Acute Host-Mediated Endothelial Injury After Adenoviral Gene Transfer in Normal Rabbit Arteries
Impact on Transgene Expression and Endothelial Function
Abstract—Acute injury after adenoviral vascular gene transfer remains incompletely characterized. Here, we describe the early response (≤days) in 52 New Zealand White rabbits undergoing gene transfer (β-galactosidase or empty vector) or sham procedures to both carotid arteries. After gene transfer, arteries were either left in vivo for 1 hour to 3 days (in vivo arteries) or were excised immediately after gene transfer and cultured (ex vivo arteries). Within 1 hour, in vivo arteries receiving infectious titers of ≥4×109 plaque-forming units (pfu)/mL showed endothelial activation, with an acute inflammatory infiltrate developing by 6 hours. Ex vivo arteries showed endothelial activation but no inflammatory infiltrate. There were also significant differences in transgene expression between in vivo and ex vivo arteries. Ex vivo arteries showed titer-dependent increases in β-galactosidase expression through 2×1010 pfu/mL, whereas in in vivo arteries, titers above 4×109 pfu/mL merely increased acute inflammatory response, without increasing transgene expression. In vivo arteries showed significant time- and titer-dependent impairment in endothelium-dependent relaxation, with no effect on contraction or nitroprusside-induced relaxation. Interestingly, however, if rabbits were made neutropenic with vinblastine, their arteries maintained full endothelium-dependent relaxation, even after very high titer vascular infection (up to 1×1011 pfu/mL). These findings show that recombinant adenovirus triggers an early inflammatory response, and it is the inflammatory response that in turn causes functional endothelial injury. This occurs at much lower titers than previously appreciated (though the precise threshold will undoubtedly vary between laboratories). However, titers below the inflammatory threshold produce excellent transgene expression without inflammation or vascular injury.
Recombinant adenovirus shows promise as a tool for vascular research and therapy,1 2 3 but significant limitations are clearly apparent. Most prominent are the long-term problems4,5: in vivo transgene expression declines from initially high levels to virtually nothing within ≈2 weeks,6 and an adenovirus-induced chronic inflammatory response produces significant endothelial injury and neointimal proliferation that confound data interpretation and counter any potential therapeutic effect.7 These untoward effects are largely attributable to a virus-induced host immune response.
Acute vascular effects of adenoviral gene transfer—those occurring within 3 days of gene transfer—have also been described but are less well characterized: (1) in normal rat arteries, complete endothelial denudation after high-titer (2×1011 pfu/mL) gene transfer but no apparent injury at lower titers (≤1011 pfu/mL),8 (2) in balloon-injured rat arteries, an acute medial inflammatory infiltrate and smooth muscle cell loss in arteries receiving 1011 pfu/mL,9 and (3) in normal rabbit carotid arteries, a medial inflammatory infiltrate, an impaired response to contractile agonists, and complete loss of endothelium-dependent vasodilation after moderately high-titer (4×1010 pfu/mL) gene transfer.10
These observations have led to 2 general hypotheses about adenovirus-induced acute injury. One is that the adenovirus itself is directly toxic to the vessel (“direct viral toxicity”),8 11 a view supported by the cytotoxic effects of high-multiplicity recombinant adenovirus infections in cell culture and the known in vitro cytotoxicity of some adenoviral proteins, such as the penton base.12 13 The second is that the observed injury is a host-mediated effect, largely secondary to acute inflammatory cell infiltrate, that develops after high-titer gene transfer.14 A better understanding of adenovirus-induced acute effects and the cellular mechanisms through which they are produced may be particularly important for vascular gene transfer, since these effects may have an untoward impact on transgene expression8 and the subsequent development of neointimal proliferation.7
Accordingly, we undertook the present study to define more thoroughly the acute vascular effects of adenovirus and to answer the following general questions about adenoviral gene transfer in normal rabbit arteries: (1) What is the extent of acute vascular inflammation, and what is its time course and titer dependence? (2) What is the impact of acute vascular inflammation on transgene expression? (3) To what extent does adenoviral gene transfer functionally impair the vessel, and if functional impairment is present, what is its time course and titer dependence? (4) What is the relative contribution of “direct viral toxicity” and host-mediated inflammatory effects to the observed functional deficits?
The present study shows that a titer-dependent acute inflammatory response that significantly limits transgene expression and endothelial vasomotor function develops within hours of vascular gene transfer. Moreover, our data suggest that the host response, rather than direct viral toxic effects, is largely responsible for the observed endothelial injury.
