Adenovirus Transduction: More complicated than receptor expression

Abstract

The abundance and accessibility of a primary virus receptor are critical factors that impact the susceptibility of a host cell to virus infection. The Coxsackievirus and adenovirus receptor (CAR) has two transmembrane isoforms that occur due to alternative splicing and differ in localization and function in polarized epithelia. To determine the relevance of isoform-specific expression across cell types, the abundance and localization of both isoforms were determined in ten common cell lines, and correlated with susceptibility to adenovirus transduction relative to polarized primary human airway epithelia. Data show that the gene and protein expression for each isoform of CAR varies significantly between cell lines and polarization, as indicated by high transepithelial resistance, is inversely related to adenovirus transduction. In summary, the variability of polarity and isoform-specific expression among model cells are critical parameters that must be considered when evaluating the clinical relevance of potential adenovirus-mediated gene therapy and anti-adenovirus strategies.

Graphical abstract

Differential accessibility of CAR for adenovirus infection of non-polarized or polarized cells. A. CAR^Ex7 and CAR^Ex8 are exposed and accessible to mediate the infection of incoming adenovirus in non-polarized epithelia. B. CAR^Ex7 localizes beneath the tight junctions at the basolateral surface of polarized epithelia and is not accessible to incoming adenovirus. C. CAR^Ex8 localizes at the apical surface and in the recycling endosomes of polarized epithelia. Apical CAR^Ex8 allows adenovirus to bind, internalize, and infect cells from the apical surface of polarized epithelia.

1. Introduction

Understanding the development, maintenance, and composition of the lung epithelial barrier is of great importance to human respiratory health and disease. This barrier segregates the microbial infested external environment from the body’s sterile internal environment. Each polarized epithelial cell has two distinct compartments: an apical or mucosal surface exposed to the air, and a basolateral surface in close communication with the internal environment. Investigations into the mechanisms of pathogenic microbial penetration of the lung epithelial barrier provide insights into its structure and regulation (Zihni et al., 2014). Interestingly, many viruses use basolateral adhesion proteins as their primary receptors. Three groups of viruses, coxsackie B viruses (CVB), all but group B human adenoviruses, and swine vesicular disease virus, utilize the coxsackievirus and adenovirus receptor (CAR) as their primary receptor (Bergelson et al., 1997; Carson et al., 1997; Tomko et al., 1997).

Several protein isoforms have been described for CAR (Excoffon et al., 2014; Excoffon et al., 2010; Raschperger et al., 2006). Alternative splicing not only regulates protein expression, but also allows multiple proteins to be expressed from the same gene resulting in significant proteomic diversity. Human CAR was initially isolated as a seven exon encoded protein (Bergelson et al., 1997). In contrast to other species, mouse CAR (mCAR) was initially cloned as a protein composed of eight exons (Tomko et al., 1997). Transcripts for the seven-exon mouse and eight-exon human forms were subsequently identified (Bergelson et al., 1998). A detailed analysis of protein expression and localization in mice has revealed differential tissue dependent expression and localization for the mCAR^Ex7 (mCAR2) and mCAR^Ex8 (mCAR1) isoforms (Excoffon et al., 2010; Mirza et al., 2006; Nalbantoglu et al., 2001; Nalbantoglu et al., 1999; Raschperger et al., 2006). This suggests that protein-protein interactions and potentially the functional importance of these two isoforms may be distinct. We are the first group to investigate the importance of human CAR^Ex8 in polarized epithelia (Excoffon et al., 2010; Kolawole et al., 2012; Kotha et al., 2015; Sharma et al., 2012a; Sharma et al., 2012b; Yan et al., 2015). The alternative splicing event that creates CAR^Ex8 occurs at a cryptic splice site within the seventh exon. Thus, these two isoforms contain identical extracellular and transmembrane domains. The majority of the cytoplasmic C-terminal domain is also identical except for the last 26 amino acids encoded by exon 7 (CAR^Ex7) that are replaced by 13 distinct amino acids encoded by exon 8 (CAR^Ex8).

