In vivo high-resolution synchrotron radiation imaging of collagen-induced arthritis in a rodent model
aDepartment of Orthopedic Surgery, School of Medicine, Catholic University of Taegu, Taegu, Republic of Korea, bDepartment of Anatomy, School of Medicine, Catholic University of Taegu, Taegu, Republic of Korea, cDepartment of Internal Medicine, School of Medicine, Catholic University of Taegu, Taegu, Republic of Korea, dDepartment of Neurosurgery, School of Medicine, Catholic University of Taegu, Taegu, Republic of Korea, eDepartment of Thoracic Surgery, Seoul National University, Seoul, Republic of Korea, and fDepartment of Biomedical Engineering, School of Medicine, Catholic University of Taegu, Taegu, Republic of Korea
*Correspondence e-mail: email@example.com
In vivo microstructures of the affected feet of collagen-induced arthritic (CIA) mice were examined using a high-resolution synchrotron radiation (SR) X-ray refraction technique with a polychromatic beam issued from a bending magnet. The CIA models were obtained from six-week-old DBA/1J mice that were immunized with bovine type II collagen and grouped as grades 0–3 according to a clinical scoring for the severity of arthritis. An X-ray shadow of a specimen was converted into a visual image on the surface of a CdWO4 scintillator that was magnified using a microscopic objective lens before being captured with a digital charge-coupled-device camera. Various changes in the joint microstructure, including cartilage destruction, periosteal born formation, articular bone thinning and erosion, marrow invasion by pannus progression, and widening joint space, were clearly identified at each level of arthritis severity with an equivalent pixel size of 2.7 µm. These high-resolution features of destruction in the CIA models have not previously been available from any other conventional imaging modalities except histological light microscopy. However, thickening of the synovial membrane was not resolved in composite images by the SR refraction imaging method. In conclusion, in vivo SR X-ray microscopic imaging may have potential as a diagnostic tool in small animals that does not require a histochemical preparation stage in examining microstructural changes in joints affected with arthritis. The findings from the SR images are comparable with standard histopathology findings.
Rheumatoid arthritis (RA) is the most common inflammatory joint disorder, affecting about 0.5% of adults. It is a chronic disease involving bone and joint destruction (Smith & Haynes, 2002). Much of the knowledge that has been acquired for determining new RA treatments is based on studies using experimental arthritis models. The collagen-induced arthritis (CIA) model is a very useful experimental tool for understanding the pathogenesis and therapeutic challenges of RA since it presents a destructive polyarthritis that has many clinical and radiological similarities to rheumatoid arthritis (Trentham et al., 1977). Therefore, the CIA model has been used to examine the degree of synovial inflammation, periarticular change and bony structural damage, such as bony erosion, using various imaging methods (Carpenter et al., 1994; Barck et al., 2004; Beckmann et al., 1998; Clavel et al., 2008).
The evaluation of potential therapies requires techniques that quantify the severity of disease in experimental animals. Joint inflammation can be visually observed, but there are no standard methods for quantifying the degree of bone destruction in this modality. Although radiography is the gold standard for the quantification of joint destruction, it does not allow for assessing damage in soft tissues like cartilage. Radiographic criteria of RA is mostly attributed to bone erosions (Ando et al., 2006; Goldring, 2002). However, the histopathological features of RA are characterized by diverse pathological changes in the cartilage, bone and surrounding soft tissues of the joint that have been well established through histological analysis of the CIA mice model (Beckmann et al., 1998; Ando et al., 2006; Barck et al., 2004; Larsson et al., 2004; Clavel et al., 2008). It is clinically important to understand these multiform pathological features in order to develop either new imaging modalities to detect changes at early stages of RA or new treatment to stop the evolution of the illness.
Histological examination is an accurate method of assessing the severity of RA lesions in humans and animals; however, histological examination requires biopsy in the patient, which is an invasive procedure. In order to carry out histological analysis in animal models of RA the animal needs to be sacrificed. This allows for evaluation of the disease at only one point in the illness evolution and prevents follow-up throughout the rest of the disease course. Although magnetic resonance imaging (MRI) may be an attractive method in the RA diagnosis, poor results have been found in some studies that attempt to demonstrate a correlation between MRI and histopathological observations (Ostergaard et al., 1997, 1998; Hunter & Conaghan, 2006). Recently, the high beam coherence determined by the small X-ray source size at a third-generation SR facility has permitted the development of new imaging techniques like the `phase-contrast' technique, which allows images of biological samples to be recorded showing superior contrast than that obtainable using conventional absorption-based X-ray imaging (Wilkins et al., 1996).
