Thyroid hormone suppression in feeder pigs following polymicrobial or porcine reproductive and respiratory syndrome virus-2 challenge

Abstract

Thyroid hormones are powerful regulators of growth, development, and basal metabolic rate and can be dysregulated under conditions of severe stress or illness. To understand the role of these hormones in porcine disease response, serum samples were obtained from three batches of nursery-aged pigs (n = 208) exposed to a natural polymicrobial disease challenge with an array of bacterial and viral pathogens. Levels of total thyroxin (T4) and triiodothyronine (T3) assessed in sera by radioimmunoassay, decreased significantly by 14 days post-exposure (DPE). Levels of T3 partially rebounded by 48 DPE, while T4 levels remain depressed. Post-exposure T3 and T4 levels were positively correlated with acute and long-term average daily gain (ADG). Cross-sectional sampling of animals maintained at the high health source farms, showed no equivalent change in either hormone when managed under standard industrial conditions. To further elucidate the effect of porcine reproductive and respiratory syndrome virus (PRRSV)-infection on thyroid hormone levels, archived sera over 42 days post inoculation (DPI) from nursery pigs (N = 190) challenged with one of two PRRSV2 strains by the PRRS Host Genetics Consortium were similarly assessed, with animals selected in a two-by-two design, to investigate biological extremes in ADG and viral load (VL). All animals showed a similar decrease in both thyroid hormones reaching a minimum at 7 DPI and returning to near pre-challenge levels by 42 DPI. Post-challenge T3 and T4 levels were significantly greater in high ADG groups, with no significant association with VL or strain. The results of this study demonstrate porcine susceptibility to thyroid disruption in response to disease challenge and demonstrate a relationship between this response and growth performance.

Introduction

Across all sectors of animal agriculture, infectious disease remains one of the primary determinants of profitability, and health outcomes are increasingly recognized as a critical component of animal welfare. The intensive nature of the modern swine industry makes it particularly vulnerable to disease, with a wide array of endemic and emerging pathogens having had disastrous effects on production. A prime example is porcine reproductive and respiratory syndrome virus (PRRSV), which has previously been estimated to cost some $664M/yr to the United States industry alone, with roughly 55% of this cost incurred in the grow finish phase (Holtkamp et al., 2013). Within the swine industry, the increased cost of production associated with disease results from the cost of treatment, as well as decreased average daily gain (ADG) and average daily feed intake (Pastorelli et al., 2012) which collectively result in an increased weaning to slaughter interval. Under standard industry conditions, the associated cost also scales with the severity of disease and increases substantially under the pressure of multiple pathogens (Cornelison et al., 2018). These changes in production benchmarks can, at least in part, be linked to the substantial energetics cost of mounting an effective immune response (Lochmiller and Deerenberg, 2000). In addition, a whole host of immune insults have been shown to have a negative impact on feed consumption (Pastorelli et al., 2012; Helm et al., 2019). However, the role of endocrinology in the porcine response to disease remains poorly characterized. An improved understanding of the physiological mechanisms that link morbidity to performance phenotypes, such as growth rate, feed intake, and carcass quality, may provide the insights necessary to help identify genetic markers of disease resilience or to develop management or treatment strategies capable of augmenting this high value trait.

A trait as complex as disease resilience is most likely the product of simultaneous variation in numerous physiological systems, and such a widespread response is likely coordinated by at least one aspect of the endocrine system. Thyroid hormones are widely recognized for their primary function of regulating basal metabolic rate (Mullur et al., 2014), but they possess additional functionality with regards to controlling growth, immune responses (De Luca et al., 2020), body composition (Scow, 1959), and appetite (Amin et al., 2011). These hormones are initially produced via quadruple iodination and subsequent cleavage of tyrosine residues from a thyroglobulin precursor within the thyroid gland. The resulting primary hormone, thyroxin (T4), can then be acted on by deiodinase enzymes to produce the derivative triiodothyronine (T3), or other metabolites such as reverse T3 (rT3) and diiodothyronines (T2). Although some of this deiodination occurs within the thyroid gland, the majority of the hormone is released in the form of T4, which exhibits relatively low bioactivity, and which is largely converted to the more bioactive T3 within peripheral target tissues. Production and release of thyroid hormone from the thyroid gland is governed by thyroid-stimulating hormone (TSH) that is canonically produced in the pituitary in response to thyrotropin-releasing hormone released from the hypothalamus, in what is termed the hypothalamic–pituitary–thyroid axis (HPT). Negative feedback within the HPT typically results in homeostatic regulation, however, in response to severe and/or chronic illness the physiological set points in the HPT can be altered (Chatzitomaris et al., 2017). The resulting suppression in circulating thyroid hormones has been previously referred to as euthyroid sick syndrome and more recently as non-thyroidal illness syndrome (NTIS). The endocrinological hallmarks of this condition depend on both the animal species and the type or scale of the physiological insult, however, the most persistent symptom is a large scale decreases in total circulating T3 and T4, often with limited alteration in the levels of TSH (Warner and Beckett, 2010).

In the pig, the basic biology and concentration of thyroid hormones during in utero development has been well described (Krysin, 1995; Brzezińska-Slebodzińska and Slebodziński, 2004). In addition, postnatal alterations in the HPT axis in response to changes in environmental conditions (Pathak et al., 2018), feed intake (Buonomo and Baile, 1991), diet (Carroll et al., 1998), and anti-nutritional factors present in feed stuff (Spiegel et al., 1993) have been reported. Further, disruptions in the porcine thyroid systems, consistent with NTIS, have also been reported following immune stimulation with intravenous lipopolysaccharide (Castro et al., 2013). However, the impact of an infectious disease state on the function of the HPT axis has only recently been described, when we demonstrated that both dam and fetus experience significant hypothyroidism following infection with PRRSV2 (Pasternak et al., 2020a). The endocrinological and immunological adaptations required during both pregnancy and fetal development may have created a unique physiological environment making these animals uniquely susceptible to PRRSV-induced disruption of thyroid homeostasis. Therefore, to further expand on this novel finding, the present study was undertaken to determine if a similar decrease in the circulating thyroid hormones occurs in feeder pigs following: 1) polymicrobial infection resulting from natural exposure to an assortment of viral and bacterial pathogens and 2) a standardized experimental inoculation with PRRSV2. We further compared the temporal dynamics observed in each of these pathogen challenge models with that of healthy animals of similar genetic backgrounds. Finally, we evaluated the relationship between the degree of suppression in circulating thyroid hormone levels and growth performance metrics.

