Adenoviral Vector Transfection into the Pulmonary Epithelium after Cecal Ligation and Puncture in Rats

All animal studies were approved by the University Laboratory Animal Resources committee (University of Pennsylvania, Philadelphia, PA) and conformed to National Institutes of Health standards. During methoxyflurane anesthesia, fulminant sepsis was induced in male Sprague-Dawley rats (Charles River, Boston, MA; weight, 250–275 g) using 2CLP with an 18-gauge needle as previously described. 5SO animals and animals not undergoing abdominal surgery served as controls. After the procedure, rats underwent fluid resuscitation with 40 ml/kg of subcutaneously injected sterile saline. Animals were awakened and allowed free access to water and food. Fluid resuscitation was repeated every 24 h until they were killed. At 24, 48, and 72 h after surgery, respiratory rate was determined, and animals were reanesthetized with 40 mg/kg intraperitoneal pentobarbital. Blood for arterial oxygen tension (Pao²) analysis was obtained, and the rats were killed via  exsanguination. A separate group of animals were killed at 0 (three animals), 3, 6, 16, 24, and 48 h after 2CLP (27 animals) or SO (10 animals). In these animals, lung tissue was prepared for immunoblot analysis of CAR and integrin expression.

In one group of 62 animals (3 nonoperated controls, 47 subjected to 2CLP, and 12 SO animals), bronchoalveolar lavage was performed just after they were killed. Lungs were infused with 1.5 ml of room-temperature phosphate-buffered saline (PBS) until fully distended. Fluid was withdrawn and saved. This process was repeated three times, volumes were pooled and recorded, and protein content was determined using the Bradford method (Pierce, Rockford, IL).

Myeloperoxidase activity, an index of neutrophil content in tissue, was determined using modifications of the methods described by Calderon et al.  12Briefly, lung tissue obtained from a second group of 62 animals immediately after they were killed was excised, homogenized in 0.1 m potassium phosphate buffer, centrifuged, resuspended, sonicated, and recentrifuged. The supernatant was treated with o-dianisodine and H²O², and changes in absorbance over a 3-min time period at 460 nm were determined.

Just after killing in a third group of 62 rats, lungs were removed en bloc , inflated and fixed overnight in 10% neutral buffered formalin, sliced sagittally, paraffin embedded, cut into 5 μm sections, and stained with hematoxylin and eosin. Light microscopy was performed.

Recombinant E1-deleted adenoviruses expressing green fluorescent protein (GFP) or bacterial Lac-Z  genes with a cytomegalovirus promoter were supplied by the vector core, Institute for Human Gene Therapy, University of Pennsylvania School of Medicine (Philadelphia, PA). Vector was resuspended within 1 h of administration to avoid a decline in viral titer.

Viral particles were suspended in PBS and injected via  a 24-gauge tracheal cannula inserted immediately after sham operation (12 animals) or 2CLP (76 animals), or into a group of six nonoperated animals. Over 10 min, one of two viral doses (3 × 10¹²or 3 × 10¹¹viral plaque-forming units [pfu] in PBS, total volume of 300 μl) were delivered in three divided aliquots. Three control animals received PBS without virus. Animals were allowed to recover and were killed at 0, 24, and 48 h after injection.

Green fluorescent protein–treated lung tissue excised from rats was formalin-fixed, paraffin-embedded, and sectioned as described above. GFP expression was determined on representative 5-μm sections using fluorescence microscopy. Three representative sections from each of three animals at each time point were chosen for examination. Total fluorescence per high-power field was determined electronically. In each representative section, fluorescence from 10 different high-power fields was quantified. The number of counts per high-power field was averaged for each animal. Fluorescence levels per high-power field for three SO and three 2CLP animals at each time point were averaged, and SEs were determined. Heart, kidney, and liver tissue was excised and fixed, and GFP expression in these tissues was determined using similar methodology.

A pathologist unaware of treatment evaluated sections. Hematoxylin and eosin–stained sections were examined for evidence of congestion, atelectasis, inflammatory cell accumulation, alveolar edema, hyaline membrane formation, and lymphocytic infiltration. Fluorescence microscopy was used to evaluate sections for the presence and distribution of GFP staining. GFP was evaluated using a dual-pass filter that transmitted the appropriate wavelengths of light to excite both rhodamine and fluorescein isothiocyanate conjugate and eliminated autofluorescence. Sections were evaluated and scored on the basis of ARDS-like pathology and the number of GFP-positive cells per 40× high-powered field (three fields per animal).

