Download pdf - Expression of exogenous tissue factor pathway inhibitor in vivo suppresses thrombus formation in injured rabbit carotid arteries

Expression of Exogenous Tissue FactorPathway Inhibitor In Vivo Suppresses ThrombusFormation in Injured Rabbit Carotid ArteriesPaolo Golino, MD, PHD,* Plinio Cirillo, MD,* Paolo Calabro’, MD,* Massimo Ragni, MD,*Davide D’Andrea, MD,* Enrico V. Avvedimento, MD,† Francesco Vigorito, MD,* Nicola Corcione, MD,*Francesco Loffredo, BS,* Massimo Chiariello, MD*Naples and Catanzaro, Italy

OBJECTIVES The aim of the present study was to test the hypothesis that retrovirus-mediated in vivo tissuefactor pathway inhibitor (TFPI) gene transfer to the arterial wall would efficiently inhibitthrombosis without causing significant changes in systemic hemostatic variables.

BACKGROUND Acute coronary syndromes (unstable angina and acute myocardial infarction) are usuallycaused by atherosclerotic plaque rupture, with consequent activation of the coagulationcascade and circulating platelets. Tissue factor (TF) exposure represents an early event in thispathophysiologic sequence, leading to activation of the extrinsic coagulation pathway andthrombin formation. Tissue factor pathway inhibitor is a naturally occurring inhibitor of theextrinsic pathway.

METHODS In the present study, the gene coding for rabbit TFPI was inserted in a retroviral vector undercontrol of a tetracycline-inducible promoter. Replication-defective, infectious, recombinantretroviruses were used to transfect rabbit carotid arteries with either TFPI or a reportergene—green fluorescent protein (GFP).

RESULTS Retroviral-mediated arterial gene transfer of TFPI resulted in potent inhibition of intravas-cular thrombus formation in stenotic and injured rabbit carotid arteries, whereas transfectionof the contralateral carotid artery with GFP had no effect on thrombosis. No significantchanges in systemic hemostatic variables (prothrombin time and partial thromboplastin time)were observed when thrombosis was inhibited.

CONCLUSIONS These data suggest that retroviral-mediated transfection of the arterial wall with TFPI mightrepresent an attractive approach for the treatment of thrombotic disorders. (J Am CollCardiol 2001;38:569–76) © 2001 by the American College of Cardiology

Plaque rupture or ulceration represents a threatening com-plication of coronary atherosclerosis, even when the plaqueitself does not significantly obstruct the arterial lumen (1).Plaque complication is usually associated with formation ofan intracoronary thrombus, which may lead to the abruptconversion from chronic to acute coronary syndromes, suchas unstable angina, acute myocardial infarction and suddencoronary death (2).

Currently, treatment of acute coronary syndromes isbased on the systemic administration of compounds thatinhibit platelet aggregation and thrombin activity or pro-mote fibrinolysis (1). However, because these compoundsrequire systemic administration and do not act selectivelyonly at the site of thrombus formation, a major drawback ofthis approach is related to the possible occurrence of seriousside effects—namely, bleeding complications at sites distantfrom their intended site of action.

One of the key events responsible for thrombus formationafter plaque rupture is the activation of the extrinsic coag-

ulation pathway, triggered by exposure of tissue factor (TF)to flowing blood. Tissue factor, a transmembrane glycopro-tein, complexes with circulating factors VII and VIIa,permitting enzymatic activation of factors X and IX, whichare the substrates for factor VIIa (3), and ultimately leadingto thrombin formation. Several experimental (4–6) and clinical(7–9) studies have clearly established that TF-dependent acti-vation of the coagulation cascade represents an importantdeterminant of intravascular thrombus generation.

Tissue factor procoagulant activity is modulated in vivoby a natural inhibitor that is produced by endothelial cells,known as tissue factor pathway inhibitor (TFPI), a Kunitz-type serine protease inhibitor. The extrinsic coagulationpathway is inhibited by TFPI in two different steps: first,TFPI binds to the active site of factor Xa, thus blocking itsenzymatic activity. Subsequently, the complex TFPI–factorXa is capable of binding to the previously formed complexTF–factor VIIa, thus preventing further formation of factorXa (10,11). Recombinant human TFPI (rTFPI) has beensuccessfully produced and has been shown to be effective inpreventing thrombus formation in a variety of experimentalmodels (12–14). However, systemic administration of rTFPIas an antithrombotic intervention requires the use of largedoses, which may entail significant hemorrhagic risks (14).

The rapid progress of in vivo gene transfer technologieshas created powerful new tools to transfer foreign genes into

From the *Department of Internal Medicine, Division of Cardiology, University ofNaples “Federico II,” Naples, Italy; and †Department of Experimental Medicine,School of Medicine at Catanzaro, University of Reggio Calabria, Italy. Presented inpart at the 49th Annual Scientific Sessions of the American College of Cardiology,Anaheim, California, March 12–15, 2000.

Manuscript received July 28, 2000; revised manuscript received April 2, 2001,accepted April 11, 2001.

