J Mol Cell Cardiol 30, 867–877 (1998)

Intramyocardial Infusion of FGF-1 MimicsIschemic Preconditioning in PigMyocardiumPatrik Htun, Wulf D. Ito, Imo E. Hoefer, Jutta Schaper andWolfgang SchaperMax-Planck-Institute for Physiological and Clinical Research, W.G. Kerckhoff-Institute,Department of Experimental Cardiology, Benekestraße 2, D-61231 Bad Nauheim, Germany

(Received 4 September 1997, accepted in revised form 30 January 1998)

P. H, W. D. I, I. E. H, J. S W. S. Intramyocardial Infusion of FGF-1 Mimics IschemicPreconditioning in Pig Myocardium. Journal of Molecular and Cellular Cardiology (1998) 30, 867–877. Previousstudies on the mRNA and protein level suggested a cardioprotective role of FGF-1. These presumed actions ofFGF-1 and FGF-2, as well as the underlying mechanisms, were investigated in this study. Human recombinantFGF-1 (0.5 lg/ml, 20 ll/min) and FGF-2 (2 lg/ml) were applied by means of direct intramyocardial infusion (IM)for 60 min prior to a 60 min LAD-occlusion and 120 min reperfusion. Myocardial infarction compared to theregion at risk was significantly decreased by FGF-1 and FGF-2 treatment (FGF-1: 51.8±7.7%, respectively, FGF-2: 57.3±6.5% v control 83.4±2.8%, P<0.05). The increase in survival time was about 33 min, and equalledthat of ischemic preconditioning. This effect was caused by the mitogenic part of the molecule, since infusion ofa truncated version of FGF-1 (0.5–1 lg/ml), lacking mitogenicity but maintaining hemodynamic activity, did notinduce cardioprotection (78.3±0.73% v control 83.4±2.8%). Suramin (0.5 lg/ml) prevented the observedcardioprotection (77.0±1.2% v control 83.4±2.8%) proving that the cardioprotective effect is receptor-mediated.Genistein (0.5 lg/ml), an inhibitor of tyrosine kinases, abolished the cardioprotection as well (77.2±2.4% vcontrol: 83.4±2.8%). Immunohistochemical staining revealed an uptake and translocation of exogenous FGF-1to a (peri-)nuclear localization in myocytes and into non-myocytes for FGF-2. We conclude that both FGF-1 andFGF-2 are cardioprotective (FGF-1 being more active on a molar basis), and mimic ischemic preconditioning.Their actions are receptor-mediated and receptor activation is involved. Uptake and transport to a (peri-)nuclearlocalization, seems to be a pathway of minor relevance, since it could not be blocked by tyrosine kinase receptorinhibition. Tyrosine kinase-coupled receptor occupation in general is not protective as demonstrated by the lackof effect with VEGF-infusion. 1998 Academic Press Limited

K W: Growth factors; Signal transduction; Infarct size.

tyrosine receptor kinase (TRK) system as a mech-Introductionanism of protection (Sellke et al., 1996), and wehave demonstrated expression and protective actionRepetitive short-term coronary occlusions have a

cardioprotective effect against a subsequent long of TRK-ligands following brief ischemia-reperfusion(Vogt et al., 1997). Improved functional recoveryperiod of ischemia (Murry et al., 1986). This is

called ischemic preconditioning and is considered from ischemia of isolated buffer perfused rat heartsunder the influence of FGF-2 was recently reportedas an endogenous protection against myocardial

infarction for the in situ beating heart. Recent by Padua et al. (1995).In a previous study (Schaper et al., 1994), wecommunications suggest the involvement of the

Please address all correspondence to: Prof. Dr Wolfgang Schaper, Max-Planck-Institute for Physiological and Clinical Research, W.G.Kerckhoff-Institute, Department of Experimental Cardiology, Benekestraße 2, D-61231 Bad Nauheim, Germany.

0022–2828/98/040867+11 $25.00/0 mc980654 1998 Academic Press Limited

P. Htun et al.868

found changes in myocardial gene expression fol- Guide for care and use of laboratory animals pub-lished by the US National Institutes of Health.lowing brief coronary occlusions under open chest

conditions in the pig. Among other upregulatedgenes, we found an increased mRNA content of theTRK-ligands FGF-1, VEGF and insulin-like growthfactor (IGF-II). The cardioprotective effect of IGF-II Animal preparationpeptide was confirmed previously (Vogt et al., 1997).

