Neuroimmune Origins of Behavioral Disorders

NEUROIMMUNE ORIGINS OF BEHAVIORAL DISORDERS

ADRIENNE ANTONSON

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Animal Sciences in the Graduate College of the University of Illinois at Urbana-Champaign, 2018

Urbana, Illinois

Doctoral Committee:

Professor Rodney Johnson, Chair and Director of Research Assistant Professor Andrew Steelman Professor Romana Nowak Professor Susan Schantz

ABSTRACT Neuropsychiatric illnesses pose a large burden on society, including health care costs, the need for special education and counseling, and emotional hardship for both sufferers and care givers. For many, these burdens are lifelong. The etiopathologies of some psychiatric illnesses, such as Autism Spectrum Disorder (ASD) and schizophrenia, appear to originate during key stages of prenatal or early postnatal neurodevelopment, offering a time window for therapeutic intervention. Epidemiological data has revealed that infections during pregnancy increase the risk of psychiatric illnesses, such as ASD and schizophrenia, in children. Indeed, similarities in psycho- and neuropathologies between the two disorders suggest that there may be a common etiology. Animal models of maternal immune activation (MIA), developed to investigate this link, suggest that maternally derived cytokines mediate fetal neurodevelopment by signaling across the placenta and upregulating inflammatory pathways within the fetal compartment. Microglia, the resident immune cells of the brain and robust responders to cytokine signaling, have been implicated in the neuroimmune pathogenesis of psychiatric disorders, but their role in fetal neurodevelopment during maternal infection has yet to be elucidated. Thus, the primary goal of the work presented in this dissertation is to delineate the activity of prenatal and neonatal microglial cells during and immediately following maternal infection, using a highly-translatable swine model. The investigations undertaken to obtain this goal are divided into three major sections within this dissertation. First, we established a prenatal MIA paradigm in swine by inoculating pregnant gilts with a live virus, porcine reproductive and respiratory syndrome virus (PRRSV), during late gestation to induce circulating pro-inflammatory cytokines and initiate classical sickness behaviors. Using this animal model, we demonstrated that maternal viral infection altered social behaviors, but not learning and memory, in neonatal piglet offspring. Postnatal microglia isolated from these animals did not differ from controls, indicating that aberrant sociability cannot be attributed to over-activation of these cells. Thus, we hypothesized that changes in neurodevelopment, regulated by microglial cells, may be occurring during the prenatal period, manifesting as altered behaviors later in life. The second and third sections of this dissertation were designed to test this hypothesis, and focused exclusively on neurodevelopment during the fetal time point. First, we analyzed gene expression and morphology of the developing porcine hippocampus, a brain region that displays ongoing concentrated neurogenesis, making it exceptionally vulnerable to maternal insults. We found that maternal infection reduced neuron number and caused astrocyte-specific gliosis in conjunction with increased expression of select inflammatory genes, three days before anticipated parturition (five weeks-post-inoculation). Additionally, MIA resulted in a reduction in overall

