- Email: [email protected]
Three-dimensional reconstruction of chick calvarial osteocytes and their cell processes using confocal microscopy
Bone 36 (2005) 877 – 883 www.elsevier.com/locate/bone
Three-dimensional reconstruction of chick calvarial osteocytes and their cell processes using confocal microscopy Yasuyo Sugawaraa, Hiroshi Kamiokaa, Tadashi Honjoa, Ken-ichi Tezukab, Teruko Takano-Yamamotoa,* a
Department of Orthodontics and Dentofacial Orthopedics, Graduate School of Medicine and Dentistry, Okayama University, 2-5-1 Shikata, Okayama City, Okayama 700-8525, Japan b Department of Tissue Organ Development, Graduate School of Medicine, Gifu University, Tsukasa-machi 40, Gifu City, Gifu 500-8705, Japan Received 4 June 2004; revised 10 September 2004; accepted 8 October 2004
Abstract Osteocytes are surrounded by hard bone matrix. Therefore, it has not previously been possible to demonstrate the real architecture of the osteocyte network in bone. We previously reported that it is possible to observe osteocytes in bone by labeling the cells with fluorescence and using confocal laser scanning (CLS) microscopy. In this study, we for the first time conducted an extensive analysis of the morphology and morphometry of the three-dimensional (3D) osteocyte structure using three-dimensionally reconstructed fluorescent images. Sixteen-day-old embryonic chick calvariae were stained with fluorescently labeled phalloidin and observed using a confocal laser scanning microscope. Morphometry of osteocytes in the calvaria was analyzed using extensive three-dimensional reconstructing software IMARIS, process length measuring software NEURON TRACER and cell surface area-/cell volume-analyzing software SURPASS. From the IMARIS-derived images, we found that the average of 10 osteocytes is 52.7 F 5.7 processes, and the point-to-point distance between centers of the osteocytes was 24.1 F 2.8 Am. In addition, we could calculate that each osteocyte spans an average of 4180 F 673 Am3 of bone volume. NEURON TRACER showed that the length of osteocyte processes was 0.26 F 0.02 Am per 1 Am3 bone compartment. In addition, SURPASS indicated that the surface area of osteocytes was 0.36 F 0.03 Am2 per 1 Am3 bone compartment and that the volume ratio of osteocyte cell body to bone compartment was 9.42% F 1.18%. Together, the average total length of the processes, the average surface area, and the average volume of one osteocyte were 1070 F 145 Am, 1509 F 113 Am2, and 394 F 49 Am3, respectively. It is possible to reconstruct the real architecture of the osteocyte network and obtain morphometric data from fluorescently labeled osteocytes in chick calvaria. D 2004 Elsevier Inc. All rights reserved. Keywords: Osteocytes; Bone; Three-dimensional morphometry; Actin; Confocal laser scanning microscopy
Introduction Osteocytes are the most numerous cells in bone. They represent the final differentiation step in the osteoblastic lineage and arise when bone-forming osteoblasts become encased within the calcified matrix [15]. Osteoblasts lose a
* Corresponding author. Fax: +81 86 235 6694. E-mail addresses: [email protected] (Y. Sugawara)8 [email protected] (H. Kamioka)8 [email protected] (T. Honjo)8 [email protected] (K. Tezuka), [email protected] (T. Takano-Yamamoto). 8756-3282/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2004.10.008
large part of their cell organelles but gain long, slender cell processes by which the cells remain in contact with earlier incorporated osteocytes and with osteoblasts lining the bone surface [1]. The extended network of osteocyte processes joined at gap junctions [3,4]. Gap junctions are known to be involved in intercellular communication. In vitro, osteocytic and osteoblastic cells are functionally coupled to one another via gap junctions as shown by the ability of calcein to pass between cells and the ability of cells to communicate a mechanically induced calcium response [20]. The cellular network in vivo might play an important role for bone remodeling in response to mechanical loading [10,18]. Therefore, it is significant to
878
Y. Sugawara et al. / Bone 36 (2005) 877–883
analyze the whole figure of the osteocyte network and to obtain morphometric data. However, osteocytes are surrounded by hard bone matrix; hence it has been impossible to visualize the three-dimensional (3D) osteocyte network and to obtain morphometric data. To date, most previous studies on the 3D structure of osteocytes have been done with transmission electron microscope (TEM) serial sections, where the tissue is cut into 70- to 80-nm-thin slices [13,14]. Although the morphometrical data from TEM are very reliable because the margins of the cell surface are clearly distinguished by the dense uniform line of cell membrane, it is not suitable to the follow extended complex run of osteocyte processes. Therefore, morphometric analysis of osteocytes including processes has not previously been performed. Recently, we found that it was possible to observe the osteocyte network in chick calvaria by labeling the cells with fluorescence and using confocal laser scanning (CLS) microscopy [6]. As chick calvariae were very thin and flat, we could stain and visualize clearly. In the present study, we applied this technique for making a 3D-reconstructed fluorescent model of the osteocyte network. Observation of the osteocyte network in chick calvaria as well as morphometric analysis of the osteocyte network was performed using a 3D-reconstructed fluorescent model. IMARIS software, which reconstructed extensive 3D images [2,19], was employed to visualize the architecture of osteocyte network as a shadow projection and for counting the number of osteocyte processes, measuring the distance between osteocytes and calculating the volume of bone compartment occupied by a single osteocyte. NEURON TRACER software, which automatically traced the fluorescence images [11,16], was used to measure the length of osteocyte processes. In addition, SURPASS software was used to quantify the surface area and volume of osteocyte including osteocyte processes. As is well known, it is difficult to determine where one osteocyte process ends and contacts with another osteocyte. To report morphometric data for a single osteocyte, we analyzed all data in bone compartment units of 1 Am3. As a result, we could for the first time visualize a computer-reconstructed osteocyte network and obtain 3D morphometric data for a single osteocyte in bone including the full length of osteocyte processes.
