RESEARCH ARTICLE
THE JOURNAL OF GENE MEDICINE J Gene Med 2001; 3: 42±50.
Increasing endothelial cell permeability improves the ef®ciency of myocyte adenoviral vector infection Nathalie Nevo* Nathalie Chossat Willy Gosgnach Damien Logeart Jean-Jacques Mercadier Jean-Baptiste Michel INSERM U460, Faculte de MeÂdecine Xavier Bichat, 16 rue Henri Huchard, 75018 Paris, France *Correspondence to: N. Nevo, INSERM U460, CHU Xavier Bichat, 16 rue Henri Huchard, 75018 Paris, France. E-mail:
[email protected]
Abstract Background Gene delivery to the myocardium using blood-borne adenoviral vectors is hindered by the endothelium, which represents a barrier limiting the infection rate of underlying myocytes. However, endothelial permeability may be modulated by pharmacological agents. Methods In the present study, we modeled the endothelial barrier in vitro using a human umbilical vein endothelial cell (HUVEC) monolayer seeded on a Transwell membrane as a support and diffusion of ¯uorescent dextrans as a permeability index. We used a-thrombin (100 nM) as a pharmacological agent known to increase endothelial permeability and tested the barrier function of the endothelial cell monolayer on adenovector-mediated luciferase gene transfer to underlying isolated cardiac myocytes. Results A con¯uent HUVEC monolayer represented a considerable physical barrier to dextran diffusion; it reduced the permeability of the micropore membrane alone to ¯uorescein isothiocyanate (FITC)-labeled dextrans of molecular weights 4, 70, 150 and 2000 kDa by approximately 54, 78, 88 and 98%, respectively. a-Thrombin (100 nM) increased the permeability coef®cients (PEC) by 276, 264, 562 and 4166% for the same dextrans, respectively. A con¯uent HUVEC monolayer represented a major impediment to adenovector-mediated luciferase gene transfer to cardiac myocytes, largely reducing gene transfer ef®ciency. However thrombin induced a nine-fold increase in myocyte infection. Conclusion In our model, the endothelial cell monolayer represents a major impediment to myocyte adenovector-mediated gene transfer which can be partially improved by pharmacologically increasing endothelial permeability. The Transwell model is therefore particularly useful for testing the ef®ciency of pharmacological agents in modulating adenovector passage through the endothelial barrier. Copyright # 2000 John Wiley & Sons, Ltd. Keywords endothelium; cardiac myocytes; adenoviral vector; permeability; thrombin; gene therapy
Introduction Received: 22 May 2000 Revised: 18 September 2000 Accepted: 29 September 2000 Published online: 3 October 2000
Copyright # 2000 John Wiley & Sons, Ltd.
Gene delivery to cardiac myocytes offers new perspectives for in vivo pathophysiological studies and represents a new therapeutic approach to heart diseases [1,2]. Adenovectors have proven to be very ef®cient vectors for transferring genes to cardiac myocytes, with in vitro transfer ef®ciencies
Endothelial Permeability and Myocyte Infection
approaching 100% [3,4]. However, their use in vivo is still limited by various methodological obstacles. One major limitation in their in vivo bioavailability concerns the route of vector delivery. Intravenous systemic administration is limited by adenovector trapping in the liver and a very low transfer ef®ciency to the heart [5,6] whereas direct intramyocardical injection results in gene expression limited to a small area surrounding the needle track [7,8]. The injection of recombinant adenovectors into the coronary arteries will probably offer the best opportunity for homogenous gene transfer to the myocardium but its ef®ciency appears to be highly variable [9,10]. Based on these results and according to Fechner et al. it appears that the selective barrier between blood and tissue plays a major role in limiting blood-borne adenovector access to cardiac myocytes, which are otherwise potentially transfectable [11]. This barrier consists mainly of the endothelium [12] and multiple factors can modulate in vivo the passage of blood-borne macromolecules from the vessel lumen to the organs [13]. In this context, the development of models of this endothelial barrier with more rigorously de®ned experimental settings could help to determine conditions which would increase the passage of macromolecules such as adenovectors. To assess more precisely the role of the endothelium as a barrier to adenovector gene transfer to underlying myocytes we have established an in vitro assay system of diffusion through an endothelial cell monolayer, using the Transwell system as a support and ¯uorescent dextrans as permeability markers. We then tested the effects of a pharmacological agent, a-thrombin, on endothelial permeability, and ®nally examined the in¯uence of a-thrombin-induced increased permeability on adenovector gene transfer to underlying myocytes.
