Left Ventricular Eccentric Remodeling and Matrix Loss Are Mediated by Bradykinin and Precede Cardiomyocyte Elongation in Rats With Volume Overload

Animal preparation

Abdominal ACF was performed in male Sprague-Dawley rats (200 to 250 g) as previously described (4). Age-matched sham- and ACF-operated rats were generated for echocardiographic and hemodynamic study at 12 h; 2 and 5 days; and 4, 8, and 15 weeks, after which time they were killed and tissues were collected for morphometry, immunohistochemistry, and protein analysis (n values given in Table 1).A separate group of rats was killed at similar time points for isolated myocyte studies (n values given in Table 2).In a third group of animals, the BK receptor type 2 (BK₂R) antagonist Hoe 140 (0.5 mg/kg/day, subcutaneous, Sigma/RBI, Natick, Massachusetts) was started after ACF or sham surgery, and animals were killed at 2 days (n values given in Table 3).A subset of rats (n = 8) was treated with angiotensin II (1 μg/kg/min, osmotic mini-pump, Bachem Bioscience Inc., King of Prussia, Pennsylvania) for 2 days after ACF to generate a blood pressure increase similar to that with ACF + Hoe 140. In a final group, the ACE inhibitor was started in the drinking water (ramipril at 10 mg/kg/day, Hoechst, Frankfurt, Germany) after induction of ACF or sham, and animals were killed at 4 weeks. Hemodynamic and echocardiographic measurements were made on all drug-treated animals. This protocol was approved by the Animal Resource Program at the University of Alabama at Birmingham.

Hemodynamic and echocardiographic measurements

Rats were anesthetized with ketamine (80 mg/kg intraperitoneally) and xylazine (10 mg/kg intraperitoneally). High-fidelity LV pressures (SPR-249A Millar Mikro-Tip catheter transducer, Millar Instruments, Houston, Texas) were recorded concurrently with echocardiography (Agilent Sonos 5500, Philips, Bothell, Washington). The LV wall stress and function were calculated as described previously (8).

Collagen analysis

Hearts were immersion fixed in 10% buffered formalin, embedded in paraffin, sectioned at 3-μm thickness, and stained with picric acid Sirius red F3BA. Interstitial collagen volume percent was evaluated with light microscopy at high power (40× objective, 1,220× video screen magnification) in 30 to 40 randomly selected fields. Collagen fiber thickness was measured using laser-confocal images (40× objective) of 15-μm-thick sections and fluorescein isothiocyanate-labeled phalloidin. Image analysis was performed using a line orthogonal grid overlay at 250 × 250 μm for each field (approximately 5 fields per animal). Average width was calculated at each grid intersection point.

Adult rat LV myocyte isolation and measurement of single-cell contractility

Left ventricular myocytes were isolated by cardiac retrograde aortic perfusion as previously described (9), with minor modifications. Briefly, hearts from anesthetized rats (phenobarbital 50 mg/kg intraperitoneally) were rapidly dissected and perfused using a Langendorf apparatus with Kreb’s buffer (NaCl 118 mM, KCl 4.7 mM, NaHCO₃25 mM, Ca Cl₂1.8 mM, Mg₂SO₄1.2 mM, KH₂PO₄1.2 mM, glucose 11), followed by Ca²⁺-free Kreb’s buffer, and Ca²⁺free-Kreb’s buffer containing 100 U/ml type II collagenase (Worthington Biochemical, Lakewood, New Jersey). Left ventricular tissue was isolated and further digested in collagenase/body surface area (BSA) Kreb’s buffer containing 25 μM CaCl₂. The resulting cells were resuspended in perfusion buffer with 3% BSA, in which the concentration of CaCl₂was stepwise increased to 1 mM. Rod-shaped cells were resuspended in Medium199 (Sigma, St. Louis, Missouri) supplemented with 5 mM creatine, 2 mM L-carnitine, and 5 mM taurine and plated on laminin-coated glass coverslips.