ii fetal brain weight, but not body weight, emphasizing the targeted impact of maternal viral infection on the central nervous system. Notably, there was an absence of pronounced neuroinflammation at this time, including an apparent reduction in microglial cell activation, which indicated that classical neuroinflammatory pathways, if activated by MIA, are mostly resolved by parturition. The third section of this dissertation, therefore, focused on investigating neuroinflammation, and specifically microglia activation, at earlier time points: peak maternal viral infection (7 days-post-inoculation [dpi]) and immediately following the resolution of maternal symptoms (21 dpi). At 21 dpi, we observed a reduction in fetal brain weights, but not body weights, due to maternal viral infection, replicating our previous findings. Assessment of gene expression revealed increased inflammation at the maternal-fetal interface (endometrium and placenta) at both 7 and 21 dpi. Evaluation of fetal microglia revealed an increase in the classical activation marker MHCII and altered levels of phagocytic and chemotactic activity at both time points. Using high throughput quantitative real time- PCR, we observed that genes involved in neurodevelopment, the microglia sensome, and inflammation (enriched in microglia) were differentially regulated in fetal microglia. Given that reduced sociability was evident in neonatal piglets, we also examined gene expression patterns in the fetal amygdala, a brain region integral to the regulation of social behavior, and found that a similar subset of genes was altered by MIA. We also observed that the effects of MIA on inflammatory and neurodevelopmental processes appears to be temporally regulated, evidenced by diverging patterns of microglial and amygdalar gene expression from 7 to 21 dpi. Interestingly, sexual dimorphisms were evident in gene expression patterns across both tissues and time points, mimicking similar patterns observed in human neuropsychiatric disorders. Finally, we assessed microglia number and morphology in the fetal hippocampus and amygdala. While both number and morphology were relatively unchanged in the hippocampus at 7 and 21 dpi, an increase in microglia number was evident in the amygdala at 7 dpi, though this did not continue to 21 dpi. Contrary to our hypothesis, maternal viral infection did not alter microglia morphology. In conclusion, this dissertation provides evidence that maternal viral infection in swine results in altered social behaviors postnatally, and that changes in postnatal behaviors could be contributed to aberrant neurodevelopmental processes during acute maternal immune activation. Specifically, we show that fetal microglial cells are transiently sensitized during peak maternal viral infection and that insults incurred during peak infection persist after inflammation has resolved. Overall, these data emphasize the importance of prenatal interventions and provide a foundation for fetal microglial cells as potential therapeutic targets in neurodevelopmental disorders.

iii

ACKNOWLEDGEMENTS This work would not have been possible without the support, insight, and friendship of an extraordinary group of people. This includes my committee members and of course, my mentor, Dr. Rod Johnson, who taught me to love the pursuit of knowledge and who challenged me every day to become a better research scientist. This also includes my lab mates, who taught me the importance of humor and loyalty, especially in the face of struggle, and with whom I have shared a special friendship for the last five years; I will miss you all and I wish everyone the best as we each move on to the next step in our lives. I would like to give a special thanks to two of my biggest mentors and supporters, Dr. Emily Radlowski and especially Dr. Marcus Lawson, whose guidance and mentorship throughout the years was indispensable. A very special thanks goes out to my family and friends, who believed in me from the start and who were always there for late night cocktails or phone calls when the going got rough. Finally, the biggest thank you goes out to my fiancé, Jacob Allen, who continues to be my chief ally, my strongest supporter, and my greatest challenger; you brought out a confidence and determination that I didn’t know was in me. THANK YOU!

For my mom, who has conquered in the face of strife all her life.

You are my biggest idol.

LIST OF ABBREVIATIONS

ANOVA Analysis of Variance ASD Autism Spectrum Disorders BSA Bovine Serum Albumin DEX Dexamethasone (synthetic glucocorticoid) DPI Days Post Infection DSM Diagnostic and Statistical Manual cDNA Complementary DNA CNS Central Nervous System CSF Cerebral Spinal Fluid E Embryonic Day ELISA Enzyme-Linked Immunosorbent Assay FBS Fetal Bovine Serum GC Glucocorticoid GD Gestational Day Iba1 Ionized Calcium-Binding Adaptor Molecule 1 (marker for microglia) Ig Immunoglobulin IL Interleukin i.m. Intramuscular i.p. Intraperitoneal i.v. Intravenous LPS Lipopolysaccharide (Bacterial Mimetic) M1 Classical (Pro-Inflammatory) Macrophage Activation Category M2 Alternative (Anti-Inflammatory) Macrophage Activation Category MIA Maternal Immune Activation mRNA Messenger RNA PBS Phosphate Buffered Saline PD Postnatal Day Poly I:C Polyinosinic:Polycytidylic Acid (Viral Mimetic) qPCR Quantitative Polymerase Chain Reaction (Gene Expression) RT-PCR Real-Time Polymerase Chain Reaction (Gene Expression) TCID Tissue Culture Infective Dose