Materials and Methods
Male New Zealand White rabbits (2 to 2.5 kg) were maintained on a normal diet. Anesthesia was induced using ketamine (Ketastat, 60 mg/kg, Bristol Laboratories) and xylazine (Anased, 6 mg/kg, Lloyd Laboratories) subcutaneously. All animal care and procedures were approved by Duke University Institutional Animal Care and Use Committee and complied with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication 80-23, revised 1985).
Preparation of Adenovirus Vector
We generated a recombinant adenovirus, Ad.β-Gal, derived from the in340 mutant strain of Ad5, containing a nuclear-targeted β-Gal transgene and expression cassette in the E1 region, as previously described.15 Virus was purified from infected 293 cells by lysis in virus storage buffer (20 mmol/L Tris [pH 7.4], 150 mmol/L NaCl, 5 mmol/L KCl, and 1 mmol/L MgCl2), followed by ultracentrifugation of the lysate on a 1.3 to 1.4 g/mL cesium chloride step gradient at 100 000g for 2.5 hours at 4°C. Virus was harvested, and residual cellular RNA was removed by incubation with RNase A (100 μg/mL, Sigma Chemical Co) for 30 minutes at room temperature. Pooled virus was ultracentrifuged overnight on a second cesium chloride gradient at 4°C. Pure virus was harvested and desalted by serial gel filtration on Sepharose CL-6B spin columns (Pharmacia) in virus storage buffer. Gel-filtered virus was diluted 1:1 with 80% normal rabbit serum/20% glycerol (virus storage medium) and immediately frozen in aliquots. Viral concentration was initially estimated spectrophotometrically by absorbance at 260 nm, and infectious titer of all viral stocks was determined by at least 3 independent plaque assays on 293 cells using standard techniques (1-hour infection at 37°C, without rocking).16
Screening of Preparations to Exclude Wild-Type Adenovirus Contamination
Wild-type contamination was shown to be <1 in 107 by PCR analysis, using primers designed to amplify a 560-bp product corresponding to Ad5 base pairs 690 to 1250 (sense primer, 5′ ACGAACTGTATGATTTAGACG 3′ [Ad5 base pairs 690 to 711]; antisense primer, 5′ AGGCTCAGGTTCAGACACAGG 3′ [Ad5 base pairs 1250 to 1229]). As a positive control, serial 10-fold dilutions of Ad5 adenovirus DNA, ranging from 1 ng to 1 fg (106 to 1 copy of the in340 adenovirus genome) were used as template in the PCR. A single copy of Ad5 DNA produced a clearly visible 560-bp band. No recombinant preparations produced a positive PCR result, unless >107 copies of the recombinant adenovirus DNA were present in the PCR. In addition, no recombinant preparation produced cell death on cultured human vascular smooth muscle cells, despite high multiplicity of infection levels (≥100 pfu/cell) and observation periods of up to 10 days. In contrast, Ad5 preparations rapidly induced cell death in human vascular smooth muscle cells (within 3 days at a multiplicity of infection of 1).
Carotid Artery Gene Transfer
Rabbits underwent gene transfer to both carotid arteries in an identical manner. High-titer viral stock, maintained on dry ice until immediately before use, was thawed and diluted with DMEM/virus storage medium to ensure equal composition of virus solutions at different viral titers. Mock infections were carried out using DMEM/virus storage medium alone. At surgery, the right and left carotid arteries were carefully exposed through a midline neck incision. Heparin (700 IU IV) was administered, and gene transfer was performed by isolating the exposed vessel segment between microvascular clamps. A 24-gauge polytetrafluoroethylene (Teflon) cannula was inserted through a proximal arteriotomy, and the lumen of the vessel was cleared of blood by gentle flushing with DMEM. Approximately 200 μL of virus solution was instilled into the vessel lumen and incubated for 20 minutes. The arteriotomy was then repaired using 10/0 nylon sutures.
Induction of Neutropenia With Vinblastine
To induce neutropenia, animals received intravenous vinblastine according to previous protocols.17 18 Two doses of 1.5 mg vinblastine were given: 3 days before and, again, 1 day before surgery and gene transfer. A peripheral blood sample was taken immediately before each dose, on the day of surgery, and at the harvest of the vessels to ensure that profound neutropenia had been induced by the time of surgery and had been maintained throughout the experimental period. Peripheral blood counts were determined by Coulter counter, and absolute counts of neutrophils and lymphocytes were also determined by manual counting of blood smears. During vinblastine treatment, animals received a single daily dose of enrofloxacin (Baytril, 15 mg, Miles Inc) subcutaneously.