FGF-1, also known as acidic fibroblast growth Fifty-three male castrated German landrace-typedomestic pigs with body weights betweenfactor (aFGF), is a member of the FGF family, that

consists of nine structurally related polypeptides 32.2–39.6 kg (35.9±3.8 kg) were premedicatedwith 2 mg/kg body weight i.m. azaperone and 2 mg/that play a key role in numerous aspects of em-

bryogenesis, growth, angiogenesis and cell survival kg body weight subcutaneous piritramid 30 minprior to the initiation of anesthesia with 10 mg/(Cuevas et al., 1994a, 1995; Fu et al., 1995). In

neuronal models and in the skeletal musculature, kg body weight metomidate. After endotrachealintubation, a bolus of 25 mg/kg body weight of a-a trophic and protective effect in the setting of

ischemia and reperfusion has been suggested chloralose was given intravenously. Anesthesia wasmaintained by a continuous intravenous infusion of(Cuevas et al., 1994b, 1995; Fu et al., 1995).

Because upregulation of cardiac FGF-1 mRNA was 25 mg/kg a-chloralose. The animals were ventilatedartificially with a pressure-controlled respiratorobserved following an ischemic insult, we assumed

a cardioprotective effect. This assumption was tested (Stephan Respirator ABV, F. Stephan GmbH, Quick-born, Germany), with room air enriched with 2 l/by our method of infusing the compound directly

into the myocardium. FGF-2, a related protein, was min oxygen. Arterial blood gases were analysedfrequently to guide adjustment of the respiratorincluded in our investigations as well, since it shows

similar biological activities. The advantage of our settings. Additional doses of piritramid (10 mg) weregiven i.v. every 60 min. Both internal jugular veinsmethod is that drug-related systemic hemodynamic

effects that could influence infarct size can be ex- were cannulated with polyethylene tubes for ad-ministration of saline, piritramid and a-chloralose.cluded (Vogt et al., 1996). Furthermore, we in-

vestigated the various pathways used by the FGFs Arterial sheath catheters (7F) were inserted intoboth common carotid arteries. To measure aorticthat could lead to cardioprotection. The first ques-

tion was whether the effect is receptor-mediated. blood pressure, the left sheath was advanced intothe aortic arch and connected with a StathamThis question was investigated by infusing a ligand

antagonist, in this case suramin, prior to the FGF-1 transducer (P23XL, Statham, Puerto Rico). A 5Fhigh fidelity catheter tipped manometer (Millar In-or FGF-2 infusion. Second, we investigated whether

receptor activation is necessary to induce car- struments, Houston, TX, USA) was inserted via theright common carotid artery into the left ventricledioprotection by applying genistein, a tyrosine kin-

ase inhibitor, prior to the FGF infusions. Third, to measure left-ventricular pressure and to calculateits first derivative (LV dP/dt ). The chest was openedwe investigated endocytosis of the receptor/ligand

complex and translocation to the nucleus, as de- by a midsternal thoracotomy and the heart wassuspended in a pericardial cradle. A loose ligaturescribed previously in in vitro experiments (Wied-

locha et al., 1994). And fourth, we investigated was placed halfway around the left anterior des-cending coronary artery (LAD), and was sub-whether or not the mitogenic part of the molecule

is responsible for myocardial protection. sequently tightened to occlude the vessel. In thepigs subjected to intramyocardial infusion, eight 26-gauge needles connected by tubing with a peristalticpump (Minipulse, Gilson, Germany) were placed inpairs along the LAD into the myocardium, per-pendicular to the epicardial surface. One tube wasMaterials and Methodsreserved for the infusion of Krebs–Henseleit-buffer(vehicle), which served as a control. ConcerningThe experimental protocol described in this study

was approved by the Bioethical Committee of the the test compounds suramin and genistein, similarproceedings were performed, i.e. each compoundDistrict of Darmstadt, Germany. Furthermore, all

animals in this study were handled in accordance was infused by a single line into non-ischemicmyocardium to detect potentially cytotoxic effects.with the guiding principles in care and use of

animals, as approved by the American Physiological This technique was described in detail by Vogt etal. (1996). After preparation, a stabilization periodSociety and the investigation conformed with the

aFGF and Cardioprotection 869

of 30 min was allowed and the different ex-perimental protocols were started.

Chemicals

Azaperone, metomidate and piritramid were pur-chased from Janssen Pharmaceutica, Neuss, Ger-many. a-chloralose, TTC and genistein wereobtained from Sigma Chemical Co. FGF-1 and FGF-2 were purchased from Biotrend. Suramin wasobtained from Research Biochemicals, Inc., Natick,MA, USA. All test compounds were dissolved inKrebs–Henseleit buffer (pH 7.4). TTC was dissolvedin 100 mmol/l phosphate buffer (pH 7.0). Thefluorescent zinc-cadmium sulfide microspheres (dia-meter 2–15 lm) were purchased from Duke Sci-entific Corporation, Palo Alto, CA, USA.

