Massive Mobilization of Dendritic Cells During Influenza A Virus Subtype H5N1 Infection of Nonhuman Primates

Abstract

Highly pathogenic avian influenza virus infection is characterized by a marked inflammatory response, but the impact of infection on dendritic cells (DCs) is unknown. We show that influenza A virus subtype H5N1 infection rapidly and profoundly impacts DCs in cynomolgus macaques, increasing the number of blood myeloid and plasmacytoid DCs by 16- and 60-fold, respectively. Infection was associated with recruitment, activation, and apoptosis of DCs in lung-draining lymph nodes; granulocyte and macrophage infiltration in lungs was also detected, together with expression of CXCL10. This degree of DC mobilization is unprecedented in viral infection and suggests a potential role for DCs in the pathogenesis of highly pathogenic avian influenza virus.

Highly pathogenic avian influenza (HPAI) viruses cause serious and often fatal disease in humans, but the mechanism for their high pathogenicity is poorly understood. The association of severe disease with markedly increased levels of serum proinflammatory cytokines and macrophage infiltration into lungs in patients infected with HPAI viruses suggest that innate immune dysregulation may contribute to disease pathogenesis [1]. However, relatively little is known about the effect of HPAI virus infection on dendritic cells (DCs), which play a key role in innate immune recognition of virus infection.

Infection of cynomolgus macaques (Macaca fascicularis) with HPAI viruses produces acute pneumonia with production of proinflammatory mediators consistent with infection in humans and offers a powerful model to study disease pathogenesis [2–5]. The 2 major DC subsets in humans, myeloid DCs and plasmacytoid DCs are similar both in phenotype and function in macaques [6]. In this study, we documented the impact of infection with two different H5N1 influenza virus strains on DCs in the blood and lymph nodes of cynomolgus macaques and reveal a profound response previously unseen in acute viral infection.

MATERIALS AND METHODS

Animals, Viruses, and Sample Collection

Eight adult male cynomolgus macaques housed at the Regional Biocontainment Laboratory of the University of Pittsburgh or at Bioqual (Rockville, MD) were used in this study. All procedures were performed using enhanced biosafety level 3 containment according to protocols approved by the University of Pittsburgh Institutional Animal Care and Use Committee and Institutional Biosafety Committee. Six macaques were inoculated with the clade 2.2 influenza A virus subtype H5N1 isolate A/Whooper Swan/Mongolia/244/2005 (WS/05) and were used as controls to test the efficacy of an experimental influenza A(H5N1) vaccine as previously described [7]. These macaques were euthanized on day 3 (animals 118 and 120), day 6 (animals 111 and 112), or day 7 (animals 217 and 234) after infection. Two additional macaques (animals 220 and 223) were inoculated with the clade 1 influenza A(H5N1) isolate A/Vietnam/1203/2004 (VN/04) as controls for a separate vaccination trial and were euthanized on day 7 after infection. Virus inoculations were done via ocular, nasal, and tracheal routes, using 1 × 10⁶ plaque-forming units of virus per location [7]. Virus titers were measured in tracheal washes, using a standard plaque assay. Tissues from 6 uninfected macaques were used as naive controls.

Flow Cytometric and Histopathologic Analyses

Antibody labeling of whole blood, peripheral blood leukocytes, and lymph node cell suspensions and enumeration of different cell subsets in blood by use of TruCOUNT tubes (BD Biosciences) were done as described elsewhere [8]. Data were collected using a BD LSRII or FACSAria flow cytometer and analyzed with BD FACSDiva or FlowJo (Tree Star) software. Lung tissue was inflated and fixed in 10% formalin at room temperature for 1 week and assigned a pneumonia score, ranging from 0 (<10% involvement) to 3 (>50% involvement), on the basis of findings from hematoxylin and eosin staining [7]. Immunohistochemistry analysis using antibody against myeloperoxidase (Abcam) and in situ hybridization using 35S-labeled riboprobes against influenza A virus and CXCL10 were done as described elsewhere [7, 9].

RESULTS

Lung Inflammation and CXCL10 Expression in Macaques Infected With Influenza A(H5N1)

To confirm that mucosal inoculation with the clade 2.2 isolate WS/05 and the clade 1 isolate VN/04 produced infection and characteristic lung inflammation in macaques, we measured virus titers in tracheal washes and examined lung tissue histologically. Both viruses established productive infection, with variable virus titers recovered from tracheal washes from days 1 to 5 after infection; virus was undetectable in 4 of 4 animals by day 7 (Figure 1A). Similarly, viral RNA was detected in lung tissue on days 3 and 6 after infection by in situ hybridization but was absent on day 7 (Figure 1B). Infection resulted in acute focal pneumonia associated with significant infiltration of neutrophils and macrophages, together with variable but, in some cases, marked expression of the inflammatory chemokine CXCL10, particularly on day 7 after VN/04 infection (animal 223; Figure 1B). These findings reflect sublethal pneumonia characterized by significant inflammation, consistent with previous reports [4, 5].

Figure 1.

Inflammation and CXCL10 expression in lungs of influenza A virus subtype H5N1–infected macaques. A, Virus titers in tracheal washes taken at intervals after mucosal inoculation with influenza A(H5N1) isolate WS/05 or VN/04. B, Lung sections were stained with hematoxylin and eosin (H&E), with an antibody to myeloperoxidase by using immunohistochemistry (IHC), or with riboprobes to influenza virus or CXCL10 by using in situ hybridization (ISH) in the indicated animals. Numbers in images of H&E-stained specimens represent pneumonia scores. Scale bar, 100 µm; original magnification, 200×. Abbreviation: PFU, plaque-forming units.

Profound Increases in Blood Myeloid DCs and Plasmacytoid DCs in Influenza A(H5N1) Infection

To determine the impact of influenza A(H5N1) infection on circulating DC subsets, we used established flow cytometry methods to identify myeloid DCs and plasmacytoid DCs as lineage (CD3/CD14/CD20)⁻ major histocompatibility complex (MHC) class II⁺ CD11c⁺ CD123⁻ cells and lineage⁻ MHC class II⁺ CD11c⁻ CD123⁺ cells, respectively, within the CD45⁺ gate of peripheral blood leukocytes. Absolute numbers of myeloid DCs and plasmacytoid DCs were then calculated by multiplying the proportion of each cell type within the CD45⁺ gate by the number of CD45⁺ cells in whole blood (Figure 2A and 2B). Blood samples were available from 4 animals for this analysis and were obtained on day 3 (for animals 118 and 120) or day 6 (for animals 111 and 112) after WS/05 infection. Infection resulted in marked increases both in the percentage of lineage⁻ MHC class II⁺ cells within the CD45⁺ population, and the percentage of myeloid DCs and plasmacytoid DCs within this lineage⁻ MHC class II⁺ fraction (Figure 2B), and as a consequence the absolute number of circulating DCs was strikingly increased. Myeloid DCs increased from a mean of 25 cells/µL before infection to >400 cells/µL after infection, representing a 16-fold increase, whereas plasmacytoid DCs increased from a mean of 5 cells/µL to >280 cells/µL, representing a 60-fold increase. Similar increases were noted both 3 and 6 days after infection (Figure 2C). There was no detectable impact of WS/05 infection on the number of circulating monocytes (CD14⁺CD3⁻CD8⁻CD20⁻), natural killer cells (CD3⁻CD8α⁺NKG2A⁺CD16⁺), or T cells (CD3⁺), suggesting that the changes were limited to the DC subsets (Figure 2D).

Figure 2.

Massive mobilization of blood myeloid dendritic cells (mDCs) and plasmacytoid DCs (pDCs) in macaques infected with WS/05 virus. A, Dot plots showing staining of whole blood and peripheral blood leukocytes (PBLs) with antibody to CD45. Fluorescent beads in the whole-blood sample are indicated. B, Representative dot plots showing gating strategy to identify DC subsets in blood from animal 111 before and after WS/05 virus infection. Numbers represent the percentage of cells in each gate. C, Changes in the absolute number of myeloid DCs and plasmacytoid DCs in blood from before infection to 3 days (for animals 118 and 120) or 6 days (for animals 111 and 112) after WS/05 virus infection. Because results from both days were similar, the data for all animals were combined. Fold-increases in the absolute number of myeloid DCs and plasmacytoid DCs (mean ± standard error of the mean) are also shown. D, Number of T cells, monocytes, and natural killer (NK) cells in blood before and after infection. P values were determined by the Mann–Whitney U test. Abbreviation: NS, not significant.

