INTRODUCTION
Influenza virus, a member of the
Orthomyxoviridae family of viruses, is a human pathogen of great interest that causes respiratory tract infections that result in yearly epidemics and occasional severe pandemics (
7). Influenza virus primarily infects epithelial cells of the upper respiratory tract, but certain highly pathogenic strains can infect pneumocytes of the lower respiratory tract as well (
26,
34,
38,
41). Additionally, we have recently shown that murine respiratory dendritic cells (RDC) are susceptible to infection with at least one influenza virus strain (A/Japan/305/57) (
13). It is has not been established whether different influenza virus strains can infect RDC with comparable efficiencies. Previous studies evaluating DC susceptibility to infection by influenza virus have utilized splenic and bone marrow-derived DC, both of which differ in properties from DC resident in the respiratory tract (
4,
27,
28). Therefore, very little is currently known about the interaction of influenza virus with tissue-resident RDC that encounter infectious influenza virus during natural infection in the respiratory tract.
Dendritic cells (DC) are a distinct lineage of hematopoietic cells derived from myeloid and lymphoid progenitors in the bone marrow that reside in both lymphoid organs and peripheral tissues, where they play key roles in the immune response to viral infection (
3,
11). As innate defenders in the periphery, DC recognize virus-associated molecular patterns via endosomal Toll-like receptors and cytoplasmic RIG-I-like receptors and subsequently produce a variety of inflammatory cytokines and chemokines involved in immune cell activation and recruitment to the site of infection. Additionally, specific subsets of DC have been shown to produce molecules, i.e., type I interferons, with direct antiviral activity (
14,
16,
22,
25). Furthermore, maturation of DC that acquire viral antigens (Ag) in the periphery, either through phagocytosis of infected cells or direct infection by the virus, is followed by migration to draining lymphoid organs, where the DC serve as antigen-presenting cells (APC) that stimulate the adaptive immune response (
12,
15).
The plethora of functions exhibited by DC is highlighted by the recent identification of a number of unique tissue-specific subsets of these cells. Within the respiratory tract, there are at least four distinct RDC subsets that have been identified thus far (
18,
35,
39,
40). Airway CD103
+ RDC are localized to intraepithelial and subepithelial spaces (and perivascular sites), and recent work in our laboratory demonstrated that, following influenza virus infection, these cells acquire influenza virus Ag and migrate to lung-draining lymph nodes, where they present Ag to both CD4
+ and CD8
+ T cells and induce their proliferation and differentiation into effector cells (
15). A subset of CD103
− CD11b
hi RDC localized to the lung interstitium and perivascular regions also present influenza Ag to CD4
+ T cells and induce effector differentiation in lung-draining lymph nodes (
15). Plasmacytoid DC (pDC) in the lung parenchyma, although less abundant than CD103
+ and CD11b
hi RDC, function as potent producers of type I interferons (
9,
17) and are required for influenza virus-specific antibody production (
12). Finally, a substantial population of monocytic RDC (MoRDC) also resides in the lung interstitium, but the function of these cells has not been well characterized. Because many, if not all, of these RDC subsets play a major role in the anti-influenza virus immune response, it is vital that we gain a better understanding of their interaction with influenza virus.
The influenza virus hemagglutinin (HA) serves as the receptor for virus attachment to cells, a necessary first step for virus infection. The HA recognizes and binds terminal sialic acids in certain preferred carbohydrate linkages potentially displayed by a diverse array of glycolipids and glycoproteins on the cell surface (
10,
33,
36), although there is evidence that influenza virus may preferentially interact with one or more host cell N-linked glycoproteins for efficient entry and infection (
6). In this study, we isolated murine RDC from healthy (uninfected) lungs (
13,
24) as a source of cells for
in vitro analysis of susceptibility to infection by type A influenza virus strains of different subtypes. We found that, among the major RDC subsets, only the CD103
+ and CD11b
hi RDC subsets were susceptible to infection by type A influenza virus either
in vitro or
in vivo. Of note, these two RDC subsets also express the highest basal level of major histocompatibility complex (MHC) class II molecules. Furthermore, only type A influenza virus strains of the H2N2 subtype could efficiently infect these RDC subsets. A link between influenza virus infectivity and MHC class II expression was suggested by the finding that binding of monoclonal antibody to the I-E MHC class II locus product substantially inhibited infection of these two RDC subsets by the H2N2 subtype viruses. The potential significance of these findings with respect to influenza virus infection in the respiratory tract is discussed here.
