Alternation in the gap-junctional intercellular communication capacity during the maturation of osteocytes in the embryonic chick calvaria

Bone 91 (2016) 20–29

Contents lists available at ScienceDirect

Bone journal homepage: www.elsevier.com/locate/bone

Full Length Article

Alternation in the gap-junctional intercellular communication capacity during the maturation of osteocytes in the embryonic chick calvaria Ziyi Wang, Naoya Odagaki, Tomoyo Tanaka, Mana Hashimoto, Masahiro Nakamura, Satoru Hayano, Yoshihito Ishihara, Noriaki Kawanabe, Hiroshi Kamioka ⁎ Department of Orthodontics, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, Okayama, Japan

a r t i c l e

i n f o

Article history: Received 26 February 2016 Revised 18 June 2016 Accepted 27 June 2016 Available online 29 June 2016 Keywords: Osteocytes Gap-junctional intercellular communication Osteocyte transformation Fluorescence recovery after photobleaching Connexins Mathematic model of simple diffusion

a b s t r a c t Introduction: The intercellular network of cell-cell communication among osteocytes is mediated by gap junctions. Gap junctional intercellular communication (GJIC) is thought to play an important role in the integration and synchronization of bone remodeling. To further understand the mechanism of bone development it is important to quantify the difference in the GJIC capacity of young and developmentally mature osteocytes. Materials and methods: We first established an embryonic chick calvaria growth model to show the growth of the calvaria in embryos at 13 to 21 days of age. We then applied a fluorescence recovery after photobleaching (FRAP) technique to compare the difference in the GJIC capacity of young osteocytes with that of developmentally mature osteocytes. Finally, we quantified the dye (Calcein) diffusion from the FRAP data using a mathematic model of simple diffusion which was also used to identify simple diffusion GJIC pattern cells (fitted model) and accelerated diffusion GJIC pattern cells (non-fitted model). Results: The relationship between the longest medial-lateral length of the calvaria (frontal bone) and the embryonic age fit a logarithmic growth model: length = 5.144 × ln(day) − 11.340. The morphometric data during osteocyte differentiation showed that the cellular body becomes more spindle-shaped and that the cell body volume decreased by approximately 22% with an increase in the length of the processes between the cells. However, there were no significant differences in the cellular body surface area or in the distance between the mass centres of the cells. The dye-displacement rate in young osteocytes was significantly higher than that in developmentally mature osteocytes: dye displacement only occurred in 26.88% of the developmentally mature osteocytes, while it occurred in 64.38% of the young osteocytes. Additionally, in all recovered osteocytes, 36% of the developmentally mature osteocytes comprised non-fitted model cells while 53.19% of the young osteocytes were the non-fitted model, which indicates the active transduction of dye molecules. However, there were no statistically significant differences between the young and developmentally mature osteocytes with regard to the diffusion coefficient, permeability coefficient, or permeance of the osteocyte processes, which were 3.93 ± 3.77 (× 10−8 cm2/s), 5.12 ± 4.56 (×10−5 cm2/s) and 2.99 ± 2.47 (×10−13 cm2/s) (mean ± SD), respectively. Conclusions: These experiments comprehensively quantified the GJIC capacity in the embryonic chick calvaria and indicated that the cell-cell communication capacity of the osteocytes in the embryonic chick calvaria was related to their development. © 2016 Elsevier Inc. All rights reserved.

1. Introduction Osteocytes are derived from a select group of osteoblasts that have undergone a final differentiation. Osteocytes represent 90–95% of the cells in the bone. Each osteocyte radiates approximately 50 long and branched cellular processes that contact other osteocytes and possibly other cell types, by means of gap junctions [8]. Gap junctional intercellular communication (GJIC) is crucial for bone formation, resorption ⁎ Corresponding author at: Department of Orthodontics, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, 2-5-1 Shikata, Kita-ku, Okayama 700-8525, Japan. E-mail address: [email protected] (H. Kamioka).

http://dx.doi.org/10.1016/j.bone.2016.06.016 8756-3282/© 2016 Elsevier Inc. All rights reserved.

and development [4,6,38]. Previous reports have shown that gap junctions exist in all types of bone cells [9,19]. Our previous experiments have shown that GJIC can occur between isolated chick osteocytes as well as osteocytes in the embryonic chick calvaria [17,21]. Connexins are integral membrane proteins that are composed of four transmembrane domains. They are passive channels that allow for the passive transport of intracellular signaling molecules of b 1 kDa in molecular weight [47]. Changes in cytosolic pH, voltage, intracellular or extracellular Ca2 +, as well as oncogenes and growth factors have been shown to open or close the connexin channels, likely resulting from changes in the state of phosphorylation of the connexin molecules [17,36]. Cx43 is the most abundant gap junction protein expressed in bone cells. Its function includes the formation of gap junctions and

Z. Wang et al. / Bone 91 (2016) 20–29

hemi-channels. In addition, through its cytoplasmic C-terminus domain, Cx43 can serve as a scaffolding protein that associates with structural and signaling molecules leading to regulation of intracellular signaling [36]. Cx43, which is crucial for skeletal homeostasis, regulates bone cell function by actively participating in the regulation of signaling, gene expression, cell survival and the ability to respond to mechanical stress [6, 36]. Additionally, several findings [25,28,46] in other organs suggest that Cx43 plays a role in cell polarized transformation and cell migration by means of the crosstalk with the cell signaling pathways via its ion channel function, especially its C-terminal domains. In addition, our previous study showed the age-related morphological transformation of osteocytes in the chick calvaria [22]. Plotkin and Bellido reviewed the findings in a large number of studies and suggested that Cx43 expression in osteoblast precursors (but not in mature osteoblasts or osteocytes) is required for full skeletal development, and that Cx43 expression is also needed for the proper function of mature osteoblasts and osteocytes [36]. Briefly, this suggests that GJIC can modulate cell growth, transformation and differentiation. To further understand the mechanism of bone development, it is therefore important to quantify the difference in GJIC capacity between young and developmentally mature osteocytes. For this aim, we established an embryonic chick calvaria growth model, which reflected the growth of the calvaria in embryos at 13 to 21 days of age. This model can also reveal the relationship between the osteocyte location and its age. So this model actually quantified and confirmed our previous finding that different aged osteocytes have different morphologies, with the young osteocytes having a more spherical shape and developmentally mature osteocytes being more spindle shaped [22]. Next, the FRAP technique was applied to calculate the dye displacement ratio in cell bodies and permeability in the cellular process network. In addition, a comparison was made between the young and developmentally mature osteocytes.

