Actin and Tropomyosin Variants in Smooth Muscles DEPENDENCE ON TISSUE TYPE*

Actin was found to be the major source of myofibrillar protein heterogeneity in smooth muscles. Three isoelectric variants, a-smooth muscle (a-SM), 8-non-muscle (8-NM), and y-actins (y-SM and y-NM) were measured in 15 different smooth muscles. a-SM and y-actin contents displayed an inverse relationship in a given smooth muscle, some of which contained primarily a-SM actin while y-actins dominated in others. a-SM actin and y-actin distributions were tissue-specific, independent of species. A greater proportion of a-SM actin appears to be associated with tissues having a high degree of tonic activity. 8-Nonmuscle actin was a significant, and relatively constant, component of all smooth muscle tissues. The high NM-actin content of these tissues may reflect the importance of proliferative, synthetic, or secretory activities in smooth muscle, because the a-SM actin disappeared in tissue culture with a time course paralleling the modulation of phenotype from a contractile to a proliferative cell. Two tropomyosin subunits were present in approxi-mately equal amounts in all smooth muscle tissues studied. One tropomyosin subunit exhibited identical mobility on two-dimensional gel electrophoresis, while the other was characterized by some species-specific variation which was unrelated to actin variant distribution. No variants of the 20,000-dalton regulatory light chain of myosin were observed. These results suggest that SM-specific actin


Actin and Tropomyosin Variants in Smooth Muscles
Actin was found to be the major source of myofibrillar protein heterogeneity in smooth muscles. Three isoelectric variants, a-smooth muscle (a-SM), 8-nonmuscle (8-NM), and y-actins (y-SM and y-NM) were measured in 15 different smooth muscles. a-SM and yactin contents displayed an inverse relationship in a given smooth muscle, some of which contained primarily a-SM actin while y-actins dominated in others. a-SM actin and y-actin distributions were tissue-specific, independent of species. A greater proportion of a-SM actin appears to be associated with tissues having a high degree of tonic activity. 8-Nonmuscle actin was a significant, and relatively constant, component of all smooth muscle tissues. The high NM-actin content of these tissues may reflect the importance of proliferative, synthetic, or secretory activities in smooth muscle, because the a-SM actin disappeared in tissue culture with a time course paralleling the modulation of phenotype from a contractile to a proliferative cell. Two tropomyosin subunits were present in approximately equal amounts in all smooth muscle tissues studied. One tropomyosin subunit exhibited identical mobility on two-dimensional gel electrophoresis, while the other was characterized by some species-specific variation which was unrelated to actin variant distribution. No variants of the 20,000-dalton regulatory light chain of myosin were observed. These results suggest that SM-specific actin variants are associated with functional diversity among smooth muscles. Myofibrillar protein variants have been observed in smooth and striated muscles. Myosin variants exhibit differences in specific ATPase activity in striated muscles which are proportional to the maximum velocity of shortening for zero load (Barany, 1967;Weeds, 1978). Differences in the ratio of a:@ tropomyosin subunits have also been observed between fast and slow striated muscles Perry, 1973, 1974;Johnson, 1974), although the biological significance of this is unknown. Skeletal and cardiac muscles exhibit no significant variability in the actin content or type.
There is little evidence for myosin or tropomyosin variants in smooth muscle . However, four forms of actin have been detected in several smooth muscle tissues Vandekerckhove and Weber, 1978). These include two var-* This work was supported by National Institutes of Health Grant 5 PO1 HL19242. This study describes portions of a Ph. D. dissertation (University of Virginia, 1983) by V. F. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. iants specific to smooth muscle (a-SM' and y-SM actin) plus appreciable amounts of the two cytoplasmic actins found in virtually all eukaryotic cells (P-NM and y-NM actin). The possible relationship of actin variants to smooth muscle function is not known.
The aim of this study was to determine whether the distribution of actin variants and/or the presence of myosin light chain or tropomyosin subunit variants is associated with differences in smooth muscle function in a series of smooth muscles, many of which have been the subject of extensive mechanical characterization.