MATERIALS AND METHODS
Mice.
Eight- to 10-week-old female BALB/c (H-2d) and C57BL/6 (H-2b) mice were purchased from Taconic Farms (Germantown, NY) and maintained in a pathogen-free environment. All experiments were performed in accordance with regulatory standards and guidelines approved by the University of Virginia Animal Care and Use Committee.
Antibodies.
The following monoclonal antibodies were used in the analysis of RDC populations: anti-MHC class II-fluorescein isothiocyanate (FITC) (2G9), I-Ad-FITC (AMS-32.1), I-E-phycoerythrin (PE) (14-4-4S), CD11c-PE-Cy7 (HL3), CD45R/B220-PerCP (RA3-6B2), and CD11b-PerCP-Cy5.5 (M1/70) (all purchased from BD Pharmingen, San Diego, CA). CD103-PE (2E7) was purchased from eBioscience (San Diego, CA); monoclonal antinucleoprotein (anti-NP) (H16) was a gift from Walter Gerhard (Wistar Institute, Philadelphia, PA). The anti-NP antibody was conjugated to Alexa Fluor 647 dye in accordance with the instructions provided with a conjugation kit purchased from Molecular Probes (Eugene, OR).
For antibody blockade experiments, RDC were resuspended in fluorescence-activated cell sorter (FACS) buffer without NaN3 and Fc blocked for 5 min at 4°C before being incubated for 30 min in the presence of 5 μg of one of the following unconjugated antibodies: anti-I-Ed (14-4-4s), anti-I-Ad (AMS-32.1), anti-H2-Dd (34.2.12), or mouse IgG2a isotype control (MOPC-173).
Cell lines.
Madin-Darby canine kidney (MDCK) cells were obtained from the American Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle medium (DMEM) (Gibco BRL, Gaithersburg, MD) supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals, Norcross, GA), 2 mM l-glutamine, 100 U/ml penicillin (Gibco BRL), 100 μg/ml streptomycin (Gibco BRL), and 50 mM 2-ME (2-mercaptoethanol). Cells were passaged at 80% confluence.
Preparation and administration of Flt3L plasmid.
The pUMVC3-hFlex plasmid encoding Flt3 ligand (Flt3L) was obtained from Hardy Kornfeld (University of Massachusetts Medical School, Worcester, MA). Escherichia coli One Shot Top10 competent cells (Invitrogen Life Technologies, Carlsbad, CA) were transformed with the plasmid and grown to a large quantity. The Flt3L plasmid was purified from the transformed bacteria by the use of an EndoFree plasmid Giga kit (Qiagen, Valencia, CA). No residual endotoxin was detected in the plasmid preparation, as measured by the Limulus amebocyte lysate assay (Charles River Laboratories, Wilmington, MA). Flt3L-encoding plasmid was resuspended in sterile 0.9% saline solution at a final concentration of 5 μg/ml, and 2 ml of the preparation was rapidly injected intravenously (i.v.) via the lateral tail vein of each mouse over 5 to 10 s in order to achieve hypotonic in vivo transfection. This procedure was repeated 6 days later, and the lungs of plasmid-treated mice were isolated for RDC preparation 6 days after the second plasmid administration.
Preparation of lung single-cell suspension.
Mice were sacrificed by cervical dislocation. The lungs were perfused via the right ventricle of the heart with 5 ml of sterile phosphate-buffered saline (PBS) to remove the intravascular pool of cells from the lung vasculature. Lungs were minced and digested in 183 U/ml type II collagenase (Worthington, Lakewood, NJ) in Iscove's modified Dulbecco's medium (IMDM) (Gibco BRL, Gaithersburg, MD) at 37°C in 7% CO2 for 40 min. Afterward, the minced lung tissue was homogenized, passed through cell strainers (BD), and washed twice with IMDM. Lung cell pellets were resuspended either in magnetism-activated cell sorter (MACS) buffer (PBS supplemented with 0.5% bovine serum albumin [BSA] and 2 mM EDTA [Promega Corporation, Madison, WI]) for RDC isolation or in FACS lysing solution (BD, San Jose, CA) for flow cytometry staining.
