Heart and Circulatory Physiology

Accelerated LV remodeling after myocardial infarction in TIMP-1-deficient mice: effects of exogenous MMP inhibition

John S. Ikonomidis, Jennifer W. Hendrick, Andrea M. Parkhurst, Amanda R. Herron, Patricia G. Escobar, Kathryn B. Dowdy, Robert E. Stroud, Elizabeth Hapke, Michael R. Zile, Francis G. Spinale


Alterations in matrix metalloproteinases (MMPs) and tissue inhibitors of MMPs (TIMPs) have been implicated in adverse left ventricular (LV) remodeling after myocardial infarction (MI). However, the direct mechanistic role of TIMPs in the post-MI remodeling process has not been completely established. The goal of this project was to define the effects of altering endogenous MMP inhibitory control through combined genetic and pharmacological approaches on post-MI remodeling in mice. This study examined the effects of MMP inhibition (MMPi) with PD-166793 (30 mg·kg−1·day−1) on LV geometry and function (conductance volumetry) after MI in wild-type (WT) mice and mice deficient in the TIMP-1 gene [TIMP-1 knockout (TIMP1-KO)]. At 3 days after MI (coronary ligation), mice were randomized into four groups: WT-MI/MMPi (n = 10), TIMP1-KO-MI/MMPi (n = 10), WT-MI (n = 22), and TIMP1-KO-MI (n = 23). LV end-diastolic volume (EDV) and ejection fraction were determined 14 days after MI. Age-matched WT (n = 20) and TIMP1-KO (n = 28) mice served as reference controls. LVEDV was similar under control conditions in WT and TIMP1-KO mice (36 ± 2 and 40 ± 2 μl, respectively) but was greater in TIMP1-KO-MI than in WT-MI mice (48 ± 2 vs. 61 ± 5 μl, P < 0.05). LVEDV was reduced from MI-only values in WT-MI/MMPi and TIMP1-KO-MI/MMPi mice (42 ± 2 and 36 ± 2 μl, respectively, P < 0.05) but was reduced to the greatest degree in TIMP1-KO mice (P < 0.05). LV ejection fraction was reduced in both groups after MI and increased in TIMP1-KO-MI/MMPi, but not in WT-MI/MMPi, mice. These unique results demonstrated that myocardial TIMP-1 plays a regulatory role in post-MI remodeling and that the accelerated myocardial remodeling induced by TIMP-1 gene deletion can be pharmacologically “rescued” by MMP inhibition. These results define the importance of local endogenous control of MMP activity with respect to regulating LV structure and function after MI.

  • transgenic model
  • heart failure
  • myocardial hypertrophy

myocardial infarction (MI) evokes changes within the architecture of the left ventricular (LV) wall leading to chamber dilation. The matrix metalloproteinases (MMPs) are a family of endopeptidases known to degrade all components of the myocardial extracellular matrix. In experimental models of MI, gene deletion of certain MMPs or pharmacological inhibition of MMP activity directly modifies the LV remodeling process after MI (8, 9, 11, 18, 19, 24, 27). Thus a cause-effect relation between induction of MMP activity and LV remodeling after MI is beginning to emerge. However, little is known about the putative role of the endogenous tissue inhibitors of the MMPs (TIMPs) in the post-MI remodeling process. Of the four known TIMPs, TIMP-1 has been the best characterized (16, 40). TIMP-1 binds to active MMPs in a 1:1 stoichiometric relation and, thereby, provides a localized control point for MMP activational states. In end-stage cardiomyopathic disease, an increase in the abundance of MMP species is not paralleled by an increase in myocardial TIMP levels (33). These observations would suggest a defect in local MMP inhibitory control in the myocardial remodeling process. In an initial report in which a TIMP-1-deficient mouse construct was used, LV remodeling occurred as a function of age (32). Moreover, a recent study suggested that post-MI remodeling may be accelerated in TIMP-1-deficient mice (8). However, it remains to be established whether TIMP-1 deletion directly results in enhanced myocardial MMP activity after MI. To address this issue, the present study examined LV geometry and function in TIMP-1-deficient mice after surgically induced MI and determined whether pharmacological MMP inhibition would directly modify the remodeling process in this transgenic model.


Animal model and surgical induction of MI.