Experimental groups

The present study consisted of five experimentalgroups (Fig. 1). Group I (the control group) wassubjected to 60 min of occlusion and 2 h of re-

Figure 1 Experimental groups. Nine groups of animalsperfusion. In group II (FGF-1, dose: 0.5 lg/ml) andwere studied: Control animals (group I, n=6) were sub-group III (FGF-2, dose: 2.0 lg/ml), the peptides were jected to 60 min LAD-occlusion (CO) and 120 min re-

administered 60 min prior to the index ischemia of perfusion (REP). Groups II (n=7) and III (n=4) received60 min and the following reperfusion period of 2 h. various compounds: FGF-1 (0.5–1 lg/ml) and FGF-2

(2 lg/ml) by means of intramyocardial microinfusionGroup IV was treated with suramin (0.5 lg/ml), a(IM) for 60 min prior to the LAD occlusion. Groups IVnon-specific FGF antagonist, 60 min prior to the(n=6) and V (n=4) were treated with the growth factor60 min microinfusion of FGF-1 or FGF-2. Group V antagonist suramin (0.5 lg/ml) or the tyrosine kinase

was treated with genistein (0.3 lg/ml) for 60 min inhibitor genisteine (0.35 lg/ml) prior to FGF-1/2 in-prior to the FGF-1 or FGF-2 microinfusion. Cy- fusion. Group VI (n=3) was treated with the truncated

FGF-1 (0.5–1 lg/ml). To test the duration of protection,clohexyladenosine was locally infused as a positiveGroup VII (n=3) received FGF-1 prior to an extentedcontrol and Krebs–Henseleit buffer (KHB) as a neg-occlusion period (70–90 min). Animals of group VIII (n=ative control. 3) were subjected to 45 min LAD-occlusion. Group IX(n=3) received VEGF for 60 min prior to LAD occlusion.

Additional experimentswith FGF-1/-2, suramin and genistein. The activeconcentrations of suramin and genistein were alsoTo test the duration of the protective effect of FGF-1,

three additional experiments were carried out under tested regarding cytotoxicity and hemodynamic sideeffects as parameters. To test another potent tyrosine-(group VII) the influence of FGF-1 where the index

ischemia was prolonged to 90 min. All other cir- kinase-dependent growth factor, three additional ex-periments were carried out with local infusions ofcumstances of the experiment were kept the same.

Since we suspected that 60 min of index ischemia VEGF (dose range from 0.5–2 lg/ml), 60 min beforeindex ischemia (group IX).was longer than necessary to produce a transmural

infarction, we performed three additional control ex- Three additional experiments were carried out totest the anti-ischemic effect of FGF-1 v that byperiments with 45 min of index ischemia (group

VIII). Three additional experiments were done to test ischemic preconditioning. Sixty minutes of indexischemia were preceded by two 10-min occlusionthe influence of a truncated version of FGF-1 that

lacks the mitogenic part of the molecule (a kind gift of and 30-min reperfusion periods. The resulting in-farct size was compared with that of the controlDr Gimenez-Gallego, group VI). Fourteen additional

experiments were conducted as dose-finding studies group.

P. Htun et al.870

Determination of infarct size antibodies against human recombinant FGF-1, re-spectively, mouse monoclonal antibodies against

At the end of the experimental protocol, con- human recombinant FGF-2, both obtained fromSigma) for 12 h at 4°C. The sections were washedcentrated fluoresceine (1%) was injected i.v. as a

reperfusion marker (hearts with non-reperfused risk in PBS three times (3-min each) and incubatedwith the biotinylated second antibody (Biotin SP-regions were excluded). Thereafter, the LAD (at the

point of previous occlusion) and the thoracic aorta conjugated antimouse, Dianova) for 1 h. After re-peated rinsing, the third incubation was carried outwere occluded and 200 mg of zinc-cadmium fluor-

escent microspheres (Duke Scientific) in 20 ml Ring- with Cy-2-conjugated streptavidin (Rockland) for30 min. Nuclei were stained with Amino-er’s solution were injected into the ascending aorta.

After uptake of the mirospheres (at about 2 min actinomycin D (Molecular Probes, Eugene, USA)diluted 1:100; contractile proteins were stainedafter injection), the animals were killed with an i.v.

bolus of 20% potassium chloride to achieve cardiac with Phalloidin (Sigma, Chemical Co.) diluted1:200 for 30 min. After rinsing in PBS, the sectionsarrest. The heart was excised and both atria and

the right ventricle were removed. The left ventricle were covered with Mowiol (Hoechst A.G., Frankfurt,Germany) and coverslipped.was cut into slices along the pairwise inserted

microinfusion-needles perpendicular to the LAD. Omission of the first antibody served as negativecontrol to check for non-specific binding of theHeart slices were weighed and then incubated at

37°C in triphenyltetrazolium chloride (TTC) (1%) second antibody system.in PBS, pH 7.0 for 15 min. Myocardium at risk ofinfarction was identified by the presence of fluor-esceine and by the absence of fluorescent mi- Resultscrospheres at a wavelength of 366 nm. Theinfarcted area was demarcated by the absence of Exclusion criteriatetrazolium precipitation. The slices were pho-tographed under uv- and tungsten-lamp light by Perfusion sites were excluded from evaluation ifdouble exposure, and the color slides were used systolic–diastolic cardiac movements caused dis-for further planimetric evaluation. Infarct size is location of the needles, or if the TTC-staining areasexpressed as a percentage of the risk region. of protected and control tissue were not clearly

