List of participants:
Valérie Cormier-Daire, Harry C. (Hal) Dietz, Thomas Doetschman, Reinhard Faessler, Penelope (Penny) Handford, Caroline Hill, Boriz Hinz, Catherine (Cay) Kielty, Katri Koli, Marene Landström, Bart Loeys, Aristidis Moustakas, Francesco Ramirez, Dieter P. Reinhardt, Daniel B. (Dan) Rifkin (organizer), Dean Sheppard, Arnoud Sonnenberg, Timothy A. Springer
Summary: The meeting at Les Treilles on Regulation of Transforming Growth Factor Beta (TGFb) Activity in Cardiovascular Disease and Related Conditions focused on recent advances in the understanding of TGFb signaling in the development of vascular and pulmonary pathology. New results and extensive discussions focused on the relationship of TGFb and genetic diseases of connective tissue including fibrillin abnormalities, new mechanisms of TGFb signaling, matrix cell interactions, and the interactions of additional matrix proteins with elasto-fibrillar structures.
Key Words; TGFb, connective tissue, aneurysm, Smad, LTBP, fibrillin, smooth muscle cells.
On September 30 – October 5, a meeting was held on the topic of Regulation of Transforming Growth Factor Beta (TGFb) Activity in Cardiovascular Disease and Related Conditions. Recent evidence has indicated that inappropriate TGFb signaling is associated with a variety of cardiovascular manifestations observed in rare genetic syndromes. This information has sparked interest in the potential utilization of TGFb inhibitors as therapeutics. However, the mechanisms regulating active TGFb production are poorly understood for several reasons including the presence of multiple isoforms of TGFb, the complex controls governing TGFb signaling intracellularly and extracellularly, and unique latent complexes in which TGF-b is secreted. Therefore, the goal of the meeting was to clarify what is known about TGF-b regulation and to outline strategies for approaching molecular intervention.
Much of the interest in dysregulation of TGFb in the vascular system came from analysis of aneurysms in people and mice with Marfan syndrome (MFS), an autosomal dominant multi-system condition caused by mutations in the extracellular matrix (ECM) protein fibrillin-1. Dr. B. Loeys (University of Antwerp) presented studies illustrating that a number of additional genetic diseases associated with aortic aneurysms result from mutations in proteins that modulate TGFb action. These conditions include Loeys-Dietz syndrome (LDS), caused by mutations in TGFb receptor 1 or 2 genes, arterial tortuosity syndrome, caused by a deficiency of the glucose transporter GLUT10, cutis laxa type AB, caused by a deficiency in fibulin-1, as well as conditions caused by mutations in the SMAD3, TGFb2, and SK1 genes that are all involved with TGFb intracellular signal transduction. These observations suggest that normal function of TGFb is crucial for aortic wall homeostasis, in addition to illustrating that aneurysms may also result from mutations in genes other than those coding for structural proteins.
Dr. H. Dietz (Johns Hopkins Medical School) described further characterization of signaling events that contribute to aneurysm progression and wall tear in fibrillin-1 deficient mouse models of MFS. Conditional provocations that accelerate aortic disease show strong correlation with enhanced canonical (Smad) and non-canonical (ERK) TGFb signaling and the phenotype can be rescued using specific antagonists of these pathways. More recent experiments revealed crosstalk with the PLC-IP3-PKC signaling axis. To address the paradoxical enhancement of TGFb signaling seen in vasculopathies associated with primary heterozygous loss-of-function mutations in genes encoding positive regulators of signaling, the Dietz lab has identified genetic modifiers of disease in both patients and mouse models. In keeping with the hypothesis that increased TGFb signaling drives postnatal aneurysm progression, the protective modifier alleles act by blunting the TGFb response. In apparent contrast to the paradoxical enhancement of TGFb signaling, no mitigation of aortic disease was noted in LDS mice systemically treated with the pan-TGFb neutralizing antibody 1D11. The Dietz’ laboratory also reported their recent characterization of the mechanism of skin fibrosis in mouse models of stiff skin syndrome (SSS), a condition caused by mutations in the sole domain of fibrillin-1 that harbors an RGD sequence needed to mediate cell-matrix attachment via integrin binding. SSS mice show fully penetrant dense dermal fibrosis by 3 months of age. Remarkably, SSS point mutations phenocopy autoimmune and auto inflammatory events more typical of systemic sclerosis (scleroderma) including autoantibody production, Th2 and Th17 skewing, and the recruitment and activation of plasma cells. All of these abnormalities, including fibrosis and autoimmunity, are prevented by treatment of mice with a b1 integrin-activating antibody and are reversed upon treatment of older SSS mice with TGFb neutralizing antibody. Evidence suggests a critical role for plasmacytoid dendritic cells and relevance of these findings to more common but complex presentations of scleroderma.
