The ultrastructure of the bone-hydroxyapatite interface in vitro J. D. de Bruijn: C. P. A.T. Klein, K. de Groot, and C. A. van Blitterswijk Laboratory for Otobiology and Biocompatibility, ENT-dept., Biomaterials Research Group, University of Leiden, Building 25, Rijnsburgerweg 10, 2333 AA Leiden, The Netherlands Rat bone marrow cells were cultured on plasma-sprayed hydroxyapatite (HA).The cells formed a mineralized extracellular matrix (ECM) that exhibited several characteristics of bone tissue. The interface between this mineralized ECM and the HA was studied at the ultrastructural level w i t h scanning and transmission electron microscopy and x-ray microanalysis. Initially, the deposition of a globular, afibrillar matrix was observed on HA. This was followed by the integration of collagen fibers in this matrix and their subsequent mineralization. At the bone-HA interface two distinctly different interfacial structures were observed. An electron-dense layer with a thickness of 20-60 nm was regularly present, which
contained both organic and inorganic material and was rich in glycosaminoglycans. The interfaces differed however, in the presence or absence of an amorphous zone which was free of collagen fibers and had an average thickness of 0.7-0.8 pm. It was frequently seen interposed between the electron-dense layer and the hydroxyapatite. Similar interfacial structures have also been described in the in vivo environment, where they were referred to as lamina limitans-like or cement linelike. From the results of this study, it can be concluded that the described in vitro system is a suitable model to study bonebiomaterial interactions. 0 1992 John Wiley & Sons, Inc.
INTRODUCTION
Certain calcium phosphate (Ca-P) ceramics, such as hydroxyapatite (HA) and tricalcium phosphate, are materials that can form a bond with bone tisThis bone-bonding capacity has also been described for other biomaterials such as glass-ceramics, Bioglas~;-~and recently for a copolymer.* The bone-implant interfaces of these materials all comprise a so-called bonding zone which is composed of a calcium and phosphorous rich proteinaceous m a t r i ~ . ” ’This ~ bonding zone varies in thickness from practically zero to 1000 nm, and collagen fibers are attached to it. It is either referred to as being similar to the cementing-substance16 or the lamina limitans’O’”of bone, but with regard to HA its nature still has to be elucidated. Although little is known of the processes that take place at the interface which result in bone bonding, it is suggested that bone formation on Ca-P ceramics is enhanced by Ca2+ion release from the implant material. This *To whom correspondence should be addressed. Journal of Biomedical Materials Research, Vol. 26, 1365-1382 (1992) CCC 0021-9304/92/l01365-18$4.00 0 1992 John Wiley & Sons, Inc.
DE BRUIJN ET AL.
1366
would result in a locally increased calcium concentration followed by reprecipitation of Ca2+with physiologically present PO, ions, or secondary nucleation involving an epitactic growing p r o ~ e s s . ' ~Protein "~ adsorption onto implant materials may also be of importance for the bond formation. Implantation studies have shown the presence of an electron-dense layer at the interface between hydroxyapatite and b~ne.','",'','~-~~ This layer was continuous with the lamina limitans of the surrounding and was composed of both organic and inorganic material. It is unknown whether this layer is the result of cellular activity, physiological processes, or both. However, it is believed to play a key role in the strong bonding between bone and implant. This is confirmed by specimen examination after push-out studies, which revealed that fracture occurred either in the bone or in the implant but seldom at the interfa~e.'~ There is a need to further elucidate the mechanism underlying bone bonding, and examination of the bone-biomaterial interface is therefore a prerequisite. To study the interfacial bonding phenomena, in vitro cell culture ~ystems'~ have - ~ ~been developed that can deconvolute the complexities of the in nino environment.23The bone forming system described by Maniatopoulos et al.'* is an example of such a system with which both the early stages of bone formation24and interfacial bonding phenomena with HAz5 and titanium" have been examined. The objective of the present study is to investigate the initial bonding phenomena at the bone-hydroxyapatite interface in this in nitro bone-forming system.
MATERIALS A N D METHODS
Hydroxyapatite ceramics Using the plasma-spray a coating of hydroxyapatite (HA) was applied onto 13-mm round coverslips (Thermanox'"). Figure l(a) shows the x-ray diffraction pattern, and Figure l(b) shows a scanning electron micrograph of the HA coating. For culture experiments, HA plasma-sprayed coverslips were sterilized by 6"Cogamma-irradiation (2.5 MRad) and placed in 6- or 24-well plates (Costar). Uncoated coverslips served as a control material in order to assess the osteogenic capacity (osteogenicity) of the bone-forming system. Cell isolation and culture According to the method described by Maniatopoulos et al.", bone marrow cells were obtained from femora of 100-120-g young adult male Wistar rats. The diaphyses were flushed out and the cells were grown in a-minimal essential medium (a-MEM DNA/RNA, Gibco) supplemented with 15% fetal calf serum (FCS, Gibco), antibiotics (100 U/mL penicillin and 100 pg/mL streptomycin, Boehringer-Mannheim, FRG) and freshly added lO-*M dexamethasone (Sigma), 10 mM Na-P-glycerophosphate (Gibco) and 50 pg/mL ascorbic acid (Gibco). Cultures were incubated in a humidified atmosphere
BONE-HA INTERFACE IN VlTRO
26
1367
30
35
40
2-8 (deg)
Figure 1. X-ray diffraction pattern (A), and scanning electron micrograph (B) of the plasma-sprayed HA coating. Bar = 19 pm.
of 90% air, 10% COz at 37°C. After 5 days of primary culture, cells were subcultured using trypsin. Primary or second passage cells were plated at a density of 1 X lo4 cells/cm2 into 6- or 24-well tissue culture plates that contained either HA-coated or uncoated coverslips, and cultured for 1, 2, and 4 weeks. The medium was changed every 48 h. As a control, HA-coated coverslips were placed in cell-free, supplemented culture medium in order to examine possible medium mediated alterations of the H A substratum.
Light microscopy Cells were fixed in 1.5% glutaraldehyde in 0.14M sodium cacodylate buffer
(pH 7.4, 4°C) for 30 min, dehydrated through a graded series of ethanol and embedded in glycolmethacrylate. Semithin sections were cut and stained with toluidine blue or alizarin red.
Alkaline phosphatase cytochemistry Alkaline phosphatase activity was detected using the AZO-dye coupling method of Gomori.28The substrate solution was composed of 2 mg/mL Na
DE BRUIJN ET AL.
1368
a-naphthyl phosphate and 1 mg/mL Fast Blue RR salt dissolved in 0.1M Nabarbiturate buffer pH 9.2. Fixed cells were incubated for 10 min with the substrate solution and then thoroughly rinsed in tap water. Specificity for alkaline phosphatase activity was determined by incubating cells with a control substrate solution that lacked a-naphthyl phosphate.
Electron microscopy and x-ray microanalysis
Transmission electron microscopy ( T E M ) For routine TEM, cells were fixed according to the light microscopical procedures. After rinsing in 0.14M sodium cacodylate buffer pH 7.4, post-fixation was carried out in an aqueous solution of 1.5% potassium ferrocyanide and 1%Os04for 16 h at 4°C. The presence of glycosaminoglycans was visualized by fixing the cells in the presence of ruthenium red, according to the method described by G r ~ o t . ’Cells ~ were dehydrated through a graded series of ethanol and embedded in Epon. Ultrathin sections were prepared on a LKB Ultramicrotome, stained with uranyl acetate and lead citrate and examined at 80 kV in a Philips EM 201 or 400.
Scanning electron microscopy (SEM) Specimens were fixed and dehydrated according to the routine TEM procedure and critical-point dried from carbon dioxide in a Balzers model CPD 030 Critical Point Dryer. A layer of gold or carbon was sputter-coated with a Balzers sputter coater model MED 010 onto the specimens and they were examined in a Philips S 525 scanning electron microscope at an accelerating voltage of 15 kV. In order to examine the interface, the cell layers were removed with compressed air or adhesive tape.