Materials and Methods

Experiment 1a: natural polymicrobial disease challenge

The natural disease challenge model (NDCM) in Quebec, Canada, is operated by the Centre de développement du porc du Québec (CDPQ) in accordance with the guidelines of the Canadian Council on Animal Care. The experimental protocol was approved by the Protection committee of the Centre de recherche en sciences animales de Deschambault (15PO283) and the Animal Care and Use Committee at the University of Alberta (AUP 00002227). Establishment and management of the NDCM has been previously described (Putz et al., 2019; Harding et al., 2020). In short, CDPQ in conjunction with PigGen Canada and researchers from the Universities of Alberta, Saskatchewan and Iowa State, strategically seeded the NDCM in late 2015 with pathogens of critical importance in the North American swine industry. A continuous flow of animals was then maintained through late 2019 by entering a batch of 60 to 75 high health Landrace × Yorkshire cross nursery barrows every 3 wk. Detailed performance phenotypes were collected during the pathogen challenge period. Since inception, routine serology and PCR monitoring has demonstrated the presence of viral and bacterial pathogens including at least three strains of PRRSV2 as well as swine influenza A (IAV-S) and Mycoplasma hyopneumoniae. In addition, necropsies and other target surveillance detected other common swine pathogens including porcine circovirus type 2 (PCV2), hemagglutinating encephalomyelitis virus, pathogenic strains of Escherichia coli and Salmonella spp., Lawsonia intracellularis, Streptococcus suis, Glaesserella parasuis, rotavirus, Mycoplasma hyorrhinis, Brachyspira hampsonii, Ascaris suum, and Cystoisospora suis. While the level of challenge varies over time, exposure to PRRSV2 and M. hyopneumoniae was consistent across all batches and different intervention strategies were employed to maintain treatment and mortality rates at appropriate levels to balance animal welfare with the research objectives (Harding et al., 2020). For the purposes of the present study, sera were collected from three distinct batches of animals that entered the challenge unit between March and June 2018 (Table 1). Pigs in each of these batches were single sourced at weaning ($\overline{x}$ = 21.7 d, range 19 to 29) from multiplier farms for three unique genetic suppliers and placed in a quarantine unit 1 km from the NDCM to acclimate for 19 d during which PRRSV negative status was confirmed by RT-PCR. Upon arrival, all pigs were vaccinated against PCV2 with a single dose of Circovac (Boehringer Ingelheim) as per label instructions. From nursery to finish, all animals were fed standard commercial diets formulated to meet or exceed the requirements. Blood samples (5 to 10 mL) were collected from each animal in serum separator tubes at −14 and 0 d during the quarantine period prior to challenge, and at 14 and 42 days post pathogen exposure (DPE) and centrifuged before overnight transport to the University of Saskatchewan to be aliquoted and frozen for subsequent analysis. Individual bodyweights were measured on −21, 0, 28, 49, 70, 91, and 112 DPE and used to calculate ADG in incremental periods (e.g., −21 to 0, 0 to 28, 28 to 49, etc.) and for the complete post weaning growth period (−14 to 119 DPE). Infection status for each pathogen on individual animals within the NDCB is not routinely collected. Therefore, as an indicator of disease severity, treatments, defined as instances of parenteral injection of antibiotic ± anti-inflammatory in response to clinical signs suggestive of a disease process in accordance with predetermined protocol developed by the herd health veterinarian, were tracked throughout the trial period. Treatment rates were then determined as the percentage of animals receiving one or more such injections within the 42 d period covering the present study. Finally, antemortem loin depth and back fat were measured ultrasonically between the 3rd and 4th last rib at 112 DPE using a Vetko Plus ultrasound and 3.5 MHz linear probe (DFG Equipment Specialists, QC).

Table 1.

Experimental groups and characteristics for each experiment

Trial	Batch or farm	Challenge	Breed cross1	N
Experiment 1a: NDCM	41	Polymicrobial2	LW/LR	68
	43		LW/LR	63
	45		LW/LR	68
Experiment 1b: Healthy	A	None	LW/LR	45
	B		LW/LR	50
	C		LW/LR	60
Experiment 2: PHGC	10	PRRSV2KS-2006-72109	Pietrain × LW	37
	13		Duroc × LW/LR	22
	14		Duroc × LR/LW	24
	5	PRRSV2NVSL 97-7895	Duroc × LR/LW	34
	8		Duroc × LW/LR	37
	15		Pietrain × LW	36

¹Large White (LW) and Landrace (LR).

²Natural exposure to PRRSV2, PCV2, IAV-S, M. hyopneumoniae, E. coli, L. intracellularis, S. suis, and G. parasuis and others.