Materials and methods Preparation of bone fragments Calvariae were obtained from 16-day-old embryonic chickens and washed with PHEM (60 mmol/l piperazineN, NV-bis [2-ethane-sulfonic acid], 25 mmol/l N-[2-hydroxyetyl] piperazine-NV-[2-ethanesulfonic acid], 10 mmol/l ethylene glycol-bis [2-amino-ethyl ether]-N, N, NV, NV-tetra-
acetic acid, 2 mmol/l magnesium chloride, pH 6.9). After stripping off the periosteum, the calvariae were trimmed into 3 3-mm pieces for further use. The average thickness of the sample ranged from 60 to 80 Am. Fluorescence staining and confocal laser scanning images Calvarial fragments were fixed with 3% paraformaldehyde in PHEM overnight, decalcified with 5% EDTA overnight, then permeabilized by incubation in 0.3% Triton X-100 in phosphate-buffered saline (PBS) for 10 min. The fragments were rinsed and stained for 2 days at 48C with a 1:200 dilution of Texas red-X-conjugated phalloidin (excitation wavelength = 595 nm; emission wavelength = 615 nm, Molecular Probes Inc., Eugene, OR) in PBS containing 1% BSA. After rinsing with PBS, the samples were embedded in fluorescence mounting medium (Dako, Carpinteria, CA) containing 1 mg/ml p-phenylenediamine dihydrochloride (Sigma, St. Louis, MO), then viewed immediately. Confocal optical sectioning was performed with an LSM510 confocal laser scanning (CLS) microscopy system (Carl Zeiss, Oberkochen, Germany) coupled to an upright microscopy (Zeiss) with PlanFluor objective (63, N.A. = 1.4) at Central Research Laboratory, Okayama University Medical School. The refraction index of immersion media (Zeiss 518F) was 1.518. Theoretical xy- and z-axes resolutions were 0.263 and 0.604 Am, respectively. Frame size of the image was 146.2 146.2 Am with an 8-bit color depth. Confocal images were taken with 0.3-Am step size and were processed four times with Kalman averaging. After obtaining confocal images, the images were processed for noise reduction with Huggens software (Bitplane AG, Zurich, Switzerland). We have showed previously study that the distribution of osteocyte processes coincided with that of canaliculi in the depth of 20 Am from osteoblast layer [6]. In this study, we focused on the osteocytes in osteoid layer (only 0 to 7.2 Am in depth). Confocal optical sectioning was initiated from the bone side of the osteoblast layer. 3D reconstruction of the osteocyte network by IMARIS The 3D structure of the osteocyte network was reconstructed from CLS images using IMARIS software (Bitplane). CLS images from 1.8 to 7.2 Am in depth were reconstructed for visualizing the osteocyte network. Computer analysis was performed by an Octane Workstation (Silicon Graphics, Inc.). The reconstructed images were digitally processed by Adobe Photoshop Version 5.5 software (Adobe Systems Inc., Mountain View, CA). Counting the number of processes that radiate from one osteocyte was performed by rotating the 3D-reconstructed fluorescent model. In addition, the volume of bone compartment occupied by one osteocyte was calculated by dividing the volume of bone compartment by the
Y. Sugawara et al. / Bone 36 (2005) 877–883
number of osteocytes in the compartment. IMARIS allows us to recognize the center of mass in each osteocyte. Thus, point-to-point distances between osteocytes were calculated. Quantitative analysis of length of osteocyte processes by NEURON TRACER NEURON TRACER (Bitplane AG), originally developed to analyze neurite morphology, was especially designed to automatically map areas of fluorescence on confocal images and construct 3D dendritic models to evaluate the total length of the model. We employed this software to obtain the topology of dendritic models of the osteocyte network and to determine the total length of osteocyte processes. We quantified 10 different regions, which contain several osteocytes, to analyze the length of the processes. The accuracy of the software was within 3%. To report the length of osteocyte processes, we divided the total length of osteocyte processes obtained using NEURON TRACER with the volume of bone compartment that was analyzed to measure the length of osteocyte processes by the software. Therefore, the length of osteocyte processes was reported by the length per 1 Am3 of bone compartment. Quantitative analysis of surface area and volume of osteocytes by SURPASS Surface area and volume of osteocytes per 1 Am3 of bone compartment were determined from SURPASS software. Eight-bit IMARIS images were converted to binary images to make 3D wire frame models. First, we determined the threshold for the binary models of osteocytes by superimposing binary images to IMARIS images until the images overlapped. To separate the cell body and cell processes, the length of wire frame was regulated by manual object segmentation in the SURPASS software. The average bone volume that was analyzed for surface area and volume of osteocytes was 22905 F 4026 Am3. This region contained five to eight osteocytes. Surface area and volume were automatically analyzed by SURPASS. To test the accuracy of the software automation, we confirmed control errors between red-fluorescent Fluosphere beads (2.0, 1.0, and 0.5 Am in diameter; Molecular Probes Inc.) and acquired fluorescent model in SURPASS software. Calculation of morphometrical data for a single osteocyte The average total length of the processes, the average surface area, and the average volume per osteocyte were determined by calculating the bone compartment volume occupied by one osteocyte, multiplying the length of processes, the surface area, and volume per unit volume of bone, respectively.
879
Result Observation of the osteocyte network by using IMARIS Confocal optical sectioning was performed from the bone side of osteoblast surface to bone compartment. Based on our previous study, the layer chosen was the osteoid layer. Fig. 1 shows the serial confocal images 0 Am (A), 1.8 Am (B), and 7.2 Am (C) in depth from the bone side of the osteoblast layer. Fig. 1A shows both osteocytes and osteoblasts because the surface of the calvaria showed slight undulations. Osteoblasts were seen as strong flat signals (asterisk) because the intensity of fluorescence in osteoblasts was much higher than that in osteocytes. In addition, halfembedded osteocytes were observed, as indicated by the arrowheads. At the depth of 1.8 Am, osteoblasts were mostly excluded and the observed cells were mainly osteocytes (Fig. 1B). Strong fluorescent signals in the processes were still recognizable even at 7.2 Am in depth (Fig. 1C). Fig. 2 shows IMARIS imaging of the osteocyte network reconstructed from confocal images from 1.8 to 7.2 Am in depth obtained with a 0.3-Am step size. Fig. 2A shows the view from the periosteal side, and Fig. 2B, that from the bone side at the same location. The network of osteocytes was visualized as a shadow projection. Osteocytes clearly appeared as spindle-shaped cells with numerous processes (arrows). Osteocytes in calvaria were regularly distributed in the same direction. The point-to-point distances between centers of osteocytes were measured. As a result, the distance from one cell to the adjacent cell was 24.1 F 2.8 Am (n = 10). Interestingly, we could recognize osteocytes that were covered by osteoblasts (asterisk in Fig. 2B). Although osteocyte processes and their branches were intersected with each other, we were able to trace the processes to the last detail (Figs. 2C, D). Then, the number of osteocyte processes were counted by IMARIS. The average number of processes radiating from single osteocyte was 52.7 F 5.7 (n = 10). In addition, we could calculate a single osteocyte occupied 4180 F 673 Am3 of bone compartment by dividing the volume of bone compartment by the number of osteocytes in the compartment. Topology of the dendritic tree and determination of the length of osteocyte processes by NEURON TRACER NEURON TRACER software was developed to form dendritic models of neurons and to measure the length of neuronal dendrites [11,16]. To obtain the length of osteocyte processes, we used this software. Fig. 3 shows the process of making the topological dendritic model (yellow line) from fluorescence-positive sites (white region). From the dendritic model, NEURON TRACER automatically measured the total length of osteocyte processes. It was difficult to determine where one osteocyte process ended. Therefore, we analyzed the length of
880
Y. Sugawara et al. / Bone 36 (2005) 877–883
Fig. 1. (A–C) Serial histotomography taken by CLS microscopy. Fluorescent images showed bone cells stained with Texas red-X phalloidin (red). Images were 0 Am (A), 1.8 Am (B), and 7.2 Am (C) in depth from the bone side of the osteoblast layer. Arrowheads in A show half-embedded osteocytes, and asterisks in A show osteoblast layer. Bar = 20 Am.
osteocyte processes in 1 Am3 bone compartment. The length of osteocyte processes per 1 Am3 bone compartment was 0.26 F 0.02 Am (n = 10). Morphometrical analysis of osteocyte in unit bone matrix by SURPASS The original IMARIS image in Fig. 4A was converted to a binary image using SURPASS software (Fig. 4B). We confirmed that the binary model maintained most of the osteocyte processes by comparison with those in the original IMARIS image (compare Figs. 4A and B).
