The authors have no conflict of interest.
Earliest Mineral and Matrix Changes in Force-Induced Musculoskeletal Disease as Revealed by Raman Microspectroscopic Imaging†
Article first published online: 15 DEC 2003
Copyright © 2004 ASBMR
Journal of Bone and Mineral Research
Volume 19, Issue 1, pages 64–71, January 2004
How to Cite
Tarnowski, C. P., Ignelzi, M. A., Wang, W., Taboas, J. M., Goldstein, S. A. and Morris, M. D. (2004), Earliest Mineral and Matrix Changes in Force-Induced Musculoskeletal Disease as Revealed by Raman Microspectroscopic Imaging. J Bone Miner Res, 19: 64–71. doi: 10.1359/jbmr.0301201
- Issue published online: 2 DEC 2009
- Article first published online: 15 DEC 2003
- Manuscript Accepted: 8 AUG 2003
- Manuscript Revised: 25 JUN 2003
- Manuscript Received: 2 DEC 2002
- mechanical force
Craniosynostosis, premature fusion of the skull bones at the sutures, is the second most common human birth defect in the skull. Raman microspectroscopy was used to examine the composition, relative amounts, and locations of the mineral and matrix produced in mouse skulls undergoing force-induced craniosynostosis. Raman imaging revealed decreased relative mineral content in skulls undergoing craniosynostosis compared with unloaded specimens.
Introduction: Raman microspectroscopy, a nondestructive vibrational spectroscopic technique, was used to examine the composition, relative amounts, and locations of the mineral and matrix produced in mouse skulls undergoing force-induced craniosynostosis. Craniosynostosis, premature fusion of the skull bones at the sutures, is the second most common birth defect in the face and skull. The calvaria, or flat bones that comprise the top of the skull, are most often affected, and craniosynostosis is a feature of over 100 human syndromes and conditions.
Materials and Methods: Raman images of the suture, the tips immediately adjacent to the suture (osteogenic fronts), and mature parietal bones of loaded and unloaded calvaria were acquired. Images were acquired at 2.6 × 2.6 μm spatial resolution and ranged in a field of view from 180 × 210 μm to 180 × 325 μm.
Results and Conclusions: This study found that osteogenic fronts subjected to uniaxial compression had decreased relative mineral content compared with unloaded osteogenic fronts, presumably because of new and incomplete mineral deposition. Increased matrix production in osteogenic fronts undergoing craniosynostosis was observed. Understanding how force affects the composition, relative amounts, and location of the mineral and matrix provides insight into musculoskeletal disease in general and craniosynostosis in particular. This is the first report in which Raman microspectroscopy was used to study musculoskeletal disease. These data show how Raman microspectroscopy can be used to study subtle changes that occur in disease.
Craniosynostosis, premature fusion of the calvarial bones at the sutures, often results in an abnormally shaped skull and has been recognized as a craniofacial anomaly for centuries. Craniosynostosis is the second most common birth defect in the face and skull. Craniosynostosis occurs in 1 in 2400 live births as a result of genetic and nongenetic causes.(1) Craniosynostosis occurs in all races and ethnic groups and presents most often as a sporadic anomaly, but can also present as a component of a syndrome. The flat bones that comprise the top of the skull, or calvaria, are most often affected, and craniosynostosis is a feature of over 100 human syndromes and conditions. This birth defect is a significant biomedical burden because surgery is the only option to separate the prematurely fused skull bones. Premature fusion of the skull bones at the sutures (the nonmineralized fibrous connective tissues that separate the calvarial bones) restricts the growth of the expanding brain and may result in increased intracranial pressure, exophthalmia, midface hypoplasia, and an abnormally shaped skull.(2)
Although five human genes have been associated with craniosynostosis,(3) the role of abnormal mechanical force in the etiology of craniosynostosis is less clear. In response to mechanical loads, sutures have exhibited increased bone volume and increased mineral apposition rate,(4–8) suture osteoprogenitor cell number,(9) suture fibroblast cell number,(10) alkaline phosphatase activity,(7,11) proline incorporation,(7,11) collagen matrix production,(8,12) and matrix degradation.(13) In addition, rat sutures that normally fuse failed to do so after the falx cerebri, or tracts that connect the cranial base to the calvarial suture, were severed, presumably through an alteration in the tensile forces transmitted by the cranial base.(14) A higher incidence of craniosynostosis occurred in mice pups after birth was delayed for 3 days, presumably the result of abnormally high intrauterine compression.(15) Altered masticatory force is also thought to play a role in craniosynostosis as osteopetrotic mice display fusion of the sagittal suture,(16) and rats on soft diets exhibited premature fusion of the internasal suture.(17) These studies suggest that altered mechanical force environments may affect calvarial bone patterning and/or development and may result in premature fusion.
