Gene Profiling Changes in Cytoskeletal Proteins During Clinical Recovery After Left Ventricular–Assist Device Support
Background— After left ventricular–assist device (LVAD) support, a proportion of patients recover sufficient ventricular function to enable explantation of the device. The exact molecular mechanisms involved in myocardial recovery remain unknown. Cytoskeletal proteins are essential for the structure and function of the cardiac myocyte and might play a major role.
Methods and Results— A total of 15 patients with nonischemic cardiomyopathy who required LVAD implantation were studied; 6 recovered sufficiently to allow explantation of the device compared with 9 who did not recover and required transplantation. LV myocardial samples were collected at implantation and explantation/transplantation. Affymetrix microarray analysis was performed on the paired samples and analyzed with reference to sarcomeric and nonsarcomeric cytoskeletal proteins. In the recovery group, of the nonsarcomeric proteins, lamin A/C increased 1.5-fold (P<0.05) and spectrin 1.6-fold (P<0.05) between the times of implantation and explantation. Integrins β1, β6, and α7 decreased 1.7-fold (P<0.05), 2.4-fold (P<0.05), and 1.5-fold (P<0.05), respectively, but integrins α5 and β5 increased 2.3-fold (P<0.01) and 1.2-fold (P<0.01) at explantation. The following sarcomeric proteins changed in the recovered group only: β-actin increased 1.4-fold (P<0.05); α-tropomyosin, 1.3-fold (P<0.05); α1-actinin, 1.8-fold (P<0.01); and α-filamin A, 1.6-fold (P<0.05). Both troponin T3 and α2-actinin decreased by 1.6-fold at the time of explantation (P<0.05). Vinculin decreased 1.7-fold (P=0.001) in the recovered group but increased by 1.7-fold (P<0.05) in the nonrecovered group. Vinculin protein levels decreased 4.1-fold in the recovered group.
Conclusions— Myocardial recovery was associated with a specific pattern of changes in sarcomeric, nonsarcomeric, and membrane-associated proteins, which could have important implications in understanding the mechanisms involved.
After support with a left ventricular–assist device (LVAD), some patients recover sufficient ventricular function to allow explantation of the device.1 The proportion of patients in whom this occurs has previously been thought to be small.2 We have used a combination therapy in which we first induce structural reverse remodeling (with the use of β-blockers, angiotensin-converting enzyme inhibiters, angiotensin II blockers, and the aldosterone receptor blocker spironolactone) followed by induction of physiologic hypertrophy with use of the β2-agonist clenbuterol.3,4 Using this combination therapy, we have observed recovery in approximately two thirds of patients with nonischemic cardiomyopathy. The exact molecular mechanisms involved in the recovery of myocardial function in these patients, however, remain unknown.
Cytoskeletal proteins play a critical role in myocardial structure and function and have been shown to be altered in heart failure and/or familial dilated cardiomyopathy.5–9 Previous studies have examined changes in individual cytoskeletal proteins in patients supported with an LVAD who are bridged to transplantation, but global changes in cytoskeletal proteins in patients showing sufficient myocardial function to allow the device to be explanted have not previously been studied. In this study, we investigated changes in both the sarcomeric and nonsarcomeric cytoskeletal proteins in patients who recovered ventricular function and underwent explantation of the device compared with bridge-to-transplant patients who did not recover and were transplanted.
We compared 6 LVAD patients who received the combination therapy and showed sufficient myocardial recovery to allow explantation of the device with 9 LVAD patients who did not recover (and did not receive the combination therapy) and were transplanted. All patients had undergone coronary angiography within 1 year of LVAD implantation to exclude coronary stenosis.
Of the 6 recovered patients, 5 were male and 1 was female; their mean age was 42.5 years. At the time of implantation, all were inotrope dependent and 1 required intra-aortic balloon pump support. Mean ejection fraction (EF) was 9.2% (range, 5% to 14%), pulmonary capillary wedge pressure was 24.3 mm Hg (range, 12 to 33), and cardiac index was 1.63. All had dilated cardiomyopathy. None had myocarditis (mean duration of heart failure symptoms was 66.2 months; range, 4 to 156), creatine kinase levels were normal, and histologic analysis of the LV core samples removed at implantation all showed dilated cardiomyopathy and no evidence of myocarditis (by Dallas criteria).
