Cyclic Stretch Enhances Gap Junctional Communication Between Osteoblastic Cells


  • Konstantinos Ziambaras,

    1. Division of Bone and Mineral Diseases, Departments of Internal Medicine and Cell Biology and Physiology, Washington University School of Medicine, Barnes-Jewish Hospital, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Fernando Lecanda,

    1. Division of Bone and Mineral Diseases, Departments of Internal Medicine and Cell Biology and Physiology, Washington University School of Medicine, Barnes-Jewish Hospital, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Thomas H. Steinberg,

    1. Division of Infectious Diseases, Departments of Internal Medicine and Cell Biology and Physiology, Washington University School of Medicine, Barnes-Jewish Hospital, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Roberto Civitelli M.D.

    Corresponding author
    1. Division of Bone and Mineral Diseases, Departments of Internal Medicine and Cell Biology and Physiology, Washington University School of Medicine, Barnes-Jewish Hospital, St. Louis, Missouri, U.S.A.
    • Division of Bone and Mineral Diseases, Department of Internal Medicine, Barnes-Jewish Hospital, North Campus, 216 S. Kingshighway Boulevard, St. Louis, MO 63110 U.S.A.
    Search for more papers by this author

  • Part of this work was presented in abstract form at the 18th annual meeting of the American Society for Bone and Mineral Research, Seattle, WA, U.S.A., September 7–11, 1996, abstract M341.


Mechanical loading is essential to maintain skeletal integrity. Because gap junctions in bone are affected by mechanical factors, we studied whether stretch, an anabolic stimulus for osteoblasts, modulates direct intercellular communication in these cells. Gap junctional communication during stretch was assessed using a newly developed method, the “parachute assay,” which allows monitoring of dye diffusion without disruption of the plasma membrane. Application of cyclic stretch for 2 or 24 h to well-coupled ROS 17/2.8 cells resulted in a 56.5% and 30.4% increase in dye coupling, respectively, compared with resting conditions. Stretch increased dye diffusion less dramatically (12.4% compared with unstimulated cells) in the poorly coupled UMR 106-01 cells. The stretch-induced increase of cell coupling was abolished in the presence of the gap junctional inhibitor, heptanol. Steady-state mRNA levels of connexin43 (Cx43), the gap junction protein that mediates cell-to-cell diffusion of negatively charged dyes between osteoblasts, were not different between control and stretched ROS 17/2.8 or UMR 106-01 cultures after various periods of cyclic stretch. However, phosphorylated forms of Cx43 protein were more abundant in stretched ROS 17/2.8 than in controls. This was associated with increased punctate Cx43-specific immunostain at appositional membranes of stretched cells. Thus, cyclic stretch increases gap junctional communication between osteoblastic cells by modulating intracellular localization of Cx43.


Mechanical factors are essential for the development and maintenance of skeletal integrity. Absence of gravity or mechanical load results in a very rapid and severe uncoupling between bone formation and resorption, with fast bone loss.(1,2) Although weightlessness increases osteoclast number and bone resorption in short-term flights,(3,4) for longer periods of weightlessness bone turnover decreases.(5) Growing rats in microgravity conditions experience a remarkable reduction of periosteal bone formation,(6–8) and a decrease of osteocalcin and type I collagen is observed after 7 days of space flight.(9) Therefore, removal of physical strain decreases osteoblast function. However, application of mechanical strain stimulates osteoblast activity. Using the commercially available equipment, Flexercell, Buckley et al.(10) observed an increase of alkaline phosphatase (ALP) activity and collagen synthesis in chick osteoblastic cells subjected to physical strain. These results were later confirmed in the human osteosarcoma cells, OHS-4, which responded to cyclic mechanical strain with an increased production and secretion of type I collagen and osteopontin.(11) Other types of mechanical forces are also effective stimulators of osteoblasts. Increased ALP activity and osteopontin expression have been obtained by application of hydrodynamic force in the form of compressed air to ROS 17/2.8 cell cultures.(12)

The mechanisms by which mechanical stimulation of osteoblasts is transduced into intracellular signals and physiologic effects are not well understood. Stretch-activated membrane channels have been demonstrated in osteoblasts,(13) and chronic cyclic mechanical strain, applied using the Flexercell system, increases the sensitivity and the number of open cation-selective channels in the osteoblast-like UMR 106-01 cells.(14) Transient increases of inositol polyphosphates have also been observed within 20 s to 2 minutes after the onset of mechanical stimulation,(15) and intercellular “calcium waves” can be induced in osteoblastic networks by mechanical stimulation of the plasma membrane of a single cell.(16) Although calcium waves in osteoblastic networks can be propagated via diverse mechanisms, some types of calcium waves require the presence of functional gap junctions.(16,17) Interestingly, earlier work in animals suggested that the number of gap junctions in bone declines in weightlessness conditions,(18) although a more recent preliminary report indicates an increase of Cx43 in unloaded limbs.(19) Therefore, regulation of cell-to-cell communication via gap junctions is conceivably of critical importance for the elaboration and propagation of intercellular signals generated by mechanical stimuli in osteoblast networks.(13,20)

