INTRODUCTION

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

The transcriptional regulatory mechanisms that control gene expression during differentiation and contractile protein accumulation are becoming well understood in skeletal and cardiac muscle lineages. Current understanding of smooth muscle-specific gene transcription is much more limited, though recent studies have begun to shed light on this topic (1). In this review, we summarize some of the themes emerging from these studies, and suggest the possibility of therapeutic intervention through manipulation of smooth muscle-specific gene expression in asthma or pulmonary arterial hypertension.

	EMBRYOLOGIC ORIGINS OF SMOOTH MUSCLE

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

Skeletal muscle arises from myotomes that develop within axially arranged somites, which themselves stem from mesoderm flanking the neural tube; cardiac muscle develops from cells in the anterior lateral plate mesoderm that migrate to the midline to form the primitive heart tube (2). Thus, both skeletal and cardiac muscles arise from particular populations of precursor cells. In contrast, smooth muscle cells have a much more heterogeneous and dispersed origin. Arterial myocyte precursors of the proximal aorta are derived from the cardiac neural crest, and so are ectodermal in origin (3). In contrast, arterial myocytes in more distal vessels (and in the coronary arteries) arise by condensation of local lateral mesoderm- derived mesenchyme surrounding the developing elastic arteries (3). Smooth muscle cells within veins are of less certain origin but probably also stem from mesodermal mesenchymal cells (4). The smooth muscles of viscera develop from local mesenchyme, under the influence of the epithelial buds they invest. During embryogenesis, airway smooth muscle markers can be detected early in the pseudoglandular stage (5) in newly differentiated smooth muscle cells that appear close to the tip of the advancing lung bud (6). The recently demonstrated functional innervation of these cells suggests that trophic factors they might release could possibly influence mesenchymal differentiation (6). These diverse origins of visceral and vascular smooth muscle cells are now thought to contribute to heterogeneity in their structural and functional properties (7, 8) and contractile apparatus gene expression (9).

	PHENOTYPIC HETEROGENEITY OF SMOOTH MUSCLE CELLS

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

In contrast to striated muscle cells, which undergo terminal differentiation from myoblast precursors to mature myocytes, smooth muscle cells retain an ability to modulate between mature and immature phenotypes. This phenotypic plasticity is well established in cell culture (12). Smooth muscle cells of vascular or visceral origin plated at low density in the presence of growth factor-replete fetal serum typically proliferate rapidly and express a range of matrix proteins, growth factors, and cytoskeletal proteins characteristic of the more immature, "synthetic-proliferative" phenotype. In contrast, confluent smooth muscle cells, especially when deprived of serum, characteristically exit the cell cycle and begin to re-express markers of the more differentiated "contractile" phenotype, especially proteins of the contractile apparatus. In fact, prolonged serum deprivation of cultured, confluent smooth muscle cells causes a subset of these cells to develop morphologic, structural, and functional features that more closely resemble the fully contractile phenotype typical of myocytes found within the arterial media or bronchial smooth muscle layer (13).

Phenotypic heterogeneity among smooth muscle cells appears important in both normal physiology and disease. In the pulmonary arterial wall, Frid and colleagues (7) have identified at least four types of smooth muscle cells that differ in their proliferative potential, pattern of gene expression, morphology, and location among the layers of the normal pulmonary arterial wall. Heterogeneity in velocity of contraction among individual systemic arterial myocytes has also been found, and appears related in part to differences in contractile protein contents (14). Halayko and coworkers (8) categorized at least three subsets of smooth muscle cells freshly isolated from normal canine airways, based upon their diverse sizes, contractile features, and proliferative responses in cell culture. Beyond this normal repertoire of smooth muscle cell phenotypic variation, phenotypic modulation of smooth muscle cells occurs in disease. Abnormal accumulation of synthetic-proliferative arterial myocytes leads to neointimal thickening in atherosclerotic plaques and during post-angioplasty vascular restenosis. It appears that phenotypic modulation, proliferation (possibly clonal [15]), and migration of previously contractile myocytes from the arterial media provides the source of these neointimal cells. Furthermore, arterial smooth muscle hypertrophy occurs in patients with systemic hypertension, and angiotensin II has been implicated in this process (16). In asthma, bronchial smooth muscle accumulates through hypertrophy and hyperplasia of resident cells (17), and both animal models (18) and theoretical considerations (19) suggest that these changes contribute causally to the constrictor hyperresponsiveness characteristic of this disease. This evidence for plasticity of smooth muscle structure and function clearly establishes that expression of the contractile apparatus is not fixed, but rather is regulated by cellular responses to normal and pathologic environmental cues.