Along with others, we have previously shown that CAR^Ex7 resides on the basolateral surface of polarized epithelia where it behaves as a homophilic epithelial junction adhesion protein and a heterophilic leukocyte adhesion protein that plays a role in neutrophil transepithelial migration and gammadelta T cell activation (Cohen et al., 2001; Excoffon et al., 2010; Kotha et al., 2015; Verdino et al., 2010; Walters et al., 2002; Witherden et al., 2010; Zen et al., 2005). CAR^Ex7 also plays a role in trafficking proteins to cell-cell junctions (Coyne et al., 2004; Excoffon et al., 2004; Excoffon et al., 2012; Lim et al., 2008; Mirza et al., 2005; Sollerbrant et al., 2003; Weber et al., 2014; Yan et al., 2015).

CAR^Ex8 can also behave as a homo- and heterophilic adhesion protein; however, in contrast to CAR^Ex7, CAR^Ex8 is found at the apical surface of polarized epithelia where it is unlikely to interact with CAR molecules on adjacent cells (Excoffon et al., 2010). Instead, we have shown that CAR^Ex8 mediates heterophilic adhesion with neutrophils that have transmigrated from the basal to apical surface (Kotha et al., 2015). The apical CAR^Ex8 isoform also mediates adenovirus entry from the apical surface and, surprisingly, neutrophils further enhance adenovirus entry. The protein levels of apical CAR^Ex8 are tightly regulated by intracellular proteins, such as the cytoplasmic scaffolding protein MAGI-1, and are increased by extracellular signals such as interleukin-8 (IL-8) and side-stream cigarette smoke (Excoffon et al., 2010; Kolawole et al., 2012; Kotha et al., 2015; Sharma et al., 2012a).

The abundance and accessibility of a primary viral receptor is a critical factor that defines the susceptibility of a host cell to viral infection. Many distinct model cells have been investigated for CAR-binding virus infection. In this manuscript we characterize cellular polarity and isoform-specific expression of both transmembrane isoforms of CAR in eleven commonly studied cell types. The cell lines tested were divided in two groups based on their polarity; 1) non-polarized cells (CHO-K1, COS-7, HeLa, HEK-293, 293T and A549) that are unable to form tight junctions and hence unable to polarize at the air-liquid interface; 2) cells able to form tight junctions (MDCK, Caco-2, Calu-3, NuLi-1 and HAE) and therefore polarize on a semipermeable membrane resulting in two distinct compartments (apical and basolateral). Together, these data will allow those investigating CAR-binding virus infections a clearer understanding of cellular polarity and receptor availability that will enhance data interpretation and advance the understanding of the factors that regulate susceptibility to infection.

2. Materials and methods

2.1 Cell Lines

Ten immortal cell lines and primary human airway epithelia were investigated: Chinese hamster ovary (CHO-K1, ATCC, CCL-61), transformed monkey kidney (COS-7, ATCC, CRL-1651), human cervical adenocarcinoma (HeLa, ATCC- CCL-2), human embryonic kidney (HEK-293, ATCC, CRL-1573), subclone of human embryonic kidney (293T, Clontech, Mountain View, CA), human lung carcinoma (A549, ATCC, CCL-185), Madin Darby canine kidney (MDCK, ATCC, CCL-34), human colonic adenocarcinoma (Caco-2, ATCC, HTB-37), human lung adenocarcinoma (Calu-3, ATCC, HTB-55), normal lung transformed epithelial cells (NuLi-1, a kind gift from Dr. Joseph Zabner, University of Iowa) and human airway epithelial (HAE, Human Donors)

HAE were originally isolated under approval by the Institutional Review Board of the University of Iowa (IRB ID No. 9507432) from discarded and deidentified trachea and bronchi of donor lungs. This study used discarded lung tissue, thus IRB deemed consent was not needed. HAE were cultured and differentiated as previously described (Karp et al., 2002; Kotha et al., 2015; Liu et al., 2012).

2.2 Culture Media

The cells were cultured in appropriate sterile culture medium (Gibco™, Invitrogen Corporation, Grand Island, NY) supplemented with 10% fetal bovine serum (except MDCK, supplemented with 5% FBS). All the culture media were further supplemented with 0.5% penicillin and streptomycin antibiotics and with 3.024 g/L sodium bicarbonate, 0.5% by appropriate volume percentage.

2.3 Total Cellular RNA Extraction

Total cellular RNA was extracted from cells using TRIzol Reagent (Invitrogen Life Technologies, Carlsbad, CA) according to the manufacturer’s protocol. The RNA pellet was dissolved in diethyl pyrocarbonate (DEPC)-treated ddH₂O and the purity and concentration of the extracted RNA were determined with a nanodrop spectrophotometer (Thermo Scientific, Waltham, MA).