These observations raise the possibility that mechanical stimulation may not only produce signals that can be propagated via gap junctions, but it can also modulate the function of gap junctions. Gap junctions are clusters of aqueous intercellular channels formed by hexamers of gap junction proteins, connexins. In bone, gap junctions are formed primarily by connexin43 (Cx43), and these channels allow diffusion of negatively charged dyes from cell to cell (dye coupling).(21–23) Because mechanical stimulus applied in the form of cyclical stretch was found to be anabolic for osteoblasts,(11) we asked whether cyclic stretch regulates gap junctional intercellular communication among osteoblastic cells. We found that this type of mechanical stimulation rapidly up-regulates gap junctional communication via mechanisms that involve modulation of intracellular processing of Cx43 protein.


Reagents and cells

Calcein acetoxymethyl ester (calcein-AM) was obtained from Molecular Probes (Eugene, OR, U.S.A.), dissolved in dimethyl sulfoxide (DMSO) to a concentration of 1 mg/ml (1 mM), aliquoted, and stored at −20°C in the dark. The membrane permanent dye, PKH-26 was purchased from Zynaxis Cell Science (Malvern, PA, U.S.A.), and it was dissolved in aqueous solutions, according to the manufacturer's instructions, as described below. Heptanol was dissolved in ethanol (1:4) and prepared fresh just before use. The mixture was added directly to the culture media, at a final concentration of 3 mM heptanol and 0.15% ethanol.(21) Dextran Texas red (Molecular Probes) was dissolved in phosphate-buffered saline (PBS) at a final concentration of 10 mg/ml. Polyclonal rabbit antiserum directed against Cx43 was produced by immunizing rabbits with a synthetic peptide corresponding to amino acids 252–271 of rat Cx43.(24) All other chemicals and the tissue culture media were purchased from Sigma Chemicals (St. Louis, MO, U.S.A.), unless otherwise indicated.

The UMR 106-01 cells were a gift of Dr. Nicola C. Partridge, St. Louis University. These cells are derived from the rat osteogenic sarcoma cell line UMR 106, which has been characterized as having an osteoblastic phenotype.(25,26) They were maintained in minimum essential Eagle's medium (MEM), supplemented with 10% heat-inactivated fetal bovine serum (FBS; Summit Biotechnology, Greely, CO, U.S.A.), penicillin, and streptomycin. Subcultures until passage 30, relative to the original UMR 106-01 subclone,(25) were used in these studies. The ROS 17/2.8 cell line was provided by Dr. Gideon Rodan (Merck Research Laboratories, West Point, PA, U.S.A.) and were also maintained in MEM, supplemented with 10% FBS and antibiotics. They also express several osteoblastic features, including production of osteocalcin and other matrix proteins.(27,28)

Mechanical strain

The Flexercell Strain unit (Flexcell Corp., McKeesport, PA, U.S.A.) was used for application of mechanical stretch to osteoblasts.(29–31) Cells to be subjected to stretch were grown on specially designed six-well tissue culture dishes with collagen-coated flexible silicone bottoms (Flex I, Flexcell). Through an air pump, a negative pressure of 12 kPa was applied to the flexible bottomed wells at three cycles per minute. During each cycle, a 10-s stretch period was followed by a 10 s relaxation. As the flexible bottoms are pulled downward by the negative pressure, the cells attached to their upper surface are stretched by the deformation of the rubber.(30,32) Control cultures were grown under the same conditions on dishes of the same size containing a collagen-coated silicone disc identical to the one used for stretch sitting on a rigid plastic bottom (Flex II, Flexcell). One of the limitations of the Flexercell instrument is that the magnitude of the strain delivered across the surface of the plate is nonuniform.(32) Based on the strain profile reported for the Flex I dishes at 22 kPa,(11) the degree of strain applied to the cells using 12 kPa negative pressure can been calculated to range between ∼160,000 microstrain (μE) at the edge to less than 20,000 μE at the center of the well, and the average strain along the radius of the dish is approximately 50,000 μE. Because microstrain is defined as a millionth of the change in length divided by the initial length of the cell, cells in the concentric area between 0.5 and 2 cm from the center are subjected to ∼5% average elongation, with ∼12% maximal elongation at the edge, down to a minimum 1–2% elongation at the center of the dish.(11)

Dye coupling

To monitor the diffusion of fluorescent molecules through gap junctions (dye coupling) while stretch is applied, we developed a new technique, the “parachute” assay, which allows the study of intercellular dye transfer without requiring micromanipulation. Microinjections cannot be performed while cells are stretched in the Flexercell apparatus, and therefore this method is unsuitable for monitoring effects on gap junction function during application of the mechanical stimulus. The principle of the parachute assay is similar to methods described by Tomasetto et al.(33) and Goldberg et al.(34) and consists in preloading cells with a gap junction permeant dye, such as calcein, using the acetoxymethyl ester derivative of the dye, and allowing the loaded cell to adhere to cells in an unloaded monolayer. After a short time, the fluorescent dye will pass from the “donor” cell to “acceptor” cells. Second, third, and higher order cells will also take the dye, depending upon the degree of coupling, which is monitored using two different techniques, as explained below.