Journal of the American College of Cardiology Vol. 38, No. 2, 2001© 2001 by the American College of Cardiology ISSN 0735-1097/01/$20.00Published by Elsevier Science Inc. PII S0735-1097(01)01350-X

the cells of a variety of organs, including the vascular wall. In1989, Nabel et al. (15) and Wilson et al. (16) demonstratedthe feasibility of transfecting blood vessels with foreigndeoxyribonucleic acid (DNA) in vivo, so that local geneexpression can be readily achieved in defined segments of agiven artery. Since then, different vectors have been devel-oped to efficiently transfect target cells, including retroviral,adenoviral and direct DNA transfer (for a recent review ongene therapy, see reference 17). Thus, the aim of the presentstudy was to test the hypothesis that retrovirus-mediated invivo TFPI gene transfer to the arterial wall would efficientlyinhibit thrombosis without causing significant changes insystemic hemostatic variables.

METHODS

Preparation of recombinant retroviruses. Rabbit coronaryendothelial cells were harvested and cultured as previouslydescribed (18). Total ribonucleic acid (RNA) was extractedfrom rabbit coronary endothelial cells according to the meth-ods of Chomczynski and Sacchi (19). The complete comple-mentary DNA (cDNA) of rabbit TFPI was obtained byreverse transcription polymerase chain reaction (PCR) usingesanucleotide random primers and by PCR using the followingspecific primers: forward 59AGGGATCCAGTTTTACT-CAGTT-GATCT-39; reverse 39-CGAAGCTTGCTTTT-CTATTACAACAAAC-59, corresponding to the full-lengthrabbit TFPI molecule, including the pre-pro signal peptide.The underlined sequences correspond to BamHI and HindIIIrestriction sites, respectively. The PCR product was subclonedinto pT7 blue vector (Novagen, San Diego, California), di-gested with HindIII and BamHI and cloned into the prokary-otic expression vector, pRSET-C (Invitrogen, San Diego,California). This vector contains a sequence coding for6-histidine residues, so that a fusion protein is obtained withthe 6-histidine tag at the C-terminal. The histidine tag waslater used to differentiate the wild-type TFPI from the recom-binant protein by immunohistochemistry and by a two-antibody sandwich assay (see subsequent text).

The TFPI sequence plus the 6-histidine sequence, con-tained in the pRSET-C vector, was again amplified by PCRusing the following primers: forward 59GGGATATCA-CTCAGCATAAT-39; reverse 39CGCGGCCGCCTTT-TCTATTACAA-CAAAC-59. The underlined sequence

corresponds to a NotI restriction site. The PCR product wassubcloned into pCR 2.1 TOPO vector (Invitrogen) anddigested with NotI, thus giving rise to a DNA fragmentwith two NotI ends containing the TFPI and the histidinetag sequence. This fragment was finally cloned into aretroviral vector, pRETRO-ON (Clontech, Palo Alto, Cal-ifornia), which is a Moloney murine leukemia virus-derivedretroviral vector expressing the reverse tetracycline-controlled transactivator (rtTA) from the SV40 promoter.RtTA is a fusion of amino acids 1-207 of the tetracyclinerepressor and the negatively charged C-terminal activationdomain of the VP16 protein of herpes simplex virus. Thetetracycline-responsive element controls expression of thegene of interest cloned into the multiple cloning site. In thepresence of the tetracycline derivative (i.e., doxycycline),rtTA binds the tetracycline-responsive element and acti-vates transcription of the gene of interest, although whendoxycycline is removed from the medium, transcription isturned off in a highly dose-dependent manner (20). The59-long terminal repeat controls expression of the transcriptthat contains psi (C), the extended retroviral packagingsignal. pRETRO-ON can be transfected into a high-titerpackaging cell line and can thereby mediate production ofinfectious, replication-incompetent retroviral particles (21).The pRETRO-ON retroviral genome lacks the viral genesgag, pol and env, which are supplied by the packaging cellline. The transcript produced by the pRETRO-ON con-struct is recognized by the viral structural proteins in apackaging cell line and is packaged into infectious retroviralparticles. Because the RNA transcript packaged in theseparticles does not contain the viral genes, it cannot replicatein the target cells that it infects (21).

The DNA sequence coding for the green fluorescentprotein (GFP), as a reporter gene, was obtained by digestingpEGFP-N1 (Clontech) with NotI and BamHI and clonedinto pRETRO-ON. Both recombinant retroviruses (retro-TFPI and retro-GFP) were produced by in vitro transfec-tion of Phoenix packaging cells (purchased from AmericanType Culture Collection), yielding high-titer, recombinant,infectious, replication-defective retroviruses. Virus titer wasassessed using rabbit smooth muscle cells (SMCs) as targetcells and expressed as colony forming units (cfu)/ml. Ret-roviral stocks with a titer of 1 3 107 cfu/ml were collectedand kept at 280°C until use.In vitro infection of rabbit SMCs and induction ofrecombinant proteins. Rabbit SMCs were harvested fromthe thoracic aorta and cultivated in 100-mm tissue culturedishes in Dulbecco’s modified Eagles medium (DMEM)supplemented with 10% heat-inactivated calf serum,100 U/ml penicillin and 100 U/ml streptomycin, as previ-ously described (22). Twenty-four hours before the experi-ment, the SMCs were removed from culture dishes by briefexposure to 0.3% trypsin/0.02% egtazic acid and seededonto 24-well culture plates at a density of 3 to 5 3 104/cm2.On the day of the experiment, the medium was removedand the retrovirus-containing medium was diluted 1:2 with

