Surg Today (2005) 35:855–860DOI 10.1007/s00595-005-3042-3

Therapeutic Angiogenesis Induced by Injecting Hepatocyte GrowthFactor in Ischemic Canine Hearts

Takahiro Yamaguchi1, Yoshiki Sawa1, Yuji Miyamoto1, Toshiki Takahashi1, Chung Chung Jau1,Ismayil Ahmet1, Toshikazu Nakamura2, and Hikaru Matsuda1

1 First Department of Surgery and 2 Biomedical Research Center, Osaka University Medical School, 2-2 Yamada-oka, Suita, Osaka 565-0871,Japan

Key words Angiogenesis · Ischemic heart disease ·Revascularization · Experimental surgery · Surgicaloperation

Introduction

Most patients with angina pectoris caused by coronaryartery disease respond to treatment with antianginalmedication, coronary intervention, or coronary arterybypass surgery. However, some patients have anginathat is refractory to those treatments, generally becausetheir coronary artery disease is diffuse, especially in thedistal part of their coronary circulation. Transmyo-cardial laser revascularization (TMLR) has beengiven to alleviate angina in some patients withungraftable ischemic heart disease, resulting in relief ofangina with less nitrates.1 It is speculated that the mainmechanism of symptomatic recovery depends on theangiogenesis.1,2 Therapeutic angiogenesis induced bythe direct injection of angiogenic growth factors suchas vascular endothelial growth factor (VEGF) or fibro-blast growth factor (FGF) was been reported as apotential new therapeutic strategy for end-stage coro-nary artery disease.3 However, few studies have beendone to compare the angiogenic effects of TMLR andVEGF.4

Hepatocyte growth factor (HGF), a novel potent an-giogenic growth factor, has been identified as a cytokineand a principal mediator of mesenchymal–epithelialand mesenchymal–endothelial interactions contributingto the migration and proliferation of endothelial cells.It also promotes angiogenesis. These effects of HGFare reported to be stronger than those of VEGF orbasic fibroblast growth factor (FGF).5–8 We describedthe angiogenic effect of HGF in ischemic myocardiumin a previous study9 and speculate that it has moreadvantages as therapeutic angiogenesis than TMLR. In

AbstractPurpose. Therapeutic angiogenesis, induced by thedirect injection of angiogenic growth factors or bytransmyocardial laser revascularization (TMLR), hasshown great potential as a new therapeutic strategyfor end-stage coronary artery disease. However, nosignificant differences in angiogenic effects of TMLRand vascular endothelial growth factor (VEGF) havebeen reported. We compared the effects of theintramyocardial injection of hepatocyte growth factor(HGF), a novel angiogenic factor, with those of TMLR,by evaluating the improvement in regional blood flowand regional function in a canine heart model of chronicischemia.Methods. To create a model of chronic ischemia, weligated the left anterior descending artery (LAD) in 15beagles. We divided the dogs into three groups accord-ing to the treatment given 1 month after ligation. Fourdogs were given an intracardial injection of human re-combinant HGF (H group), six dogs were given TMLR(T group), and five dogs were used as a control (Cgroup). We compared the degree of improvement inregional blood flow and regional function 1 month afterthe treatment.Results. The regional myocardial blood flow and func-tion were significantly better in the H group than in theT or C groups (P < 0.05). Histologically, there weresignificantly more von Willebrand factor-positive cellsin the LAD region in the H group than in the T or Cgroups.Conclusion. The intramural injection of recombinanthuman HGF resulted in therapeutic angiogenesis withan intrinsic contractile state, and it may have greateradvantages than TMLR for the treatment of chronicischemic heart disease.

Reprint requests to: H. MatsudaReceived: March 16, 2004 / Accepted: January 18, 2005

856 T. Yamaguchi et al.: Angiogenesis in Ischemic Canine Hearts

the present study, we evaluated the effect of theintramyocardial injection of recombinant HGF in com-parison with that of TMLR, for improving regionalcardiac function and blood flow in chronic ischemicmyocardium.

