Chronic heart failure is a common consequence of various heart diseases. Mechanical force is known to play a key role in heart failure development through regulating cardiomyocyte hypertrophy. In order to understand the complex disease mechanism, this article discussed a multi-disciplinary approach that may aid the illustration of heart failure molecular process.
Heart disease is a major cause of mortality in modern society [1,2]. Chronic heart failure (HF) represents 8% of the disease population , and more than 70% of which had a history of arterial hypertension and 25% were preceded by ischaemic infarction [4,5]. Studies demonstrated that HF is an end result of interactions of a range of molecular pathways from mechanical signalling, Ca2+ handling to energy metabolism [6–10]. Mechanical force is suggested to be a key player in HF development [11,12]; therefore, this mini review will focus on the HF mechanisms associated with mechanical forces and how a simple in vivo model, such as fruit fly (Drosophila melanogaster), may aid the studies using an empirical data-based in silico approach.
Cardiac hypertrophic transition and mechanical forces
Blood flow by exerting haemodynamic force helps the heart to evolve from its initial tubular shape to the round mature shape with separate chambers during embryonic development [13–15]. During postnatal growth, blood volumetric increment, presented as a load ‘stretch’ force as well as a shear force, continuously stimulates heart hypertrophy [11,12] by adding extra sarcomeres without increasing the cell number. By doing so, the heart reduces its strain in counter-balancing the imposed forces, so the process is also known as adaptive or compensative hypertrophy. The forces are shown to have differentiated effects on muscle cells elongating or myofibrils lengthening , for example, only the myofibrils in a sharp angle to the flow direction had the most hypertrophic change versus those in a parallel or perpendicular position to the flow direction. This effect of the force on myofibrils is apparently crucial to both the formation and keeping of the heart shape. Myofibrils in Drosophila melanogaster heart also run spirally along the long axis of the heart so in a range of angles to the flow direction of the haemolymph in the heart. This indicates that there is a shared mechanical regulatory mechanism for heart growth across species.
In addition to meet the demand of body's physiological activities, adaptive hypertrophy also occurs in early arterial hypertensive heart due to pressure overload . Therefore, it has been a long time research topic on the cause of the transition from the adaptive to maladaptive hypertrophy and eventually HF over the hypertension development. Establishment of clinical hypertension means that arterial condition has become rigid or less flexible, suggesting the formation of vascular fibrosis . In the excess pressure resistance hypertension mouse model, the failing heart displays a significant reduction in capillary density . This implies a poor blood supply to some areas of the heart, where cardiac fibrogenesis and tissue damage as widened interstitial space and degeneration of hypertrophied myocardial cells in the failing heart may occur [19,20]. These pathological changes provide valuable insights for understanding the hypertrophic transition.
Heart failure due to organ survival strategy via integrin regulation?
Heart failure (HF) is marked by tissue fibrosis. Studies suggest that the fibrogenesis may start from localized poor blood perfusion or insufficient oxygen supply. This condition will cause ATP production reduction and, in turn, results in uneven magnitude of contractions of the cells across the border of the affected and unaffected areas. As a result, mixed signals will be sensed by integrins and cause varied cellular responses from hypertrophy to atrophy. Consequently, fibrogenesis is triggered for strengthening the heart against the persistent pressure overload. Without rhythmic contractility, the fibrotic tissue causes HF development. This hypothesized route for hypertrophic–atrophic transition is discussed below.
Enhanced contractile magnitude by ‘stretch’ intensified Ca2+ transient
One of the immediate early cardiomyocyte responses to mechanical stretch is a quick calcium influx through highly Ca2+-selective mechanic-sensitive transient receptor potential (TRP) cation channels [21–23]. The initial Ca2+ current induces a strong Ca2+ transient through activating the cell membrane-located L-type channels and SR (sarcoplasmic reticulum)-located RyR to release Ca2+ into the cytoplasm (called CIC), as illustrated in Figure 1I–II. The amount of Ca2+ delivered to and the rate of its removal from cytoplasm determine the intensity and duration of myocyte contraction  by making more troponin C available from calcium binding to troponin T. Such strong contraction through activating integrins  will trigger the cascade of integrin pathways to carry out hypertrophic activities (detailed later).
A schematic diagram of a potential route of HF development under persistent pressure overload.