Following two 1-h incubation periods, with medium changed between hours, coverslips were placed in a perfusion chamber with internal stimulating platinum electrodes (Warner Instruments, Hamden, Connecticut) attached to a Grass stimulator, and imaged using an inverted microscope (Olympus America Inc, Melville, New York) and CCD camera (Philips, Andover, Massachusetts). A rod-shaped cell without blebbing or spontaneous contractions was chosen for imaging. The chamber was perfused with Tyrode basic salt solution and paced (3-ms duration) for 2 min at a starting frequency of 0.5 Hz. The stimulating frequency was stepwise increased from 0.5 to 3 Hz. The contractility of each cell was recorded with a video edge detection system (Crescent Electronics, Sandy, Utah) and analyzed using Digi-Med System Integrator Model 200/1 and Cell Data Analysis software program (MicroMed Technology, Inc., Houston, Texas). At each frequency, data were analyzed after 1 min of pacing.

MMP analysis

Left ventricular tissue was homogenized in lysis buffer (0.2% Triton X-100, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 0.5 mM PMSF, 500 μM Na₃VO₄) and centrifuged at 14,000 rpm for 10 min at 4°C. Protein concentrations were assessed with a bicinchonic acid assay. Western blot analysis was performed with monoclonal anti-rat MMP-13, monoclonal anti-rat tissue inhibitor of MMP-1 (TIMP-1), and rabbit polyclonal anti-rat TIMP-4 (all antibodies Chemicon, Temecula, California). Membranes were washed and treated with horseradish peroxidase-conjugated secondary antibodies (goat anti-mouse or anti-rabbit immunoglobulin G). Immunoreactive bands were visualized using enhanced chemiluminescence reagents exposed to Hyperfilm at the linear range of film density. Films were scanned, and densitometric analysis was performed with National Institutes of Health image software.

Statistical analyses

All statistical analyses were done using SAS software (SAS Institute Inc., Cary, North Carolina). Generalized linear models (two-factor with interaction and one-factor) were used to analyze all variables. We used the Kolmogorov-Smirnov test and the Levene test on the residuals to check normality and constancy of variance, respectively. If all p values for testing these assumptions were >0.05, we considered the model a good fit. Otherwise, we did a transformation (natural logarithm or ranks) and then chose the appropriate transformation that would resolve the assumption(s) violated. Problems of non-normality and/or nonconstant variance of the error terms were observed in a majority of the variables. If transformations worked, p values for these goodness-of-fit tests were all >0.10. Transformations applied to the variables are indicated in the tables and figures.

In a few variables, neither log nor rank transformation remedied the goodness-of-fit problem observed. In these cases, we used the Kruskal-Wallis test for 1-factor designs and variance component analyses (via SAS PROC MIXED) to accommodate the unequal variances across the factor combinations for 2-factor designs. All tests were performed using a 5% significance level. The Bonferroni method was used in the pair-wise comparisons for each variable to ensure an overall 5% significance level. Appropriate decision rules based on the Bonferroni adjustment are indicated in the tables and figures.

LV remodeling in ACF

Twelve hours after the induction of volume overload, LVEDD had increased (p < 0.05) by 20%, whereas LVEDD/wall thickness was increased by 30% at 2 days after ACF (Figs. 1Aand 1B). There was no increase in cardiomyocyte length at either time (Fig. 1C). Both LVEDD and LVEDD/wall thickness were increased by 40% at 4 to 15 weeks after ACF (Figs. 1A and 1B). Mean LV wall thickness decreased (p < 0.05) at 15 weeks after ACF (Fig. 1D). Cardiomyocyte length increased by 9% and 18% above shams only at the 8- and 15-week post-ACF time points, respectively (Fig. 1C), whereas LV weight, which had increased by 40% at 4 weeks after ACF, did not increase further after this point (Fig. 1E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