Materials and Methods

Fifteen adult beagles (mean weight 10 kg)10 were pre-medicated with ketamine hydrochloride (25mg/kgintramuscularly) and anesthetized with intravenous so-dium pentobarbital (30 mg/kg) through a leg vein. Allthe animals were tracheally intubated and mechanicallyventilated with 50% oxygen (Ventilator 710, Siemens-Elema, Solna, Sweden). Anesthesia was maintainedwith 1%–2% sevoflurane. To create cardiac ischemia,we performed a left lateral thoracotomy through thefifth intercostal space, and ligated the left anteriordescending coronary artery (LAD) after branching thefirst diagonal branch (D1).

We performed another left lateral thoracotomy 1month later. Then four dogs were given an intramuralinfection of hepatocyte growth factor (HGF) 1mgaround the border area of the ischemic myocardium(H group), and six dogs were given transmyocardiallaser revascularization (TMLR), achieved carefullywith a CO2 laser (20 J/shot, mean 10 shots/dog), aroundthe same area (T group). The remaining five dogs wereused as the control (C) group. Human recombinantHGF was purified from the culture medium of Chinesehamster ovary cells or C-127 cells transfected with ex-pression plasmid containing human HGF cDNA.11 TheHGF solution consisted of 1 mg HGF in 1ml of salinesolution, which was injected intramuscularly through a26-gauge needle attached to a 1-ml syringe. Transmyo-cardial laser revascularization was performed as pre-viously described.12 Transmyocardial channels werecreated in the beating heart with a CO2 laser (TheHeart Laser, PLC Medical Systems, Milford, MA,USA) coupled to an articulating arm. The output en-ergy was 20J. Laser activation was synchronized withthe R wave of the electrocardiogram to avoid creatingarrhythmias by stimulating the heart during the electri-cally vulnerable ventricular repolarization period(T wave). The articulating arm of the laser was con-nected to a probe, which was placed on the epicardialsurface of the heart within the area at risk. The channelsproduced were 1 mm in diameter and 1mm apart. Aftereach channel was formed, gentle manual pressure to theepicardium was applied to secure hemostasis. Transmu-ral full-thickness channels were confirmed by the detec-tion of turbulent flow by echocardiogram as describedpreviously,11 and the average number of channels was10 ± 2.

The pericardium and the incision were closed in theusual manner, and the animals were given routine post-operative care. Cephalothin sodium (20mg/kg) wasgiven intramuscularly twice daily for 3 days after theoperation. A third left lateral thoracotomy was per-formed 1 month after the second procedure and hemo-dynamic state and cardiac function were evaluated ineach dog. A catheter was positioned in the femoralartery for blood sampling. The sonomicrometers of asingle Doppler probe were fixed on the left circumflexcoronary artery (LCX) region and the border area ofthe ischemic myocardium to measure wall thickening(%Wth; 20Hz, Crystal Biotech, Northboro, MA, USA).A Millar high-fidelity micromanometer (MPC500;Millar, Houston, TX, USA) was passed into the leftventricle via the apex to measure left ventricular pres-sure. To measure regional cardiac function, a catheterwas positioned in the left atrial appendage and about 4.5¥ 106 colored microspheres, 15.5 ± 0.2mm in diameter(Triton, San Diego, CA, USA) were injected throughthe catheter. At the same time, arterial blood was with-drawn from the femoral artery for about 2min at therate of 6 ml/min. The fully anesthetized animals werekilled by an intravenous injection of potassium chloride,and the hearts were excised and placed in cold salinesolution in preparation for histological evaluation.Standard immunohistochemical techniques using VonWillebrand factor were used to stain the tissue for vas-cular endothelium and help identify vascular structures.All animals received humane care in accordance withthe “Principles of Laboratory Animal Care” formulatedby the National Society for Medical Research and the“Guide for the Care and Use of Laboratory Animals”prepared by the National Academy of Sciences (NIHpublication 85–23, revised 1985).