This biomedical imaging method has produced high-resolution images of the microstructures of various tissues, including breast, lung and brain tumor (Bravin et al., 2007; Jheon et al., 2006; Pfeiffer et al., 2007). Therefore, in experimental models of RA, the synchrotron radiation (SR) phase-contrast imaging method, combined with a high-resolution CCD detector, can be useful in assessing longitudinal and translational outcomes in CIA lesions in vivo.
In our present work, we examine the in vivo microstructures of affected feet joints in CIA mice using a SR phase-contrast imaging technique.
Fourteen DBA/1J mice (six weeks of age; Jackson Laboratories, Bar Harbor, ME, USA) including two normal mice were used for these studies. This study protocol was approved by the ethics committee of the Catholic University Hospital of Taegu. The experimental animals were housed in a controlled environment and provided with standard rodent food and water. Bovine type II collagen (CII) (Chondrex, Redmond, WA, USA) was dissolved in 0.05 M acetic acid at a concentration of 2 mg ml−1 by stirring overnight at 277 K. The dissolved CII was then frozen at 203 K until needed. Freund's complete adjuvant (CFA) was prepared by adding Mycobacterium tuberculosis H37Ra at a concentration of 4 mg ml−1. Before the injection, CII was emulsified with an equal volume of CFA. The CIA was induced as follows: 100 µl of the emulsion (containing 100 µg of CII) was injected intradermally at the base of the tail. After three weeks a second injection of CII in CFA was administered. After the second immunization the degree of joint inflammation was observed every other day. The severity of arthritis in each paw was assessed using a four-point clinical scoring system from 0 to 3 with 0 indicating no sign of inflammation, 1 for slight swelling and redness of the paw with the involvement of less than two joints, 2 for pronounced edema and involvement of more than two joints, and 3 for severe arthritis of the entire paw and digits with joint rigidity and ankylosis (Butler et al., 1999). Two investigators (CHC and JYC) independently assessed the degree of arthritis for each mouse. Mice were grouped according to severity assessment. Two or three mice were assigned to each group.
Normal control and grade 3 mice were sacrificed after SR imaging in order to assess corresponding structures in the histological sections. The removed forefoot was fixed in 10% neutral buffered formalin, decalcified in 10% EDTA and embedded in paraffin. The prepared tissue blocks of the mouse paw were then sectioned at a 10 µm thickness. The sections were stained with hematoxylin and eosin and observed by LM (light microscopy).
The SR imaging studies were performed at the 1B2/microprobe beamline of the Pohang Light Source (PLS) in Korea. The PLS is designed to provide SR with continuous wavelengths down to 10 nm at 2.5 GeV. The experimental set-up is similar to that presented by Jheon et al. (2006) and Kim et al. (2008), and is schematically shown in Fig. 1.
Mice were anesthetized by injecting 10 ml kg−1 body weight of ketamine. They were imaged live attached to an acrylic plate in a supine position, thus facing the horizontal beam. The sample was positioned 25 m away from a bending-magnet source. In order to allow for mesh scanning of the sample, it was mounted onto a motorized XY stage and scanned. An aluminium filter (1.0 mm) was placed in the beam. Under these conditions polychromatic synchrotron X-rays were used in an energy range of 7–14 keV with a calculated dose rate between 1.6 and 3 Gy s−1. The exposure time was taken between 0.5 and 1 s for scanning a view. The total dose per view was varied between 0.8 and 3 Gy.
The X-ray shadow of the specimen was converted into a visual signal on the surface of a thin (100 µm) scintillation crystal (CdWO4), which was placed 5 cm away from the specimen to take advantage of the increase in phase contrast. The visual image was magnified using a ×10 microscope objective and captured using a digital CCD camera (Megaplus II ES 2001).
The fields of view of these images were determined by the size of the CCD chip and the optical magnification. For a 2.54 cm CCD camera with a ×10 microscope objective, the field of view was 1.18 mm × 0.89 mm because the CCD camera has an active area of 11.8 mm × 8.9 mm with 1600 × 1200 pixels. Therefore, each pixel corresponded to approximately 0.74 µm × 0.74 µm. The spatial resolution of our image was about 1.48 µm after a 4:1 binning process.
