OUP user menu

TGF-β and atherosclerosis in man

David J. Grainger
DOI: http://dx.doi.org/10.1016/j.cardiores.2007.02.022 213-222 First published online: 1 May 2007


The transforming growth factor type-β (TGF-β) superfamily of ligands, receptors, binding proteins and ligand traps together plays a key role in the maintenance of normal blood vessel wall structure. Specific defects in genes encoding superfamily members have now been linked to a range of cardiovascular syndromes involving loss of healthy vessel architecture, including hypertension and aneurysm. However the contribution of TGF-β to the development of atherosclerosis is simultaneously more subtle and more complex. TGF-β ligands are produced by a range of different cell types, which also regulate release of the active cytokine that, in turn, signals through multiple receptor complexes on different cell types. Recent evidence suggests that the T cell may be both a key source of TGF-β1 and a key target for its effects during atherogenesis, as in other chronic inflammatory disorders. Here we review the evidence for the role of TGF-β in the human vasculature during atherogenesis, and evaluate the available data in the context of our knowledge from animal models of the disease.

  • Blood vessel
  • T cell
  • Autoimmunity
  • Inflammation

The development of atherosclerosis is a complex process that can begin in the first years of life (possibly even before birth) and then continues progressively for decades [1]. In the early stages, the concentric three-layer structure of the healthy vessel wall is gradually lost as smooth muscle cells encroach into the intima (either through migration from the vessel media or recruitment of circulating precursor cells). This expanded intima can become a site of chronic (and probably episodic) inflammation, with recruitment of macrophages and T cells, likely as a result of gene expression changes in the overlying endothelium. Gradually, if the conditions are right, cholesterol-rich lipoproteins can accumulate at the site of the developing lesion, further promoting local inflammation. Over many years, a plaque can form which encroaches on the vessel lumen and inhibits blood flow, which in the coronary arteries can lead to myocardial ischemia, pain and shortness of breath that typify angina pectoris [1].

In some individuals, for reasons that remain incompletely understood, plaques can become unstable (associated histologically with increased macrophage content and calcification, as well as reduced extracellular matrix (ECM) deposition) [1,2]. Unstable plaques are prone to acute rupture, leading to thrombus formation and sudden closure of the affected vessel. Unless the lumen is rapidly recanalised (either naturally or through medical intervention) infarction can result, which, with sufficient loss of tissue, may be fatal.

Cytokines have long been considered to play a central role in orchestrating the varied cellular changes that contribute to the gradual architectural changes typical of atherogenesis. In the 1970s and 80s, considerable attention was focussed on growth factors, particularly for smooth muscle cells (such as PDGF and FGF) [3]. The original concept that smooth muscle proliferation was a major contributor to the development of atherosclerosis is somewhat simplistic [4,5]. Nevertheless many of the same molecules are now known to play important roles, either as growth factors for other cell types, or as survival factors or pro-migratory cues. Through the 1990s, focus shifted to the complex web of immunomodulatory cytokines: chemokines, various interleukins and TNF-α all apparently promote vascular inflammation [6]. In contrast, members of the TGF-β superfamily, together with IL-10, exert a constitutive anti-inflammatory action [7]. Ultimately, the balance of pro- and anti-inflammatory signals within the vessel wall will determine the extent of vascular inflammation.

Among these immunomodulatory cytokines, TGF-β is of particular interest to cardiovascular biologists because of its wide range of actions on all the different cell types that compose the blood vessel wall [8–11]. There are three closely related isoforms of TGF-β, designated TGF-β1, TGF-β2 and TGF-β3, in mammals, all of which bind to the same receptors and, in most cases where comparative data is available, exert similar functions. The major difference between the isoforms seems to be the spatiotemporal control of their expression patterns. In addition, there are more than 20 cytokines, which together form the TGF-β superfamily, a structurally related group of proteins with diverse functions particularly during embryonic development. Most of the data presented here relates to TGF-β1, which has been studied most extensively, particularly in relation to atherogenesis, although the studies of TGF-β receptor modulation do not distinguish which ligand was responsible for the reported effects.