Before calculating the volume and surface area of osteocytes, we measured 10 red-fluorescent beads (2.0, 1.0, and 0.5 Am diameter) as a test phantom to evaluate the accuracy of SURPASS software. The maximal diameters of the x-, y-, and z-axes in fluorescent beads were detected. The accuracy of our estimates of beads diameter is summarized in Table 1. Based on the calibration, a correction factor was derived for each axis. The residual of 2.0-Am diameter beads was 21% in the x and y, 97% in the z-axis, that of 1.0-Am diameter beads was 45% in the x and y, 103% in the z-axis, and that of 0.5-Am diameter beads was 71% in the x and y, 95% in the z-axis. Therefore, cross-
Fig. 2. Three-dimensional reconstruction of osteocyte network by IMARIS software. We used CLS images from 1.8 Am (Fig. 1B) to 7.2 Am (Fig. 1C) for the reconstruction. Left column of photos (A and C) shows the appearance from the periosteal side, and the right column (B and D), that from the bone side at the same location. C and D are magnified images of the rectangular areas in A and B. Arrows in A show osteocyte. Asterisks in B show osteocytes covered by osteoblasts. Bar in A = 20 Am, bar in C = 10 Am.
Y. Sugawara et al. / Bone 36 (2005) 877–883
881
Fig. 3. Image of a dendritic tree made by NEURON TRACER software. Fluorescence-positive sites (white region) showing the histotomography of osteocytes obtained by CLS microscopy. The yellow line shows the topology of the dendritic tree of osteocytes made by NEURON TRACER. (A–C) View from the vascular side, (D–F) sagittal views of A to C. Bar = 10 Am.
sections of a circle were thought to be distorted to an ellipse with elongation of the x-, y-, and z-axis in SURPASS data. Based on these results, we calculated the approximate volume and surface area correction as follows: Volume correction ¼ area of circle=area of ellipse ¼ pr 2 =pab Surface area correction ¼ circumference of circle=circumference of ellipse h pffiffiffiffiffii 2pr=p 1:5ða þ bÞ ab r: radius of fluorescent beads a: short axis of ellipse as x- or y-axis b: long axis of ellipse as z-axis As a result, the volume corrections for 0.5-, 1.0-, and 2.0Am beads were 0.30, 0.34, 0.41, respectively. The surface area corrections for 0.5-, 1.0-, and 2.0-Am beads were 0.54, 0.57, and 0.62, respectively. Most of the osteocyte-process diameters were less than 0.5 Am, while the short axis of the osteocyte cell body was approximately 2–5 Am as shown in
our previous study [17]. Therefore, we decided to use different size beads for calibration in order to determine correction factors to be used to accurately determine length and volume of processes and cell body: that obtained from 0.5-Am beads was used for process correction and that from 2.0-Am beads for cell body correction. Next, processes having less than 0.5-Am diameters were excluded by manual object segmentation (Fig. 4C). After this procedure, we separated the osteocyte cell body and the processes. Then the volume and surface area of the osteocyte cell body were calculated using corrections obtained from 2.0-Am beads. Then, the volumes and surface areas of osteocyte cell processes were calculated using corrections obtained from 0.5-Am beads. The average surface area of osteocytes per 1 Am3 of bone compartment was 0.36 F 0.03 Am2 (n = 10), and the volume ratio of osteocytes to bone compartment was 9.42% F 1.18% (n = 10). The percentage processes volume was 35%, and the percentage cell body excluding processes volume was 65%. On the other hand, the percentage processes surface area was 52%, and the percentage cell body excluding processes surface area was 48%.
Fig. 4. Images of binary model of osteocytes in calvaria made by SURPASS software. (A) Three-dimensional reconstruction of osteocyte network by IMARIS software. (B) A binary image of osteocytes using pseudocolors (green), which was processed from image A. (C) A binary image of osteocytes whose processes were excluded. Bar = 10 Am.
882
Y. Sugawara et al. / Bone 36 (2005) 877–883
Table 1 Summary of the evaluation of the effective resolution of fluorescent beads Beads diameter (accurate value) (Am) 2.0 1.0 0.5
Beads diameter from red-fluorescent beads (Am) [magnification (%)] x
y
z
2.42 F 0.17 [121 F 8.5] 1.45 F 0.12 [145 F 12.0] 0.85 F 0.05 [170 F 10.0]
2.42 F 0.17 [121 F 8.5] 1.45 F 0.12 [145 F 12.0] 0.85 F 0.05 [170 F 10.0]
3.90 F 0.20 [195 F 10.0] 2.03 F 0.20 [203 F 20.0] 0.98 F 0.04 [196 F 8.0]
Total length of processes, cell surface area, and volume of one osteocyte As described before, we could calculate that each osteocyte spans an average of 4180 F 673 Am3 of bone volume. Therefore, the morphometric data per unit bone compartment were multiplied by the mean volume of bone compartment occupied by each osteocyte (4180 Am3) to obtain morphometric data per single osteocyte. As a result, the average values for one osteocyte were calculated as follows: cell processes, 1070 F 145 Am; surface area 1509 F 113 Am2 (cellular body 724 Am2; cytoplasmic processes 785 Am2); and cell volume, 394 F 49 Am3 (cellular body 257 Am3; cytoplasmic processes 137 Am3) (Table 2).
Discussion Ejiri and Ozawa [5] first developed the three-dimensional structure of osteocytes under the scanning electron microscope (SEM) after removing the bone matrix by the HClcollagenase method. This intriguing study showed that osteocytes near osteoblasts were flattened, spindly shaped, and were connected to each other by fine cytoplasmic processes. In our study, we observed a similar pattern of osteocytes and showed fine processes between osteocytes. In addition, point-to-point distances between surfaces of osteocytes ranged from 10 to 15 Am in their study. However, the point-to-point distance between centers of osteocytes was 24.1 F 2.78 Am in our study. Considering that osteocyte cell size ranges from 10 to 15 Am, the cell-to-cell distance was close to the data from the previous SEM study. The number of osteocyte processes was counted in the present study. We have previously reported that a single osteocyte radiates 26.1 F 5.3 processes to the periosteal side [6]. The previous data did not include the number of processes that run parallel to the bone surface or on the bone side. In the present study, the average of 10 osteocytes was 52.7 F 5.7 processes. Based on these findings, the number of osteocyte processes facing the periosteal side might be higher than that on bone side. Marotti et al. [12] reported Table 2 Morphometrical data of single osteocyte Total length of cell processes Cell surface area Cell volume
1070 F 145 Am 1509 F 113 Am2 394 F 49 Am3
that the investigation was carried out by counting the number of canaliculi departing from the osteocyte lacunae using SEM, which showed that the number of canaliculi departing from periosteal side was significantly higher than that from the bone side. Therefore, morphological observations and several morphometrical analyses which were obtained from the osteocyte model-reconstructed confocal images could be reliable as same as those by using SEM. To our knowledge, analysis regarding the length of osteocyte processes was first performed in this study. As a result, one osteocyte had approximately a 1-mm total length of the processes. The total length of all processes in the average osteocyte (1 mm) was approximately 100 times greater than average length of the osteocyte cell body (10–15Am). As osteocytes have such a total length of processes forming intercellular network in bone, the question of how the osteocyte maintains such a total length of processes arises. We previously reported that osteocyte processes were rich in actin filaments [17]. Recently, we further analyzed the role of the cytoskeleton by exposing mechanical stress on the osteocytes [7]. During the course of fluid shear, osteoblasts slowly retracted, and the actinbinding protein, fimbrin, was dramatically reorganized, while osteocytes showed few changes in shape and stable distribution of fimbrin. Thus, the results described above suggest the importance of actin-binding protein for the structure and stability of the osteocyte processes. To analyze the surface area and volume of osteocytes, SURPASS software required a binary-image conversion process. Whenever binary images are reconstructed, there is always a problem of how to select the threshold of fluorescence intensity [9]. In this study, we employed fluorescently labeled phalloidin to stain the actin filaments in osteocytes because we previously reported that osteocyte cell body and cell processes are rich in actin filaments [17]. As a result, both cell body and cell processes showed strong fluorescent signals. Consequently, the process to distinguish fluorescence-positive sites from the background was easily performed. In addition, poorer resolution on the z axial direction than that on the xy plane is a typical problem of confocal 3D microscopy. This effect leads to elongation of both small and large structures, but the elongation becomes more pronounced as structures become smaller. The diameters of the processes are less than 0.5 Am. Therefore, it was necessary to determine the volume correction and surface area correction, and thus, we analyzed the fluorescently labeled gauge beads. Using the correction values obtained, we performed morphometrical analysis in osteocytes in bone.