LV Remodeling in Age-Matched Sham and ACF Rats

Left ventricular (LV) remodeling in aortocaval fistula (ACF) (open circles)at 12 h; 2 and 5 days; and 4, 8, and 15 weeks compared with age-matched shams (closed circles). Left ventricular end-diastolic dimension (LVEDD) (A), LVEDD/wall thickness (B), isolated cardiomyocyte length (C), left ventricular end-diastolic (LVED) wall thickness (D), and LV wet weight (E). Alternating black and white boxeseach represent one 24-h period. Data are presented as mean ± SEM, n values given in Tables 1 and 2except: 4-week sham LVEDD, LVEDD/wall thickness, LVED wall thickness (n = 9); 4-week ACF echocardiographic parameters and LV wet weight (n = 9); 8-week sham echocardiographic parameters and LV wet weight (n = 8); 15-week sham LV wet weight (n = 12); and 15-week ACF LV wet weight (n = 9). *p < 0.008 (= 0.05/6) for sham versus ACF at each time of the 6 time points except for isolated cardiomyocyte length, where p < 0.01 (= 0.05/5) for each of the 5 time points. Analyses were based on ranks of all the variables in this figure.

Acutely, there was a 75% decrease in collagen volume percent 12 h after ACF that persisted at day 2 (Figs. 2Ato 2E). Collagen volume percent returned to sham values by 5 days after ACF (Fig. 2D); however, the average diameter of collagen fibers at 5 days was increased compared with sham (Fig. 2F). Chronically, collagen volume percent was decreased 30% in ACF rats at 4, 8, and 15 weeks (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Representative Examples of LV Interstitial Collagen in Control and ACF Rats

Representative examples of LV interstitial collagen in control (CTL) (n = 8) (A)and ACF at 12 h (n = 5) (B), 2 days (n = 8) (C), and 5 days (n = 4) (D). Mean collagen volume percent in 2-day CTL vs. ACF at 12 h, 2 days, and 5 days, where *p < 0.0167 (= 0.05/3) using variance component analysis (E). Mean collagen fiber width at 5 days of ACF versus CTL (F)where *p < 0.05 using ttest. Data are presented as mean ± SEM. Abbreviations as in Figure 1.

Indexes of systolic function in ACF

Mean arterial pressure was decreased and heart rate increased both 12 h and 5 days after induction of ACF, but did not differ from shams at 4, 8, and 15 weeks after ACF (Table 1). Left ventricular end-systolic dimension (LVESD) did not differ from shams at 12 h and 2 and 5 days after ACF. However, at 4 weeks after ACF, LVESD increased by 40% and remained elevated in 8- and 15-week post-ACF rats despite the fact that left ventricular end-systolic (LVES) pressure did not differ from sham (Figs. 3Aand 3B). Left ventricular fractional shortening and velocity of circumferential shortening (VCF_r) were not different from sham values until 15 weeks after ACF, when there was a decrease in VCF_rconcomitant with an increase in LVES stress (Figs. 3C to 3E). This decrease in VCF_roccurred in the face of normal cardiomyocyte fractional shortening and velocity of shortening (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Indexes of LV Systolic Function in Age-Matched Sham and ACF Rats

Left ventricular (LV) function in ACF (open circles)at 12 h; 2 and 5 days; and 4, 8, and 15 weeks compared with age-matched shams (closed circles). Left ventricular end-systolic (LVES) pressure (A), left ventricular end-systolic dimension (LVESD) (B), LV fractional shortening (FS) (C), LVES wall stress (σ) (D), and LV velocity of circumferential shortening (VCF_r) (E). Alternating black and white boxeseach represent one 24-h period. Data are presented as mean ± SEM, n values given in Table 1except: 4-week sham LVESD, LV fractional shortening, and LV VCF_r(n = 9); 8-week sham LVESD, LV fractional shortening, and LV VCF_r(n = 8). *p < 0.008 (= 0.05/6) sham versus ACF at each of the 6 time points. Analyses for left ventricular end-systolic pressure (LVESP), LVESD, and LVVCFr were based on their ranks, whereas analyses for LV fractional shortening and LVES σ were based on their natural logarithms.

