Our study characterized genes and pathways associated with variation in VL and preservation status in fetuses. Our main findings were: 1) the host immune response was initiated only in fetuses with detectable levels of PRRSV; 2) upon infection of fetal thymus, a set of core responsive IFN-inducible genes (CXCL10, IFIH1, IFIT1, IFIT3, ISG15, and MX1) were strongly upregulated in both tissues. V - F (contrasted with V - F) had numerous pathways deactivated including Acute Phase Response, B Cell Receptor Signaling, HMGB1 Signaling, iNOS Signaling, NFKB Signaling, Production of RO and NOS in macrophages, Senescence Pathway, and TLR Signaling in both tissues; 3) gene expression associated with VL was more accurately assessed in the thymus than placenta; the strong downregulation of genes in the B Cell Development pathway may be a major mechanism of PRRSV dysregulation of host immunity associated with high VL; and 4) gene expression associated with fetal demise was more accurately assessed in the placenta than the thymus; potential biomarkers of susceptibility (NFKB2, NFKBIA, and FASLG) were identified that may contribute to fetal death. These results advance our understanding of the fetal response to PRRSV infection one step further by characterizing tissue specific gene expression patterns associated with host resistance, resilience, and susceptibility. The fetal classifications used in the current study are based on previous work in our group [16]. We classified the UNINF, PLCO-VIA, PLCO-MEC groups as resistant as they have avoided or minimized infection suggesting some capacity to prevent viral entry/replication [17]. The LVL-VIA and HVL-VIA were classified as tolerant/resilient as they remain uncompromised in the face of viral infection [18] whereas the LVL-MEC and HVL-MEC were considered susceptible as they are neither able to limit viral replication and/or survive it [17]. A limitation of our classifications is that we are looking at a snapshot in time at 12 DPI. We cannot be certain what the ultimate outcome (i.e., VL and meconium staining status) of these fetuses would be if the pregnancy went to term. Additionally, the placenta has been infected with PRRSV (except for the UNINF group) for a longer period compared to the thymus due to the nature of the transplacental infection model. Thus, interpretation of the results of our study carefully considered the experimental design.
Rational for tissue selection and experimental methods
The placenta functions to protect the fetus while providing critical nutrients and oxygen from the dam to the fetus. We chose to investigate the placenta to gain a better understanding of the role it plays in fetal protection and transmission of PRRSV, critical questions that remain unanswered. The thymus primarily functions as the site for T cell progenitor production and is an important site of PRRSV replication in the fetus [19], indicating the importance of this tissue in understanding host response to infection. We used a targeted gene approach assaying 286 genes using NanoString technology, which is a more sensitive measure of gene expression and less dependent on RNA quality as compared to RNAseq. NanoString analyses are highly dependent on the codeset used. They require considerable expertise to assemble immune gene codesets properly and may miss genes that are not expected to be associated with PRRSV infection. Thus, NanoString analyses may be less useful for hypothesis generation when compared to RNAseq. Future RNAseq analyses of these samples, and potentially additional fetal tissues from these individuals, could identify additional genes and pathways involved in this complex of host-pathogen interactions. All 22 pathways we investigated were enriched in one or more contrasts in this study, providing evidence for the value of the selected set of genes and their associated pathways.