Vessel Harvesting and Analysis
Vessels were harvested at time points from 1 hour to 72 hours after surgery. Animals were anesthetized and heparinized (700 IU IV), and the carotid arteries were dissected free. Animals were killed with an intravenous overdose of pentobarbital sodium, and vessels were excised and washed in PBS. At the same time, ex vivo vessel rings were removed from the incubator and processed in parallel. Segments from all vessels were immediately frozen at −80°C for protein extraction or were rapidly processed for frozen sections and immunohistochemistry.
β-Gal and total protein concentration was quantified in frozen vessel segments using an ELISA kit (5 Prime-3 Prime) and a Bradford protein assay. β-Gal protein quantity was determined as nanograms β-Gal per milligram of vessel protein.
Immunohistochemistry and Image Analysis
Vessel segments were briefly equilibrated in 30% sucrose in PBS at 4°C, embedded in optimal cutting temperature compound (Miles Scientific), frozen in liquid nitrogen, and sectioned (6 μm) onto silane-coated glass microscope slides. Vessel sections were thawed, dried, and then stained for β-Gal by incubation at room temperature in X-Gal solution for 4 hours (20 mmol/L Tris [pH 7.4], 150 mmol/L NaCl, 5 mmol/L potassium ferricyanide, 5 mmol/L potassium ferrocyanide, and 2 mmol/L MgCl2 containing 0.5 mg/mL X-Gal). Immunohistochemistry to identify CD18+ leukocytes, T lymphocytes, or macrophages was performed using primary antibodies directed against rabbit CD18 (Serotec),19 rabbit CD43 (T11/135, Serotec),7 or rabbit RAM 11 (Dako),19 respectively. VCAM-1 and ICAM-1 were identified using monoclonal antibodies raised against rabbit VCAM-1 and ICAM-120 (a generous gift of Dr M. Cybulsky, University of Toronto). Immunostaining of smooth muscle cells (HHF 35, Dako) and endothelial cells (anti–von Willebrand factor, Atlantic Antibodies) was also performed. Briefly, frozen sections were fixed for 10 minutes in cold acetone and then equilibrated in PBS. Blocking solution (1.5% horse serum in PBS) was applied for 1 hour at room temperature. Antibodies were diluted in blocking solution at concentrations determined in preliminary experiments and were applied to tissue sections for 1 hour. This was followed by sequential incubation with biotinylated anti-mouse IgG (except for von Willebrand factor staining) and ABC reagent (Vectastain ABC kit, Vector Laboratories, Inc). Immune complexes were localized using the chromogenic alkaline phosphatase substrate Vector Red (Vector Laboratories, Inc). The sections were lightly counterstained with hematoxylin, dehydrated, and mounted with Permount (Fisher Scientific). In all experiments an adjacent section was incubated with an irrelevant murine IgG monoclonal antibody to serve as a negative control. Staining intensity was quantified using an image analysis system (Olympus IX70 inverted microscope, Optronics DEI-750 image capturing hardware, PowerTowerPro 180 CPU). Images were captured using Adobe Premiere and quantified using NIH Image 1.61 software.
Endothelial inflammatory cell infiltration was assessed by counting the total number of CD18+ or CD43+ leukocytes on the luminal surface of arterial cross sections. Counting was performed at ×200 magnification, in a blinded fashion, using a pointing device and a manual counter. The mean number of luminal inflammatory cells was calculated from counting multiple vessel cross sections (range, 4 to 20; median, 8 sections/vessel).
Freshly harvested vessels were cleaned of fat and connective tissues, cut into helical strips, and mounted in 30-mL organ baths containing Krebs-Henseleit buffer (mmol/L: NaCl 120, KCl 4.7, CaCl2 2.5, MgSO4 1.2, KH2PO4, NaHCO3, and glucose 5.5 mmol/L, pH 7.4) maintained at 37°C and oxygenated with 95% O2/5% CO2. Vessels were equilibrated for 60 minutes, with changes of bathing fluid every 15 minutes. Isometric tension studies were performed using a Grass model 7D polygraph. Optimal resting tension was determined in baseline studies, and the response to vasoactive drugs was then determined. Cumulative dose-response curves to PE (10−9 to 10−4 mol/L) were established. The vessels were then submaximally precontracted with PE (typically 3×10−6 mol/L), and endothelial function was evaluated by vascular relaxation to ACh (10−8 to 10−4 mol/L). NO mediation of ACh responses was confirmed by blocking ACh-induced relaxation by N-methyl-l-arginine (1 mmol/L), a specific competitive inhibitor of NO synthase. After they were washed, the vessels were again precontracted, and endothelium-independent relaxation responses to SNP (10−8 to 10−4 mol/L) were determined. Contractile responses were measured from the polygraph chart and expressed as a percentage of the maximal contraction or, for relaxations, as a percentage of the precontracted tension. Statistical significance of the differences between responses was assessed by ANOVA.