Regional Cardiac Function

Regional myocardial function was assessed withsonomicrometer dimension gauges. Using a Millar high-fidelity micromanometer, the ejection phase was de-fined as the time of onset of the rise of dp/dt until 20msbefore the peak negative dp/dt. The transit time ofsound traveling between the ultrasonic crystalswas measured as the distance between the crystals.Systolic wall thickening (%Wth) was defined as themaximal systolic increase in wall thickness from theend-diastolic value and was expressed as the percentsystolic thickening fraction; that is, the ratio of systolicthickening to end-diastolic wall thickness multipliedby 100. The measurements in the LAD region areexpressed as the percent change from those in theindividual LCX region.

857T. Yamaguchi et al.: Angiogenesis in Ischemic Canine Hearts

Regional Blood Flow

Regional myocardial blood flow was measured by in-jecting approximately 4.5 ¥ 106 colored microspheres,15.5 ± 0.2 mm in diameter (Triton) into the left atrium.Spheres were suspended in 10% dextran and 0.01%Tween-80 solvent, and were agitated with a vortexmixer before injection. Reference blood samples werewithdrawn from the femoral artery catheter by defusionat 6ml/min over 2 min. Myocardial and blood sampleswere analyzed for absorbance using an UV-visiblerecording spectrophotometer (Shimadzu UV-160A,Tokyo, Japan). Regional myocardial blood flow (Qm)was computed using the formula Qm = Qr *Cm/Cr, whereQr is the reference flow rate (in ml/min), Cm is the absor-bance from the myocardial tissue sample, and Cr is theabsorbance from the reference blood sample. Eachblood flow measurement was normalized for the weightof the myocardial tissue sample. The measurements inthe LAD region are expressed as the percent changefrom those in the individual LCX region.

Assessment of Histology Samples

After cardiectomy, three central short-axis slices wereremoved from the left ventricle, each of which wasdivided into seven circumferential wedges without aseptal wall. The basal side was fixed in neutral bufferedformalin, embedded in paraffin, sectioned at thick-nesses of 5mm, and stained with hematoxylin — eosin.The apex side was used for microsphere blood flowanalysis. The other slice was embedded in OCT com-pound (Tissue Tek, Miles, Elkhart, IN, USA), quicklyfrozen in liquid N2, and stored at -80°C. Cryostat sec-tions were cut at a thickness of 5mm, dried at roomtemperature, and fixed with acetone. Endogenous per-oxidase activity was blocked by incubating the sectionswith 3% hydroperoxide. Sections were incubated withrabbit antihuman Von Willebrand factor and horserad-ish peroxidase coupled to an insert polymer backbone(Dako, Glostrup, Denmark) for 60min. Each sectionwas examined at ¥400, and the size of the area viewedwas defined as one unit area, being 440 ¥ 340mm. Thus,vessels £ 10mm in diameter were considered to becapillaries. We calculated the number of capillaries perunit area.

Statistical Analysis

Data are expressed as mean ± standard deviation. Inter-group comparisons were made using Student’s t-testwhere appropriate. Significance was defined as P valuesof less than 0.05.

Results

Hemodynamics

There were no hemodynamic changes associated withthe ligation of the LAD. During the infusion ofdobutamine, the heart rate increased from 120 ± 10 to162 ± 8 beats/min (P < 0.05), and the mean arterialblood pressure increased from 110 ± 7 to 153 ±14mmHg (P < 0.05). No significant differences werefound in baseline values among any of the experimentalgroups or their respective control. No laser-inducedarrhythmias developed.