Importance of fetal infection status on the initiation of host response
A major finding of our study was that PRRSV must be detected in the fetus to initiate a host immunologic response in both the placenta and thymus. Our results show that once fetuses are infected, and especially in HVL fetuses, a strong immune response is initiated as reported previously [13, 19]. However, we detected no DEG in virus negative fetuses (i.e., UNINF, PLCO-VIA, and PLCO-MEC each contrasted to CTRL) in either tissue, indicating the regulation of immune-related gene expression, at least at 12 DPI, is not responsible for generating the resistant fetal phenotype. We anticipated low responsiveness in the thymus of virus negative fetuses but not in the placenta because: 1) virus was detected in the placenta in PLCO groups, 2) the MFI plays an important role limiting fetal exposure to PRRSV, and 3) the placenta has innate immune response capabilities and produces cytokines that play an important role in all stages of pregnancy in pigs [20, 21]. The placenta is also rich in trophoblast cells, which cover most of the placental surface and function to recognize pathogens through TLRs, produce cytokines, and recruit immune cells [22]. In contrast to our results, a previous study using RNAseq reported 864 and 121 DEG in the MFI (endometrium plus placenta) and thymus, respectively, in the UNINF vs CTRL contrast using RNA-seq at 21 DPI [11]. The contrasting results between our study and the Wilkinson study may be explained by timepoint (12 DPI vs 21 DPI), gene expression platform (NanoString vs RNAseq), and/or tissues analyzed (placenta vs MFI, the combined endometrium and placenta). The presence of the endometrial tissue, with its own unique transcriptomic signature, may have not only added additional DEGs but also further obscured others due to relative abundance in such a sample. Alternatively, the lack of response of the placenta, while fetuses are still uninfected in the current study, may be caused by PRRSV modulating the host immune response [1, 23]. More research should be done to probe this mechanism. Although our study did not identify genes and their expression patterns that are associated with resistance, we do report those associated with tolerance and susceptibility.
IFN response and importance in host immunity
Our results show in virus infected fetuses (LVL-VIA, LVL-MEC, HVL-VIA, and HVL-MEC compared to CTRL) a coordinated and strong upregulation of IFN-inducible genes regardless of fetal VL or preservation status in both placenta and thymus, indicating potential crosstalk in immune activation between the two tissues. A hallmark of host response to viral infection, IFNs are cytokines released from virus infected cells that activate host immunity to combat infection. The pig has IFNs of Types I, II, and III, all of which mainly function to induce a cascade of cellular responses that result in transcriptional activation of immune-related genes via STAT and NF-κB. The pig has many duplicated and unique IFN genes, many of which were tested herein (IFNA, IFNB, IFND, IFNE, IFNG, IFNK, and IFNW) along with the receptor IFNGR1. Previous research has shown that the level of type I IFNA protein in growing pigs is correlated with favorable immune response to PRRSV infection [24]. Additionally, IFNB protein is induced variably in PRRSV infected porcine alveolar macrophages in vitro [25].
While IFNA and IFNB are the most often studied with regards to viral response, many of the other type I IFNs found in swine have been shown to exhibit antiviral properties with regards to PRRSV [26]. It is, however, worth noting that during at least the early stages of pregnancy the porcine conceptus, like other livestock species, uses type I (IFND) and type II (IFNG) IFN along with the canonical Interferon Signaling pathway to communicate with and alter the behavior of the endometrium [27]. The role of such signaling during later gestations remains unclear, however, it stands to reason such signaling pathways may still modulate functionality at the MFI during late gestation. Here we observed two type I IFNs, IFNW4/5 and IFNK, among the top loaded genes in the placenta principal component 2. Component 2 was found to rather clearly differentiate between VIA and MEC fetuses regardless of VL. In the context of this component both genes were upregulated in viable fetuses and downregulated in their MEC counterparts. It is thus possible that expression of these IFNs may offset the observed significant decrease in placental IFNE associated with viral infection of the fetus and, thereby, play a vital role in mediating fetal resilience at the placenta. Another study found high levels of gene expression of type I IFNs (IFNA and IFNB) in PRRSV infected fetuses compared to non-infected fetuses and the expression was poorly correlated with serum type I IFNs at the protein level [13], indicating a post-transcriptional modification is occurring. In our study, we found a non-consistent pattern of IFN expression across fetal groups and tissues, indicating IFNs themselves may not be good predictors of fetal response to infection. In the placenta, IFNG was upregulated in the HVL-VIA fetuses while IFNE was downregulated in the fetuses of V + F contrast. We also found IFNB1 to be increased in placenta HVL-MEC, in the thymus HVL-VIA. In the placenta, IFNG1R was downregulated in LVL-MEC, HVL-VIA, HVL-MEC, and in V + F contrast.
