SEARCH

SEARCH BY CITATION

Keywords:

  • PLGA;
  • magic-angle spinning NMR;
  • order parameters;
  • bone collagen;
  • bone apatite

Abstract

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Purpose

The influence of the pore size of biodegradable poly(lactic-co-glycolic acid) scaffolds on bone regeneration was investigated.

Methods

Cylindrical poly(lactic-co-glycolic acid) scaffolds were implanted into a defect in the tibial head of rats. Pore sizes of 100–300, 300–500, and 500–710 μm were tested and compared to untreated defects as control. Two and four weeks after implantation, the specimens were explanted and defect regeneration and de novo extracellular matrix generation were investigated by MRI, quantitative solid-state NMR, and mass spectrometry.

Results

The pore size of the scaffolds had a pronounced influence on the quantity of the extracellular matrix synthesized in the graft; most collagen was synthesized within the first 2 weeks of implantation, while the amount of hydroxyapatite increased in the second 2 weeks. After 4 weeks, the scaffolds contained large quantities of newly formed lamellar bone while the control defects were filled by inhomogenous woven bone. Best results were obtained for scaffolds of a pore size of 300–500 μm.

Conclusion

Our analysis showed that the structure and dynamics of the regenerated extracellular matrix was very similar to that of the native bone, suggesting that biomineralization was significantly enhanced by the choice of the most appropriate implant material. Magn Reson Med, 70:925–935, 2013. © 2012 Wiley Periodicals, Inc.

The repair of bone defects remains challenging in the treatment of conditions such as osteomyelitis, osteonecrosis, bone cysts, bone tumors, or implant failure caused by osteolysis or nonunion [1]. Generally, bone has a very good self-healing capacity and acute fractures typically heal without additional interventions [2]. However, for defects larger than a critical size spontaneous healing is not observed. The gold standard for the treatment of bone defects is autologous bone grafting but the amount of tissue that can be obtained from the iliac crest and other sources is naturally limited and donor site morbidity remains an important issue [3]. Therefore, research has focused on the development of bone scaffold materials that can be implanted into a defect of a given size and would stimulate the formation of new bone tissue in the defect along with the gradual resorption of the scaffold.

Among the promising scaffolding materials, partially synthetic bone substitute materials have been developed, such as hydroxyapatite [4], glass-reinforced hydroxyapatite [5], brushite [6], tricalcium phosphate [7], and mixtures of these materials (composites) [8]. Hydroxyapatite ceramics and cements represent the most important clinical scaffolds. The structure of such materials is characterized by a system of interconnecting pores that induces osseous integration [9]. In addition, bioceramics can be seeded with mesenchymal stem cells for regenerative therapies to utilize their proliferation capacities and abilities to build new tissue in a defect [10].

Biodegradable polymers represent a promising class of materials designed to match the mechanical properties of hard tissues. The advantage of polymer structures is that they can be chemically tailored to be resorbable in a time frame appropriate for defect healing and remodelling [11]. The most frequently applied biodegradable biopolymers are poly(D,L-lactide-co-glycolide) (PLGA) scaffolds. These materials are approved by the U. S. Food and Drug Administration for medical use both experimentally and clinically [12]. PLGA polymers degrade in the presence of water to lactic and glycolic acid, which both represent natural by-products in several metabolic pathways. Although these acidic degradation products may not be ideal as they can cause local inflammation, these materials have been successfully implanted in bone defects numerous times and their compatibility and biosafety have been proven [11, 13, 14]. Typical fabrication techniques for PLGA scaffolds further allows structural modifications, for instance to adjust the surface of the grafts for optimal cell-biomaterial interactions or to blend the materials with mineral particles to adjust mechanical properties and bioactivity [15, 16]. Porosity and pore size of the material can be controlled as they play a crucial role in bone formation [17]. The pore size of the materials can be adjusted to provide optimal conditions for cells to colonize the scaffolds and produce the extracellular matrix (ECM) for optimal osseous integration. Therefore, we studied the influence of the pore size of the PLGA scaffolds on healing of a tibia defect in rats.