Abbreviations and AcronymsCFV 5 cyclic flow variationDMEM 5 Dulbecco’s modified Eagles mediumGFP 5 green fluorescent proteinPBS 5 phosphate-buffered salinePCR 5 polymerase chain reactionPT 5 prothrombin timePTT 5 partial thromboplastin timeSMC 5 smooth muscle cellrtTA 5 reverse tetracycline-controlled transactivator(TF)PI 5 (tissue factor) pathway inhibitor

570 Golino et al. JACC Vol. 38, No. 2, 2001Expression of TFPI In Vivo August 2001:569–76

fresh DMEM containing polybrene (4 mg/ml) and added tothe SMC culture. Six hours later, the retrovirus was re-moved and fresh DMEM was added. Twenty-four hourslater, recombinant protein expression (either TFPI or GFP)was induced in infected SMCs by addition of doxycycline ata final concentration of 2 mg/ml. Conditioned media fromthese cultures were removed and used later for measurementof TFPI activity by a two-antibody sandwich assay (seesubsequent text), whereas SMCs from the same cultureswere fixed in 2% buffered paraformaldehyde at the followingtime points: 30 min and 1, 2, 4, 8, 12 and 24 h after additionof doxycycline. In SMCs infected with retro-TFPI, immu-nofluorescence was performed using a mouse monoclonalantibody against the 6-histidine tag, as the primary anti-body, and an anti-mouse immunoglobulin (Ig)G goat poly-clonal antibody was conjugated with fluorescein. The SMCsinfected with retro-GFP were directly observed under anultraviolet fluorescence microscope.Measurements of rTFPI levels in the medium ofretroviral-transfected SMCs in vitro. Levels of rTFPI inthe medium obtained from SMCs transfected in vitro withretro-TFPI were measured using a two-antibody sandwichassay. Briefly, a 20-mg/ml solution of a purified goat IgGanti-rabbit TFPI (a generous gift from Dr. Mirella Ezban,Novo Nordisk A/S, Gentofte, Denmark) was incubated inenzyme-linked immunosorbent assay microtiter plates for2 h at room temperature. After washing the plates fourtimes with phosphate-buffered saline (PBS), the plates wereblocked with a 3% bovine serum albumin solution in PBSovernight at room temperature. After washing with PBS,50 ml of the culture medium obtained from the SMCstransfected in vitro with retro-TFPI were added to the wellsand incubated for 2 h at room temperature. The wells werewashed again four times with PBS, and a 1:2,000 dilution ofa peroxidase-conjugated mouse monoclonal antibody di-rected against the 6-histidine tag was added to the wells for2 h. After extensive washing with PBS, the reaction wasdeveloped by adding tetramethylbenzidine (10 mg/ml in 0.1mol/liter sodium acetate, pH 6.0) and H2O2 0.01%. After30 min, the reaction was stopped by adding 50 ml of 1mol/liter H2SO4, and the plates were read in a microtiterplate reader at 450 nm.Local in vivo delivery of recombinant retroviruses. NewZealand White rabbits were anesthetized with ketamine(35 mg/kg) and xylazine (10 mg/kg) intramuscularly. In afirst group of six rabbits, both carotid arteries were exposedthrough a median incision of the neck, and a segment ofboth arteries was injured by gently squeezing it between apair of rubber-covered forceps. Damage of the carotidarteries was performed to induce proliferation of the cells ofthe arterial wall, because it is known that retroviruses infectproliferating cells more efficiently (17). Heparin was admin-istered (100 IU/kg, intravenous bolus) to avoid thrombosisof the damaged vessels, the surgical wounds were suturedand the animals were allowed to recover. The following day,the animals were anesthetized again, both carotid arteries

were exposed and the damaged segment was isolated fromthe circulation through atraumatic microvascular clamps. Asmall polyethylene catheter (25-gauge outer diameter) wasplaced into the carotid artery, the content of the lumen wasaspirated and the lumen was washed with 200 ml of PBS.Then, 200 ml of 1 3 107 cfu/ml retro-TFPI or retro-GFPwas injected into the clamped vascular segments and kept inplace for 30 min. Randomly, one carotid artery was infectedwith retro-TFPI and the other one with retro-GFP. After30 min, the retrovirus-containing medium was aspiratedback into the catheter, and the vascular clamps wereremoved. Heparin was administered again (100 IU/kg), andthe animals were allowed to recover.

