Transgenic Expression in Mouse Lung Reveals Distinct Biological Roles for the Adenovirus Type 5 E1A 243- and 289-Amino-Acid Proteins

Abstract

Little is known about the biological significance of human adenovirus type 5 (Ad5) E1A in vivo. However, Ad5 E1A is well defined in vitro and can be detected frequently in the lungs of patients with pulmonary disease. Transgenic expression of the Ad5 E1A gene targeted to the mouse lung reveals distinct biological effects caused by two Ad5 E1A products. Either of two Ad5 E1A proteins was preferentially expressed in vivo in the transgenic lungs. The preferential expression of the Ad5 E1A 243-amino-acid (aa) protein at a moderate level was associated with cellular hyperplasia, nodular lesions of proliferating lymphocyte-like cells, and a low level of p53-dependent apoptosis in the lungs of transgenic mice. In contrast, the preferential expression of the Ad5 E1A 289-aa protein at a moderate level resulted in a proapoptotic injury and an acute pulmonary proinflammation in the lungs of transgenic mice, mediated by multiple apoptotic pathways, as well as an enhancement of the host immune cell response. Expression of the Ad5 E1A 243-aa protein resulted in proliferation-stimulated p53 upregulation, while expression of the Ad5 E1A 289-aa protein led to DNA damage-induced p53 activation. These data suggest that the Ad5 E1A 243- and 289-aa proteins lead to distinct biological roles in vivo.

To date, nearly all studies on human adenovirus type 5 (Ad5) early region 1A (E1A) have been performed in vitro. Alternative splicing converts a single Ad5 E1A transcript into either 12S or 13S mRNA to encode proteins of 243 or 289 amino acids (aa), respectively. Four functional domains of the E1A proteins, one N-terminal region (NTR) and three conserved regions (CR1, CR2, and CR3), have been defined in a variety of adenoviruses. Both E1A proteins have identical amino and carboxyl termini but differ by 46 aa (CR3) unique to the 289-aa protein (Fig. 1). Adenovirus E1A proteins are known to regulate transcriptional activation through interaction with a variety of cellular proteins. The NTR, CR1, and CR2 domains are thought to be involved in the inhibition of differentiation-associated gene expression (1, 3, 49, 50, 57) and the induction of cellular DNA synthesis (30, 40, 41), transformation (45, 46, 51, 60), and immune suppression (2, 31, 53) by binding to some transcriptional factors, such as the tumor suppressor pRb family of proteins, p300/CBP, activation factors (APs), TATA-binding protein (TBP), and STAT. The CR3 domain is responsible only for specific E1A-dependent trans-activation of the E1A-responsive genes by binding to other transcriptional regulators, such as activated-transcription factors (ATFs), APs, TBP, and TBP-associated factors TFII and TFIII (11, 12, 20, 26, 29). Some of these E1A-responsive genes are necessary for the activation of gene expression during cell differentiation. In addition, the Ad5 E1A gene seems to have both oncogenic and antioncogenic activities in a diversity of target cells (7, 16). As an oncogene, Ad5 E1A can induce cell proliferation and transform rodent cells in vitro under certain conditions (30, 41, 51, 60). However, recent evidence indicates that Ad5 E1A can also upregulate p53, trigger apoptosis, partially reverse the transformed morphology of cultured cells, and induce Ad5-infected cell susceptibility to the host cellular immune response (2, 10, 30, 52, 56). Notwithstanding that individual functions of the E1A proteins are well defined in vitro, we lack knowledge about their in vivo roles since E1A proteins target diverse promoters through interaction with different cellular DNA-binding proteins. Therefore, this study was aimed at further characterization of the in vivo potential effects of the E1A proteins in a mouse model.

FIG. 1.

Schematic diagram of the functional domains of the E1A 243- and 289-aa proteins and their interactions with cellular transcription factors. The Ad5 E1A gene contains two exons and encodes the 243- and 289-aa proteins mainly through alternative splicing. The functional domains of the E1A proteins are indicated as NTR, CR1, CR2, and CR3. Those domains responsible for interaction with the cellular transcription factors STAT, p300/CBP, p53, pRb, TBP, and ATFs are also indicated.

Infection with human adenovirus is thought to be involved in a variety of pediatric and adult pulmonary diseases, including acute or chronic respiratory pneumonia, pulmonary inflammation, and fibrosis. In addition, Ad5 can lead to latent infection in the human pulmonary system (15). Recent experiments reveal that both Ad5 E1A DNA and its gene products can be detected in the respiratory systems of patients with a pulmonary disease (14, 23, 24, 33), suggesting that latent Ad5 E1A may play a partial role in the pathogenesis of chronic pulmonary disease. Adenovirus-associated pulmonary disease has been studied in several animal models, including mice, in which, interestingly, only the Ad5 early gene (especially E1A and E1B) expression is activated in the absence of viral replication or viral late gene expression. The histopathological response in the lungs of those animals is similar to that reported to occur in humans (39, 55). However, the potential effect of Ad5 E1A latency in the lungs is unclear. The objective of this study was to gain insight into in vivo effects of Ad5 E1A, which, furthermore, could help in understanding the pathogenesis associated with latent Ad5 E1A in respiratory inflammation and pulmonary disease. We have used a transgenic approach to limit the expression of Ad5 E1A to mouse lung and have shown that the Ad5 E1A gene is preferentially expressed in vivo. The preferential expression of the Ad5 243-aa protein in our E1A transgenic lines induced abnormal cell proliferation and nodular lesions of proliferating lymphocyte-like cells. The preferential expression of the Ad5 289-aa protein in our other transgenic line, in contrast, led to a proapoptotic injury and pulmonary proinflammation, reflecting Ad5 E1A-associated proapoptosis and upregulation of the host cellular immune response. Thus, tissue-specific transgenic expression reveals the distinct pathophysiology caused by preferential expression of each E1A protein.

MATERIALS AND METHODS

Generation of transgenic mice.

