Impaired myocardial angiogenesis and ischemic cardiomyopathy in ...

© 1999 Nature America Inc. • http://medicine.nature.com

ARTICLES

Impaired myocardial angiogenesis and ischemic cardiomyopathy in mice lacking the vascular endothelial growth factor isoforms VEGF164 and VEGF188 PETER CARMELIET1, YIN-SHAN NG2, DIETER NUYENS1, GREGOR THEILMEIER1, KOEN BRUSSELMANS1, IVO CORNELISSEN1, ELISABETH EHLER3, VIJAY V. KAKKAR4, INGEBORG STALMANS1, VIRGINIE MATTOT1, JEAN-CLAUDE PERRIARD3, MIEKE DEWERCHIN1, WILLEM FLAMENG5, ANDRAS NAGY6, FLOREA LUPU4, LIEVE MOONS1, DÉSIRÉ COLLEN1, PATRICIA A. D’AMORE2 & DAVID T. SHIMA2,7


1

The Center for Transgene Technology and Gene Therapy, Flanders Interuniversity Institute for Biotechnology, KU Leuven, Leuven, B-3000, Belgium 2 Schepens Eye Research Institute, Harvard Medical School, Boston, Massachusetts 02115, USA 3 Institute for Cell Biology, Swiss Federal Institute of Technology, Zürich, CH-8093, Switzerland 4 Vascular Biology Laboratory, Weston Experimental Research Center, Thrombosis Research Institute, London SW3 6LR, UK 5 Laboratory of Experimental Cardiac Surgery, KU Leuven, Leuven, B-3000, Belgium 6 Samuel Lunenfeld Institute, Mount Sinai Hospital, Toronto, ON M5 1X5, Canada 7 Cell Biology, Imperial Cancer Research Fund, London WC2A 3PX, UK Correspondence should be addressed to P.C.; email: [email protected]

The gene for vascular endothelial growth factor (VEGF) encodes three spliced isoforms. Although the heparin binding capacities of these isoforms differ, little is known about their differential functions in vivo. We generated mice expressing exclusively the VEGF120 isoform (VEGF120/120 mice) by Cre/loxP-mediated removal of exons 6 and 7, which encode the isoforms of 164 and 188 amino acids. VEGF120/120 mice had impaired postnatal myocardial angiogenesis, resulting in ischemic cardiomyopathy characterized by reduced contractility and sarcomere breakdown, but normal stores of high-energy phosphates. VEGF120/120 mice ultimately died of cardiac failure. The VEGF120/120 mouse model may be useful for studying the molecular mechanisms of the myocardial response to ischemia and for testing the angiogenic properties of different VEGF isoforms. Vascular endothelial growth factor (VEGF) is involved in embryonic and pathological vascular development1. It may also be involved in the angiogenic response to myocardial ischemia and is now being tested for use in gene therapy of ischemic heart and tissue disease2. The mouse VEGF gene is alternatively transcribed to produce at least three isoforms: VEGF120, VEGF164 and VEGF188. VEGF120 is diffusible in the extracellular milieu, whereas the longer isoforms show increasing binding to heparan sulfate-rich matrix1,3–5. These isoforms differ in their mitogenicity, chemotactic properties, receptor binding characteristics and tissue-specific expression5–8. However, it remains controversial whether the isoforms differ in specificity, potency or quality for their role in normal, pathological or therapeutical angiogenesis in vivo5,7,9–11. Nevertheless, gene therapy protocols now use VEGF120 or VEGF164 indiscriminately for improvement of tissue ischemia. As the loss of a single VEGF allele results in embryonic lethality due to severe vascular defects12,13, it is not possible to study the role of VEGF during postnatal angiogenic processes by conventional transgene technology. Therefore, we generated mice expressing only VEGF120 (VEGF120/120 mice) using the Cre/loxP system to remove exons 6 and 7, which encode NATURE MEDICINE • VOLUME 5 • NUMBER 5 • MAY 1999