Experiment 1b: healthy animals

Due to the nature of the facility and the experimental design, the NDCM does not include unchallenged control animals. Therefore, to characterize the long-term temporal patterns in serum thyroid hormone abundance in equivalent healthy feeder pigs, a separate set of samples was collected from the three high health status Canadian multiplier farms (Table 1) that supplied animals for each of the NDCM batches assessed in experiment 1a. Source farm sampling of animals occurred from February to April of 2018, nearly simultaneous with entry of corresponding animals into the NDCB from March to May 2019. These animals were maintained under standard commercial conditions, as defined by the Canadian Code of Practice for swine production, and fed standard commercial diets designed to meet their requirements. On-farm blood sampling from a random assortment of gilts and barrows was approved by the University of Saskatchewan’s Animal Research Ethics Board (AUP 20180006). Blood samples were collected in serum separator tubes and shipped overnight to Queen’s University, where the sera were aliquoted and frozen for later transport to the University of Saskatchewan. For biosecurity reasons, sampling on each farm was conducted by the herd health veterinarians. For logistical reasons, a cross-sectional profile was performed, thus, serum was collected from unique sets of 10 to 12 animals per farm at each time point (4, 8, 12, 16, and 20 wk of age) over 1 or 2 visits.

Experiment 2: PRRS experimental challenge

To evaluate the impact of PRRSV2 alone on thyroid hormone levels, we drew on archived samples from the challenge experiments undertaken by the PRRS Host Genetics Consortium (PHGC). The collection of challenge experiments and associated protocols for the PHGC trials were approved by the Kansas State University Institutional Animal Care and Use Committee (IACUC 2647 for PHGC 5 to 8; IACUC 3000 for PHGC 9 to 15). The specifics of this challenge model along with an accumulation of phenotypic and genetic findings have been well described previously (Lunney et al., 2011; Hess et al., 2016; Dekkers et al., 2017). In short, groups of ~200 weaned pigs were transported from high health commercial to the animal biocontainment facility at Kansas State University. Source farms were free of PRRSV, IAV-S, and M. hyopneumoniae, and where possibly either free of PCV2 or protected by colostral antibodies following maternal vaccination (Lunney et al., 2011). Throughout the study, animals were provided with ad libitum access to feed and water, with nonmedicated diets formulated to meet their age-appropriate requirements. After 7 d acclimation, the pigs were challenged with 10⁵ TCID₅₀ PRRSV2 by a combination of intranasal and intramuscular inoculation. Serum samples were collected from each animal on 0, 4, 7, 11, 14, 21, 28, 35, and 42 days post inoculation (DPI) and used to quantify viral load (VL). Post-challenge growth rate was evaluated with individual body weights measured on a weekly schedule. For the purposes of the present study, archived serum samples were selected from six different batches of animals of which three were inoculated with NVSL 97-7895 (NVSL) and three with KS-2006-72109 (KSO6; Hess et al., 2016). Initially, a group of 190 animals from 6 batches (Table 1) were identified based on 2 factor selection for biological extremes of resistance and resilience as determined by within-group Z-scores for VL area under the curve (AUC) for 0 to 21 DPI (Boddicker et al., 2012) and weight gain (WG) from 0 to 42 DPI, respectively. This selection procedure resulted in subpopulations of animals exhibiting high and low WG and VL (Figure 1). In total, 87 animals selected exhibited high WG of which 62 and 25 were classified as low and high VL, respectively. Similarly, a total of 103 animals were selected for low WG, of which 29 and 74 were classified as low and high VL, respectively. Thyroid hormone levels in this primary group of animals were assessed on 0, 4, 7, and 11 DPI. Additional samples collected at 14, 21, 28, and 42 DPI from a further subset of these animals (N = 49), drawn from one batch (#10 and #15) for each viral strain were assessed to further investigate the recovery/rebound/persistence phase of viremia.

Figure 1.

Two factor selection of pigs from a total of N = 932 animals across 6 PGHC batches used in experiment 2, of which 3 were challenged with the NVSL and 3 with the KSO6 PRRSV2 strains. A total subset of N = 190 pigs were selected based on within batch Z-scores (|Z| > 0.75) for weight gain from 0 to 42 DPI and area under the curve (AUC) for viral load from 0 to 21 DPI. The resulting population included N = 87 animals classified as high weight gain (orange and purple) and N = 103 low weight gain (red and light blue). In addition, N = 91 were classified a low viral load (orange and teal) and N = 99 as high viral load (purple and red). DPI, days post inoculation.

Serum total T3 and T4

Concentrations of total triiodothyronine (T3) and thyroxine (T4) in serum were determined using analyte specific commercial radioimmunoassay (RIAs; MP Biomedical, Santa Ana, CA), as previously used porcine samples (Pasternak et al., 2020a). A total of 1,875 samples from a total of 544 unique animals across the three experiments (1a, 1b, and 2) were tested in duplicate (Table 1). Inter-assay variability across all three experiments was 8.01% and 9.50% for T3 and T4, respectively, when calculated using three unique pools of pig serum previously shown to span the measurable range of both assays. Intra-assay variability across all samples was 4.95% and 4.96% for T3 and T4, respectively. For each analyte and AUC was calculated using the trapezoidal method.

Statistical analyses

Data processing and analyses were conducted in R 3.5.0 (R Core Team, 2019) using the nlme package (Pinheiro et al., 2019). Temporal differences of thyroid hormone concentration in pigs following natural disease challenge (experiment 1a) and experimental PRRSV challenge (experiment 2) were assessed utilizing a mixed effects linear regression model accounting for animal as a repeat measure and included batch/trial as a random effect. To assess the cross-sectional time course in unchallenged, healthy animals (experiment 1b) a similar model was used which included source farm as a random effect. Within each of the three time course models, pairwise comparisons were made using the emmeans (Lenth, 2019) package with the Šidák correction for multiple testing. For the natural disease challenge dataset (experiment 1a), to evaluate the relationship between thyroid hormone levels and animal growth performance phenotypes as dependent variables, a post infection day (0 to 42 DPE) was used as a fixed effect and batch as a random effect. For the PHGC data set (experiment 2), the impact of PRRSV2 viral strain (NVSL vs. KS06) and pre-selected phenotypic resilience groupings (Figure 1) based on WG and VL were made relative to thyroid hormone AUC from 0 to 11 DPI which included the full cohort of animals tested and included batch as a random effect. The concentration of both thyroid hormones are presented in the form of emmean (lsmean) ± SD from the corresponding statistical model and in nmol/L. Graphical presentations of the data were generated with ggplot2 package (Wickham, 2016) using boxplots overlaid with individual data points and with observed statistical differences (P < 0.05), where present, marked with unique superscripts.