All animal work was carried out in accordance with institutional guidelines. SPCE1A was constructed by cloning an entire Ad5 E1A genomic fragment (40) into the 3.7SPC/SV40 vector (17). A 5.6-kb fragment (NdeI/NotI generated) of the SPCE1A transgene (Fig. 2A) was isolated and microinjected into fertilized B6/SJL mouse eggs to generate transgenic mice. Transgenic founder mice were identified from genomic tail DNA by PCR analysis with a pair of primers (F1, CGG AGG TGA TCG ATC TTA CC; R2, AAG GCG TTA ACC ACA CACG) spanning Ad5 E1A mRNA splice sites and genotyped by Southern blot analysis with a probe of the ³²P-labeled Ad5 E1A genomic fragment. HindIII and EcoRI digestions were used to determine copy numbers and inserts, respectively, for the SPCE1A transgene by performing DNA blot analysis on founder mouse tail DNA (Fig. 2B). The intensity of each HindIII- or EcoRI-generated band was measured with a densitometer (Amersham Pharmacia Biotech). The EcoRI-generated band of lowest intensity was assumed to represent one copy of the SPCE1A transgene. The copy numbers of the transgene in each founder were calculated by comparing the intensity of the HindIII-generated band with that of the EcoRI-generated band of lowest intensity. The inserts were calculated as the numbers of EcoRI-generated bands. Founders were backcrossed to C57BL/6 mice, and transgene-heterozygous offspring from at least three mice of each line were used for each analysis.

FIG. 2.

The SPCE1A transgene and its expression in transgenic mice. (A) The 3.7k-SPC/Ad5 E1A constructor used for generation of the transgenic mice and two transcripts, 12S and 13S mRNAs, resulted from alternative splicing. (B) Southern blot analyses of tail genomic DNAs from A30 and A97 founder (F0) mice. A 5-kb HindIII-digested band (lanes H) represents an intact SPCE1A transgene in the A97 founder. A rearranged 3.5-kb band or the other larger HindIII-digested bands also contain an intact E1A transgene in the A30 founder, as confirmed by PCR. The EcoRI-digested bands (lanes E) indicate the existence of one or two insertion sites in the A97 or A30 founders. The preferential expression of the E1A transgene in the lungs of A97 and A30 mice was determined at the level of mRNA by RT-PCR (C) and at the level of protein by immunoblotting (D), and it seems that the levels of transgene products in female mice (labeled F) were higher than those in male mice (labeled M). (E) The preferential expression of each E1A protein was quantified by determination of relative levels of the 289- and 243-aa proteins on immune blots in a densitometer (n = 4). (F) The E1A proteins were localized in the nuclei of pulmonary epithelial cells of alveoli, bronchiole, and trachea in both transgenic lines by immunostaining frozen lung sections with anti-Ad5 E1A antibodies, but the proteins were not found in nontransgenic littermates (Non-Tg). Scale bar, 8 μm.

mRNA analysis.

Total RNA was isolated from transgenic or nontransgenic mouse lung. Five micrograms of the total RNA from each sample was used in 50 μl of reverse transcription reaction buffer containing 25 ng of oligo(dT)/μl, 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 2.5 mM MgCl₂, 10 mM dithiothreitol, 0.5 mM deoxynucleoside triphosphate, and 10 U of SUPERSCRIPT (GIBCO) and incubated at 42°C for 50 min. Then 5 μl of reverse transcription product was used for PCR amplification in 50 μl of PCR buffer with 100 mM primers F1 and R2, 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl₂, 0.2 mM deoxynucleoside triphosphate, and 2 U of Taq DNA Polymerase (GIBCO). Amplification was performed for 35 cycles. Cycling conditions were 94°C for 40 s, 56°C for 50 s, and 72°C for 90 s. The reverse transcription-PCR (RT-PCR) products were analyzed on 1.5% agarose gel.

Western blot analysis.

Fresh mouse lung was homogenized in tissue extraction buffer (phosphate-buffered saline [PBS], 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 1 mM 2-mercaptoethanol, 0.1 mg of phenylmethylsulfonyl fluoride/ml, 30 μg of aprotinin/ml, 100 mM sodium orthovanadate, and 200 ng of leupeptin/ml). Lung lysate was spun twice at 4°C and 12,000 × g for 20 min, separated by sodium dodecyl sulfate-10 to 20% polyacrylamide gel electrophoresis, and blotted onto a polyvinylidene difluoride membrane. Immunoblotting was performed with anti-Ad5 E1A monoclonal (M58; Pharmingen) and polyclonal (13S-5; Santa Cruz) antibodies. The levels of p53 and Fas were analyzed by immunoblotting with anti-p53 and anti-Fas polyclonal antibodies (Santa Cruz), respectively. The amount of total protein in each line was controlled with antiactin antibody (Santa Cruz). The bound antibodies were detected by ECL (Amersham Life Science). The level of expression was quantified with a densitometer.

Immunohistochemistry.

Fresh lung was recovered by cardiac perfusion and inflated by intratracheal injection with cold PBS. One lobe of the lung was fixed in 10% neutral-buffered formalin overnight, embedded in paraffin, and sectioned at 4 μm for histopathological and immunohistochemical analyses, and the rest of the lung was fixed in fresh 4% paraformaldehyde in PBS for 4 h, soaked in 30% sucrose overnight at 4°C, and then embedded in O.C.T. Compound (Miles Inc.) and sectioned at 5 μm. Serial frozen sections from each lung were blocked and incubated at 4°C overnight with anti-Ad5 E1A primary rabbit antibody (13S-5; Santa Cruz) and visualized by biotinylated anti-rabbit secondary antibody and ABC staining (Santa Cruz). At least six serial lung sections from at least three mice of each line were used in E1A immunostaining, 5-bromo-2′-deoxyuridine (BrdU; Sigma) incorporation, terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) DNA fragmentation assay, and leukocyte marker assay.

BrdU incorporation.

