|Home | About | Journals | Submit | Contact Us | Français|
Carcinoid heart disease was one of the first valvular pathologies studied in molecular detail, and early research identified serotonin produced by oncogenic enterochromaffin cells as the likely culprit in causing changes in heart valve tissue. Researchers and physicians in the mid-1960s noted a connection between the use of several ergot-derived medications with structures similar to serotonin and the development of heart valve pathologies similar to those observed in carcinoid patients. The exact serotonergic target that mediated valvular pathogenesis remained a mystery for many years until similar cases were reported in patients using the popular diet drug Fen-Phen in the late 1990s. The Fen-Phen episode sparked renewed interest in serotonin-mediated valve disease, and studies led to the identification of the 5-HT2B receptor as the likely molecular target leading to heart valve tissue fibrosis. Subsequent studies have identified numerous other activators of the 5-HT2B receptor, and consequently, the use of many of these molecules has been linked to heart valve disease. Herein, we: review the molecular properties of the 5-HT2B receptor including factors that differentiate the 5-HT2B receptor from other 5-HT receptor subtypes, discuss the studies that led to the identification of the 5-HT2B receptor as the mediator of heart valve disease, present current efforts to identify potential valvulopathogens by screening for 5-HT2B receptor activity, and speculate on potential therapeutic benefits of 5-HT2B receptor targeting.
Serotonin or 5-hydroxytryptamine (5-HT; Fig. 1) is enzymatically transformed from the essential amino acid tryptophan following hydroxylation and decarboxylation. Serotonin was discovered and isolated from serum 60 years ago (Rapport, Green et al. 1948), and shortly after, the molecule was determined to originate from the enterochromaffin (or Kulchitsky) cells that are found throughout the gastrointestinal and bronchopulmonary system (Erspamer and Asero 1952). High concentrations of 5-HT are found in blood platelets and enterochromaffin cells of the gut; lesser amounts are found around neurons located along the raphé nuclei of the brainstem. The human brain has evolved a sophisticated arrangement of axons stemming from the raphé nuclei to innervate nearly every brain region.
5-HT is involved in a diverse array of physiologic and biologic processes. In the brain, 5-HT has been found to affect sleep, mood, appetite, anxiety, aggression, perception, pain, and cognition (Roth, Willins et al. 1998; Edited by Roth 2006; Berger, Gray et al. 2009). Systemically, 5-HT contributes to vascular and non-vascular smooth muscle contraction and platelet aggregation. Serotonin released from neurons is recaptured by an active reuptake pump (serotonin transporter), and is then inactivated by monoamine oxidase and converted to 5-hydroxyindoleacetic acid (Guyton and Hall 1996). In vivo, the rate of urinary 5-hydroxyindoleacetic acid output is commonly used as an index of 5-HT metabolism.
Signaling of 5-HT is mediated by receptors that are located on the cell membrane of neurons and most other cells in the body (Edited by Roth 2006; Berger, Gray et al. 2009). There are six classes of G protein-coupled 5-HT receptors (5-HT1,2,4,5,6,7) that can be subdivided into 14 unique subtypes. The 5-HT3 receptor is unique among 5-HT receptors in that it is a ligand gated ion channel (Maricq, Peterson et al. 1991; Edited by Roth 2006). Heterotrimeric guanine nucleotide-binding protein G protein-coupled receptors (GPCRs) are well characterized and have been described extensively (Kroeze, Kristiansen et al. 2002; Kroeze, Sheffler et al. 2003). Briefly, GPCRs are transmembrane proteins consisting of seven membrane-spanning α-helical segments with an extracellular N-terminus and an intracellular C-terminus. The binding of 5-HT to one of its receptors is thought to elicit a conformational change that activates associated heterotrimeric G proteins and recruits other downstream signaling/scaffolding molecules, such as GPCR kinases and β-arrestins (Armbruster and Roth 2005; Allen and Roth 2010). Upon activation by an agonist-occupied GPCR, G proteins release guanosine diphosphate, which is constituitively bound to the α subunit of the heterotrimer, and bind guanosine triphosphate (GTP). Binding of GTP to the α subunit causes it to dissociate from the βγ subunits (which remain associated to each other); free Gα then interacts with nearby, downstream effectors (e.g., adenylate cyclase for Gαs/olf- and Gαi/o/z-types or phospholipase C for Gαq/11-types), generating second messengers (e.g., cAMP produced by adenylate cyclase or inositol 1,4,5-trisphosphate and diacylglycerol produced by phospholipase C) that modulate downstream effectors inside the cell (e.g., protein kinases A and C activated by cAMP and diacylglycerol) [see (Raymond, Mukhin et al. 2001) for review of 5-HT signaling pathways]
Because of the systemic presence of 5-HT and the multitude of receptor types found throughout the body that can elicit a myriad of cellular responses, drugs targeting 5-HT receptors are effective treatments for a variety of conditions. Each 5-HT receptor subtype contains at least one important therapeutic target. For instance, antimigraine medications (e.g., ergotamine, sumatriptan) activate 5-HT1B/D receptors. Clinically effective antipsychotics block the activation of 5-HT2A, 5-HT2C (Roth et al, 1992), 5-HT6 (Roth, Craigo et al. 1994), and 5-HT7 (Roth, Craigo et al. 1994; Abbas, Hedlund et al. 2009) receptors [see (Roth and Xia 2004) for review]. Antagonism of 5-HT6 receptors has been postulated to enhance memory and learning in healthy individuals (Lindner, Hodges et al. 2003). There are many pharmaceuticals used to target the multitude of serotonergic GPCRs; however the 5-HT2 receptors are among the most frequently targeted, highlighting their important role in physiological and pathophysiological processes see [(Roth 2010) for recent review].