X-ray microanalysis ( X R M A ) Using unstained ultrathin sections, single spot or line XRMA was performed with a Tracor Northern (TN) 2000 x-ray microanalyzer attached to a Philips EM 400 scanning transmission electron microscope. The spot diameter was 100 nm, accelerating voltage 80 kV, and measurements were performed during 100 s livetime. RESULTS
Osteogenicity Rat bone marrow cells (RBMC) that were cultured on uncoated coverslips reached confluency after 4 to 6 days and nodule formation was observed af-
BONE-HA INTERFACE IN VlTXO
1369
Figure 2. Light micrograph of a nodule (N), stained in situ for alkaline phosphatase activity. Bar = 31 pm.
ter 1 week. In the culture dishes, nodules were seen at different stages of maturation, and the central areas of individual nodules showed a later stage of development than their peripheries. Light microscopy showed that cells at the periphery of the nodules resembled osteoblasts, as they were cuboidal in shape and exhibited an intense alkaline phosphatase activity (Fig. 2). Cells in the more central part of the nodules were osteocyte-like in appearance and were surrounded by a birefringent, collagenous matrix which was observed using polarized light microscopy. Mineralization of this matrix was light microscopically observed from 2 weeks onward. Due to their osteogenic character, the nodules were further examined at the ultrastructural level. TEM revealed that the nodules were composed of cells surrounded by an extracellular matrix (ECM) which consisted of collagen fibers with a banding pattern of 64-67 nm. Cells in the unmineralized ECM showed morphological similarities to osteoblasts by having a well developed cytoplasm that was of ten rich in rough endoplasmic reticulum, Golgi complexes, microfilaments,
Figure 3. Osteocyte (OC) in a lacunae in the mineralized extracellular matrix (ECM). Note the electron-dense layer at the periphery of both the OC lacunae and the mineralized ECM. Bar = 2.6 pm.
1370
DE BRUIJN ET AL.
Figure 4. Mineralization foci in different stages of development. Transmission electron micrograph showing small electron-dense foci composed of needle-shaped crystals (arrow) and more mature foci with a granular center and an electron-dense outer layer (arrowhead), containing neealeshaped crystals (A). XRMA of the periphery of these latter foci shows the Staining for glycosaminoglycans presence of calcium and phosphorous (8). shows a positive reaction at the periphery of the foci (arrow) (C). Bar: (A) 0.35 pm, (C) 0.42 p m .
BONE-HA INTERFACE rN
vrmo
1371
Figure 5. Incorporation of collagen fibers by mineralization foci. Note the parallel arrangement of the mineralized areas and needle-shaped crystals to the collagen fiber axis (arrow). Bar = 1.4 pm.
intracellular glycogen, and lipid inclusions. Mineralization of the ECM was observed after 2 weeks, and was only associated with nodules. In contrast to cells in the unmineralized ECM, cells that were surrounded by mineralized ECM had an osteocyte-like appearance as they were spindle-shaped and often displayed cell processes. The cytoplasm was less well developed as compared to the cytoplasm in osteoblasts, and the cells often contained deposits of glycogen and fat inclusions. Osteoblast-like cells were frequently separated from the mineralized ECM by an unmineralized, osteoid-like border. At the interface between this border and the mineralized ECM an electron-dense layer was observed that contained both organic and inorganic material (Fig. 3). Mineralization of the nodules began with formation of small, electrondense, globular mineralization foci [Fig. 4(a)], the periphery of which stained positive for glycosaminoglycans [Fig. 4(c)]. They were randomly scattered in the ECM and did not appear to be related to the collagen fibers. Initially, distinct needle-shaped crystals were visible throughout the foci. These crystals were composed of calcium and phosphorous as was revealed with XRMA [Fig. 4(b)]. Growth and maturation of the foci resulted in the presence of the needle-shaped crystals only in the periphery, whereas the center became less electron-dense and more granular in appearance. As mineralization proceeded and the mineralized foci increased in size, collagen fibers were integrated (Fig. 5). Needle-shaped crystals were aligned parallel to the collagen fiber axis. Hydroxyapatite-mineralized tissue interface Cells adhered and spread out over the plasma-sprayed HA and cell multilayers were formed within 2 weeks of culture. In order to examine the boneHA interface with SEM, the cell multilayers were removed with either adhesive tape or compressed air. Using either method, cell multilayers were
DE BRUIJN ET AL.
1372
C
afibrillar
matrix
mineralized extracellular matrix centre
Figure 6. Scanning electron micrograph of a transition zone from extranodular (left) to intranodular (right) (A) and a higher magnification of the area where collagen fibers become integrated (B). Note fused mineralization foci (arrowheads) and collagen fibers (arrow). (C) is a schematic drawing of this transition zone. Bar: (A) 10 pm, (B) 2.1 pm.
more easily removed from uncoated than from HA-coated coverslips. Cells in the center of the nodules that were associated with mineralized ECM were especially difficult to remove from the HA surface. A clear difference was seen therefore between areas where mineralization had, and had not, taken place. Mineralization was only observed in the nodules and, as they graduaIIy increased in size with time, it was more developed in the center than in
BONE-HA INTERFACE IN V I T A 0
1373
Figure 7. Cross section of the mineralized ECM-HA interface. Note the deposited layer (L) on the HA surface. Bar: (A) 6.31 pm, (B) 1.0 pm.
the periphery. A gradual transition was seen between unmineralized extranodular and mineralized intranodular areas. Figure 6(c) is a schematic drawing of a cross-section of the transition zone between both areas as seen in the SEM micrograph of Figure 6(a). The transition zone was about 250 pm wide and was, from extranodular to intranodular, composed of afibrillar globules, approximately 1 pm in diameter, that became closely packed and finally formed a homogeneous afibrillar layer onto the HA in which collagen fibers were attached [Fig. 6(b)]. In time, growth and maturation of the nodule caused a movement of the transition zone in a peripheral direction. Figure 7 shows a scanning electron micrograph of a cross section of the interface. The mineralized layer that is deposited onto the H A is about 1-1.3 p m thick and collagen fibers can be seen attached to it. TEM revealed two major types of interface in the mineralized areas. Figure 8 shows a schematic representation of both types of interface. The first consisted of an amorphous zone situated between the mineralized ECM and the HA surface [Figs. 8(a), 9(a)]. This zone had an average thickness of 0.7 to 0.8 p m and stained darkly with toluidine blue. TEM examination revealed a 20- to 60nm-thick electron-dense layer, frequently interposed between the amorphous
DE BRUIJN ET AL.
1374
mineralized ECM electron-dense layer amorphous zone
A mineralized ECM electron-deme layer
B Figure 8. Schematic drawing of the two types of observed interfaces associated with mineralized areas (see text).
Figure 9. Transmission electron micrograph of an interface in which an amorphous zone (AZ) is situated between the mineralized ECM and the H A surface (A). Note also the electron-dense layer (LL) interposed between the amorphous zone and the mineralized ECM (B). Bar: (A) 0.25 pm, (B) 0.1 pm.
BONE-HA INTERFACE IN VITRO
1375
zone and the mineralized ECM [Fig. 9(b)]. In this electron-dense layer needle-shaped crystals were present and collagen fibers were attached. The absence of the amorphous zone in the second type of interface resulted in the electron-dense layer interposed between the ECM and the HA surface [Figs. 8(b), 101. A direct contact was seen with crystals in the electron-dense layer and both the HA surface and crystals in the bone tissue. Only the mineralized ECM and the electron-dense zone of the interfaces showed a positive reaction when stained for glycosaminoglycans, the amorphous zone did not display a distinct reaction. In Figure 11 it can be distinguished from the mineralized ECM and the (decalcified) HA by only a weak staining reaction. Line XRMA showed the presence of calcium and phosphorous at the interface. However, the signal was higher in the mineralized ECM than in the electron-dense layer or in the amorphous zone. Sulfur was also detected in both the calcified ECM and in the electron-dense layer (not shown).
Figure 10. Transmission electron micrograph showing an electron-dense layer (arrow) at the interface between the mineralized ECM and the HA. Bar = 1.0 pm.
Figure 11. Transmission electron micrograph of a decalcified HA-bone interface stained for glycosaminoglycans. The mineralized ECM shows distinct positive staining. Also note the less densely stained area (asterisk). Bar = 0.6 pm.
DE BRUIJN ET AL.
1376
In nonmineralized areas of the culture, collagen fibers were seen deposited on the HA surface from 1 week onward. A parallel alignment of fibers onto the HA surface was generally observed (Fig. 12), however, collagen fibers interdigitating with the HA surface have also been seen. Both TEM and SEM revealed that specimens soaked in supplemented culture medium in the absence of cells did not show an alteration of the HA surface as compared to the surface prior to this treatment.
DISCUSSION
Several in uitro systems have been described in which interactions between biomaterials and cultured bone cells were s t ~ d i e d . ~ " - ~To ~ , investigate ~",~' the physicochemical bonding phenomena of biomaterials with bonelike tissue in vitro, we chose the system described by Maniatopoulos et al.," who demonstrated the osteogenic potential of this system by the presence of alkaline phosphatase activity, bone GLA-protein (osteocalcin), osteonectin, and collagen-type I. Our observations are mainly based on morphological characteristics of bone cells and bone tissue32-35and the intense alkaline phosphatase activity which indicated the presence of active o s t e ~ b l a s t s . ~ ~ The results of the present study demonstrate that the initial bone formation observed in the nodules (Figs. 4 and 5) is similar in many ways to that found in embryonic mouse radii in vi110.~~ These observations correspond with in uitro observations reported by Davies et al.,24who studied this phenomenon on different polymeric substrata. Small afibrillar mineralization foci are formed either on the substrata or in the ECM, prior to incorporation of collagen fibers. We have observed three distinct interfacial structures on HA, two of which were associated with mineralized areas. The first comprises a collagen-free, 0.7-0.8-pm-wide amorphous zone interposed between the electron-dense layer and the HA surface. The second lacks the amorphous zone and is com-
Figure 12. Transmission electron micrograph showing collagen fibers running predominantly parallel to the HA surface. Bar = 0.33 pm.