Preparations
The criteria for tissue selection were as follows: 1) ability to dissect smooth muscle layers relatively free of nonmuscle cells which would bias contractile protein measurements; 2) availability of mechanical data and/or estimates of total actin, tropomyosin, and myosin contents; and 3) variety, allowing species-and tissue-specific comparisons.
Tissue Preparation-Tissues were collected from animals at slaughter or when experiments with animals were terminated and stored in a zwitterionic-buffered physiological salt solution at 4 "C until used (0-2) days. Complete viability of the tissues as assessed by measurements of mechanical or contractile properties can be maintained over this period (Dillon and Murphy, 1982). The composition (mM) of the physiological salt solution was NaC1, 140; KCl, 4.7; MgS04, 1.2; CaC12, 1.6; Na2HP04, 1.2; D-glucose, 5.6; EDTA, 0.02; and MOPS, 2.0 (pH 7.4 a t 37 " C ) . The smooth muscle layers were carefully dissected, lightly blotted, and weighed. Tissue samples were frozen in an acetone/dry ice slush and homogenized in 1% SDS, 10% glycerol, 20 mM dithiothreitol (IO mg of tissue/ml of SDS bomogenization medium) according to the method described by Aksoy et al. (1983). Polypeptide resolution was optimal when the samples were immediately subjected to electrophoresis.
Isolated Smooth Muscle Cell.s"Smooth muscle layers from swine aorta were aseptically dissected from the tissues. The cells were enzymatically isolated from minced samples (Chamley-Campbell et at., 1979). Cells were plated at a density of 5 X IO4 cells/cm2 for culture and grown in "199 medium (Gibco) containing 10% fetal calf serum (Gibco), 100 units/ml of penicillin and 100 pg/ml of streptomycin (Gibco) (Chamley-Campbell et al., 1979). Cells were observed at different times from plating until confluency. Harvested cultured primary smooth muscle cell isolates were washed in phosphate-buffered saline and counted using a hemocytometer. The cell pellet was resuspended in SDS homogenization medium at 37 "C (2,000 to 20,000 cells/pl) and sonicated.

Resolution of Protein Variants
A modification of O'Farrell's (1975) two-dimensional electrophoresis method was used in which IEF was followed by electrophoresis on a polyacrylamide slab containing SDS. See Fatigati (1983) for additional details regarding optimization of these procedures.

Protein Variants in Smooth Muscles
Isoelectric Focusing-The IEF gels (3 mm X 18 cm) were cast according to O'Farrell's method (1975) except that the ampholyte mixture was restricted to the 4-6.5 interval. The gels were allowed to polymerize for at least 2 h and were prefocused at 18 'C using a SDS homogenization medium overlay for 15 min at 300 V, 30 min at 650 V, and 30 min at 1000 V. Water at 18 "C was circulated through the cooling core during prefocusing and focusing. Samples were loaded onto the rinsed, prefocused gel in varying amounts to optimize visualization of specific proteins as follows: 3-20 p1 to estimate actin, desmin, and vimentin; 40 pl for tropomyosin; and 150-200 pl for the 20,000-dalton myosin light chain. The sample was overlaid with 20 p1 of SDS homogenization medium, and fresh 0.02 M NaOH was used to fill the tubes. The samples were focused at 1000 V for 16 h. The gel tube was shattered with a Bio-Rad Gel Tube Eliminator and the gel segment containing the proteinfs) to be studied was either placed immediately on a SDS slab gel or sealed in a glass tube and rapidly frozen in an acetone/dry ice slush for analysis within two weeks. for "nonequilibrated gels" (Driska et al., 1981) was followed except that electrophoresis with the "high-SDS-running buffer" was reduced from 20 to 10 min. The gels were stained overnight in 1 liter of 25% isopropanol, 10% acetic acid, and 0.04% Coomassie Brilliant Blue R (Fairbanks et al., 1971). Gels were first destained in 25% isopropanol and 10% acetic acid for 5 h and then transferred to 10% acetic acid for complete destaining. Gels were either photographed or dried after densitometry. Scanning Densitometry and Duta Analysis-The gels were scanned with a Quick Scan Jr. densitometer which was custom modified with high resolution optics (Helena Laboratories) at a wavelength of 540 nm. A pinpoint light source was used. The protein spot must be scanned through the center in both electrophoretic dimensions to obtain the maximum optical density. Several scans were made to insure that the maximum optical density was obtained. Gels were excluded from data analysis if the scans were not symmetrical.
The densitometer peak heights were measured together with the peak widths at one-half peak height. The optical density value ( A ) was calculated by the following equation (Fatigati, 1983), A = 4/3 H X WIEF X WSDS, where H = maximum peak height, which should be equal for scans in the IEF and SDS directions. The larger value was used if slight variations occurred. WIEF and WSDS are the widths of the peaks a t one-half maximum height for the scan in the IEF and SDS direction, respectively. The optical density values for each variant were converted into a per cent of the total protein optical density for all variants of a specific protein and the results were averaged (f S.E.). Table I illustrates the reproducibility of data obtained for actin with this method. Comparison of the errors for estimates for individual tissues (A) with errors calculated from data obtained with different samples (B) indicates that the method is reasonably reproducible, but it is a major source of variability. Comparable results were obtained for tropomyosin subunit comparisons.

