ACE2 Receptor Expression and Severe Acute Respiratory Syndrome Coronavirus Infection Depend on Differentiation of Human Airway Epithelia

Primary culturesof human airway epithelia were prepared from trachea or bronchi cellsand grown at the air-liquid interface (ALI) on collagen-coated porousfilters as described previously(17). In selectexperiments, the apical surface of primary cells was submerged under500 μl of cell culture medium for 7 days to inducededifferentiation or grown in nonpolarized fashion on tissue cultureplastic. This study was approved by the Institutional Review Board atthe University of Iowa. A549 cells (ATCC CCL-185) were maintained inDulbecco's modified Eagle's medium/F12 (Gibco) containing10% fetal bovine serum, penicillin (100 U/ml), and streptomycin (100mg/ml).

Cells or tissues were lysed in 0.1%Triton X-100 in phosphate-buffered saline (PBS), and total protein wasseparated on a sodium dodecyl sulfate-7.5% polyacrylamide gelelectrophoresis gel and transferred to polyvinylidenedifluoride membrane. Goat anti-ACE2 polyclonal primaryantibody (catalog no. AF933, R & D Systems,Minneapolis, MN), horseradish peroxidase-conjugated donkey anti-goatsecondary antibody, mouse anti-foxj1 monoclonal primary antibody (giftof S. Brody, Washington University), and a horseradishperoxidase-conjugated goat anti-mouse secondary antibody were used.Primary antibody binding was visualized using SuperSignalchemiluminescent substrate (Pierce, Rockford,IL).

Airway epitheliawere fixed in 4% paraformaldehyde for 5 min at room temperature andwashed with PBS. Five percent bovine serum albumin in PBS was used toblock nonspecific antibody binding. An anti-ACE2 monoclonal antibody(MAB933; R & D Systems, Minneapolis, MN) was applied to cells (bothapical and basolateral surfaces for air-liquid interface-culturedcells). Epithelia were then incubated at 37°C for 1 hand again washed with PBS. Rabbit anti-mouse fluoresceinisothiocyanate-conjugated secondary antibody (F-4143;Sigma, St. Louis, MO) was added to cells and incubated at 4°Covernight. Cells were then washed with PBS, and a Cy3-labeled mouseanti-β-tubulin IV monoclonal antibody (C-4585; Sigma) wasadded. Cells were incubated at 37°C for 1 h, followedby PBS washes, and then mounted with DAPI(4′,6′-diamidino-2-phenylindole) VectaShield (VectorLabs, Burlingame, CA). Nuclei were stained with To-pro-3 (T3605; Sigma,St. Louis, MO). The SARS-CoV nsp1 protein was localized in cells fixedwith 100% methanol using a rabbit polyclonal anti-nsp1 primary antibody(gift of M. Denison) (33)and a mouse anti-rabbit fluorescein isothiocyanate-conjugated secondaryantibody (catalog no. A11088, Molecular Probes, Eugene,OR).

The apical or basolateral surfaces ofairway epithelia were treated with 1 mg/mlN-hydroxysulfosuccinimidobiotin (Pierce, Rockford, IL) in PBSfor 30 min at 25°C. The epithelial surface was washed and thenincubated with 100 mM glycine in PBS for 20 min at 25°C toquench unreacted biotin. Epithelial protein lysates were prepared bysonication in lysis buffer, and biotinylated proteins were precipitatedusing neutravidin covalently linked to immobilized diaminodipropylamine(Pierce). Biotinylated proteins were released in 8% sodium dodecylsulfate-containing loading buffer, boiled, and analyzed byimmunoblotting for ACE2. ErB2 was detected as a control basolateralmembrane marker (43)using rabbit polyclonal anti-c-erbB2 antibody (Dako Corporation,Carpinteria, CA).

