Calcif Tissue Int (1992) 50:42-48

Calcified Tissue International 9 1992 Springer-Verlag New York Inc.

Laboratory Investigations Characterization of Very Young Mineral Phases of Bone by Solid State 31phosphorus Magic Angle Sample Spinning Nuclear Magnetic Resonance and X-Ray Diffraction J. E. Roberts, l'z L. C. Bonar, 2 R. G. Griffin, ~ and M. J. Glimcher 2 ~Francis Bitter National Magnet Laboratory, Massachusetts Institute of Technology, Cambridge 02139, and 2Laboratory for the Study of Skeletal Disorders and Rehabilitation, Department of Orthopedic Surgery, Harvard Medical School and Children's Hospital, 300 Longwood Avenue, Enders 11, Boston, Massachusetts 02115, USA Received May 18, 1990, and in revised form December 18, 1990

Summary. The properties of bone mineral change with age and maturation. Several investigators have suggested the presence of an initial or "precursor" calcium phosphate phase to help explain these differences. We have used solid state 31p magic angle sample spinning (MASS) nuclear magnetic resonance (NMR) and X-ray radial distribution function (RDF) analyses to characterize 11- and 17-day-old embryonic chick bone and fractions obtained from them by density fractionation. Density fractionation provides samples of bone containing Ca-P solid-phase deposits even younger and more homogeneous with respect to the age of mineral than the calcium phosphate (Ca-P) deposits in the whole bone samples. The analytical techniques yield no evidence for any distinct phase other than the poorly crystalline hydroxyapatite phase characteristic of mature bone mineral. In particular, there is no detectable crystalline brushite [DCPD, CaHPO4 2H20 <1%] or amorphous calcium phosphate (<8-10%) in the most recently formed bone mineral. A sizeable portion of the phosphate groups exist as HPO42- in a brushite (DCPD)-like configuration. These acid phosphate moieties are apparently incorporated into the apatitic lattice. The most likely site for the brushite-like configuration is probably on the surface of the crystals.

The extracellular mineral component of bone is generally considered to be a poorly crystalline material resembling hydroxyapatite (HAp), but with very small crystalline size and extensive substitution, vacancies, and impurity ions [1-3]. The youngest embryonic bone described to date has a CaJP ratio considerably below the 1.67 value of stoichiomettic HAp or mature biological calcium-phosphate apatite, and slowly increases in average crystal size, approaching more closely the stoichiometric chemical composition of apatite with maturation. The calcium phosphate (Ca-P) in young bone is thought to be more soluble than the Ca in mature bone and synthetic HAp. To account for these properties of young bone, various authors have suggested that bone rain-

Offprint requests to: M. J. Glimcher

eral is initially deposited as some phase other than poorly crystalline HAp (PCHAp), which is converted to or replaced by PCHAp as maturation proceeds [4, 5]. Three potential precursor phases extensively discussed in the literature are amorphous calcium phosphate (ACP) [4], brushite (DCPD, CaHPO 4 2H20 ) [5], and octacalcium phosphate (OCP) [CasH2(PO4)6 5H20] [6]. We have been studying the process of bone mineral formation and development by a variety of physical and chemical techniques, with particular emphasis on the possible occurrence of a precursor calcium phosphate phase [7-9]. We report here on the extension of these studies to even younger, more recently deposited bone mineral than heretofore examined [7-9j--bone mineral from unfractionated 11and 17-day-old embryonic chick bone and density fractions obtained from them. The samples examined in this work are the youngest, most recently formed bone mineral to be studied in this way, and from a practical standpoint probably represent the youngest bone mineral that can be studied by such sample-intensive techniques. Results were similar to those reported for the slightly older bone studied previously [7-9]. Neither ACP nor crystalline DCPD was found in measurable amounts even in the newest, most recently formed bone. A significant proportion of the phosphate in the mineral of young bones is present as HPO42- , which was suggested by magic angle spinning sample (MASS) alp NMR techniques to be in a configuration similar to that in DCPD. The acid phosphate groups are not in discrete crystalline domains, and may exist within and/or on the surface of the PCHAp crystallites of bone. We also conclude that there is no evidence for the presence of OCP in the bone mineral. Materials and Methods Preparation o f Bone

The midportions of tibial diaphyses of 11- and 17-day-oldembryos were dissected immediately after sacrifice. Tissue was quickly cleaned of soft tissue and periosteum, frozen in liquid nitrogen and lyophilized. The lyophilized bone was then powdered at liquid nitrogen temperature. The bone powder was studied unfractionated and after separation into density fractions by flotation in bromoform-toluene solutions. Powdered bone, sieved to pass a screen with l0 I~mopenings,

J. E. Roberts et al.: Characterization of Youngest Bone Mineral was stirred in bromoform-toluene fractionating solutions of desired density, and the resultant suspension was centrifuged. The sedimented material, representing bone with a density equal or lower than the fractionating solution, was similarly fractionated further in progressively lower density solutions. All samples were washed with ethanol to remove residual bromoform and toluene, lyophilized, and stored in dessicators at room temperature except during measurements. More detailed experimental procedures are described in recent publications [10, I 1].