The 5-HT2 family consists of three GPCRs: 5-HT2A, 5-HT2B, and 5-HT2C. 5-HT induces an increase in inositol 1,4,5-trisphosphate (which leads to release of intracellular calcium stores and diacylglycerol production (Conn and Sanders-Bush 1984; Roth, Nakaki et al. 1984; Kursar, Nelson et al. 1994). In addition to these known signaling mechanisms, 5-HT2 receptors also generate second messenger signaling that leads to cell-type specific responses depending on the organ under consideration. Some of the most notable effects of 5-HT2 receptor-preferring drugs involve the brain, and these activities are exploited therapeutically. Two important, common examples are atypical antipsychotics and anorexigens. These drugs—and/or their metabolites—display activity at 5-HT2A (viz: atypical antipsychotics are inverse agonists) (Meltzer, Matsubara et al. 1989; Weiner, Burstein et al. 2001) and 5-HT2C receptors (anorexigens and putative atypical antipsychotic drugs are agonists) (Kozikowski, Cho et al. 2010). 5-HT2A and 5-HT2C receptors are highly abundant in various human brain regions with 5-HT2A being highly concentrated in cortical regions and 5-HT2C more broadly distributed (Pazos, Cortes et al. 1985; Abramowski, Rigo et al. 1995). The 5-HT2B receptor subtype displays a lower expression in the brain (Kursar, Nelson et al. 1994), and thus, it plays a lesser role in the effects of psychoactive agents. Nevertheless, recent genetic and pharmacologic studies have implicated 5-HT2B receptors in the biological activities of the recreational psychostimulant 3,4-methylenedioxymethamphetamine (MDMA, ecstasy) (Doly, Valjent et al. 2008; Doly, Bertran-Gonzalez et al. 2009) and the anorexigen fenfluramine (Banas, Doly et al. 2010).
The putative cardiovascular action of the 5-HT2A and 5-HT2B receptors may be similar. Both of these receptors have been shown to elicit mitogenic and secretory responses in ventricular and heart valve fibroblasts (Xu, Jian et al. 2002; Setola, Hufeisen et al. 2003; Jaffre, Callebert et al. 2004; Yabanoglu, Akkiki et al. 2009), indicating a possible role for each in cardiac development and disease. An issue in isolating functional differences between the 5-HT2 receptors has been the lack of specificity in pharmacological agents used to target the receptors. In fact, many clinically used agents, particularly antipsychotics and anorexigens, display some activity at all three 5-HT2 receptor subtypes. One reason for low selectivity among 5-HT2 receptor-active compounds is the high degree of amino acid sequence homology among the three subtypes (Roth, Willins et al. 1998; Barnes and Sharp 1999) with a structural similarity of 45–50% between the receptors (Pytliak, Vargova et al. 2010). This homology is of significant consequence as drugs intended for the 5-HT2A or 5-HT2C receptors located in the brain may also bind to the 5-HT2B receptors expressed in the brain or in other tissues. Therefore, special attention should be given to differential properties of these receptor subtypes to identify functional differences and enhance understanding of target specificity.
In 2005, we examined non-conserved residues in the transmembrane helices of the 5-HT2 receptors to identify ones that might participate in the preferential binding of (+)-norfenfluramine to 5-HT2B receptors. We generated a series of 5-HT2B receptor point mutants that contained one 5-HT2A- and/or 5-HT2C-like putative ligand binding residue, and we determined whether any of the point mutations affected (+)-norfenfluramine binding affinity. Mutation of a valine in TM2, V2.53, to leucine (the analogous residue in the 5-HT2A receptor, caused a 17-fold decrease in the Ki of (+)-norfenfluramine. Residue 2.53 in the 5-HT2C receptor is also a valine, and the V2.53L point mutation caused a 12-fold decrease in the Ki of (+)-norfenfluramine. The reciprocal point mutation (L2.53V) in the 5-HT2A receptor had no effect on the Ki of (+)-norfenfluramine. The preceding observations suggest that V2.53 in the 5-HT2B receptor contributes to the high-affinity binding of (+)-norfenfluramine.
Using a rhodopsin-based 5-HT2B receptor homology model, we performed in silico ligand docking, and molecular dynamics simulations to predict how V2.53 might contribute to (+)-norfenfluramine binding. One result suggested that both terminal methyl groups of V2.53 formed stabilizing van der Waals (vdW) interactions with the α-methyl group of (+)-norfenfluramine, and that the V2.53L mutation resulted in the loss of one of these interactions. To test that prediction, we generated additional point mutants and norfenfluramine analogs. First, we reasoned that a V2.53A mutation would eliminate both vdW interactions, further decreasing (+)-norfenfluramine affinity. In fact, the mutation caused a 150-fold reduction in the Ki of (+)-norfenfluramine. Second, we synthesized a norfenfluramine analog lacking an α-methyl group. The affinity of α-desmethyl-norfenfluramine for the wild type 5-HT2B receptor was reduced three-fold compared with (+)-norfenfluramine. Further, α-desmethyl-norfenfluramine binding was less sensitive to the V2.53L mutation than was (+)-norfenfluramine. Our molecular dynamics simulations also predicted that a V2.53I mutation would permit two vdW interactions between the terminal γ- and δ-methyl groups of I2.53 and the α-methyl group of (+)-norfenfluramine. Upon experimental validation, the Ki of (+)-norfenfluramine binding to the V2.53I 5-HT2B receptor was 35 nM, compared with 22 at the wild type 5-HT2B receptor. Together, our in vitro and in silico studies of the 5-HT2B receptor (+)-norfenfluramine binding provide evidence linking V2.53 to the high-affinity and subtype-selective binding of the valvulopathogenic anorexigen.