Cell hypertrophic response induced by the calcium ion influx has been studied at various levels of cardiac functions. First, a decreased amplitude of Ca2+ transient (indicators of reduced Ca2+ release from SR [26,27]) in cells and tissues isolated from failing human hearts correlated with decreased EC coupling gain  and reduced myocardial contraction [29,30]. Ca2+ influx is also found to play a role in fibrogenesis, for example, the family member TRPV4 opens and causes Ca2+ influx in response to changes in plasma membrane stretch and promotes differentiation of fibroblasts and tissue fibrosis [31–34]. Furthermore, TRPC1 and TRPC6 may be activated directly by stretch and implicated to be associated with HF [35–37]. Mice without TRPC1 fail to produce the maladaptive cardiac phenotype , indicating that calcium influx is a necessary step of maladaptive hypertrophy in HF development. These findings lead to the proposal  that pathological remodelling of calcium handling is one of the routes that can lead to human HF.
Mechanically activated integrin pathways promote cardiac hypertrophy
Integrins are heterodimeric membrane protein, consisting of one α and one β subunit with a single transmembrane domain. There are 24 mammalian integrins formed from different combinations of 8β and 18α isoforms, serving cellular proliferation, differentiation and cell migration, as reviewed elsewhere [25,39–44]. Integrins are a constitutive part of cell membrane-associated focal adhesion complex of multiple proteins. In myocardium, the adhesion complex locates between the sarcolemmal membrane and z-discs (called costamere) and specialized intercalated discs (IDs) at the end of longitudinal myocytes [10,41,45]. In response to mechanical stretch force, the extracellular N-terminus of β-integrin subunit binds to extracellular matrix (ECM) collagen, which allows α-subunit to free the β-intracellular C-terminus to interact with other proteins, such as cytoskeleton protein talin and signalling molecules ILK and FAK, directly or indirectly [46–48]. Through these interactions, integrins activate hypertrophic activities in the heart.
Integrin ‘involvement’ is another immediate early cardiomyocyte response to physical stimulations . Upon activation by the force generated from internal muscle contraction (inside-out signalling) , integrins start clustering within seconds  and by 5–10 min, the focal adhesion components or adaptor proteins are connected to the cytoplasmic domain of β-integrins (β1 is a major isoform in mammalian myocardium) and start recruiting actins to sarcomere for hypertrophy or maintaining process [50–53] (Figure 1II). By force-sensitive assembling and disassembling of focal adhesion components , including mRNA and ribosome mobilizations [41,42,54,55], integrin-mediated adhesion complex plays a pivotal role in heart hypertrophy.
Integrin protein levels increased during the adaptive hypertrophy in arterial constricted mice hearts at early hypertension . This corresponds to the elevated contractility and strengthened IDs  as well as forming new IDs at the lateral site with neighbouring membrane of the longitudinal cells observed by Maron and Ferrans . The newly developed lateral IDs provide sites for new sarcomeres to be added to widen the cells. This event described as lateral process and the mechanism postulated by the authors that localized mechanical tensions induce their growth. These tensions are due to shearing forces exerted at side-to-side junction when contraction occurs at different rates or magnitudes in adjacent cells, leading to asymmetric and complementation of these junctions and to the eventual formation of the lateral processes.
In contrast with the stretch-induced calcium influx that has no influence on hypertrophic gene expression , in addition to expanding sarcomeres, integrins also stimulate hypertrophic gene expression, such as those for α- and β-myosin heavy chains [59,60], cardiac troponin C and T  and α-cardiac actin . This activity of integrin is suggested to be carried out via tyrosine kinase pathways, involving MAP kinase pathways . When β1 integrin is excised, the heart fails to withstand excessive pressure load , presumably due partly to an insufficient production of sarcomere proteins. Furthermore, defects in ECM ligand laminin-α4 and ILK caused dilated cardiomyopathy . A similar observation is also reported in patients with atrial fibrillations . Taken together, these evidences demonstrate that by both, or either, of new sarcomere building and hypertrophic gene expressions, integrins play a key role in matching the cardiac capacity to the demand of body's physiological activities.
Hypertrophy to atrophy — via inside-out integrin down-regulation?