Preparation of Synthetic Calcium Phosphates ACP was prepared by addition of 0.016 M CaC12 to 0.0l M (NH4)2HPO4 adjusted to pH 10.5, rapidly washing the precipitate with dilute ammonia water, then lyophilizing the precipitate and storing it in a dessicator [16]. DCPD was prepared by rapid addition of 1 M (NH4)2HPO4 to an equal volume of 1 M Ca(NO3)z, stirring the resultant dense precipitate for 4 hours, then removing the precipitate by filtration and washing it with distilled water. The DCPD was air-dried and stored under ambient conditions. HAp was prepared by slow precipitation from CO2-t'ree solutions of CaCIz and KH2P04 at pH 9.0-9.5 at boiling temperature. The precipitate was filtered, washed with deionized water, and then refluxed for 45 days in frequent changes of deionized water. It was then dried at 105~ and stored in a dessicator. OCP was prepared by hydrolysis of DCPD in sodium acetate at pH 5.5 for several days, followed by brief washing with deionized water and air drying; the OCP was stored under ambient conditions. Synthetic carbonated apatites were prepared by rapidly adding 250 ml of 0.3 M Ca (NO3)2.4H20 to 500 ml of a solution of 0.6 M (NH4)2HPO4 plus 0.24 M NaHCO3 containing 1 ml of concentrated NH4OH (specific gravity 0.90). The resulting solution was agitated to assure mixing, then allowed to stand undisturbed for various periods of time. The precipitates were filtered and washed in demineralized water, then lyophilized. In all cases, the identity of the synthetic calcium phosphates was confirmed by X-ray powder diffraction.

Determination of Chemical Composition Calcium was determined by atomic absorbance, and phosphorus by the phosphomolybdate method [12] on ashed or HNO3/HC10 4 digested samples. HPO42- was determined by heating samples for 2 hours at 150, 200, 250, 300, and 350~ and measuring the amount of pyrophosphate formed 113]. Pyrolysis appeared complete after heating at 300~ with no further change in appearance with further heating. The PzO7-4 content after heating was lower when samples were heated above 350%The HPO42- content was taken as the equivalent of the maximum PzO74 content. Pyrophosphate was determined by stepwise elution from Dowex 1 • 10 resin with 0.5 N HC1 followed by hydrolysis at 100~ and measurement as the orthophosphate. Carbonate was determined using a micro-Conway procedure [2] consisting of liberating the CO2 in the mineralized samples by addition of dilute HzSQ, absorbing the C02 in standardized KOH, and back titrating the excess potasium hydroxide.

X-ray Measurements X-ray diffraction measurements were made with Cu K (wavelength 1.5418 A) radiation using a diffractometer in a step-scan mode, and with Debye-Scherrer and Guinier cameras. Radial distribution functions were calculated from the diffraction data as described by Wagner [14]. Further details of the X-ray diffraction procedures and calculations are presented elsewhere [15, 16].

NMR Measurements The 31p MASS NMR spectra were obtained on custom-built pulse spectrometers operated at 119.1 and 128.6 MHz for phosphorus (294.4 and 317.7 MHz for protons). The radiofrequency field strengths were 2.4--3.0 mT for phosphorus and 1.3-2.0 mT for pro-

43 tons. Powdered samples (30-150 mg) were tightly packed into delrin Andrew-Beams rotors (119.1 MHz) or ceramic double-bearing rotors (128.6 MHz). The sample spinning rates were 2.0 kHz (119.1 MHz) and 2.16 kHz (128.6 MHz) to facilitate comparisons of line intensities among samples and between spectrometers. Under these conditions, the 3~p MASS NMR spectra of DCPD obtained from the two spectrometers were identical within experimental error. The four pulse sequences used are described in detail elsewhere [7]. Briefly, they are (1) a 3~p Bloch decay with proton decoupling, with a 60 second recycle delay between acquisitions; (2) the normal cross-polarization [17] with proton decoupling and a l0 second recycle delay; (3) and (4) normal cross-polarization with one and two rotor periods (i.e., 0.5 ms and 1.0 ms) of no proton decoupling before proton decoupled data acquisition. This dipolar suppression sequence [ 18,19] quenches signals arising from phosphates with high local proton density (Le., HPO42- groups). The recycle delay for these sequences was also l0 seconds.