What is now called the 5-HT2B receptor (Fig. 2) was first recognized 50 years ago relating to the putative role of a specific 5-HT receptor subtype in the contraction of the gastric fundus from rat stomach (Vane 1959). Although there was controversy prior to the cloning of the 5-HT2B receptor whether the stomach fundus receptor was pharmacologically distinct from the 5-HT2A and 5-HT2C receptors (Kaumann and Levy 2006), this disappeared once all three were cloned and their tissue distribution illuminated. Since then, 5-HT2B receptors have been found to be present in both rodent and human tissues, particularly in the cardiovascular system, gastrointestinal tract, bone, and central nervous system (Ullmer, Schmuck et al. 1995; Choi and Maroteaux 1996; Fitzgerald, Burn et al. 2000). Importantly, the tissue distribution of 5-HT2B receptor protein in rodents and humans is similar, as are their pharmacologies (Manivet, Schneider et al. 2002); this observation facilitates the extrapolation of physiological and pharmacological results from rodent studies of the 5-HT2B receptor to humans. Accordingly, Sprague-Dawley rats have been used in several studies to gain insight into the mechanism of 5-HT-induced valvular alterations (Elangbam, Lightfoot et al. 2005; Gustafsson, Tommeras et al. 2005; Elangbam, Wehe et al. 2006; Hauso, Gustafsson et al. 2007; Elangbam, Job et al. 2008). Long-term serotonin administration in these rats has been shown to induce aortic valve insufficiency and histopathological changes similar to those observed in human carcinoid patients (discussed in Section 3.1) (Gustafsson, Tommeras et al. 2005). The valvular changes seem to be associated with an increase in 5-HT2B receptor expression and a decrease in the expression of 5-HT transporter in both mitral and aortic valves (Elangbam, Job et al. 2008). These observations may indicate an indirect interaction between these two membrane proteins; whereby, in normal physiology, the 5-HT transporter may control homeostatic 5-HT levels, preventing 5-HT2B receptor over-stimulation (Elangbam 2010). In the disease state, the cooperation between these receptors may be lost, leading to persistent 5-HT2B receptor stimulation and the resultant valvulopathies discussed in Section 3.
The complexity and variety of 5-HT2B receptor expression is paralleled by its signal transduction. Activation of 5-HT2B has been found to stimulate phospholipase C (Kellermann, Loric et al. 1996) and phospholipase A2 (Tournois, Mutel et al. 1998), both of which increase intracellular calcium levels. Activation of 5-HT2B receptors in some cell types has also been shown to stimulate nitric oxide synthase (Manivet, Mouillet-Richard et al. 2000). In fibroblasts and smooth muscle cells, the biological result of activating 5-HT2B receptors is mitosis (Fitzgerald, Burn et al. 2000; Nebigil, Choi et al. 2000; Nebigil, Launay et al. 2000; Setola, Hufeisen et al. 2003) and secretion of inflammatory cytokines and extracellular matrix (ECM) components (Jaffre, Bonnin et al. 2009). As such, 5-HT2B receptors appear to play a crucial role in allowing these cells to maintain the structural homeostasis of the tissues comprising them (e.g., myocardium, heart valves, blood vessels). For example, over-expression of 5-HT2B receptors in hearts of transgenic mice results in cardiac hypertrophy and decreased ventricular function due to enhanced ECM and remodeling (Nebigil, Jaffre et al. 2003). Likewise, genetic deletion of 5-HT2B receptors was shown to lead to ventricular dilation and incomplete cardiac development (Nebigil, Choi et al. 2000).
The molecular signaling pathways associated with these matrix remodeling responses appears to involve typical mitogenic signal transduction as well as coordination between the signaling of 5-HT2B receptors and various growth factor and cytokine receptors (Nebigil, Choi et al. 2000; Nebigil, Launay et al. 2000; Jaffre, Callebert et al. 2004). Specifically, 5-HT2B activation in mouse fibroblasts was initially shown to lead to the p21Ras- and heterotrimeric G protein-dependent activation of mitogen activated protein kinase (MAPK) (Launay, Birraux et al. 1996). Further studies in fibroblasts revealed that 5-HT2B activation also leads to phosphorylation of the cytoplasmic tyrosine kinase Src (Nebigil, Launay et al. 2000). Src phosphorylation appears to enhance signaling of the platelet derived growth factor and epidermal growth factor (Nebigil, Launay et al. 2000; Li, Zhang et al. 2008); Src inhibitors have been found to negatively affect signaling at both these receptors (Hsu, Persons et al. 1991; Nebigil, Launay et al. 2000). A more recent study connected these 5-HT2B signaling pathways using pharmacological agents to selectively inhibit key signal transduction proteins in cardiac fibroblasts (Jaffre, Bonnin et al. 2009). Accordingly, 5-HT2B receptor-mediated Src activation was found to lead to an increase in matrix metalloproteinase activity, which leads to a release of heparin-bound epidermal growth factor. In turn, epidermal growth factor signaling leads to MAPK activation and subsequent upregulation of various cytokines. In addition, Jaffré et al. (Jaffre, Bonnin et al. 2009) demonstrated that 5-HT2B receptors work in concert with the angiotensin II type 1 receptor (AT1R) to mediate hypertrophic signaling in cardiac fibroblasts. These results indicate that, in cardiac fibroblasts at least, the function of both GPCRs is required for the activity of either one. Consequently, by inhibiting one receptor the action of the other receptor is likewise inhibited.
Many of the cardiac-related studies into the molecular signaling pathways of the 5-HT2B receptors have focused on the role of this receptor in ventricular fibroblasts. In this review, we will highlight the role 5-HT2B receptors may play in mediating changes in heart valves. Therefore, to make the connection between the action of 5-HT2B receptors in ventricles and heart valves, it is interesting to note that the signaling pathways discussed previously seem to play a major role in both regulating and responding to the biomechanical properties of ventricular tissue, and as noted below, biomechanical integrity is crucial to the appropriate function of heart valves. Thus, both AT1 and 5-HT2B receptors demonstrate a mechano-dependent upregulation and signaling activation during ventricular pressure overload that results in an increase in tissue stress levels (Rosenkranz 2004; Liang, Lai et al. 2006). Moreover, agonist signaling at these GPCRs has been shown to lead to an increase in expression of the cytokine transforming growth factor-β1 (TGF-β1) (Jian, Xu et al. 2002; Rosenkranz 2004), which has also been shown to exhibit mechano-responsive signaling (Wipff, Rifkin et al. 2007) and seems to be a key mediator in the tissue changes that lead to heart valve disease (Walker, Masters et al. 2004; Merryman, Lukoff et al. 2007).