Host Factors Limit Adenovirus-Mediated Transgene Expression In Vivo
We measured β-Gal transgene expression as a function of infectious titer and compared expression between (1) “in vivo arteries,” ie, carotid arteries left in situ after gene transfer until harvest 3 days later, and (2) “ex vivo arteries,” ie, the contralateral artery from the same rabbit, which was excised immediately after gene transfer and cultured ex vivo for 3 days.
The in vivo arteries showed a striking plateau in transgene expression. β-Gal protein was detectable at the lowest titer evaluated and rose to near-maximal levels at a titer of 1.6×109 pfu/mL. However, β-Gal levels showed no significant increase from 2×109 through 1×1011 pfu/mL, and a further increase in titer (to 5×1011 pfu/mL) significantly reduced β-Gal expression. In ex vivo arteries, transgene expression was higher than in the paired contralateral in vivo arteries at all titers studied. In addition, expression in ex vivo arteries increased steadily as infecting titers were increased from 1.6×109 to 2×1010 pfu/mL. Thus, the gap between ex vivo and in vivo expression became progressively larger in this titer range (reaching a 6-fold difference at 2×1010 pfu/mL). This suggested that some host factor(s) may be limiting transgene expression in vivo.
Adenovirus-Mediated Gene Transfer Provokes Time- and Titer-Dependent Endothelial Activation
We examined arteries for evidence of endothelial activation and acute inflammation after from 1 to 72 hours after infection with recombinant adenovirus. Cryosections were evaluated by immunohistochemistry to identify VCAM-1, ICAM-1, CD18+, and CD43+ leukocytes (T lymphocytes). CD18+ leukocytes were identified as PMNs, as they had multilobed nuclei and did not stain with RAM 11, which would identify CD18+ monocytes.
ICAM-1 was detectable at 6 hours after adenoviral infection in both the adventitia and endothelium (Figure 2⇓). ICAM-1 immunostaining increased through 24 hours and remained at high levels at 72 hours. VCAM-1 was detectable as early as 1 hour and, like ICAM-1, increased through 24 hours and remained at high levels at 72 hours. However, VCAM-1 staining was confined to the endothelium. Mock-infected arteries showed no endothelial ICAM-1 or VCAM-1 expression (Figure 2⇓, bottom panels), nor did arteries infected with low-titer adenovirus (1.6×109 pfu/mL, not shown). Thus, high-titer recombinant adenoviral gene transfer induces endothelial activation very soon after infection.
The degree of endothelial activation was also titer dependent, as judged by immunohistochemistry combined with image analysis (Figure 3⇓). When the infectious titer was held below 4×109 pfu/mL, neither ICAM-1 nor VCAM-1 showed significantly increased expression above that observed in sham-infected arteries. However, as titers met or exceeded that threshold, there was a steady increase in endothelial expression of both adhesion molecules. We observed a sigmoidal relationship between the log of infectious titer used and relative staining intensity, with the steepest increase in staining intensity occurring between 4×109 and 2×1010 pfu/mL.
We next compared the degree of inflammation in mock-infected vessels and in low-titer (1.6×109 pfu/mL) and moderate-titer (2×1010 pfu/mL) adenovirus-infected arteries. Arteries were harvested 72 hours after gene transfer, then cryosectioned, and (1) stained with X-Gal to demonstrate β-Gal–expressing cells, followed by (2) immunostaining for CD18 (Figure 4⇓, all panels). We observed the following: (1) Mock-infected arteries showed no β-Gal–positive nuclei (Figure 4A⇓), whereas arteries infected with moderate-titer (Figure 4B⇓) and low-titer (Figure 4C⇓) adenovirus showed approximately equal amounts of transduced endothelial cells (≥80% endothelial cell nuclei positive for β-Gal). (2) In mock-infected arteries (Figure 4A⇓) and low-titer–infected arteries (Figure 4C⇓), endothelial inflammatory cells were observed only occasionally. (3) In contrast, vessels infected with moderate-titer adenovirus (Figure 4B⇓) showed extensive endothelial infiltration by PMNs. Thus, low-titer– and moderate-titer–infected vessels differed primarily in the degree of endothelial inflammatory cell infiltrate and were not distinguishable in terms of transgene expression.