The activation of the Interferon Signaling Pathway reported herein was associated with the upregulation of core IFN-inducible genes, CXCL10, IFIH1, IFIT1, IFIT3, ISG15, and MX1, in both tissues in every fetal group that is virus positive (LVL-VIA, LVL-MEC, HVL-VIA, and HVL-MEC each compared to CTRL). While we do not know which IFNs may be driving the expression of these IFN-inducible genes, it may be that IFNs produced in the blood or other lymphatic tissues are transported to the placenta and thymus causing increased expression of IFN inducible genes. This dysregulation of host immunity by PRRSV infection, supports previous works in this area [13, 28,29,30].
Dysregulation of immune-related pathways
Our data show that the placenta and thymus of V + F (vs V - F) deactivate Acute Phase Response, B Cell Receptor Signaling, HMGB1 Signaling, iNOS Signaling, NFKB Signaling, Production of RO and NOS in Macrophages, Senescence Pathway, and TLR Signaling. Previous work, typically performed in vitro, has shown that PRRSV can induce many of these pathways [31,32,33,34,35,36,37]. Our study is a single snapshot in time (12 DPI) and it may be that the dysregulation of these pathways is time dependent. A commonality of several of these pathways identified in the current study is cellular signaling through the JAK-STAT, MAPK/ERK, and PI3K/AKT/mTOR pathways. Previous work has shown that PRRSV inhibits JAK-STAT signaling through dysregulation of cytokines reviewed in [38]. These complex pathways are integrated via JAK proteins that phosphorylate cytokine receptors, activate MAPK, NFKB and PI3K proteins, and transcriptionally regulate themselves or other proteins downstream [39]. Specifically, in both placenta and thymus V + F groups we report a downregulation at the receptor (TLR4), cytoplasmic signaling (JAK2, SOCS3, SOCS6, PIK3CG/PIK3C2B), and nuclear signaling levels (MAP2K4, MAP3K7, MAPK14, and NFKB). These data suggest that PRRSV infected fetuses have multilevel and widespread silencing of transcriptional immune-related regulatory pathways, which may prevent efficient response to virus.
Differences between tissues in the V + F (vs V - F) include the dysregulation of the Antigen Presentation Pathway and the TR/RXR Activation Pathway in the placenta and thymus, respectively. Antigen presenting cells (APCs) play a crucial role in bridging innate and adaptive immunity by pathogen recognition, processing, and stimulation of T cells via major histocompatibility complex (MHC) presentation [40]. Our data show the dysregulation in the placenta of V + F of Antigen Presentation Pathway through upregulation of B2M, PSMB8, PSMB9, TAP1, and TAP2 but downregulation in the receptor MHCI-related. The APCs present in the placenta in V + F (vs V - F) may be infected with PRRSV but may not be interacting with T cells efficiently due to low MHCI protein expression and lack of IFNG induction. Interestingly, VL in fetal thymus is positively correlated with the number of PRRSV infected CD163+ and CD169+ cells in the MFI but, unexpectedly, the relationship of CD163+ cell counts in placenta was negatively correlated with fetal thymus VL [8]. Our study design investigates gene expression at the tissue level and thus, the changes in gene expression observed herein may be due to changes in numbers of specific cells (e.g., APCs and T cells), transcriptional changes in cells already present, or a combination of both. Regardless, our data show in detail that PRRSV must be replicating in the fetus to initiate changes in the Antigen Presentation Pathway in the placenta. Future research could use single cell gene sequencing to investigate APCs located in the placenta in virus infected fetuses.
