The wall of hollow organs of vertebrates is a unique structure able to generate active tension and maintain a nearly constant passive stiffness over a large volume range. These properties are predominantly attributable to the smooth muscle cells that line the organ wall. Although smooth muscle is known to possess plasticity (i.e., the ability to adapt to large changes in cell length through structural remodeling of contractile apparatus and cytoskeleton), the detailed structural basis for the plasticity is largely unknown. Dense bodies, one of the most prominent structures in smooth muscle cells, have been regarded as the anchoring sites for actin filaments, similar to the Z-disks in striated muscle. Here, we show that the dense bodies and intermediate filaments formed cable-like structures inside airway smooth muscle cells and were able to adjust the cable length according to cell length and tension. Stretching the muscle cell bundle in the relaxed state caused the cables to straighten, indicating that these intracellular structures were connected to the extracellular matrix and could support passive tension. These plastic structures may be responsible for the ability of smooth muscle to maintain a nearly constant tensile stiffness over a large length range. The finding suggests that the structural plasticity of hollow organs may originate from the dense-body cables within the smooth muscle cells.
- muscle mechanics
- three-dimensional electron microscopy
hollow organs such as the stomach, bladder, blood vessel, and airways undergo reversible volume or diameter changes in carrying out their physiological function. The volume changes are not simply passive reactions to internal or transmural pressures of the organs, but to a large degree determined by the activities of the smooth muscle cells lining the organ wall. Large changes in organ volume require smooth muscle cells to have a large functional length range. The ability of smooth muscle to carry out its function is aided by its structural plasticity (6, 10, 14, 22, 28, 35). It has been shown that smooth muscle cells are able to achieve multiple-fold length changes while maintaining substantial ability to generate force (11, 30), especially when the muscle is allowed to recover after a length change (12, 24). It has been shown that after a recovery period that normally lasts tens of minutes (during which the muscle is periodically activated and relaxed), the muscle regains its ability to generate maximal or near-maximal active tension (12, 22, 32). This process is called length adaptation (1, 34a). It has been further shown that length adaptation may involve addition or deletion of contractile units (akin to sarcomeres) within smooth muscle cells, leading to preservation of optimal overlap between the contractile filaments and maintenance of nearly constant force generation over the range of adapted lengths (12, 14, 22). Interestingly, the passive tension of relaxed smooth muscle is also affected by the process of length adaptation. Specifically, a slackened smooth muscle cell will gradually straighten and regain its resting tension during length adaptation, and the resting tension of a stretched muscle will return to its prestretch level after length adaptation (26, 33). That is, the underlying structure responsible for the passive tension in smooth muscle exhibits plasticity in addition to viscoelasticity. The exact subcellular structure that bears passive tension in smooth muscle is not known.
The current understanding of the structure of the contractile apparatus in smooth muscle is that it is made of contractile units, similar to sarcomeres in striated muscle, consisting of myosin-containing thick filaments, actin-containing thin filaments, and dense bodies acting as the equivalents of striated muscle Z-disks. The dense bodies are often depicted as small, discrete anchorage sites in a network of thin and intermediate filaments, distributed in a way that the thick filaments can be accommodated in between dense bodies (3, 7, 13, 29, 34a). This conventional depiction is in stark contrast with dense-body (DB) aggregates revealed in high-resolution 3-dimensional (3-D) images showing long, cable-like structures that do not contain thick filaments (19–21). The abundance of the DB cables and their parallel arrangement with the contractile filaments suggest that these structures could play a role in bearing passive tension in smooth muscle. In this study, we examined the ability of the DB cables to withstand externally applied tension and determined whether they display plastic behavior.
MATERIALS AND METHODS
Sheep tracheas collected from a local abattoir (Pitt Meadows Meats) were used in the experiments. The use of animal tissue for this study was approved by the Animal Care Committee of the University of British Columbia. The tracheas were stored in physiological saline solution (PSS) at 4°C (pH 7.4, 118 mM NaCl, 5 mM KCl, 1.2 mM NaH2PO4, 22.5 mM NaHCO3, 2 mM MgSO4, 2 mM CaCl2, 2 g/l dextrose). Before an experiment, a tracheal ring (∼9 mm in width) was removed from the whole trachea. Before the tracheal ring was cut open, the in situ length of the tracheal smooth muscle bundles that connect the C-shaped cartilage was measured. A rectangular piece of smooth muscle (∼9 × ∼10 × ∼0.2 mm in dimension, mostly free of connective tissues) was obtained. From each trachea, six strips (∼1.5 × ∼10 × ∼0.2 mm) were dissected and clipped on both ends with aluminum foil clips for attachment to the force transducer. The resting lengths of all six strips were the same.
