The Migration of T cells in Response to Influenza Virus is Altered in Neonatal Mice

Abstract

Influenza virus is a significant cause of mortality and morbidity in children, however, little is known about the T cell response in infant lungs. Neonatal mice are highly vulnerable to influenza and only control very low doses of virus. We compared the T cell response to influenza virus infection between mice infected as adults or at 2 days old, and observed defective migration into the lungs of the neonatal mice. In the adult mice, the numbers of T cells in the lung interstitia peaked at 10 days post-infection, whereas neonatal T cell infiltration, activation and expression of TNFα was delayed until 2 weeks post-infection. While T cell numbers ultimately reached adult levels in the interstitia, they were not detected in the alveoli of neonatal lungs. Instead, the alveoli contained eosinophils and neutrophils. This altered infiltrate was consistent with reduced or delayed expression of type 1 cytokines in the neonatal lung and differential chemokine expression. In influenza-infected neonates, CXCL2, CCL5 and CCL3 were expressed at adult levels, while the chemokines CXCL1, CXCL9 and CCL2 remained at baseline levels and CCL11 was highly elevated. Intranasal administration of CCL2, IFNγ, or CXCL9 was unable to draw the neonatal T cells into the airways. Together, these data suggest that the T cell response to influenza virus is qualitatively different in neonatal mice and may contribute to an increased morbidity.

Introduction

Influenza virus infection is particularly dangerous during the first few months of life (1). Indeed, hospitalization rates for children under one year of age are as high as for the elderly (2). Rates of symptomatic infection are also highest among children (3), who shed virus for longer periods and at higher titers than adults (2). Consequently, children are thought to act as a human reservoir for influenza infection and may have an important role in virus dissemination throughout the community (4). Vaccination is the best means of preventing influenza infection (2), though the factors that are important for inducing effective immunity to influenza in children are not well established due to limited data. In fact, stochastic modeling suggests that vaccination of just 20% of children between 6–23 months of age would reduce the incidence of influenza in the US by 46% (4). While adult mice have been extensively analyzed, informative studies for analyzing T cell responses to influenza virus in neonatal animals are minimal. Other than studies examining the T cell response to DNA vaccines (5), the neonatal immune response to influenza virus has not been studied in great detail.

A Th1 biased response is required for clearance of influenza virus in adult mice (6); however, neonatal T cells preferentially mount Th2 responses to some pathogens including viruses (7). It has been shown that neonatal mice are capable of mounting robust Th1 responses with appropriate costimulation (8) including unmethylated CpG motifs (9), IL-12 and/or IFNγ (10, 11). This Th2 bias can be retained upon transfer in adult mice (12) suggesting it is at least in part T cell intrinsic. In this regard, a link has also been demonstrated between reduced IFNγ production and hypermethylation of CpG and non-CpG sites in human neonatal T cells (13).While IFNγ is not required for clearance of influenza virus, it promotes T cell migration to the lungs (14). Type 2 biased cells, which are not protective, are associated with eosinophilia and delayed viral clearance (6). Interestingly, the lungs of children infected with either respiratory syncytial virus (RSV) or influenza virus contain type 2 cytokines and eosinophils (7). However, a recent study in which neonatal mice were infected with influenza virus failed to find type 2 biased immunity (15).

The alveolar spaces of the lungs are a site of restricted T cell access. Although T cells continuously traffic through the lung interstitium in response to CCL5 (16), further signals are required for the cells to cross the epithelial cell barrier into the airways (17, 18). The criteria that are required for this migration are poorly understood. Our studies show that although neonatal T cells proliferate and migrate into the lungs, in contrast to adults, most T cells remain in the interstitium and fail to reach the airways. Instead, the alveolar infiltrate of neonates contained macrophages, eosinophils and neutrophils. These differences correspond with different cytokine and chemokine expression than adults.

Materials and Methods

Mice and viral stocks

Breeder C57BL/6J and BALB/cJ and mice were purchased at 5 – 7 weeks of age from Jackson Labs (Bar Harbor, ME). Female mice were co-housed together for two weeks to synchronize estrus for timed pregnancies. The A/Puerto Rico/8/34 (PR8) strain of influenza virus was grown in the allantoic fluid of 10-day-old embryonated specific-pathogen-free chicken eggs as previously described (19). Viral stocks were tested for common mouse pathogens and were shown to contain only influenza virus. Viral yield was quantified by titration in eggs to determine the 50% egg-infectious dose (EID₅₀). Unless noted otherwise (Fig. 1), sublethal doses of approximate LD₁₀ were used. Mice are considered neonates at less than 7 days of age and will be referred to as pups thereafter. At two days of age, neonates were infected intranasally (i.n.) under halothane or isoflurane anesthesia with an LD₁₀ of PR8 virus (0.2 EID₅₀ /g bodyweight) in a 10μl volume of PBS. Adult mice were simultaneously infected with an LD₁₀ of PR8 virus (2.0 EID₅₀ /g bodyweight) in a 50μl volume. Following infection, mice were monitored daily for weight loss.

FIGURE 1.