Y. Sugawara et al. / Bone 36 (2005) 877–883
Previously, the surface area and volume of chick osteocytes in tibia were calculated by a three-dimensional ultrastructural study using serial TEM photographs [13]. They measured morphometrical data for an osteoid–osteocyte: cell cytoplasm 237 Am3 (cellular body 213 Am3; cytoplasmic processes 24 Am3), cell surface area 209 Am2 (including cytoplasmic process surface of 64 Am2). Although the morphometric data from TEM are very reliable because the margins of the cell surface are clearly distinguished by a dense uniform line of the cell membrane, this method is not suitable for following the extended complex course of osteocyte processes. In the present study, the cell volume and surface area for one osteocyte were 394 F 49 Am3 and 1509 F 113 Am2, respectively. Interestingly, despite the similar cellular body volume between TEM-based 3D reconstructed cells and fluorescently reconstructed cell of 213 to 257 Am3, the surface area of fluorescently reconstructed cell was almost seven times larger than that of TEM-based 3D reconstructed cells. Thus, the great increments in the surface area might be related to the ability to trace long osteocyte processes because the total length of osteocyte processes was shown to be over 1 mm in our study. The characteristic dendritic morphology of an osteocyte may contribute to enlarging the surface area in order to sense extracellular stiQ muli such as cytokines, hormones, and mechanical loading. In summary, previous osteocyte morphologic and morphometric studies were mainly performed by SEM or TEM. Although SEM observation of osteocyte network is possible after HCl-collagenase treatment, this method is not suitable for analyzing the length of the processes, volume, and surface area of osteocytes. Although 3D reconstruction from TEM serial images can provide morphometrical information, the TEM technique involves intricate pretreatment, such as decalcification, embedding, and sectioning and is not suitable for tracing the complex course of osteocyte processes. Recently, it is reported that osteocyte network changes in disease states such as osteoporosis, osteoarthritis, and osteomalacia [8]. Our approach to accurately determining osteocyte morphology could be used for this purpose of characterizing osteocytes in various disease states such as osteoporosis, osteoarthritis, etc. We could also use our measurements to design models of osteocyte function and response to stress.
Acknowledgments We thank Mr. Hiroshi Okamoto and Mr. Tetsuji Iwasa (Central Research Laboratory, Okayama University Medical School) for their technical assistance. This study was supported in part by grants-in-aid (#14207092, #15359491, #16390606) for scientific research from Japan Society for the promotion of Science.
883
References [1] Aarden EM, Burger EH, Nijweide PJ. Function of osteocytes in bone. J Cell Biochem 1994;55:287 – 99. [2] Arber S, Hunter JJ, Ross J, Hongo M, Sansig G, Borg J, et al. MLPdeficient mice exhibit a disruption of cardiac cytoarchitectural organization, dilated cardiomyopathy, and heart failure. Cell 1997; 88:393 – 403. [3] Donahue HJ. Gap junctions and biophysical regulation of bone cell differentiation. Bone 2000;26:417 – 22. [4] Doty SB. Morphological evidence of gap junctions between bone cells. Calcif Tissue Int 1981;33:509 – 12. [5] Ejiri S, Ozawa H. Scanning electron microscopic observations of rat tibia using the HCl-collagenase method. Arch Histol Jpn 1982;45: 399 – 404. [6] Kamioka H, Honjo T, Takano-Yamamoto T. A three-dimensional distribution of osteocyte processes revealed by the combination of confocal laser scanning microscopy and differential interference contrast microscopy. Bone 2001;28:145 – 9. [7] Kamioka H, Sugawara Y, Honjo T, Yamashiro T, Takano-Yamamoto T. Terminal differentiation of osteoblasts to osteocytes is accompanied by dramatic changes in the distribution of actin-binding proteins. J Bone Miner Res 2004;19:471 – 8. [8] Knothe Tate ML, Adamson JR, Tami AE, Bauer TW. The osteocyte. Int J Biochem Cell Biol 2004;36:1 – 8. [9] Kubinova L, Janacek J. Confocal microscopy and stereology: estimating volume, number, surface area and length by virtual test probes applied to three-dimensional images. Microsc Res Tech 2001; 53:425 – 35. [10] Lanyon LE. Osteocytes, strain detection, bone modeling and remodeling. Calcif Tissue Int 1993;53:S102–7. [11] Luthi A, Schwyzer L, Mateos JM, Gahwiler BH, McKinney RA. NMDA receptor activation limits the number of synaptic connections during hippocampal development. Nat Neurosci 2001;4:1102 – 7. [12] Marotti G, Remaggi F, Zaffe D. Quantitative investigation on osteocyte canaliculi in human compact and spongy bone. Bone 1985;6:335 – 7. [13] Palumbo C. A three-dimensional ultrastructural study of osteoid– osteocytes in the tibia of chick embryos. Cell Tissue Res 1986;246: 125 – 31. [14] Palumbo C, Silvana P, Davide Z, Gastone M. Osteocyte differentiation in the tibia of newborn rabbit: an ultrastructural study of the formation of cytoplasmic processes. Acta Anat 1990;137:350 – 8. [15] Power J, Loveridge N, Rushton N, Parker M, Reeve J. Osteocyte density in aging subjects is enhanced in bone adjacent to remodeling haversian systems. Bone 2002;30:859 – 65. [16] Schwyzer L, Marteos J, Abegg M, Rietschin L, Heeb L, Thompson SM, et al. Physiological and morphological plasticity induced by chronic treatment with NT-3 or NT-4/5 in hippocampal slice cultures. Eur J Neurosci 2002;16:1939 – 48. [17] Tanaka-Kamioka K, Kamioka H, Hans R, Lim SS. Osteocyte shape is dependent on actin filaments and osteocyte processes are unique actinrich projections. J Bone Miner Res 1998;13:1555 – 68. [18] Terai K, Takano-Yamamoto T, Ohba Y, Hiura K, Sugimoto M, Sato M, et al. Role of osteopontin in bone remodeling caused by mechanical stress. J Bone Miner Res 1999;14:839 – 49. [19] Winkler DT, Bondolfi L, Herzig MC, Jann L, Calhoun ME, Wiederhold K, et al. Spontaneous hemorrhagic stroke in a mouse model of cerebral amyloid angiopathy. J Neurosci 2001;21:1619 – 27. [20] Yellowley CE, Li Z, Zhou Z, Jacobs CR, Donahue HJ. Functional gap junctions between osteocytic and osteoblastic cells. J Bone Miner Res 2000;15:209 – 17.