43
Materials and methods Human umbilical vein endothelial cell (HUVEC) isolation and culture Endothelial cells were isolated from fresh umbilical cord veins according to the method of Jaffe et al. [14]. Cells were cultured on ®bronectin-coated ¯asks in Medium 199 (M199; Biomedia, Boussens, France) supplemented with 10% fetal calf serum (Biomedia), 10% human serum Ãpital Bichat, Paris, (Centre de Transfusion Sanguine, Ho France), 0.1 ng/ml epidermal growth factor, 1 ng/ml basic ®broblast growth factor, 0.4% endothelial cell growth supplement/heparin (PromoCell, Heidelberg, Germany) and 1% antibiotic antimycotic solution (Sigma-Aldrich, St. Louis, MD, USA). Cells were incubated in a humidi®ed atmosphere containing 5% CO2 at 37uC. The purity of the endothelial cells was assessed by their typical cobblestone appearance under light microscopy and immunostaining with rabbit anti-human von Willebrand factor (Dako Carpinteria, CA, USA) [15].
Transwell system We used ®bronectin-coated (10 mg/cm2) Transwell polycarbonate ®lters (pore size 0.4 mm, exposed area 1 cm2; Costar, Brumath, France). HUVEC (®rst passage) were seeded at high density (150 000/cm2) on the ®lter surface. The cell-seeded Transwell membrane separates the well into two chambers (Figure 1). The upper chamber was ®lled with 0.5 ml culture medium and the lower one contained 1.5 ml of the same medium. Volumes in both chambers were adjusted to avoid hydrostatic pressure gradients. The culture medium was replaced
Figure 1. (A) Schematic representation of the Transwell system including (B) a con¯uent HUVEC monolayer seeded on a permeable membrane in the upper compartment (U) ®lled with 0.5 ml culture medium and (C) freshly isolated cardiac myocytes in the lower compartment (L) ®lled with 1.5 ml of the same medium. Scale bar=50 mm Copyright # 2000 John Wiley & Sons, Ltd.
J Gene Med 2001; 3: 42±50.
44
every day. In order to achieve a functional endothelial barrier, experiments were performed 7 days after seeding.
HUVEC monolayer permeability to FITC-dextrans Endothelial cells were deprived of cell growth factors for 24 h. Cells were pre-incubated in M199, 1% BSA (Fraction V; Sigma) for 1 h. Four different sizes of ¯uorescein isothiocyanate (FITC)-labeled dextran [molecular weights (MW) of 4, 70, 150 and 2000 kDa; Sigma] were used as an index of macromolecular diffusion. At the beginning of the experiment, 1 mg/ml FITC-dextran in 1% BSA-M199, with or without human a-thrombin (100 nM), was added to the upper chamber of the Transwell system. Samples (100 ml) were taken from the lower chamber at various time intervals (30, 60, 90 and 120 min), and the same volume of 1% BSA-M199 was replaced in this chamber to prevent ¯uid permeation due to hydrostatic pressure. The ¯uorescence was measured with a spectrophotometer (BMG Lab Technologies FLUOstar 0.403), using 480 nm and 520 nm as the excitation and emission wavelengths, respectively. The quantities of all abluminal dextrans were estimated using a standardization curve. The endothelial permeability coef®cient has been calculated for each molecular weight dextran as previously described [16]. The quantity of dextran transported from the upper to the lower chamber was transformed into a volume cleared for each time point. The average volume cleared was plotted against the time, and the slope was estimated by linear regression analysis to give the mean and standard error. The slope of the clearance curve with ®lter alone is equal to PSF, where PS=permeabilityrsurface area. The slope of the clearance curve for the wells containing the ®lter plus endothelial cells was denoted PSF+EC. The slope of the clearance curve was linear up to 2 h for all compounds. The PS value for the endothelial monolayer, PSEC, was calculated as follows: 1/PSEC=1/PSF+ECx1/PSF. As standard errors of both PSF+EC and PSF were estimated in the linear regression analysis, the standard calculated as follows: error ofq the PSEC mean was CVEC =
CVF EC2
CVF 2 where CV is the coef®cient of variation (=SEM/mean). The endothelial permeability coef®cient (PEC in cm/sec) was calculated as follows: PEC=PSEC/S where S is the surface area of the Transwell membrane.
Adenovectors Ad.CMV.luc are recombinant, replication-defective adenovectors derived from human adenovector type 5. They contain a luciferase gene driven by the cytomegalovirus (CMV) promoter. Titers of virus stocks were determined by plaque titration on the 293 cell line as described previously [17], and are expressed in plaque-forming units (PFU). The assay for the presence of replicationcompetent adenoviruses (RCA) was carried out by Copyright # 2000 John Wiley & Sons, Ltd.