Results

Experiment 1a

Temporal response in thyroid hormone levels following natural polymicrobial disease challenge

We initially evaluated the impact of substantial disease pressure on thyroid hormone levels in postnatal pig serum from three batches of animals from the polymicrobial environment of the NDCM (Figure 2). In the three batches of pigs used in this study, total mortality rates across the grow finish period of 13.3%, 31.5%, and 16.2% were observed, with 55% of this mortality on average occurring by 42 DPE across the three batches. Treatment rates, defined as the proportion of animals requiring at least one parenteral pharmaceutical administration, among these three batches of animals reached 64%, 100%, and 70% by 42 DPI, respectively, with rationale for said treatments including respiratory symptoms (70%), diarrhea (10%), lameness (6%), and lack of thriftiness (5%). As a further indicator of pathogen load, commercial testing of 10 animals from each of the three batches indicated the presence of serum antibody for PRRSV in 80% to 100% of the population by 42 DPE, and M. hyopneumoniae in 90% of the population by 126 DPE. During the initial 14-d acclimation period following weaning and transport to the quarantine barn, concentrations of T3 increased significantly (P < 0.001) from 1.23 ± 0.06 at −14 DPE (5 d following transportation) to 1.98 ± 0.06 nmol/L at 0 DPE, immediately prior to pathogen exposure. During this acclimation period, concentrations of T4 also increased but to a lesser extent from 69.2 ± 1.1 to 74.0 ± 1.1 nmol/L (P = 0.002). At 14 d following pathogen exposure, concentrations of circulating T3 and T4 had decreased to 0.54 ± 0.06 (P < 0.001) and 38.3 ± 1.1 nmol/L (P < 0.001), roughly 70% and 50% of pre-challenge levels, respectively. Concentrations of T3 rebounded significantly by 42 DPE, reaching 1.22 ± 0.06 nmol/L (P < 0.001), which was equivalent to concentrations at −14 DPE but remained significantly depressed relative to concentrations at 0 DPE (P < 0.001). In contrast, no significant rebound in T4 concentration was observed (P = 0.117), with mean serum concentration at 42 DPE (41.5 ± 1.1 nmol/L) similar to 14 DPE levels.

Figure 2.

Concentrations of (A) total triiodothyronine (T3) and (B) thyroxine (T4) in sera from animals (N = 199) in experiment 1a, relative to the days post-exposure to polymicrobial challenge in a natural disease challenge model. Unique superscripts denote statistical differences (P < 0.05) over time.

Association of thyroid hormone response in the NDCM with disease resilience

As thyroid hormones are involved in regulating basal metabolic rate, energy balance, appetite and more specifically muscle deposition, we next evaluated the relationship between circulating thyroid hormone concentrations, expressed as AUC for the 42 d following exposure in the NDCM, and ADG, using existing phenotypic data on growth rate and body composition (Table 2). When ADG was calculated in incremental windows, significant relationships between ADG and thyroid hormone concentrations were detected from pre-exposure (−21 DPE) to 70 DPE for T3, and to 49 DPE for T4 (Table 2). For the entire post weaning growth period (−21 to 112 DPE), ADG was significantly related to both T3 (P = 0.001) and T4 (P = 0.030), with slope estimates of 1.911 and 0.0376, respectively. Further evaluation of the relationship between thyroid hormone concentrations and body composition, in the form of back fat and loin depth as measured by ultrasound at 112 DPE, identified only the latter as being significantly related to T3 levels (P = 0.0104; Table 2).

Table 2.

Relationship between thyroid hormones and growth in experiment 1a

			Serum T3 (AUC)1		Serum T4 (AUC)1
	Performance metric	Mean (SE)	Estimate2 (SE)	P-value	Estimate2 (SE)	P-value
ADG3, g	−21 to 0 DPE	239.0 (5.3)	1.005 (0.324)	0.002	0.0254 (0.0096)	0.008
	0 to 28 DPE	259.4 (8.1)	1.390 (0.670)	0.040	0.0513 (0.0191)	0.008
	28 to 49 DPE	595.2 (13.6)	4.237 (1.190)	0.001	0.0789 (0.0347)	0.024
	49 to 70 DPE	809.4 (14.4)	2.999 (1.292)	0.021	0.0663 (0.0366)	0.072
	70 to 91 DPE	917.8 (16.9)	2.573 (1.544)	0.098	0.0021 (0.0446)	0.962
	91 to 112 DPE	944.8 (16.5)	0.211 (1.426)	0.883	−0.0028 (0.0418)	0.946
	−21 to 112 DPE	616.2 (6.4)	1.911 (0.571)	0.001	0.0376 (0.0172)	0.030
Loin muscle depth, mm		52.7 (0.4)	0.093 (0.036)	0.010	0.0012 (0.0011)	0.266
Fat depth, mm		11.9 (0.2)	0.003 (0.015)	0.842	0.0001 (0.0004)	0.809

¹AUC, area under the curve in (nmol/Lּ days) for serum total triiodothyronine (T3) and thyroxine (T4) and metrics of animal performance.

²Statistical estimates and standard error from each linear model, bold text highlights significant relationships (P <0.05).

³ADG, average daily gain calculated for incremental windows and total growth period.