Purification of RDC.
Lung single-cell suspensions were magnetically labeled with anti-CD11c microbeads (N418) purchased from Miltenyi Biotec GmbH and isolated using a MACS according to the protocols of the manufacturer. Preparations were consistently >98% CD11c+ (data not shown).
Viruses and infection.
Mouse-adapted influenza A/PR/8 virus (H1N1) and A/Japan/305/57 (H2N2) virus were grown in hen eggs that had been embryonated for 10 days and stored at −80°C. Other wild-type influenza A virus strains used included A/Bel (H1N1), A/Taiwan/1/64 (H2N2), A/HK/68 (H3N2), and A/Mem/1/71 (H3N2). Reassortant influenza virus strains used included A/Jap/Bel (H2N1), A/Nip/Bel (H2N1), A/Ned/Bel (H2N1), and A/Mem/Bel (H3N1). The recombinant influenza A virus PR/8-Japan HA strain was generated using a reverse-genetics approach and pPOLI plasmids carrying the hemagglutinin of influenza A/Japan/305/57 virus and the remaining seven gene segments of influenza A/PR/8 virus.
For infection of RDC in vitro, purified RDC from either untreated or Flt3L plasmid-treated mice were seeded at a concentration of 1.0e6 cells/ml in a 6-well flat-bottom plate (Corning Incorporated, Corning, NY) and incubated for 1 h in complete medium (IMDM supplemented with 2 mM l-glutamine, 10% fetal bovine serum [Atlanta Biologicals, Norcross, GA], 100 U/ml penicillin [Gibco BRL], 100 μg/ml streptomycin [Gibco BRL], and 50 mM 2-ME). Following the 1-h incubation, RDC were washed once with prewarmed IMDM to remove residual serum proteins and nonadherent cells in the cultures. Serum-free IMDM (1 ml) was added to the wells, and virus was added at a multiplicity of infection (MOI) of 10. RDC were then incubated on ice for 10 min, followed by 1 h of incubation at 37°C in 7% CO2. Following this incubation, RDC were washed twice with serum-free IMDM and then cultured in 1 ml of complete medium for the indicated time period. In some experiments, virus was subjected to heat inactivation at 55°C for 30 min prior to infection; in those cases, heat inactivation reduced the hemagglutinating activity of the virus 2-fold (data not shown), and we compensated for this reduction by doubling the inoculum size (i.e., equivalent to an MOI of 20) used to treat the RDC.
For in vivo infections, mice were lightly anesthetized with halothane prior to intranasal (i.n.) infection with a 50-μl volume of influenza virus diluted in serum-free IMDM. Lungs of infected mice were harvested as described at 24 h postinfection (p.i.).
MDCK cells were infected in serum-free IMDM suspensions as indicated for 10 min on ice followed by 1 h at 37°C in 7% CO2 with light agitation every 5 to 10 min. Following infection, cells were washed twice with serum-free IMDM, resuspended in MDCK culture media, and plated in 6-well flat-bottom plates at a concentration of 1.0e6 cells/ml for the indicated time period.
Intracellular and surface staining for flow cytometry analysis.
Purified RDC or MDCK cells from in vitro cultures or total lung cell suspensions from infected mice were fixed in FACS lysing solution for 10 min at room temperature, followed by resuspension in FACS buffer (PBS supplemented with 0.5% BSA and 0.02% NaN3). Cells were Fc blocked for 5 min at 4°C and then surface stained by incubation with the indicated monoclonal antibodies for 30 min at 4°C in the dark. Cells were then washed with FACS buffer and permeabilized with Perm/Wash (BD). To block nonspecific intracellular antibody binding, cells were incubated with Perm/Wash containing 5% healthy goat serum for 20 min on ice in the dark. Following this incubation, Alexa Fluor 647-conjugated anti-NP antibody was added directly to the cells for 30 min at 4°C in the dark. Labeled cells were detected by flow cytometry using either a FACSCalibur system (BD, Mountain View, CA) or a FACS Canto system (BD, Mountain View, CA) and were analyzed using Flowjo software (Tree Star).