2.4 Quantitative reverse transcription and PCR (qPCR)

Complementary DNA (cDNA) was synthesized using Quanta First Strand Kit (Quanta BioSciences, Gaithersburg, MD). qPCR was performed using SYBR Green with low ROX (Quanta BioSciences) in Stratagene's Real Time PCR System (Agilent Technologies, Santa Clara, CA) using 3000p software v2.0.1 for data analysis as previously described (Kotha et al., 2015; Sharma et al., 2012a; Sharma et al., 2012b). Primers used were: CAR-qPCR-F: 5’ TCGGCAGTAATCATTCATCCCTGG; CAR^Ex7-qPCR-R: 5′ ATAGACCCATCCTTGCTCTGTGCT; CAR^EX8-qPCR-R: 5’ ACTGTAATTCCATCAGTCTTGTAAGGG. Abundance relative to GAPDH gene expression was calculated for each gene of interest in human cells. GAPDH-F: 5’ CACCCTGTTGCTGTAGCCAAA; GAPDH-R: 5’ CAACAGCGACACCCACTCCT. In dog cells, CAR was normalized to MDCK-Actin-F: 5’ AAGATCTGGCACCACACCTTCTAC; MDCK-Actin-R: 5’ ATCTGGGTCATCTTCTCACGGTTG (Kotha et al., 2015). In CHO cells, CAR was normalized to CHO-Actin-F-5’ TGGCATCCACGAAACTACAT; CHO-Actin-R-5’ TGGTACCACCAGACAGCACT.

2.5 Immunocytochemistry

Cells seeded and polarized on millicells (0.4 µm pore; Millipore, St. Louis, MO) were kept on ice for 5 min, and washed 3 times with ice-cold PBS supplemented with Mg²⁺ and Ca²⁺ (PBS +/+). The cells were then fixed with ice-cold methanol containing 1% paraformaldehyde for 20 min at −20°C. Cells were rinsed with ice-cold PBS +/+, allowed to come to room temperature, and blocked with 2% bovine serum albumin (BSA) in SuperBlock (Pierce, Rockford, IL) for 45 min. Primary antibodies, anti- rabbit 1605 (rα1605) for total CAR, anti-rabbit 5678 (rα5678) and anti-mouse actin as previously published (Excoffon et al., 2014; Excoffon et al., 2010; Excoffon et al., 2005) were added to cells for 2.5 h at 37°C or overnight at 4°C. Cells were rinsed, reblocked, and then incubated with secondary antibodies (Invitrogen, Carlsbad, CA). The cells were rinsed and mounted onto glass slides using Vectashield mounting media with DAPI (Vector Laboratories Inc., Burlingame, CA). Staining was evaluated by laser scanning confocal microscopy (Olympus FV1000) with a 60× oil immersion lens; images are shown as single X-Y sections.

2.6 Cell Polarization

For polarization studies, 2 × 10⁴ – 2 × 10⁶ cells per well were seeded on 12 mm diameter polyester Millicell filters consisting of a semi-permeable membrane with a pore size of 0.4 µm (Millipore, St. Louis, MO). Media on the apical surface of cells was removed every alternate day in order to establish and maintain an air-liquid interface. Polarized cells actively transport fluid from the apical to the basolateral surface and thus maintain a defined apical surface fluid composition.

2.7 Transepithelial electrical resistance (TER) measurement

Transepithelial electrical resistance was measured with a chopstick ohmmeter (World Precision Instruments, Sarasota, FL) every other day. Media was aspirated from the wells and replaced with 400 µL fresh media at the basolateral surface. The same amount of PBS +/+ was applied to the apical surface. The background electrical resistance was determined by adding media to the basolateral chamber and PBS +/+ to the apical chamber of a blank Millicell filter. The TER measurements were recorded as Ω•cm².