Estimation of Smooth Muscle Cell Fraction
Smooth muscle tissues were fixed for 2 h in 2.5% (v/v) glutaraldehyde, 0.1 M cacodylate buffer (pH 7.2, 25 "C), and 6.6% (w/v) sucrose. They were postfixed in 1% osmium tetroxide and 0.1 M cacodylate buffer (pH 7.2) for 2 h and stained with 3% (w/v) uranyl acetate and 28 mM veronyl acetate for 2 h. The tissues were embedded in Poly/Bed 812 (Polysciences, Warrington, PA) after dehydration and sectioned perpendicularly to the long axis of the muscle cells.
The sections were placed on Formvar-coated copper grids. Lowmagnification electron micrographs (X 10,962) were taken of randomly chosen fields representing the entire tissue cross-section (Gerthoffer and . Smooth muscle cell fractions were determined from the electron micrographs by point-count analysis (Underwood, 1970). This estimate of cell fraction allowed calculation of cellular protein content from tissue data but is subject to error produced by shrinkage artifacts. We assume that such errors were relatively uniform among the tissues, allowing valid comparisons,

Identification of Proteins after Electrophoresis
The myofibrillar proteins are the major polypeptides present in smooth muscle cells and are readily observed (Fig. 1). Three isoelectric variants of actin can be separated. Desmin, vimentin, tropomyosin (two subunits), and the 20,000-dalton myosin light chain are also resolved with the appropriate loading.
The actin variants were tentatively identified by comparison with published results (Garrels and Gibson, 1976;Izant and Lazarides, 1977). The following studies were conducted to insure that the three spots were different actins and not charge modifications. Rabbit askeletal muscle actin, prepared according to the method of Spudich and Watt (1971), appeared as one spot in the gel. Thus, the method did not lead to detectable charge modifications which would be confused with true isoelectric variants. The a-skelatal muscle actin coelectrophoresed with the most acidic actin variant from smooth muscle tissue, thereby identifying it as a-SM actin. The consistency in results (Table I) also argues against significant charge modifications during tissue processing and electrophoresis. The 0-NM and the unresolved y S M and y N M variants were confirmed by coelectrophoresis of various smooth muscle tissues and endothelial cells. Tropomyosin and the phosphorylated and nonphosphorylated forms of the 20,000-dalton myosin light chain were previously identified in this gel system (Driska et al., 1981).

Statistics
The t-test was used to determine if values were members of the same population (Snedecor, 1956). Mean values were considered significantly different at p 5 0.05. Calculation of the actin variant content as mg/g of cell wet weight required multiplying the total actin content (either individual tissue values or average arterial or nonarterial values) by the relative actin variant composition (as per cent of total actin). The S.E. for actin variant contents was calculated from the following equation (Day and Underwood, 1967): where x = total actin estimate (mg) and y = per cent of total actin estimate for each variant.