Five patients were supported with a Thoratec Heartmate I and 1 with a Heartmate II. Mean duration of support was 366 days (range, 156 to 503). Before explantation, the EF (measured after the pump had been turned off for 15 minutes) was 65. 8% (range, 60% to 72%). Mean cardiac output measured with the pump turned off was 5.8 L/min, pulmonary capillary wedge pressure was 10.8 mm Hg, and cardiac index was 2.9. The recovery in these patients has since been sustained, and the mean time after explantation is now 1121 days. One year after explantation, the EF was 61.8% (range, 44% to 74%), cardiac output was 4.9 L/min, and cardiac index was 2.54.
Of the 9 nonischemic cardiomyopathy patients who did not recover (not treated with combination therapy and not treated with intention to recover), 5 were female and 4 were male. The mean age was 43.3 years. At the time of implantation, 5 were on inotropic support and the mean EF was 15.9%. All were supported with a Thoratec Heartmate I.
Hypertension, which was absent preoperatively, developed in 6 patients while on the device and was treated by a β-blocker in 6 and an angiotensin-converting enzyme inhibitor in 5. The mean duration of support was 121.5 days. Before transplantation, mean right atrial pressure was 11.1 mm Hg and mean (pump on) cardiac index was 2.95.
In addition, we studied 2 patients with nonischemic cardiomyopathy who received the combination therapy but did not achieve sufficient restoration of function for LVAD removal and were transplanted. Both were male, ages 18 and 41 years. Both required inotropic and 1 required intra-aortic balloon pump support before implantation. Periods of support were 601 and 828 days. Before transplantation, their EFs with the pump off for 15 minutes were 30% and 45%, and cardiac indices were 1.7 and 1.8, respectively.
LV apical core myocardial tissue was taken at implantation from all patients and then at either explantation or transplantation, immediately frozen in LN2, and stored in a −80°C freezer. The ethics/institutional review board–approved informed consent was obtained from all participating patients before tissue collection at both institutions.
RNA was extracted, and the Affymetrix human genome U133A gene chip was used to perform the array, as previously described.10–12 All chips were processed through Gene Expressionist (GeneData AG), which uses a hierarchical clustering algorithm to assess similarity. A cutoff of 30% similarity was applied to create groups of similar chips for detection of abnormal signal behavior and masking. For each group of similar chips, a reference chip was computed by a robust, feature-wise average. The algorithm for abnormal signal behavior identifies localized areas of light or dark features, ie, “defects.” Abnormal signal behavior is determined by comparison of each experiment to the reference experiment. Before comparison, the experiment is normalized in memory with respect to the reference experiment. This normalization applies a nonlinear transformation so that the signal response of the experiment equals that of the reference. This ensures the comparability of the experiment to the reference experiment. The difference signal is computed as the logarithm of the ratio of the experiment and the reference experiment. Differentially expressed genes will also lead to positive or negative differences and might be interpreted as defects. This is avoided by using the information of the chip layout: feature sets that belong to the same gene are excluded from defect detection if they show a common deviation from the reference experiment. Feature intensities for each chip were condensed into single intensity values per gene by using the Affymetrix statistical algorithm (MAS 5.0), with ô=0.015, α1=0.04, α2=0.06, and a target intensity of 500. Normalized expression values were used for the paired Student’s t tests. Pathway analysis was assessed with the use of Ingenuity. The Affymetrix microarray data were analyzed with particular reference to sarcomeric and nonsarcomeric proteins.
The quantity of tissue available did not allow for mRNA or protein detection of all cytoskeletal proteins, so real-time polymerase chain reaction (PCR) and Western blotting were performed on a small number for which there was enough tissue available to verify the array results.
Quantitative real-time PCR for α1-actinin and integrin β5 was performed as previously described11 on the implant and explant samples of the recovered group (n=6) to confirm array findings. The primers used were as follows: human actinin-α1, 5′-TCATCTCAGGTGAACGCTTG-3′ and 5′-AGATGTCCTG-GATGGCAAAG-3′; human integrin-β5, 5′-GGAAGCTCACCA-GCAACTTC-3′ and 5′-CTGTCTGTGAGAGGCAGCAG-3′; human cyclophilin A, 5′-TTCATCTGCACTGCCAAGAC-3′ and 5′-TCGAGTTGTCCACAGTCAGC-3′; and human glyceraldehydes 3-phosphate dehydrogenase (GAPDH), 5′-ACCACAGT-CCATGCCATCAC-3′ and 5′-TCCACCACCCTGTTGCTGTA-3′.
Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis and Western Blotting
Vinculin and metavinculin protein expression was analyzed by standard gel electrophoresis followed by Western blotting and immunoprobing with anti-vinculin antibody (Sigma), as previously described,13 on samples from the 5 recovered patients for whom there was enough tissue available. This anti-vinculin antibody recognizes both 120-kD vinculin and 150-kD metavinculin. Levels of expression of these were normalized to levels of GAPDH (Chemicon), which was used as a loading control. Quantification was carried out with Quantity One.
We first used a compendium of paired gene profiles from the 6 recovered-patient nonischemic myocardium samples (at implantation and explantation of the LVAD) and compared them with paired gene profiles from the 9 nonischemic myocardium samples obtained at the time of LVAD implantation and transplantation. A paired Students t test was used to determine statistical significance between pre-LVAD and post-LVAD heart samples within each cohort.
The 2 groups had no significant differences at baseline, as determined by 3 different approaches. First, a simple boxplot demonstrating the range and distribution of normalized expression values showed no significant differences between samples from the recovered and nonrecovered cohorts at baseline. Second, we assessed the coefficient of variation between the pre-LVAD samples in the recovered versus nonrecovered cohorts and found that 93% of the genes in these 15 samples had a variance coefficient of <0.1. Finally, a Student’s t test revealed that with a value of P<0.01 and no fold-change cutoff, only 1.3% of the genes were differentially expressed.
Changes in the Recovered Group Only
In the recovery group, of the nonsarcomeric proteins, the nuclear laminar protein lamin A/C increased 1.5-fold (P<0.05) between the times of implantation and explantation (Figure 1 and the Table). Using a distance metric designed to identify genes whose expression pattern was most closely positively correlated with lamin A/C (Figure 2), we identified tissue inhibitor of metalloproteinase (TIMP)-1, actin, RAB31 (a member of the RAS oncogene family), and chondroitin sulfate proteoglycan (CSPG)-2.
Of the costamere proteins, spectrin increased 1.6-fold (P<0.05), and ankyrin-1 and -3 decreased 1.9-fold (P<0.05) and 1.7-fold (P<0.005), respectively, in the recovered group. The dystrophin-like proteins β-sarcoglycan increased 1.3-fold (P<0.05) and dystrophin decreased 1.4-fold (P<0.05). The transmembrane integrins β1, β6, and α7 decreased 1.7-fold (P<0.05), 2.4-fold (P<0.05), and 1.5-fold (P<0.05), respectively, but integrin β5 and α5 increased 1.2-fold (P<0.01) and 2.3-fold (P<0.01), respectively, between the times of implantation and explantation.
The following sarcomeric proteins significantly changed in the recovered group only: β-actin increased 1.4-fold (P<0.05), α-tropomyosin increased 1.3-fold (P<0.05), α1-actinin increased 1.8-fold (P<0.01), and α-filamin A increased 1.6-fold (P<0.05). Both troponin T3 and α2-actinin decreased 1.6-fold from implantation to explantation (P<0.05).
Divergent Changes in Both Groups
Vinculin decreased 1.7-fold (P=0.001) in the recovered group but increased 1.7-fold (P<0.05) in the nonrecovered group. Syntrophin (a dystrophin-like protein) decreased 1.7-fold (P<0.05) in the recovered group but increased 2.2-fold (P<0.05) in the nonrecovered patients (the Table).
Changes in the Nonrecovered Group Only
In the nonrecovered group, α2-tubulin increased 1.6-fold (P<0.005) and β2- and β5-tubulin increased 1.4-fold (P<0.05). Of the dystrophin-like proteins, dystroglycan increased 1.4-fold (P<0.05); syntrophin-α, 1.5-fold (P<0.01); and syntrophin-β, 2.0-fold (P<0.05) in the nonrecovered group. LIM domain–binding-3 increased 2.2-fold (P<0.05) and actin-binding LIM protein-1 increased 1.9-fold (P<0.05) between implantation and transplantation. LIM domains 1 and 7 increased 1.7-fold (P<0.05) and decreased 1.2-fold (P<0.05), respectively. These did not change in the recovered group. β1-Laminin increased 4.7-fold (P<0.05) and α3-laminin increased 3.4-fold (P<0.05). Troponin I increased 9.2-fold (P<0.005). Myosin 5C increased 1.5-fold (P<0.005), and myosin IXB increased 1.8-fold (P<0.05) in the nonrecovered group. Integrin-α10 increased 1.7-fold (P<0.005; the Table).