Dr. V. Cormier-Daire (INSERM U781, Paris Descartes University) described her work analyzing the phenotypes and genotypes of patients from the acromelic dysplasia group. These patients are characterized by short stature, restricted joint mobility and heart defects. Genetic analysis of patients within this group revealed mutations in fibrillin-1 (Weill-Marchesani syndrome), SMAD4 (Myhre syndrome), ADAMTSL2 (Geleophysic dysplasia), and fibrillin-1 (Acromicric dysplasia). Cells from patients with these syndromes displayed enhanced TGFb signaling further emphasizing the relationship of TGFb and proper matrix organization.
Next, Dr. T. Doetschman (University of Arizona) reported on his studies on the role of TGFb in heart valve development. Using exquisite in situ labeling of TGFb isoform expression coupled with analysis of embryos with null mutations in the three individual TGFb genes, Doetschman teased out the requirements for TGFb2 in valvulogenesis. In the absence of TGFb2 endocardial cell transition into pre-valvular mesenchyme (EnMT) is delayed, but does not cease, leading to enlarged valves that are unable to remodel the pre-valvular mesenchyme into the highly structured elastic and collagen layers of the mature valve. Expression of ECM components of the developing valves is nearly halved, including those involved with integrin signaling and TGFb activation. It is unclear whether the TGFb2 conditionally null phenotypes result directly from the absence of TGFb2 signaling or indirectly from altered signaling of other TGFbs/BMPs that reside in the dysregulated matrix.
Although TGFb has been suggested to be the root cause of many MFS phenotypes, including both vascular and bone disorders, Dr. F. Ramirez (Mt. Sinai School of Medicine) presented data on the dilated cardiomyopathy observed in MFS mice, indicating that much of the pathology is due to mechanical activation of the AT1 receptor and not to excessive levels of active TGFb. By using a genetic approach with mice with null mutations in genes that regulate the mechanosensing activity of the AT1 receptor versus pharmacological inhibition of TGFb, Ramirez demonstrated in an elegant and persuasive way that the heart defect is caused by non-TGFb mechanisms. This heart phenotype is one that may be of increasing medical significance as MFS patients reach older ages necessitating improved treatment modalities for conditions previously not considered germane. By using losartan and TGFb neutralizing antibody 1D11, his group correlated abnormal TGFb and AT1 receptor signaling to aneurysm growth and medial degeneration, respectively, in mice with severe MFS (Fbn1 hypomorphic mice). Interestingly, administration of TGFb neutralizing antibodies to MFS mice soon after birth exacerbated aortic wall degeneration thereby leading to earlier death than untreated MFS mice. Ramirez also demonstrated that fibrillin-1 mutations in MFS mice influence the differentiation of bone precursor cells through a combination of TGFb and BMP signaling. This novel effect of fibrillin-1 mutations on bone cell differentiation demonstrates the pervasive contribution of microfibers to cell phenotypes in multiple environments and highlights the contextual specificity of ECM function.
Microfibril defects and their relation to TGFb are likely to be more complicated than what has been described, as emphasized by Dr. D. B. Rifkin (New York University Langone Medical School) in his presentation describing the contributions of the latent TGFb binding proteins (LTBP) to aortic aneurysms. LTBPs bind both TGFb and fibrillin and target latent TGFb to the microfibers. Rifkin presented data indicating that the loss of LTBP-3 is protective in Fbn1 hypomorphic mice that die because of dissecting aortic aneurysms. These mice live almost as long as WT mice, have minimal dilation of the aorta, and have virtually no enhanced TGFb signaling; all of which are observed in MFS animals. These results indicate that in the absence of a specific carrier of TGFb, the aortic aneurysmal phenotype is ablated seemingly in agreement with the earlier work from Dietz and in contradiction to certain findings of Ramirez. The resolution of this conflict will probably highlight the spatiotemporal specificity of TGFb actions or additional roles of LTBP-3.