RESULTS
Distribution of Actin in Smooth M u c k Tissues-For a given tissue type, irrespective of species (i.e. all arterial smooth muscles, all tracheal smooth muscles), the a-SM actin was a fairly constant fraction of the total actin (Table 11). Across tissue types, a continuum of a-SM actin fractions was observed (Table 11). Interestingly, there is a difference in a-SM actin values between the lower esophageal sphincter and the adjacent, anatomically indistinguishable circular layer of the esophagus in cat and opossum. These tissues can only be    Cohen and Murphy (1978). Cohen and Murphy (1979).
distinguished on the basis of contractile activity. The y-(7-SM and y-NM) actin content also exhibited a continuum of values across tissue types (Table 11) and increased as the a-SM actin fractions fell. There was more variation in y-actin content for a given tissue type than in the a-SM fraction. This variability may arise from the combination of y-NM and y-SM actins which are not resolved. P-NM actin constituted a significant component in all tissues. The overall tissue average (18.6 & 1.7%) appears to be too high to be totally ascribed to the actin content of nonmuscle cells in the tissues.
Actin Content of Smooth Muscle CelLs-The basis for the high P-NM content in smooth muscle tissues (compared to striated muscle, where it is virtually undetectable) prompted two sets of experiments. Primary cell isolates prepared for standard tissue culture of smooth muscle should have a reduced fraction of nerves or perhaps other cell types which might contribute to a high NM-actin content. However, swine aortic primary isolates showed no differences from the dissected tissue preparations (Table 111). This was also true for swine trachealis primary isolates subjected to further fractionation on a Percol gradient (Table 111). The aortic smooth muscle cells lost the a-SM actin variant in culture over the period associated with the proliferation and growth to confluence (Fig. 2). However, passaged cells did contain a-SM actin (approximately 18%, n = 1). These experiments suggest that the NM actins are a normal constituent of smooth muscle cells and that their presence can be reasonably estimated in tissues selected according to the criteria established for this study.
The absolute content (mg/g of smooth muscle cell, wet weight) for the actin variants was estimated by multiplying the per cent composition of each actin variant by the previously estimated total cellular actin content (mg/g of smooth muscle cell, wet weight) for each tissue (Fig. 3). The error for each value used in this calculation was taken into account (see "Materials and Methods"). The variability reported for the total mg of actin/g of cell, wet weight, values can be large due to propagation of errors in content and in cell fraction estimates (Table 11). Arteries have a significantly higher actin

Protein Variants in Smooth
Muscles content than have other smooth muscle cells Murphy, 1978, 1979). However, no significance can be attributed to differences between total tissue actin contents in other tissues.
The "extra" actin in arterial smooth.muscle compared with all other smooth muscle types examined Murphy, 1978, 1979;Fatigati et al., 1982) was primarily due to a high a-SM actin content (Fig. 3).
Tropomyosin-The two tropomyosin subunits differed in their isoelectric points as well as in their molecular weights and were labeled + (basic) and -(acidic) accordingly (Fig. 4).
The subunits were present in equal amounts in the rabbit trachealis and cat jejunum (Table IV). The results for other tissues showed slightly greater amounts of the positive subunit. This imbalance may reflect relative staining differences, the presence of another protein co-migrating with the positive subunit, excess synthesis of the positive subunit, differential  Murphy, 4o 1978, 1979;Fatigati et al., 1982). The length of each segment within a bar indicates the amount of each actin variant .
(? S.E.). The smooth muscle cell frac-30 tions used for these estimates are listed .E in Table 11. proteolysis during sample processing, or the presence of tropomyosin molecules composed of two positive chains. However, the molar ratio of actin to tropomyosin for all tissues was 5.94 f 0.74. This is comparable to the value determined for thin filaments from striated muscles Murphy, 1978, 1979) and suggests that the overall quantification of actin and tropomyosin is reasonable.
The more positive tropomyosin subunit was common to all tissues studied (Fig. 4), as judged by coelectrophoresis. Swine carotid, trachealis, aorta, and esophageal smooth muscle, together with the cat jejunum, rabbit trachealis, and guinea pig taenia coli all appeared to have the same two subunits. Dog carotid media and trachealis had identical subunits. The negative subunit of dog tissues had a slightly higher molecular weight than the negative subunit of the other tissues, and it may represent a species-specific variant. No tissue-specific variation or correlation with the distribution of actin variants was apparent in the tropomyosin studies.  limb skeletal muscle (not illustrated) showed that smooth and skeletal muscle tropomyosins had no subunits in common.
Myosin-Coelectrophoretic studies on the following tissues indicated that the 20,000-dalton myosin light chain did not exhibit any tissue or species variants: dog trachealis and carotid media, cat jejuneum, guinea pig taenia coli, swine carotid and aorta media, trachealis, and esophagus. The phosphorylated form of the 20,000-dalton myosin light chain was only seen occasionally.
This was not surprising since the tissues were dissected and homogenized in solutions which did not cause contraction. The 17,000-dalton light chain and the heavy chains were poorly resolved on the gel system and were not analyzed.