BONE-HA INTERFACE IN VITRO
1377
posed of a glycosaminoglycan-rich, 20-60-nm-thick electron-dense layer interposed between the HA surface and the mineralized ECM. The third type of interface was observed in nonmineralized areas, in which collagen fibers were seen directly deposited onto the HA surface. Much evidence is presented that accounts for cellular activity in the formation of the interface. Mineralized interfaces have only been observed in association with nodules and not in other areas of the cultures. Therefore, mineral precipitation from the tissue culture fluid is not a likely reason for the interfacial formation. Possible formation mechanisms of the different types of interface will be discussed in the following paragraphs. The structure of the amorphous zone observed on HA resembles that found in mineralization foci produced at the beginning of bone formation in ~ i v oBoth . ~ ~comprise a granular electron lucent zone surrounded by a more electron-dense, glycosaminoglycan-rich zone. In the latter, calcium- and phosphorous-containing needle-shaped crystals are present and collagen fibers are integrated. The interfacial formation might therefore be explained as a fusion of globular, afibrillar mineralization foci with the HA surface. Thus, from a morphological point of view, the amorphous zone may be an early stage in interfacial bone formation. Another phenomenon that enhances this explanation is that the amorphous zone shows similarities to the ground substance in cement lines as has been suggested by Davies et al.23,24 There are several indications that naturally occurring cement lines are mineralized areas, several microns wide and composed of a sulfur-rich, noncollagenous matrix.37 Furthermore, by studying the bone-remodeling sequence in the rat, Tran Van et al.38initially observed the formation of a dense, granular collagen-free layer that later became calcified and formed the cement line. However, the nature of cement line composition is still controversial with regard to its sulfur, mineral, and protein ~ o n t e n t . ~ " ~ ~ The above is a possible explanation of the amorphous zone, formed as a result of the deposition of afibrillar mineralization foci. However, as the plasma-sprayed HA is not fully crystalline, as can be seen in the x-ray diffraction pattern of Figure l(a), it can be expected that it will also contain amorphous phases which will show a higher degradation rate than the crystalline areas. These amorphous phases will mainly be present at the surface of the plasma-sprayed HA, which was cooled down relatively quickly, resulting in incomplete recrystallization. Therefore, another possibility is that the amorphous zone is due to a partial degradation of HA, followed by protein adsorption.4oi41 In vivo evidence for these adsorption phenomena have been reported by van Blitterswijk et al.," who found floccular material of moderate electron density at the interface between HA and the rat middle-ear lumen. They speculated it might be adsorbed proteins or mucosubstances. Hence, both cellular and physiological processes may be involved in the formation of the amorphous zone. However, because we did not observe an alteration of the HA surface in a cell-free culture, protein adsorption alone, without the inf luence of cellular activity, cannot be the reason for the presence of the amorphous zone.
1378
DE BRUIJN ET AL.
A similar phenomenon of degradation and protein adsorption may have been responsible for the observations done by Sautier et al.42in a recent publication. They cultured enzymatically released rat calvarial cells onto synthetic H A and showed an electron-dense layer on the HA after which amorphous, granular material extended between the HA crystals to the central part of the HA mass. This amorphous, granular material may have been caused by degradation of the bulk HA material, as it was calcined at 900°C and not sintered. The specific surface would therefore have been higher than sintered resulting in a high degradation rate. Thus, our and their observations implicate that a degradation and reprecipitation process, involving protein adsorption, is an important factor for the formation of the amorphous zone. The absence of the amorphous zone in the second type of interface may be explained in two ways. First, as a follow-up of the above, the HA surface might contain both amorphous and crystalline areas. Therefore, as the crystalline areas will show a slower degradation rate than the amorphous areas, this type of interface may have been formed on crystalline parts of the ceramic. Second, with regard to the afibrillar mineralization foci, it may have been formed at the periphery of the nodules, where collagen fibers have been deposited onto the H A prior to mineralization (this premineralization stage can be seen in Figure 12). Growth of the nodule in a peripheral direction and a subsequent mineralization of these areas would result in the absence of the amorphous zone, and the presence of mineralizing collagen fibers directly onto the HA surface. The 20-60-nm-thick electron-dense layer can in both cases be explained as adsorbed proteins that are at least partially cell mediated and have an affinity for HA, as it contained glycosaminoglycans. Similar interfaces have also been reported by several authors after implantation of HA in osseous sites. There are several reports describing the presence of an amorphous zone of variable thickness on HA in bony implantation sites. Denissen et al.' described an amorphous zone between a dense HA implant and newly formed bone tissue after a 6-month implantation period, that varied in thickness from almost zero to 1000 nm. They speculated that this zone might be chemically bonded to both the HA implant and the living bone and that its existence could be due to an interaction between organic components and hydroxyapatites from biological and implant origin. Another study, in which HA was implanted into the rat middle ear,'" reported the presence of an electron-lucent amorphous layer between the HA implant surface and the lamina limitans of the surrounding bone. The electron-dense layer we observed at the interface resembles the structure that has been observed in vivo at the periphery of calcification islands and at the interface between HA and In these ilz vim studies the interfacial electron-dense layer showed a continuity with the lamina limitans of the surrounding bone. It has been described by de Lange et al." as a purely organic layer, interconnecting mineralized bone and implant crystals. However, van Blitterswijk et a1." observed that it was composed of both organic and inorganic material, and stated that its presence points to an active contribution in normal bone metabolism. The easy removal of the cell multilayer
BONE-HA INTERFACE IN VITXO
1379
from uncoated coverslips, in contrast to its more difficult removal from HAcoated coverslips, suggests that the mineralized interfaces and the collagen fibers must somehow have been bonded to the HA. This may be achieved by a chemical bond, a mechanical bond due to spacial interlocking, or both phenomena may be involved. The mainly parallel arrangement of the collagen fibers could be explained as an interaction with the phosphate groups of the calcium phosphate ceramics5 and may therefore be involved in a chemical bond. In fact this is comparable to the parallel arrangement between the needle-shaped H A crystals and the collagen fiber axis. However, due to a distinct difference in size between needle-shaped bone crystals and ceramic surfaces, and the possible sites for interaction, care has to be taken to make this c~mparison.’~ To gain more insight into the processes that take place at the interface, the inf hence of material properties on protein adsorption, interfacial mineralization, and osteoblast phenotypic expression have to be studied. The significance of our observations, however, is that the different interfacial structures may have been caused either by the time in which bone formation on the ceramic was started (afibrillar mineralization foci on HA prior to collagen fiber integration, in the center of the nodules vs. collagen fibers on HA followed by their mineralization, at the periphery of the nodules), or by the somewhat biphasic (amorphous vs. crystalline) character of the HA coating. The results of this paper show that this in vitro assay is able to mimic the biological reactions known to occur at biomaterials surfaces in vivo, and is therefore a useful tool to study bone-biomaterial interface reactions. Currently, bone-biomaterial interface reactions are being studied with Ca-P ceramics that either differ in chemical structure or in crystallinity, in order to gain more insight into the processes that are responsible for the interface formation. The authors gratefully acknowledge the help of Bart van der Lans (ENT-dept., Leiden) and Lambert Verschragen (Laboratory for Electron Microscopy, Leiden) for printing the photographic material, Joop Wolke and Jim Flach (Biomaterials Research Group, CAM-implants B.V., Leiden) for technical assistance, and they wish to express their gratitude to Yvonne Bovell (Centre for Biomaterials, University of Toronto, Canada) for making the manuscript more readable.
References 1. M. Jarcho, ”Calcium phosphate ceramics as hard tissue prosthetics,” Clin. Orthop. Rel. Res., 157, 259-278 (1981). 2. J. F. Osborn and H. Newesely, ”Bonding osteogenesis induced by calcium phosphate ceramic implants,” in Biomaterials 1980, G. D. Winter, D. F. Gibbons, and H. Plenck (eds.), John Wiley & Sons, New York, 1980, pp. 51-58. 3. T. Fujiu and M. Ogino, “Difference of bone bonding behavior among surface active glasses and sintered apatite,“ J. Biomed. Muter. Res., 18, 845-859 (1984). 4. L. L. Hench and J. Wilson, ”Surface-active biomaterials,” Science, 226, 630-636 (1984).
DE BRUIJN ET AL.
1380 5. 6.
7.
8.
9.
10. 11.
12. 13.
14.
15. 16. 17.