2.8 Western blotting

Cell culture plates were placed on ice for 5 min, washed with ice-cold PBS +/+, and lysed in buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% Triton X-100, protease inhibitors (20 µg/ml, leupeptin, aprotinin, 10 µg/ml pepstatin, and 17.4 µg/ml phenylmethylsulfonyl fluoride)) by rocking at 4°C for 10 min. Cells were scraped into a tube, sonicated five times and centrifuged at 17000 × g for 10 minutes at 4°C. The supernatant was transferred to fresh tubes and subjected to protein estimation by Bio-Rad protein assay according to manufacturer’s instructions. Aliquots of each lysate were used for SDS polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride (PVDF) membrane (Millipore, Bedford, MA), blocked with 5% BSA in TBST, washed, and probed with primary antibody, as indicated in the text, followed by HRP-conjugated secondary antibodies (Jackson Immuno Research, West Grove, PA). Protein bands were detected with ECL reagents (Pierce, Rockford, IL) and imaged on a Fuji LAS 4000 and/or developed in an X-ray Medical film processor (Konica SRX 101). Blots shown where chosen to show bands in all wells. Quantitation was performed at multiple exposures for blots from three separate experiments (see also Table 1).

Table 1.

Calu-3 and NuLi-1 cells have similar CAR^Ex7 and CAR^Ex8 gene and protein expression relative to HAE. Ratio of CAREx7:CAREx8 gene and protein expression in cell lines as determined by qRT-PCR and Western blot.

	Cell lines CAR Isoform Gene Expression (Relative to HAE)			CAR Isoform Protein Expression (Relative to actin)
	Ex7	Ex8	Ex7:Ex8	Ex7	Ex8	Ex7:Ex8
CHO-K1	N.D.	N.D.	N.D.	N.D.	N.D	N.D.
COS-7	1.5	0.6	2.6	2.83	0.52	5.49
HeLa	0.4	0.8	0.4	3.70	1.67	2.21
HEK-293	0.4	0.7	0.6	6.38	3.27	1.95
293T	0.4	0.7	0.6	5.24	6.30	0.83
A549	0.1	0.6	0.2	0.98	1.37	0.72
MDCK	4.8	0.0	641.4	3.69	0.39	9.48
Caco-2	0.5	1.7	0.3	3.43	0.73	4.66
Calu-3	0.7	0.6	1.1	3.67	1.47	2.50
NuLi-1	0.5	0.4	1.4	0.83	0.63	1.33
HAE	1.0	1.0	1.0	0.86	1.08	0.80

2.9 Adenovirus transduction

The apical surfaces of cells, seeded on millicells, were rinsed with PBS+/+. Recombinant human adenovirus type 5 containing the β-galactosidase gene (AdV5-β-Gal) driven by the CMV-IE promoter (University of Iowa Vector Core, Iowa City, IA), diluted in Opti-Mem serum-free culture media (Gibco, Invitrogen), was added to the apical surface at a multiplicity of infection (MOI) of 100 pfu/cell and returned to the incubator. After 1 h, the inoculum was removed and cells were rinsed once with Opti-MEM. Cells were lysed 24 h later and analyzed by β-galactosidase assay (Galactolight Plus β-Galalactosidase Reporter Gene Assay, Fisher Scientific). Luminescence values were normalized to the amount of protein (Bio-Rad protein assay). Error bars represent standard error over three experiments with 3–5 replicates per cell line per experiment.

3. Results

3.1 MDCK, Caco-2, Calu-3, NuLi-1 and HAE cells polarize into epithelia

Many cells transform from non-polarized to polarized states during development. Epithelial polarization occurs by the formation of tight junctions and membrane segregation into apical and basolateral surfaces, each with a distinct distribution of proteins and lipids. Furthermore, these unique surfaces are known to mediate differential susceptibility to pathogens (Vogelmann et al., 2004). The formation, polarization, and integrity of epithelial tight junctions of various commonly used immortal cell lines were determined and compared to primary human airway epithelial cells. The cells were seeded onto semi-permeable membranes and transepithelial electrical resistance (TER) was measured and recorded over a 10 day period. Of the epithelial cells known to polarize (MDCK, Caco-2, Calu-3, NuLi-1 and HAE), MDCK and Caco-2 cells displayed the most rapid increase in TER after seeding with a marked TER within 2 days after seeding (Fig. 1). Calu-3, NuLi-1, and HAE cells were also able to polarize although the polarization was not as rapid as MDCK and Caco-2 cells. It took about 5 days for these other epithelial cell types to attain a resistance above background (> 300 Ω·cm²), indicative of a functional tight junction. Furthermore, upon polarization, primary human airway epithelial cells (HAE) were able to differentiate and be maintained at an air-liquid interface for more than 2 months with TER between 1000–2500 Ω·cm². By contrast, CHO-K1, COS-7, HeLa, HEK-293, 293T, and A549 never increased above the background levels (dotted line) when grown on a semipermeable membrane, indicating that these cells fail to fully polarize (Fig. 1).