The mice used in these studies were adult inbred 129 Sv mice in which a stable line of TIMP-1-deficient mice [TIMP-1 gene knockout (TIMP1-KO)] was created through the use of a replacement vector carrying a stop codon containing oligonucleotide and the neo resistance cassette with the 5′ end of the TIMP-1 coding frame (20). The original breeding pairs used to develop the mice for this study were a kind gift from Dr. Paul J. Soloway (Roswell Park, Buffalo, NY). For this study, age- and gender-matched wild-type (WT) mice and TIMP1-KO mice underwent surgical induction of MI or served as reference controls. Under isoflurane anesthesia (3% in oxygen), mice were placed in a supine position, and the trachea was intubated with a 1.1-mm steel intubation tube. The mice were then placed on a rodent ventilator and ventilated at a tidal volume of 1 ml and 200 cycles/min. Sterile technique was used to perform a left thoracotomy in the fourth intercostal space. After the pericardium was opened, the left anterior descending artery was ligated near its origin with 6-0 prolene and an atraumatic needle (model K801, Ethicon). The incisions were then closed. After extubation, the mice were given buprenorphine (2–2.5 mg/kg ip), oxygen was administered by mask, and the mice were covered with a warming blanket. The surgical mortality rate (due to sudden death or requirement for euthanasia because of deteriorating status) was 15%. At 3 days after operation, a transthoracic echocardiogram was obtained to confirm the presence of an MI by clear defects in LV posterior wall motion. For these studies, the mice were anesthetized with isoflurane (1.5–2% in oxygen) and maintained at ambient body temperature with a heating blanket. Heart rate was determined from a surface ECG, and our results confirmed that this regimen maintains an ambient heart rate of 400–500 beats/min and that the mice remain normothermic throughout the procedure. Two-dimensional targeted M-mode echocardiographic recordings were obtained using a high-band linear 15.7-MHz transducer (Sonos 5500, Hewlett-Packard/Agilent Technology). From these postoperative screening echocardiographic studies, any mice that did not display a clear LV wall motion abnormality were excluded from the study. The exclusion rate was 5%. All mice were then randomized to the experimental protocol described below. Isoflurane was chosen for the anesthesia to be utilized during the MI surgery as well as for LV function measurements because of the stable hemodynamic profile achieved with this inhalation anesthetic (20). However, it must be recognized that isoflurane can cause coronary vasodilation (28). Equivalent isoflurane concentrations were utilized for the LV function measurements to allow for comparisons between groups. All animals were treated and cared for in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Research Council, Washington, DC, 1996).

Experimental design.

At 3 days after MI (coronary ligation), mice were assigned to the following groups: WT-MI (n = 22), TIMP1-KO-MI (n = 23), WT-MI with MMP inhibition (WT-MI/MMPi, n = 10), and TIMP1-KO-MI/MMPi (n = 10). The first 20 mice were subjected to MI only to generate a large database for LV function and geometry. The mice were then randomized to each treatment group. Age-matched WT (n = 20) and TIMP1-KO (n = 28) mice served as reference controls. MMP inhibition with PD-166793 (30 mg·kg−1·day−1) was instituted at randomization on postoperative day 3 and continued until postoperative day 14. The dose of MMP inhibitor used in this study achieved a steady-state plasma level of 8.5 ± 2.2 mg/ml, which has been demonstrated previously in ex vivo studies to achieve a significant broad-spectrum MMP inhibitory effect (31). Although this compound effectively inhibits MMP activity, it does not affect other metalloproteinases such as angiotensin-converting enzyme or tumor necrosis factor-α-converting enzyme (31). With respect to MMP inhibitory specificity, the inhibitory potency of PD-166793 for the MMP catalytic domain [effective inhibitory concentration (EC50)] ranges from 7.9 μmol/l for MMP-9 to 0.008 μmol/l for MMP-13 (31). Thus the plasma levels achieved in the present study significantly exceeded the EC50 for all major MMP types. In a previous study, this MMP inhibitor exhibited a high level of myocardial penetration and significantly inhibited myocardial MMP activity (27). The rationale for initiating MMP inhibition 3 days after MI was twofold: 1) to allow for screening and confirmation of a significant MI in this murine model and 2) to examine post-MI remodeling and not to interfere with the acute phase of the myocardial wound-healing process. In terminal studies 14 days after MI, LV function and geometry were studied by conductance volumetry and tissue was collected for histomorphometry and biochemistry.

LV conductance volumetry.

The mice were anesthetized and prepared as described above. The mice were positioned on a feedback temperature-controlled operating table (Vestavia Scientific, Birmingham, AL). Under microscopic guidance (model OPMI, Zeiss), the right carotid artery was exposed. A precalibrated four-electrode pressure-sensor catheter (1.4-F, model SPR-839, Millar Instruments, Houston, TX) was positioned in the LV. The catheter was interfaced to a pressure-conductance unit (ARIA, MPCU-200, Millar), in which electrical excitation was performed under digital control (DAQ, PV Analysis Software, Millar). The continuous digital pressure and conductance signals were integrated with an ECG signal (PowerLab, AD Instruments) and displayed in real time using dual-display flat-screen monitors (Sony Electronics). This allowed for optimal placement of the LV catheter with respect to the LV conductance signal. An algorithm described previously (13, 15) was used to convert the LV conductance signal to an LV volume. One of the important factors in this algorithm is the determination of a parallel conductance volume coefficient (Vp). In the present study, Vp was routinely computed by using an isotonic saline injection in terminal studies in mice (8). Briefly, a 5-μl bolus of 15% hypertonic saline was injected into the external jugular vein, and the LV pressure-conductance signals were continuously acquired. The saline bolus caused a significant LV volume shift without a marked change in LV systolic pressure. The isochronal LV systolic and diastolic volumes were plotted along with the line of identity. From the intersection of these two lines, we computed a mean value of 26 μl for Vp, which is consistent with values reported previously (8, 13). This factor was used in the software algorithms (PVAN, Millar) to compute absolute LV volumes. This saline calibration was repeated on a weekly basis in WT and TIMP1-KO mice to ensure that the computed Vp remained within 10% of predicted values. In addition, weekly ex vivo cuvette calibrations were performed for WT and TIMP1-KO mice with use of whole blood samples from WT and transgenic mice to establish the linear relation between volumes and conductance (Fig. 1). Steady-state LV pressures and conductance volumetry were determined with the ventilator suspended and the signals digitized for ≥12 consecutive cardiac cycles. With continuous recording of the LV pressure-conductance volume signal, gentle digital pressure was placed on the abdomen to reduce venous return and then released. An adequate reduction in venous return was defined as an ∼50% reduction in LV systolic pressure. This resulted in a family of LV pressure-volume loops in which definable points of the LV end-systolic pressure-volume relation could be determined in non-MI and MI animals (Fig. 1). The isochronal LV pressure-volume points were used to compute the slope of the preload recruitable stroke work (PRSW) relation (13, 15). To confirm that the changes in LV volumes induced by altering LV preload did not significantly affect the conductance volumetric measurements, simultaneous transthoracic echocardiography and conductance catheter measurements were performed in a WT mouse preparation. A high degree of concordance between these two methods of measuring LV end-diastolic volume suggests that the preload-induced changes in LV volumes did not artifactually influence the conductance volumetry readings (r = 0.98, P < 0.001).