Normal joint structures visible in SR images were compared with histological slices. The results showed that the microstructures including cartilage, subchondral bone and surrounding soft tissue structures in the normal feet joints could be clearly identified in the synchrotron images and had good correlation with the histological findings (Fig. 2). Phase-contrast SR high-resolution radiography allowed us to identify the amount of articular cartilage covering a subchondral bone. In addition, each chondrocyte lacuna with a diameter of 10–30 µm was identified as a small hole. SR images exhibited well defined carpal bones, metatarsal heads and styloid processes of radial bone in normal joints that were comparable with that of conventional histological examination by light microscopy. Structural information in a slice direction was overlaid on a present propagation-based phase-contrast SR image. A composite radiographic image made of images of tissue thickening will provide more structural information than will a single histological section, as shown by comparing Figs. 2(b) and 2(c). On the other hand, such an overlaid picture may make it difficult to resolve fine structural features like synovial membranes, owing to overlapping and partial volume effects in the slice direction.
The high-resolution X-ray images of the CIA model grouped according to the severity of arthritis showed aggravated changes of joint structures and bony structures as arthritis grade increased, as shown in Fig. 3. In SR images of front limb joints and bones in higher-grade CIA mice, we observed destructive changes including loss of subchondral bone and articular cartilage, irregularities of the joint surface, and decreased radio-opacity.
Thinning and erosion of cortical and carpal bones, styloid processes of the radius, and metacarpal head loss were visualized more conspicuously in higher grades, as shown in the expanded view images of each part of the paw (Figs. 4–6).
Morphological alterations in the joint and bone of an SR image were compared with the histological images as demonstrated in Figs. 4(B) and 5(A). Destructive changes of articular structures including loss of articular cartilage (Figs. 4–6), irregularities in joint surfaces, widening of joint spaces (Fig. 6) and marrow invasions (Fig. 4B) were identified in the SR images. An increase of joint space that may be associated with the growth of the synovium and complete infiltration of inflammatory cells was delineated more conspicuously in the high-resolution X-ray images than in the MRI. There was decreased radio-opacity in the joint space of the grade 3 group when compared with lower severity grades. This contrast may be associated with growth of the synovium and filling of inflammatory cells into the joint space where synovial fluid is normally present. We observed detailed loss of subchondral bone, decreased radio-opacity owing to porosity, cortical bone thinning, and erosion in SR images that were also seen in the histologic images (Figs. 4B and Fig. 5A).
Small radio-opaque fragments around thinned cortical bone were observed in the body of the phalanx in a severely diseased paw in both SR and histological images of the CIA grade 3 mice. These were confirmed as bony trabecules that had been produced from new periosteal bone formation (Figs. 5 and 6). We found that subchondral bone thinning occurred in grade 1 CIA mice, while bone was later missing in grade 3 mice (Figs. 5 and 6). Decreased radio-opacity of the marrow cavity of the fifth digit was observed and is shown in Fig. 4B. This may have been caused by the invasion of pannus tissue into the marrow which increased the severity of the arthritis.
The pathological findings that were identified in the SR images of each clinical grade are summarized below and are compared with the grade criteria of other imaging modalities in Table 1. Similar patterns of destruction were observed in the SR images of other imaging modalities of clinical scoring-based grades 2 and 3. However, structural damage appeared more extensive in the grade 3 group.
‡Ando et al. (2006).
Grade 1. The cortical bone of the proximal phalanx, the subchondral bone of the metacarpophalangeal joint of the third digit, and the articular structure of the distal interphalangeal joint were still intact. However, destructive changes of the joint and bone, including irregularity of the joint surfaces between carpal bones, cortical bone erosion and thinning, periosteal new bone formation, and subchondral bone thinning had already begun.
Grade 2. Severe destructive changes in joints and bones were observed in almost all parts of the paw. In Fig. 4 it was difficult to define carpal bones, metacarpal heads, and styloid processes owing to the extensive destruction. Cortical bone thinning and erosion that are typical of CIA appeared in most of the phalanges and metacarpal bones. Destructive changes of articular structures including loss of articular cartilage, irregularity in joint surface, and marrow invasion of pannus tissue could be clearly identified in all joints including distal interphalangeal joints.