18. 19.
20.
21.
22. 23.
L. L. Hench, R. J. Splinter, W.C. Allen, and T. K. Greenlee, ”Bonding mechanisms at the interface of ceramic prosthetic materials,” J. Biomed. Muter. Res. Symp., 2, 117-141 (1972). L. L. Hench and H. A. Paschall, “Histochemical responses at a biomaterial’s interface,” 1. Biomed. Muter. Res. Symp., 5, 49-64 (1974). U. M. Gross, J. Brandes, V. Strunz, I. Bab, and J. Sela, ”The ultrastructure of the interface between a glass ceramic and bone,” J. Biomed. Muter. Res., 15, 291-305 (1981). D. Bakker, C. A. van Blitterswijk, S.C. Hesseling, W.Th. Daems, and J. J. Grote, ”Tissue/biomaterial interface characteristics of four elastomers. A transmission electron microscopical study,” 1. Bionzed. Muter. Res., 24, 277-293 (1990). H.W. Denissen, K. de Groot, P.Ch. Makkes, A. van den Hooff, and P. J. Klopper, “Tissue response to dense apatite implants in rats,” J. Biomed. Muter. Res., 14, 713-721 (1980). C. A. van Blitterswijk, J. J. Grote, W. Kuijpers, C. J.G. Blok-van Hoek, and W.Th. Daems, ”Bioreactions at the tissue/hydroxyapatite interface,” Biomaterials, 6, 243-251 (1985). G. L. d e Lange, C. de Putter, and F. L. J. A. de Wijs, “Histological and ultrastructural appearance of the hydroxyapatite-bone interface,” I. Biomed. Muter. Res., 24, 829-845 (1990). G. Daculsi, R. Z . LeGeros, J.C. Heughebaert, and I. Barbieux, ”Formation of carbonate-apatite crystals after implantation of calcium phosphate ceramics,” Calcif. Tiss. Int., 46, 20-27 (1990). I. Orly, M. Gregoire, J. Menanteau, M. Heughebaert, and B. Kerebel, ”Chemical changes in hydroxyapatite biomaterial under in vivo and in vitro biological conditions,” Calcif. Tiss. Int., 45, 20-26 (1989). C. A. van Blitterswijk, S.C. Hesseling, J. J. Grote, H. K. Koerten, and K. de Groot, ”The biocompatibility of hydroxyapatite ceramic: A study of retrieved human middle ear implants,” J. Biomed. Muter. Res., 24, 433543 (1990). B. M. Tracy and R. H. Doremus, “Direct electron microscopy studies of the bone-hydroxylapatite,” J. Biomed. Muter. Res., 18, 719-726 (1984). J. Ganeles, M. A. Listgarten, and C.I. Evian, “Ultrastructure of durapatite-periodontal tissue interface in human intrabony defects,” J. Periodontol., 57, 133-139 (1985). M. Jarcho, J. F. Kay, I. Kenneth, K.I. Gumaer, R. H. Doremus, and H. I? Drobeck, ”Tissue, cellular and subcellular events at a bone-ceramic hydroxylapatite interface,” 1. Bioeng., 1, 79-92 (1977). C. Maniatopoulos, J. Sodek, and A.H. Melcher, ”Bone formation in vitro by stromal cells obtained from bone marrow of young adult rats,” Cell Tiss.Res., 254, 317-330 (1988). J. E. Davies and T. Matsuda, ”Extracellular matrix production by osteoblasts on bioactive substrata in vitro,“ Scanning Microsc., 2, 14451452 (1988). M. F. Harmand, L. Bordenave, R. Duphil, R. Jeandot, and D. Ducassou, ”Human differentiated cell cultures: “in vitro” models for characterisation of cell/biomaterial interface,” in Biological and Biomechanical Performance of Biomaterials, P. Christel (ed.), Elsevier Science Publisher B.V., Amsterdam, 1987, pp. 361-366. M. Gregoire, I. Orly, and J. Menanteau, “The influence of calcium phosphate biomaterials on human bone cell activities. An in vitro approach,” J. Biomed. Muter. Res., 24, 165-177 (1990). A. Uchida, S. Nade, E. McCartney, and W. Ching, “Growth of bone marrow cells on porous ceramics in vitro,” J. Biomed. Muter. Res., 21,l-10 (1987). J. E. Davies, “The use of cell and tissue culture to investigate bone cell reactions to bioactive materials,“ in Handbook of Bioactive Ceramics,
BONE-HA INTERFACE IN VITRO
24.
25.
26.
27.
28. 29.
30.
31. 32. 33.
34. 35. 36.
37. 38.
39.
40.
41.
1381
VoZ. 1, T. Yamamuro, L.L. Hench, and J. Wilson (eds.), CRC Press, Boca Raton, 1990, pp. 1-65. J. E. Davies, R. Chernecky, B. Lowenberg, and A. Shiga, "Deposition and resorption of calcified matrix in vitro by rat bone marrow cells," Cells b Muter., 1(1), 3-15 (1991). J. D. de Bruijn, C. P.A.T. Klein, R.A. Terpstra, K. de Groot, and C. A . van Blitterswijk, "Study of the bone-biomaterial interface reactions in an in vitro bone forming system: A preliminary report," in Interfaces i n Medicine and Mechanics 2, K. R. Williams, A. Toni, J. Middleton, and G. Pallotti (eds.), Elsevier Applied Science, Amsterdam, 1991, pp. 420-429. B. Lowenberg, R. Chernecky, A. Shiga, and J. E. Davies, "Mineralized matrix production by osteoblasts on solid titanium in vitro," Cells b Muter., 1(3), 177-187 (1991). K. de Groot, R. G.T. Geesink, P. Serekian, and C. P. A.T. Klein, "Plasma sprayed coatings of hydroxylapatite," J. Biomed. Muter. Res., 21, 13751381 (1987). J.D. Bancroft and A. Stevens (eds.), Theory and Practice of Histological Techniques 2nd ed., Churchill Livingstone, London, 1982, pp. 383-384. C.G. Groot, 'Acid groups in the organic matrix of foetal bone. An electron microscopical study," PhD Thesis, University of Leiden, The Netherlands, 1982. J. E. Davies, B. Causton, Y. Bovell, K. Davy, and C. S. Sturt, "The migration of osteoblasts over substrata of discrete surface charge," Biomaferials, 7, 231-233 (1986). T. Matsuda and J. E. Davies, "The in vitro response of osteoblasts to bioactive glass," Biomaterials, 8, 275-284 (1987). B. L. Scott and M. J. Glimcher, "Distribution of glycogen in osteoblasts of the fetal rat," 1. Ultrastruct. Res., 36, 565-586 (1971). J. P. Scherft, "The ultrastructure of the organic matrix of calcified cartilage and bone in embryonic mouse radii," J. Ultrastruct. Res., 23, 333343 (1968). G. Boivin, C . Anthoine-Terrier, and K. J. Obrant, "Transmission electron microscopy of bone tissue," Acta Orthop. Scand., 61, 170-180 (1990). J. P. Scherft, "The lamina limitans of the organic bone matrix: Formation in vitro," J. Ultrastruct. Res., 64, 173-181 (1978). P. Bianco, P. Ballanti, and E. Bonucci, "Tartrate-resistant acid phosphatase activity in rat osteoblasts and osteocytes," Calcif. Tiss. Int., 43, 167-171 (1988). M. B. Schaffler, D. B. Burr, and R.G. Frederickson, "Morphology of the osteonal cement line in human bone," Anat. Rec., 217, 223-228 (1987). P. Tran Van, A. Vignery, and R. Baron, 'An electron microscopic study of the bone remodeling sequence in the rat," Cell Tiss. Xes., 225, 283292 (1982). P. Frasca, "Scanning electron microscopy studies of "ground substance" in the cement lines, resting lines, hypercalcified rings and reversal lines of human cortical bone," Acta Anat., 109, 115-121 (1981). E.C. I. Veerman, R. J. F. Suppers, C. P. A.T. Klein, K. de Groot, and A.V. Nieuw Amerongen, "Comparison of in vivo and in vitro protein adsorption to bone substitutes," in lmplant Materials in Biofunction, Vol. 8, C. de Putter, G. L. de Lange, K. de Groot, and A. J.C. Lee (eds.), Elsevier Science Publishers, Amsterdam, 1988, pp. 337-342. E.C. I. Veerman, R. J. F. Suppers, C. P. A.T. Klein, K. de Groot, and A.V. Nieuw Amerongen, "Immunochemical identification of the protein layers adsorbed onto hydroxyapatite after in vivo and in vitro incubation with serum and plasma," in Implant Materials in Biofunction, Vol. 8, C. de Putter, G.L. de Lange, K. de Groot, and A. J.C. Lee (eds.), Elsevier Science Publishers, Amsterdam, 1988, pp. 331-335.
DE BRUIJN ET AL.