Three-dimensional reconstruction of chick calvarial osteocytes and their cell processes using confocal microscopy Yasuyo Sugawaraa, Hiroshi Kamiokaa, Tadashi Honjoa, Ken-ichi Tezukab, Teruko Takano-Yamamotoa,* a
Department of Orthodontics and Dentofacial Orthopedics, Graduate School of Medicine and Dentistry, Okayama University, 2-5-1 Shikata, Okayama City, Okayama 700-8525, Japan b Department of Tissue Organ Development, Graduate School of Medicine, Gifu University, Tsukasa-machi 40, Gifu City, Gifu 500-8705, Japan Received 4 June 2004; revised 10 September 2004; accepted 8 October 2004
Abstract Osteocytes are surrounded by hard bone matrix. Therefore, it has not previously been possible to demonstrate the real architecture of the osteocyte network in bone. We previously reported that it is possible to observe osteocytes in bone by labeling the cells with fluorescence and using confocal laser scanning (CLS) microscopy. In this study, we for the first time conducted an extensive analysis of the morphology and morphometry of the three-dimensional (3D) osteocyte structure using three-dimensionally reconstructed fluorescent images. Sixteen-day-old embryonic chick calvariae were stained with fluorescently labeled phalloidin and observed using a confocal laser scanning microscope. Morphometry of osteocytes in the calvaria was analyzed using extensive three-dimensional reconstructing software IMARIS, process length measuring software NEURON TRACER and cell surface area-/cell volume-analyzing software SURPASS. From the IMARIS-derived images, we found that the average of 10 osteocytes is 52.7 F 5.7 processes, and the point-to-point distance between centers of the osteocytes was 24.1 F 2.8 Am. In addition, we could calculate that each osteocyte spans an average of 4180 F 673 Am3 of bone volume. NEURON TRACER showed that the length of osteocyte processes was 0.26 F 0.02 Am per 1 Am3 bone compartment. In addition, SURPASS indicated that the surface area of osteocytes was 0.36 F 0.03 Am2 per 1 Am3 bone compartment and that the volume ratio of osteocyte cell body to bone compartment was 9.42% F 1.18%. Together, the average total length of the processes, the average surface area, and the average volume of one osteocyte were 1070 F 145 Am, 1509 F 113 Am2, and 394 F 49 Am3, respectively. It is possible to reconstruct the real architecture of the osteocyte network and obtain morphometric data from fluorescently labeled osteocytes in chick calvaria. D 2004 Elsevier Inc. All rights reserved. Keywords: Osteocytes; Bone; Three-dimensional morphometry; Actin; Confocal laser scanning microscopy
Introduction Osteocytes are the most numerous cells in bone. They represent the final differentiation step in the osteoblastic lineage and arise when bone-forming osteoblasts become encased within the calcified matrix [15]. Osteoblasts lose a
* Corresponding author. Fax: +81 86 235 6694. E-mail addresses: [email protected] (Y. Sugawara)8 [email protected] (H. Kamioka)8 [email protected] (T. Honjo)8 [email protected] (K. Tezuka), [email protected] (T. Takano-Yamamoto). 8756-3282/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2004.10.008
large part of their cell organelles but gain long, slender cell processes by which the cells remain in contact with earlier incorporated osteocytes and with osteoblasts lining the bone surface [1]. The extended network of osteocyte processes joined at gap junctions [3,4]. Gap junctions are known to be involved in intercellular communication. In vitro, osteocytic and osteoblastic cells are functionally coupled to one another via gap junctions as shown by the ability of calcein to pass between cells and the ability of cells to communicate a mechanically induced calcium response [20]. The cellular network in vivo might play an important role for bone remodeling in response to mechanical loading [10,18]. Therefore, it is significant to
878
Y. Sugawara et al. / Bone 36 (2005) 877–883
analyze the whole figure of the osteocyte network and to obtain morphometric data. However, osteocytes are surrounded by hard bone matrix; hence it has been impossible to visualize the three-dimensional (3D) osteocyte network and to obtain morphometric data. To date, most previous studies on the 3D structure of osteocytes have been done with transmission electron microscope (TEM) serial sections, where the tissue is cut into 70- to 80-nm-thin slices [13,14]. Although the morphometrical data from TEM are very reliable because the margins of the cell surface are clearly distinguished by the dense uniform line of cell membrane, it is not suitable to the follow extended complex run of osteocyte processes. Therefore, morphometric analysis of osteocytes including processes has not previously been performed. Recently, we found that it was possible to observe the osteocyte network in chick calvaria by labeling the cells with fluorescence and using confocal laser scanning (CLS) microscopy [6]. As chick calvariae were very thin and flat, we could stain and visualize clearly. In the present study, we applied this technique for making a 3D-reconstructed fluorescent model of the osteocyte network. Observation of the osteocyte network in chick calvaria as well as morphometric analysis of the osteocyte network was performed using a 3D-reconstructed fluorescent model. IMARIS software, which reconstructed extensive 3D images [2,19], was employed to visualize the architecture of osteocyte network as a shadow projection and for counting the number of osteocyte processes, measuring the distance between osteocytes and calculating the volume of bone compartment occupied by a single osteocyte. NEURON TRACER software, which automatically traced the fluorescence images [11,16], was used to measure the length of osteocyte processes. In addition, SURPASS software was used to quantify the surface area and volume of osteocyte including osteocyte processes. As is well known, it is difficult to determine where one osteocyte process ends and contacts with another osteocyte. To report morphometric data for a single osteocyte, we analyzed all data in bone compartment units of 1 Am3. As a result, we could for the first time visualize a computer-reconstructed osteocyte network and obtain 3D morphometric data for a single osteocyte in bone including the full length of osteocyte processes.