21

more closely resembles the base of the human skull than that used in other studies. Where most studies have used the top and lateral portions of the chick skull (shown in blue in Fig. 1 A), we used the skull base frontal bone, located under and in front of the chick brain and behind the eye (shown in red in Fig. 1 A). This part of the frontal bone is harder and thicker than other portions, allowing us to conveniently and precisely measure its growth change. After decapitating the embryonic chick (LOHMANN LSL-CLASSIC; LOHMANN TIERZUCHT, Germany), the calvariae were obtained from both gender and washed with commercial α-MEM (Catalog Number 12561-056, GIBCO, Waltham, Massachusetts, USA) to remove non-adherent cells. After stripping off the periosteum, the frontal bone was resected for further use (Fig. 1 A, B). First, the frontal bone was loaded with 2.5 μM calcein acetoxymethyl ester (Calcein-AM, Molecular Probes Inc., Eugene, OR), in α-MEM for 15 min in an incubator (at 37 °C in a 5% CO2 atmosphere). After Calcein-AM permeates into the cytoplasm, it is hydrolyzed by esterases to calcein (molecular weight = 622 Da), which remains inside the cell. Next, the sample was washed several times in α-MEM to remove excess dye. The frontal bone was then cut into about 1.5 mm (lateralmedial) × 2.5 mm (anterior-posterior) bone fragments (Fig. 1 C), and placed in glass-bottomed plastic dishes. They were held in place by a coverslip, which was secured using adhesive grease. During the experiment, the fragments were incubated with α-MEM containing 5% FBS, and the sealed dish was placed on a microscope heating stage (Model: MI-IBC-I-2, OLYMPUS, Japan) at 37 °C. All of the instruments and dishes had been sterilized, and all of the steps were performed on a clean bench. No antibiotics were used in the culture media. The study was approved by the institutional ethics committee on animal research at Okayama University (study number: OKU-2013260).

2.2. The establishment of the embryonic chick calvaria growth model 2. Materials and methods 2.1. The preparation of bone fragments and the loading of Calcein-AM into osteocytes in chick calvariae The method was modified from that of Ishihara et al. in 2008 [17]. However, the part of the chick skull bone that we used in our study

As seen in Fig. 1A, the frontal bone was selected from embryos at 13 to 21 days of age (8 eggs per day; total: 72 eggs). The distance from the peak of the medial side to the peak of the lateral side of the frontal bone was then measured using a slide gauge (precision: 0.05 mm). Non-linear regression analyses (an inverse model, a logistic model and a logarithmic model) were then performed to fit the data.

Fig. 1. The preparation of bone fragments and the loading of Calcein-AM. (A) The red area shows the skull base part (frontal bone) of the calvaria. The blue area shows the traditional portion used in most previous studies. (B) The red line segment shows how growth length is measured (from peak to peak of each side). (C) The preparation of bone fragments (B&C: Scale bar = 1 mm). (D) Spindle-shaped osteocytes stained with Calcein-AM. (E) Sphere-shaped osteocytes stained with Calcein-AM. (D&E: The red asterisks represent the target osteocyte, Scale bar = 20 μm). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

22

Z. Wang et al. / Bone 91 (2016) 20–29

2.3. Determining the selection region of the young and mature osteocytes using the growth model Nine frontal bones from chick embryos of 19 to 21 days of age (this age was selected because it was considered that the growth pattern of the osteocytes in the frontal bone in this age range could be predicted by the growth model) were chosen for Calcein-AM staining using the above-described method (see in Section 2.1). The fragments from these bones were then visualized with a FLUOVIEW FV500 confocal laser scanning microscopy system (CLCP system; FluoviewFV500, Olympus, Tokyo), which was coupled to a microscope with a × 40 (N.A. = 1.35) oil-immersion objective lens. After taking images with a 0.3-μm step along the z-axis at a resolution of 512 × 512 pixels (each side of the image was 417 μm in length) in the medial to lateral direction through the whole sample, all of the image data were analysed using the Imaris software program (Bitplane AG, Zurich, Switzerland). To measure the change in the diameter ratio (long/short) of each cell and the ratio of the spindle osteocytes (spindle osteocytes/amount) in the lateral–medial direction, the sample was divided into 13 rectangles with a 100-μm lateral-medial width and a 200-μm anterior-posterior length. We then defined the young region as a spindle osteocyte ratio of b 10% and a mean diameter ratio of b2.5; and the developmentally mature region as a spindle osteocyte ratio of N50% and a mean diameter ratio of N2.5 (this parameter set was based on our observation of pervious data [22]. Then used the growth model that obtained from Section 2.2 to calculate the age difference between the young and developmentally mature regions for confirmation of the age difference. 2.4. The FRAP analysis of the osteocytes in chick calvariae The frontal bones from chick embryos of 19 to 21 days of age were chosen for Calcein-AM staining using the above-described method (see in Section 2.1). Fluorescence-labeled osteocytes in these bone fragments were then visualized using a CLS system (1.66 s/scan). An osteocyte in lacuna, which was confirmed by differential contrast microscopy (DIC), was chosen for photobleaching. We chose an osteocyte surrounded by other osteocytes as a target cell (Fig. 1D and E). The boundary was enclosed and outlined with a rectangular region-of-interest tool. According to the distribution of the osteocyte diameter ratio (Table 1) and the growth model (Fig. 2), we chose sphere-shaped osteocytes and spindle-shaped osteocytes that were located 100–450 μm and 850–1300 μm, respectively, from the medial edge (our growth model showed an age difference of 2 to 3 days between these two regions, Table 2). First, using a low laser intensity (acousto-optic tunable filter [AOTF] = 1%, zoom = × 2) with a × 40 (N.A. = 1.35) oil-

Fig. 2. A growth model of embryonic chick front bone growth from 13 to 21 days of age. The model reveals a relationship between the longest medial-lateral length of the calvaria (frontal bone). The embryonic age fit a logarithmic growth model, as follows: length = 5.144 × ln(day) − 11.340.

immersion objective lens and a 0.3-μm step along the z-axis, a 3D image was taken for the collection of morphometric data of the target cell. Next, after modifying the protocol used by Ishihara [17] in 2008, a predefined three-step FRAP procedure was automatically executed at a scan speed of 1.66 s/scan on the largest cross section of the target cell. In the first step, a prebleached image of the whole field was taken using a low laser intensity (AOTF = 1%, zoom = × 2) with a × 40 (N.A. = 1.35) oil-immersion objective lens. The laser intensity was then increased ×100 (AOTF = 100%, zoom = ×40), and the target osteocyte was photobleached 3 times for about 0.9–3.3 s (depending on the shape of target cell). Finally, the laser intensity and zoom were immediately reset to the prebleach levels (AOTF = 1%, zoom = ×2). In each experiment, 16-bit images were acquired prior to and immediately after bleaching, and at 30 s intervals to record the recovery phase based on our previously experiments. The FRAP experiments were performed 4–6 times for each bone fragment (depending on the fragment condition; the FRAP experiments were performed an equal number of times in spindle- and sphere-shaped osteocytes). Fragments which showed at least one recovery in a FRAP experiment were then counted as valid samples (as a result, 14% of the bone fragment data were discarded). During the capture of the time-lapse images, some of the samples lost focus for reasons such as the loosening of the grease (approximately 5% of the FRAP experiments were discarded for this reason); some osteocytes were recovered too quickly and reached a plateau in just 1 min, which indicated that they were not wellphotobleached due to some unknown reason (approximately 5% of