Fig. 1.

Ex vivo calibration utilizing murine blood samples and conductance measurements in cuvettes of known volumes. A significant linear relation was obtained: y = 5x (r = 0.96, P < 0.05). RVU, relative volumetric units.

Myocardial sampling and histomorphometric and biochemical measurements.

At the completion of the hemodynamic studies, 0.5 ml of 0.1 mM cadmium chloride was injected into the LV to arrest the heart in diastole. After perfusion with an ice-cold phosphate buffer solution, the heart was excised and quickly weighed. The sections were fixed in a 4% formalin solution overnight and then embedded in paraffin. Sections (5 μm) were stained with hematoxylin and eosin for measurement of MI size and myocyte cross-sectional area using computer-assisted methods described previously (8, 27). Additional LV sections were stained with picrosirius red for fibrillar collagen, and the percent area of collagen within the remote and MI regions of the LV was computed (8, 27). To examine whether relative changes in MMP or TIMP profiles occurred with TIMP1-KO or MI, substrate zymography was performed to assess the relative content of the gelatinases MMP-2 and MMP-9. Immunoblotting was performed for MMP-3, MMP-13, and MT1-MMP (MMP-14), as well as for TIMP-2, TIMP-3, and TIMP-4, using methods described previously (27, 39). Briefly, for zymography, myocardial extracts (10 μg of total protein) were subjected to electrophoretic separation with a denatured collagen substrate (1 mg/ml type III gelatin; Sigma, St. Louis, MO). For immunoblotting, myocardial extracts (10 μg) were loaded onto 4–12% Bis-Tris gels and subjected to electrophoretic separation. The separated proteins were then transferred to a nitrocellulose membrane. After a blocking-and-washing step, the membranes were incubated for 1 h in antiserum (1:5,000 dilution) corresponding to MMP-3 (AB810, Chemicon), MMP-13 (AB8114, Chemicon), MT1-MMP (AB815, Chemicon), TIMP-2 (RP2T2, Triple Point Biologics), TIMP-3 (CL2T3, Cedarlane Laboratories), or TIMP-4 (AB816, Chemicon). Recombinant standards (Chemicon) were included in all immunoblots as a positive control. The relative degree of oxidative stress was determined in myocardial samples by measurement of the changes in protein carbonyl groups through derivatization of dinitrophenylhydrazine as well as by quantitation of superoxide dismutase (SOD) activity (1, 5, 10, 25, 38). Flash-frozen myocardial samples were homogenized in a 12% SDS solution, and an aliquot (5 μg) was derivatized with dinitrophenylhydrazine (Oxyblot Protein Oxidation Detection Kit, S7150, Chemicon) for 15 min, yielding a stable dinitrophenyl (DNP) product (5, 10). After neutralization, the myocardial extract was subjected to electrophoresis and immunoblotting as described above utilizing an antiserum to DNP (1:300 dilution; 90451, Chemicon). For the SOD activity assay, myocardial homogenates (1 μg) were incubated with a xanthine oxidase mixture (SOD assay; 706002, Cayman Chemical), and the colorimetric reaction was read at 450 nm and converted to units of SOD activity per milligram of myocardial protein.

Data analysis.

Indexes of LV function and geometry were compared between the groups initially using a multiway analysis of variance (ANOVA), in which the main treatment effects were the presence or absence of the TIMP-1 gene, the presence or absence of MI, and MMP inhibition treatment. If the ANOVA identified a significant treatment effect, then individual group mean comparisons were performed using Bonferroni’s pairwise corrections. For the morphometric data, the measurements of cross-sectional area and collagen area were first confirmed to conform to a Gaussian distribution and subjected to ANOVA and, finally, Tukey’s test for mean separation. For the zymography and immunoblotting studies, all measurements were performed in duplicate, and the zymographic/immunoreactive signals were analyzed using densitometric methods (Gel Pro Analyzer, Media Cybernetics) to obtain two-dimensional integrated optical density values. The integrated optical density values were then computed as a percentage of WT control values, where WT values were set to 100%. Comparisons to WT values were performed by a separate t-test, and potential treatment effects were examined by a two-way ANOVA followed by Bonferroni’s corrected t-test as appropriate. The SOD measurements were compared by ANOVA and then by Tukey’s test. All statistical procedures were performed using Systat statistical software (SPSS, Chicago, IL). Values are means ± SE. P < 0.05 was considered statistically significant.