Heart valves (HVs) control the direction of blood flow through the heart during the cardiac cycle by directing oxygen-poor blood to the lungs and oxygen-rich blood to the vital organs via the systemic circulation. There are four valves in the heart: two on pulmonic side and two on the systemic side (Fig. 3). Progressing through the heart from the vena cava, deoxygenated blood enters the right atrium and passes through the tricuspid valve to the right ventricle, and then from the right ventricle through the pulmonary valve into the pulmonary artery, which directs blood flow to the lungs. Upon reoxygenation by perfusion through the lungs, the blood returns to the left atrium of the heart via the pulmonary vein, passes through the mitral valve into the left ventricle, and then enters the aorta through the aortic valve. The anatomies of the pulmonary and aortic valves are very similar, as are the anatomies of the tricuspid and mitral valves. The pulmonary and aortic valves are referred to as semilunar valves because the three leaflets of the valves look like a half-moon when excised (Fig. 4a). Semilunar valves open and close solely due to inertial forces from the blood during the cardiac cycle; whereas, the structure and function of the tricuspid and mitral valves, referred to as atrioventricular valves, are more complex than the semilunar valves (Fig. 4b). Briefly, their leaflets are tethered to the papillary muscles of ventricular wall via chordae tendineae, which act as anchors to prevent the leaflets from prolapsing into the atria during ventricular contraction. The primary difference between the tricuspid and mitral valve is the number of leaflets: the tricuspid has three and the mitral only two. The atrioventricular valves ensure that blood does not back-flow during myocardial contraction (systole), and the semilunar valves keep blood from re-entering the heart during relaxation (diastole).
All of the valves rely on hemodynamic forces to regulate the opening and closing of the HV leaflets. As such, the biomechanical integrity of the leaflet tissues is crucial to appropriate function of the HVs. Two distinct cell types have been shown to play a role in regulating tissue homeostasis within the HV: valve endothelial cells and valve interstitial cells (VICs) (Chester and Taylor 2007). Valve endothelial cells exhibit regional heterogeneity along the surface of the HV leaflets and may function to direct spatial-specific extracellular matrix (ECM) production through paracrine signaling to the VICs (Simmons, Grant et al. 2005). During HV development, activated VICs synthesize ECM components to form the leaflet structure that is crucial to correct HV function; however, during disease, over-activation of VICs can lead to inappropriate ECM accumulation and/or valvular architecture and subsequent loss of HV function that leads to heart valve disease (HVD). Under healthy physiologic conditions, the left side VICs are much more active and dynamic with greater levels of cytoskeletal proteins and ECM biosynthesis, likely due to the increased pressures imposed on the leaflets (Merryman, Huang et al. 2006; Merryman, Youn et al. 2006; Merryman, Liao et al. 2007).
There exist two types of HVD: congenital and acquired (Otto 2004). Congenital valve disease is an abnormality that develops before birth and may be related to improper valve size, malformed leaflets, or an irregularity in the way the leaflets are attached. This most often affects the pulmonary and aortic valves. A very common congenital defect is bicuspid aortic valve disease where instead of the normal three leaflets or cusps, the bicuspid aortic valve has only two. Without the third leaflet, the valve may be stenotic (narrowing, obstructing flow) or regurgitant (‘floppy,’ allowing backward flow). It is estimated that some form of bicuspid aortic valve disease affects about 1.4% of the population (Lloyd-Jones, Adams et al. 2010), but this is difficult to verify since in some cases it goes undetected and does not result in significant changes in cardiovascular function. Acquired HVD pertains to problems that develop within valves that were at one time normal. This is typically referred to as age-related degenerative valve disease. In the early part of the 20th century, the primary cause of acquired valve disease stemmed from rheumatic valve disease; however, this trend has changed dramatically with a decrease in rheumatic valve disease and a concomitant increase in age-related degenerative valve disease (Otto 2004).
The risk factors for HVD are similar to traditional clinical risk factors for atherosclerosis and coronary artery disease and include age, male gender, hypertension, diabetes, triglycerides, and smoking (Mohler, Sheridan et al. 1991; Stewart, Siscovick et al. 1997; Agmon, Khandheria et al. 2001). In fact, the similarities between the two have led to the hypothesis that acquired HVD is primarily a manifestation of atherosclerosis (O’Brien 2006). The hallmark of HVD pathogenesis is the formation of lesions containing cell types that are characteristic of chronic inflammation (Olsson, Dalsgaard et al. 1994; Olsson, Rosenqvist et al. 1994; Otto, Kuusisto et al. 1994). These include macrophages, T lymphocytes, and mast cells. Additionally, there are lipoproteins [LDL and Lp(a)] found in human diseased valve lesions, and accumulation of these lipoproteins is mediated, in part, by ECM proteoglycans (Walton, Williamson et al. 1970; O’Brien, Reichenbach et al. 1996). Besides lipoproteins, recent evidence has shown that the renin-angiotensin system, particularly angiotensin converting enzyme and angiotensin II, may play a role in HVD pathogenesis (O’Brien, Shavelle et al. 2002). Angiotensin II has a number of potential lesion-forming effects, including inflammation and macrophage and cholesterol accumulation. Therefore, there are multiple mechanisms that initiate HVD and further perpetuate the disease in otherwise normally functioning valve.
As noted previously, degenerative HVD is believed to be caused by an over-activation of VICs through a process that depends on signaling associated with TGF-β1. This cytokine is secreted from cells in a latent form and stored in the ECM. Various ECM cues can result in the activation of TGF-β1, which then binds to and activates the TGF-β1 serine/threonine kinase receptors (TβRs). Following cytokine binding, TβRs modulate a wide variety of cellular processes by signaling through a family of transcription factors known as Smads (Howe 2003). In VICs, active TGF-β1 leads to myofibroblast activation in vitro (Walker, Masters et al. 2004), and the combination of active TGF-β1 and mechanical strain leads to a synergistic increase in hallmarks of HVD within porcine HV leaflets (Merryman, Lukoff et al. 2007). Confounding the problems associated with its mediation of cellular changes that lead to HVD, TGF-β1 signaling also leads to increased synthesis of the latent form of TGF-β1, which in turn may be activated by VICs (Merryman 2008). This feed-forward signaling mechanism may drive the progression of HVD.
