npb #12-113 revised€¦ · title: the effects of disease challenge (prrs) on pig growth and...
TRANSCRIPT
Title: The effects of disease challenge (PRRS) on pig growth and metabolic pathways NPB #12-113
revised
Investigator: Nicholas Gabler
Institution: Iowa State University
Date Submitted: June 10, 2014
Industry Summary. While significant advances have been made through research efforts to enhance the
health, productivity, and well-being of swine, the detrimental effects associated with disease outbreaks remains
a significant challenge to the U.S. pork industry. Therefore, the goal of this research was to characterize over
time the changes in blood metabolic and immunological parameters of pigs infected with porcine reproductive
and reproductive syndrome (PRRS) virus using canonical discriminant analysis (CDA) model. A total of 76
gilts (16.1 ± 0.43 kg BW, 47-60 d old; EBX Pure; Choice Genetics, West Des Moines, IA) were selected and
randomly assigned to one of four pens (19 pigs/pen) in a commercial finishing unit in central Iowa. Blood
samples were collected before selection to verify that all gilts were PPRSV naïve. Body weights and blood
samples were collected on each animal at 0, 7, 14, 28 and 56 days post-inoculation (dpi). Retrospectively, the
12 highest and 12 lowest growth rate gilts over a 56-day PRRS challenge study were selected for mutivariant
blood analysis. Their serum was analyzed for 52 metabolites, 17 complete blood count (CBC), and
inflammatory traits on dpi 0, 7, 14, 28 and 56. Our CDA analysis revealed no major differences in blood
parameters between dpi 0 and 56. However, during early stages of PRRS infection (dpi 7 and 14), amino acid
mobilization markers increased potentially for immune protein synthesis and energy requirements, whereas in
later stages (dpi 28) we observed an increase in protein catabolism markers. Altogether, these results indicate
dynamic changes in immune and energy requirements for pigs growing through a PRRS challenge. By
understanding these changes or shifts in metabolism as a result of immune stress, long term, we hope to develop
management strategies to optimize pig performance in the face of disease challenge.
Key Findings:
Within a given population, pigs handle a PRRS virus challenge differently. Even though all pigs were
infected with PRRS, the low growth rate population had a 26% reduction in ADG compared to the high
growth gilt population.
Mutivariant blood analysis of these PRRS infected pigs analyzed for over 52 metabolites, 17 complete
blood count (CBC) and inflammation traits revealed dynamic changes over a 56 day challenge period.
During early stages of PRRS infection, amino acid mobilization is increased for immune protein
synthesis and energy requirements.
Whereas in later stages of infection indicated increased protein catabolism.
2
Keywords: Porcine Reproductive and Respiratory Syndrome, Metabolism, Growth
Scientific Abstract. The metabolic and immune response in pigs inoculated with porcine reproductive and
reproductive syndrome (PRRS) was assessed using canonical discriminant analysis (CDA) of multiple blood
parameters. Twenty-four gilts (BW 16±4.4 kg) were selected based on different residual growth rate gain over a
56-day growth study following an I.M. PRRS virus challenge in a commercial setting. Each animal had blood
samples collected on days 0, 7, 14, and 28 post-inoculation (dpi) for measurements on 52 metabolites, 17
complete blood count (CBC), and inflammatory traits. Prior to the CDA, traits were analyzed in a univariate
manner in order to identify traits that could potentially discriminate the different phases of PRRS progression:
7, 14, and 28 dpi. The univariate analyses statistical model included the fixed effects of growth rate group
(GRG), infection status (IS), dpi and their interactions, the initial age as covariate, and pen as random effect.
Traits were included for CDA when P<0.1 for the effects dpi or dpi interactions. Thirty-three traits had a P<0.1
for dpi or its interactions were included in the CDA. The first and second canonical variables (CAN1 and
CAN2, respectively) were significant (P<0.01 and P=0.02, respectively) and showed squared canonical
correlations of 0.95 and 0.86, respectively. While CAN1 discriminated dpi 7 and 14 from 28, CAN2
discriminated dpi 7 from 14. The best discriminators in CAN1 were alkaline phosphatase (ALP) and
haptoglobin, whereas for CAN2 were C-reactive protein (CRP), glucose, and insulin. CAN1 and CAN2 were
significant of PRRS infection showed high values of haptoglobin and low values of ALP. However, 7 dpi had
low values of CRP, glucose and insulin compared to dpi 14. Additionally, urea showed a potential
discrimination power, with high values on dpi 7 & 14 compared to 28 (CAN1), whereas the amino acids
alanine, proline, and threonine were good discriminators of dpi 7 and 14 (CAN2). This may suggest that during
early stages of PRRS infection (dpi 7 and 14), amino acid mobilization is increased (CAN2) for immune protein
synthesis and energy requirements, whereas in later stages (dpi 28) we observed increased protein catabolism
(CAN1). Altogether, these results indicate dynamic changes in immune and energy requirements for pigs
growing through a PRRS challenge.
Introduction. Porcine reproductive and respiratory syndrome virus is a major swine virus that causes
reproductive impairment in sows and respiratory distress in all ages of pigs. This virus is estimated to cost the
pork industry $664 million annually, mainly stemming from loss of production (Holtkamp et al., 2013). Porcine
Reproductive and Respiratory Syndrome virus (PRRSV) infections induce both humoral and cellular immune
responses. Recent reports indicate that serum pro-inflammatory cytokine concentrations are elevated early after
infection (Lunney et al., 2010). In addition, Che et al., (2012) observed that PRRSV-infected pigs experience
increases in pro-inflammatory cytokines (i.e., TNF-alpha) as well as acute phase proteins such as C-reactive
protein and haptoglobin at d 7 and 14 post-infection. This continuous immune stimulation in the rearing
environment results in the production of potent pro-inflammatory cytokines, which antagonize anabolic growth
factors, and suppresses growth (Johnson, 1997; Spurlock, 1997; Broussard et al., 2003).
