined because the low pH
required for dissolving the apatitic phase led to denaturation of rhBMP-2, as discussed in
Section 3.2.3.)
We surmised that the unbound rhBMP-2 contributed to burst release.
Although seemingly small, this proportion of rhBMP-2 was significant when compared to the
actual amount of growth factor released. Depending on the particle formulation, composite
microparticles released between 2.5 w/w% and 15 w/w% of the theoretical BMP loading over
4 weeks (Section 3.3.3). Therefore, an initial burst was expected to be discernible in the
release profiles of rhBMP-2 from the composite microparticles. In contrast, BSA-loaded
composite particles described in Chapter 2 did not exhibit significant burst release because the
BSA-apatite complexes were washed repeatedly to remove unbound BSA before polymer
encapsulation.
.3.3.3 Effect of Material and ProcessingParameterson the In VitroRelease of rhBMP-2
from CompositeMicroparticles
Based on the in vitro release results reported in Chapter 2, a composite particle
formulation of protein-sCAP-59 kD PLGA was hypothesized to be suitable for BMP delivery.
However, since BMP-2 is a smaller molecule (26 kD) and has a very different IEP (- 9) from
BSA, it is reasonable to expect differences in the release profiles of the two proteins. Hence,
we verified the effect on rhBMP-2 release of some of the parameters that were described in
Section 2.3.6.
75
3.3.3.1
PolbvmerMolecular Weight
Polymers of lower molecular weight degraded more quickly, and hence, enhanced the
release of proteins from composite microparticles. This observation, which was discussed in
Section 2.3.6.1 for composite particles encapsulating BSA [23], was further verified with
particles loaded with rhBMP-2 (see Fig. 3.1). By using an equi-portion blend of PLGA of
different molecular weights, the production of acidic degradation products was equalized over
time, leading to a more gradual and sustained release of rhBMP-2.
e~
0
.r4) 7
W
0
-5
a(I)
(U
EM
O,0
24) 40)
=1
0
20
40
60
80
100
Time (days)
Fig. 3.1. Effect of PLGA molecular weight on rhBMP-2 release from composite
microparticles. CAP loading = 0.08 mg/mg of polymer; rhBMP-2 loading = 65 ng/mg of
carrier. Data shown are mean values for n = 2.
.3.3.3.2
Apatite Particle Size
RhBMP-2 was adsorbed onto sCAP (1 tm) and CAP (7 ptm). The rhBMP-2-apatite
complexes formed were subsequently incorporated into separate sets of composite
microparticles.
Apatite particle size was found to have a pronounced
effect on rhBMP-2
release, similar to that on the release of BSA (Section 2.3.6.3) [23]. The smaller apatite
particles presented a larger proportion of rhBMP-2 on their surface than within their pores due
tlotheir higher surface/volume ratio. As the apatitic surface was eroded by acidic degradation
products, sCAP released larger amounts of rhBMP-2 than CAP, accounting for the difference
in release profiles shown in Fig. 3.2.
76
2.5
1
o 2.0
0n
E
0'
-S
o 1.5
U)
C, 1.0
,
0.5
E
0.0
0
5
10
15
20
Time (days)
25
30
Fig. 3.2. Release of rhBMP-2 from composite particles fabricated from 59 kD PLGA, and ()
1-pm sCAP and () 7-pm CAP. Apatite loading = 0.08 mg/mg of polymer; rhBMP-2 loading
= 65 ng/mg of carrier. Data shown are mean values for n = 2.
3.3.3.3
Apatite Loading
When the amount of rhBMP-2 was held constant while the loading of sCAP in the
composite microparticles was increased from 18 to 45 mg per 250 mg of polymer, rhBMP-2
release was found to decrease (Fig. 3.3). This result was in agreement with the observations
reported in Section 2.3.6.4 for BSA release [23]. For the release of a fixed amount of protein
from apatite, a greater extent of polymer degradation would be needed to dissolve a higher
apatite loading.
In addition, the apatite served as a buffer, mitigating the acidity within the
composite particles, and diminishing the autocatalytic effect of pH on polymer hydrolysis.
Hence, the apatite served towards dampening the protein release.
77
3
a)
M
0
cm
E
0 2
00
0
C"
ax
0L
1
0
(U
M
o
E
0o
0
5
10
15
20
Time (days)
25
30
Fig 3.3. Effect of sCAP loading (per 250 mg of polymer) on rhBMP-2 release. Composite
particles were fabricated from 59 kD PLGA, and contained 65 ng of rhBMP-2/mg of carrier.
Data shown are mean values for n = 2.
3.3.3.4
PolymerHydrophobicity
Increasing polymer hydrophobicity by the addition of PLA led to a surprising
observation: rhBMP-2 release was enhanced with increasing polymer hydrophobicity (Fig.
3.4), contrary tlo the results for BSA release discussed in Section 2.3.6.2.1 [23]. Polymer
hydrophobicity was expected to reduce water penetration and polymer degradation, and hence,
decrease protein release. The unexpected result could be explained by considering the release
of structurally intact, biologically active protein versus the total release of both intact and
degraded protein.
The determination of rhBMP-2 concentration by ELISA was highly
specific; rhBM:P-2 molecules had to be of the appropriate conformation
for binding to
antibodies. Denatured rhBMP-2 molecules were not detected, as confirmed by our previous
experience with evaluating the encapsulation efficiency of composite particles whereby
rhBMP-2 was extracted in citrate buffers of low pH (Section 3.2.3) - the low pH led to
rhBMP-2 denaturation and poor detection by ELISA.
Therefore, higher polymer
hydrophobicity could have led to a less aggressive pH environment within the composite
microparticles, contributing to less protein denaturation and a greater release of bioactive
rhBMP-2.
78
30
'E 25
E
m 20
a)
va)
Cl
'D 15
o.,
m 10
5)
0ES
0
0
5
10
15
20
Time (days)
25
30
Fig. 3.4. Effect of polymer hydrophobicity on rhBMP-2 release. Composite particles were
fabricated from 59 kD PLGA and/or 25 kD PLA. sCAP loading = 0.08 mg/mg of polymer;
rhBMP-2 loading = 145 ng/mg of carrier. Data shown are mean values for n = 2.
3.3.3.5
Release Medium
The medium in which composite particles were incubated had a significant impact on
rhBMP-2 release (Fig. 3.5). BES buffer, which was low in ionic strength and contained no
proteins or amino acids, showed the highest release. Release was considerably diminished in
complete BME., which contained a host of amino acids, serum proteins, salts, antibiotics and
other supplements. The variation in rhBMP-2 release might arise from differences in polymer
degradation, apatite dissolution and/or protein deactivation in the different media. To test the
first of these possibilities, degradation of the polymeric component of composite
microparticles in BES buffer and complete BME was tracked by GPC, and found to be similar
(Fig. 3.6).
Differences in apatite dissolution were not examined due to the difficulty in
monitoring calcium release in Ca-rich cell culture medium. The possibility of enzymatic
degradation of rhBMP-2 by serum proteases contained in complete BME was checked by
comparing rhBMP-2 release in serum-free BME. In the latter, rhBMP-2 release was further
suppressed (see Figure 3.5), suggesting that other factors, such as protein precipitation due to
high ionic strength, were at work. This result highlighted the difficulty in extrapolating in
vilro data for in vivo applications since the in vivo milieu would be even more complex. All
79
release studies discussed in this chapter were performed in complete BME unless otherwise
specified.
250
a,
L
o 200
o 150
a) 0100
""
E
0
0
5
10
15
Time (days)
Fig. 3.5. RhBMP-2 release in (m) BES buffer, (*) complete BME, and (e) serum-free BME,
all at pH 7.4. Composite microparticles were fabricated from 59 kD PLGA. sCAP loading =
0.08 mg/mg of polymer; rhBMP-2 loading = 65 ng/mg of carrier. Data shown are mean
values for n = 2.
35
0o
E 30
o9 25
o
20
E 15
0
a
o10
E
5
0
0
20
40
60
Time (days)
Fig. 3.6. Change in PLGA molecular weight with time in () BES buffer and () complete
BME, both at pH 7.4. Composite microparticles were fabricated from 59 kD PLGA. sCAP
loading = 0.08 mg/mg of polymer.
80
3.3.4 Effect of ConditionedMedium Collectedat Each Time Point
The induction of ALP activity in C3H10T1/2 cells by different concentrations of
rhBMP-2 over 4 days was examined. Increasing the rhBMP-2 concentration from 0 ng/well
to 200 ng/well (400 ng/ml) was found to elevate ALP activity (Fig. 3.7). The effect reached
saturation at a rhBMP-2 concentration of 200 ng/well. Thus, there appeared to be an optimum
dosage of rhBMP-2 for evoking ALP expression, which was related to the commitment of
undifferentiated cells to the osteoblast lineage.
.
o,
'
E
20
E
20
15
o
E
>
.>
10
-
5
0
0
5
50
100 200 400 600 800 1000
Amountof rhBMP-2per Well (ng/well)
Fig. 3.7. Effect of rhBMP-2 concentration on induced ALP activity in C3H10T1/2 cells.
Each well contained 0.5 ml of medium; 50 ng/well is equivalent to 100 ng/ml of rhBMP-2.
Asterisks denote statistically significant differences of p < 0.05 compared to the control (0 ng
of rhBMP-2/well). Data shown are mean + standard deviation (SD) for n = 5.
Levels of ALP activity induced by release media collected from rhBMP-2-loaded
composite microparticles are illustrated in Figs. 3.8-3.10.
Cumulative release is given in
units of 'ng/well' to reflect the actual amounts to which the cells were exposed, and to
facilitate comparison with Fig. 3.7.
Fig. 3.8 shows that the release media collected over a period of 100 days were able to
induce elevated ALP expression in C3H10T1/2 cells. These encouraging results suggested
that the released rhBMP-2 retained at least part of its biological activity, and that the release
of bioactive rhBMP-2 over extended periods of time was possible. The experiment was
halted at 100 days, but this set of composite microparticles could conceivably release
81
bioactive rhBMP-2 for even longer periods of time since the cumulative release and ALP
inductive capacity of rhBMP-2 appeared to be experiencing an upturn at 100 days (Fig. 3.8).
50
4.0
45
3.5
40
3.0
0
35
E
E
2.5
30 a,)
0
QE
2.0
20
25
m
a)
.-0
1.5
20
a
<x
C
Q
.
II
a,
1.0
10
0.5
E
5
0
0.0
0
4
10
18
26
34
43
50
60
70
80
90
100
Time (days)
Fig. 3.8. ALP activity induced by release medium collected at each time point from rhBMP2-loaded composite microparticles fabricated from 59 kD PLGA. sCAP loading = 0.08
mg/mg of PLGA; rhBMP-2 loading = 145 ng/mg of carrier. Asterisks denote statistically
significant differences of p < 0.05 compared to control (blank composite particles) at each
time point. Data shown are mean ± SD for n = 5.
When a higher apatite loading was used in the composite microparticles, rhBMP-2
release was dampened, and the capacity for ALP induction declined within the first week (Fig.
3.9). In contrast, when rhBMP-2 release was augmented and prolonged by using a 3:2 blend
of PLGA and PLA in the composite microparticles, ALP expression was significantly
elevated throughout the course of the study (Fig. 3.10).
82
4.0
r
0
a
50
45
3.5
40
35(D
0)
E 2.5
Cn
C,
30
C
E
o
.C
2.0
25
a)
a)
cN
Q.
20
1.5
.)
-_
?
3.0
2
M
15
.a)
10
:
E
5
°
1.0
.J
<~
0.5
0
0.0
0
1
4
7
10
14
18
22
26
Time (days)
Fig. 3.9. ALP activity induced by release medium collected at each time point from rhBMP2-loaded composite microparticles fabricated from 59 kD PLGA. sCAP loading = 0.18
mg/mg PLGA; rhBMP-2 loading = 145 ng/mg of carrier. Asterisks denote statistically
significant differences of p < 0.05 compared to control (blank composite particles) at each
time point. Data shown are mean ± SD for n = 5.
8
160
7
140
6
120
3
0
E
5
100
a)
Co
a)
E
0
4
80
Ci
03
3
60
_
a)
03
C
a)
->
E
<r
2
40
X
E
20
0
O
0
0
4
10
18
26
34
42
50
60
70
Time (days)
Fig. 3.10. ALP activity induced by release medium collected at each time point from rhBMP2-loaded composite microparticles fabricated from a 3:2 blend of 59 kD PLGA and 25 kD
PLA. sCAP loading = 0.08 mg/mg of polymer; rhBMP-2 loading = 145 ng/mg of carrier.
Asterisks denote statistically significant differences of p < 0.05 compared to control (blank
composite particles) at each time point. Data shown are mean ± SD for n = 5.
83
160~~~~~~~~~~__I
160
8
8
BMP-LoadedSponge
EZ1Emptv
7
140
Release per Time Point
-
'~
Sponge
A CumulativeRelease
120
6
I&-
cL
0)
a)
c')
E
-
5
100
X
a)
4
80
L
E
0
E
E
LU
.3 _
3
60
_
2
40
a)
3
4
+ *
~~
~ ~
O
~ ~ ~ ~
20
1
0
0
0
4
10
18
26
34
42
50
60
70
80
90
Time (days)
Fig. 3.11. ALP activity induced by release medium collected at each time point from rhBMP2-loaded Helistat sponges. rhBMP-2 loading = 286 ng/mg of carrier. Asterisks denote
statistically significant differences of p < 0.05 compared to control (empty sponge) at each
time point. Data shown are mean + SD for n = 5.
Howeve:r, the induction of ALP activity was not a simple function of the amount of
rhBMP-2 released. This was made apparent by our experiments with Helistat® collagen
sponges, which were the conventional carriers of BMPs used in spinal fusions. RhBMP-2
release from Helistat® sponges (Fig. 3.11) was similar to that from the composite
microparticles in Fig. 3.10. Nevertheless, the ALP expression level was lower in the former,
and tapered off after 2 weeks despite the continued release of rhBMP-2. The cause of the
lower bioactivity of rhBMP-2 released from these sponges was unclear. The experiment was
repeated to validate these results; the same drop-off in ALP activity was observed in the
second trial as well. In comparison, the set of composite microparticles formulated from a 3:2
blend of PLGA and PLA (Fig. 3.10) appeared to have remarkable bioactivity, and should hold
promise for future in vitro and in vivo investigations.
The higher bioactivity of rhBMP-2 released from composite microparticles was
possibly due to synergy between rhBMP-2 and the calcium and phosphate ions released from
apatite dissolution. These ions alone were not sufficient to increase ALP activity, as inferred
84
from the levels induced by release media from blank composite particles. However, calcium
and phosphate might have boosted the activity of rhBMP-2 by priming the cells for
osteoblastic differentiation. High extracellular calcium levels have been reported to stimulate
osteoblastic proliferation and differentiation through mediation by calcium-sensing receptors
(CaRs) [27, 28-1.