Currently, the only viable long-term treatment option for HVD is valve replacement surgery. Typically, the native diseased HV is either replaced by a decellularized HV of bovine or porcine origin or supplanted by a mechanical HV. While the techniques and resultant risks associated with HV replacement procedures have dramatically improved, surgery still remains a poor option for the two patient populations most affected by HVD: children and elderly.
In pediatric patients, the main concern is the integration of replacement HVs within tissues that are rapidly growing and changing. Compounding the problem in children, replacement HVs tend to wear quickly and typically need to be replaced within 20 years (Merryman, Engelmayr Jr et al. 2006). To circumvent these problems, many surgeons elect to perform the Ross procedure whereby the pulmonic HV is resected and placed in the aortic position. The replacement HV is then placed in the position formerly occupied by the pulmonic HV. The advantage of this procedure is that the pulmonic HV is able to incorporate within native tissue in the aortic position, and the replacement HV wears more slowly due to the lower pressures associated with the pulmonic circulation. The major disadvantage of the Ross procedure is that it requires two very invasive surgical procedures (i.e., resection of both the aortic and pulmonic HVs), and often, the pulmonic HV is not able to adapt to the increased pressure associated with systemic circulation. For these reasons, the Ross procedure is rarely performed in elderly patients. Faced with the problems associated with HV replacement, physicians often delay surgical options until they are absolutely necessary; therefore, many patients would greatly benefit from an early treatment to delay or prevent the onset of HVD. Elderly patients face a difficult recovery following open-chest surgery. While these procedures may improve the cardiac function of these patients, their quality of life declines for a prolonged period of time (up to a year). Thus, elderly patients would benefit greatly from a pharmacological strategy against HVD.
The hopeful story in recent years has been that lipid-lowering drugs (i.e., statins) might prevent HVD (Rajamannan, Subramaniam et al. 2002; Rajamannan, Edwards et al. 2003; Benton, Kern et al. 2009), similarly to their promise against atherosclerosis. In 2001, two retrospective studies indicated that statins may inhibit the progression of aortic valve stenosis (Aronow, Ahn et al. 2001; Novaro, Tiong et al. 2001), and it was suggested that large prospective studies be conducted. When these prospective studies were finally completed, the efficacy of statins proved disappointing. In 2005, a small double-blind, placebo controlled study showed no benefit of statins to reduce HVD (Cowell, Newby et al. 2005). Moreover, a recent report of a large clinical study demonstrated, rather conclusively, that statins do not reduce major cardiovascular outcomes, including aortic valve replacement, in patients with aortic valve stenosis (Rossebo, Pedersen et al. 2008). In light of these findings, the enthusiasm for statin therapy as a potential preventive treatment for HVD has been severely dampened. Thus, there is no pharmacological strategy currently available or being developed (to our knowledge) that has the potential to prevent or delay HVD progression.
Carcinoid syndrome occurs as a result of the formation of neuroendocrine tumors arising from oncogenic enterochromaffin cells. These cells synthesize and deliver 5-HT to the bloodstream for controlled systemic distribution; however, when tumor-forming enterochromaffin cells metastasize to the liver, 5-HT more readily enters systemic circulation resulting in an increase in plasma 5-HT levels and leading to the cardiac changes associated with carcinoid heart disease (CHD) (Druce, Rockall et al. 2009). CHD was first noted 80 years ago by the Dutch pathologist A.J. Scholte, who observed that a patient with a carcinoid tumor also had thickened tricuspid valve leaflets (Gustafsson, Hauso et al. 2008). Interestingly, CHD differs from most forms of HVD in that it mostly affects right-side HVs (i.e., the pre-lung tricuspid and pulmonary valves). Elevated blood levels of 5-HT are carried to the right side of the heart through the inferior vena cava, where it is believed that interaction with 5-HT2B receptors on cells in the tricuspid and pulmonic HVs leads to the extracellular matrix secretion and the thickening of the HV leaflets that characterize HVD. As the blood continues into pulmonary circulation, the 5-HT is inactivated by monoaminoxigenase in the lungs and, therefore, does not induce HV changes when the blood returns to the left side of the heart. For many years, the serotonergic receptor subtype responsible for CHD was not known due to the fact that serotonin has a similar affinity for many of its receptors, and several of the subtypes are expressed throughout the cardiovascular system. The current hypotheses identifying 5-HT2B receptor as the major HV target came about only as a result of inadvertent targeting of the 5-HT2B receptors with Fen-Phen.
Among all the serotonergic drugs on the market, there may be none more recognizable to the general public than fenfluramine, one of the components of the ‘Fen-Phen’ anorexigen combination. This popular diet drug regimen consisting of fenfluramine and phenteramine was shown to be better tolerated than either alone in 1984 and the drug combination was subsequently and widely prescribed (Weintraub, Hasday et al. 1984). However, a study published in late 1997 identified both right-sided and left-sided HV defects in a number of patients who had been taking Fen-Phen for an average of 12 months (Connolly, Crary et al. 1997). Soon thereafter, the drugs (including other fenfluramine formulations) were voluntarily withdrawn from the market based upon recommendations by the FDA. In a retrospective study, the highest incidence of HVD development from Fen-Phen was found to be 25.2% among patients treated for an average of 20 months (Khan, Herzog et al. 1998).
Early histological analyses of diseased tissue from Fen-Phen patients indicated a pathophysiology similar to that observed in carcinoid patients and from tissues of subjects who had taken specific ergot-derived formulations. In this regard, in the mid-1960s, researchers and physicians noticed a strong connection between the ergot agents methysergide and ergotamine and the development of fibrotic pathologies including HVD (Graham, Suby et al. 1966; Graham 1967; Mason, Billingham et al. 1977). Some even went as far as to note that the “similarities in chemical structure of serotonin, methysergide, and ergotamine” may “suggest a common pathophysiologic mechanism for ergot alkaloid-associated valve disease and carcinoid valve disease” (Redfield, Nicholson et al. 1992). Much like with CHD, however, the exact molecular cause of the tissue changes that lead to HVD remained undetermined.