In a second group of six animals, the carotid arteries wereinfected with retro-TFPI or retro-GFP, as previously de-scribed, except that the arterial damage was not performed.These experiments were performed to establish whether therecombinant retroviruses could also infect intact arterial walls.In vivo model of intravascular thrombus formation. Theantithrombotic effects of transfecting the arterial wall withretro-TFPI were evaluated using a rabbit model of recurrentcarotid artery thrombosis, as described in detail elsewhere(23,24). Briefly, 48 h after transfection, the rabbits wereanesthetized again with a mixture of ketamine (35 mg/kg)and xylazine (5 mg/kg) administered intramuscularly. An-esthesia was maintained during the course of the experimentby an intravenous infusion of ketamine sufficient to abolishthe corneal reflex. Through a median incision of the neck, theleft and right common carotid arteries were exposed again.Polyethylene catheters were inserted into a jugular vein and afemoral artery for both drug administration and blood pressuremonitoring. Segments of the exposed vessels distal to the sitesof transfection were injured again by gently squeezing thearteries between a pair of rubber-covered forceps. An externalplastic constrictor was placed around each damaged site.Carotid blood flow velocities in both arteries were simulta-neously and continuously measured by Doppler flow probespositioned proximal to the constrictor. After instrumentation,the animals developed cyclic fluctuations of carotid blood flow(cyclic flow variations [CFVs]), characterized by gradual de-creases of flow to almost zero values, followed by spontaneousor induced restorations of flow. Previous studies have shownthat CFVs are caused by recurrent cycles of thrombus forma-tion and subsequent dislodgment (23,24).

After observing CFVs in both carotid arteries for 30 min,doxycycline was administered as an intravenous bolus of3 mg/kg, followed by a continuous infusion of 300 mg/kgbody weight per min to induce expression of the recombi-nant proteins (i.e., TFPI and GFP). Carotid blood flowvelocities in both arteries were continuously measured untilCFVs were abolished, at which time the animals wereobserved for an additional hour.Immunohistochemical evaluation of recombinant pro-tein expression. At the end of the experiment, both carotidarteries were isolated, rinsed with saline, fixed in 4%buffered neutral formalin and embedded in paraffin. Serial

571JACC Vol. 38, No. 2, 2001 Golino et al.August 2001:569–76 Expression of TFPI In Vivo

sections (5 mm) were mounted on coated slides and, oncedeparaffinized, immunostained using a peroxidase-conjugated monoclonal antibody against the 6-histidine tag(Sigma Chemicals, St. Louis, Missouri) for 1 h at 37°C.Peroxidase activity was visualized using 3-amino-9-ethyl-carbazole (AEC, Sigma) as substrate. After immunostain-ing, the sections were lightly counterstained with hematox-ylin. As negative control samples, arterial sections wereincubated with an unrelated antibody (Anti-Xpress, Invitro-gen)—a mouse monoclonal antibody directed against theleader peptide, Asp-Leu-Tyr-Asp-Asp-Asp-Asp-Lys, in-stead of the one against the 6-histidine tag—and werefurther processed according to the same procedure describedearlier. Arterial sections were obtained from animals sub-jected or not to previous arterial injury. As an additionalinternal control, the sections were obtained from animalsunder uninduced conditions (i.e., no doxycycline infusion)or from animals receiving doxycycline infusion.

RESULTS

In vitro infection of rabbit SMCs and induction ofrecombinant proteins. Immunofluorescence performed onrabbit SMCs transfected in vitro with either a recombinant

retrovirus containing the gene of rabbit TFPI (retro-TFPI)or a control retrovirus containing a reporter gene (i.e.,retro-GFP) is shown in Figures 1 and 2. At baseline, that is,in SMCs transfected but under uninduced conditions, nodetectable expression of recombinant TFPI or GFP couldbe detected. In contrast, a time-dependent induction ofrecombinant TFPI was observed after addition of doxycy-cline to the medium, peaking at 12 to 24 h. The use of aspecific antibody directed against the 6-histidine tag,present only in the recombinant TFPI, allowed us to detectonly the recombinant protein and not the wild-type TFPI,which is known to be produced also by SMCs (25). Inaddition, no fluorescence was detected in SMCs transfectedwith retro-GFP or with a retrovirus not containing anyrecombinant gene 24 h after induction (Fig. 1), thusindicating a lack of nonspecific cross reactivity of theantibody with a cellular or retroviral component. The SMCstransfected with GFP showed a similar pattern of GFPexpression (Fig. 2).

Measurement of TFPI levels in the culture medium ofSMCs transfected with retro-TFPI, by a two-antibodysandwich assay, also revealed a time-dependent increase inTFPI levels after induction with doxycycline, which also

Figure 1. Time course of recombinant tissue factor pathway inhibitor (TFPI) expression in rabbit smooth muscle cells (SMCs) transfected in vitro witha recombinant retrovirus containing the gene of rabbit TFPI. The SMCs were probed with a monoclonal antibody directed against a 6-histidine tag presentonly on the recombinant protein, thus allowing differentiation of rTFPI from the wild-type protein. After induction of rTFPI with the addition ofdoxycycline in the culture medium, a progressive increase in rTFPI expression was observed, peaking at 12 to 24 h, although no significant expression wasobserved at baseline (i.e., uninduced conditions) or in SMCs transfected with a retrovirus containing the gene of GFP or a control retrovirus (pRETRO).See text for details.