3.3.5 Effect of Prolonged Exposure to Conditioned Medium
3.3.5.1
Measurementof InducedAlkalinePhosphataseActivity
Compared against controls with no exposure to rhBMP-2, prolonged exposure of
C3H10T1/2 cells to the growth factor resulted in elevated levels of ALP expression (Fig.
3.12). However, the increase was not monotonic; ALP activity reached a maximum at - 2
weeks, and then dipped to levels that were maintained over the remainder of the experiment.
Similar profiles were reported by Shea et al., who studied the effect of BMP-7 on C3H1OT1/2
cells, and observed a peak in ALP activity at 8 days [18]. In general, ALP expression would
rise with increasing
osteoblast
differentiation
from stem cells to osteoprogenitors
to
preosteoblasts and finally to osteoblasts. However, ALP activity would then decrease with
the progression of mineralization [29]. The culture of C3H10T1/2 cells with release media
collected from rhBMP-2-loaded composite microparticles produced a similar effect. ALP
activity was considerably raised in the first 2 weeks, and then slowly declined to levels
comparable to the control at the end of 8 weeks (Fig. 3.13).
Fig. 3.12 also shows a dose-
dependent response to rhBMP-2; 50 ng of rhBMP-2/well induced higher levels of ALP
activity than 5 ng/well.
85
25_-
1
o 50 ng
20
')
-
m5 ng
o Control
E
E
E
15
E
10
.5
0
7
14
21
28
35
42
49
56
Time (days)
Fig. 3.12. Effect of length of exposure to rhBMP-2 on ALP activity. RhBMP-2
concentrations of 0 (control), 5 and 50 ng/well were used. Asterisks denote statistically
significant differences of p < 0.05 compared to control at each time point. Data shown are
mean ± SD for n = 4.
n...
uU
o
C
·~
0 50
0)
E 40
C
E
0 30
E
.> 20
0
<
10
0
7
14
21
28
42
56
Time (days)
Fig. 3.13. Effect on ALP activity of prolonged exposure to release media collected from
rhBMP-2-loaded composite microparticles fabricated from 59 kD PLGA. sCAP loading =
0.08 mg/mg of PLGA; rhBMP-2 loading = 145 ng/mg of carrier. Asterisks denote
statistically significant differences of p < 0.05 compared to control (blank particles) at each
time point. Data shown are mean
SD for n = 4.
86
3.3.5.2
Measurement of Osteocalcin Expression
Osteocalcin is a later stage marker of osteoblast differentiation associated with matrix
maturation and mineralization [12].
RhBMP-2 was found to upregulate osteocalcin
expression in C3HIOT1/2 cells in a time- and dose-dependent manner (Fig. 3.14).
Control
cells with no exposure to rhBMP-2 showed low osteocalcin levels, particularly at the start of
the experiment. 5 ng of rhBMP-2/well produced a weak response, which was statistically
insignificant (p > 0.05) over the 8-week study. A more robust response was obtained with 50
ng of rhBMP-2/well; osteocalcin levels were doubled in the first 3 weeks, held steady for the
next 4 weeks, and then peaked again at week 8.
Release media collected from rhBMP-2-loaded composite microparticles had a
remarkable effect on osteocalcin upregulation. Osteocalcin levels increased steadily with
time over 4 weeks, then showed an exponential rise from weeks 4 to 7 (Fig. 3.15). The level
reached at week 7 was 10-fold that of the response to rhBMP-2-enriched BME (50 ng/well)
shown in Fig. 3.14. These results suggested that the cells were experiencing a significant
amount of mineralization activity, indicative of an osteoblastic phenotype.
IU
9
.
8
E
o~
0)77
>
a)
6
C
5
o4
co
0
3
0 2
1
0
7
14
21
28
35
42
49
56
Time (days)
Fig. 3.14. Effect of length of exposure to rhBMP-2 on osteocalcin levels. RhBMP-2
concentrations of 0 (control), 5 and 50 ng/well were used. Asterisks denote statistically
significant differences of p < 0.05 compared to control at each time point.
mean ± SD forln = 4.
87
Data shown are
70
60
E 50
0)
a
40
-1
C
'_3 30
(U
00
Cn
0
20
10
0
7
14
21
28
35
42
49
56
Time (days)
Fig. 3.15. Effect on osteocalcin levels of prolonged exposure to release media collected from
rhBMP-2-loaded composite microparticles fabricated from 59 kD PLGA. sCAP loading =
0.08 mg/mg of PLGA; rhBMP-2 loading = 145 ng/mg of carrier.
Asterisks denote
statistically significant differences of p < 0.05 compared to control (blank particles) at each
time point. Data shown are mean ± SD for n = 4.
3.3.5.3
Evaluationof CellProliferation
RhBMP-2 appeared to have a proliferative effect on C3H10T1/2 cells. In the first 4
weeks, treatment with 50 ng of rhBMP-2 per well led to cell numbers that surpassed those of
treatment with 0 ng (control) and 5 ng of rhBMP-2 per well (Fig. 3.16). By week 5, however,
cell numbers for all 3 groups reached similar levels and began to decline, suggesting that
diffusion and spatial limitations of static cell culture were being manifested.
Release media collected from rhBMP-2-loaded composite microparticles also induced
significant increases in cell number (Fig. 3.17), demonstrating the proliferative capacity of the
released rhBMP-2.
Puleo proposed that rhBMP-2 overrides the contact inhibition of
C3H10T1/2 cells, allowing them to grow in multilayers necessary for in vitro mineralization
[1711. He found that the withdrawal of rhBMP-2 led to a decrease in cell number, purportedly
due to the loss of the ability to maintain cell multilayers.
88
25
20
0
CD
00
.0
E
z
15
10
)
o
5
0
7
14
21
28
35
42
49
56
Time (days)
Fig. 3.16. Effect of length of exposure to rhBMP-2 on cell number. RhBMP-2 concentrations
of 0 (control), 5 and 50 ng/well were used. Asterisks denote statistically significant
differences of p < 0.05 compared to control at each time point. Data shown are mean + SD
for n = 4.
30
25
00
0q
0'X,
2
20
15
E
z
oa)
10
O
5
0
7
14
21
28
42
56
Time (days)
Fig. 3.17. Effect on cell number of prolonged exposure to release media collected from
rhBMP-2-loaded composite microparticles fabricated from 59 kD PLGA. sCAP loading =
0.08 mg/mg of PLGA; rhBMP-2 loading = 145 ng/mg of carrier. Asterisks denote
statistically significant differences of p < 0.05 compared to blank particles at each time point.
Data shown are mean SD for n = 4.
89
Fig. 3.1 7 further shows that between weeks 2 and 6, cell population was smaller for
cultures using release media collected from blank composite particles than for cultures with
no exposure to composite particles or rhBMP-2 (control). This difference in cell proliferation
might be caused by composite particles promoting the adsorption or precipitation of proteins
in the cell culture medium, rendering these proteins unavailable for utilization by cells.
Another possibility was the lower pH caused by acidic PLGA degradation products being less
conducive to cell proliferation. However, the fact that cell number did increase albeit slowly
over 8 weeks suggested that the composite microparticles did not compromise cell viability.
More importantly, the delivery of rhBMP-2 with these composite microparticles led to
significant cell proliferation, overwhelming any negative impact the composite particles might
have had alone. Hence, these particles, especially in combination with growth factors, were
biocompatible, i.e. they were non-toxic and generated an appropriate cell response.
3.4
Summary
Apatite-PLGA composite microparticles were applied to the delivery of rhBMP-2. A
comparison with the results in Section 2.3.6 [23] shows that despite the size and charge
differences between rhBMP-2 and BSA, polymer molecular weight, apatite loading and
apatite particle size had similar effects on the release of both proteins.
However, an
unexpected observation was the increase in rhBMP-2 release with increasing polymer
hydrophobicity created by the addition of PLA (Fig. 3.4). The slower degradation of PLA
would have contributed
to less acid generation
and less severe pH conditions
within
composite microparticles, which might have assuaged the denaturation of rhBMP-2 molecules.
Therefore, while the total amount of rhBMP-2 released might be lower with the introduction
of PLA, the amount of structurally intact rhBMP-2 released was higher. Herein lies the
difference between rhBMP-2 detection by ELISA, which is sensitive to the conformation of
the protein, and BSA detection by Coomassie total protein assay, which is sensitive to certain
amino acid residues (primarily arginine) of proteins and does not discriminate readily between
denatured and intact proteins. Hence, the Coomassie assay measured the total amount of
protein (BSA) released, whereas ELISA measured the fraction of rhBMP-2 with sufficiently
preserved confirmations.
Coomassie was not used to measure the total concentration of
90
rhBMP-2 released since the concentrations, on the order of- ng/ml, were below its detection
limit (- 0.5 ptg/ml).
The bioactivity of rhBMP-2 released from composites microparticles was assessed by
measuring the expression of osteoblastic markers, ALP and osteocalcin, in C3H10T1/2
murine pluripotent fibroblasts after a certain period of exposure to release medium collected
from the composite particles. In the first set of experiments, cells were exposed for 4 days to
release medium from a particular time point. This experimental design facilitated screening
of different sets of particles and minimized the amount of composite particles needed. In the
second set of experiments, cells were cultured for 8 weeks during which the culture medium
was continually replenished with release medium collected from a selected set of particles
with the formulation, rhBMP-2-sCAP-59
kD PLGA.
In the first set of experiments, the
elevation of ALP activity in response to release media implied that the released rhBMP-2 had
retained at least part of its bioactivity. Furthermore, rhBMP-2 released after 70 days, and
even at 100 clays appeared to be active, demonstrating
the potential of apatite-PLGA
composite particles for releasing protein in the bioactive form over extended periods of time.
The expression profiles of ALP and osteocalcin, which were tracked in the second experiment,
were in agreement with those reported in the literature. ALP, an early stage marker, was
found to peak at 2 weeks and then decline (Fig. 3.13), whereas osteocalcin, a later stage
marker, was upregulated significantly after 4 weeks (Fig. 3.15).
A notable observation was the remarkable bioactivity of composite particles prepared
with rhBMP-2-sCAP
and a 3:2 blend of 59 kD PLGA and 25 kD PLA. Even slight levels of
rhBMP-2 released from these particles were able to markedly increase ALP activity in
C3H10T1/2 cells (Fig. 3.10).
Hence, this particle formulation might be superior to the
rhBMP-2-sCAP-59 kD PLGA formulation that was proposed before the unusual effect of
PLA was discovered.
Helistat"' collagen sponges, which were used as a positive reference, were found to
release rhBMP-2 with a higher initial burst than composite particles (16.7% vs. < 2.7%).
Release media collected from these sponges after 2 weeks were not able to raise ALP activity
above the control levels (Fig. 3.11). However, the same amount of rhBMP-2 contained in
release media from composite particles had an inductive effect (Figs. 3.8 and 3.11).
The
higher bioactivity of composite particles might be ascribed to possible synergy between
91
rhBMP-2 and Ca2+/PO43 - ions released from apatite dissolution. In the absence of rhBMP-2,
these ions, as released from blank composite particles, did not exhibit an inductive effect. A
synergistic effect of Ca2+/PO43- seemed reasonable since elevated calcium concentrations are
present in VGivoduring the resorptive phase of bone remodeling and might act on calcium3 - ions might also have
sensing receptors to initiate steps for depositing bone. The Ca2+/PO4
contributed to raising the levels of osteocalcin expressed by cells continually exposed to
release media from composite particles (Fig. 3.14). The strikingly high levels, indicative of
significant mineralization, were 6-12 times higher than the levels expressed by cells
continually exposed to 50 ng of rhBMP-2 per well (100 ng/ml). Over 8 weeks, 50 ng of
rhBMP-2 replenished twice a week added up to a cumulative exposure to 750 ng of rhBMP-2
per well of cells. In contrast, the total amount of rhBMP-2 released by the composite particles,
and hence, the total amount that the cells were exposed to over 8 weeks was less than 40 ng
per well (see Fig. 3.8). This served to highlight the remarkable bioactivity of the rhBMP-2-
loaded composite particles.
,3.5
References
[1]
Hoffman A, Weich HA, Gross G, Hillmann G. Perspectives in the biological function,
the technical and therapeutic application of bone morphogenetic proteins. Appl Microbiol
Biotechnol 2001 ;57:294-308.
[2]
Urist MR. Bone: Formation by autoinduction. Science 1965;150:893-899.
[3]
Sykaras N, Oppermann LA. Bone morphogenetic proteins (BMPs): how do they
function and what can they offer the clinician? J Oral Sci 2003;45:57-73.
[4]
Wozney JM. Overview of bone morphogenetic proteins. Spine 2002;27:S2-S8.
[15]
Zegzula HD, Buck DC, Brekke J, Wozney JM, Hollinger JO. Bone formation with use
of rhBMP-2. J Bone Joint Surg [Am] 1997;79A: 1778-1790.
1-6]
Cheng 1H,Jiang W, Phillips FM, Haydon RC, Peng Y, Zhou L, Luu HH, An N, Breyer
B, Vanichakarn P, Szatkowski JP, Park JY, He TC. Osteogenic activity of the fourteen types
of human bone morphogenetic proteins (BMPs). J Bone Joint Surg [Am] 2003;85A:15441552.
[7]
ten Dijke P, Fu JY, Schaap P, Roelen BAJ. Signal transduction of bone morphogenetic
proteins in osteoblast differentiation. J Bone Joint Surg [Am] 2003;85A:34-38.
92
[8]
Groeneveld EHJ, Burger EH. Bone morphogenetic proteins in human bone
regeneration. Eur J Endocrinol 2000;142:9-21.
[9]
Sodek .1, Cheifetz S. Molecular recognition of osteogenesis. In: Davies JE, editor.
Bone Engineering. Toronto: Em squared incorporated; 2000. p. 31-43.
[10]
Takagi M, Kamiya N, Takahashi T, Ito S, Hasegawa M, Suzuki N, Nakanishi K.
Effects of bone morphogenetic protein-2 and transforming growth factor beta 1 on gene
expression of transcription factors, AJ18 and Runx2 in cultured osteoblastic cells. J Mol Hist
2004;35:81-90.
[11]
Rawadi G, Vayssiere B, Dunn F, Baron R, Roman-Roman S. BMP-2 controls alkaline
phosphatase expression and osteoblast mineralization by a Wnt autocrine loop. J Bone Miner
Res 2003;18:1842-1853.
[12]
Boskey AL, Paschalis E. Matrix proteins and biomineralization.
In: Davies JE, editor.
Bone Engineering. Toronto: Em squared incorporated; 2000. p. 44-62.
