Innate immune responses to human rotavirus in the neonatal gnotobiotic piglet disease model

Abstract

Intestinal and systemic dendritic cell (DC) frequencies, serum and small intestinal content cytokines and uptake/binding of human rotavirus (HRV) virus-like particles (VLP) were studied in HRV acutely infected or mock-inoculated neonatal gnotobiotic piglets. Intestinal, mesenteric lymph node (MLN) and splenic plasmacytoid DCs (pDCs), conventional DCs (cDCs) and macrophages/monocytes were assessed by flow cytometry. In infected pigs, serum and small intestinal content interferon-α (IFN-α) were highest, interleukin-12 (IL-12) was lower and IL-10, tumour necrosis factor-α and IL-6 were minimal. Compared with mock-inoculated piglets, frequencies of total intestinal DCs were higher; splenic and MLN DC frequencies were lower. Most intestinal pDCs, but few cDCs, were IFN-α⁺ and intestinal macrophages/monocytes were negative for IFN-α. Serum IFN-α levels and IFN-α⁺ intestinal pDCs were highly correlated, suggesting IFN-α production in vivo by intestinal pDCs (r = 0·8; P < 0·01). The intestinal pDCs and cDCs, but not intestinal macrophages/monocytes, of HRV-infected piglets showed significantly lower VLP uptake/binding compared with mock-inoculated piglets, suggesting higher activation of pDCs and cDCs in infected piglets. Both intestinal pDCs and cDCs were activated (IFN-α⁺ and lower VLP binding) after HRV infection, suggesting their role in induction of HRV-specific immunity. Dose-effects of HRV on serum IFN-α and IFN-α⁺ DCs were studied by infecting piglets with 100-fold higher HRV dose. A high dose increased parameters associated with inflammation (diarrhoea, intestinal pathology) but serum IFN-α and IFN-α⁺ DCs were similar between both groups. The pDCs have both anti- and pro-inflammatory functions. Stimulation of the anti-inflammatory effects of pDCs after the high dose, without increasing their pro-inflammatory impacts, may be critical to reduce further immunopathology during HRV infection.

Introduction

Rotavirus (RV) is an enteric pathogen affecting nearly all children by the age of 5 years.^1,2 Complications caused by secondary untreated or refractory dehydration lead to high mortality rates in developing countries.^3–5 The neonatal gnotobiotic (Gn) pig is susceptible to heterotypic RV infection [e.g. human RV (HRV)], and unlike any adult animal (e.g. mice, rats, rabbits, monkeys) model, it is susceptible to HRV-induced diarrhoea.⁶ Furthermore, the gastrointestinal and immune systems of neonatal Gn piglets resemble those of human infants,⁶ so responses detected in neonatal piglets better mimic those after natural RV infection of infants. The piglets are colostrum-deprived, maintained in isolator units and are fed sterile infant formula, assuring a pathogen/microbe-free and antibody-free status, making them an ideal animal model to study antigen-induced innate immunity in vivo.

The types of dendritic cells (DCs) have been characterized in detail in mice and humans. Two main types are described: conventional DCs (cDCs) and plasmacytoid DCs (pDCs),⁷ but within each type, many subtypes exist with varied tissue distributions.^8,9 Previous studies of circulating pDCs in piglets¹⁰ characterize them as SWC3^low CD4⁺ interleukin-3 (IL-3) receptor⁺ and negative for lineage markers such as CD8, CD3, CD21 and CD14. Also, pDCs express low levels of major histocompatibility complex class II and CD80/86 and secrete high amounts of interferon-α (IFN-α) after exposure to various antigens such as viruses and CpG.¹¹ In piglets, cDCs express the myeloid marker SWC3, they lack CD4, are CD11b⁺ or CD11b ⁻ and secrete lower amounts of IFN-α after antigen stimulation compared with pDCs.¹² Monocytes are SWC3^high CD14⁺ CD16⁺ and lack CD4 and CD11b. Macrophages are SWC3^high CD14^low and also lack CD4.^10,13,14

The type I IFN, IFN-α is mainly secreted by pDCs, and in lower amounts by cDCs. It has been shown that type I IFNs have both pro-inflammatory and anti-inflammatory functions.^15–17 There is in vivo evidence of an anti-inflammatory role of type I IFNs during RV infection. It is known that neonatal mice infected with Rhesus RV (RRV) develop biliary duct inflammation and subsequent biliary atresia. Studies of type I IFN receptor knockout (KO) mice infected with RV demonstrated that mortality as a result of biliary atresia was greatly increased compared with that in infected wild-type (WT) mice.^18,19 However, RV clearance was similar in infected WT mice compared with type I or II KO mice²⁰ and signal transducer and activator of transcription 1 (STAT1) KO suckling mice,²¹ suggesting that type I IFNs do not prevent RV replication but regulate the inflammation induced in vivo. However, there is also evidence of type I IFN control of RV replication. In STAT1 KO adult mice,²¹ RV shedding was increased, although not prolonged, compared with WT mice. In double type I/II IFN signaling (IFN-αβ/γ receptor and STAT1) KO mice,²² heterologous RRV shedding was increased and development of lethal pancreatitis, hepatitis and biliary atresia was observed. Nevertheless replication of homologous murine RV (EC strain) in the KO mice did not differ from that of WT mice. Therefore in mice, the diverse effects of type I IFNs seem to depend on the infecting RV strain. It is also known that RV is capable of preventing type I IFN production by certain cell types. The RV non-structural protein 1 (NSP1) can degrade the IFN regulatory factor 3 (IRF-3) and prevent type I IFN production.²³ Mouse embryonic fibroblasts infected with heterologous RV (UK, NCDV, OSU) fail to degrade IRF-3, resulting in IFN-β secretion and reduced RV replication. On the other hand, homologous murine RV (ETD, EHP strains) and heterologous RRV degrade IRF-3 in mouse embryonic fibroblasts, diminishing IFN-β secretion and achieving effective RV replication.²⁴

There is in vitro evidence for pDCs and myeloid (or conventional) DCs inducing HRV-specific memory T helper type 1 (Th1) cellular immunity.^25,26 Studies of human sera showed that IFN-α is detected by 2 days after onset of RV diarrhoea.²⁷ Other studies of innate immunity to RV have demonstrated that RV-derived proteins were detected in intestinal and extra-intestinal murine macrophages.²⁸ Evidence for the importance of intestinal DCs in RV immunity derive from studies in CCR6 KO mice showing that the KO mice lack CD11b⁺ or CD11c⁺ DCs in the subepithelial dome of the Peyer's patches, leading to diminished humoral immunity to RV.²⁹

Rotavirus infects the small intestinal epithelium and also causes transient viraemia/antigenaemia in humans and different animal models including Gn piglets.^30–33 The source of early serum or intestinal cytokines after HRV infection is unknown. Also, whether intestinal or systemic DC activation occurs in vivo early after HRV infection remains to be determined. To examine the types of DCs induced after RV infection and their localization, we characterized the ex vivo DCs locally in the intestine and the mesenteric lymph nodes (MLN) and systemically in the spleen after HRV infection of neonatal Gn piglets. Piglets were inoculated with 10 focus-forming units (ffu) of HRV M strain (P1A[8]G3) and IFN-α, IL-12, IL-6, tumour necrosis factor-α (TNF-α) and IL-10 were measured in serum and small intestinal contents (SIC) by enzyme-linked immunosorbent assay (ELISA) for 7 days after inoculation. Intestinal, MLN and splenic porcine pDCs and cDCs, and macrophages/monocytes, were tested for uptake/binding of HRV virus particles bound to green fluorescent protein (GFP) and IFN-α production. We show that the main DC populations activated after HRV infection were intestinal pDCs followed by intestinal cDCs. To determine the association between IFN-α production and clinical outcome, a subset of piglets were infected with 100 times (1000 ffu) the original dose. A dose effect was observed related to earlier induction of diarrhoea and intestinal pathology, but the higher dose did not alter HRV titres in faeces or IFN-α levels in serum, suggesting complex interactions related to diarrhoea, intestinal pathology, IFN-α and the infecting HRV dose. To our knowledge this is the first study to characterize the ex-vivo-induced intestinal and systemic DC responses (and the dose effect) of an enteropathogenic viral infection in a neonatal animal disease model.

Materials and methods

Virus

The virus inoculum was the HRV M strain (P1A[8]G3). A pool of intestinal contents was obtained from the eleventh Gn pig passage and diluted in minimal essential medium (MEM; Invitrogen, Carlsbad, CA) to a final concentration of 10 ffu/ml or 1000 ffu and stored at − 70° until use. The 50% disease dose (DD₅₀) and infectious dose (ID₅₀) of the HRV M strain in gnotobiotic piglets are < 0·1 ffu.

Animal

The Gn piglets were derived by hysterectomy and maintained in isolator units as previously described.³⁴ At 3 days of age, rectal swabs were collected to assess the piglets’ microbial sterility. At 3–5 days of age, piglets were bled and inoculated. The piglets received 2% sodium bicarbonate orally to neutralize gastric acidity followed by 10 ffu or 1000 ffu of virulent M HRV in 3 ml MEM (Invitrogen). Controls were inoculated with MEM only. Piglets were bled on the day of inoculation and at post-inoculation day (PID) 2, 4 and 7; subsets of piglets were then euthanized on PID 2, 4 and 7. All procedures were conducted in accordance with protocols reviewed by the Ohio State University's Institutional Laboratory Animal Care and Use Committee (ILALUC).

Assessment of diarrhoea and RV shedding

After HRV inoculation, diarrhoea was assessed by scoring faecal consistency. Shedding of infectious virus was quantified by a cell culture immunofluorescence (CCIF) assay after inoculation until the day of euthanasia as previously described.³⁵ Briefly, faecal consistency was scored as follows: 0 = solid; 1 = pasty; 2 = semi-liquid; 3 = liquid, with scores of 2 or more considered diarrhoeic. Rectal swabs were collected every day, including the day of euthanasia and were resuspended in 8 ml MEM and tested by RV ELISA for assessment of antigen shedding and CCIF assay for detection of infectious virus shed, as previously described.⁶ Briefly, a sandwich ELISA was performed to detect RV antigen. Half of a 96-well plate was coated with RV-specific hyperimmune serum and the other half was coated with control serum. Then, duplicates of diluted rectal swab samples were added to the plate. Positive (RV-infected cell culture supernatant) and negative (negative rectal swab fluids) controls were added to each test. The cut-off value was calculated as the absorbance of the negative serum-coated wells plus three standard deviations. Absorbance values higher than the cut-off value were considered positive. For CCIF assay, 96-well culture plates were incubated with MA104 cells with fetal bovine serum (FBS) until a complete monolayer was formed (usually after 72 hr). The cells were washed with FBS-free medium and then incubated with 10-fold serial dilutions of the diluted rectal swab samples. After 18 hr, the cells were washed and stained with fluorescein isothiocyanate-conjugated antiserum. The final CCIF titre was calculated as the reciprocal of the last dilution of the sample that showed fluorescing cells viewed using a fluorescence microscope. Virus shedding was considered positive if the sample was positive by CCIF assay or ELISA.

Preparation of HRV virus-like particles

Recombinant baculoviruses expressing VP6 (Wa, G1), VP7 (CJN, G1), VP4 (Ku, P1A[8]) and VP2 were constructed as previously described.^35,36 Recombinant baculovirus expressing VP2 (first 92 amino acids deleted Δ92VP2, RF, bovine RV) with green fluorescent protein (GFP) was kindly provided by the late Dr Jean Cohen (L’Institut National de la Recherche Agronomique-INRA, Paris, France).³⁷ For the construction of fluorescent virus-like particles (VLPs; using the VP2-GFP construct) used in the uptake/binding assay, Trichoplusia ni cells (H5 cells) in Express Five serum-free media (Invitrogen) were infected with various combinations of recombinant baculoviruses to generate 2/4/6/7VLP-GFP at a total multiplicity of infection of 5.³⁷ At PID 7, infected cell lysates were collected and stored at 4°. Particles were purified by sucrose-CsCl purification, resuspended in TNC (Tris–HCl 10 mm, NaCl 140 mm, CaCl₂ 10 mm) buffer, pH 7·4 and tested by electron microscopy for particle integrity. Western blotting to confirm the composition of the VLPs and microbial testing using anaerobic and aerobic culture conditions at 37° and 27° to detect contaminating bacteria or fungi were performed.^35,37 A RV antigen ELISA and insect cell (H5) protein ELISA were conducted to determine the RV antigen and insect cell protein titres from each VLP preparation to assess the purity of the VLPs. To assure that most of the assessed protein was from the VLP and not of insect cell protein origin, only samples with a ratio of RV antigen : insect cell protein > 3000 were used for the VLP-DC uptake/binding studies. Protein concentrations were determined by Bradford protein assay (Bio-Rad, Hercules, CA) and endotoxin levels were determined by Limulus amoebocyte assay (Associates of Capecod Inc., Woods Hole, MA) as previously described.³⁵ The VLP preparations used were at least 80% or more intact 2/4/6/7VLP-GFP particles as determined by electron microscopy and were stored for < 3 months after purification.

Isolation of mononuclear cells

Spleen, MLN and ileum–jejunum tissue fragments were collected on the day the piglets were euthanized. Spleen, MLN and intestine were processed for isolation of mononuclear cells (MNCs) as previously described.³⁴ Isolation of intestinal MNCs included diffuse small intestinal Peyer's patches, lamina propria and intraepithelial MNCs. After isolation the cells were diluted in phenol-red-free Dulbecco's Eagle's modified MEM (Invitrogen) supplemented with 10% filtered porcine serum (Sigma Aldrich, St Louis, MO), 1%l-glutamine, sodium pyruvate, non-essential amino acids and antibiotic–antimycotic (Invitrogen; DC media) and kept at room temperature until testing in the VLP uptake/binding assay and by flow cytometric staining. Both assays were performed on the same day, immediately after the isolation of all tissue-derived cells.

Flow cytometry for detection of intracellular cytokine-positive pDCs

After isolation of the MNCs, approximately 2·4 × 10⁷ cells were added to 15 ml polypropylene conical tubes (Becton Dickinson Labware, Franklin Lakes, NJ), cells were diluted in cold-filtered (0·1-μm filter) phosphate-buffered saline (PBS) with 0·5% bovine serum albumin (BSA) and 0·02% sodium azide (Sigma Aldrich; wash buffer) and centrifuged at 400 g for 5 min. After the supernatant was discarded, 20 μl of a 1/100 dilution of uninfected Gn pig control serum was added and cells were incubated for 10 min at 4°. Cells were washed, centrifuged and stained. For staining of surface markers, cells were incubated for 10 min at 4° protected from light, washed with wash buffer and centrifuged at 400 g for 5 min at 4°. To define DC subpopulations that secrete IFN-α, cells were stained with monoclonal antibody (mAb) to porcine SWC3 [immunoglobulin G2b (IgG2b) isotype, clone 742215A; Becton Dickinson, San Jose, CA], which is a pan myeloid marker (including pDCs and cDCs) in piglets and a mAb to porcine CD4 (IgG2a isotype, clone PT90A; VMRD, Pullman, WA) followed by fluorescein isothiocyanate-conjugated anti-mouse IgG2b (Becton Dickinson) and purified phycoerythrin (PE) -conjugated goat anti-mouse IgG2a (Southern Biotech, Birmingham AL). Then, biotin-conjugated recombinant human cytotoxic T-lymphocyte-associated antigen-4 (hCTLA-4) (Ancell Corp, Bayport, MN) was added followed by conjugated peridinin-chlorophyll-protein (PerCP)-streptavidin (SA). The hCTLA-4 binds to porcine CD80/86,¹¹ which is up-regulated on activated DCs. After staining of the cell surface markers, Cytofix-Cytoperm (Becton Dickinson) was added for fixation and permeabilization of the cells. For intracellular staining, the cells were washed with 1 ml of perm-wash (Becton Dickinson), the incubation time was 10 min at room temperature protected from light and centrifugation was performed at 500 g for 5 min. After permeabilization, cells were washed and 50 μl uninfected Gn pig serum diluted 1 : 100 in perm-wash was added for 5 min at room temperature to block non-specific antibody binding. Then cells were washed and stained with mAbs to IFN-α (IgG1 isotype, clone 27105-1; R&D Systems, Minneapolis, MN). After incubation, cells were washed twice and allophycocyanin-conjugated anti-mouse IgG1 (Becton Dickinson) was added. All cells were washed twice and diluted in 200 μl perm-wash. Control tubes were stained as follows: one set of controls included all the antibodies to cell surface markers except the cytokine antibodies and only the allophycocyanin-conjugated anti-mouse IgG1 was added. Another set of controls included all the isotype controls for the cell surface mAbs, the conjugated secondary antibody or SA-PerCP only. Acquisition of 100 000 events was performed by a FACScalibur flow cytometer (Becton Dickinson). Analyses were performed using Flowjo software (Treestar Inc., Ashland, OR).

DC uptake/binding assay

Immediately after isolation, cells were placed in DC media (see Isolation of MNC) and kept at room temperature until the assay was performed on the day of isolation of the cells. Approximately 10⁶ cells were distributed to 5 ml sterile polypropylene tubes and resuspended in a maximum volume of 50 μl DC medium. Then, 2·5 μg 2/4/6/7VLP-GFP was added and cells were incubated at 37° for 20 min. After incubation with VLPs, cells were washed with cold-wash media (PBS/0·5% BSA/0·02% sodium azide) and centrifuged at 400 g for 5 min at 4°. The cells were stained with PE-conjugated mAb to porcine CD4 (isotype IgG2b, clone 74124; Becton Dickinson), the biotin-conjugated mAb to porcine SWC3 (clone 742215; Southern Biotechnologies) and mAb to allophycocyanin-conjugated human CD21 (clone B-ly4; Becton Dickinson), which cross-reacts with porcine CD21 (C3d complement receptor and a porcine B-cell marker), were added followed by SA-PerCP. Incubations for each step were at 4° for 10 min protected from the light. The cells were fixed and resuspended in 300 μl of 1% paraformaldehyde and stored at 4° until cell acquisition by flow cytometry. Acquisition of 100 000 events was performed by a FACScalibur flow cytometer (Becton Dickinson) and analyses were performed using Flowjo software (Treestar Inc).

ELISA for cytokine quantification

At the time of euthanasia SIC and blood samples without anticoagulant were taken for serum collection and stored at − 20° until testing. Cytokines in serum and SIC were detected as previously described.³⁸ Briefly, Nunc Maxisorp 96-well microtitre plates were coated with anti-IL-12, anti-TNF-α, anti-IL-6, anti-IL-10 or anti-IFN-α (R&D Systems) overnight at room temperature. The plates were blocked with PBS/0·1% Tween/0·5% BSA for 2 hr at room temperature. Diluted samples (1 : 1 in PBS/1% BSA) and cytokine standard dilutions were added to the plates at a total volume of 100 μl. Plates were incubated at room temperature for 2 hr and then washed five times with PBS/0·1% Tween. Biotinylated mAbs to IL-12 ,TNF-α, IL-6, IL-10 and IFN-α (R&D Systems) were diluted and added for a 2-hr incubation at room temperature. Plates were washed and horseradish peroxidase-conjugated SA (0·1 μg/ml) was added, followed by incubation of the plates for 45 min at room temperature. Tetramethylbenzidine (TMB) with H₂O₂ peroxidase system (KPL Inc., Gaithersburg, MA) was added for development of the plates. Standard curves for each cytokine were constructed using a computer-generated four-parameter curve-fit with the dilutions of the recombinant cytokines used as positive controls. The detection limits of the ELISAs were 78 pg/ml for IFN-α, 15 pg/ml for TNF-α and IL-6 and 7·8 pg/ml for IL-12 and IL-10.

Histopathological evaluation of intestinal sections

For evaluation of intestinal lesions after HRV infection, piglets infected with HRV at 10 ffu or 1000 ffu and control piglets were killed at PID 2, 4 and 7. Specimens from the middle and distal jejunum and ileum were collected. The intestinal samples were rapidly fixed in 10% neutral buffered formalin overnight at room temperature. The tissues were dehydrated in an ethanol gradient and embedded in paraffin for histopathological examination after haematoxylin & eosin staining. A blinded examination was performed. Villous height and crypt depth (VH : CD) ratios were measured as previously described,³⁹ and intestinal tissues were scored according to the VH : CD ratios as follows: 0 = normal intestinal villi (VH : CD ratio 6–7); 1 = mild intestinal villous atrophy (VH : CD ratio 5–6); 2 = moderate intestinal villous atrophy (VH : CD ratio 4–5); 3 = severe intestinal villous atrophy (VH : CD ratio 2–4); and 4 = very severe intestinal villous atrophy (VH : CD ratio 1–2).

Purification of intestinal SWC3⁺ cells

Intestinal MNCs from HRV-infected and control piglets were collected³⁴ at PID 2, 4 and 7 and frozen in liquid nitrogen rapidly after isolation. Cells were thawed using 20 ml cold degassed and filtered 1 × PBS/0·5% BSA/2 mm ethylenediamine tetraacetic acid pH 7·4 buffer and centrifuged 300 g for 5 min. The cell pellet was resuspended in 1 ml of buffer and filtered to eliminate contaminating mucus and then the cells were counted using an automated counter. The supernatant was discarded and mouse anti-porcine SWC3-PE (Southern Biotech) was added. After incubation for 10 min at 4°, the cells were washed with 5 ml of cold buffer and centrifuged at 300 g for 5 min. Anti-PE beads (Miltenyi Biotech Inc., Auburn, CA) were added, incubated for 10 min at 4° and washed/centrifuged as described earlier. Cells were resuspended in 1 ml of cold buffer and the cell suspension was applied to LS columns (Miltenyi Biotec Inc.,) for positive selection according to the manufacturer's instructions. The columns were washed four times for the collection of the negative fraction. The SWC3⁺ cells were collected in 1 ml of cold buffer. The SWC3⁺ and SWC3⁻ cells were centrifuged, resuspended in 200 μl of cold buffer and counted using an automated cell counter. Approximately 5 × 10⁴ SWC3⁺ and SWC3⁻ cells were analysed by flow cytometry for purity assessment. Mean purity of the cells was 83·5% for SWC3⁺ cells and 88·5% for SWC3⁻ cells. The remaining cells (SWC3⁺ and SWC3⁻ fractions) were frozen at − 70° until RNA extraction to perform a quantitative reverse transcription–polymerase chain reaction (qRT-PCR) to detect evidence of HRV replication as described below.

qRT-PCR for detection of HRV NSP3

Cell extracts of known numbers of intestinal SWC3⁺ and SWC3⁻ cells were used for quantification of HRV RNA using qRT-PCR. The total RNA was extracted from SWC3⁺ and SWC3⁻ cell suspensions using TRIZOL LS reagent (Gibco, Life Tech, Grand Island, NY) according to the manufacturer's instructions. The PCR primers (HRV NVP3-F accatctacacatgaccctc, nucleotides 963–982, and HRV NVP3-R ggtcacataacgcccc, nucleotides 1034–1049) and the TaqMan probe (atgagcacaatagttaaaagctaacactgtcaa, nucleotides 984–1016)^40,41 recognized a highly conserved region of the HRV NSP3 and the expected size of amplicon was 87 base pairs. The fluorogenic probe was labelled with a FAM reporter at the 5′-end and a BHQ1 quencher at the 3′-end. A single-step real-time RT-PCR was performed using a Qiagen OneStep RT-PCR Kit (Qiagen Inc., Valencia, CA) and a Smart Cycler II System (Cepheid, Sunnyvale, CA).

The RT-PCR was performed in a 25-μl volume containing 5 μl 5 × Qiagen OneStep RT-PCR Buffer, 400 μm of each dNTP, 10 U RNase inhibitor (Promega, Madison, WI), 10 μm of each primer, 0·2 μm of probe and 2 μl RNA. RNA was mixed with 0·5 μl dimethylsulphoxide, incubated for 10 min at 97° and then snap-chilled on ice.

The RT was performed at 50° for 30 min, followed by 15 min incubation at 95° to activate HotStartTaq DNA polymerase and 45 three-step thermal cycles of 94° for 20 seconds, 60° for 30 seconds and 72° for 20 seconds. Two negative controls and a standard (RNA from HRV M strain, purified by CsCl density gradient) at 10⁰, 10⁻¹, 10⁻², 10⁻³ dilutions (giving target RNA concentrations in the range of 30 ng/μl to 30 pg/μl) were used for each run to generate a standard curve for HRV RNA quantification.

Statistical analyses

Statistical analyses were performed using the SAS program (SAS Institute Inc., Cary, NC). Differences in diarrhoea scores and shedding titres among groups were analysed by analysis of variance followed by Duncan's test. The percentage of piglets that developed diarrhoea and shedding at each PID was analysed by Fisher's exact test. Cytokines in serum and SIC, frequencies of IFN-α⁺ DCs and numbers of 2/4/6/7VLP-GFP⁺ DCs were analysed by Kruskall–Wallis rank-sum test. Spearman's correlation coefficient was used to correlate frequencies of cytokine-producing pDCs with cytokines in serum. A P-value of ≤ 0·05 was considered significant.

Results

The main cytokine detected in serum was IFN-α

At PID 2, 4 and 7 post-HRV infection, the innate IFN-α, TNF-α, IL-6, the Th1 inducer IL-12 and the Th2 inducer IL-10 were measured in serum and SIC (Fig. 1). The most abundant cytokine in serum was IFN-α, which peaked at PID 2, with levels significantly higher at PID 2 and 4 compared with controls. In the SIC, the IFN-α was lower and differed significantly from controls only at PID 4. Lower amounts of IL-12 were also observed in serum at PID 2 and 4, but they were significantly higher than in controls. Intestinal contents had lower amounts of IL-12, which also differed from controls at PID 2. Minimal TNF-α, IL-6 and IL-10 were observed in serum and SIC and concentrations of these cytokines did not differ significantly from controls. We concluded that after M strain HRV infection, the main cytokine secreted in serum and SIC is IFN-α.

Figure 1.

Serum and small intestinal content (SIC) cytokines after human rotavirus (HRV) infection. Serum and SIC cytokines were measured by enzyme-linked immunosorbent assay. Graphs to the left represent serum cytokines and graphs to the right represent SIC cytokines at post-infection days (PID) 2, 4 and 7 from top to bottom: interferon-α (IFN-α), interleukin-12 (IL-12), tumour necrosis factor-α (TNF-α), IL-6 and IL-10. Open circles represent the cytokine amount in serum or SIC in HRV-infected piglets and black squares represent cytokine levels in serum or SIC in mock-inoculated piglets and at day 0. The number of piglets used per PID is shown at the bottom of each graph. The error bars denote the standard error of the mean (SEM) and time-points with a bar and an asterisk differ significantly from the controls at each PID noted (Kruskall–Wallis rank sum test P ≤ 0.05).

After HRV infection total intestinal DC frequencies were dominant over splenic and MLN DC frequencies

Total frequencies of intestinal, MLN and splenic pDCs, cDCs and macrophages/monocytes were quantified (Figs 2 and 3). Frequencies of intestinal DC frequencies (pDCs, cDCs and macrophages/monocytes) were higher than those for splenic and MLN DCs and significantly higher at PID 2 and PID 4 compared with controls. In general, total splenic and MLN DC frequencies were consistently lower than intestinal DC frequencies and more similar to controls (significantly higher than controls only for splenic pDCs at PID 2 and 4, splenic macrophages/monocytes at PID 2 and MLN pDCs and cDCs at PID 2, Fig. 3). In conclusion, after acute HRV infection, DC frequencies were higher in the gut and lower in the spleen and MLN.

Figure 2.

Total intestinal frequencies of plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs) and macrophages/monocytes after human rotavirus (HRV) infection. (a) HRV-inoculated piglets and (b) medium-inoculated piglets, representative dot plots of SWC3 versus CD4 at post-infection day (PID) 2, 4 and 7 and isotype control plots are shown. Gates and percentages of SWC3^low CD4⁺ DCs (pDCs), SWC3^high CD4⁻ DCs (macrophages/monocytes) and SWC3^low CD4⁻ DCs (cDCs) are shown. Mean total frequencies of intestinal pDCs (c), macrophages/monocytes (d) and cDCs (e) in piglets that received HRV (○) or media alone (▪) are summarized. For each experiment 100 000 events were acquired by the flow cytometer. The number of piglets studied at each PID is shown at the bottom of each graph. The error bars denote the standard error of the mean (SEM) and time-points with an asterisk differ significantly from the controls at each PID (Kruskall–Wallis rank sum test P ≤ 0.05).

Figure 3.

Total frequencies of splenic and mesenteric lymph node (MLN) plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs) and macrophage/monocytes after human rotavirus (HRV) infection. (a) spleen and (e) MLN, representative dot plots of SWC3 versus CD4 at post-infection day (PID) 2, 4 and 7 after HRV inoculation (left graph) or inoculation with medium alone (right graph) are shown. Gates and percentages of SWC3^low CD4⁺ DCs (pDCs), SWC3^high CD4⁻ DCs (macrophage/monocytes) and SWC3^low CD4⁻ DCs (cDCs) are shown. Mean total frequencies of splenic pDCs (b), macrophages/monocytes (c) and cDCs (d) or MLN pDCs (f), macrophages/monocytes (g) and cDCs (h) in piglets that received HRV (○) or media alone (▪) are summarized. For each experiment 100 000 events were acquired by the flow cytometer. The number of piglets studied at each PID is shown at the bottom of each graph. The error bars denote the standard error of the mean (SEM) and time-points with an asterisk differ significantly from the controls at each PID (Kruskall-Wallis rank sum test P ≤ 0.05).

The main IFN-α⁺ DC population detected after HRV infection was intestinal pDCs followed by intestinal cDCs

To characterize the DC populations that produced the serum or SIC IFN-α detected after HRV infection, intracellular IFN-α staining was performed (Fig. 4). We observed that mainly intestinal pDCs followed by intestinal cDCs were IFN-α⁺. Intestinal macrophages/monocytes (Fig. 4), splenic and MLN DCs were IFN-α⁻ (data not shown). A high and significant correlation coefficient was observed between serum IFN-α and IFN-α⁺ pDCs (r = 0·8, P < 0·0001, Fig. 5), suggesting intestinal pDCs as the in vivo source of IFN-α. A lower but significant correlation was recorded between serum IFN-α and IFN-α⁺ cDCs (r = 0·6, P = < 0·001, data not shown). We also detected correlations with serum and SIC IL-12 and IL-12⁺ cDCs (r = 0·42 P < 0·01 and r = 0·36 P < 0·01, respectively and data not shown). Minimal up-regulation of CD80/86 on the surface of DCs was observed (data not shown). Only after incubation of peripheral blood mononuclear cells with live attenuated HRV was a significant up-regulation of CD80/86 on SWC3⁺ cells observed (data not shown). In conclusion, intestinal pDCs and cDCs and not systemic pDCs, cDCs or intestinal macrophages/monocytes were probably the main source of serum IFN-α after HRV infection.

Figure 4.

Intestinal frequencies of interferon-α (IFN-α) positive plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs) and macrophages/monocytes after human rotavirus (HRV) infection. (a) HRV inoculated and (b) medium alone inoculated representative dot plots of SWC3 versus CD4 at post-infection days (PID) 2, 4 and 7 are shown. Gates on SWC3^low CD4⁺ DCs (pDCs), SWC3^high CD4⁻ DCs (macrophage/monocytes) and SWC3^low CD4⁻ DCs (cDCs) and plots on each gate show IFN-α⁺ DCs. Mean frequencies of IFN-α⁺ pDCs (c), IFN-α⁺ cDCs (d) and IFN-α⁺ macrophages/monocytes (e) in piglets that received HRV (○) or medium alone (▪) are summarized. For each experiment 100 000 events were acquired by the flow cytometer. The number of piglets studied at each PID is shown at the bottom of each graph. The error bars denote the standard error of the mean (SEM) and bars with an asterisk differ significantly from the controls at each PID (Kruskall–Wallis rank sum test P ≤ 0.05).

Figure 5.

Correlation coefficient between interferon-α (IFN-α) positive plasmacytoid dendritic cells (pDCs) and serum IFN-α. A Spearman's correlation coefficient was calculated between IFN-α⁺ pDCs and serum IFN-α. The R, P and number of piglets included in the analysis are shown in the graph.

Indicative of an activated state, intestinal pDCs and cDCs, but not intestinal macrophages/monocytes, had lower 2/4/6/7VLP-GFP uptake/binding compared with controls

Immature DCs capture antigens by macropinocytosis or mannose-receptor-mediated endocytosis⁴² and soon after activation, DCs rapidly reduce antigen uptake.¹⁰ Therefore, lower antigen uptake after DC activation is a measure of their functional maturity.⁴³ To determine the DC functional maturity after HRV infection, we studied the numbers of freshly isolated DCs diluted in DC medium and their uptake/binding of 2/4/6/7VLP-GFP (Fig. 6a–e). A window on CD21^– cells was selected to analyse the numbers of 2/4/6/7VLP-GFP⁺ cells to avoid binding of naive B cells to the VLPs.^44,45 By analyzing the number of DCs (Fig. 6a–e) or the frequencies of DCs (data not shown) both intestinal pDCs and cDCs showed significantly lower uptake/binding of 2/4/6/7VLP-GFP at PID 4 compared with controls, suggesting activation of these DC populations. Macrophages/monocytes did not differ from controls at any time-point. In conclusion, only intestinal pDCs and cDCs and not intestinal macrophages/monocytes showed signs of activation (IFN-α production and decreased 2/4/6/7VLP-GFP uptake/binding).

Figure 6.

Intestinal numbers of 2/4/6/7VLP-GFP⁺ plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs) and macrophages/monocytes after human rotavirus (HRV) infection. (a) HRV-inoculated and (b) medium-alone-inoculated representative dot plots of SWC3 versus CD4 at post-infection day (PID) 2, 4 and 7 are shown. Gates on SWC3^low CD4⁺ DCs (pDCs), SWC3^high CD4⁻ DCs (macrophages/monocytes) and SWC3^low CD4⁻ DCs (cDCs) and plots on each gate show 2/4/6/7VLP-GFP⁺ DCs. Numbers of 2/4/6/7VLP-GFP⁺ pDCs (c), 2/4/6/7VLP-GFP⁺ cDCs (d) and 2/4/6/7VLP-GFP⁺ macrophages/monocytes (e) in piglets that received HRV (○) or medium alone (▪) are summarized. For each experiment 100 000 events were acquired by the flow cytometer. The number of piglets studied at each PID is shown at the bottom of each graph. The error bars denote the standard error of the mean (SEM) and bars with an asterisk differ significantly from the controls at each PID (Kruskall–Wallis rank sum test P ≤ 0.05).

HRV-dose effect on shedding, diarrhoea, intestinal pathology and serum IFN-α

Faecal HRV shedding, diarrhoea scores and intestinal pathology were measured after infection with 10 ffu of HRV (Fig. 7a–d). To assure the development of higher rates of diarrhoea and to determine the dose effect of HRV on the IFN-α levels in serum, we infected a set of animals with 100 times the original HRV infection dose (1000 ffu). Diarrhoea scores were higher for the high-dose group at PID 1 and 2 (Fig. 7a). Jejunal intestinal pathology was evident in 100% of the piglets that received a high dose at PID 2 and 4 whereas at PID 2 and PID 4, 25% and 40% of the 10-ffu dose piglets, respectively, developed jejunal intestinal pathology (Fig. 7b). Despite the 100-fold difference in HRV dose between the two groups, no significant difference in serum IFN-α was observed at any PID (Fig. 7c). Also, virus shedding was statistically similar between both groups at most PIDs except at PID3 (Fig. 7d). Like the serum IFN-α, the frequencies of intestinal IFN-α⁺ pDCs did not differ significantly between doses at any PID (10 ffu versus 1000 ffu: PID 2–5·1% ± 0·9 versus 2·7% ± 0·9; PID 4–0·45% ± 0·2 versus 0·57% ± 0·2; PID 7–0·14% ± 0·08 versus 0% ± 0, respectively). Total frequencies of DCs in the gut, spleen or MLN did not differ significantly between doses (10 ffu versus 1000 ffu: PID 2–1·51% ± 0·4 versus 0·93% ± 0·1; PID 4–5·42% ± 2·2 versus 4·73% ± 1·17; PID 7–3·31% ± 2·2 versus 2·62% ± 1·0, respectively).

Figure 7.

Diarrhoea scores, percentage of piglets with jejunal pathology, serum interferon-α (IFN-α) and infectious virus shed after 10 focus-forming units (ffu) or 1000 ffu of human rotavirus (HRV). Diarrhoea scores (a), percentage of piglets with jejunal pathology (b), serum IFN-α (pg/ml, c) and infectious virus shedding (ffu/ml, d) are shown after 10 ffu (○) and 1000 ffu (▪) from post-infection day (PID) 0–6 (x-axis). The number of piglets studied and the percentage of piglets that developed diarrhoea, shedding and jejunal pathology at each time-point is shown in each graph. Data-points with an asterisk differ significantly between doses at each time-point. The error bars denote the standard error of the mean (SEM) and values with an asterisk denote significant differences between groups at each PID. The HRV faecal shedding and diarrhoea scores were analysed using analysis of variance followed by Duncan's rank sum test, the percentage of pigs with diarrhoea and RV shedding was analysed using Fisher's exact test. Serum IFN-α was analysed using Kruskall–Wallis rank sum test. A P < 0.05 was considered the limit of statistical significance.

To investigate the lack of a dose effect between 1000 ffu versus 10 ffu of HRV and serum IFN-α, we assessed ex vivo HRV-infected SWC3⁺ DCs to determine if a higher HRV dose induced increased infection of DCs, affecting their IFN-α production. The amount of NSP3 RNA, reflecting replicating HRV in SWC3⁺ DCs, at PID 2 (peak of serum IFN-α) was measured. No difference of NSP3 transcripts in SWC3⁺ DCs was observed after 1000 ffu or 10 ffu (0·93 ± 0·48 versus 1·11 ± 0·7, respectively). A higher dose of HRV induced significantly higher diarrhoea scores early (PID 1 and 2) in infection and a higher percentage of piglets with jejunal pathology (PID 2 and 4), but similar amounts of faecal infectious virus shedding, serum IFN-α (Fig. 7) and frequencies of intestinal IFN-α⁺ pDCs. A higher dose did not induce higher replication of HRV in DCs as measured by quantification of NSP3 RNA, which could explain the lack of dose effect on serum IFN-α. In conclusion, a high dose increased parameters associated with inflammation (diarrhoea and intestinal pathology) but did not induce increased IFN-α by DCs compared with a lower dose. This phenomenon was not explained by higher frequencies of RV-infected DCs associated with a higher dose.

Discussion

The responses of DCs ex vivo were studied after HRV infection of Gn piglets. The most abundant cytokine in serum and SIC detected after HRV infection of neonatal piglets was IFN-α. After an intestinal viral infection it is not surprising to detect IFN-α in serum or intestinal contents. However, it was striking to find that IFN-α was considerably more abundant than any other cytokine measured. Influx of total DCs was mainly to the intestine and minimally to the spleen and MLN. The main IFN-α⁺ DCs were intestinal pDCs followed by intestinal cDCs. A high correlation with serum IFN-α and IFN-α⁺ pDCs suggests that these DCs were the main producers of this cytokine. Systemic or MLN DCs did not produce any detectable IFN-αex vivo. The limited uptake/binding of 2/4/6/7VLP-GFP demonstrated that intestinal pDCs and cDCs were functionally mature at PID 4. Intestinal macrophages/monocytes, despite migrating in abundance to the gut, did not produce IFN-α or mature functionally as measured by the uptake/binding of 2/4/6/7 VLP-GFP compared with the other DC populations. The influx and cytokine production by intestinal and not systemic DCs probably reflects the fact that HRV is mainly enteropathogenic despite the transient viraemia observed.³³

In a previous study from our laboratory, serum and SIC cytokines were quantified in neonatal piglets after infection with 5 × 10⁷ ffu of Wa HRV. As described in the current study, after infection with 10 ffu of M strain HRV, only minimal amounts of the innate IL-6, TNF-α and the Th2 inducer IL-10 were detected acutely. Similar amounts of the Th1 inducer IL-12 were detected in serum, but IFN-α was not measured in this previous work.³⁸ In a different publication from our laboratory, Gn piglets were orally inoculated with 10⁵ ffu Wa HRV and the main responses were at PID 28, derived from intestinal CD8⁺ and CD4⁺ T cells and Th1 biased.⁴⁶ Minimal amounts of IL-13⁺ T cells were detected. Lower splenic T-cell frequencies were observed and only the intestinal IFN-γ⁺ T cells were correlated with protection against diarrhoea. Antibody-secreting cells (ASCs) induced after infection with 10⁵ ffu Wa HRV first appeared at PID 8, and were mainly intestinal IgA and IgG ASCs with minimal induction of splenic ASCs. The intestinal localization of HRV-specific T and B cells detected after HRV infection corresponds to the dominance and activation patterns of intestinal DCs that we observed in the current study. The lower systemic B-cell and T-cell HRV-specific responses previously observed after virulent HRV infection were probably induced by the activated systemic DCs that we observed at much lower frequencies.

Infection of neonatal piglets with 10 ffu of M strain HRV mainly induced IFN-α followed by IL-12 in serum. It is known that both IFN-α and IL-12 are Th1 inducers.^16,47 However, after HRV infection of children and Gn piglets it is known that both Th1 and Th2 types of immunity are induced.^38,48 In the current study we did not observe increases in the serum Th2 inducer IL-10 after low-dose HRV infection. However, in previous studies³⁸ using high HRV doses, we observed that as early as PID 3, the Th2 cytokine IL-4 increased in serum. The type of cell producing this early cytokine response remains to be determined. Also, infection with M strain HRV induced primarily intestinal IFN-α⁺ pDCs and cDCs and not macrophages/monocytes. The same trend was observed in terms of functional maturity measured with the VLP uptake/binding assay. In humans, myeloid DCs (or cDCs) and pDCs were able to induce RV-specific effector T cells in vitro.^25,26 Whether intestinal pDCs and cDCs are able to induce in vivo RV-specific effector T cells that help to clear the infection needs to be explored further, but the human studies and our data suggest that in vivo, both pDCs and cDCs may play a major role in the induction of immunity to RV.