Acceptor cells were grown in the flexible bottom or control plates to confluence. On the day of the experiment, cells from a parallel culture grown in a 100-mm dish were loaded with calcein-AM (donor cells), by incubating in freshly made solution of 2 μM calcein-AM and 2.5 mM probenecid in growth medium, for 45 minutes, at 37°C and pH 7.4. Probenecid, an organic anion transport blocker, was used to prevent sequestration of calcein within cytoplasmic vacuoles that may affect its transfer rate from cell to cell.(35) Unincorporated dye was removed by three consecutive washes in culture medium. The cells were harvested in a single cell suspension by trypsinization and transferred to a polypropylene tube. After centrifugation, the cell pellet was gently resuspended in serum-containing medium. Then, a small number of the calcein-loaded donor cells (∼10 × 103/well) was added to the monolayer of acceptor cells in flexible and control dishes. The “parachuted” donor cells attach to the cell monolayer within 10–15 minutes. Thereafter, the flexible-bottomed plates were seated on the gaskets of the Flexercell baseplate and maintained in a humidified atmosphere of 95% air and 5% CO2, at 37°C in a cell culture incubator. Stretch was applied at a rate of 3 cycles/minute for 2 h as detailed above. In most experiments, calcein fluorescence was monitored by epifluorescence microscopy (Zeiss Axioscope, Zeiss, Jena, Germany) using a fluorescein filter set. Dye coupling was then assessed in digitized images of several microscopic fields taken from each well within 30 minutes after the 2 h incubation by counting the number of cells acquiring dye per parachuted donor cell. Particular care was taken to exclude from the analysis those cells that were located near the edge of the dish, the site where the flexible bottoms are attached to the plate, so that the cells are either not stretched or receive the maximal degree of stretch. Likewise, cells grown in the center of the plates, where the strain is minimal, were also excluded from analysis of dye coupling (see above). For the 24-h stretch experiments, cells were subjected to stretch for 22 h before parachuting calcein-loaded donor cells. After allowing the donor cells to settle, the cultures were returned to the Flexercell apparatus for the remaining 2 h of stimulation. Thus, in all experiments the number of coupled cells was estimated approximately between 2.25 and 3 h after parachuting the donor cells.

Dye coupling also was assessed by fluorescence activated cell sorting (FACS), using modifications of a method we have previously described.(36) For these experiments, the acceptor cells were prelabeled with the membrane permanent, nontransferable dye PKH-26 on the day before the parachute assay. PKH-26 labeling was performed according to the manufacturer's recommendations by exposing a cell suspension to 2 μM PKH-26 for 2 minutes at room temperature. After the addition of heat-inactivated FBS to terminate the reaction, cells were replated in either flexible or control bottom plates at a density of about 7 × 105 cells/cm2, which yields confluent cultures after about 24 h. These PKH-26–labeled cultures constituted the acceptor cells, on which calcein-loaded donor cells were parachuted using the same procedures and timing described above. To assess the degree of dye coupling, after the appropriate stretch periods, cells were harvested in a single cell suspension by trypsin digestion, and fluorescence was detected in a Coulter EPICS/XL flow cytometer (Coulter, Irving, TX, U.S.A.), using a 525-nm band pass filter (green) for calcein and a 575-nm band pass filter (red) for PKH-26. The gates for each channel (green and red) were established using a suspension of calcein-loaded and PKH-26–labeled cells mixed in equal amounts just prior to the measurement, before any dye diffusion may have occurred. This also ensured that a good separation of the two populations was obtained in each experiment. Because with this method all donor and acceptor cells are counted, the number of acceptor cells that will ultimately take the dye depends on the initial donor/acceptor ratio. We found that a 1:10 donor/acceptor ratio was optimal for ROS 17/2.8 cells cultured on the tissue culture dishes used in this study, in that it provided adequate amounts of cells for flow cytometry, and at the same time it allowed good separation of parachuted cells, thus minimizing overlap of dye diffusion from more than one cell. Therefore, in this condition, dye transfer is mostly dependent on second-order transfer of calcein between acceptor cells.(36)

RNA blots

RNA blots were performed as previously described.(21) Briefly, cells were grown in flexible and control dishes at different densities, so that at the end of the stretching period they would reach confluence. After application of stretch for 2, 6, or 24 h, total cellular RNA was isolated using guanidinium isothiocyanate and separated on formaldehyde agarose gels. The gels were blotted onto nylon membranes and UV cross-linked. The membranes were hybridized using32P-labeled cDNA probe for Cx43 and washed twice under low and one time under high stringency conditions for 30 minutes each time.(21) The relative amount of Cx43 mRNA was quantitated by densitometric analysis of the autoradiographic bands, after normalization for the intensity of glyceraldehyde-3-phosphate dehydrogenase bands (Clontech, Palo Alto, CA, U.S.A.).