To quantify further the extent of inflammation, we counted endothelium-adherent CD18+ and CD43+ (pan T-lymphocyte) cells in cryosections (Figure 5⇓). As was observed for endothelial activation, the degree of inflammatory infiltrate showed both time and titer dependence. A significant inflammatory infiltrate developed as early as 6 hours after high-titer adenovirus infection (1×1011 pfu/mL) and progressed to near-maximal levels by 12 hours after gene transfer (Figure 5⇓, left). Next, we examined the effect of viral titer on inflammation at 72 hours after infection. At or below 1.6×109 pfu/mL, the number of endothelial inflammatory cells was not increased relative to mock-infected vessels (Figure 5⇓, right). However, at ≥4×109 pfu/mL, the inflammatory infiltrate progressively increased. Numbers of PMNs continued to increase with titers up to 5×1011 pfu/mL, whereas CD43+ T lymphocytes were less numerous than PMNs and did not show progressive increase with viral titer.
To determine whether β-Gal transgene expression was contributing to the observed inflammation, we infected 4 arteries with 5×1010 pfu/mL of an “empty vector” (identical to Ad.β-Gal but the β-Gal transgene was absent) and 4 arteries with 5×1010 pfu/mL of Ad.β-Gal and compared the degree of adhesion molecule expression and inflammatory cell infiltration between the 2 groups (Table 1⇓). Empty vector-infected arteries and Ad.β-Gal–infected arteries were not distinguishable with regard to any of the parameters studied, suggesting that β-Gal expression did not contribute significantly to the observed inflammation.
Adenovirus-Mediated Vascular Gene Transfer Results in Functional Endothelial Injury
To assess the functional impact of adenovirus-mediated gene transfer, we excised rabbit carotid arteries 3 days after gene transfer and evaluated their response to vasoactive agonists. Regardless of the adenovirus titer received, all vessels showed essentially identical contraction in response to PE (Figure 6A⇓). Similarly, all vessels relaxed in response to treatment with the NO donor SNP (Figure 6C⇓). These observations show that adenoviral gene transfer did not harm the contraction and relaxation responses of vascular smooth muscle in the medial layer. However, moderate- and high-titer gene transfer significantly impaired endothelium-dependent relaxation (Figure 6B⇓). Mock-infected arteries relaxed by up to 70% in response to 10−5 mol/L ACh, as did arteries infected with low-titer adenovirus (3×109 pfu/mL). In contrast, arteries infected with moderate- and high-titer adenovirus (8×109 and 1×1011 pfu/mL, respectively) showed progressively impaired relaxation responses to ACh (maximal relaxation, 50% and 30% of precontracted tension, respectively). An empty vector (5×1010 pfu/mL) produced the same degree of vasomotor impairment as did high-dose Ad.β-Gal, showing that β-Gal expression itself did not cause the observed functional injury. Arteries stripped of endothelium by balloon denudation and then allowed to recover for 3 days showed minimal relaxation in response to ACh. Relaxations to ACh were inhibited by N-methyl-l-arginine, thus confirming their dependence on NO production (data not shown).
We also evaluated the time course of vasomotor function at 6, 12, 24, and 72 hours after infection with 1×1011 pfu/mL. ACh-induced relaxation was significantly impaired as early as 6 hours after infection, with further injury progressing through 72 hours (Figure 7⇓). There was no significant impact on either PE-induced contraction or SNP-induced relaxation at any time after gene transfer (data not shown). Thus, adenovirus induces time- and titer-dependent endothelial dysfunction but does not impair vascular smooth muscle function.
Neutrophil Inflammation Mediates Functional Endothelial Injury
We hypothesized that inflammatory cell infiltration may play a role in the pathogenesis of adenovirus-related endothelial injury. To investigate this possibility, we induced neutropenia in rabbits undergoing carotid artery gene transfer. Four rabbits received vinblastine (1.5 mg IV) on days 3 and 1 before surgery. All animals survived until the designated time of harvest without evidence of infection. At surgery, all vinblastine-treated rabbits were profoundly neutropenic (15-fold reduction compared with pretreatment) and remained so at the time of harvest (Table 2⇓). Peripheral blood lymphocyte and platelet counts were modestly reduced at surgery (2-fold and 2.5-fold, respectively), with further reductions noted by the time of harvest.
Vinblastine pretreatment had no apparent effect on either VCAM-1 or ICAM-1 expression after gene transfer; arteries from vinblastine-treated animals still mounted a vigorous adhesion molecule response after adenoviral gene transfer (Figure 8A⇓ and 8B⇓; compare with Figure 2⇑ at 24 hours). However, vinblastine pretreatment strikingly reduced endothelial inflammatory cell infiltration (Figure 8C⇓ and 8D⇓; compare with Figure 4B⇑). Endothelium-adherent PMNs were reduced by 25-fold (Figure 9A⇓; P<0.01), whereas endothelium-adherent T-lymphocytes were reduced by 2-fold (P=NS) 3 days after adenoviral gene transfer. Thus, vinblastine treatment did not affect endothelial cell activation and only modestly reduced T-lymphocyte infiltration but virtually abrogated endothelial PMN infiltration in infected arteries.