	BRIEF OVERVIEW OF TRANSCRIPTIONAL REGULATION OF GENE EXPRESSION

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

Gene transcription is only one step in a cascade of events that leads to the presence or activity of a given protein within a particular cell. The cell's whole life experienceits ancestral embryonic origin, exposure to tissue matrices, other cells or trophic factors, and its current environmentdetermines its ability to express any of the 100,000 human genes at all, as well as its propensity to modulate expression of that gene in response to internal or external cues. Having arrived at its present state, the cell can regulate protein abundance or activity through control of: (1) gene transcription to mRNA; (2) mRNA abundance/ availability through mRNA processing, editing, survival, or sequestration; (3) mRNA translation to protein; (4) post-translational covalent or noncovalent protein modification; and (5) protein degradation or sequestration. Within muscle cells, contractile apparatus proteins appear to be controlled at each of these levels (20), including tissue-specific gene transcription.

In general, transcription of a particular gene in a given cell depends upon: (1) its chromatin structure at that time and place; (2) the presence of cis-acting DNA sequences within its promoter, and positive or negative transcriptional regulatory elements within its transcriptional regulatory regions; (3) the activator or repressor trans-acting nuclear proteins (transcription factors) that bind to these elements; and (4) binding of RNA polymerase II and the transcription initiation complex. (For review of gene transcription see Reference 23.) About 160 base pairs (bp) of chromosomal DNA are wound upon histone octamer cores in structures called nucleosomes, which themselves are joined by the DNA chain between nucleosomes (linker DNA) and a histone H1 molecule that constrains the linker DNA (24). This chromatin is further arranged in a helix, allowing for tight packing of chromosomal DNA into a small space. During gene transcription, histone H1 is released and the doubly wound DNA structure unwinds, making the DNA strand available to the nuclear proteins that participate in transcription. In some genes, cytosine residues within 5'-CG-3' (CpG) sequences can be methylated by nuclear methylases. Especially when clustered, methylated CpGs can bind repressor proteins (especially MeCP2) that inhibit gene transcription through mechanisms now under study (24), but which appear to prevent chromatin unwinding. Importantly, DNA methylation can be maintained through replication, and so provides one mechanism for long-term gene silencing. The recent identification of CpG islands within the human h1-calponin promoter (25) raises the possibility that methylation contributes to its smooth muscle-restricted expression.

Now uncoiled, the chromosomal DNA encoding a gene to be transcribed contains the gene itself (i.e., the exons and introns that are transcribed) plus additional transcriptional regulatory regions. DNA just 5' of the gene is called the promoter. Usually, the promoter contains a TATA sequence about 25 bp upstream from the transcription start site; this TATA box binds the ubiquitous TATA-binding protein (TBP) needed to anchor the transcription pre-initiation complex composed of RNA polymerase II and a variety of other protein factors. Characteristically, the promoter region also contains other DNA sequence motifs that provide specific binding sites for their target nuclear proteins. These DNA sequences (often 4-10 bp in length, often palindromic) are said to act in cis, because they are located on the same double helix as the gene. (The nuclear proteins that bind specifically to these sequences thus act in trans.) Some of these motifs bind proteins that upregulate transcription (activators), and others bind negatively acting transcription factors (repressors). In some cases, transcription factor DNA binding sequences are adjacent or directly overlap, so that their respective nuclear proteins compete for binding and transcriptional influence.

Nuclear proteins that serve as transcription factors typically have distinct domains with different functions (26). The DNA binding domain is usually an alpha helix with positively charged residues positioned to contact DNA. A separate domain may influence transcription itself or may interact with the same or another nuclear protein, generating a dimer whose composite structure influences transcription. Sometimes, even higher order complexes are formed (e.g., ternary complexes involving serum response factor). An example of combinatorial regulation of transcription factor activity is the interaction of the myogenic basic helix-loop-helix (bHLH) transcription factor, MyoD, to one of its several partners. Binding of a transcriptionally active heterodimer of the muscle-restricted MyoD with a ubiquitous HLH partner, E12, to the regulatory regions of several muscle-specific genes increases their transcription in differentiated myotubes (27). MyoD and E12 are actually also present in precursor myoblasts, but this more immature cell type also contains abundant Id, another HLH transcription factor that lacks an activation domain. Id competes with MyoD for E12 binding, and effectively sequesters it in a transcriptionally inactive E12-Id dimer. In addition to controlling transcriptional activity, such partnering with different transcription factors can apparently confer cell specificity to ubiquitous factors (e.g., serum response factor).