In smooth muscle cell cultures, TGF-β1 dramatically stimulates ECM production [12,13], and in particular various collagens, as well as (under most conditions) inhibiting cell proliferation [14–16] and increasing expression of contractile proteins [16]. In endothelial cell cultures, TGF-β1 inhibits proliferation and migration, as well as expression of adhesion molecules involved in leukocyte recruitment [17]. In addition to an indirect anti-inflammatory activity mediated through suppression of endothelial cell activation, TGF-β1 has a range of direct effects on immune cells in culture, including inhibition of foam cell formation in cultured macrophages [18]. Taken together, these activities in vitro are consistent with the hypothesis that TGF-β1 plays a role in the maintenance of normal blood vessel wall architecture.

A range of strategies have been employed to modulate the levels of TGF-β1 in vivo, in order to test this hypothesis. These animal model experiments, which have been comprehensively reviewed elsewhere [10,11,19,20], are here summarised, and the available evidence from human studies is systematically reviewed to assess to what extent the findings in animals are likely also to be true in man.

1 Defects in ligand production and activation

In animals, reduced availability of the TGF-β1 ligand leads to pro-atherogenic changes in the blood vessel wall. Deletion of a single allele of the tgfb1 gene (which reduces the amount of TGF-β1 protein in the vessel media by ∼ 50%) results in reduced SMC differentiation, marked by levels of SM-α-actin and SM-myosin heavy chain, and increased susceptibility to endothelial cell activation and vascular lipid lesion formation in response to pro-atherogenic stimuli such as a lipid-rich diet [21]. Similarly, treatment with neutralising anti-TGF-β1 antibodies led to increased vascular inflammation, accelerated lipid lesion formation and a shift in plaque morphology towards an unstable phenotype [22].

In man, where similar specific interventions to lower TGF-β1 ligand availability are not possible, several studies have instead examined whether natural variation in TGF-β1 ligand levels is associated with increased risk of heart disease. One approach is to directly measure levels of TGF-β1 ligand either in blood fractions or in tissues, comparing individuals with and without heart disease, and the results from these studies are summarised in Table 1. Such studies are hampered by the presence of multiple TGF-β1-containing protein complexes, including various latent forms (with or without disulphide bonded binding proteins) and many non-covalent complexes with a range of serum components and ECM proteins [23]. We reported an increase in latent TGF-β1 and a marked decrease in receptor-binding (or ‘active’) TGF-β1 among individuals with severe atherosclerosis, identified by angiography [24], an observation consistent with observations by a number of other laboratories [25,26]. However, yet other groups have reported either no statistically significant difference or even an increase in TGF-β activity [27], possibly reflecting differences in assay methodology [23]. As a result, no firm conclusion can be reached, at least until we have a definitive statement of which TGF-β1 complexes are detected by each assay methodology, which remains elusive.

View this table:
Table 1

Summary of published studies comparing levels of TGF-β 1 in blood fractions from patients with varying degrees of cardiovascular disease

Definition of diseaseSample sizeBlood fractionAssayMean (ng/ml)Reference
Angiography; >50% stenosis in 3 arteries61SerumaRef. [100] (antigen)5.68.5[24]
Ref. [100] (active)4.0<1*[24]
Angiography; 50% stenosis of at least 1 artery371PlasmaPromega (antigen)54.656.9[27]
Promega (active)0.961.74 *[27]
Angiography; 50% stenosis of at least 1 artery109PlasmaRef. [101] (antigen)6.36.4[25]
Amersham (active)0.100.05 *[25]
Angiography; correlation with severity89SerumR&D (antigen)b27 (p for trend=0.058)c[26]
MI, positive angiogram or angina+exercise test155SerumR&D (antigen)35.126.1*[102]
  • For each study, the criteria used to define the extent of coronary artery disease are shown, together with the total number of patients (all sub-groups) analysed. A wide variety of different assay systems have been used: where the assay is not commercially available, the reference describing the assay is given; for commercial assays, the manufacturer is shown. The mean value reported for controls and diseased individuals is shown (without errors, because of the lack of consistent reporting of errors in the primary publications). *=p<0.05, disease versus control.

  • a This assay is reported not to measure platelet TGF-β1 and therefore to report the same value in both serum and plasma.

  • b The primary publication reports these values as active TGF-β1, but the methods used would measure antigen levels.

  • c All the individuals in this study had CHD of varying severity, and the p value quoted is for a linear trend of decreasing TGF-β1 from mild to severe disease.