1382
J. M. Sautier, J. R. Nefussi, and N. Forest, ”Ultrastructural study of the bone formation on synthetic hydroxyapatite in osteoblast cultures,” Cells b Muter., 1(3), 209-217 (1991). 43. C. P. A.T. Klein, K. de Groot, and G. van Kamp, ”Activation of complement C 3by different calcium phosphate powders,” Biomateriuls, 4, 181189 (1983). 42.
Received August 21,1991 Accepted February 21, 1992
contained both organic and inorganic material and was rich in glycosaminoglycans. The interfaces differed however, in the presence or absence of an amorphous zone which was free of collagen fibers and had an average thickness of 0.7-0.8 pm. It was frequently seen interposed between the electron-dense layer and the hydroxyapatite. Similar interfacial structures have also been described in the in vivo environment, where they were referred to as lamina limitans-like or cement linelike. From the results of this study, it can be concluded that the described in vitro system is a suitable model to study bonebiomaterial interactions. 0 1992 John Wiley & Sons, Inc.
INTRODUCTION
Certain calcium phosphate (Ca-P) ceramics, such as hydroxyapatite (HA) and tricalcium phosphate, are materials that can form a bond with bone tisThis bone-bonding capacity has also been described for other biomaterials such as glass-ceramics, Bioglas~;-~and recently for a copolymer.* The bone-implant interfaces of these materials all comprise a so-called bonding zone which is composed of a calcium and phosphorous rich proteinaceous m a t r i ~ . ” ’This ~ bonding zone varies in thickness from practically zero to 1000 nm, and collagen fibers are attached to it. It is either referred to as being similar to the cementing-substance16 or the lamina limitans’O’”of bone, but with regard to HA its nature still has to be elucidated. Although little is known of the processes that take place at the interface which result in bone bonding, it is suggested that bone formation on Ca-P ceramics is enhanced by Ca2+ion release from the implant material. This *To whom correspondence should be addressed. Journal of Biomedical Materials Research, Vol. 26, 1365-1382 (1992) CCC 0021-9304/92/l01365-18$4.00 0 1992 John Wiley & Sons, Inc.
DE BRUIJN ET AL.
1366
would result in a locally increased calcium concentration followed by reprecipitation of Ca2+with physiologically present PO, ions, or secondary nucleation involving an epitactic growing p r o ~ e s s . ' ~Protein "~ adsorption onto implant materials may also be of importance for the bond formation. Implantation studies have shown the presence of an electron-dense layer at the interface between hydroxyapatite and b~ne.','",'','~-~~ This layer was continuous with the lamina limitans of the surrounding and was composed of both organic and inorganic material. It is unknown whether this layer is the result of cellular activity, physiological processes, or both. However, it is believed to play a key role in the strong bonding between bone and implant. This is confirmed by specimen examination after push-out studies, which revealed that fracture occurred either in the bone or in the implant but seldom at the interfa~e.'~ There is a need to further elucidate the mechanism underlying bone bonding, and examination of the bone-biomaterial interface is therefore a prerequisite. To study the interfacial bonding phenomena, in vitro cell culture ~ystems'~ have - ~ ~been developed that can deconvolute the complexities of the in nino environment.23The bone forming system described by Maniatopoulos et al.'* is an example of such a system with which both the early stages of bone formation24and interfacial bonding phenomena with HAz5 and titanium" have been examined. The objective of the present study is to investigate the initial bonding phenomena at the bone-hydroxyapatite interface in this in nitro bone-forming system.
MATERIALS A N D METHODS
Hydroxyapatite ceramics Using the plasma-spray a coating of hydroxyapatite (HA) was applied onto 13-mm round coverslips (Thermanox'"). Figure l(a) shows the x-ray diffraction pattern, and Figure l(b) shows a scanning electron micrograph of the HA coating. For culture experiments, HA plasma-sprayed coverslips were sterilized by 6"Cogamma-irradiation (2.5 MRad) and placed in 6- or 24-well plates (Costar). Uncoated coverslips served as a control material in order to assess the osteogenic capacity (osteogenicity) of the bone-forming system. Cell isolation and culture According to the method described by Maniatopoulos et al.", bone marrow cells were obtained from femora of 100-120-g young adult male Wistar rats. The diaphyses were flushed out and the cells were grown in a-minimal essential medium (a-MEM DNA/RNA, Gibco) supplemented with 15% fetal calf serum (FCS, Gibco), antibiotics (100 U/mL penicillin and 100 pg/mL streptomycin, Boehringer-Mannheim, FRG) and freshly added lO-*M dexamethasone (Sigma), 10 mM Na-P-glycerophosphate (Gibco) and 50 pg/mL ascorbic acid (Gibco). Cultures were incubated in a humidified atmosphere
BONE-HA INTERFACE IN VlTRO
26
1367
30
35
40
2-8 (deg)
Figure 1. X-ray diffraction pattern (A), and scanning electron micrograph (B) of the plasma-sprayed HA coating. Bar = 19 pm.
of 90% air, 10% COz at 37°C. After 5 days of primary culture, cells were subcultured using trypsin. Primary or second passage cells were plated at a density of 1 X lo4 cells/cm2 into 6- or 24-well tissue culture plates that contained either HA-coated or uncoated coverslips, and cultured for 1, 2, and 4 weeks. The medium was changed every 48 h. As a control, HA-coated coverslips were placed in cell-free, supplemented culture medium in order to examine possible medium mediated alterations of the H A substratum.
Light microscopy Cells were fixed in 1.5% glutaraldehyde in 0.14M sodium cacodylate buffer
(pH 7.4, 4°C) for 30 min, dehydrated through a graded series of ethanol and embedded in glycolmethacrylate. Semithin sections were cut and stained with toluidine blue or alizarin red.
Alkaline phosphatase cytochemistry Alkaline phosphatase activity was detected using the AZO-dye coupling method of Gomori.28The substrate solution was composed of 2 mg/mL Na
DE BRUIJN ET AL.
1368
a-naphthyl phosphate and 1 mg/mL Fast Blue RR salt dissolved in 0.1M Nabarbiturate buffer pH 9.2. Fixed cells were incubated for 10 min with the substrate solution and then thoroughly rinsed in tap water. Specificity for alkaline phosphatase activity was determined by incubating cells with a control substrate solution that lacked a-naphthyl phosphate.
Electron microscopy and x-ray microanalysis
Transmission electron microscopy ( T E M ) For routine TEM, cells were fixed according to the light microscopical procedures. After rinsing in 0.14M sodium cacodylate buffer pH 7.4, post-fixation was carried out in an aqueous solution of 1.5% potassium ferrocyanide and 1%Os04for 16 h at 4°C. The presence of glycosaminoglycans was visualized by fixing the cells in the presence of ruthenium red, according to the method described by G r ~ o t . ’Cells ~ were dehydrated through a graded series of ethanol and embedded in Epon. Ultrathin sections were prepared on a LKB Ultramicrotome, stained with uranyl acetate and lead citrate and examined at 80 kV in a Philips EM 201 or 400.
Scanning electron microscopy (SEM) Specimens were fixed and dehydrated according to the routine TEM procedure and critical-point dried from carbon dioxide in a Balzers model CPD 030 Critical Point Dryer. A layer of gold or carbon was sputter-coated with a Balzers sputter coater model MED 010 onto the specimens and they were examined in a Philips S 525 scanning electron microscope at an accelerating voltage of 15 kV. In order to examine the interface, the cell layers were removed with compressed air or adhesive tape.
X-ray microanalysis ( X R M A ) Using unstained ultrathin sections, single spot or line XRMA was performed with a Tracor Northern (TN) 2000 x-ray microanalyzer attached to a Philips EM 400 scanning transmission electron microscope. The spot diameter was 100 nm, accelerating voltage 80 kV, and measurements were performed during 100 s livetime. RESULTS
Osteogenicity Rat bone marrow cells (RBMC) that were cultured on uncoated coverslips reached confluency after 4 to 6 days and nodule formation was observed af-
BONE-HA INTERFACE IN VlTXO
1369
Figure 2. Light micrograph of a nodule (N), stained in situ for alkaline phosphatase activity. Bar = 31 pm.
ter 1 week. In the culture dishes, nodules were seen at different stages of maturation, and the central areas of individual nodules showed a later stage of development than their peripheries. Light microscopy showed that cells at the periphery of the nodules resembled osteoblasts, as they were cuboidal in shape and exhibited an intense alkaline phosphatase activity (Fig. 2). Cells in the more central part of the nodules were osteocyte-like in appearance and were surrounded by a birefringent, collagenous matrix which was observed using polarized light microscopy. Mineralization of this matrix was light microscopically observed from 2 weeks onward. Due to their osteogenic character, the nodules were further examined at the ultrastructural level. TEM revealed that the nodules were composed of cells surrounded by an extracellular matrix (ECM) which consisted of collagen fibers with a banding pattern of 64-67 nm. Cells in the unmineralized ECM showed morphological similarities to osteoblasts by having a well developed cytoplasm that was of ten rich in rough endoplasmic reticulum, Golgi complexes, microfilaments,
Figure 3. Osteocyte (OC) in a lacunae in the mineralized extracellular matrix (ECM). Note the electron-dense layer at the periphery of both the OC lacunae and the mineralized ECM. Bar = 2.6 pm.