Mice received intraperitoneal injection of 100 μg of BrdU/g and 10 μg of 5-fluoro-2′-deoxyuridine (Sigma)/g. After 16 h, lung tissues were perfused by a heart perfusion, inflated intratracheally, and fixed in 10% neutral-buffered formalin. Serial paraffin-embedded lung sections (4 μm) were deparaffinized, rehydrated, and treated with 70 μg of pronase E/ml for 20 min at 37°C and with 2 N HCl for 30 min at 37°C, followed by blocking and incubation overnight at 4°C with 1:300 diluted rat anti-BrdU antibody (Accurate Chemical Science). BrdU-labeled nuclei were visualized by using biotinylated rabbit anti-rat immunoglobulin G antibody and ABC staining (Genzyme). BrdU-labeled nuclei from more than 2,000 lung cells (from at least three mice) were scored to estimate the BrdU labeling index under the light microscope at ×400 magnification.

TUNEL DNA fragmentation assay.

Fresh frozen lung sections (5 μm) were fixed in cold acetone, blocked in 10% bovine serum albumin in PBS, and then incubated with TUNEL labeling mix and TdT enzyme (In Situ Cell Death Detection Kit; Boehringer Mannheim) for 1 h at 37°C. After being blocked in 10% bovine serum albumin and 10% sheep serum in PBS, TUNEL-labeled nuclei were visualized by using a sheep anti-fluorescein antibody conjugated with horseradish peroxidase (Boehringer Mannheim) and AEC (Genzyme). Cellular apoptosis characteristics of condensed or fragmented nuclei from at least 2,000 lung cells (from at least three mice) were analyzed and estimated for proapoptotic injury scored under the light microscope at ×400 magnification.

Immunostaining Fas/FasL and leukocyte markers.

Serial paraffin-embedded lung sections were deparaffinized, rehydrated, and treated with 0.0025% trypsin for 10 min at room temperature. After blocking, the sections were incubated overnight at 4°C with rabbit anti-mouse Fas or FasL (Santa Cruz) and rat anti-CD3, -CD8, and -Mac-1 antibodies (Serotec) and then visualized by using biotinylated secondary antibodies and ABC staining (Genzyme). The Fas or FasL staining index was estimated by analyses of more than 2,000 lung cells (from at least three mice) under the light microscope at ×400 magnification. The population of CD8⁺ cells or Mac-1⁺ cells (from at least 2,000 lung cells of at least three mice of each line) was quantified under the light microscope at ×400 magnification.

Cytometric analyses.

One- and two-color flow cytometric analyses were carried out as described previously (63). The cells were prepared by disrupting the lung tissues or cervical lymph nodes from at least three mice of each line through a wire mesh, and then red blood cells were removed by passing the cell suspension through Lympholyte-M (Cedarlane, Hornby, Ontario, Canada). Approximately 1 × 10⁵ to 3 × 10⁵ cells were stained, of which 3,000 to 10,000 cells were analyzed on a Becton Dickinson FACScan flow cytometer. A series of monoclonal antibodies was used as follows: pan anti-γδ-TCR/FITC conjugate, anti-CD4/PE and -CD8/biotin conjugate, anti-NK/biotin conjugate, anti-MHC class I and II-A/biotin conjugate (Pharmingen), anti-Mac-1 (M1/70), and anti-CD45R/B220/PE conjugate (Sigma) were used to identify γδ-TCR T lymphocytes, subsets of αβ-TCR T cells, natural killer cells, major histocompatibility complex (MHC) class I- and II-expressing cells, macrophages and monocytes, granulocytes, and B lymphocytes. Biotin-conjugated antibodies were detected with streptavidin-fluorescein isothiocyanate (SA-FITC) or SA-R-phycoerythrin (Pharmingen). Unconjugated antibodies were detected with FITC-conjugated F(ab′)2 goat anti-mouse immunoglobulin G (Fc specific) (Accurate Chemical and Scientific).

Pathological analyses.

All animals used in this research project were cared for and used humanely according to the Guides for the Care and Use of Laboratory Animals of the Hospital for Sick Children (1995). All mice were housed within the animal facility of the Hospital for Sick Children, which is routinely monitored and maintained as a pathogen-free facility. Both outbred and inbred offspring were observed daily for abnormal phenotypes. Nontransgenic littermates were used as controls for all analyses. Embryos of E11 to E19 were dissected to analyze developmental defects or lethality. For histopathological analysis, mice were euthanized, lungs were examined grossly, and tissues were immediately collected and fixed in 10% neutral-buffered formalin, embedded in paraffin, sectioned at 4 μm, and stained with hematoxylin and eosin.

Statistical analyses.

A Student's t test (SigmaStat) was used to statistically analyze the data obtained from each transgenic line and from nontransgenic mice.

RESULTS

Preferential expression of either Ad5 E1A 243- or 289-aa protein in the transgenic mice.