Materials and methods Preparation of bone fragments Calvariae were obtained from 16-day-old embryonic chickens and washed with PHEM (60 mmol/l piperazineN, NV-bis [2-ethane-sulfonic acid], 25 mmol/l N-[2-hydroxyetyl] piperazine-NV-[2-ethanesulfonic acid], 10 mmol/l ethylene glycol-bis [2-amino-ethyl ether]-N, N, NV, NV-tetra-
acetic acid, 2 mmol/l magnesium chloride, pH 6.9). After stripping off the periosteum, the calvariae were trimmed into 3 3-mm pieces for further use. The average thickness of the sample ranged from 60 to 80 Am. Fluorescence staining and confocal laser scanning images Calvarial fragments were fixed with 3% paraformaldehyde in PHEM overnight, decalcified with 5% EDTA overnight, then permeabilized by incubation in 0.3% Triton X-100 in phosphate-buffered saline (PBS) for 10 min. The fragments were rinsed and stained for 2 days at 48C with a 1:200 dilution of Texas red-X-conjugated phalloidin (excitation wavelength = 595 nm; emission wavelength = 615 nm, Molecular Probes Inc., Eugene, OR) in PBS containing 1% BSA. After rinsing with PBS, the samples were embedded in fluorescence mounting medium (Dako, Carpinteria, CA) containing 1 mg/ml p-phenylenediamine dihydrochloride (Sigma, St. Louis, MO), then viewed immediately. Confocal optical sectioning was performed with an LSM510 confocal laser scanning (CLS) microscopy system (Carl Zeiss, Oberkochen, Germany) coupled to an upright microscopy (Zeiss) with PlanFluor objective (63, N.A. = 1.4) at Central Research Laboratory, Okayama University Medical School. The refraction index of immersion media (Zeiss 518F) was 1.518. Theoretical xy- and z-axes resolutions were 0.263 and 0.604 Am, respectively. Frame size of the image was 146.2 146.2 Am with an 8-bit color depth. Confocal images were taken with 0.3-Am step size and were processed four times with Kalman averaging. After obtaining confocal images, the images were processed for noise reduction with Huggens software (Bitplane AG, Zurich, Switzerland). We have showed previously study that the distribution of osteocyte processes coincided with that of canaliculi in the depth of 20 Am from osteoblast layer [6]. In this study, we focused on the osteocytes in osteoid layer (only 0 to 7.2 Am in depth). Confocal optical sectioning was initiated from the bone side of the osteoblast layer. 3D reconstruction of the osteocyte network by IMARIS The 3D structure of the osteocyte network was reconstructed from CLS images using IMARIS software (Bitplane). CLS images from 1.8 to 7.2 Am in depth were reconstructed for visualizing the osteocyte network. Computer analysis was performed by an Octane Workstation (Silicon Graphics, Inc.). The reconstructed images were digitally processed by Adobe Photoshop Version 5.5 software (Adobe Systems Inc., Mountain View, CA). Counting the number of processes that radiate from one osteocyte was performed by rotating the 3D-reconstructed fluorescent model. In addition, the volume of bone compartment occupied by one osteocyte was calculated by dividing the volume of bone compartment by the
Y. Sugawara et al. / Bone 36 (2005) 877–883
number of osteocytes in the compartment. IMARIS allows us to recognize the center of mass in each osteocyte. Thus, point-to-point distances between osteocytes were calculated. Quantitative analysis of length of osteocyte processes by NEURON TRACER NEURON TRACER (Bitplane AG), originally developed to analyze neurite morphology, was especially designed to automatically map areas of fluorescence on confocal images and construct 3D dendritic models to evaluate the total length of the model. We employed this software to obtain the topology of dendritic models of the osteocyte network and to determine the total length of osteocyte processes. We quantified 10 different regions, which contain several osteocytes, to analyze the length of the processes. The accuracy of the software was within 3%. To report the length of osteocyte processes, we divided the total length of osteocyte processes obtained using NEURON TRACER with the volume of bone compartment that was analyzed to measure the length of osteocyte processes by the software. Therefore, the length of osteocyte processes was reported by the length per 1 Am3 of bone compartment. Quantitative analysis of surface area and volume of osteocytes by SURPASS Surface area and volume of osteocytes per 1 Am3 of bone compartment were determined from SURPASS software. Eight-bit IMARIS images were converted to binary images to make 3D wire frame models. First, we determined the threshold for the binary models of osteocytes by superimposing binary images to IMARIS images until the images overlapped. To separate the cell body and cell processes, the length of wire frame was regulated by manual object segmentation in the SURPASS software. The average bone volume that was analyzed for surface area and volume of osteocytes was 22905 F 4026 Am3. This region contained five to eight osteocytes. Surface area and volume were automatically analyzed by SURPASS. To test the accuracy of the software automation, we confirmed control errors between red-fluorescent Fluosphere beads (2.0, 1.0, and 0.5 Am in diameter; Molecular Probes Inc.) and acquired fluorescent model in SURPASS software. Calculation of morphometrical data for a single osteocyte The average total length of the processes, the average surface area, and the average volume per osteocyte were determined by calculating the bone compartment volume occupied by one osteocyte, multiplying the length of processes, the surface area, and volume per unit volume of bone, respectively.
879
Result Observation of the osteocyte network by using IMARIS Confocal optical sectioning was performed from the bone side of osteoblast surface to bone compartment. Based on our previous study, the layer chosen was the osteoid layer. Fig. 1 shows the serial confocal images 0 Am (A), 1.8 Am (B), and 7.2 Am (C) in depth from the bone side of the osteoblast layer. Fig. 1A shows both osteocytes and osteoblasts because the surface of the calvaria showed slight undulations. Osteoblasts were seen as strong flat signals (asterisk) because the intensity of fluorescence in osteoblasts was much higher than that in osteocytes. In addition, halfembedded osteocytes were observed, as indicated by the arrowheads. At the depth of 1.8 Am, osteoblasts were mostly excluded and the observed cells were mainly osteocytes (Fig. 1B). Strong fluorescent signals in the processes were still recognizable even at 7.2 Am in depth (Fig. 1C). Fig. 2 shows IMARIS imaging of the osteocyte network reconstructed from confocal images from 1.8 to 7.2 Am in depth obtained with a 0.3-Am step size. Fig. 2A shows the view from the periosteal side, and Fig. 2B, that from the bone side at the same location. The network of osteocytes was visualized as a shadow projection. Osteocytes clearly appeared as spindle-shaped cells with numerous processes (arrows). Osteocytes in calvaria were regularly distributed in the same direction. The point-to-point distances between centers of osteocytes were measured. As a result, the distance from one cell to the adjacent cell was 24.1 F 2.8 Am (n = 10). Interestingly, we could recognize osteocytes that were covered by osteoblasts (asterisk in Fig. 2B). Although osteocyte processes and their branches were intersected with each other, we were able to trace the processes to the last detail (Figs. 2C, D). Then, the number of osteocyte processes were counted by IMARIS. The average number of processes radiating from single osteocyte was 52.7 F 5.7 (n = 10). In addition, we could calculate a single osteocyte occupied 4180 F 673 Am3 of bone compartment by dividing the volume of bone compartment by the number of osteocytes in the compartment. Topology of the dendritic tree and determination of the length of osteocyte processes by NEURON TRACER NEURON TRACER software was developed to form dendritic models of neurons and to measure the length of neuronal dendrites [11,16]. To obtain the length of osteocyte processes, we used this software. Fig. 3 shows the process of making the topological dendritic model (yellow line) from fluorescence-positive sites (white region). From the dendritic model, NEURON TRACER automatically measured the total length of osteocyte processes. It was difficult to determine where one osteocyte process ended. Therefore, we analyzed the length of
880
Y. Sugawara et al. / Bone 36 (2005) 877–883
Fig. 1. (A–C) Serial histotomography taken by CLS microscopy. Fluorescent images showed bone cells stained with Texas red-X phalloidin (red). Images were 0 Am (A), 1.8 Am (B), and 7.2 Am (C) in depth from the bone side of the osteoblast layer. Arrowheads in A show half-embedded osteocytes, and asterisks in A show osteoblast layer. Bar = 20 Am.
osteocyte processes in 1 Am3 bone compartment. The length of osteocyte processes per 1 Am3 bone compartment was 0.26 F 0.02 Am (n = 10). Morphometrical analysis of osteocyte in unit bone matrix by SURPASS The original IMARIS image in Fig. 4A was converted to a binary image using SURPASS software (Fig. 4B). We confirmed that the binary model maintained most of the osteocyte processes by comparison with those in the original IMARIS image (compare Figs. 4A and B).