Therefore, the goal of this research project was to characterize the metabolic impact that immune system
activation and disease has on a growing pig’s metabolism and growth. This characterization will help lead to the
develop mitigation and management strategies to alleviate the negative effects disease and inflammation have
on pig performance and carcass quality. While significant advances have been made through research efforts to
enhance the health, productivity, and well-being of swine, the detrimental effects associated with disease
outbreaks remains a significant challenge to the U.S. pork industry. The objective of this research project was
to challenge grower pigs in a commercial barn with a live field strain of PRRS and to then monitor and
characterize both the short to medium term impact of this challenge on the pig’s immune system, metabolism,
and growth performance. We will use non-invasive blood sampling to identify changes in the pigs’ clinical
pathology and metabolome over an eight week challenge period. Although important, we are not interested in
characterizing the acute changes in immunity and immune competent proteins directly post challenge (0-72
hours), but the impact downstream on immunity, metabolism and performance.
3
Materials & Methods. Animals, Housing and Experimental Design
The Iowa State University Institutional Animal Care and Use Committee approved the experimental
protocol used in this study. A total of 76 gilts (16.1 ± 0.43 kg BW, 47-60 d old; EBX Pure; Choice Genetics,
West Des Moines, IA) were selected and randomly assigned to one of four pens (19 pigs/pen) in a commercial
finishing unit in central Iowa. Blood samples were collected before selection to verify that all gilts were PPRSV
naïve. These gilts where fed a standard Midwest corn-soybean-corn DDGS diet in two phases that met or
exceeded NRC (2012) requirements for swine. Pigs had free access to water and were fed ad libitum, unless
otherwise stated. All pigs were inoculated intramuscularly with 2 mL of a high-virulence field strain of PRRSV
isolated from central Iowa. Following a 6-12 h overnight feed withdrawal, all pigs were snared and bled via
venipuncture from the jugular vein for serum and plasma collections at 0, 7, 14, 28 and 56 d post infection (dpi).
Plasma and serum from all pigs was used for determination of viral load, antibody titer and complete blood cell
counts. Body weights were also collected at 0, 7, 14, 21, 28, 42, and 56 dpi to calculate individual pig ADG
over the 56 day infection period.
Blood Diagnostics
Fresh plasma samples were stored on ice and immediately sent to the Iowa State University Clinical
Pathology Laboratory for complete blood cell count (CBC) analysis within 6 h of collection. The CBC traits
measured included the number of white blood cells (WBC), lymphocytes, basophils, neutrophils, monocytes,
and eosinophils, red blood cell count (RBC), hemoglobin concentration (Hg), hematocrit percent, mean
corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin
concentration (MCHC), and red blood cell distribution width (RDW), platelet count and mean platelet volume
(MPV). Additionally, plasma and serum samples were centrifuged at 1,000 × g for 10 minutes and the aliquots
of serum and plasma were stored at −80°C until further analysis was conducted.
A serum sample from all pigs was assayed for PRRSV antibodies at the Iowa State University
Veterinary Diagnostics Laboratory using a commercial indirect ELISA (PRRSX3 Ab Test, IDEXX
Laboratories, Inc., Westbrook, ME USA) performed according to the manufacturer’s instruction. As
recommended by the manufacturer, a positive result was defined as a sample-to-positive (S/P) ratio ≥ 0.4.
PRRS virus load was also determined by routine qRT-PCR protocol for serum at the Iowa State University
Veterinary Diagnostics Laboratory. In brief, nucleic acid extraction from serum samples was performed using a
commercial RNA extraction kit (Ambion® MagMax™-96 Viral RNA isolation kit, Applied Biosystems™,
Foster City, CA USA). Real-time PCR was performed with commercial reagent sets (TaqMAN® NA and EU
PRRSV Reagents and TaqMAN® NA and EU PRRSV Controls, Applied Biosystems™) using the following
cycling conditions: 1 cycle at 45°C for 10 minutes, 1 cycle at 95°C for 10 minutes, 40 cycles of: 97°C for
2 seconds, 60°C for 40 seconds. Eight 10-fold serially-diluted (100 to 107 copies/μl) plasmid-derived
commercial standards (TaqMAN® NA and EU PRRSV RNA controls, Applied Biosystems™) were run on
each PCR plate and their Ct values used to derive a standard curve. Samples were quantified as genome
equivalents per μl (geq/μl) by fitting the sample Cts to the standard curve using the AB7500 Fast System SDS
Software (Applied Biosystems™).
Selection of pigs for multivariate analysis
Based on analysis of the entire test population, 12 high residual and 12 low residual growth rate pigs
were selected for in depth blood metabolite and immune function analysis on samples collected at 0, 7, 14, 28
and 56 d dpi. Therefore, 24 pigs were longitudinally analyzed for blood cytokine, acute phase proteins,
lipidomics and 1H-NMR metabolomics over the 56 d PRRSV challenge.
Serum endotoxin concentrations were measured by an end point fluorescent assay using the recombinant
factor C (rFC) system (Lonza, Basel, Switzerland). Briefly, the serum samples were diluted 1000X in pyrogen
free water and 100µL of the samples and standards were added to a 96 well round bottom plate and incubated at
4
37°C for 10 min. After incubation, 100 µL of rFC enzyme, rFC assay buffer and rFC substrate were added at a
ratio of 1:4:5 to the plate and an initial reading were taken followed by 1h incubation at 37°C. Thereafter, the
relative fluorescence unit (RFU) for each well was determined (excitation 380 nm and emission 440 nm). The
concentration of the endotoxin was interpolated from the standard curve constructed from the standards and
corrected for sample dilution.
Alkaline phosphatase (ALP) activity was measured using the Quantichrom ALP assay kit (DALP-250,
Gentaur, Bioassay systems, Hayward, CA). Briefly, 50 µL of serum was added to a 150 µL working solution
containing magnesium acetate, p-nitrophenyl phosphate and assay buffer in a 96 well plate. The optical density
at 405 nm was measured at time 0 and after 4 minutes using a Synergy 4 microplate reader (Bio-Tek, Winooski,
VT) and ALP activity was calculated according to the manufacturer’s instructions. Plasma lysozyme activity
using the EnzChek fluorescent assay which compares sample lysozyme activity to lysozyme activity on
Micrococcus Lysodeikticus cell walls (Invitrogen-Molecular Probes, Carlsbad, California). Samples were
diluted and fluorescence was measured using excitation emission wavelengths of 485 and 530nm and the
lysozyme activity were interpolated from the standard curve constructed from the standards and corrected for
sample dilution.