We wished to investigate the in vivo effects of Ad5 E1A proteins and the possibility that the latent E1A gene might be one of the factors causing pulmonary disease during latent adenovirus infection in the respiratory tract. Since ubiquitous or non-lung tissue-specific expression of Ad5 E1A proteins in transgenic mice led to intrauterine or newborn lethality (58), we used the lung-specific promoter, human surfactant protein C (SPC) regulatory elements, to generate viable transgenic mice expressing Ad5 E1A proteins in the lung, mimicking the latency of Ad5 E1A in the human respiratory system. Of about 200 mice born, four animal founders, SPCE1A/A30, SPCE1A/A97, SPCE1A/A2-20, and SPCE1A/A3-02 (referred to as the A30, A97, A2-20, and A3-02 lines, respectively), showed by Southern blotting, PCR, and sequencing analyses that they carried an entire Ad5 E1A coding region. The low frequency of transgenic founders may be due to embryonic lethality caused by expression of the E1A transgene. The expression of the E1A transgene in the lungs of all lines was determined by RT-PCR analysis and immunoblotting. Three lines, A97, A2-20, and A3-02, shared the same phenotype and expression pattern, predominantly expressing the Ad5 E1A 243-aa protein; one line, A30, showed a distinct phenotype and a different expression pattern, predominantly expressing the Ad5 E1A 289-aa protein. Two transgenic lines, A30 and A97, were characterized in detail. In the A30 founder, the SPCE1A transgene had approximately two copies inserted in tandem at one site in the genome. In the A97 founder, approximately five copies of the SPCE1A transgene were integrated at two or three sites in the genome (Fig. 2B). The copy number and insertion of the SPCE1A transgene were determined by analyzing the HindIII- and EcoRI-cleaved tail DNA. A 5-kb HindIII-generated fragment indicated the full size of the SPCE1A transgene, while each EcoRI-generated fragment consisted of a 5.6-kb SPCE1A transgene and a flanking cellular DNA fragment which varied in size because another EcoRI cutting sequence was located in the cellular DNA and represented an integration site of the transgene (Fig. 2B). Cleavage of the A30 founder tail DNA with HindIII produced a 3.5-kb fragment, 1.5 kb smaller than the expected 5-kb fragment, indicating a possible rearrangement and/or deletion of part of the SPC promoter region in those mice. In addition, the other larger fragment, generated by cleavage of the A30 founder tail DNA with HindIII, indicated that the upstream HindIII site of the SPCE1A transgene appeared to be deleted (Fig. 2B). No obvious expression was detected in thymus, heart, liver, spleen, or kidney at the mRNA level, while low-level expression of the E1A 243-aa protein appeared in the brain of 3-month-old A97 mice (data not shown). A30 mice preferentially expressed the 289-aa protein in the lung (approximately 94% of the total E1A protein), while A97 mice preferentially expressed the 243-aa protein in the lung (approximately 90% of the total E1A protein) (Fig. 2C to E). The E1A transgene products were further localized, by immunostaining lung sections with anti-Ad5 E1A antibody, in the epithelial cells of trachea, bronchiole, and alveoli in both lines of transgenic mice (Fig. 2F).

Distinct lung pathophysiological alterations in the transgenic mice preferentially expressing either 243- or 289-aa protein.

Examination of embryos at 13.5 to 15.5 days postcoitum indicated underdevelopment and prenatal death of some of the transgenic animals (Fig. 3D). A higher level of the E1A protein was detected in the lungs of transgenic embryos at 15.5 days postcoitum than in the lungs of 1-month-old A97 mice (unpublished data). Some transgenic mice died shortly after birth due to thickened alveolar septa, shrunken alveolar sacs, and marked accumulation of mucus in the airspace and airway in fetal lung (data not shown). The A97 mice that survived had histopathologic changes that included regional thickening of alveolar walls, abundant strands of mucus, and prominent linear crystals compared with normal lung of the nontransgenic littermates (Fig. 3A and B). At 12 months of age, the lungs of A97 mice were characterized by multiple perivascular and peribronchiolar nodules of lymphocyte-like cells (Fig. 3C), and the mice became ill or died due to the widespread nodular lesions of lymphocyte-like cells in the lungs after 14 months of age (Fig. 3E). The A97 founder mouse died from a lymphoreticular tumor at 22 months of age. In contrast, preferential expression of the 289-aa protein did not lead to these pathological abnormalities. Instead, active pulmonary proinflammation was found in the lung and respiratory tract of A30 mice as young as 1 month old, characterized by abnormal increases in the numbers of leukocytes in the lungs compared with those of the nontransgenic littermates (Fig. 3B and C). At 3 months of age, most A30 mice were frail and demonstrated signs of respiratory distress, swollen nose, enlarged cervical lymph nodes, and shortened life span (less than 12 months) (Fig. 3A).

FIG. 3.

Phenotypic differences of A97 and A30 mice. (A) In comparison with the normal appearance of A97 mice and nontransgenic littermates (Non-Tg), 3-month-old A30 mice showed a frail phenotype with signs of respiratory distress, swollen nose, enlarged cervical lymph nodes, and a shortened life span. The lung histological sections were stained with hematoxylin and eosin in panels B, C, and E. (B) Thickened alveolar septa and marked accumulation of mucus in airspace and airway in the lungs of 5-week-old A97 mice, a mild preinflammatory response in the lungs of 5-week-old A30 mice, and normal appearance of the lungs of 5-week-old nontransgenic littermates. (C) Normal lungs in 12-month-old nontransgenic littermates, nodular lesions of lymphocyte-like cells in the lungs of 12-month-old A97 mice, and an inflammatory response or dysplastic changes in the lungs of 12-month-old A30 mice. Scale bars, 40 μm. (D) Embryonic lethality caused by the expression of the Ad5 E1A transgene during embryonic development was examined by dissection of A97 F2 embryos at 15.5 days postcoitum. Some transgenic embryos were underdeveloped compared with those of nontransgenic littermates. (E) Nodular lesions of lymphocyte-like cells developed in the lungs of A97 mice at 16 months of age.

Cell proliferation in the lungs preferentially expressing the E1A 243-aa protein.

We next determined whether the induction of cell proliferation was correlated with the in vivo activities of either the 243- or 289-aa protein. The proliferating cells in the lung sections from 3-month-old mice were detected by BrdU labeling. About 12% of BrdU-positive cells appeared in the lungs of A97 mice, whereas only 2.3% of BrdU-positive cells were seen in the lungs of nontransgenic mice (P < 0.01) (Fig. 4A and C). The population of BrdU-positive cells correlated with the thickness of the alveolar septa in the lungs of A97 mice (Fig. 4A). A result of double staining of lung sections with anti-BrdU and anti-pulmonary epithelial cell marker SPC antibody showed that most BrdU-positive cells were the epithelial cells, which colocalized with the cells expressing the E1A proteins (data not shown). In addition, most BrdU-positive cells in the nodules of lymphocyte-like cells in the lungs of 1-year-old A97 mice were CD3⁺ lymphocytes (data not shown). Cell proliferation induced by constitutive expression of the Ad5 E1A 243-aa protein seemed persistent and promoted the nodular lesions of lymphocyte-like cells in the lungs of A97 mice. These results were confirmed by immunostaining with another cell proliferation marker, K_i-67, suggesting that the E1A 243-aa protein is responsible for the induction of cell proliferation in the lungs of the A97 mice. In contrast, a modest increase in cell proliferation (approximately 4% of BrdU-positive cells) was detected in the lungs of 3- and 12-month-old A30 mice in comparison with the level in nontransgenic littermates (Fig. 4A and C).