Table 1 The distribution of the differently shaped osteocytes. Distance from the medial edge (μm)

Mean diameter ratio (19 days)

Mean diameter ratio (20 days)

Mean diameter ratio (21 days)

Kruskal-Wallis P value

Mean diameter ratio

Spindle osteocyte ratio (diameter ratio N 2.5)

100–200 200–300 317–417 417–517 517–617 634–734 734–834 834–934 951–1051 1051–1151 1151–1251 1268–1368 1368–1468

1.56 1.60 1.71 1.81 2.00 2.32 2.45 2.45 2.70 2.39 2.49 2.76 2.83

1.53 1.59 1.74 1.75 1.94 2.13 2.40 2.58 2.58 2.66 2.64 2.83 2.87

1.59 1.64 1.65 1.86 2.10 2.33 2.31 2.45 2.81 2.83 2.84 2.99 3.07

0.720 0.468 0.456 0.412 0.360 0.144 0.487 0.840 0.134 0.001⁎⁎ 0.012⁎ 0.205 0.211

1.56 1.61 1.70 1.81 2.01 2.26 2.38 2.50 2.70 2.63 2.66 2.86 2.92

3.68% 4.69% 7.66% 12.96% 22.87% 33.91% 39.94% 48.41% 57.91% 56.06% 58.02% 69.77% 67.91%

The Kruskal-Wallis test was performed using samples from the three days. ⁎ P b 0.05. ⁎⁎ P b 0.01.

Z. Wang et al. / Bone 91 (2016) 20–29 Table 2 The estimated age in the regions of young and developmentally mature osteocytes. Age (days)

Estimated age in young region 100–450 μm (day)

Estimated age in developmentally mature region 850–1300 μm (day)

Age difference (day)

19 20 21 Mean

0.35–1.57 0.36–1.65 0.50–1.85 1.05

2.93–4.28 3.00–4.50 3.34–4.82 3.82

2.58–2.70 2.72–2.85 2.84–2.97 2.78

23

body surfaces were obtained from the 3D images of the FRAP experiments using the Imaris software program. Based on our previous study [40], the values of the length, volume and area were multiplied by the correction factors of 0.83, 0.41 and 0.62, respectively. As Fig. 3C shows, the number of processes for each osteocyte was counted from the prebleached image of the Texas Red-X-conjugated phalloidinstained sample (method described below).

2.6. The number of cell processes contributing to dye transport

the FRAP experiments were discarded for this reason). The analysis was performed with the valid experiments (after removing the above-described invalid experiments) (Table 3). The item “no recovered cells” in Table 3 included the cells that were recovered at a rate of b10%. The 3D images were then used to obtain morphologic measurements (see Section 2.5. The collection of the morphometric data). Fluorescence replacement was assessed at 30 s intervals. There was a similar number of adjacent cells around the photobleached cell in each of the experiments (3–7). We performed the experiments at 37 °C within 45 min after finishing the dye loading. The fluorescence intensity (a grey-scale, ranging from 0 to 4095 relative integer units) of each cell was obtained using the Fiji software program (LOCI, Madison, Wisconsin, USA). The double normalization method [35] was then applied to normalize all of the digitized data. Double normalization involves normalization of the recovery signal to the average prebleach signal and, at the same time, takes into account the loss of total signal due to the bleach pulse and bleaching during postbleach imaging. For double normalization, the average intensity at each imaging time point must be measured for three regions of interest: the bleached osteocyte (It), all other osteocytes in the image field (Tt), and a random region outside of all of the cells for background subtraction (BG). The following equation was used [35]: Relative fluorescence
signal of bleached cell T prebleach −BG ðI t −BGÞ
¼ ðT t −BGÞ Iprebleach −BG

2.5. The collection of the morphometric data The shortest diameter, longest diameter, diameter ratio (long/ short), cell body volume, cell body surface area, cell-cell mass center distance and the shortest distance between the target and neighbour cell

As the target osteocyte is photobleached with the high-intensity scanning laser beam, the cell body and the processes above and below the target plan also lie within the path of the laser (shaded area in Fig. 3A) and are photobleached. Thus, these processes cannot deliver fluorescent tracers to the photobleached processes, and only the processes outside the light path contribute to the subsequent cell body fluorescence recovery. To measure the fraction of the contributing processes, 19-day-old calvarial fragments were fixed with 3% paraformaldehyde in phosphate-buffered saline (PBS) overnight, decalcified with 5% EDTA overnight, then permeabilized by incubation in 0.3% Triton X-100 in PBS for 10 min. The fragments were rinsed and then stained for 2 days at 4 °C with a 1:200 dilution of Texas Red-X-conjugated phalloidin (1:200 dilution; 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), then viewed immediately. Images of 20 cells (10 spindle-shaped cells and 10 sphere-shaped cells) from the cross-sections were chosen. In theory, the photobleaching laser paths will form a 54-degree conical angle emanating from the edges of the osteocyte, according to the numerical aperture (0.9) of the lens and the refraction index of the immersion media 1.518 that were used in the FRAP experiments (Fig. 3A). In order to count the number of contributing processes, a FRAP experiment was performed using Texas Red-X-conjugated phalloidin-stained samples. All of the parameters of the experiment were the same as those in the Calcein-AMstained experiments, with the exception that the bleaching period was increased to 1 min. A 3D superimposition image was then created (Fig. 3 B, C, D and E) using the prebleaching and postbleaching images in order to count the number of cell processes that were not influenced by bleaching. There was no significant difference in the number of contributing processes that lay outside the laser path in the spindle- and sphere-shaped osteocytes; the average number was 17.25 ± 3.23 (mean ± SD) per osteocyte.

Table 3 A summary of the FRAP experiments. Amount GLOBAL Total number of fragments Number of recovered fragments Total number of osteocyte FRAP experiments Invalid osteocyte FRAP experiments (loss of focus) Invalid osteocyte FRAP experiments (not well-photobleached) VALID Valid FRAP experiments Recovered cells Fitted model cells R2 of mathematical model in non-fitted model cells Non-fitted model cells R2 of mathematical model in fitted model cells No recovered cells RATIO Rate of recovered cells Non-fitted model cells/recovered cells