1370
DE BRUIJN ET AL.
Figure 4. Mineralization foci in different stages of development. Transmission electron micrograph showing small electron-dense foci composed of needle-shaped crystals (arrow) and more mature foci with a granular center and an electron-dense outer layer (arrowhead), containing neealeshaped crystals (A). XRMA of the periphery of these latter foci shows the Staining for glycosaminoglycans presence of calcium and phosphorous (8). shows a positive reaction at the periphery of the foci (arrow) (C). Bar: (A) 0.35 pm, (C) 0.42 p m .
BONE-HA INTERFACE rN
vrmo
1371
Figure 5. Incorporation of collagen fibers by mineralization foci. Note the parallel arrangement of the mineralized areas and needle-shaped crystals to the collagen fiber axis (arrow). Bar = 1.4 pm.
intracellular glycogen, and lipid inclusions. Mineralization of the ECM was observed after 2 weeks, and was only associated with nodules. In contrast to cells in the unmineralized ECM, cells that were surrounded by mineralized ECM had an osteocyte-like appearance as they were spindle-shaped and often displayed cell processes. The cytoplasm was less well developed as compared to the cytoplasm in osteoblasts, and the cells often contained deposits of glycogen and fat inclusions. Osteoblast-like cells were frequently separated from the mineralized ECM by an unmineralized, osteoid-like border. At the interface between this border and the mineralized ECM an electron-dense layer was observed that contained both organic and inorganic material (Fig. 3). Mineralization of the nodules began with formation of small, electrondense, globular mineralization foci [Fig. 4(a)], the periphery of which stained positive for glycosaminoglycans [Fig. 4(c)]. They were randomly scattered in the ECM and did not appear to be related to the collagen fibers. Initially, distinct needle-shaped crystals were visible throughout the foci. These crystals were composed of calcium and phosphorous as was revealed with XRMA [Fig. 4(b)]. Growth and maturation of the foci resulted in the presence of the needle-shaped crystals only in the periphery, whereas the center became less electron-dense and more granular in appearance. As mineralization proceeded and the mineralized foci increased in size, collagen fibers were integrated (Fig. 5). Needle-shaped crystals were aligned parallel to the collagen fiber axis. Hydroxyapatite-mineralized tissue interface Cells adhered and spread out over the plasma-sprayed HA and cell multilayers were formed within 2 weeks of culture. In order to examine the boneHA interface with SEM, the cell multilayers were removed with either adhesive tape or compressed air. Using either method, cell multilayers were
DE BRUIJN ET AL.
1372
C
afibrillar
matrix
mineralized extracellular matrix centre
Figure 6. Scanning electron micrograph of a transition zone from extranodular (left) to intranodular (right) (A) and a higher magnification of the area where collagen fibers become integrated (B). Note fused mineralization foci (arrowheads) and collagen fibers (arrow). (C) is a schematic drawing of this transition zone. Bar: (A) 10 pm, (B) 2.1 pm.
more easily removed from uncoated than from HA-coated coverslips. Cells in the center of the nodules that were associated with mineralized ECM were especially difficult to remove from the HA surface. A clear difference was seen therefore between areas where mineralization had, and had not, taken place. Mineralization was only observed in the nodules and, as they graduaIIy increased in size with time, it was more developed in the center than in
BONE-HA INTERFACE IN V I T A 0
1373
Figure 7. Cross section of the mineralized ECM-HA interface. Note the deposited layer (L) on the HA surface. Bar: (A) 6.31 pm, (B) 1.0 pm.
the periphery. A gradual transition was seen between unmineralized extranodular and mineralized intranodular areas. Figure 6(c) is a schematic drawing of a cross-section of the transition zone between both areas as seen in the SEM micrograph of Figure 6(a). The transition zone was about 250 pm wide and was, from extranodular to intranodular, composed of afibrillar globules, approximately 1 pm in diameter, that became closely packed and finally formed a homogeneous afibrillar layer onto the HA in which collagen fibers were attached [Fig. 6(b)]. In time, growth and maturation of the nodule caused a movement of the transition zone in a peripheral direction. Figure 7 shows a scanning electron micrograph of a cross section of the interface. The mineralized layer that is deposited onto the H A is about 1-1.3 p m thick and collagen fibers can be seen attached to it. TEM revealed two major types of interface in the mineralized areas. Figure 8 shows a schematic representation of both types of interface. The first consisted of an amorphous zone situated between the mineralized ECM and the HA surface [Figs. 8(a), 9(a)]. This zone had an average thickness of 0.7 to 0.8 p m and stained darkly with toluidine blue. TEM examination revealed a 20- to 60nm-thick electron-dense layer, frequently interposed between the amorphous
DE BRUIJN ET AL.
1374
mineralized ECM electron-dense layer amorphous zone
A mineralized ECM electron-deme layer
B Figure 8. Schematic drawing of the two types of observed interfaces associated with mineralized areas (see text).
Figure 9. Transmission electron micrograph of an interface in which an amorphous zone (AZ) is situated between the mineralized ECM and the H A surface (A). Note also the electron-dense layer (LL) interposed between the amorphous zone and the mineralized ECM (B). Bar: (A) 0.25 pm, (B) 0.1 pm.
BONE-HA INTERFACE IN VITRO
1375
zone and the mineralized ECM [Fig. 9(b)]. In this electron-dense layer needle-shaped crystals were present and collagen fibers were attached. The absence of the amorphous zone in the second type of interface resulted in the electron-dense layer interposed between the ECM and the HA surface [Figs. 8(b), 101. A direct contact was seen with crystals in the electron-dense layer and both the HA surface and crystals in the bone tissue. Only the mineralized ECM and the electron-dense zone of the interfaces showed a positive reaction when stained for glycosaminoglycans, the amorphous zone did not display a distinct reaction. In Figure 11 it can be distinguished from the mineralized ECM and the (decalcified) HA by only a weak staining reaction. Line XRMA showed the presence of calcium and phosphorous at the interface. However, the signal was higher in the mineralized ECM than in the electron-dense layer or in the amorphous zone. Sulfur was also detected in both the calcified ECM and in the electron-dense layer (not shown).
Figure 10. Transmission electron micrograph showing an electron-dense layer (arrow) at the interface between the mineralized ECM and the HA. Bar = 1.0 pm.
Figure 11. Transmission electron micrograph of a decalcified HA-bone interface stained for glycosaminoglycans. The mineralized ECM shows distinct positive staining. Also note the less densely stained area (asterisk). Bar = 0.6 pm.
DE BRUIJN ET AL.
1376
In nonmineralized areas of the culture, collagen fibers were seen deposited on the HA surface from 1 week onward. A parallel alignment of fibers onto the HA surface was generally observed (Fig. 12), however, collagen fibers interdigitating with the HA surface have also been seen. Both TEM and SEM revealed that specimens soaked in supplemented culture medium in the absence of cells did not show an alteration of the HA surface as compared to the surface prior to this treatment.
DISCUSSION
Several in uitro systems have been described in which interactions between biomaterials and cultured bone cells were s t ~ d i e d . ~ " - ~To ~ , investigate ~",~' the physicochemical bonding phenomena of biomaterials with bonelike tissue in vitro, we chose the system described by Maniatopoulos et al.," who demonstrated the osteogenic potential of this system by the presence of alkaline phosphatase activity, bone GLA-protein (osteocalcin), osteonectin, and collagen-type I. Our observations are mainly based on morphological characteristics of bone cells and bone tissue32-35and the intense alkaline phosphatase activity which indicated the presence of active o s t e ~ b l a s t s . ~ ~ The results of the present study demonstrate that the initial bone formation observed in the nodules (Figs. 4 and 5) is similar in many ways to that found in embryonic mouse radii in vi110.~~ These observations correspond with in uitro observations reported by Davies et al.,24who studied this phenomenon on different polymeric substrata. Small afibrillar mineralization foci are formed either on the substrata or in the ECM, prior to incorporation of collagen fibers. We have observed three distinct interfacial structures on HA, two of which were associated with mineralized areas. The first comprises a collagen-free, 0.7-0.8-pm-wide amorphous zone interposed between the electron-dense layer and the HA surface. The second lacks the amorphous zone and is com-
Figure 12. Transmission electron micrograph showing collagen fibers running predominantly parallel to the HA surface. Bar = 0.33 pm.