Representative LV pressure-volume loops for WT and TIMP1-KO mice are shown in Fig. 2. At 14 days after MI, a significant shift to the right occurred in the LV pressure-volume relation, indicative of LV dilation. The degree of LV dilation was most pronounced in TIMP1-KO mice after MI (TIMP1-KO-MI). With MMP inhibition, the degree of LV dilation was attenuated in WT and TIMP1-KO mice. However, the attenuation in LV dilation after MI was more pronounced in the TIMP1-KO mice. Summary LV volumetric and pressure data are presented in Table 1. The heart rate was higher and mean blood pressure lower after MI in both groups of mice; this elevation in heart rate and mean blood pressure persisted in the MMP inhibition groups. LV end-diastolic volume, in absolute terms or indexed to tibial length, was increased in both groups after MI and was increased to a further degree in TIMP1-KO mice. With MMP inhibition, LV end-diastolic volume was reduced in both groups, with a greater relative reduction in TIMP1-KO mice. The LV ejection fraction was reduced in all groups after MI but was increased in TIMP1-KO mice after MMP inhibition compared with respective post-MI WT mice. LV peak pressure was reduced after MI and was reduced to a further degree in the MMPi groups. LV end-diastolic pressure was increased after MI and reduced in both MMPi groups. The LV PRSW relation normalized to LV end-diastolic volume (dyn·cm·mg·μl−1) was similar in WT and TIMP1-KO mice under control conditions (1.98 ± 0.42 and 1.79 ± 0.39, respectively). After MI, the PRSW relation decreased in WT mice (0.63 ± 0.06, P < 0.05). Although the PRSW relation fell in TIMP1-KO mice after MI, this change did not reach statistical significance (1.49 ± 0.21, P = 0.20). In WT-MI/MMPi mice, the PRSW relation increased from MI-only values but remained reduced compared with reference control values (1.52 ± 0.20, P < 0.05). In TIMP1-KO-MI/MMPi mice, the PRSW relation returned to within control values (1.61 ± 0.20).

Fig. 2.

A: representative steady-state left ventricular (LV) pressure-volume loops for a wild-type mouse under control [no myocardial infarction (MI)] condition, 14 days after MI, and 14 days after MI with matrix metalloproteinase (MMP) inhibition (MMPi). LV pressure-volume loop shifted to the right in the MI mouse, indicative of greater operative LV volumes, and shifted toward normal LV volumes with MMPi. B: steady-state LV pressure-volume loops in tissue inhibitor of MMP (TIMP-1) gene knockout (TIMP1-KO) mice without MI, after surgical induction of MI, and with concomitant MMPi. A significant rightward shift in TIMP1-KO mice after MI returned to within normal values with MMPi.

View this table:
Table 1.

LV function and hemodynamics after MI In mice: effects of TIMP-1 gene deletion and MMP inhibition

Total heart mass-to-body weight ratio increased in all MI groups compared with respective control non-MI values (Table 1). MI size computed by planimetry (27) was similar between WT and TIMP1-KO mice (38 ± 4 and 34 ± 4%, respectively), with no difference between WT-MI/MMPi and TIMP1-KO-MI/MMPi mice (36 ± 3 and 39 ± 4%, respectively). The frequency distribution for myocyte cross-sectional area is shown in Fig. 3. LV myocyte cross-sectional area increased from respective control non-MI values in all MI groups. In the TIMP1-KO group, myocyte cross-sectional area was higher than respective WT values, consistent with previous reports (32). With MMP inhibition, the degree of myocyte hypertrophic response after MI was similar to untreated MI values. Relative collagen volume fractions were determined for the MI region and the remote viable myocardium. Within the MI region, percent fibrillar collagen was 80 ± 2% for the WT-MI group and 65 ± 4% (P < 0.05) for the TIMP1-KO group. In both MMPi groups, percent collagen within the MI region was similar to WT-MI values (78 ± 4 and 84 ± 5%, respectively). Representative photomicrographs for fibrillar collagen and summary data for the remote region are shown in Fig. 4. The relative collagen fraction was reduced in the TIMP1-KO mice compared with WT values in the absence of MI. The collagen volume fraction within the remote region increased in the WT mice after MI, but this pattern was not observed in the TIMP1-KO mice after MI. The collagen volume fraction within the remote region was reduced in the WT-MI/MMPi group compared with reference non-MI values and MI-only values. In the TIMP1-KO-MI/MMPi group, the collagen volume fraction within the remote region was similar to respective non-MI values.

Fig. 3.

Frequency-distribution plots for LV myocyte cross-sectional area in wild-type and TIMP1-KO mice without MI (no MI), 14 days after MI, and 14 days after MI with MMPi. LV myocyte cross-sectional area increased after MI and remained increased with MMPi. LV myocyte cross-sectional area was higher in TIMP1-KO than in wild-type mice, indicative of greater myocyte remodeling. *P < 0.05 vs. no MI. #P < 0.05 vs. wild type.

Fig. 4.

Left: representative photomicrographs of picrosirius red-stained LV myocardial sections from wild-type and TIMP1-KO mice without MI, 14 days after MI, and 14 days after MI with MMPi. Right: relative increase in collagen volume fraction within the remote viable myocardium in wild-type mice after MI was reduced with MMPi. In TIMP1-KO mice, relative collagen volume fraction was reduced from respective wild-type values, and magnitude of collagen increase after MI was reduced. *P < 0.05 vs. no MI. #P < 0.05 vs. wild-type.