Current information indicates that, like fibrillin, fibronectin is needed for LTBP-1 deposition, and that LTBP-1 co-localizes with microfibrils. Dr. C. Kielty (University of Manchester) presented an extensive set of biochemical and cell culture experiments examining cell-type specific differences in the mechanisms governing LTBP-1 deposition, the contributions of fibronectin and heparan sulfate to LTBP-1 matrix accumulation, interactions of LTBP-1 with fibulin-4, and the mechanisms regulating stable LTBP-1 microfiber association. The theme of heparin and fibronectin interactions with matrix proteins was also considered by D. Reinhardt (McGill University), who discussed the requirement for a fibronectin foundation for the deposition of fibrillin, LTBP-1, and other proteins. Using a fibronectin conditional null mouse model, Reinhardt observed preliminary evidence suggesting impaired smooth muscle interaction with elastic lamellae and disrupted architecture of the vessel wall in the absence of fibronectin. Reinhardt also discussed the role of heparin and its effects on matrix assembly.
The next two sessions focused on x-ray analysis of fibrillin, LTBP-1, and latent TGFb. Dr. P. Handford (Oxford University) reported on structural studies of fibrillin-1 modules from both the N- and C-terminal regions. While most fibrillin domains are rigid, the N-terminal region fragment contains two flexible sequences of unknown function. The C-terminal region has a sequence critical for secretion, as deletion of the sequence blocks release of the protein into the medium. Handford also described an inventive system in which an EGFP sequence is inserted into the flexible region of the fibrillin-1 N-terminal sequence. This tagged protein may have utility in analysis of fibrillin-1 secretion and incorporation into the ECM. Dr. T. Springer (Harvard Medical School) described the crystal structure of the small latent complex of TGFb1. In this structure, the cytokine is enveloped by its cleaved pro-peptide in a manner that prohibits receptor binding to ligand. By integrin binding to the RGD sequence in the pro-peptide, force can be applied since the opposite end of the complex contains an LTBP tethered to the matrix. This stretching liberates the mature TGFb. In some cells, especially T cells, LTBP is replaced in the latent complex by a trans membrane protein GARP that may function in a manner similar to that of LTBP. Interestingly, Springer reported that BMP-pro-peptide interactions, as illustrated by BMP-7, are quite different than that of TGF-b, as regions of the growth factor that interact with its receptor are not shielded by the pro-peptide. The biological significance of these differences in TGFb and BMP propeptide interactions is not yet clear.
The question of matrix rigidity and the activation of latent TGFb in conditions such as fibrosis was considered by Dr. B. Hinz (University of Toronto). Hinz’ previous research on latent TGF-b1 activation demonstrated a mechanical pulling mechanism, requiring integrins, cell contraction, and binding of the LTBP to a stiff ECM. His recent results show that myofibroblasts can organize fibrilar structures from LTBP-1 that is either 1) endogenously produced, 2) offered as non-organized ECM produced by different cell-expression systems, or 3) added in a soluble form to the culture medium of myofibroblasts. The percentage of contraction-activated TGF-b1 out of total TGF-b1 correlated with the degree of organization of LTBP-1-ECM in all conditions. Moreover, pre-straining cell free LTBP-1-ECM using a mechanical device enhanced activation of TGF-b1 by myofibroblast contraction compared with relaxed LTBP-1-ECM. Thus, myofibroblasts remodel LTBP-1 into strained fibrilar structures. The organization level/pre-strain of the TGF-b1 complex will determine the amount of active TGF-b1 released from the ECM by myofibroblast contraction. Dr. A. Moustakas (Ludwig Institute, Uppsala University) discussed his work on the control of epithelial-mesenchymal transition (EMT) mediated by TGFb. Moustakas described how the high mobility group A (HMGA2) protein, whose expression is regulated by TGFb3, is a critical factor and induces the upregulation of EMT master genes. Moustakas also linked TGFb-HMGA2 to breast cancer stem cell biology, as cancer stem cells express high levels of HMGA2. In addition, HMGA2 negatively regulates dicer, which effects cell renewal.