Figure 2. Time course of green fluorescent protein (GFP) expression in rabbit smooth muscle cells (SMCs) transfected in vitro with a recombinantretrovirus containing the gene of GFP. Because GFP is spontaneously fluorescent when observed under ultraviolet light, the SMCs were directly observedunder an ultraviolet microscope. After induction of GFP with the addition of doxycycline in the culture medium, a progressive increase in GFP expressionwas observed, peaking at 12 to 24 h, although no significant expression was observed at baseline (i.e., uninduced conditions). See text for details.


peaked at 12 to 24 h (Fig. 3). Also, in this case, by using theantibody against the 6-histidine tag, we were able to measureonly the concentration of rTFPI in the medium. In addition,rTFPI levels in the medium obtained from SMCs transfectedwith retro-GFP were unmeasurable (data not shown).In vivo antithrombotic effects of retroviral TFPI trans-fection. In group I rabbits (i.e., animals subjected to vesselinjury before retroviral transfection), CFVs, which are anexpression of recurrent thrombosis, were induced in bothvessels in all animals after the second endothelial damage,followed by placement of the constrictor around the carotidartery. The CFV frequency averaged 12.3 6 4.1 and 11.4 63.4 cycles/hour in the carotid arteries transfected withretro-TFPI and retro-GFP, respectively (p 5 NS). Afterinduction of the recombinant proteins by doxycycline ad-ministration (3 mg/kg bolus plus 300 mg/kg per min), CFVswere completely inhibited only in the vessels previouslytransfected with retro-TFPI (n 5 6), whereas they re-mained unchanged in the vessels transfected with retro-GFP (n 5 6, p , 0.01) (Fig. 4 and 5). The time necessaryto achieve complete inhibition of CFVs after startingdoxycycline infusion averaged 2.5 6 0.6 h. This findingdemonstrates that the antithrombotic effects of retro-TFPIare related to the secretion of TFPI, because in the sameanimal, the contralateral vessel transfected with retro-GFP didnot show any significant change in CFVs. Similar results wereobserved in group II animals (n 5 6) that were transfectedwithout previously damaging the artery (data not shown).

The antithrombotic effects of retro-TFPI were obtainedwithout significant changes in systemic prothrombin time(PT) and partial thromboplastin time (PTT), as indicated inFigure 6.Immunohistochemical evaluation of recombinant pro-tein expression. Immunohistochemical studies aimed atidentifying the presence of rTFPI in the transfected arteriesrevealed no rTFPI expression in vessels obtained from

animals under uninduced conditions (i.e., not receivingdoxycycline) (Fig. 7A), whereas an intense and diffusepositivity was evident after administration of doxycycline(Fig. 7B). No significant differences were observed in rTFPIexpression levels in arterial sections obtained from animalssubjected or not to previous injury (Fig. 7C), indicating thatin this model, previous arterial injury was not necessary toobtain a significant level of transfection. Arterial sectionsprobed with an unrelated antibody did not show anysignificant positivity (Fig. 7D). Similarly, vessels transfectedwith retro-GFP did not show any positivity for the presenceof rTFPI, although they were spontaneously fluorescent dueto the presence of GFP (Fig. 8).

DISCUSSION

The main finding of the present study is that local arterialexpression of TFPI through retrovirus-mediated gene trans-fer is associated with complete inhibition of intravascularthrombus formation at the site of arterial damage, withoutinducing significant changes in systemic hemostatic vari-ables, such as PT and PTT. These observations also confirmthe important role of TF-induced activation of coagulationin the pathophysiology of intravascular thrombus formationat sites of arterial injury and underline the antithromboticefficacy of gene transfer techniques targeted at modulatingTF procoagulant activity.

Accumulating evidence indicates that the complex TF–factor VIIa is a key initiator of arterial thrombosis in vivo.Coronary thrombosis generally occurs at sites of a preexist-ing atherosclerotic plaque, often precipitated by fissurationor ulceration of the plaque (1). Because atheroscleroticplaques are rich with TF-synthesizing cells, such as mono-cytes, foam cells and mesenchymal-like cells, plaque rupturemay lead to exposure of TF to flowing blood (7), binding offactor VII/VIIa to TF and activation of TF-dependentcoagulation. Several studies have identified both TF antigenand activity within human atherosclerotic plaques and havesuggested that they may represent an important determinant

Figure 3. Ex vivo secretion of recombinant tissue factor pathway inhibitor(TFPI) by rabbit smooth muscle cells (SMCs). The rTFPI levels weremeasured by a two-antibody sandwich assay, using a monoclonal antibodyagainst a 6-histidine tag present only in the recombinant protein, thusexcluding the contribution of the wild-type TFPI. A progressive increase inrTFPI levels was observed, peaking at 12 to 24 h after doxycycline induction.