The Fen-Phen episodes sparked new interest in identifying the specific 5-HT receptor subtype involved in drug-induced HVD. Due to the known mitogenic roles of the 5-HT2 subfamily, Fitzgerald et al. (Fitzgerald, Burn et al. 2000) examined the interaction of fenfluramine, norfenfluramine (the main metabolite of fenfluramine), ergotamine, and methysergide on human 5-HT2A, 5-HT2B, and 5-HT2C receptors. The ergot-derived compounds were found to possess high affinity for all three receptor subtypes. In contrast, fenfluramine rotamers demonstrated weak affinities for the 5-HT2 receptors; however, the norfenfluramine rotamers exhibited relatively high affinity for both 5-HT2B and 5-HT2C receptors (Table 1) and slightly lower affinity for 5-HT2A receptors. These results suggested that norfenfluramine, the main metabolite of fenfluramine, was chiefly responsible for 5-HT2 receptor activation, and thus, the most likely candidate for causing Fen-Phen-mediated HVD. Furthermore, norfenfluramine was found to be two orders of magnitude more potent at 5-HT2B and 5-HT2C receptors compared with 5-HT2A receptors (Table 2). Additionally, 5-HT2C receptors, which are restricted to the central nervous system, were found to be expressed at exceedingly low levels within HV tissues (Fig. 5); whereas 5-HT2B receptors are expressed in relatively high quantities within the HV leaflets. Simultaneously, Rothman et al. (2000) compared the in vitro pharmacology of norfenfluramine with that of known inducers of HVD (5-HT, methysergide), as well as with that of serotonergic drugs not associated with HVD (negative controls), and found that the HVD-associated compounds were, unlike the negative controls, potent 5-HT2B receptor agonists (see below). Taken together with the expression of 5-HT2B receptors in heart valve tissue (Fitzgerald, Burn et al. 2000) and the mitogenic effect of 5-HT2B receptors in cardiac and other cell types (Nebigil, Choi et al. 2000; Nebigil, Launay et al. 2000), these results pointed to the 5-HT2B receptor as the serotonergic target leading to HVD.
As noted above, the results of the Fitzgerald et al. study were supported in a seminal report published simultaneously by Rothman et al. (Rothman, Baumann et al. 2000) in which the affinities and potencies of 15 different molecules were tested at 11 different 5-HT receptor subtypes: 5-HT1A, 5-HT1B, 5-HT1D, 5-HT1E, 5-HT2A, 5-HT2B, 5-HT2C, 5-HT5A, 5-HT6, and 5-HT7. In this study, all but the 5-HT2B receptor subtypes were systematically ruled out based on pharmacological similarity with HVD-associated molecules and pharmacological differences from negative control molecules that were not suspected mediators of HVD (e.g., fluoxetine and its metabolite norfluoxetine). The 5-HT2 receptor subtypes were the only receptors to display relatively high affinity (Table 1) and high potency (Table 2) for both the norfenfluramine rotamers and for the HVD-associated ergot-derived compounds (Tables 1 and and2,2, italicized). Similar to the Fitzgerald et al. study, the data indicate both high affinity (Ki) and potency (shown by phosphatidylinositol hydrolysis, Kact) at 5-HT2B over 5-HT2A for norfenfluramines and the known inducers of HVD. Taken together, these studies indicate that norfenfluramine, the main fenfluramine metabolite, is the likely culprit in fenfluramine-mediated HVD. It is important to note that in both the Fitzgerald et al. (2000) and Rothman et al. (2000) studies, phenteramine did not exhibit high affinity agonism of 5-HT2B receptors (Table 1); correspondingly, formulations consisting of phenteramine alone (i.e., in the absence of fenfluramine)—in use for decades prior to the Fen-Phen HVD outbreak—have not been associated with the disease (Rothman, Baumann et al. 2000).
Following the Fen-Phen episode, a number of additional HVD-inducing drugs have been identified. First among these were the non-specific dopamine agonists pergolide and cabergoline, prescribed for Parkinson’s disease (Antonini and Poewe 2007; Schade, Andersohn et al. 2007; Zanettini, Antonini et al. 2007). The stories of pergolide and cabergoline are similar to that of fenfluramine. The first preliminary account of potential pergolide-mediated HVD was reported in 2002 (Pritchett, Morrison et al. 2002). Dopamine agonists had become popular anti-Parkinsonian therapeutics, and many feared that these drugs would suffer the same fate as fenfluramine. Around the same period of time, many drugs were beginning to be profiled at 5-HT2B receptors including anti-Parkinsonian drugs, amphetamine derivatives and other medications (Newman-Tancredi, Cussac et al. 2002; Setola, Hufeisen et al. 2003). Of the large number of drugs tested, pergolide, cabergoline, MDMA and its active N-demethylated metabolite 3,4-methylenedioxyamphetamine (MDA) were found to exhibit potent agonist activity at 5-HT2B (Newman-Tancredi, Cussac et al. 2002; Setola, Hufeisen et al. 2003). It is interesting to note in this regard that Newman-Tancredi and colleagues were solely interested in differentiating activity of several Parkinsonian drugs at various serotonergic receptor subtypes and suggested that 5-HT2B activity might be associated with some sort of therapeutic action of pergolide and cabergoline rather than mediating side-effects. On the other hand, Setola et al. correctly predicted that pergolide and other drugs which were 5-HT2B agonists (including MDMA and MDA) would be associated with a high incidence of HVD. Accordingly, the incidence of HVD from pergolide and cabergoline has been found to be 23.4% and 28.6%, respectively (Zanettini, Antonini et al. 2007); no significant increase in HVD incidence has been associated with any of the other dopamine agonists. Further, three of the other drugs (lisuride, bromocripine, and terguride) were found to be potent agonists at 5-HT2A receptors (Newman-Tancredi, Cussac et al. 2002), but have little to no activity at 5-HT2B receptors (Huang, Setola et al. 2009). None of these drugs has been subsequently reported to be associated with HVD (Elangbam 2010). These findings lend further credence to the theory that the 5-HT2B receptor is the molecular culprit for drug-mediated HVD. Tables 1 and and22 contain partial compilation of the results from these early studies and offer a representative view of the data that first indicated agonist activity at 5-HT2B receptors leads to the development of HVD.