Failing hearts at the end-stage dilated cardiomyopathy typically showed hypertrophy, atrophy of myocytes and an increased amount of fibrosis with alteration of decrease in myofilaments, ranging from rarefication to the complete absence of sarcomeres in cells filled with unspecified cytoplasm . Efforts, since decades ago , in understanding the adaptive hypertrophic transition to the pathological setting have revealed that integrins, among other cardiac genes, show an expression pattern with an increased level in early adaptive hypertrophy, but a decreased level in later maladaptive hypertrophy in hypertension development . Re-expressed foetal isoform α-skeletal actin and increased β-myosin heavy chain mRNAs may fall back to previous levels following the hypertrophy regress within weeks of stimulus removal in physiological hypertrophy [11,67–69], for instance, in trained athletes [70,71]. In non-compensative hypertrophy and HF, however, elevated α-skeletal actin reduces back to control animal level while the pressure condition is still in place [19,68]. Furthermore, short-time (2 weeks) conditional cardiac expression of AKT(1), a serine/threonine protein kinase that mediates cellular growth responses, in transgenic mouse induced a reversible compensative hypertrophy, whereas its continuous expression for 6 weeks caused HF . Explanation of the prolonged hypertrophy that promotes an adverse opposite effect is incomplete yet.
Persistent pressure overload should continuously stimulate cardiac hypertrophy. But, evidence indicates that myocytes could expand out of proportion to the interstitial space, where vascular network lies, in the confined heart and lead to poor supply of nutrients and oxygen . This is supported by a significant reduction in capillary density  and reduced energy production in the failing heart . As a result, decreased contractility of myofilaments in affected tissue areas is expected, which through inside-out signalling of integrin could weaken the focal adhesions or IDs and cause localized atrophy. This role of the inside-out regulation of integrin is believed here because outside-in signal of pressure overload is unchanged if not stronger, so that the stretch-enhanced calcium transient and resulted strengthened contractility should not decrease. Furthermore, fatty acid oxidation, which produces up to 90% ATP in adult, is decreased in the failing hearts  with accompanied alteration of mitochondrial structure [58,75] and abnormality of metabolism pathway that result in decreased energy production, energy transfer and energy utilization [74,76]. Altogether, it is suggested that during hypertension development, cardiac persistent pressure overload is through the inside-out down-regulation of integrin to turn the initial adaptive hypertrophy to atrophy of the heart.
Cardiac fibrogenesis compensates for weakened failing heart?
In the heart, interconnected fibroblasts with the deposited ECM, consisting of collagens, elastin, fibronectin and ground material, hold cardiomyocytes together . Hypertrophied left ventricles produced total more collagens with increased type III in the early hypertension , which has a strong bond (160pN) to resist rupture forces [57,79]. Excess pressureload on the heart as well as hypoxia can trigger cardiac fibroblasts differentiated into myofibroblasts [80,81], and the latter produces different proportion of collagen III and collagen I from the former and matrix metalloproteinases with the balance to promote cardiac fibrogenesis . In advanced hypertension in human and the aortic constricted mice, the hearts have significantly increased pericellular collagen weave fibres [18,83], where they are proposed to provide support to the collapsed or weakened intercalated structure that built up in early phase of compensative hypertrophy in the hypertension patients . Those fibrotic areas form barriers or gaps, so that the normal passage of electrical current of action potentials in the heart is blocked. As a result, arrhythmia develops. Fibrotic tissue does not contract, thus giving the impaired contractility of the heart. Both the functional aspects provide the traditional diagnostic basis of a (chronic) failing heart.
Cardiac interstitial fibrosis formation induced in arterial hypertension by excess pressure does not involve cell injury compared with the scar formation induced in coronary arterial disease by cell death followed by inflammatory response [7,84]. But HF development post the formation of the-scar in coronary disease likely shares the same developing route with the hypertension-induced HF, which is also strongly associated with ageing population . Thus, to balance the myocyte hypertrophy to a level that does not trigger the transition of fibroblasts to myofibroblasts is believed a way to halt further pathological and age-related cardiac regression and potentially reverse the pathological remodelling.
Modern imaging technology can localize the region of the fibrotic tissue by its increased stiffness and using the relationship analysis of stress (pressure imposed on heart wall) and strain (heart deformation) to predict a deleterious cardiac failure [85,86]. For example, slowed relaxing period of a local area of a heart can be captured from congenital cardiomyopathy patients by the imaging-based strain (deformation) rate analysis . Sensitivity of such analysis can even capture the congenital disease carrier before the formation of fibrotic scar tissue, that is, in the state of increasing in myocardial collagen synthesis without overt disease [85,87]. Therefore, the imaging technologies promise an early diagnosis tool for the transition of adaptive hypertrophy to HF.