Regional Cardiac Function and Blood Flow

By 1 month after TMLR or HGF injection, the regionalcardiac function in the H group was significantly higherthan those in the C or T groups, at 67.7% ± 19.1%,vs 21.0% ± 36.4%, and 35.8% ± 11.0%, respectively(Fig. 1). However, there was no significant differencebetween the T and C groups. The regional blood flow inthe H group was significantly higher than those in the Tgroup or the C group, at 101% ± 6%, 73% ± 8%, and59% ± 5%, respectively (P < 0.05). There was no

Fig. 1. Systolic wall thickening (%Wth) was defined as themaximal systolic increase in wall thickness from the end-diastolic value, and expressed as the “percent systolic thic-kening fraction,” being the ratio of systolic thickening toend-diastolic wall thickness multiplied by 100. The measure-ments in the left anterior descending artery (LAD) regionare expressed as the percent change from those in the indi-vidual left circumflex artery (LCX) region. By 1 month aftertransmyocardial laser revascularization (TMLR) or hepato-cyte growth factor (HGF) injection, the regional cardiac func-tion in the H group was significantly higher than that in the Cor T groups (H group, 67.7% ± 19.1%; T group, 21.0% ±36.4%; C group, 35.8% ± 11.0%)

858 T. Yamaguchi et al.: Angiogenesis in Ischemic Canine Hearts

significant difference between the T and C groups(Fig. 2).

Histology

Histological examination revealed dense fibrous tissuewith a blood-filled capillary network and small veins inthe original laser-created channels in the T-grouphearts. Conversely, the H-group hearts had more VonWillebrand factor-positive cells, the number of capillar-ies which stained with anti-Von Willebrand factorantibody being 48 ± 6 per unit area (440 by 340mm),whereas the corresponding values in the T group andthe C group were 37 ± 4/unit and 35 ± 3/unit, re-spectively (Fig. 3). There were significantly more capil-laries in the H group than in the T or C groups (Fig. 4)(P < 0.05).

Discussion

We conducted this experimental study to evaluate theeffects of angiogenesis in dogs treated with TMLR orHGF, and a control group, after 1 month of coronaryartery ligation causing ischemic myocardium, focusingon local cardiac performance and local myocardialblood flow. The regional cardiac performance was sig-nificantly better in the HGF group than in the other twogroups. The regional myocardial blood flow was alsosignificantly better in the HGF group than in the other

two groups. Histological examination revealed that thechannels in the TMLR group were occluded 1 monthafter the operation, whereas significant proliferation ofsmall blood vessels were detected in the HGF group.There was no significant difference between the TMLRand the control groups in regional cardiac performanceor regional myocardial blood flow. Thus, TMLR did noteffectively improve cardiac performance and regionalmyocardial blood flow 1 month after the operation.Conversely, intramyocardial HGF induced earlier an-giogenesis, resulting in better regional myocardialblood flow, and subsequently improving regional car-diac performance. These data suggest that the adminis-tration of human recombinant HGF may be moreeffective for angiogenesis to treat ischemic myocardiumthan TMLR after 1 month of treatment. However, theeffectiveness of TMLR may not have been evident atthis point.

Many experimental studies on TMLR have shownearly-phase occlusion of the channel created byTMLR.13 Although clinical studies reported relief ofanginal pain postoperatively, nuclear medicine exami-nations could not confirm improved myocardial bloodflow.14 Therefore, the mechanism of symptomatic recov-ery of TMLR may depend on the contribution of angio-genic factors or denervation, rather than the directperfusion of blood from the left ventricle to the cardiacmuscle.2,15 Recent studies have proven that VEGF isinvolved in the postoperative therapeutic effects ofTMLR. Expression of VEGF becomes apparent about6 weeks after laser therapy and coincides with the histo-logical evidence of angiogenesis. However, we did notobserve significant capillary growth as a result of lasertherapy in the early postoperative phase in this study.Thus, TMLR may need longer than 1 month to achievetherapeutic angiogenesis because of time-dependentangiogenic growth factors.

Experimental and clinical studies have shown thatseveral growth factors, including VEGF16,17 and FGF,18

contribute to angiogenesis; however, HGF has beenfound to be effective not only as a regenerative factorfor hepatic cells, but also for inducing angiogenesis andas an antiapoptosis factor in various organs. Moreover,VEGF has an accentuated effect on membrane perme-ability causing tissue edema, when used in the treatmentof ischemic heart disease. On the contrary, HGF has notonly a stronger angiogenic effect than VEGF, but alsoantifibrotic and antiapoptoic effects, preventing theonset of myocardial edema (data not shown). In thepresent study, we observed remarkable vessels 1 monthafter HGF infusion, which led to a significant improve-ment in regional cardiac performance and myocardialblood flow without histological evidence of interstitialedema in the myocardium. This shows that HGF in-duces more potent angiogenic effects than TMLR.