Recently, we have shown that the controlled application of mechanical force leads to craniosynostosis within mouse calvaria in vitro. Cyclic uniaxial compression of calvarial explants led to increased osteoid production, increased bone marker gene expression, increased alkaline phosphatase activity, and fusion in loaded sutures and unloaded sutures that were cocultured with these loaded sutures (MA Ignelzi Jr, JM Taboas, and SA Goldstein, unpublished observations, 1999). These coculture results suggest that loaded sutures produce soluble factors and that these factors have the ability to initiate new bone formation in unloaded tissues. Fusion was observed in the sagittal suture, a suture that does not normally fuse, and was observed in discrete foci, consistent with the human disease. Suture fusion in this organ culture model was not the result of cyclic loading pushing the parietal bone fronts together. Rather, fusion was the result of new bone formation in the suture.
This organ culture model was used in conjunction with Raman microspectroscopy to elucidate the pathogenesis of force-induced craniosynostosis. Raman microspectroscopy was used to determine the composition, relative amounts, and locations of the tissue matrix and mineral components during suture fusion. Raman spectroscopy has been used to study a wide variety of tissues, and it has several inherent advantages over other techniques.(18,19) Specimen preparation is minimal, and tissues that have been fixed, processed, and embedded can be examined, and intact specimens that have not been fixed, processed, or embedded can also be examined. Raman spectroscopy can probe both thick and thin, as well as hydrated, specimens. The only preparation requirement is that the tissues be blood-free, because the heme-groups in the blood fluoresce under the near-infrared laser excitation required for these studies. Raman spectroscopy has excellent spatial resolution (up to 0.5 μm) and spectral resolution (1–2 cm−1). Because Raman spectroscopy is a nondestructive technique, the same specimen can be examined over time and by other methods after spectral acquisition.
Raman spectroscopy has been used in the study of both hard and soft tissues.(19) It has been used in the analysis of diseased soft tissues including atherosclerosis,(20–24) cervical cancer,(25,26) breast cancer,(27) oral dysplasia,(28) and gastrointestinal cancer.(29) However, Raman spectroscopy has not yet been used to examine compositional changes in bone tissue caused by disease.
Raman spectroscopy is particularly well suited to study bone because it probes the molecular and ionic vibrations of the mineral and matrix components.(18,19) In the mineral lattice, the phosphate (ν1, 945–964 cm−1 and ν4, 580 cm−1), carbonate (ν1, 1065–1070 cm−1), and monohydrogen phosphate (1010 cm−1) symmetric vibrations are the most intense vibrations observed. The collagen-dominated, proteinaceous matrix also produces several prominent vibrations (see Table 1).
We have used Raman spectroscopy previously to examine mouse calvaria from the onset of mineralization (fetal day 15.5, 4 days before birth) to 6 months postnatal.(30) Using Raman spectroscopy, we were able to detect the earliest mineralization (fetal day 15.5) and observed that the earliest mineral deposition results in the presence of a homogeneous, poorly carbonated apatite mineral environment. With maturation, there was variation in the composition of the mineral environments present throughout the tissue. Using traditional univariate techniques, we showed that the mineral:matrix ratio increases during the first 3 days of mineralization, remains constant, and increases between 14 days postnatal and 6 months, indicating rapid mineral deposition at the onset of mineralization with subsequent mineral and matrix production. We also demonstrated that the carbonate:phosphate ratio remains constant during late fetal and early postnatal life, but increases with age beginning at 3 days postnatal, suggesting increased ionic substitution with age. Using a multivariate analysis technique called factor analysis (FA), we also were able to monitor compositional changes in the mineral during the first weeks after birth. While a carbonated apatite was always the most prominent mineral environment found in the calvaria, we found increased heterogeneity in the mineral lattice composition postnatally, with the greatest amount of heterogeneity found during the 3 days after birth. Changes in the mineral composition are presumably a reflection of changes in the mineral deposition and remodeling processes.