DISCUSSION
The molecular basis for functional differences among vertebrate striated muscles lies in the expression of different isoenzymatic forms of myosin. Smooth muscles exhibit even greater diversity in activity patterns. Nevertheless, there is no compelling evidence for the existence of different isoenzymes of myosin among smooth muscles, or that they contribute to functional specializations . This study has demonstrated that smooth muscles exhibit a unique type of molecular diversity in the contractile proteins, i.e. tissue-specific patterns in the contents of isoactins. The possible functional significance of these distributions is considered below.
Significance of Nonmuscle Actins in Smooth Muscle Tissues-The relative p-NM actin content ranged from 10.3-33.3% of the total actin. This range may reflect differences among smooth muscle cells and/or the actin content of varying amounts of nonmuscle cell types in the tissues. The particularly high fractional amounts of P-NM actin observed in some tissues may reflect nonmuscle cells. For example, nonpregnant myometrium (33.3% P-NM actin) is perhaps the tissue most likely to have the highest fraction of nonmuscle cells among those studied. Furthermore, the estimated P-NM actin content decreased with the size of the three trachealis tissues, where greater size facilitated dissection of the more purely muscular portions of the tissue. At the other extreme, the cat intestinal circular muscle had a very high smooth muscle cell content and a comparatively low P-NM fraction (Table 11). These observations are consistent with the hypothesis that nonmuscle cells contribute significantly to the 8-NM actin content of many tissues. The P-NM actin content is not tissue-specific, as shown by the variation observed among smooth muscles in a given type (10.3-23.6% for gastrointestinal smooth muscle, 13.6-22.4% for arterial smooth muscle, and 14.0-29.3% for trachealis smooth muscle).
While it seems probable that nonmuscle elements may contribute to estimates of P-NM actin, the results also provide fairly strong evidence that P-NM actin is a normal constituent of smooth muscle cells because ( a ) many of the tissues appear to contain almost entirely smooth muscle cells on a volume basis, and (b) isolated smooth muscle cells (Table 111) have appreciable amounts of 8-NM actin.
In contrast to smooth muscle, skeletal muscle has virtually no nonmuscle actins (less than 1%; Pardo et al., 1983). It is possible that nonmuscle actins reflect aspects of smooth muscle function. Smooth muscle is not a dedicated contractile cell and has significant proliferative, synthetic, and secretory functions (Wissler, 1968). The nonmuscle actin may therefore be associated with these noncontractile functions. This interpretation was supported by the cell culture experiment (Fig.  2). Vascular smooth muscle cells in primary culture lose their thick filaments (and apparently their ability to contract normally) and develop an increased content of organelles associated with protein synthesis and secretion during the proliferative phase (Chamley-Campbell et al., 1979). The time course of change in actin distribution during culture of swine aortic smooth muscle cells (Fig. 2) appears to reflect this change in cell phenotype and function (Franke et al., 1980). The most reasonable interpretation of our results is that some 10, and perhaps 25%, of the actin in smooth muscle may reflect the expression of genes involved with basic cellular activities not associated with the contractile apparatus. This interpretation is consistent with observations suggesting that appreciable amounts of actin in smooth muscle are not polymerized (Murphy and Megerman, 1977;Seidel et al., 1981) and presumably complexed with various actin-binding proteins. The analysis omits the contribution of y-NM actin which was not resolved from 7-SM actin by isoelectric focusing. However, the @-variant is likely to be the predominant nonmuscle actin form (see below). Cell culture studies of smooth muscle have been seriously hampered by the absence of readily detectable markers for the state of cytodifferentiation. The isoactin distribution estimated by two-dimensional gel electrophoresis may be of value as such a marker.
Actin Variant Distribution and Smooth Muscle Contractile Function-The estimated distribution of the a-SM actin and y (7-SM and y-NM) actins exhibit striking variations both in total (Fig. 3) and in relative amounts (Table 11). One uncertainty in our results concerns the relative amounts of y-SM and y-NM actins which were not resolved. However, previous studies indicate that the y-NM content averages only about one-third of the p-NM actin content in mammalian tissues including a number of smooth muscles Weber, 1979, 1981;Gabbiani et al., 1981;summarized in Table 15 in Fatigati, 1983). Therefore, it should be safe to assume that tissues with high y-actin contain mainly y-SM actin and that the a-SM/y-actin ratio (Table  11) provides a useful estimate of the relative a-SM/y-SM content.