Real-time RT-PCR.
For analysis of influenza virus NP gene expression in infected RDC, in vitro-cultured RDC were harvested and lysed with buffer RLT for RNA extraction using an RNeasy kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. Extracts were digested with amplification grade DNase I (Invitrogen Life Technologies, Carlsbad, CA), and total RNA was quantified using a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE). Total RNA (0.5 μg) was used for reverse transcription-PCR (RT-PCR) with random hexamer primers (Invitrogen Life Technologies, Carlsbad, CA) to generate cDNA. Real-time PCR was performed using an ABI 7000 instrument with cDNA samples, specific influenza virus NP primers, and SYBR green technology (Applied Biosystems, Foster City, CA). The A/Japan/305/57 NP primers used were as follows: forward, 5′-TCTTGTCTGCCTGCCTGTGT-3′; reverse, 5′-GTGTGCTGGATTCTCGTTCG-3′. The influenza A/PR/8 virus NP primers used were as follows: forward, 5′-GAACAACCGTTATGGCAGCA-3′; reverse, 5′-CATGTCAAAGGAAGGCACGA-3′. The 18s rRNA primers used were as follows: forward, 5′-TCGAACGTCTGCCCTATCAA-3′; reverse, 5′-GGGTCGGGAGTGGGTAATTT-3′. Relative NP gene expression levels were calculated by normalizing influenza virus NP threshold cycle (CT) values against 18s rRNA CT values for each sample.
DISCUSSION
RDC represent a heterogeneous set of peripheral tissue-resident DC that both function as innate effector cells and orchestrate the induction of adaptive immune responses to microorganisms and inert antigens present in the respiratory tract. In this report, we characterize the interaction between RDC and influenza virus in vitro and in vivo in the murine model. We demonstrate that specific RDC subsets are differentially susceptible to infection by influenza virus and that infection of these RDC is dependent on the virus strain. The MHC class IIhi intrapepithelial and subepithelial CD103+ RDC and the interstitial CD103− CD11bhi RDC are efficiently infected by the influenza A/Japan/305/57 virus strain both in vitro and in vivo, whereas MHC class IIlo MoRDC and pDC are not susceptible to infection following exposure to this virus. In contrast, influenza A/PR/8 virus does not efficiently infect any of these RDC subsets. Interestingly, we found efficient infection of RDC by a number of H2N2 influenza virus strains and poor infection by several H1N1 and H3N2 strains. Evidence from the infection of RDC by reassortant influenza virus strains and a reverse genetics-constructed influenza A/PR/8-Japan HA recombinant virus demonstrates that expression of HA of the H2N2 subtype facilitates efficient infection of murine RDC by type A influenza viruses. The susceptibility of RDC to infection by H2N2 virus was dependent upon their expression of MHC class II molecules, with the MHC class IIhi CD103+ and CD11bhi RDC subsets displaying the greatest susceptibility to infection. Importantly, susceptibility was further shown to be associated with expression by RDC of the murine MHC class II locus I-E; also of interest, binding of antibody to I-E on RDC prior to virus exposure inhibited virus infectivity.
The restriction of infection of RDC to influenza virus strains expressing HA of the H2N2 subtype was unexpected, as was the requirement for expression of the murine MHC class II I-E locus product by the RDC. Influenza viruses bind to cell surfaces through HA-mediated recognition of terminal sialic acids displayed by cell surface molecules. Although the HA molecules of different type A influenza virus subtypes differ in specificity of sialic acid recognition according to sialic acid linkage to subterminal sugars (
34), it is unlikely that the HA of an H1N1 influenza virus strain such as influenza A/PR/8 virus would differ sufficiently from influenza A/Japan/305/57 virus HA in sialic acid binding specificity to render the virus incapable of binding murine RDC. In this regard, it is noteworthy that both of these mouse-adapted type A influenza virus strains are capable of high-level replication in the murine respiratory tract (
19,
20,
21), again suggesting that the influenza A/PR/8 virus is fully capable of efficiently binding to murine cells. We did not formally evaluate the impact of binding of antibody to MHC class II on the extent of virus binding to the RDC. Since RDC display abundant sialic acids on their surface glycoproteins and glycolipids, we considered it unlikely that blocking the interaction of virus with MHC class II molecules would lead to a diminution of virus binding to the RDC cell surface that was sufficient to be detected by standard quantitative means. Instead, the link between virus infectivity and MHC class II expression by RDC could argue for a role of MHC molecules in facilitating infection of cells by influenza virus. However, most cell types susceptible to influenza virus infection, most notably ciliated and nonciliated respiratory (airway) epithelial cells, do not express MHC class II molecules. Consequently, the MHC class II dependence of susceptibility to influenza virus infection observed in this study would most likely be restricted to infection of DC and related cell types.