Each muscle strip was mounted to the force transducer and bathed in PSS at 37°C. Before the muscle was chemically fixed for electron microscopy (EM), it was equilibrated (preconditioned) in the PSS for 1 h and followed by measurements described below. During the equilibration period, the muscle was stimulated by electrical field stimulation (EFS) for 10 s at 5-min intervals.
A reference length of the muscle (Lref) was established by stretching the muscle to the in situ length (±5%). The passive force in the muscle strip was ∼1.0 mN at Lref. The muscle was considered equilibrated when it developed a stable maximal active force. The maximal isometric force (Fmax) of each muscle at Lref was recorded; the strip was then ready for the next step of the experiment.
Six muscle strips (of identical initial length) from the same trachea were used for each set of experiments. Three such sets of experiments were carried out for this study. A total of two tracheas were used, one used in one set of experiments, the other used in two sets of experiments. For each set of experiments, five muscle strips were fixed under conditions A–E, as described in results. The sixth strip was not fixed; it was used to assess irreversible damage due to stretching of muscle. No damage to the muscle strips was found after length changes; this assessment was based on the full recovery of active force after returning the muscle to Lref.
Measurement of passive stiffness.
The stiffness of resting muscle (passive stiffness) was assessed by applying a quick stretch to the muscle and measuring the force response. The size of the step increase in length was arbitrarily set at 10% of Lref, and the speed of stretch was fixed at 10% Lref per 100 ms. The duration of the stretch was 100 ms, and the muscle was maintained at the stretched length for 1 s before it was released back to its original length. The peak force response (which occurred at the end of the step increase in length, i.e., 100 ms after onset of stretch) divided by 10% Lref was used as a measurement of muscle stiffness.
Primary fixation (15 min) was carried out while the muscle strips were still attached to the apparatus. The fixing solution (1% paraformaldehyde, 2.5% glutaraldehyde, and 2% tannic acid in 0.1 M sodium cacodylate buffer) was preheated to 37°C. After primary fixation, the muscle strip was removed from the apparatus and cut into small blocks (∼2 × ∼0.5 × ∼0.2 mm in dimension) in cold fixation buffer and kept in the same fixative for 2 h at 4°C on a shaker.
For secondary fixation, the tissue blocks were transferred to 2% osmium buffer for 1.5 h at 4°C on the shaker, followed by three washes with distilled water (10 min per wash).
Standard methods for block staining, dehydration, embedding, and sectioning were used as previously described (12, 15). Images of the cross-sections of the muscle cells were obtained with an electron microscope (Philips FEI Tecnai 12 TEM). For 3-D reconstruction, images from ∼100 cell cross-sections (50-nm thick) for each of the 5 groups (fixed under conditions A–E, as described in results) were collected. All the images were taken with a digital camera (Gatan 792) at a magnification of ×37,000. To capture the whole cross-section of a single cell, it often required taking multiple images of different parts of a cell cross-section and merging the parts together by eye with the help of Adobe Photoshop.
3-D image reconstruction.
The serial sections in each of the 5 groups fixed for EM were aligned using Adobe Photoshop. The alignment was needed because each one of the serial EM images was taken from different sections, and these sections had slightly different orientations. Intracellular organelles and some cell features, but not dense bodies or intermediate filaments (e.g., mitochondria, endoplasmic reticulum, caveolae, thick filament, and microtubules), were used as markers for the alignment. Usually 3 or more markers from each of the adjacent sections were aligned to allow orientation of the image to be determined. The alignment was carried out manually by eye. Perfect alignment for multiple thick filaments adjacent to a chosen dense body could be achieved most of the time for consecutive 50-nm thin sections, indicating that the thick filaments were perpendicular to the sections. No adjustment was made for differential shrinkages, i.e., the magnification factor was the same for all images in the series. A great majority of the consecutive images could be aligned satisfactorily by simple shifts and rotation of the images without any additional adjustment. After the alignment, dense bodies and intermediate filaments were traced, and the 2-D traced sections were assembled by a program (Volocity) into 3-D images. Unlike the case with thick filaments, alignment of multiple intermediate filaments was not possible most of the time, indicating that individual intermediate filaments may not have the same orientation and may not be perpendicular to the cut, thin sections.