After mechanical stretch, cells cultured on flexible and control plates were washed twice with PBS and scraped into a solution of PBS containing a cocktail of protease and phosphatase inhibitors (1 mM phenylmethylsulforyl fluoride [PMSF], 1 mM sodium orthovanadate (Na2VO4), 10 mM sodium fluoride (NaF), 10 mM N-ethylmaleimide (NEM), 2 μg/ml leupeptin, 1 μg/ml pepstatin), at 4°C and sonicated twice for 30 s.(21,37) Protein concentration in each sample was determined before electrophoresis using the method of Bradford,(38) and appropriate dilutions were made to ensure that equal amounts of protein were loaded in each lane. Proteins were separated by sodium dodecyl sulfide polyacrylamide gel electrophoresis (SDS-PAGE) on 10% polyacrylamide gel, using Promega molecular weight protein standards (Madison, WI, U.S.A.), and transferred to polyvinylidene fluoride (PDVF) membranes (Immobilon P, Millipore Corp., Bedford, MA, U.S.A.) using a tank transfer apparatus (Trans-blot Cell, Bio-Rad, Richmond, CA, U.S.A.). After blocking with 5% nonfat milk in PBS containing 0.1% Tween-20, pH 7.4, overnight, the membranes were incubated with anti-Cx43 antibody (1:1000 dilution) in blocking buffer for 2 h at room temperature, then washed in PBS, and incubated for 1 h with horseradish peroxidase–conjugated goat anti-rabbit IgG (Tago, Burlingame, CA, U.S.A.). The immune reaction was detected by exposing the membranes to autoradiography film (Hyperfilm, Amersham Int., Buckinghamshire, U.K.) in the presence of luminol using the ECL kit (Amersham). In some experiments, gap junction proteins were partially purified by extraction in alkaline conditions, using a method described by Hertzberg,(39) with modifications. Briefly, cells were scraped from five confluent flexible and control dishes in PBS with 1% PMSF, centrifuged, resuspended in 1 ml of 1 mM NaHCO3 containing protease and phosphatase inhibitors as mentioned above, and 1 M NaOH to reach pH 8.0 at 4°C. The suspension was then sonicated for 30 s, incubated on ice for 60 minutes, and then centrifuged at 35,000g for 30 minutes. The resulting pellet was resuspended in bicarbonate buffer for protein determination and PAGE as described above.


After application of stretch, cells were fixed in methanol/acetone (1:1) for 2 minutes at room temperature, washed, and incubated in PBS containing 2% heat-inactivated goat serum and 0.5% Triton X-100 for 15 minutes.(21,37) The samples were then incubated with anti-Cx43 serum (1:500 dilution) for 45 minutes at room temperature. After three washes in PBS, rhodamine-conjugated goat anti-rabbit IgG antibody (Boehringer Mannheim Biochemicals, Indianapolis, IN, U.S.A.) was added (1:500 dilution) for 45 minutes at room temperature. Cells were subsequently washed three times in PBS, and the silicone bottom of each well was removed and immediately examined by epifluorescence microscopy.

Statistical analysis

Data represent average ± SD, unless otherwise indicated. When normally distributed, groups means were compared by Student's t-test for unpaired samples. Because dye coupling expressed as a number of coupled cells per cell yields data sets highly skewed toward low coupling (standardized skewness > 2 in all conditions), the nonparametric Mann–Whitney (Wilcoxon) test was used to compare group medians, and the Kolmogorov–Smirnov test was employed to assess equality of frequency distributions. Data were analyzed using the statistical software package, Statgraphics Plus for Windows 2.1 (Manugistic, Inc., Rockville, MD, U.S.A.).