Studies in computational modelling of mammalian heart
Primary elemental aspects of the heart to be brought together to computationally simulate or faithfully analyze heart behaviour were described by Hanna : 1) description of the shape and size of the left ventricle (LV), 2) the orientation and motion of muscle fibres in the LV wall, 3) the mechanics and thermodynamics of isolated muscle fibre contraction, 4) the haemodynamic load imposed on the LV by the systemic circulation and 5) the assimilation of these four elements into a functional model. Subsequently, a finite-element mathematical method is employed to build a heart model that can analyze those elemental aspects and more, such as cardiac tissue's inhomogeneity, anisotropy and non-linear material properties . Such a model with detailed 3D cardiac geometry and myofibre architecture  has been continuously improved in parametric sensitivity [91–93], for example, analyzing tissue contractile properties between fibrosis and myocardium. Improvement in modelling cardiac response to imposed mechanical forces is expected to be supported from clinical capabilities, in analyzing localized strain rate of hearts with congenital disease .
Heart modelling is currently an active research field. A wide range of structural and functional modelling methods and their applications have been developed in the past decade. To name a few, there are models for investigating structural and molecular determinants of ventricular electromechanical function ; anatomically detailed continuum models for integrating biophysical processes into simulations of regional-specific mechanics and electrophysiology of the intact heart ; models for molecular and subcellular-scale study of energy consumption, such as nucleotide diffusion in the cardiac myofilament lattice ; models for multi-scale modelling of large biomolecular complexes at atomic resolution ; models for exploring arrhythmia mechanism with respect to transmembrane ionic current flow in cardiomyocytes [98,99] and models for detailed description of histoanatomy up to micro-scale resolution [100,101]. An up-to-date overview of the evolution of the methodologies, used in constructing heart models and their applications, is provided in an excellent review by Lopez-Perez et al. .
It is noticeable, however, that there seems to be no heart model available in the area of mechanical signalling transduction despite the essentiality of this area in the study of heart development and disease. However, relevant in vitro studies in modelling the dynamic behaviour of cellular micro-structural networks influenced by genes in the mechanotransduction pathways, such as integrins, have generated interesting insights [103,104]. The challenge is how to best use the rich in vitro observations, that is how to correctly integrate the in vitro findings into the corresponding state of heart physiology or disease to enhance heart-modelling capability. This issue is discussed in the last section.
D. melanogaster heart studies
Drosophila has a single tubular cardiovascular system, also known as dorsal vessel . During postembryonic larval development, it is formed by a posterior resided heart (∼1–2 mm long), which has rhythmic contractility and three pairs of ostia opening at both sides of the heart, and a narrow aorta that spans most of the remaining body length, which passively guides haemolymph (blood equivalent) flow out of the cardiovascular system . In the open circulatory system, haemolymph flows through extravascular interstitials and back into the heart again via the ostia . Two rows of cardioblast join up at cell membrane along the body midline by specialized focal adhesion junctions, which resemble the heart's ‘IDs’ of vertebrates. Myofilaments mainly form longitudinal bundles, inserted at the boundary between adjacent cardioblasts . In the study, the heart wall is shown, on both the luminal and the outer surfaces, to be covered by a layer of ECM. The ECM layers are connected to the cardiomyocyte layer by hemi-adhesion junctions, presumably at the z-lines like the vertebrate's costamere.
Studies of D. melanogaster heart have made valuable contributions towards our understanding of heart biology and disease in the last two decades [108–110]. Identification of heart fate determination gene tinman led to the discovery of vertebrate homologue Nkx 2.5 [109,111]. Structural and functional homology of many genes between D. melanogaster and vertebrate hearts, for example potassium channel, involves cardiac rhythmic activity ; βPS integrin, the human β1 homologue, influences heart ageing [113,114]; NOT3 regulates heart function  and tropomyosin mutation causes arrhythmia . Furthermore, expressing mutated human gene (δ-sarcoglycan) in fruit fly is capable of recapitulating human dilated cardiomyopathy , supporting the observation that heart physiology is homology between mammalian and D. melanogaster heart . These data provide the basis that haemodynamic force's regulation of fly heart behaviour is highly interspecies interpretable.
Since βPS integrin, the D. melanogaster homologue of human β1 , also expresses in the fly heart, it is intriguing to learn whether the important mechanotransduction pathway is shared by D. melanogaster. Thus, we developed a computational modelling method  based on heart volumetric dynamics. With the method, we estimated the shear force imposed on the wall of the larval heart by haemolymph flow when the expression of βPS gene (mys) was knocked down using the Gal4–UAS system . The force estimation is based on the analysis for wall shear stress (τ) in the arterial system: τ = 4ηq/πr3 , where η is haemolymph viscosity (similar to human plasma); q is the haemolymph flow rate, estimated based on the volumetric dynamics of 3D computational model heart that was geometrically created using the imaging of third instar larval heart with transgene 1029-Gal4;UAS-RedStinger;UAS-mysRNAi and r is the average radius of the model heart.