The most striking observation was the tissue specificity of the isoactin distributions. Errors propagate rapidly when ratios or absolute tissue contents are calculated, so tissue specificity can be most clearly seen when data for a-SM actin as per cent total actin are examined (Table 11). The values for three arteries (58-59%), three trachealis muscles (37-40%), two lower esophageal sphincters (31-33%), and three esophageal body tissues (13-21%) exhibited no significant differences although five species were involved. A comparable pattern is evident in the y-variants and in the a / y ratios, although compromised by inability to discriminate the y-SM variant and error propagation. We previously distinguished two groups of smooth muscles in terms of total actin and tropomyosin contents, i.e. arterial and nonarterial smooth muscles Murphy, 1978,1979;Murphy et al., 1983). It is evident that the "extra" actin in arteries is due to a high content of a-SM actin. It is this difference in content which produces high a l y actin ratios in arteries. The experimental data (Table 11) do not provide any clear basis for defining any further general categories based on isoactin distributions. The following analysis considers whether any pattern in functional behavior parallels the data.
Actin variant distributions or total actin contents do not show any obvious relationships with published estimates of contractile function such as maximum active stress or shortening velocity. However, most of the velocity and stress values in the literature probably do not reflect the true contractile system capacity. Stress measurements are typically made at submaximal activation, and shortening velocity is affected by the level of myosin phosphorylation . These factors may obscure any correlations with parameters reflecting aspects of the cross-bridge interaction such as maximal stress generating capacities or maximal shortening velocities.
There does appear to be a general relationship between activity patterns and isoactin distributions, however. Muscles that are normally relaxed (i.e. esophageal body muscles and intestine) contain primarily y-actin (presumably 7-SM). In contrast, muscles which are normally contracted (arteries and lower esophageal sphincter) contain greater amounts of a-SM actin. The terms "normally relaxed and "normally contracted were used rather than phasic and tonic, because the latter classification is usually based on the in vitro behavior of tissues. The normal physiological contractile state is uncertain for some of the muscles tested. For example, the trachealis may normally exhibit tone or, alternatively, its primary activity may involve phasic contractions as part of reflexes triggered by airway irritants.' Comparison of the lower esophageal sphincter (normally contracted) and the adjacent, anatomically indistinguishable circular layer of the esophagus in cat and opossum provided a test of the general relationship. As predicted, the normally contracted lower esophageal sphincter had a-SM contents twice those of the normally relaxed esophageal body muscle and correspondingly higher a-SM/y-actin ratios (Table 11). The sphincter ratios are somewhat lower than those of the tonic arteries. However, sphincter tissues are identified only by the development of tone when placed in a tissue bath, and our tissue samples undoubtedly contained some esophageal body muscle, which would lower the ratio.
The results are consistent with the hypothesis that the tissue-specific actin distribution is correlated with the amount or pattern of the contractile activity of the smooth muscle tissue. In skeletal muscle there is fairly strong evidence that myosin gene expression is regulated by the amount of contractile activity, with the fast myosin isoenzyme being expressed in cells which contract infrequently (Weeds, 1978). There is also considerable evidence suggesting that the expression of myosin isoenzymes in the heart depends on loading (Litten et al., 1982). It may be that stress or some cellular factor associated with contractile activity determines actin gene expression in smooth muscles. If this is true, it is a new aspect of differentiation and diversity in muscle. An important unanswered question is whether specific actin variants are associated with differences in the cross-bridge cycle and contractile function, and the nature of any such differences. * N. L. Stephens, personal communication.