FA, just one of many multivariate techniques used widely in the spectroscopy and imaging communities, is an extremely useful data analysis technique.(31–34) The greatest advantages that FA provides in spectroscopic imaging analysis are an inherent signal averaging effect and the ability to resolve spatially and spectrally distinct components. Fundamentally, FA of a dataset results in a set of factors and scores. Here, the factors resemble Raman spectroscopic signatures of either tissue or background components or noise in the system. Each factor has a corresponding score that reports the relative amount of that component at each location probed throughout the image. The benefit of using a FA is that if the relative amounts of two components vary spatially throughout the regions probed, the FA will resolve the components into two factors. For example, if the ratio of the mineral and the matrix components is different throughout an imaged area, the spectral signatures of the mineral component and the matrix component will be resolved and separated into two different factors. The relative amount of each component found in the imaged area will be reflected in its score image.
In this work, we have used Raman microspectroscopic imaging and FA to examine the mineral and matrix changes caused by the controlled application of mechanical force that leads to craniosynostosis. Understanding how mechanical signals are translated into distinct changes in the composition, relative amount, and location of the mineral and matrix components of bone will shed insight into the pathogenesis of a broad spectrum of musculoskeletal diseases.
MATERIALS AND METHODS
All experiments were approved by the University of Michigan Committee on the Use and Care of Animals. Calvaria were excised from postnatal 21-day B6CBA F1/J wildtype mice (Jackson Laboratories, Bar Harbor, ME, USA) and rinsed of debris using PBS. A 4 × 12 mm area was taken from the calvaria containing parts of the parietal bones and the sagittal suture (Fig. 1). The areas were cultured in serum-free media for either 4, 7, 10, or 14 days. For each time-point, six specimens were examined: three specimens were loaded in a uniaxial compression device and three specimens were not loaded. Loaded sutures were subjected to a 0.3g uniaxial compressive load for 30 minutes each day at a frequency of 1 Hz. The loaded and unloaded tissues were fixed in 100% ethanol for 30–60 minutes, soaked in 95% ethanol for 20 minutes, and transferred to 70% ethanol for storage at 4°C. Before the spectroscopic analysis, the specimens were transferred to 35% ethanol and soaked in PBS to avoid dehydration during spectral acquisition. The calvarial areas were cut perpendicular to the long axis of the suture into 1-mm sections, turned on their side, and placed on a quartz slide (Esco Products, Inc., Oak Ridge, NJ, USA) for Raman imaging.
Raman spectroscopic imaging
The method by which Raman spectra were obtained and the instrument design have been described previously.(35) Briefly, a 785-nm diode laser in a line configuration (50 mW at specimen, Invictus NIR Laser; Kaiser Optical Systems, Inc., Ann Arbor, MI, USA) was focused onto the specimens using a 10×/0.5 NA objective (Fluar Series; Zeiss, Thornwood, NY, USA) mounted on a BH-2 microscope frame (Olympus, Inc., Melville, NY, USA). In the epi-illumination configuration, the Raman scatter was collected by the same objective and focused into a NIR-optimized spectrograph (HoloSpec f/1.8I; Kaiser Optical Systems, Inc., Ann Arbor, MI, USA). The spectrograph dispersed the Raman scatter onto a thermoelectrically cooled, back-thinned, deep depletion camera (Andor Technology, Belfast, Ireland). A 25-μm entrance slit was used to set the spectral resolution of the system to 3–4 cm−1. Images were created line-by-line by acquiring transects (lines of point spectra) and moving the specimen under the focused laser line with a motorized translation stage (Danaher Precision Systems; New England Affiliated Technologies, Salem, NH, USA) at 1.3-μm increments to provide 2.6-μm spatial resolution.
For each specimen, we acquired one Raman image that encompassed the tips of the parietal bones, or osteogenic fronts (OFs), the suture mesenchyme, and mature bone tissue. All Raman images were acquired with 2.6 × 2.6 μm spatial resolution and ranged in a rectangular field of view from 180 × 210 μm to 180 × 325 μm.
All data analysis was performed using MatLab 5.3 (MathWorks, Inc., Natick, MA, USA) using vendor-supplied and custom written scripts. All spectra were preprocessed by removing detector-generated artifacts (spikes) and subtracting the detector dark current.