Beyond the expansive repertoire provided by transcription factor combinatorial interactions, other modifications to these nuclear proteins (e.g., phosphorylation, binding of hormones, etc.) can alter their affinity for one another, for their DNA binding sites, and even their presence within the nucleus. For example, nuclear factor kappa B (NF-B) is a transcription factor commonly activated by oxidant stress or exposure to various cytokines (28). It is a dimer composed of proteins from the rel family. Prior to activation, the NF-B dimer resides within the cytoplasm and is bound to a third protein IB, which covers the dimer's nuclear translocation signal and so prevents it from entering the nucleus. After appropriate stimulation, the cell activates NF-B by phosphorylating IB; IB becomes ubiquinated and degraded in the proteosome, thus releasing NF-B and exposing its nuclear localization signal, allowing NF-B translocation to the nucleus and DNA binding.

Individual nuclear factor binding sites within DNA also act in concert with other, often nonadjacent sites to form functional modules that enhance or repress gene transcription (2). Transcription of a single gene may thus be controlled in different ways in diverse cell types, or in different ways in a given cell under different conditions. While these transcriptional regulatory elements are often in the 5' flanking DNA of a gene, they sometimes occur within the gene or in the 3' flanking DNA. In some genes (e.g., the gene encoding skeletal myosin light chain 1/3), a critical enhancer element is located quite far from the gene (~ 24 kilobases [kb] downstream). Given the flexibility of the single helix of DNA that comprises each chromosome, it is easy to envision DNA looping and bending that facilitates approximation of nuclear proteins bound to widely separated DNA sites.

	SOME TRANSCRIPTION FACTORS FOUND TO PLAY IMPORTANT ROLES IN SPECIFIC SMOOTH MUSCLE GENE EXPRESSION

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

Serum Response Factor

Serum response factor (SRF) is a 67-kilodalton (kD) protein of the MADS family described initially for its role in activating the c-fos promoter in cultured cells stimulated with fetal serum. The c-fos promoter contains a cis-acting positive regulatory sequence, called the serum response element (SRE) (29). Dimerized SRF monomers bind in a ternary complex with a third 62-kD protein to the SRE, and this binding transactivates the promoter (30). The p62 ternary complex factor (TCF) has subsequently been identified as one of at least two proteins, Elk-1 and SAP-1, members of the Ets family of transcription factors. Since SRF seems to bind to the c-fos SRE before, during, and after serum stimulation, it seems likely that SRF binding per se is insufficient for transactivation. Rather, ternary complex formation and/or activation of Elk-1, SAP-1, or SRF by phosphorylation may enhance transcriptional activity (29). Phosphorylation of SRF can increase its affinity for its DNA binding site, but hyperphosphorylation in senscent cells decreases its binding (31). Also, association with other nuclear factors, such as the paired-like homeodomain protein, Phox1, or its murine counterpart, MHox, can increase the binding of SRF to its binding site (32). As such, the influence of SRF on a gene's transcription can be modulated in multiple ways.

The c-fos SRE has the CArG box sequenceCC(A/ T)₆GGat its core, and contains an adjacent Ets binding site just 5'. The Ets site is important for TCF binding and SRF- dependent transactivation of c-fos. Indeed, nuclear factor binding motifs that appear adjacent to, or even overlap, a CArG box in several genes often regulate their transcription by facilitating positive interactions with SRF (which at least in vitro binds a variety of other transcription factors, including p65relA [33]) or by competing for DNA binding. For example, the c-fos SRE can also bind YY1, NF-IL6, E12, and Phox1 (29). In several gene promoters, including that of the skeletal -actin gene, YY1 binding inhibits SRF binding and transcriptional activation (34).

SRF appears at low levels in most cell types, and can be induced as a "delayed" early response gene by serum stimulation. However, recent development studies in birds (35) and mice (36) localized substantial SRF expression primarily to the myogenic lineages. Many genes expressed selectively in skeletal or cardiac muscle (or indeed, in smooth muscle; see below) contain within their promoters CArG sequences whose binding to SRF is essential for maximal (or in some cases, any) transcription. SRF expression and activity are critical for the differentiation of skeletal myoblasts into mature myotubes (37); indeed, SRF mRNA and abundance among nuclear proteins increase 40-fold during avian embryonic myoblast differentiation (35). How does SRF expression increase so much? The SRF gene promoter itself requires the binding of SRF to either of two CArG boxes for maximal activity, implying that an autoregulatory positive feedback mechanism maintains high-level expression (36, 38). Inhibition of SRF expression (37) or activity (39) in skeletal myoblasts can prevent their differentiation into mature myotubes, and can even inhibit transcription of contractile protein genes in already differentiated skeletal myocytes (39).

