Advanced Electron Microscopy to Study the Collagen-Mineral Nanocomposite in Human Bone: From On-Axis Electron Tomography to 4D-STEM

From a materials perspective, the bones in our skeleton are a composite of proteins, primarily type I collagen, and a calcium-phosphate mineral, hierarchically assembled across several length scales [1]. At the smallest level of bone architecture, collagen fibrils and mineral particles form an intricate nanocomposite whose precise organization has puzzled researchers for decades. Different interpretations of the collagen-mineral spatial relationship have largely stemmed from technical limitations of the imaging tools used, especially when inferring three-dimensional (3D) information from two-dimensional (2D) (scanning) transmission electron microscopy (S/TEM) images. Adding to the complexity, bone structure appears different depending on the orientation of collagen fibrils with respect to the image plane. When imaged in-plane, collagen fibrils display a characteristic banding pattern due to their periodic staggering [2-4]. However, when visualized out-of-plane, seemingly empty or low-contrast circular regions, hereinafter referred to as “holes”, become visible [2-4]. Are these “holes” cross-sectional images of collagen fibrils, or are they nanopores, possibly occupied by non-collagenous proteins? Extensive debate has also taken place around the nature of bone mineral, particularly regarding the distinction between mineral located within the collagen fibrils (intra-fibrillar [5]) or on their outer surface (inter- or extra-fibrillar [6]), as well as the mineral habit; are mineral particles needle-like or plate-like? Recently, we combined on-axis Z-contrast electron tomography (ET) with correlative energy-dispersive X-ray spectroscopy (EDX) tomography to shed light on the nature of the alleged nanoporosity and on the shape of mineral in human bone [7]. For this, we employed rod-shaped samples up to 700 nm in diameter prepared from the femoral cortex by focused ion beam (FIB) annular milling, and acquired tilt series of both STEM images and EDX maps of representative elements (Ca and P as mineral signatures, and C and N as proxies for collagen) over a ±90° range. This allowed us to combine structural and compositional information at high resolution over a larger volume than conventional electron tomography, where lamellar-shaped samples are employed. Moreover, rotating the samples over a complete ±90° range in on-axis ET enabled artifact-free 3D reconstructions, overcoming the missing-wedge sampling limitation of conventional ET. Analysis of these 3D reconstructions via virtual re-slicing revealed that the “holes” observed in one plane correspond to the banding pattern in mutually orthogonal planes, demonstrating that the apparent nanoporosity actually represents collagen fibrils in cross-section. Segmentation of extra-fibrillar mineral identified platelets a few tens of nanometers in length and width and a few nanometers in thickness, splaying over multiple fibrils in a cross-fibrillar fashion. EDX tomography confirmed mineralization both within and outside the collagen fibrils, but highlighted an enrichment in mineral content in the extra-fibrillar phase [7]. While this work provided spatial and compositional information on the collagen-mineral nanocomposite, some questions on the mineral itself remain. Bone mineral is generally accepted to be a carbonate-substituted apatite deficient in hydroxyl ions [8], but other phases such as octacalcium phosphate have also been reported, especially at mineralization sites [9]. Some debate also persists on eventual differences between intra-fibrillar vs. extra-fibrillar mineral, with some authors proposing the former to be in a less crystalline state [10]. To obtain crystallographic information (degree of crystallinity, orientation, and morphology) on bone mineral at the individual collagen fibril level, we employed four-dimensional (4D)-STEM to overcome the limitations of conventional selected area electron diffraction (SAED). In SAED, diffraction patterns (DPs) contain crystallographic information averaged over the area defined by the SA aperture, typically a few micrometers in diameter, hence encompassing several mineralized collagen fibrils. In contrast, in 4D-STEM, a focused electron beam 1-10 nm in diameter is rastered across the sample, and a DP is recorded at each position of the probe. This results in a 4D dataset where a 2D DP corresponds to each location of a 2D raster of the beam. This enables virtual dark-field imaging post-acquisition. For our 4D-STEM experiments, we used FIB milling to prepare electron-transparent lamellae of human femoral cortical bone in different orientations with respect to the long axis of the femur. This made it possible to have collagen fibrils with varying orientations ranging from in-plane to out-of-plane. High-angle annular dark-field (HAADF)-STEM images and 4D-STEM data were acquired in each sample in a S/TEM instrument operated in scanning mode at 300 kV with a convergence semi-angle of 0.7 mrad, and equipped with an ultrafast 4D Camera [11] (TEAM 0.5, National Center for Electron Microscopy, Lawrence Berkeley National Laboratory, CA, USA). Over one million DPs were collected in 2 × 2 µm2 regions at a probe step size smaller than 2 nm, while limiting beam exposure thanks to the ultrafast acquisition (87,000 frames per second). Thanks to 4D-STEM, we were able to resolve and map crystallographic information in greater detail than SAED, thus achieving a more precise spatial correlation between mineral crystallinity and bone structure at the nanometer scale. When collagen fibrils were in-plane, we observed the 30°-wide arc typically reported for the {002} reflections of bone mineral (c-axis of apatite crystals). We then divided this arc into three sub-components by placing distinct virtual apertures in the diffraction space, in turn revealing patches of similarly oriented crystals, mostly within the inter-fibrillar spaces (Fig. 1). On the other hand, no {002} reflections were detected in samples where collagen fibrils were out-of-plane, as expected since the c-axis of apatite is normal to the image plane. In this case, when considering virtual dark-field images obtained from distinct reflections in the DPs, different sets of plate-shaped regions were illuminated, suggesting that the mineral is organized into straight single-crystalline domains arranged around individual collagen fibrils (Fig. 2). Overall, on-axis correlative ET and EDX tomographies, together with 4D-STEM, can expand the characterization toolbox to probe bone structure at the level of its building block unit, the mineralized collagen fibril. The combination of these techniques is potentially helpful beyond its application to normal, healthy bone in the study of pathological conditions that affect bone at the most fundamental level. The information provided by these techniques can indeed reveal how various diseases affect the structure and composition of mineral aggregates, as well as mineral organization and crystallinity in relation to collagen fibrils. HAADF-STEM image of the human bone sample oriented with collagen fibrils in-plane (A) and corresponding dark-field images overlaid on the HAADF-STEM image (B). These dark-field images were obtained by placing virtual apertures (red, green, and blue circles) on the {002} reflections in the global DP (virtual apertures and dark-field images are colour-coded) (C). HAADF-STEM image of the human bone sample oriented with collagen fibrils out-of-plane (A), and virtual dark-field images obtained from the region marked in A (B). These dark-field images were obtained by placing virtual apertures on distinct reflections in the DPs (virtual apertures and dark-field images are colour-coded) (C).

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.