Animal model studies increasingly suggest that local (rather than systemic) alterations in TGF-β activity may be important during atherogenesis. As a result, examining TGF-β levels in tissues may be more informative than in blood fractions. Several studies suggest that TGF-β levels are reduced at sites of atherosclerotic plaque development [28], and also at sites of low haemodynamic sheer stress which are susceptible to development of lesions [29] (consistent with more comprehensive studies in animal models [30,31]), while other studies show increased levels of TGF-β1 protein [32,33] or a complex association depending on the stage of plaque development [34]. However, once again doubts about the selectivity of antibody probes for various TGF-β1 complexes [23] coupled with significant problems in normalising for changes in cellularity as vessel wall architecture changes preclude a definitive conclusion. Unfortunately, measurement of mRNA levels (commonly used to circumvent problems with antibody specificities) do not materially clarify the situation because there is no reason to suppose the majority of the TGF-β present in the vessel wall is locally synthesised.

More recently, focus has shifted to the leukocyte as a major source of TGF-β, particularly since manipulations that increase leukocyte production of TGF-β reduce transplant-induced atherosclerosis in mice [35]. Kempf and colleagues demonstrated that TGF-β1 expression in circulating leukocytes was decreased in individuals who suffered acute myocardial infarction [36]. Further studies of this kind, although challenging in large cohorts, seem likely to be informative.

Since direct measurement of ligand availability is so technically demanding, then associations between genetic polymorphisms known to modulate ligand production and heart disease might prove more accessible. Furthermore any such associations would support a causal relationship between altered TGF-β production and disease. Fortunately, several common genetic variants in the tgfb1 promoter (such as A-800G and T-509C) have been associated with altered plasma levels of TGF-b1 ligand [37], and (in the case of T-509C) with differential transcription factor binding and transcriptional activity [38,39]. Two other polymorphisms give rise to amino acid substitutions in the TGF-β1 signal sequence (Leu10Pro and Arg25Pro), and have been shown to modulate TGF-β1 secretion by various cell types [40–43]. A number of studies have now examined the association between these polymorphisms and cardiovascular disease status (summarised in Table 2), but the results have been as complex to interpret as the attempts to directly measure TGF-β1 ligand levels.

View this table:
Table 2

Summary of published studies examining the association between genotype at the tgfb1 locus and coronary heart disease

PopulationDefinition of diseaseSample sizePolymorphisms tested (significance)Reference
EuropeanAngiography563629−988 (ns);−800 (ns),−509 (ns);+72 (ns); L10P (ns); R25P (ns); T263I (ns)[44]
MI− 988 (ns);− 800 (ns),− 509 (ns);+72 (ns); L10P (ns); R25P (p<0.05); T263I (ns)[44]
EuropeanAngiography655244−800 (ns);−509 (ns); L10P (ns); R25P (ns); T263I (ns)[45]
AustralianAngiography3710−509 (ns)a[46]
JapaneseMI315591L10P (p<0.001)[43]
EuropeanMI36571211−509 (p<0.01); L10P, R25P, T263I (p<0.05 as haplotype with − 509)[48]
EuropeanMI3585558−800 (ns);−509 (ns); L10P (ns); R25P (ns); T263I (ns)[49]
JapaneseAutopsy2481255L10P (p<0.05)[51]
  • For each study, the technique used to define the presence of coronary heart disease is shown, together with the number of patients and controls analysed. Every polymorphism studied is listed (whether significantly associated with the presence of disease or not): polymorphisms that do not affect coding are given by position relative to the major transcriptional start site of the tgfb1 gene; non-synonymous polymorphisms are described by their position in preproTGF-β1, using single-letter amino acid designations. The significance of each association is shown (ns = non-significant), with significant associations highlighted in bold. All significant associations reported are in the same direction, with the haplotypes previously linked with lower TGF-β1 production and/or plasma levels associated with an increased risk of disease. Additional associations with non-coronary phenotypes reported in the same manuscripts (e.g. hypertension, stroke) are omitted for clarity.

  • a This study tested for association with severity of angiographic disease among patients.

The first genetic association study, by Cambien et al. in the ECTIM cohort, found an association between tgfb1 polymorphisms and myocardial infarction, but no association with stenotic lesion burden by angiography and an inverse relationship with hypertension [44] (somewhat surprisingly, since hypertension is a risk factor for myocardial infarction). Follow-up studies have largely confirmed these findings: no association between five different polymorphisms and angiographic disease in three different cohorts [45–47], but a strong association between the same polymorphisms and myocardial infarction in a Japanese population [43] and a large German population [48]. However, the largest study to date (in more than 6000 individuals from the Rotterdam study [49]) found no association between the same polymorphisms and myocardial infarction, although a significant association with stroke (another pathology associated with plaque rupture, but in a different vascular field) was identified [49,50]. Most recently, using an autopsy definition of atherosclerosis in a Japanese population, Oda and colleagues report a significant association with the only tgfb1 gene polymorphism they studied, at least in some artery fields [51].