[13]
Ahmed N, Sammons J, Khokher MA, Hassan HT. Effect of bone morphogenetic
protein-5 on hematopoietic stem cells and cytokine production in normal human bone marrow
stroma. Exp Hematol 2000;28:1491-1505.
[14]
Dragoo JL, Choi JY, Lieberman JR, Huang J, Zuk PA, Zhang J, Hedrick MH,
Benhaim P. Bone induction by BMP-2 transduced stem cells derived from human fat. J
Orthop Res 2003;21:622-629.
[15]
Jorgensen NR, Henriksen Z, Sorensen OH, Civitelli R. Dexamethasone, BMP-2, and
1,25-dihydroxyvitamin D enhance a more differentiated osteoblast phenotype: validation of
an in vitro model for human bone marrow-derived primary osteoblasts. Steroids 2004;69:219226.
[16]
Katagiri T, Yamaguchi A, Ikeda T, Yoshiki S, Wozney JM, Rosen V, Wang EA,
Tanaka H, Omura S, Suda T. The non-osteogenic mouse pluripotent cell line, C3H10T1/2, is
induced to differentiate into osteoblastic cells by recombinant human bone morphogenetic
protein-2. Biochem Biophys Res Commun 1990;172:295-299.
[117] Puleo DA. Dependence of mesenchymal cell responses on duration of exposure to
bone morphogenetic protein-2 in vitro. J Cell Physiol 1997;173:93-101.
93
[18]
Shea C-M, Edgar CM, Einhorn TA, Gerstenfeld LC. BMP treatment of C3H10T1/2
mesenchymal stem cells induces both chondrogenesis and osteogenesis. J Cell Biochem
2003;90:1112-1127.
[19]
Van der Horst G, Van Bezooijen RL, Deckers MML, Hoogendam J, Visser A, Lowik
CWGM, Karperien M. Differentiation of murine preosteoblastic KS483 cells depends on
autocrine bone! morphogenetic protein signaling during all phases of osteoblast formation.
Bone 2002;31:661-669.
[20]
Wang EA, Israel DI, Kelly S, Luxenberg DP. Bone morphogenetic protein-2 causes
commitment and differentiation in C3H10T1/2 and 3T3 cells. Growth Factors 1993;9:57-71.
[21]
Kim HD, Valentini RF. Retention and activity of BMP-2 in hyaluronic acid-based
scaffolds in vitro. J Biomed Mater Res 2002;59:573-584.
[22]
Li RH, Wozney JM. Delivering on the promise of bone morphogenetic proteins.
Trends Biotech-nol2001;19:255-265.
[23]
Yong TH, Hager EA, Ying JY. New controlled delivery strategy for proteins based on
apatite-polymer composites. To be submitted to Biomaterials.
[24]
Geiger M, Li RH, Friess W. Collagen sponges for bone regeneration with rhBMP-2.
Adv Drug Deliv Rev 2003;55:1613-1629.
[25]
Uludag H, Gao T, Porter TJ, Friess W, Wozney JM. Delivery systems for BMPs:
factors contributing
to protein retention at an application site. J Bone Joint Surg [Am]
2001;83A:S1-128.
[26]
Lowry OH, Roberts NR, Wu ML, Hixon WS, Crawford EJ. The quantitative
histochemistry of brain. J Biol Chem 1954;207:19-37.
[27]
Chattopadhyay N, Yano S, Tfelt-Hansen J, Rooney P, Kanuparthi D, Bandyopadhyay
S, Ren XH, Terwilliger E, Brown EM. Mitogenic action of calcium-sensing receptor on rat
calvarial osteoblasts. Endocrinology 2004; 145:3451-3462.
[28]
Dvorak MM, Riccardi D. Ca2+ as an extracellular
signal in bone. Cell Calcium
2004;35:249-255.
1-29]
Aubin JE. Osteogenic cell differentiation. In: Davies JE, editor. Bone Engineering.
Toronto: Em squared incorporated; 2000. p. 19-30.
94
Chapter 4 - Incorporation of Apatite-Polymer Composite Microparticles into Bulk
Matrices
4.1
Introduction
Chapter 3 [1] demonstrated that apatite-polymer composite microparticles [2] are
capable of extended release of bioactive rhBMP-2. Being on the order of tens of microns,
these microparticles are small enough to be injected into bone fractures and cracks. To fill
large bone defiects, the particles have to be formed into bulk implants.
One possibility is to
compress the particles into the desired shape and size. However, the compaction would
disrupt the microparticles, create low porosity (less than 20%), and alter the release profile of
the therapeutic agent [3-5]. Laurencin et al. have sintered microparticles of 100-300 tm to
form highly interconnected scaffolds with 30-40% porosity [6, 7].
These scaffolds are
capable of supporting osteoblast attachment and proliferation throughout the pore system.
One concern, however, is that the heat treatment may cause protein denaturation; their system,
which did not contain proteins, was sintered at - 150C.
An attractive option is to disperse protein-loaded composite microparticles in a
secondary matrix that constitutes the tissue engineering scaffold.
The composite
microparticles would serve as carriers of growth factors, while the scaffold would provide the
desired mechanical and chemical properties for cell adhesion, migration, and proliferation.
Elisseeff et al. have applied such a concept to cartilage engineering by co-encapsulating
chondrocytes and PLGA microspheres containing insulin growth factor-1 and transforming
growth factor-j3 in PLA-PEG hydrogels formed by photopolymerization [8, 9]. The release
profiles of the growth factors were not specifically controlled, but their presence in the
hydrogel led to increased cell proliferation.
Similarly, Woo et
al. dispersed rhBMP-2-loaded PLGA microspheres
in
carboxymethylcellulose [10]. The release profile of the system was adjusted by varying the
proportion of "immediate release" and "sustained release" microspheres in the implant.
The greater tunability of our composite microparticles should allow increased
flexibility in the design of growth factor release from tissue engineering scaffolds.
Multifactor release at distinct rates can be achieved by dispersing different combinations of
composite microparticles containing different growth factors in a single scaffold.
95
4.1.1 Tissue EngineeringScaffolds
Desirable characteristics of scaffolds for bone tissue engineering include (i)
biocompatibility - for favorable interactions with cells, (ii) biodegradability - to allow
ultimate replacement by natural bone, (iii) high porosity and pore interconnectivity - for cell
migration
and vascularization,
and (iv) mechanical
integrity - to maintain
a stable 3-
dimensional scaffold to support cell growth at the defect site and to prevent encroachment of
soft tissue [11-14].
The dispersion of BMP-loaded apatite-polymer composite microparticles in the
scaffold should preferably be conducted under aqueous conditions since the polymeric
constituent of the microparticles, PLGA, is soluble in many organic solvents.
Hence,
biocompatible and biodegradable hydrogels were selected as the scaffold matrix.
The
particles could be dispersed in an aqueous solution of the polymer, which would be
subsequently crosslinked to entrap the particles in the hydrogel network. Although these
matrices have limited mechanical strength, they would be able to maintain their shape if the
network density were high enough. Particle loading in the hydrogels might also enhance their
mechanical integrity. Hydrogels of interest included naturally derived materials such as
gelatin [15], chitosan [16], cellulose [17], dextran [18] and alginate [19], and synthetic gels
such as PLGA-PEG [20] and PLGA-polyvinyl
alcohol (PLGA-PVA) [21]. Protein delivery
from hydrogels is often rapid due to facile diffusion through the aqueous channels, and due to
the low affinity of some gels (e.g. PEG) for proteins [22]. This rapid diffusion could be
countered by encapsulating the proteins in composite microparticles before dispersion in the
gel.
Gelatin and collagen were selected for their biosafety [15] and their similarity to the
extracellular matrix of bone. Composite gelatin scaffolds were prepared by crosslinking a
suspension of BMP-loaded microparticles in gelatin.
The commercially available and
clinically used Helistat®collagen sponge (Integra Life Sciences) was also used as the scaffold.
BMP-loaded microparticles were introduced into the collagen sponges by injection. The use
of collagen sponges would facilitate comparison with the current clinical method. In addition,
the low pH required for collagen dissolution in water, which might accelerate PLGA
hydrolysis, could be avoided with the prefabricated collagen sponges.
96
4.1.1.1 Gelatin Scaffolds
Gelatin is derived from collagen. Type A and type B gelatins refer to gelatins derived
by acidic and basic denaturation of collagen, respectively. Type A gelatin has an isoelectric
point (IEP) of - 7-9 as opposed to - 3-5 for type B gelatin. BMPs typically have IEPs of- 9.
Hence, at the physiological pH of 7.4, type A gelatin and BMPs are positively charged,
whereas type B gelatin is negatively charged. Through electrostatic interactions, type B
gelatin can retain BMPs, and indeed, such interactions are utilized for delayed protein release
from gelatin [15, 23]. However, our intention was for the scaffold to sequester the composite
particles and interfere minimally with protein release. Therefore, a scaffold material with low
interactions with BMPs was desired, and type A gelatin was selected for our experiments.
Gelatin could be chemically crosslinked by aldehydes or carbodiimides. More highly
crosslinked gels have lower water content and would undergo slower enzymatic degradation
in the body [15]. To extend the life span in vivo, a gel with higher crosslinking and higher
molecular weight should be chosen [24]. Porous gels with large pore sizes could be created
by controlling the freeze drying process [25]. A slower freezing rate would promote growth
rather than nucleation of ice crystals, leading to larger pores after sublimation of the ice.
4.1.1.2
Collagen Sponges
Helistat: absorbable collagen sponges, originally designed as hemostatic agents, are
prepared from collagen obtained from the bovine Achilles tendon. These porous sponges are
weakly crosslinked and are resorbed within - 8 weeks after subcutaneous implantation in rats,
according to the product insert.
4.2
Experimental
4.2.1
Preparation of Composite Gelatin Scaffolds
Apatite-PLGA composite microparticles loaded with either BSA (model protein) or
rhBMP-2 (therapeutic protein) were prepared under aseptic conditions according to the
procedure in Section 3.2.2 [1]. Cell culture grade gelatin powder (Type A, 300 Bloom; Sigma)
was added to sterile water to form a 10 w/v% solution. The gelatin solution was sterilized by
autoclaving. A typical synthesis involved adding 250 mg of protein-loaded apatite-PLGA
composite particles to 5 ml of gelatin solution, giving a particle loading of 33.3 w/w%. The
97
mixture was vortexed briefly, and 0.5 ml was transferred to a chamber on a chamber slide
(Lab-TekTM II, Nunc; 8 chambers total, - 0.7 cm 2 /chamber).
The gelatin mixture was crosslinked with either glutaraldehyde or N-ethyl-N'-(3dimethylaminopropyl)carbodiimide
hydrochloride
(EDAC;
Sigma)
and
N-
hydroxysuccinimide (NHS; Sigma). Glutaraldehyde bridges the amino groups on gelatin to
form an imine linkage whereas EDAC/NHS catalyzes the reaction between a carboxyl group
and an amino group to form an amide linkage. In the case of crosslinking by glutaraldehyde,
20 tl of a 25 w/v% aqueous glutaraldehyde solution was added to the gelatin solution,
resulting in a concentration of 10 w/w% glutaraldehyde in gelatin. The gel was left to set at
room temperature for 3 h. It was then removed from the chamber slide and washed 3 times
with 5 w/v% glycine in water. Glycine was used to cap any unreacted glutaraldehyde, which
is highly cytotoxic. The glycine washes were followed by 3 aqueous washes. All washes
took at least 1 h at 37°C.
Crosslinking by EDAC/NHS was conducted by adding 50 tl of
0.70 mM NHS and 50 ptl of 0.98 mM EDAC to 0.5 ml of gelatin suspension in the chamber.
Crosslinking was allowed to proceed at room temperature for 2 h, after which the composite
gels were rinsed 3 times with water to remove the non-toxic side-product, urea, and unreacted
EDAC/NHS.
The composite gels were frozen at -20 0 C overnight and then freeze dried for 2
days.
For dual release of proteins from these composite gels, co-encapsulation of two sets of
-BSA-loadedcomposite particles was performed. Fast-releasing FITC-BSA-loaded composite
particles were constructed from 250 mg of 6 kD PLGA and 50 mg of sCAP. Tetramethyl
rhodamine isothiocyanate (TRITC)-BSA-loaded composite particles were fabricated from 250
mg of 59 kD PIL[,GA
and 50 mg of sCAP, and had a slower release profile. Protein loading in
both sets of particles was 20 tg/mg of particle. The two sets of particles were dispersed in
equal proportion in gelatin at a particle loading of 28.5 w/w% prior to crosslinking.
RhBMP-2-loaded composite particles constructed from 250 mg of 59 kD PLGA and
40 mg of sCAP were also loaded into the composite gels at a particle loading of 28.5 w/w%.
RhBMP-2 loading was 145 ng per mg of particle. The details of each system are included
with the results in Section 4.3.
98
4.2.2 Preparation of CompositeCollagenSponges
Helistat® collagen sponges were cut into the desired size. For characterization, larger
pieces of-
13 mm x 13 mm x 6 mm and - 14 mg were used. For in vitro studies, sponges of
- 6 mm x 13 mm x 4 mm and - 2.3 mg were used. Protein-loaded composite microparticles
were suspended in sterile water at a concentration of 0.03-0.1 g/ml. Using a 21-gauge needle,
the suspension was carefully injected along the midline of each collagen sponge. Only one
set of composite particles was introduced into the sponges for the release of a single protein;
whereas for dual release of proteins, two sets of protein-loaded composite particles were
mixed in a single suspension before injection. The resulting composite collagen sponges were
frozen at --200 C overnight and freeze dried for 2 days.
Particle loading in the sponges was varied from 0 to 78.5 w/w%. A particle loading
above 50 w/w°%O
implies a larger proportion of particles than matrix; for example, the loading
of 78.5 w/w% was derived from 8.4 mg of particles in 2.3 mg of sponge. These low-density
sponges were able to contain more particles than their own mass. Blank composite particles
as well as composite particles encapsulating FITC-BSA, rhBMP-2 or rhBMP-6 were injected
into the sponges. The details of each system are noted with the results in Section 4.3.
It was apparent that this method of loading collagen sponges with composite particles
suffered from inefficiencies due to loss of particles in the dead volume of the syringe and
needle, and possible leakage of particles from the sponge during injection.
To obtain an
average particle loading efficiency, 12 pre-weighed sponges were injected with a fixed mass
of blank composite particles. After freeze drying, the composite sponges were weighed to
determine the mass of particles loaded and the particle loading efficiency.
4.2.3 Characterizationof Scaffolds and Sponges
4.2.3.1 Mechanical Testing
Composite gelatin scaffolds and composite collagen sponges were swollen in water.
The wet gels and sponges were trimmed to rectangular blocks and their dimensions were
measured with a pair of vernier calipers. For each type of gel or sponge, a minimum of 3
specimen blocks were tested with a Zwick/Roell series Z010 machine equipped with a 10-kN
load cell. A pre-load stress of 5 N and a crosshead speed of 1 mm/min were used. The
machine recorded loading and displacement data, which were converted into stress-strain
99
curves for estimating the compressive modulus of the specimens.