basic domains that are only present in VEGF164 and/or VEGF188 (ref. 14). The absence of VEGF164 and VEGF188 impaired myocardial angiogenesis, leading to ischemic cardiomyopathy. Myocardial ischemia is the principal cause of mortality and morbidity in Westernized societies. However, the response of cardiomyocytes to reduced oxygen supply remains poorly characterized. Although cardiomyocytes die of severe acute ischemia, they may survive chronic ischemia by ‘hibernating’; for example, by reversibly downgrading contractile function to reequilibrate energy expenditure and supply15–19. However, cardiomyocytes may also fail to adapt to ischemia and undergo progressive degeneration, atrophy and interstitial fibrosis17,20, resulting in incomplete functional recovery after revascularization21. Despite its medical importance, the mechanisms of the myocardial response to ischemia remain poorly understood, in part because of lack of informative animal models16,22. Animal models of more progressive coronary artery occlusion rarely combine sustained or intermittent hypoperfusion with ventricular dysfunction, structural adaptation (myolysis) and intact metabolic activity, or simply lack any adaptation to ischemia because of well-developed collateral vessels16,22,23. The VEGF120/120 mouse model, with expression of only VEGF120, may allow the elucidation of the molecular mechanisms of the cardiac response to ischemia, and permit testing of gene therapy strategies for ischemic heart disease. Generation of VEGF120/120 mice The VEGF120 isoform is produced by alternative splicing around exons 6 and 7 (refs. 1,14). We achieved site-specific removal of VEGF exons 6 and 7 in embryonic stem (ES) cells using the Cre/LoxP system (Fig. 1a–c). We used targeted ES cells to generate VEGF+/120 mice, which seemed normal and healthy. Neonates expressing exclusively VEGF120 (VEGF120/120), sired by VEGF+/120 breeding pairs, were recovered at birth at a normal Mendelian frequency: Of 120 neonates, 26% were VEGF+/+, 51% were 495


ARTICLES


VEGF+/120 and 24% were VEGF120/120. RNA analysis confirmed correct gene targeting, as only VEGF120, but not VEGF164 or VEGF188, was detectable in VEGF120/120 mice, at levels similar to the total amount of VEGF120, VEGF164 and VEGF188 in VEGF+/+ mice (Fig. 1d). About half the VEGF120/120 neonates died within a few hours after birth because of bleeding in several organs (data not shown). The VEGF120/120 neonates that survived this perinatal period had normal body weights at birth, but failed to gain weight. The body weights for VEGF+/+ and VEGF120/120 mice were: on postnatal day 0.5 (P0.5), 1.5 ± 0.03 g and 1.3 ± 0.03 g; on P6, 4.6 ± 0.02 g and 2.9 ± 0.1 g; and on P12, 8.1 ± 0.2 g and 4.4 ± 0.3 g (mean ± s.e.m.; n = 10–13; P < 0.05 for P6 and P12). VEGF120/120 mice became progressively lethargic and died before postnatal day 14 (of 230 neonates: 51% survival at P1; 27%, at P4–P6; 6%, at P9; and 0.5%, at P12). Impaired cardiac performance in VEGF120/120 mice Dissection of VEGF120/120 mice showed they had enlarged hearts, irregular heart beats and dysmorphic, weak heart contractions. As the anesthetized mice were not ventilated, they died of pulmonary collapse shortly after their thoracic cavities were opened. VEGF120/120 mice survived this stress for a much shorter period (about 2 minutes) than their VEGF+/+ littermates (about 15 minutes). Heart rates were slower in conscious P6 VEGF120/120 mice (360 ± 30 beats per minute) than in VEGF+/+ mice (530 ± 20 beats per minute)(n = 8–14; P < 0.001). Compared with their VEGF+/+ littermates, P6 VEGF120/120 mice had depressed left ventricular (LV) systolic pressure (73 ± 3 mm Hg in VEGF+/+ and 54 ± 2 mm Hg in VEGF120/120; P < 0.05), LV contractility (LV dP/dtmax: 2800 ± 110 mm Hg/s in VEGF+/+ and 1500 ± 100 mm Hg/s in VEGF120/120; P < 0.05) and LV relaxation (LV dP/dtmin: 2100 ± 170 496