Grade 3. The destructive changes in bones and joints were similar to the findings of grade 2, but more extensive damage occurred over entire parts of the paw. Extensive destruction included pannus progression, bony trabecules from periosteal new bone formation, and widening of the joint space, as shown in Figs. 5 and 6.
Conventional radiographic scorings were assessed mainly by the degree of bone erosion found in the images (Devauchelle et al., 2004). SR imaging findings included not only bone destruction but damage in cartilage and surrounding tissue. This is quite advantageous compared with ultrasound imaging where destruction of the joint cannot always be detected (Szkudlarek et al., 2004; Clavel et al., 2008). Furthermore, this high-resolution imaging modality produced comparable findings with those of the LM investigation with regard to reflecting the degree of damage and micro-architectural changes to the joints. Observations in this study demonstrated that high-resolution SR X-ray refraction imaging can identify early inflammatory changes in the affected joint of a CIA model. In CIA, the only current way to evaluate arthritis within the paw joint is through histological analysis, which requires the animal to be sacrificed. This method of analysis only allows for the examination of a given mouse at one point in time. In this study, SR imaging allows us to evaluate arthritis in the paws of many animals for the entire duration of the course of arthritis. Therefore it is anticipated that this SR radiographic technique might play a potential role as a follow-up tool for disease activity or treatment response in RA.
Thickening of the synovial membrane was not directly visualized in this SR refraction imaging study, although a phase-contrast image was acquired with high resolution of 2.7 µm. This may be due to the overlaying nature and low contrast from synovial membrane hyperplasia (SMH) for the propagation-based two-dimensional refraction imaging technique. Presently, SMH can be identified only by histological analysis. MRI microscopy and ultrasound imaging cannot demonstrate SMH in their CIA model studies owing to the low resolution of their modalities that is typically around 80 µm (Beckmann et al., 1995; Clavel et al., 2008). Imaging of SMH would be highly important because of its relevance to the early diagnosis of RA. Tomographic images with mono-energetic X-rays issued from a SR source would be suitable to resolve SMH of the CIA model, by using one of the three phase-contrast imaging techniques: interferometry, analyzer-based imaging (ABI) or grating interferometry. X-ray interferometry is the most sensitive method for detecting the small differences in refractive index of soft tissues (Momose et al., 1998). ABI and grating interferometry are capable of revealing the contrast generated not only from absorption but also from refraction and extinction arising from the internal structures of the object examined (Chapman et al., 1997; Pfeiffer et al., 2006). These structures may provide either refraction-sensitive or diffraction-sensitive phase contrast at the tissues interfaces.
Rheumatoid arthritis is accompanied by three types of bone loss: focal articular bone erosion, juxta-articular osteopenia adjacent to inflamed joints, and systemic osteoporosis (Goldring & Gravallese, 2000). It is clear that radiographic joint destruction is strongly associated with systemic osteoporosis (Forsblad D'Elia et al., 2003; Sambrook, 2000). A general decrease in the radio-opacity, owing to increasing porosity in higher-grade CIA, was observed in the front limb bones of CIA mice. This suggests that phase-contrast SR imaging can potentially diagnose juxta-articular osteopenia adjacent to inflamed joints more clearly and at an earlier stage than can conventional radiographic techniques. In this regard, mono-energetic X-ray beams should be used in SR imaging to quantitatively detect opacity changes as a measure of bone porosity or bone mineral density (BMD) without the beam-hardening effect of conventional BMD techniques. Although this was not realised in this propagation-based two-dimensional imaging experiment, recent SR micro-CT studies demonstrated quantitative analysis of bone mineral density and micro-architecture in vivo in a rodent model (Bayat et al., 2005; Matsumoto et al., 2006). These estimations showed the feasibility of this methodology by measuring fractional bone volume from high-resolution three-dimensional imaging of the trabecular bone architecture and calculating BMD based on linear attenuation coefficient values.
The radiation doses delivered in this study were quite high compared with those of conventional radiography. This may constitute a real limitation to the use of polychromatic SR in non-invasive in vivo imaging. The main reason for using polychromatic SR was to obtain a sufficient X-ray flux penetrating the sample thickness for imaging with a detector system. A 7 keV monochromatic X-ray beam, the highest X-ray energy with maximum flux, was available from a monochromator at the current 1B2 beamline of Pohang Accelerator Laboratory (PAL), but this energy is not able to penetrate a sample thicker than 2 mm. Although delivered doses could be reduced by using an automatic fast shutter to stop the beam between two exposures or a detector system with higher detective quantum efficiency (Bravin et al., 2003), they will remain very high because of the ultra-high resolution applied for this examination.