4.2.3.2 PorosityMeasurements
The pore size distribution and porosity of composite gelatin scaffolds and composite
collagen sponges were assessed by mercury porosimetry (Quantachrome PoreMaster 33).
Pore sizes below 10 ltmand greater than 200 tm were not detected with this equipment due
to the experimental settings used. For each type of gel or sponge, a minimum of 2 specimens
was tested.
4.2.4
Evaluation of Release Kinetics
4.2.4.1 Release of BSA
For each type of composite gelatin scaffold or composite collagen sponge tested, two
pieces of the gel or sponge were placed in separate 1.5-ml centrifuge tubes (i.e. n = 2).
Composite gelatin scaffolds were pre-swollen with 10 dtlof sterile water per mg of scaffold.
1 ml of BES buffer at pH 7.4, which was sufficient to keep the gels and sponges submerged,
was added to each tube. The tubes were incubated at 37C.
At 1, 4, 7, 14, 18, 22, 26 days, up
to 42 days, 0.75 ml of release medium was withdrawn from the tubes and replaced with 0.75
ml of fresh buffer. The fluorescence of the release medium was measured at 485/538 nm
(ex/em) for FITC-BSA and 544/590 (ex/em) for TRITC-BSA. Using standard calibration
curves, fluorescence was converted to protein concentration, which was used to construct
cumulative release profiles.
4.2.4.2 Release of rhBMP-2
For the release of rhBMP-2 from the gels and sponges, the same procedure as above
was used except for the following modifications. The release medium was 1 ml of complete
BME instead of BES buffer. At each time point, the release medium was withdrawn entirely,
and replaced with 1 ml of fresh complete BME. RhBMP-2 concentration in the release
medium was determined with an ELISA kit.
4.2.5
Cellular Response
Following the evaluation of release kinetics, it was decided that in vitro testing with
100
pluripotent murine embryonic fibroblasts (C3H10T1/2 line; ATCC) would be conducted on
composite collagen sponges only.
4.2.5.1 Evaluation of Cell Viability by Microscopy
To evaluate cytotoxicity, collagen sponges injected with BSA-loaded composite
particles were used. It was postulated that any cytotoxicity would stem from the materials
involved (collagen, PLGA, apatite, BSA, remnant solvent and surfactant), and that BMPs
were unlikely to have a negative influence on cell viability. Therefore, in the interest of
conserving the expensive cytokine, BSA was used in place of rhBMP-2.
Each piece of composite sponge was placed in a well on a 24-well plate. C3H10T1/2
cells were seeded directly onto the sponges by slowly dripping - 50 I1of complete BME
containing 50,000 cells onto the dry composite sponges. This small volume was chosen to
ensure that the sponge soaked up the entire cell solution with minimal spillage onto the plate.
The sponges were incubated at 370 C for 30 min to allow cell attachment before 1 ml of
complete BME was added to the well in which the sponge was contained.
Cells seeded in 24-
well plates with no exposure to the composite sponges served as controls. The medium in all
wells was renewed twice a week.
Cell viability was evaluated using a Live/Dead cytotoxicity kit (Molecular Probes).
The basis of this assay was the use of two dye agents, calcein AM and ethidium homodimer,
to distinguish between live and dead cells. Cell-permeant calcein AM was converted to bright
green fluorescent
calcein (495/515 ex/em) by enzymes in live cells, whereas ethidium
homodimer entered dead cells with damaged membranes, bound to nucleic acids, and
experienced a significant increase in red fluorescence (495/635 ex/em). Thus, live and dead
cells could be detected simultaneously by their different colors with a fluorescence
microscope.
At day 4 and day 7 of culture, 3 wells of each group were aspirated and washed 2
times with Dulbecco's phosphate buffered saline (DPBS; Sigma). 100 l of a 0.5-PM calcein
AM/ethidium homodimer dye solution were added to each well, and the plate was incubated
at 37°C for 40 min. The wells were then visualized using fluorescence microscopy (Olympus
FV 300 laser scanning confocal microscope with IX-71 base unit).
101
4.2.5.2 Evaluationof CellProliferation
Collagen sponges injected with BSA-loaded composite particles were evaluated for
cell proliferation.
This study was complementary to the visualization of cell viability
described in the previous section. Cells were seeded by two methods.
The first method was
described in Section 4.2.5.1, whereby cells were seeded onto the sponges directly. The
second method involved seeding cells at a density of 6000 cells/cm 2 in each well of a 24-well
plate, and allowing cells to attach overnight before the composite collagen sponge was added
to each well.
In this case, the cells were attached to the bottom of the plate, and not to the
sponge, which floated above the cell layer. Cells with no exposure to composite sponges
served as controls.
Cell culture was conducted for 3 weeks for all groups. Each week, cells in 3 wells of
each group (except controls, where n = 4) were lysed and analyzed for cell number by the
CyQUANT assay according to the procedure described in Section 3.2.5.2 [1]. For samples
whereby cells were seeded onto the sponge, the sponge was transferred to lysis buffer and
subjected to freeze-thaw cycles to release cellular contents. Hence, cells that had grown onto
the bottom of the plate were not included in the cell count.
4.2.5.3 Assessmentof OsteoblasticMarkers
Four groups of composite collagen sponges were tested: (i) sponges injected with
rhBMP-2-loaded composite particles, (ii) sponges injected with rhBMP-6-loaded composite
particles, (iii) sponges injected with an equi-mixture of rhBMP-2-loaded and rhBMP-6-loaded
composite particles, and (iv) sponges injected with blank composite particles. Each sponge
contained the same mass of composite particles; the particle loading was - 66 w/w% (4.4 mg
of particles per 2.3 mg of sponge). For sponges injected with rhBMP-containing composite
particles, the total amount of rhBMP was 638 ng per sponge. All sets of composite particles
were prepared from 250 mg of 59 kD PLGA and 20 mg of sCAP, and contained 145 ng of
BMP per mg of' particle.
C3H10T1/2 cells were seeded in 24-well plates at a density of 6000 cells/cm2 with 1
ml of complete BME per well. After 1 day of culture, a composite sponge was introduced
into each well. The plates were incubated at 37°C. The medium was renewed twice per week.
Every week for 6 weeks, 4 wells of each group were analyzed for ALP activity and
102
osteocalcin level according to the procedures described in Sections 3.2.5.1 and 3.2.5.2,
respectively [1]. In addition, to compare the in vitro potency of rhBMP-6 versus rhBMP-2,
C3H10T1/2 cells were cultured in complete BME enriched with different concentrations of
rhBMP-6 ranging from 0 to 1000 ng/well for 4 days. For each concentration, 5 wells of cells
were tested.
ALP activity was compared to that induced with rhBMP-2 at the same
concentration (using results from Section 3.3.4 [1]).
4.2.5.4 Statistical Analysis
Statistical analysis was conducted using analysis of variance (ANOVA) with respect
to treatment or time, followed by Tukey-Kramer HSD post hoc test for multiple comparisons
(JMP v.5.0.1). Data were plotted as mean ± standard deviation. A significance level of p <
0.05 was used.
4.2.6
Preliminary In Vivo Testing in a Rat Ectopic Model
Two groups of BMP-loaded collagen scaffolds were tested. The dimensions of the
Helistat® collagen sponges used were - 6 mm x 4 mm x 13 mm, weighing - 2.3 mg. The
first group comprised of sponges injected with - 2.4 mg of rhBMP-2-loaded composite
microparticles (with - 1.4 jtg of rhBMP-2 per mg of particle). Thus, the total amount of
rhBMP-2 in the scaffold was - 3 tg. The second group was a positive reference, prepared by
soaking the collagen sponges with 3 ptg of rhBMP-2 for 15-20 min. Controls were either
® sponges (without composite particles) or sponges containing - 2.4 mg of
blank Helistat'R
blank composite microparticles.
Animal studies were conducted in collaboration with the Institute of Bioengineering
and Nanotechnology in Singapore in accordance with the Institute's guidelines for the care
and use of animals.
Male Wistar rats of 7-8 weeks old were anesthetized with an
intraperitoneal injection of sodium pentobarbital (- 50 mg/kg) before each procedure. The
back of each rat was wiped with iodine and 70% ethanol solution. Four incisions, each - 1
cm long, were made on either side of the rat lateral to the spine along the dorsum. Four
sponges of the same group were inserted into the subcutaneous pouches on one side of the rat,
while the other side received the appropriate control (either blank sponge or sponge with
blank particles). Each experimental group consisted of 6 rats. After surgery, food was given
103
ad libitum. No adverse clinical reactions were observed, and all surgical incisions healed
without infection.
At time points of 1, 2 and 4 weeks, 2 rats from each group were anesthetized and
subjected to radiography. The rats were sacrificed, and the collagen scaffolds were removed
and fixed in 10% formaldehyde solution. The specimens were dehydrated, embedded in
paraffin, sectioned, and stained with hematoxylin and eosin (H&E). Histological sections
were examined under a light microscope (Leica).
Based on results from radiography and histology, a second experiment was conducted
to determine the minimum dosage of rhBMP-2 and rhBMP-6 required to induce ectopic bone
formation in vivo. To increase the persistence time of the collagen sponges in vivo, the extent
of crosslinking of the sponges was increased by soaking them in 10 ml of a solution of 0.35
mM EDAC and 0.49 mM NHS for 1 h. The sponges were then washed with sterile water and
freeze-dried. RhBMP-2 and rhBMP-6 were tested independently at dosages of 2, 5, 10, 20
and 50 tg per sponge. Intramuscular sites, which were more vascularized and might be more
responsive than subcutaneous sites, were tested. Each BMP-containing sponge was implanted
in one quadricep of a rat, while the other quadriceps was implanted with a control. For each
dosage, two rats were tested. At 2, 4, 6 and 8 weeks, the rats were anesthetized and X-rayed
to evaluate the development, if any, of a radiopaque area indicative of bone formation.
4.3
Results and Discussion
4.3.1
Mechanical Properties of Scaffolds
A typical stress-strain curve for composite gelatin scaffolds is shown in Fig. 4.1. The
slope of the initial linear segment of the curve was taken as the compressive modulus of the
sample.
104
600
500
400
CL
n, 300
a)
u,
200
100
0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Strain
Fig. 4.1. Typical stress-strain curve for composite gelatin scaffolds. The composite particle
loading of this sample was 33 w/w%.
The compressive moduli of gelatin scaffolds containing different amounts of
composite microparticles are presented in Fig. 4.2. The results indicated that the introduction
of composite particles at low weight percentages led to a reduction in compressive modulus.
As the composite particle loading was raised to 33 w/w% and 43 w/w%, the modulus
increased two-fold compared to that of blank gelatin scaffolds. The composite particles were
believed to act as non-adherent fillers because of the poor bonding between hydrophilic
gelatin and hydrophobic PLGA. Fillers that were capable of strong interactions with the
matrix would be more effective enhancers of mechanical properties [26]. However, even
'without such interfacial interactions, an increase in the volume fraction of the composite
particles was expected to increase the modulus of the composite scaffold because the particles
were stiffer than gelatin [27-29]. The higher modulus also stemmed from the lower porosity
of the scaffold with increasing particle loading (see Section 4.3.2.). The apparent anomaly at
low particle loading was possibly due to the differential swelling of gelatin and composite
particles [30], resulting in the disruption of the gelatin network and the possible introduction
of cracks.
However, beyond a critical particle loading, the composite particles were able to
stiffen the matrix. Other factors that might influence mechanical properties but were not
investigated included composite particle size and particle morphology. In this study, the
composite particles used were spherical and polydisperse with an average size of 45 ± 15 pm.
105
700
600
CL
z-
500
U)
23
-
400
a)
Xm
300
E
200
Cn
100
0
0
5
10
15
33
43
Composite Particle Loading in Gel (w/w%)
Fig. 4.2. Compressive moduli of composite gelatin scaffolds with different composite particle
loadings. 'Composite particle loading' referred to the weight percent of composite particles
with respect to the total weight of the composite gelatin. For example, a scaffold weighing
150 mg with a composite particle loading of 33 w/w% was comprised of 100 mg of gelatin
and 50 mg of composite particles. Data shown are mean ± standard deviation (SD) for n = 3.
Helistat' collagen sponges were found to exhibit very poor mechanical properties.
When wet, the sponges showed a compressive modulus of only 11.5 ± 8.3 kPa. A composite
particle loading of 33 w/w% did not significantly alter the modulus of the sponge (7.4 ± 4.8
kPa).
4.3.2 Porosity and Pore Size Distribution of Scaffolds
To allow for cell migration and proliferation, tissue scaffolds must be porous [12, 25,
131]. Higher porosity would accommodate more cells in the scaffold and enhance cell
distribution, but would impact mechanical properties adversely and reduce surface area for
cell attachment [6, 31]. The porosities of composite gelatin scaffolds and composite collagen
sponges were measured by mercury porosimetry. As the composite particle loading in the
gelatin scaffold was increased from 0 to 33 w/w%, porosity was reduced from 72.0% to
35.5% (Table 4.1).
The porosity of Helistat® collagen sponges, however, appeared to be
unaffected by the presence of composite particles. The particles occluded only a small
fraction of the sponge, leaving the porosity and structure of the rest of the sponge unchanged.
106
Table 4.1 . Porosity of composite gelatin scaffolds and composite collagen sponges.
-
Scaffold
Gelatin
Gelatin
Gelatin
Helistat® Collagen
Composite Particle Loading (w/w%)
Porosity (%)
0
72.0
63.3
35.5
10
33
0
10
33
Helista.t® Collagen
Helistat® Collagen
58.9
57.3
57.6
35
35
30
30
(b)
'B
3
25
25
)
0,
E
>
"o
U) 1 5
') 15
N
E
.N
E1 0
10
0
Z
Z
5
0
0
10
30
50
70
90
110
130
10
30
50
Pore Size (m)
35
__
30
110
130
110
130
..............
(d)
30
e) 25
70
90
Pore Size (m)
325
C.,
a)
E 20
E
E 20
o
V
15
() 15
.N
CN
E
,0 10
0 10
z
z
5
5
0
0
10
30
50
70
90
110
130
10
Pore Size (lm)
30
50
70
90
Pore Size (m)
Fig. 4.3. Pore size distributions of composite gelatin scaffolds loaded with (a) 0 w/w%, (b) 10
w/w%, (c) 33 w/w% and (d) 43 w/w% of composite microparticles.