An index ischemia of 60 min followed by a demarcated by necrotic tissue in between. Suc-120 min of reperfusion leads to a confluent area of cessful countershock defibrillation was not a cri-transmural infarction. Non-infarcted myocardium terium for exclusion. In one animal treated withnear the needle tip in an index ischemia position FGF-1, countershocking caused dislocation of someis a definite sign of protection, because it is never microinfusion needles. These infusion sites wereobserved spontaneously or with KHB-infusions. excluded from evaluation. None of the animals died

of ventricular fibrillation. Countershocks were onlyapplied to the chest wall and not directly to the

Immunofluorescence microscopy heart.

In a separate experimental setting, tissue was ob-tained after a 90-min infusion period of FGF-1 and

Hemodynamic data-2, and after genistein pretreatment prior to FGF-1application (each group n=2), by rapidly excising

The hemodynamic parameters remained un-tissue around the needle in a radius of 1 cm, after thechanged during intramyocardial microinfusion ofanimals death. These tissue samples were mountedFGF-1 (Fig 3). Compared to control values, nowith Tissue Tek (OCT compound, Miles Inc., USA)significant change was observed in any measuredand cryosections with a thickness of 5 lm were cutparameters. In addition, no ventricular prematurewith a cryostat (Leica, Germany). The sections werebeats were detected during the infusion.collected on gelatine-covered glass slides, air dried

and fixed with 4% paraformaldehyde for 10 min.Thereafter, the slides were incubated in a BSA(0.5%) and glycin (0.5%)/Triton (0.05%) solution Infarct sizefor 20 min each. Rinsing in phosphate bufferedsaline (PBS, pH 7.4) was followed by incubation Infarct size after a 60-min index ischemia followed by

2 h of reperfusion is transmural as seen in Figure 2with diluted primary antibodies (mouse monoclonal

aFGF and Cardioprotection 871

left-ventricular cavity and because of the survivingsubepicardial layer that is also salvaged by diffusion.Figure 2 depicts the effect of intramyocardial layer thatis also salvaged by diffusion. Figure 2 depicts the effectof intramyocardial microinfusion of FGF-1 and FGF-2compared to the control group. Both compounds wereadministered for 60 min before index ischemia. FGF-1 induced an infarct size reduction of 51.8±7.7% vcontrol 83.4±2.8%, P<0.05 (Fig 4). To induce car-dioprotection by FGF-2 (Fig 5), a four-fold higher con-centration (2 lg/ml) was needed (57.3±6.5 v83.4±2.8%, P<0.05). Truncated FGF-1 did not in-duce cardioprotection (78.3±0.73%). Treatment

90

100

Duration of index ischemia (min)

80

60

40

20

0

IA/R

A (

%)

15 30 45 60 75

with suramin (77.0±1.2 v 83.4±2.8%) or genisteinFigure 2 Infarct areas v duration of index ischemia. The

(77.2±2.4 v 83.4±2.8%) in the presence of FGF-figure shows the shift of the infarct-time-course curve1/-2 abolished cardioprotection [Figs 6(A) and 6(B)].towards longer ischemia times by ischemic pre-

conditioning (Χ) and by FGF infusion (Ε). The effect is Index ischemia of 45 min (82.5±2.7%) was not dif-neutralized by suramin (Ο) and genistein (Η). Pre- ferent from that of 60 min (83.4±6.3%), and indexconditioned myocardium after 90 min of index ischemia ischemia of 90 min following pretreatment with FGF-is not different from control (Β).

1 (85%) was also not different from control index isch-Treatment with the growth factor antagonist suraminemia of 60 min. Intramyocardial infusion of VEGF didand the tyrosine kinase inhibitor genistein in comparison

with the FGF-1 and FGF-2 induced cardioprotection. not change infarct size following 60 min of index isch-Control, 83.4±2.8%; FGF-1, 51.8±7.7; FGF-2, emia (75.1±6.4%).57.2±6.5%; suramin, 77.0±1.2%; genistein, The growth factors were infused into 42 areas of77.2±2.4%; truncated aFGF, 78.3±0.73%.

potentially ischemic myocardium. Of these, six be-came dislocated during the experiment (de-fibrillation), and were hence excluded from analysis.

and covers 83.4±2.8% of the region at risk of in- Of the 36 remaining infusion sites, 32 showed a halofarction. It is not 100%, because of the lateral bor- of protected myocardium that was about 1 ml in vol-derzone of surviving but non-perfused tissue which is ume. This area of protected myocardium was sur-about 1–2 mm wide, because of the rim of sub- rounded by infarcted myocardium and was

always located within the region at-risk of infarction.endocardial tissue that survives by diffusion from the