Following the Fen-Phen and pergoline/cabergolide findings, many other drugs have been tested for agonist activity at 5-HT2B. Specifically, the NIMH Psychoactive Drug Screening Program has sought to screen large numbers of FDA-approved drugs and drug-like compounds for agonist activity at 5-HT2B receptors in order to identify drugs/chemotypes with HVD liability (Setola, Dukat et al. 2005; Roth 2007; Huang, Setola et al. 2009). The most recent findings in this effort identified 27 5-HT2B receptor agonists among 2200 drugs and drug-like compounds (Table 3; (Huang, Setola et al. 2009)). Of these, 14 had previously been identified as 5-HT2B receptor agonists, including 7 known HVD-associated drugs (Table 3). Six of the compounds that had not previously been identified as 5-HT2B receptor agonists are currently approved medications (Table 3, italics), with five of these acting as potent agonists: guanfacine, oxymetazoline, quinidine, xylometazoline, and fenoldopam. Guanfacine and quinidine were noted to be of particular concern. Guanfacine is often prescribed as an antihypertensive agent, and quinidine is used to treat arrhythmia. As such, both tend to be prescribed for sustained periods and could potentially lead to valvular problems. The recent indication of guanfacine to treat attention deficit hyperactivity disorder in pediatric patients is of particular concern (Huang, Setola et al. 2009). Of less concern (perhaps) are xylometazoline, oxymetazoline, and fenoldopam, which are typically prescribed for relatively short time periods. However, all of these compounds should be studied closely to ensure that there is no link to HVD (Bhattacharyya, Schapira et al. 2009).
One of the identified agonists, ropinirole, was found to be much less potent than the other compounds (active only at 1 μM) and therefore may not be associated with inducing HVD. Indeed, ropinirole, which has been used for several years to treat restless legs syndrome, has been considered ‘safe’ vis-à-vis HVD. However, a very recent study documented a case of retroperitoneal fibrosis in a patient using ropinirole (Parissis, Papachristodoulou et al. 2010). Additionally, at least four other cases of HVD have been reported in patients using ropinirole (Parissis, Papachristodoulou et al. 2010). While these relatively few reports may not indicate a statistically increased risk of HVD with ropinirole use, these types of cases should be continuously monitored to ensure that no causal link exists.
Though unfortunate in terms of human and economic loss, the fenfluramine case has led to some important insights. The first and most impactful of these is that screening drugs for 5-HT2B receptor agonist activity is the best in vitro method for identifying potential drug-inducing HVD candidates (Rothman, Baumann et al. 2000; Setola, Hufeisen et al. 2003; Roth 2007; Elangbam 2010). The case of MDMA supports this recommendation. In 2003, we (VS and BLR), in collaboration with Richard Rothman’s group at NIDA, profiled the serotonergic amphetamine derivative MDMA to identify targets other than the monoamine plasmalemmal transporters. The screening campaign revealed that MDMA and MDA were both moderately potent 5-HT2B receptor agonists, and that they stimulated the proliferation of primary human VIC cultures in a 5-HT2B receptor-dependent manner (Setola, Hufeisen et al. 2003). These findings led us to predict that MDMA use may be a risk factor for HVD. In 2007, a clinical study reported an increased prevalence of HVD in MDMA users (Droogmans, Cosyns et al. 2007). Similarly, recent clinical findings have indicated that long-term usage of the fenfluramine-derivative benfluorex can lead to HVD (Noize, Sauer et al. 2006; Boutet, Frachon et al. 2009; Le Ven, Tribouilloy et al. 2010; Weill, Paita et al. 2010). Benfluorex is an anorexigen and hypolipidemic agent commonly prescribed to overweight diabetic patients in Asia, Europe and South America; however, the severity of the clinical case reports led the European Medicines Agency to withdraw benfluorex from the European market in December 2009 (Droogmans, Cosyns et al. 2010). Given that norfenfluramine is a known metabolite of benfluorex, the association of this drug with HVD should not have come as a surprise (Droogmans, Cosyns et al. 2010). Thus, the aforementioned studies support the claim that in vitro screening of drugs at 5-HT2B receptors can identify previously unknown drug-inducing HVD candidates.
Another lesson that can be learned from the role of 5-HT2B receptors in HVD is that 5-HT2B receptor antagonists may be potential prophylactics and/or treatments. Indeed, if the activation of 5-HT2B receptors (on heart valve interstitial cells and/or other cells) contributes to HVD, it is likely that blockade of 5-HT2B receptors may antagonize the onset and/or progression of the disease (Roth 2007). Furthermore, there are a number of 5-HT2B receptor antagonists that are FDA-approved medications (e.g., antidepressants, antipsychotics, antihistamines) with established records of safety and tolerability (Rothman, Baumann et al. 2000; Setola, Hufeisen et al. 2003; Roth 2007; Huang, Setola et al. 2009). Along these lines, Droogmans and colleagues recently reported that the antihistamine cyproheptadine—which is also a potent 5-HT2B receptor antagonist (Young, Khorana et al. 2005)—mitigated pergolide-induced HVD in rats (Droogmans, Roosens et al. 2009), as predicted by Roth (2007). Similar experiments using other 5-HT2B receptor antagonists, and analyses of HVD prevalence among large numbers of patients taking medications with 5-HT2B receptor antagonist activity (e.g., antihistamines, antidepressants, antipsychotics), will be informative in terms of HVD prevention/treatment strategies.