In resting, control conditions ROS 17/2.8 cells, which express abundant Cx43 on their cell surface,(37) exhibited a high degree of cell coupling with ample diffusion of calcein to second- and third-order neighbors (Fig. 1A). Application of stretch for either 2 or 24 h increased the extension of calcein diffusion to cells further removed from the parachuted cell (Figs. 1B and 1C). In all the micrographs of Figs. 1 and 3, the “parachuted,” calcein-loaded, donor cells are easily distinguishable from the acceptor cells that have taken the dye because of their higher brightness and the different plane of focus which blurs their fluorescence. In most cases, calcein spread radially from the parachuted cell, although in some instances dye diffusion appeared more directional, but this was unrelated to application of stretch. Dye diffusion primarily occurred between two cell bodies, although occasionally diffusion occurred via cytoplasmic processes. Overall, control donor ROS 17/2.8 cells diffused dye to an average of 6.9 ± 5.8 acceptor cells (n = 1414, median = 6.0), whereas after application of cyclic stretch for 2 h calcein diffused to an average of 10.8 ± 9.6 acceptor cells (n = 469; median = 8.0), corresponding to a highly significant (p < 0.001) 56.5% increase of dye coupling. As is evident from Fig. 1, dye coupling was still higher than baseline after 24 h of cyclic stretch, with an average of 9.0 ± 8.0 coupled cells per donor cell (n = 1203, median = 7.0). Thus, although the degree of dye coupling was lower at 24 h than after 2 h of stretch (p = 0.001), it remained ∼30% higher as compared with resting conditions (p < 0.001), suggesting that part of the stimulatory effect is maintained for as long as mechanical strain is present.

Figure FIG. 1.

Stretch increases calcein diffusion among ROS 17/1.8 osteoblastic cells. A small number of calcein-labeled ROS 17/2.8 cells were “parachuted” on a monolayer of unlabeled cells grown on either (A) rigid-bottomed control wells, or flexible-bottomed wells, which were subjected to cyclic mechanical stretch for (B) 2 or (C) 24 h. Fluorescence was detected by epifluorescence microscopy (×10 objective magnification) using a fluorescein filter set. Bar = 40 μm.

Figure 2 illustrates the same data on ROS 17/2.8 cells plotted as frequency histograms of the number of coupled cells per donor cell. Compared with unstimulated cells, there was a clear shift toward higher coupling in the frequency histograms after 2 and 24 h of stretch, although the shift was clearly more prominent at the earlier time point (Fig. 2). In both cases, the distributions were significantly different relative to control conditions (p < 0.001, Kolmogorov–Smirnov test). In more simplified terms, approximately 26% of control ROS 17/2.8 cells were coupled to more than 10 cells, and 5% to more than 20 cells, whereas after 2 h stretch 41% of the parachuted cells were coupled to >10 cells, and 15% to >20 cells. Therefore, more ROS 17/2.8 cells were coupled to a high number of cells than in control conditions.

Figure FIG. 2.

Frequency distribution of dye coupling as assessed by the “parachute” assay in control and stretched ROS 17/2.8 cells. Calcein-labeled ROS 17/2.8 cells were “parachuted” on a monolayer of unlabeled cells grown on either rigid-bottomed control wells, or flexible-bottomed wells, which were subjected to cyclic mechanical stretch for 2 or 24 h. Dye coupling was assessed by counting the number of coupled cells to each parachuted cell and normalizing to 100 for each condition. Data are derived from a total of 1414, 469, and 1203 parachuted cells for control, 2 and 24 h stretch, respectively. The 2 h stretch and 24 h stretch distributions were significantly different than the control distribution by Kolmogorov–Smirnov test (p < 0.001).

In agreement with previous studies,(36,37) we found that UMR 106-01 cells were coupled to fewer neighboring cells (5.2 ± 3.7, n = 492, median = 4.0) than were ROS 17/2.8 (Fig. 3). A 2-h stretch increased calcein transfer by 12.4% compared with unstimulated cells, to an average of 5.9 ± 4.8 (n = 642, median = 5.0) coupled cells per donor cell, a difference that was not significant (p > 0.10). Although the frequency histograms of control and stretched UMR 106-01 cells appeared very similar (Fig. 4), they were found to be significantly unequal by the Kolmogorov–Smirnov test (p < 0.001). We interpret these results to indicate that UMR 106-01 cells, which express abundant Cx45 and very little Cx43,(37) responded less vigorously to mechanical stimulation than did ROS 17/2.8 cells. The effect of stretch on these cells may cause a significant shift of coupling toward higher values when a large number of cells are scored, but it is insufficient to alter the average number or median of coupled cells, probably because of the skewness of the variable measured.

Figure FIG. 3.

Stretch increases calcein diffusion among UMR 106-01 osteoblastic cells. A small number of calcein-labeled UMR 106-01 cells were “parachuted” on a monolayer of unlabeled cells grown on either (A) rigid-bottomed control wells, (B) or flexible-bottomed wells, which were subjected to cyclic mechanical stretch for 2 h. Fluorescence was detected by epifluorescence microscopy (×10 objective magnification) using a fluorescein filter set, immediately after application of stretch. Bar = 40 μm.

Figure FIG. 4.