Indexes of diastolic function in ACF

Left ventricular end-diastolic (LVED) pressure and LVED wall stress (Figs. 4Aand 4B) were increased at 2 days and remained elevated throughout the time course, whereas lung wet weight (Fig. 4C) was increased at 4 to 15 weeks and right ventricular weight (Fig. 4E) was increased from 5 days after ACF onward. Isolated cardiomyocyte relaxation velocity and time to 50% relaxation did not differ in ACF and age-matched shams across all time points, except at 5 days after ACF (Table 2), whereas LV − rate of left ventricular pressure decline (dP/dt_max) in ACF did not differ from shams at all time points (Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Indexes of Diastolic LV Remodeling in Age-Matched Sham and ACF Rats

Indexes of diastolic LV remodeling in ACF (open circles)at 12 h; 2 and 5 days; and 4, 8, and 15 weeks compared with age-matched shams (closed circles). LVED pressure (A), LVED wall stress (σ) (B), lung wet weight (C), and right ventricular (RV) wet weight (D). Alternating black and white boxeseach represent one 24-h period. Data are presented as mean ± SEM, n values given in Table 1except: 4-week sham LVED pressure and LVED σ (n = 5); 4-week sham lung wet weight and RV wet weight (n = 8); 4-week ACF lung wet weight and RV wet weight (n = 9); 8-week sham lung wet weight and RV wet weight (n = 8); 15-week sham lung wet weight and RV wet weight (n = 12); and 15-week ACF lung wet weight and RV wet weight (n = 9). *p < 0.008 (= 0.05/6) sham versus ACF at each of the 6 time points. Analyses were based on ranks of all the variables in this figure. Abbreviations as in Figure 1.

Hoe 140 and LV remodeling and function in 2-day ACF rats

Hoe 140 prevented the decrease in LV interstitial collagen (Fig. 5A)and the increase in LVEDD and LVEDD/wall thickness at 2 days after ACF (Table 3). There was an increase in LVES pressure in both sham + Hoe 140 and ACF + Hoe 140 rats (Table 3); however, LVES wall stress did not differ among all groups (data not shown). Left ventricular fractional shortening was increased in sham + Hoe 140 and ACF + Hoe 140 despite an increased left ventricular end-systolic pressure compared with untreated sham and ACF cases (Table 3). Left ventricular weight/body weight did not increase in either ACF group; total heart weight/body weight increased in ACF versus age-matched shams. In contrast, total heart weight/body weight did not increase in ACF + Hoe 140 versus sham + Hoe 140 and sham rats (Table 3). At 2 days after ACF, MMP-13 (rodent collagenase) was increased, whereas TIMP-1 and -4 protein levels were not altered. Hoe 140 prevented the increase in MMP-13 but had no effect on TIMP-1 and -4 levels (Figs. 5B to 5D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Indexes of Collagen Homeostasis in Hoe 140-Treated Sham and ACF Rats

Collagen volume percent (A), matrix metalloproteinase (MMP)-13 (B), tissue inhibitor of matrix metalloproteinase (TIMP)-1 (C), and TIMP-4 (D)levels at 2 days in sham (SHM) and ACF rats with and without Hoe 140 treatment. Data are presented as mean ± SEM, n values given in Table 3. *p < 0.008 (= 0.05/6) versus SHM; †p < 0.008 versus ACF. Collagen volume and MMP-13 were analyzed via variance component, whereas TIMP-1 and -4 were analyzed using a Kruskal-Wallis test. Abbreviations as in Figure 1.

To assess whether the effects of Hoe 140 were through a direct mechanism on the heart or an indirect mechanism through increased blood pressure, we infused angiotensin II and achieved a similar increase in MAP as seen with Hoe 140 infusion (Table 4).In comparison with ACF alone, angiotensin II partially prevented the decrease in collagen and the increase in LVEDD/wall thickness. However, there was an increase in cardiac hypertrophy in ACF + angiotensin II when compared with the ACF + Hoe 140 (Table 4).

Effect of ramipril on LV remodeling and function in 4-week ACF rats

Ramipril decreased LVES pressure and heart weight/body weight in 4-week post-ACF rats. However, the decrease in LVES pressure was not accompanied by an improvement in fractional shortening, LV dilation, and LVEDD/wall thickness (Table 5).Ramipril was associated with a further decrease in LV interstitial collagen compared with untreated ACF (Fig. 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Volume Percent of Collagen in Ramipril-Treated Sham and ACF Rats

Collagen volume percent in 4-week sham (SHM) and aortocaval fistula (ACF), 4-week sham + ramipril, and 4-week ACF + ramipril. Data are presented as mean ± SEM, n values given in Table 5. *p < 0.008 (= 0.05/6) versus sham, †p < 0.008 versus ACF, ‡p < 0.008 versus sham + ramipril. Variance components analysis was used.