The development of a suitable therapeutic to prevent/retard HVD depends on the ability to target the root cause of the disease, which ultimately manifests itself as thickening and stiffening of HV leaflets, which diminishes the ability of the HV to maintain directionality in blood flow. At a cellular level, HV stiffening is believed to be caused by activation of VICs to a myofibroblast phenotype (Rabkin, Aikawa et al. 2001; Rabkin, Hoerstrup et al. 2002; Rabkin-Aikawa, Farber et al. 2004; Aikawa, Whittaker et al. 2006). Once activated, these cells increase ECM deposition, which directly leads to the decreased compliance of the leaflets observed in HVD. Additionally, increased mechanical strain has been shown to exacerbate VIC activation (Balachandran, Konduri et al. 2006; Merryman, Lukoff et al. 2007; Balachandran, Sucosky et al. 2009; Balachandran, Sucosky et al. 2010). In order to develop a strategy to prevent or treat HVD, a more thorough understanding of the cellular signaling and subsequent tissue-level changes involved in the progression of HVD is needed to elucidate the relevant molecular targets. Given the large number of currently-approved medications that exhibit antagonist/inverse agonist activity at 5-HT2B receptors (Huang, Setola et al. 2009), we believe that the 5-HT2B receptor is a tractable target to achieve these therapeutic goals.
Given its negative history, the idea of targeting 5-HT2B for therapeutic gain may initially seem counter-intuitive; however, studies have begun to explore the potential benefits of controlling 5-HT2B receptor signaling (Fabre, Marchal-Somme et al. 2008; Monassier, Laplante et al. 2008; Porvasnik, Germain et al. 2010). Just as agonists of the 5-HT2B receptor have been observed to lead to HVD, many of these agonists have also been implicated in fibrotic responses and ECM alterations that lead to other pathologies such as ventricular hypertrophy (Jaffre, Callebert et al. 2004; Jaffre, Bonnin et al. 2009) and pulmonary arterial fibrosis and hypertension (Esteve, Launay et al. 2007). Correspondingly, genetic deletion of 5-HT2B receptor expression in mice has been shown to lead to incomplete cardiac development characterized by ventricular dilation and a lack of tissue integrity (Nebigil, Choi et al. 2000; Nebigil, Hickel et al. 2001). Taken together, these results indicate that 5-HT2B receptors play a crucial role in the maintenance of ECM homeostasis in cardiac tissues, and with a better understanding of the downstream effectors of these receptors, the pathways may be able to be manipulated to therapeutically target cardiac fibrotic diseases (Hauso, Gustafsson et al. 2007).
As an illustration of this, studies in pulmonary fibrosis have shown that 5-HT2B receptor antagonists can effectively reduce fibrotic lesions in a mouse model (Fabre, Marchal-Somme et al. 2008), and the selective 5-HT2B receptor antagonist PRX-08066 has been shown to increase ventricular ejection fraction and reduce hypertrophy and vascular remodeling in a rat model of pulmonary arterial hypertension (Porvasnik, Germain et al. 2010). A 5-HT2B receptor antagonist may be able to function similarly in preventing HVD by blocking the fibrotic response of VICs to other, non-serotonergic stimuli.
As mentioned previously, TGF-β1 is believed to be a key mediator of the cellular changes that lead to HVD. Unfortunately, the ubiquity of TGF-β1 signaling makes this molecule a poor therapeutic target. A more appropriate therapeutic goal would be to interrupt TGF-β1 signaling through a separate pathway that is more localized to HV tissues, and the 5-HT2B receptor signaling pathway may be a promising candidate to achieve this goal. Evidence suggests that the signaling pathway from 5-HT2B receptors may crosstalk with TGF-β1 signaling pathways (Jian, Xu et al. 2002), which may be mediated in part by the tyrosine kinase Src (Fig. 6). Src is known to play an important role in both 5-HT2B and TGF-β1 receptor signaling pathways and has been shown to be involved in the 5-HT2B receptor-dependent regulation of the platelet derived growth factor receptor (Nebigil, Launay et al. 2000; Xu, Jian et al. 2002; Mishra, Zhu et al. 2007; Pechkovsky, Scaffidi et al. 2008; Samarakoon, Higgins et al. 2008; Skhirtladze, Distler et al. 2008). Therefore, 5-HT2B receptor blockade has the potential to function in two ways: 1) preventing proliferation and fibrotic ECM accumulation by VICs directly, and 2) interacting with TGF- β1 signaling pathways to prevent VIC myofibroblastic differentiation (Fig. 6).
Additionally, AT1R antagonists have been observed to inhibit TGF-β1 signaling in Marfan’s syndrome (Habashi, Judge et al. 2006; Cohn, van Erp et al. 2007; Brooke, Habashi et al. 2008), and as such, the AT1R antagonist losartan has shown tremendous clinical promise in treating afflicted patients. Given that AT1R and 5-HT2B receptors may be functionally linked in other cell types (Jaffre, Bonnin et al. 2009), 5-HT2B receptor antagonists may work in a similar manner to mitigate TGF-β1 signaling. Therefore, 5-HT2B receptors may provide a localized pharmacological target to prevent the VIC-mediated fibrotic changes that characterize HVD. As noted above, many 5-HT2B receptor antagonists are currently FDA-approved and used clinically to treat other diseases, with acceptable tolerance (including but not limited to: Amesergide, Amisulpride, Bromocryptine, Lisuride, Chlorpromazine, Clozapine, Cyproheptidine, Metergoline, Mianserin, Olanzapine, Aripiprazole, Risperidone, 9-OH-risperidone, Terguride, Yohimbine, Ziprasidone, Roxindole). Furthermore, it is interesting to note that lisuride has been shown to be an antagonist at 5-HT2B receptors (Newman-Tancredi, Cussac et al. 2002; Elangbam 2010) and was prescribed for more than 30 years without a single known report of HVD (Hofmann, Penner et al. 2006). While the absence of documented cases does not necessarily lead to the conclusion that lisuride prevented HVD in these patients, it does seem reasonable to believe that given the large population of patients that received lisuride and its known serotonergic activity, even background levels (i.e., not significantly higher than the total population) would have been reported. In conclusion, we believe that these types of molecules should be tested for efficacy in preventing or treating HVD, as has been suggested (Roth 2007).
JDH was supported by an AHA Pre-doctoral Fellowship (10PRE4290020). BLR and VS were supported by the NIMH Psychoactive Drug Screening Program, U19MH82441, RO1MH61887. BLR was also supported by the Michael Hooker Distinguished Professorship in Pharmacology. WDM was supported by the AHA (0835496N and 09GRNT2010125), Wallace H. Coulter Foundation (Early Career Award), and NIH (HL094707).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.