Figure 1.

Transepithelial electrical resistance (TER) increases over time in cell lines able to polarize into epithelia. No increase in the TER above background level (grey dotted line) was observed in cell lines unable to polarize on semipermeable membranes (CHO-K1, COS-7, HeLa, HEK-293, 293T and A549). By contrast, a rapid increase in TER over time was observed for most epithelial cell lines (MDCK, Caco-2, Calu-3, NuLi-1, and primary human airway epithelia (HAE)).

3.2 CAR^Ex7 isoform-specific RNA is more abundant than CAR^Ex8 RNA in cells expressing both isoforms

Previously we have shown that mRNA for CAR^Ex7 is more abundant than mRNA for CAR^Ex8 in primary human airway epithelia (HAE) (Excoffon et al., 2010). To determine the relative abundance of these two isoforms in other cell lines, RNA was isolated from each cell line and gene expression was quantified by quantitative RT-PCR (qPCR). RNA levels were measured with respect to β-actin and normalized to the CAR negative CHO-K1 cell line (Table 1) or HAE (Fig. 2 A and B and Table 1). With the exception of CHO-K1 cells, all the cell lines tested expressed RNA specific for each isoform. We first compared relative gene expression to CHO-K1 in order to illustrate the difference from baseline. Among the cell lines expressing both CAR isoforms, similar to HAE, CAR^Ex7 transcript levels were more abundant than CAR^Ex8 transcript levels (Table 1). However, the relative abundance of CAR isoforms varied significantly between cell lines, with the greatest isoform difference found in MDCK cells. We next compared expression relative to HAE and found that MDCK and COS-7 cells have much more CAR^Ex7 relative to CAR^Ex8 than HAE, Calu-3 and NuLi-1 were similar to HAE, and the other cell lines had a lower ratio.

Figure 2.

CAR^Ex7 transcript and protein levels are more abundant than CAR^Ex8 transcript and protein levels. A) CAR^Ex7 or B) CAR^Ex8 gene expression was quantified by qPCR with exon specific primers after cDNA synthesis from total RNA obtained from non-polarized and polarized cell types grown on semi-permeable membranes. Gene expression shown is relative to HAE (see also Table 1). A comparison to CHO-K1 cells is shown in Supplementary Table 1. C–F) Lysate from each cell type was subjected to SDS–PAGE and analyzed by Western blotting. Total CAR expression was detected in C) the cell lines that do not polarize or D) polarized cells with a CAR-specific polyclonal antibody (rabbit anti-CAR-1605p) that recognizes the C-terminus of both CAR isoforms. CAR^Ex8 specific protein expression was detected in E) the cell lines that do not polarize or F) polarized cells by a CAR^Ex8-specific antibody (rabbit anti-CAR^Ex8-5678p). As expected, no CAR^Ex7 or CAR^Ex8 protein was detectable in CHO-K1 cells.

3.3 CAR protein expression varies between cell lines

CAR^Ex7 and CAR^Ex8 protein expression was determined in ten different cell lines and HAE. Western blot with a rabbit polyclonal Ab that detects both isoforms (rα1605) (Excoffon et al., 2005; Sharma et al., 2012a) demonstrated a CAR-specific band at approximately 46 kDa in all cell lines except CHO-K1 (Fig. 2). We found that CAR protein expression levels vary greatly between the investigated cell lines. In non-polarized cells, HEK-293 cells expressed the highest level of total CAR (Fig. 2C) and CAR^Ex8 protein (Fig. 2D). In the case of polarized cells, MDCK cells had the highest protein expression of the CAR^Ex7 isoform (Fig. 2D) and, consistent with the mRNA expression results, MDCK cells also had very low endogenous protein levels for the apical CAR^Ex8 isoform (Fig. 2E and F). These data support the use of the MDCK cell line as a suitable cell line for studying CAR^Ex7 with little CAR^Ex8 to confound results. The protein ratio of the two transmembrane isoforms studied in all the cell lines is shown in Table 1 and suggests that the CAR gene transcript levels in MDCK, NuLi-1 and HAE correlate relatively well to the amount of protein expression, but do not correlate in the other cell lines. These data suggest that while all tested cell lines are candidates for CAR expression studies, the variability of CAR isoform gene and protein expression should be considered when choosing a cell line for the desired experiment.