Serum C-reactive protein (CRP) and haptoglobin were analyzed using commercially available ELISA
kits (ALPCO Diagnostics, Salem, NH). Briefly, serum samples were added to wells adsorbed with anti-porcine
haptoglobin, CRP and LBP antibodies. After washing, horseradish peroxidase (HRP) conjugated anti
haptoglobin, CRP and LBP antibodies were added to the plate. After another washing, the HRP was assayed by
the addition of the chromogenic substrate 3,3’,5,5’-tetramethylbenzidine (TMB) and the absorbance was
measured at 450 nm. The quantity of haptoglobin and CRP in the test samples was interpolated from the
standard curve constructed from the standards and corrected for sample dilution. Further, serum tumor necrosis
factor (TNF)-α was measured also using a commercially available ELISA kit (Quantikine® Porcine TNF-α,
catalog number PTA00, R&D systems, Minneapolis, MN, USA).
Plasma insulin was analyzed in duplicate using an ELISA kit solid phase two-site enzyme immunoassay
based on the sandwich technique (Mercodia Porcine Insulin ELISA, ALPCO Diagnostics, Salem, NH). Plasma
non-esterified fatty acids (NEFA) concentrations were determined using a commercially available kit (Wako
HR Series NEFA-HR, Wako Diagnostics, Richmond, VA). Plasma urea nitrogen was analyzed using a
commercially available kit (Quantichrom Urea Assay Kit, BioAssay Systems, Hayward, CA).
Blood 1H-NMR and lipidomic analysis was conducted at the David H. Murdock Research Institute
(Raleigh, NC). The serum samples were thawed and 300 microliters of serum was added to 300 microliters
100mM phosphate buffer (pH of 7.4 in D2O, Sigma-Aldrich) in a centrifuge tube. Ten microliters of 100mM
DSS (Cambridge Isotope Laboratories) solution was added as an internal standard. These samples were
vortexed and transferred to NMR tubes for experiments. After experiments were performed, samples were
restored at -80oC. One sample was randomly chosen to obtain a rough estimate of concentration. Four months
after initial analysis, sample 572 T7 was removed from the -80oC. 60 microliters of 1mM formate solution was
added to the sample as an internal standard. The sample was then vortexed and transferred to an NMR tube for
analysis. NMR analysis was performed using a 950 MHz Bruker NMR spectrometer equipped with
cryogenically cooled TCI probe and Avance III console (the statement of work called for use of a 600 MHz
NMR spectrometer; however, the 950 MHz spectrometer is more sensitive and was available, therefore it was
used. This was a decision of the DHMRI; therefore the customer will be charged at the 600 MHz rate). A one-
dimensional CPMG experiment with pre-saturation water suppression was performed on each sample. The
CPMG sequence was used to suppress any large compounds such as proteins that may have still been present in
the sample. The water suppression was used to attenuate the water signal. Experiments were performed using
256 scans and a delay of two seconds. All experiments were performed at 25oC. One sample, 163T14, was
missing. All spectra were aligned by setting the doublet of the methyl group of valine to 1.036 ppm in all
spectra. The internal standard DSS could not be used due to its interaction with compound within the serum.
Data was analyzed in by integrating peaks in each spectrum. The same integral regions were used for all
spectra and 114 integral regions were evaluated. Integral values were normalized by dividing by the sum of the
5
integral region of 8.5 ppm to 5.15 ppm and the integral region 4.65 ppm to 0.5 ppm (excluding water and the
internal standard). Integral regions were compared to an internal spectral library of 170 compounds in order to
relate the regions to the correct metabolites. Data was then analyzed using multivariate statistical analysis.
For lipidomic analysis, methanol (JT Baker, LC/MS), acetonitrile (JT Baker, LC/MS), and 2-propanol
(JT Baker, LC/MS) were purchased from Mallinkrodt-Baker (St. Louis, MO). Dichloromethane was purchased
from Applied Biosystems (Forester City, CA). Ammonium Acetate was purchased from Sigma-Aldrich (St.
Louis, MO). The internal standards were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Leucine
enkephalin solid (Waters Corp., Milford, MA) was used to prepare a lockmass solution at 50 pg/μL in
acetonitrile:water:formic acid (500:500:1, v/v/v) and a sodium formate solution (0.05 M NaOH + 0.5% formic
acid in 90:10 2-propanol: water, Waters Corp., Milford, MA) was used for instrument tuning and calibration.
Stored serum samples were thawed and serum samples, pooled QC samples and blank controls were prepared
simultaneously. A pooled QC sample was prepared by combining 30 μL aliquots from each sample, vortexing
to mix, and preparing 15 individual QC samples. This pooled QC sample was analyzed throughout the study
sequence to monitor for system variation. For the serum and pooled QC sample, a 30 μL aliquot of each serum
sample was mixed with 60 μL of internal standard solution (250 μg/mL each of 1-heptadecanoyl-2-hydroxy-sn-
glycero-3-phosphocholine, 1-nonadecanoyl-2-hydroxy-snglycero-3-phosphocholine, 1,2-diheptadecanoyl-sn-
gycero-3-phosphoethanolamine, 1,2-diheptadecanoyl-sn-glycero-3-[phosphor-rac-(1-glycerol)], and 1,2-
diheptadecanoyl-sn-glycero-3-phosphocholine in 2:1 DCM:MeOH). The diluted sample was extracted with 190
μL of MeOH and vortexed for 20s, 380 μL of DCM and vortexed for 20s, and 120 μL of Milli-Q water and
vortexed for 20s. The solution was allowed to sit at room temp for 10 minutes and was then centrifuged for 10
minutes at 4,000 x g and 4°C. The lower organic layer was collected (250 μL) and diluted five times with
methanol for analysis. A blank solution was simultaneously made from the above steps omitting the serum and
standard solutions. The injection volume for all samples was 10 μL.