One of the most striking differences between the recovered and nonrecovered cohorts was the change in β-integrin signaling networks (Figures 3 and 4⇓). In particular, recovered hearts showed significant decreases in vinculin, Wiskott-Aldrich syndrome protein, p21-activated kinase, Rho, and Graf. In contrast, in the nonrecovered cohort, vinculin, Rho, Graf, src, and focal adhesion kinase were upregulated. Recovered hearts also exhibited a significant increase in actin, v-crk sarcoma virus CT10 homologue, Ar2/3, Cdc42, myosin light-chain kinase, and myosin light-chain phosphatase, whereas these genes were not affected in the nonrecovered cohort.
We also analyzed implant and transplant myocardial samples from 2 patients who received clenbuterol but did not recover. Although this small sample number restricts a meaningful statistical analysis, they showed very similar changes to those of other nonrecovered patients, with only a few exceptions (integrin-α10, β1-laminin, and myosin 5C did not rise between the time of implantation and transplantation).
In the recovered group (n=6), actinin-α1 mRNA expression was increased 1.7±0.4-fold in the explant versus implant samples when normalized to cyclophilin and 2.2±0.5-fold when normalized to GAPDH. Integrin-β5 mRNA expression was increased 1.5±0.34-fold in the explant versus implant samples when normalized to cyclophilin A and 1.8±0.27-fold when normalized to GAPDH. These confirmed array findings (see previous section).
Levels of vinculin decreased between the time of implantation and explantation in every recovered patient examined (n=5; Figure 5). Overall, the mean vinculin level significantly decreased, from 0.33±0.09 OD units at implantation to 0.08±0.06 at explantation (4.1-fold, P<0.01). Metavinculin also decreased between implantation and explantation in every recovered patient examined, from 0.76±0.07 to 0.43±0.13 OD units, respectively, P<0.01.
This is the first study to describe the pattern of cytoskeletal protein changes in patients who recover sufficient ventricular function to allow device explantation with subsequently maintained ventricular function. Previous studies have shown changes in individual cytoskeletal proteins after LVAD unloading as a bridge to transplantation.
Dilated cardiomyopathy is a major cause of congestive heart failure worldwide, and despite advances in medical therapy, its mortality remains high.14 It is often associated with abnormalities of the myocyte’s contractile apparatus and cytoskeleton; however, the underlying mechanisms remain poorly understood. The ability to induce reverse remodeling with an LVAD resulting in myocardial recovery offers a unique opportunity to study the mechanisms of heart failure and its recovery in these patients.
This study has shown that recovery is associated with specific changes in the expression of genes encoding several sarcomeric and nonsarcomeric proteins. There was an increase in the gene expression of lamin A/C in the recovered but not the nonrecovered patients. Mutations in the lamin A/C gene have been previously shown in families with dilated cardiomyopathy.6 Lamins A and C are components of the nuclear envelope and are located in the lamina, a structure associated with the nucleoplasmic surface of the inner nuclear membrane. These highly conserved proteins are transcribed from a single gene, are structurally homologous with other intermediate filaments, and consist of a central rod domain flanked by globular amino and carboxyl domains.15 The protein forms a complex assembly of the filaments. Nuclear lamins contribute to the structural integrity of the nuclear envelope and provide mechanical support for the nucleus. In nondividing cells, lamins may participate in signal transduction by mediating movement between the cytoplasm and the nucleus.16 Lamin mutations are thought to interrupt nuclear function, resulting in cell death.6 In addition, mutations in the rod domain of the lamin A/C gene are thought to alter interactions with cytoplasmic proteins, in particular, intermediate filaments of the sarcomere, the actin-based cytoskeleton, and the sarcolemma.6 In our study, the increased expression of lamin A/C suggests that it might be restored in recovery. Interestingly, we identified several genes whose pattern of expression mirrored that of lamin A/C in the recovery cohort, including TIMP1, actin, RAB31, and CSPG-2.