107
For the ingrowth of bone cells, the optimal pore size has been suggested to be in the
range of 200-400 pm [12, 31]. Unfortunately, the Quantachrome PoreMaster could only
detect pores up to a maximum of 200 pm at the settings that were used. However, the
presence of pores larger than 200 tlm in composite gelatin scaffolds could be deduced from
examining the graphs in Fig. 4.3. With increasing composite particle loading, the fraction of
pores larger than 100 tm seemed to increase. This could be due to occlusion of some of the
larger pores by the composite particles, bringing their pore size down to the range measurable
by the instrument.
The presence of pores greater than 200 pm was confirmed by
environmental scanning electron microscopy (ESEM) (Fig. 4.4). Thus, the gelatin scaffolds
possessed pores that would be suitable for the migration of bone cells. Fig. 4.4 also shows
composite miciroparticles dispersed in the gelatin scaffold. These particles were detected
throughout the scaffold by electron microscopy.
Fig. 4.4. ESEMImicrographs of a composite gelatin scaffold (10 w/w% particles) showing (a)
a. pore greater than 200 pim, and (b) composite microparticles dispersed in the scaffold.
The pore size distribution of Helistat® sponges remained essentially the same despite
the introduction of increasing amounts of composite microparticles (Fig. 4.5). It appeared
that neither porosity (Table 4.1) nor pore size distribution (Fig. 4.5) were much affected by
the presence of the particles, which were confined to a small region of the sponge (particles
were typically injected along the midline of the sponge).
A further observation was that
Helistat® collagen sponges had a wide range of pore sizes (Fig. 4.5), and contained large
pores (> 100 im) that would allow the migration of bone cells. Such pores were also detected
by ESEM (Fig. 4.6a). The micrograph in Fig. 4.6b shows the composite microparticles that
108
had been injected into the sponge along its midline. Regions away from the midline were
devoid of particles. These particles were weakly held in the collagen matrix, and left behind
spherical impressions when they were dislodged by the sectioning of the sponge.
35
35
M.............
a)
30
'"
25
E
20
0
30
Is
c,
8a)
0
C,
I
15
15
N
>
-a) 15
N
CU
E 10
10
z0
Z
So
:
5
5
0
0
10
30
50
90
70
110
10
130
30
50
70
90
110
130
Pore Size (lm)
Pore Size (rm)
....... . .... ............. ........ ..... ......... ....
35
(c)
30
25
C.
E 20
0
a) 15
N
t
z
10
5
0
10
30
50
70
90
110
130
Pore Size (m)
Fig. 4.5. Pore size distributions of composite Helistat® sponges loaded with (a) 0 w/w%, (b)
10 w/w% and (c) 33 w/w% of composite microparticles.
109
Fig. 4.6. ESEM micrographs of the cross-section of a composite collagen sponge showing (a)
pores greater than 200 pm, and (b) composite microparticles held in the sponge.
4.3.3 In VitroRelease of Proteinsfrom Scaffolds
4.3.3.1 Composite Gelatin Scaffolds
Two sets of composite microparticles were prepared, each with different release
kinetics.
One set encapsulating FITC-BSA was fabricated from 6 kD PLGA and sCAP, and
had a fast release profile [2]. The other set encapsulating TRITC-BSA was prepared from 59
kD PLGA and sCAP, and exhibited a slower release [2]. The two sets of composite particles
were dispersed in separate gelatin scaffolds ('solo gels'), as well as combined in equiproportions in a single gelatin scaffold ('mixed gel'). The release profiles of BSA from the
composite particles, 'solo gels', and 'mixed gels' are shown in Fig. 4.7.
Entrapment of the composite particles in a gelatin scaffold led to protein release of
lower magnitude. The gel posed an additional diffusion barrier to the proteins, and served to
smoothen out the release profile, particularly for the set of composite particles encapsulating
FITC-BSA with a significant initial burst. The co-entrapment of two sets of composite
particles in a single gelatin scaffold demonstrated the feasibility of dual protein release.
FITC-BSA and TRITC-BSA were released from the 'mixed gel' with distinct release profiles.
Whether the particles were combined in a 'mixed gel' or kept separate in 'solo gels' made
little difference to their release profiles. Each set of composite particles functioned essentially
on its own in gelatin. Hence, the aggregate protein release profile of a scaffold is the sum of
its parts, which facilitates the design and testing of multifactor release.
110
16
O,- 14
- - -
FTC (narticlre
--
FITC(sologel)
--
---
FITC(nixed nael
TRITC (particle)
- -- - TRITC
(ixed gel)- -e- - TRITC
(sologel)
co 12
C,
co-
,
4'
0
5
10
15
20
25
30
Time (days)
Fig. 4.7. BSA release from composite particles and composite gelatin scaffolds. Protein
release was normalized by the mass of the corresponding set of composite particles. Solid
curves indicate release from the composite particles alone. Composite particles encapsulating
FITC-BSA were fabricated from 250 mg of 6 kD PLGA and 50 mg of sCAP. Composite
particles encapsulating TRITC-BSA were fabricated from 250 mg of 59 kD PLGA and 50 mg
of sCAP. 'Solo gel' refers to release from composite gelatin scaffolds containing one set of
composite particles, whereas 'mixed gel' refers to release from composite gelatin scaffolds
containing an equi-mixture of the two sets of composite particles. The particle loading (28.5
w/w%), and the protein content (20 itgper mg of particle, 200 pIgper gel) of 'solo gels' and
'mixed gels' were the same. Data shown are mean values for n = 2.
Composite particles encapsulating rhBMP-2 were also dispersed in these gelatin
scaffolds.
However, BMP release was found to be nearly undetectable by ELISA.
It is
suspected that the proteins were denatured by the crosslinking reaction during gel formation.
Crosslinking involved the reaction of amino (and carboxyl) groups on peptides; hence, gelatin
and BMPs were similarly susceptible. It was hoped that the BMPs might be shielded from the
crosslinkers by their encapsulation in composite particles, but it appeared that the crosslinkers,
which were small molecules, were able to permeate into the composite particles. Using
glutaraldehyde and EDAC/NHS as crosslinkers led to non-release of rhBMP-2. As further
confirmation, release medium from these rhBMP-2-loaded composite gelatin scaffolds were
used to culture C3H10T1/2 cells. No inductive effect was found, indicating the absence of
bioactive rhBMP-2 in the release medium. This finding highlighted a potential pitfall of using
proteins with no biological
activity, such as BSA, as surrogates in release studies.
111
The
detection of proteins in release media would not constitute a positive event unless their
structure and biological activity were successfully preserved.
Since the gelatin scaffolds were deemed to be unsuitable for BMP delivery, no further
experiments were conducted on these scaffolds. However, if a milder crosslinking reaction
could be devised, these scaffolds would have potential in multifactor release. Alternatively,
protein-loaded composite particles could be dispersed in hydrogels of other chemical
compositions formed under more benign conditions.
For example, Anseth et al. have
dispersed polymeric microparticles containing insulin growth factor-i and transforming
growth factor-3 in PEG hydrogels formed by photopolymerization for cartilage tissue
engineering
[811.
4.3.3.2 Composite Collagen Sponges
Composite microparticles encapsulating either FITC-BSA or rhBMP-2 were injected
into porous Helistat® collagen sponges. Since the sponges were prefabricated, loading of
these sponges with composite particles did not involve a potentially denaturing crosslinking
reaction. The efficiency of loading these sponges with composite particles was estimated to
be 49
26% based on the weighing of 12 sponges before and after particle injection and
freeze drying.
Releases of FITC-BSA and rhBMP-2 from composite collagen sponges are shown in
Figs. 4.8 and 4.9, respectively. As the composite particles were not embedded in the collagen
matrix but were held in the voids of the sponge (see Fig. 4.6b), proteins released from the
composite particles could diffuse rapidly out of the porous sponge.
Consequently, the
releases of FITC-BSA and rhBMP-2 from the composite collagen sponges closely matched
their respective releases from the composite particles alone.
112
16
.0
14
(U
o 12
E
-
v 10
a)
U)
C
8
0
6
C)
a)
.> 4
E
2
0
0
5
10
15
20
25
30
Time (days)
Fig. 4.8. BSA release from (*) composite particles and () composite collagen sponge.
Protein release was normalized by the mass of the composite particles. Composite particles
encapsulating FITC-BSA were fabricated from 250 mg of 6 kD PLGA and 50 mg of sCAP.
The particle loading in the sponge was 78.5 w/w% (8.4 mg of particles in 2.3 mg of sponge),
and BSA content was 25 pg/mg particle. Data shown are mean values for n = 2.
20
Q)
a1
18
16
E,) 14
C
a) 12
(n
10
8
m
a)
CU
r2
6
0
4
2
0
0
20
40
60
80
Time (days)
Fig. 4.9. RhBMP-2 release from (*) composite particles and (A) composite collagen sponge.
Protein release was normalized by the mass of the composite particles. Composite particles
encapsulating rhBMP-2 were fabricated from 150 mg of 59 kD PLGA, 100 mg of 20 kD PLA,
and 20 mg of sCAP. The particle loading in the sponge was 68.5 w/w% (5 mg of particles in
2.3 mg of sponge), and rhBMP-2 content was 145 ng/mg particle. Data shown are mean
values for n
2..
113
4.3.4 Cellular Response
4.3.4.1 Evaluation of Cell Viability by Microscopy
Viable cells were detected on the outer perimeters of the composite collagen sponges
at days 4 and 7 (Fig. 4.10). The interior of the sponges was essentially devoid of cells despite
the seeding of 50,000 cells per sponge. One explanation for the sparse cell population might
be the sieving effect of the sponge. As the pores were not large enough, cells were prevented
from percolating into the center of the sponge and were mostly left on the surface. Cells that
did not adhere well to the sponge before the medium addition were eventually washed onto
the bottom of the well. An examination of the bottom of the wells did reveal a confluent layer
of cells at day 7. Even if the interior of the sponge had been seeded with cells initially,
diffusion limitations of such a static cell culture might have reduced cell viability. Beyond
3.5 mm from the nutrient supply, cell survival has been reported to be poor [32, 33]. For this
reason, dynamic cell culture conditions such as in perfusion bioreactors, which enhance
transport of nutrients and removal of metabolic wastes, are used to create tissue constructs ex
vivo [34-36].
-I,
Fig. 4.10. Fluorescent cell viability staining of cell-seeded composite collagen sponges at (a)
4 days (20x magnification) and (b) 7 days (10x magnification).
Live C3H1OTl/2 cells
appeared green, whereas dead cells and composite particles were stained red. In each figure,
the top left quadrant displays green fluorescence, the top right quadrant displays red
fluorescence, the bottom left quadrant shows the light micrograph of the scaffold, and the
bottom right quadrant shows the superposition of red and green fluorescent images.
114
Cells that did attach to the scaffold appeared viable, and very few dead cells were
detected (Fig. 4.10). In addition, the composite particles were stained red by the ethidium
homodimer dye, allowing their visualization in the collagen sponge.
4.3.4.2 Evaluationof CellProliferation
For cells seeded onto the bottom of wells, the presence of composite collagen sponges
in the medium did not affect cell proliferation.
Over 3 weeks of culture, cell number
exceeded or was comparable to that of wells with no exposure to collagen sponges (Fig. 4. 1).
No cytotoxic substances appeared to have leached from the sponges.
..
9o
100
o
0
x
80
E
60
20
20
0
7
14
21
Time (days)
Fig. 4.11. Number of C3H10T1/2 cells growing () on composite collagen sponges or ( ) on
the bottom of wells containing composite collagen sponges. ( ) Controls were cells with no
exposure to composite collagen sponges. Asterisks denote statistically significant differences
of p < 0.05 compared to control.
Data shown are mean ± SD for n = 3 (samples) or n = 4
(controls).
In the case of sponges directly seeded with cells, the cell population on the sponges
was one-fifth of that on the bottom of wells in the other 2 groups, despite the much higher
seeding density of the former (50,000 cells/sponge vs. 12,000 cells/well). The lower cell
proliferation on the sponges might be due to inefficient seeding and poor cell survival
associated with diffusion limitations, as discussed in the previous section. The number of
cells growing on the composite collagen sponge 7 days after seeding was - 13,200, which
115
translated into a cell seeding efficiency of 26%.
This number was probably an over-
estimation of the actual seeding efficiency since it encompassed 7 days of cell proliferation on
the sponge.
4.3.4.3 Measurementof InducedAlkalinePhosphataseActivity
To determine the effect of the composite collagen sponges on osteoblastic
differentiation, C3H10T1/2 cells were plated in 24-well plates, and sponges were
subsequently introduced to the wells. The sponges were not in direct contact with the cells,
but released BMPs into the milieu, simulating what cells a distance away from the implant
might experience in vivo. Cells were not seeded directly onto the sponges due to the low cell
proliferation (see Section 4.3.4.2).
Four types of composite collagen sponges were investigated: (i) sponges containing
rhBMP-2-loaded composite particles, (ii) sponges containing rhBMP-6-loaded composite
particles, (iii) sponges containing an equi-mixture of rhBMP-2-loaded and rhBMP-6-loaded
composite particles, and (iv) sponges containing blank composite particles.
The particle
loading was the same in all sponges. The composite particle formulation was also the same in
terms of the amount and type of polymer and apatite used. Fig. 4.12 shows the effect on ALP
activity of prolonged cell culture in the presence of the composite collagen sponges. Within
each group, the significant variation in particle loading in the sponges (- 49 ± 26%) led to
large experimental variance in cellular response. For example, ALP activity levels for the
sponge contain:ing rhBMP-2 and rhBMP-6 at day 7 and for the sponge containing rhBMP-2 at
day 14 seemed to be considerably higher than that of the control, and yet were not statistically
significant by ANOVA and Tukey-Kramer post hoc test because of the small sample size (n =
13)
and large experimental variance.
In general, ALP activity of BMP-loaded sponges was
elevated over controls, particularly for sponges containing rhBMP-2 in the first 2 weeks.
Since ALP is an early osteoblastic marker, its levels are expected to diminish with time
(Section 3.3.5.1 [1]).
116
r
',
4
c-
E
0
E
0
E
3
, 2
.t
.>
-
1
<
7
14
21
28
35
42
Time (days)
Fig. 4.12. ALP activity induced by prolonged exposure to composite collagen sponges
containing composite particles loaded with () rhBMP-2, () rhBMP-6, ( ) rhBMP-2 and
rhBMP-6, and (-) no BMP (control). Particles were fabricated from 250 mg of 59 kD PLGA
and 20 mg of sCAP. BMP content was 145 ng/mg particle. Particle loading in the sponges
was 66 w/w%.
Asterisks denote statistically significant differences of p < 0.05 compared to
control. Data shown are mean ± SD for n = 4.
Fig. 4.12 suggests that rhBMP-2 was more potent than rhBMP-6 in the stimulation of
ALP activity over the first 2 weeks of the experiment. However, Cheng et al. reported that in
their experiments, ALP activity induced by BMP-6 was almost 5 times that induced by BMP2 [37]. Instead of adding growth factors exogenously to the culture medium, Cheng et al.
used recombinant adenoviruses expressing various BMPs to transfect C3HIOT1/2 cells,
leading to the endogenous production of the growth factors.