In addition to standard histological examination, we applied a comprehensive set of MR techniques and mass spectrometry (MS) for the analysis of the de novo bone formation in the implanted scaffolds. NMR spectroscopy is capable of quantitatively monitoring the development of the major components of the organic (∼20 wt % of bone) and inorganic bone matrix (∼70 wt %), in particular collagen and hydroxyapatite [18, 19]. In addition, NMR spectroscopy allows for comparison of the molecular structure and dynamics of the de novo synthesized molecules in the implants with those in the natural tissue. Thus, a quantitative and atomistic picture of the essential molecular properties of the bone tissue in the implants can be provided [20]. In particular, the molecular dynamics of collagen is a sensitive marker for the degree of mineralization and the state of the development of the bone tissue [21]. While previous NMR studies could only be carried out in isotopically enriched tissues [21], modern techniques now allow for the detection of the collagen moiety in natural tissues already at natural abundance of useful NMR isotopes (particularly 13C) [22, 23] enabling a quality assessment of the de novo formed bone even from a biopsy without any further sample treatment. Although various MR parameters had been shown to be dependent to the cartilage glycosaminoglycan (GAG) content [24, 25], MALDI-TOF (matrix-assisted laser desorption and ionization time-of-flight) MS is particularly sensitive to the GAG of bone, which can be studied quantitatively [26, 27].

The aim of this study was to analyze the properties of the de novo formed bone and ECM in PGLA scaffolds of different pore sizes at 2 and 4 weeks after implantation into the rat tibia using histology, MRI, quantitative NMR spectroscopy, and MALDI-TOF MS.

METHODS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Preparation of PLGA Scaffolds

PLGA (Resomer® RG756S, kindly provided by Boehringer Ingelheim, Germany) at a 75:25 molar ratio of lactic and glycolic acid was processed into macroporous polymer cylinders with constant porosity by solid lipid templating [28]. Pore sizes of ∼100–300, ∼300–500, or ∼500–710 μm were generated by lipid microparticles of the corresponding size fraction. From these scaffolds, small cylinders with a diameter of 2.5 mm and 6 mm length were cut out using a surgical tool.

Animal Experiments

Female adult Wistar rats (300 g average body weight) and a standardized cylindrical defect model in the proximal tibia [29] were used for the in vivo experiments. Overall, eight different groups were investigated. Group I contained five native tibiae and Group II contained three untreated defects. Groups III to V was represented by samples at 2 weeks of implantation time with four samples at pore size 100–300 μm (Group III), four samples at pore size 300–500 μm (Group IV), and four samples at pore size 500–710 μm (Group V). Finally, Groups VI to VIII contained samples at 4 weeks of implantation time with five samples at pore size 100–300 μm (Group VI), five samples at pore size 300–500 μm (Group VII), and five samples at pore size 500–710 μm (Group VIII). In addition, two samples for each pore size and the untreated defect at 4 weeks implantation time were used for histology. The exact number of samples investigated by each method is given in the respective sections of the article. The guidelines of the Federal State of Saxony, Germany, for the care and use of laboratory animals were observed, and approval was obtained beforehand from the local veterinary ethics committee. Surgery was performed under a combination of xylazine (10 mg/kg body weight) and ketamine (100 mg/kg body weight) applied intraperitoneally. The hind limbs of the animals were shaved, disinfected, and draped free. A 5 mm longitudinal anterior skin incision was made over the tibial tuberosity. The anterior rim of the tibia was freed from adhering soft tissues. A sagittal monocortical defect was created with a 2.5 mm drill. Scaffolds were disinfected and wetted in diluted ethanol (70%) and washed three times with sterile water prior to implantation. Subsequently, scaffolds were fitted into the defect but not pressed to conserve the macroporous properties. Overlaying material was cut off flush to the bone surface prior to closure. The skin was closed with absorbable sutures and disinfected again. Postoperatively, tramadol was added to the drinking water for 2 days. The animals were allowed to move freely in their cages without restrictions. Animals were sacrificed 2 or 4 weeks postoperatively using a lethal dose of carbon dioxide gas after CO2/O2 anaesthesia. The tibiae were exarticulated at the knee joint and freed from all adhering soft tissues. The implants were removed from the surrounding bone using a surgical hole puncher that provided exactly the diameter of the implant (2.5 mm). Placing the puncher onto the implant could be carried out very precisely and was further guided by the MR images. The uncertainty in extracting surrounding bone tissue rather than the implant was therefore marginal.