In our study, 6 dystrophin-like proteins changed. Dystroglycan, syntrophin-α, and syntrophin-β increased in the nonrecovered patients. Sarcoglycan increased but dystrophin decreased in the recovered patients, and syntrophin decreased in the recovered patients but increased in the nonrecovered patients. Dystrophin is thought to provide structural support for the myocyte and the cardiomyocyte membrane. It links actin at the N-terminus with the dystrophin-associated protein complex and sarcolemma at the C-terminus and the extracellular matrix. Mutations in dystrophin can result in skeletal myopathy and cardiomyopathy. Vatta et al17 identified a disruption in the N-terminus of dystrophin by both immunocytochemistry and Western blotting in 18 of 20 patients with end-stage cardiomyopathy (dilated and ischemic), and this disruption reversed after LVAD unloading in 4 of 6 LVAD-supported patients. In a subsequent study, the same group18 demonstrated disruption of the N-terminal dystrophin in both left and right ventricles of failing hearts, which was reversed in 12 of 14 patients using either pulsatile or continuous devices. This suggests that dystrophin may play an important role in reverse remodeling.
One of the most striking findings from our analysis was the high percentage of unique and significant changes in β-integrin signaling. Integrins are heterodimeric cell-surface transmembrane receptors that link the extracellular matrix to the intracellular cytoskeleton. They are expressed on many cells in the cardiovascular system and have multiple functions, including roles in the regulation of gene expression, cell migration, cell proliferation, differentiation, and cell death.19 Interestingly, they have also been shown to function as mechanotransducers20 in response to physiologic and pathophysiologic signals.21 They consist of α- and β-subunits, and ≈24 paired integrins have been identified, although only 12 have been identified in the myocardium.21 The repertoire of integrins expressed on a particular cell can vary temporally in response to various stimuli, eg, hemodynamic load. Integrins bind directly to components of the cytoskeleton, such as α-actinin, through their cytoplasmic tails, and can orchestrate changes in cellular cytoarchitecture. They also signal through molecules such as focal adhesion kinase and the GTPase Rho.
A high percentage of integrins changed in our study. There were also specific changes that support results of previous animal studies. Ross et al22 have previously shown overexpression of β1-integrin to induce a hypertrophic response (and inhibition of β1-integrin to reduce adrenergically mediated hypertrophy) in neonatal ventricular myocytes, which is interesting because β1-integrin decreased in the recovered group in our study. In rodents undergoing morphological cardiac hypertrophy with aortic constriction, increased expression of α7-integrin occurred in another study,23 and patients with Duchenne and Becker dystrophies have increased α7-integrin expression, which is interesting because α7-integrin decreased in our recovered patients.
In our study, alterations also occurred distal to the integrins in the signaling pathway (Figures 3 and 4⇑). Vinculin decreased 1.7-fold (P=0.001) in the recovered group but increased 1.7-fold (P<0.05) in the nonrecovered patients. The decrease in vinculin was confirmed at the protein level in the recovered group. Vinculin is situated at the costameres of the lateral sarcolemma and at the intercalated disc and is 1 of the major components of the linkage system that connects, via the integrins, the intracellular milieu with the extracellular matrix. That vinculin decreased in the recovered patients but increased in the nonrecovered patients is in agreement with a previous study in which vinculin protein was found to be elevated in the myocardium of patients with advanced heart failure due to dilated cardiomyopathy who were undergoing heart transplantation.8
De Jonge et al24 studied the myocardium of patients before and after LVAD support as a bridge to transplantation and found an improvement, but not normalization, in the architecture of actin, tropomyosin, troponins C and T, and titin by immunocytochemistry after LVAD support. Aquila et al25 isolated myocytes from LVAD-supported (bridge-to-transplant) patients and compared them with myocytes from those without LVAD support undergoing transplantation and with myocytes from donor hearts. α-Actinin protein expression was no different among donor, failing, and LVAD-supported hearts, but immunofluorescence showed the LVAD-unloaded hearts to have a pattern of α-actinin resembling that of nonfailing cells.25 In our study, α1-actinin increased but α2-actinin decreased in the recovered patients.
The recovered group of patients was treated with the drug clenbuterol, whereas the nonrecovered group was not, and the effect of the drug on cytoskeletal proteins remains unknown; however, the changes measured in a small number of clenbuterol-treated, nonrecovered patients showed changes very similar to those of other nonrecovered patients.
In conclusion, in this study we have documented a specific pattern of changes in both sarcomeric and nonsarcomeric cytoskeletal proteins in the myocardium of patients who recovered compared with those who did not, suggesting that the cytoskeleton might play an important role in myocardial recovery. This could be helpful in monitoring patients for recovery and in identifying new therapeutic targets for heart failure.