Myocyte Enhancer Factor 2

Myocyte enhancer factor 2 (MEF2) is a group of four nuclear proteins (MEF2A-MEF2D) of the MADS family that share homology with SRF in its DNA-binding MADS domain (40). Similar to that for SRF, the MEF2 consensus binding motif is A + T-rich, CTA(A/ T)₄TA(G/A); MEF2 factors bind as homodimers or heterodimers to these sites, which are found in many muscle-specific genes (41). MEF2 expression appears critical for the development of all three myogenic lineages, because knockout of the single MEF2 homolog in Drosophila (D-MEF2) disrupts all muscle formation (42). MEF2 nuclear factors play important roles in the expression of skeletal and cardiac genes, but although their binding motifs are present within the transcriptional regulatory elements of smooth muscle contractile protein genes, no clearcut role for MEF2 had been established in the control of their expression until very recently. However, Katoh and colleagues (43) have now reported that MEF2B binds to an A + T-rich region in the rat smooth muscle myosin heavy chain promoter, thereby transactivating transcription from this gene.

Sp1

Sp1 is one member of a family of four zinc finger transcription factors (Sp1-Sp4). It was initially recognized for its role in activating the early promoter of the simian virus 40 (SV40). Expressed ubiquitously (though not constantly; see below), it plays an important role in the transcriptional activation of a wide range of genes, including "housekeeping" genes, cell proliferation genes, growth factors and their receptors, extracellular matrix protein genes, and muscle genes. In addition to its usual role as an activator of gene transcription, Sp1 binding can apparently inhibit gene expression, as suggested by mutational analysis of the upstream enhancer in the rat smooth muscle myosin heavy chain regulatory region (44, 45). Sp1 binds to a G + C-rich motif, the GC box, whose consensus sequence is (G/A)(G/A)GGCG(G/ T)(G/A)(G/A)(G/ T) (46). Like SRF and MEF2, Sp1 binding can be regulated through competition for DNA occupancy with other nuclear factors (e.g., Sp3 [47]) or post-translational modification of Sp1 protein through phosphorylation at its C-terminus by casein kinase II (which decreases in vitro DNA binding activity [48]), or phosphorylation by cAMP-dependent protein kinase A (which increases in vitro DNA binding activity [49]). Sp1 also contains O-linked N-acetylglucosamine residues; although glycosylation seems not to alter DNA binding of Sp1, hypoglycosylation (as occurs during glucose deprivation) does lead to accelerated proteosome degradation of Sp1 (50). Sp1 activity can also be reduced by interaction with other nuclear factors, including factors bound by the retinoblastoma gene product, Rb (51).

Recent studies implicate Sp1 in maintaining cells in a less differentiated state. Upon terminal differentiation of hepatocytes, Sp1 phosphorylation increases and its DNA binding activity is reduced (52). During skeletal myoblast differentiation to mature myotubes, Sp1 abundance becomes reduced, and this change can be mimicked by artificial overexpression of MyoD (53). This reduction in Sp1 has important functional consequences for gene expression. For example, transcription of the GLUT1 glucose transporter is reduced upon myoblast differentiation, likely due to the reduced binding and transactivation of the GLUT1 gene promoter by Sp1 (53). Interestingly, Id, the inhibitor of skeletal muscle differentiation, is also positively regulated by Sp1 (54). The fall in Sp1 in differentiating skeletal myocytes could therefore reduce Id expression and promote MyoD activity. However, it is uncertain whether such a mechanism pertains in smooth muscle. Recent studies show increased Sp1 immunoreactivity in neointimal smooth muscle cells, which express the less differentiated synthetic-proliferative phenotype (45).

Activator Protein 2

Activator protein 2 (AP2) is a 52-kD nuclear protein identified initially through its interactions with SV40 and metallothionein regulatory DNA sequences. During embryogenesis it is preferentially expressed in neural crest cells and their derivatives, surface ectoderm, spinal cord and hind-brain, and limb bud mesenchyme (55), and it plays a critical role in morphogenesis of the face, skeleton, and skin (59, 60), where it plays an important role in epidermal-specific gene expression (55). AP2 activity can be induced by retinoic acid exposure (61), and this transcription factor is thought to be one signaling intermediate for cAMP-mediated nuclear responses (57). It is unknown whether AP2 has any functional role in smooth muscle gene expression. Its consensus sequence has been variously identified, but includes G + C-rich sites akin to G(G/C)(G/C)(A/ T)G(G/C)CC.