At present, the mechanism accounting for the dependence of RDC infection by influenza virus on MHC class II expression and on hemagglutinin of the H2N2 subtype is not as yet known with certainty. As discussed above, it is unlikely that any subtle differences in the sialic acid specificity of the influenza A/PR/8 virus or influenza A/Japan/305/57 virus HA would affect the initial virus attachment to cells. However, although it seems unlikely, we cannot formally exclude the possibility that the MHC class II I-E molecule displays a specific carbohydrate array recognized and found only by HAs of the H2N2 subtype. Nevertheless, the impact of MHC class II and the subtype of the type A influenza virus HA would most likely be at the level of virus uptake by RDC or postentry events associated with virus replication. In this regard, following exposure of RDC to virus strains (such as influenza A/PR/8 virus) whose HA is nonpermissive, we were unable to detect any de novo viral gene expression and amplification in RDC exposed to such viruses. Thus, a block to nonpermissive influenza A/PR/8 virus replication postentry into RDC would be at an extremely early step, presumably following virus uncoating and viral genome entry into the cell cytoplasm and certainly prior to any de novo viral gene expression in the cell nucleus. Also the finding that the susceptibility of RDC to infection maps to the HA gene (as demonstrated by the efficient infection of RDC by the influenza A/PR/8-Japan HA recombinant virus) further argues against a direct effect in the postfusion function of the RNP and polymerase complexes in the cytosol or nucleus, as these genes and their products are identical in the wild-type and the recombinant virus.
An alternative explanation is suggested by the requirement for I-E expression by RDC to support infection by H2N2-subtype influenza virus as well as by the ability of antibody to the I-E gene products to inhibit infection. These findings raise the possibility that the I-E molecule may serve as a novel receptor or coreceptor required to initiate and/or support infection of murine RDC by H2N2-subtype influenza viruses. MHC class II has been reported to serve as a receptor for other viruses (
8,
23). Furthermore, recent evidence has emerged to suggest that, in addition to the requirement for binding to sialic acid ligands, influenza virus must also interact with one or more cell surface N-linked glycoproteins as a requirement for infection. These N-linked glycosylated glycoproteins are not required for influenza virus binding but are required for virus internalization (
6). MHC class II I-A and I-E locus-encoded molecules are known to differ with respect to the sites and diversity of N-linked glycosylation (
37). Since, as noted above, MHC class II expression is not required for the infection of most cell types by type A influenza viruses, the role of MHC class II as a coreceptor and the link to the H2N2-subtype HA may reflect a feature of virus uptake and endocytosis unique to dendritic cells.
The idea of an interaction between influenza viruses (specifically influenza virus HA) and MHC class II locus molecules is not without precedent. An earlier study (
5) demonstrated that type A influenza viruses, and influenza virus strains of the H2N2 subtype in particular, serve as lymphocyte mitogens. Subsequent studies refined this phenomenon and demonstrated that influenza virus HA (and specifically the HA of the H2 and H6 subtypes) provided a strong mitogenic stimulus to murine B cells (
1,
2,
30,
31,
32). This mitogenic activity was dependent on a direct interaction between the influenza HA and the MHC class II I-E gene product. The earlier observations raise the possibility that in RDC, the H2N2-subtype HA may, through an interaction with the I-E molecule, provide the activation stimulus necessary to support virus replication in RDC. To address this issue, we infected RDC simultaneously with heat-inactivated influenza A/Japan/305/57 virus, which does not replicate in RDC but retains a functional HA molecule, and wild-type influenza A/PR/8 virus to determine whether the interaction between H2 HA from influenza A/Japan/305/57 virus and I-E
d allowed the RDC to support replication of A/PR/8. However, influenza A/PR/8 virus failed to replicate efficiently in the RDC exposed to the influenza A/Japan/305/57 H2 virus HA (data not shown). In companion experiments, we were likewise unable to prevent infection of RDC by influenza A/Japan/305/57 virus following pretreatment of the cells with influenza A/PR/8 virus, implying that the encounter of RDC with this virus strain does not result in active suppression of influenza replication. Thus, we are unable to identify a virus strain-dependent activating or suppressive signal triggered by a virus interaction with RDC which could account for this restricted virus tropism.