Marked Increase in DC Activation and Proliferation in Lung-Draining Lymph Nodes

We next analyzed suspensions of lung-draining (hilar) and peripheral lymph nodes from all 8 infected macaques and compared them to lymph nodes from 6 naive control macaques. In hilar lymph nodes, the proportion of myeloid DCs and plasmacytoid DCs within the lineage⁻ MHC class II⁺ was not altered as a consequence of infection. However, the frequency of myeloid DCs and plasmacytoid DCs that had recently divided, based on detection of Ki67 expression, was significantly increased, as was the fraction of each subset that expressed active caspase 3 and thus was apoptotic (Supplementary Figure 1). In addition, both DC subsets were activated, with significant increases in expression of the costimulatory molecules CD80 and CD86. In general, the highest expression of Ki67, active caspase 3, and costimulatory molecules on either DC subset was seen in animal 111, euthanized on day 6 after WS/05 infection, and in animals 220 and 223, euthanized on day 7 after VN/04 infection (Supplementary Figure 1). There was a positive correlation between myeloid DC expression of Ki67 in hilar lymph nodes and pneumonia score that approached statistical significance (P = .06; Spearman rank correlation coefficient, 0.604), suggesting a relationship between the magnitude of the DC response and the severity of disease. Similar trends were seen in myeloid DCs and plasmacytoid DCs in peripheral lymph nodes, although the changes were less marked than in hilar lymph nodes (Supplementary Figure 1).

DISCUSSION

Our findings show that within days of mucosal inoculation of cynomolgus macaques with influenza A(H5N1), there is a marked mobilization of myeloid DCs and plasmacytoid DCs into blood, with increases above preinfection numbers that averaged 16- and 60-fold, respectively. To our knowledge, this is the first report of a substantial increase in the number of circulating DCs in acute viral infection. There are no known published data on the impact of influenza A(H5N1) infection on blood DC subsets in either humans or nonhuman primates, although significant activation and apoptosis of DCs in the lung of infected macaques has been described [2]. Myeloid DCs and plasmacytoid DCs are recruited to mucosal tissues upon respiratory syncytial virus and seasonal influenza virus infection of children, but this is associated with a loss of cells from the circulation [10], as is 2009 pandemic influenza A(H1N1) infection of adults [11]. Our data therefore suggest that influenza A(H5N1) infection induces a more extreme DC response than other respiratory virus infections. Whether this response is specific to HPAI viruses and could be a potential contributing factor in their high pathogenicity or a function of influenza virus infection in general in this model is a key question that awaits additional comparative studies.

Our data showing granulocyte and macrophage infiltration and increased cellular expression of CXCL10 in the lung after influenza A(H5N1) infection, particularly in response to VN/04 virus, are consistent with previous findings [2, 5] and suggest a potential role for inflammatory cytokines and chemokines in mediating DC mobilization. In murine models, tumor necrosis factor α mediates rapid mobilization of myeloid DCs and plasmacytoid DCs into blood within hours of an inflammatory insult, and CXCL10 and CXCL11 drive the subsequent recruitment of DCs into inflamed tissues [12]. The relationship between DCs described in our study and inflammatory DCs, particularly tumor necrosis factor α/inducible NO synthase–producing DCs, which are known to be increased in murine models of influenza A(H5N1) infection [13], and whether select subsets of DCs are infected with HPAI viruses in vivo [14] remain to be determined.

Our finding lends itself to therapeutic intervention and has clinical relevance. Blockade of the DC response to influenza virus in vivo by using select antagonists of Toll-like receptors [15] would determine whether this response is beneficial or detrimental to the host. If DC blockade abrogated disease, this would support the use of antagonists to modulate disease severity during HPAI outbreaks.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.