+/+

+/neotk

BamHI 120/120

c

EcoRI

+/120

b +/+

a

9.5kb

7.5kb

7.8kb

2.5kb

d Abundance of individual VEGF mRNA Isoforms (% of total)

Fig. 1 Targeting of the VEGF gene. a, Modification of the VEGF gene. Top box, Targeting vector pPNT.VEGF120 and wildtype VEGF allele (WT VEGF). Middle box, Homologously recombined VEGF allele (VEGFneo/tk). Bottom box, Modified VEGF120 allele after Cre-excision of the floxed neo/tk cassette. Filled boxes, numbered exons; grey shaded bars, intron sequences; lettered bars external hybridization probe A (0.65-kb Sma1–BglI fragment) and internal hybridization probe B (1.7kb NcoI–EcoRI fragment).The observed sizes of the diagnostic restriction fragments, used to distinguish the wild-type and mutant alleles, correspond to their expected sizes. b, Southern blot analysis of EcoRI-digested genomic DNA from VEGF+/+ and VEGFneotk/+ ES cell clones using the external 5’ probe A, which generates a 7.5-kb wild-type fragment and a 2.5-kb VEGFneotk fragment. c, Southern blot analysis of BamHI-digested genomic DNA from VEGF+/+, VEGF+/120 and VEGF120/120 mice using the external 5’ probe A, which generates a 9.5-kb wild-type fragment and a 7.8-kb VEGF120 fragment. d, RNAse

VEGF120

VEGF164

VEGF188

protection analysis of P0.5 VEGF+/+ and VEGF120/120 organs. The relative abundance of each isoform is expressed as a percent of the total amount of VEGF, indicated in arbitrary units (AU) above each bar.

mm Hg/s in VEGF+/+ and 1100 ± 100 mm Hg/s in VEGF120/120; P < 0.05)(Fig. 2a). During periods of spontaneous arrhythmia in VEGF120/120 mice, LV pressure was reduced even further (Fig. 2a). Transthoracic echocardiography showed that, compared with VEGF+/+ mice, VEGF120/120 mice had enlarged systolic LV diameters (39 ± 4 µm in VEGF+/+ and 70 ± 4 µm in VEGF120/120; P < 0.05) and diastolic LV diameters (140 ± 6 µm in VEGF+/+ and 165 ± 9 µm in VEGF120/120; P < 0.05), impaired fractional shortening (72 ± 3 µm in VEGF+/+ and 56 ± 2 µm in VEGF120/120; P < 0.05) and reduced aortic outflow velocity (80 ± 5 ml/s in VEGF+/+ and 51 ± 7 ml/s in VEGF120/120; P < 0.05) (Fig. 2b). Thus, VEGF120/120 mice suffered severe LV pump failure. They also showed signs of right ventricular failure, including accumulation of ascitic fluid and increased liver-to-body weight ratio (31 ± 1 mg/g and 31 ± 1 mg/g in VEGF+/+ mice, and 37 ± 2 mg/g and 56 ± 1 mg/g in VEGF120/120 mice at P6 and P12, respectively; n = 7–13; P < 0.001 compared with VEGF+/+ mice). The cardiac dysfunction was not due to anemia (mice had normal hematological profiles) or renal failure (mice had normal ionograms) (not shown). Moreover, macroscopic and histological analysis failed to demonstrate any congenital heart defects, anomalies of the large vessels or abnormal closure of the ductus arteriosus (not shown). Impaired myocardial angiogenesis in VEGF120/120 mice Cardiomyocytes from P0.5 VEGF120/120 mice, cultured for 4 days, differentiated normally, indicating that loss of VEGF164 and VEGF188 did not directly affect cardiomyocytes (not shown). In VEGF+/+ mice, the number of capillaries and coronary vessels (surrounded by smooth muscle α-actin stained cells) increased during the first 3 postnatal weeks 300% and 1,000%, respectively (Fig. 2c and h), presumably to match the increasing metaNATURE MEDICINE • VOLUME 5 • NUMBER 5 • MAY 1999