Mhox

Mhox is a homeobox-containing transcription factor analogous to the paired gene in Drosophila. It plays a key role in vertebrate skeletogenesis and participates in the control of skeletal muscle-specific gene transcription. It binds to A + T-rich DNA regions and appears to facilitate binding of SRF to CArG sequences in muscle genes (62), though it can inhibit ternary complex formation on the serum response element of the c-fos promoter (69). Mhox appears to participate in the angiotension II-induced increase of smooth muscle -actin gene expression in quiescent smooth muscle cells (70), but its more general role in smooth muscle, if any, is uncertain.

Yin-Yang 1

Yin-Yang 1 (YY1) is a zinc finger containing transcription factor whose name stems from its dual ability to enhance or repress transcription in various genes; YY1 is the same protein as NF-E1, UCRBP, , and nuclear matrix protein-1 (71, 72). Its consensus motifs include two core sequences: CCAT or ACAT (73). YY1 can suppress gene transcription through multiple mechanisms, including competition with other activators for DNA binding at embedded or overlapping sites. As an example, YY1 can inhibit SRF binding by such competition (74), but it can also facilitate SRF association with the c-fos serum response element; apparently YY1 contacts DNA in the major groove while SRF contacts the same sequence in the minor groove (75). In this regard, it is interesting that YY1 overexpression can transactivate the mouse SM22 promoter (11). YY1 can initiate transcription; apparently YY1 binds to TFIIB, stabilizing its binding to DNA, and also contacts the large subunit of RNA polymerase II (76). YY1 can also interact with a wide variety of other nuclear factors, including c-myc (77) and Sp1 (78).

	COMMON MOTIFS IN SMOOTH MUSCLE CONTRACTILE APPARATUS GENE PROMOTER/REGULATORY REGIONS

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

The structure of several smooth muscle contractile apparatus genes have been studied in varying degrees of detail. SM22 is a 22-kD actin-binding protein that is ubiquitous in smooth muscle tissues, but is also expressed in senescent fibroblasts and as an immediate early gene in diverse cell types exposed to fetal serum. Although the function of SM22 remains unknown, immunocytochemistry reveals its presence with smooth muscle -actin in longitudinal cables in contractile cultured airway myocytes (13), suggesting it may participate in contraction. Smooth muscle -actin (sm -actin) and smooth muscle myosin heavy chain (sm-MHC) are, of course, central components of the contractile motor, and caldesmon is a thin filament protein that may have a role in the inhibition of smooth muscle contraction (79). Caldesmon is transcribed from two promoters, more active in brain or in gizzard (smooth muscle), respectively; we will focus on the promoter active in smooth muscle, which transcribes a longer mRNA that encodes the heavy h-caldesmon isoform. Telokin, also known as kinase- related peptide, is transcribed from a promoter within the third intron of the smooth muscle myosin light chain kinase (sm-MLCK) gene. Telokin protein is identical to the C-terminal 155 amino acids of sm-MLCK, and is thought to help maintain myosin filament polymerization even in the absence of myosin phosphorylation. Calponin is another thin filament protein thought to participate in relaxation of smooth muscle. In adult animals, SM22, sm-MHC, h-caldesmon, telokin, and calponin are highly restricted to smooth muscle tissues. Smooth muscle -actin is most prevalent in smooth muscle but can also be found in fibroblasts or myofibroblasts.

During development, and during phenotypic modulation of smooth muscle, each of these genes is expressed in its own characteristic temporal and spatial pattern. As such, it appears that there is no absolute coordination of their expressions. Nonetheless, under some important circumstances, levels of these proteins and their mRNAs do change roughly in parallel. For example, each species is more abundant in contractile phenotype smooth muscle cells than in those of the synthetic proliferative phenotype.

To gain insight into how this general coordination of gene expression might occur, it is instructive to examine these genes' promoters and regulatory regions (specifically, their 5' flanking DNA) for motifs known to bind the nuclear factors discussed above. First, we should note that the DNA sequences in important regulatory regions are often conserved across species (Figure 1). Thus, sequence homology within bp 350 to 1 in the SM22 promoters of humans, mice, and rats suggest the (now proven) importance of this region (80, 81). Likewise, sequence conservation of two regions of 5' flanking DNA of the sm-MHC promoters from rabbit and rat are also consistent with their known role in sm-MHC transcriptional regulation (82).