FIG. 4.

Distinct effects of the E1A proteins on cell proliferation and proapoptotic injury. (A) The cell proliferation induced by expression of the E1A transgene was detected by BrdU incorporation, and BrdU-labeled nuclei (arrows) present in proliferating epithelial cells in the paraffin-embedded lung sections of 3-month-old transgenic mice are indicated. Increased BrdU labeling index of the epithelial cells appeared in the alveolar region in A97 mice. Original magnification, ×200. (B) The apoptosis induced by expression of the E1A transgene was detected by TUNEL DNA fragmentation assay with frozen lung sections. TUNEL-labeled positive cells (arrows) were seen in apoptotic epithelial cells of 3-month-old transgenic mice. Elevated numbers of TUNEL-labeled cells and cells exhibiting nucleus condensation or fragmentation were seen in the alveolar and bronchiolar regions in A30 mice, and most proapoptotic cells were undergoing an early stage of apoptosis. Original magnification, ×200. (C) The levels of BrdU-labeled positive cells were quantified for 3-month-old A97 and A30 mice and nontransgenic littermates (n = 3). (D) The ratio of cells undergoing proapoptosis from the total lung cells was quantified based on TUNEL DNA fragmentation assay and nucleus condensation or fragmentation analyses for 3-month-old A97 and A30 mice and nontransgenic littermates (n = 3). (E) The expression of the E1A transgene stimulated expression of p53, and the levels of p53 in the lung lysates of 3-month-old nontransgenic littermates and A30 and A97 mice were detected by immunoblotting assay with anti-mouse p53 antibody. Non-Tg, nontransgenic.

Proapoptotic injury in the lungs predominantly expressing the 289-aa protein.

The incidence of proapoptotic injury in the lung sections from both 3-month-old A30 and A97 transgenic mice was assessed by using a TUNEL DNA fragmentation assay and directly detected by analyzing the condensed or fragmented nuclei under the microscope. The TUNEL-labeled positive cells were seen in the alveoli, bronchioles, and tracheas of 3-month-old A97 and A30 mice (Fig. 4B). Double staining of lung sections with TUNEL labeling and anti-pulmonary epithelial cell marker SPC antibody revealed that over half of the TUNEL-labeled positive cells were pulmonary epithelial cells in both transgenic mice (data not shown). Notably, majorities of TUNEL-labeled nuclei in the lungs of A30 mice were faintly stained and modestly condensed, suggesting that those cells were undergoing slight DNA damage. The percentages of TUNEL-labeled positive lung cells were measured as 4.3% for the A97 mice (P < 0.01) and 12.6% for the A30 mice (P < 0.01) compared with 0.5% for the nontransgenic littermates (Fig. 4D). These results suggest that the 289-aa protein correlates with the induction of a proapoptotic injury in the lungs of A30 mice. The increase in apoptotic cells could explain why the proliferating cells in the lungs of A97 mice did not frequently lead to a carcinoma. Rather, this abnormal cell proliferation appeared to lead to nodular lesions of lymphocyte-like cells that occurred with increasing frequency as the mice aged.

Expression of p53 was induced in the lungs of both lines of transgenic mice.

We evaluated the expression of p53 tumor suppressor to investigate whether p53-dependent apoptosis contributed to apoptosis seen in the lungs of A97 mice and proapoptotic injury in the lungs of A30 mice. Western blot analysis of the total lung lysates from 3-month-old mice demonstrated a 2.6-fold increase in the levels of p53 in the lungs of A97 mice compared with those for nontransgenic littermates (Fig. 4E). Moreover, there seemed to be a correlation between the increase in p53 expression and the level of the E1A 243-aa protein; i.e., more p53 protein was detected in the lungs of younger A97 mice expressing larger amounts of the 243-aa protein. Furthermore, a higher ratio of BrdU incorporation correlated with the higher levels of p53 in A97 mice (data not shown), suggesting that the increase in p53 was the consequence of a higher level of cell proliferation induced by expression of the 243-aa protein. An even higher (approximately four- to sevenfold) increase in p53 was detected in the lung lysates of A30 mice compared with the level in nontransgenic littermates (Fig. 4E). The presence of increased levels of p53 in the lungs of A30 mice was in agreement with the higher level of proapoptotic cells in these mouse lungs, suggesting that increased cellular DNA damage, induced by expression of the 289-aa protein, may have resulted in the increased p53 expression.

p53-independent apoptosis pathways were also activated in the transgenic lungs predominantly expressing the 289-aa protein.

To further elucidate other apoptosis pathways involved in the proapoptotic injury that occurred in the lungs of A30 mice, the levels of the apoptotic cytokine tumor necrosis factor alpha (TNF-α) in the sera of mice were examined by using an enzyme-linked immunosorbent assay kit for mouse TNF-α (Genzyme). At 3 months of age, the level of TNF-α was significantly increased in A30 mice (858 pg/ml) by greater than fourfold compared with that in A97 mice (P < 0.01), in which the level of TNF-α (210 pg/ml) was not significantly different from that in nontransgenic littermates (165 pg/ml) (P = 0.59) (Fig. 5A). In addition to TNF-α, the expression of the Fas molecule was measured by immunoblotting. The level of Fas expressed in the lungs of 3-month-old A30 mice was two- to fourfold higher than that in nontransgenic littermates on the basis of the immunoblotting assay, while a slight increase (1.4-fold) in the level of Fas was detected in the lungs of 3-month-old A97 mice (Fig. 5B). The expression of FasL was also increased in some leukocytes and in certain bronchiolar and alveolar epithelial cells in the lungs of A30 mice, while a moderate increase of FasL was detected in the lungs of A97 mice (Fig. 5C). In addition, neither FasL nor Fas was detected in the nodular lesions of lymphocyte-like cells in the lungs of 1-year-old A97 mice (data not shown), suggesting that those rapidly proliferating lymphocyte-like cells were anergizing and lacked cytotoxicity. Overall, these results indicate that TNF-α-mediated proapoptosis and Fas/FasL-mediated cellular injuries have been involved in the proapoptotic injury detected in the lungs of A30 mice.