Computer Modeling of ~ R

Spectra

To estimate contributions from more than one calcium phosphate phase to the 31p MASS NMR spectra of bone, the spectra of various model compounds were added together in different proportions. This was accomplished by appropriate exponential multiplication of the time-domain signal from each synthetic standard, followed by Fourier transformation. To obtain enhancement factors for a particular experiment, both components of the 1:1 integrated intensity spectrum were scaled to match the experimental spectrum, with the ratio of the two scaling factors being taken as the enhancement factor [8]. Results

Density Fractionation and Composition Measurements Figure 1 shows density distribution histograms of the 1l- and 17-day embryonic chick bone. Chemical composition data on the fractions are given in Table 1. HPO4 z - determinations were not performed on 11-day, embryonic, bone density fractions, because of limitation in sample quantity, it is evident that the major fraction of the 11-day embryonic chick bone is at a very early stage of mineralization, and that considerable additional mineralization of the bone occurs between 11 and 17 days in ovo.

31p Mass NMR Figure 2 presents spectra obtained from the 1.4-1.5 g/cm 3 and 2.0--2.1 g/cm 3 fractions of 17-day in ovo embryonic chick bone, acquired under the four sets of experimental conditions discussed above. (These fractions were chosen to allow comparison of bone from a given chronological age having as widely varying stages of maturation as possible; because of the small amount of bone available from the 11-day embryos, it was not practical to prepare sufficient quantities of the tower-density fractions of this age bone for sample-intensive measurements such as NMR.) These spectra illustrate the extremes of the samples examined in the present study, and are qualitatively similar to the spectra of all bone specimens studied to date [8, 9]. Although the sets of spectra from the two different samples are qualitatively similar, several significant differences are evident. The Bloch decay spectra (Fig. 2A) are virtually superimposable, and are almost identical to Bloch decay spectra previously obtained from fully mature or high-density fractions of chick bone. This similarity indicates the presence of a predominant phosphate species of apatitic nature: in all of these samples examined in this and previous studies, a poorly crystalline HAp. This conclusion is consistent with the observed relaxation times for the synthetic phosphate. In particular, unprotonated phosphates (i.e., PO43- groups) in synthetic apatites have 31pT 1 relaxation times smaller than

J. E. Roberts et al.: Characterization of Youngest Bone Mineral

44 l 1-DAY EMBRYONIC CHICK

Table 1. Composition of embryonic chick bone and bone density fractions

BONE

Sample fraction density (g/cm3)

80

32

1l-day embryonic bone Unfractionateda <1.5 1.5-1.6 1.6-1.7 1.7-1.8 t.8-1.9 1.9-2.0 17-day embryonic bone Unfractionated~ 1.4-1.5

60

4o N.

20

0 t_ 1.3

1.4

1,5

1.6

1.7

1S

DENSITY

17-DAY

1.9

2~

2.1

2.2

g/cm 3

EMBRYONIC CHICK

BONE

80

6O *-4

40 20

~

13

14

1.5

1.6

1.7

DENSITY

1.8

1,9

20

2.1

1.6-1.7 1.7-1.8 1.8-1.9 1.9-2.0 2.0-2.1

HPO4-2 content % of total PO4-3 (chemical)

CO2 content (%)

40.8 8.5 23.6 35.5 42.7 47.0 49.7

N.D. N.D. N.D. N.D. N.D. N.D. N.D.

N.D. N.D. N.D. N.D. 1.3 N.D. N.D.

48.4 10.9 20.8 32.2 43.4 47.2 54.4 56.8

13.2 36.3 21.5 16.7 13.7 14.9 11.1 10.3

3.3 1.3 N.D. 2.1 2.9 3.3 3.2 4.0

N.D. = not determined Calculated from density-fraction data

I00

0

1.5-1.6

Mineral content, Ca + PO4 (%)

22_

g/cm 3

Fig. 1. Density distribution histograms obtained from 11- and 17day embryonic chick bone by density centrifugation. The lowest density fractions contain the newest bone and the youngest crystals deposited during the earliest stages of mineralization. approximately 30 seconds. The synthetic protonated phosphates (i.e., HPO42- groups) have 31PTl values greater than several minutes. Thus, the Bloch decay experiment with a 60 second repetition is expected to discriminate against any protonated species. The observation of similar Bloch decay spectra from all bone samples suggests that the PCHAp component does not undergo any major qualitative change with age (vide infra), although the relative amounts of this phase do vary with the age and density of the bone. The cross-polarization spectra (Fig. 2B) for the two samples are also similar, although significant differences from the Bloch decay spectra (2A) are apparent. The Bloch decay spectra comprise strong centerbands flanked by two sets of weak rotational sidebands. In contrast, the crosspolarization spectra show four sets of rotational sidebands of greater intensity with a pronounced characteristic left-right asymmetry within each set. These spectra, taken together, contain features that must arise from at least two components present in bone mineral: the PCHAp component which dominates the Bloch decay experiment, and a new species