The authors are grateful to the Royal Brompton and Harefield Charitable Trustees, Thoratec Corp, and the Lillehei Heart Institute University of Minnesota for supporting this work.
Mancini DM, Beniaminovitz A, Levin H, Catanese K, Flannery M, DiTullio M, Savin S, Cordisco ME, Rose E, Oz M. Low incidence of myocardial recovery after left ventricular assist device implantation in patients with chronic heart failure. Circulation. 1998; 98: 2383–2389.
Yacoub MH. A novel strategy to maximise the efficacy of LVADs as a bridge to recovery. Eur Heart J. 2001; 22: 534–540.
Yacoub MH, Birks EJ, Tansley P, Henein MY, Bowles CT. Bridge to recovery: the Harefield approach. J Congest Heart Failure Circ Support. 2001; 2: 27–30.
Fatkin D, MacRae C, Sasaki T, Wolff MR, Porcu M, Frenneaux M, Atherton J, Vidaillet HJ, Spudich S, De Girolami U, Seidman JG, Seidman C, Muntoni F, Muehle G, Johnson W, McDonagh B. Missense mutations in the rod domain of the lamin A/C gene as causes of dilated cardiomyopathy and conduction-system disease. N Engl J Med. 1999; 341: 1715–1724.
Fatkin D, Graham RM. Molecular mechanisms of inherited cardiomyopathies. Physical Rev. 2002; 82: 945–980.
Heling A, Zimmermann R, Kostin S, Maeno Y, Hein S, Devaux B, Bauer E, Klovekorn WP, Schlepper M, Schaper W, Schaper J. Increased Expression of cytoskeletal linkage and extracellular proteins in failing human myocardium. Circ Res. 2000; 86: 846–853.
Huang X, Pan W, Park S, Han X, Miller LW, Hall J. Modeling the relationship between LVAD support time and gene expression changes in the human heart by penalized partial least squares. Bioinformatics. 2004; 20: 888–894.
Chen Y, Park S, Li Y, Missov E, Hou M, Han X, Hall JL, Miller LW, Bache RJ. Alterations of gene expression in failing myocardium following left ventricular assist device support. Physiol Genomics. 2003; 14: 251–260.
Hall JL, Grindle S, Han X, Fermin D, Park S, Chen Y, Bache RJ, Mariash A, Guan Z, Ormaza S, Thompson J, Graziano J, de Sam Lazaro SE, Pan S, Simari RD, Miller LW. Genomic profiling of the human heart before and after mechanical support with a ventricular assist device reveals alterations in vascular signaling networks. Physiol Genomics. 2004; 17: 283–291.
Birks EJ, Latif N, Owen V, Bowles C, Felkin LE, Mullen AJ, Khaghani A, Barton PJ, Polak JM, Pepper JR, Banner NR, Yacoub MH. Quantitative myocardial cytokine expression and activation of the apoptotic pathway in patients who require left ventricular assist devices. Circulation. 2001; 104 (suppl I): I-233–I-240.
Fisher DZ, Chaudhary N, Blobel G. cDNA sequencing of nuclear lamins A and C reveals primary and secondary structural homology to intermediate filament proteins. Proc Natl Acad Sci U S A. 1986; 83: 6450–6454.
Johnson NA, Sengupta S, Saidi SA, Lessan K, Charnock-Jones SD, Scott L, Stephens R, Freeman TC, Tom BDM, Harris TM, Denyer G, Sundaram M, Sasisekharan R, Smith SK. Endothelial cells preparing to die by apoptosis initiate a programme of transcription and glycome regulation. FASEB J. 2004; 18: 188–190.
Simpson DG, Majeski M, Borg TK, Terracio L. Regulation of cardiac myocyte protein turnover and myofibrillar structure in vitro by specific directions of stretch. Circ Res. 1998; 85: e59–e69.
Ross RS, Pham C, Shai SY, Goldhaber JI, Fenczik C, Glembotski CC, Ginsberg MH, Loftus JC. β1-Integrins participate in the hypertrophic response of rat ventricular myocytes. Circ Res. 1998; 82: 1160–1172.
De Jonge N, Van Wichen DF, Schipper M, Lahpor JR, Gmelig-Myeling FHJ, Robles de Medina EO, de Weger RA. Left ventricular assist device in end-stage heart failure: persistence of structural myocyte damage after unloading: an immunohistochemical analysis of the contractile myofilaments. J Am Coll Cardiol. 2002; 39: 963–969.