Neonatal mice are more susceptible to influenza A/PR/8/34 virus than adult mice. Mice at 8 weeks (A, C), or 2 days (B, D) of age were infected i.n. with the indicated dose of the PR8 strain of influenza virus based on body weight. Mice were weighed daily (C, D) and humanely killed when moribund (A, B). Viral burdens of mice infected with an LD₁₀ dose of PR8 were determined by plaque assay of lung homogenates on MDCK cells (E). The dashed line indicates the limit of detection in this assay. (A–D) Data represent the mean ± SD of 12 mice per group and are representative of 2 separate experiments. (E) Data represent the mean ± SD of at least 10 mice per group. * p<0.05 compared to pups at the same time point.

Determination of Lung Virus Titers

Lungs were sterilely isolated from infected mice and frozen at −80°C until analysis. Viral burdens were determined by plaque assay on Madin Darby Canine Kidney (MDCK) cells (ATCC, Manassas, VA). Cells were grown to confluency in 6-well plates in DMEM (ATCC) media supplemented with non-essential amino acids and 10% heat inactivated FCS (Atlanta Biologicals, Lawrenceville, GA). 10-fold dilutions of lung homogenate were incubated with the cells for 1 hour at 37°C. The cells were then washed and overlaid with DMEM media in 1% Bacto Agar with 1% trypsin (Sigma-Aldrich, St. Louis, MO). Three days later the cells were fixed with 20% acetic acid and the overlay removed. Plaques were visualized by staining with crystal violet and counted.

Cell isolation

Lungs were lavaged with 5 volumes of cold HBSS / 3mM EDTA to isolate alveolar cells. Blood was removed from the lungs by perfusion of HBSS into the right side of the heart. Lungs and tracheobronchial lymph nodes (TBLN) were excised and minced and the lungs were incubated with 50 U/ml DNase (Sigma-Aldrich), and 1 mg/ml collagenase A (Sigma-Aldrich) in RPMI 1640 containing 3% FCS. Digested lung tissue and lymph nodes were pushed through cell strainers to obtain a single cell suspension. Finally, RBCs were lysed by treatment with a hypotonic solution, and the remaining cells were washed and resuspended in HBSS.

Flow cytometry

For surface staining, 5 × 10⁵ to 10⁶ cells were stained with fluorochrome-conjugated Abs specific for murine CD4, CD8, CD44, CD62L and CD69 (eBioscience, San Diego, CA) in PBS phosphate-buffered saline containing 0.1% BSA and 0.02% NaN₃. Cells were then fixed in 5% formalin for 20 minutes at room temperature and resuspended in HBSS for multiparameter analysis using a FACSCalibur or LSR II cytofluorometer (BD Biosciences, Mountain View, CA).

For intracellular cytokine staining, cells from digested lung tissue were stimulated for 4 h with 50 ng/ml PMA and 1 μg/ml ionomycin. Brefeldin A (10 μg/ml) was added for the final 2 h of incubation to inhibit secretion of cytokines. Cells were surface-stained with anti-CD8 as described above, fixed in 5% formalin, and permeabilized with PBS/BSA/azide containing 0.5% saponin. Non-specific binding sites were blocked with anti-CD16/CD32 (eBioscience), and then cells were stained with anti-IFNγ (BD biosciences) and analyzed by flow cytometry.

Histology and microscopy

10 days post-infection (dpi), mice were euthanized and lungs were inflated and fixed with 10% formalin. Lung lobes were then embedded in paraffin, cut in 5-μm-thick sections, and hydrated through graded alcohol washes. Slides were stained with hematoxylin and eosin (H&E), dehydrated, and mounted.

For differential counts, an aliquot of 3 × 10⁴ bronchial alveolar lavage (BAL) cells were spun onto glass slides fixed in methanol, and stained with Diff-Quik (Dade International, Miami, FL). Images were obtained using a Spot digital camera attached to an Eclipse microscope (Nikon, Melville, NY).

Antigen Presentation Assay

To make bone marrow dendritic cells (BMDC), bone marrow progenitor cells from adult C57BL/6 mice were cultured with 90% complete media (RPMI/5%HIFBS/2-ME and antibiotics) and 10% supernatant from culture of GM-CSF-producing B16/F10-9GM cells for 14 days (20). BMDCs were infected with PR8 virus at 0.1 MOI multiplicity of infection (MOI) for one hour, or were left uninfected, then washed and cultured with lymphocytes isolated from day 7 post-infection lymph nodes at a 1:5 ratio in complete media. After 3 days, culture supernatants were harvested and assayed for IFNγ by ELISA (eBioscience).

Cytokine and Chemokine Analysis

Lungs were lavaged with 1ml HBSS / 3mM EDTA and BAL fluid (BALF) was stored at −80°C until analysis. Lungs from separate mice were homogenized in HBSS containing a 100-fold dilution of Sigma protease inhibitor cocktail (Sigma, St. Louis, MO) with the Glas-Col® Tissue Homogenizing System (Daigger, Vernon Hills, IL). ELISAs for TNFα, IFNγ, IL-4, IL-5, CCL5 and CXCL9 were performed on BALF and lung homogenate using ELISA kits according to manufacturers' instructions (eBioscience). CCL11, CXCL1, CXCL2, and CCL3 levels in BALF were determined using Milliplex cytokine kits (Millipore, Billerica, MA) and xMAP technology (Luminex, Austin, TX), according to the manufacturer's instructions. Cytokine levels in the lung homogenates were normalized using the RC DC Protein Assay (Bio-Rad, Hercules, CA).