Developmentally mature osteocytes

Young osteocytes

93 25 16 98.35% ± 1.24% 9 79.44% ± 11.34% 68

73 47 22 97.86% ± 1.28% 25 86.12% ± 11.57% 26

26.88% 36.00%

64.38% 53.19%

52 38 186 10 10 166 72 38 98.07% ± 1.27% 34 84.36% ± 11.73% 94 43.37% 47.22%

24

Z. Wang et al. / Bone 91 (2016) 20–29

Fig. 3. A schematic illustration of the laser pathway and a 3D reconstructed image. (A) The schematic illustration shows that the processes above and below the target osteocyte also lay within the laser path. (B) A 3D reconstructed image shows bone cells stained with Texas red-X phalloidin (red). (C) A prebleaching image which was used to count the total number of processes. (D) A postbleaching image of the same target cell as (A) and (B), which was used for 3D superimposition. (E) A 3D superimposition image which was used to count the number of contributing processes. The green part comes from the prebleaching image, the red part comes from the postbleaching image. (The white asterisks represent the target osteocyte, scale bar = 3um). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

2.7. The fluorescence replacement rate The replacement of fluorescence within a bleached osteocyte was calculated using the following equation [17]: Percent replacement ¼ ½ð F t −F 0 Þ=ð F i −F 0 Þ  100ð%Þ The percent replacement was defined as the fraction of molecules that were replaced during the time-course of the experiment. Ft was the normalized fluorescence intensity after photobleaching. F0 was the theoretical fluorescence intensity immediately after photobleaching. Fi was the initial fluorescence intensity before photobleaching. 2.8. The mathematical model for dye diffusion We used a two-compartment model that was developed by Liyun Wang et al. in 2005 [45]. Briefly, this model described the movement of unbleached tracer molecules into the photobleached osteocyte (a pool) from the neighbouring osteocytes (a source reservoir) through the connecting channel. The diffusion coefficient (D) can be approximated using the following formulae: Vr ¼ 

NAfc d Vs

I0 −Iðt 0 Þ ln I0 −Ib

ð1Þ  ¼−

V r t0 D 2

d

ð2Þ

where I0 and Ib are the fluorescence intensity immediately before and after photobleaching, respectively; I(t′) is the fluorescence intensity during recovery; t′ is the time after photobleaching; D is the tracer diffusion coefficient; Afc is the cross-section area in one osteocyte process; Vs is the volume of the photobleached osteocyte; d is the mean distance between the photobleached osteocyte and its neighbours (process lengths are estimated by the shortest length from the target cell body surface to neighbour cell surface). Vs and d were obtained from the 3D prebleaching image, and were based on our previous study; the values were multiplied by the correction factors of 0.41 and 0.82 respectively [40]. The average process diameter (200 nm) [48], was used in our calculations of the diffused area. Since this model predicts that all of the data points should fall on a straight line (Eq. (2)). The fit model cell was therefore defined as the coefficient of determination (R2) was N0.96. Please notice that this mathematic model was based on a simple

diffusion procedure [45] and assumes that the tracer moves evenly in processes. This means that in all of the fitted model cells, the speed of tracer through the interface between two processes was the same as the tracer movement in the cell process channels. This model also can therefore be used to identify the simple diffusion GJIC pattern cells (fitted model) and the accelerated diffusion GJIC pattern cells (nonfitted model). The most appropriate measure for the specific description of the ease of diffusion through a gap junction is permeance [3,10,15]. Since the number and cross sectional area of the contributing processes were known, permeance (PA) contains two terms, the permeability coefficient (P), and the junctional area (A) [3]. For this experiment, A is the cross-sectional area of the osteocyte processes. Since the number and cross sectional area of the contributing processes were known, P and PA can be calculated using the following formulae [3,43]: P ¼ D=distance PA ¼ P  A where distance is the shortest distance between two cell body surfaces. 2.9. Statistical analysis Student's t-test was used to analyse the data that were normally distributed in two groups. Welch's t-test was used to analyse the data of two groups that showed different variance. The Mann-Whitney U test was used for the data were not normally distributed. Asterisks indicate statistically significant differences (⁎p b 0.05; ⁎⁎p b 0.01; ⁎⁎⁎p b 0.0001; ⁎⁎⁎⁎p b 0.0001). Statistical significance was determined using a statistical analysis software program (SPSS, Chicago, Ill). All of the analyses were performed with a significance level of α = 0.05. 3. Results 3.1. The embryonic chick calvaria growth model Fig. 2A shows the mathematic growth model that best fit the growth data (R2 = 0.99). This model revealed the relationship between the skull base bone length and the embryonic age of the samples from the 13-day-old to 21-day-old chick embryos. It followed a simple equation: length = 5.144 × ln(day) − 11.340 (Fig. 2). Using this model, the age of the different shapes of osteocytes can be calculated.

Z. Wang et al. / Bone 91 (2016) 20–29

Table 1 shows the distribution of the differently shaped osteocytes. The data from the 9 samples included a total 4252 cells. The data showed the relationship between the distance of target cell to the medial edge and the cell's diameter ratio. Based on these results, the distance of 100–450 μm (diameter ratio, b2; spindle-shaped osteocyte ratio, b10%) and 850–1300 μm (diameter ratio, N2.5; spindle-shaped osteocyte ratio, N50%) from the medial edge were chosen to select the sphere- and spindle-shaped osteocytes, respectively (Table 1). As Table 2 shows, the age difference between the sphere- and spindle-shaped osteocytes was approximately 2–3 days, and approximately 3–4 days are required for sphere-shaped osteocytes to transform into spindle-shaped osteocytes (Table 2). 3.2. A summary of the FRAP experiments Within 5 min after photobleaching (Fig. 4A) the mean fluorescence replacement in the spindle- and sphere-shaped cells were 34.44 ± 3.40% and 46.45 ± 2.50% (mean ± SEM), respectively (an example is shown in Fig. 5). There was a significant difference in the recovery phase of the spindle- and sphere-shaped osteocytes. Spindle-shaped osteocytes, or osteocytes that are developmentally mature, have a lower GJIC capacity. Meanwhile, the rate of recovered cells in developmentally

25

mature osteocytes was only 26.88%, while that in young osteocytes was 64.38% (Table 3). 3.3. The morphometric results As Table 4 shows, there were significant differences between the developmentally mature osteocytes and the young osteocytes regarding the shortest diameter of the cell body, the longest diameter of the cell body, the diameter ratio (long/short), cell body volume and the shortest distance between the target and neighbour cell body surfaces, while the mass-centre distance between the cells, the cell body surface area, the number of processes and the number of contributing processes did not differ to a statistically significant extent. The shortest diameter and cell volume decreased from young osteocytes to mature osteocytes, but the longest diameter, the diameter ratio and the shortest boundaryboundary distance increased (Table 4). 3.4. The permeability results Among all valid data (Table 3, 38 fitted model cells and 34 non-fitted model cells), there were significant differences between the fitted model cells and the non-fitted model cells throughout the entire

Fig. 4. The displacement rates in the different groups. (A) Young and mature osteocytes in all of the recovered samples; (B) Fitted and non-fitted model samples that contained both young and mature osteocytes; (C) Young and mature osteocytes that only contained fitted model cells; (D) Young and mature osteocytes that only contained non-fitted model cells; (E) Fitted and non-fitted model cells in young osteocytes. (F) Fitted and non-fitted model cells in mature osteocytes. The data are expressed as the mean ± SEM.