Taken together, these studies suggest lowered TGF-β1 ligand production might favour unstable lesion phenotype (and hence myocardial infarction) without affecting plaque burden, once again highlighting the need to select carefully the cardiovascular end-point under study, and not assume studies with an angiographic end-point will necessarily have the same outcome as those with an event-based end-point. Interestingly, both the Japanese [43] and German [48] studies suggested the strength of the association differed between men and women, but in both studies the number of events in women was 3-fold lower than in men and the studies were inadequately powered to determine whether the association between tgfb1 polymorphisms and myocardial infarction really differs between the sexes.

To date, therefore, both direct measurement of TGF-β1 ligand and genetic association studies, provide some indication that lower levels of TGF-β1 activity predispose to unstable atherosclerotic disease in man, as in animals, but the dataset cannot be considered conclusive. This is not, perhaps, very surprising: TGF-β1 (like the superfamily as a whole) has diverse impacts on many organ systems, and global modulation of TGF-β1 levels will likely have a wide range of different outcomes. Even in mouse, we know that heterozygous deletion of tgfb1 increases susceptibility to cancer [52] as well as atherosclerosis [21]. In the human population, with mixed genetic backgrounds and myriad environmental influences, associations between systemic TGF-β1 and atherosclerotic heart disease will be weakened by the competition between diverse pathological outcomes, even if the molecular link is as strong in man as in rodents.

2 Defects in TGF-β receptors and signalling

Even if ligand levels remain normal, then disrupted or altered cellular signalling could contribute to changes in blood vessel wall architecture in cardiovascular pathologies. In mouse, suppression of TGF-β signalling through expression of a dominant negative type II TGF-β receptor (dnTβRII), either systemically [53] or selectively in T cells [54,55], results in accelerated lipid lesions formation, increased vascular inflammation and a shift to an unstable lesion phenotype similar to that observed by ligand reduction strategies [21,22].

Productive TGF-β signalling requires the formation of a multimeric receptor complex involving both type II and type I TGF-β receptors [56]. Although there is only a single type II receptor protein (TβRII), there are multiple type I receptors (such as alk1–alk6) with different expression patterns. This raises the possibility that the same ligand can elicit a multitude of different effects depending on the particular portfolio of type I receptors present on a particular cell at a particular time [56]. Worse still, assembly of the productive signalling complex probably also requires low-affinity accessory receptors which also differ between cell types (such as endoglin, decorin and betaglycan). Intracellular signalling in response to TGF-β1 has been extensively reviewed elsewhere [57]: the Smad family of transcription factors are important for mediating many of the cellular response to TGF-β1, but Smad-independent effects (perhaps mediated directly by kinase cascades such as MAP kinase or mTOR pathways) have also been described [58] (Fig. 1).

Fig. 1

Major TGF-β signalling pathways. Various latent TGF-β complexes (such as the small latent complex, consisting of the 25 kDa mature dimer (yellow diamonds) and the LAP dimer (red lines)) are activated, most likely at the cell surface and bind to accessory type III receptors (blue ovals), such as endoglin, betaglycan or decorin. The activated dimer then assembles the productive signalling heteromeric receptor complex. This process is thought to be similar for all three TGF-b isoforms. In the signalling receptor complex, the type II receptor (TbR-II) then phosphorylates the TBR-I (such as alk5) and activates its intracellular serine/threonine kinase activity, resulting in multiple intracellular signalling cascades. For example, the R-Smads (primarily Smad2 and Smad3; green oval) are phosphorylated, bind to the co-Smad, Smad4 (orange oval) and are translocated to the nucleus, where (together with various co-transcription factors; grey box) they modulate transcription of target genes. In response, the I-Smads (primarily Smad6) and Smurf are exported to the cytoplasm, bind to TbR-I and suppress further signalling, as well as promoting receptor degradation. Various non-Smad pathways are also activated including the small GTPase Rho, which promotes actin polymerisation, and various members of the MAP kinase family (including p38 and JNK; blue squares). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Perhaps because of the complexity associated with type I receptor and accessory receptor expression patterns, there has been much focus on the level of TβRII in human atherosclerosis. MacCaffrey and colleagues [59] demonstrated that SMCs isolated from atherosclerotic plaque tissue expressed lower levels TβRII than SMCs from healthy vessel wall (which correlated with a shift from a pro-differentiation response to TGF-β ligand to a powerful fibrotic response). They went on to suggest that somatic mutation of an A10 run in the TβRII gene sequence, which had previously been identified as common in replication error-rate positive (RER+) colon cancers [60], was responsible for this loss of TβRII expression in plaque [61]. A careful re-evaluation of the methodology used to detect frameshift mutations in the polyadenosine tract casts considerable doubt on this somatic mutation origin for TGF-β resistance during plaque development [62], but the observation of lower type II receptor expression nevertheless seems robust.