Figure 4. Representative carotid Doppler flow recordings obtained from arabbit after local treatment of one carotid artery with retro-tissue factorpathway inhibitor (TFPI) and the other one with retro-green fluorescentprotein (GFP). At baseline (i.e., before induction of the recombinant proteins),cyclic flow variations (CFVs) developed similarly in both carotid arteries. Incontrast, 2 h after induction with doxycycline (3 mg/kg intravenous bolus plus300 mg/kg per min), CFVs were completely inhibited only in the carotid arterypreviously transfected with retro-TFPI, although they remained unchanged inthe vessel previously transfected with retro-GFP.


of thrombogenicity after plaque rupture (8,9,26). Further-more, previous studies from our own group have demon-strated that inhibition of the TF–factor VIIa axis results inmarked antithrombotic effects (4–6).

An important regulator of TF-mediated coagulationpathway is TFPI, a multivalent protease inhibitor with threeKunitz-type domains; TFPI inhibits the TF-dependentcoagulation cascade initially by binding factor Xa throughits Kunitz II domain, and subsequently by binding to theTF–factor VIIa catalytic complex through its Kunitz Idomain (10). This prevents further production of factor IXaand Xa, as well as auto-activation of factor VII in the factorVII–TF complex (11). The formation of a quaternarycomplex—TF–factor VIIa–TFPI–factor Xa—dampens on-going coagulation and may allow modulation of thrombosisin vivo. Important evidence for TFPI as a natural antico-agulant in vivo is offered by a recent study from ourlaboratory, in which we have demonstrated that duringthrombus formation, plasma TFPI activity measured distalto the site of thrombus formation was significantly lowerthan that of samples obtained at a proximal site, suggestinga direct involvement of TFPI at the site of thrombosis (27).

Furthermore, in that study, we also observed that systemicplasma TFPI activity decreased during recurrent thrombo-sis, indicating a progressive consumption of this proteinduring the process of thrombus formation (27). Otherstudies have shown that exogenous rTFPI was effective ininhibiting intravascular thrombosis, but this effect wasachieved at doses far higher than those present physiologi-cally in plasma (12–14). Not surprisingly, considering thefactor Xa inhibitory effects of TFPI, Oltrona et al. (14)found that systemic administration of rTFPI led to amarked prolongation in PT, suggesting that rTFPI at doseseffective in preventing arterial thrombosis is associated witha substantial risk of bleeding.

The main advantage of increasing local TFPI concentra-tions by gene transfer to the arterial wall is that therapeuticTFPI levels can be achieved only where they are needed (i.e.,at the damaged arterial site where TF is exposed), withoutconcomitant prolongation of PT and PTT. In search ofeffective strategies to treat intravascular thrombosis, manyantithrombotic or fibrinolytic substances have been adminis-tered systemically. Unfortunately, such an approach, althoughproven to be effective in exerting significant inhibitory effectson thrombosis, also tends to induce systemic bleeding (28).This is obviously an important factor in the determination ofthe clinical therapeutic window and underlines the potentialadvantage of local gene transfer into the arterial wall as aneffective means of treating intravascular thrombosis.Previous studies on arterial gene transfer as antithromboticinterventions. In addition to recombinant hirudin genetransfer (29), transfection of the arterial wall with cycloox-ygenase (30), tissue-type plasminogen activator (31), nitricoxide synthase (32) and thrombomodulin gene (33) hasbeen demonstrated to be effective in inhibiting intravascularthrombosis in a variety of experimental models. Two otherstudies recently published while the present work was inprogress warrant further discussion, however. Nishida et al.(34) and Zoldhelyi et al. (35) have reported the antithrom-botic efficacy of arterial TFPI gene transfer in a rabbit andporcine model of balloon-injured carotid arteries.

However, important limitations can be identified in theselatter studies. First, Zoldhelyi et al. (35) have transfected

Figure 5. Frequency of cyclic flow variations (CFVs) (cycles/hour) in carotid arteries transfected with retro-tissue factor pathway inhibitor (TFPI) (A) andcarotid arteries transfected with retro-green fluorescent protein (GFP) (B). After induction of the recombinant proteins with doxycycline, CFVs werecompletely inhibited only in retro-TFPI transfected vessels, although they remained unchanged in the contralateral arteries transfected with retro-GFP.Each dot represents one animal.

Figure 6. The prothrombin times (PTs) and partial thromboplastin times(PTTs) measured at baseline (i.e., uninduced conditions) and at the time ofinhibition of cyclic flow variations (CFVs) after administration of doxycy-cline. No significant changes in PTs and PTTs were observed wheninduction of recombinant human tissue factor pathway inhibitor (rTFPI)exerted its antithrombotic effects.


two groups of pigs with either an adenoviral vector contain-ing the gene of human TFPI or with a control adenovirus.Thus, this protocol did not allow the investigators to useeach animal as its own control. Second, and perhaps mostimportant, transfection of the arterial wall with both adeno-viruses used in the study of Nishida et al. (34) and Zoldhelyiet al. (35) resulted in constitutive expression of rTFPI. As aconsequence, rTFPI expression was already started at thetime of induction of CFVs, resulting in the impossibility ofinducing CFVs in the vessels transfected with rTFPI.However, some differences in inducing arterial damage inTFPI-transfected vessels, compared with the control arter-ies, cannot be excluded. Finally, Nishida et al. (34) did notprovide unequivocal evidence that transfection of the vessels

was indeed associated with an increase in rTFPI secretion.In contrast, in the present study, all of these potentiallyimportant limitations have been overcome. For example, anadvantage of the retroviral vector used in the present studyis that the gene of interest is under control of a tetracycline-responsive element, so that it can be switched on in thepresence of the tetracycline derivative (i.e., doxycycline).This peculiarity of the vector allowed us to induce theexpression of rTFPI after intravascular thrombosis wasinitiated, thus using each animal as its own control. Thisalso allowed us to clearly establish that the antithromboticeffects were indeed caused by local expression of rTFPI.Two lines of evidence indicate that this was the case: first,after starting doxycycline infusion, CFVs were inhibited