Currently PAL is undergoing an upgrade project to increase injector energy up to 3 GeV, and to prepare a multipole-wiggler-type insertion device that will provide an enhanced X-ray flux over a large energy range (20–60 keV) for biomedical imaging application. When this project is completed, the radiation dose necessary for producing monochromatic SR X-ray images in the 20–60 keV energy range would be similar to that necessary to produce images of the same samples with a detector combined with an anti-scatter grid as used in clinical imaging (Burattini et al., 1995).
Observations in this study demonstrated that high-resolution SR X-ray refraction imaging can identify early inflammatory changes and associated bone erosion in the affected joints of a CIA model in mice with fine pathological features that are comparable with those of histological analysis. Further developments in SR imaging may be necessary to resolve synovial membrane hyperplasia using more sophisticated phase-contrast tomographic techniques with a mono-energetic beam.
This work was supported by a grant (2009-0088454) from the National Research Foundation of Korea and a grant from Institute of Medical Science, Catholic University of Taegu.
Ando, W., Hashimoto, J., Nampei, A., Tsuboi, H., Tateishi, K., Ono, T., Nakamura, N., Ochi, T. & Yoshikawa, H. (2006). J. Bone Miner. Metab. 24, 274–282. Web of Science CrossRef PubMed CAS Google Scholar
Barck, K. H., Lee, W. P., Diehl, L. J., Ross, J., Gribling, P., Zhang, Y., Nquven, K., van Bruqqen, N., Hurst, S. & Carano, R. A. (2004). Arthritis Rheum. 50, 3377–3386. Web of Science CrossRef PubMed Google Scholar
Bayat, S., Apostol, L., Boller, E., Brochard, T. & Peyrin, F. (2005). Nucl. Instrum. Methods Phys. Res. A, 548, 247–252. Web of Science CrossRef CAS Google Scholar
Beckmann, N., Bruttel, K., Mir, A. & Rudin, M. (1995). Magn. Res. Imaging, 13, 1013–1017. CrossRef CAS Web of Science Google Scholar
Beckmann, N., Bruttel, K., Schuurman, H. & Mir, A. (1998). J. Magn. Res. 131, 8–16. Web of Science CrossRef CAS Google Scholar
Bravin, A., Fiedler, S., Coan, P., Labiche, J.-C., Ponchut, C., Peterzol, A. & Thomlinson, W. (2003). Nucl. Instrum. Methods Phys. Res. A, 510, 35–40. Web of Science CrossRef CAS Google Scholar
Bravin, A., Keyriläinen, J., Fernández, M., Fiedler, S., Nemoz, C., Karjalainen-Lindsberg, M.-L., Tenhunen, M., Virkkunen, P., Leidenius, M., von Smitten, K., Sipil, P. & Suortti, P. (2007). Phys. Med. Biol. 52, 2197–2211. Web of Science CrossRef PubMed Google Scholar
Burattini, E., Cossu, E., Maggio, C. D., Gambaccini, M., Indovina, P. L., Marziani, M., Pocek, M., Simeoni, S. & Simonetti, G. (1995). Radiology, 195, 239–244. CrossRef CAS PubMed Web of Science Google Scholar
Butler, D. M., Malfait, A. M., Maini, R. N., Brennan, F. M. & Feldmann, M. (1999). Eur. J. Immunol. 29, 2205–2212. CrossRef PubMed CAS Google Scholar
Carpenter, T. A., Everett, J. R., Hall, L. D., Harper, G. P., Hodgson, R. J., James, M. F. & Watson, P. J. (1994). Skelet. Radiol. 23, 429–CCCC437. CrossRef CAS Google Scholar
Chapman, D., Thomlinson, W., Johnston, R. E., Washburn, D., Pisano, E., Gműr, N., Zhong, Z., Menk, R., Arfelli, F. & Sayers, D. (1997). Phys. Med. Biol. 42, 2015–2025. CrossRef CAS PubMed Web of Science Google Scholar