Representative MMP zymograms and immunoblots are shown in Fig. 5. A strong proteolytic signal for MMP-2 was observed in all myocardial extracts and was increased by 172 ± 16% in all MI groups compared with WT values (P < 0.05); similar values were obtained in the TIMP1-KO and MMPi groups after MI (2-way ANOVA). A proteolytic signal for MMP-9 was observed in all MI groups and appeared to be greater than WT-only values (166 ± 41%), but the increase did not reach statistical significance (P = 0.114), with no difference between any of the MI groups. Relative MMP-3 levels were increased by 269 ± 27% with MI (P < 0.05) and were equivalently increased in all post-MI groups. For MMP-13, a latent pro-MMP-13 immunoreactive signal, as well as a lower active MMP-13 signal, was detected. In contrast to the other MMPs measured, differences in MMP-13 levels were observed in the non-MI TIMP1-KO group compared with the WT group. Specifically, pro-MMP-13 was increased by 163 ± 21% (P < 0.05) and active MMP-13 was increased by 161 ± 30% (P = 0.058) from WT values. In all MI groups, pro-MMP-13 increased by 131 ± 12% and active MMP-13 increased by 202 ± 24% from WT values (both P < 0.05). MMP-13, both latent and active, increased to a similar degree in all post-MI groups. MT1-MMP increased by 168 ± 20% from WT values in all MI groups (P < 0.05); no treatment differences were observed. Representative immunoblots for TIMP-2, TIMP-3, and TIMP-4 are shown in Fig. 5. In the TIMP1-KO group, basal differences in TIMP-2 levels were observed compared with WT values. Specifically, in the non-MI TIMP1-KO group, TIMP-2 was increased by 274 ± 21% from WT values (P < 0.05). After MI, TIMP-2 levels were equivalently increased from WT values in the WT and TIMP1-KO groups (243 ± 27 and 193 ± 37%, respectively, P < 0.05). TIMP-2 levels were increased in the WT-MI/MMPi and TIMP1-KO-MI/MMPi groups (248 ± 26 and 166 ± 20%, respectively, P < 0.05) but were lower in the TIMP1-KO-MI/MMPi group (P < 0.05). TIMP-3 levels were equivalent in the WT and TIMP1-KO groups under basal conditions. TIMP-3 levels significantly fell in the WT-MI group (57 ± 9%, P < 0.05) and were reduced from WT and WT-MI values in the TIMP1-KO-MI group (38 ± 7%, P < 0.05). Similar reductions in TIMP-3 were observed in the WT-MI/MMPi (50 ± 6%, P < 0.05) and TIMP1-KO-MI/MMPi (36 ± 7%,P < 0.05) mice, with TIMP1-KO values being lower than respective WT values (P < 0.05). A robust immunoreactive signal for TIMP-4 in all myocardial extracts was unaffected by TIMP1-KO (98 ± 9%), MI (108 ± 3%), or MI/MMPi (107 ± 4%).

Fig. 5.

Left: representative zymograms for MMP-2 and MMP-9 and immunoblots for MMP-3, MMP-13, and MT1-MMP in wild-type (WT) and TIMP1-KO mice 14 days after MI and 14 days after MI with MMPi. Relative levels for MMP-2 significantly increased after MI. A detectable signal for MMP-9 was observed in all myocardial extracts. An immunoreactive signal for MMP-13 could be detected for the proform (∼60 kDa) and an active form (∼48 kDa). Active band for MMP-13 was increased in TIMP1-KO mice and in mice after MI. MT1-MMP band was increased in all groups after MI. Right: relative levels of TIMP-2, TIMP-3, and TIMP-4 were determined by immunoblotting and revealed a robust band for TIMP-2 in TIMP1-KO mice under basal non-MI conditions. TIMP-2 levels were increased in WT and TIMP1-KO mice after MI. TIMP-3 levels decreased in all groups after MI. TIMP-4 levels were similar in all groups of mice.

Representative immunoblots for the derivatized DNP product are shown in Fig. 6. Quantitative measurements revealed that DNP levels increased by 307 ± 25% from WT values in the TIMP1-KO group (P < 0.05). DNP levels increased to a similar degree in the WT and TIMP1-KO groups after MI (234 ± 18 and 325 ± 41%, respectively, P < 0.05) and remained increased in both groups with MMP inhibition (272 ± 33 and 304 ± 38%, respectively, P < 0.05). LV myocardial SOD activity followed a similar pattern (Fig. 7). SOD activity was higher in the TIMP1-KO group and was higher in all groups after MI. However, SOD activity was higher in the WT-MI/MMPi group than in the respective MI-only group, and this trend was also observed in the TIMP1-KO-MI/MMPi group but did not reach statistical significance (P = 0.26).

Fig. 6.

Relative measure of oxidative stress determined in myocardial extracts by measurement of degree of protein carbonylation through derivatization to a stable dinitrophenyl (DNP) product (19). Representative immunoblots are shown for DNP in myocardial extracts from WT and TIMP1-KO mice 14 days after MI and 14 days after MI with MMPi. Relative DNP levels were increased in TIMP1-KO mice under basal non-MI conditions and increased in all groups after MI. Nonderivatized myocardial samples were run on adjoining lanes (-) and demonstrated an absence of a DNP signal.

Fig. 7.

Indication of significant changes in oxidative stress through measurement of superoxide dismutase (SOD) activity (21–23). SOD activity was measured in myocardial extracts from WT and TIMP1-KO mice 14 days after MI and 14 days after MI with MMPi. SOD activity was increased in TIMP1-KO mice under basal non-MI conditions and increased in all groups after MI. SOD activity was higher in WT-MI/MMPi than in MI mice. *P < 0.05 vs. WT. #P < 0.05 vs. MI.