For IL-4 mRNA analysis, lungs from infected mice were isolated into 2ml of RNAlater stabilization reagent (Qiagen, Valencia, CA), and stored at −80°C until analysis. Lungs were homogenized on ice, and total RNA was isolated using RNeasy kits (Qiagen). Reverse transcriptase reactions were performed using the High Capacity cDNA Reverse Transcriptase kit from Applied Biosystems (Foster city, CA). cDNA was purified using QIAquick PCR purification kits (Qiagen) and quantified with a ND-1000 (NanoDrop Technologies, Wilmington, DE). Real-time PCR was performed using the TaqMan® Gene Expression Assay Mm00445259_m1 (IL-4) and the TaqMan® Rodent GAPDH Control Reagents with ABI Universal PCR Master Mix on a 7500 Fast Real-Time PCR System (All from Applied Biosystems).

Administration of intranasal cytokines and chemokines

At 9 days post-infection with an LD₁₀ dose of PR8 virus, 16ng/g IFNγ, 2ng/g CXCL9, or 2ng/g, 400ng/g, or 2μg/g CCL2 (Peprotech, Rocky Hill, NJ) was administered in a 10μl volume of PBS to pups i.n. under isoflurane anesthesia. Control mice were given 10μl of PBS.

Statistics

Data were expressed as the mean ± SD of 5 mice per group, and each experiment was repeated at least twice. Statistics were performed using SigmaStat software (San Jose, CA) to run a two-way analysis of variance (ANOVA) followed by the Holm-Sidak post-hoc test for pairwise comparisons. If variance or normality tests failed, then Kruskal-Wallis one-way ANOVA on ranks were performed at each individual time point followed by a Dunn's pairwise post-hoc test. Differences were considered statistically significant with p<0.05.

Results

Neonatal mice are more susceptible to influenza PR8 virus than adult mice

To establish a dose of PR8 virus that was not lethal to neonatal mice, a dose response was performed as shown in Fig. 1. This study showed that neonatal C57BL/6 mice were more susceptible to lethal influenza infection than adults. Adult mice succumbed to 20 EID₅₀/g of body weight whereas 2 EID₅₀/g was lethal for neonatal mice (Fig. 1A and B). This corresponded to an absolute dose of 500 EID₅₀ in adults and 4 EID₅₀ in pups. For both adults and pups, 10-fold lower doses of virus (2 EID₅₀/g and 0.2 EID₅₀/g, respectively) were lethal to only 10% of the mice (LD₁₀, Fig. 1A and B). Morbidity was monitored by weight loss in adults and failure to gain weight in pups (Fig. 1C and D). Plaque assays performed on lungs from mice infected with an LD₁₀ dose of PR8 showed that the neonates were productively infected, and viral burdens were as high in neonatal mice as in adults within a few dpi (Fig. 1E). Additionally, neonates cleared virus more slowly than adult mice (Fig. 1E).

Neonatal T cells distribute differently within the lung

We first examined the T cell response in the lungs of PR8 infected C57BL/6 mice. The development of the T cell response in the lung interstitium of the neonatal mice was delayed by about 4 days as compared to adults, peaking at 14dpi (Fig. 2A and B). The CD8⁺ T cell response ultimately peaked at a lower magnitude than in adults (Fig. 2B). Beyond 14dpi, CD4⁺ and CD8⁺ T cell responses in the lung interstitium matched adult mice in both total numbers and percentage of lung infiltrate (Fig. 2A and B and data not shown).

FIGURE 2.

Neonatal T cells are excluded from lung airways. Adult or 2-day-old mice were infected i.n. with an LD₁₀ dose of the PR8 strain of influenza. BALF cells (C, D) and cells in digested lung tissue (A, B) were stained with anti-CD4 (A, C) and anti-CD8 (B, D) and analyzed by flow cytometry. Aliquots of BALF cells were spun onto slides and stained with DiffQuik. Differential counts were performed to determine the percentage of eosinophils (E) and neutrophils (F). Data represent the mean ± SD of 5 mice per group and are representative of 8 separate experiments. * p<0.05 compared to pups at the same time point.

Interestingly, T cell migration from the interstitium into the alveolar spaces was highly defective in the neonates. Although the numbers of CD4⁺ cells in the lung interstitium reached adult levels by 14dpi, only trace numbers entered the lung airways of the neonates (Fig. 2C). Similarly, only late in the inflammatory response, were small numbers of CD8⁺ T cells observed in the BAL (Fig. 2D).

While T cells failed to enter the alveolar spaces, cytospin analysis showed that large numbers of polymorphonuclear (PMN) leukocytes were able to enter the alveolar spaces (Fig. 3A and B). The neutrophil infiltration was slightly delayed in the neonates, but ultimately reached a similar magnitude as in adult mice (Fig. 2E). Interestingly, while eosinophils form a relatively minor fraction of the adult BALF, the numbers are highly elevated in neonates, peaking 14dpi when they accounted for 30–40 % of the total infiltrate (Fig. 2F). These data indicate that defective migration into the alveolar spaces does not extend to all leukocytes.

FIGURE 3.

Neonatal mice respond to influenza infection with interstitial inflammation. Adult (A, C, E) or 2-day-old (B, D, F) mice were infected i.n. with an LD₁₀ dose of the PR8 strain of influenza virus (A, B, C, D) or left uninfected (E, F). 10 days later, mice were euthanized and aliquots of BALF cells were spun onto slides and stained with DiffQuik (A, B) or fixed lung sections were stained with H&E (B, C, D, E). Representative fields are shown under 60× magnification.

The absence of T cells in the lung airways of the neonatal mice was suggestive of interstitial inflammation. To confirm this idea, sections of infected lung were stained with H&E at 10dpi (Fig. 3). Since much of the lung alveolarization and septation in mouse lungs occurs post-natally (21), uninfected pups have larger alveoli and broader septa than adult mice (Fig. 3E and F). Our studies showed that during influenza infection, inflamed areas of the neonatal interstitium were considerably thickened with inflammatory cells, while adult alveoli remain thin-walled and open (Fig. 3C and D). By flow cytometry, we determined that the cells in the lung interstitium of infected pups are made up of about 10% T cells, 30% monocytes, macrophages, and dendritic cells, 8% NK cells, and 10% PMN (data not shown).

Early T cell activation is normal in the lungs of neonatal mice infected with influenza virus

CD8⁺ T cells in the airways of influenza-infected adult mice are phenotypically different from those in the lung interstitium, and tend to be uniformly antigen-specific and highly activated (18). When we examined the activation status of the T cells in the lung interstitia of neonatal mice we found that the early activation markers CD69 and CD11a were expressed at similar levels on adult and pup CD4⁺ and CD8⁺ T cells throughout the course of infection (Fig. 4A, B, and data not shown). However, early in infection, a significantly lower proportion of pup interstitial CD4⁺ and CD8⁺ T cells were CD44^hiCD62L^lo antigen-experienced cells as compared to adults (Fig. 4C and D). Interestingly, by 14dpi the size of the lung interstitial infiltrate in the pups matched that of adults, and there were equivalent proportions of antigen-experienced T cells in the two groups of mice (Fig. 4C and D). Additionally, lymph node cells that were isolated 7dpi and cultured with PR8-infected BMDCs produced IFNγ to levels comparable with adult lymph node cells (Fig. 4E), indicating that they were predominantly virus-specific T cells.

FIGURE 4.

The proportion of activated T cells in the lungs of neonatal mice is significantly reduced compared to adult mice. Adult or 2-day-old mice were infected i.n. with an LD₁₀ dose of the PR8 strain of influenza. Cells from digested lung tissue were stained with anti-CD4, anti-CD8 and either anti-CD69 (A, B) or anti-CD44, and anti-CD62L (C, D) and were analyzed by flow cytometry. Data represent the mean ± SD of 5 mice per group and are representative of 3 to 8 separate experiments. * p<0.05 compared to pups at the same time point. (E) In a separate experiment, lymph node cells were isolated from adults or pups at day 7 post-infection and were cultured at a 1:5 ratio with BMDCs infected with 0.1 or 0 MOI of PR8 virus. After 3 days, culture supernatants were assayed for IFNγ production by ELISA.

Cytokine expression is altered in influenza-infected neonates

Since the neonates consistently had eosinophilic airway infiltrates we used PCR to assay for IL-4 by RNA expression. Both adult and neonatal T cells had the capacity to produce IFNγ upon re-stimulation ex vivo (Fig 5A), but IL-4 mRNA was detected at much higher levels in the neonatal lung than in adult lung (Fig 5B). Although intracellular cytokine staining showed that the neonatal T cells had the potential to make IFNγ upon ex vivo re-stimulation, we wanted to confirm whether IFNγ and IL-4 were produced in the lungs during infection. ELISA assay showed that upon infection, IFNγ was upregulated in lung homogenates of both pups and adults, however IFNγ was slightly delayed and reached lower levels in pups (Fig. 5C). Additionally, no IFNγ was detectable in the BALF from the neonates at any time point examined (Fig. 5E), corresponding with the lack of IFNγ-producing T cells in the airways. Concomitant with the peak in IFNγ, and a shift to a type 1 biased lung environment, the levels of IL-4 protein dropped in the lungs of adult mice (Fig. 5D). This drop was absent in neonatal mice, and instead IL-4 protein levels doubled (Fig. 5D). In contrast, IL-5 levels in the lungs of adults and neonates did not increase significantly above baseline levels at any of the time points examined (data not shown).

FIGURE 5.

The neonatal cytokine response is altered in influenza-infected neonates. Adult or 2-day-old mice were infected i.n. with an LD₁₀ dose of the PR8 strain of influenza. (A) Intracellular expression of IFNγ in lung cells was determined by flow cytometry after restimulation with PMA and ionomycin. Data is expressed as the percentage of CD8⁺ T cells that are IFNγ⁺. (B) IL-4 mRNA levels were determined in lung homogenate by RT-PCR and are expressed as fold increase relative to age-matched uninfected controls. Lung homogenate (C, D) and BALF (E, F) were analyzed for IL-4 (C) IFNγ (D, E), or TNFα (F) by ELISA. Data represent the mean ± SD of 5 mice per group and are representative of 2 separate experiments. * p<0.05 compared to pups at the same time point.

In adult mice, TNFα is produced rapidly after infection and activates endothelial and epithelial cells to up-regulate adhesion molecules and produce chemokines (22). While high levels of TNFα were detectable in BALF from adult mice 7dpi, levels did not peak in the pups until 21dpi (Fig. 5F). Notably, the increase in levels of TNFα in the BALF correlated with the appearance of CD4⁺ T cells in the lung interstitium of pups (Fig. 2A and 5F).

Chemokine expression is altered in influenza-infected neonates

We hypothesized that delayed production of TNFα and a lack of BALF IFNγ would lead to a delay in TNFα- and IFNγ-induced chemokines, such as CXCL9 and CXCL10, which may in part explain the defect in CD4⁺ T cell infiltration into the airways of influenza virus-infected neonatal mice.

CCL5 (RANTES) and CCL3 (MIP-1α) are highly upregulated during influenza infection and are important for T cell migration to the lungs (23–25). CCL3 is also chemotactic for eosinophils and is associated with pulmonary eosinophilia (26, 27). Levels of CCL5 protein did not differ in either BALF or lung homogenate from infected adults or pups (Fig. 6A). CCL3 was expressed in the neonatal lung, but was delayed and ultimately peaked at a lower level than in adult mice (Fig. 6B). CCL3 and CCL5 may be important in drawing T cells and eosinophils into the lung.

FIGURE 6.

Chemokines are differentially expressed in adult and neonatal influenza-infected mice. Adult or 2-day-old mice were infected with PR8 at LD₁₀ doses. Lung homogenate (A, E) and BALF (B, C, D, F, G) were analyzed for CCL5 (A), CCL3 (B), CXCL9 (C), CCL2 (D), CCL11 (E), CXCL1 (F), or CXCL2 (G) by ELISA or luminex multiplex assay. Data for uninfected adults is shown at day 0 and data for uninfected pups is shown at time points corresponding to when infected littermates were analyzed. Data for uninfected controls is not shown for CCL11 (E). Data represent the mean ± SD of 5 mice per group, and are representative of at least 2 separate experiments. * p<0.05 compared to pups.

We next examined the CXCR3 ligand CXCL9 (MIG) and the CCR2 ligand CCL2 (MCP-1), both of which are important for T cell migration and function (25, 28). The adult mice expressed high levels of both CXCL9 and CCL2, which peaked in BAL 7dpi. In contrast, CXCL9 and CCL2 were not detected in the neonatal lungs (Fig. 6C and D). This may be partly responsible for the defective migration of neonatal T cells into BALF, and may be linked to reduced IFNγ in the alveolar spaces.

In contrast to the T cells, neutrophils and eosinophils were able to enter the alveolar spaces of neonates (Fig. 3). We therefore examined the lungs for the neutrophil-attracting chemokines CXCL1 (KC) and CXCL2 (MIP-2) as well as the eosinophil chemokine CCL11 (eotaxin). Both CXCL1 and CXCL2 peaked in adult BALF prior to 10dpi (Fig. 6F and G). Interestingly, CXCL1 concentrations were markedly reduced in pup BALF, while CXCL2 was expressed at similar levels as in adult BALF (Fig. 6F and G). CXCL2 expression corresponded with the neutrophil influx seen in Fig. 2E. Consistent with the low level of eosinophils in the lungs of adult mice, CCL11 was not expressed in adult lung homogenate at significant levels (Fig. 6E). In contrast, CCL11 was expressed highly in the lungs of influenza virus infected neonates, consistent with the entry of eosinophils (Fig. 6E and 3F). This is also in agreement with the presence of CCL11 and eosinophils in the lungs of infants infected with influenza virus (7).

Overall, the neonatal lung expresses an array of chemokines distinct to those expressed in adults, consistent with differences in the lung infiltrate. Neonates express CCR5 and CCL3 that may draw T cells into the lungs, but lack CCL2 and CXCL9, which may be linked to the absence of T cells in the alveolar spaces. Further, pups express CXCL2 for neutrophil migration, and high levels of CCL11 (eotaxin) that likely is responsible for the elevation in eosinophils.

Intranasal CCL2, IFNγ, or CXCL9 treatment is insufficient to attract neonatal T cells into the alveolar spaces

CD8⁺ T cells that are biased toward type 2 cytokines show altered migration patterns in the lungs during influenza infection (29), while IFNγ^−/− mice have reduced numbers of CD8 T cells in the BALF (14). To determine if IFNγ played a role in migration of neonatal T cells to the lungs murine IFNγ (16ng/g) or PBS was administered 7, 8 and 9dpi. This resulted in concentrations of IFNγ in BALF that were comparable to levels found in influenza virus infected adult lungs at 7dpi (Figs. 7B and 5E). Flow cytometry and differential counts were performed on BALF cells 10dpi. This time period was selected because large numbers of T cells accumulated in the neonatal lung interstitium, but did not enter the alveolar spaces (Fig. 2). The cytokine treatment led to increased total cellularity in both BALF and lung homogenate (data not shown), but did not increase the proportion of lymphocytes in the BALF of neonatal mice (Fig 7A). The percent of CD4⁺ and CD8⁺ T cells remained in the range observed in the BALF of uninfected pups. However, there was a small increase in myeloid cells in the BALF and increased numbers of T cells and myeloid cells in the lung digests of IFNγ treated pups (data not shown). These data show that IFNγ is unable to correct the migration defect of neonatal T cells across the epithelium and into the alveolar space.

FIGURE 7.

Intranasal IFNγ, CXCL9 or CCL2 treatment is not sufficient to attract neonatal T cells into alveolar spaces. 2-day-old mice were infected with PR8 at LD₁₀ doses, and then were treated with PBS or 16ng/g IFNγ (A), 2ng/g CXCL9 (C) or 400ng/g CCL2 (E) at days 7, 8 and 9 post-infection (A) or 9 post-infection (B, C). At day 10 post-infection, BALF cells were stained with anti-CD4 and anti-CD8 for analysis by flow cytometry. (A, B, C) and BALF was stored for subsequent quantitation of IFNγ (B), CXCL9 (D), or CCL2 (F) by ELISA. Data for panels A,C, and E represent the mean ± SD of 5 mice per group, and are representative of 2 separate experiments. Data for panels B,D, and F show data from individual mice. Bars represent means of the groups. Data for panel F represents mock controls from 3 experiments, low dose (2ng/g) MCP-1, or high dose (400ng/g) MCP-1 treatment from single experiments. *p<0.05 compared to mock treated controls. Note that the outlier in the mock group in panel F was eliminated from statistical analysis based on the Dixon outlier test.

We also administered CCL2 or CXCL9 to the lungs, which were both absent from the BALF of influenza virus-infected neonates (Fig. 6G and H). Although CXCL9 increased the overall cellularity of the BALF of neonatal mice, the numbers of CD4⁺ or CD8⁺ T cells did not change (Fig. 7C and data not shown). Intratracheal CCL2 given to uninfected adult mice was reported to induce a monocytic infiltrate into the alveolar spaces and a ten-fold increase in CD4⁺ lymphocytes (30). Because we could not detect increased concentrations of CCL2 in the BALF of the treated pups 24 hrs after a dose of 2ng/g CCL2, we next tested a much higher dose (400ng/g). Though the higher dose resulted in increased levels of CCL2 in the BALF, it did not approach endogenous concentrations found in adult mice nor did it stimulate infiltration of T cells into the alveolar spaces (Fig. 7E,F and Fig. 6D). To try to arrive at a dose that would get closer to CCL2 levels found in adult mice, we inoculated 2μg/g which was similar to the single dose used by Maus et. al. to induce monocytic infiltration into the lungs of uninfected adult mice (31). This higher dose of CCL2 did not have a significant affect on T cell migration into the alveolar spaces, though in one experiment lung CCL2 levels were nearly that of what we see in adult lungs infected with influenza virus (data not shown).

Unlike CCL2, CXCL9 administration resulted in a significant increase in the proportion of PMNs in the BALF (Fig. 8A). Since eosinophils and neutrophils both have anti-viral activity (31, 32), we investigated whether PMNs have a role in controlling the viral burden in neonates. Pups were treated with CXCL9 or diluent at 7dpi, and viral loads were analyzed 3 days later. Although not statistically significant, there was a trend towards higher viral load in the CXCL9-treated group (Fig. 8B). This suggests that increased numbers of PMNs were not able to enhance viral clearance in influenza virus infected neonatal mice.

FIGURE 8.

Intranasal CXCL9 treatment induces infiltration of PMNs, which do not enhance viral clearance. 2-day-old mice were infected with PR8 at LD₁₀ doses, and then were treated with PBS or 2ng/g CXCL9 at day 7 post-infection. (A) 24-hours post-treatment, BALF cells were spun onto glass slides for differential counts. Data represent the mean ± SD of 5 mice per group, and is representative of 2 separate experiments. * p<0.05 compared to mock-treated. (B) In separate mice, at 3 days post-treatment, lungs were collected and snap-frozen. Viral burden was measured by plaque assays on MDCK cells. Means are indicated by horizontal bars.

Together, these data indicate that the cytokine response to influenza in neonatal mice is a mixture of type 1 and type-2 cytokines rather than the typical type-1 response of adult mice. This is associated with infiltration of eosinophils but not T cells into the BAL, and increased susceptibility to influenza virus infection. Moreover, treatment with IFNγ or IFNγ-induced chemokines did not alter the neonatal response to influenza virus.

Discussion

Although neonatal mice are highly susceptible to the lethal affects of influenza virus infection, they were able to clear an LD_10, albeit with some delay, compared to adults (Fig. 1). Importantly, while activation and entry of neonatal T cells into the lungs was delayed, pups were capable of generating robust interstitial T cell responses (Fig. 2). The delay in the T cell response corresponded with delayed and / or reduced expression of the pro-inflammatory cytokines TNFα and IFNγ (Fig. 5D and F). Although neutrophil migration was normal and eosinophil numbers were elevated very few T cells entered the alveolar spaces of the neonatal mice (Fig. 2). Our data suggest that the altered composition of the BALF infiltrate was linked to altered chemokine expression. While CCL5, CCL3 and CXCL2 were upregulated normally in the neonatal lung, the chemokines CXCL9 and CCL2 were markedly reduced (Fig. 6).

The data presented here are complicated by the fact that neonates are more susceptible to lethal influenza infection than adults, and only tolerate small doses of virus. Here adult and neonatal B6 mice were infected with a LD₁₀ dose, corresponding to a 2-fold difference per gram of bodyweight, and a large difference in total virus. Titration experiments that were performed in adult mice suggested that the kinetic differences in the T cell response were unlikely to be only dose related (33). Others have also found that adult mice, which are infected with a 100-fold range of influenza virus do not exhibit significant differences in kinetics of T cell infiltration into the lungs (33). Moreover, our own data shows a similar lung burden of virus in pups and adults through 7 dpi and even higher levels through 10 dpi without a concomitant infiltration of T cells into the alveoli. These data indicate that there is an intrinsic defect in migration of T cells into the alveoli of neonatal mice that is unrelated to the level of antigen in the lungs.

In earlier studies we showed delayed entry of T cells into the lungs of neonatal mice that were infected with Pneumocystis (34). In this model, the delay in the T cell response was also independent of the dose of the inoculum and corresponded with delayed TNFα expression as well as slow upregulation of VCAM-1 and ICAM-1 (35). These studies were consistent with the influenza model which showed delayed expression of TNFα in the BAL of neonatal mice (Fig. 5F). Others have shown that cord blood cells have a markedly reduced capacity to produce TNFa in response to TLR ligands (36, 37) which may be partly explained by reduced MyD88 expression on monocytes (37). Defects in TNFα production by adult blood monocytes have also been attributed to factors that are found in cord blood plasma (36). The importance of TNFα in T cell entry of the lungs is illustrated by the fact that neutralization of TNFα with a TNFα-specific antibody reduces cell recruitment to the lungs of adult mice in response to influenza (38). However, even though TNFα production was delayed in our model of neonatal influenza virus infection, the appearance of this cytokine after 7 dpi did not significantly influence the migration of T cells from the lung interstitium into the alveolar spaces leaving us to surmise that the lack of IFNγ may have been responsible for the failure of T cells to migrate across the epithelial barrier into the alveoli.

The type 1-biased cytokine, IFNγ, was reduced in the neonatal BALF (Fig. 5E). We found that the IFNγ-induced chemokine CXCL9 was also markedly reduced in neonatal BALF (Fig. 6C). A number of studies have demonstrated that T cells from both human and murine newborns are defective in IFNγ production (39–41), and that this may be due to defects in IL-12 and IL-18 expression (42). IFNγ signaling is not required for clearance of influenza, and IFNγ^−/− mice do not show significantly increased mortality (14, 43). However, IFNγ has an important role in ameliorating immunopathology, and IFNγ-deficient responses are associated with delayed T cell recruitment and increased PMN influx (44). Additionally, it appears that IFNγ receptor 1 deficient mice show defects in migration of CD8⁺ T cells into the alveolar spaces (14). This was interesting given our finding that neonatal T cells are able to reach adult cell densities in the lung interstitium, but do not enter the alveolar spaces (Fig. 2). To better examine the role of IFNγ in the distribution of T cells in the lung we administered exogenous IFNγ to mice infected as neonates. However, this treatment was unable to induce T cell recruitment to the alveolar spaces even though we measured adult IFNγ levels in the alveolar spaces of treated pups infected with influenza virus (Fig. 7A).

Although neonatal T cells can produce IFNγ during influenza virus infection, the lungs also contain IL-4 mRNA and increased IL-4 protein suggesting that the response is less strongly type 1 biased (Fig. 5B). The bias of cytokines that T cells produce is known to correspond with their migratory properties. For example, the chemokine receptors CCR2, CCR5 and CXCR3 are expressed on type 1 biased T cells, while type 2 biased T cells tend to express CCR3 and CCR4 (45). Additionally, type 2 biased T cells also show defects in migration, and cluster in the lung parenchyma at locations distinct from influenza virus infected epithelial cells (29). The interstitial inflammation observed in neonates infected with influenza virus may be linked to the less type 1 biased T cell response. In support of such an idea, interstitial pneumonia has been associated with type 2 biased cytokines in humans (46). Further, studies in adult mice have shown that type 2 biased T cells are ineffective against influenza virus infection, and are associated with eosinophilia, increased morbidity and delayed clearance (6). We observed these features in influenza infected neonatal mice (Figs. 1 to 3). The lack of a clear type 1 bias in neonates is consistent with data demonstrating that neonatal T cells have more of an intrinsic type 2 bias to a number of antigens (47–49), even upon adoptive transfer into adult hosts (12). Additionally, since the strength of TCR interactions influence Th1 differentiation (50), and neonates have a more restricted TCR repertoire (51–53), this may hinder full Th1 biasing.

In addition to T cell-intrinsic factors, the neonatal lung environment and / or APC may influence the cytokine profile of the T cells. For example, a lack of dendritic cell (DC) maturation is associated with failure to induce good Th1 biased responses (54, 55). It has also been shown that CD8α⁺CD4⁻ DCs that produce IL-12 are significantly reduced in number prior to 6 days of age (56). In this vein, the respiratory tract of rats lacks MHCII⁺ DCs until day 2–3 after birth, and mature DCs do not reach adult levels until after weaning (57). DC recruitment to the lung, and DC maturation are deficient in neonates as compared to adult rodents during infection with aerosolized Moraxilla catarrhalis in rats and Pneumocystis carinii in mice (34, 58). During influenza infection, lung DCs are critical for T cell activation and function, so DC deficiency could potentially impact T cell function during neonatal influenza virus infection. This could account for the delayed migration of T cells into the lungs of pups compared to adults (Fig. 2) though early activation marker expression and robust IFNγ production ex vivo (Fig. 4) argue that initial activation events may be normal in neonates in response to influenza virus.