giving rise to the stronger rotational sidebands. The latter species is less abundant than the PCHAp (vide infra). The remaining spectra in Figure 2 result from a "dipolar suppression" experiment with one and two rotational periods (i.e., -0.5 and 1 ms) with no proton decoupling before data acquisition [18, t9]. The absence of proton decoupling for these time periods suppresses lines arising from protonated phosphates and phosphate groups in regions of high proton density. The observed spectra, Figure 2C and D, are similar to the Bloch decay spectra (Fig. 2A). In particular, the strong rotational sidebands observed in normal crosspolarization spectra (Fig. 2B) are absent, suggesting that the cross-polarization spectrum contains significant contributions from protonated phosphates and/or phosphate moieties with a large local proton density. Similar experiments on a variety of synthetic calcium phosphates confirm this interpretation [7]. Three species with protonated phosphates which might be expected to occur in bone mineral are OCP, DCPA (monetite), and DCPD. In addition, because ACP has frequently been suggested as a possible component in bone mineral [2], it was also included in the list of possibilities. Synthetic preparations of ACP frequently occur in a highly hydrated state, which could conceivably provide the proton-rich environment indicated by the 3~P-NMR cross-polarization spectra. The observed differences between the Bloch decay and cross-polarization spectra from a single sample suggest that subtracting the Bloch decay spectrum from the crosspolarization spectrum might reveal the spectrum due to the protonated portion of the sample. Figure 3 presents the result of this subtraction for unfractionated 11-day in ovo embryonic chick bone. This subtraction was performed for the 11-day sample, which is typical of lightly mineralized bone, because of the better signal-to-noise characteristics of its 31P-NMR spectra compared with the less mineralized sample illustrated in Figure 2. The solid line is the difference spectrum which emphasizes contributions from the protonated species. The superimposed dot spectra represent the sideband intensities of the four synthetic candidates listed above. (The unusual appearance of the centerband in the difference spectrum results from the different isotropic

45

J. E. Roberts et a~l.: Characterization of Youngest Bone Mineral 1 7 - D a y Embryonic Chick Bone 1.4-1.5

g/cm 3

2.0-2.1

An attempt was made to estimate the amount of the DCPD-tike constituent present in the different samples studied by simulating the observed spectra with computer additions of line-broadened spectra of DCPD and PCHAp, as previously described [8]. Estimates of the amounts of the H P O 4 2 - species were made, and are listed in Table 2, which compares the amounts estimated by this procedure with amounts estimated by conventional chemical analysis. A strong correlation can be seen between estimates derived by the two techniques, but the estimates derived from 31p NMR measurements are lower than the chemically determined HPO42- content by a factor of approximately two.

g/cm 3

X-ray Diffraction and Radial Distribution Function Analysis

c

C

D

D A.__

t____ 15

1 0 kHz

] [ - 1 5 15

[. 0

1 -15

kHz

Fig. 2. Proton decoupled 31p MASS NMR spectra of the 1.4-1.5 g/cm 3 and the 2.0-2.1 g/cm3 fractions of 17-day in ovo embryonic

chick bone. Each column presents data obtained under the four sets of conditions described in the text: (A) Bloch decay, (B) crosspolarization, (C) and (D) cross-polarization with 0.5 and 1.0 ms proton dipolar coupling between 31p magnetization and data acquisition. For comparison purposes, A and B are presented on the same vertical scale, whereas C and D are displayed on an absolute intensity scale. chemical shifts and linewidths of the two principal components.) It is evident from Figure 3A that the difference spectrum obtained from bone mineral bears no resemblance to the cross-polarization spectrum obtained from ACP. This mismatch is a sufficient condition to eliminate ACP as a possible second major component giving rise to the four sets of rotational sidebands observed (i.e., the difference spectrum). Similarly, the comparison of OCP and DCPA spectra (Fig. 3B and 3C, respectively) serve to rule out these two protonated phosphates as the second phosphate component in chick bone. All three spectra obtained from these synthetic materials do not show enough intensity in the outer rotational sidebands to fit the observed difference spectrum of bone mineral. In contrast, the DCPD spectrum fits the difference spectrum quite well (Fig. 3D). In particular, each of the four sideband pairs mimics the difference spectrum within the limits of the poor signal-to-noise ratio. Thus, the observed 31p MASS NMR spectra of lightly mineralized bone indicate the presence of two principal components in the bone: a PCHAp major component, and a minor component whose spectra resembles the spectra of DCPD, and is emphasized by cross-polarization. The variation in apparent proportion of the DCPD-Iike component with bone age or maturity is illustrated in Figure 4. As bone density increases for a given chick age in ovo, the intensity of the spectral contribution from the DCPD-like constituent decreases, presumably indicating a decrease in the amount of the DCPD-like species.