N. Nevo et al.
infecting A549 cells. One RCA was detected at a dose of 7r108 PFU. Virus stocks were aliquoted in small volumes and stored in phosphate-buffered saline (PBS) with 10% glycerol at x80uC until use. All experiments were carried out using aliquots of the same virus stock.
Isolation and culture of cardiac myocytes Adult male Wistar rat (Iffa Credo, Abresle, France) ventricular myocytes were isolated as described previously [18]. Brie¯y, hearts were quickly excised and perfused retrogradely (8 ml/min) using a Langendorff apparatus with Krebs buffer (KCl 4.75 mM, NaCl 35 mM, KH2PO4 1.2 mM, Na2HPO4 16 mM, NaHCO3 25 mM, glucose 10 mM, sucrose 134 mM, HEPES 10 mM, pH 7.4) containing collagenase (0.62 IU/ml; Boehringer Mannheim, Indianapolis, IN, USA) and hyaluronidase (147 IU/ml; Sigma) at 37uC for about 12 min. After several low-speed centrifugation (10 g) and sedimentation steps, myocytes were resuspended in DMEM (Life Technologies, Cergy Pontoise, France), supplemented with 10% fetal calf serum (BioWhittaker, Emerainville, France), 4% non-essential amino acids (Life Technologies), 1 nM insulin (Sigma), 10 mM cytosine arabinose (Sigma), and antibiotics. Cells were seeded on laminincoated (10 mg/ml) 12-well plates at a density of 90 000 myocytes/well. One hour after seeding, non-attached cells (about 30%) were removed.
Adenoviral vector infection Before the adenoviral vector infection, the HUVEC monolayer permeability was tested using 70 000 Da FITC-labeled dextran. One hour before the beginning of the experiment, the culture media were replaced by 1% BSA-M199. During this time, the myocytes in the bottom wells (covered with 1.5 ml 1% BSA-M199) were topped with the Transwell membranes alone or with the HUVECseeded Transwell membranes (covered with 0.5 ml of the same medium). At the beginning of the experiment, a-thrombin (100 nM) containing 1% BSA-M199 was added to the upper chamber of HUVEC-seeded Transwell membranes, except in control conditions, for 1 h. One hour after this pretreatment, Ad.CMV.luc at 1, 10 and 100 PFU per myocyte was added to the upper chambers for 1 h. For comparison, the infection of the myocytes by direct contact was achieved in the same conditions. Then the medium was removed and each type of cell was maintained in culture, as described above, for 48 h before luciferase activity measurement.
Luciferase assay Forty-eight hours after exposure to Ad.CMV.luc, infected and control cells were washed twice in PBS (pH 7.4) and lysed in 200 ml cell culture lysis reagent r1 (Promega, Madison, WI, USA). The lysate (20 ml) was diluted 1 : 20 and the luciferase activity was determined. The luciferase J Gene Med 2001; 3: 42±50.
Endothelial Permeability and Myocyte Infection
45
assay reagent (100 ml) containing D-luciferin (1.48 mg) (Promega) was added to the sample and peak light emission was measured at 25uC using a luminometer (Berthold Bioluminat 9501) for 10 s. Background values from control cells (not infected) were subtracted for each sample. Total cell protein content was determined for each sample (DC Protein Assay; Bio-Rad, Richmond, CA, USA) and the results were normalized to yield luciferase activity in light units per milligram of cell protein.
Study of endothelial morphology The 7-day con¯uent HUVEC seeded on the micropore membranes were exposed to the different treatments cited in the `Adenoviral vector infection' section. Cells were then ®xed with 4% formaldehyde and permeabilized with 0.1% TritonX-100 (Sigma). To localize f-actin ®laments, cells were incubated with TRITC-phalloidin (tetramethylrhodamine B isothyocyanate-phalloidin) (Sigma) for 30 min at 37uC in the dark. The preparations were excited with a mercury lamp at 550 nm (TRITC ®lter; Nikon Eclipse E800 Microscope). In order to exclude toxic effects of the adenoviral vector (associated or not with a-thrombin) on the HUVEC monolayer, we used TRITC-phalloidin staining of the f-actin cytoskeleton to evaluate the integrity of the monolayer.
Statistical analysis All results were expressed as meanststandard errors of the mean (SEM). Signi®cance between two groups was calculated using the unpaired Student's t-test. One- or two-factor analysis of variance (ANOVA) was applied to the data as appropriate, and subsequent group-to-group comparisons were made using the Fisher test. Linear regression and correlation coef®cients were obtained by the least-squares method. Differences were considered signi®cant for values of p