Fig. 2. Each blood flow measurement in the LAD region isexpressed as the percent change from that in the individualLCX region. By 1 month after TMLR or HGF injection, theregional blood flow in the H group was significantly higherthan that in the T group or the C group (P < 0.05), while thatin the T group was higher than that in the C group (H group,101% ± 6%; T group, 73% ± 8%; C group, 59% ± 5%)

859T. Yamaguchi et al.: Angiogenesis in Ischemic Canine Hearts

Although TMLR has been reported to relieve post-operative clinical symptoms in some patients, the chan-nels are almost all occluded immediately after theoperation, casting doubt on the efficacy of TMLR at thispoint. Moreover, TMLR procedures require expensiveequipment. In contrast, the development of therapeuticangiogenesis using recombinant growth factors mightlead to highly effective and economical therapy forserious ischemic heart disease. Further investigationson the clinical application of HGF may be required toeliminate the risk of inducing tumorigenesis, includingangioma, and exacerbating diabetic retinopathy as a

result of higher tissue or serum concentrations anddoses of HGF. Although gene therapy with HGFseems to have a prolonged and beneficial effect, thereare still ethical problems and the production of HGFis uncontrollable. From this viewpoint, recombinantHGF is more useful and convenient for clinicalapplication.

A recent study suggests that combined therapy ofTMLR and VEGF stimulates the development of suffi-cient collateral circulation to preserve cardiac function.4

We also recently reported that intramyocardial HGFimproved the effect of TMLR on the myocardial blood

Fig. 3. Dense fibrous tissue with a blood-filled capillary net-work and small veins was seen in the original laser-createdchannels. In the H group, the number of capillaries thatstained with anti-Von Willebrand factor antibody was 48 ± 6per unit area (440 by 340mm), whereas that in the T groupwas 37 ± 4/unit, and that in the C group was 35 ± 3/unit (¥200)

860 T. Yamaguchi et al.: Angiogenesis in Ischemic Canine Hearts

flow and regional cardiac function in chronic ischemicmyocardium. This combined therapy may be betterthan the combination of TMLR and VEGF, becauseHGF is a multipoint growth factor for myocardium,unlike VEGF. Further investigations are needed forclinical application.

In summary, the intramural injection of recombinantHGF induced therapeutic angiogenesis with an intrinsiccontractile state in the canine ischemic heart. The ef-fects of intramural recombinant HGF seemed to be bet-ter than those of TMLR 1 month after treatment.

References

1. Horvath KA, Cohn LH, Cooley DA, Crew JR, Frazier OH,Griffith BP, et al. Transmyocardial laser revascularization: resultsof a multicenter trial with transmyocardial laser revascularizationused as solo therapy for end-stage coronary artery disease.J Thorac Cardiovasc Surg 1997;113:645–54.

2. Horvath KA, Chiu E, Maun DC, Lomasney JW, Greene R,Pearce WH, Fullerton DA. Up-regulation of vascular endothelialgrowth factor mRNA and angiogenesis after transmyocardiallaser revascularization. Ann Thorac Surg 1999;68:825–9.

3. Symes JF, Losordo DW, Vale PR, Lathi KG, Esakof DD, IsnerJM, et al. Gene therapy with vascular endothelial growth factorfor inoperable coronary artery disease. Ann Thorac Surg1999;68:830–6.

4. Heilman CA, Attmann T, Samson P, Gobel H, Marme D,Beyersdorf F, et al. Transmyocardial laser revascularization com-bined with vascular endothelial growth factor121 (VEGF121) genetherapy for chronic myocardial ischemia — do the effects reallyadd up? Eur J Cardiothorac Surg 2003;23:74–80.