An FA was performed on the raw data; the covariance matrix (the data set matrix multiplied by its transpose) of the image spectra dataset was calculated,(34) and a principal component analysis was performed, enhancing any small changes in the Raman spectra throughout the dataset. No pure component spectra or training sets were used in the analysis. The principal component analysis resulted in a set of eigenvectors that describes the original image dataset. Most of these eigenvectors describe noise in the system; these noise-describing eigenvectors are discarded. The rest of the non-noise eigenvectors are manually rotated resulting in “factors.” These factors describe the original spectral signatures of the various tissue components. Non-negativity and band shapes were used as rotational constraints.(36)
To gather information regarding changes in the amounts of the mineral and matrix found throughout a specimen, a second FA for each image was performed. For each image, the original dataset was truncated into spectral subregions representative of the mineral and matrix. The phosphate ν1 envelope (900–995 cm−1, mineral metric) and the amide I envelope (1500–1800 cm−1, matrix metric) of each spectrum of an image were joined together (concatenated). The rest of the spectrum was discarded to reduce variance that did not contain the desired spectroscopic information. An FA as described above was performed on the new concatenated spectra dataset. In 20 of 24 specimens examined, the mineral and matrix could be resolved into two factors. Comparisons of the amounts of the mineral found throughout the tissue were made for these specimens; similar comparisons were also made for the matrix components. Using the mineral score image as a guide, a mature bone area (away from the OF), an OF area, and a suture area were defined for each specimen. The defined areas all were 507 μm2; the shape of the area depended on the morphology of each specimen, but typically was either a 19.5 × 26 μm or a 13 × 39 μm area. The mean mineral and the mean matrix values and SDs in each area were calculated.
After Raman imaging and data analysis, selected specimens were fixed, decalcified, embedded, sectioned at 7-μm intervals, and stained with hemotoxylin and eosin.
Separate FAs of the loaded and unloaded specimens revealed that the only resolvable mineral factor was a carbonated apatite (Fig. 2A; PO43−ν1, 959 cm−1; PO43−ν4, 580 cm−1; CO32−ν1, 1072 cm−1), and the only resolvable matrix factor was a collagen-dominated protein (Fig. 2B; amide I, 1662 cm−1; amide III, 1242 and 1269 cm−1; CH2 wag, 1446 cm−1; hydroxyproline, 853 and 875 cm−1; proline, 919 cm−1; C-C backbone stretch, 936 cm−1; phenylalanine ring-breathing mode, 1000 cm−1). A raw spectrum from an image is shown in Fig. 2C (band assignments are the same as for Figs. 2A and 2B).
The relative amount and location of the mineral and matrix components in the loaded specimens differed from the unloaded specimens. The score images presented in Fig. 3 represent the relative amount—red corresponds to high; blue corresponds to low—and the location of either the mineral or the matrix. In the mineral Raman score image from a 4-day loaded specimen, one of the OFs has a decreased amount of mineral (seen as light green in Fig. 3A, boxed area) compared with the more mature tissue farther away from the OFs (viewed as yellow, orange, and red in Fig. 3A). This same finding was observed for 10 of 12 loaded specimens in one or both OFs. The matrix Raman score image shows that an increased amount of collagen-dominated protein was found in the loaded specimen in areas where an OF was in close proximity to the opposing parietal bone (Fig. 3B, arrows). The mineral Raman score image shows that there is little mineral in the same areas as the increased protein signal (Fig. 3A, arrows). To compare the locations of the mineral and matrix, an RGB image was constructed by overlaying the mineral score image (blue) with the matrix score image (green). Areas of mineral and matrix overlap are seen as light blue-green. The loaded specimen RGB image in Fig. 3C shows that the matrix extends beyond the mineral and is increased at the OFs (arrows). A hematoxylin and eosin-stained section from the same specimen (acquired after Raman imaging) is shown in Fig. 3D. Increased hematoxylin staining at each bone OF (arrows) reveals increased cell density. This increased cell density at the OFs is not found in unloaded specimens. The increased amount of cells and collagen found in the loaded specimen are also seen in the Raman matrix score image (Fig. 3B, arrows). This increase in cell density and collagen concentration coupled with mineralization pattern is indicative of osteoid production.