Histology

The tibiae were fixed in 4% buffered formaldehyde for 24 h at room temperature. Subsequently, they were washed and decalcified for 3 weeks in ethylenediaminetetraacetic acid at pH 7.4–7.6. The specimens were dehydrated overnight (STP 420, Microm, Walldorf, Germany), embedded in Technovit 9100N (Heraeus-Kulzer, Friedrichsdorf, Germany) and dried overnight. Sections of 3 μm were cut in the frontal plane parallel to the longitudinal axis of the tibia and mounted on silane-coated slides. Hematoxylin and eosin (H&E), Goldner-Masson, and Periodic acid-Schiff reaction/Alcian blue staining was performed in all specimens. Three consecutive sections were used for morphological evaluation.

MRI

MRI measurements were performed at 288 K on a Bruker DRX 300 (7.1 T) wide bore spectrometer equipped with a microimaging unit (Bruker BioSpin GmbH, Rheinstetten, Germany) with a maximal gradient strength of 0.73 T/m (limits for phase encoding were ±71%). The tibia was placed in a 10 mm NMR tube, mechanically fixed and coated with a wet piece of paper towel to protect the bone from dehydration. No fixation treatment of the bone specimen with formaldehyde was performed. A 15-mm diameter birdcage coil was used for imaging. A effective spectral width of 50 kHz was used. The resonance line width was ∼1 kHz. Multiple slices in coronal and sagittal planes were acquired using a fast spin-echo sequence (rapid acquisition with relaxation enhancement, RARE) with a RARE factor of 4 (effective echo time TEeff = 20 ms). The slice thickness was 0.5 mm and the repetition time TR typically 1500 ms. The field of view (FOV) and, thus, the resolution was adjusted according to the size of the specimens. An in plane digital pixel size of 47 × 47 μm2 was achieved (FOV 12 × 12 mm2, matrix size 256 × 256).

Solid-State NMR Spectroscopy

For solid-state NMR, the PLGA implants were extracted from the tibia as described in the Methods section. Control samples were obtained from native tibiae, where the bone was extracted in a manner resulting in the same cavity as with implant treatment. Subsequently, each specimen was transferred into a 4 mm MAS (magic angle spinning) rotor sealed with a Teflon plug to provide a volume of ∼50 μL. After the 13C MAS NMR measurements, each specimen was split up; one part was used for MS, the other for the 31P NMR analysis. As an internal standard, samples were supplemented with aliquots of adenosine-5′-triphosphate disodium salt (ATP, Fluka) and filled into 2.5 mm MAS rotors. All NMR measurements were carried out on Bruker Avance I 750 (17.6 T) and Avance III 600 NMR (14 T) spectrometers (Bruker BioSpin GmbH, Rheinstetten, Germany), equipped with double-resonance MAS probes with 4 mm or 2.5 mm spinning modules.

The 13C MAS NMR measurements were carried out at 277 K and a MAS frequency of 7 kHz. Directly polarized 13C MAS NMR spectra were recorded using a Hahn echo pulse sequence with high power proton decoupling during acquisition, a 5 μs 90° 13C excitation pulse and a recycle delay of 5 s. For the 13C cross-polarized (CP) MAS spectra, a 1H excitation pulse length of 4 μs, a contact time of 0.7 ms, a CP spin lock field of ∼50 kHz, 65 kHz TPPM decoupling during detection, and a recycle delay of 2.5 s were used. 13C chemical shift were referenced externally relative to TMS.

The strength of the 1H-13C dipolar couplings was determined using the dipolar coupling and chemical shift (DIPSHIFT) experiment [30]. During spin evolution under 13C-1H dipolar coupling, 1H-1H homonuclear decoupling was applied using the FSLG [31] with an effective decoupling field of 80 kHz. As the dipolar-induced signal decay is periodic with the rotor period, the acquired signal over one rotor period was simulated [22] to determine the motionally averaged coupling values inline image. The molecular order parameters of the CH bonds (SCH) were calculated according to inline image. The rigid limit values were obtained from measurements of crystalline amino acids as reference for the full dipolar coupling δCH [32].