Epitheliawere fixed in 2.5% glutaraldehyde in Na cacodylate buffer for 30 minand rinsed with 0.1 M Na cacodylate buffer three times. Samples werepostfixed in 1% OsO₄ for 1 h and sequentiallyrinsed with 0.1 M Na cacodylate buffer and H₂O. Samples werethen serially dehydrated using 25% to 100% ethanol. Aftercritical-point drying, samples were mounted on stubs, sputter coated,and examined using a Hitachi F-4000 electronmicroscope.

Total cellular RNA was isolated usingTRI-Reagent (MRC, Cincinnati, OH) or a commercial spin column isolationkit (Stratagene, La Jolla, CA), and 1 μg was reversetranscribed using a reverse transcription (RT)-PCR kit (Ambion, Austin,TX). An aliquot of cDNA was subjected to PCR using an iCycler iQfluorescence thermocycler (Bio-Rad, Hercules, CA) with SYBR green I DNAdye (Molecular Probes, Eugene, OR) and Platinum Taq DNApolymerase (iQ SYBR green Supermix kit, Bio-Rad). PCR conditionsincluded denaturation at 95°C for 3 min and then for 35 cyclesof 94°C for 30 s, 60°C for 30 s,and 72°C for 30 s, followed by 5 min at 72°Cfor elongation. The following primers were used: (i) human ACE2 forward5′-GGACCCAGGAAATGTTCAGA-3′ andreverse5′-GGCTGCAGAAAGTGACATGA-3′,(ii) SARS-CoV N gene forward/leader5′-ATATTAGGTTTTTACCCAGG-3′and reverse5′-CTTGCCCCATTGCGTCCTCC-3′, (iii)SARS-CoV S gene forward/leader5′-ATATTAGGTTTTTACCCAGG-3′ andreverse5′-CTCCTGAGGGAACAACCCTG-3′, and(iv) human hypoxanthine phosphoribosyltransferase (HPRT) forward5′-CCTCATGGACTGATTATGGAC-3′and reverse5′-CAGATTCAACTTGCGCTCATC-3′.Fluorescence data was captured during the 10-s dwell to ensure thatprimer dimers were not contributing to the fluorescence signal, andspecificity of the amplification was confirmed using melting curveanalysis. Data was collected and recorded by iCycler iQ software(Bio-Rad) and initially determined as a function of threshold cycle(C_T). Levels of mRNA were expressed relative tocontrol HPRT levels, which were calculated as2^ΔCT. In some samples, PCR products were visualized on 2% agarose gels withethidium bromide.

E1-deleted replicationincompetent adenoviral vectors expressing human ACE2, Escherichiacoli β-galactosidase, and foxj1 under the control of thecytomegalovirus promoter were produced as previouslydescribed(1).

The SARS-CoV Sprotein cDNA (Urbani strain “S-H2,” as describedreference 21) was used topseudotype feline immunodeficiency virus (FIV) expressing anuclear-targeted β-galactosidase by using previously describedmethods (45).The virus was concentrated 250-fold by centrifugation, andtiters were determined on HT1080 cells, obtaining titers of ∼2×10⁵ to 4 × 10⁶transducing units/ml. Well-differentiated human airway epithelia weretransduced with the pseudotyped FIV by applying 100 μl ofsolution to the apical surface of airway epithelia. After 1 hof incubation at 37°C, virus was removed and cells wereincubated at 37°C for 2 days. To infect epithelia from thebasolateral side, the Millicell insert was turned over, and virus wasapplied to the basolateral surface for 1 h in100 μl of medium. Following the 1-h infection,virus was removed and the insert was turned upright and incubated at37°C in 5% CO₂ for 2days.

The Galacto-Lightchemiluminescent reporter assay (Tropix, Bedford, MA) was used toquantify β-galactosidase activity following the manufacturer'sprotocol. Relative light units were quantified using a luminometer(Monolight 3010; Pharmingen, San Diego, CA) and standardized to totalprotein in the sample.