A shorter image of the same specimen shown in Figs. 3A–3D (180 × 325 μm in length) is shown in Figs. 3E–3G (180 × 230 μm); this image was acquired to present the mineral and matrix content at the OF (marked with an arrowhead) on a smaller relative scale to enhance detail. In the shorter Raman mineral score image (Fig. 3E), decreased amounts of mineral were more clearly seen in the loaded specimen OF compared with the more mature tissue farther away from the OF, suggesting new growth at the OF. Increased matrix production is seen just beyond the lightly mineralized OF in the Raman matrix score image (Fig. 3F, right of arrowhead). The corresponding RGB image in Fig. 3G again reveals matrix production extending past the OF (areas of green in Fig. 3G).
FA of the individual images of both the loaded and unloaded specimens showed that the mineral and matrix were not colocalized in most specimens and that there were common areas of increased matrix presence. The FAs of 20 of 24 specimen images resulted in the separation of the mineral factor and the matrix factor, revealing that the mineral and matrix were dispersed in different ratios throughout the imaged area. In all of the specimens that had separated mineral and matrix factors, the Raman matrix score image showed that there was an increased amount of protein along the surfaces of the calvaria, particularly true for the endocranial (or dural) surface of the calvaria (Fig. 3I, asterisks). This increase in protein along the endocranial surface in a 4-day unloaded specimen is also seen in the RGB image in Fig. 3J. The endocranial surface is distinctly blue-green, and the matrix (green) extends only slightly beyond the bone front. In the unloaded specimens, mineral content was evenly distributed throughout the tissue, especially at the OFs (Fig. 3H, arrowheads, note red and yellow).
To quantitate differences in the amount of mineral at the OF compared with the mature parietal bone between the loaded and unloaded specimens, we calculated the ratio of the mineral in the OF area to the mature bone area for loaded and unloaded specimens. The mean OF to mature bone mineral ratio for the loaded specimens was 0.6407 ± 0.1000 compared with 0.9892 ± 0.1166 for the unloaded specimens (Fig. 4). The decreased mineral content at the OF of the loaded specimens is presumably a result of the new growth and incomplete mineral deposition at the OF across the suture as the suture was undergoing fusion. The loaded and unloaded specimen ratios were found to be statistically different at the 99.5% confidence level.
The images acquired during these experiments are the first spectroscopic images of murine craniosynostosis and provide insight into the composition, relative amounts, and location of the mineral and matrix in areas of early suture mineralization. We found that in 20 of 24 specimens, the mineral and matrix tissue components were resolved into two factors. The failure to separate the mineral and matrix components in four of the specimens could be because of the fact that the mineral and matrix were found in the same ratio throughout the tissue. However, it is more likely that the inability to resolve the mineral and the matrix was caused by either increased background fluorescence or poor signal-to-noise in the spectra.
The one mineral environment found in all of the calvaria specimens was a carbonated apatite, similar to that found in other bone tissue. In the 10 of 12 loaded specimens that had resolvable mineral and matrix factors, we observed decreased relative mineral content at one or both OFs compared with control specimens. The OF to mature bone mineral ratio was significantly decreased for the loaded tissue (mean ratio = 0.6407 ± 0.1000) compared with the unloaded tissue (mean ratio = 0.9892 ± 0.1166). The decreased amount of mineral in the OFs after loading suggests that there is new mineral deposition at the bone OFs caused by the uniaxial compression. The 0.9892 OF to mature bone mineral ratio in the unloaded specimens reveals that there were almost equal amounts of mineral in both the OFs and the mature bone. The observation of new mineral deposition at the OFs after uniaxial compressive loading supports previous studies that have shown increased bone production in tissue experiencing increased stress.(37) This is the first study to report new areas of mineral presence after the controlled application of mechanical force.
We also observed an increase in the protein content in areas where a loaded specimen OF was in the process of fusing with the opposing parietal bone. However, the ratio of the suture matrix content to the mature bone matrix content is not statistically different for the loaded versus the unloaded specimens. This lack of difference is most probably because of a concomitant increase in matrix content in both the mature bone and the suture for the loaded specimens, resulting in comparable matrix ratios for the two sets of specimens. The observed protein increase beyond the loaded specimen OFs is caused by an increase in the amount of collagen and cell density, an observation consistent with other craniosynostosis studies. Previous studies have found that in calvaria undergoing craniosynostosis, there is cell proliferation at the OFs,(38,39) cell differentiation in the suture,(40,41) and an increase in collagen synthesis.(38) We have shown increased alkaline phosphatase activity, Runx2 expression, and osteocalcin expression in loaded sutures (MA Ignelzi Jr, W Wang, JM Toboas, and SA Goldstein, unpublished observations, 2000).