Very little is known about changes in the expression pattern of type I receptors during atherogenesis, although the major type I receptor alk5 is known to be expressed in both healthy vessel wall and plaque intima, at least in fibrofatty lesions [32]. In contrast, significant disease-associated increases in the level of the accessory receptors endoglin and decorin have been reported, both in tissue studies [33,63] and in the circulation [64]. Unfortunately, since these receptors can apparently act as both positive and negative regulators of TGF-β signalling (either by assisting formation of the productive signalling receptor complex, or by sequestering TGF-β into inactive compartments) it is difficult to interpret the impact of these changes.

Changes in other components of the TGF-β signalling system have also been described. Kalinina and colleagues [65] report a marked reduction in the levels of the R-Smads, Smad2 and Smad3, as well as the ubiquitous co-Smad, Smad4, in SMCs in the intima of fibrofatty lesions. Again, these findings suggest that even if TGF-β activity is unaffected, the capacity to respond to the protective signal is modulated or lost during atherogenesis.

Interestingly, a range of germ-line mutations in the various TGF-β receptors cause defects in vessel wall structure, underlining the importance of the TGF-β system in patterning the vessel wall during development and in maintaining that architecture in the adult. The first such association to be described was between endoglin mutations and hereditary telangectasia [66], where capillary patterning is derailed leading to knots of malformed capillaries prone to leakage and bleeding. The TGF-β type II receptor is apparently vital for vasculogenesis, since deletion of the gene encoding TβRII in mice results in early embryonic lethality due to failure to vascularise the yolk-sac [67], suggesting loss-of-function mutations would also be unviable in man. However, point mutations in the high-affinity TGF-β receptors have recently been associated with more subtle changes in vessel wall architecture: both Loeys–Deitz Syndrome and a vascular form of the Ehlers–Danlos Syndrome (both of which are characterised by aggressive arterial aneurysms which lead to death usually in the third decade) have been associated with a range of point mutations in the either the type I or type II TGF-β receptor [68]. Other point mutations in these receptors have been associated with Marfan Syndrome-related disorders [69,70] (again characterised by aortic root dilatation and aneurysm), which is particularly interesting since classical Marfan Syndrome is caused by mutations in the fibrillin-1 (or-2) gene [71–73], and fibrillins are members of the Latent TGF-β Binding Protein (LTBP) family implicated in localising TGF-β to the ECM [74]. However, the most recent study associates excess TGF-β activity (presumably resulting from the failure to localise it into the ECM) with the impaired vessel wall repair underlying aneurysm susceptibility in Marfan Syndrome [75] (a conclusion which does not sit comfortably with the association between multiple point mutations in TGF-β signalling receptors and aneurysm formation [68]).

Clearly, other TGF-β-related cytokines also have important functions in regulating vessel wall architecture, since primary pulmonary hypertension has been associated with mutations in the type II receptor for bone morphogenetic proteins (BMP-RII) [76,77], as well as, in rare cases, with the alk1 type I receptor [77], although non-vascular effects (such as on lung fibroblast myotransdifferentiation) may also contribute to the pathological consequences of receptor loss.

Taken together, these observations are consistent with a key role for TGF-β (and possibly other members of the TGF-β superfamily) in developmental patterning of the vasculature, and probably also in maintenance of that structure in the adult. Rare mutations which result in significant reductions in TGF-β signalling seem to cause aneurysms presumably as a result of reduced ECM synthesis. However, the consequences of polymorphic variants in the receptors and signalling proteins of the TGF-β system, which may cause much milder variations in signalling capacity, remain almost entirely unexplored.