FIG. 5.

p53-independent apoptosis pathways. (A) The expression of the E1A transgene induced TNF-α-mediated proapoptotic injury and/or proinflammation, and the levels of TNF-α in the blood of 3-month-old nontransgenic, A30, and A97 mice were quantified by TNF-α enzyme-linked immunosorbent assay (n = 2 to 4). (B) The expression of the E1A transgene induced Fas/FasL-mediated proapoptosis. The levels of Fas molecules in the lung lysates of 3-month-old nontransgenic (Non-Tg), A30, and A97 mice were detected by immunoblotting assay with anti-mouse Fas antibody. (C) The expression of FasL in the epithelial cells of bronchiole and alveoli of 3-month-old nontransgenic, A30, and A97 mice was detected by immunohistochemistry with anti-mouse FasL antibody. (D) The expression of E1A induced cell-mediated immune apoptosis, which was evaluated by cytometric analysis (FACS) (Table 1). The FACS results were confirmed by staining NK cells and macrophages with anti-Mac-1 antibody and by staining T cells with anti-CD8⁺ antibody in the paraffin-embedded lung sections of 3-month-old mice (arrows). The small windows in the top right corners show stained cells in blood vessels. (E) The ratio of the cells with Mac-1 mark or CD8⁺ mark was quantified (n = 3) to confirm that cell-mediated immune apoptotic injury occurred in the lungs of 3-month-old mice. Scale bars, 8 μm.

Cell-mediated immune apoptotic injury activated in the lung preferentially expressing the 289-aa protein.

The numbers of leukocytes were significantly increased in the lungs of A30 mice (Fig. 3B and C). In addition, swollen cervical necks with enlarged lymph nodes appeared in A30 mice (Fig. 3A). In A97 mice, the nodular lesions of lymphocyte-like cells frequently appeared in the lungs (Fig. 3C). To determine whether there was a correlation between the preferential expression of each Ad5 E1A protein and the host immune cell response, flow cytometry analysis (FACS) was used to quantify the leukocyte populations present in the lungs and the cervical neck lymph nodes (NLN) in transgenic mice and nontransgenic littermates. Relative to the level in nontransgenic littermates, by 3 months of age there were 1.3- and 3- to 4-fold increases in the total number of leukocytes recovered from the lungs and NLN of A30 mice, respectively. Although there was some variation among individual transgenic mice, over half of the A30 mice analyzed showed significant increases in the relative proportions of natural killer (NK) cells, macrophages/monocytes, gamma/delta TcR⁺ T cells, and the cells expressing MHC class I in the lungs and NLN, while the number of alpha/beta TcR⁺ T cells was relatively decreased (Table 1). Similar analyses for 11-month-old A97 mice revealed a 1.5-fold decrease and up to a 1.8-fold increase in the total number of leukocytes recovered from lungs and NLN, respectively. Among these cells there were no significant differences in the relative abundance of any subpopulation compared with those of nontransgenic littermates, except for significant decreases in the population of macrophages/monocytes and granulocytes and a decrease in the relative population of NK cells in the lungs (Table 1). The FACS results were confirmed by immunostaining of 3-month-old mouse lung sections with a number of comparable anti-leukocyte marker antibodies. In A30 mice, the percentages of macrophages/monocytes and NK cells were significantly higher (16.7% of the total lung cells stained for Mac-1 marker), while the percentage of CD8⁺ T cells was relatively lower (3.8% of the total lung cells stained for CD8 marker) than those in A97 mice (P < 0.01 for both values) (Fig. 5D and E). In A97 mice, the percentages of macrophages/monocytes and NK cells were relatively lower (5.6% of the total lung cells stained for Mac-1 marker) while the percentage of CD8⁺ T cells was relatively higher (9.4% of the total lung cells stained for CD8 marker) (Fig. 5D and E). Since TNF-α is one of the main cytotoxic cytokines used by activated macrophages or NK cells in macrophage- or NK cell-mediated cytolysis pathways (9, 10, 21), the increased proportion of macrophages and NK cells was comparable to the increased level of TNF-α in A30 mice. These results indicate a cell-mediated immune apoptotic response that involves the activation of cytolytic lymphocytes, NK cells, gamma/delta T cells, macrophages, and the cells expressing MHC class I in the lungs of A30 mice. In contrast, the decreased proportion of macrophages/monocytes, granulocytes, and NK cells and the increased proportion of T cells suggest nodular lesions of impaired lymphocyte-like cells in the lungs of A97 mice.

TABLE 1.

Altered subpopulations of leukocytes^a

Cell/staining marker	Fold increase over wild-type level
	A30 (3 mo old)		A97 (11 mo old)
	Lung	NLN	Lung	NLN
NK cell/NK1.1+	2.6	3.1	0.56	0.98
Monocyte/Mac-1+	2.1	8.2	0.20	0.49
Granulocyte/Gr-1+	1.2	5.9	0.16	0.96
γδ T cell/γδ TcR+	6.5	3.7	0.96	0.64
αβ T cell/αβTcR+	0.60	0.44	0.71	1.2
MHC class I/H2Lb+	2.1	3.9	1.1	0.76
MHC class II/I-A+	1.4	2.8	1.2	1.7

^aThe expression of Ad5 E1A transgene-induced cell-mediated immune apoptosis. The activated cytolytic leukocytes were evaluated by FACS. A population of each leukocyte in the transgenic lung or NLN greater or lower by 50% than those in nontransgenic lung or NLN was considered significant.