A UPLC-QTOF system (ACQUITY UPLC-SYNAPT HDMS, Waters Corp., Milford, MA) was used to
profile global lipids in the porcine plasma samples. The system was operated in both electrospray ionization
(ESI) positive and negative modes to provide complementary information. The optimized instrument settings
consisted of a ACQUITY UPLC HSS T3 5 μm VanGuard pre-column (2.1×5 mm) and ACQUITY UPLC HSS
T3 1.8 μm analytical column (2.1 × 100 mm), column temperature of 55 oC and a flow rate of 0.4 mL/min. The
gradient conditions consisted of Gradient A (10 mM Ammonium Acetate in 4:6 ACN:H2O) and gradient B (10
mM Ammonium Acetate in 1:9 ACN:IPA). The raw data files generated by the UPLC-QTOFMS system were
processed using the MarkerLynx XS application manager (Waters Corp., Milford, MA). Using this software,
data extraction (including baseline correction, peak detection, and peak alignment) was performed. Software
parameters included: retention time for data acquisition from 0.5 min to 12.5 min, mass range from 50 to 1200
Da, and mass tolerance within 0.05 Da. The parameters for Apex Track Peak were automatically determined by
the MarkerLynx XS software. The minimal peak signal intensity was 50 ion counts with a noise elimination
level of 6. The resulting data set was organized into a matrix including sample information, arbitrary identities
for all the detected peaks (Retention Time-Mass pairs), and an intensity determination for each detected peak.
Statistical Analysis
All statistical procedures were performed in SAS 9.3 (Statistical Analysis System Institute, Inc., Cary, NC,
USA). Due to no differences between dpi 0 and dpi 56, all data for dpi 56 was dropped from the analysis. Pig
was considered the experimental unit and the metabolomics, cytokine and complete blood count (CBC) data
was analyzed in an univariate mixed model including the fixed effects of growth rate group (GRG; high and low
gain), days post infection (dpi; 7, 14, and 28) and all possible interactions between these effects, and the
covariate age at beginning of the trial, and the random effect of pen. A first-order autoregressive (ar[1])
covariance structure was used to account for the repeated measurements across dpi. Linear and quadratic
contrasts were also tested.
Canonical discriminant analyses (CDA) were performed in order to characterize the stages of PRRS in
pigs with different growth performance. Two CDAs were performed: one to discriminate uninfected from
6
infected animals (CDAIS), and one to discriminate PRRS progression (7, 14, and 28 dpi; CDAdpi). The traits
used in this analysis were selected according to the significance (P<0.1) of Infection and of PRRS progression,
for CDAinf and DCAprog, respectively. The effect of Infection was estimated as a contrast between dpi 0 and the
other dpi (7, 14, and 28). The effect of PRRS progression was estimated as the linear and quadratic effects of
dpi 7, 14, and 28. For the CDAinf, traits were included when showing significant effect of infection and/or
interactions between GRG and Infection, whereas for CDAprog, traits were included in the analysis when
significant for PRRS progression and/or interaction between GRG and PRRS progression.
Results Pig Performance and PRRS titers. Over the 56 dpi period, PRRS infected pigs grew from 16.1 kg BW to 55.6
kg BW (Figure 1). Within this population of PRRS infected gilts, after correcting for age at infection, BW at dpi
0 and pen, two GRG were selected that contained 12 high and 12 low growth rate gilts (Figure 1). These two
GRG were then used for multivariate blood analysis. Even though both groups were infected with PRRS, the
low growth group had a 26% reduction in ADG compared to the high growth gilts (0.57 vs. 0.78 kg/d,
respectively, P < 0.01). As expected the population (all 76 pigs) viremia and antibody responses to inoculation
significantly changed over time (Figure 2, P < 0.001). However, we reported no differences in PRRS viremia
responses and antibody titers between GRG (data not shown).
Blood Metabolites, Inflammatory and Immune Markers. Complete blood cell (CBC) counts were assessed over
the time course of the PRRS challenge as are shown in Table 1. Surprisingly, none of the 17 CBC traits
measured were significantly different between the GRG or for the interaction between GRG and dpi. However,
basophils, eosinophils, hemoglobin, MCH, monocytes, platelets and white blood cells significantly changed
over time (dpi P < 0.05). While there was a tendency for lymphocytes, MCV and red blood cells to change over
dpi (P < 0.10). Fitting this data to either quadratic or linear models suggested that most of these changes across
time where quadratic (P < 0.05, Table 1).
Blood immune and metabolite marker are presented in Table 2. Interestingly, the low GRG had higher
blood endotoxin concentrations (P = 0.06) and concentrations of the acute phase protein haptoglobin (P =
0.033), compared to the high GRG gilts. There was no significant group by dpi interactions reported (Table 2).
From dpi 7 to 28, endotoxin and haptoglobin concentrations decreased (P < 0.01). However, C-reactive protein,
tumor necrosis factor alpha, alkaline phosphatase and insulin concentrations all increased over dpi and in a
linear and quadratic manner (P < 0.05). The antimicrobial factor, lysozyme, was not effected by PRRS infection
(P > 0.05). Non-esterified free fatty acids and glucose insulin ratio tended to change in a quadratic manner,
with dpi 14 having lower concentrations (P < 0.07). The proxy protein catabolism marker of blood urea
nitrogen also tended to change across time in a quadratic manner with peak concentrations at dpi 14 (P = 0.06).
Serum 1H-NMR metabolic analysis identified 24 metabolites (Table 3). Blood concentrations of valine,
leucine, hypoxanthine and formate were significantly different between low and high growth rate groups. While
carnitine, N-acetyl-Asparate and threonine tended to be different between GRG (P < 0.10). Surprisingly, only 7
fasting blood metabolites were significant (histidine, hypoxanthine and urea) or tended (hydroxyproline, N-
acetyl-Asparate, proline and lysine) to be different across dpi (Table 3).
Further lipidomic analysis utilization a LC-MS platform identified 16 lipid metabolites in blood samples of
gilts infected with PRRS (Table 4). None of these metabolites were significantly different when examined by
dpi by group. However, retinoic acid was the only LC-MS metabolite that tended to differ between GRG (P =
0.08). Generally, prostaglandin D3 (P = 0.02), eicosapentaenoic acid (P = 0.08), Prostaglandin E2 and I2 (P =
0.02) all increased as dpi increased. 15-deoxy-d-12,14-prostaglandin J2 (P = 0.02) and 2,3-dimor-6-keto-
prostaglandin (P = 0.10) both decreased over time (Table 4).