All 31P NMR spectra were recorded at 303 K and a MAS frequency of 25 kHz. Directly polarized 31P MAS NMR spectra were acquired using a Hahn echo pulse sequence. Broadband proton decoupling (∼30 kHz) during acquisition, a 4 μs 90° excitation pulse and a recycle delay of 380 s were used. For quantification, peaks were integrated and related to the integral intensity of the ATP standard (including the first order spinning sidebands).

Mass Spectrometry

The GAG content of the de novo formed bone tissue in the defect was detected by MALDI-TOF MS [27]. Samples were digested with chondroitinase ABC (AMS Biotechnology, Oxon, UK) and the concentration of the generated unsaturated chondroitin sulphate disaccharide was determined by negative ion MALDI-TOF MS on a Bruker Autoflex device (equipped with a N2 laser emitting at 337 nm) [27]. 2,5-Dihydroxy-benzoic acid in methanol served as the matrix and all measurements were performed in the reflector mode under delayed extraction conditions in order to improve the MS resolution.

Statistics

As experiments involved the use of animals, there was a clear limitation to the number of possible measurements, rendering rigorous statistical testing unfeasible. Assuming a normal distribution of observed measurements and equal variances within each measured parameter, we used simple t-tests solely to illustrate potential differences between treatments and time points. For comparison of time points within one pore size, a one-sided test was used, assuming that parameters would only increase with time. For comparison between pore sizes at the same time point, a two-sided test was used. P-values were not corrected for multiple testing.

RESULTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Histological Examination of the PLGA Bone Implants

Histological staining of the bone defects in the coronal plane provided an overview over the implantation site (Fig. 1). Two samples for the untreated defect and two for each pore size after 4 weeks of implantation were investigated. The remnants of the PLGA-implants had been dissolved during the fixation and decalcification process and appear as white areas. All visible tissue in the defects had been newly formed during the observation period. H&E staining showed numerous bone lamellae (red stain) without intervening fibrous tissue. Bone marrow cells (nuclei stained in blue) invaded the implantation site from the margins. There was no difference in cell invasion for the three different pore sizes investigated. With the Goldner-Masson stain the nuclei appeared in black, the cytoplasm in red, and the connective tissue in green. Bone invasion had well progressed within the scaffold material and appeared to be similar to the native bone. After 4 weeks, the scaffolds were invaded completely by newly formed lamellar bone that was morphologically similar to the surrounding metaphyseal host bone. In contrast, the control defects were filled only partially by inhomogeneous and less mature woven bone with intervening strands of fibrous tissue and patches of fibrocartilage (Fig. 1d). No visible negative effects like persisting invasion of granulocytes or foreign body giant cells of the scaffold treatment could be detected by histological methods.

image

Figure 1. Representative images of histological specimens of rat tibiae 4 weeks after defect placement and treatment with and without PLGA implants. H&E (a), Periodic acid-Schiff reaction/Alcian blue (b), Goldner-Masson (c) stained images of a tibia treated with a PLGA scaffold (pore size 300–500 μm) after 4 weeks of implantation is shown. The scaffold (S) which has been dissolved during the decalcification and fixation process is completely invaded with and partially replaced by more mature lamellar bone (L). d: The Goldner-Masson stained image reveal that the corresponding control defect (D) is partially filled with woven bone (W) that displays an inhomogeneous structure. Original magnification ×25. Scale bar represents 0.5 mm.

Download figure to PowerPoint

MRI of the PLGA Implants

In contrast to the histological stains, the PLGA implants could still be identified in the MRI scans. Figure 2a,b show characteristic slices of the sagittal and coronal planes of the tibia head with the PLGA scaffolds after an implantation time of 4 weeks. The implant could clearly be delineated from the surrounding native bone of the tibia. Additionally, the de novo generated ECM in the implant showed no oriented trabecular structure as in the case of the native bone. When the defect was left empty, large areas in the images appeared black, indicating that no ECM had been deposited in the defect (Fig. 2c,d). Only a poorly organized macroscopic ECM structure was observed. The boundary of the control defect appearing in black could clearly be delineated.

image

Figure 2. Representative MR images of rat tibiae 4 weeks after defect placement and treatment with and without PLGA implants. Panels ad display MRI scans performed in sagittal (a,c) and coronal (b,d) orientation. Slices with a thickness of 0.5 mm were acquired using a RARE sequence with an in plane digital pixel size between 43 × 66 μm2 and 39 × 39 μm2. The RARE factor was 4 and the effective echo time TEeff = 20 ms. Panels (a) and (b) correspond to the MR images of a tibia treated with a PLGA implant (pore size 300–500 μm) and (c) and (d) show corresponding images without implant treatment.