Determination of the length of DB cables.
A ∼5-μm segment for each muscle strip fixed for EM was cut into 50-nm-thin sections for 3-D reconstruction of the dense body structure. The length (along the long axis of the cell) of a DB cable within a cell segment was measured in 3-D space using the following method. The centroid of each dense body was first measured using ImageJ (version 4.0). In the 3-D analysis, we defined the transverse section of the cell as the x-y plane, and the z-axis was parallel to the long axis of the cell (and perpendicular to the x-y plane). Each dense body (centroid) measured was given an x-y coordinate with reference to the centroid on the first section (x = y = 0). The distance (D) between the centroids of two adjacent dense bodies in section i and i+1 of the same cable was the length of the DB cable in the ith section, as illustrated in Fig. 1. It followed that the length of DB cable in section i is Di = , where L is the section thickness (50 nm), and di is the projected distance on the x-y plane between the centroids of the two dense bodies in sequence in section i and i+1 and is calculated as di2 = . The total length of the DB cable in the cell segment was the sum of the Ds calculated for all serial sections. To obtain a numerical index of how straight (or slack) the cables were, we divided the 3-D cable length by the length of the muscle segment (i.e., 50 nm multiplied by the number of sections). This length ratio of DB cable-to-cell segment was used as an index of cable “slackness”.
In this study, a total of 15 sets of serial sections (∼100 for each set) were obtained from 15 cells. That is, 3 cells for each of the experimental conditions (A–E) were randomly chosen. For each cell, 3 DB cables were reconstructed. The total number of merged electron micrographs used in this study was 1,500. The “n” number referred to the number of cells from which the data were collected. Unless otherwise stated, data are shown as means ± SE. Statistical difference was considered significant at P < 0.05 using 1-way repeated-measures ANOVA followed by Tukey a posteriori test.
Serial transverse sections of ovine airway (tracheal) smooth muscle were cut for EM. Figure 2A shows an example of such a section. Dense bodies and the adjacent intermediate filaments were identified and traced (Fig. 2B) so that a 3-D image of these structures could be reconstructed from consecutive sections. Figure 3 shows dense bodies and intermediate filaments in a 750-nm cell segment consisting of 15 thin (50-nm) sections. In the 3-D image (Fig. 3), the dense bodies appeared to form elongated structures flanked by intermediate filaments that ran approximately parallel to the long axis of the muscle cell. We call these cable-like structures DB cables. The intermediate filaments in Fig. 3 appeared to be discontinuous. This is likely an artifact stemming from our 3-D reconstruction where we assumed the filaments were perpendicular to the cross-sections. The discontinuous appearance suggests that the intermediate filaments are “wavy,” perhaps undulating in the same manner as the dense body structure along the long axis of the cell.
To determine whether the DB cables were mechanically interconnected in series throughout the cells within a muscle bundle and to the extracellular matrix, we examined their response to a stretch. We dissected 5 muscle strips (A–E) with the same initial length from a single trachea and fixed them for EM after adaptation at different lengths. Each strip was first adapted to its in situ length, and the maximal active force (Fmax) measured at that length was used to normalize other force measurements. The mean value for Fmax (in terms of force per muscle cross-sectional area, i.e., stress) was 143.3 ± 11.6 kPa (n = 3). Muscle A was fixed at its in situ length in the relaxed state and used as a reference. The length of muscle A (Lref) was used as the normalizing unit for all length measurements. Muscle B was fixed at 0.5 Lref after the muscle was passively shortened by 50% and then given a brief (1-s) EFS to shorten and straighten the muscle strip. The muscle was allowed to relax before it was fixed. The shortened strip remained straight after muscle relaxation. Longitudinal sections were examined to confirm that individual cells remained straight and aligned with the axis of force transmission. An example is shown in Fig. 4. Muscle C was fixed at 0.5 Lref after it was fully adapted to the shortened length, at which point it had recovered most of the force lost immediately after it was shortened. The protocol for adaptation consisted of periodic short EFS (10 s) applied to the muscle once every 5 min for 30 min. Full adaptation occurred when the EFS-induced force reached a maximal plateau. Muscle D was fixed at 1.5 Lref immediately after the stretch in the relaxed state. Muscle E was fixed at 1.5 Lref after the muscle was fully adapted to the stretched length. Passive tension in the muscle preparations just before they were fixed was recorded. This experiment was repeated two more times using tracheas from the same sheep and one additional sheep.