View larger version (29K): [in this window] [in a new window]   Figure 1.   Schematic illustration of potential transcription factor binding motifs found by sequence analysis of the promoter/regulatory regions of six smooth muscle contractile apparatus-related genes, along with a proposed "stereotype" promoter incorporating common features of these genes. Potential regulatory elements are coded according to fill pattern. Elements in the six smooth muscle gene promoters with known functional activity have no asterisk beneath; those elements with an asterisk have untested or no known activity. The number above each motif indicates its location relative to the most 5' transcriptional start site. Neg indicates that this element negatively regulates transcription of the rat smooth muscle myosin heavy chain gene. Question mark indicates that the putative element identified differs from the consensus sequence. Not all known or potential nuclear factor binding sites are shown.

Besides this close nucleotide sequence homology across species, the gene promoters for several smooth muscle contractile apparatus proteins also share "motif homology" to some extent. Figure 1 illustrates schematically the presence of several potential transcription factor binding sites within their 5' flanking regions. At the outset, several caveats must be acknowledged: (1) The motifs shown represent potential nuclear factor binding sites, and were identified based on their similarities to known consensus sequences. Some, but not all, of the potential sites identified have been shown to bind nuclear factors in smooth muscle, and only some of these sites have been found to regulate transcription in the six genes. Other sites have not been studied or are known not to influence transcription in the circumstances studied. (2) Not all cis-acting sequences are shown within each promoter. Undoubtedly there are additional motifs that bind nuclear proteins and influence gene expression, including DNA sequences that bind known nuclear proteins and those that bind unknown transcription factors. Furthermore, as noted above, multiple nuclear factors can compete for the same or nearby binding sites. Thus, trans-acting factors other than those implied by the motif name may exert influence from these sites. (3) Each potential transcriptional regulatory site is shown in ordinal arrangement (and the site location is designated relative to the most 5' transcriptional start site), but distances between sites are not represented proportionally within or among genes. For example, the pair of CArG boxes in the sm-MHC promoter is positioned much further upstream than are CArG sites in the other promoters.

Within these several limitations, Figure 1 shows the structure of a smooth muscle contractile gene promoter stereotype, which includes a pair of CArG boxes, with Sp1 and AP2 sites near the more 5' CArG site, and with a transforming growth factor beta (TGF-) control element (TGE) near the more 3' CArG. The core promoter region includes Sp1 and AP2 sites, with a TATA sequence located about 30 bp upstream of the transcriptional start site. Some of these genes contain potential YY1 binding sites, and some contain potential Mhox motifs.

	TRANSCRIPTION REGULATION OF SMOOTH MUSCLE CONTRACTILE GENE EXPRESSION

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

Investigators use several tools to elucidate transcriptional control mechanisms. One can discern which cis-acting DNA sequences bind nuclear proteins extracted from cells under circumstances of interest, using DNase I footprinting or electrophoretic mobility shift assay (EMSA) in vitro, or by performing DNase footprinting in vivo. Footprinting in vitro exploits the inability of DNase I to contact and digest double-stranded DNA bound to nuclear proteins. A segment of DNA from the region of interest is radiolabeled at one end (5' or 3'), incubated with nuclear protein extract, and partially digested with DNase I, then the resulting radiolabeled fragments (of various lengths) are resolved by gel chromatography. Sequences protected by DNase I digestion are revealed by their underrepresentation when compared against the same DNA region digested in the absence of nuclear proteins. In EMSA, a short segment of radiolabeled double-stranded DNA spanning one or several potential nuclear factor binding sites is incubated with nuclear extract. Nuclear proteins may form one or more complexes with the DNA probe and retard its mobility during gel electrophoresis. The identity of nuclear proteins within the complexes can often be established by further reduction of electrophoretic mobility (supershifting) or ablation of complex formations, including antibodies directed against a particular protein. An excess of unlabeled competitor DNA oligonucleotides whose protein binding preference is known can also inhibit radiolabeled probe complex formation. In both approaches, in vitro conditions may differ from those inside the cell nucleus, and results could possibly be misleading. In vivo footprinting is performed using genomic DNA inside the cell nucleus, and may reduce in vitro artifacts. Additional methods are available for identifying unknown proteins binding to DNA (83).

The functional role of cis-acting DNA sequences can be established by deleting or mutating these sites within the promoter or regulatory regions. Often, deletion-mutant or point-mutant promoters are linked to reporter genes that are easily measured (e.g., firefly luciferase or -galactosidase); then changes in abundance of the reporter imply changes in promoter activity. Recombinant promoter-reporter transgenes can be introduced into cells during transient or stable transfection, or used to generate transgenic animals in which reporter activity can reveal the temporal and spatial pattern of promoter/regulatory region activity throughout life.