We also observed that
in situ (
in vivo) infection of RDC by influenza A/Japan/305/57 virus and influenza A/PR/8 influenza virus strains directly paralleled the pattern of infection observed
in vitro, i.e., RDC were readily infected by the prototype H2N2 strain but only minimally infected by the prototype H1N1 strain, as determined by NP expression in a flow cytometry-based analysis performed 24 h following intranasal influenza virus infection. Since, as we and others have demonstrated (
12,
15), RDC which have migrated during influenza virus infection from the respiratory tract to the draining lymph nodes play an essential role in initiating the adaptive immune T cell response, these results might imply that infection of RDC by influenza A/PR/8 virus is not essential for efficient presentation of viral Ag to naïve T cells in the draining nodes. In this connection, we had previously reported in studies that employed infections by both influenza A/Japan/305/57 virus (
21) and influenza A/PR/8 virus (
15) that NP
+ migrant RDC are demonstrable in the lung draining lymph nodes at day 3 following infection. While influenza A/PR/8 virus only minimally infects lung-resident RDC
in vivo during the first 24 h following intranasal virus inoculation, we have observed in preliminary studies that, by day 3 postinfection, NP
+ lung-resident DC are readily demonstrable in the influenza A/PR/8 virus-infected respiratory tract (K. M. Hargadon and T. J. Braciale, unpublished data). The reason for this transition in susceptibility status within the RDC population from NP
− (uninfected) to NP
+ (and presumably infected) is currently unknown. It has, however, been well documented that, following infection of the respiratory tract, there is an influx of DC into the lungs in response to this inflammatory stimulus (
12,
15), and the properties of these infiltrating DC are likely to be different from those resident at the onset of an infection.
This report provides the first demonstration that bona fide RDC can differ with respect to susceptibility to infection by different influenza virus strains. Moreover, the link between RDC expression of the I-E MHC class II locus and infection by influenza virus strains of the H2N2 subtype suggests a potential role for MHC polymorphisms in regulating the susceptibility of RDC and other cells of the respiratory tract to infection by different influenza virus strains. Such a role would have significant implications for the relevance of this finding to human infection by influenza virus. Therefore, the impact of RDC infection by influenza virus on the immune response to the virus is of great interest, and we are currently exploring whether differences in the susceptibilities of RDC to infection by different influenza virus strains have an impact on the efficiency of virus clearance or, alternatively, viral virulence. The differences in susceptibility of RDC to infection with the influenza A/Japan/305/57 virus and the influenza A/PR/8 virus strains do not appear to have a significant effect on the capacity of RDC to serve as APC for the induction of the adaptive immune T cell response to these two viruses. Infection of RDC may, on the other hand, significantly affect the early innate immune response in the respiratory tract. Indeed, we have observed that type 1 interferon gene expression is dramatically upregulated in vivo following infection by influenza A/Japan/305/57 virus but is absent following infection by influenza A/PR/8 virus; this finding correlates with the type I interferon gene and protein expression profiles of RDC exposed to these virus strains in vitro (Hargadon and Braciale, unpublished). It should be interesting to evaluate additional cytokine and chemokine profiles for RDC following exposure to different influenza virus strains. The mechanistic basis for the virus strain-dependent differences in immune responses made to infection, the role of these responses in immune-mediated virus clearance versus pathogenesis, and the role of RDC in regulating the activities of other cells within the respiratory tract during influenza virus infection are topics for future analysis.