26

Z. Wang et al. / Bone 91 (2016) 20–29

Fig. 5. Examples of the FRAP experiment recordings and the data analysis for an osteocyte. (A) Time-lapse confocal images of the photobleached osteocytes in a representative FRAP experiment. (B) The fluorescence intensity of the photobleached osteocytes increases exponentially after photobleaching due to the influx of the tracer, reaching a plateau of ≈45% of the prebleach level by 300 s. This partial recovery is possibly due to a net loss of fluorescence tracer due to photobleaching. (C) The experimental data ln[(I0 − I) / (I0 − Ib)] fit well with a straight line as predicted by the two-compartment model (Eq. 2). The diffusion coefficient (D) of Calcein fluorescein was calculated from the slope of the line.

recovery phase (Fig. 4B). The data (example shown in Fig. 5) from 38 (Table 3, 16 developmentally mature osteocytes and 22 young osteocytes) cells fit a straight line with an average R2 of 98.07% ± 1.27% (mean ± SD) (Table 3). There were no significant differences between the developmentally mature osteocytes and young osteocytes in the recovery phase of these 38 osteocytes (Fig. 4C); within 5 min after

photobleaching, these 38 cells reached a plateau with a mean value of 37.14 ± 2.69% (mean ± SEM). Interestingly, in the cells of the non-fitted model cells (Table 3, 9 developmentally mature osteocytes and 25 young osteocytes), there were significant differences between the developmentally mature osteocytes and the young osteocytes throughout the entire recovery phase (Fig. 4D). In young osteocytes there were

Table 4 The morphometric data of the differently shaped osteocytes.

b

Shortest diameter (μm) Longest diameter (μm)a Diameter ratio, long/shortc Cell volume (μm3)b Cell surface area (μm2)b Cell-cell distance, mass center (μm)a Boundary-boundary shortest length (μm)c Number of processesb Number of contributing processesb The values show the mean ± SD. a Welch's t-test. b Student's t-test. c Mann-Whitney U test. ⁎ P b 0.05. ⁎⁎ P b 0.01. ⁎⁎⁎ P b 0.001. ⁎⁎⁎⁎ P b 0.0001.

Young osteocytes (n = 47)

Developmentally mature osteocytes (n = 25)

6.14 ± 1.01 10.48 ± 1.32 1.75 ± 0.34 197.70 ± 70.98 196.60 ± 47.01 From neighbour-target cells couple (n = 243) 16.04 ± 3.91 6.42 ± 2.87 From phalloidin-stained samples (n = 10) 50.40 ± 4.5 17.40 ± 2.59

4.50 ± 0.99⁎⁎⁎⁎ 12.90 ± 2.50⁎⁎⁎ 2.91 ± 0.46⁎⁎⁎⁎ 154.90 ± 78.51⁎ 179.70 ± 62.72P N 0.09 From neighbour-target cells couple (n = 101) 16.30 ± 6.04P N 0.95 7.70 ± 3.59⁎⁎ From phalloidin-stained samples (n = 10) 50.40 ± 6.20P N 0.99 17.10 ± 3.90P N 0.84

Z. Wang et al. / Bone 91 (2016) 20–29

27

significant differences between the fitted and non-fitted model cells throughout the entire recovery phase (Fig. 4E). In developmentally mature osteocytes, a statistically significant difference between the fitted and non-fitted model cells was only observed at the first time point (Fig. 4F). Additionally, developmentally mature osteocytes and young osteocytes accounted for 36% and 53.19%, respectively, of the osteocytes in the non-fitted model (Table 3). There were no significant differences in the diffusion coefficient (D), permeability coefficient (P) or permeance (PA) of mature osteocytes and young osteocytes; the average values were 3.93 ± 3.77 (× 10−8 cm2/s), 5.51 ± 4.56 (× 10− 5 cm/s) and 2.99 ± 2.47 (×10−13 cm3/s), respectively (Table 5). 4. Discussion 4.1. The growth model and the morphological changes in vivo in the embryonic chick calvaria A previous study showed that the transformation from osteoblasts to osteocytes takes about three days in the femoral metaphysis of 2week-old rabbits [34]. In their review, Franz-Odendaal et al. [11], combined different observations to propose four recognizable stages of osteocyte differentiation: osteoblastic osteocytes (Type I preosteocytes that are just starting to embed in hard tissue), osteoid-osteocyte (Type II preosteocytes), Type III preosteocytes, young osteocytes and old osteocytes. According to their description and combined with our growth model (Fig. 2) the sphere-shaped osteocytes corresponded to Type I, while the spindle-shaped osteocytes corresponded to Type III or the fourth stage of development. Our results indicated that 3–4 days are required for the osteoblasts to transform themselves into developmentally mature osteocytes in embryonic chicks (Table 2). This result was closely similar to the results of a previous study in rabbits [34]. We concluded that at least 3–4 days are required for sphere-shaped osteocytes to transform themselves into spindle-shaped osteocytes. Our morphometric data revealed the morphological changes from Type I to Type III or older osteocytes. During the transformation of osteocytes in the embryonic chick calvaria, the cellular body becomes more spindle-shaped and the cell body volume decreases by about 22%. In addition, there is no significant change in the cellular body surface (Table 4). The process length can be estimated by the shortest distance between cell surfaces. The decrease in cellular body volume and the unchanged mass centre distance (16.04 μm in sphere and 16.30 μm in spindle) therefore contribute to the increase in the estimated shortest process length (Table 4). The shortest process length increases by approximately 1.28 μm during development. Given the complicated nature (the increase in branches and length) of the cellular process network [16,41], this increase may thus be over underestimated. Taken together we propose the following growth pattern of the calvaria (front bone) at 19 to 21 days of age (Fig. 6). As Fig. 6 shows, this pattern includes some assumptions: a) only osteoblasts can proliferate; b) all osteocytes have almost the same speed of differentiation; c) the growth direction is from the surface to the inside. Thus, before the plateau phase, the area of the sphere-shaped osteocyte region will

Table 5 The permeability of the osteocyte pr ocesses.

D (×10−8 cm2/s)c P (×10−5 cm/s)c PA (×10−13 cm3/s)c

Young osteocytes (n = 22)

Developmentally mature osteocytes (n = 16)

Global (n = 38)

4.12 ± 3.20 6.29 ± 4.24 3.41 ± 2.30

3.68 ± 4.54P N 0.28 4.44 ± 4.89P N 0.05 2.41 ± 2.66P N 0.06

3.93 ± 3.77 5.51 ± 4.56 2.99 ± 2.47

D: diffusion coefficient; P: permeability coefficient; PA: permeance. The values show the mean ± SD. c Mann-Whitney U test.

Fig. 6. The growth pattern of calvaria (front bone). According to this pattern, the growth direction is from the surface to inside and the area of the sphere-shaped osteocyte region will maintain its size (middle yellow layer in this schematic illustration). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

maintain its size. These assumptions were almost the same as our observations (Table 1). As Table 1 shows, there were no significant differences in the diameter ratios at 19, 20 and 21 days in almost all of the regions, except for 1051–1151 and 1151–1251. This exception may be caused by a sample morphometric structure. As a result, the thickness of the sample increases from the medial edge to the lateral edge (Fig. 1B and C). Thus, as the distance increases, the medial side the image does not show the entire thickness, it also becomes difficult for the laser to pass and the tissue becomes harder and thicker. When the sample size is considered (9 bone fragments and 4252 cells), it makes sense that the conditions were not ideal in some of the regions. 4.2. The difference in the displacement rate in GJIC capacity between younger and more mature osteocytes in vivo in the embryonic chick calvaria Among other cell-labeling reagents, including carboxy-fluorescein diacetate (CFDA) and fluorescein diacetate (FDA), Calcein-AM is particularly suitable as a fluorescent probe for staining living cells because of its low cytotoxicity [44]. Calcein-AM can permeate into the cell noninvasively. Furthermore, the Calcein-AM that remains outside the cells is a non-fluorescent compound, and the Calcein-AM that infiltrates the cells will be hydrolyzed by esterases to calcein, a fluorescent anion that cannot pass through the cellular membrane. All of these features helped us create a nearly natural bone environment for this experiment. This experiment only focuses on what is mostly a single layer of osteocytes, because the chosen lens formed a laser pathway with a small angle (54o, see in Section 2.6) at the cellular body edge (Fig. 3A). This meant that the processes and the neighbouring cells above and below the target osteocyte that also lay within the laser path were bleached. A limitation associated with this study is that there was a very large number of lacunae that did not contain any labeled cells in heavily mineralized areas (Fig. 1D). This could be due to the slower dye penetration in heavily mineralized areas compared to less mineralized (more permeable) areas for the initial immersion of the dye. Although this

28

Z. Wang et al. / Bone 91 (2016) 20–29

experiment only focused on what is mostly a single layer of osteocytes, it is still possible that the reduced dye transfer could have been due to the fact that there are dramatically fewer labeled osteocytes that connected to the target cells. The combination of all 166 FRAP experiments (93 spindle-shaped and 73 sphere-shaped osteocytes, Table 3) showed that there were significant differences at each time point between the young and developmentally mature osteocytes during the recovery phase. The developmentally mature osteocytes had a lower displacement rate (Fig. 4A). Furthermore, only 26.88% of mature osteocytes had a GJIC with neighbouring cells, while this value was 64.38% in young osteocytes (Table 3). All of these results suggested that developmental maturation caused a decrease in the GJIC capacity between osteocytes. This correlates with the findings of previous studies. In their review, Buo and Stains [6] combined several findings [1,7,12–14,18,26,27,33,37,42] and suggested that the resistance of osteoblasts/osteocytes to IGF-1 and/or PTH signaling can alter the Cx43 function caused by aging. Solan and Lampe described that in different stages of the cell cycle, the level of phosphorylation and the rate of Cx43 assembly is different. In the G0 phase Cx43 has the lowest phosphorylation level but the highest assembly rate, and the GJIC capacity is highest in the S and G2 phases. This review also suggested a PKC-mediated increase in phosphorylation as cells progress from the G0 phase to the S phase [39]. Basically, developmentally mature osteocytes are known as in G0 phase cells and the mechanism of underlying the high dye displacement rate in the remaining mature cells requires further study (Fig. 4C and D). 4.3. The permeability of the osteocyte processes in vivo in the embryonic chick calvaria In their review, Zhang et al. [49] summarized a number of studies to sketch a Cx43 lifecycle. They suggested that Cx43 is a short-lived protein with a half-life of only 1–5 h, suggesting that its intracellular movements are tightly regulated; the dependence of Gap Junction (GJ) formation and the maintenance of actin and the F-actin network determines that the Cx43 cargo moves toward the cellular junction, in other words, microtubules bring the Cx43 cargo from the Golgi network and the actin conjugate the positive end of the microtubule to determine the direction of the Cx43 cargo. The movement of the Cx43 cargo slows in association with actin (b0.25 μm/s). Our previous study [20] showed that microtubules are distributed at the end of the process base but that actin is distributed throughout the entire process. Thus, even when Cx43 expression in the bones or cells from old rodents appears to be unaltered [13,18], the increase in processes length (Table 4, see above) can still reduce the GJIC capacity during osteocyte development. This may partly explain why the age-related changes in the GJIC cause the alteration in cell morphology. 4.4. The number of opened Cx43 channels in each cell Many studies have suggested that the permeability, channel pore size and open-state probability of the Cx43 gap junction could be regulated by changes in phosphorylation [5,10,23,30–32,36]. Thus, changes in the phosphorylation patterns of Cx43 have been correlated with alterations in GJIC [24]. The mathematic model that we used in the present study was based on a simple diffusion procedure [45] in which it was assumed that the tracer moves evenly in the processes. This means that in all of the cells of the fitted model, the speed of the tracer through the interface between two processes was the same as tracer movement in the channels of the cell processes. This reason could explain why there were significant differences between the young and developmentally mature cells in the non-fitted model but not in the fitted model (Fig. 4C and D). This suggests that if the Calcein diffusion coefficient of Cx43 singlechannel were to be known, the number of open Cx43 channels on the

interface could be calculated for the fitted model cells. Fortunately, in 2006, Eckert [10] used high Cx43-expressing HeLa cells and BICR/M1Rk cells to determine the diffusion coefficients of several dyes through the Cx43 single-channel, including Calcein. In Reiner's study, the results were not normally distributed. Their results showed two peaks, the lower permeance peaks of the two cell types (HeLa cells vs. BICR/M1Rk cells) were very similar (2.0 ± 2.4 × 10−15 cm3/s vs. 2.2 ± 2.0 × 10−15 cm3/s) but there was a large difference in the higher permeance peaks (15 ± 25 × 10−15 cm3/s vs. 7.5 ± 9.5 × 10−15 cm3/s). One of the author's assumptions was that the low permeance came from the cells in which the phosphorylation level of the Cx43 gap junction was low. In our experiment, there were no significant differences between the young and developmentally mature osteocytes in the fitted model cells (Fig. 4C), while there was a clear and significant difference in the nonfitted model cells (Fig. 4D). A reasonable assumption is that the fitted model cells had a low phosphorylation level; it could also indicate the passive transduction of dye molecules. The average PA value at the low permeance peak of Calcein from Eckert's study was 2.1 × 10−15 cm3/s. Heyman et al. [15] had already estimated the Cx43 channel diameter (10 × 10−8 cm) and Cx43 channel length (160 × 10−8 cm) using a dye transfer experiment. Taken together, we can assume that the PA of the processes was equal to the PA of interface (details in Section 2.8). The combined relationship is shown in the formula below [32]: PAinterface ¼ Npore  PApore