In the thymus of V + F (vs V - F) we show a dysregulation of the TR/RXR Activation pathway. Thyroid hormones regulate a wide range of processes such as growth, development, and metabolism; this pathway has recently been shown to be dysregulated during both maternal and fetal response to PRRSV infection. The triiodothyronine (T3) hormone initiates cellular response by binding the nuclear receptors, THRa and THRb, to directly regulate gene expression. Our data show in the thymus of V + F, downregulation of thyroid hormone receptors (THRa and THRb), along with the critical outer ring deiodinase (DIO2) required to convert T4 to the more bioactive T3. In addition, three of the transmembrane transporters known to aid in the traffic of T3 and T4 into the cytoplasm (SLC16A10, SLCO1C1, SLC16A2) were also down regulated in V + F. Interestingly, we found a strong upregulation in the DIO3, an inner ring deiodinase which inactivates thyroid hormone by converting T4 to a comparatively inert rT3 and converting highly bioactive T3 to the metabolite T2, in various placenta contrasts but not in the thymus. Because of thyroid hormone’s role in growth, it could be that fetuses infected with PRRSV are allocating more resources to immunity rather than growth. Complicating matters further is the finding that increased fetal intrauterine growth is associated with a more susceptible phenotype to PRRSV infection [41], possibly indicating a delicate balance in resource allocation of fetuses infected with PRRSV. Alternatively, non-intrauterine growth restricted (IUGR) fetuses could begin infection sooner than IUGR fetuses. More research should be done on fetal thyroid hormone dysregulation to better understand the interplay between immunity and growth.
Our unique dataset allowed characterization of the gene expression differences related to fetal VL (LVL-VIA vs HVL-VIA) providing insight into mechanisms of resilience and/or susceptibility. We predicted that HVL fetuses would have a poorer outcome compared to LVL based on previous studies [10], indicating that fetal compromise and death is strongly related to VL. Despite virus detected in fetuses within the LVL-VIA, Interferon Signaling is the only pathway impacted in both the placenta and thymus. It could be that an immune response at an earlier timepoint limited the VL in these fetuses. Alternatively, delayed infection is a viable alternative, in that the LVL fetuses could have been infected a few days later than the HVL fetuses and the virus had not had enough time to replicate. By comparison, in the HVL-VIA we observe a strong downregulation of numerous immune pathways (vs CTRL), especially in the thymus. In the HVL-VIA group we did not detect as clear of a picture of host dysregulation in placenta as in the thymus.
Assessment of fetal demise and viral levels by gene expression
The downregulation of pathways and genes gives insight into ways the fetal host fails to limit viral replication. Dysregulation of B Cell Receptor Signaling, NFKB Signaling, iNOS Signaling, and HMGB1 Signaling in the thymus may contribute to increases in VL. We found a large number (N = 58) of DEG identified exclusively in the thymus of HVL-VIA fetuses (vs LVL-VIA, LVL-MEC, and HVL-MEC fetuses), all of which were downregulated compared to CTRL. Among these downregulated genes were CD19, CD25B, CD55, CD79A, CD79B, and CD83. The cluster differentiation (CD) genes have pleotropic immunologic roles but are well known as receptors often used to differentiate immune cell subtypes. The pathway analysis for this group of fetuses showed a 70% enrichment in B Cell Development that was largely unimpacted in any other fetal group. Our study investigated fetal response at 12 DPI and therefore, the fetuses could have been exposed to replicating PRRSV for 5 days already [16]. These genes may be considered markers of the inability to limit viral replication which could be associated with invasion of the thymus by immune cell populations or maturation of resident immune cells.