Lung sections were deparaffinized, rehydrated, and heat-treated in citrate buffer (Antigen Unmasking Solution, Vector Labs, Burlingame, CA). In animals not treated with virus, sections were washed with PBS, incubated with a primary mouse antibody to human-CAR (a gift from Jeffrey Bergelson, M.D., Department of Pediatrics, Children’s Hospital of Philadelphia, Philadelphia, PA) or primary goat polyclonal antibodies to α^vβ³or α^vβ⁵integrins (Chemicon International, Temicula, CA) followed by secondary goat–antimouse immunoglobulin G or rabbit antigoat immunoglobulin G (Vector Labs). Human CAR is known to cross-react with both rat and mouse antigens (Bergelson J, verbal communication, January 1999). After washing, sections were treated with avidin and biotinylated alkaline phosphatase, followed by itroblueterazolium–5-bromo,4-chloro,3-indolyphosphate (Boerhinger, Mannheim, Germany).

In a separate group of rats subjected to 2CLP (27 animals) or SO (10 animals), total lung protein was isolated at 0, 3, 6, 16, 24, and 48 h, and immunoblot analysis was performed as previously described. 6The Bradford technique (Pierce, Rockford, IL) was used to determine homogenate protein concentration. Human CAR and α^vβ^3/5integrins were detected using the same primary and secondary antibodies described above. Detection was performed via  enhanced chemiluminescence (Amersham Biotech, Buckinghamshire, United Kingdom). Blotting was performed on purified integrin (Chemicon International, Burlingame, CA) and CAR samples (gift of Dr. Bergelson) to assure that binding was specific (data not shown). To further assure even loading, blots were also probed with an antibody to GAPHD (Chemicon International). Concentrations were quantified with scanning laser densitometry. 7Densities for CAR or the integrins were divided by the density for GAPHD at the same time point in the same blot. This corrected density was then normalized to the density at T0 on the same blot.

Lungs from animals injected with Lac-Z  containing adenovirus were removed en bloc . Tissue was processed using the method of Byrne et al.  13The lung tissue was minced into 0.5-cm cubes, incubated with fresh X-Gal at room temperature for 4 h, washed with PBS, fixed with 2.5% glutaraldehyde for 4 h, and washed with sodium cacodylate buffer. Samples were trimmed under a dissecting microscope, stained en bloc  with 1% uranyl acetate for 1 h in the dark, dehydrated with alcohol, embedded in epoxy LX-112, and polymerized at 70°C for 72 h. Fixed tissue was sectioned to a thickness of 8 nm for electron microscopy. Semithin sections (1.5 μm) were cut with a histo-diamond knife, stained briefly with toluidine blue, and examined by light microscopy.

Analysis of variance with the Bonferroni correction (P < 0.05) was used to identify significant differences between SO and 2CLP and over time.

Hypoxemia, tachypnea, and neutrophilic infiltration characterize clinical ALI–ARDS. Figure 1depicts changes in Pao², respiratory rate, bronchoalveolar lavage protein content, and myeloperoxidase concentrations over time after 2CLP and SO. Compared with sham operation, 2CLP resulted in a 50% increase in respiratory rate and a 50% decrease in Pao². Myeloperoxidase concentrations after 2CLP were 25% above baseline, significantly different from SO, and indicative of neutrophilic infiltration. Bronchoalveolar lavage protein content after 2CLP doubled, reflecting capillary leak. Hematoxylin and eosin–stained sections (fig. 2) confirmed the presence of a significant neutrophilic infiltrate and also revealed septal thickening, increased cellularity, and proteinacious exudate, typical of the lung pathology observed during clinical ALI–ARDS. Within each lung there was a spectrum of disease, ranging from mild injury to severe atelectasis and honeycombing. Minor levels of atelectasis were noted in lungs from SO and nonoperated animals.

To determine the extent to which adenoviruses were taken up by pulmonary epithelial cells after 2CLP or sham operation, we examined GFP fluorescence. Figure 3Adepicts representative sections obtained from nonoperated animals 48 h after virus administration without operation or 48 h after virus administration followed by sham operation or 2CLP. Sections were viewed under fluorescence microscopy. Figure 3Bis a graphic depiction of the number of stained cells per high-powered field 48 h after virus administration, sham operation and virus administration or 2CLP and virus administration. Data were obtained from three nonoperated animals that were administered PBS without virus, three nonoperated animals that were administered GFP-expressing virus, three SO animals that were administered GFP-expressing virus, and five 2CLP animals that were administered GFF-expressing virus. Ten fields in each section were evaluated. Relative to nonoperated and SO controls, there was a statistically significant (P < 0.001) increase in GFP expression in the lungs after 2CLP. The majority of this expression was observed in small airways and surrounding alveoli with little expression in the trachea, bronchi, and bronchioles. GFP fluorescence was most prominent in mildly to moderately diseased areas of 2CLP lung but was not observed in severely diseased regions (data not shown). No GFP expression was noted in heart, liver, and kidney sections examined after 2CLP (data not shown).