Figure 5 shows snapshots of reconstructed 3-D images of DB cables in a 5.25-μm segment of sheep trachealis containing 105 serial transverse EM sections. This particular muscle strip was fixed immediately after stretching the muscle from Lref to 1.5 Lref (muscle D described above). Only 3 out of ∼20 cables in the cell segment were traced and reconstructed. Figure 6 shows snapshots of 3-D images of DB cables from 2 cell segments fixed under a stretched condition (muscle D) and a shortened condition (muscle C). The images are shown mostly in the z-x plane, illustrating the general parallel feature of the cables with the horizontal axis (axis of force transmission); they also show that under the stretched condition, the DB cables appear straighter compared with those in shortened muscle. The appearance of discrete structures in an array that characterizes the DB cables we observed (e.g., Figs. 5 and 6) was at least partially if not all due to the way the 3-D images were constructed with relatively thick (50-nm) sections, just like the case of intermediate filaments shown in Fig. 3. The real structure of a DB cable is probably much less “discrete” than that shown in Figs. 5 and 6.
The passive tensile stress associated with muscle strips A–E is shown in Fig. 7A. Passive shortening of the muscle did not decrease the resting tension further because at Lref there was minimal tension to start. Stretching the muscle caused significant increase in resting tension. Figure 7B shows the length ratio of DB cable-to-cell segment in conditions A–E. The length ratio is an index of slackness in the cables because it was calculated as the measured length of DB cables in 3-D space divided by the length of the muscle segment within which the cables were embedded (see more details of the calculation in materials and methods). Stretching the muscle strips (D and E compared with A) significantly straightened the cables. Figure 7C shows the correlation between resting tensile stress in the muscle strips and the slackness of the cables. A significant correlation was found in muscle strips stretched from Lref (P < 0.05, goodness of fit r2 = 0.99). It should be pointed out that under all conditions (A–E), the same segment length (∼5 μm) was chosen for 3-D reconstruction. That means in the muscle strips fixed at 0.5 Lref, a 5-μm segment would contain twice as much “cell material” as a 5-μm segment from muscles fixed at Lref and 3 times more material as a 5-μm segment from muscles fixed at 1.5 Lref.
Passive stiffness (i.e., stiffness of resting muscle) of muscle strips was measured under different conditions (Fig. 8). At Lref, passive stiffness was measured after the muscle had been fully adapted to the length; this stiffness value was used as a reference for normalizing the subsequently measured values at 0.5 Lref. The peak force response to a 10% Lref quick stretch divided by the length change was used as a measure of muscle stiffness. The reference value of muscle stiffness at Lref was 3.21 ± 0.28 (Fmax/Lref) (n = 3). Under the “not adapted” condition, the stretch was applied to the muscle immediately after it had been passively shortened by half. The muscle was slack and wavy and gave no resistance to the stretch. However, after one brief stimulation (1-s EFS), the muscle became straight and remained straight after relaxation (Fig. 4). A stretch was applied to the muscle 30 s after complete relaxation, and the stiffness measured was plotted in Fig. 8 under “partially adapted” condition. Under the “fully adapted” condition (Fig. 8), stiffness measurement was carried out after the muscle had been adapted at 0.5 Lref for 30 min.
Compared with skeletal muscle, smooth muscle operates over a much larger length range to carry out its physiological function because of the large volume change normally associated with the function of hollow organs. Although its in vivo function does not require airway smooth muscle to generate force over a large length range, it nevertheless possesses the ability to do so (12, 14, 22, 24, 33), just like bladder (25) and uterine smooth muscle (35). Interestingly, smooth muscle maintains an approximately constant passive stiffness over a wide range of lengths (4, 26, 33). This property has perhaps evolved from the need to avoid excessively high or low vesicular pressure in hollow organs at large and small volumes, respectively. It has been known for a long time that all muscle cells are able to withstand some level of tensile stress even when they are not activated to produce active tension. The origin of the passive stiffness is believed to reside in the cell cytoskeleton, a structural domain separated (from a functional point of view) from the contractile domain. Titin filaments are believed to bear resting tension in striated muscle (2, 16). The role of titin in smooth muscle is less certain because of its low expression level and smaller isoform size in smooth muscle (17). Passive stiffness in skeletal muscle is not known to be adjustable lengthwise; perhaps this is because the working length range in skeletal muscle is small and limited by the skeleton. It is likely that the structure giving rise to passive stiffness in smooth muscle is different from that of striated muscle and that this structure possesses a unique material property, i.e., plasticity. The viscoelastic extracellular matrix (mostly made of collagen and elastin fibers) is therefore not likely to be the factor determining resting tensile stress in the wall of hollow organs; the adjustable passive stiffness is likely the results of an active process that involves plastic restructuring of subcellular elements within the smooth muscle cells. The most significant finding of this study is that we have identified one potential subcellular structure that could be (at least partially) responsible for the plasticity of resting smooth muscle and the wall of hollow organs.
Figure 7C shows a linear relationship between the amount of stretch applied to the muscle and the “straightness” of the DB cables, suggesting that these highly abundant parallel longitudinal structures may be responsible for bearing passive tension in smooth muscle. Figure 7 also demonstrates the plastic behavior of shortened DB cables capable of adjusting their length according to the cell length. This point may not be obvious from the figure and requires additional explanation as follows. Because the cell volume is conserved (14), when the cell length is changed from Lref to 0.5 Lref, the cell will become wider in diameter so that a 2.5-μm segment in the shortened cell contains the same amount of material as a 5-μm segment does in a cell at Lref. If the cable behaves like a string and does not change its length when the cell shortens by half, the average length of the DB cables in a 2.5-μm segment in muscles fixed at 0.5 Lref would have the same length as that in the 5-μm segment from a cell fixed at Lref. In other words, the DB cables in the 5-μm segment from muscles fixed at 0.5 Lref shown in Fig. 7B would be twice as long as that in the 5-μm segment from muscles fixed at Lref. Results (Fig. 7) show that the average length of DB cables from cell segments of muscles fixed at 0.5 Lref was in fact not different from that in cell segments of muscles fixed at Lref, indicating that the cables shortened by approximately half as the muscle length was halved. Because the cell volume is constant at different cell lengths, it is reasonable to assume that the DB material is conserved at different cell lengths, although it may appear in different forms including some that are not visible to EM. In this study, we did not quantify the volume of all dense bodies and the total number of DB cables in the cell segment and therefore cannot speculate on whether or how the volume of dense bodies is conserved (i.e., more volume per DB cable or more cables in parallel or a mix of both). On the premise that the DB cables are able to bear tension, findings shown in Fig. 8 suggest that plastic restructuring initially rendered the cables less able to support tension, but with length adaptation, the tensile strength of the cable returned to its previous level (before its length was halved).
Stiffness of airway smooth muscle (whether it is active or passive) directly contributes to airway wall stiffness, which in turn determines how effectively lung volume fluctuation (due to tidal breathing and deep inspiration) results in bronchodilation (9). Reducing stiffness of airway smooth muscle is therefore a potentially effective way of combating airway hyperresponsiveness as that seen in asthma. Although we know more about the molecular mechanism underlying smooth muscle stiffness associated with muscle activation (i.e., active stiffness) (8), much less is known about the origin of the passive stiffness. The general behavior of passive stiffness is predicted by the current theories on cytoskeletal rheology of living cells (5–6, 28) and network of actin filaments (23). Presumably, these theories also apply to the DB cables.