SARS-CoV (Urbani strain) was producedin Vero E6 cells under BSL3 containment conditions. Virus titers weredetermined on Vero E6 cells with typical yields of ∼4 ×10⁶ PFU/ml. Epithelia derived from three donors wereinfected in duplicate with SARS-CoV from the apical surface(multiplicity of infection [MOI], 0.8). Following a 30-min incubationat 37°C, virus was removed and 10 washes with PBS wereperformed. A sample was collected by adding and removing500 μl of medium from the apical side of oneepithelium from each donor. A total of 500 μl of medium wasadded to the basolateral side of the remaining epithelia, and24 h later, samples were collected for titerdeterminations.

ACE2 serves as the SARS-CoV receptor,at least for tissue culture cell entry(21). To determine ifACE2 is also the SARS-CoV receptor in the respiratory tract, we firstlooked for evidence of ACE2 protein expression in human lung tissue byWestern blotting. An immunoreactive band consistent with theglycosylated form of ACE2 was identified in lysates from human airwayand distal lung tissue (Fig.1A). However, the cell type expressing ACE2 could not be identified in thisexperiment, as both endothelial and epithelial cells can express ACE2.Accordingly, we next evaluated ACE2 protein expression inwell-differentiated primary cultures of airway epithelia usingimmunohistochemistry and observed airway epithelial cell expression,with the most abundant signal for ACE2 on the apical rather than thebasolateral surface(Fig. 1B).Furthermore, the signal intensity was strongest on ciliated cells, asdemonstrated by colocalization with the β-tubulin IV marker ofcilia (23), suggestingthat ciliated cells express ACE2 abundantly. To confirm a polardistribution of ACE2 in differentiated epithelia, selective apical orbasolateral surface biotinylation with subsequent precipitation wasperformed (Fig. 1C).Immunoblot analysis of precipitated proteins confirmed that ACE2 isexpressed in greater abundance on the apical surface of conductingairway epithelia, although in some experiments a weak ACE2 signal wasalso detected basolaterally. In contrast, ErbB2, the receptor forheregulin-alpha, was more abundant on the basolateral surface aspreviously reported (43),confirming selective biotinylation.

Results from polarizedepithelia suggested that ACE2 expression might depend on the state ofcellular differentiation. To validate a model to test this hypothesis,we used SEM to compare the apical surface morphology ofwell-differentiated epithelia with that of well-differentiated cellssubsequently grown with medium present on their apical surfaces for 7days. Of note, submersion of the apical surfaces of primary cellscaused loss of cilia (Fig.2A). Loss of this important marker of cell differentiationwith resubmersion was associated with markedly diminished expression ofACE2 mRNA and protein (Fig. 2B andC). In contrast to results in polarized epithelia, poorlydifferentiated primary human tracheobronchial epithelia (hTBE) or A549cells grown on tissue culture plastic expressed little ACE2 mRNA orprotein. Interestingly, foxj1, a transcription factor required for awell-differentiated ciliated epithelia phenotype(3) was also coordinatelyexpressed with ACE2, indicating that ACE2 positively correlates withthe state of epithelial differentiation. We also asked whether foxj1might directly regulate ACE2 expression in airway epithelia. Primarytracheobronchial cells grown in submersion culture were transduced withan adenoviral vector expressing ACE2, a negative-controlβ-galactosidase, or foxj1. Only transduction with the ACE2vector conferred ACE2 expression, suggesting that foxj1 expressionalone does not regulate ACE2 (Fig.2D and data notshown).