Before calculating the volume and surface area of osteocytes, we measured 10 red-fluorescent beads (2.0, 1.0, and 0.5 Am diameter) as a test phantom to evaluate the accuracy of SURPASS software. The maximal diameters of the x-, y-, and z-axes in fluorescent beads were detected. The accuracy of our estimates of beads diameter is summarized in Table 1. Based on the calibration, a correction factor was derived for each axis. The residual of 2.0-Am diameter beads was 21% in the x and y, 97% in the z-axis, that of 1.0-Am diameter beads was 45% in the x and y, 103% in the z-axis, and that of 0.5-Am diameter beads was 71% in the x and y, 95% in the z-axis. Therefore, cross-
Fig. 2. Three-dimensional reconstruction of osteocyte network by IMARIS software. We used CLS images from 1.8 Am (Fig. 1B) to 7.2 Am (Fig. 1C) for the reconstruction. Left column of photos (A and C) shows the appearance from the periosteal side, and the right column (B and D), that from the bone side at the same location. C and D are magnified images of the rectangular areas in A and B. Arrows in A show osteocyte. Asterisks in B show osteocytes covered by osteoblasts. Bar in A = 20 Am, bar in C = 10 Am.
Y. Sugawara et al. / Bone 36 (2005) 877–883
881
Fig. 3. Image of a dendritic tree made by NEURON TRACER software. Fluorescence-positive sites (white region) showing the histotomography of osteocytes obtained by CLS microscopy. The yellow line shows the topology of the dendritic tree of osteocytes made by NEURON TRACER. (A–C) View from the vascular side, (D–F) sagittal views of A to C. Bar = 10 Am.
sections of a circle were thought to be distorted to an ellipse with elongation of the x-, y-, and z-axis in SURPASS data. Based on these results, we calculated the approximate volume and surface area correction as follows: Volume correction ¼ area of circle=area of ellipse ¼ pr 2 =pab Surface area correction ¼ circumference of circle=circumference of ellipse h pffiffiffiffiffii 2pr=p 1:5ða þ bÞ ab r: radius of fluorescent beads a: short axis of ellipse as x- or y-axis b: long axis of ellipse as z-axis As a result, the volume corrections for 0.5-, 1.0-, and 2.0Am beads were 0.30, 0.34, 0.41, respectively. The surface area corrections for 0.5-, 1.0-, and 2.0-Am beads were 0.54, 0.57, and 0.62, respectively. Most of the osteocyte-process diameters were less than 0.5 Am, while the short axis of the osteocyte cell body was approximately 2–5 Am as shown in
our previous study [17]. Therefore, we decided to use different size beads for calibration in order to determine correction factors to be used to accurately determine length and volume of processes and cell body: that obtained from 0.5-Am beads was used for process correction and that from 2.0-Am beads for cell body correction. Next, processes having less than 0.5-Am diameters were excluded by manual object segmentation (Fig. 4C). After this procedure, we separated the osteocyte cell body and the processes. Then the volume and surface area of the osteocyte cell body were calculated using corrections obtained from 2.0-Am beads. Then, the volumes and surface areas of osteocyte cell processes were calculated using corrections obtained from 0.5-Am beads. The average surface area of osteocytes per 1 Am3 of bone compartment was 0.36 F 0.03 Am2 (n = 10), and the volume ratio of osteocytes to bone compartment was 9.42% F 1.18% (n = 10). The percentage processes volume was 35%, and the percentage cell body excluding processes volume was 65%. On the other hand, the percentage processes surface area was 52%, and the percentage cell body excluding processes surface area was 48%.
Fig. 4. Images of binary model of osteocytes in calvaria made by SURPASS software. (A) Three-dimensional reconstruction of osteocyte network by IMARIS software. (B) A binary image of osteocytes using pseudocolors (green), which was processed from image A. (C) A binary image of osteocytes whose processes were excluded. Bar = 10 Am.
882
Y. Sugawara et al. / Bone 36 (2005) 877–883
Table 1 Summary of the evaluation of the effective resolution of fluorescent beads Beads diameter (accurate value) (Am) 2.0 1.0 0.5
Beads diameter from red-fluorescent beads (Am) [magnification (%)] x
y
z
2.42 F 0.17 [121 F 8.5] 1.45 F 0.12 [145 F 12.0] 0.85 F 0.05 [170 F 10.0]
2.42 F 0.17 [121 F 8.5] 1.45 F 0.12 [145 F 12.0] 0.85 F 0.05 [170 F 10.0]
3.90 F 0.20 [195 F 10.0] 2.03 F 0.20 [203 F 20.0] 0.98 F 0.04 [196 F 8.0]
Total length of processes, cell surface area, and volume of one osteocyte As described before, we could calculate that each osteocyte spans an average of 4180 F 673 Am3 of bone volume. Therefore, the morphometric data per unit bone compartment were multiplied by the mean volume of bone compartment occupied by each osteocyte (4180 Am3) to obtain morphometric data per single osteocyte. As a result, the average values for one osteocyte were calculated as follows: cell processes, 1070 F 145 Am; surface area 1509 F 113 Am2 (cellular body 724 Am2; cytoplasmic processes 785 Am2); and cell volume, 394 F 49 Am3 (cellular body 257 Am3; cytoplasmic processes 137 Am3) (Table 2).
Discussion Ejiri and Ozawa [5] first developed the three-dimensional structure of osteocytes under the scanning electron microscope (SEM) after removing the bone matrix by the HClcollagenase method. This intriguing study showed that osteocytes near osteoblasts were flattened, spindly shaped, and were connected to each other by fine cytoplasmic processes. In our study, we observed a similar pattern of osteocytes and showed fine processes between osteocytes. In addition, point-to-point distances between surfaces of osteocytes ranged from 10 to 15 Am in their study. However, the point-to-point distance between centers of osteocytes was 24.1 F 2.78 Am in our study. Considering that osteocyte cell size ranges from 10 to 15 Am, the cell-to-cell distance was close to the data from the previous SEM study. The number of osteocyte processes was counted in the present study. We have previously reported that a single osteocyte radiates 26.1 F 5.3 processes to the periosteal side [6]. The previous data did not include the number of processes that run parallel to the bone surface or on the bone side. In the present study, the average of 10 osteocytes was 52.7 F 5.7 processes. Based on these findings, the number of osteocyte processes facing the periosteal side might be higher than that on bone side. Marotti et al. [12] reported Table 2 Morphometrical data of single osteocyte Total length of cell processes Cell surface area Cell volume
1070 F 145 Am 1509 F 113 Am2 394 F 49 Am3
that the investigation was carried out by counting the number of canaliculi departing from the osteocyte lacunae using SEM, which showed that the number of canaliculi departing from periosteal side was significantly higher than that from the bone side. Therefore, morphological observations and several morphometrical analyses which were obtained from the osteocyte model-reconstructed confocal images could be reliable as same as those by using SEM. To our knowledge, analysis regarding the length of osteocyte processes was first performed in this study. As a result, one osteocyte had approximately a 1-mm total length of the processes. The total length of all processes in the average osteocyte (1 mm) was approximately 100 times greater than average length of the osteocyte cell body (10–15Am). As osteocytes have such a total length of processes forming intercellular network in bone, the question of how the osteocyte maintains such a total length of processes arises. We previously reported that osteocyte processes were rich in actin filaments [17]. Recently, we further analyzed the role of the cytoskeleton by exposing mechanical stress on the osteocytes [7]. During the course of fluid shear, osteoblasts slowly retracted, and the actinbinding protein, fimbrin, was dramatically reorganized, while osteocytes showed few changes in shape and stable distribution of fimbrin. Thus, the results described above suggest the importance of actin-binding protein for the structure and stability of the osteocyte processes. To analyze the surface area and volume of osteocytes, SURPASS software required a binary-image conversion process. Whenever binary images are reconstructed, there is always a problem of how to select the threshold of fluorescence intensity [9]. In this study, we employed fluorescently labeled phalloidin to stain the actin filaments in osteocytes because we previously reported that osteocyte cell body and cell processes are rich in actin filaments [17]. As a result, both cell body and cell processes showed strong fluorescent signals. Consequently, the process to distinguish fluorescence-positive sites from the background was easily performed. In addition, poorer resolution on the z axial direction than that on the xy plane is a typical problem of confocal 3D microscopy. This effect leads to elongation of both small and large structures, but the elongation becomes more pronounced as structures become smaller. The diameters of the processes are less than 0.5 Am. Therefore, it was necessary to determine the volume correction and surface area correction, and thus, we analyzed the fluorescently labeled gauge beads. Using the correction values obtained, we performed morphometrical analysis in osteocytes in bone.