Administration of 3 × 10¹¹or 3 × 10¹²pfu resulted in comparable levels of GFP expression. However, 48 and 72 h after 3 × 10¹²viral pfu was administered, a degree of viral-mediated lymphocytic infiltration was observed. Despite similar levels of GFP expression, little lymphocytic infiltration was observed after administration of the lower viral dose.

Light microscopy strongly suggested that most GFP expression after 2CLP occurred in type II cells. However, 48 h after adenovirus administration and 2CLP, fluorescence was detected lining the alveoli (fig. 3A). This is consistent with either secretion of GFP or with GFP expression within type I cells. Indeed, some type I cells appeared to be GFP-positive. To determine more precisely which cells were taking up adenoviruses and expressing gene product, we used electron microscopy and semithin section light microscopy. One nonoperated animal was administered PBS without virus, one nonoperated animal received 10¹¹pfu Lac-Z -expressing adenovirus, two SO operated animals were given a like dose of LacZ -expressing viruses, and four 2CLP animals were administered LacZ -expressing virus. Electron and semithin section light microscopy were performed 48 h after intervention. A representative electron micrograph from a 2CLP animal is shown in figure 4. This localized Lac-Z  gene product 48 h after 2CLP primarily to lamellar body–containing type II cells, although staining also was observed in some type I cells (not shown). Similar results were obtained in all four 2CLP animals and in the two SO rats. Viral uptake and LacZ  expression in nonoperated animals was not sufficient to assess localization. Semithin sections examined under 60× magnification confirmed that type II cells were the primary sites of expression (fig. 5, black arrow) although again, some type I cells contained LacZ  gene product (fig. 5, grey arrow). Some expression also was observed in interstitial macrophages; indeed, X-Gal staining was observed in macrophage phagosomes (fig. 5, blue arrow).

One possible mechanism to explain enhanced GFP expression after 2CLP is an up-regulation of one or all of the receptors involved in adenovirus attachment and internalization. To determine changes in the relative abundance and location of these receptors after 2CLP, immunohistochemistry was performed (fig. 6). These studies demonstrated an increase in α^vβ^3/5integrin and CAR expression after 2CLP (fig. 6D) relative to SO (fig. 6C) or nonoperated (fig. 6B) controls. Treatment with only secondary antibodies followed by avidin, biotinylated alkaline phosphatase and NBT–BCIP did not result in any staining (results not shown).

Increased apical abundance can result from either augmented production or from translocation of intracellular or basal membrane receptors. Therefore, protein abundance in lung homogenate was determined using immunoblotting. Representative immunoblots are reproduced in figure 7A. Data from autoradiograms were quantified using laser densitometry (fig. 7B). Relative to both nonoperated controls and SO animals, a statistically significant increase in both α^vβ³integrin and CAR abundance was observed after 2CLP. α^vβ³integrin concentrations were increase at 3, 6, 16, 24, and 48 h, whereas CAR abundance was significantly greater only at 2 and 6 h. No change was noted in α^vβ⁵integrin concentrations (data not shown).

In this report we demonstrated that 2CLP in rats results in a clinical and pathologic picture consistent with ALI–ARDS. We found high levels of expression after intratracheal administration of GFP-containing adenoviruses, with marker protein primarily localized to type II pneumocytes. The mechanism leading to increased expression after 2CLP appears to involve an increase in the abundance and apical density of α^vβ³integrin and CAR receptors, which are responsible in part for adenovirus attachment and internalization on the cell surface.

Our interest in the therapeutic modulation of gene expression after 2CLP arises from previous studies that revealed an inappropriate, CLP-associated decrease in the expression of key hepatic and pulmonary genes. 5,6Demonstration of enhanced viral uptake and expression after 2CLP indicates that replacement of underexpressed proteins may be possible. The distribution of GFP expression in moderately diseased areas and the localization of this marker protein to type II pneumocytes is key. Pelosi et al.  14proposed that pathologic changes in the acutely injured lung are heterogeneous, i.e. , in clinical ARDS, some lung regions are normal, other areas are irreversibly consolidated, and still others reversibly diseased. GFP expression after 2CLP localizes primarily to areas in which pathologic changes appear to be reversible. This increases the likelihood of repair strategies being successful.