3.4 CAR is localized at cell junctions

To determine the localization of CAR, cells were seeded and polarized on semi-permeable membranes, after which the cells were fixed and subjected to immunocytochemistry for total CAR or CAR^Ex8 (Fig. 3 A, B, C and D). Both CAR isoforms (green) were localized at the cell junctions of non-polarized A549 and COS-7 cells (total CAR, Fig. 3A; CAR^Ex8, Fig. 3B). Polarized cell lines showed robust basolateral junctional localization of CAR^Ex7 (Fig. 3A and C). The smooth “chickenwire” outline surrounding the cells suggests that tight junctions are well formed in these cells. This is in contrast to A549 and COS-7 cells where CAR is much less organized and does not form a smooth chickenwire pattern around each cell suggesting a lack of polarization (Sharma et al., 2012b). CAR^Ex8 was mainly found in within the cytoplasm region consistent with the sub-apical compartment (Fig. 3D). In summary, in non-polarized immortal cell lines, both CAR^Ex7 and CAR^Ex8 isoforms are localized at cell junctions and no differential localization is observed. By contrast, immortal polarizable cell lines express both CAR isoforms in a similar manner to primary human airway epithelia. The characteristics shown by these cell lines confirm that they are suitable epithelial cells lines for investigating endogenous CAR isoform-specific expression in polarized epithelia.

Figure 3.

CAR^Ex7 localizes at basolateral junctions while CAR^Ex8 localizes within the subapical and apical compartments and epithelial polarization decreases susceptibility to adenovirus transduction. A) and B) Immunocytochemistry for total CAR staining (rabbit anti-CAR-1605p) in green and nuclei in blue (DAPI). A) In non-polarized cells (A549 and COS-7), CAR^Ex7 staining and localization is discontinuous due to lack of formation of tight junction, B) whereas CAR^Ex7 localization in polarized cells (MDCK, Caco-2, Calu-3, and HAE) presents as a “chicken wire” pattern due to the formation of tight junctions. C) and D) CAR^Ex8 localization (rabbit anti-CAR^Ex8-5678p) is shown in green and nuclei in blue (DAPI). C) CAR^Ex8 localization is similar to CAR^Ex7 in non-polarized cells. D) However, CAR^Ex8 localization in well differentiated and polarized epithelial cells is distinct and is found in the cytoplasm consistent with the subapical compartment. (60× oil immersion confocal microscopy, white bar equals 10 µm. E) All cell types were seeded onto semi-permeable membranes and maintained at the air-liquid interface prior to infection from the “apical” surface with recombinant adenovirus carrying the gene for beta-galactosidase (AdV5-Beta-Gal) and Beta-galactosidase activity evaluated 24 h post-transduction.

3.5 Adenovirus transduction depends on primary receptor expression and cell polarity

We next correlated the expression of the CAR isoforms to adenovirus transduction. All of the cell lines were seeded on semipermeable membranes, allowed to polarize, infected with recombinant adenovirus containing beta-galactosidase gene (AdV5-β-Gal), and assayed using a luminescence assay relative to the amount of protein (Fig. 3E). With the exception of CAR-negative CHO-K1 cells, all the non-polarized cells were much more susceptible to adenovirus infection when compared to polarized cells. Among the non-polarized cells, HEK-293 and 293T demonstrated the highest expression of β-Galactosidase, which directly correlates with the primary receptor expression. Also, HEK-293 and 293T cells have an insertion of adenovirus E1A and E1B genes that transcomplement the production of first generation E1/E3 deleted recombinant vectors, such as the AdV5-β-Gal vector used here (Graham et al., 1977), and therefore also reflect viral replication. In the case of polarized cells, epithelia were approximately 3 log less susceptible to adenovirus infection and demonstrated little variability in apical adenovirus infection.

4. Discussion

In contrast to previous beliefs that adenoviral infection occurs only from the basolateral surface of polarized cells, recent evidence shows that viral infection can occur at the apical surface of polarized epithelia (Excoffon et al., 2014; Excoffon et al., 2010; Kolawole et al., 2012; Kotha et al., 2015; Lutschg et al., 2011; Sharma et al., 2012a; Sharma et al., 2012b). CAR has two transmembrane isoforms, CAR^Ex7 and CAR^Ex8, each having a unique ability to initiate viral entry from distinct epithelial surfaces (Excoffon et al., 2014). This raises the question of the abundance and cellular localization of the two CAR isoforms, and their relative contribution to viral infection of commonly used cells lines. Ten immortal cell lines that have previously been used to study the function of viral receptors were investigated including: CHO-K1, COS-7, HeLa, HEK-293, 293T, A549, MDCK, Caco-2, Calu-3, NuLi-1, and compared to primary HAE. As expected, the two airway epithelial model cell lines, Calu-3 and NuLi-1, were most similar to HAE.