BK-mediated collagen dissolution exacerbates acute LV remodeling

Important sequential changes in LV chamber and cardiomyocyte remodeling with ACF are shown in Figure 1. Within 2 days, there is an increase in LVEDD and LVEDD/wall thickness with no change in cardiomyocyte length. At the same time, there is a marked decrease in interstitial collagen, showing that cardiomyocyte elongation is not the earliest step in LV eccentric remodeling in response to volume overload. This rapid decrease in interstitial collagen is likely to be multifactorial. Simple stretch of cardiac fibroblasts in vitro increases collagenase activity (10) and membrane type 1-MMP expression (11). In addition, we and others have found an increase in mast cells and BK in this early phase of ACF (4,12). Mast cell secretory products such as trypsin, chymase, stromelysin (MMP-3), and cytokines, such as tumor necrosis factor-α, are potent in vitro activators of MMPs (13–16). Inhibition of ACE in ACF rats significantly increased interstitial fluid BK but did not attenuate increases in mast cell density, which was blocked by the addition of BK₂R blockade, supporting the role of BK as an inflammatory mediator (4).

In addition to its effects on inflammatory cell recruitment, BK decreases collagen I and III mRNA production in cardiac fibroblasts (5), modulates MMP activation through the plasminogen activator system (17), and increases MMP-2 (6) and MMP-9 (7) expression. Two days after the induction of ACF, MMP-13 levels are increased in the absence of changes in TIMP-1 and -4 levels (Fig. 5) and collagen I and III expression (data not shown), suggesting that the decrease in interstitial collagen is caused by increased degradation, but not decreased synthesis. This acute loss of collagen in ACF stands in stark contrast to a marked increase in collagen synthesis that has been reported after 2 days of acute pressure overload in the rat (18). Thus, an acute inflammatory response [increased BK and mast cell density (4)] coupled with a stretch stimulus, which occurs in the absence of pressure overload, may induce a rapid loss of the collagen matrix.

Blockade of BK₂R prevents an increase in MMP-13 and a decrease of interstitial collagen, and normalizes LVEDD/wall thickness at 2 days of ACF stress. Further, LVES pressure increases as LVESD trends down 25% vs. untreated ACF rats (Table 3), suggesting that BK₂R blockade improves LV function. This functional improvement could be a consequence of more efficient transmission of forces between adjacent myocytes and myofibers because of a preservation of the matrix. Interestingly, antagonism of BK₂R in ACF animals is associated with a decrease in cardiac hypertrophy despite the increase in MAP. To test whether the local cardiac effects of BK₂R blockade are important, or whether functional and remodeling improvements during ACF are attributable to the increase in blood pressure caused by BK₂R blockade, we examined the effect of a hypertensive challenge that was similar in magnitude to that seen with Hoe 140 treatment. Using angiotensin II infusion, we achieved an increase in blood pressure at 2 days after ACF equivalent to that seen with Hoe 140 treatment. Similar to Hoe 140, angiotensin II increased collagen matrix and normalized LVEDD/wall thickness compared with ACF alone; unlike Hoe 140, it also increased cardiac hypertrophy. Because BK is upregulated acutely in the LV during ACF, it is tempting to speculate that its blockade in the heart prevents the loss of collagen through a direct mechanism rather than through an indirect effect on blood pressure.

At 5 days after ACF, the collagen volume percent returns to normal; however, this rebound is abnormal because there are fewer collagen fibers and these fibers are thicker than normal (Fig. 2). This rebound in collagen occurs in the absence of an increase in MAP, and may also be a response to a local increase in cardiac interstitial angiotensin II (4). The change in collagen structure could be a temporary adaptive response because it is associated with a cessation of further LV dilation at 5 days. However, at 4 weeks after ACF there is a subsequent decrease in extracellular matrix and a marked increase in LVEDD/wall thickness.