10

120

0

Time (min)

80

60

40

20

2500

1002000

1500

1000

500

0

dP/dt(mmHg/s)

HR(beats/min)

AOP(mmHg)

LVP(mmHg)

5010501050

Figure 3 Hemodynamic data during IM. Systemic hemodynamics (mean±..., error bars hidden behind the usedsymbols) remained unchanged during intramyocardial microinfusion. Compared to baseline values (t=0), none of theregistered parameters were changed significantly, as determined by Bonferroni-adjustment. LVP, left-ventricular pressure;AOP, aortic pressure; HR, heart rate (Ο); dP/dt, first derivative of left-ventricular pressure (Ν); (Β) systolic; (Χ)diastolic; (Φ) EDP.

P. Htun et al.872

Figure 5 Reduction of infarcted areas by FGF-2. Treat-ment with FGF-2 significantly reduced infarcted area (IA)normalized to ischemic area (RA), as determined byFigure 4 Intramyocardial microinfusion of FGF-1. The TTC-staining and planimetry (shown in double exposureneedles for IM (arrows, 26 gauge) were placed in pairs technique).into the subsequent ischemic part of the left ventricle. The

fluorescent microspheres demarcate the none fluorescentarea of risk. After TTC-staining, myocardial protectionwas defined as stained tissue surrounding the mi-croinfusion-needles in transmurally infarcted myo-cardium. Salvaged myocardium to the right and leftof the needles, with zones of infarcted myocardium inbetween, were under the influence of other infusion sitesin slices below or above.

Figure 6 Prevention of cardioprotection. Infusion of (A) suramin and (B) genistein prior to the FGF-1/-2 treatmentprevented cardioprotection. Fluorescent microspheres demarcate the none fluorescent risk area, whereas TTC-stainingshows the infarcted area (shown in double exposure technique). The area around the needles does not show anycardioprotection.

The seven infusions sites that received KHB were Immunohistochemistryalways surrounded by infarction. The four infusionsites that received cyclohexyladenosine as positive In control myocardium, FGF-1 was localized mainly

in the extracellular space and occasionally also bycontrols showed always cardioprotection.

aFGF and Cardioprotection 873

Figure 7 Localization of FGF-1 (green), counterstaining with phalloidin (red). (A) In control tissue, endogenous FGF-1was detected in the ECM (arrow) and rarely in perinuclear localization of myocytes (double arrow). (B) Accumulationof FGF-1 after exogenous administration was in the ECM (arrow) and within numerous myocytes (double arrow). (C)Accumulation of FGF-1 after administration of genistein prior to FGF-1 application shows a similar localization as in(B).

Figure 8 Localization of FGF-2 (green). (A) In control tissue, FGF-2 was found in endothelial and interstitial cells ina perinuclear localization. (B) Accumulation of exogenous FGF-2 in the perinuclear space of interstitial cells. Thislocalization is comparable to that seen in control tissue.

P. Htun et al.874

myocytes [Fig. 7(A)]. After administration of FGF-1 are prominently expressed in neural and in cardiactissue. In previous studies, we had shown that FGF-1it was present in the perinuclear space of numerous

myocytes but also in the extracellular matrix (ECM) is constitutively expressed in the normal heart andthat its expression is upregulated in chronic isch-[Fig. 7(B)]. Similar results were obtained for pre-

treatment with genistein prior to FGF-1 infusion emia following coronary microembolization in thepig (Quinkler et al., 1989; Bernotat-Danielowski et[Fig. 7(C)]. FGF-2 was found mainly in endothelial

and interstitial cells, and, like FGF-1, in a peri- al., 1993). Increased expression following acutebrief ischemia was also observed for VEGF and IGF-nuclear localization [Figs 8(A), (B)]. Accumulation

of FGF-2 in cardiac myocytes could not be detected. II mRNA (Sharma et al., 1992; Kluge et al., 1995).Since VEGF, FGF-1/-2 and IGF-II are all tyrosineThe peptide localizations were aided for FGF-2 by

counterstaining of the nuclei with amino- kinase receptor binding ligands with trophic effectsin cultured cells and in ischemic neural tissues, weactinomycin and for FGF-1 of the cytoplasm with

phalloidin. had assumed that these ligands may exert also acardioprotective effect, which was indeed the casefor IGF-II (Vogt et al., 1997). Tyrosine kinase coupledreceptors are discussed as mediators for ischemicStatistical analysispreconditioning, and inhibition of these receptorswas found to block myocardial protection (BainesHemodynamic data are expressed as

mean±standard error of mean (...). The test of et al., 1996). Here, we show that stimulation oftyrosine kinase receptors protect the myocardium.statistical differences of the hemodynamic meas-

urements was done by Bonferroni-adjustments and This, however, is not a general property of all TKRs,since VEGF infusion has no protective effect in ourby pairwise mean differences (ANOVA, Scheffe-test).