Cell lines were first tested for their ability to polarize and form tight junctions. Cells grown on semipermeable membranes were effaced every other day in order to determine the ability for each line to form electrically tight junctions and a unique “dry” apical surface at the air-liquid interface. As expected, media from the basolateral surface leaked into the apical compartment of millicells containing CHO-K1, COS-7, HeLa, HEK-293, 293T, or A549 cells. By contrast MDCK, Caco-2, Calu-3, NuLi-1, and HAE developed a “dry” apical surface and high transepithelial resistance. All the immortal polarized cell lines were found to polarize similarly to airway epithelia but the number of days they took to achieve polarity is different (Fig. 1). Moreover, of these 3 cell lines, MDCK cells polarize most rapidly, confirming that MDCK cells are an ideal line for investigating rapid epithelial polarization.

We next examined the expression and localization of both CAR isoforms (Fig. 2, 3, Table 1, S1). With the exception of CHO-K1 cells, all cell lines were found to contain a sufficient amount of RNA for CAR^Ex8 protein production. CAR^Ex7 transcript and protein levels were consistently more abundant than CAR^Ex8 (Fig. 2, Table S1), both isoforms are expressed in multiple cell lines. Although most cell types had similar CAR^Ex7 and CAR^Ex8 transcript levels, higher CAR^Ex7 transcript levels were observed in COS-7 and MDCK, lower CAR^Ex7 transcript levels were found in A549 cells, and very low CAR^Ex8 transcript levels were found in MDCK. A few studies have shown that the STAT1 pathway and transcription factors, SP1, SNAIL1, and SMAD3/4, transcriptionally regulate CAR (Chung et al., 2011; Ruppert et al., 2008; Vincent et al., 2009). The only cell line overlapping with our study is HeLa, and it was shown that deletion of the SP1 promoter motif impaired CXADR (CAR gene) promoter activity (Chung et al., 2011). How each of these pathways impact transcriptional regulation of CXADR in the other cell lines in our study is currently unclear. Moreover, to our knowledge, no studies have investigated the regulation of alternative mRNA splicing for CAR. Given the variability in CAR isoform transcript levels in COS-7, A549, and MDCK cells, these likely represent candidate cell lines to elucidate novel CAR-isoform specific transcriptional regulation pathways.

With the exception of MDCK, NuLi-1, and HAE, transcript levels did not directly correlate with protein levels (Fig. 2). Multiple factors could play a role in translational regulation. For example, we have previously shown that the activation of the Akt pathway via IL-8 stimulation increases CAR^Ex8 protein synthesis (Kotha et al., 2015), whereas inhibition of GSK-3β increased both transcript and proteins levels for CAR^Ex8 (Sharma et al., 2012a). Both pathways lead to increased apical localization of CAR^Ex8 and increased adenovirus transduction.

Both isoforms localize to cell-cell junctions in non-polarized cells (Fig. 3.). In polarized MDCK, Caco-2, Calu-3 and HAE epithelia, CAR^Ex7 shows robust junctional localization, whereas little CAR^Ex8 is present at basolateral junctions. Differential isoform expression and localization has previously been investigated in mice. Raschperger et al compared CAR transmembrane isoform expression in tissues from adult mice using isoform specific antibodies (Raschperger et al., 2006). The murine CAR^Ex8 equivalent, mCAR-1 (TVV), and CAR^Ex7, mCAR-2 (SIV), were predominately expressed in epithelial cells lining organs, although immunoreactivity was also found in endothelial cells, at the neuromuscular junction, and at cardiac intercalated discs. Although differences in localization were primarily observed in liver and kidney tissues, large differences in the relative amount of mCAR-1 protein expression was found between liver, kidney, pancreas and colon, and relative to mCAR-2. Differential isoform localization has also been shown in skeletal muscle, where mCAR-2 is the primary isoform at the neuromuscular junction in mice and both CAR^Ex7 and CAR^Ex8 are present in human (Shaw et al., 2004). Perhaps the most striking difference in isoform localization has been observed in mature spermatozoa where each isoform was present in a distinct region of the acrosome (Mirza et al., 2006), however, subsequent studies that conditionally knocked out both isoforms in mice did not show any defect in spermatogenesis or fertilization (Sultana et al., 2014). These findings are consistent with differential localization of the two CAR transmembrane isoforms in multiple cell types in mice and in humans.