Alterations in MMPs and TIMPs have been identified in end-stage human heart failure and animal models of MI (9, 24, 27, 33). Previous animal studies demonstrated a causal link between changes in MMP and TIMP levels in response to adverse LV myocardial remodeling after MI (8, 18). In a previous report, TIMP-1 gene deletion resulted in an accelerated LV dilation after MI (8). However, from this study (8) as well as other post-MI transgenic models (11, 18, 19), it was unclear whether and to what degree this phenotype could be altered through pharmacological MMP inhibition. The present study utilized the TIMP1-KO mouse model to examine the effects of MMP inhibition on LV geometry and function in the context of post-MI remodeling. The new findings from the present study are as follows: 1) the increased LV dilation that occurred in the TIMP1-KO mouse after MI could be significantly abrogated with MMP inhibition and was associated with improved LV pump function; 2) MMP inhibition did not prevent the adaptive myocyte hypertrophic response in WT or TIMP1-KO mice and was not associated with a robust fibrotic response; and 3) in TIMP1-KO mice, differential profiles of the primary rodent interstitial MMP MMP-13 were associated with increased markers of oxidative stress. These unique findings demonstrated that, in a murine model in which a “gain of MMP function” was achieved through TIMP-1 gene deletion, pharmacological MMP inhibition altered the LV phenotype in this transgenic model after MI. These results emphasize the importance of MMP inhibitory control within the myocardial compartment after MI with respect to the adverse LV remodeling process.

Previous studies utilized different transgenic constructs to demonstrate that the deletion of specific MMP genes can alter the course of LV remodeling after MI (11, 18, 19). For example, MMP-9 gene deletion in mice caused a relative reduction in the degree of LV remodeling after MI and was associated with increased survival (11). More recently, in mice devoid of the MMP-2 gene, a similar effect was noted, in that the degree of LV dilation was reduced and survival was increased after MI (18). TIMP-1 binds to the active form of all MMPs, including MMP-9 and MMP-2. Thus deletion of TIMP-1 would be expected to result in an acceleration of the LV remodeling process. Indeed, it was demonstrated previously that TIMP-1 gene deletion in mice was associated with increased LV dilation after MI (8). Taken together, these previous studies suggest that alterations in MMP:TIMP stoichiometry, which would favor increased myocardial MMP activity, adversely affect the post-MI remodeling process. However, these previous transgenic studies result in a global gene deletion that is manifested at development; therefore, the effects of other confounding variables, such as defects in the inflammatory response, cannot be completely eliminated. To address this issue, the present study utilized a pharmacological MMP inhibitor coincident with surgical induction of MI to examine the effects of this pharmacological approach superimposed on global TIMP-1 gene deletion. The results from this study demonstrated that the post-MI phenotype in TIMP1-KO mice could be successfully modified with pharmacological MMP inhibition. These results emphasize the important role of TIMP-1 in the early post-MI remodeling process. Furthermore, these results provide the justification for future studies in which selective induction of TIMP-1 through gene transfer is performed after MI.

A number of previous studies demonstrated that pharmacological MMP inhibition can alter the course of post-MI remodeling (9, 24, 27, 31). For example, it has been demonstrated that the degree of LV dilation can be reduced in a mouse model of MI when MMP inhibition is introduced in the post-MI period (5). In large animal models of post-MI remodeling, it was demonstrated previously that MMP inhibition can reduce the degree of infarct expansion, which in turn was translated to a relative reduction in LV dilation in the post-MI period (27). The present study builds on these past pharmacological studies by examining the effects of pharmacological MMP inhibition superimposed on a genetic lesion in endogenous inhibitory control (TIMP-1 deletion). This provided for a more careful assessment of the potential effects of a loss in TIMP-1-mediated MMP inhibition with respect to LV structure and function. In this study, LV dilation occurred in WT and TIMP1-KO mice 2 wk after coronary ligation. This LV dilation after MI was accompanied by LV pump dysfunction, defined as a relative decrease in LV ejection fraction and peak systolic pressure and a relative increase in LV end-diastolic pressure. When MMP inhibition was introduced 3 days after MI, the degree of LV dilation was reduced in WT and TIMP1-KO mice and was associated with a concomitant increase in LV ejection fraction. In a previous report from this laboratory, the induction of MI in this mouse model was not associated with a reduction in LV ejection fraction (8). We speculated that this may have been due to the superimposition of mitral regurgitation, which would result in the overestimation of LV ejection fraction (8). Before the initiation of the present study, a pilot study was performed using transmitral pulse-wave Doppler echocardiography to assess whether regurgitation could be detected utilizing the coronary artery ligation approach described previously (8). Indeed, mitral regurgitation could be visualized and appeared to be exacerbated by placement of the LV catheter. Accordingly, in the present study, the MI procedure was modified such that only the left anterior descending artery was ligated, which resulted in only minimal mitral regurgitation. This resulted in consistent, but slightly smaller, MIs than those reported previously (8) and a uniform reduction in LV ejection fraction 2 wk after MI. However, it must be recognized that mitral regurgitation is not an uncommon consequence of MI; therefore, quantification of this potentially confounding variable, particularly in murine models of MI, would be appropriate in future studies (2). Several other interesting outcomes were observed from this portion of the study. First, LV peak systolic pressure remained reduced with MMP inhibition, despite improvements in LV geometry and function. This was likely due to the fact that the degree of neurohormonal activity and ventricular-arterial coupling remained abnormal in this post-MI period, independent of MMP inhibition. In a previous report, MMP inhibition instituted in a model of heart failure reduced the degree of LV dilation but did not significantly improve LV fractional shortening (34). Thus it is also possible that MMP inhibition instituted in this murine model of MI may have neutral or negative effects on LV systolic function. On the basis of these observations, future studies that employ MMP inhibition for longer time periods and assess LV geometry and function are warranted. Second, the relative changes in myocardial contractility, as assessed by the PRSW relation, increased to a greater degree in TIMP1-KO mice after MMP inhibition. This was unlikely due to differences in MI size, because MMP inhibition was instituted 3 days after coronary ligation; therefore, the degree of myocardial salvage with this pharmacological intervention would have been minimal. Rather, this relative improvement in LV contractility in TIMP1-KO mice was likely due to a greater degree of hypertrophy of residual myocytes than in WT mice and improved extracellular structure. Another possibility for this apparent differential effect of MMP inhibition in TIMP1-KO mice is the fact that the TIMP-1 protein may influence biological processes other than MMP activity (21, 40). For example, TIMPs have been reported to alter cell growth and biological responsiveness to extracellular stimuli in vitro (5). Thus, in the WT mice, these pleotropic effects of TIMP-1, which are independent of MMP activity, would be operative. On the basis of the observations from this study, future studies that address the potential divergent effects of TIMP-1 within the myocardium are warranted.