Integrated Analysis
Discriminant analysis was used to identify those variables that better separate among dpi, and it helps to
7
identifies how many of these variables are required for the best classification possible. The squared canonical
correlations (R2), proportion of the variance, and significance levels of the two canonical variables (CAN) used
to discriminate PRRS progression are shown in Table 5. The first CAN (CAN1) was significant (P < 0.0001),
accounted for 75% of the total variance, and had a high R2 (95.04%), indicating great potential in discriminating
PRRS progression. The second CAN (CAN2) was also significant (P = 0.0224), with a high R2 (86.44%) and
accounted for the remaining of the total variance (25%). Therefore, both CAN have potential do discriminate
different stages of PRRS progression,
The canonical scores resulted from the discriminant functions CAN1 and CAN2 are depicted in Figure
3. Accounting for most of the variation and with a high R2, CAN1 (Table 5) discriminated infection stages (7
and 14 dpi) from end of infection (28 dpi). For CAN2, the canonical scores from dpi 28 were around 0,
indicating that this dpi is not being discriminated in this CAN. Interestingly, this CAN was able to discriminate
early infection stage (7 dpi) from later stage (dpi 14). The discrimination power for both CAN was high. In
CAN1, all canonical scores from dpi 7 and 14 had negative values, whereas all positive canonical scores were
observed for the samples from dpi 28. Although samples from dpi 28 were around 0 in CAN2, all negative
canonical scores were from dpi 7, and positive from dpi 14.
The correlations between each variable and each CAN, i.e. total canonical structure, are presented in
Table 6. These correlations can be used to see how closely a trait is related to each CAN. In CAN1, traits with
positive correlations indicate high phenotypic values on dpi 28, whereas those with negative correlation have
high phenotypic values on dpi 7 and 14, and hence, greater discriminant power. In this manner, the traits that
best characterized PRRS progression in CAN1 were: endotoxin, haptoglobin, eosinophil, insulin, and alanine
(Ala), with high values for dpi 7 and 14, and urea, C-reactive protein (CRP), TNF, and alkaline phosphatase
(ALP) for dpi 28. These traits indicate that on dpi 7 and 14 there is high innate immune (endotoxin,
haptoglobin, and eosinophil) and small metabolic (insulin and Ala) responses, whereas on dpi 28 we observed
high protein catabolism (urea) and adaptive immune response (CRP, TNF, and ALP). In CAN2, different traits
appeared with high correlation with this discriminant function. For instance, three amino acids (proline,
threonine, and Ala) had similar negative correlations with CAN2, indicating that there is protein mobilization
on dpi 7 for energy and immune response requirements. For dpi 14, high correlation values for glucose and
insulin suggest a shift in the metabolic state of animals growing gilts infected with PRRSV.
Discussion Altogether, our results indicate dynamic changes in immune and energy requirements for pigs growing
through a PRRS challenge. We showed that CAN1 discriminated dpi 7 and 14 from 28, CAN2 discriminated
dpi 7 from 14. The best discriminators in CAN1 were alkaline phosphatase (ALP) and haptoglobin. Alkaline
phosphatase is important in regulating endotoxin detoxification (Mani et al., 2012), pH and bicarbonate
secretion (Lalles, 2010). Serum concentrations of acute-phase proteins, such as Haptoglobin and C-reactive
protein, are known to increase after stress and inflammation (Pineiro et al., 2007; Pineiro et al., 2009) and the
magnitude of the response is generally related to the severity of the stress or disease (Hall et al., 1992; Murata et
al., 2004; Williams et al., 2009). On the other hand, the best discriminators in CAN2 were C-reactive protein
(CRP), glucose, and insulin. CAN1 and CAN2 were significant for PRRS infection and showed high values of
haptoglobin and low values of ALP. However, 7 dpi had low values of CRP, glucose and insulin compared to
dpi 14. Additionally, urea showed a potential discrimination power, with high values on dpi 7 & 14 compared to
28 (CAN1), whereas the amino acids alanine, proline, and threonine were good discriminators of dpi 7 and 14
(CAN2). This may suggest that during early stages of PRRS infection (dpi 7 and 14), amino acid mobilization is
increased (CAN2) for immune protein synthesis and energy requirements, whereas in later stages (dpi 28) we
observed increased protein catabolism (CAN1).
Both the innate and adaptive immune responses are integrated with cellular bioenergetics, metabolism,
and inflammation at a whole animal, tissue and cellular levels. However, we still have minimal knowledge on
how PRRSv and PEDv infection may regulate a pig’s metabolism. By understanding the metabolic
ramifications of viral infections, better mitigation strategies can be employed to reduce the severity of health
8
challenges to pig and business performance. Homeorhetic switches in bioenergy intimately links metabolism
with immune function and inflammation to restore homeostasis and protect cells. Profound metabolic
reprogramming occurs in which there is a heightened reliance on glycolysis and a down regulation of oxidative
phosphorylation (O'Neill, 2011; Tannahill et al., 2013). Specifically, Tannahill et al., (2013) reported that the
activation of Toll-like receptors in the innate immune response leads to a switch in metabolism from oxidative
phosphorylation to glycolytic. These authors have stated that glycolysis and glucose metabolism is preferred so
the cell can generate more ATP per unit time for immune and homeostatic needs, even though it is not as
efficient in terms of ATP generation as the process of oxidative phosphorylation in mitochondria. Meta-analysis
of PPRS pigs microarray studies partially support these notions in that mitochondrial dysfunction and oxidative
phosphorylation pathways, innate immune response, apoptosis and cell homeostasis pathways are differentially
regulated under PRRSV infection (Badaoui et al., 2013). Based on this meta-analysis and our observed
reduction in performance measures, innate and adaptive responses to intense, prolonged and poorly-contained
immunological stimuli are most likely responsible for the reduced growth rates observed. Furthermore, our data
clearly agrees with previous results that there is an integral link between metabolism and an immune response
in pigs.