Y. Sugawara et al. / Bone 36 (2005) 877–883
Previously, the surface area and volume of chick osteocytes in tibia were calculated by a three-dimensional ultrastructural study using serial TEM photographs [13]. They measured morphometrical data for an osteoid–osteocyte: cell cytoplasm 237 Am3 (cellular body 213 Am3; cytoplasmic processes 24 Am3), cell surface area 209 Am2 (including cytoplasmic process surface of 64 Am2). Although the morphometric data from TEM are very reliable because the margins of the cell surface are clearly distinguished by a dense uniform line of the cell membrane, this method is not suitable for following the extended complex course of osteocyte processes. In the present study, the cell volume and surface area for one osteocyte were 394 F 49 Am3 and 1509 F 113 Am2, respectively. Interestingly, despite the similar cellular body volume between TEM-based 3D reconstructed cells and fluorescently reconstructed cell of 213 to 257 Am3, the surface area of fluorescently reconstructed cell was almost seven times larger than that of TEM-based 3D reconstructed cells. Thus, the great increments in the surface area might be related to the ability to trace long osteocyte processes because the total length of osteocyte processes was shown to be over 1 mm in our study. The characteristic dendritic morphology of an osteocyte may contribute to enlarging the surface area in order to sense extracellular stiQ muli such as cytokines, hormones, and mechanical loading. In summary, previous osteocyte morphologic and morphometric studies were mainly performed by SEM or TEM. Although SEM observation of osteocyte network is possible after HCl-collagenase treatment, this method is not suitable for analyzing the length of the processes, volume, and surface area of osteocytes. Although 3D reconstruction from TEM serial images can provide morphometrical information, the TEM technique involves intricate pretreatment, such as decalcification, embedding, and sectioning and is not suitable for tracing the complex course of osteocyte processes. Recently, it is reported that osteocyte network changes in disease states such as osteoporosis, osteoarthritis, and osteomalacia [8]. Our approach to accurately determining osteocyte morphology could be used for this purpose of characterizing osteocytes in various disease states such as osteoporosis, osteoarthritis, etc. We could also use our measurements to design models of osteocyte function and response to stress.
Acknowledgments We thank Mr. Hiroshi Okamoto and Mr. Tetsuji Iwasa (Central Research Laboratory, Okayama University Medical School) for their technical assistance. This study was supported in part by grants-in-aid (#14207092, #15359491, #16390606) for scientific research from Japan Society for the promotion of Science.
883
References [1] Aarden EM, Burger EH, Nijweide PJ. Function of osteocytes in bone. J Cell Biochem 1994;55:287 – 99. [2] Arber S, Hunter JJ, Ross J, Hongo M, Sansig G, Borg J, et al. MLPdeficient mice exhibit a disruption of cardiac cytoarchitectural organization, dilated cardiomyopathy, and heart failure. Cell 1997; 88:393 – 403. [3] Donahue HJ. Gap junctions and biophysical regulation of bone cell differentiation. Bone 2000;26:417 – 22. [4] Doty SB. Morphological evidence of gap junctions between bone cells. Calcif Tissue Int 1981;33:509 – 12. [5] Ejiri S, Ozawa H. Scanning electron microscopic observations of rat tibia using the HCl-collagenase method. Arch Histol Jpn 1982;45: 399 – 404. [6] Kamioka H, Honjo T, Takano-Yamamoto T. A three-dimensional distribution of osteocyte processes revealed by the combination of confocal laser scanning microscopy and differential interference contrast microscopy. Bone 2001;28:145 – 9. [7] Kamioka H, Sugawara Y, Honjo T, Yamashiro T, Takano-Yamamoto T. Terminal differentiation of osteoblasts to osteocytes is accompanied by dramatic changes in the distribution of actin-binding proteins. J Bone Miner Res 2004;19:471 – 8. [8] Knothe Tate ML, Adamson JR, Tami AE, Bauer TW. The osteocyte. Int J Biochem Cell Biol 2004;36:1 – 8. [9] Kubinova L, Janacek J. Confocal microscopy and stereology: estimating volume, number, surface area and length by virtual test probes applied to three-dimensional images. Microsc Res Tech 2001; 53:425 – 35. [10] Lanyon LE. Osteocytes, strain detection, bone modeling and remodeling. Calcif Tissue Int 1993;53:S102–7. [11] Luthi A, Schwyzer L, Mateos JM, Gahwiler BH, McKinney RA. NMDA receptor activation limits the number of synaptic connections during hippocampal development. Nat Neurosci 2001;4:1102 – 7. [12] Marotti G, Remaggi F, Zaffe D. Quantitative investigation on osteocyte canaliculi in human compact and spongy bone. Bone 1985;6:335 – 7. [13] Palumbo C. A three-dimensional ultrastructural study of osteoid– osteocytes in the tibia of chick embryos. Cell Tissue Res 1986;246: 125 – 31. [14] Palumbo C, Silvana P, Davide Z, Gastone M. Osteocyte differentiation in the tibia of newborn rabbit: an ultrastructural study of the formation of cytoplasmic processes. Acta Anat 1990;137:350 – 8. [15] Power J, Loveridge N, Rushton N, Parker M, Reeve J. Osteocyte density in aging subjects is enhanced in bone adjacent to remodeling haversian systems. Bone 2002;30:859 – 65. [16] Schwyzer L, Marteos J, Abegg M, Rietschin L, Heeb L, Thompson SM, et al. Physiological and morphological plasticity induced by chronic treatment with NT-3 or NT-4/5 in hippocampal slice cultures. Eur J Neurosci 2002;16:1939 – 48. [17] Tanaka-Kamioka K, Kamioka H, Hans R, Lim SS. Osteocyte shape is dependent on actin filaments and osteocyte processes are unique actinrich projections. J Bone Miner Res 1998;13:1555 – 68. [18] Terai K, Takano-Yamamoto T, Ohba Y, Hiura K, Sugimoto M, Sato M, et al. Role of osteopontin in bone remodeling caused by mechanical stress. J Bone Miner Res 1999;14:839 – 49. [19] Winkler DT, Bondolfi L, Herzig MC, Jann L, Calhoun ME, Wiederhold K, et al. Spontaneous hemorrhagic stroke in a mouse model of cerebral amyloid angiopathy. J Neurosci 2001;21:1619 – 27. [20] Yellowley CE, Li Z, Zhou Z, Jacobs CR, Donahue HJ. Functional gap junctions between osteocytic and osteoblastic cells. J Bone Miner Res 2000;15:209 – 17.