We next sought to determine whether the marked reduction in endothelial PMNs favorably affected endothelial vasomotor function after adenoviral gene transfer. We examined vasomotor function in arteries from 4 groups of rabbits: (1) no vinblastine, mock-infected; (2) no vinblastine, high-titer gene transfer (1×1011 pfu/mL); (3) vinblastine-treated, mock- infected; and (4) vinblastine-treated, high titer gene transfer. Vinblastine treatment did not alter sensitivity to PE and modestly reduced sensitivity to SNP, reflecting a slight reduction in sensitivity to NO (for unknown reasons). Therefore, to standardize comparisons between vinblastine-treated and untreated animals, we expressed ACh-induced relaxation as a percentage of maximal SNP-induced relaxation. With this correction, ACh-induced relaxation was virtually identical in mock-infected arteries, regardless of vinblastine treatment.
We next compared ACh-induced relaxation in arteries from untreated high-titer gene transfer rabbits with: (1) the contralateral sham-infected carotid from the same animal and (2) arteries from vinblastine, high-titer gene transfer rabbits (Figure 9B⇑). In untreated rabbits, sham-infected arteries relaxed by >75% to ACh, whereas high-titer adenovirus markedly impaired ACh-induced relaxation. But in vinblastine-treated rabbits, ACh-induced relaxation was virtually identical, regardless of whether the artery was mock-infected or uninfected. Thus, high-titer adenovirus by itself did not impair endothelial vasomotor function, if the rabbit was first made neutropenic with vinblastine. It is host-mediated inflammation, rather than direct effects of the virus, that induces the observed deficit in endothelial vasomotor function.
The present study shows that adenoviral gene transfer acutely induces endothelial activation, inflammatory cell infiltration, and endothelial dysfunction. All of these effects are directly related to infectious titer, and all become apparent within 6 hours of gene transfer. Although superficially discouraging, these findings are tempered considerably by the existence of a therapeutic window—a range of infectious titers that induce near-maximal transgene expression but little or no acute vascular injury. In our hands, this window lies roughly between 1 and 4×109 pfu/mL. Previously published therapeutic window estimates, based on expression data and morphological observations (ie, abrupt loss of transgene expression and endothelial denudation after very high-titer gene transfer8), were substantially higher; our downward revision is made possible by more detailed immunohistochemistry and studies of vasomotor function. These approaches appear to be substantially more sensitive than morphology alone as indicators of acute vascular injury.
The precise boundaries of the therapeutic window will be affected by a number of variables. For example, our boundaries apply to rabbit carotids exposed to virus for 15 minutes, but they may be quite different, for example, for other species, shorter or longer exposures, and different organ systems. Moreover, the titer determinations themselves may vary considerably between laboratories. Thus, therapeutic window boundaries must be determined on a case-by-case basis.
We show that adenovirus-induced inflammation is well developed as early as 6 hours after vascular gene transfer. This onset is too soon to attribute to 2 key adenovirus life cycle events: adenoviral viral DNA replication and late antigen expression.21 This has important implications for vascular gene transfer with newer adenoviral vectors, which are designed to reduce or eliminate viral replication and antigen expression.22 23 Our data suggest that acute vascular injury may be primarily determined by events surrounding the infection itself. If so, second-generation vectors will continue to induce titer-dependent acute inflammation, and strict attention to the therapeutic window will continue to be important.
Reduced ACh-induced vascular relaxation reflects loss of endothelial NO production,24 and that, in turn, may facilitate the long-term chronic inflammation and vascular injury that is known to occur in this model.7 NO and related molecules reduce the expression of adhesion molecules and proinflammatory cytokines,25 26 inhibit platelet aggregation and adhesion,27 28 and directly inhibit vascular smooth muscle cell proliferation29 and migration.30 Thus, the detrimental effects of early endothelial injury may extend beyond acute loss of ACh-induced relaxation and may contribute to the long-term vascular injury induced by recombinant adenovirus.