Clavel, G., Marchiol-Fournigault, C., Renault, G., Boissier, M.-C., Fradelizi, D. & Bessis, N. (2008). Ann. Rheum. Dis. 67, 1765–1772. Web of Science CrossRef PubMed CAS Google Scholar
Devauchelle, P. V., Saraux, A., Berthelot, J. M., Alapetite, S., Jousse, S., Chales, G., Thorel, J. B., Hoang, S., Nouy-Troll, I., Martin, A., Chiocchia, G., Youinou, P. & LeGoff, P. (2004). J. Rheumatol. 31, 66–70. PubMed Google Scholar
Forsblad D'Elia, H., Larsen, A. & Waltbrand, E. (2003). Ann. Rheum. Dis. 62, 617–623. CrossRef PubMed CAS Google Scholar
Goldring, S. R. (2002). Curr. Opin. Rheumatol. 14, 406–410. Web of Science CrossRef PubMed Google Scholar
Goldring, S. R. & Gravallese, E. M. (2000). Arthritis Res. 2, 33–37. Web of Science CrossRef PubMed CAS Google Scholar
Hunter, D. J. & Conaghan, P. G. (2006). Curr. Opin. Rheumatol. 18, 157–162. Web of Science CrossRef PubMed Google Scholar
Jheon, S. H., Youn, H.-S., Kim, H.-T., Choi, G.-H. & Kim, J.-K. (2006). Microsc. Res. Tech. 69, 656–659. Web of Science CrossRef PubMed Google Scholar
Kim, B.-I., Kim, K.-H., Youn, H.-S., Jheon, S., Kim, J.-K. & Kim, H.-T. (2008). Microsc. Res. Tech. 71, 443–447. Web of Science CrossRef PubMed Google Scholar
Larsson, E., Erlandsson Harris, H., Larsson, A., Månsson, B., Saxne, T. & Klareskog, L. (2004). Rheumatology, 43, 428–434. Web of Science CrossRef PubMed CAS Google Scholar
Matsumoto, T., Yoshino, M., Asano, T., Uesugi, K., Todoh, M. & Tanaka, M. (2006). J. Appl. Physiol. 100, 274–280. Web of Science CrossRef PubMed Google Scholar
Momose, A., Takeda, T., Itai, Y., Yoneyama, A. & Hirano, K. (1998). Medical Applications of Synchrotron Radiation, edited by M. Ando and C. Uyama, pp. 63–71. Tokyo: Springer-Verlag. Google Scholar
Ostergaard, M., Stoltenberg, M., Lovgreen-Nielsen, P., Volck, B., Jensen, C. H. & Lorenzen, I. (1997). Arthritis Rheum. 40, 1856–1867. CrossRef CAS PubMed Web of Science Google Scholar
Ostergaard, M., Stoltenberg, M., Lovgreen-Nielsen, P., Volck, B., Sonne-Holm, S. & Lorenzen, I. (1998). Magn. Reson. Imaging, 16, 743–754. Web of Science CrossRef CAS PubMed Google Scholar
Pfeiffer, F., Bunk, O., David, C., Bech, M., Le Duc, G., Bravin, A. & Cloetens, P. (2007). Phys. Med. Biol. 52, 6923–6930. Web of Science CrossRef PubMed CAS Google Scholar
Pfeiffer, F., Weitkamp, T., Bunk, O. & David, C. (2006). Nat. Phys. 2, 258–261. Web of Science CrossRef CAS Google Scholar
Sambrook, P. (2000). J. Rheumatol. 27, 2541–2542. Web of Science PubMed CAS Google Scholar
Smith, J. B. & Haynes, M. K. (2002). Ann. Intern. Med. 136, 908–922. Web of Science CrossRef PubMed Google Scholar
Szkudlarek, M., Narvestad, E., Klarlund, M., Court-Payen, M., Thomsen, H. S. & Ostergaard, M. (2004). Arthritis Rheum. 50, 2103–2112. Web of Science CrossRef PubMed Google Scholar
Trentham, D. E., Townes, A. S. & Kang, A. H. (1977). J. Exp. Med. 146, 857–868. CrossRef CAS PubMed Web of Science Google Scholar
Wilkins, S. W., Gureyev, T. E., Gao, D., Pogany, A. & Stevenson, A. W. (1996). Nature (London), 384, 335–338. CrossRef CAS Web of Science Google Scholar
© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.