ð3Þ

where Npore is the number of open channels on the interface between two processes; PAinterface is equal to the process permeability; and PApore indicates the single gap junction channel permeability. Based on Eq. 3, the number of open Cx43 channels in interface can be calculated. The combination of young and old fitted model cells was 149.4; this sums up all of the contributing processes. 4.5. Indirect evidence of the active transduction of molecules through the Cx43 channel Many studies have suggested that the permeability of Cx43 to dyes depends more on the probe size than its charge [2,15,29]. In our experiment, 52.78% of the recovered cells fitted our simple diffusion mathematic model (Table 3) – possibly because an even channel had the most productive performance in the case of simple diffusion. However, 47.22% of the cells did not fit the model (Table 3) and possibly showed ionophoretic current-aided transfer (Fig. 4B). There is a hypothesis about this phenomenon. To explain the very high single-channel permeance found in Xenopus oocytes by an ab initio pore model, Nitsche et al. [32] introduced an affinity term to their diffusion model. This assumption seems plausible, since larger solutes can be expected to interact more strongly with the pore structure than small ions. The structural features at the pore wall or at the pore entrance could create preferential “binding sites” or energy wells that could enhance the partitioning of the solute into the pore. However, further study is required to gain an accurate understanding of the mechanism underlying this phenomenon. 5. Conclusions The present study suggested that in ex vivo embryonic chick calvaria: • GJIC was decreased during the developmental maturation of osteocytes. • GJIC between osteocytes could be regulated and almost half of the gap junction channels have the same permeability as the cell processes for Calcein transfer. • Indirect evidence of the active transduction of molecules through the Cx43 channel.

Z. Wang et al. / Bone 91 (2016) 20–29

Acknowledgements We thank Dr. Yoshitaka Kameo, Department of Biomechanics, Research Center for Nano Medical Engineering, Institute for Frontier Medical Sciences, Kyoto University, for fruitful discussions on the mathematical model of simple diffusion. This work was supported by the Japan Society for the Promotion of Science in the form of a Grant-inAid for Scientific Research (#25293419, #16H05549).

[23]

[24]

[25] [26]

References [1] M. Almeida, L. Han, M. Martin-Millan, L.I. Plotkin, S.A. Stewart, P.K. Roberson, S. Kousteni, C.A. O'Brien, T. Bellidlo, A.M. Parfitt, R.S. Weinstein, R.L. Jilka, S.C. Manolagas, Skeletal involution by age-associated oxidative stress and its acceleration by loss of sex steroids, J. Biol. Chem. 282 (2007) 27285–27297. [2] C. Bathany, D. Beahm, J.D. Felske, F. Sachs, S.Z. Hua, High throughput assay of diffusion through Cx43 gap junction channels with a microfluidic chip, Anal. Chem. 83 (2011) 933–939. [3] R.P. Biegon, M.M. Atkinson, T.F. Liu, E.Y. Kam, J.D. Sheridan, D. Judson, Permeance of novikoff hepatoma gap junctions: quantitative video analysis of dye transfer, J. Membr. Biol. 96 (1987) 225–233. [4] N. Bivi, K.W. Condon, M.R. Allen, N. Farlow, G. Passeri, L.R. Brun, Y. Rhee, T. Bellido, L.I. Plotkin, Cell autonomous requirement of connexin 43 for osteocyte survival: consequences for endocortical resorption and periosteal bone formation, J. Bone Miner. Res. 27 (2012) 374–389. [5] F.F. Bukauskas, Conductance and permeability of the residual state of connexin43 gap junction channels, J. Gen. Physiol. 119 (2002) 171–185. [6] A.M. Buo, J.P. Stains, Gap junctional regulation of signal transduction in bone cells, FEBS Lett. 588 (2014) 1315–1321. [7] J.J. Cao, P. Kurimoto, B. Boudignon, C. Rosen, F. Lima, B.P. Halloran, Aging impairs IGF-I receptor activation and induces skeletal resistance to IGF-I, J. Bone Miner. Res. 22 (2007) 1271–1279. [8] M. Capulli, R. Paone, N. Rucci, Osteoblast and osteocyte: games without frontiers, Arch. Biochem. Biophys. 561 (2014) 3–12. [9] S.B. Doty, Morphological evidence of gap junctions between bone cells, Calcif. Tissue Int. 33 (1981) 509–512, http://dx.doi.org/10.1007/BF02409482. [10] R. Eckert, Gap-junctional single-channel permeability for fluorescent tracers in mammalian cell cultures, Biophys. J. 91 (2006) 565–579. [11] T.A. Franz-Odendaal, B.K. Hall, P.E. Witten, Buried alive: how osteoblasts become osteocytes, Dev. Dyn. 235 (2006) 176–190. [12] R. Garzon, G.A. Calin, C.M. Croce, MicroRNAs in cancer, Annu. Rev. Med. 60 (2009) 167–179. [13] D.C. Genetos, Z. Zhou, Z. Li, H.J. Donahue, Age-related changes in gap junctional intercellular communication in osteoblastic cells, J. Orthop. Res. 30 (2012) 1979–1984. [14] A. Grey, Q. Chen, X. Xin, K. Callon, J. Cornish, Parallel phosphatidylinositol-3 kinase and p42/44 mitogen-activated protein kinase signaling pathways subserve the mitogenic and antiapoptotic actions of insulin-like growth factor I in osteoblastic cells, Endocrinology 144 (2003) 4886–4893. [15] N.S. Heyman, J.M. Burt, Hindered diffusion through an aqueous pore describes invariant dye selectivity of Cx43 junctions, Biophys. J. 94 (2008) 840–854. [16] A. Himeno-Ando, Y. Izumi, A. Yamaguchi, T. Iimura, Structural differences in the osteocyte network between the calvaria and long bone revealed by three-dimensional fluorescence morphometry, possibly reflecting distinct mechano-adaptations and sensitivities, Biochem. Biophys. Res. Commun. 417 (2012) 765–770. [17] Y. Ishihara, H. Kamioka, T. Honjo, H. Ueda, T. Takano-Yamamoto, T. Yamashiro, Hormonal, pH, and calcium regulation of connexin 43-mediated dye transfer in osteocytes in chick calvaria, J. Bone Miner. Res. 23 (2008) 350–360. [18] R.L. Jilka, M. Almeida, E. Ambrogini, L. Han, P.K. Roberson, R.S. Weinstein, S.C. Manolagas, Decreased oxidative stress and greater bone anabolism in the aged, when compared to the young, murine skeleton with parathyroid hormone administration, Aging Cell 9 (2010) 851–867. [19] S.J. Jones, C. Gray, H. Sakamaki, M. Arora, A. Boyde, R. Gourdie, C. Green, The incidence and size of gap junctions between the bone cells in rat calvaria, Anat. Embryol. (Berl) 187 (1993) 343–352. [20] H. Kamioka, Y. Sugawara, T. Honjo, T. Yamashiro, T. Takano-Yamamoto, Terminal differentiation of osteoblasts to osteocytes is accompanied by dramatic changes in the distribution of actin-binding proteins, J. Bone Miner. Res. 19 (2004) 471–478. [21] H. Kamioka, Y. Ishihara, H. Ris, S.A. Murshid, Y. Sugawara, T. Takano-Yamamoto, S.S. Lim, Primary cultures of chick osteocytes retain functional gap junctions between osteocytes and between osteocytes and osteoblasts, Microsc. Microanal. 13 (2007) 108–117. [22] H. Kamioka, S.A. Murshid, Y. Ishihara, N. Kajimura, T. Hasegawa, R. Ando, Y. Sugawara, T. Yamashiro, A. Takaoka, T. Takano-Yamamoto, A method for observing