For the statistical differences in infarct size meas- experimental model. Infused FGF not only binds toFGF receptors, but is also taken up by the cells (FGF-1urements the Kruskal–Wallis test was performed.

A P value smaller than 0.05 was considered into myocytes and FGF-2 into non-myocytes) andis transported to the perinuclear space (see Figs 7statistically significant.and 8). Immunofluorescent material in the peri-nuclear space is most probably located in lysosomes,and constitutes internalized ligand-receptor com-Discussionplexes. Genistein had no influence on this process.Both FGF-1 and FGF-2 produced cardiac protectionWe show in the present study that local in-

tramyocardial infusion of acidic (FGF-1) and basic which manifested itself as a delay of cell death. Wetested the question of the duration of the protective(FGF-2) fibroblast growth factor exert an anti-isch-

emic structural protection that is comparable in effect by increasing the index ischemia from 60 to90 min, and reperfused the myocardium thereafterintensity and duration with that obtained by isch-

emic preconditioning. This protection is receptor- for 2 h, which abolished the protection. Since a 45-min occlusion is sufficient to produce a transmuralmediated and can be blocked by suramin and

genistein, and the mitogenic epitope of the molecule infarction, the time benefit of cardioprotection ismore than 15 min but probably less than 45 min.is necessary for the effect, since a truncated version

of the molecule with a deletion of the mitogenic Thus, FGF infusion has approximately doubled thetime required for complete infarction.portion was not protective. Improved functional

recovery from ischemia of isolated buffer perfused Investigating the signal transduction, we infusedsuramin, a growth factor antagonist, prior to therat hearts under the influence of FGF-2 was pre-

viously reported by Padua et al. (1995), but our FGF-1 or FGF-2 infusion to determine whether theobserved effects are receptor mediated. Suraminreport is, to the best of our knowledge, the first to

show structural protection against lethal ischemic prevented FGF-induced cardioprotection.So far, our findings showed that the observedinsult in myocardium in vivo. A cytoprotective effect

for FGF-2 has been described previously in various cardioprotection is receptor mediated, but this doesnot necessarily imply a receptor activation, sinceanimal models of neuronal ischemia (Fisher et al.,

1995; Jiang et al., 1996; Bethel et al., 1997). Since various pathways do exist. FGF ligand-binding in-duces receptor dimerization, resulting in auto-phos-both neurons and cardiac myocytes are terminally

differentiated cells that cannot compensate cell loss phorylation of the intracellular tyrosine residues(Coughlin et al., 1988), and initiation of down-by cell division, there seems to be a concordance

in their need for cytoprotection. The FGFs can meet stream signaling via the MAP Kinases ERK-1 andERK-2 ending in the activation of transcriptionthe requirements for cytoprotection because they

aFGF and Cardioprotection 875

factors. To elucidate the role of tyrosine kinases FGF-FGFR complex to the nucleus. Wiedlocha et al.(1994) showed that stimulation of DNA synthesiswe tried to block the signal cascade by applying

genistein, a tyrosine kinase inhibitor, prior to the correlates with the transport of FGF-1 to the nucleusand is independent of tyrosine phosphorylation. InFGF infusion. We are aware that genistein might

also have some impact on PKC and PKA, but it is our experiments, FGF-1 becomes redistributed froman extracellular to a (peri-)nuclear localization inthe most potent classical tyrosine kinase inhibitor

available at the moment. The result of this study myocytes. These show a brighter staining patterncompared to the endogenous FGF-1 and to that inshowed that tyrosine kinase inhibition prevented

the cardioprotection by FGF-1 and that receptor the extracellular matrix (Weiner and Swain, 1989).Since the uptake and intracellular localization ofactivation is involved as well. Another receptor-

mediated pathway is PKC activation via the PLCc- exogenous FGF-1 after tyrosine kinase inhibitionby genistein was still evident, it is unlikely that thispathway (Burgess and Maciag, 1989; Zhan et al.,

1993). Studies performed by (Bogoyevitch et al., pathway is important for the cardioprotection.Fibroblast growth factors can ameliorate isch-1994) showed that FGF can induce an upregulation

of PKC activity, but this effect is very small compared emia-induced cell death. However, the question thatarises is by which molecular mechanisms does theto that on MAPK activity.