3 Immune system function: a mirror of TGF-β signalling?

A number of factors hamper direct investigation of TGF-β function in man. Firstly, specific interventions to modulate TGF-β levels are not possible in man. Secondly, we are usually limited to measurements on blood fractions (which likely only reflect systemic disturbances in the TGF-β system) or to post-mortem tissue. Finally, there is a range of definitions of cardiovascular disease (such as the presence of angiographic stenosis or the occurrence of acute myocardial infarction events) which may result from different underlying factors.

An alternative, complementary, approach to direct observation is analysis of the downstream effects of TGF-β on the immune system. Although complex, the role of TGF-β in the immune system is gradually being teased out in animal model studies, and there is now good evidence that a certain functions are heavily dependent on local TGF-β activity. For example, antibody class switching during B cell maturation proceeds down different paths depending on the local cytokine profile: in the presence of high levels of TGF-β activity heavy chain switching to IgA predominates [78], while IgG1/IgE is indicative of a Th2 cytokine profile and IgG2a marks a Th1 profile (at least in mice) [79]. The relative levels of each isotype in the circulation may therefore be a mirror of the recent cytokine profile of the individual.

Unfortunately, few studies have examined the relative levels of different isotypes, instead concentrating on the total levels of each, which may reflect overall activation of antibody production rather than class switching [80]. However, our detailed analysis of antibody isotypes against thymus-independent antigens revealed a major difference in isotype switching between individuals with and without angiographically-defined heart disease [81]. This seems likely to reflect a substantial difference in the cytokine profile of the individuals, but whether TGF-β in particular contributes to this difference remains the subject of further investigation. A much larger study of isotype switching is underway in the Metabonomics and Genomics in Coronary Artery Disease (MaGiCAD) cohort [82] to determine whether such measures can be used clinically to aid in the diagnosis of coronary heart disease.

Other immune system parameters may also act as indirect markers for TGF-β activity. TGF-β suppresses T cell activation, and the number of mature helper T cells may be an indication of local TGF-β activity. Although there has been considerable focus on the importance of polarisation of the T-helper cell population (towards a Th1 pattern of cytokine secretion) in atherosclerosis in recent years [83,84], there is much less information on the overall extent of T cell activation. Alternatively, the number of natural regulatory T cells (the nTreg population) may be a surrogate marker for local TGF-β activity, since this cell population are a major source of immunoregulatory TGF-β in other tissues and diseases. Low nTreg numbers may be indicative of reduced local TGF-β activity, and in turn contribute to the many pro-inflammatory aspects of atherogenesis, including autoimmune responses to plaque components.

Drawing inferences about the status of the TGF-β system from immune system parameters is clearly fraught with difficulties: other cytokines, such as IL-10 have overlapping activities with TGF-β (for example, both cytokines suppress T cell activation, although only TGF-β apparently regulates IgA class switching). Despite these difficulties, it seems that such indirect assessment of immune system parameters can yield insight into TGF-β activity during atherogenesis, particularly if combined with parallel studies in animal models (such as tgfb1+/−mice) where the causal relationships can be definitively identified.

4 Upregulating TGF-β

If depressed TGF-β signalling causally contributes to the vessel wall changes in atherosclerosis, then upregulating TGF-β to the right degree, in the right place at the right time should be atheroprotective. In animal models, evidence is now accumulating that this is indeed the case, since adenoviral transfer of an activated TGF-β1-expressing construct, resulting in a systemic increase in TGF-β1, has been shown to prevent lipid lesion formation in mice [85] and overexpression of TGF-β by T cells reduces transplant-induced arteriosclerosis [35,86]. Excessive TGF-β activity, however, can be as detrimental to blood vessel wall structure as low levels and direct over-expression of activated TGF-β in the vessel media leads to expansion of a fibrotic intima [87], emphasising the importance of spatial and temporal context for the TGF-β signal.

In man, studies have been limited to non-specific pharmacological agents that happen to stimulate TGF-β activity in addition to other effects. The most widely cited such examples are the Selective Estrogen Receptor Modulators (SERMs) of the triphenylethylene class, such as Tamoxifen [88]. These drugs increase TGF-β in many cell types in culture, including SMCs of rodent [89] and human origin [90], as well as in rodents in vivo [91–93]. Unfortunately, it is much more difficult to determine directly whether they stimulate TGF-β activity in man. However, an array of indirect markers of TGF-β activity including endothelial function and immune system parameters [94] in a study of men treated with Tamoxifen for 3 months suggest that locally TGF-β activity may indeed have been increased as in the rodents.