Figure 7. Immunohistochemical staining for recombinant human tissue factor pathway inhibitor (rTFPI) in arterial sections obtained from carotid arteriestransfected with retro-TFPI under uninduced conditions (i.e., no infusion of doxycycline) (A) and after administration of doxycycline to an animal subjectedto a previous injury (B) or after administration of doxycycline to an animal not subjected to a previous arterial injury (C). After doxycycline infusion, a diffuseand intense reactivity was present across the arterial wall, although no staining was present under uninduced conditions. No significant differences in rTFPIexpression were observed in the arterial wall between the two groups of animals. Recombinant TFPI was probed with a monoclonal antibody against a6-histidine tag present only in the recombinant protein, thus allowing detection of rTFPI from the wild-type protein. (D) Histologic section of an arteryobtained during doxycycline infusion and probed with an unrelated mouse monoclonal antibody (Anti-Xpress, Invitrogen). No cross reactivity is presentin this section (see Methods for further details).

Figure 8. Detection of green fluorescent protein (GFP) in the arterial wall after transfection with retro-GFP and induction with doxycycline. (A) Sectionstained with hematoxylin-eosin. (B) Adjacent section not stained, observed under the ultraviolet microscope equipped with a filter specific for GFP fluorescence.


only in the carotid artery transfected with retro-TFPI,although they continued to be unchanged in the contralat-eral carotid artery transfected with retro-GFP. Second,immunohistochemistry studies using a specific antibodyagainst rTFPI that did not cross react with wild-type TFPIindicated a marked rTFPI expression in the arterial wallafter doxycycline infusion, whereas no expression could bedetected in arteries transfected with retro-GFP. Thesefindings strongly point toward a specific effect of rTFPIexpression in inhibiting intravascular thrombus formation.Finally, an additional interesting feature of retroviral vectorsis represented by the fact that they stably integrate into thegenomic DNA of target cells, giving the theoretical advan-tage over other vectors, such as adenovirus, that expressionof the transfected genes can be maintained for long periods,probably indefinitely, thus eliciting the possibility of induc-ing the transfected gene more than once, as needed. How-ever, the safety of a retroviral approach as a means oftransferring genes into humans remains to be elucidated.Conclusions. Our study provides evidence that retrovirus-mediated site-specific TFPI gene transfer into rabbit carotidarteries eliminates intravascular thrombosis without sys-temic side effects. The validity of this approach in treatinghuman thrombotic disorders needs to be determined.

Reprint requests and correspondence: Dr. Paolo Golino, Divi-sion of Cardiology, University of Naples “Federico II,” via SergioPansini 5, 80131 Naples, Italy. E-mail: [email protected].

REFERENCES

1. Badimon JJ, Zaman A, Helft G, et al. Acute coronary syndromes:pathophysiology and preventive priorities. Thromb Haemost 1999;82:997–1004.

2. Willerson J, Golino P, Eidt J, et al. Specific platelet mediators andunstable coronary artery lesions. Circulation 1989;80:198–205.

3. Nemerson Y. Tissue factor and hemostasis. Blood 1988;71:1–8.4. Pawashe AB, Golino P, Ambrosio G, et al. A monoclonal antibody

against rabbit tissue factor inhibits thrombus formation in stenoticinjured rabbit carotid arteries. Circ Res 1994;74:56–63.

5. Ragni M, Cirillo P, Pascucci I, et al. A monoclonal antibody againsttissue factor shortens tissue-plasminogen activator lysis time andprevents reocclusion in a rabbit model of carotid artery thrombosis.Circulation 1996;93:1913–20.

6. Golino P, Ragni M, Cirillo P, et al. Antithrombotic effects ofrecombinant human, active site-blocked factor VIIa in a rabbit modelof recurrent arterial thrombosis. Circ Res 1998;82:39–46.

7. Wilcox JN, Smith KM, Schwartz S, et al. Localization of tissue factorin the normal vessel wall and in the atherosclerotic plaque. Proc NatlAcad Sci USA 1989;86:2839–43.

8. Annex BH, Denning SM, Channon KM, et al. Differential expressionof tissue factor protein in directional atherectomy specimens frompatients with stable and unstable coronary syndromes. Circulation1995;91:619–22.

9. Ardissino D, Merlini PA, Ariens R, et al. Tissue factor antigen andactivity in human coronary atherosclerotic plaques. Lancet 1997;349:769–71.