ARTICLES bolic demands of the hypertrophying cardiomyocytes24 (the cross-sectional area of VEGF+/+ cardiomyocytes increased from 37 ± 2 µm2 at birth to 89 ± 8 µm2 at P12). In contrast, the capillary density did not change in VEGF120/120 mice (Fig. 2c). The 400% reduced capillary density in VEGF120/120 hearts compared with VEGF+/+ hearts was mainly due to impaired angiogenesis, as the size of VEGF120/120 myocytes at P12 (127 ± 4 µm2) was only about 40% larger than that of VEGF+/+ myocytes (89 ± 8 µm2). The defective capillary vessel formation resulted in larger intercapillary distances in VEGF120/120 hearts (15 ± 1.5 µm) than in VEGF+/+ hearts (6 ± 0.4 µm)(n = 5; P < 0.05), and in increased myocyte-to-capillary ratios at P12 in VEGF120/120 hearts (3.1 ± 0.4) compared with VEGF+/+ hearts (1.1 ± 0.03)(n = 5; P < 0.05; Fig. 2d and e), indicating that oxygen delivery to individual VEGF120/120 cardiomyocytes was reduced. The angiogenic defects (as well as the ischemia) were most profound in the subendomyocardium, consistent with previous observations that angiogenic sprouting occurs in an epicardial-to-endocardial gradient25. Capillaries in VEGF120/120 hearts were also more irregular, tortuous and slightly dilated, indicative of incomplete vessel remodeling. Expression of the VEGF receptor FLT1 was similar and that of FLK1 was only slightly reduced, whereas expression of neuropilin-1 (a VEGF164-specific receptor5) and of TIE2 and VEcadherin (other endothelial markers) was significantly reduced by P6 in VEGF120/120 compared with VEGF+/+ mice (Table and Fig. 2f and g). There were similar levels of VEGF-B in VEGF+/+ and VEGF120/120 mice, whereas levels of VEGF-C were reduced in VEGF120/120 mice by P6 (Table). Hearts from VEGF120/120 mice also contained fewer coronary vessels, which showed less smooth muscle α-actin staining (Fig. 2h–j). In addition, a smaller fraction of VEGF120/120 vessels was covered by α-actin stained smooth muscle cells at P0.5 and P3. Compared with VEGF+/+ hearts, VEGF120/120 hearts expressed reduced levels of platelet-derived growth factor-B and its receptor type-β (molecules involved in smooth muscle recruitment26), whereas levels of angiopoietin-1 (suspected to affect (peri)endothelial cells27,28) were similar (Fig. 2k). Capillary and coronary angiogenesis were also impaired in the right ventricle (not shown) and in the atria (numbers of capillaries per mm2 in the left atrium: 2,200 ± 140 in VEGF+/+ mice and 980 ± 160 in VEGF120/120 mice at P12; P < 0.05). Ischemic cardiomyopathy in VEGF120/120 mice The defects in myocardial angiogenesis in VEGF120/120 mice resulted in myocardial ischemia. Regional blood flow (ml/min per g) through the myocardium at P6 was 2.9 ± 0.4 in VEGF+/+ mice and 0.7 ± 0.3 in VEGF120/120 mice (n = 3–7; P < 0.05). ECG recordings showed ST-segment depressions and T-wave inversions in nine of ten P6 VEGF120/120 mice at rest (Fig. 3a and b). VEGF120/120 mice, but not their VEGF+/+ littermates, rapidly developed considerable ST-segment elevations when ventilation was stopped (Fig. 3c and d). Hypoxic zones, visualized by EF5 staining29, were found in VEGF120/120 but not in VEGF+/+ subendomyocardium (Fig. 3e and f). Expression of known hypoxia-inducible genes of anaerobic glycolysis30 was increased in VEGF120/120 mice (Fig. 3g). Propagation of the electrical pulse in VEGF120/120 hearts was abnormal. Indeed, the QRS interval at P6 was 12 ± 1 milliseconds in VEGF+/+ mice and 20 ± 2 milliseconds in VEGF120/120 mice (P < 0.05; Fig. 3a and b), and the QTc-interval at P6 was 150 ± 8 milliseconds in VEGF120/120 mice and 120 ± 4 milliseconds in VEGF+/+ mice (n = 8–14; P