5. Sonnenberg E, Meyer D, Weidner KM, Birchmeier C. Scatterfactor/hepatocyte growth factor and its receptor, the c-mettyrosine kinase, can mediate a signal exchange between mesen-chyme and epithelia during mouse development. J Cell Biol1993;123:223–35.

6. Jennische E, Ekberg S, Matejka GL. Expression of hepatocytegrowth factor in growing and regenerating rat skeltal muscle. AmJ Physiol 1993;265:C122–8.

7. Igawa T, Matsumoto K, Kanda S, Saito Y, Nakamura T. Hepato-cyte growth factor may act as a renotropic factor for regenerationin rats with acute renal injury. Am J Physiol 1993;265:F61–9.

8. Grant SD, Kleinman HK, Goldberg ID, Bhargava MM, NickoloffBJ, Kinsella JL, et al. Scatter factor induces blood vessel forma-tion in vivo. Proc Natl Acad Sci USA 1993;90:1937–41.

9. Ahmet I, Sawa Y, Yamaguchi T, Matsuda H. Gene transfer ofhepatocyte growth factor improves angiogenesis and function ofchronic ischemic myocardium in canine heart. Ann Thorac Surg2003;75:1283–7.

10. Funatsu T, Sawa Y, Ohtake S, Takahashi T, Nakamura T,Matsuda H, et al. Therapeutic angiogenesis in the ischemic canineheart induced by myocardial injection of naked complementaryDNA plasmid encoding hepatocyte growth factor. J ThoracCardiovasc Surg 2002;124:1099–105.

11. Nakamura T, Nishizawa T, Hagiya M, Seki T, Shimonishi M,Sugimura A, et al. Molecular cloning and expression of humanhepatocyte growth factor. Nature 1989;342:440–3.

12. Horvath KA, Mannting F, Cummings N, Shernan SK, Cohn LH.Transmyocardial laser revascularization: operative techniquesand clinical results at two years. J Thorac Cardiovasc Surg1996;11:1047–53.

13. Genyk IA, Frenz M, Ott B, Walpoth BH, Schaffner T, Carrel TP.Acute and chronic effects of transmyocardial laserrevascularization in the nonischemic pig myocardium by usingthree laser systems. Lasers Surg Med 2000;27:438–50.

14. Landolfo CK, Landolfo KP, Hughes GC, Coleman ER, ColemanRB, Lowe JE. Intermediate-term clinical outcome followingtransmyocardial laser revascularization in patients with refractoryangina pectoris. Circulation 1999;100(suppl II):II128–33.

15. Sheikh TA, Allen KB, Straka SP, Heimansohn DA, Fain RL,Hutchins GD, et al. Cardiac sympathetic denervation aftertransmyocardial laser revascularization. Circulation 1999;100:135–40.

16. Su H, Lu R, Kan YW. Adeno-associated viral vector-mediatedvascular endothelial growth factor gene transfer inducesneovascular formation in ischemic heart. Proc Natl Acad Sci USA2000;97:13801–6.

17. Banai S, Jaklitsch MT, Shou M, Lazarous DF, Scheinowitz M,Biro S, et al. Angiogenic-induced enhancement of collateralblood flow to ischemic myocardium by vascular endothelialgrowth factor in dogs. Circulation 1994;89:2183–9.

18. Schumacher B, Pecher P, von Specht BU, Stegman T. Inductionof neoangiogenesis in ischemic myocardium by human growthfactors: first clinical results of a new treatment of coronary heartdisease. Circulation 1998;97:645–50.

Fig. 4. In the H group, the number of capillaries stained withanti-Von Willebrand factor antibody was 48 ± 6 per unit area(440 by 340 mm), whereas that in the T group was 37 ± 4/unit,and that in the C group was 35 ± 3/unit. The number of vonWillebrand factor-positive capillaries was significantly higherin the H group than in the T and C groups (P < 0.05)