These methods have been used to evaluate the function of several potential nuclear factor binding sites shown in Figure 1. To date, mouse SM22 expression has been the most comprehensively analyzed of the smooth muscle genes. Initial studies demonstrated that 445 bp of 5' flanking DNA were sufficient to program high-level, smooth muscle-specific reporter transcription in cultured cells (84). This same DNA segment restricts -galactosidase expression to arterial smooth muscle in adult transgenic mice (9); transient expression in the heart and somites has also been found. Interestingly, no reporter expression was found in visceral or venous smooth muscle even though myocytes in these tissues express abundant endogenous SM22 (4). This discrepancy has been attributed (9) to differences in myogenic lineages among these smooth muscle cells, which do have different embryonic origins. In this view, the nuclear factors within arterial smooth muscle cells are sufficient to activate the transgene promoter, but those within visceral smooth muscle are not; instead, factors acting at other regulatory regions not contained within the 445 bp must be required. Confounding this view, though, are the findings that: (1) the 445-bp SM22 promoter is highly active in cultured tracheal smooth muscle cells (85); (2) adenoviral delivery of the lacZ gene (which encodes -galactosidase) results in prominent, smooth muscle- restricted transgene expression in both arterial and visceral smooth muscle (86); and (3) including up to 5,000-bp 5' flanking DNA and/or the entire first intron of the SM22 gene (~ 4 kb) did not alter expression pattern in transgenic mice (9). While it is conceivable that the "missing" regulatory region that activates visceral muscle reporter expression falls outside this zone, an alternative explanation could possibly be that the transgene becomes methylated and thereby repressed in visceral and venous myocyte progenitors during development. Methylation of the human SM22 promoter prominently reduces its activity in vitro (81), and transgene methylation accounts for the discrepant axial distributions of transgene versus endogenous promoter activities in mice harboring a skeletal MLC1/3 promoter-chloramphenicol acyltransferase reporter transgene (87).

In both cultured cells and transgenic animals in vivo, the more 3' CArG box in the SM22 promoter plays an especially important positive regulatory role. Mutation of either CArG box within the 445-bp SM22 promoter reduced its activity by half during transient transfection in smooth muscle cells (11), though deletion of the more 5' CArG site in the human SM22 promoter had little, if any, effect (80). In transgenic mice, disruption of the 3' CArG site abolished promoter activity, but disruption of the 5' site appeared to reduce only embryonic myotomal expression and did not change the smooth muscle expression pattern (88). This greater importance of the more 3' CArG box in the SM22 gene is a theme that recurs in other smooth muscle gene promoters as well.

Footprinting and DNase I protection studies have begun to identify the nuclear factors in smooth muscle that transactivate the SM22 promoter (11). These include SRF, YY1, and Sp1, at least, and it seems likely that other factors participate as well. While it has been suspected that a ternary factor, possibly smooth muscle-specific, may interact with SRF during SM22 gene activation, no such factor has yet been identified. Although YY1 is often a negative regulator, it appears to have a positive regulatory role in SM22 transcription (11).

The smooth muscle -actin gene is interesting in that its transcriptional control in non-smooth muscle cells relies on repression of gene expression (89), and this repression was effected by different regions upstream of the promoter region shown in Figure 1 among diverse non-smooth muscle cell types. Thus, transcriptional regulation of the sm -actin gene is especially dependent upon the cell type in which it is studied. In cultured smooth muscle cells, both CArG boxes are critical for promoter activity (70, 90); mutation of either abolishes reporter expression. Two other features of this promoter may also have relevance for the panel of smooth muscle genes we are considering. First, an Mhox consensus sequence near the 5' CArG box has been shown to bind Mhox, and is necessary for the full increase in promoter activity in quiescent arterial myocytes exposed to angiotensin II (70). This motif is found near or within the 5' CArG sequences of the sm-MHC and telokin promoters as well, though it is unknown what role the site plays in their regulation. Second, a TCE appears 3' of the proximal CArG site (90). Though it is often inhibitory in non-muscle gene promoters (91), mutation of the TCE in the sm -actin promoter both reduces basal activity and virtually abolishes TGF--induced promoter activation. This TGF- effect also requires protein synthesis and intact CArG boxes, and appears to involve increased SRF synthesis and CArG box binding. Interestingly, the TCE consensus sequence can be found in the SM22, sm-MHC, and calponin promoters, and a near-consensus sequence is also present at the anticipated position in the chicken gizzard caldesmon promoter as well. It is intriguing that deletion of a DNA fragment including the potential TCE within the caldesmon gene ablates the increment in promoter activity found in differentiated (versus de-differentiated) smooth muscle cells (92). Thus, the TCE might contribute to the increased expression of contractile apparatus proteins in contractile phenotype smooth muscle cells.