Frequency distribution of dye coupling as assessed by the “parachute” assay in control and stretched UMR 106-01 cells. Calcein-labeled UMR 106-01 cells were “parachuted” on a monolayer of unlabeled cells grown on either rigid-bottomed control wells, or flexible-bottomed wells, which were subjected to cyclic mechanical stretch for 2 h. Dye coupling was assessed immediately after application of stretch by counting the number of coupled cells to each parachuted cell, and normalizing to 100 for each condition. Data are derived from a total of 492 and 642 parachuted cells for control and 2 h stretch, respectively. The 2 h stretch and control distributions were significantly different by Kolmogorov–Smirnov test (p < 0.001).

In both control and 2-h stretched ROS 17/2.8 and UMR 106-01 cells, calcein transfer was almost completely abolished (more than 90% of cells remained uncoupled) when heptanol, an inhibitor of gap junctional communication, was added just after parachuting (data not shown). However, Dextran Texas red (Molecular Probes), a large (∼10,000 Da) molecular weight dye impermeable to gap junctions did not pass from scrape-loaded, parachuted cells to acceptor cells (manuscript in preparation), further proving that the parachute assay reports gap junctional permeability, rather than membrane fusion or other phenomena. Thus, the increased calcein diffusion observed after application of stretch was the consequence of increased gap junctional communication.

To corroborate these findings, the degree of cell coupling was also assessed after stretch in ROS 17/2.8 cells by FACS. In the plots of Fig. 5, the PKH-26–stained and calcein-loaded cells are located in quadrants 1 and 4, respectively, and they are fully separated when mixed just before fluorescence measurement (baseline). In cell suspensions obtained after parachuting calcein-loaded cells on a PKH-26–labeled cell layer and stretching for 2 h, a third population of cells appears in quadrant 2. This represents acceptor cells that have taken calcein from donor cells. The number of these double-labeled cells (quadrant 2) as a percent of the total number of potential acceptor, PKH-26–labeled cells (quadrants 2 + 1) is defined as the transfer ratio, and it represents a direct measurement of the degree of dye coupling. Consistent with the data obtained by direct cell counting, application of stretch to ROS 17/2.8 cell cultures for 2 h resulted in a significantly larger double-loaded cell population (transfer ratio, 29.4 ± 9.9%) compared with unstimulated conditions (transfer ratio, 12.4 ± 3.7%) (Fig. 5). This corresponds to a 2.4-fold increase in dye coupling (p < 0.001, n = 12). It is important to recognize that the degrees of dye coupling obtained from the two scoring methods (i.e., direct cell counting and flow cytometry) are not directly comparable in numerical terms. In particular, the transfer ratio obtained by FACS is strictly dependent on the initial donor/acceptor ratio, a factor that is less relevant when scoring coupling on the tissue culture dish. Furthermore, in the latter case cell coupling is estimated in a certain number of randomly chosen microscopic fields, and in selected areas of the dish, whereas the FACS assay reports the percentage of all acceptor cells that have finally received dye. While on the one hand scoring all cells for coupling removes any potential biases in counting, on the other hand lack of selectivity may dilute the effect of stretch in the flow cytometry method, considering the heterogeneity of microstrain produced by deformation of the bottom in different points of the flexible dish. In any case, these data indicate that regardless of the scoring method used, mechanical strain increases the ability of osteoblastic cells to diffuse calcein from cell to cell.

Figure FIG. 5.

FACS analysis of calcein transfer in control and stretched ROS 17/2.8 cells. Cells loaded with calcein (green) were parachuted on a layer of PKH-26-labeled (red) cells grown on either rigid-bottomed (control) or flexible-bottomed (stretch) plates, to a ratio of 1:10 donor (green) to acceptor (red) cells. After application of stretch for 2 h, cells were released from the dish and subjected to FACS. Separate aliquots of calcein-loaded and PKH-26–labeled cells were mixed in equal amounts just before flow cytometry (baseline) to define the gates for the green (x-axis) and the red channels (y-axis). PKH-26–labeled cells are represented by the dots in quadrant 1, and calcein-loaded cells are in quadrant 4. In quadrant 2 are PKH-26–labeled cells that have taken calcein from donor cells via gap junctions.

Because Cx43 mediates dye coupling in ROS 17/2.8 and UMR 106-01 cells, we then asked whether the effect of mechanical stimulation of gap junctional communication was related to regulation of Cx43 expression. This expectation was also borne out of the observation that the effect of stretch was clearly less evident in the UMR 106-01 cells, which expresses primarily Cx45 and little Cx43. However, there was no difference in steady-state Cx43 mRNA levels between ROS 17/2.8 in resting conditions and after cyclic stretch for 2, 6, or even 24 h. Likewise, changes in Cx43 mRNA were not detected after a 2-h stretch applied to UMR 106-01 cells (Fig. 6).

Figure FIG. 6.