We provide evidence that the cytotrophic effect of cardioprotection occur? One attractive hypothesis,since the ERK-1/-2 are assumed to be involved,FGF-1 is conferred by the growth factor’s mitogenic

epitope, and that the hemodynamic component would be the phosphorylation of cytoplasmic andnuclear proteins (Davis, 1993; Clerk et al., 1994).plays no part in it. This assumption was verified in

a series of experiments infusing a truncated FGF-1 MAPKs are important mediators of signal trans-duction from the cell surface to the nucleus being(a kind gift from Dr Gimenez-Gallego), i.e. the non-

mitogenic but hemodynamically active epitope involved not only in the regulation of cell hyper-trophy but also in the response to cellular stresses(Cuevas et al., 1994a), under the same conditions

as the fully active growth factor. The results of these such as hypoxia or ischemia.With the present study, we have shown thatexperiments showed no cardioprotection, excluding

that the effect might be due to vasodilatation. the cardioprotective effect of the FGFs is receptor-mediated, but we have not been able so far toAlthough FGF-1 and FGF-2 are structurally-re-

lated polypeptides, we needed a four-fold higher completely identify the cytosolic signal transductionand the possible effector mechanisms. Our pre-concentration of FGF-2 to induce a cardioprotective

effect equivalent to FGF-1. This is interesting, in as liminary experiments have shown (Barancik et al.,1997) that brief ischemia and FGF upregulate themuch as in in-vitro experiments, both FGFs can

bind to FGFR-1 and FGFR-2 equally and, moreover, MAPKs ERK1 and 2, especially during reperfusion.These may be linked to the potential effector thatcross-phosphorylation between the heterologous

FGFs has been described (Bellot et al., 1991). Quink- may become active upon phosphorylation.ler et al. (1989) could not detect measurableamounts of FGF-2 protein in porcine heart, andCasscells et al. (1990) showed that there is quant- Limitations of the study, critique of methodologyitatively more FGF-1 than FGF-2 in the rat heart,both groups emphasizing a greater importance of Despite its obvious advantages (absence of systemic

effects, strictly local mode of action, small amountsFGF-1. Another possible explanation for the dif-ference in concentration would be that variations in of often expensive tool drugs), the method of in-

tramyocardial infusion has several shortcomings.extracellular immunoglobulin-like domains mightselectively influence receptor affinity for specific The quantification of the protective effect being

one, and the distribution of effective concentrationsmembers of the FGF family. We therefore concludethat the differences in protective concentrations being the other. The ideal way to express the pro-

tective effect is to determine the true region of risk,and immunohistochemical results may be due tothe different ligand affinities to the receptors, since as seen by the tip of the needle. Experiments with

the direct infusion of tracers that have differentthe mechanism of protection for both FGFs areprobably the same. spaces of distribution (sucrose, water, albumen,

peptides, dyes) exhibit different ‘‘risk regions’’. TheIn the present study, a perinuclear localizationof FGF-1 was observed after application of this region that is reached by the infusion settings (flow

rate, concentration) is also influenced by the lymphgrowth factor. This phenomenon is in accordancewith findings by Prudovsky et al. (1994), who flow and by the blood flow, depending on the type

of indicator used. Our validation experiments showdescribed the translocation of added FGF as

P. Htun et al.876

C W, S E, S J, K M, E S,that a volume of between 1 and 2 ml is reached by1990. Isolation, characterization and localization ofthe infusion, and is kept in a dynamic equilibriumheparin-binding growth factors in the heart. J Clinwith lymph- and blood flow during the infusion Invest 85: 433–441.

period of 60 min. For this reason, we do not infuse C A, B M, A M, S P, 1994.test substances during coronary occlusion. Differential activation of protein kinase C isoforms

by endothelin-1 and phenylephrine and subsequentWhen there is a substantial halo of protectedstimulation of p42 and p44-mitogen activated proteinmyocardium around the needle tip, the entire truekinases in ventricular myocytes cultured from neonatalrisk region of the infusion area appears to be pro-rat hearts. J Biol Chem 269: 32848–32857.

tected, but that, as discussed, is difficult to quantify. C S, B P, C L, F L, WThe degree of protection, as expressed by the con- L, 1988. Acidic and basic fibroblast growth factors

stimulate tyrosine kinase activity in vivo. J Biol Chemventional infarct size determination, is therefore263: 988–993.underestimated. It is also underestimated because

C P, C F, O S, Z M, N I,the deeper (subendocardial) layers of the risk regionG-G G, 1994a. Hypotensive activity of

are not reached by the infusate. We opted for the fibroblast growth factor. Science 254: 1208–1210.conventional expression of infarct size because the C P, R C, H O, L C,total risk region is easy to measure, and strict G-G G, 1994b. Neuroprotective effect of

acidic fibroblast growth factor on seizure-associatedstandardization of the experiment (always sixbrain damage. Neurol Res 16: 365–369.needles in the ischemic region) leads to predictable

C P, C F, G-G G, 1995. Acidicresults. Another difficulty of our method is that thefibroblast growth factor prevents post-axotomy neur-

concentration of the infused peptide is known only onal death of the newborn rat facial nerve. Neuroscinear the tip of the needle. A fall of actual con- Lett 197: 183–186.