The expression level of the
growth factors was unclear and unavailable for our comparison. In addition, the bioactivity of
growth factors produced endogenously was likely to be different from that added exogenously,
since the commercially obtained proteins were subjected to multiple processing steps.
To compare the bioactivities of rhBMP-2 and rhBMP-6 as received from R&D
Systems, the induction of ALP activity in C3H10T1/2 cells by different concentrations of the
two growth factors was examined.
Both growth factors showed a dose-dependent effect; the
optimum dosage was - 400 ng/well at which rhBMP-6 had a markedly stronger effect than
rhBMP-2 (Fig. 4.13). However, at almost all other concentrations, rhBMP-2 stimulated a
117
slightly higher effect, though the difference was not statistically significant. Based on these
results, a conclusion could not be reached on the relative bioactivity of rhBMP-2 versus
rhBMP-6, but it likely depended on their concentrations in the environment.
Is..
*#
30
C
a- 25
0
1
O,
E
C
-
20
C
*
*
0
E
,7-
rf
I=R
-
II
TI1
>1
C
rT
-
0-
-j
-
r
5
0-
7
0
50
100
200
400
600
800
1000
Amount of BMP per well (ng/well)
Fig. 4.13. Effects of (i) rhBMP-2 and () rhBMP-6 concentrations on induced ALP activity
in C3H10T1/2 cells. Each well was seeded with 12,000 cells and contained 0.5 ml of medium.
Asterisks denote statistically significant differences of p < 0.05 compared to control (0
ng/well of rhBMP-2 or rhBMP-6). Hashes denote statistically significant differences of p <
0.05 between rhBMP-2 and rhBMP-6 treatment of the same concentration. Data shown are
mean ± SD for n = 5.
4.3.4.4 Measurement of Osteocalcin Expression
Exposure to composite collagen sponges releasing rhBMPs led to elevated levels of
osteocalcin after 7 days (Fig. 4.14). In contrast, control cells cultured in the presence of blank
composite collagen sponges expressed osteocalcin weakly over the course of the experiment.
The osteoinductive effect of BMPs was more discernible by the evaluation of osteocalcin as
the osteoblastic marker than by ALP activity.
In agreement with the ALP activity assessement (Fig. 4.12), rhBMP-6 showed lower
inductivity than rhBMP-2. Osteocalcin levels were consistently lower for cells treated with
rhBMP-6. There also appeared to be little or no synergy between BMP-2 and BMP-6 in the
induction of the osteoblastic phenotype at the dosages used. The selection of this combination
of growth factors was arbitrary and was based on availability in our laboratory.
118
A better
choice might have been rhBMP-4 and vascular endothelial growth factor (VEGF) [38]. In
Peng et al.'s study, VEGF alone did not induce bone formation, but by promoting
vascularization, and through its supposed anti-apoptotic effect, VEGF aided in enhancing
nutrient circulation, cell survival, cell recruitment, and ultimately, bone healing [38]. Another
possible combination would be transforming growth factor-33 and rhBMP-2, which was
reported to evoke endochondral ossification in a synergistic fashion [39]. Currently, to
determine an optimal combination of growth factors to use in bone tissue engineering,
multiple hit-or-miss experiments have to be performed. Further elucidation of the bone
healing process, especially the signaling pathways involved, might facilitate a more informed
development of the appropriate cocktail to use.
1U
9
8
7
0)
0a)
-j
5
C
(/3
04
n
0
3
2
1
0
7
14
21
28
35
42
Time (days)
Fig. 4.14. Osteocalcin levels induced by prolonged exposure to composite collagen sponges
containing composite particles loaded with () rhBMP-2, () rhBMP-6, ( ) rhBMP-2 and
rhBMP-6, and ( ) no BMP (control). Particles were fabricated from 250 mg of 59 kD PLGA
and 20 mg of sCAP. BMP content was 145 ng/mg particle. Particle loading in the sponges
was 66 w/w%. Data shown are mean ± SD for n = 4.
4.3.5
Preliminary In Vivo Testing
After 1 and 2 weeks, collagen scaffolds were excised from the rats.
Scaffolds
containing rhBMP-2, either encapsulated within composite microparticles or directly loaded
onto the sponge, were fully resorbed at 4 weeks. Only remnants of the control - blank
composite collagen sponges and blank collagen sponges - were found at 4 weeks.
119
The
explants from week 1 and 2 were stained with hematoxylin and eosin (H&E) (Figs. 4.15 and
4.16). In the histological sections, nuclei appeared blue/blue-black, cytoplasm appeared pink,
red blood cells were red, and the collagen scaffold was stained brown (Mayer's hematoxylin)
or blue (Harris' hematoxylin).
Fig. 4.15 shows histological sections of excised composite collagen sponges, which
had been injected with composite particles. Fig. 4.16 shows histological sections of excised
collagen sponges without composite particles. The presence of rhBMP-2 in the sponges in
both groups led to greater cellularity and vascularity, which were evident 1 week post
implantation.
The higher density of nuclei and lower proportion of scaffold remnants
indicated greater tissue ingrowth. The presence of red blood cells and the distinctive outline
of blood vessells particularly in the 2-week samples suggested that angiogenesis had occurred
in the rhBMP-.2-loaded sponges. The controls without rhBMP-2 showed tissue lining the
perimeter of the scaffolds, but the centers remained largely undegraded and unpopulated.
Examination of the histology slides under higher magnification revealed the presence
of inflammatory cells, such as macrophages and lymphocytes, in areas adjacent to the scaffold
material for boothBMP-loaded and control sponges. An inflammatory response to the scaffold
material was expected, as it was xenographic (bovine origin).
Most of the cells in the
infiltrated tissue were fibroblasts, and the inflammatory response was relatively mild. In
sponges containing composite particles, inflammatory cells were also found in the perimeter
of some of the voids where the composite particles had resided (the particles were dissolved
during the preparation of the histology slides). Regions resembling granulation tissue, which
would typically be present in the early stages of a tissue healing response, were also observed
in BMP-loaded sponges.
No significant difference could be discerned between collagen sponges with rhBMP2-loaded composite particles and collagen sponges directly loaded with rhBMP-2. Both
groups showed approximately the same amount of tissue ingrowth and vascularization,
suggesting that rhBMP-2 delivered from the composite microparticles had comparable
bioactivity to the protein released from the sponges loaded by conventional method. Staining
by von Kossa for calcium deposits was negative for both groups, implying that no bone
formation had occurred. A possible explanation was that the rhBMP-2 dosage (3 itg)was too
low for osteoinluction.
Osteoinduction has been shown to be dose-dependent, and to occur at
120
dosages as low as 1 tg of rhBMP-2 [40-43].
It was postulated that a lower but still
efficacious dosage of rhBMP-2 would lead to more perceptible differences in the
osteoinductivity of composite collagen sponges and conventionally loaded collagen sponges.
Hence, a dosage of 3 iagwas chosen. However, the bioactivity of the protein we employed
might have been lower than those used by other groups, necessitating a higher dosage.
In addition, the rapid resorption of the collagen sponges excluded observations at later
time points when the effect of more sustained BMP delivery might be apparent. Without the
scaffold material, it was extremely difficult to locate in the rats the few milligrams of
composite particles, which were likely undegraded at 4 weeks. Thus, scaffolds with longer
persistence times are needed to contain the particles.
The vascularity observed in the explants might be an indication of the angiogenic
effect of rhBMP-2. BMPs have been reported to promote angiogenesis both in vitro and in
vivo [44-47]. This step, which precedes bone formation, is crucial to bone healing as the
vasculature supplies stem cells, nutrients, mineral elements, and cytokines needed for
osteogenesis.
Of particular interest is the study by Deckers et al. showing that BMPs
stimulate angiogenesis through enhancing the production of vascular endothelial growth
factor A (VEGF-A) by osteoblastic cells [45]. VEGF-A subsequently acts on endothelial
cells that form blood vessels. They found that treatment of 25 ng/ml and 100 ng/ml of
rhBMP-2, rhBMP-4 or rhBMP-6, but not 10 ng/ml, led to calcium deposition by
preosteoblasts. However, all 3 doses stimulated the production of VEGF-A after 72 h of
treatment. Within 14 days, VEGF-A levels of cells treated with 10 ng/ml and 25 ng/ml of
BMP-4 became indistinguishable [45]. These results suggest that low doses of BMPs may be
sufficient for initiating angiogenesis but not osteogenesis, and may explain the vascularity of
the BMP-loaded sponges that we observed. Although the 3 lagdose we had used was too low
to induce ectopic bone formation, the angiogenic activity of rhBMP-2, which might occur at
lower doses, led to increased vascularity that was evident by histology.
121
Blank Composite Collagen Sponge (Control)
RhBMP-2-Loaded Composite Collagen Sponge
'''-"' JfK~~~~~~~~~~~~·
..
...
`"'
"·-r=.:fe·L·;E.·
't
·-
-fa
-. I
r·
lc.
"
·*....*--
·
rri-
;·r
'
~~~~~-~ ~ ~ ~ ~ ~ ~~~~~~·-
~;
C~~
,
.
I..' ~,,i/~:..:.,,
"-
r
..
3
.
7i
r"l"'"
·'
". i
f'~n.
.
·i
.·
Il
4
'r.
-----
.\`' =L;I·· · · -·'
1-
r'
''
;
'
::..'.
-::
.:
-'.'-.:",,
': v .-. >.".'~ .
-,,.,;
f:::'.,.2
:,:,:;,R,'
~.,'"~,~
-:7.,-.'
~:'i:f;;,
,"..~d,!,!:.'.:'~9;~
:..~-.~ :-':?"~, ',..'i"."4'...'.:.
"~,'.'~::f'4~'~~'~*:v~-,"~
'.':c
*.=~!"'",
~ '.":_ "":,.I'{ .c,/
..
£.
` ·
'
ri
-'.(.Y--.`.
r
;
-
Sl
:
t-,;!
:
t.,
.7'.
_
.
.;
_
~~
_,,
IC
-.
c
x
!
-1,
r.·,,
l
"i
,-
;
.
.1.
*
s
.
J
.
X
_~
se
.
*
p··
s
,
e
!"4
_,0.,
2.it
.) ?
· 14
t
V_\1 -
-r*
I
I
i
-r' -
I -,,
-
..· .'......
,. t-!....
~*["
:',: _blood
':'
'
:''.:': .-'- ' L , (~''
.; ',,,
,,i,.,
·
,,,
.
,~~"'
.
.
.
..
Fig. 4.15. H&E staining of excised composite collagen sponges at 1 week (a-d) and 2 weeks
(e, f). Micrographs for rhBMP-2-loaded composite collagen sponges are in the left column (a,
c, and e); micrographs for blank composite collagen sponges are in the right column (b, d, f).
Magnification: lOx (a, b, e, and f) and 20x (c and d). Mayer's hematoxylin was used for all
sections except (e), which was stained with Harris' hematoxylin.
122
RhBMP-2-Loaded Collagen Sponge
_ __
Blank Collagen Sponge (Control)
I- " ' I' _. ii tc20
x
,-1~~~~~~~-,
.. -.- I,-.7.
-
' -' ;-.
.
.~;
I--d
4%Ij·-a1
"'T'"'-
r;
.
' ':
·. .c~
~~~
i: . · :
:..
~ ~.
JI'~I'
,
.
..
,
2 wee~, 1'O
.
..
.. ,.
'I'
· ~'~'
-<t, ..'~,', .1'P~ -',
'""' ,I
:
t~~~~~~~~~~~~~
·
,¥ ' blq
'· ,.,/..'
:' '
'
)
'
~":
'
'
.
.
-.
...
,'.
~:
,'-...-,:,.... _,.': - . ,.':
:'! :',:.A<,
,':,-,., ....,
,.. . .....
.:,.'
'%,',,·
.)
:.?.,
~',.
~'- -, ',
.,
',
'..
· ~ri ,~¥
·..
'
..
,
-.
,...
.,,· i ' .,:...
'.,
-r·~~~~~,:
~,.`..'":':4.'.,
, ,"'.,'
Fig. 4.16. H&E staining of excised collagen sponges at 1 week (a-d) and 2 weeks (e, f).
Micrographs for rhBMP-2-loaded collagen sponges are in the left column (a, c, and e);
micrographs for blank collagen sponges are in the right column (b, d, f). Magnification: 10 x
(a, b, e, and f) and 20 X (c and d). Mayer's hematoxylin was used for all sections except (f),
which was stained with Harris' hematoxylin.
123
4.3.5.1 Evaluationof DosageRequiredfor EctopicBone Formation
To find, the minimum dosage required for osteoinduction in an ectopic site, rhBMP-2
and rhBMP-6 were separately loaded into collagen sponges (no composite particles) at
different concentrations and implanted in the quadriceps of rats. Intramuscular sites were
more vascularized, and might promote faster cellular response. Radiographs of the rats at 2
and 4 weeks are shown in Figs. 4.17 and 4.18 for rhBMP-2 treatment and rhBMP-6 treatment,
respectively.
Radiopaque areas were deemed to signify mineralization, although bone
fonnation has to be confirmed by histological examination. The results indicated that the
minimum dosage associated with radiopacity at 2 weeks was 10 fig of rhBMP-2 or 50 .tg of
rhBMP-6.
At 4 weeks, 5 [ig of rhBMP-2 or 10 ptg of rhBMP-6 was able to produce
radiopacity. These values exceeded the dosage of 3 g that was used in our initial study
(Section 4.3.5), and should be used for future studies. The lower effective dosage of rhBMP2 than rhBMP-6 confirmed the in vitro results (Figs. 4.12 and 4.14) that rhBMP-2 was the
more potent of the two growth factors.
124
2 weeks
4 weeks
- --------- ----
Fig. 4.17. Radiographs at 2 weeks (left column) and 4 weeks (right column) of rats implanted
with collagen sponges loaded with rhBMP-2 of (a, b) 2 gtg, (c, d) 5 gig, (e, f) 10 gg, (g, h) 20
jig, and (i) 50 pg. Black rings highlight radiopaque areas.
125
4 weeks
2 weeks
Fig. 4.18. Radiographs at 2 weeks (left column) and 4 weeks (right column) of rats implanted
with collagen sponges loaded with rhBMP-6 of (a, b) 2 Ctg, (c, d) 5 gtg, (e, f) 10 gg, (g, h) 20
p1g,and (i) 50 jg. Black rings highlight radiopaque areas.
126
4.4
Summary
Protein-loaded apatite-PLGA composite microparticles that we have prepared are
sufficiently small for injection into fractures and small voids in bone. However, for larger
bone defects, the particles need to be formulated into a bulk material that can fill the defect
space. We have chosen to disperse the composite particles in secondary matrices of either
gelatin or collagen.
The intention was for the protein-loaded composite particles to deliver
molecular signals to guide bone formation, while the matrices would provide suitable porosity,
mechanical integrity, and attachment sites for cells. If so desired, multifactor release could be
realized by incorporating different sets of composite particles with distinct profiles and
growth factors in a single scaffold.
Composite gelatin scaffolds were prepared by crosslinking a suspension of proteinloaded composite particles in a gelatin solution. The encapsulation of a model protein, BSA,
led to promising results: dual delivery of FITC-BSA and TRITC-BSA at different rates with a
single scaffold. However, when rhBMP-2-loaded composite particles were dispersed in the
scaffolds, no protein release was detected by ELISA and no osteoinductive effect was seen
when the release medium was used to culture C3H10T1/2 cells.
It appeared that the
crosslinking reaction used to form the gelatin network was deleterious to rhBMP-2.
Glutaraldehyde or EDAC/NHS was used for crosslinking the amino (and carboxyl) groups on
gelatin chains. Those functional groups exist on almost all proteins, hence, rhBMP-2 was also
susceptible to crosslinking and denaturation.
Encapsulation of rhBMP-2 in composite
particles could not shield the proteins from the crosslinkers, possibly because the composite
particles were porous (see Fig. 2.7). These results also point to a caveat in using model
proteins with no measurable bioactivity, such as BSA, as surrogates for therapeutic proteins.
A commonly used model protein with bioactivity is lysozyme, whose ability to hydrolyze [31,4 glycosidic linkages in the cell walls of m. lysodeikticus can be assessed. However, the
adsorption of lysozyme onto apatite was previously found to be low (see Table 2.4, [2]),
which might limit its encapsulation efficiency in composite particles and might not provide
f'or the release of a sufficient amount for testing.
Protein-loaded composite particles were also incorporated into pre-fabricated HelistatO
collagen sponges by injection along the midline of the sponges. Due to losses in the dead
volume of the syringe and leakage during injection, only - 50% of the particles drawn into the
127
syringe were actually deposited in the sponge. Particle distribution was non-uniform and was
concentrated in the midline of the sponge, as was expected for such a method of loading.
These sponges are commercially used as carriers of rhBMP-2 in spinal fusion, and have been
proven to be suitable substrates for cell adhesion, migration and proliferation. However, they
were found to be mechanically very weak (compressive modulus - 10 kPa), and hence, might
not hold up well in load-bearing defects. In spinal fusion, mechanical strength is provided by
metal cages into which BMP-loaded Helistat® sponges are inserted before implantation.
Nevertheless, the use of these pre-fabricated sponges in containing composite particles did not
involve a potentially denaturing crosslinking reaction, and allowed the rapid diffusion of the
proteins out of the sponges once they were released from the composite particles.
The salfety and efficacy of collagen sponges injected with BMP-loaded composite
particles were verified in vitro.
ALP and osteocalcin levels were elevated by sponges
containing rhBMP-2, rhBMP-6, or an equi-mixture of the two growth factors loaded in
composite particles. Osteoinductivity was more readily distinguishable by the assessment of
osteocalcin as the osteoblastic marker. RhBMP-2 appeared to possess higher activity than
rhBMP-6, contrary to a recent study by Cheng et al. [37]. However, others have cited BMP2 and BMP-7 as the most potent of the fourteen BMPs [48, 49]. These two proteins are also
the only two members of the BMP family currently approved by FDA for use in orthopedic
applications.
Preliminary in vivo studies were conducted with rats involving subcutaneous
implantation of collagen sponges injected with BMP-loaded composite particles (composite
sponges) and conventionally loaded collagen sponges.
An ectopic model was used to
disaggregate osteoinduction from osteoconduction. Bone formation in an orthotopic model
might be due to existing osteoblasts migrating into the scaffold and laying down new matrix
(osteoconduction), or due to stem cells and pre-osteoblasts differentiating into osteoblasts that
then deposit new bone (osteoinduction). Since the activity of BMPs is osteoinductive, an
ectopic model would serve to evaluate the bioactivity of BMPs delivered by the sponges in
vivo.
Both groups of BMP-loaded sponges (either conventionally loaded or composite
sponges) showed significantly enhanced cellularity and vascularity over blank controls at 1
and 2 weeks post implantation. Although no mineralization was observed in the BMP-loaded
128
sponges by von Kossa staining and radiography, the marked difference in tissue infiltration
and vasculariz;ation between samples and controls suggested that the BMPs were bioactive in
vivo. The rhBMP-2 dosage used, 3 tg, might have been too low to induce ectopic bone
formation. Studies have shown that the angiogenic effect of BMPs might be manifested at
lower doses than required for osteogenic activity [45].
Therefore, the increased
vascularization that was observed might be an indication of rhBMP-2 activity at low
concentrations..
Our suspicion that the 3 itg dosage was insufficient for evoking osteogenesis was
confirmed by a subsequent investigation showing that 10 tg of rhBMP-2 or 50 [Ig of rhBMP6 were needed to induce radiopacity of the implants at 2 weeks. At 4 weeks, lower dosages of
5 pg of rhBMP-2 or 10 tg of rhBMP-6 were able to elicit a similar response. This study also
affirmed the hi gher potency of rhBMP-2 over rhBMP-6.
In this preliminary in vivo study, comparable results were obtained with
conventionally loaded sponges and composite sponges. With the composite sponges, we had
hoped to demonstrate that a more sustained release of rhBMP-2 might produce a different
pattern or amount of bone formation. Unfortunately, the complete resorption of the collagen
sponges by 4 weeks excluded observations at later time points. Moreover, as aforementioned,
the BMP dosage used was probably inadequate for ectopic bone formation even after an
extended period of time. For future in vivo studies, scaffolds with longer persistence times
and higher BM-P dosages would aid in the evaluation of the effect of different BMP release
-kineticson osteogenesis.
4.5
References
[1]
Yong TH, Ying JY. Apatite-polymer composite microparticles for the controlled
delivery of bone morphogenetic proteins. To be submitted to Biomaterials.
1-2]
Yong T-H,Hager EA, Ying JY. New controlled delivery strategy for proteins based on
apatite-polymer composites. To be submitted to Biomaterials.
13]
Comoglu T, Gonul N, Baykara T. The effects of pressure and direct compression on
tabletting of microsponges. Int J Pharm 2002;242:191-195.
[41
Durucan
C, Brown PW.
Low temperature formation of
calcium-deficient
hydroxyapatite--PLA/PLGAcomposites. J Biomed Mater Res 2000;51:717-725.
129
[5]
Murakami H, Kobayashi M, Takeuchi H, Kawashima Y. Utilization of poly(DL-
lactide-co-glycolide) nanoparticles for preparation of mini-depot tablets by direct compression.
J Control Release 2000;67:29-36.
[6]
Borden MD, El-Amin SF, Attawia A, Laurencin CT. Structural and human cellular
assessment of a novel microsphere-based tissue engineered scaffold for bone repair.
Biomaterials 2003;24:597-609.
[7]
Ambrosio AMA, Sahota JS, Khan Y, Laurencin CT. A novel amorphous calcium
phosphate polymer ceramic for bone repair: I. Synthesis and characterization. J Biomed Mater
Res 2001;58:295.
[8]
Anseth KS, Metters AT, Bryant SJ, Martens PJ, Elisseeff JH, Bowman CN. In situ
forming degradable networks and their application in tissue engineering and drug delivery. J
Control Release 2002;78:199-209.
[9]
Elisseeff J, McIntosh W, Fu K, Blunk BT, Langer R. Controlled-release of IGF-I and
TGF-31 in a photopolymerizing hydrogel for cartilage tissue engineering. J Orthop Res
2001;19:1098-1104.
[10]
Woo BiH, Fink BF, Page R, Schrier JA, Jo YW, Jiang G, DeLuca M, Vasconez HC,
DeLuca PP. Enhancement of bone growth by sustained delivery of recombinant human bone
morphogenetic protein-2 in a polymer matrix. Pharm Res 2001;18:1747-1753.
[11]
Kneser U, Schaefer DJ, Munder B, Klemt C, Andree C, Stark GB. Tissue engineering
of bone. Min Invas Ther & Allied Technol 2002;11:107-116.
[12]
Laurencin CT, Ambrosio AMA, Borden MD, Cooper JA. Tissue engineering:
Orthopedic applications. Annu Rev Biomed Eng 1999;1:19-46.
[13]
Rose FRAJ, Oreffo ROC. Bone tissue engineering: Hope vs. hype. Biochem Biophys
Res Commun 2002;292:1-7.
[14]
Yang S, Leong K-F, Du Z, Chua C-K. The design of scaffolds for use in tissue
engineering. Part I. Traditional factors. Tissue Eng 2001;7:679-689.
[15]
Tabata Y, Ikada Y. Protein release from gelatin matrices. Adv Drug Deliv Rev
1998;31:287-301.
116]
Khor E, Lim LY. Implantable applications of chitin and chitosan. Biomaterials
2003 ;24:2339-2349.
[17]
Fricain JC, Granja PL, Barbosa MA, de Jeso B, Barthe N, Baquey C. Cellulose
130
phosphates as biomaterials. In vivo biocompatibility studies. Biomaterials 2002;23:971-980.
[18]
van Dijk-Wolthuis WNE, Hoogeboom JAM, van Steenbergen MJ, Tsang SKY,
Hennink WE. Degradation and release behavior of dextran-based hydrogels. Macromolecules
1998;30:4639-4645.
[19]
Gombotz WR, Wee SF. Protein release from alginate matrices. Adv Drug Deliv Rev
1998;31:267-285.
[20]
Jeong B, Bae YH, Kim SW. Drug release from biodegradable injectable
thermosensitive hydrogel of PEG-PLGA-PEG triblock copolymers. J Control Release
2000;63:155-1163.
[21]
Mandal TK, Bostanian LA, Graves RA, Chapman SR. Poly(d,l-lactide-co-glycolide)
encapsulated
poly(vinyl
alcohol)
hydrogel
as a drug
delivery
system.
Pharm
Res
2002;19:1713-1719.
[22]
Drury JL, Mooney DJ. Hydrogels for tissue engineering: Scaffold design variables and
applications. Biomaterials 2003;24:4337-4351.
[23]
Tabata Y, Yamada K, Miyamoto S, Nagata I, Kikuchi H, Aoyama I, Tamura M, Ikada
Y. Bone regeneration by basic fibroblast growth factor complexed with biodegradable
hydrogels. Biomaterials 1998;19:807-815.
[24]
Hager EA. Composite gelatin delivery system for bone regeneration [Senior Thesis].
Cambridge, MA: Massachusetts Institute of Technology; 2004.
[25-] O'Brien FJ, Harley BA, Yannas IV, Gibson L. Influence of freezing rate on pore
structure in freeze-dried collagen-GAG scaffolds. Biomaterials 2004;25:1077-1086.
[26]
Stricker F, Bruch M, Mulhaupt R. Mechanical and thermal properties of syndiotactic
polypropene filled with glass beads and talcum. Polymer 1997;38:5347-5353.
1[27]
Chen Y-C, Lin H-C, Lee Y-D. The effects of filler content and size on the properties
of PTFE/SiO 2 composites. J Polym Res 2003;10:247-258.
[128] Juhasz JA, Best SM, Brooks R, Kawashita M, Miyata N, Kokubo T, Nakamura T,
Bonfield W. Mechanical properties of glass-ceramic A-W-polyethylene composites: Effect of
filler content and particle size. Biomaterials 2004;25:949-955.
129]
Maquet V, Boccaccini
AR, Pravata
L, Notingher
I, Jerome R. Porous poly(cc-
hydroxyacid)/Bioglass(R) composite scaffolds for bone tissue engineering. I: Preparation and
in vitro characterisation. Biomaterials 2004;25:4185-4194.
131
[30]
Holland TA, Tabata Y, Mikos AG. In vitro release of transforming growth factor- 1
from gelatin imicroparticles encapsulated in biodegradable, injectable oligo(poly(ethylene
glycol) fumarate) hydrogels. J Control Release 2003;91:299-313.
[31]
Burg KJL, Porter S, Kellam JF. Biomaterial developments for bone tissue engineering.
Biomaterials 2000;21:2347-2359.
[32]
Goldberg VM, Bos GD, Heiple KG, Zika JM, Powell AE. Improved acceptance of
frozen bone allografts in genetically mismatched dogs by immunosuppression. J Bone Joint
Surg [Am] 1984;66A:937-950.
[33]
Kruyt MC, Dhert WJA, Yuan H, Wilson CE, van Blitterswijk CA, Verbout AJ, de
Bruijn JD. Bone tissue engineering in a critical size defect compared to ectopic implantations
in the goat. J Orthop Res 2004;22:544-551.
[34]
Martin I, Wendt D, Heberer M. The role of bioreactors in tissue engineering. Trends
Biotechnol 2004;22:80-86.
[35]
Carrier RL, Rupnick M, Langer R, Schoen FJ, Freed LE, Vunjak-Novakovic
G.
Perfusion improves tissue architecture of engineered cardiac muscle. Tissue Eng 2002;8: 175188.
[3611 Bancroft GN, Sikavitsas VI, Mikos AG. Design of a flow perfusion bioreactor system
for bone tissue-engineering applications. Tissue Eng 2003;9:549-554.
[37]
Cheng II, Jiang W, Phillips FM, Haydon RC, Peng Y, Zhou L, Luu HH, An N, Breyer
B, Vanichakarn P, Szatkowski JP, Park JY, He TC. Osteogenic activity of the fourteen types
of human bone morphogenetic proteins (BMPs). J Bone Joint Surg [Am] 2003;85A:15441552.
[38]
Peng H., Wright V, Usas A, Gearhart B, Shen H-C, Cummins J, Huard J. Synergistic
enhancement of bone formation and healing by stem cell-expressed VEGF and bone
morphogenetic protein-4. J Clin Invest 2002;110:751-759.
1-39] Simmons CA, Alsberg E, Hsiong S, Kim WJ, Mooney DJ. Dual growth factor
delivery and controlled scaffold degradation enhance in vivo bone formation by transplanted
bone marrow stromal cells. Bone 2004;35:562-569.
[140] Bessho K, Carnes DL, Cavin R, Ong JL. Experimental studies on bone induction
using low-molecular-weight poly(DL-lactide-co-glycolide) as a carrier for recombinant
human bone morphogenetic protein-2. J Biomed Mater Res 2002;61:62-65.
132
[41]
Saito N, Okada T, Horiuchi H, Narumichi M, Takahashi J, Nawata M, Ota H, Nozaki
K, Takaoka K. A biodegradable polymer as a cytokine delivery system for inducing bone
formation. Nat Biotechnol 2001;19:332-335.