Experiment 1b

Temporal changes in thyroid hormone in feeder pigs under standard commercial conditions

To better understand the temporal dynamics in thyroid hormone levels during the nursery and feeder periods in the absence of such severe disease challenge, blood samples were collected at intervals from 4 to 20 wk of age (Figure 3) from pigs on high health farms free of common major pathogens including, PRRSV and M. hyopneumoniae. Samples were collected from the same three high health farms which contributed animals to experiment 1a with the initial sampling at 4 and 8 wk of age, spanning the equivalent exposure period. Serum T3 concentrations in these objectively healthy pigs were stable, averaging 1.71 ± 1.9 and 1.64 ± 0.186 nmol/L at 4 and 8 wk, respectively. Serum T4 was similarly stable during this time, averaging 57.3 ± 3.84 nmol/L at 4 wk and 52.4 ± 3.96 nmol/L at 8 wk. No significant difference in T3 concentration (P = 0.177) was detected over the entire time course (Figure 3). While serum T4 levels demonstrated similar temporal stability, a steady numerical, and likely age-associated increase from 8 wk onward was observed, with a significant increase (P = 0.015) detected between 8 and 20 wk of age (Figure 3).

Figure 3.

Concentrations of (A) total triiodothyronine (T3) and (B) thyroxine (T4) in sera from healthy animals (N = 155) in experiment 1b, sampled in three farms supplying animals to the natural disease challenge model. Unique superscripts denote statistical differences (P < 0.05) over time.

Experiment 2

Temporal response in thyroid hormone concentrations following PRRSV2 challenge

To determine what proportion of the dysregulation observed in the polymicrobial natural challenge pigs might be associated with PRRSV2, we next evaluated the concentrations of thyroid hormone in serum following controlled PRRSV2 infection (Figure 4). Serum T3 concentrations decreased significantly (P < 0.001) by 4 DPI (0.67 ± 0.05 nmol/L) relative to their pre-inoculation level (1.17 ± 0.05 nmol/L) and continued to decrease significantly (P < 0.001) to concentrations less than 50% of baseline on 7 (0.512 ± 0.05 nmol/L) and 11 DPI (0.561 ± 0.05 nmol/L). Concentrations subsequently rebounded, increasing significantly (P < 0.001) from 11 to 21 DPI (0.776 ± 0.05 nmol/L) after which they reached a plateau at 28 to 42 DPI (0.947 ± 0.05 nmol/L) at a level that was still significantly below preinfection concentrations (P < 0.001). Serum T4 concentrations initially showed a similar pattern, decreasing significantly (P < 0.001) from 0 DPI (71.1 ± 2.54 nmol/L) to their lowest level at 7 DPI (40.9 ± 2.53 nmol/L) before rebounding significantly (P < 0.001) to preinfection equivalent levels by 14 DPI (66.9 ± 2.54 nmol/L). However, unlike T3, concentrations of T4 showed a second wave of suppression decreasing significantly to a minimal value at 28 DPI (48.2 ± 2.54 nmol/L) before rising again by 42 DPI (60.1 ± 2.54 nmol/L) but remaining significantly lower than pre-challenge values (P < 0.001).

Figure 4.

Concentrations of (A) total triiodothyronine (T3) and (B) thyroxine (T4) in sera, relative to the days post inoculation (DPI) in six PRRS Host Genetics Consortium trials used in experiment 2. Samples collected between 0 and 11 DPI from N = 190 (Red and Blue) animals were initially assayed, with a further subset of N = 49 (Blue) were assayed for samples collected though to 42 DPI. Unique superscripts denote statistical differences (P < 0.05) over time.

Thyroid hormone response and PRRSV2 challenge phenotypes

To put clinical context on these results, we next assessed the thyroid hormone levels following PRRSV2-specific challenge relative to the phenotypic groups (High/Low) for VL (AUC 0 to 21 DPI) and WG (0 to 42 DPI; Figure 1), pre-selected based on within-group Z-scores (Figure 5A and D). Neither T3 (P = 0.1409) nor T4 (P = 0.2355) differed between high and low VL animals. By contrast, high WG (resilient) animals had higher concentrations of both T3 (P = 0.005) and T4 (P = 0.006) relative to their more susceptible low WG counterparts. Furthermore, no significant interaction between VL and WG categories was identified suggesting the effect on WG is independent of the degree of viral replication. Finally, we assessed the impact of viral strain (KSO6 and NVSL) and identified no significant differences in T3 (P = 0.725) or T4 (P = 0.393), likely owing to significant variation in the response within virus strain.

Figure 5.

Relationship between post infection area under the curve (AUC 0 to 11 DPI) for (A–C) total triiodothyronine (T3) and (D–F) thyroxine (T4) viral load (AUC 0 to 21 DPI) (A and D), weight gain (0 to 42 DPI) (B and E), and viral strain (C and F), using raw data from N = 190 animals in experiment 2. Unique superscripts denote statistical differences (P < 0.05) as determined using multivariate statistical model. DPI, days post inoculation.

Discussion

The presence of infectious disease in swine typically results in a combination of undesirable effects on ADG, feed conversion rate and carcass quality while simultaneously increasing variability and extending the weaning to slaughter interval (Cornelison et al., 2018). Understandably, depression in any of these metrics, either alone or in combination, can result in substantial economic losses at the farm level. Progress in this area is dependent on our understanding the nature of the host–pathogen interactions and the underlying physiological factors which mediate host response. Historically, much of the interest in this area has focused on the host immune response following infection and its relationship to pathology and pathogen load (Ladinig et al., 2014; Hess et al., 2016; Walker et al., 2018; Pasternak et al., 2020c). Pathogenic infection of pigs is also known cause a significant decrease in feed intake, however, pair feeding models have demonstrated that hypophagia alone cannot entirely explain the reduced growth or metabolic changes observed following challenge (Helm et al., 2019). Although a portion of the residual decrease in performance following infection can obviously be attributed to the energetic cost of mounting a robust immune response (Maciolek et al., 2014), no direct mechanistic or physiological linkages between infection and key performance metrics such as ADG or FCR have yet been identified. As a result, the well-established residual variation in pathogen resilience that exists within the swine herd remains unexplained. Herein, we investigated the potential role of the endocrine system in conferring resilience (and/or tolerance), and more specifically, alterations in the thyroid hormone pathway following disease challenge.