D RJ, 1993. The mitogen-activated protein kinasecentrations farther off is to be expected. However,signal transduction pathway. J Biol Chem 268: 14553–this would mean that protection noticed at some14556.distance from the needle point must have been

F M, M ME, D T, W J, T V,caused by concentrations lower than that of the C M, F SP, 1995. Delayed treatmentinfusate, i.e. we may underestimate the efficiency with intravenous basic fibroblast growth factor reducesof the infusate concentrations, but we do not know infarct size following permanent focal cerebral ischemia

in rats. J Cereb Blood Flow Metab 15: 953–959.by how much.F X, C P, G-G G, 1995. Acidic fibro-

blast growth factor reduces rat skeletal muscle damagecaused by ischemia and reperfusion. Chin Medical JReferences108: 209–214.

J N, F SP, D T, C CG, C M,B CP, C MV, D JM, 1996. Protein tyrosineC M, 1996. Delayed intravenous administrationkinase inhibitor, genistein, blocks preconditioning inof basic fibroblast growth factor (bFGF) reduces infarctisolated rabbit hearts. Circulation 94: I–661.volume in a model of focal cerebral ischemia/re-B M, H P, M Y, Z R, Sperfusion in the rat. J Neurol Sci 139: 173–179.W, 1997. Differential regulation of distinct protein

K A, Z R, M B, V PD,kinase cascades by ischemia and ischemia/reperfusionS J, S W, 1995. Insulin-like growth factorin porcine heart. Circulation 96: I–252.II is an experimental stress inducible gene in a porcineB F, C G, K J, S J, J M,model of brief coronary occlusions. Cardiovasc Res 29:D C, 1991. Ligand-induced transphosphorylation708–716.between different FGF receptors. EMBO J 10: 2849–

M C, J R, R K, 1986. Preconditioning2854.with ischemia: a delay of lethal cell injury in ischemicB-D S, S HS, S RJ, S-myocardium. Circ 74: 1124–1136. W, 1993. Generation and localisation of mono-

P RR, S R, D NS, K E, 1995. Basicclonal antibodies against fibroblast growth factors infibroblast growth factor is cardioprotective in ischemia-ischaemic collateralised porcine myocardium. Car-reperfusion injury. Mol Cell Biochem 143: 129–135.diovasc Res 27: 1220–1228.

P I, N S, Z X, F R, X J, HB A, K JR, K RC, F SP, T-J, M W, M T, 1994. Intact functional RJ, 1997. Intravenous basic fibroblast growthfibroblast growth factor (FGF) receptor-1 trafficks nearfactor decreases brain injury resulting from focal isch-the nucleus in response to FGF-1. J Biol Chem 269:emia in cats. Stroke 28: 609–615.31720–31724.B MA, G PE, A MB, C A,

Q W, M M, B-D S,L A, M CJ, P PJ, S PH, 1994.L N, S H, S W, 1989. Isolation ofEndothelin-1 and fibroblast growth factors stimulateheparin-binding growth factors from bovine, porcinethe mitogen-activated protein kinase signaling cascadeand canine hearts. Eur J Biochem 181: 67–73.myocytes. J Biol Chem 269: 1110–1119.

S W, Z R, K A, A J, SB W, M T, 1989. The heparin-binding (fibro-H, F O, K R, W B, V O, 1994.blast) growth factor family of proteins. Annu Rev Bio-

chem 58: 575–606. Pattern of myocardial gene expression after cycles of

aFGF and Cardioprotection 877

brief coronary occlusion and reperfusion. Ann NY Acad V A, H P, K A, Z R, S W,1997. Insulin like growth factor II delays myocardialSci 723: 284–291.

S FW, W SY, S A, L JJ, L J, L J, infarction in experimental coronary occlusion. Car-diovasc Res 33: 469–477.S M, 1996. Enhanced microvascular relaxations

to VEGF and bFGF in chronically ischemic myo- W H, S J, 1989. Acidic fibroblast growth factormRNA is expressed by cardiac myocytes in culture andcardium. Am J Physiol 271: H713–H720.

S HS, W M, S S, S W, 1992. the protein is localized to the extracellular mattrix.Proc Natl Acad Sci USA 86: 2683–2687.Vascular growth in the intermittently ischemic heart:

a study on growth factors expression. In: Maragoudakis W A, F P, M I, S K, O S,1994. Dual mode of signal transduction by externallyME (ed.). Angiogenesis in Health and Disease. New York:

Plenum Press, 195–206. added acidic fibroblast growth factor. Cell 76: 1039–1051.Z X, H X, F R, M T, 1993. Long term growthV A, H P, A M, P T, S W,

1996. Myocardial infusion of tool drugs for the study factor exposure and differential tyrosine phosphorylationare required for DNA synthesis in BALB/c3T3 cells. J Biolof molecular mechanisms in ischemic preconditioning.

Basic Res Cardiol 91: 389–400. Chem 268: 9611–9620.