[27] [28]

[29] [30]

[31]

[32]

[33] [34] [35] [36] [37]

[38] [39] [40]

[41]

[42]

[43]

[44]

[45]

[46]

[47] [48]

[49]

29

silver-stained osteocytes in situ in 3-microm sections using ultra-high voltage electron microscopy tomography, Microsc. Microanal. 15 (2009) 377–383. D.W. Laird, M. Castillo, L. Kasprzak, Gap junction turnover, intracellular trafficking, and phosphorylation of connexin43 in brefeldin A-treated rat mammary tumor cells, J. Cell Biol. 131 (1995) 1193–1203. R. Lin, B.J. Warn-Cramer, W.E. Kurata, A.F. Lau, v-Src phosphorylation of connexin 43 on Tyr247 and Tyr265 disrupts gap junctional communication, J. Cell Biol. 154 (2001) 815–827. X. Liu, L. Sun, M. Torii, P. Rakic, Connexin 43 controls the multipolar phase of neuronal migration to the cerebral cortex, Proc. Natl. Acad. Sci. 109 (2012) 8280–8285. X. Ma, M. Kumar, S.N. Choudhury, L.E. Becker Buscaglia, J.R. Barker, K. Kanakamedala, M.F. Liu, Y. Li, Loss of the miR-21 allele elevates the expression of its target genes and reduces tumorigenesis, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 10144–10149. X. Ma, L.E. Becker Buscaglia, J.R. Barker, Y. Li, MicroRNAs in NF-kaapaB signaling, J. Mol. Cell Biol. 3 (2011) 159–166. S. Machtaler, K. Choi, M. Dang-Lawson, L. Falk, F. Pournia, C.C. Naus, L. Matsuuchi, The role of the gap junction protein connexin43 in B lymphocyte motility and migration, FEBS Lett. 588 (2014) 1249–1258. S.L. Mills, S.C. Massey, The kinetics of tracer movement through homologous gap junctions in the rabbit retina, Vis. Neurosci. 15 (1998) 765–777. A. Moreno, G. Fishman, D. Spray, Phosphorylation shifts unitary conductance and modifies voltage dependent kinetics of human connexin43 gap junction channels, Biophys. J. 62 (1992) 51–53. A.P. Moreno, J.C. Sáez, G.I. Fishman, D.C. Spray, Human connexin43 gap junction channels. Regulation of unitary conductances by phosphorylation, Circ. Res. 74 (1994) 1050–1057. J.M. Nitsche, H.-C. Chang, P.A. Weber, B.J. Nicholson, A transient diffusion model yields unitary gap junctional permeabilities from images of cell-to-cell fluorescent dye transfer between xenopus oocytes, Biophys. J. 86 (2004) 2058–2077. D.V. Novack, Role of NF-κB in the skeleton, Cell Res. 21 (2011) 169–182. M. Owen, S. MacPherson, Cell population kinetics of an osteogenic tissue II, J. Cell Biol. 19 (1963) 33–44. R.D. Phair, Measurement of dynamic protein binding using photobleaching microscopy, Methods Enzymol. 375 (2004) 393. L.I. Plotkin, T. Bellido, Beyond gap junctions: connexin43 and bone cell signaling, Bone 52 (2013) 157–166. S. Qiu, D.S. Rao, S. Palnitkar, A.M. Parfitt, Age and distance from the surface but not menopause reduce osteocyte density in human cancellous bone, Bone 31 (2002) 313–318. H. Reers, A. Jacot, W. Forstmeier, Do zebra finch parents fail to recognize their own offspring? PLoS One 6 (2011), e18466. J.L. Solan, P.D. Lampe, Connexin43 phosphorylation: structural changes and biological effects, Biochem. J. 419 (2009) 261–272. Y. Sugawara, H. Kamioka, T. Honjo, K.I. Tezuka, T. Takano-Yamamoto, Three-dimensional reconstruction of chick calvarial osteocytes and their cell processes using confocal microscopy, Bone 36 (2005) 877–883. Y. Sugawara, R. Ando, H. Kamioka, Y. Ishihara, T. Honjo, N. Kawanabe, H. Kurosaka, T. Takano-Yamamoto, The three-dimensional morphometry and cell-cell communication of the osteocyte network in chick and mouse embryonic calvaria, Calcif. Tissue Int. 88 (2011) 416–424. S. Tatarano, T. Chiyomaru, K. Kawakami, H. Enokida, H. Yoshino, H. Hidaka, T. Yamasaki, K. Kawahara, K. Nishiyama, N. Seki, M. Nakagawa, miR-218 on the genomic loss region of chromosome 4p15.31 functions as a tumor suppressor in bladder cancer, Int. J. Oncol. 39 (2011) 13–21. M.H. Wade, J.E. Trosko, M. Schindler, A fluorescence photobleaching assay of gap junction-mediated communication between human cells, Science 232 (1986) 525–528. X.M. Wang, P.I. Terasaki, G.W. Rankin, D. Chia, H.P. Zhong, S. Hardy, A new microcellular cytotoxicity test based on calcein AM release, Hum. Immunol. 37 (1993) 264–270. L. Wang, Y. Wang, Y. Han, S.C. Henderson, R.J. Majeska, S. Weinbaum, M.B. Schaffler, In situ measurement of solute transport in the bone lacunar-canalicular system, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 11911–11916. X. Xu, R. Francis, C.J. Wei, K.L. Linask, C.W. Lo, Connexin 43-mediated modulation of polarized cell movement and the directional migration of cardiac neural crest cells, Development 133 (2006) 3629–3639. M. Yeager, Structure of cardiac gap junction intercellular channels, J. Struct. Biol. 121 (1998) 231–245. L.-D. You, S. Weinbaum, S.C. Cowin, M.B. Schaffler, Ultrastructure of the osteocyte process and its pericellular matrix, Anat. Rec. A. Discov. Mol. Cell. Evol. Biol. 278 (2004) 505–513. S.-S. Zhang, R.M. Shaw, Trafficking highways to the intercalated disc: new insights unlocking the specificity of connexin 43 localization, Cell Commun. Adhes. 21 (2014) 43–54.