The increased amount of collagen-dominated protein along the ectocranial and endocranial surfaces of both the loaded and unloaded tissue specimens indicates an increased amount of cells and fibrous tissue lining the calvaria. This is especially noticeable in Fig. 3J, where the light blue-green is prominent along the calvaria endocranial surface. The larger increase in new matrix on the endocranial surface is likely caused by the dura mater, the outermost meningeal membrane of the brain, which lies immediately adjacent to the endocranial surface of the calvaria. The dura mater is thought to secrete factors that instruct the skull to expand to accommodate the growing brain.(38,42) The increased cellular activity found in the dura mater would result in increased matrix content on the endocranial surface in the Raman score image.
In these studies, the calvaria specimens were cut to 1-mm-thick sections, and as a result, only four sections per specimen were available for spectroscopic analysis. None of the specimens showed evidence of complete suture fusion; however, this is not unexpected because suture fusion occurs as discrete foci along the entire length of the suture.(1,42) We also observed just one mineral environment in all of the specimens, rather than a heterogeneous mineral composition. The absence of a heterogeneous mineral composition could be a result of decreased mineral intensity caused by decreased mineral presence or decreased power density, but is most likely the result of homogeneous mineral composition within the spatial resolution of our images.
This is the first study to use Raman imaging to examine a mouse model of a human musculoskeletal disease. These studies provide insight into the composition and relationship of the mineral and matrix at bone OFs in both loaded and unloaded sutures. The earliest pathological changes that accompany craniosynostosis—incomplete mineralization and increased matrix production at the OFs of loaded specimens—were observed using Raman imaging. Raman microspectroscopy's ability to provide both qualification of the tissue composition and quantitative results regarding the location and relative amount of mineral and matrix components in the tissue will likely elucidate the pathogenesis of a wide variety of bone diseases.
CPT thanks Dr Jeremy Shaver (Eigenvector Research, Inc.) for helpful discussions on data analysis and Dr Jerilyn Timlin (Sandia National Laboratories) for assistance with the RGB contrast images. This work was supported in part by National Institutes of Health Grants R29 DE 11530 (MAI), P60 AR 20557 (MAI and SAG), and P30 AR 46024 (SAG, MAI, and MDM).
- 12000 Craniosynostosis: Diagnosis, Evaluation, and Management. Oxford University Press, New York, NY, USA.
- 21993 Sutural biology and the correlates of craniosynostosis. Am J Med Genet 47:581–616.
- 31999 Craniosynostosis syndromes: From genes to premature fusion of skull bones. Mol Genet Metab 68:139–151.,
- 41999 Implant-borne suture expansion in rabbits: A histomorphometric study of the supporting bone. J Biomed Mater Res 45:1–10., , , ,
- 51997 Sutural expansion using rigidly integrated endosseous implants: An experimental study in rabbits. Angle Orthod 67:283–290., , , ,
- 61995 Frontonasal suture expansion in the rabbit using titanium screws. J Oral Maxillofac Surg 53:1033–1043., ,
- 71987 The morphologic and biochemical effects of tensile force application to the interparietal suture of the Sprague-Dawley rat. Am J Orthod Dentofacial Orthop 92:123–133.,
- 81996 Effects of expansive force on the differentiation of midpalatal suture cartilage in rats. Bone 18:341–348., , , , , ,
- 91992 Force magnitude effects upon osteoprogenitor cells during premaxillary expansion in rats. Angle Orthod 62:197–202.,
- 101990 The effect of duration and magnitude of tensile mechanical forces on sutural tissue in vivo. Eur J Orthod 12:330–339., ,
- 111988 The effects of force magnitude on a sutural model: A quantitative approach. Am J Orthod Dentofacial Orthop 93:460–466.,
- 121989 Effect of force level on synthesis of type III and type I collagen in mouse interparietal suture. J Dent Res 68:1746–1751., ,
- 131990 Immunolocalization of collagenase and tissue inhibitor of metalloproteinases (TIMP) in mechanically deformed fibrous joints. Am J Orthod Dentofacial Orthop 97:281–288., , , ,
- 141959 The pathogenesis of premature cranial synostosis in man. Acta Anat (Basel) 37:351–370.