Finally we investigated the susceptibility of each cell line to adenovirus transduction after growth on a semipermeable membrane (Fig. 3). Adenovirus transduction was approximately 3 logs lower in polarized epithelia in comparison to non-polarized cells. The variability in adenovirus-mediated gene expression was primarily observed in non-polarized cells and ranged from no transduction of CHO-K1 cells (CAR negative) to the highest expression in HEK-293 and 293T cells. HEK-293 and 293T contain adenovirus 5 DNA and transcomplement the production of adenovectors that have genomic deletions that render them replication defective in the absence of transcomplementation. Since a high level of AdV5-β-Gal replication will have occurred in these cells by 24 h, β-galactosidase activity does not solely reflect viral transduction as it does in the other cell types. Moreover, the CMV promoter is known to vary from cell type to cell type by up to a log difference in gene expression (Qin et al., 2010; Xia et al., 2006). Given the large 3 log difference between polarized and non-polarized cells, it is unlikely that a difference in promoter or enzymatic activity would explain our findings. Transduction was evaluated rather than viral binding since binding may not lead to viral entry. Moreover, not all adenovirus genomes that enter cells end up in the nucleus where genome transcription occurs. Transduction is a readout for the genomes that reach the nucleus and can successfully express a transgene – a critical parameter for gene therapy and wild type infection alike. Future studies will focus on the correlations between binding, entry, transduction, and wild type adenovirus infection in polarized cell types.

The cell lines in this study were chosen based on their instrumental roles in the initial discovery and advancement of our understanding of CAR and adenovirus infection. For example, CHO-K1 and HeLa cells were critical in the discovery of CAR and its alternative splice products, adenovirus co-receptors, and the step-wise dismantling of adenovirus during infection (Bergelson et al., 1997; Carson et al., 1997; Greber et al., 1993; Thoelen et al., 2001; Wickham et al., 1993). A549 cells have been used extensively in the production, characterization, and titering of wild type adenovirus. Interpretation of studies in this cell line should take into account that A549 cells demonstrate only partial polarization and therefore do not reflect infection from distinct apical or basolateral surfaces. MDCK cells have been used to elucidate the clathrin-mediated trafficking of CAR^Ex7 in polarized cells (Carvajal-Gonzalez et al., 2012; Diaz et al., 2009). We consider MDCK to be a particularly strong experimental system given the high amount of endogenous CAR^Ex7, but low levels of CAR^Ex8. We have taken advantage of this imbalance and created doxycycline-inducible MDCK cell lines that contain the cDNAs for human CAR^Ex8, CAR^Ex7, or mCherry, to investigate the important physiological effects of CAR^Ex8 expression on viral infection and innate immunity (Kotha et al., 2015). CAR^Ex8 overexpression leads to increased apical trafficking of CAR^Ex8, which, in turn, increases adenoviral infection from the apical surface. This evidence suggests that, contrary to a decrease in TER or a break in the junctions, apical adenovirus infection can occur by apically localized CAR^Ex8.

Applications of these data include the ability to test innovative medical treatments for respiratory infection. The effect that CAR^Ex8 exerts on viral entry in polarized cells could potentially be targeted to develop novel therapies that regulate receptor expression and/or trafficking. Down-regulation of CAR^Ex8 could prevent adenovirus-mediated exacerbation of diseases, such as asthma, cystic fibrosis (CF), and chronic obstructive pulmonary disease (COPD), and protect immunosuppressed populations. Moreover, the up-regulation of CAR^Ex8 could be used to facilitate viral entry during gene therapy using adenovirus as vectors. Understanding CAR expression and function in epithelial cells may facilitate the development and evaluation of novel treatment options for both new and persistent respiratory and gastrointestinal infections in humans.

Highlights.

• -Variability in CAR expression in model cell lines is compared to primary epithelia
• -Polarization reduces adenovirus transduction efficiency
• -Localization of CAR transmembrane isoforms impacts adenovirus transduction