Download figure to PowerPoint

Solid-State NMR of the Organic Bone Matrix

To reveal the structure and dynamics of the de novo generated ECM within the PLGA implants on a molecular level, solid-state NMR investigations were performed. The organic moiety of the bone ECM can be investigated in detail by 13C solid-state MAS NMR [22, 33, 34]. Figure 3 shows 13C CP MAS spectra of native rat tibial bone (a), where the typical spectral signature of collagen was observed. Signal assignments agreed with literature [22, 35, 36].

image

Figure 3. 1H decoupled 188.6 MHz 13C cross-polarized MAS NMR spectra of native bone (a), a pure PLGA scaffold (b), a PLGA scaffold after an implantation time of 4 weeks (c), and the difference spectrum of the spectra c and b (d). Trace (e) shows a directly excited 13C MAS NMR spectrum of the sample. All spectra were scaled to fit the same vertical space. *spinning sidebands.

Download figure to PowerPoint

Further, the corresponding NMR spectra of the PLGA scaffold before (b) and after implantation (c) are shown. The PLGA spectrum was dominated by the signals of lactic and glycolic acid: CO (169.6 ppm), CH (69.0 ppm), CH2 (60.7 ppm), and CH3 (16.4 ppm). The NMR spectra obtained from the implant (c) could be described as a superposition of the NMR spectra of the de novo generated collagen [37], which represents the most abundant organic component of the ECM, and the characteristic resonances of the PLGA implant. The characteristic collagen resonances are unequivocally detectable, but partially superimposed by the PLGA signals. This was a clear indication that scaffold desorption had not taken place during an implantation time of either 2 or 4 weeks. This agreed with previous studies, where the onset of PLGA degradation in vitro was determined to be about 6 to 8 weeks [38]. This was also confirmed by the absence of any sharp peaks indicative of degradation products of the PLGA matrix in directly polarized 13C NMR spectra of the implants (Fig. 3e).

For a better comparison, the NMR spectrum of the PLGA implant (b) was subtracted from the NMR spectrum of the bone implant (c) to yield a difference spectrum shown in trace (d), which was very similar to the NMR spectrum of native bone (a). Unfortunately, the resonance of Hyp Cγ at 71.1 ppm, representing the most reliable fingerprint of collagen, was superimposed with a PLGA matrix signal and therefore strongly influenced by the subtraction procedure. Although peak intensities differed between the individual samples, characteristic collagen resonances [37] could be easily identified and no major changes of their chemical shifts were observed. This indicated that the de novo formed collagen already had its native structure.

Next, we quantified the 13C CPMAS NMR spectra obtained for the different PLGA pore sizes. Because there was no indication for the degradation of the scaffold during 4 weeks of implantation, the relative amount of collagen synthesized in the implants could be monitored. To this end, the CH signal of lactic acid from the scaffold (69.0 ppm) was used as an internal standard to compare the intensities of the Cα signals of Gly (42.4 ppm) between the samples.

The results obtained from the NMR spectra quantification are summarized in Figure 4. Because there was no degradation of the scaffold, the relative amount of collagen synthesized in the implants could be monitored. The plot consistently showed that the largest amount of collagen was synthesized in the implants of the middle pore size (300–500 μm) after 4 weeks. Four samples were investigated for each pore size at an implantation time of 2 weeks and five samples for each pore size at an implantation time of 4 weeks. Statistical analysis by simple t-test revealed significant differences between the middle and the largest and between the smallest and the largest pore size (P < 0.05). Further, it is interesting to note that most of the collagen synthesis (∼75–95%) was completed already within the first 2 weeks after implantation.

image

Figure 4. Comparison of the relative Glycin Cα peak integrals as a measure of collagen generated in the PLGA implants after 2 weeks (gray) and 4 weeks (black). The collagen amount was determined by using the ratio of the Gly Cα peak of collagen and CH peak of lactic acid of the PLGA scaffold. Error bars represent the standard deviation (n = 4 for 2 weeks of implantation time and n = 5 for 4 weeks of implantation time). Statistical significance according to the standard Student's t-test is indicated (*P < 0.05).