- 151989 Sutures and intrauterine deformation. In: PersingJA, EdgertonMT, JaneJA (eds.) Scientific Foundations and Surgical Treatment of Craniosynostosis. Williams & Wilkins, Baltimore, MD, USA, pp. 96–106.,
- 161998 Midpalatal suture of osteopetrotic (op/op) mice exhibits immature fusion. Exp Anim 47:277–281., , , , , ,
- 171986 Masticatory muscle function and craniofacial morphology. An experimental study in the growing rat fed a soft diet. Swed Dent J Suppl 36:1–55.
- 181995 Infrared and Raman Spectroscopy: Methods and Applications. VCH Publishers, Inc., New York, NY, USA.
- 192000 Application of vibrational spectroscopy to the study of mineralized tissues. J Biomed Opt 5:259–268.,
- 201999 Histological classification of Raman spectra of human coronary artery atherosclerosis using principal component analysis. Appl Spectrosc 53:938–942., , , , ,
- 211992 Quantitative histochemical analysis of human artery using Raman spectroscopy. J Photochem Photobiol B 16:211–233., , ,
- 221992 In situ optical histochemistry of human artery using near infrared Fourier transform Raman spectroscopy. Proc Natl Acad Sci USA 89:3473–3477., ,
- 231998 Raman spectroscopy for quantifying cholesterol in intact coronary artery wall. Atherosclerosis 141:117–124., , , , , , , ,
- 242000 In vivo determination of the molecular composition of artery wall by intravascular Raman spectroscopy. Anal Chem 72:3771–3775., , , , , , , ,
- 251998 Near-infrared Raman spectroscopy for in vitro detection of cervical precancers. Photochem Photobiol 68:123–132., , , , , ,
- 262001 Near-infrared Raman spectroscopy for in vivo detection of cervical precancers. Appl Spectrosc 55:955–959., , , , ,
- 271997 Raman chemical imaging of breast tissue. J Raman Spectrosc 28:119–124.,
- 282000 In vivo detection of dyplastic tissue by Raman spectroscopy. Anal Chem 72:6010–6018., , , , , , ,
- 292000 In vivo near-infrared Raman spectroscopy: Demonstration of feasibility during clinical gastrointestinal endoscopy. Photochem Photobiol 72:146–150., , ,
- 302002 Mineralization of developing mouse calvaria as revealed by Raman microspectroscopy. J Bone Miner Res 17:1118–1126., ,
- 311996 Multivariate Image Analysis. John Wiley & Sons Ltd, West Sussex, UK.,
- 321996 Applied Factor Analysis in the Natural Sciences. Cambridge University Press, Cambridge, UK.,
- 331995 Chemometrics in Analytical Spectroscopy. The Royal Society of Chemistry, Cambridge, UK.
- 341999 Chemometrics: Statistics and Computer Application in Analytical Chemistry. Wiley-VCH, New York, NY, USA.
- 352000 Raman spectroscopic imaging markers for fatigue-related microdamage in bovine bone. Anal Chem 72:2229–2236., , , ,
- 361998 Hyperspectral Raman line imaging on an aluminosilicate glass. Appl Spectrosc 52:64–69., ,
- 371998 Is all bone the same? Distinctive distributions and properties of non-collagenous matrix proteins in lamellar vs. woven bone imply the existence of different underlying osteogenic mechanisms. Crit Rev Oral Biol Med 9:201–223.
- 381998 Dura mater maintains rat cranial sutures in vitro by regulating suture cell proliferation and collagen production. J Craniofac Genet Dev Biol 18:150–158., , , ,
- 391999 Msx2 gene dosage influences the number of proliferative osteogenic cells in growth centers of the developing murine skull: A possible mechanism for MSX2-mediated craniosynostosis in humans. Dev Biol 295:260–274., , , , , , , ,
- 402003 Fibroblast growth factors lead to increased Msx2 expression and fusion in calvarial sutures. J Bone Miner Res 18:751–759., ,
- 411996 Increased bone formation and osteoblastic cell phenotype in premature cranial suture ossification (craniosynostosis). J Bone Miner Res 11:401–407., , ,
- 421995 Cranial sutures require tissue interactions with dura-mater to resist osseous obliteration in vitro. J Bone Miner Res 10:1978–1987., , , ,