BONE-HA INTERFACE IN VITRO
1377
posed of a glycosaminoglycan-rich, 20-60-nm-thick electron-dense layer interposed between the HA surface and the mineralized ECM. The third type of interface was observed in nonmineralized areas, in which collagen fibers were seen directly deposited onto the HA surface. Much evidence is presented that accounts for cellular activity in the formation of the interface. Mineralized interfaces have only been observed in association with nodules and not in other areas of the cultures. Therefore, mineral precipitation from the tissue culture fluid is not a likely reason for the interfacial formation. Possible formation mechanisms of the different types of interface will be discussed in the following paragraphs. The structure of the amorphous zone observed on HA resembles that found in mineralization foci produced at the beginning of bone formation in ~ i v oBoth . ~ ~comprise a granular electron lucent zone surrounded by a more electron-dense, glycosaminoglycan-rich zone. In the latter, calcium- and phosphorous-containing needle-shaped crystals are present and collagen fibers are integrated. The interfacial formation might therefore be explained as a fusion of globular, afibrillar mineralization foci with the HA surface. Thus, from a morphological point of view, the amorphous zone may be an early stage in interfacial bone formation. Another phenomenon that enhances this explanation is that the amorphous zone shows similarities to the ground substance in cement lines as has been suggested by Davies et al.23,24 There are several indications that naturally occurring cement lines are mineralized areas, several microns wide and composed of a sulfur-rich, noncollagenous matrix.37 Furthermore, by studying the bone-remodeling sequence in the rat, Tran Van et al.38initially observed the formation of a dense, granular collagen-free layer that later became calcified and formed the cement line. However, the nature of cement line composition is still controversial with regard to its sulfur, mineral, and protein ~ o n t e n t . ~ " ~ ~ The above is a possible explanation of the amorphous zone, formed as a result of the deposition of afibrillar mineralization foci. However, as the plasma-sprayed HA is not fully crystalline, as can be seen in the x-ray diffraction pattern of Figure l(a), it can be expected that it will also contain amorphous phases which will show a higher degradation rate than the crystalline areas. These amorphous phases will mainly be present at the surface of the plasma-sprayed HA, which was cooled down relatively quickly, resulting in incomplete recrystallization. Therefore, another possibility is that the amorphous zone is due to a partial degradation of HA, followed by protein adsorption.4oi41 In vivo evidence for these adsorption phenomena have been reported by van Blitterswijk et al.," who found floccular material of moderate electron density at the interface between HA and the rat middle-ear lumen. They speculated it might be adsorbed proteins or mucosubstances. Hence, both cellular and physiological processes may be involved in the formation of the amorphous zone. However, because we did not observe an alteration of the HA surface in a cell-free culture, protein adsorption alone, without the inf luence of cellular activity, cannot be the reason for the presence of the amorphous zone.
1378
DE BRUIJN ET AL.
A similar phenomenon of degradation and protein adsorption may have been responsible for the observations done by Sautier et al.42in a recent publication. They cultured enzymatically released rat calvarial cells onto synthetic H A and showed an electron-dense layer on the HA after which amorphous, granular material extended between the HA crystals to the central part of the HA mass. This amorphous, granular material may have been caused by degradation of the bulk HA material, as it was calcined at 900°C and not sintered. The specific surface would therefore have been higher than sintered resulting in a high degradation rate. Thus, our and their observations implicate that a degradation and reprecipitation process, involving protein adsorption, is an important factor for the formation of the amorphous zone. The absence of the amorphous zone in the second type of interface may be explained in two ways. First, as a follow-up of the above, the HA surface might contain both amorphous and crystalline areas. Therefore, as the crystalline areas will show a slower degradation rate than the amorphous areas, this type of interface may have been formed on crystalline parts of the ceramic. Second, with regard to the afibrillar mineralization foci, it may have been formed at the periphery of the nodules, where collagen fibers have been deposited onto the H A prior to mineralization (this premineralization stage can be seen in Figure 12). Growth of the nodule in a peripheral direction and a subsequent mineralization of these areas would result in the absence of the amorphous zone, and the presence of mineralizing collagen fibers directly onto the HA surface. The 20-60-nm-thick electron-dense layer can in both cases be explained as adsorbed proteins that are at least partially cell mediated and have an affinity for HA, as it contained glycosaminoglycans. Similar interfaces have also been reported by several authors after implantation of HA in osseous sites. There are several reports describing the presence of an amorphous zone of variable thickness on HA in bony implantation sites. Denissen et al.' described an amorphous zone between a dense HA implant and newly formed bone tissue after a 6-month implantation period, that varied in thickness from almost zero to 1000 nm. They speculated that this zone might be chemically bonded to both the HA implant and the living bone and that its existence could be due to an interaction between organic components and hydroxyapatites from biological and implant origin. Another study, in which HA was implanted into the rat middle ear,'" reported the presence of an electron-lucent amorphous layer between the HA implant surface and the lamina limitans of the surrounding bone. The electron-dense layer we observed at the interface resembles the structure that has been observed in vivo at the periphery of calcification islands and at the interface between HA and In these ilz vim studies the interfacial electron-dense layer showed a continuity with the lamina limitans of the surrounding bone. It has been described by de Lange et al." as a purely organic layer, interconnecting mineralized bone and implant crystals. However, van Blitterswijk et a1." observed that it was composed of both organic and inorganic material, and stated that its presence points to an active contribution in normal bone metabolism. The easy removal of the cell multilayer
BONE-HA INTERFACE IN VITXO
1379
from uncoated coverslips, in contrast to its more difficult removal from HAcoated coverslips, suggests that the mineralized interfaces and the collagen fibers must somehow have been bonded to the HA. This may be achieved by a chemical bond, a mechanical bond due to spacial interlocking, or both phenomena may be involved. The mainly parallel arrangement of the collagen fibers could be explained as an interaction with the phosphate groups of the calcium phosphate ceramics5 and may therefore be involved in a chemical bond. In fact this is comparable to the parallel arrangement between the needle-shaped H A crystals and the collagen fiber axis. However, due to a distinct difference in size between needle-shaped bone crystals and ceramic surfaces, and the possible sites for interaction, care has to be taken to make this c~mparison.’~ To gain more insight into the processes that take place at the interface, the inf hence of material properties on protein adsorption, interfacial mineralization, and osteoblast phenotypic expression have to be studied. The significance of our observations, however, is that the different interfacial structures may have been caused either by the time in which bone formation on the ceramic was started (afibrillar mineralization foci on HA prior to collagen fiber integration, in the center of the nodules vs. collagen fibers on HA followed by their mineralization, at the periphery of the nodules), or by the somewhat biphasic (amorphous vs. crystalline) character of the HA coating. The results of this paper show that this in vitro assay is able to mimic the biological reactions known to occur at biomaterials surfaces in vivo, and is therefore a useful tool to study bone-biomaterial interface reactions. Currently, bone-biomaterial interface reactions are being studied with Ca-P ceramics that either differ in chemical structure or in crystallinity, in order to gain more insight into the processes that are responsible for the interface formation. The authors gratefully acknowledge the help of Bart van der Lans (ENT-dept., Leiden) and Lambert Verschragen (Laboratory for Electron Microscopy, Leiden) for printing the photographic material, Joop Wolke and Jim Flach (Biomaterials Research Group, CAM-implants B.V., Leiden) for technical assistance, and they wish to express their gratitude to Yvonne Bovell (Centre for Biomaterials, University of Toronto, Canada) for making the manuscript more readable.
References 1. M. Jarcho, ”Calcium phosphate ceramics as hard tissue prosthetics,” Clin. Orthop. Rel. Res., 157, 259-278 (1981). 2. J. F. Osborn and H. Newesely, ”Bonding osteogenesis induced by calcium phosphate ceramic implants,” in Biomaterials 1980, G. D. Winter, D. F. Gibbons, and H. Plenck (eds.), John Wiley & Sons, New York, 1980, pp. 51-58. 3. T. Fujiu and M. Ogino, “Difference of bone bonding behavior among surface active glasses and sintered apatite,“ J. Biomed. Muter. Res., 18, 845-859 (1984). 4. L. L. Hench and J. Wilson, ”Surface-active biomaterials,” Science, 226, 630-636 (1984).
DE BRUIJN ET AL.
1380 5. 6.
7.
8.
9.
10. 11.
12. 13.
14.
15. 16. 17.
18. 19.
20.
21.
22. 23.