Pathogenic and infection challenges causes a state of pseudo-starvation due to suppressed feed intake
and increase energetic demands. This is evident during the adaptation phase to infection and inflammation
which is more catabolic and requires fatty acid oxidation (Liu et al., 2012). Activation of AMP-activated
protein kinase (AMPK) would promote the switch to an anti-inflammatory phenotype and cause a switch away
from glycolysis towards mitochondrial oxidative phosphorylation and fatty acid oxidation (O'Neill and Hardie,
2013). Interestingly, AMPK-mediated pathways have been shown to be involved in the antiviral response to
PRRSv infection (Sang et al., 2014).
Immuno-metabolism, metabolic homeostasis and homeohresis is a relatively new frontier that focuses
on the integration and interaction of immune and metabolic systems. This work in livestock species is limited.
However, by understanding the metabolic pathways activated and the coordinated hormonal response (insulin,
glucagon etc…) to infection, metabolism targeted interventions and management strategies for poor health
livestock could be developed to mitigate production losses. Finally, although our n was small, there are clearly
differences in how pigs within a given population respond to a disease challenge. This may be associated with
genetics (Boddicker et al., 2014) and the ability of pigs to alter their metabolism. Better understanding of these
areas will allow us to enhance growth rates and feed efficiency of swine
References Badaoui, B., C. K. Tuggle, Z. Hu, J. M. Reecy, T. Ait-Ali, A. Anselmo, and S. Botti. 2013. Pig immune
response to general stimulus and to porcine reproductive and respiratory syndrome virus infection: a
meta-analysis approach. BMC genomics 14: 220.
Boddicker, N. J., A. Bjorkquist, R. R. Rowland, J. K. Lunney, J. M. Reecy, and J. C. Dekkers. 2014. Genome-
wide association and genomic prediction for host response to porcine reproductive and respiratory
syndrome virus infection. Genetics, selection, evolution : GSE 46: 18.
Broussard, S. R., R. H. McCusker, J. E. Novakofski, K. Strle, W. H. Shen, R. W. Johnson, G. G. Freund, R.
Dantzer, and K. W. Kelley. 2003. Cytokine-hormone interactions: tumor necrosis factor alpha impairs
biologic activity and downstream activation signals of the insulin-like growth factor I receptor in
myoblasts. Endocrinology 144: 2988-2996.
Che, T. M., M. Song, Y. Liu, R. W. Johnson, K. W. Kelley, W. G. Van Alstine, K. A. Dawson, and J. E.
Pettigrew. 2012. Mannan oligosaccharide increases serum concentrations of antibodies and
inflammatory mediators in weanling pigs experimentally infected with porcine reproductive and
9
respiratory syndrome virus. J Anim Sci.
Hall, W. F., T. E. Eurell, R. D. Hansen, and L. G. Herr. 1992. Serum haptoglobin concentration in swine
naturally or experimentally infected with Actinobacillus pleuropneumoniae. J Am Vet Med Assoc 201:
1730-1733.
Holtkamp, D. J., J. B. Kliebenstein, E. J. Neumann, J. J. Zimmerman, H. F. Rotto, T. K. Yoder, C. Wang, P. E.
Yeske, C. L. Mowrer, and C. A. Haley. 2013. Assessment of the economic impact of porcine
reproductive and respiratory syndrome virus on United States pork producers. Journal of Swine Health
and Production 21: 72-84.
Johnson, R. W. 1997. Inhibition of growth by pro-inflammatory cytokines: an integrated view. J. Anim. Sci. 75:
1244-1255.
Lalles, J. P. 2010. Intestinal alkaline phosphatase: multiple biological roles in maintenance of intestinal
homeostasis and modulation by diet. Nutrition reviews 68: 323-332.
Liu, T. F., C. M. Brown, M. El Gazzar, L. McPhail, P. Millet, A. Rao, V. T. Vachharajani, B. K. Yoza, and C.
E. McCall. 2012. Fueling the flame: bioenergy couples metabolism and inflammation. Journal of
leukocyte biology 92: 499-507.
Lunney, J. K., E. R. Fritz, J. M. Reecy, D. Kuhar, E. Prucnal, R. Molina, J. Christopher-Hennings, J.
Zimmerman, and R. R. Rowland. 2010. Interleukin-8, interleukin-1beta, and interferon-gamma levels
are linked to PRRS virus clearance. Viral Immunol 23: 127-134.
Mani, V., T. Weber, L. Baumgard, and N. Gabler. 2012. GROWTH AND DEVELOPMENT SYMPOSIUM:
Endotoxin, inflammation, and intestinal function in livestock. Journal of Animal Science 90: 1452-1465.
Murata, H., N. Shimada, and M. Yoshioka. 2004. Current research on acute phase proteins in veterinary
diagnosis: an overview. Vet J 168: 28-40.
O'Neill, L. A. 2011. A critical role for citrate metabolism in LPS signalling. The Biochemical journal 438: e5-6.
O'Neill, L. A., and D. G. Hardie. 2013. Metabolism of inflammation limited by AMPK and pseudo-starvation.
Nature 493: 346-355.
Pineiro, C., M. Pineiro, J. Morales, M. Andres, E. Lorenzo, M. D. Pozo, M. A. Alava, and F. Lampreave. 2009.
Pig-MAP and haptoglobin concentration reference values in swine from commercial farms. Vet J 179:
78-84.
Pineiro, M., C. Pineiro, R. Carpintero, J. Morales, F. M. Campbell, P. D. Eckersall, M. J. Toussaint, and F.
Lampreave. 2007. Characterisation of the pig acute phase protein response to road transport. Vet J 173:
669-674.
Sang, Y., W. Brichalli, R. R. Rowland, and F. Blecha. 2014. Genome-wide analysis of antiviral signature genes
in porcine macrophages at different activation statuses. PloS one 9: e87613.
Spurlock, M. E. 1997. Regulation of metabolism and growth during immune challenge: an overview of cytokine
function. J. Anim. Sci. 75: 1773-1783.
Tannahill, G. M. et al. 2013. Succinate is an inflammatory signal that induces IL-1beta through HIF-1alpha.
10
Nature 496: 238-242.
Williams, P. N., C. T. Collier, J. A. Carroll, T. H. Welsh, Jr., and J. C. Laurenz. 2009. Temporal pattern and
effect of sex on lipopolysaccharide-induced stress hormone and cytokine response in pigs. Domest Anim
Endocrinol 37: 139-147.
11
Figure 1. Growth performance of PRRS infected pigs.
Figure 2. PRRS viremia and antibody titers of infected pigs.
Viremia dpi P < 0.001
Antibody dpi P < 0.001
12
Table 1. Complete blood cell counts
Trait
Days Post Inoculation (dpi)
SEM
P-value
7 14 28 dpi Group dpi*Group dpi
Linear
dpi
Quadratic
Basophils 0.20 0.10 0.12 0.036 0.042 0.21 0.39 0.11 0.038
Eosinophils 0.26 0.38 0.18 0.048 0.018 0.18 0.45 0.066 0.027
Hemocrit 41.7 36.9 40.5 2.05 0.15 0.70 0.71 0.94 0.057
Hemoglobin 17.4 9.9 15.5 2.71 0.047 0.56 0.63 0.88 0.016
Lymphocytes 7.51 10.19 7.97 0.954 0.063 0.74 0.66 0.95 0.022
MCH 19.4 15.7 19.0 1.31 0.035 0.87 0.77 0.79 0.014
MCHC 25.8 28.3 37.7 1.03 0.19 0.74 0.48 0.30 0.11
MCV 47.7 58.4 49.7 4.26 0.052 0.27 0.34 0.98 0.019
Monocytes 0.95 1.67 1.29 0.227 0.024 0.54 0.91 0.44 0.009
MPV 9.7 10.0 8.9 0.57 0.30 0.46 0.15 0.21 0.40
Neutrophils 6.25 7.19 7.23 0.825 0.64 0.89 0.19 0.48 0.50
N:L ratio 7.0 0.8 8.0 3.36 0.25 0.48 0.65 0.65 0.13
Platelets 430 631 443 63.4 0.015 0.22 0.43 0.74 0.006
Red Blood Cells 7.73 6.55 7.27 0.455 0.083 0.90 0.53 0.68 0.030
RDC 142 30 99 48.6 0.12 0.30 0.40 0.841 0.044
White Blood Cells 14.7 19.9 15.3 1.45 0.022 0.15 0.81 0.80 0.007
Unidentified cells 0.43 0.30 0.38 0.079 0.033 0.97 0.82 0.83 0.14
13
Table 2. Blood immune and metabolite markers
Trait
Days Post Inoculation (dpi)
SEM
P-value
7 14 28 dpi Group dpi*Group dpi
Linear
dpi
Quadratic
Haptoglobin, ng/mL 259 138 111 20.2 0.0001 0.033 0.11 0.0001 0.0035
C-reactive protein, ng/mL 91 148 142 9.1 0.0001 0.23 0.44 0.0001 0.0001
Tumor necrosis-α, pg/mL 40 135 315 25.5 0.019 0.17 0.62 0.025 0.053
Blood urea nitrogen, mg/dL 29 36 27 3.8 0.11 0.98 0.86 0.42 0.060
Non esterified fatty acids, mM 3.76 0.19 4.23 1.86 0.15 0.55 0.59 0.58 0.066
Alkaline phosphatase, U/L 40 36 53 3.0 0.0001 0.71 0.84 0.0005 0.0027
Insulin, pM 5.4 13.4 8.7 1.94 0.0008 0.69 0.33 0.83 0.003
Endotoxin, AU 139 108 59 11.5 0.01 0.060 0.78 0.001 0.75
Lysozyme, U/mL 168 177 185 11.9 0.50 0.18 0.36 0.25 0.74
Glucose:Insulin 7.3 2.9 6.4 2.66 0.13 0.17 0.48 0.90 0.050
14
Table 3. Blood 1H-NMR data
Trait Days Post Inoculation (dpi)
SEM P-value
7 14 28 dpi Group dpi*Group dpi Linear dpi Quadratic
Acetate 2.86 1.60 1.98 0.657 0.37 0.28 0.36 0.48 0.22 Adenosine 0.07 0.10 0.09 0.019 0.28 0.42 0.75 0.33 0.22
Alanine 26.16 19.98 18.76 3.160 0.22 0.34 0.27 0.14 0.33
Allantoin 2.28 0.31 0.29 0.991 0.26 0.57 0.23 0.21 0.25 Asparagine 1.38 1.30 1.33 0.040 0.30 0.58 0.92 0.52 0.16
Betaine 5.78 5.42 6.23 0.375 0.23 0.44 0.84 0.23 0.22
Carnitine 4.98 4.89 5.15 0.368 0.75 0.083 0.71 0.59 0.24 CH2 Lipids 119 120 106 9.2 0.47 0.22 0.09 0.28 0.60
Choline 5.4 5.9 6.1 0.23 0.13 0.80 0.21 0.067 0.29
Cis-Aconitate 0.05 0.06 0.06 0.008 0.34 0.85 0.27 0.66 0.17 Citrate 3.75 2.99 3.00 0.386 0.29 0.12 0.44 0.24 0.26
Creatine 5.96 6.90 7.07 0.825 0.56 0.29 0.55 0.36 0.59
Creatinine 1.79 1.69 1.76 0.078 0.61 0.86 0.21 0.98 0.33 Formate 0.05 0.05 0.05 0.015 0.76 0.004 0.52 0.95 0.47
Glucose 15.2 13.0 14.1 1.18 0.39 0.84 0.73 0.65 0.20
Glutamine 11.2 11.1 10.8 0.65 0.86 0.20 0.62 0.59 0.96 Glycerol 7.6 7.2 7.0 0.54 0.72 0.44 0.48 0.46 0.71
Glycine 20.6 20.8 22.4 1.26 0.54 0.44 0.71 0.29 0.77
Histidine 0.98 1.14 1.21 0.077 0.034 0.88 0.97 0.016 0.29 Hydroxyproline 3.72 3.22 3.36 0.165 0.092 0.57 0.75 0.25 0.062
Hypoxanthine 0.19 0.24 0.25 0.018 0.033 0.017 0.26 0.020 0.20
Imidazole 0.45 0.40 0.39 0.033 0.23 0.57 0.32 0.13 0.39 Isobutyrate 0.65 0.82 0.76 0.060 0.14 0.62 0.76 0.27 0.10
Isoleucine 16.77 16.13 15.75 1.320 0.84 0.24 0.43 0.57 0.84