Convincing evidence exists that several other drugs in common use also stimulate TGF-β activity, including the immunosuppressant antibiotic Rapamycin [95] and a number of HMG-CoA reductase inhibitors of the statin class [96,97]. Increased TGF-β activity could plausible be responsible for the immunomodulatory activity which contributes to the clinical efficacy of these agents [98], but (as for SERMs) direct evidence implicating elevated TGF-β in the atheroprotective effects observed will be difficult to obtain.

Taken together, the available evidence suggests that upregulating TGF-β activity is a viable therapeutic option, either to prevent or reduce atherosclerotic plaque formation, or more likely to induce a transition from a macrophage-rich unstable plaque phenotype to a more stable, fibrous lesion [20,99]. Partly because the potential number of contributing mechanisms are manifold (including direct stimulation of ECM production by intimal SMCs, altered macrophage function and recruitment, inhibitory activities on helper T cells, suppression of endothelial expression of adhesion molecules and so forth), it is difficult to determine what the best approach to increasing TGF-β activity might be. Would local administration be better than systemic, or would this favour intimal expansion rather than immunosuppression? Would stimulating activation be better than stimulating production of the latent precursor, or would that bypass the mechanisms (such as increased PAI-1 production) which normally prevent TGF-β activity exceeding ‘safe’ levels? Clearly there remains considerable scope for further investigations in this area.

5 Conclusions

Many studies over the last decade have combined to elucidate the important role of TGF-β family members in the maintenance of normal blood vessel wall architecture in rodents, and demonstrated that reduction of this protective signal results in an increased propensity to form lipid lesions in response to pro-atherogenic stimuli, as well as a switch to a more vulnerable plaque phenotype. Moreover, strategies which elevate TGF-β activity (for the most part) protect against atherogenic changes in the blood vessel wall.

The corresponding situation in man, however, is less clear. Certainly, most of the evidence reviewed here is consistent with the concept that TGF-β plays broadly the same role in the vasculature in man as in rodents, but any kind of direct, conclusive evidence is completely lacking. This is unsurprising, and surely reflects the difficulty of probing complex systems in man where the kind of interventional experiments commonplace in animal models are almost without exception impossible to perform, rather than any intrinsic difference in the regulatory systems in human vasculature.

Much remains to be done, however, to fully understand many aspects of TGF-β biology in man, and it seems certain that many differences (some subtle, others more central) will emerge between the picture in rodents and that in man. Perhaps the most important of these will be the relative contribution of TGF-β regulation versus other factors (both environmental and genetic) which also regulate vessel wall structure. Addressing this issue is a big challenge, with few well accepted experimental paradigms for approaching such a problem, not least because the relative importance of different pathways will differ considerably between individuals, but the prize (in terms of improved therapeutic options) is sufficient to drive us forward in that endeavour.


D.J.G. is a British Heart Foundation Senior Research Fellow. I am grateful to the British Heart Foundation for their support over more than a decade, allowing my laboratory to contribute to the field reviewed here.


  • Time for primary review 34 days

  • Abbreviations:
    extracellular matrix
    low density lipoprotein
    smooth muscle cell
    transforming growth factor type-β


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
  32. [32]
  33. [33]
  34. [34]
  35. [35]
  36. [36]
  37. [37]
  38. [38]
  39. [39]
  40. [40]
  41. [41]
  42. [42]
  43. [43]
  44. [44]
  45. [45]
  46. [46]
  47. [47]
  48. [48]
  49. [49]
  50. [50]
  51. [51]
  52. [52]
  53. [53]
  54. [54]
  55. [55]
  56. [56]
  57. [57]
  58. [58]
  59. [59]
  60. [60]
  61. [61]
  62. [62]
  63. [63]
  64. [64]
  65. [65]
  66. [66]
  67. [67]
  68. [68]
  69. [69]
  70. [70]
  71. [71]
  72. [72]
  73. [73]
  74. [74]
  75. [75]
  76. [76]
  77. [77]
  78. [78]
  79. [79]
  80. [80]
  81. [81]
  82. [82]
  83. [83]
  84. [84]
  85. [85]
  86. [86]
  87. [87]
  88. [88]
  89. [89]
  90. [90]
  91. [91]
  92. [92]
  93. [93]
  94. [94]
  95. [95]
  96. [96]
  97. [97]
  98. [98]
  99. [99]
  100. [100]
  101. [101]
  102. [102]
View Abstract