Finally, a major finding of our study was that gene expression profiles in the placenta more accurately assessed fetal demise compared to gene expression patterns in the thymus. Although we detected no DEG in either tissue in the comparison of PLCO-MEC versus CTRL, we observed a clear separation of the PLCO-MEC fetal groups across component 2 in both tissues. This fetal grouping is classified in the resistant category for the ability to limit/prevent viral infection. However, fetuses are clearly compromised via meconium staining for which the cause is not completely understood. The clear separation across component 2 may be indicative of conserved gene expression patterns consistent with fetal demise, regardless of fetal viral load. Although PLCO-MEC represents a relatively small proportion (i.e., in the current study N = 4 and N = 5 for the placenta and thymus, respectively) of the overall array of fetal outcomes, exploring their response further may be warranted. Our results clearly show the placenta has more unique changes in pathways between VIA and MEC fetuses compared to the thymus. Interestingly, we detected a robust gene expression response in the placenta LVL-MEC group, while the thymus remained relatively silent in both LVL-VIA and LVL-MEC groups. Our data revealed the clear separation of MEC from VIA fetuses on component 2 of the PCA for the placenta but not for the thymus. The top negative loading genes in the placenta for this component were IFNW4/W5, IFNK, MBL1, MASP1, and IL12B (Supplemental Table 3). The IFNW4/5 and IFNK were discussed above. The MBL1 protein may play a role in Complement activation [42]. MASP1 and IL12B proteins function in the Complement System and have pleotropic functions (HMGB1, Th1 and Th2 Activation, and TLR Signaling), respectively. Interestingly, all these genes were more lowly expressed in MEC fetuses compared to VIA fetuses. The direction of expression could be interpreted in two ways; reduced gene expression contributes to a susceptible phenotype (MEC) or higher gene expression contributes to a resistant/resilient phenotype (VIA). It could be that higher gene expression, especially that of the IFNs, may have a protective effect in the placenta. However, none of these genes were independently identified as DEG in any comparison groups indicating more research should be done on these biomarkers. In addition, in the differential expression analysis of MEC + F we identified 7 and 3 DEG in the placenta and thymus, respectively. Of these 7 DEG in the placenta, we found NFKB2, NFKBIA, and FASLG to have consistently higher expression and statistically significant patterns in fetuses with meconium staining (PLCO-MEC, LVL-MEC, HVL-MEC) compared to viable (UNINF, PLCO-VIA, LVL-VIA, and HVL-VIA) fetuses. The NFKB2 and NFKBIA proteins are highly pleotropic but are best known for their involvement in the NFKB Pathway, a well-characterized pathway initiated during host infection [43]. Interestingly, all three of these genes are in the Apoptosis Pathway. Across all groups of virus positive fetuses in the placenta, we see a strong activation and enrichment of genes in the Apoptosis and Ubiquitination Pathways. PRRSV causes apoptosis in the maternal uterine epithelium and fetal trophoblast epithelium as well as surrounding cells in late gestation [6, 44].
The question of PRRSV infection in the fetus as the cause of demise has been previously questioned based on the understanding that fetal lesions are infrequently observed [19]. However, it is now understood that the fetus responds physiologically and immunologically to PRRSV infection in the absence of pathology [11, 13, 14, 45]. Previous reports show a positive association between fetal survival during PRRSV infection and reduced intrauterine growth [16], of which increased apoptosis and increased cellular senescence could contribute to reduced fetal growth. Interestingly, we report changes in gene expression that downregulate genes in the Cellular Senescence Pathway in the thymus while Apoptosis Pathway in the placenta were activated. Cellular senescence is the reduction of cell division through various cell cycle checkpoints and is typically activated by decreased telomere length, or cellular stress, while it is sometimes decreased in cancerous cells via mTOR regulatory mechanism. Our results show that very different cellular survival/proliferation mechanisms are possibly occurring between the two tissues, with high cell death in the placenta and high cell proliferation in the thymus, giving further evidence to support that gene expression in the placenta more accurately assesses fetal demise while gene expression in the thymus more accurately assesses VL. However, the temporal aspect of our study must be considered because the placenta is infected with virus for a longer period in transplacental challenge compared to the thymus. For the first time, we report NFKB2, NFKBIA, and FASLG as potential biomarkers in the placenta that may contribute to fetal demise (MEC). Taken together, our study provides unprecedented insight into fetal response to PRRSV infection with a complex interplay between placenta and thymus.