Stretch does not alter Cx43 mRNA levels in osteoblastic cells. Confluent cultures of either ROS 17/2.8 (left) or UMR 106-01 cells (right) were grown on either rigid-bottomed (control) or flexible-bottomed plates which were subjected to mechanical strain (stretch) for the indicated times. Total RNA was extracted, separated in agarose gel, and blotted onto nylon membranes. Membranes were hybridized with32P-labeled cDNA probe for Cx43, then washed and rehybridized with a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe, to control for RNA abundance and integrity.

To determine whether the effect of mechanical stimulation could be related to post-transcriptional regulation of Cx43 protein, immunoblots of ROS 17/2.8 extracts after 2 h of cyclic stretch were performed. In both whole cell lysates and alkali extracted material, two specific bands migrating at around 41 and 43–44 kDa were detected, and the intensity of both of these bands increased after 2 h of mechanical stimulation (Fig. 7). A faster migrating band (∼40 kDa) was also present in the whole cell lysates, and it was detectable in the alkali extracted material only after prolonged exposure of the autoradiograph. A corresponding very faint band was also present in the preimmune lane, although the signal in the Cx43 blots (whole cell lysates) was clearly much greater (Fig. 7). Based on previous observations in the same and other cell lines, the ∼41 and ∼43 kDa bands represent phosphorylated Cx43, whereas the faster migrating band corresponds to nonphosphorylated Cx43.(21,37) By densitometric analysis, the intensity of the two slowly migrating bands was ∼40% and ∼30% higher than in nonstimulated cells after 2 and 24 h stretch, respectively, whereas the intensity of the ∼40 kDa band was not different. The degree of change in the higher molecular weight bands was similar using either extraction method. These data suggest that mechanical stimulation increases the abundance of Cx43 protein, and primarily of phosphorylated forms of Cx43, without changes in total mRNA expression.

Figure FIG. 7.

Mechanical stretch increases Cx43 protein levels in ROS 17/2.8 cells. Confluent cultures of ROS 17/2.8 cells were grown on either rigid-bottomed (control) or flexible-bottomed plates which were subjected to mechanical strain (stretch) for 2 or 24 h, as indicated. Whole cell lysates or alkali extracted material were separated by PAGE and blotted with a Cx43-specific antibody. A lane (third from left) with whole cell lysate blotted with preimmune serum is also shown. The 24-h blot was run on a separate gel.

Consistent with this conclusion, immunofluorescence staining revealed a higher abundance of Cx43-specific fluorescence signal in ROS 17/2.8 cultures subjected to a 2-h cyclic stretch compared with nonstretched cells (Fig. 8). Careful inspection of the fluorescence micrographs also revealed that in cells subjected to stretch the fluorescent signal was most frequently localized at appositional membranes in a linear, punctate distribution, whereas in control cells the less abundant Cx43-specific stain was not as clearly localized as in stretched cells, and it was frequently observed in perinuclear or in other cytoplasmic areas (Fig. 8). Although in a qualitative fashion, these immunofluorescence images indicate that application of mechanical strain increases Cx43 abundance on the cell surface of adjoining ROS 17/2.8 cells, compatible with a mechanically induced increase of functional gap junctions.

Figure FIG. 8.

Mechanical stretch increases the abundance of Cx43-specific immunofluerescence staining in ROS 17/2.8 cells. Cells were grown on either rigid-bottomed (A and C) or flexible-bottomed plates, which were subjected to mechanical strain for 2 h (B and D). After fixation, cells were immunostained with rabbit anti-Cx43 antibody, followed by incubation with rhodamine-conjugated goat anti-rabbit IgG. Arrows indicate linear, punctate distribution of Cx43-specific stain at appositional membranes; arrowheads point to cytoplasmic and perinuclear stain. Bar = 20 μm.


This report demonstrates that short-term application of cyclic mechanical strain increases gap junctional intercellular communication among osteoblastic cells. This effect is independent of Cx43 gene expression, presumably the result of increased Cx43 assembly into gap junctions.

To study dye coupling in osteoblast-like cells during application of mechanical strain, we have developed and applied a new technique, the “parachute assay.” The development of a novel approach was required to overcome the limitations of the currently used methods to assess gap junctional communication based on microinjection of a fluorescent tracer into one cell. Microinjections must be performed on the stage of a microscope, and therefore mechanical stimulation must be terminated several minutes before dye coupling can be assessed. This approach is thus unsuitable for detecting effects that are either very rapid or require a continuous application of the stimulus. However, with the parachute assay, dye coupling can be examined in very large numbers of cells while the mechanical stimulus is applied. In addition, the parachute assay does not require disruption of the plasma membrane. This point is generally disregarded, but it may be critical for the estimation of dye coupling when microinjections are performed. Cell impalement causes sudden and dramatic changes in intracellular homeostasis, including changes in membrane potential or intercellular ionic concentration, which in turn can alter gap junctional communication.(40,41) Such consideration may in theory provide an explanation for the relatively high dye coupling obtained with the UMR 106-01 cells using the parachute assay, as compared with previous experience with microinjections.(22,37) Conceivably, gap junctions between UMR 106-01 cells, which are formed by both Cx43 and Cx45, may be more sensitive to plasma membrane manipulation than are Cx43 junctions.