A much greater degree of myocardial hypertrophy occurred in the TIMP-1−/− mice after MI, as evidenced by greater ventricular mass and myocyte cross-sectional area. Because a greater degree of LV dilation occurred in the TIMP-1−/− mice, this robust hypertrophic response was likely due to increased stress on the viable myocardium. With MMP inhibition, myocyte cross-sectional area remained similar to that observed in the untreated MI groups. This observation is likely due to the fact that MMP inhibition was initiated 3 days after MI, which provided for manifestation of the adaptive hypertrophic response. However, the relative degree of LV dilation after MI was reduced with MMP inhibition in WT and TIMP1-KO mice. These results provide the basis for the following conclusions. First, there may be a temporal disparity between the type of myocyte hypertrophy that occurs after MI. Specifically, myocyte cross-sectional area increased (addition of sarcomeres in parallel) soon after MI, and myocyte length increased (addition of sarcomeres in series) at a later time point after MI. Second, the putative changes in myocyte length after MI were likely facilitated by a loss of normal extracellular collagen support.

The myocardial fibrillar collagen matrix provides for architectural support for myocytes and other resident cells within the myocardium. This network imparts specific physical characteristics to the myocardium and also contributes to the maintenance of myocyte alignment and geometry (4, 41). Several previous studies demonstrated that disruption of the normal structural support mediated by the fibrillar collagen matrix can alter myocardial structure and function (23, 35). For example, Kim and colleagues (23) demonstrated that direct disruption of the myocardial matrix by overexpression of MMP-1 resulted in marked changes in LV geometry. This laboratory demonstrated previously that, in TIMP1-KO mice, alterations in fibrillar collagens occurred as a function of age and were associated with progressive LV dilation (32). Consistent with this previous report, relative collagen volume fractions were reduced in the TIMP1-KO mice compared with the WT mice under control conditions as well as after MI. This observation is likely due to enhanced MMP activity secondary to a loss of local inhibitory control. With MMP inhibition, increased fibrillar collagen was not observed in WT or TIMP1-KO mice. This finding is consistent in other model systems in which chronic MMP inhibition did not cause increased collagen accumulation (24, 27, 31). Collagen degradation can result in the formation of small peptides, which can actually stimulate collagen synthesis (14, 36). Thus, in the present study, MMP inhibition may have interrupted a “feedforward” stimulus for collagen synthesis and, ultimately, fibrosis within the viable remote myocardium. Although the mechanism(s) for this effect of MMP inhibition remains to be elucidated, the present study demonstrated that this pharmacological intervention after MI was not associated with adverse myocardial fibrosis.

Previous studies in MI models demonstrated that alterations in the relative myocardial levels of MMPs and TIMPs result in a change in MMP:TIMP stoichiometry favoring heightened matrix proteolytic activity within the myocardium (9, 27, 33, 39). In the present study, representative MMP types from each subfamily of MMPs were measured: the interstitial collagenase MMP-13, the gelatinases MMP-2 and MMP-9, the stromelysin MMP-3, and the membrane-type MMP MT1-MMP. The first important observation from these studies was that a greater amount of active MMP-13 was observed under control conditions with TIMP-1 deletion. This was likely due to several factors, including increased oxidative stress, as well as a loss of TIMP-1 inhibitory control (see below). The active band for MMP-13 was increased in WT and TIMP1-KO mice after MI. In addition, 14 days after MI, higher levels of MMP-2, MMP-3, and MT1-MMP were observed in WT and TIMP1-KO mice. However, it is likely that the proteolytic activity of MMP-13, as well as these other MMP types, would be more prolonged in the absence of TIMP-1 and, in turn, result in a greater degree of matrix degradation and turnover in TIMP1-KO mice. Indeed, fibrillar collagen content was lower in TIMP1-KO than in WT mice after MI. In the present study, the relative levels of MMP-9 were increased 14 days after MI, but this did not reach statistical significance. This is likely because MMP-9 is primarily released by neutrophils and other inflammatory cells during the acute phase of an MI (9, 19, 24), and this acute inflammatory phase was likely resolving when MMP type was measured in the present study. With MMP inhibition, the relative myocardial MMP levels remained similar to those observed in untreated MI. Thus the effects of MMP inhibition in this study were likely due to inhibition of active MMPs rather than modulatory effects on endogenous MMP levels. However, previous reports documented that myocardial levels of certain MMP types are affected, with longer durations of MMP inhibition after MI (27). The time-dependent effects of MMP inhibition on local MMP levels, within the myocardium as well as systemically, warrant further study if this treatment approach is to be considered for clinical application (9, 33).