Download figure to PowerPoint

In addition to the molecular structure of collagen, the material properties of the de novo formed bone also depend on the dynamics of the macromolecules in the tissue. Here, we studied the amplitudes of the fast motions (correlation times in the submicrosecond window) of the CH bond vectors in the most abundant amino acids of collagen. As the 13C-1H dipolar coupling is partially averaged by these motions, an order parameter (S) can be determined. A fully rigid C-H-bond would lead to an order parameter of S = 1, while a value of S = 0 corresponds to fully isotropic motion expressed by a vanishing dipolar coupling.

The order parameters are shown in Figure 5 for all resolved sites of the collagen. In agreement with literature [22], relatively high order parameters of S = 0.58–0.87 were determined for native bone, indicating that the mineralized collagen is essentially rigid with only small amplitude fluctuations of the C-H bond vectors. There was a general trend for the order parameters of the de novo generated collagen in the bone implants to be lower than that of native bone 2 weeks after implantation, while the order parameter typically reached the value of native bone 4 weeks after implantation. We investigated three samples for each group. Some statistically significant differences were seen for these comparisons by a simple t-test. However, there was no reliable trend as to what implant pore size provided the best results with regard to collagen dynamics. Overall, from the dynamics perspective, the de novo synthesized collagen also closely resembles that of the native bone.

image

Figure 5. Comparison of the measured C-H order parameters (S) for various sites of native bone collagen (white bars) and from the de novo formed collagen in PLGA implants analyzed after either 2 weeks (gray bars) or 4 weeks (black bars) of implantation. Error bars represent the standard deviation of three different samples investigated. Statistical significance according to the standard Student's t-test is indicated (*P < 0.05).

Download figure to PowerPoint

Solid-State NMR of the Inorganic Bone Matrix

To quantify the formation of inorganic matrix in the bone implants, 31P MAS NMR was performed. For higher sensitivity and to avoid the superposition of resonances, high MAS frequencies were used. 31P NMR spectra of bone and the bone implants could be excited both by direct polarization as well as cross-polarization, indicating the close proximity of hydroxyl ions in the bone mineral. For the quantification of the bone apatite, we added ATP as internal standard. From these spectra (Fig. 6, upper panel), the hydroxyapatite content of the implants was calculated and referenced to native bone shown in Figure 6 (lower panel). Four samples were investigated for each pore size at an implantation time of 2 weeks and five samples for each pore size at an implantation time of 4 weeks along with five native samples. Best results were once again obtained for the middle pore size (300–500 μm pores), however, statistical analysis (simple t-test) did not show significance for this observation. Only the differences between 2 and 4 weeks of implantation time were significant for the middle pore size (P < 0.05). In contrast to the synthesis of collagen, a significant portion (33–56%) of the apatite was synthesized during the second half of the animal experiment.

image

Figure 6. In the upper panel, a typical 31P solid-state NMR spectrum of an extracted PLGA implant and the added ATP standard is shown. From this, the total fraction of hydroxyapatite in the implants after either 2 weeks (gray bars) or 4 weeks (black bars) of implantation was calculated with respect to native bone (white bar), and is shown in the lower panel. Error bars represent the standard deviations (n = 4 for 2 weeks of implantation time, n = 5 for 4 weeks of implantation time, and n = 5 for the native sample). Statistical significance using the standard Student's t test is indicated (*P < 0.05).

Download figure to PowerPoint

Glycosaminoglycan Content Within the Bone Implants

Bone development progresses from a cartilaginous state which is increasingly mineralized through the replacement of the hydration water by apatite crystallites [21]. MS was used to determine the concentration of GAGs [39]. These molecules are ubiquitously present in cartilage and their concentration is gradually decreased as biomineralization progresses.