Evidence supporting the notion of dense bodies as the anchorage sites for thin filaments, analogous to the Z disks in striated muscle, has been provided by a series of elegantly designed and carefully carried-out studies (3, 7, 13, 29). The present study was not designed to examine the structural role of dense bodies in the contractile apparatus; it is possible that dense bodies play a dual role in smooth muscle, that is, they function as Z-disk equivalents and also support passive tension as part of the cytoskeleton. However, there are discrepancies between the findings of the present and the previous studies that need to be reconciled. One obvious discrepancy is in the description of the structural relationship between dense bodies. In the early studies, dense bodies were shown as discrete elements in series along the long axis of the muscle cell (3, 7, 13, 29), and the distance between them decreased in actively shortened muscle, presumably due to filament sliding caused by actomyosin cross-bridge interaction (7, 13), whereas in the present study, dense bodies were shown as longitudinal aggregates of (possibly) individual bodies linked by intermediate filaments (and likely thin filaments as well), without thick filaments in between them. The discrepancy could be due to the fact that 2-D structural information was used in the early studies in deriving the dense body structure. If we were to take a 2-D longitudinal thin section of the dense body cables shown in Fig. 6 (especially Fig. 6B), a continuous structure would not be revealed. Another possible explanation of the discrepancy is that the distribution of α-actinin along a DB cable is not even but punctuated. That would account for the discrete appearance of dense bodies seen by Fay and coworkers (7). However, this would not reconcile with the findings by Kargacin et al. (13). In their studies, phase-contrast microscopy rather than antibody staining was used to visualize dense bodies (13). Yet another possible explanation is that there are two types of dense bodies: one possesses long, cable-like structure and belongs to the cytoskeletal domain, and the other has the classic “fusiform” appearance, is smaller, and belongs to the contractile domain. In our 3-D reconstruction, we tended to follow larger dense bodies. It should be pointed out that we did not examine the relationship between dense bodies and the contractile (i.e., thin and thick) filaments. It is possible that the contractile filaments run obliquely between adjacent DB cables, as proposed by North et al. (21). However, we could not find evidence of obliquely aligned thick and thin filaments (15). Taken together, reconciliation of the disparate findings will require further studies that specifically examine the role of dense bodies as part of the contractile apparatus, keeping in mind that many, if not all dense bodies exist in long, cable-like structures. The DB cables observed in our 3-D reconstruction were likely longer than 5 μm. A smooth muscle cell is, however, much longer than 5 μm; it can be 100–500 μm, depending on how much it is stretched. It is possible that the DB cables do not span from one end of the cell to the other but are punctuated with gaps. These gaps could be occupied by thin, intermediate, and thick filaments. If this is the case, our finding is consistent with the classic model of the contractile apparatus of smooth muscle (3, 7, 13, 29, 34a), with a minor discrepancy regarding the dimension of the dense bodies.
The present finding defines a starting point of our understanding of intracellular structures responsible for smooth muscle cell and organ stiffness and plasticity. Besides the evidence shown above, the argument that DB cables are able to bear passive tension in smooth muscle tissue is also supported by a previous finding from our laboratory (15) that dense plaques on the cell membrane are physically connected to dense bodies and extracellular matrix. This explains how the force from an external stretch can be transmitted to the DB cables. In a similar experiment using magnetic beads bound to integrin receptors (in dense plaque areas), Deng et al. (6) measured the bead movements in a twisting magnetic field and concluded that the cell was made of soft glass-like material possessing plasticity. However, they did not identify the exact molecular structure responsible for the material property.
Recent findings regarding vimentin phosphorylation during smooth muscle activation (18, 27, 34) may shed some light on the molecular mechanism of DB cable restructuring. It appears that on activation and vimentin phosphorylation, disassembly of intermediate filaments occurs. This may allow the length of the DB cables to be reestablished when the muscle settles to its final resting length on relaxation and dephosphorylation of vimentin, followed by reassembly of intermediate filaments.
In conclusion, the present study provides evidence supporting DB cables as an intracellular structure capable of bearing passive tension. Furthermore, the plastic nature of the DB cables could be responsible for the length-independent passive stiffness and the reversible shifts in the passive length-tension relationship in smooth muscle. These smooth muscle properties in turn confer plasticity on the wall of hollow organs.
This work was supported by operating grants from Canadian Institutes of Health Research (CIHR; MOP-13271 and MOP-4725).
No conflicts of interest, financial or otherwise, are declared by the author(s).
We give special thanks to Pitt Meadows Meats (Pitt Meadows, British Columbia, Canada) for the supply of fresh sheep tracheas in kind support of this research project.
Present address of A. M. Herrera: Investigacion en Ciencias Básicas, Faculdad de Medicina, Universidad CES, Calle 10A #22-04, Medellín, Colombia.
- Copyright © 2010 the American Physiological Society