A study by You et al. described infection of 7-day-old mice with influenza virus (15). In this model, they saw infiltration of only small numbers of neutrophils, and no eosinophils. Additionally, they saw little IFNγ or IL-4 production by neonatal T cells in the lung (15). These differences are likely due to the older age of the pups at infection and differences in experimental methods. We did not see peaks in eosinophil or T cell responses until after 10dpi (Fig. 2). You et al. did not examine time points between 10dpi and 109dpi, and generally did not compare pup to adult responses (15). The authors indicated that both adults and pups cleared virus by 7dpi, while at this time point both adults and pups were still infected in our study (Fig. 1), suggesting that they may have used lower infecting doses. The lack of a clear type 1 biased response in our model is consistent with the existence of type 2 cytokines and eosinophils in the lungs of children infected with influenza virus (7). Additionally, the interstitial pneumonia that we observed is consistent with lung specimens of children that have died from influenza virus infection (59). Our model may be more comparable with severe childhood infections, while the You et al. study may recapitulate sub-clinical infections.

It is known that central memory T cells are only found in the parenchyma of the lung and not within the airways. This has been suggested to indicate the existence of further selection criteria for airway entry, in which chemokines and chemokine receptors may be important (18). Recently, it was demonstrated that CCR5 is critical for the early recruitment of memory CD8⁺ T cells into the alveolar spaces, but is dispensable for interstitial recruitment (17). In contrast, CCR5 was not necessary for entry of T cells into alveolar spaces during acute infections. In vitro, migration across epithelial cells is dependent on CCL2 but is independent of CCL5 (28). Other possible candidates are the CXCR3 ligands, which are expressed at the epithelial cell surface (60). CXCR3^−/− mice have decreased T cell infiltration of the BAL (25), suggesting that CXCR3 ligands, such as CXCL9 may be important for alveolar T cell responses. A role for CCL2 and CXCL9 in alveolar trafficking is consistent with the lack of these chemokines and T cells in neonatal BALF (Fig. 2 and 6C and D). However, these chemokines appeared not to be the only defect in alveolar trafficking, since the alveolar administration of these chemokines was unable to induce T cell migration into the airways (Fig. 7). We previously reported that neonatal T cells failed to migrate across an endothelial cell line in vitro in response to either CCL2 or CCL5 (36) indicative of an intrinsic defect in migration of neonatal T cells to the alveoli. Notably, endothelial cells produce chemokines such as CCL2 and the differential migration of T cells to the neonatal lung interstitium but not to the alveolar space may be due to differential expression of chemokines by endothelial cells and not epithelial cells or alveolar macrophages. We are currently examining chemokine production by endothelial versus epithelial cells as well as chemokine receptor expression in these mice.

The differential chemokine expression in the lungs of influenza virus-infected neonates appeared to lead to a different assembly of leukocytes. The main chemokines we observed in the neonatal lungs were CCL5 and CCL3 (Fig 6A and B). These chemokines are also detected in nasal secretions of children infected with respiratory viruses including RSV, adenovirus, parainfluenza and influenza virus (61). Both of these chemokines are produced by macrophages and epithelial cells and are potent chemoattractants for eosinophils (28, 62), which we observed at high levels in the BALF of neonates (Fig 2F) along with eotaxin (CCL11, Fig. 6E). CCL3 and CCL5 are also able to activate eosinophils and induce degranulation and adhesion molecule expression (63). Neutrophil influx into the lungs of neonates may be a result of CXCL2 in the BALF (Fig. 6G), a known chemoattractant for neutrophils. We also have preliminary data that IL-17 is elevated in the alveoli of influenza infected neonates which may also contribute to neutrophil attraction to the lungs of neonates.

In addition to chemokines, adhesion molecules both on T cells and epithelial cells may be important selection criteria for migration of leukocytes into the alveolar spaces. We have previously reported that expression levels of both ICAM-1 and VCAM-1 on Pneumocystis-infected lungs is significantly reduced in neonates compared to adults and exogenous TNFα can upregulate these adhesion molecules (36). We have observed that ICAM-1 expression is also reduced in neonatal lungs infected with influenza virus despite the presence of TNFα in the BALF late during influenza virus infection (data not shown). Crossing the endothelial barrier requires expression of the integrins CD18 and CD11a (LFA-1) while for the epithelial barrier, VLA-4 may be more important (64). However, during Pneumocystis infection, the levels of VLA-4 and LFA-1 were comparable between adult and pup T cells (35). Despite this, Pneumocystis-specific neonatal T cells were less responsive to in vitro migration across an endothelial cell monolayer than adult cells, indicating that there may be some T cell-intrinsic defects in chemokine and/or integrin signaling. We are currently examining T cell integrins during influenza infection in neonates.

Overall, our data indicate that there are fundamental differences in the migration pattern of neonatal T cells into the lungs in response to influenza virus. Similar results have recently been reported in autopsy specimens from young children that died from influenza virus infection (59). As in human children (7), the lungs of neonatal mice contained elevated levels of eosinophils and neutrophils and were less strongly Th1 biased. We postulate that failure of T cells to appropriately bias to type 1 responses in infected infants may contribute to increased susceptibility and poor outcome.

Abbreviations used in this paper

(BALF)

bronchiolar lavage fluid

(BMDC)

bone marrow dendritic cells

(DC)

dendritic cells

(dpi)

days post-infection

(EID₅₀)

50% egg-infecting dose

(i.n.)

intranasal

(HIFBS)

heat-inactivated fetal bovine serum

(PR8)

influenza A/PR/8/34

(2-ME)

2-mercaptoethanol

(MOI)

Multiplicity of infection

(MDCK) cells

Madin Darby Canine Kidney

(RSV)

respiratory syncytial virus

(TBLN)

tracheobronchial lymph nodes