The smooth muscle myosin heavy chain promoter has been studied with some variation in results among investigators (44, 45, 82, 93, 94), using promoters from different species and diverse cell culture conditions. The rat sm-MHC promoter shown in Figure 1 follows the "smooth muscle promoter stereotype" well. Mutation of either CArG site reduces activity substantially during transient transfection into aortic smooth muscle cells, and mutation of both sites has an additive effect (45). (However, mutation of a third, even more 5' CArG site [not shown in Figure 1] was without effect [44].) Interestingly, the Sp1 motif adjacent to the functional 5' CArG motif behaves as a negative regulatory element in that mutation of this site increases promoter activity. EMSA analysis showed that this GC-rich motif does bind Sp1 in vitro and footprinting showed it binds nuclear factors in vivo (45). Furthermore, Sp1 immunoreactivity is increased in the neointima of arterial lesions, where sm-MHC expression is reduced (45). Whether this is a principal mechanism for downregulation of sm-MHC transcription (or transcription of any other smooth muscle contractile gene) in synthetic-proliferative myocytes is unknown.

The promoters of the gizzard caldesmon (92), telokin (95, 96), and calponin (25, 97, 98) genes have been studied much less extensively than those of SM22, sm -actin, or sm-MHC. The 3' CArG box in the caldesmon and telokin genes are critical for substantial promoter activity, but deleting the 5' CArG of either promoter is without effect in transient transfection assay. Although previous reports (92, 93) have noted the absence of CArG motifs in the mouse calponin promoter, there are two sequences (CCATAgAgGG and CCAgAATAAGG, located at bp 413 and bp 203, respectively) that might conceivably bind SRF. Supporting this possibility, deletion of bp 549 to 116 virtually ablates calponin promoter activity in smooth muscle cells (98).

	SPECULATION ON THE INHIBITION OF CONTRACTILE APPARATUS EXPRESSION AS A THERAPY FOR ASTHMA

TOP ABSTRACT INTRODUCTION EMBRYOLOGIC ORIGINS OF SMOOTH PHENOTYPIC HETEROGENEITY OF BRIEF OVERVIEW OF SOME TRANSCRIPTION FACTORS FOUND COMMON MOTIFS IN SMOOTH MUSCLE TRANSCRIPTION REGULATION OF SPECULATION ON THE INHIBITION OF REFERENCES

It seems likely that, in the six genes analyzed, SRF plays a substantial regulatory role, both in activating basal gene expression and in further augmenting or maintaining smooth muscle contractile apparatus transcription in contractile phenotype cells. Given this central role, SRF abundance and DNA-binding activity could provide the mechanism through which smooth muscle cells coordinate transcription of these genes.

This central role of SRF in activating multiple contractile apparatus genes makes it a tempting target for pharmacologic intervention to suppress contractile apparatus protein expression, which would undoubtedly weaken airway or pulmonary vascular smooth muscle, and could possibly ameliorate bronchoconstriction in asthma or vasoconstriction in pulmonary hypertension. In asthma, it seems obvious that preventing bronchoconstriction would prevent acute asthma attacks, and that without acute airway narrowing asthma would be a much more tolerable disease. Ablating pulmonary arterial tone in patients with pulmonary vascular disease could also be expected to reduce pulmonary arterial pressure in some patients, and this should also be of notable benefit. While in normal individuals there may be some physiologic advantage for maintaining airway or pulmonary vascular tone, the potential therapeutic benefits of weakening airway or pulmonary vascular smooth muscle in disease also seem obvious. Whether the potential detrimental effects of such weakening outweigh the potential therapeutic advantage is now purely speculation.

Could antitranscriptional therapy actually work? Could one inhibit smooth muscle contractile apparatus expression by inhibiting SRF activity? Though untried in smooth muscle, expression of anti-sense mRNA in skeletal myocytes blocked their differentiation (and their proliferation as myoblasts) (37). Also, distamycin A (an antibiotic that binds DNA in the minor groove in A + T-rich regions) has been shown to inhibit SRF binding to its consensus CArG motif, to inhibit differentiation of cultured skeletal myoblasts in a reversible fashion, and to inhibit myosin and actin gene transcription in differentiated myotubes (39). Of course, nonselective intervention against SRF would be ill-advised, given its key role in cell proliferation and striated muscle gene expression. However, perhaps selective intervention against binding sites peculiar to smooth muscle genes (e.g., regions flanking CArG boxes) could be designed. Then it might be possible to downregulate one or multiple smooth muscle contractile protein genes in a less toxic way. Perhaps such a therapy could hold promise in chronic, severe asthma.