Cardiomyocyte cyclooxygenase-2 influences cardiac rhythm and function

Abstract

Nonsteroidal anti-inflammatory drugs selective for inhibition of COX-2 increase heart failure and elevate blood pressure. The COX-2 gene was floxed and crossed into merCremer mice under the alpha-myosin heavy-chain promoter. Tamoxifen induced selective deletion of COX-2 in cardiomyocytes depressed cardiac output, and resulted in weight loss, diminished exercise tolerance, and enhanced susceptibility to induced arrhythmogenesis. The cardiac dysfunction subsequent to pressure overload recovered progressively in the knockouts coincident with increasing cardiomyocyte hypertrophy and interstitial and perivascular fibrosis. Inhibition of COX-2 in cardiomyocytes may contribute to heart failure in patients receiving nonsteroidal anti-inflammatory drugs specific for inhibition of COX-2.

Fig. 1.

Cardiomyocyte-specific COX-2 gene deletion in vivo. (A) Exons 6, 7, and 8 of the COX-2 gene are flanked by 2 loxp sites to generate COX-2 floxed mice; cre enzyme activation may then permit time- and site-specific deletion of COX-2. (B) Peritoneal macrophages were harvested from WT COX-2 and floxed COX-2 mice and were treated with lipopolysaccharide (5 μg/ml) or vehicle control. COX-2-dependent PGE₂ production in peritoneal macrophages (n = 3) was similar in the floxed mice and in age- and gender-matched controls. (C) Primers were designed to flank exons 6, 7, and 8 of the COX-2 gene. PCR of genomic DNA results in a ≈2-kb product from the WT allele and a ≈0.4-kb DNA product from the COX-2 KO allele. (D) The COX-2-mutant band is detected in heart, but not kidney of CM COX-2 knockout mice. The COX-2 mutant band was absent in cardiac tissue from control mice.

Fig. 2.

CM-specific deletion of COX-2 impairs cardiac function. (A) Left ventricular (LV) ejection fraction (EF) was reduced by an average 8% (from 0.65 ± 0.02 cm³ to 0.60 ± 0.02 cm³; *, P < 0.05) in CM-COX-2 KOs and left ventricular end systolic volume (ESV) was increased 24% (from 0.017 ± 0.002 cm³ to 0.021 ± 0.004 cm³; *, P < 0.05). Heart rate fell from 443 ± 21 in controls to 411 ± 30 beats per min in CM-COX-2 KOs (n = 9–11; *, P < 0.05). The estimates of ventricular volume are expressed in cm³, those for ventricular mass in grams. (B) Mice were subjected to 3 weeks of treadmill exercise. The decline in body weight at the end of week 3 was greater in CM-COX-2 KOs than in controls (1.22 ± 0.21 grams vs. 0.44 ± 0.23 grams; n = 9; **, P < 0.01). (C) After 3 weeks, exercise tolerance on a 10° incline was reduced on average by 19% from 3,087 ± 96 to 2,500 ± 327 (body weight × running time) in the CM-COX-2 KOs but not in controls (n = 9; * P < 0.05). (D) Ventricular arrhythmia in a cardiomyocyte COX-2 null animal. Shown from top to bottom are surface ECG leads I, II, III, aVR, aVL, aVF, and intracardiac recordings from the right atrium (RA) and right ventricle (RV). Shown is an episode of polymorphic ventricular tachycardia induced by burst stimulation (S) at a cycle length of 50-ms followed by 3 extrastimuli coupled at 30 ms in a CM COX-2 CKO mouse. The rapid ventricular beats (v) with dissociation from the atrial beats (a) are pathognomonic of ventricular tachycardia. A sinus beat is shown upon termination of the episode with normal atrial (p) and ventricular (QRS) depolarization.

Fig. 3.

Deletion of CM COX-2 modulates the response to aortic banding. (A) The impairment in left ventricular ejection fraction in CM-COX-2 KOs was still evident 1 week after aortic banding, being reduced on average by 12% in the CM-COX-2 KOs (0.58 ± 0.02) compared to controls (0.66 ± 0.02 in.; n = 6–8, **, P < 0.01). However, 6 weeks after aortic banding, function had recovered significantly (n = 13) versus 1 week after banding in the CM-COX-KOs, and now did not differ from the values in controls. (B) Wheat germ staining was performed to analyze the size of cardiac myocytes. An example illustrates that CM hypertrophy in response to banding is exaggerated in CM-COX-2 KOs. (C) CM size was not significantly altered by CM-COX-2 deletion alone. However, 6 weeks after banding, CM size was increased on average by 28% (4,440 ± 361 versus 3,459 ± 205) in the CM-COX-2 KOs compared to controls (n = 7–9, *, P < 0.05).

Fig. 4.

Deletion of cardiomyocyte COX-2 augments interstitial and perivascular fibrosis in the heart. Collagen deposition in the heart from 6-week aortic-binding mice by Picro-Sirius red staining (A) and Trichrome staining (B). Both interstitial and perivascular fibrosis was apparent. (C) Fibrosis was increased significantly (2.028 ± 0.209 vs. 0.963 ± 0.116 arbitrary unit/pixel; n = 3, ****P < 0.0001) in CM-COX-2 KOs. This difference was exaggerated further when analyzed 6 weeks after the mice had been subject to aortic banding (5.220 ± 0.989 vs. 1.339 ± 0.156 arbitrary unit/pixel; n = 3, *** P < 0.0002).