In response to natural polymicrobial and experimental PRRSV2 challenge, we observed significant decreases in circulating T3 and T4 in feeder pigs. The concentrations of T3 and T4 were numerically equivalent between animals in the NDCM prior to exposure and healthy animals assayed at the source farms at an equivalent age. Cross-sectional data from these healthy populations demonstrated no significant change with age in circulating T3 and a steady increase in T4 concentrations between 8 and 20 wk of age, consistent with the previous observations (Chapel et al., 2017) and in contrast to animals in the NDCM. Thus, the apparent temporal stability observed within this healthy population confirms that the changes following disease challenge cannot be attributed to either animal age or standard husbandry practices. It is worth noting that concentrations of thyroid hormone observed prior to exposure and those within the healthy populations were found to be substantially higher than those of the control animals in much older experimental reports (Spiegel et al., 1993; Schöne et al., 1997), and more consistent with contemporary reports (Chapel et al., 2017) of experiments utilizing similar high performance animals. Given the fundamental role of these hormones in regulating metabolic rate and fat mobilization, this apparent increase in resting thyroid hormone concentrations in commercial pigs over time is likely the result of the substantial selection pressure applied to increase growth rate and lean deposition.

Substantial variation in both hormones was identified within all three experimental populations used in the present study regardless of challenge status or age. This variability may result from environmental factors such as temperature, humidity, and feed availability, all of which have previously been shown to alter thyroid hormone levels in swine (Macari et al., 1986; Buonomo and Baile, 1991). Alternatively, a portion of this variation may be genetically determined, as each farm sampled for the healthy animal data set (experiment 1b), as well as each batch used in the challenge trials (experiments 1a and 2), were of unique genetic backgrounds tied to their source breeding companies. This is consistent with studies conducted in humans that found low intra-individual compared to inter-individual serum thyroid hormone variation (Hansen et al., 2004), suggesting a genetic component in determining an individual set point within the HPT axis. The large variability in thyroid hormone concentrations observed in all three of the present studies, both pre and post disease challenge, suggest there is opportunity to utilize genetic selection to increase the this set point within commercial swine populations. Given the role of these hormones in regulating metabolism, and the observed association with growth rate under challenge conditions demonstrated herein, selection for higher baseline levels in the absence of disease may be beneficial to improve disease resilience.

Both T3 and T4 decreased acutely and profoundly following PRRSV2 challenge, with both reaching a nadir between 7 and 10 DPI. An increase in pro-inflammatory cytokines such as IL1β, TNFα, and IL6 are thought to play a role in suppression of the thyroid hormone system in instances of human NTIS (Wajner and Maia, 2012). Following PRRSV2 infection in the present study, the decrease in thyroid hormones occurs prior to the peak abundance of pro-inflammatory cytokines such as IL-8 and IL-1β which do not reach maximum levels in serum until 14 DPI (Lunney et al., 2010). While some less virulent strains of PRRSV2 induce small amounts of INFα by 2 DPI, peak levels of TNFα are not typically observed until 11 DPI (Renson et al., 2017). By contrast, the depression in thyroid hormone concentrations observed in the present study corresponds well with the temporal pattern in viremia observed for NVSL and KSO6 which reach peak levels at 7 and 9.5 DPI, respectively (Hess et al., 2016). Although concentrations of T3 and T4 in the PRRSV-infected PHGC pigs eventually return to near-baseline levels by 42 DPI, T4 shows a biphasic temporal pattern with a second depression observed after 28 DPI. This latter result may be consistent with one of three possible viremia profiles previously identified for PHGC pigs, referred to as rebound, where similar to the observed thyroid hormone levels, viremia shows a biphasic pattern within the first 42 DPI for a subset of PHGC pigs (Islam et al., 2013). However, we found no significant relationship between the AUC for T4 from 14 to 28 DPI and previously established categorical phenotypes (Islam et al., 2013) of viral persistence and rebound (P > 0.05). The validity of such an analysis in the present study is questionable as the distribution of these phenotypes was skewed by the a priori selection for biological extremes in VL AUC, which created a heavy bias toward persistent (N = 36/49) vs. rebound (N = 4/49) phenotypes. Further analysis of large numbers of unselected animals will be required to determine if this biphasic patterning in T4 does in fact result from viral rebound or some other aspect of the host response to PRRSV2.

In either of the present disease challenge paradigms, the scale of endocrine disruption was considerable with concentrations of T3 and T4 following polymicrobial challenge decreasing to just 27.7% and 51% of pre-exposure levels, respectively, and T3 and T4 levels in PRRSV2-infected PHGC pigs reaching a minimum of 43% and 57% of pre-inoculation levels, respectively. Admittedly, a portion of the difference in the scale of disruption observed between the two experiments, may be, at least in part, associated with the higher pre-challenge levels of both hormones observed in the NDCM pigs. However, unlike PHGC pigs where concentrations of both hormones appeared to return to near-baseline levels by 42 DPI, animals in the NDCM showed a sustained disruption with no increase in T4 and only a partial increase in T3 concentrations by 42 DPE. Given the combination of high mortality (13.3% to 31.5%) and treatment rates (64% to 100%) in the three NCDM batches used for this study, the prolonged suppression, may simply result from greater disease pressure. The degree of thyroid suppression in NTIS has been linked to the severity of illness in both horses (Hilderbran et al., 2014) and dogs (Nishii et al., 2019) suggesting that the difference in thyroid response between PHGC and NDCM may be associated with the relative scale of the pathophysiological insult as well as the polymicrobial nature of the challenge for NDCM pigs. In addition to the obvious immunological response brought about by the full array of viral and bacterial pathogens present within the NDCM, both PCV2 and PRRSV specifically target cells of the immune system and deplete the capacity of their hosts to combat other pathogens (Nauwynck et al., 2012; Butler et al., 2014), further exacerbating the physiological burden. The present study identified a strong relationship between circulating thyroid hormone concentration and growth performance in both the NDCM and PHGC challenge models. In both studies animals which maintained higher levels of thyroid hormone following disease challenge exhibited higher levels of growth. This relationship is perhaps unsurprising given nonpathogenic models of hypothyroidism in rats have demonstrated a significant reduction in body WG (Nambiar et al., 2014). While growth performance of pigs following methimazole treatment has only been evaluated in a small number of animals, a significant decrease in ADG was observed in treated animals (Schöne et al., 1997). In the present experiment, the association between serum thyroid hormone levels evaluated in the first 42 d following exposure in the NDCB and ADG extended well into the grower period, with a significant relationship between ADG and T3 observed out to 77 DPE for T3 and 56 DPE for T4. Although the relationships in subsequent growth periods approached significance for both analytes, the relationship with ADG over the entire grow finish period was statistically significant indicating that compensatory gain was insufficient to make up for earlier losses. Interestingly the present study also finds a significant relationship between loin muscle depth and concentrations of T3. This finding is consistent with previous studies in mice which demonstrated that T3 supplementation was able to counteract muscle atrophy following starvation.

In addition to negatively impacting growth rate, a variety of immune challenges have been shown to decrease feed intake in swine (Pastorelli et al., 2012; Helm et al., 2019). Although individual feed intake was not measured during the sampling time period of either challenge trial, a positive relationship is known to exist between dietary intake and circulating thyroid hormones (T3 and T4) within the context of ambient temperature (Christon, 1988; Pathak et al., 2018). For instance, feed restriction of pigs for as little as 24 h has been shown to cause a significant downregulation in the circulating T3 and T4 concentrations in swine (Buonomo and Baile, 1991). In addition, acute starvation, in a rat model, has also been shown to suppress thyroid deiodinase rate and decrease thyroid hormone excretion though both the urinary and fecal routes (Nathanielsz, 1970). Such an acute effect, resulting from disease induce changes in feed intake, may play a minor role in the observed thyroid disruption, as previous analysis of feed intake data from the finisher phase of the natural disease challenge barn identified periods of decreased feed intake and duration that were found to be associated with resilience (Putz et al., 2019). The relationship between thyroid hormones and feed intake has been shown to be rather acute, with return to baseline levels within 24 h after the restoration of normal feed intake (Buonomo and Baile, 1991). The acute relationship between feed intake and thyroid hormone is most evident in horses where T3 concentrations have been shown to decrease in response to off feed periods of more than 1 d, but if feed intake is suppressed for more than 3 d T3 levels appeared to rebound to normal (Hilderbran et al., 2014).

We also evaluated the relationship between thyroid hormone and viremia within the pre-selected PHGC population and found no significant relationship. This result was surprising given the well-established capacity of these hormones to modulate activity in the innate (Montesinos and Pellizas, 2019) and adaptive (Jara et al., 2017) immune systems. Controlled studies with herpes simples virus-1 in rats demonstrated that chemical induction of a hypothyroid state significantly increased splenic VL, while excessive supplementation with levothyroxine to produce a hyperthyroid state resulted in a considerable decrease in VL (Varedi et al., 2014). The impact of a hypothyroid state on macrophages and dendritic cells in particular, may be of interest in the context of PRRSV infection as the virus has restricted tropism to cells of the myeloid lineage which express the virus’ obligate receptor, CD163. Regardless of polarity, macrophages have been shown to express detectable levels of the THRβ receptor and are sensitive to T3 in particular; which has been shown to alter their differentiation into M1 vs. M2 phenotypes (Perrotta et al., 2014). Interestingly, T3 supplementation of unpolarized macrophages increased expression of an array of M1 macrophage markers including CCL5 (Perrotta et al., 2014). Our previous assessment of the fetal immune response to PRRSV infection identified upregulation of both thymic and splenic CCL5 in meconium-stained (compromised) infected fetuses (Pasternak et al., 2020c), which we later showed to also have elevated concentrations of T3 compared with their viable (non-meconium-stained) high-VL cohorts (Pasternak et al., 2020a). PRRSV has been shown to modulate the host immune system, including induction of specific cytokines such as CCL5 (Wang et al., 2011). Given the well-established interaction between the thyroid hormone and immune systems, a more in-depth assessment of cytokine response profiles of animals at the biological extremes of thyroid hormone response may be valuable in understanding the variability in post-challenge performance.

A febrile response resulting in elevated body temperatures is often observed in swine following disease challenges with pathogens such as PRRSV (Loving et al., 2008) or IAV-S (Van Reeth et al., 2002). This physiological response results from the production of a wide array of pro-inflammatory cytokines which act as endogenous pyrogens, altering the hypothalamic regulation of body temperature (Zampronio et al., 2015). Given the established role of thyroid hormones in upregulating metabolic function and endogenous heat production, one would expect an increase in circulating thyroid hormone in response to these pyrogenic signals. This is, however, contradicted by the accumulated literature describing the NTIS response to critical illness which typically involves a decrease in free and total forms of both T3 and T4 with no corresponding upregulation in TSH, suggesting the set point of the HPT axis has been decreased (Chatzitomaris et al., 2017). Body temperature was not routinely measured in either of the present challenge experiments, but if the typical febrile response associated with PRRSV occurred, it would have done so despite significant decreases in circulating T3 and T4.