As shown in Figure 2B, the shear stress corresponding to peak systolic contraction is the highest, and it reduced seven times in the knockdown hearts compared with the control. Immunoconfocal microscopic analysis revealed that this shear stress reduction corresponded to the discontinuation or detachment of myofilaments along the heart ventral midline (Figure 2A). An accompanied change is the row of discrete integrin-intense ‘bars’ (presumably the fly's ‘IDs’) disappeared in the knockdown hearts. Further knockdown (by exposing the larvae carrying the same Gal4-UAS integrin knockdown construct to higher temperature) of the gene resulted in a massive loss of the myofibrils (unpublished data not shown). The correlation of the downgradient of integrin levels with the severer myofibril's damage appears to resemble the disappearing of myocardium in dilated cardiomyopathy in failing human hearts.
Reduced shear force in larval D. melanogaster heart with myofilaments disconnected between cells due to βPS integrin knockdown.
Shear stress estimated here is close to the reported value in prepupa using a different method , which validates the approximation capacity of our approach. An integrin level-dependent myofibril phenotype, from disorganization to massive loss of the myofibril content, provides evidence that larval development is a highly sensitive period for studying the force regulation of heart behaviour, particularly at the indicated transition between physiopathology that uniquely allows exploration of the cardiomyopathy at its starting moment. Consider the relation between myofibril orientation and the direction of the shear force , fly's larval heart lends itself a powerful in vivo tool for exploring the force regulation mechanisms.
Quantitative mechanic forces as reference for interspecies data interpretation
It is hoped that the above discussions have argued for the need to employ a spatial–temporal computational modelling approach to illustrate the dynamic hypertrophic gene networks underlying heart physio-pathological process. A current major issue in doing so is lacking a linker or standard reference to correctly integrate vastly available in vitro-generated molecular data into a 3D model of defined state of physiology or disease at organ's organization level. Phenotypes in a quantitative correlation with mechanic forces allow the integration of the former based on the latter as a reference system. But the correlations between individual molecules or genes and organ states are hard to obtain from in vivo studies in vertebrates because of the tissue organizational complexity-associated poor experimental accessibility. Such hurdle can be overcome by translating D. melanogaster heart studies into the vertebrate system by taking the advantage of physiological homology between the two species. This is because the D. melanogaster heart exists as both a perfect organ and a single layer of cardiomyocytes, which, to a certain extent, resembles mammalian cardiomyocytes in culture, and the advantage of which is that the single sheet of cells allows force-phenotype to be studied at (sub)cellular level in situ.
As outlined in the diagram below, this proposed data intertranslation across species is not expected to be a straightforward one. Arrows indicate the direction that data feed to, and two directional arrows indicate the verification steps needed in the data-intertranslational process.
This mini review focused the discussions on our current understanding of HF development. Overgrowing by hypertrophy of the myocytes under persistent pressure overload in advanced hypertension is believed to cause a chain reaction that leads to down-regulation of integrins and pathways by weakened contractility due to insufficient energy productions. As a result, atrophy followed by HF develops. A major issue in tackling the chronic HF is the poor accessibility of the complex cardiac system in vertebrates. For this reason, D. melanogaster larval heart studies have been proved capable of providing an indirect alternative research approach to aid a correct integration of mammalian in vitro findings of hypertrophic molecular genetics by in silico means. Together with a combined effort of physiology, molecular genetics, biofluid dynamics, biomechanics, imaging technology and computational biology, ultimate control of HF may be achieved sooner.
EC coupling, excitation-contraction coupling; ECM, extracellular matrix; FAK, focal adhesion kinase; HF, heart failure; IDs, intercalated discs; ILK, integrin linked kinase; LV, left ventricle; MAP kinase, mitogen-activated protein kinase; MMPs, matrix metalloproteinases; RyR, ryanodine receptor; SR, sarcoplasmic reticulum; TRP, transient receptor potential.
I thank Dr Jian Chang (The Medium School, Bournemouth University) and Dr Mandun Zhang (School of Computer Science and Engineering, Hebei University of Technology, Tianjin, China) for designing and executing the 3D larval heart modelling and haemodynamic calculations. My thanks also go to Dr David A. Johnston (Biomedical Imaging Unit, Faculty of Medicine, University of Southampton) for the excellent immunoconfocal microscopy work.
The Author declares that there are no competing interests associated with this manuscript.