The differences in MMP levels and activation patterns observed in the present study must be considered in the context of changes in the endogenous inhibition of MMPs. In the present study, TIMP-2, TIMP-3, and the relatively myocardium-specific TIMP-4 (17) were assessed. TIMP-1 was not measured in the present study, because this TIMP type was totally absent in TIMP1-KO mice, and preliminary studies revealed that TIMP-1 levels appeared unchanged in WT-MI mice (data not shown). Although basal levels of TIMP-3 and TIMP-4 remained unchanged in TIMP1-KO mice, increased TIMP-2 levels were observed. This suggests that the deletion of TIMP-1 caused a compensatory increase in TIMP-2. Despite the increased abundance of TIMP-2, the active form of MMP-13 remained increased in TIMP1-KO mice, and LV remodeling occurred. With the superimposition of an MI, the relative levels of TIMP-2 increased equivalently in WT and TIMP1-KO mice. Furthermore, TIMP-3 levels significantly fell after MI and were reduced to a greater degree in TIMP1-KO mice. Finally, TIMP-4 levels were unaffected after MI. Thus, after the induction of MI, the summated loss in TIMP-mediated MMP inhibition was greater in TIMP1-KO mice, in that TIMP-1 was absent and TIMP-3 levels were reduced to a greater degree. These findings are significant in several ways. First, with the induction of an MI, a compensatory increase in alternative TIMPs did not occur with TIMP-1 deletion. It has been demonstrated that TIMP-1 gene deletion is not associated with a concomitant increase of alternative TIMPs in other organ injury models such as the kidney (12). Second, the relative increase in TIMP-2 and the relative reduction in TIMP-3 after MI suggest differential regulation of these TIMP types. Indeed, differences in the TIMP gene promoter regions have been identified and may contribute to differential expression in disease states (37). Finally, in TIMP1-KO mice and in all mice after MI, there was no compensatory increase in myocardium-specific TIMP-4. A previous study documented that the relative levels for all TIMP types are reduced within the remodeling myocardium 2 mo after MI. Taken together, the findings from the present study and previous reports would suggest a differential temporal profile in TIMP levels within the myocardium after MI. This differential TIMP profile may influence the time course of adverse LV remodeling after MI and may hold important considerations for the deployment of pharmacological MMP inhibition after MI. In the present study, MMP inhibition was instituted 3 days after MI and continued for 14 days after MI. This MMP inhibition protocol did not change relative TIMP levels from MI-only values. However, previous studies suggested that MMP inhibition may influence TIMP levels (3, 27); therefore, whether and to what degree more prolonged MMP inhibition would influence endogenous TIMP levels in this murine MI model remain to be addressed.

The generation of reactive oxygen species after MI is likely one of several biological events that causes the induction of a number of molecules associated with the wound-healing response as well as adverse LV remodeling (5, 29, 38). One approach in measuring the relative changes in oxidative stress is measurement of DNP labeling of derivatized LV myocardial extracts (5, 10). In addition, the present study measured relative SOD activity in LV myocardial extracts, which would reflect a persistent formation of reactive oxygen species (1, 25, 29, 38). In TIMP1-KO mice, increased DNP labeling and SOD activity, which are biomarkers for oxidative stress, were observed under basal, non-MI conditions. These observations would imply increased oxidative stress in this transgenic construct in the absence of an exogenous pathological stimulus. Previous studies demonstrated that the generation of reactive oxygen species can cause MMP activation (22, 26). Moreover, Cox and colleagues (6, 7) demonstrated that biomarkers of oxidative stress are associated with MMP activation and subsequent LV remodeling. In the present study, increased myocardial levels of the active form of MMP-13 were observed in the TIMP1-KO mouse and may have been due to increased oxidative stress. However, it remains unclear from this study whether increased oxidative stress is a major inciting stimulus for the MMP activation and subsequent LV remodeling in this transgenic construct (9, 32). As expected, these biomarkers of oxidative stress increased in the post-MI period. The relative magnitude of this increase was equivalent in WT and TIMP1-KO mice. However, although DNP labeling was unchanged with MMP inhibition, SOD activity was higher than MI-only values. Whether and to what degree these higher SOD levels can be translated into an overall reduction in myocardial oxidative stress remain to be established. However, a recent report documented that exogenous delivery of a TIMP construct can reduce biomarkers of oxidative stress, reduce relative MMP activity, and modify LV remodeling in a rodent model of volume overload (7).

In summary, we have shown that MMPs have a regulatory role in post-MI remodeling and that the accelerated adverse LV myocardial remodeling induced by TIMP-1 gene deletion can be pharmacologically “rescued” by MMP inhibition. These results define the importance of local endogenous control of MMP activity with respect to regulating LV structure and function after MI.


This study was supported by National Institutes of Health Grants HL-59165, PO1 HL-48788-08, and P20 RR-16434 and a Career Development Award from the Veterans Affairs Health Administration.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract