- Email: [email protected]
Imaging collagen packing dynamics during mineralization of engineered bone tissue
Acta Biomaterialia 23 (2015) 309–316
Contents lists available at ScienceDirect
Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat
Imaging collagen packing dynamics during mineralization of engineered bone tissue G. Campi a,⇑, M. Fratini b,c, I. Bukreeva d, G. Ciasca e, M. Burghammer f, F. Brun g, G. Tromba h, M. Mastrogiacomo i, A. Cedola d a
Institute of Crystallography, CNR, Via Salaria km 29.300, I-00015 Monterotondo, Roma, Italy Centro Fermi – Museo Storico della Fisica e Centro Studi e Ricerche ‘‘Enrico Fermi’’, Roma, Italy Dipartimento di Scienze, Università di Roma Tre, Roma, Italy d Institute for Chemical and Physical Process, CNR, c/o Physics Dep. at Sapienza University, P-le A. Moro 5, 00185 Roma, Italy e Istituto di Fisica, Universitá Cattolica SC, L.go Francesco Vito 1, I-00168 Roma, Italy f European Synchrotron Radiation Facility, B. P. 220, F-38043 Grenoble Cedex, France g Department of Engineering and Architecture, University of Trieste, Via A. Valerio 10, 34127 Trieste Italy h Sincrotrone Trieste SCpA, 34149 Basovizza, Trieste, Italy i Istituto Nazionale per la Ricerca sul Cancro, and Dipartimento di Medicina Sperimentale dell’Università di Genova & AUO San Martino Istituto Nazionale per la Ricerca sul Cancro, Largo R. Benzi 10, 16132 Genova, Italy b c
a r t i c l e
i n f o
Article history: Received 25 February 2015 Received in revised form 8 May 2015 Accepted 28 May 2015 Available online 3 June 2015 Keywords: Collagen Biomineralization Fluctuations Spatial statistics Scanning X-ray micro-diffraction
a b s t r a c t The structure and organization of the Type I collagen microfibrils during mineral nanoparticle formation appear as the key factor for a deeper understanding of the biomineralization mechanism and for governing the bone tissue physical properties. In this work we investigated the dynamics of collagen packing during ex-vivo mineralization of ceramic porous hydroxyapatite implant scaffolds using synchrotron high resolution X-ray phase contrast micro-tomography (XPClT) and synchrotron scanning micro X-ray diffraction (SlXRD). While XPClT provides the direct 3D image of the collagen fibers network organization with micrometer spatial resolution, SlXRD allows to probe the structural statistical fluctuations of the collagen fibrils at nanoscale. In particular we imaged the lateral spacing and orientation of collagen fibrils during the anisotropic growth of mineral nanocrystals. Beyond throwing light on the bone regeneration multiscale process, this approach can provide important information in the characterization of tissue in health, aging and degeneration conditions. Statement of Significance: BONE grafts are the most common transplants after the blood transfusions. This makes the bone-tissue regeneration research of pressing scientific and social impact. Statement of Significance: Bone is a complex hierarchical structure, where the interplay of organic and inorganic mineral phases at different length scale (from micron to atomic scale) affect its functionality and health. Thus, the understanding of bone tissue regeneration requires to image its spatial-temporal evolution (i) with high spatial resolution and (ii) at different length scale. Statement of Significance: We exploited high spatial resolution X-ray Phase Contrast micro Tomography and Scanning micro X-ray Diffraction in order to get new insight on the engineered tissue formation mechanisms. This approach could open novel routes for the early detection of different degenerative conditions of tissue. Ó 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
1. Introduction Bone is a complex composite and nanostructured material formed via dynamic interactions between an organic matrix of biomacromolecules and an inorganic mineral phase [1,2]. The
⇑ Corresponding author. Tel.: +39 06 90672624; fax: +39 06 90672630. E-mail address: [email protected] (G. Campi). http://dx.doi.org/10.1016/j.actbio.2015.05.033 1742-7061/Ó 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
inorganic phase consists of carbonate apatite nanocrystals growing in the organic matrix composed by collagen (type I) proteins, with some minor noncollagenous proteins and minor amounts of lipids and osteogenic factors [3]. Collagen 3D arrangement constitutes a basic issue to be addressed since it plays a fundamental role in the bone tissue biomechanical properties. Collagen structure and organization has been investigated and visualized by several techniques such as magnetic resonance imaging (MRI) [4], ultrasound [5], electron [6] and optical imaging [7,8]. The obtained results
310
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
indicate that the collagen molecule is a right-handed triple-helix, called fibril, about 300 nm long and 1.0–1.5 nm diameter. Collagen fibers are composed from about 100 up to more than 1000 tightly packed fibrils. X-ray diffraction (XRD) is an effective tool for studying the fibrillar periodic assembly of collagen. Typically, X-ray scattering patterns from collagen present Bragg spots along the axial direction, indicative of a long-range ordered phase made of staggered parallel fibrils [9,10]. This results in a repeating axial D-spacing providing open sites for cross-links and mineral nucleation. Alongside the axial Bragg peaks, collagen XRD measurements show a diffuse equatorial scattering characteristic of a short-range order typical of fluid-like arrangement of the molecules [11–13]. Hulmes et al. interpreted XRD data as a molecular crystal with quasi-hexagonal molecular packing [14] while an arrangement in concentric layers was assumed by Parry et al. [15]. Doucet et al. [16] showed that the equatorial X-ray scattering patterns of collagen can easily be modeled using the paracrystal organization of fibers into parallel rods packed in a hexagonal or pseudo-hexagonal lateral network. Unfortunately, conventional XRD as well as neutron diffraction [17] has a strong limitation since it provides insight only into the average concentration and structure of the collagen [12,13,18]. However collagen is, from a morphological point of view, a fluctuating and inhomogeneous tissue [19,20]: this makes quite difficult to model its structural features with the conventional experimental approaches, requiring high spatial resolution probes to map spatial changes in structural organization. In this framework the importance of the collagen axial D-spacing fluctuations has been recently recognized; indeed, significant alterations in bone collagen axial D-spacing distribution have been found in Osteogenesis Imperfecta [21] and long term estrogen depletion [22]. A quantitative method for space resolved axial D-spacing analysis at sub-micrometer scale, used Atomic Force Microscopy (AFM) and two-dimensional Fast Fourier Transform (2D FFT) analysis [23]. Anyway, being the AFM a surface approach, it does not allow to investigate the interplay of collagen spacing and minerals arrangement in the bulk composite material. Indeed, it is little known about the spatial distribution of both collagen and mineral crystals packing in the radial direction of the collagen fibrils. This is also due to the fact that conventional imaging techniques, such as electron microscopic tomography, can reach resolution of 4–6 nm that cannot resolve either the mineral nanocrystal size during the early stage of mineralization [24] either the collagen molecular lateral spacing (<2 nm) [9,25]. In this paper we overcome both the limitation of the 2D techniques and the limited resolution of the conventional tomography, by combining the 3D visualization of the collagen bundles provided by high resolution XPClT with the high spatial resolution structural information obtained by SlXRD. This multiscale approach is needed due to the fact that collagen fibrils organize into bundles at micrometer scale via interfibrillar cross-links. At this scale, high resolution XPClT is an appropriate tool to visualize collagen bundles since unlike conventional radiography and tomography it is able to image small density variations in weakly absorbing materials such as biological samples [26,27]. On the other hand, SlXRD technique allows to probe both the k-space and real space, visualizing structural features at atomic scale and nanoscale such as the lateral packing of collagen molecules and the size of mineral particles. The spatial mapping of the simultaneously collected Small Angle X-ray Scattering (SAXS) and Wide Angle X-ray Scattering (WAXS) signals, can readily monitor the mineralization process according to its temporal evolution. Although SlXRD has been already used for investigating ultrastructure in biological tissue, [28–30] in this work it is used to study the dynamics of the collagen fibrils during the HA nanocrystal nucleation and growth with a spatial statistical approach. More specifically, we measured the fluctuations of (i) the
lateral spacing and preferred orientation of collagen fibrils (ii) the longitudinal and transverse size of mineral nanocrystals. Thus, the order degree of collagen arrangement and mineral growth during the bone tissue formation has been quantified by applying basics spatial statistics tools to the measured quantities on the large amount of collected data from different samples. This approach has been recently used to get key information on systems presenting structural fluctuations and heterogeneity on (sub)micrometric scales in different research fields ranging from material science to biomedicine [31–34]. 2. Materials and methods All experimental animal procedures were carried out in the IRCCS AOU San Martino –IST Animal Facility (Genoa, Italy), in the respect of the national current regulations regarding the protection of animals used for scientific purpose (D.lgsvo 27/01/1992, n. 116). Research protocols have been evaluated and approved by the IRCCS AOU San Martino –IST Ethical Committee for animal experimentation (CSEA) as Animal use project n. 336 communicated to The Italian Ministry of Health, having regard to the article 7 of the D.lgs 116/92. Here we investigated collagen packing dynamics during nucleation and growth of bone mineral nanocrystals in ex vivo conditions: expanded bone marrow mesenchymal stem cells (BMSC) seeded onto porous ceramic scaffolds and subcutaneously implanted in the mouse [35]. After four weeks the scaffolds are removed from the host animals and the newly formed bone analyzed. To improve the statistical significance 5 samples extracted from different animals were studied. 2.1. Sample preparation Marrow aspirates were obtained from the iliac crest of the experimental sheep as part of a protocol approved by the competent ethical authority. Bone marrow stromal cell (BMSC) cultures were counted with a nuclear stain, suspended in Coon’s modified Ham’s F12 medium supplemented with 10% FCS, 1 ng/ml of human recombinant FGF-2 and subsequently plated in 100-mm dishes at 0.5–1.0 107 cells per dish. Cultures were incubated at 37 °C in a humidified atmosphere containing 95% air and 5% CO2. Before reaching confluence, cells were enzymatically detached and passaged. Experiments were performed using a pool of first passage BMSC derived from marrows of four animals. Osteogenic properties of the BMSC were evaluated by an ‘‘in vivo’’ assay in an immuno-deficient mice model. First passage cells expanded in standard medium were tripsynized, resuspended in a fibrinogen solution (Tissuecol; Baxter, Italia) to a final concentration of 62.5 106 cells/ml and loaded onto highly porous ceramic scaffolds (100% hydroxyapatite cubes, 3 3 3 mm3; FinCeramica, Faenza, Italy); an appropriate volume of 20 ll of Thrombin was added to ignite the enzymatic cleavage that originates a fibrin clot around and within the ceramic, entrapping the cells. Two samples were implanted subcutaneously in each mouse. After four weeks the samples dedicated to XlPCT were harvested, washed in Phosphate Buffered Saline (PBS) three times and fixed in paraformahaldehyde (4% in PBS) for 2–3 h at 4 °C. Additional washes in PBS removed the residual fixative. The samples for SlXRD were additionally dehydrated in ethanol at increasing concentration, embedded in methylmetacrilate and transversally cut using a diamond saw (Gillings- Hamco, Hamco Machines, Inc., Rochester, N.Y., U.S.A) in serial sections of about 100 lm thick. We choose 100 microns as a well suited thickness value, quite less than the general mineral bone attenuation length (about 350 lm at X-ray incident energy of 12.7 keV).
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
311
2.2. Synchrotron X-ray phase contrast microscopy measurements and data analysis The X-ray phase contrast imaging measurements were carried out on the TOMCAT beamline at the Swiss Light Source, SLS, Switzerland. The incident beam energy was set at 0.729 Å (17 keV) by using a multilayer based monochromator. The beam size was 2 2 mm2 at the sample placed about 30 m downstream the monochromator. A (2048 2048) pixels CCD camera optically coupled with a microscope system mounted with 1.25x objective providing a voxel size of 0.64 lm3 was set at a distance of 5 cm from the sample. The real resolution, i.e. the minimal detectable detail, was about 3 voxels which means about 2 2 2 lm3. A layout of the experimental setup is illustrated in Fig. 1. We collected 1601 projections covering a total angle range of 360° for the 3D tomographic reconstruction. The sample was protected by a mylar foil and measured without any further preparation. A single distance phase retrieval algorithm [36] has been applied to all projections of the tomographic measurements, using the ANKAphase code [37]. When applied to all tomographic projections, the retrieved phase maps can be fed to a standard filtered back-projection algorithm to obtain phase tomograms. A typical image obtained from the tomographic reconstructed volume of a sample ROI is reported in Fig. 2. 2.3. Synchrotron X-ray micro-diffraction measurements and data analysis Bone sections with thickness of about 100 lm were measured on the ID13 beamline of the European Synchrotron Radiation Facility, ESRF, France. The scanning micro-diffraction setup (Fig. 1), equipped with a double-crystal monochromator and a Kirkpatrick-Baez mirror as focusing system, supplied a beam size of 1 1 lm2 with a wavelength of 0.976 Å (12.703 keV). The sample was mounted at a distance from the detector of 110 cm. The beam center, detector tilt and sample-to-detector distance were calibrated using a silver behenate standard. The samples were scanned by piezo-scanning stage with 0.1 micron repeatability using a step size of 5 lm in both vertical, z, and horizontal, y, direction. Diffraction patterns were recorded in transmission by a FreLon CCD detector (2048 2048 pixels of 50 lm2) with a typical acquisition time of 5 s and corrected for parasitic scattering by taking into account the transmission, Tr, measured by a photodiode using the same scan parameters. The 2D diffraction patterns have been radially and azimuthally integrated to provide 1D profiles
Fig. 2. (A) Tomographic section from a reconstructed volume of a sample implanted for 4 weeks. The different gray-levels represent different densities in the sample. The thickness is of 512 lm. We can distinguish the three different types of tissues: bone, B, the collagenous soft tissue, ST and the scaffold, SC. The collagen fiber compaction (e.g. indicated by asterisk) can be seen between the formed B tissue and the ST tissue. (B) Collagen fibrils intensity in a typical ROI. (C) Collagen fibrils Intensity profile trough the red line in (B); the thickness of the B/ST interface in found to be about 30 lm.
of intensity, I(q), vs. transfer moment, q = 4psin(h)/k, and angular intensity distribution I(U). The achieved q range of (0.5–30) nm1 allowed us to measure simultaneously both the SAXS and the WAXS signal for investigating structure from nano to atomic scale. In each sample we performed X-ray microdiffraction measurements on two regions of interest (ROI) including ceramic scaffold, SC, newly formed bone, B, and organic collagenous soft tissue, ST.
Fig. 1. Layout of experimental setup adopted for high resolution synchrotron X-ray phase contrast micro-tomography and synchrotron scanning X-ray micro-diffraction.
312
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
Typical 2D diffraction patterns and integrated profiles, recorded at different positions and characterizing different tissues SC, B and ST, are shown in Figure S1. The SC tissue shows XRD patterns with sharp peaks, due to micrometric size of HA crystallites. In B tissue, the peaks become broadened, as typically occurs in nanocrystals. Finally, ST, presents diffuse features from the organic matrix. We classified and mapped these three basic tissues by using neural network pattern recognition approach as described in detail elsewhere [38]. Briefly, we applied a pattern recognition feed-forward back-propagation network [39]. After a training procedure on inputs profiles constituting about the 1% of the total MxN profiles, the network was simulated on all the MxN diffraction patterns. For each profile we obtained the probability to belong to a specific tissue. This allowed us to define a mask for each tissue type: MSC for SC, MB for B and MST for ST. The mineralization degree of the newly formed bone, is R qðSAXS Þ measured by the SAXS invariant M ¼ qðSAXS12Þ q2 ½IðqÞ IBKG ðqÞdq, to defined in the region from q(SAXS1) = 0.5 nm1 q(SAXS2) = 3.5 nm1; the background intensity, IBKG, corresponds to the inorganic matrix signal in the pore. During the mineralization, collagen packing has been monitored by measuring the lateral spacing of the diffuse peak at about qC = 5.6 nm1 [40,41], modeled with a Gaussian added to a second order polynomial background (see Figure S2). The peak position gives the lateral spacing, D = 2p/qC, while the lateral spacing fluctuations are given by DD/D where DD corresponds to the collagen peak FWHM. Information on the preferred orientation of collagen, averaged throughout the tissue thickness passed by the X-rays is contained in the azimuthal plot, I(U), given by the integrated intensity along the 2p angle around the collagen reflection from q = 4 to q = 6.5 nm1. It shows two peaks separated by about 180° (see Figure S2). The total area under the I(U) curve is the sum of the area under the peaks, AU, which is proportional to the fraction of aligned particles, and the area under the constant background, ABKG, which is proportional to the fraction of randomly oriented particles. The orientation degree, corresponding to the fraction of the aligned fibrils is defined as the Rho-parameter, q = AU/ABKG. Figure S2 shows I(q) and I(U) profiles at the points (1–7) indicated by full circles along the linear selected (y0, z) path in Fig. 3A, where the transmitted intensity in a typical ROI of one of the analyzed samples is mapped. The crystallinity of the mineral nanoparticles is monitored and quantified by the evolution of the WAXS signal from q(WAXS1) = 17 nm1 to q(WAXS2) = 30 nm1 corresponding to the
crystallographic hexagonal P6/m structure with cell parameters a = 9.4162 and c = 6.8791 Å. In parallel to the collagen lateral spacing fluctuations, we mapped the crystallite size normal to two distinct crystallographic planes (i.e. 002 and 310) in the formed bone tissue. The width of the (002) and (310) peaks in the diffraction patterns are inversely proportional to the mean crystallite size, namely L002 and T310, along the longitudinal c-axis and planar transverse directions, respectively, as is explained by the Scherrer Equation [42], Lhkl = kk/Bcoshhkl. Here Lhkl is the crystal length along the (hkl) direction, B the full width at the half maximum (FWHM) of the (hkl)-peak, k is a constant related to the crystallite shape (in the range of 0.87–1.0), and hhkl represents the Bragg angle of the (hkl) peak. The broadening, B, is carried out by fitting of the HA (002) and (310) peaks with a Gaussian function added to a linear background, which is corrected with respect to the instrumental broadening contribution determined by the measurement of the (002) and (310) peak width in HA synthetic powder. Figure S3 reports typical profiles of the 002 and 310 reflections, along the best fitted curves, collected at the indicated (1–7) points of the (y0, z) pathway in Fig. 3A.
3. Results and discussion 3.1. Mapping collagen lateral spacing fluctuations before and after HA crystal nucleation: from isotropic to lyotropic molecular organization In the tomographic section of Fig. 2A, one pore of the scaffold is imaged and the SC, B and ST, stand out well. The mineralization proceeds from the pore toward the scaffold edge, where we detect the older, firstly formed bone. In the pore, the mineralized collagen matrix is composed of collagen fibers quite randomly arranged, without a preferred orientation. In the newly formed bone at the B/ST interface the collagen matrix becomes packed and oriented following the scaffold edges as indicated by asterisks in Fig. 2A. In Fig. 2B we show the collagen fibril intensity in a typical ROI where the B/ST interface is well imaged. The cut of Fig. 2C trough the red line in Fig. 2B, reports the intensity profile allowing us to quantify the thickness of the B/ST interface which is about 30 lm. This dynamic behavior has been deepened by SlXRD to investigate structural fluctuations of the forming B tissue at nano and atomic scale. In the microradiography of Fig. 3A the red region, in which Tr is close to 1, represents the ST region in the scaffold pore lacking in mineralization. Moving toward the scaffold, SC, (blue region),
Fig. 3. (A) Typical map of transmission intensity, Tr, measured by photodiode in a sample ROI. The bar corresponds to 20 lm. (B) Log–log X-ray scattering profiles measured on the (1–7) points along the (y0, z) vertical pathway indicated in A, with a resolution of 25 lm.
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
Tr decreases. This is due to the X-ray absorption by the newly formed bone, B, growing from the scaffold interface and gradually filling the pore. The X-ray scattering profiles I(q), collected on the (1–7) points of the (y0, z) pathway indicated by the white line in Fig 3A are shown in Fig. 3B. We analyzed all the collected X-ray scattering profiles in order to carry out and map the IC, qC, lateral spacing (D), DD/D quantities, defined in par. 2.2, both in the soft and in the bone tissue; moreover we mapped the mineralization degree M, the T310, and L002 size of nanocrystals in the ex-vivo formed bone tissue. Fig. 4A reports a typical mineralization degree evolution, M, measured along the linear pathway (y0, z) starting from the soft tissue, crossing the newly formed bone and ending on the scaffold. During the mineralization, the evolution of collagen amount, IC, lateral spacing, spacing fluctuations DD/D and orientation degree, qC,
313
along the same path, (y0, z), are reported in Fig. 4B–E, respectively. In agreement with XRPClT data, we can locate a ST/B interface, about 30 lm thick, corresponding to the zone where the HA nanocrystals nucleate and collagen fibrils become more compact and aligned. Indeed, at this ST/B interface, we observe that the collagen amount, IC, and orientation degree, qc, increase (Fig. 4A and E) while collagen fibrils’ lateral spacing tends to contract (Fig. 4C). At the same time the system rigidity increases, as indicated by the DD/D fluctuations’ reduction (Fig. 4D). This behavior is quite similar in all vertical linear profiles crossing the newly formed bone tissue, in all measured samples. This compaction and alignment of collagen molecules recalls a lyotropic behavior that typically occurs in a drying process [13,43]. Onsager [44] theoretically studied lyotropic property claiming that elongated particles align spontaneously above a critical volume fraction. Beyond this value, the isotropic phase becomes unstable and the system undergoes an isotropic/nematic first-order transition assuming an orientational order. Several works in vitro tested this liquid crystalline character in collagen solutions for increasing concentration [45–48]. In vivo, the situation becomes more complex since collagen molecules interact with different molecules [49,50]. Second Harmonic Generation measurements highlighted the role of tensile forces generated by fibroblasts and the scaffold geometry in collagen fiber orientation [51]. Our observations are consistent with a drying mechanism recalling the so called polymer-induced liquid-precursor (PILP) model of Olszta et al. [52]. In this model the collagen works as a charged polymer converting the conventional solution into a precursor mineral amorphous liquid-phase containing several ions such as Ca and P. This precursor liquid phase crystallizes and fills the gap sites leaving free space between fibrils; as a consequence the collagen fibrils contract and orient along the scaffold edges. Here we visualize the lyotropic character of collagen dynamics by the probability density functions of IC, lateral spacing (D), DD/D and qC, calculated in all sample ROIs and reported in Fig. 5. The insets represent typical maps of IC, D, DD/D and qC just for one ROI. Once more are evident the differences in the collagen molecules arrangement passing from the soft tissue, to the mineralized tissue (see the arrows in Fig. 4 and Fig. 5): the collagen structural organization becomes more packed and oriented with the early nucleation of HA nanocrystals. It is worth to note that this behavior, as seen at larger micrometric scale, from XPClT measurements appears occurring also at nanoscale. This assures that the dehydration and the embedding in methyl methacrylate, in our sample preparation for SlXRD measurements, do not create significant artifacts. After the collagen fibril compaction, we observe an increase in the lateral spacing indicating that the mineral is growing outside of the gap zones and pushing the collagen apart. We will deepen this dynamics in the following section, describing the mineral particle nucleation and growth. 3.2. Collagen extracellular matrix mineralization: size dynamics of HA growing nanocrystals
Fig. 4. (A) Mineralization degree, M, calculated along the (y0, z) line in Fig. 3A. The collagen amount, lateral spacing, lateral spacing fluctuations and orientation degree are calculated along the same line and represented by the empty circles in B, C, D, E, respectively. The full red circles represent the same quantities calculated on the (1– 7) points along the (y0, z) path in A. The shadowed rectangle represents the early nucleation stage at the ST/B interface, in which the collagen packing is observed; the arrows in these areas represent the increasing of (B) density and (E) orientation degree, q, accompanying the fibers (C) contraction and (D) ordering.
The tomographic section of Fig. 2 shows the bone particles deposited toward the interface with the scaffold. Thanks to the sensitivity of the imaging technique, the micro-granular structure of the deposited bone crystals is evident. Major details at nanoscale of the HA nanocrystal nucleation and growth have been performed by mapping crystallite size along the transverse and longitudinal crystallographic directions 310 and 002. Typical SlXRD maps of HA nanocrystals in plane transversal length, T310, longitudinal length, L002, in the B tissues of a selected ROI are reported in Fig. 6A. A general characterization of all measured quantities in all samples, during the collagen mineralization, is achieved through their
314
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
Fig. 5. Probability density functions of collagen IC, lateral spacing D, DD/D and qC. Full blue circles and red empty squares represent the IC, D, DD/D and qC distributions calculated in the B and ST regions defined by neural network pattern recognition algorithm, as described in 2.3 and in SI. The continuous lines represent the best fits of distributions assumed as Normal for D, DD/D and qC; the collagen amount IC shows a quite skewed distribution, modeled with a lognormal shape line. The insets represent typical maps of the same quantities IC, lateral D-spacing, DD/D fluctuations and qC measured in typical selected ROI. We also note the terminal fatter tail in the lateral spacing distribution arising from enlarging values of D at the B/SC interface, presumably due to the adhesion forces on the scaffold.
Fig. 6. (A) Typical maps of nanocrystal thickness, T310, and L002 length in a ROI of a measured sample. Scatter plot of (B) average transversal thickness of the nanocrystals, T310 and the crystal longitudinal length L002. T310, shows a monotonic increase from less than 3 nm in the early mineralized regions, to more than 5.0 nm at the late mineralized regions toward the scaffold interface B/SC. At the same time, L002, i.e., the dimension along the c-axis direction, decreases from the lightly mineralized region (>20 nm) to the heavily mineralized regions (<15 nm). (C) Collagen lateral spacing and orientation degree as function of the mineralization, M. The vertical dashed line indicates the first stage of mineralization, M0, where nanocrystals grow in the collagen gaps and reach the like-saturation value; here collagen does not change significantly, acting as a stable template. After M0, in the late mineralization stage, nanocrystal growth strive the collagen fibers that change their arrangement, compressing and dilating, to guest the nanocrystal growth exceeding the gaps.
Table 1 Correlation coefficients of size T310, L002, lateral spacing D, lateral spacing fluctuations and orientation degree. Stronger negatively and positively correlated coefficients are indicated by gray and black cells, respectively.
Pearson correlation coefficients, c(i,j) reported in Table 1. Upon inspection of correlation table, it is immediately found the stronger negative correlations c(T310, L002) between the crystallite size in the plane versus the out of plane direction. These negative correlations are well visualized by the evolution of size as a function of the mineralization degree, M, illustrated in the scatter plots of Fig. 6B. In the first mineralization stage (M < M0) longer and thinner nanocrystals form in the axial direction of the fibrils gaps. After this first step crystals grow anisotropically, thickening in the transversal direction, 310, and shortening in the longitudinal 002 direction. Once filled the gap zones, the crystals extend in the overlap zones of collagen [10] that change its stable configuration. Indeed, for M > M0, we find zones with collagen fibrils dilated and disoriented coexisting with zones in which fibrils compress enhancing their
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
orientation degree, as shown in Fig. 6C. This collagen behavior in the mineralized bone tissue can be also described quantitatively by the Pearson correlation coefficients, c(i,j) reported in Table 1. The lateral spacing is positively correlated with its own fluctuations; at the same time, lateral spacing values and DD/D spacing fluctuations are negatively correlated with the collagen fibrils’ orientation degree. 4. Conclusions Spatial–temporal variations of collagen packing during mineral nanoparticle nucleation and growth have been measured by a multiscale approach combining high resolution X-ray phase contrast tomography with scanning X-ray micro-diffraction. We applied basics space statistics to the mapped structural fluctuations of (i) size of the mineral nanoparticles along the longitudinal and transverse directions; (ii) lateral spacing and orientation of the interposed collagen molecules. Spatial correlations and distribution of these structural fluctuations are interpreted as being due to temporal changes’ comprehensive information: first, the early mineral particle nucleation fills the collagen gaps producing an isotropic-lyotropic transition of collagen; second, after having filled the gaps, nanocrystals thicken and shorten recalling a needle-platelets like shape transition. The described experimental and methodological approach can provide a valid contribution for the characterization and treatment of aging and disease in tissues and cells. Indeed, the majority of biomedical tissues are complex, inhomogeneous materials where structural fluctuations and inhomogeneity distribution at mesoscale affect their functionality and health. Therefore, the need of probing tissue structural organization with higher spatial resolution in 3D becomes crucial. In this context the feasibility of 3D microdiffraction scanning techniques [53] could allow to get a deeper insight on the material functionality. Acknowledgement The authors would like to thank Peter Modregger from TOMCAT beamline at Swiss Light Source-PSI for his fundamental assistance during the X-ray Phase Contrast Tomography measurement. Appendix A. Figures with essential color discrimination Certain figures in this article, particularly Figs. 1–6 are difficult to interpret in black and white. The full color images can be found in the on-line version, at 10.1016/j.actbio.2015.05.033. Appendix B. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.actbio.2015.05. 033. References [1] S. Weiner, I. Sagi, L. Addadi, Choosing the crystallization path less traveled, Science 309 (5737) (2005) 1027–1028. [2] M.J. Glimcher, Composition, structure and organization of bone and other mineralized tissue and the mechanism of calcification, in: R.O. Greep, E.B. Astwood (Eds.), Handbook of Physiology: Endocrinology, American Association for the Advancement of Science, Washington, DC, 1976. [3] B. Ganss, R.H. Kim, J. Sodek, Critical reviews in oral biology and medicine: an official publication of the american association of oral biologists, Bone sialoprotein 10 (1) (1999) 79–98. [4] P. Caravan, B. Das, S. Dumas, F.H. Epstein, P.A. Helm, V. Jacques, S. Koerner, A. Kolodziej, L. Shen, W.C. Sun, Z. Zhang, Collagen-targeted MRI contrast agentfor molecular Imaging of fibrosis. Angewandte Chemie-International Edition; 2007; 46: 8171–3.
315
[5] X.Y. He, E. Spoerl, J.H. Tang, J. Liu, Measurement of corneal changes after collagen cross linking using a noninvasive ultrasound system, J. Cataract Refractive Surgery 36 (2010) 1207–1212. [6] D.J.S. Hulmes, J.C. Jesior, A. Miller, C. Berthet-Colominas, C. Wolff, Electronmicroscopy shows periodic structure in collagen fibril cross-sections. In: Proceedings of the national academy of sciences of the united states of America-biological sciences 78, 1981, pp. 3567–3571. [7] G. Cox, E. Kable, A. Jones, I.K. Fraser, F. Manconi, M.D. Gorrell, 3-dimensional imaging of collagen using second harmonic generation, J. Struct. Biology 141 (2003) 53–62. [8] W.L. Chen, T.H. Li, P.J. Su, C.K. Chou, P.T. Fwu, S.J. Lin, D. Kim, P.T.C. So, C.Y. Dong, Second harmonic generation chi tensor microscopy for tissue imaging, Appl. Phys. Lett. 94 (2009) 183902. [9] G.J. Cameron, I.L. Alberts, J.H. Laing, T.J. Wess, Structure of Type I and Type III heterotypic collagen fibrils: an X-ray diffraction study, J. Struct. Biology 137 (2002) 15–22. [10] A.J. Hodge, J.A. Petruska, Recent studies with the electron microscope on ordered aggregates of the tropocollagen molecule, in: G.N. Ramachandran (Ed.), Aspects of Protein Structure, Academic Press, New York, 1963. [11] J. Woodhead-Galloway, P.A. Machin, Modern theories of liquids and the diffuse equatorial X-ray scattering from collagen, Acta Crystallogr. A 32 (1976) 368– 372. [12] P. Fratzl, N. Fratzl-Zelman, K. Klaushofer, G. Vogl, K. Koller, Nucleation and growth of mineral crystals in bone studied by small-angle X-ray scattering, Calcif. Tissue Int. 48 (6) (1991) 407–413. [13] P. Fratzl, N. Fratzl-Zelman, K. Klaushofer, Collagen packing and mineralizationAn X-ray scattering investigation of turkey leg tendon, Biophys. J. 64 (1993) 260–266. [14] D.J.S. Hulmes, A. Miller, Quasi-hexagonal molecular packing in collagen fibrils, Nature 282 (1979) 878–880. [15] D.A. Parry, A.S. Craig, Electron microscope evidence for an 80 A unit in collagen fibrils, Nature 282 (5735) (1979) 213–215. [16] J. Doucet, F. Briki, A. Gourrier, C. Pichon, L. Gumez, S. Bensamoun, J.-F. Sadoc, Modeling the lateral organization of collagen molecules in fibrils using the paracrystal concept, J. Struct. Biology 173 (2011) 197–201. [17] S. Lees, L.C. Bonar, H.A. Mook, A study of dense mineralized tissue by neutron diffraction, Int. J. Biol. Macromol. 6 (1984) 321–326. [18] J.P.R.O. Orgel, T.C. Irving, A. Miller, T.J. Wess, Microfibrillar structure of type I collagen in situ, PNAS 103 (24) (2005) 9001–9005. [19] J. Woodhead-Galloway, P.A. Machin, X-ray scattering from a gas of uniform hard discs using the Percus–Yevick approximation: an application to a ‘planar liquid’, Moi. Phys. 32 (1976) 41–48. [20] J.M. Wallace, B.G. Orr, J.C. Marini, M.M.B. Banaszak Holl, Nanoscale morphology of Type I collagen is altered in the Brtl mouse model of osteogenesis imperfecta, J. Struct. Biol. 173 (2011) 146–152. [21] M. Fang, K.G. Liroff, A.S. Turner, C.M. Les, B.G. Orr, M.M. Holl, Estrogen depletion results in nanoscale morphology changes in dermal collagen, J. Invest. Dermatology 132 (7) (2012) 1791–1797. [22] J.M. Wallace, Q. Chen, M. Fang, B. Erickson, B.G. Orr, M.M. Banaszak Holl, Type I collagen exists as a distribution of nanoscale morphologies in teeth, bones, and tendons, Langmuir. 26 (2010) 7349–7354. [23] B. Erickson, M. Fang, J.M. Wallace, B.G. Orr, C.M. Les, M.M. Banaszak Holl, Nanoscale structure of Type I collagen fibrils: quantitative measurement of Dspacing, Biotech. J. 8 (1) (2013) 117–126. [24] A. Guagliardi, A. Cedola, C. Giannini, M. Ladisa, A. Cervellino, A. Sorrentino, S. Lagomarsino, R. Cancedda, M. Mastrogiacomo, Debye function analysis and 2D imaging of nanoscaled engineered bone, Biomaterials 31 (32) (2010) 8289–8298. [25] K. Grandfield, Bone, implants, and their interfaces, Phys. Today 68 (4) (2015) 40–45. [26] S. Di Fonzo, W. Jark, G. Soullie, A. Cedola, S. Lagomarsino, P. Cloetens, C. Riekel, J. Synchrotron Radiat. 5 (3) (1998) 376–378. [27] A. Cedola, G. Campi, D. Pelliccia, I. Bukreeva, M. Fratini, M. Burghammer, L. Rigon, F. Arfelli, R.C. Chen, D. Dreossi, N. Sodini, S. Mohammadi, G. Tromba, R. Cancedda, M. Mastrogiacomo, Three dimensional visualization of engineered bone and soft tissue by combined X-ray micro-diffraction and phase contrast tomography, Phys. Med. Biol. 59 (2014) 189–201. [28] R. Seidel, A. Gourrier, M. Burghammer, C. Riekel, G. Jeronimidis, O. Paris, Mapping fibre orientation in complex-shaped biological systems with micrometre resolution by scanning X-ray microdiffraction, Micron 39 (2008) 198–205. [29] Y. Ma, B. Aichmayer, O. Paris, P. Fratzl, A. Meibom, R.A. Metzler, Y. Politi, L. Addadi, PUPA Gilbert, S. Weiner, The grinding tip of the sea urchin tooth exhibits exquisite control over calcite crystal orientation and Mg distribution, PNAS 106 (2009) 6048–6053. [30] G. Campi, A. Ricci, A. Guagliardi, C. Giannini, S. Lagomarsino, R. Cancedda, M. Mastrogiacomo, A. Cedola, Early stage mineralization in tissue engineering mapped by high resolution X-ray microdiffraction, Acta Biomater. 8 (2012) 3411–3418. [31] G. Campi, A. Ricci, N. Poccia, L. Barba, G. Arrighetti, M. Burghammer, A.S. Caporale, A. Bianconi, Scanning micro-X-ray diffraction unveils the distribution of oxygen chain nanoscale puddles in YBa2Cu3O6.33, Phys. Rev. B 8 (2013) 014517. [32] Y. Hiraoka, Y. Tanaka, T. Kojima, Y. Takata, M. Oura, Y. Senba, H. Ohashi, Y. Wakabayashi, S. Shin, T. Kimura, Spin-chiral domains in Ba0.5Sr1.5Zn2Fe12O22 observed by scanning resonant X-ray microdiffraction, Phys. Rev. B 84 (2011) 064418.
316
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
[33] G. Campi, G. Ciasca, N. Poccia, A. Ricci, M. Fratini, A. Bianconi, Controlling photoinduced electron transfer via defects self-organization for novel functional macromolecular systems, Curr. Protein Pept. Sci. 15 (4) (2014) 394–399. [34] N. Poccia, G. Campi, A. Ricci, A.S. Caporale, E. Di Cola, T.A. Hawkins, A. Bianconi, Changes of statistical structural fluctuations unveils an early compacted degraded stage of PNS myelin, Sci. Rep. 4 (2014) 5430. [35] R. Quarto, M. Mastrogiacomo, R. Cancedda, S.M. Kutepov, V. Mukhachev, A. Lavroukov, E. Kon, M. Marcacci, Repair of large bone defects with the use of autologous bone marrow stromal cells, N. Eng. J. Med. 344 (5) (2001) 385–386. [36] D. Paganin, S.C. Mayo, T.E. Gureyev, P.R. Miller, S.W. Wilkins, Simultaneous phase and amplitude extraction from a single defocused image of a homogeneous object, J. Microsc. 206 (2002) 33–40. [37] T. Weitkamp, D. Haas, D. Wegrzynek, A. Rack, ANKAphase: software for singledistance phase-retrieval from inline X-ray phase contrast radiographs, J. Synchrotron Radiat. 18 (4) (2011) 617–629. [38] G. Campi, G. Pezzotti, M. Fratini, A. Ricci, M. Burghammer, R. Cancedda, M. Mastrogiacomo, I. Bukreeva, A. Cedola, Imaging regenerating bone tissue based on neural networks applied to micro-diffraction measurements, Appl. Phys. Lett. 103 (2013) 253703. [39] C.M. Bishop, Pattern Recognition and Machine Learning, vol. 1, Springer, New York, 2006. [40] S. Rinnerthaler, P. Roschger, H.F. Jakob, A. Nader, K. Klaushofer, P. Fratzl, Scanning small angle X-ray scattering analysis of human bone sections, Calcif. Tissue Int. 64 (5) (1999) 422–429. [41] A. Cedola, M. Mastrogiacomo, S. Lagomarsino, R. Cancedda, C. Giannini, A. Guagliardi, M. Ladisa, M. Burghammer, F. Rustichelli, V. Komlev, Orientation of mineral crystals by collagen fibers during in vivo bone engineering: an X-ray diffraction imaging study, Spectrochimica Acta B: Atomic Spectroscopy 62 (6) (2007) 642–647.
[42] A. Patterson, The Scherrer formula for X-Ray particle size determination, Phys. Rev. 56 (10) (1939) 978–982. [43] M.M. Giraud-Guille, Liquid crystallinity in condensed type I collagen solutions. a clue to the packing of collagen in extracellular matrices, J. Mol. Biol. 224 (3) (1992) 861–873. [44] L. Onsager, The effects of shape on the interactions of colloidal particles, Ann. N.Y. Acad. Sci. 51 (1949) 627–659. [45] A.R. Khokhlov, A.N. Semenov, Liquid-crystalline ordering in the solution of partially flexible macromolecules, Phys. A 112 (1982) 605–614. [46] F. Henry, M. Nestler, S. Hvidt, J.D. Ferry, A. Veis, Flexibility of collagen determined from dilute solution viscoelastic measurements, Biopolymers 22 (1983) 1747–1758. [47] K. Claire, R. Pecora, Translational and rotational dynamics of collagen in dilute solution, J. Phys. Chem. 101 (1997) 746–753. [48] Z. Dogic, K.R. Purdy, E. Grelet, M. Adams, S. Fraden, Isotropic–nematic phase transition in suspensions of filamentous virus and neutral polymer dextran, Phys. Rev. E 69 (2004) 051702. [49] J.E. Scott, Proteoglycan-fibrillar collagen interactions, Biochem. J. 252 (2) (1988) 313–323. [50] J.E. Scott, Proteoglycan: collagen interactions and subfibrillar structure in collagen fibrils. Implications in the development and ageing of connective tissues, J. Anat. 169 (1990) 23–35. [51] R.M. Delaine-Smith, N.H. Green, S.J. Matcher, S. MacNeil, G.C. Reilly, Monitoring fibrous scaffold guidance of three-dimensional collagen organisation using minimally invasive second harmonic generation, PLOS ONE 9 (2) (2014) 1–14. [52] M.J. Olszta, X.G. Cheng, S.S. Jee, R. Kumar, Y.Y. Kim, M.J. Kaufman, E.P. Douglas, L.B. Gower, Bone structure and formation: a new perspective, Mater. Sci. Eng. R-Reports 58 (3–5) (2007) 77–116. [53] A.R. Gerson, 3D-microdiffraction: a facility for the Australian synchrotron, Curr. Appl. Phys. 8 (2008) 463–466.
Contents lists available at ScienceDirect
Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat
Imaging collagen packing dynamics during mineralization of engineered bone tissue G. Campi a,⇑, M. Fratini b,c, I. Bukreeva d, G. Ciasca e, M. Burghammer f, F. Brun g, G. Tromba h, M. Mastrogiacomo i, A. Cedola d a
Institute of Crystallography, CNR, Via Salaria km 29.300, I-00015 Monterotondo, Roma, Italy Centro Fermi – Museo Storico della Fisica e Centro Studi e Ricerche ‘‘Enrico Fermi’’, Roma, Italy Dipartimento di Scienze, Università di Roma Tre, Roma, Italy d Institute for Chemical and Physical Process, CNR, c/o Physics Dep. at Sapienza University, P-le A. Moro 5, 00185 Roma, Italy e Istituto di Fisica, Universitá Cattolica SC, L.go Francesco Vito 1, I-00168 Roma, Italy f European Synchrotron Radiation Facility, B. P. 220, F-38043 Grenoble Cedex, France g Department of Engineering and Architecture, University of Trieste, Via A. Valerio 10, 34127 Trieste Italy h Sincrotrone Trieste SCpA, 34149 Basovizza, Trieste, Italy i Istituto Nazionale per la Ricerca sul Cancro, and Dipartimento di Medicina Sperimentale dell’Università di Genova & AUO San Martino Istituto Nazionale per la Ricerca sul Cancro, Largo R. Benzi 10, 16132 Genova, Italy b c
a r t i c l e
i n f o
Article history: Received 25 February 2015 Received in revised form 8 May 2015 Accepted 28 May 2015 Available online 3 June 2015 Keywords: Collagen Biomineralization Fluctuations Spatial statistics Scanning X-ray micro-diffraction
a b s t r a c t The structure and organization of the Type I collagen microfibrils during mineral nanoparticle formation appear as the key factor for a deeper understanding of the biomineralization mechanism and for governing the bone tissue physical properties. In this work we investigated the dynamics of collagen packing during ex-vivo mineralization of ceramic porous hydroxyapatite implant scaffolds using synchrotron high resolution X-ray phase contrast micro-tomography (XPClT) and synchrotron scanning micro X-ray diffraction (SlXRD). While XPClT provides the direct 3D image of the collagen fibers network organization with micrometer spatial resolution, SlXRD allows to probe the structural statistical fluctuations of the collagen fibrils at nanoscale. In particular we imaged the lateral spacing and orientation of collagen fibrils during the anisotropic growth of mineral nanocrystals. Beyond throwing light on the bone regeneration multiscale process, this approach can provide important information in the characterization of tissue in health, aging and degeneration conditions. Statement of Significance: BONE grafts are the most common transplants after the blood transfusions. This makes the bone-tissue regeneration research of pressing scientific and social impact. Statement of Significance: Bone is a complex hierarchical structure, where the interplay of organic and inorganic mineral phases at different length scale (from micron to atomic scale) affect its functionality and health. Thus, the understanding of bone tissue regeneration requires to image its spatial-temporal evolution (i) with high spatial resolution and (ii) at different length scale. Statement of Significance: We exploited high spatial resolution X-ray Phase Contrast micro Tomography and Scanning micro X-ray Diffraction in order to get new insight on the engineered tissue formation mechanisms. This approach could open novel routes for the early detection of different degenerative conditions of tissue. Ó 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
1. Introduction Bone is a complex composite and nanostructured material formed via dynamic interactions between an organic matrix of biomacromolecules and an inorganic mineral phase [1,2]. The
⇑ Corresponding author. Tel.: +39 06 90672624; fax: +39 06 90672630. E-mail address: [email protected] (G. Campi). http://dx.doi.org/10.1016/j.actbio.2015.05.033 1742-7061/Ó 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
inorganic phase consists of carbonate apatite nanocrystals growing in the organic matrix composed by collagen (type I) proteins, with some minor noncollagenous proteins and minor amounts of lipids and osteogenic factors [3]. Collagen 3D arrangement constitutes a basic issue to be addressed since it plays a fundamental role in the bone tissue biomechanical properties. Collagen structure and organization has been investigated and visualized by several techniques such as magnetic resonance imaging (MRI) [4], ultrasound [5], electron [6] and optical imaging [7,8]. The obtained results
310
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
indicate that the collagen molecule is a right-handed triple-helix, called fibril, about 300 nm long and 1.0–1.5 nm diameter. Collagen fibers are composed from about 100 up to more than 1000 tightly packed fibrils. X-ray diffraction (XRD) is an effective tool for studying the fibrillar periodic assembly of collagen. Typically, X-ray scattering patterns from collagen present Bragg spots along the axial direction, indicative of a long-range ordered phase made of staggered parallel fibrils [9,10]. This results in a repeating axial D-spacing providing open sites for cross-links and mineral nucleation. Alongside the axial Bragg peaks, collagen XRD measurements show a diffuse equatorial scattering characteristic of a short-range order typical of fluid-like arrangement of the molecules [11–13]. Hulmes et al. interpreted XRD data as a molecular crystal with quasi-hexagonal molecular packing [14] while an arrangement in concentric layers was assumed by Parry et al. [15]. Doucet et al. [16] showed that the equatorial X-ray scattering patterns of collagen can easily be modeled using the paracrystal organization of fibers into parallel rods packed in a hexagonal or pseudo-hexagonal lateral network. Unfortunately, conventional XRD as well as neutron diffraction [17] has a strong limitation since it provides insight only into the average concentration and structure of the collagen [12,13,18]. However collagen is, from a morphological point of view, a fluctuating and inhomogeneous tissue [19,20]: this makes quite difficult to model its structural features with the conventional experimental approaches, requiring high spatial resolution probes to map spatial changes in structural organization. In this framework the importance of the collagen axial D-spacing fluctuations has been recently recognized; indeed, significant alterations in bone collagen axial D-spacing distribution have been found in Osteogenesis Imperfecta [21] and long term estrogen depletion [22]. A quantitative method for space resolved axial D-spacing analysis at sub-micrometer scale, used Atomic Force Microscopy (AFM) and two-dimensional Fast Fourier Transform (2D FFT) analysis [23]. Anyway, being the AFM a surface approach, it does not allow to investigate the interplay of collagen spacing and minerals arrangement in the bulk composite material. Indeed, it is little known about the spatial distribution of both collagen and mineral crystals packing in the radial direction of the collagen fibrils. This is also due to the fact that conventional imaging techniques, such as electron microscopic tomography, can reach resolution of 4–6 nm that cannot resolve either the mineral nanocrystal size during the early stage of mineralization [24] either the collagen molecular lateral spacing (<2 nm) [9,25]. In this paper we overcome both the limitation of the 2D techniques and the limited resolution of the conventional tomography, by combining the 3D visualization of the collagen bundles provided by high resolution XPClT with the high spatial resolution structural information obtained by SlXRD. This multiscale approach is needed due to the fact that collagen fibrils organize into bundles at micrometer scale via interfibrillar cross-links. At this scale, high resolution XPClT is an appropriate tool to visualize collagen bundles since unlike conventional radiography and tomography it is able to image small density variations in weakly absorbing materials such as biological samples [26,27]. On the other hand, SlXRD technique allows to probe both the k-space and real space, visualizing structural features at atomic scale and nanoscale such as the lateral packing of collagen molecules and the size of mineral particles. The spatial mapping of the simultaneously collected Small Angle X-ray Scattering (SAXS) and Wide Angle X-ray Scattering (WAXS) signals, can readily monitor the mineralization process according to its temporal evolution. Although SlXRD has been already used for investigating ultrastructure in biological tissue, [28–30] in this work it is used to study the dynamics of the collagen fibrils during the HA nanocrystal nucleation and growth with a spatial statistical approach. More specifically, we measured the fluctuations of (i) the
lateral spacing and preferred orientation of collagen fibrils (ii) the longitudinal and transverse size of mineral nanocrystals. Thus, the order degree of collagen arrangement and mineral growth during the bone tissue formation has been quantified by applying basics spatial statistics tools to the measured quantities on the large amount of collected data from different samples. This approach has been recently used to get key information on systems presenting structural fluctuations and heterogeneity on (sub)micrometric scales in different research fields ranging from material science to biomedicine [31–34]. 2. Materials and methods All experimental animal procedures were carried out in the IRCCS AOU San Martino –IST Animal Facility (Genoa, Italy), in the respect of the national current regulations regarding the protection of animals used for scientific purpose (D.lgsvo 27/01/1992, n. 116). Research protocols have been evaluated and approved by the IRCCS AOU San Martino –IST Ethical Committee for animal experimentation (CSEA) as Animal use project n. 336 communicated to The Italian Ministry of Health, having regard to the article 7 of the D.lgs 116/92. Here we investigated collagen packing dynamics during nucleation and growth of bone mineral nanocrystals in ex vivo conditions: expanded bone marrow mesenchymal stem cells (BMSC) seeded onto porous ceramic scaffolds and subcutaneously implanted in the mouse [35]. After four weeks the scaffolds are removed from the host animals and the newly formed bone analyzed. To improve the statistical significance 5 samples extracted from different animals were studied. 2.1. Sample preparation Marrow aspirates were obtained from the iliac crest of the experimental sheep as part of a protocol approved by the competent ethical authority. Bone marrow stromal cell (BMSC) cultures were counted with a nuclear stain, suspended in Coon’s modified Ham’s F12 medium supplemented with 10% FCS, 1 ng/ml of human recombinant FGF-2 and subsequently plated in 100-mm dishes at 0.5–1.0 107 cells per dish. Cultures were incubated at 37 °C in a humidified atmosphere containing 95% air and 5% CO2. Before reaching confluence, cells were enzymatically detached and passaged. Experiments were performed using a pool of first passage BMSC derived from marrows of four animals. Osteogenic properties of the BMSC were evaluated by an ‘‘in vivo’’ assay in an immuno-deficient mice model. First passage cells expanded in standard medium were tripsynized, resuspended in a fibrinogen solution (Tissuecol; Baxter, Italia) to a final concentration of 62.5 106 cells/ml and loaded onto highly porous ceramic scaffolds (100% hydroxyapatite cubes, 3 3 3 mm3; FinCeramica, Faenza, Italy); an appropriate volume of 20 ll of Thrombin was added to ignite the enzymatic cleavage that originates a fibrin clot around and within the ceramic, entrapping the cells. Two samples were implanted subcutaneously in each mouse. After four weeks the samples dedicated to XlPCT were harvested, washed in Phosphate Buffered Saline (PBS) three times and fixed in paraformahaldehyde (4% in PBS) for 2–3 h at 4 °C. Additional washes in PBS removed the residual fixative. The samples for SlXRD were additionally dehydrated in ethanol at increasing concentration, embedded in methylmetacrilate and transversally cut using a diamond saw (Gillings- Hamco, Hamco Machines, Inc., Rochester, N.Y., U.S.A) in serial sections of about 100 lm thick. We choose 100 microns as a well suited thickness value, quite less than the general mineral bone attenuation length (about 350 lm at X-ray incident energy of 12.7 keV).
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
311
2.2. Synchrotron X-ray phase contrast microscopy measurements and data analysis The X-ray phase contrast imaging measurements were carried out on the TOMCAT beamline at the Swiss Light Source, SLS, Switzerland. The incident beam energy was set at 0.729 Å (17 keV) by using a multilayer based monochromator. The beam size was 2 2 mm2 at the sample placed about 30 m downstream the monochromator. A (2048 2048) pixels CCD camera optically coupled with a microscope system mounted with 1.25x objective providing a voxel size of 0.64 lm3 was set at a distance of 5 cm from the sample. The real resolution, i.e. the minimal detectable detail, was about 3 voxels which means about 2 2 2 lm3. A layout of the experimental setup is illustrated in Fig. 1. We collected 1601 projections covering a total angle range of 360° for the 3D tomographic reconstruction. The sample was protected by a mylar foil and measured without any further preparation. A single distance phase retrieval algorithm [36] has been applied to all projections of the tomographic measurements, using the ANKAphase code [37]. When applied to all tomographic projections, the retrieved phase maps can be fed to a standard filtered back-projection algorithm to obtain phase tomograms. A typical image obtained from the tomographic reconstructed volume of a sample ROI is reported in Fig. 2. 2.3. Synchrotron X-ray micro-diffraction measurements and data analysis Bone sections with thickness of about 100 lm were measured on the ID13 beamline of the European Synchrotron Radiation Facility, ESRF, France. The scanning micro-diffraction setup (Fig. 1), equipped with a double-crystal monochromator and a Kirkpatrick-Baez mirror as focusing system, supplied a beam size of 1 1 lm2 with a wavelength of 0.976 Å (12.703 keV). The sample was mounted at a distance from the detector of 110 cm. The beam center, detector tilt and sample-to-detector distance were calibrated using a silver behenate standard. The samples were scanned by piezo-scanning stage with 0.1 micron repeatability using a step size of 5 lm in both vertical, z, and horizontal, y, direction. Diffraction patterns were recorded in transmission by a FreLon CCD detector (2048 2048 pixels of 50 lm2) with a typical acquisition time of 5 s and corrected for parasitic scattering by taking into account the transmission, Tr, measured by a photodiode using the same scan parameters. The 2D diffraction patterns have been radially and azimuthally integrated to provide 1D profiles
Fig. 2. (A) Tomographic section from a reconstructed volume of a sample implanted for 4 weeks. The different gray-levels represent different densities in the sample. The thickness is of 512 lm. We can distinguish the three different types of tissues: bone, B, the collagenous soft tissue, ST and the scaffold, SC. The collagen fiber compaction (e.g. indicated by asterisk) can be seen between the formed B tissue and the ST tissue. (B) Collagen fibrils intensity in a typical ROI. (C) Collagen fibrils Intensity profile trough the red line in (B); the thickness of the B/ST interface in found to be about 30 lm.
of intensity, I(q), vs. transfer moment, q = 4psin(h)/k, and angular intensity distribution I(U). The achieved q range of (0.5–30) nm1 allowed us to measure simultaneously both the SAXS and the WAXS signal for investigating structure from nano to atomic scale. In each sample we performed X-ray microdiffraction measurements on two regions of interest (ROI) including ceramic scaffold, SC, newly formed bone, B, and organic collagenous soft tissue, ST.
Fig. 1. Layout of experimental setup adopted for high resolution synchrotron X-ray phase contrast micro-tomography and synchrotron scanning X-ray micro-diffraction.
312
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
Typical 2D diffraction patterns and integrated profiles, recorded at different positions and characterizing different tissues SC, B and ST, are shown in Figure S1. The SC tissue shows XRD patterns with sharp peaks, due to micrometric size of HA crystallites. In B tissue, the peaks become broadened, as typically occurs in nanocrystals. Finally, ST, presents diffuse features from the organic matrix. We classified and mapped these three basic tissues by using neural network pattern recognition approach as described in detail elsewhere [38]. Briefly, we applied a pattern recognition feed-forward back-propagation network [39]. After a training procedure on inputs profiles constituting about the 1% of the total MxN profiles, the network was simulated on all the MxN diffraction patterns. For each profile we obtained the probability to belong to a specific tissue. This allowed us to define a mask for each tissue type: MSC for SC, MB for B and MST for ST. The mineralization degree of the newly formed bone, is R qðSAXS Þ measured by the SAXS invariant M ¼ qðSAXS12Þ q2 ½IðqÞ IBKG ðqÞdq, to defined in the region from q(SAXS1) = 0.5 nm1 q(SAXS2) = 3.5 nm1; the background intensity, IBKG, corresponds to the inorganic matrix signal in the pore. During the mineralization, collagen packing has been monitored by measuring the lateral spacing of the diffuse peak at about qC = 5.6 nm1 [40,41], modeled with a Gaussian added to a second order polynomial background (see Figure S2). The peak position gives the lateral spacing, D = 2p/qC, while the lateral spacing fluctuations are given by DD/D where DD corresponds to the collagen peak FWHM. Information on the preferred orientation of collagen, averaged throughout the tissue thickness passed by the X-rays is contained in the azimuthal plot, I(U), given by the integrated intensity along the 2p angle around the collagen reflection from q = 4 to q = 6.5 nm1. It shows two peaks separated by about 180° (see Figure S2). The total area under the I(U) curve is the sum of the area under the peaks, AU, which is proportional to the fraction of aligned particles, and the area under the constant background, ABKG, which is proportional to the fraction of randomly oriented particles. The orientation degree, corresponding to the fraction of the aligned fibrils is defined as the Rho-parameter, q = AU/ABKG. Figure S2 shows I(q) and I(U) profiles at the points (1–7) indicated by full circles along the linear selected (y0, z) path in Fig. 3A, where the transmitted intensity in a typical ROI of one of the analyzed samples is mapped. The crystallinity of the mineral nanoparticles is monitored and quantified by the evolution of the WAXS signal from q(WAXS1) = 17 nm1 to q(WAXS2) = 30 nm1 corresponding to the
crystallographic hexagonal P6/m structure with cell parameters a = 9.4162 and c = 6.8791 Å. In parallel to the collagen lateral spacing fluctuations, we mapped the crystallite size normal to two distinct crystallographic planes (i.e. 002 and 310) in the formed bone tissue. The width of the (002) and (310) peaks in the diffraction patterns are inversely proportional to the mean crystallite size, namely L002 and T310, along the longitudinal c-axis and planar transverse directions, respectively, as is explained by the Scherrer Equation [42], Lhkl = kk/Bcoshhkl. Here Lhkl is the crystal length along the (hkl) direction, B the full width at the half maximum (FWHM) of the (hkl)-peak, k is a constant related to the crystallite shape (in the range of 0.87–1.0), and hhkl represents the Bragg angle of the (hkl) peak. The broadening, B, is carried out by fitting of the HA (002) and (310) peaks with a Gaussian function added to a linear background, which is corrected with respect to the instrumental broadening contribution determined by the measurement of the (002) and (310) peak width in HA synthetic powder. Figure S3 reports typical profiles of the 002 and 310 reflections, along the best fitted curves, collected at the indicated (1–7) points of the (y0, z) pathway in Fig. 3A.
3. Results and discussion 3.1. Mapping collagen lateral spacing fluctuations before and after HA crystal nucleation: from isotropic to lyotropic molecular organization In the tomographic section of Fig. 2A, one pore of the scaffold is imaged and the SC, B and ST, stand out well. The mineralization proceeds from the pore toward the scaffold edge, where we detect the older, firstly formed bone. In the pore, the mineralized collagen matrix is composed of collagen fibers quite randomly arranged, without a preferred orientation. In the newly formed bone at the B/ST interface the collagen matrix becomes packed and oriented following the scaffold edges as indicated by asterisks in Fig. 2A. In Fig. 2B we show the collagen fibril intensity in a typical ROI where the B/ST interface is well imaged. The cut of Fig. 2C trough the red line in Fig. 2B, reports the intensity profile allowing us to quantify the thickness of the B/ST interface which is about 30 lm. This dynamic behavior has been deepened by SlXRD to investigate structural fluctuations of the forming B tissue at nano and atomic scale. In the microradiography of Fig. 3A the red region, in which Tr is close to 1, represents the ST region in the scaffold pore lacking in mineralization. Moving toward the scaffold, SC, (blue region),
Fig. 3. (A) Typical map of transmission intensity, Tr, measured by photodiode in a sample ROI. The bar corresponds to 20 lm. (B) Log–log X-ray scattering profiles measured on the (1–7) points along the (y0, z) vertical pathway indicated in A, with a resolution of 25 lm.
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
Tr decreases. This is due to the X-ray absorption by the newly formed bone, B, growing from the scaffold interface and gradually filling the pore. The X-ray scattering profiles I(q), collected on the (1–7) points of the (y0, z) pathway indicated by the white line in Fig 3A are shown in Fig. 3B. We analyzed all the collected X-ray scattering profiles in order to carry out and map the IC, qC, lateral spacing (D), DD/D quantities, defined in par. 2.2, both in the soft and in the bone tissue; moreover we mapped the mineralization degree M, the T310, and L002 size of nanocrystals in the ex-vivo formed bone tissue. Fig. 4A reports a typical mineralization degree evolution, M, measured along the linear pathway (y0, z) starting from the soft tissue, crossing the newly formed bone and ending on the scaffold. During the mineralization, the evolution of collagen amount, IC, lateral spacing, spacing fluctuations DD/D and orientation degree, qC,
313
along the same path, (y0, z), are reported in Fig. 4B–E, respectively. In agreement with XRPClT data, we can locate a ST/B interface, about 30 lm thick, corresponding to the zone where the HA nanocrystals nucleate and collagen fibrils become more compact and aligned. Indeed, at this ST/B interface, we observe that the collagen amount, IC, and orientation degree, qc, increase (Fig. 4A and E) while collagen fibrils’ lateral spacing tends to contract (Fig. 4C). At the same time the system rigidity increases, as indicated by the DD/D fluctuations’ reduction (Fig. 4D). This behavior is quite similar in all vertical linear profiles crossing the newly formed bone tissue, in all measured samples. This compaction and alignment of collagen molecules recalls a lyotropic behavior that typically occurs in a drying process [13,43]. Onsager [44] theoretically studied lyotropic property claiming that elongated particles align spontaneously above a critical volume fraction. Beyond this value, the isotropic phase becomes unstable and the system undergoes an isotropic/nematic first-order transition assuming an orientational order. Several works in vitro tested this liquid crystalline character in collagen solutions for increasing concentration [45–48]. In vivo, the situation becomes more complex since collagen molecules interact with different molecules [49,50]. Second Harmonic Generation measurements highlighted the role of tensile forces generated by fibroblasts and the scaffold geometry in collagen fiber orientation [51]. Our observations are consistent with a drying mechanism recalling the so called polymer-induced liquid-precursor (PILP) model of Olszta et al. [52]. In this model the collagen works as a charged polymer converting the conventional solution into a precursor mineral amorphous liquid-phase containing several ions such as Ca and P. This precursor liquid phase crystallizes and fills the gap sites leaving free space between fibrils; as a consequence the collagen fibrils contract and orient along the scaffold edges. Here we visualize the lyotropic character of collagen dynamics by the probability density functions of IC, lateral spacing (D), DD/D and qC, calculated in all sample ROIs and reported in Fig. 5. The insets represent typical maps of IC, D, DD/D and qC just for one ROI. Once more are evident the differences in the collagen molecules arrangement passing from the soft tissue, to the mineralized tissue (see the arrows in Fig. 4 and Fig. 5): the collagen structural organization becomes more packed and oriented with the early nucleation of HA nanocrystals. It is worth to note that this behavior, as seen at larger micrometric scale, from XPClT measurements appears occurring also at nanoscale. This assures that the dehydration and the embedding in methyl methacrylate, in our sample preparation for SlXRD measurements, do not create significant artifacts. After the collagen fibril compaction, we observe an increase in the lateral spacing indicating that the mineral is growing outside of the gap zones and pushing the collagen apart. We will deepen this dynamics in the following section, describing the mineral particle nucleation and growth. 3.2. Collagen extracellular matrix mineralization: size dynamics of HA growing nanocrystals
Fig. 4. (A) Mineralization degree, M, calculated along the (y0, z) line in Fig. 3A. The collagen amount, lateral spacing, lateral spacing fluctuations and orientation degree are calculated along the same line and represented by the empty circles in B, C, D, E, respectively. The full red circles represent the same quantities calculated on the (1– 7) points along the (y0, z) path in A. The shadowed rectangle represents the early nucleation stage at the ST/B interface, in which the collagen packing is observed; the arrows in these areas represent the increasing of (B) density and (E) orientation degree, q, accompanying the fibers (C) contraction and (D) ordering.
The tomographic section of Fig. 2 shows the bone particles deposited toward the interface with the scaffold. Thanks to the sensitivity of the imaging technique, the micro-granular structure of the deposited bone crystals is evident. Major details at nanoscale of the HA nanocrystal nucleation and growth have been performed by mapping crystallite size along the transverse and longitudinal crystallographic directions 310 and 002. Typical SlXRD maps of HA nanocrystals in plane transversal length, T310, longitudinal length, L002, in the B tissues of a selected ROI are reported in Fig. 6A. A general characterization of all measured quantities in all samples, during the collagen mineralization, is achieved through their
314
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
Fig. 5. Probability density functions of collagen IC, lateral spacing D, DD/D and qC. Full blue circles and red empty squares represent the IC, D, DD/D and qC distributions calculated in the B and ST regions defined by neural network pattern recognition algorithm, as described in 2.3 and in SI. The continuous lines represent the best fits of distributions assumed as Normal for D, DD/D and qC; the collagen amount IC shows a quite skewed distribution, modeled with a lognormal shape line. The insets represent typical maps of the same quantities IC, lateral D-spacing, DD/D fluctuations and qC measured in typical selected ROI. We also note the terminal fatter tail in the lateral spacing distribution arising from enlarging values of D at the B/SC interface, presumably due to the adhesion forces on the scaffold.
Fig. 6. (A) Typical maps of nanocrystal thickness, T310, and L002 length in a ROI of a measured sample. Scatter plot of (B) average transversal thickness of the nanocrystals, T310 and the crystal longitudinal length L002. T310, shows a monotonic increase from less than 3 nm in the early mineralized regions, to more than 5.0 nm at the late mineralized regions toward the scaffold interface B/SC. At the same time, L002, i.e., the dimension along the c-axis direction, decreases from the lightly mineralized region (>20 nm) to the heavily mineralized regions (<15 nm). (C) Collagen lateral spacing and orientation degree as function of the mineralization, M. The vertical dashed line indicates the first stage of mineralization, M0, where nanocrystals grow in the collagen gaps and reach the like-saturation value; here collagen does not change significantly, acting as a stable template. After M0, in the late mineralization stage, nanocrystal growth strive the collagen fibers that change their arrangement, compressing and dilating, to guest the nanocrystal growth exceeding the gaps.
Table 1 Correlation coefficients of size T310, L002, lateral spacing D, lateral spacing fluctuations and orientation degree. Stronger negatively and positively correlated coefficients are indicated by gray and black cells, respectively.
Pearson correlation coefficients, c(i,j) reported in Table 1. Upon inspection of correlation table, it is immediately found the stronger negative correlations c(T310, L002) between the crystallite size in the plane versus the out of plane direction. These negative correlations are well visualized by the evolution of size as a function of the mineralization degree, M, illustrated in the scatter plots of Fig. 6B. In the first mineralization stage (M < M0) longer and thinner nanocrystals form in the axial direction of the fibrils gaps. After this first step crystals grow anisotropically, thickening in the transversal direction, 310, and shortening in the longitudinal 002 direction. Once filled the gap zones, the crystals extend in the overlap zones of collagen [10] that change its stable configuration. Indeed, for M > M0, we find zones with collagen fibrils dilated and disoriented coexisting with zones in which fibrils compress enhancing their
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
orientation degree, as shown in Fig. 6C. This collagen behavior in the mineralized bone tissue can be also described quantitatively by the Pearson correlation coefficients, c(i,j) reported in Table 1. The lateral spacing is positively correlated with its own fluctuations; at the same time, lateral spacing values and DD/D spacing fluctuations are negatively correlated with the collagen fibrils’ orientation degree. 4. Conclusions Spatial–temporal variations of collagen packing during mineral nanoparticle nucleation and growth have been measured by a multiscale approach combining high resolution X-ray phase contrast tomography with scanning X-ray micro-diffraction. We applied basics space statistics to the mapped structural fluctuations of (i) size of the mineral nanoparticles along the longitudinal and transverse directions; (ii) lateral spacing and orientation of the interposed collagen molecules. Spatial correlations and distribution of these structural fluctuations are interpreted as being due to temporal changes’ comprehensive information: first, the early mineral particle nucleation fills the collagen gaps producing an isotropic-lyotropic transition of collagen; second, after having filled the gaps, nanocrystals thicken and shorten recalling a needle-platelets like shape transition. The described experimental and methodological approach can provide a valid contribution for the characterization and treatment of aging and disease in tissues and cells. Indeed, the majority of biomedical tissues are complex, inhomogeneous materials where structural fluctuations and inhomogeneity distribution at mesoscale affect their functionality and health. Therefore, the need of probing tissue structural organization with higher spatial resolution in 3D becomes crucial. In this context the feasibility of 3D microdiffraction scanning techniques [53] could allow to get a deeper insight on the material functionality. Acknowledgement The authors would like to thank Peter Modregger from TOMCAT beamline at Swiss Light Source-PSI for his fundamental assistance during the X-ray Phase Contrast Tomography measurement. Appendix A. Figures with essential color discrimination Certain figures in this article, particularly Figs. 1–6 are difficult to interpret in black and white. The full color images can be found in the on-line version, at 10.1016/j.actbio.2015.05.033. Appendix B. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.actbio.2015.05. 033. References [1] S. Weiner, I. Sagi, L. Addadi, Choosing the crystallization path less traveled, Science 309 (5737) (2005) 1027–1028. [2] M.J. Glimcher, Composition, structure and organization of bone and other mineralized tissue and the mechanism of calcification, in: R.O. Greep, E.B. Astwood (Eds.), Handbook of Physiology: Endocrinology, American Association for the Advancement of Science, Washington, DC, 1976. [3] B. Ganss, R.H. Kim, J. Sodek, Critical reviews in oral biology and medicine: an official publication of the american association of oral biologists, Bone sialoprotein 10 (1) (1999) 79–98. [4] P. Caravan, B. Das, S. Dumas, F.H. Epstein, P.A. Helm, V. Jacques, S. Koerner, A. Kolodziej, L. Shen, W.C. Sun, Z. Zhang, Collagen-targeted MRI contrast agentfor molecular Imaging of fibrosis. Angewandte Chemie-International Edition; 2007; 46: 8171–3.