The observed vasomotor impairment has a number of striking parallels to the inflammatory cell infiltrate in terms of temporal development, dependence on titer, and spatial distribution (ie, both inflammation and vasomotor impairment affect the endothelium and spare the vascular media). These tight parallels suggest the possibility of a causal relationship. Our results with cytopenic animals further support this hypothesis: vinblastine-induced cytopenia (1) virtually eliminated endothelial cellular infiltration and (2) prevented the loss of ACh-induced vascular relaxation that otherwise occurs after exposure to high-titer adenovirus. Moreover, it is noteworthy that vinblastine treatment had no discernible effect on endothelial ICAM-1 or VCAM-1 expression. This result shows that vasomotor impairment requires additional events beyond those necessary for endothelial activation. Again, this would implicate cellular infiltration as playing a causal role.
Our observations regarding vasomotor function contrast sharply with those of Lafont et al.10 They described complete loss of endothelium-dependent relaxation and substantial loss of PE-induced contraction in rabbit ear and femoral arteries 3 days after adenoviral gene transfer at 4×1010 pfu/mL. In contrast, our carotid artery data suggest that adenoviral gene transfer does not affect vascular smooth muscle function at any titer and does not harm endothelial function if one uses titers within the therapeutic window. Finally, our data show that vasomotor dysfunction induced by high-titer adenovirus can be prevented by prior treatment with vinblastine.
Vinblastine treatment induced neutropenia, lymphopenia, and thrombocytopenia. Any of these cytopenias may play an important role in reducing endothelial injury after gene transfer. The fact that PMNs were the most affected by vinblastine treatment, both in terms of reduced peripheral counts and reduced endothelial infiltration, suggests that neutrophils have the most significant role. Certainly, their ability to release proinflammatory mediators (eg, superoxide, tumor necrosis factor-α, and interleukin-1β31) and to generate tissue injury lends support for this view. However, a more targeted experimental approach is necessary to establish definitively whether PMNs are the agents of functional endothelial injury in this model.
Important questions remain to be answered about the pathogenesis of acute effects. What aspects of recombinant adenovirus infection trigger endothelial activation? Does the initial infection represent the sole stimulus for subsequent inflammatory events, or do viral DNA replication or protein expression contribute to the progression of injury observed from 12 to 72 hours? We show that neutrophils may mediate endothelial injury, but through what mechanisms are the neutrophils recruited and activated? Vascular selectins, C5a, interleukin-8, and platelet-activating factor all mediate acute inflammation in other models,32 33 34 35 and recent studies show that wild-type and recombinant adenoviruses rapidly upregulate interleukin-8 expression in cultured cells.36 37 Any of these mediators may be involved in adenovirus-induced acute inflammation in vivo. Clarification of these mechanisms may have important implications for vascular gene therapy.
Selected Abbreviations and Acronyms
|Ad5||=||adenovirus type 5|
|ICAM-1||=||intercellular adhesion molecule-1|
|PCR||=||polymerase chain reaction|
|VCAM-1||=||vascular cell adhesion molecule-1|
This study was supported by US Public Health Service grant HL-58105 (Dr George) and an American Heart Association-North Carolina Affiliate Grant in Aid (Dr Channon). Dr Channon is a British Heart Foundation Clinical Scientist Fellow. Dr Blazing is a Clinician Scientist of the American Heart Association, and Dr George is an Established Investigator of the American Heart Association. We are grateful to Dr Myron Cybulsky (University of Toronto) for his kind gift of antibodies to rabbit VCAM-1 and ICAM-1.
- Received March 4, 1998.
- Accepted March 30, 1998.
- © 1998 American Heart Association, Inc.
Ohno T, Gordon D, San H, Pompili VJ, Imperiale MJ, Nabel GJ, Nabel EG. Gene therapy for vascular smooth muscle cell proliferation after arterial injury. Science. 1994;265:781–784.
Chang MW, Barr E, Seltzer J, Jiang Y-Q, Nabel GJ, Nabel EG, Parmacek MS, Leiden JM. Cytostatic gene therapy for vascular proliferative disorders with a constitutively active form of the retinoblastoma gene product. Science. 1995;267:518–522.
Yang Y, Nunes FA, Berencsi K, Furth EE, Gonczol E, Wilson JM. Cellular immunity to viral antigens limits E1-deleted adenoviruses for gene therapy. Proc Natl Acad Sci U S A.. 1994;91:4407–4411.
Dai Y, Schwarz EM, Gu D, Zhang W, Sarvetnick N, Verma IM. Cellular and humoral immune responses to adenoviral vectors containing factor IX gene: tolerization of factor IX and vector antigens allows for long-term expression. Proc Natl Acad Sci U S A.. 1995;92:1401–1405.