L. L. Hench, R. J. Splinter, W.C. Allen, and T. K. Greenlee, ”Bonding mechanisms at the interface of ceramic prosthetic materials,” J. Biomed. Muter. Res. Symp., 2, 117-141 (1972). L. L. Hench and H. A. Paschall, “Histochemical responses at a biomaterial’s interface,” 1. Biomed. Muter. Res. Symp., 5, 49-64 (1974). U. M. Gross, J. Brandes, V. Strunz, I. Bab, and J. Sela, ”The ultrastructure of the interface between a glass ceramic and bone,” J. Biomed. Muter. Res., 15, 291-305 (1981). D. Bakker, C. A. van Blitterswijk, S.C. Hesseling, W.Th. Daems, and J. J. Grote, ”Tissue/biomaterial interface characteristics of four elastomers. A transmission electron microscopical study,” 1. Bionzed. Muter. Res., 24, 277-293 (1990). H.W. Denissen, K. de Groot, P.Ch. Makkes, A. van den Hooff, and P. J. Klopper, “Tissue response to dense apatite implants in rats,” J. Biomed. Muter. Res., 14, 713-721 (1980). C. A. van Blitterswijk, J. J. Grote, W. Kuijpers, C. J.G. Blok-van Hoek, and W.Th. Daems, ”Bioreactions at the tissue/hydroxyapatite interface,” Biomaterials, 6, 243-251 (1985). G. L. d e Lange, C. de Putter, and F. L. J. A. de Wijs, “Histological and ultrastructural appearance of the hydroxyapatite-bone interface,” I. Biomed. Muter. Res., 24, 829-845 (1990). G. Daculsi, R. Z . LeGeros, J.C. Heughebaert, and I. Barbieux, ”Formation of carbonate-apatite crystals after implantation of calcium phosphate ceramics,” Calcif. Tiss. Int., 46, 20-27 (1990). I. Orly, M. Gregoire, J. Menanteau, M. Heughebaert, and B. Kerebel, ”Chemical changes in hydroxyapatite biomaterial under in vivo and in vitro biological conditions,” Calcif. Tiss. Int., 45, 20-26 (1989). C. A. van Blitterswijk, S.C. Hesseling, J. J. Grote, H. K. Koerten, and K. de Groot, ”The biocompatibility of hydroxyapatite ceramic: A study of retrieved human middle ear implants,” J. Biomed. Muter. Res., 24, 433543 (1990). B. M. Tracy and R. H. Doremus, “Direct electron microscopy studies of the bone-hydroxylapatite,” J. Biomed. Muter. Res., 18, 719-726 (1984). J. Ganeles, M. A. Listgarten, and C.I. Evian, “Ultrastructure of durapatite-periodontal tissue interface in human intrabony defects,” J. Periodontol., 57, 133-139 (1985). M. Jarcho, J. F. Kay, I. Kenneth, K.I. Gumaer, R. H. Doremus, and H. I? Drobeck, ”Tissue, cellular and subcellular events at a bone-ceramic hydroxylapatite interface,” 1. Bioeng., 1, 79-92 (1977). C. Maniatopoulos, J. Sodek, and A.H. Melcher, ”Bone formation in vitro by stromal cells obtained from bone marrow of young adult rats,” Cell Tiss.Res., 254, 317-330 (1988). J. E. Davies and T. Matsuda, ”Extracellular matrix production by osteoblasts on bioactive substrata in vitro,“ Scanning Microsc., 2, 14451452 (1988). M. F. Harmand, L. Bordenave, R. Duphil, R. Jeandot, and D. Ducassou, ”Human differentiated cell cultures: “in vitro” models for characterisation of cell/biomaterial interface,” in Biological and Biomechanical Performance of Biomaterials, P. Christel (ed.), Elsevier Science Publisher B.V., Amsterdam, 1987, pp. 361-366. M. Gregoire, I. Orly, and J. Menanteau, “The influence of calcium phosphate biomaterials on human bone cell activities. An in vitro approach,” J. Biomed. Muter. Res., 24, 165-177 (1990). A. Uchida, S. Nade, E. McCartney, and W. Ching, “Growth of bone marrow cells on porous ceramics in vitro,” J. Biomed. Muter. Res., 21,l-10 (1987). J. E. Davies, “The use of cell and tissue culture to investigate bone cell reactions to bioactive materials,“ in Handbook of Bioactive Ceramics,
BONE-HA INTERFACE IN VITRO
24.
25.
26.
27.
28. 29.
30.
31. 32. 33.
34. 35. 36.
37. 38.
39.
40.
41.
1381
VoZ. 1, T. Yamamuro, L.L. Hench, and J. Wilson (eds.), CRC Press, Boca Raton, 1990, pp. 1-65. J. E. Davies, R. Chernecky, B. Lowenberg, and A. Shiga, "Deposition and resorption of calcified matrix in vitro by rat bone marrow cells," Cells b Muter., 1(1), 3-15 (1991). J. D. de Bruijn, C. P.A.T. Klein, R.A. Terpstra, K. de Groot, and C. A . van Blitterswijk, "Study of the bone-biomaterial interface reactions in an in vitro bone forming system: A preliminary report," in Interfaces i n Medicine and Mechanics 2, K. R. Williams, A. Toni, J. Middleton, and G. Pallotti (eds.), Elsevier Applied Science, Amsterdam, 1991, pp. 420-429. B. Lowenberg, R. Chernecky, A. Shiga, and J. E. Davies, "Mineralized matrix production by osteoblasts on solid titanium in vitro," Cells b Muter., 1(3), 177-187 (1991). K. de Groot, R. G.T. Geesink, P. Serekian, and C. P. A.T. Klein, "Plasma sprayed coatings of hydroxylapatite," J. Biomed. Muter. Res., 21, 13751381 (1987). J.D. Bancroft and A. Stevens (eds.), Theory and Practice of Histological Techniques 2nd ed., Churchill Livingstone, London, 1982, pp. 383-384. C.G. Groot, 'Acid groups in the organic matrix of foetal bone. An electron microscopical study," PhD Thesis, University of Leiden, The Netherlands, 1982. J. E. Davies, B. Causton, Y. Bovell, K. Davy, and C. S. Sturt, "The migration of osteoblasts over substrata of discrete surface charge," Biomaferials, 7, 231-233 (1986). T. Matsuda and J. E. Davies, "The in vitro response of osteoblasts to bioactive glass," Biomaterials, 8, 275-284 (1987). B. L. Scott and M. J. Glimcher, "Distribution of glycogen in osteoblasts of the fetal rat," 1. Ultrastruct. Res., 36, 565-586 (1971). J. P. Scherft, "The ultrastructure of the organic matrix of calcified cartilage and bone in embryonic mouse radii," J. Ultrastruct. Res., 23, 333343 (1968). G. Boivin, C . Anthoine-Terrier, and K. J. Obrant, "Transmission electron microscopy of bone tissue," Acta Orthop. Scand., 61, 170-180 (1990). J. P. Scherft, "The lamina limitans of the organic bone matrix: Formation in vitro," J. Ultrastruct. Res., 64, 173-181 (1978). P. Bianco, P. Ballanti, and E. Bonucci, "Tartrate-resistant acid phosphatase activity in rat osteoblasts and osteocytes," Calcif. Tiss. Int., 43, 167-171 (1988). M. B. Schaffler, D. B. Burr, and R.G. Frederickson, "Morphology of the osteonal cement line in human bone," Anat. Rec., 217, 223-228 (1987). P. Tran Van, A. Vignery, and R. Baron, 'An electron microscopic study of the bone remodeling sequence in the rat," Cell Tiss. Xes., 225, 283292 (1982). P. Frasca, "Scanning electron microscopy studies of "ground substance" in the cement lines, resting lines, hypercalcified rings and reversal lines of human cortical bone," Acta Anat., 109, 115-121 (1981). E.C. I. Veerman, R. J. F. Suppers, C. P. A.T. Klein, K. de Groot, and A.V. Nieuw Amerongen, "Comparison of in vivo and in vitro protein adsorption to bone substitutes," in lmplant Materials in Biofunction, Vol. 8, C. de Putter, G. L. de Lange, K. de Groot, and A. J.C. Lee (eds.), Elsevier Science Publishers, Amsterdam, 1988, pp. 337-342. E.C. I. Veerman, R. J. F. Suppers, C. P. A.T. Klein, K. de Groot, and A.V. Nieuw Amerongen, "Immunochemical identification of the protein layers adsorbed onto hydroxyapatite after in vivo and in vitro incubation with serum and plasma," in Implant Materials in Biofunction, Vol. 8, C. de Putter, G.L. de Lange, K. de Groot, and A. J.C. Lee (eds.), Elsevier Science Publishers, Amsterdam, 1988, pp. 331-335.
DE BRUIJN ET AL.
1382
J. M. Sautier, J. R. Nefussi, and N. Forest, ”Ultrastructural study of the bone formation on synthetic hydroxyapatite in osteoblast cultures,” Cells b Muter., 1(3), 209-217 (1991). 43. C. P. A.T. Klein, K. de Groot, and G. van Kamp, ”Activation of complement C 3by different calcium phosphate powders,” Biomateriuls, 4, 181189 (1983). 42.
Received August 21,1991 Accepted February 21, 1992