Uptake by type II cells is important for two reasons. First, these cells synthesize and secrete surfactant. Surfactant abnormalities contribute to impaired gas exchange in ALI–ARDS. Second, type II cells are precursors to type I cells. Because both cell types are lost in ALI–ARDS, increased numbers of type II cells are required for replacement. Therefore, preservation and proliferation of type II cells could offset type I cell dropout. Later in the disorder, there is decreasing cell proliferation accompanied by fibrosis and scaring. 15Using adenoviral transfer to enhance expression of proteins that either accelerate or interrupt the cell cycle is a potentially valuable intervention.

Demonstration of increases in the abundance of both CAR and α^vβ³integrin, which modulate adenoviral binding and internalization, could explain enhanced viral uptake after 2CLP. 10Indeed, some investigators have noted that injury may induce expression of adenoviral receptors on the cell surface. 16,17Dong et al.  16proposed that viral-mediated gene transfer is dependent not only on increased receptor density but also on physical distribution. This is consistent with work demonstrating that the majority of viral receptors reside on the basolateral surface of pneumocytes, making them inaccessible to virus. 17,18Although the increase in α^vβ³integrin protein concentrations is significant, 2CLP only modestly altered overall concentrations of CAR protein. Although CAR is responsible for viral trapping and adhesion to the cell surface, 50% increase in abundance may not be biologically meaningful. However, it is possible that 2CLP enhances viral adhesion both via  increased CAR abundance and by stimulating translocation of CAR from either an intracellular or basolateral location to the apical surface. Indeed, the differences noted in CAR immunohistochemistry are more striking than those revealed using immunoblotting. Alternatively, the increase in receptor abundance on immunohistochemistry could reflect induction of pneumocyte proliferation. Overabundant cell replication has been implicated in the pathogenesis of the abnormalities observed in the late, proliferative phase of ARDS. This explanation is consistent with data indicating that regenerating airway epithelial cells express CAR and α^vβ³integrin on all aspects of the cell surface, while expression in mature cells is restricted to the basolateral membrane. 10,19 

The manner in which integrins modulate internalization of virus is poorly understood. 10,20However, α^v-containing integrins may also be involved in the uptake of bacterial pathogens such as Pneumocystis carinii . 21Our study demonstrates an increase in the abundance of α^vβ³integrin at all time points after 2CLP. Thus, although increases in viral trapping may be limited to the early phase of sepsis, uptake may persist throughout the time course studied. Future studies will examine the role of integrin blockade in modulation of enhanced viral uptake.

The function of CAR, other than its role in trapping viral particles, also is unclear. However, the highly conserved nature of this protein, with 90% homology existing between human and murine forms, 10implies an important cellular function. Again, future investigations are required to elucidate the role of this receptor in nonpathologic states.

A major obstacle in the clinical application of gene therapy in chronic diseases has been the transient nature of transgene expression. This occurs because adenoviruses do not integrate their genome into the chromosomes of the host cells 9and because, in the high titers necessary for most gene transfer experiments, they provoke intense lymphocyte infiltration around small airways. 22Our results indicate that the pulmonary lymphocytic response to viruses is limited in this model of ARDS. We believe this reflects, in part, the ability to use lower viral loads while still achieving high levels of uptake. In this regard, our findings may have important ramifications. The nonlymphocytic inflammation characteristic of ALI–ARDS in some manner enhances adenoviral uptake. This observation could be used in other settings to improve viral entry and transgene expression. Indeed, modulation of both CAR and integrin expression might be of value in applying gene therapy to chronic diseases. However, it is important to note that the lack of a lymphocytic response may result from a sepsis-associated change in the immune response. In addition, no GFP was found in organs other then the lung. This indicates that virus does not leak into the systemic circulation despite ARDS-associated damage to epithelial cells, endothelial cells, and basement membrane, a finding that may also be important in approaching other pathologic states.

Our findings confirm that, as reported by other investigators, 2CLP is a useful model of ALI–ARDS. 23,24Lung injury in this model is induced by an indirect insult, fecal peritonitis, as occurs in most surgical patients who develop ARDS. This makes the model clinically relevant, complementing animal models of ALI that involve a direct insult to the lung. 23 

Therefore, we believe our results have important ramifications not only for the use of gene transfer in ARDS, but also for therapy of other acute inflammatory conditions and in the improvement of gene transfer for chronic disorders or inborn errors. Although a similar approach to ARDS has been proposed by other investigators, we believe the data reported here represent the first successful use of gene transfer in a model of ARDS. 25Future studies will be required to identify target genes that are suitable for enhanced expression in ARDS.