As shown in Figure 7, the de novo formed bone contained high amounts of chondroitin sulphate that typically exceeded that of native bone; only for the middle pore size, a chondroitin sulphate concentration similar to that of native bone was found. Two specimens were studied for each pore size at an implantation time of 2 weeks, three samples at an implantation time of 4 weeks, and five native samples. In spite of the large error bars that reflect the difficulties to measure such low GAG concentrations by MS, some interesting trends can be seen. Except for the largest pore size, the GAG content of cartilaginous bone decreases over time in agreement with progressing biomineralization. The middle pore size always showed the lowest GAG content in agreement with the most advanced development of the bone ECM with regard to collagen and bone apatite as determined by NMR. However, a t-test did not reveal statistically significant differences.

image

Figure 7. Comparison of the chondroitin sulphate (CS) concentration in native bone and in the PLGA implants after a regeneration time of either 2 (gray) or 4 weeks (black). Errors represent the obtained standard deviations (n = 2 for 2 weeks of implantation time and n = 3 for 4 weeks of implantation time). The absolute CS amount was determined by means of MALDI-TOF MS subsequent to a digestion of the tissue CS by the enzyme chondroitinase ABC that results in the generation of one defined disaccharide.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

PLGA scaffolds are widely applied in regenerative medicine [11, 40-42]. However, although the pore size of the material is of crucial importance for cell migration [43], studies that systematically investigate the pore size-dependence of polymer-based implants for bone regeneration are scarce [44, 45]. To the best of our knowledge, only a single study exists using similar macroporous PLGA scaffolds [46], but a completely different model (mesenterial tissue) has been used and the middle pore size has not been studied at all.

We applied a comprehensive analytics including histology, MRI, MALDI-TOF MS as well as 13C and 31P MAS NMR spectroscopy. Although NMR is a versatile and quantitative tool, it does not represent a standard technique for the application of the regenerated ECM in bone in spite of the obvious advantages of the method [22, 34-36, 47, 48]. Apart from the analytical and structural power of NMR spectroscopy with regard to the qualitative and quantitative assessment of de novo generated tissue, the NMR techniques also allow the study of the dynamics of the biopolymers. NMR studies have highlighted the impact of the molecular mobility on the macroscopic elastic properties for polymer-based materials [49]. In agreement with this aspect, the molecular dynamics of the ECM are also related to the elastic properties of bone. Thus, the application of NMR techniques for the characterization of de novo generated tissues in implants as well as the function of the related tissues is promising [39, 50].

This study investigated the de novo bone formation in PLGA implants of varying pore size using an established rat tibia model. Taken together, the results obtained indicated that the PLGA scaffolds are osteoconductive and induce bone formation and biomineralization in these untreated scaffolds. Histological evaluations of the samples 4 weeks after implantation revealed that the implanted PGLA scaffolds were homogeneously and completely invaded by newly formed lamellar bone, while the control defects were filled only partially with less mature woven bone with strands of fibrous tissue and patches of fibrocartilage characteristic for the earlier stages of enchondral ossification (Fig. 1). The good integration of the implants was confirmed by MRI (Fig. 2), although the microscopic bone structure in the implants was isotropic and similar to the cancellous bone in the host metaphyseal bone. Some MR images, particularly in the untreated defect, revealed a morphological structure at the perimeter of the defect that is reminiscent to the epiphyseal growth plate in the host bone. The resemblance of the de novo formed bone to cartilaginous tissue could also be confirmed by quantitative measurement of GAGs in the implants by MS, since bone formation can be described by a time-dependent dispersion of the different tissues [2]. Fibrous tissue is the most abundant initial component but its contribution decreases during the healing process. Furthermore, in the intermediate phase, a high amount of cartilage is present [51] providing an order of magnitude higher GAG content in comparison to bone [52]. Consequently, in the initial state, higher GAG content as in native bone is observed, while the amount of GAG decreases during a normal healing process as confirmed by MS (Fig. 7).