315
[5] X.Y. He, E. Spoerl, J.H. Tang, J. Liu, Measurement of corneal changes after collagen cross linking using a noninvasive ultrasound system, J. Cataract Refractive Surgery 36 (2010) 1207–1212. [6] D.J.S. Hulmes, J.C. Jesior, A. Miller, C. Berthet-Colominas, C. Wolff, Electronmicroscopy shows periodic structure in collagen fibril cross-sections. In: Proceedings of the national academy of sciences of the united states of America-biological sciences 78, 1981, pp. 3567–3571. [7] G. Cox, E. Kable, A. Jones, I.K. Fraser, F. Manconi, M.D. Gorrell, 3-dimensional imaging of collagen using second harmonic generation, J. Struct. Biology 141 (2003) 53–62. [8] W.L. Chen, T.H. Li, P.J. Su, C.K. Chou, P.T. Fwu, S.J. Lin, D. Kim, P.T.C. So, C.Y. Dong, Second harmonic generation chi tensor microscopy for tissue imaging, Appl. Phys. Lett. 94 (2009) 183902. [9] G.J. Cameron, I.L. Alberts, J.H. Laing, T.J. Wess, Structure of Type I and Type III heterotypic collagen fibrils: an X-ray diffraction study, J. Struct. Biology 137 (2002) 15–22. [10] A.J. Hodge, J.A. Petruska, Recent studies with the electron microscope on ordered aggregates of the tropocollagen molecule, in: G.N. Ramachandran (Ed.), Aspects of Protein Structure, Academic Press, New York, 1963. [11] J. Woodhead-Galloway, P.A. Machin, Modern theories of liquids and the diffuse equatorial X-ray scattering from collagen, Acta Crystallogr. A 32 (1976) 368– 372. [12] P. Fratzl, N. Fratzl-Zelman, K. Klaushofer, G. Vogl, K. Koller, Nucleation and growth of mineral crystals in bone studied by small-angle X-ray scattering, Calcif. Tissue Int. 48 (6) (1991) 407–413. [13] P. Fratzl, N. Fratzl-Zelman, K. Klaushofer, Collagen packing and mineralizationAn X-ray scattering investigation of turkey leg tendon, Biophys. J. 64 (1993) 260–266. [14] D.J.S. Hulmes, A. Miller, Quasi-hexagonal molecular packing in collagen fibrils, Nature 282 (1979) 878–880. [15] D.A. Parry, A.S. Craig, Electron microscope evidence for an 80 A unit in collagen fibrils, Nature 282 (5735) (1979) 213–215. [16] J. Doucet, F. Briki, A. Gourrier, C. Pichon, L. Gumez, S. Bensamoun, J.-F. Sadoc, Modeling the lateral organization of collagen molecules in fibrils using the paracrystal concept, J. Struct. Biology 173 (2011) 197–201. [17] S. Lees, L.C. Bonar, H.A. Mook, A study of dense mineralized tissue by neutron diffraction, Int. J. Biol. Macromol. 6 (1984) 321–326. [18] J.P.R.O. Orgel, T.C. Irving, A. Miller, T.J. Wess, Microfibrillar structure of type I collagen in situ, PNAS 103 (24) (2005) 9001–9005. [19] J. Woodhead-Galloway, P.A. Machin, X-ray scattering from a gas of uniform hard discs using the Percus–Yevick approximation: an application to a ‘planar liquid’, Moi. Phys. 32 (1976) 41–48. [20] J.M. Wallace, B.G. Orr, J.C. Marini, M.M.B. Banaszak Holl, Nanoscale morphology of Type I collagen is altered in the Brtl mouse model of osteogenesis imperfecta, J. Struct. Biol. 173 (2011) 146–152. [21] M. Fang, K.G. Liroff, A.S. Turner, C.M. Les, B.G. Orr, M.M. Holl, Estrogen depletion results in nanoscale morphology changes in dermal collagen, J. Invest. Dermatology 132 (7) (2012) 1791–1797. [22] J.M. Wallace, Q. Chen, M. Fang, B. Erickson, B.G. Orr, M.M. Banaszak Holl, Type I collagen exists as a distribution of nanoscale morphologies in teeth, bones, and tendons, Langmuir. 26 (2010) 7349–7354. [23] B. Erickson, M. Fang, J.M. Wallace, B.G. Orr, C.M. Les, M.M. Banaszak Holl, Nanoscale structure of Type I collagen fibrils: quantitative measurement of Dspacing, Biotech. J. 8 (1) (2013) 117–126. [24] A. Guagliardi, A. Cedola, C. Giannini, M. Ladisa, A. Cervellino, A. Sorrentino, S. Lagomarsino, R. Cancedda, M. Mastrogiacomo, Debye function analysis and 2D imaging of nanoscaled engineered bone, Biomaterials 31 (32) (2010) 8289–8298. [25] K. Grandfield, Bone, implants, and their interfaces, Phys. Today 68 (4) (2015) 40–45. [26] S. Di Fonzo, W. Jark, G. Soullie, A. Cedola, S. Lagomarsino, P. Cloetens, C. Riekel, J. Synchrotron Radiat. 5 (3) (1998) 376–378. [27] A. Cedola, G. Campi, D. Pelliccia, I. Bukreeva, M. Fratini, M. Burghammer, L. Rigon, F. Arfelli, R.C. Chen, D. Dreossi, N. Sodini, S. Mohammadi, G. Tromba, R. Cancedda, M. Mastrogiacomo, Three dimensional visualization of engineered bone and soft tissue by combined X-ray micro-diffraction and phase contrast tomography, Phys. Med. Biol. 59 (2014) 189–201. [28] R. Seidel, A. Gourrier, M. Burghammer, C. Riekel, G. Jeronimidis, O. Paris, Mapping fibre orientation in complex-shaped biological systems with micrometre resolution by scanning X-ray microdiffraction, Micron 39 (2008) 198–205. [29] Y. Ma, B. Aichmayer, O. Paris, P. Fratzl, A. Meibom, R.A. Metzler, Y. Politi, L. Addadi, PUPA Gilbert, S. Weiner, The grinding tip of the sea urchin tooth exhibits exquisite control over calcite crystal orientation and Mg distribution, PNAS 106 (2009) 6048–6053. [30] G. Campi, A. Ricci, A. Guagliardi, C. Giannini, S. Lagomarsino, R. Cancedda, M. Mastrogiacomo, A. Cedola, Early stage mineralization in tissue engineering mapped by high resolution X-ray microdiffraction, Acta Biomater. 8 (2012) 3411–3418. [31] G. Campi, A. Ricci, N. Poccia, L. Barba, G. Arrighetti, M. Burghammer, A.S. Caporale, A. Bianconi, Scanning micro-X-ray diffraction unveils the distribution of oxygen chain nanoscale puddles in YBa2Cu3O6.33, Phys. Rev. B 8 (2013) 014517. [32] Y. Hiraoka, Y. Tanaka, T. Kojima, Y. Takata, M. Oura, Y. Senba, H. Ohashi, Y. Wakabayashi, S. Shin, T. Kimura, Spin-chiral domains in Ba0.5Sr1.5Zn2Fe12O22 observed by scanning resonant X-ray microdiffraction, Phys. Rev. B 84 (2011) 064418.
316
G. Campi et al. / Acta Biomaterialia 23 (2015) 309–316
[33] G. Campi, G. Ciasca, N. Poccia, A. Ricci, M. Fratini, A. Bianconi, Controlling photoinduced electron transfer via defects self-organization for novel functional macromolecular systems, Curr. Protein Pept. Sci. 15 (4) (2014) 394–399. [34] N. Poccia, G. Campi, A. Ricci, A.S. Caporale, E. Di Cola, T.A. Hawkins, A. Bianconi, Changes of statistical structural fluctuations unveils an early compacted degraded stage of PNS myelin, Sci. Rep. 4 (2014) 5430. [35] R. Quarto, M. Mastrogiacomo, R. Cancedda, S.M. Kutepov, V. Mukhachev, A. Lavroukov, E. Kon, M. Marcacci, Repair of large bone defects with the use of autologous bone marrow stromal cells, N. Eng. J. Med. 344 (5) (2001) 385–386. [36] D. Paganin, S.C. Mayo, T.E. Gureyev, P.R. Miller, S.W. Wilkins, Simultaneous phase and amplitude extraction from a single defocused image of a homogeneous object, J. Microsc. 206 (2002) 33–40. [37] T. Weitkamp, D. Haas, D. Wegrzynek, A. Rack, ANKAphase: software for singledistance phase-retrieval from inline X-ray phase contrast radiographs, J. Synchrotron Radiat. 18 (4) (2011) 617–629. [38] G. Campi, G. Pezzotti, M. Fratini, A. Ricci, M. Burghammer, R. Cancedda, M. Mastrogiacomo, I. Bukreeva, A. Cedola, Imaging regenerating bone tissue based on neural networks applied to micro-diffraction measurements, Appl. Phys. Lett. 103 (2013) 253703. [39] C.M. Bishop, Pattern Recognition and Machine Learning, vol. 1, Springer, New York, 2006. [40] S. Rinnerthaler, P. Roschger, H.F. Jakob, A. Nader, K. Klaushofer, P. Fratzl, Scanning small angle X-ray scattering analysis of human bone sections, Calcif. Tissue Int. 64 (5) (1999) 422–429. [41] A. Cedola, M. Mastrogiacomo, S. Lagomarsino, R. Cancedda, C. Giannini, A. Guagliardi, M. Ladisa, M. Burghammer, F. Rustichelli, V. Komlev, Orientation of mineral crystals by collagen fibers during in vivo bone engineering: an X-ray diffraction imaging study, Spectrochimica Acta B: Atomic Spectroscopy 62 (6) (2007) 642–647.
[42] A. Patterson, The Scherrer formula for X-Ray particle size determination, Phys. Rev. 56 (10) (1939) 978–982. [43] M.M. Giraud-Guille, Liquid crystallinity in condensed type I collagen solutions. a clue to the packing of collagen in extracellular matrices, J. Mol. Biol. 224 (3) (1992) 861–873. [44] L. Onsager, The effects of shape on the interactions of colloidal particles, Ann. N.Y. Acad. Sci. 51 (1949) 627–659. [45] A.R. Khokhlov, A.N. Semenov, Liquid-crystalline ordering in the solution of partially flexible macromolecules, Phys. A 112 (1982) 605–614. [46] F. Henry, M. Nestler, S. Hvidt, J.D. Ferry, A. Veis, Flexibility of collagen determined from dilute solution viscoelastic measurements, Biopolymers 22 (1983) 1747–1758. [47] K. Claire, R. Pecora, Translational and rotational dynamics of collagen in dilute solution, J. Phys. Chem. 101 (1997) 746–753. [48] Z. Dogic, K.R. Purdy, E. Grelet, M. Adams, S. Fraden, Isotropic–nematic phase transition in suspensions of filamentous virus and neutral polymer dextran, Phys. Rev. E 69 (2004) 051702. [49] J.E. Scott, Proteoglycan-fibrillar collagen interactions, Biochem. J. 252 (2) (1988) 313–323. [50] J.E. Scott, Proteoglycan: collagen interactions and subfibrillar structure in collagen fibrils. Implications in the development and ageing of connective tissues, J. Anat. 169 (1990) 23–35. [51] R.M. Delaine-Smith, N.H. Green, S.J. Matcher, S. MacNeil, G.C. Reilly, Monitoring fibrous scaffold guidance of three-dimensional collagen organisation using minimally invasive second harmonic generation, PLOS ONE 9 (2) (2014) 1–14. [52] M.J. Olszta, X.G. Cheng, S.S. Jee, R. Kumar, Y.Y. Kim, M.J. Kaufman, E.P. Douglas, L.B. Gower, Bone structure and formation: a new perspective, Mater. Sci. Eng. R-Reports 58 (3–5) (2007) 77–116. [53] A.R. Gerson, 3D-microdiffraction: a facility for the Australian synchrotron, Curr. Appl. Phys. 8 (2008) 463–466.