X-Ray diffraction (XRD) patterns of the 11-day embryonic bone and density fractions prepared from it showed no indication of the presence of crystalline DCPD. An earlier communication from this laboratory [20] reported a similar lack of detectable crystalline DCPD in 17-day embryonic chick bone and bone density fractions, and indicated a detection threshold for crystalline DCPD of 1% or less. Graphs of radial distribution functions (RDFs) calculated from the XRD patterns of unfractionated 11-day embryonic chick bone and the 1.5-1.6 g/cm3 fraction of 17-day embryonic chick bone are shown in Figure 5. RDFs of highly crystalline HAp, ACP, and mature chick bone are also shown in Figure 5 for comparison. Because of severe preferential orientation effects with DCPD and OCP, we were unable to determine the RDFs of these substances. RDFs must be calculated from data gathered from a randomly oriented sample. The RDFs show that both 11-day embryonic chick bone and the 1.5-1.6 g/cm3 fraction of 17-day embryonic chick bone samples contain crystalline apatite, as shown by the persistence of RDF modulation out to at least 25A. It is apparent from these RDFs that the low-density fraction of the 17-day embryonic chick bone is considerably less crystalline than the unfractionated l 1-day embryo bone, as would be expected from their mineral contents of 20.8 and 40.8%, respectively. From a quantitative analysis of the decrease in RDF modulation with increasing R, as described by Grynpas et al. [15, 16] we conclude that there is no detectable amorphous calcium phosphate in the samples of very recently formed bone examined in the present study. The detection threshold for ACP in bone was estimated by Grynpas et al. [16] and in the current study to be about 8-10%. Discussion

The results obtained from the chemical and physical characterization of the very young bone samples analyzed in the present study show that the bone contains phosphate groups in two distinct environments: one, similar to that in PCHAp and the other similar to that in DCPD. The proportion of phosphate in this second environment appears to be greatest in the youngest bone studied, and to decrease with increasing age. The thermochemically determined HPO4z - content is also maximum in the youngest fractions and decreases with age. XRD studies indicate that no other crystallographically identifiable phases other than PCHAp is present, and suggest that there is no significant quantity of any solid amorphous calcium phosphate phase present. These results show that the very young bone studied here is qualitatively similar to the slightly older bone described in previous reports from these laboratories [7, 9].

46

J. E. Roberts et al.: Characterization of Youngest Bone Mineral Table 2. Seventeen-day embryonic chick bone HP042- Content of bone density fractions

t

15

Fig. 3. Comparisons of the difference between the cross-polarization and Bloch decay spectra of unfractionated 11-day in o v o embryonic chick bone (solid trace) and (A) ACP, (B) OCP, (C) DCPA, and (D) DCPD (dots represent sideband peak intensities). The close match in (D) indicates that HPO42- groups in a DCPD-Iike configuration form a significant proportion of young bone mineral. The centerband of the difference spectrum is distorted due to the chemical-shift, tensor characteristics of the DCPD-like and PCHAp components.

6 kHz

17-DAY

EMBRYONIC

CHICK

BONE

.25 _~ .20

.15 .t0 L 1.4

I 1.5

I 1.6

I 1.7

1.... 1 1.8 1.9

I 2.0

I 2.1

DENSITY g/era 3

Fig. 4. Plot of the intensity of the first cross-polarization sideband versus fraction density in spectra obtained from density fractionated 17-day-old embryonic chick bone. (A) First lower sideband; (g) first upper sideband.

The observation that the cross-polarization spectra of various bone minerals have a greater number of intense rotational sidebands than do the corresponding Bloch decay spectra is a sufficient condition to show the existence of two chemically distinct phosphate species in the mineral component. The identification of the predominant species as PCHAp is strongly suggested by the similarities in the Bloch decay spectra; the identification of the second component depends on analysis and comparison of other NMR data and X-ray diffraction data. To determine the molecular nature of the second component, dipolar suppression and difference spectroscopy were used. The dipolar suppression experiments indicate an HPO42- moiety, or a PO4 3 - with a high proton density in

Fraction density g/cm3

Mineral content Ca + PO4 (%)

HPO42- content as % of total PO2 Chemical

by 3ap-NMR

1.4-1.5 1.5-1.6 1.6-1.7 1.7-1.8 1.8-1.9 1.%2.0 2.0-2.1

10.9 20.8 32.2 43.4 47.2 54.4 56.8

36.3 21.5 t6.7 13.7 14.9 11.1 10.3

N.D. 12 N.D. 8 N.D. N.D. 5

N.D. = not determined

very close proximity (i.e., H20), Several observations based on these spectral techniques support the hypothesis that a "DCPD-like" HPO42- is the minor phosphate-containing component of bone mineral [8]: (1) the observed line shape is consistent with PCHAp and DCPD chemical shifts; (2) bone mineral difference spectra obtained by subtracting Bloch decay from cross-polarization spectra closely match the crosspolarization spectrum of DCPD, but not the spectra of any other candidate phase; (3) simulations obtained by adding together the 31p MASS NMR spectra obtained from synthetic preparations of PCHAp and DCPD adequately match those obtained from bone, particularly in sideband intensities; and (4) the dipolar suppression behavior of bone mineral is consistent with that expected from a mixture of PCHAp and DCPD. X-ray diffraction studies of unfractionated l 1-day embryo chick bone and the low density fractions of 17-day embryonic chick bone show no evidence for the presence of crystalline DCPD. Previous studies have shown that as little as 1% crystalline DCPD can be detected by XRD [20]. The absence of detectable crystalline DCPD indicates that the DCPD-like HPO4 a - groups unequivocally demonstrated by 31P-NMR are present in a noncrystalline state. They could be present as substituents in the HAp lattice (with compensating substitutions or vacancies to maintain electroneutrality), or in the form of HPO4a--rich domains too small to diffract X-rays coherently'. Alternatively, they could form an HPO4 z--enriched surface layer, too thin to diffract as a separate phase. This is supported by the fact that the bone mineral has a very large specific surface area (-200 cm 2) [9] and a sizeable fraction of bound water, relatively tightly bound and not removed by lyophilization. It is interesting to note that as bone matures, the crystallite size increases, and the surface area: volume ratio decreases; the data presented in Table 1 indicate that the HPO4 z - also decreases with maturation, as would be expected if the HPO42- were in fact present as a surface layer. It is unlikely that the HPO42- exists in the form of isolated H ions substituted in the apatitic lattice--the "defect apatite" structure originally suggested by Posner and Perloff [21], and more recently favored by Rothwell et al. [22]. Isolated H § ions would probably be too far from the phosphate groups, on the average, to yield the dipolar suppression behavior observed for the DCPD-like HPO4 ~- component in 31p MASS NMR spectra. Aue et al. [7] noted that a sample of synthetic HAp containing 12% of its PO43- in the form of HPO42 - , and prepared in such a way that the H + ions would be expected to be randomly substituted in the apatitic lattice, did not give DCPD-like 31p-NMR signal.

J. E. Roberts et al.: Characterization of Youngest Bone Mineral A

2 1 0

vVV-V.,,- . . . . . . .

-1 -2 I

2

.

I

.

.

I

I

.

......

B"

- ............ VVvvv V" "w 'v t

0

5

10

15

Angstr~,rns

20

25

Fig. 5. RDFs (G(r) versus r) for (A) unfractionated 11-day embryonic chick bone; (B) 1.5-1.6 g/cm3 fraction of 17-day embryonic chick bone; (C) unfractionated 10-week posthatch chick bone; (D) synthetic ACP; and (E) synthetic highly crystalline HAp.

Though the origin of the DCPD-Iike I-IPO42- groups can only be the subject of speculation at present, we believe the most likely explanation is that the HPO4 2- present in the milieu of the growing bone crystallites is adsorbed onto the surface of the apatitic crystals in the HPO4 2--rich domains. As the apatitic crystallites grow, the H P Q z- cannot be readily incorporated within the bulk HAp lattice, so it is constantly displaced to the surface and near-surface region of the crystallite. We believe that the HPO42- is present in the form of inorganic ions, and not organic phosphate moieties; the 31p-NMR DCPD "signature" is the result of the shielding effect of all the nuclei around the P nucleus, and would be expected to be very specific to a three-dimensional configuration closely mimicking DCPD. It is difficult to conceive how organic phosphate groups could form such a structure. The presence of the organic group would result in additional spatial hindrance to the phosphate group, and one of the four P-O- bonds would be occupied by a covalent link to an organic group; the shielding properties of the covalent organic group would be expected to differ from those of an ionized oxygen. The estimates of HPO4 2- content based on 31p-NMR measurements, shown in Table 2, are lower than chemically determined HPO4 2 - levels by a factor of approximately two. There are several possible reasons for this discrepancy. In calculating the different contributions to the bone

47 spectra, enhancement factors from crystalline synthetic DCPD and HAp were used. The bone mineral is poorly crystalline at best, and if the HPO42- moieties are not separate microcrystals but are absorbed on the surface of PCHAp or incorporated in the apatite lattice, then the synthetic DCPD is not the most appropriate model compound. Similarly, the enhancement factor depends critically on the overall proton density, and there is no a p r i o r i reason for assuming that this is the same in the bone mineral and synthetic samples. Proton NMR measurements indicate a much higher overall proton density in the synthetic samples (J. Roberts et al. unpublished data). In addition, the cross-polarization efficiency depends on the 31p-1H dipolar coupling, which is a function of third power of the phosphorus to hydrogen distance. Thus, a small change in this internuclear distance is magnified dramatically when comparing cross-polarization results. Given the uncertainties inherent in the NMR determination of HPO4 2- content, it is likely that the NMR method yields underestimates of the amount of HPO42- present. In addition to these spectral uncertainties, the chemical determination of HPO42- , which is based on determining the maximum amount of P2074- formed by heating bone to various temperatures, may overestimate the HPO42- content [13]. Also, it is possible that not all of the HPO42- present is in the DCPD-like form. Nevertheless, the consistent trend in estimated HPO4 2- content with maturation, and the close correlation between chemical and 31p-NMR estimates show that either method is useful for indicating variation or trends in bone mineral HPO4 z - content. Further research will be required to definitively determine the actual HPO4z- values. The RDF analysis of XRD data indicate that the very young bone samples studied here contain no detectable amorphous calcium phosphate. This is consistent with previous studies from this lab which showed that chick bone ranging in age from only slightly older than that studied here to fully mature contained no detectable ACP [16]. As a consequence of the rapid internal and external diametral growth of the middiaphysis of the long bones in the embryonic chick, existing bone is constantly being resorbed to accommodate the enlarging marrow cavity, while new bone is laid down on the surface of the bone cylinder by the periosteum. As a result, virtually all the bone tissue in the middiaphysis of the long bones of chick embryos 16-17 days in o v o is less than 48-72 hours old [23]. It is probable that the maximum age of the middiaphyseal bone from the embryos 11 days in o v o is even less than this. The use of the density fractionation technique provides fractions with the least mineralized, presumably most recently deposited, youngest bone tissue, even younger than the average age of the bone tissue in the bulk middiaphyseal bone sample. Thus, the density distribution histogram (Fig. 1) indicates that the 1.4-1.5 g/cm3 fraction of the 17-day embryo bone studied here constitutes the youngest 3%, and the 1.5-1.6 g/cm3 fraction the next-youngest 4% of bone tissue, all of which is less than 48-72 hours old. The 11-day chick embryo is the earliest stage at which sufficient bone occurs to allow dissection of reasonable amounts of cartilage free bone. Thus, the 11-day-old embryonic chick bone samples studied here comprise the chronologically youngest bone mineral samples obtained from ebryonic chickens that are amenable to study by sampleintensive techniques such as 3~p NMR and XRD. Because of the difficult nature of the dissection, and the small yield, however, it is not practical to prepare sufficient quantity of the lowest-density fractions of the 11-day embryo bone for the techniques applied here. Similarly, the 1.4-1.5 and 1.51.6 g/cm3 fractions of the 17-day embryo chick midshafts are

48 probably the youngest bone tissue that can practically be studied by slP-NMR and XRD. Any precursor phases formed during bone deposition would be expected to be present in highest proportion in the samples studied in this work. It is possible that transient precursor phases are formed during bone mineralization, which are converted very rapidly to the PCHAp phase identified in the present study and in past studies. The static techniques used here would not detect such transient phases which would be present in the tissue at extremely low concentrations. No stable precursor phases [5, 6] exist at significant, detectable levels, however, and in particular, ACP is never present as the major solid mineral phase in newly deposited bone tissue, as had been suggested [2, 4]. We conclude from this study that there is no stable detectable precursor calcium phosphate mineral phase present in this very newly synthesized bone studied here, which is the youngest bone amenable to study by bulk-sample methods. The mineral component of this bone is qualitatively similar to that of all other bone studied in this lab, and differs only in being less crystalline, and having a higher content of DCPD-like HPO42- groups and a lower content of CO 2 [24]. Bone mineral, at all stages of maturation, is best described as a poorly crystalline apatitic calcium phosphate containing significant amounts of carbonate ions, with extensive incorporation of noncrystalline I-IP042- ions in a DCPD-like configuration.

Acknowledgments. We would like to acknowledge the capable technical assistance of Irene Grubliauskas, Cynthia Lane, Dee Adams, and Janet Bonar during this study. This work was supported in part by grants from the National Institutes of Health (AM 34078, AM 34081), the National Science Foundation (PCM-7901181), the Peabody Foundation, Inc., and an institutional grant from the Orthopaedic Research and Education Foundation, funded by the BristolMyers/Zimmer Corporation.

References 1. Posner AS (1969) Crystal chemistry of bone mineral. Physiol Rev 49:760-791 2. Termine JD (1972) Mineral chemistry and skeletal biology. Clin Orthop 85:207-241 3. Glimcher MJ, Bonar LC, Grynpas MD, Landis WJ, Roufosse AH (1981) Recent studies of bone mineral: Is the amorphous calcium phosphate theory valid? J Cryst Growth 53:100-119 4. Termine JD, Posner AS (1967) Amorphous/crystalline interrelationships in bone mineral. Calcif Tissue Res 1:8-23 5. Neuman WF, Bareham BJ (1975) Evidence for the presence of secondary calcium phosphate in bone and its stabilization by acid production. Calcif Tissue Res 18:161-172

J. E. Roberts et al.: Characterization of Youngest Bone Mineral 6. Brown WE (1966) Crystal growth of bone mineral. Clin Orthop Rel Res 44:205-220 7. Aue WP, Roufosse AH, Roberts JE, Glimcher MJ, Griffin RG (1984) Solid state 31p nuclear magnetic resonance studies of synthetic solid phases of calcium phosphate. Biochemistry 23:6110-6114 8. Roufosse AH, Aue WP, Glimcher MJ, Griffin RG (1984) An investigation of the mineral phases of bone by solid state 31p magic angle sample spinning nuclear magnetic resonance. Biochemistry 23:6115-6120 9. Bonar LC, Grynpas MD, Roberts JE, Griffin RG, Glimcher MJ (1985) Physical and chemical characterization of the development and maturation of bone mineral. In: Butler William T (ed) The chemistry and biology of mineralized tissue. Ebsco Media, Birmingham, AL, pp 226-233 10. Roufosse AH, Landis WJ, Sabine WK, Glimcher MJ (1979) Identification of brushite in newly deposited bone mineral from embryonic chicks. J Ultrastr Res 68:235-255 11. Bonar LC, Roufosse AH, Sabine WK, Grynpas MD, Glimcher MJ (1983) X-ray diffraction studies of the crystallinity of bone mineral in newly synthesized and density fractionated bone. Calcif Tissue lnt 35:202-209 12. Dryer RL, Tammes AR, Routh JI (1957) The determination of phosphorus with N-phenyl-p-phenylenediamine. J Biol Chem 225:177-183 13. Pellegrino ED, Biltz RM (1972) Mineralization in the chick embryo. I. Monohydrogen phosphate and carbonate relationships during maturation of the bone crystal complex. Calcif Tissue Res 10:128-135 14. Wagner CNJ (1978) Direct methods for the determination of atomic-scale structure of amorphous solids. J Non-crystalline Solids 31:1-40 15. Grynpas MD, Bonar LC, Glimcher MJ (1984) X-ray diffraction radial distribution function studies of bone mineral and synthetic calcium phosphates. J Mater Sci 19:723-736 16. Grynpas MD, Bonar LC, Glimcher MJ (1984) Failure to detect amorphous calcium phosphate solid phase in bone mineral: a radial distribution function study. Calcif Tissue Int 36:291-309 17. Pines A, Gibby MJ, Waugh JS (1973) Proton-enhanced NMR of dilute spins in solids. J Chem Phys 59:569 18. Opella SJ, Frey MH (1979) Selection of nonprotonated carbon resonances in solid-state nuclear magnetic resonance. J Am Chem Soc 101:5854 19. Munowitz M, Griffin RG, Bodenhausen GJ, Huang TH (1981) Two-dimensional rotational spin-echo nuclear magnetic resonance in solids: correlation of chemical shift and dipolar interactions. J Am Chem Soc 103:2529 20. Bonar LC, Grynpas MD, Glimcher MJ (1984) Failure to detect crystalline brushite in embryonic chick and bovine bone by x-ray diffraction. J Ultrastr Res 86:93-99 21. Posner AS, Perloff A (1957) Apatites deficient in divalent cations. J Res Nat Bureau of Standards 58:279-286 22. RothweU WP, Waugh JS, Yesinowski JP (1980) High-resolution variable-temperature 31p-NMR of solid calcium phosphates. J Am Chem Soc 102:2637-2643 23. Tanzer ML, Hunt RD (1964) Experimental lathrysm, an autoradiographic study. J Cell Biol 22:623--631 24. Biltz RM, Pellegrino ED (1972) The hydroxyl content of calcified tissue mineral. Calcif Tissue Res 7:259-263