The formation of organic and inorganic ECM increased with time. On an atomistic level, NMR could confirm that both collagen and hydroxyapatite were generated in the scaffolds. The structure, molecular concentration, and dynamics of these de novo synthesized molecules were found to be similar to that of the native bone. Collagen type I represents the largest fraction of organic molecules in bone and its ECM [53]. Using 13C MAS NMR, collagen could be unequivocally detected in the implants. As the isotropic chemical shift of a carbon peak is indicative of its secondary structure [54] and the chemical shift of the de novo formed collagen in the implants agree with those of the bone collagen, we conclude that the characteristic triple helix structure of collagen had been formed in the PLGA implants. Interestingly, the majority of 75–95% of the collagen has already been synthesized within the first 2 weeks, while collagen synthesis was apparently downregulated between the 2nd and 4th week. Unfortunately, our quantification procedure did not allow an absolute measure of the collagen content. However, the reduced synthesis in the second half of the experiment suggests that the collagen content already reached its maximum after 2 weeks.

In contrast to collagen, the majority of the bioapatite was synthesized after 2 weeks. Biomineralization takes place after collagen is synthesized and the hydration water of collagen is then replaced by the bioapatite crystals [55]. It has been reported that the interaction of collagen with the mineral may restrict the motions in collagen [21]. This effect has also been observed here as the motional amplitudes of the collagen motions are typically higher after 2 weeks of implantation, which indicates that the collagen segments can undergo larger amplitude motions. Higher order parameters (i.e., smaller amplitude motions) were observed in the de novo formed collagen after an implantation time of 4 weeks, which are comparable to those of native bone. However, it should be noted that the collagen motions in the implant are much more restricted than in fully hydrated cartilage [56] suggesting that the newly synthesized and less mineralized collagen also experiences a significant motional restriction.

From the analytical results of our study, we conclude that the de novo formation of bone in the tibia model is already nearly complete after 4 weeks. Bone formation was most pronounced for the middle pore size of 300–500 μm. This does not appear to be a trivial effect of the geometrical parameters of the scaffolds as densities and hence porosities of the scaffolds were rather homogeneous [85.7%, 87.3%, and 87.3 % porosity for the pore sizes of 100–300, 300–500, and 500–710 μm, respectively [28]].

It is easily comprehensible why the largest pore size did not provide an optimal performance. Here, the pores are too large to provide optimal adhesion properties for the cells to migrate into the defect and fill up the empty space. The situation is then almost comparable to a free 2D cell culture that is known not to grow into the third dimension. In addition, such scaffolds would be most compromised with regard to stability. The smallest pore size also showed inferior performance. Although these pores exceeded the smallest recommended pore size for ceramic materials of 100 μm [57], literature data have also shown better osteogenesis for hydroxyapatite and titanium alloy implants with pores > 300 μm [58, 59]. While pores of 100–300 μm in PLGA materials are still much larger than the cell dimension, it is conceivable that the wetting properties of this material are compromised. Although we wetted the scaffolds via an ethanol phase to overcome the hindered entry of water into the air-filled pores, the procedure might not yield perfect results.

In conclusion, our study demonstrated the use of MR techniques and MS for the assessment of de novo bone formation in polymer-based scaffolds in an animal model. It is quite remarkable that the de novo formed bone ECM in PLGA implants quantitatively and qualitatively closely resembles that of healthy native bone according to the parameters provided by these methods. A pore size of 300 to 500 μm provides the best results regarding the formation of bone and ECM and, therefore, represents the most effective implant. Furthermore, the PLGA implants proved to be osteoconductive as shown by the generation of bone-like ECM within the PLGA treated defect as oppose to the untreated defect. It is well known that high porosity and large pores facilitate bone ingrowth and osseointegration into a scaffold after implantation [17]. However, if the pore size exceeds a critical value, mechanical stability of the scaffold is diminished. Furthermore, cell migration in scaffolds with large pores may also be limited. Consequently, PLGA implants with 300 to 500 μm pore size appear to be attractive matrices for further studies that will focus on biofunctionalization of those scaffolds, e.g. with ECM components and growth factors.

Acknowledgments

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

The authors thank the team of Drs. Jung and Speckl at the Animal Care Unit of the University of Technology, Dresden for assistance with the animal experiments. The scaffold fabrication by Mrs. Ambrosch and Feher, Institute of Pharmacy, University of Leipzig, is gratefully acknowledged. Histological workup of the specimens was done by Suzanne Manthey at the Department of Trauma and Reconstructive Surgery, University Hospital Dresden. The authors would like to thank Dr. David P. Nannemann for a critical review of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES