Animals
A transmission experiment was conducted in the ANSES level 3 biosecurity facilities using specific-pathogen-free (SPF) piglets produced in these same facilities. Those SPF pigs have a high health status and beside of being free from major diseases (Classical swine fever, African swine fever, Aujeszky disease, porcine epidemic diarrhea virus, transmissible gastro enteritis virus), they are also free from numerous endemic viruses such as porcine reproductive and respiratory syndrome virus, swIAVs, porcine circovirus type 2, porcine parvovirus and bacteria: Mycoplasma hyopneumoniae, Mycoplasma hyorhinis, Actinobacillus pleuropneumoniae, Pasteurella multocida, Bordetella bronchiseptica, Streptococcus suis 2, and Haemophilus parasuis.
SPF dams (gilts) were primo-vaccinated 6 and 3 weeks before insemination with a 2 mL intramuscular injection of an inactivated trivalent H1avN1, H1huN2, H3N2 vaccine (Respiporc Flu®3, formerly GRIPOVAC®3, IDT Biologika GmbH, Dessau-Rosslau, Germany) containing antigens representative of three enzootic lineages circulating in Europe [10, 11], followed by three boosters 6, 3 and 2 weeks before farrowing in order to induce high antibody levels in the colostrum.
Thirty-six SPF piglets born to vaccinated SPF gilts and fed as far as possible by their own dam, constituted a group of piglets with swIAV-specific MDAs (MDA+ group), whereas 36 SPF piglets without MDAs (MDA− group), born to unvaccinated SPF gilts, were also used.
Experimental design
Thirty-three piglets from each group (MDA+ and MDA−) were assigned to 6 independent rooms (2 pens per room) according to their serological status (Figure 1). Pen composition was evenly balanced according to the weight, sex, dam’s origin and the 5-week-old MDA level. In each room, 2 seeder piglets were inoculated intratracheally with 106 EID50 (Embryonic 50% Infectious Dose) of a European avian-like swine H1N1 (H1avN1) virus (A/Sw/Côtes d’Armor/0388/09 strain) from the same genetic lineage than the H1avN1 vaccine antigen [10, 11, 29], in a total volume of 5 mL, at 35 days of age (D0). Piglets to be inoculated were gathered in a different room for the inoculation and placed in contact with the other animals 1 day post-inoculation (dpi 1). In each room (3 replicates per group), the 2 seeder piglets were placed in contact with 4 piglets in the same pen to assess transmission through direct contact and 5 piglets were placed in the second pen, 30 cm apart, to assess transmission through indirect contact (no pig-to-pig contact allowed).
The remaining piglets from each group (3 MDA+ and 3 MDA− piglets) were placed in 2 different pens in a seventh independent room (Figure 1) and were mock-inoculated [5 mL of minimum essential medium (Eagle, Lonza, Belgium)], constituting the negative control groups. The experiment ended on dpi 28, corresponding to 63 days of age.
The experiment protocol was approved by the Anses/ENVA/UPEC Ethical Committee on animal experiments (agreement #16 with the National committee for Ethics in animal experimentation) and authorized by the French Ministry for Research under the legal notice 11/03/14-17.
Sampling and laboratory analyses
Biosecurity measures
Strict measures were taken to prevent viral transmission through human or fomite contacts. Samples were taken firstly in the indirect-contact pen and secondly in the pen that housed inoculated and direct-contact piglets. Inside the latter, direct-contact piglets were sampled before inoculated piglets. In each room, clothing and footwear were cleaned and gloves changed after each pen visit. Clothing and footwear were changed between each room and a shower was taken when leaving and entering another room. Individual sampling materials were used for each piglet.
swIAV serological analyses
Blood samples were collected by jugular vein puncture, using evacuated tubes without an additive (Vacuette, Dutscher SAS, Brumath, France). Vaccinated gilts were sampled 4 and 1 week before, and 3 weeks after, insemination. They were also bled at 7, 5, 3 and 2 weeks before farrowing and 1, 2, 6, 15 and 20 weeks post-farrowing to verify that high antibody levels would be obtained at farrowing and to further monitor serological titres in sera on a long-term basis. In piglets, blood samples were taken at farrowing (34 days before inoculation), 3 days before the trial began and twice a week after infection (dpi 4, 7, 11, 14, 18, 21 and 25). Sera (576 samples) were obtained by blood centrifugation for 5 min at 3500 × g and stored at −20 °C. Post-vaccination antibodies, MDAs and post-infection antibodies directed against swIAVs (NP and M antigens) were quantified in sera using LSIVet™ Porcine influenza—Serum ELISA Kit (VETSIV/I, Life Technologies, Courtabœuf, France). Antibody levels are expressed in % IRPC (Relative Index Percent). Although the positive threshold is defined by the manufacturer as 20% IRPC in field conditions, we determined the positive threshold corresponding to our experimental conditions in SPF pigs considering the results from SPF MDA− piglets. Moreover, cross-classified results with HI tests on negative and a range of positive serums were used to define categories of piglets being negative, or with low and high antibody titres. Post-vaccination antibodies in gilts were also titrated in hemagglutination inhibition (HI) tests using virus strains representative for H1avN1, H1huN2 and H3N2 lineages as antigens, following the procedure described in [15].
swIAV detection and quantification in nasal secretions
To assess viral shedding, nasal swabs (MW951 sent, Virocult®, Corsham, UK) were taken from seeders and direct and indirect contacts from both MDA+ and MDA− groups on a daily basis from D0 to dpi 14, and every 2 days thereafter until dpi 28. In control groups, nasal swabs were taken once a week throughout the duration of the experiment. All nasal swab supernatants (1317 samples) were stored at −70 °C until virological analysis.
For swIAV detection, RNA were isolated using NucleoSpin® 8 Virus (Macherey Nagel, Hoerdt, France) and submitted in quadruplicates to high-throughput analyses on LightCycler® 1536 Real-Time PCR Instrument (Roche). Briefly, real-time M gene RT-PCR [30] was conducted in duplex with the amplification of porcine β-actin gene [31] as an internal control. Each replicate comprised 1 µL RNA extract, 0.4 µM and 0.8 µM of M gene forward and reverse primers, respectively, 0.6 µM of β-actin gene primers and 0.25 µM of probes in a final volume of 2 µL. Reverse transcription (RT) was performed for 30 min at 45 °C using GoScript™ RT Mix M700A (Promega) at a 2× final concentration. After an initial activating step of 2 min at 95 °C, PCRs were run using Real Time ready DNA Probes Master Mix (Roche) at a 1× final concentration. Fluorescence data were collected at the end of each of the 40 cycles of 1 s at 95 °C and 30 s at 60 °C. One sample was interpreted positive as soon as M gene was amplified in one replicate out of four.
The swIAV genome was then quantified in selected samples from contact piglets. Five µL of RNA extracts were tested using same primers and probes as above and the GoTaq® Probe 1-Step RT-qPCR System (Promega, Madison, USA) in a total volume of 25 µL (M and β-actin probes at 0.5 µM and 0.3 µM, respectively). RT-PCR (45 °C for 30 s, 95 °C for 2 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min) was performed in a MX3005P qPCR System (Agilent Technologies, Santa Clara, USA). Serial dilutions of standardized M and β-actin mRNAs were tested similarly to generate standard curves. The swIAV genome amount was expressed as the M gene copy number per 104 copies of β-actin gene.
Virus detection in aerosols
To detect swIAV genome in aerosols, air samples were taken in each experimental room housing infected piglets 3 times a week between D-3 and dpi 25 using Coriolis®μ microbioal air sampler (Bertin Technologies, St-Quentin en Yvelines, France) (300 L/min, 10 min/room, in 15 mL of 0.005% Triton solution). Collected air samples were concentrated using Amicon® Ultra-15 30 K centrifugal filter devices (Merck Millipore Ltd, Ireland). After centrifugation for 30 min at 3900 × g, RNA were purified from 150 µL eluate using RNeasy Mini Kit© (Qiagen GmbH, Hilden, Germany) and 5 µL RNA extract were tested by real-time M gene RT-PCR (LSI VetMAX™ Swine Influenza A-A/H1N1/2009 included, Life Technologies, Lissieu, France) [32]. The conversely cycle threshold 45–Cq values were used to represent a semi-quantitative tendency of the evolution of swIAV genomic load in aerosols.
Statistical analysis and models
Duration of the persistence of maternally-derived antibodies
A nonlinear mixed-effects model (NLME) based on serological data was used to estimate individual parameters governing antibody kinetics in MDA+ piglets. The decay of antibody titres was assumed to follow an exponential decrease governed by equation dA/dt = −rA. Thus, the MDA levels depend on the initial level of MDAs A
0 and MDA decay rate r. The model describing the serological titre of the individual i at observation time t
j
(with a constant residual error model) is given by:
$$A_{ij} = A_{0}^{(i)} e^{{ - r_{i} t_{j} }} + a\varepsilon_{ij} ,$$
where A
(i)0
and r
i
are the individual parameters and ɛ
ij
a vector of standardised random variables. Individual parameters were assumed to be log-normally distributed, as described in Snoeck et al. [33]. The model for individual parameters is given by:
$$A_{0}^{(i)} = A_{0}^{pop} e^{{\eta_{A}^{(i)} }} \;{\text{and}}\;r_{i} = r_{pop} e^{{\eta_{r}^{(i)} }} ,$$
where i denotes the individual, r
pop
and A
pop0
the median decay rate and initial serological level of the global population. Vectors η
(i)
A
and η
(i)
r
are vectors of random effects assumed to be independent centred Gaussian vectors with variance Ω
A
and Ω
r
representing inter-individual variability [34]. Population parameters were estimated using MLE by the SAEM algorithm for the hierarchical nonlinear mixed-effects model analysis with Monolix software [35, 36].
Individual parameters were used for long–term projections of individual profiles. A parametric survival analysis with a gamma distribution of survival times was finally carried out to estimate the duration of MDA persistence.
Modelling within-host infectious process using viral genome loads quantified in nasal swabs from contact piglets
A model of within-host influenza dynamics [37] was used to compare shedding dynamics in MDA+ and MDA− contact piglets based on the target-cell-limited TIV model (susceptible Target cells, productively Infected cells, free Viral particles assessed by viral genome load; [38–40]):
$$\frac{dT}{dt} = - \alpha TV$$
$$\frac{dI}{dt} = \alpha TV - \delta I$$
$$\frac{dV}{dt} = pI - cV.$$
This set of differential equations describes the dynamics of susceptible target cells T becoming infected (I) at a constant rate α by contact with free infectious viral particles V. Infected cells I produce virus at an average rate p per cell until their death at a rate δ per day, giving an average lifespan for I cells of 1/δ. Free viral particles disappear at a clearance rate c per day. The effects of the immune response are implicitly included in the δ and c rates. Parameters governing the differential equations of the model were individually estimated for each group (MDA+/MDA−) by least squares minimisation using a quasi-Newton algorithm. Based on individual parameters, differences in shedding pattern according to the initial serological status of the animals were assessed by comparing the peak of viral shedding and the global amount of virus shed throughout the shedding period [areas under the fitted curves (AUC)] using ANOVA.
Estimation of between-host transmission parameters
The virus transmission process was described by an MSEIR model [41] (Piglets having MDAs—Susceptible—Exposed (i.e. latently infected)—Infectious—Removed), incorporating (i) two different transmission routes according to the contact structure between individuals (direct contact in the same pen or indirect airborne transmission) and (ii) different levels of susceptibility to infection to account for the initial serological status of the piglets (MDA+ or MDA−). A piglet was considered infectious from the first to the last positive RT-PCR nasal swab and the duration of the shedding period was estimated by parametric survival analysis. In the event of a negative RT-PCR result between two positive results, the piglet was assumed to be infectious at that time. Combining the results of previous experiments [7] and field observations [15], a latency duration of between 0.5 and 1.5 days was selected for contact piglets. Thus, as the exact time contact piglets became infected was not observed due to the latency stage, the interval in which piglets became infected was determined by subtracting 1.5 and 0.5 days (the latency period) from the first positive RT-PCR (i
1) giving an infection interval [e
1, e
2] for each contact-infected piglet calculated as [i
1—1.5, i
1—0.5] [42].
Two transmission routes were considered in this experiment, parameterised by the corresponding transmission rate parameter: (i) direct transmission β modelling pig-to-pig contacts between pen-mates and (ii) airborne transmission β
air
affecting all animals within a room. Those transmission rates were weighted by a susceptibility-to-infection factor σ which accounts for a different susceptibility to infection for piglets in presence of MDAs compared to MDA− piglets (σ = 1 in this latter case). The prevalence of shedding piglets per pen and per room was calculated for each time interval T (duration Δt). Assuming that all the shedding animals contribute to infection pressure by the airborne route, the within-room prevalence was considered as an approximation of the virus quantity in the surrounding air. The transmission rates (β and β
air
) represent the number of newly-infected piglets due to one infectious piglet per day. The global force of infection λ
k
combines the two viral transmission routes. Thus, a piglet k can become infected by direct contact with an infectious pen-mate (transmission rate parameter β) or by contact with a contaminated aerosol (parameter β
air
), weighted by the susceptibility factor σ according to its serological status. The global force of infection λ
k
for a piglet k located in pen p and room r is thus calculated at time t as follows:
$$\lambda_{k} (t) = \beta \frac{{I_{p} (t)}}{{N_{p} - 1}} + \beta_{air} \frac{{I_{r} (t)}}{{N_{r} }}$$
with I
p
(t) and I
r
(t) the (time-dependent) number of infectious piglets in pen p or room r at time t; N
p
and N
r
the total number of piglets in pen p or room r.
The probability for each piglet k to avoid infection when submitted to the global force of infection λ
k
up to time e
1 is \(e^{{ - \int_{0}^{{e_{1} ,k}} {\sigma_{k} \lambda_{k} (t)dt} }}\), with σ
k
being the susceptibility factor of individual k: \(\sigma_{k} = \left\{ {\begin{array}{*{20}c} 1 & {{\text{if}}\quad{\text{MDA}}^{ - } \,{\text{piglet}}} \\ \sigma & {{\text{if}}\quad{\text{MDA}}^{ + } \,{\text{piglet}}} \\ \end{array} } \right.\). The probability of becoming infected during time interval [e
1, e
2] is therefore \(1 - e^{{ - \int_{{e_{1} ,k}}^{{e_{2} ,k}} {\sigma_{k} \lambda_{k} (t)dt} }}\).
The product of all contact-infected piglets gives the full likelihood function L:
$$L = \prod\nolimits_{{k \in contact{ - }infected}} {e^{{ - \int_{0}^{{e_{1} ,k}} {\sigma_{k} \lambda_{k} (t)dt} }} *\left( {1 - e^{{ - \int_{{e_{1} ,k}}^{{e_{2} ,k}} {\sigma_{k} \lambda_{k} (t)dt} }} } \right)}$$
The transmission rates (β and β
air
) and the susceptibility factor (σ) were then estimated by likelihood maximisation [43]; confidence intervals were derived from the likelihood profile.
The impact of the contact structure and the initial serological status on (i) the shedding start time and (ii) shedding duration was assessed using a semi-parametric Cox model.