Adverse LV remodeling and dysfunction in chronic volume overload

Another important finding of this study is that LV systolic and diastolic dysfunction (4 to 15 weeks) is not related to cardiomyocyte dysfunction. There is evidence of LV systolic dysfunction as early as the 4-week stage of ACF (Fig. 3), manifested by an increase in LVESD, whereas LV fractional shortening does not increase despite an increase in preload and normal LVES wall stress. At 4 weeks after ACF, LVESD is increased in the presence of normal LVES pressure, resulting in a significant increase in LVES wall stress at 15-week ACF. It is also important to note that there is no incremental increase in LV hypertrophy 4 weeks after induction of ACF, despite an increase in LVES wall stress at 15 weeks. In addition, ACF LV wall thickness is decreased and cardiomyocytes are elongated at 15 weeks, accentuating adverse eccentric LV remodeling and contributing to a decrease in LV VCF_r, all in the face of normal cardiomyocyte function. At a similar ACF time point, Wang et al. (19) found enhanced LV β₁-receptor density and increased basal activity. Thus, the disparity between isolated myocyte function in vitro and LV chamber function in vivo could be caused by loss of collagen structural support.

Left ventricular end-diastolic pressure and wall stress are elevated as early as 2 days after ACF and throughout the entire time course (Fig. 4). Lung wet-weight is also increased, indicative of decreased LV compliance despite the decrease in interstitial collagen and normal indexes of cardiomyocyte relaxation throughout 15 weeks after ACF (Table 2). Previous studies in ACF rats at 6 weeks showed an increase in collagen cross-linkage and no change in collagen content, resulting in a stiffer LV chamber (20). Thus, a simple assessment of collagen content without regard to collagen cross-linking or subtype distribution, especially in volume overload, could obscure changes in chamber compliance. Our results suggest that the progressive LV remodeling and systolic dysfunction, which is disproportionate to cardiomyocyte remodeling, results from dissolution of the collagen matrix.

Chronic ACE inhibition does not attenuate LV remodeling in volume overload

We found that ACE inhibition increased collagen loss at 4 weeks after ACF (Fig. 6) and did not attenuate adverse LV chamber remodeling as summarized in Table 5. With ACF + ACE inhibitor treatment, LVEDD/wall thickness and LVED wall stress remained elevated and LVESD increased relative to ACF-untreated rats, even though there was a decrease in LVES pressure and cardiac hypertrophy. The failure to improve LV remodeling and function may be caused by an ACE inhibitor-induced loss of supporting collagen structure that results in further cardiomyocyte elongation. Our BK₂R blockade studies suggest that ACE inhibitor-induced increase in BK contributes to the acute adverse remodeling seen in volume overload after ACF. However, we do not show the role of long-term BK₂R activation in the heart on the volume overloaded LV, and therefore, the mechanism of the long-term deleterious effects of chronic ACE inhibition remains uncertain. We have shown a similar loss of collagen and MMP activity in early and late phases of mitral regurgitation in the dog (21). Indeed, ACE inhibitor or angiotensin II receptor blockade failed to attenuate LV remodeling in dogs with mitral regurgitation (8,22), and we have shown that such therapy results in further cardiomyocyte elongation (8). Finally, ACE inhibitor or angiotensin II receptor blockade have not been shown to be beneficial in patients with pure volume overload due to valvular regurgitation (2).

Summary

Here we show that an increase in LVEDD/wall thickness, previously thought to be caused by myocyte elongation, may have less to do with myocyte length and more to do with matrix changes that affect the collagen scaffolding connecting individual myocytes and collections of myocytes that comprise the laminar structure of the heart (23). We also show that the BK-mediated change in the extracellular matrix is an early event that sets in motion adverse LV remodeling and function. This study has important implications for the BK-enhancing effects of ACE inhibitor therapy in pure volume overload states. The role of matrix degradation in volume overload challenges current concepts regarding the pathophysiology of eccentric LV remodeling and explains why ACE inhibitors are effective at reversing LV remodeling during pressure overload but are ineffective in treating volume overload.