Dr. D. Sheppard (UCSF) described recent studies on the role of integrins in the activation of latent TGFb in fibrosis. The Sheppard group previously described a requirement for the integrin avb6 in lung fibrosis. However, fibrosis in non-epithelial tissues, such as liver, is not dependent upon avb6. Sheppard presented evidence that with fibroblasts the integrin avb1 activates latent TGFb1. By the use of a specific inhibitor of avb1, Sheppard blocked liver fibrosis. The discussion of integrins continued with Dr. R. Faessler (Max Planck Institute, Munich) describing how the two classes of fibronectin-binding integrins (b1 and av-class integrins) mediate specific functions. By using genetics, cell biology, and proteomics, the Faessler group demonstrated that force generation is accomplished by b1-class integrins, whereas structural adaptations to forces, as visualized by the formation of stress fibers, are mediated by av-class integrins, which in cooperation enables cells to sense the rigidity of fibronectin-bound microenvironments. Finally, Dr. A. Sonnenberg (Netherlands Cancer Institute) described the contributions of integrin a3 and tetraspanin CD151 to skin cancer production in the mouse. Sonnenberg found that a3b1 is essential for the production of chemically induced skin cancers in mice and loss of the tetraspanin CD151, which associates with the integrin, results in decreased tumorigenesis. An analysis of the phenotypes indicated that CD151 supports tumorigenesis through both a3b1 dependent and independent mechanisms. The cellular response, i.e. decreases in tumor number, to loss of a3b1 or CD151 is observed as the presence of keratin 15-positive cells outside their normal position in the hair follicle bulge and a more rapid turnover of cells. Sonnenberg suggested that rapid cell turnover leads to a loss of tumor-initiated cells before they capable of surviving independently.
The next two presentations addressed issues of TGF-b signaling. Dr. C. Hill (Cancer Research UK) addressed the process of Smad signal transduction to the nucleus upon TGF-b stimulation, as well as the known unresponsive state of cells to additional TGF-b after signaling is terminated. Using an integrative experimental and mathematical modeling approach, Hill reported the unexpected result that the model predicts that TGFb exposure desensitizes cells making them refractory to further acute stimulation. This can be explained by receptor dynamics, including the rapid depletion of TGFb receptors upon signaling and their slow replacement at the cell surface. These results may explain some of the unusual signaling responses observed in cells from LDS patients. A major question in the TGFb field is what controls signaling through the canonical Smad pathway versus the non-canonical TAK1 pathway. Dr. M. Landstrom (Umea University) described elegant experiments indicating that TGFb receptor 1 (TbR1) is modified by the ubiquitin ligase TNF receptor associated factor 6 (TRAF6). Ubiquitination results in cleavage of TbR1 by TACE and presenilin-1. The cleaved receptor cytoplasmic domain moves to the nucleus and appears to activate the Smad and MMP2 genes. This is a novel example of how TGFb may promote transcription of pro-invasive tumor genes through a non-Smad pathway.
The final speaker of the meeting was Dr. K. Koli (Helsinki University), who described recent experiments with the protein gremlin, a known BMP inhibitor. Gremlin is abundant in lung tissue from both idiopathic pulmonary fibrosis and malignant mesothelioma. Over-expression experiments with gremlin indicate that it contributes to EMT in addition to modulating cell proliferation/differentiation. These data suggest that therapeutic intervention targeting gremlin may be useful in these lung pathologies.
The meeting ended with a discussion of the current critical questions that must be addressed to understand the role of TGF-b in cardiovascular, specifically aneurysmal conditions. Briefly, questions such as what integrins are involved in the cellular responses, how is tissue specificity of matrix assembly controlled, what controls the choice of canonical versus non-canonical TGFb signaling pathways, what is the mechanism of latent TGFb activation, especially latent TGFb2, and how is TGFb auto-induction controlled were considered.