DISCUSSION

We aimed to investigate in vivo the general effects of the Ad5 E1A gene through the entire E1A protein interacting with numerous cellular proteins, rather than investigating in vitro the individual domain of the E1A protein. We wished to mimic the latent Ad5 E1A gene expression in the human lung during adenovirus latent infection in our transgenic mice. Therefore, we chose an Ad5 E1A genomic DNA as a transgene rather than Ad5 E1A cDNAs. The Ad5 E1A 243- and 289-aa proteins were unequally expressed in vivo in the transgenic lines we generated. The preferential expression of the 243-aa protein in most transgenic lines may reflect general RNA splicing and processing of the E1A pre-mRNA in vivo. More studies are required to elucidate the mechanism for the preferential expression of the 289-aa protein in the A30 line. However, we speculate that either the different integration site of the E1A transgene, which affects the regulation of splicing elements, or the apparent deletion that occurred in the SPC promoter region may have led to overaccumulation of the E1A 289-aa protein (18, 61). The evidence for selective splicing of E1A RNAs has been found during adenoviral infection of HeLa cells. During the early period of adenovirus infection, the downstream 13S mRNA 5′ splice site is preferentially used over the upstream 12S mRNA splice site, and the 13S mRNA predominates (19).

In the transgenic mice described here, the preferential expression of either the 243- or 289-aa protein has led to a distinct pathophysiology, allowing us to investigate their in vivo general effects. We believe that the phenotypic alterations observed in A97 and A30 mice are less likely to be the result of insertional mutation effects, i.e., disruption in the function of other genes, or to be specific to those mouse lines with different genetic backgrounds. This belief is based on our observation that two other transgenic lines, A2-20 and A3-02, also generated a phenotype similar to that of A97 mice. The similar phenotypic alterations were observed in heterozygote and homozygote A30 mice and not in the nontransgenic littermates. To exclude the possibility that the phenotypes developed in A30 mice were specific to this mouse line rather than occurring as a result of a general effect of the E1A 289-aa protein, the nontransgenic littermates were used as a control in our studies, so that any result specific to this line was carefully examined. The results for A30 transgenic mice were distinct from those for their nontransgenic littermates, and neither phenotypic alteration nor the result specific to this line was found in their nontransgenic littermates. Moreover, the distinct lung pathophysiological alterations observed in A97 and A30 mice are less likely to be the result of microbial respiratory infection and are more likely due to transgene effects. We believe that the lung histopathologies of Ad5 E1A transgenic mice are the consequences of the E1A protein interfering with many lung cellular functions. As a result, the E1A transgene effects may sensitize or predispose the transgenic mice to environmental stress, including respiratory infection.

The Ad5 E1A 243-aa protein appeared to promote cell proliferation in the lungs of A97 mice, which is in agreement with the results of previous in vitro studies (7, 36, 41), presumably leading to nodular lesions of proliferating lymphocyte-like cells in the lungs of A97 mice. This phenotype could be the result of an interaction of the 243-aa protein with transcriptional cofactors such as pRb family members, STAT, and p300/CBP through the NTR, CR1, and CR2 domains (Fig. 1). pRb, p300/CBP, and STAT play critical roles in the regulation of cell growth, differentiation, and the control of cellular immune response (1, 2, 27, 36, 45, 46, 48-50, 53, 57). Therefore, the putative repression of those transcriptional cofactors through binding to the E1A 243-aa protein in vivo may result in the upregulation of growth factors and/or growth cytokines and furthermore may lead to the induction of abnormal cell proliferation and nodular lesions of lymphocyte-like cells (30, 31, 42, 48, 51, 53). Compared with the significantly higher levels of Ad5 E1a 243-aa protein and carcinomas seen in the lungs of Ad5 E1A+E1B transgenic mice (data not shown), a low or moderate level of Ad5 E1A 243-aa protein alone thus far is not enough to induce carcinoma in the lungs of A97 mice. We showed that elevated levels of p53 were the consequence of increased cell proliferation. The same phenotype was also observed in Ad5 E1A+E1B transgenic mice, in which the increased levels of p53 notably followed the increased level of BrdU incorporation (data not shown). The evidence of proliferation-induced p53 is also reported in other in vitro studies, in which Ad5 E1A stimulated an unscheduled cellular DNA synthesis and then led to elevated expression of p53 (30, 40). Moreover, the proliferation-associated p53 likely leads to a p53-dependent apoptosis in the lungs of A97 mice. The E1A 243-aa-protein-accompanied apoptosis may attenuate the E1A 243-aa-protein-induced cell proliferation (52). Overall, those findings suggest that the Ad5 E1A 243-aa protein possesses an oncogenic proliferation activity in vivo.

Unlike in the case of the 243-aa protein, preferential expression of the 289-aa protein in A30 mice led to a proapoptotic injury or a proinflammation in their lungs rather than to cell proliferation. The upregulation of p53, TNF-α, Fas/FasL, and cell-mediated immune apoptosis suggests that multiple apoptosis pathways have been involved in those phenotypes caused by predominant expression of the 289-aa protein. The distinct phenotypes in the lung of A30 mice are probably due to the unique CR3 domain, presented only in the 289-aa protein, interacting with other transcription regulatory factors (Fig. 1) such as TBP, APs, and ATFs (6, 11, 12, 20, 26, 29). Therefore, extensive interaction of the 289-aa protein with those cellular factors by binding the CR3 domain may lead to release and activation of the transcriptional cofactors, such as p300/CBP, p53, pRb, and STAT, from the repressive effects of the NTR, CR1, and CR2 domains (1, 2, 20, 27, 28, 48, 50, 57). Those activated transcriptional cofactors in turn modulate the transcription of DNA damage response genes such as p53, Fas/FasL, and the proapoptosis cytokine TNF-α. The tumor suppressor p53 and the proapoptosis factors Fas/FasL play a central role in the regulation of cell proliferation and apoptosis (4, 13, 27, 37, 45, 52). The proapoptosis cytokine TNF-α mediates the proinflammatory response, induces cytokine production, suppresses proliferation, and triggers both p53-dependent and -independent apoptosis (5, 9, 10, 21, 22, 38, 47, 62). The activated NK cells, activated macrophages, gamma/delta T cells, and cells expressing MHC in the lungs of A30 mice, as components of cell-mediated immune apoptosis, may partially respond to cytolysis and upregulate the host immune antiviral response (8, 10, 25, 56). Therefore, those pathways integrally respond to the proapoptotic injury and/or proinflammation that occur in the lungs of A30 mice. The E1A 289-aa-protein-associated proapoptosis was distinct from the E1A 243-aa-protein-accompanied p53-dependent apoptosis.