Stretch increased dye transfer primarily in the highly coupled ROS 17/2.8, which express Cx43 in abundance and no Cx45, and to a much lesser extent in the UMR 106-01 cells, which express primarily Cx45 and little Cx43. Considering that Cx43 gap junctions mediate dye coupling between osteoblasts,(21,42) whereas Cx45 channels exhibit a much lower permeability,(37) these results indicate that mechanical stimulation regulates Cx43 gap junctions in ROS 17/2.8. The situation may be more complex in the UMR 106-01, whose gap junctional permeability depends upon the interaction between the two connexins. Thus, while the effect of stretch on Cx45 gap junctions awaits elucidation, we herein demonstrate increased abundance of phosphorylated Cx43 and increased Cx43 localization to the surface of ROS 17/2.8 cells after stretch application. Because stretch did not alter steady-state Cx43 mRNA, it is likely that mechanical strain interferes with Cx43 protein turnover or post-translational processing. Cx43 is synthesized as a single protein that is post-translationally converted to multiple, slower migrating species by extensive phosphorylation.(43,44) The life cycle of Cx43 includes a transient residency in the endoplasmic reticulum/Golgi apparatus, followed by formation of hemichannels and their translocation to the cell surface to form gap junction plaques.(45,46) Assembly of cytoplasmic Cx43 into gap junctions can occur independently of new protein synthesis,(45) and it is usually associated with increased phosphorylation.(47) Increased Cx43 on the cell surface has been observed after increasing cellular cyclic adenosine monophosphate in mammary tumor cells,(48) and after short-term exposure of osteoblastic cells to PGE2 or parathyroid hormone.(49) However, these effects occurred by redistribution of preformed Cx43 without changes in total Cx43 protein abundance. Therefore, it is likely that the increased amount of Cx43 protein we observed in whole lysates and alkali-extracted, plaque-enriched material, which contain almost exclusively phosphorylated Cx43, reflects an increased pool of phosphorylated Cx43 protein available for assembly into gap junctions in stretched ROS 17/2.8 cells. Because stretch did not affect the unphosphorylated form of this connexin, it seems likely that mechanical stimulation increases total cellular Cx43 by indirectly subtracting a fraction of Cx43 protein from degradation.(45) This new steady state, reflecting a changed cell activity during periods of mechanical strain, is reached rather rapidly and it is maintained, to a lesser degree, for at least 24 h with continuous application of the stimulus. The declining degree of coupling after 24 h of stretch may reflect an adaption to the constant stimulation, which has been reported to occur after prolonged stretch.(50) Adaptation has also been described for stretch-activated channels in response to continuous suction.(51)

The demonstration that stretch increases gap junctional communication among osteoblastic cells adds further evidence to the physiologic role of mechanical factors for osteoblast function. Previous work from other investigators had demonstrated that cyclic stretch exerts an anabolic effect on osteoblasts, increasing the expression and secretion of bone matrix proteins, such as alkaline phosphatase, type I collagen, osteocalcin, and osteopontin.(10,11) We have recently demonstrated that the type of gap junctional communication provided by Cx43 is permissive for expression of osteoblast-specific genes and the development of a fully differentiated osteoblastic phenotype.(52) Therefore, the results of the present studies are consistent with the hypothesis that enhanced intercellular communication is a feature of the action of anabolic factors on osteoblasts.

Although the physiologic correlates of these observations remain hypothetical, one could speculate that an increased gap junctional communication induced by mechanical stretch may facilitate an anabolic effect by enhancing metabolic coupling among networked cells, and spreading locally generated signals to more distant areas. In vivo experiments have indicated that the osteogenic response of bone cells to increased strain is not limited to the site where the strain is applied, but it is a generalized phenomenon affecting a wider area of bone.(29,53) Soluble factors generated by mechanical stimulation provide short-range selective signals that disseminate strain information from a particular area of bone to neighboring or more distant cells. As a complement to this system, gap junctions may serve as the “nodes” of an intercellular network, which allows rapid dissemination of intracellular signals generated as a local response to mechanical strain. Intercellular communication may thus synchronize cell activity within bone remodeling units, and in turn allow deposition of new mineralized matrix uniformly within each unit.

In summary, we have found that cyclic stretch increases the ability of osteoblasts to diffuse negatively charged dyes through gap junctions. Regulation of gap junctional intercellular communication represents a novel aspect of mechanotransduction in bone.


This work was supported by National Institutes of Health grants AR41255 (R.C.) and DK46686 (T.H.S. and R.C.) and in part by a grant to K.Z. from the Endocrine Fellows Foundation.