Although the present study finds a potentially valuable association between ADG and the severity of disruption in thyroid hormone under both polymicrobial and experimental PRRSV challenges, it is worth considering that such disruption may represent an intentional host adaptation to either directly combat the infection or confer some form of resilience. Under normal conditions, an animal is able to divide its energetic resources between growth, reproduction, and maintenance, however, in the diseased state, significant energy must be diverted to the immune system (Lochmiller and Deerenberg, 2000). To mount an effective defense, many cells of the immune system undergo clonal expansion and/or must produce an array of cytokines, chemotactic factors, and antibodies, all of which require energy and adaptation of metabolic pathways to produce the necessary molecular building blocks (Maciolek et al., 2014). In swine, this increased energy requirement during illness is likely further exacerbated by a concomitant decrease in feed intake associated with morbidity (Pastorelli et al., 2012). It has recently been proposed that a hypometabolic state, resulting from a trade-off between immune function and thermal regulation, allows for resilience in response to bacterial infection (Ganeshan et al., 2019). Such studies have not directly evaluated the role of thyroid hormone allostasis as a tolerogenic mechanism, however, as the central regulator of basal metabolic rate, depression of this system would create the observed hypometabolic state. Further evidence that such disruption in the thyroid pathway may represent a protective effect can be found in our recent experiments with PRRSV2-infected fetuses where we identified a lack of suppression in circulating T3 concentrations among highly infected fetuses displaying meconium staining, a marker of fetal compromise, relative to their highly infected but viable counterparts (Pasternak et al., 2020a). In the late gestation pig fetus, where thyroid hormone drives not only organ maturation but also the accretion of fetal mass, a central downregulation in thyroid hormone may reduce the metabolic requirements of peripheral organs and thus conserve the limited resources obtained from the placenta. However, in the context of extra-uterine life, metabolic resources are not limited by placental function, but rather the results of combined respiratory and digestive function. PRRSV infection alone has been shown to suppress feed intake but had no effect on apparent total tract digestibility (Schweer et al., 2016). Apparent digestibility is known to be suppressed by L. intracellularis (Visscher et al., 2018) an organism that has been detected serologically in the NDCM. However, the delayed onset of intestinal malabsorption resulting from this pathogen is inconsistent with the acute decrease in thyroid hormones observed in animals following exposure. On the other hand, postnatal PRRSV infection has, perhaps unsurprisingly, been shown to substantially depress numerous metrics of respiratory function (Wagner et al., 2011). Such an effect would only be exacerbated by the additional respiratory pathogens including M. hyopneumoniae and IAV-S that are both present in the NDCM. In this context, the reduced metabolic rate resulting from suppression of thyroid hormone may serve to reduce global oxygen requirements. Viruses and many intracellular pathogens rely on host cellular metabolism to provide not only energy in the form of ATP, but also the necessary macromolecule precursors required for pathogen replication. To meet this demand, many pathogens have been shown to modulate host metabolic activity at the cellular level, resulting in a Warburg-like effect where energy is derived from aerobic glycolysis and the TCA cycle altered to support production of nucleotide and amino acid precursors (Thaker et al., 2019). Thyroid hormones are responsible for promoting both glycolytic and mitochondrial activity (Hoch, 1962) and thus, a hypothyroid state would be antithetical to the metabolic adaptations induced by many intracellular pathogens. Therefore, the hypometabolic effect resulting from NTIS may represent a mechanism by which the host can effectively starve a pathogen of macromolecule resources required for replication.

Conclusions

Collectively, our results show that swine are particularly susceptible to disruption of the thyroid hormone system following disease challenge. We observe a significant decrease in circulating T3 and T4 following both natural polymicrobial and experimental PRRSV2 challenges that cannot be associated with normal temporal dynamics in an equivalent healthy cohort. In both instances, the maintenance of higher serum T3 or T4 following challenge was positively correlated with ADG. Although no significant relationship between thyroid hormone levels and PRRSV2 VL was identified, the observed differences in severity and temporal pattern of this response in pigs infected with PRRSV-alone (PHGC) compared with polymicrobial challenge (NDCM) suggests additional pathogens may either independently induce, or perhaps augment, this NTIS like effect. Further investigation into the role of thyroid hormone and phenotypes such as growth rate, feed intake, and carcass quality in the absence of disease, may prove valuable in identifying management or treatment strategies capable of augmenting this high value trait. In addition, the relationship between thyroid hormone levels and pathogen replication should be investigated to determine if external manipulation of this system during infection will compromise a key component of the host pathogen defense.

Glossary

Abbreviations

ADG

average daily gain

AUC

area under the curve

CDPQ

Centre de développement du porc du Québec

DPE

days post-exposure (in NDCM)

DPI

days post infection (in PHGC)

HPT

hypothalamic–pituitary–thyroid axis

IAV-S

swine influenza virus type A

LPS

lipopolysaccharide

NDCM

natural disease challenge model (polymicrobial challenge)

NTIS

non-thyroidal illness syndrome

PCV2

porcine circovirus type 2

PHGC

PRRS host genome consortium

PRRSV2

porcine reproductive and respiratory syndrome virus-2 (Strains NVSL 97-7895 and KS-2006-72109)

RIA

radioimmunoassay

rT3

reverse T3 (3,3′,5′-triiodothyronine)

T2

diiodothyronine

T3

3,3′,5-triiodothyronine

T4

thyroxine

TSH

thyroid-stimulating hormone

VL

viral load

WG

weight gain

Conflict of interest statement

The authors declare no real or perceived conflicts of interest.

Author’s Contributions

J.A.P. performed sample selection, conducted the thyroid hormone assays, statistical analysis, and drafted the manuscript with D.J.M. and J.C.S.H. J.C.S.H., G.S.P., M.K.D., P.G.C., F.F., and J.C.M.D. conceived, designed, and managed the NDCM trials and supplied the necessary sample and data for the present analysis. J.C.S.H. and J.A.P. arranged for sampling on the healthy farms. J.K.L., R.R.R.R., J.C.M.D., and G.S.P. conceived, designed, and managed the PHGC challenge trials and supplied the necessary sample and data for the present analysis. All authors read and revised the final manuscript.