10. Broze GJ Jr. TFPI and the revised theory of coagulation. Annu RevMed 1995;46:103–12.

11. Broze GJ Jr., Warren LA, Novotny WF, et al. The lipoprotein-associated coagulation inhibitor that inhibits the factor VII–tissuefactor complex also inhibits factor Xa: insight into its possiblemechanism of action. Blood 1988;71:335–43.

12. Haskel EJ, Torr SR, Day KC, et al. Prevention of arterial reocclusion

after thrombolysis with recombinant lipoprotein-associated coagula-tion inhibitor. Circulation 1991;84:821–7.

13. Jang Y, Guzman LA, Lincoff M, et al. Influence of blockade at specificlevels of the coagulation cascade on restenosis in a rabbit atheroscle-rotic femoral artery injury model. Circulation 1995;92:3041–50.

14. Oltrona L, Speidel CM, Recchia D, et al. Inhibition of tissuefactor-mediated coagulation markedly attenuates stenosis afterballoon-induced arterial injury in minipigs. Circulation 1997;96:646–52.

15. Nabel EG, Plautz G, Boyce FM, et al. Recombinant gene expressionin vivo within the endothelial cells of the arterial wall. Science1989;244:1342–4.

16. Wilson JM, Birinyi LK, Salomon RN, et al. Implantation of vasculargrafts lined with genetically modified endothelial cells. Science 1989;244:1344–6.

17. Yla-Herttuala S, Martin JF. Cardiovascular gene therapy. Lancet2000;355:213–22.

18. Cirillo P, Golino P, Ragni M, et al. A simple method for the isolation,cultivation, and characterization of endothelial cells from rabbitcoronary circulation. Thromb Res 1999;96:329–34.

19. Chomczynski P, Sacchi N. Single-step method of RNA isolation byacid guanidinium thiocynate-phenol-chloroform extraction. Anal Bio-chem 1987;162:156–9.

20. Gossen M, Bujard H. Tight control of gene expression in mammaliancells by tetracycline responsive promoters. Proc Natl Acad Sci USA1992;89:5547–51.

21. Pear W, Nolan G, Scott M, et al. Production of high-titer helper-freeretroviruses by transient transfection. Proc Natl Acad Sci USA1993;90:8392–6.

22. Cirillo P, Golino P, Ragni M, et al. Activated platelets and leucocytescooperatively stimulate smooth muscle cell proliferation and proto-oncogene expression via release of soluble growth factors. CardiovascRes 1999;43:210–7.

23. Golino P, Ambrosio G, Pascucci I, et al. Experimental carotid stenosisand endothelial injury in the rabbit: an in vivo model to studyintravascular platelet aggregation. Thromb Haemost 1992;67:302–5.

24. Folts JD, Crowell EB Jr., Rowe CG. Platelet aggregation in partiallyobstructed vessels and its elimination by aspirin. Circulation 1976;54:365–70.

25. Caplice NM, Mueske CS, Kleppe LS, et al. Expression of tissue factorpathway inhibitor in vascular smooth muscle cells and its regulation bygrowth factors. Circ Res 1998;83:1264–70.

26. Toschi V, Gallo R, Lettino M, et al. Tissue factor modulates thethrombogenicity of human atherosclerotic plaques. Circulation 1997;95:594–9.

27. Ragni M, Golino P, Cirillo P, et al. Endogenous tissue factor pathwayinhibitor modulates thrombus formation in an in vivo model of rabbitcarotid artery stenosis and endothelial injury. Circulation 2000;102:113–7.

28. Adgey AA. Bleeding complications with new antithrombotics used inischemic heart disease. Haemostasis 1996;26:237–46.

29. Rade JJ, Schulick AH, Virmani R, et al. Local adenoviral-mediatedexpression of recombinant hirudin reduces neointima formation afterarterial injury. Nature Med 1996;2:293–8.

30. Zoldhelyi P, McNatt J, Xu XM, et al. Prevention of arterial throm-bosis by adenovirus-mediated transfer of cyclooxygenase gene. Circu-lation 1996;93:10–7.

31. Dichek DA, Anderson J, Kelly AB, et al. Enhanced in vivo antithrom-botic effects of endothelial cells expressing recombinant plasminogenactivators transduced with retroviral vectors. Circulation 1996;93:301–9.

32. Chen L, Daum G, Forough R, et al. Overexpression of humanendothelial nitric oxide synthase in rat vascular smooth muscle cellsand in balloon-injured carotid artery. Circ Res 1998;82:862–70.

33. Waugh JM, Yuksel E, Li J, et al. Local overexpression of thrombo-modulin for in vivo prevention of arterial thrombosis in a rabbit model.Circ Res 1999;84:84–92.

34. Nishida T, Ueno H, Atsuchi N, et al Adenovirus-mediated localexpression of human tissue factor pathway inhibitor eliminates shearstress-induced recurrent thrombosis in the injured carotid artery of therabbit. Circ Res 1999;84:1446–52.

35. Zoldhelyi P, McNatt J, Shelat HS, et al. Thromboresistance ofballoon-injured porcine carotid arteries after local gene transfer ofhuman tissue factor pathway inhibitor. Circulation 2000;101:289–95.