The results obtained from analyses of A30 mice were strengthened by those of other Ad5 E1A function studies. Transfection of pancreatic islets with an E1A-13S gene expressing only the 289-aa protein induced extensive islet mass apoptosis or necrosis compared with that seen with nontransfected islets, while transfection of pancreatic islets with an E1A-12S gene expressing only the 243-aa protein extended the life span and functionality of pancreatic islets with few apoptotic cells (44). The elevated expression of the proapoptosis cytokine TNF-α in A30 mice was in agreement with the results of another previous in vitro study demonstrating a significant increase in TNF-α expression in inflammatory cell lines by infection with the E1A gene expressing only the 289-aa protein but not by the E1A gene expressing only the 243-aa protein (34). In addition, the upregulation of TNF-α was due to the E1A 289-aa protein trans-activating the TNF-α promoter (43). The same evidence that the induction of p53-independent apoptosis is dependent only upon CR3-mediated transactivation activity of the Ad5 E1A 289-aa protein was demonstrated in in vitro studies (59). Ad5 E1A expression sensitized mammalian cells to apoptosis triggered by cytolytic lymphocytes and TNF-α (8, 10), and we observed a similar phenotype in the lungs of A30 mice. The phenotype in which the elevated TNF-α correlated with severe pulmonary proinflammation was seen in the lungs of our A30 mice as well as in other experimental animals (21, 39, 47, 54). Moreover, the E1A 289-aa-protein-associated proapoptosis mainly appeared as DNA damage and/or nucleus fragmentation in the lung cells of A30 mice; similar results also were reported in other in vitro studies, in which Ad5 E1A expression correlated with cellular DNA damage or degradation (4, 5, 13). Overall, this evidence supports our hypothesis that the phenotypes and results with A30 mice are more likely to be an effect of the E1A 289-aa protein than to be specific to this mouse line.

Whether the increased numbers of the total leukocytes or the altered proportions of leukocyte subpopulations in the lungs and NLN contribute directly to the pathophysiology observed for those mice is not clear yet, although the effect was more pronounced and consistent in A30 mice. The prominence of the activated NK cells, activated macrophages, gamma/delta T cells, and cells expressing MHC in the resulting proinflammation suggests that these cell populations are important in adenovirus-mediated pulmonary inflammation. The influence on these cells probably results from the cytokines such as TNF-α (induced by expression of the E1A 289-aa protein). It is well-known that cytokines like TNF-α function to activate nonspecific immune cells, such as NK cells and macrophages, in response to a variety of stimuli (8, 10, 21, 38, 47, 54). The increased proapoptotic injury (approximately 12% of lung cells) in the lungs of A30 mice might explain their smaller lung size as well as their shortened life span (no more than 12 months) compared with those of their nontransgenic littermates (over 24 months) and A97 mice (over 16 months). In addition, how might these mice survive for 6 to 12 months with those stresses in the lungs? The majority of TUNEL-labeled lung epithelial cells were undergoing limited and slow DNA damage, and normally this DNA damage would be repaired and proapoptotic lung cells would still be active, not in the final phase of apoptosis. At the same time, there was still a low level of epithelial cell proliferation, which may partially compensate for some apoptotic epithelial cells. Moreover, the proapoptotic epithelial cells may be healed by constitutive regeneration of pluripotent epithelial cells. It is believed that populations of primitive pluripotent pulmonary epithelial cells are present in lung epithelium, which gives rise to both the alveolar and bronchial pulmonary epithelial cell lineages. These epithelial cell lineages proliferate transiently, although in general they maintain a quiescent state whereby under most circumstances neither proliferation nor cellular apoptosis is detected. However, after injury, the epithelium can regenerate and differentiate into terminally differentiated epithelial cells to compensate for those apoptotic epithelial cells (32).

While a few recent reports suggested a correlation existing between the pulmonary latent Ad5 E1A or E1 and the human pulmonary diseases (14, 23, 24, 33, 35), confirmation that latency of Ad5 E1A is related to clinical problems such as chronic respiratory diseases may have to await more-detailed clinic studies. However, Ad5 E1A protein was mainly presented in the pulmonary epithelial cells in either Ad5-infected human lung or Ad5-infected animal models (14, 24, 39, 55). This observation seems to be imitated in our E1A transgenic mice, in which Ad5 E1A protein was mainly expressed in the pulmonary epithelial cells. Taken together, our studies may provide some useful data regarding the in vivo general effects of each E1A protein or latent adenoviral infection. Predominant expression of the 243-aa protein would activate cell proliferation and nodular lesions of lymphocyte-like cells, which, to some extent, may potentially lead to chronic pulmonary diseases. If the 289-aa protein were preferentially expressed, a proapoptotic injury would be activated, leading to an acute pulmonary proinflammation. Overall, adenoviruses have developed both E1A functions to maintain both host and virus survival through a critical balance between these two E1A proteins. The 289-aa-protein-induced cell apoptosis restricts the 243-aa-protein-induced cell proliferation, and vice versa, cell proliferation rescues cell apoptosis. Alteration in either process may lead to susceptibility to nodular lesions of lymphocyte-like cells or proinflammation during Ad5 E1A latency in the lung.