Lactate 44.4 50.1 50.7 3.84 0.40 0.31 0.39 0.28 0.40 Leucine 19.2 20.9 21.0 0.70 0.13 0.039 0.89 0.12 0.17
Lysine 1.7 1.9 1.9 0.06 0.088 0.44 0.36 0.18 0.072
Mannose 1.65 0.64 0.60 0.555 0.30 0.52 0.22 0.23 0.30 Methionine 1.95 1.83 1.90 0.081 0.44 0.73 0.81 0.75 0.21
Methyl-hexanoic acid 1.40 1.08 1.10 0.168 0.31 0.25 0.14 0.28 0.26
Myoinositol 1.63 1.32 1.43 0.147 0.31 0.62 0.27 0.48 0.17 N-aceytl-Asparate 3.58 3.87 3.76 0.098 0.06 0.06 0.001 0.35 0.026
N,N-dimethylglycine 0.57 0.51 0.54 0.030 0.31 0.11 0.029 0.69 0.15
Phenylalanine 1.61 1.77 1.74 0.087 0.27 0.24 0.94 0.27 0.25 Phosphatidylcholine 0.13 0.06 0.05 0.068 0.32 0.57 0.24 0.25 0.30
Proline 2.15 1.58 1.61 0.168 0.053 0.51 0.14 0.21 0.034
Propionic Acid 10.6 11.2 11.8 0.79 0.65 0.97 0.71 0.37 0.85 Pyruvate 2.04 1.36 1.41 0.313 0.24 0.27 0.25 0.23 0.22
Serine 3.76 3.84 4.03 0.141 0.39 0.75 0.61 0.17 0.97
Threonine 6.35 4.20 5.52 1.049 0.27 0.08 0.28 0.76 0.12 Tyrosine 1.28 1.40 1.41 0.099 0.54 0.58 0.96 0.38 0.48
Unsaturated Lipids 27.9 29.7 26.5 1.68 0.26 0.56 0.031 0.39 0.19
Urea 0.37 0.65 0.72 0.083 0.006 0.45 0.94 0.004 0.12 Valine 8.82 9.29 9.83 0.571 0.33 0.018 0.34 0.14 0.83
Xanthine 0.10 0.11 0.11 0.017 0.64 0.54 0.15 0.34 0.910
15
Table 4. Blood LC-MS Lipidomics data
Trait
Days Post Inoculation (dpi)
SEM
P-value
7 14 28 dpi Group dpi*Group dpi
Linear
dpi
Quadratic
Prostaglandin D3 31 32 41 5.8 0.02 1.00 0.97 0.11 0.69
15-deoxy-d-12,14-PGJ2 51 47 41 5.3 0.19 0.42 0.58 0.76 0.08
2,3-dinor-6-keto-prostaglandin 157 48 109 26.9 0.02 0.65 0.96 0.50 0.01
5,8,11-Eicosatrienoic acid 68 94 95 11.7 0.10 0.37 0.31 0.19 0.22
8,11,14,17-Eicosatraenoic acid 60 47 75 9.9 0.25 0.96 0.52 0.19 0.09
8,11,14-Eicosatrienoic acid 89 94 104 11.5 0.36 0.48 0.43 0.26 0.10
Eicosadienoic acid 48 59 70 8.9 0.10 0.87 0.48 0.11 0.72
Eicosapentaenoic acid 380 466 423 47.5 0.08 0.12 0.30 0.70 0.21
Prostaglandin E2 38 32 41 6.0 0.02 0.96 0.96 0.61 0.29
Prostaglandin I2 31 32 41 5.8 0.02 1.00 0.97 0.26 0.71
Retinoic acid 475 465 479 45.1 0.52 0.08 0.49 0.93 0.83
2,3-dinor-6-keto-prostaglandin 25 4 14 4.5 0.06 0.37 0.51 0.34 0.01
8-iso-15-keto-PGE2 72 76 96 10.6 0.13 0.67 0.19 0.28 0.96
Arachidonic Acid 367 466 420 48.5 0.06 0.13 0.31 0.64 0.17
Prostaglandin D3 72 76 96 10.6 0.13 0.67 0.19 0.11 0.69
Thromboxane A2 31 32 41 5.8 0.47 0.99 0.97 0.26 0.71
16
Table 5. Squared canonical correlation (R2), proportion of the variance (%), and significance of
the canonical variables (CAN)
CAN R2 % P-value
1 0.9504 0.7504 <0.0001
2 0.8644 0.2496 0.0224
Table 6. Total canonical structure by canonical variable (CAN) of PRRS infection.
CAN
Trait 1 2
alanine -0.18 -0.23
alkaline Phosphatase 0.51 -0.21
blood urea nitrogen -0.16 0.27
citrate -0.17 -0.22
C-reactive protein 0.33 0.54
eosinophil -0.26 0.22
endotoxin -0.58 -0.09
Glucose:Insulin 0.10 -0.28
glucose 0.07 0.40
haptaglobin -0.34 -0.28
hematocrit 0.11 -0.27
hemoglobin 0.09 -0.34
hydroxyl-proline 0.01 -0.21
insulin -0.24 0.44
lymphocytes -0.13 0.27
mean corpuscular hemoglobin 0.18 -0.33
mean corpuscular hemoglobin concentration 0.02 0.20
mean corpuscular volume -0.14 0.32
methylhexanoic Acid -0.12 -0.21
monocytes -0.02 0.40
mean platelet volume -0.31 -0.13
N-aceytlAsparate 0.06 0.25
non esterified fatty acids 0.14 -0.26
neutrophil 0.19 0.07
platelet -0.12 0.40
proline -0.08 -0.31
red blood cells 0.07 -0.31
and red blood cell distribution width 0.13 -0.29
threonine -0.07 -0.24
Tumor necrosis factor 0.23 0.33
unsaturated lipid -0.18 0.13
urea 0.34 0.09
white blood cells -0.10 0.34
17
Figure 3. Graphical representation of canonical analysis of individuals of PRRS infection at 7,
14 and 21 dpi.