[42]
Uludag H, D'Augusta
D, Golden J, Li J, Timony G, Riedel R, Wozney
JM.
Implantation of recombinant human bone morphogenetic proteins with biomaterial carriers: A
correlation between protein pharmacokinetics and osteoinduction in the rat ectopic model. J
Biomed Mater Res 2000;50:227-238.
[43]
Yamamloto M, Takahashi Y, Tabata Y. Controlled release by biodegradable hydrogels
enhances the ectopic bone formation of bone morphogenetic protein. Biomaterials
2003;24:4375-4388.
[44]
Carano RAD, Filvaroff EH. Angiogenesis and bone repair. Drug Discov Today
2003;8:980-98)9.
[45]
Deckers MML, van Bezooijen RL, van der Horst G, Hoogendam J, van der Bent C,
Papapoulos SE, Lowik CWGM. Bone morphogenetic proteins stimulate angiogenesis through
osteoblast-derived vascular endothelial growth factor A. Endocrinology 2002;143:1545-1553.
[46]
Moutsatsos IK, Turgeman G, Zhou S, Kurkalli BG, Pelled G, Tzur L, Kelley P,
Stumm N, Mi S, Muller R. Exogenously regulated stem cell-mediated gene therapy for bone
regeneration. Mol Ther 2001 ;3:449-461.
[47]
Yamashita H, Shimizu A, Kato M, Nishitoh H, Ichijo H, Hanyu A, Morita I, Kimura
M, Makishima F, Miyazono K. Growth/differentiation factor-5 induces angiogenesis in vivo.
Exp Cell Res 1)97;235:218-226.
[48]
Groeneveld EHJ, Burger EH. Bone morphogenetic proteins in human bone
regeneration. Eur J Endocrinol 2000;142:9-21.
[49]
Hoffman A, Weich HA, Gross G, Hillmann G. Perspectives in the biological function,
the technical and therapeutic application of bone morphogenetic proteins. Appl Microbiol
Biotech 2001 ;57:294-308.
133
Chapter 5 - Recommendations for Future Work
5.1
Further Investigation of Release Mechanism
Protein release from apatite-PLGA composites is dependent on the interplay of
polymer degradation and apatite dissolution. If left to passive leaching from apatite-protein
complexes, the release of protein as well as calcium ions would be very slow and small in
magnitude (see Sections 2.3.3 and 2.3.9 [1]). To facilitate the design of release profiles, the
causal relationships between polymer degradation, the local pH environment, apatite
dissolution, and protein bioactivity need to be better understood. Estimations of the pH
within the degrading composite microparticles could be accomplished by co-encapsulating
pH-sensitive fluorophores in the particles, and visualizing the fluorescence intensity profiles
across the particles with a confocal microscope [2]. Our measurements of the pH of the
incubation medium (BES buffer or complete BME) suggested that the external environment
was sufficiently buffered and maintained at pH levels of-
6-7. Fu et al. noted a distinct
difference between the interior and exterior pH environment of PLGA microspheres [2].
Hence, it is the pH within degrading composite microparticles that determines the extent of
apatite dissolution, and this pH should be tracked. The results should then be correlated with
data on polymer degradation as evaluated by GPC, release of calcium ions as measured by
calcium-sensitive dyes (e.g. Arsenazo III) or inductively coupled plasma mass spectrometry
(ICP-MS), and protein release as determined by ELISA. In addition, the structure of proteins
-releasedfrom the composite microparticles could be assessed by gel electrophoresis, and their
bioactivity evaluated by binding assays.
The assembling of these various pieces of information could help determine, for
example, the extent of polymer degradation required to generate a particular pH, and the
effect of the pH environment within degrading composite microparticles on the degree of
apatite dissolution and the bioactivity of proteins.
In particular, assessment of protein
structure and bioactivity might shed some light on the difference in potency of rhBMP-2
delivered from composite microparticles and Helistat® collagen sponges (see Section 3.3.4
113])
134
5.2
Exploration of Additional Material Parameters
A number of parameters were used to adjust protein release through their effect on
polymer degradation and/or apatite dissolution. It would be fruitful to explore additional
material parameters, such as new types of polymer and basic inorganic phase. Only poly(uXhydroxy acids) were studied in this thesis; other acid-producing biodegradable polymers, such
as poly(ortho esters), poly(anhydrides) and even newly designed synthetic polymers, would
possess different degradation rates and could alter protein release rates. Since a constant rate
of acid production would facilitate zero-order release, a polymer that undergoes steady
degradation with time would be an attractive candidate.
Apatite was used as the basic inorganic phase because its dissolution results in the
release of calcium and phosphate, which could be beneficial to bone formation. Other nontoxic basic inorganic materials that would release ions that are natural substituents in bone
apatite include magnesium oxide and zinc oxide [4, 5]. These materials could have different
solubilities and affinities for proteins, and could provide another parameter for tailoring the
protein release rate.
5.3
Incorporation of Composite Microparticles into Tissue Engineering Scaffolds
In the preparation of composite gelatin scaffolds, rhBMPs were denatured by the
crosslinking reaction (see Section 4.3.3.1 [6]). However, the potential of such scaffolds in
multifactor release is attractive.
Scaffolds that can be created without deleterious chemical
reactions should be used to encapsulate protein-loaded composite microparticles. Candidate
materials include PEG-based hydrogels synthesized by photopolymerization [7], and alginate
hydrogels crosslinked by divalent ions [8, 9]. Tissue engineering scaffolds with greater
mechanical strength are also highly desirable. Examples include collagen-apatite [10, 11] and
chitin-apatite scaffolds [12].
The optimal combination of growth factors to be delivered from these scaffolds is not
yet clear. Current studies suggest that synergies exist between BMPs and VEGF or TGF-f33
in bone healing [13, 14]. These combinations could be tested with our composite scaffolds
while awaiting the elucidation of the signaling pathways in bone healing and the growth
factors involved.
135
5.4
In Vivo Studies
Preliminary in vivo studies in this thesis were performed with BMP concentrations that
were too low to induce ectopic bone formation in rats (see Section 4.3.5 [6]). These studies
should be repeated with rhBMP-2 dosages of at least 10 tg (see Section 4.3.5.1 [5]). The
effect of the rhBMP-2 release profile on bone formation could also be examined by
incorporating different sets of composite microparticles exhibiting fast, intermediate and slow
release, with accompanying evaluation of rhBMP-2 pharmacokinetics. Pharmacokinetic data
would facilitate comparison between our composites and other carriers, and elucidate the
effect of pharmacokinetics on osteoinduction.
In addition to radiographic and histological examination, biochemical assessment of
the explants for markers such as alkaline phosphatase and osteocalcin would provide further
evidence for osteoblastic differentiation.
A particular composite microparticle formulation was found to be exceptional in
inducing elevated levels of alkaline phosphatase in vitro, which warrants further investigation
in vitro and in vivo. This formulation was a 3:2 blend of 59 kD PLGA and 25 kD PLA with a
sCAP loading of 0.08 mg per mg of polymer (see Section 3.3.4 [3]).
5.5
Use of Composites in Other Shapes and Sizes
Apatite-PLGA composites have been formed into bulk scaffolds and thin films (data
not shown). The dimension of the composite, through its impact on diffusional length, is
another variable in protein release that should be examined. Furthermore, bulk scaffolds have
varying porosity and pore sizes, which influence water penetration and protein release. For
the application of composite thin films as coatings, their homogeneity and their ability to
adhere to different substrates (e.g. plastics, ceramics and metals) should be further studied.
5.6
Application to Other Therapeutic Agents
Any therapeutic agent that can be adsorbed onto apatite and withheld from premature
release is a candidate for tunable release from these apatite-polymer composites.
We foresee
the gainful application of our composite microparticles to the delivery of other growth factors,
such as VEGF, IGF, and TGF-3. Gene delivery might be possible with the encapsulation of
136
DNA-calcium phosphate complexes [15] in these composites. Small organic molecules, such
as ascorbic acid and chlorhexidine have been found to adsorb onto hydroxyapatite [16]; hence,
the delivery of small drug molecules should also be explored.
5.7
References
[1]
Yong TH, Hager EA, Ying JY. New controlled delivery strategy for proteins based on
apatite-polymer composites. To be submitted to Biomaterials.
[2]
within
Fu K, Pack DW, Klibanov AM, Langer RS. Visual evidence of acidic environment
degrading
poly(lactic-co-glycolic
acid)
(PLGA)
microspheres.
Pharm
Res
2000; 17:100-106.
[3]
Yong TH, Ying JY. Apatite-polymer composite microparticles for the controlled
delivery of bone morphogenetic proteins. To be submitted to Biomaterials.
[4]
Webster TJ, Massa-Schlueter EA, Smith JL, Slamovich EB. Osteoblast response to
hydroxyapatite doped with divalent and trivalent cations. Biomaterials 2004;25:2111-2121.
[5]
Ikeuchi M, Ito A, Dohi Y, Ohgushi H, Shimaoka H, Yonemasu K, Tateishi T.
Osteogenic differentiation of cultured rat and human bone marrow cells on the surface of
zinc-releasing calcium phosphate ceramics. J Biomed Mater Res 2003;67A: 1115-1122.
[6]
Yong TH, Gao SJ, Kris A, Ying JY. Apatite-polymer
composite microparticles
for
bone tissue engineering. To be submitted to Biomaterials.
[7]
Elisseeff J, McIntosh W, Fu K, Blunk BT, Langer R. Controlled-release
TGF-31 in a photopolymerizing
hydrogel for cartilage tissue engineering.
of IGF-I and
J Orthop Res
;2001;19:1098-1 104.
[8]
Drury J.L,Mooney DJ. Hydrogels for tissue engineering: Scaffold design variables and
applications. Biomaterials 2003;24:4337-4351.
[9]
Gombotz WR, Wee SF. Protein release from alginate matrices. Adv Drug Deliv Rev
1998;31:267-285.
[ 10]
Hu YY, Zhang C, Zhang SM, Xiong Z, Xu JQ. Development of a porous poly(L-lactic
acid)/hydroxyapatite/collagen scaffold as a BMP delivery system and its use in healing canine
segmental bone defect. J Biomed Mater Res 2003;67A:591-598.
137
[11]
Itoh S, Kikuchi M, Takakuda K, Nagaoka K, Koyama Y, Tanaka J, Shinomiya K.
Implantation study of a novel hydroxyapatite/collagen (HAp/Col) composite into weightbearing sites of dogs. J Biomed Mater Res 2002;63:507-515.
[12]
Ge Z, Baguenard S, Lim LY, Wee A, Khor E. Hydroxyapatite-chitin
materials as
potential tissue engineered bone substitutes. Biomaterials 2004;25:1049-1058.
[13]
Peng H, Wright V, Usas A, Gearhart B, Shen H-C, Cummins J, Huard J. Synergistic
enhancement of bone formation and healing by stem cell-expressed VEGF and bone
morphogenetic protein-4. J Clin Invest 2002;110:751-759.
[14]
Simmons CA, Alsberg E, Hsiong S, Kim WJ, Mooney DJ. Dual growth factor
delivery and controlled scaffold degradation enhance in vivo bone formation by transplanted
bone marrow stromal cells. Bone 2004;35:562-569.
[15]
Roy I, Mitra S, Maitra A, Mozumdar S. Calcium phosphate nanoparticles as novel
non-viral vectors for targeted gene delivery. Int J Pharm 2003;250:25-33.
[16]
Misra DIN. Adsorption from solutions on synthetic hydroxyapatite: nonaqueous vs.
aqueous solvents. J Biomed Mater Res 1999;48:848-855.
138
Chapter 6 - Conclusions
Apatite-PLGA composite microparticles were developed as a new controlled delivery
platform for proteins. The release strategy was based on the hydrolysis of the biodegradable
polymer (PLGA) in an aqueous medium in vitro or in vivo to form acidic degradation
products, which would dissolve the basic inorganic phase (apatite), leading to the desorption
and release of the proteins. Apatite-PLGA composite microparticles encapsulating BSA as a
model protein or rhBMP-2 as the therapeutic protein were fabricated by a solid-in-oil-in-water
emulsion process. The protein release profiles of these particles were varied by adjusting
parameters that affected the polymer degradation and/or apatite dissolution, such as polymer
molecular weight, polymer hydrophobicity, apatite particle size, apatite loading in the
particles, and protein loading on the apatite. Increases in polymer molecular weight, apatite
particle size and apatite loading were found to reduce the rate of protein release. A decrease
in polymer hycirophobicity through the incorporation of a more hydrophilic, non-degradable
polymer (PEG) was observed to increase the magnitude, but not the rate of protein release. In
contrast, supplementation with a more hydrophobic, biodegradable polymer (PLA) led to
diminished BSA release, but enhanced rhBMP-2 release. The effect was likely a decrease in
the total amount of protein released, but an increase in the bioactive fraction preserved due to
milder pH conditions fostered by the slower degrading hydrophobic PLA. A composite
particle formulation of protein-sCAP-59 kD PLGA was identified as suitable for sustained
rhBMP-2 delivery.
In vitro testing indicated that rhBMP-2 released from these composite microparticles
was able to induce elevated expression of alkaline phosphatase and osteocalcin, markers of
osteoblastic
differentiation,
in a mesenchymal
stem cell line.
Bioactive rhBMP-2 was
detected at late stages of the release studies - upward of 70 days - and was found to be more
potent when released from the composite microparticles than from conventional collagen
sponges.
In addition, the composite carriers were determined to be non-toxic and to support
cell proliferation.
Protein-loaded composite microparticles were dispersed in secondary matrices gelatin and collagen sponge - for bone tissue engineering. These matrices could be loaded
with multiple sets of composite microparticles, each possessing a different release profile and
139
encapsulating a different growth factor. However, the crosslinking reaction used to form the
gelatin scaffolds led to denaturation of rhBMP-2. Prefabricated collagen sponges injected
with rhBMP-2-loaded and/or rhBMP-6-loaded composite particles were found to induce
elevated levels of alkaline phosphatase and osteocalcin in vitro. Preliminary in vivo studies
indicated enhanced cellularity and vascularity when composite collagen sponges containing
rhBMPs were implanted in subcutaneous sites in rats. In contrast, blank composite collagen
sponges displayed little tissue ingrowth and scaffold degradation.
140
Room 14-0551
M ITLibraries
Document Services
77 Massachusetts Avenue
Cambridge, MA 02139
Ph: 617.253.5668 Fax: 617.253.1690
Email: docs@mit.edu
http: Il//libraries.mit. edu/docs
DISCLAIMER OF QUALITY
Due to the condition of the original material, there are unavoidable
flaws in this reproduction. We have made every effort possible to
provide you with the best copy available. If you are dissatisfied with
this product and find it unusable, please contact Document Services as
soon as possible.
Thank you.
Some pages in the original document contain color
pictures or graphics that will not scan or reproduce well.