The 1000 ffu dose induced an increased percentage of piglets with jejunal pathology (PID 2 and 4) and early diarrhoea (PID 1 and 2) compared with the lower 10 ffu dose. We did not observe, however, a dose effect between the HRV dose (10 ffu versus 1000 ffu HRV) and infectious virus shed or IFN-α in serum and frequencies of IFN-α⁺ pDCs. We also did not observe that the 1000 ffu dose infected more DCs than the 10 ffu dose. There is evidence that nitric oxide (NO) that is induced during inflammation inhibits IFN-α.⁴⁹ Therefore, increased disease and intestinal pathology, which is possibly associated with more inflammation and cell damage by-products, may have partially inhibited the activation of intestinal pDCs. However, inhibition caused by cell damage by-products, as previously shown for human pDCs, remains to be tested for porcine pDCs.⁵⁰

Irrespective of the dose, piglets generally shed similar titres of infectious virus at most PIDs. It is known that RV NSP4 induces NO after RV infection of humans.^51,52 It has been suggested that NSP4 is a structural protein and therefore present in the HRV particles that were initially inoculated.⁵³ It is possible that in our study, a higher dose induced an initially higher NO concentration than that induced by a lower dose, given the higher amount of NSP4 present in the higher dose inoculum, hence inducing a higher percentage of piglets with diarrhoea. This hypothesis remains to be tested and could explain the earlier induction of disease in piglets inoculated with a high HRV dose despite the similar infectious virus shed compared with the 10 ffu inoculated piglets.

There are two main questions that need to be answered about the in vivo interaction between type I IFNs and infectious RV. (i) Does RV infection prevent the production of IFN-αin vivo? (ii) Does IFN-α prevent RV infection in vivo? To answer (i), we observed that IFN-α is produced in high concentration after HRV infection. The effect of RV on the production of type I IFNs has been shown to depend on the type of cell producing the cytokine in vitro.⁵⁴ The IRF-3 can be rapidly degraded in mouse embryonic fibroblasts after RRV infection leading to inhibition of type I IFN production. On the other hand, RRV infection of mDCs (or cDCs) only partially prevented the production of type I IFNs. In conclusion, our evidence suggests that RV does not prevent the DC production of IFN-αin vivo. In response to (ii), in vitro studies have shown that type I IFNs interfere with RV infectivity. Treatment of human colonic cell lines with IFN-α diminished RRV infection to a maximum of 90% compared with untreated cell lines.⁵⁵ Double KO mice for IFN-α/β and γ receptors or STAT KO mice infected with heterologous RV (RRV or SA-11) had higher systemic replication and RV titres in the intestine compared with WT mice, suggesting that type I IFN inhibition of RV infection occurs in vivo.²² In the same study, however, infectious virus RNA copies did not differ between KO and WT mice infected with homologous RV (EC). In the current study, it was difficult to determine if IFN-α blocks RV infection because by infecting with two different HRV doses, similar amounts of IFN-α and infectious virus shedding were detected. However, because we observed a dose effect between HRV and intestinal pathology and diarrhoea, we speculate that there is a complex interaction among intestinal pathology, IFN-α and the dose of the infecting HRV strain. The high virus dose increased parameters associated with inflammation (diarrhoea and intestinal pathology). It is probable that the increased pathology after the higher dose led to increased inflammation by-products, which had an inhibitory effect on pDCs, such that the IFN-α levels were not increased as was expected after the higher dose, but remained similar to those in the piglets receiving a low dose. At the same time, the amount of IFN-α produced by the intestinal DCs may have prevented, to some extent, HRV replication in the gut, explaining the similar HRV infectious shedding titres observed between the two doses. The pDCs are known to have both anti-inflammatory and pro-inflammatory functions.^15–17 Stimulation of the anti-inflammatory effects of pDCs after the high dose, without increasing their pro-inflammatory impacts, may be critical to reduce further immunopathology during HRV infection. However, this needs to be investigated further and future studies will have to address if this inhibitory cycle is observed in vivo.

Previously we have shown that Gn piglets developed similar diarrhoea scores and faecal RV titres after inoculation with HRV regardless of their colonization status with lactobacillus species.⁵⁶ These findings suggest that their Gn status and lack of commensal flora per se does not influence RV replication or induction of diarrhoea. Furthermore, splenic DCs from both germ-free and pathogen-free mice displayed similar immune response parameters (CD86 /major histocompatibility complex class II expression, T-cell proliferation and ovalbumin-induced tolerance).^57,58 On the other hand, the gut microflora (including enteric viruses) plays an important role in the maturation of the local and systemic immune system of the neonate.⁵⁹ Comparative studies of germ-free versus pathogen-free piglets are lacking. Our Gn piglet studies with/without lactobacillus, HRV infection and the germ-free mice studies suggest that DCs in germ-free animals are fully functional. However, the lack of an intestinal microflora might explain the lower frequencies of DCs in the gut and MLNs observed in the control piglets. Nevertheless this immunologically naive Gn piglet model allowed us to more clearly discern the effect of HRV infection on the types and influx of DCs to the gut.

This study describes the ex vivo DC responses after HRV infection of Gn piglets. We show that intestinal and not systemic DCs predominated after HRV infection. We cannot rule out a role for macrophages/monocytes in the induction of adaptive HRV-specific immunity, but our ex vivo assay showed that these antigen-presenting cells were not producing cytokines in abundance early after low-dose or high-dose HRV infection. A lack of dose effect between 1000 ffu versus 10 ffu HRV in infectious virus shed and serum IFN-α suggest the possibility of an inhibitory cycle among inflammatory by-products, local IFN-α and the infecting HRV strain. Also, only intestinal IFN-α⁺ pDCs and cDCs were observed. Based on our data, we conclude that intestinal pDCs and cDCs may be mainly responsible for initiating innate immune responses that then trigger adaptive immunity to HRV in neonatal animals and possibly humans.

Disclosures

None of the authors have a conflict of interest including financial conflict for the research reported here.