To evaluate thepolarity of entry of the SARS-CoV in airway epithelia, we prepared FIVvirions pseudotyped with SARS S protein. The ACE2 dependence oftransduction with SARS S protein-pseudotyped FIV virions was firstevaluated on 293 cells with or without cotransfection with human ACE2cDNA. We observed that 293 cell transduction with this vector wasalmost completely ACE2-dependent (Fig.3A). To establish the ACE2 dependence of human airwayepithelia for SARS-CoV, we transduced poorly differentiated A549 cellsand submerged primary hTBE cells that do not express constitutive ACE2with increasing amounts of an adenoviral vector expressing human ACE2.After 48 h, SARS-CoV S protein-pseudotyped-FIV was applied tothe apical surface at an MOI of 0.1 (based on HT1080 titers). There wasan inoculum-dependent increase in the transduction of cells thatexpressed ACE2 (Fig. 3B).S protein-pseudotyped FIV was then used to contrast the efficiency ofentry in A549 cells, poorly differentiated (submerged) human airwayepithelia, and well-differentiated (ALI) epithelia. Onlywell-differentiated epithelial cells showed significantβ-galactosidase expression following transduction (Fig.3C). We next applied thepseudotyped virus to the apical or basolateral surfaces ofwell-differentiated primary cultures of human airway epithelia toinvestigate whether the virus preferentially entered from one cellsurface. After 2 days, the cells were harvested and entry was evaluatedby β-galactosidase activity. Transduction of human airwayepithelia by the S protein-pseudotyped virions occurred moreefficiently when applied from the apical rather than the basolateralsurface(Fig. 3D).This pattern of entry correlates with ACE2 expression on polarizedcells (Fig. 1). Incontrast, FIV pseudotyped with the vesicular stomatitisvirus-G envelope entered polarized cells better from thebasolateral surface.

We went on to perform select experimentsusing wild-type SARS-CoV (Urbani strain). Using protocols similar tothose outlined for Fig. 3,we evaluated the ability of SARS-CoV to infect multiple human airwayepithelial cell culture models. Under BSL3 containment, we applied thevirus to A549 cells, poorly differentiated (submerged) primary culturesof airway epithelia, or well-differentiated (ALI) human airwayepithelia at an MOI of 0.8. A549 and hTBE cells cultured undersubmerged conditions expressed little detectable SARS-CoV N or S genemRNA after infection (Fig.4A). In contrast, in well-differentiated cells infected with SARS-CoV fromthe apical surface, the N and S gene mRNAs were detected at highlevels. We confirmed that the gene products detected in real-timeRT-PCR assays were generated from new SARS-CoV mRNA templates ratherthan viral genome by verifying the appropriate size of the amplifiedproducts (Fig. 4B). Theprimers used included a forward primer in the conserved SARS CoV5′ leader sequence and gene-specific reverse primers. Theseresults suggest that SARS-CoV infects undifferentiated human airwayepithelial cells poorly or not at all, while well-differentiated airwayepithelia are susceptible.

To document that SARS-CoVproductively infects human airway epithelia, we applied the virus tothe apical surface of well-differentiated human airway epithelia at anMOI of ∼0.8 for 30 min and then measured therelease of virus 24 h later. As shown in Table1, these results indicate that, following apical application of SARS-CoV,a productive infection occurred and virus was preferentially releasedapically. We confirmed SARS-CoV infection of polarized epithelia byimmunostaining cells for the SARS-CoV nsp1 protein 24 hfollowing infection. Confocal microscopy revealed viral nsp1 protein inthe cytoplasm, consistent with replication complexes, following apicalapplication of the virus(Fig. 4C).Control, noninfected cells showed no staining. The infection ofpermissive Vero E6 cells showed intense cytoplasmic staining only inthe presence of SARS-CoV infection (data not shown). To better definethe cell types infected by SARS-CoV in this model, we colocalized thensp1 and β-tubulin IV proteins (Fig.4D). Nsp1 staining wasobserved in perinuclear regions and distributed throughout thecytoplasm. In addition, we observed some areas of colocalization ofnsp1 and β-tubulin IV, suggesting that replication complexesmay assemble in the cytoplasm near or within cilia. These findingsindicate that ciliated cells are the predominant cell type infected bySARS-CoV in well-differentiated airwayepithelia.