French BA, Mazur W, Ali NM, Geske RS, Finnigan JP, Rodgers GP, Roberts R, Raizner AE. Percutaneous transluminal in vivo gene transfer by recombinant adenovirus in normal porcine coronary arteries, atherosclerotic arteries, and two models of coronary restenosis. Circulation. 1994;90:2402–2413.
Newman KD, Dunn PF, Owens JW, Schulick AH, Virmani R, Sukhova G, Libby P, Dichek DA. Adenovirus-mediated gene transfer into normal rabbit arteries results in prolonged vascular cell activation, inflammation, and neointimal hyperplasia. J Clin Invest. 1995;96:2955–2965.
Schulick AH, Dong G, Newman KD, Virmani R, Dichek DA. Endothelium-specific in vivo gene transfer. Circ Res. 1995;77:475–485.
Schulick AH, Newman KD, Virmani R, Dichek DA. In vivo gene transfer into injured carotid arteries: optimization and evaluation of acute toxicity. Circulation. 1995;91:2407–2414.
Graham FL, Prevec L. Methods in Molecular Biology, Volume 7: Gene Transfer and Expression Protocols. Clifton, NJ: Humana Press Inc; 1991:109.
Rosenshein MS, Price TH, Dale DC. Neutropenia, inflammation, and the kinetics of transfused neutrophils in rabbits. J Clin Invest. 1979;64:580–585.
Langdale LA, Flaherty LC, Liggitt HD, Harlan JM, Rice CL, Winn RK. Neutrophils contribute to hepatic ischemia-reperfusion injury by a CD18-independent mechanism. J Leukoc Biol. 1993;53:511–517.
Tanaka H, Sukhova GK, Swanson SJ, Glinton SK, Ganz P, Cybulsky M, Libby P. Sustained activation of vascular cells and leukocytes in the rabbit aorta after balloon injury. Circulation. 1993;88:1788–1803.
Horwitz MS. Adenoviridae and their replication. In: Fields BN, Knipe DM, ed. Virology. New York, NY: Raven Press Publishers; 1990:1679–1721.
Engelhardt JF, Ye X, Doranz B, Wilson JM. Ablation of E2A in recombinant adenoviruses improves transgene persistence and decreases inflammatory response in mouse liver. Proc Natl Acad Sci U S A.. 1994;91:6196–6200.
Rees DD, Palmer RM, Moncada S. Role of endothelium-derived nitric oxide in the regulation of blood pressure. Proc Natl Acad Sci U S A.. 1989;86:3375–3378.
Niu X, Smith CW, Kubes P. Intracellular oxidative stress induced by nitric oxide synthesis inhibition increases endothelial cell adhesion to neutrophils. Circ Res. 1994;74:1133–1140.
De Caterina R, Libby P, Peng H-B, Thannickal VJ, Rajavashisth TB, Gimbrone MA Jr, Shin WS, Liao JK. Nitric oxide decreases cytokine-induced endothelial activation: nitric oxide selectively reduces endothelial expression of adhesion molecules and proinflammatory cytokines. J Clin Invest. 1995;96:60–68.
Radomski MW, Palmer RM, Moncada S. An L-arginine/nitric oxide pathway present in human platelets regulates aggregation. Proc Natl Acad Sci U S A.. 1990;87:5193–5197.
Garg UC, Hassid A. Nitric oxide-generating vasodilators and 8-bromo cyclic guanosine monophosphate inhibit mitogenesis and proliferation of cultured rat vascular smooth muscle cells. J Clin Invest. 1989;83:1774–1777.
Sarkar R, Meinberg EG, Stanley JC, Gordon D, Webb RC. Nitric oxide reversibly inhibits the migration of cultured vascular smooth muscle cells. Circ Res. 1996;78:225–230.
Lorant DE, Topham MK, Whatley RE, McEver RP, McIntyre TM, Prescott SM, Zimmerman GA. Inflammatory roles of P-selectin. J Clin Invest. 1993;92:559–570.
Foreman KE, Vaporciyan AA, Bonish BK, Jones ML, Johnson KJ, Glovsky MM, Eddy SM, Ward PA. C5a-induced expression of P-selectin in endothelial cells. J Clin Invest. 1994;94:1147–1155.
Ivey CL, Williams FM, Collins PD, Jose PJ, Williams TJ. Neutrophil chemoattractants generated in two phases during reperfusion of ischemic myocardium in the rabbit: evidence for a role for C5a and interleukin-8 [comment]. J Clin Invest. 1995;95:2720–2728.
Bruder JT, Kovesdi I. Adenovirus infection stimulates the Raf/MAPK signaling pathway and induces interleukin-8 expression. J Virol. 1997;71:398–404.