TH T Helper Cell TLR Toll-Like Receptor

TABLE OF CONTENTS CHAPTER 1. INTRODUCTION ...... 1

1.1 SIGNIFICANCE ...... 1 1.2 INNOVATION ...... 2 1.3 SPECIFIC AIMS ...... 3 CHAPTER 2. LITERATURE REVIEW ...... 5

2.1 MATERNAL IMMUNE ACTIVATION ...... 5 2.2 MICROGLIA IN THE CONTEXT OF MATERNAL IMMUNE ACTIVATION ...... 7 2.3 SOCIAL BEHAVIOR IN THE CONTEXT OF MATERNAL IMMUNE ACTIVATION ...... 12 2.4 SUMMARY ...... 14 CHAPTER 3. MATERNAL VIRAL INFECTION DURING PREGNANCY ELICITS ANTI-SOCIAL BEHAVIOR IN NEONATAL PIGLET OFFSPRING INDEPENDENT OF POSTNATAL MICROGLIAL CELL ACTIVATION ...... 15

3.1 ABSTRACT ...... 15 3.2 INTRODUCTION ...... 16 3.3 MATERIALS AND METHODS ...... 17 3.4 RESULTS ...... 24 3.5 DISCUSSION ...... 27 3.6 FIGURES AND TABLES ...... 31 CHAPTER 4. ALTERED HIPPOCAMPAL GENE EXPRESSION AND MORPHOLOGY IN FETAL PIGLETS FOLLOWING MATERNAL RESPIRATORY VIRAL INFECTION ...... 41

4.1 ABSTRACT ...... 41 4.2 INTRODUCTION ...... 42 4.3 MATERIALS AND METHODS ...... 44 4.4 RESULTS ...... 49 4.5 DISCUSSION ...... 51 4.6 FIGURES AND TABLES ...... 57 CHAPTER 5. MATERNAL VIRAL INFECTION CAUSES GLOBAL CHANGES IN FETAL MICROGLIA AND ALTERS GENE EXPRESSION AND MICROGLIA DENSITY IN THE FETAL AMYGDALA ...... 66

5.1 ABSTRACT ...... 66 5.2 INTRODUCTION ...... 67 5.3 MATERIALS AND METHODS ...... 68 5.4 RESULTS ...... 74 5.5 DISCUSSION ...... 80 5.6 FIGURES AND TABLES ...... 90 CHAPTER 6. CONCLUSIONS AND FUTURE DIRECTIONS ...... 114

6.1 FIGURE ...... 117 CHAPTER 7. REFERENCES ...... 118

vii

CHAPTER 1: INTRODUCTION

1.1 Significance Evidence linking altered neuroimmune mechanisms with the etiopathologies of neurodevelopmental disorders, especially during the prenatal period, has greatly expanded over the past decade. Epidemiological studies investigating this link have revealed that prenatal exposure to maternal infection is highly associated with specific psychiatric disorders like Autism Spectrum Disorders (ASD) and schizophrenia [1, 2]. It has been proposed that similarities in psycho- and neuropathology could link both disorders to a common etiology relevant for induction of either pathological phenotype; specifically, increased fetal neuroinflammation during maternal infection [3]. Interest in the link between prenatal maternal infection and the prevalence of neuropsychiatric disorders, specifically ASD and schizophrenia, has greatly expanded since the initial studies were performed, close to three decades ago. However, the etiopathologies of both disorders are still undefined, making it difficult to delineate pathways that may be involved in both prenatal maternal infection and the pathologies of these disorders. The Diagnostic and Statistical Manual (DSM-5) provides two major criteria for the diagnosis of ASD: (1) deficits in social communication and interactions, and (2) patterns of restricted, repetitive behaviors, interests, or activities. Over the past decade, the incidence of ASD diagnoses in the United States has steadily risen, though it is difficult to say whether this rise is due to more comprehensive diagnostic criteria or increased awareness [4]. The most recent report from the CDC states that one in 68 children aged 8 years may be diagnosed with ASD [5], and though therapeutic treatments such as oxytocin have seen some success [6, 7], there is still a need for long-term preventative therapies. Development of the disorder may in part be related to biological sex differences, as prevalence of ASD is characterized by a heavy male bias, with approximately 4 affected males for every 1 affected female [8]. Schizophrenia is characterized by two or more of the following five symptoms per the DSM-5: delusions, hallucinations, disorganized speech, grossly disorganized or catatonic behavior, and negative symptoms such as diminished emotional expression. The illness is also characterized by social/occupational dysfunction [9]. Though less than 1% of the population suffers from schizophrenia, the cost to society is disproportionately large considering the functional impairments associated with the illness and the variable efficacy of available treatments [10]. As with ASD, males are more likely to be diagnosed with schizophrenia [11], and although the disorder was first defined over a century ago, the exact etiology and pathophysiology of the illness eludes researchers today [9].

It appears that both ASD and schizophrenia share psycho- and neuropathologies that may be linked to prenatal maternal infection [3], so defining these pathways could be the key component to delineating a shared etiology, and thus developing efficacious preventative therapies that could be implemented prenatally. It is also important to note that epidemiological studies assessing maternal infection and the incidence of psychiatric disorders indicate that the presence of a pathogenic infection, regardless of the transmission and type, presents a threat to the offspring. For example, both transmissible (rubella virus, cytomegalovirus, herpes simplex virus, toxoplasma gondii) and non- transmissible (influenza virus) pathogens confer risk [1, 2], bolstering the idea that a robust immune response, and not the type of initiating pathogen, is the driving factor.

1.2 Innovation Though there are many established preclinical rodent models, and a few non-human primate models, for investigating maternal infection, we are the only lab to establish a maternal immune activation (MIA) model in swine (to our knowledge). Swine are particularly suited for investigating the impacts of maternal infection on fetal brain development as their brain growth trajectories, neuroanatomy, and neurochemistry closely mimic that of humans in prenatal life [12]. Beginning in the last third of gestation, the pig undergoes a dramatic brain growth spurt, and as this is also the case in humans [13-15], we chose the initiation of this growth spurt as the maternal inoculation time point, a time when essential processes such as myelination, synapse formation, and neural migration are occurring [16]. Subsequently, both pigs and humans reach ~25% total brain volume by birth [14], which is the closest chronological comparison that we are aware of. In addition, the pig is increasingly recognized as an ideal model for biomedical studies, as the porcine immunome resembles humans in 80% of the parameters analyzed (versus less than 10% for mice)[17, 18]. Next to the primate and murine immune system, the swine immune system is extremely well characterized [19, 20] and, thanks to complete sequencing of the swine genome, a wide range of established methodologies and techniques exist for swine immune research. In our model, specifically, the pig is ideal for clearly discerning maternal and fetal contributions to immune responses because the placenta and endometrium can be easily separated, allowing for specific interactions to be analyzed for each individual fetus. Additionally, the large litter size (10-12 piglets/litter) allows for sex and sibling matching, while the gestation length of 114 days in swine (compared to 270 days in humans, and 18.5 and 21 days in mice and rats, respectively) enables targeting of specific gestational time points relevant to human neurodevelopment (a more daunting task in murine models). Domestic swine also

provide specific advantages regarding the extent of surgical and non-surgical procedures that can be performed and the large amount of tissue that can be collected from a single animal. Using magnetic resonance imaging of the piglet brain, our lab has mapped brain anatomy and growth within the first several months of life [21, 22], revealing the stark similarities between brain growth trajectories in neonatal swine and neonatal human infants [23]. The highly folded gyrencephalic anatomy of the pig brain also allows for accurate comparisons between pigs and humans across brain regions, a palpable benefit compared to lissencephalic murine models. Though there are a few established non- human primate MIA models, several factors such as cost, housing logistics, biomedical research approval, and the long lifespan of non-human primates causes difficulties in obtaining data on the impacts of MIA on these gyrencephalic species across multiple time points. Thus, the gyrencephalic pig offers an affordable alternative that is still very translatable to humans. Due to its extensive use in agriculture, there is also extensive data on swine health and diseases, allowing our lab to select a widely-studied viral infection to model MIA. The use of a live viral infection, versus highly purified bacterial or viral mimetics, accounts for variation in individual maternal responses, which better replicates epidemiological data. Moreover, we aimed to investigate fetal and neonatal time points relevant to the peak and resolution of maternal infection, which represent not only essential periods of neurodevelopment, but also critical time points in delineating the immediate effects of maternal infection on neuroimmune parameters of gestating and newborn offspring; these time points have largely been ignored in the MIA literature until recently. Lastly, we aimed to specifically investigate the role of microglial cells, the resident immune cells of the brain, in altering neurodevelopmental trajectories due to maternal infection, which is an extremely important facet in defining the etiology of MIA-induced neuro-phenotypes and behavioral symptoms.

1.3 Specific Aims Aim 1. Determine how maternal viral infection during gestation impacts neurocognitive and neuroimmune function in the early postnatal period. The main objectives of this aim were to establish a maternal immune activation model in swine and to characterize the impacts of MIA on postnatal piglet behaviors and microglia cell activation. We hypothesized that MIA would elicit neurocognitive deficits, particularly reduced performance both in a hippocampal-dependent task and in a social novelty task, and would prime microglial responses such that a second inflammatory hit would result in exaggerated activation.

Aim 2. Determine how maternal viral infection during late gestation impacts neurodevelopment in the fetal hippocampus, a brain region that is particularly vulnerable to maternal insults.

The objective of this study was to assess neurodevelopmental processes and glial reactivity in the fetal hippocampus immediately prior to parturition, when maternal symptoms have resolved. We hypothesized that maternal infection would permanently alter fetal hippocampal development by decreasing neuron numbers and that fetal microglia cells would display an activated phenotype.

Aim 3. Determine how maternal viral infection during late gestation impacts fetal microglial cells and overall neuroinflammatory status, both at peak maternal infection and at the resolution of symptoms. The main objectives of this aim were to define fetal microglial cell activation and to characterize the inflammatory status of the maternal-fetal interface and fetal brain at two major time points following induction of MIA. The working hypothesis was that maternal infection would elicit acute fetal microglia activation and increased cell density, leading to heightened neuroinflammation evident at the peak of maternal infection, while at the resolution of maternal symptoms, fetal neuroinflammation and microglia pathology would be partially or fully resolved.

CHAPTER 2: LITERATURE REVIEW

2.1 Maternal Immune Activation Defining critical neurodevelopmental pathways in maternal infection studies is necessary for understanding the link to neuropsychiatric illness. Preclinical animal models allow investigators to parse out the mechanisms driving maternal infection towards altered prenatal neurodevelopment, and, ideally, to discover interventions that can be implemented before symptoms occur. Animal models of maternal immune activation (MIA), which commonly use the viral mimetic polyinosinic:polycytidylic acid (poly I:C) or the bacterial endotoxin lipopolysaccharide (LPS), demonstrate that fetal neuroinflammation is likely derived from a heightened exposure to maternal cytokines [24-29], creating an imbalance of inflammatory signaling in the fetal brain [30]. Induction of MIA in pregnant animals results in offspring that display core behavioral symptoms of both ASD (such as reduced social behavior, disruptions in vocalizations, and stereotypic behaviors [31, 32]) and schizophrenia (impaired cognition, delayed pre-pulse inhibition, altered locomotor activity, and anhedonia [24, 33]). Indeed, preliminary clinical studies support the idea that heightened maternal inflammatory cytokines are correlated with the behavioral symptoms of neurodevelopmental disorders, such as psychosis and impaired cognition, in adulthood [34, 35]. Direct mediation of offspring outcomes by maternal cytokines has been demonstrated in animal studies utilizing genetic or molecular interventions to induce or block expression of specific cytokines in pregnant rodents. Smith et al. (2007) showed that injecting pregnant dams with recombinant IL-6 resulted in behavioral abnormalities like those observed in viral infection, poly I:C, or LPS models [26]. Moreover, blocking IL-6 through genetic methods (IL-6 knock outs) or neutralizing antibodies prevented the behavioral abnormalities induced by maternal poly I:C treatment [26]. Follow-up work by the same group indicated that maternally derived IL-6 protein and IL-6 mRNA were upregulated in the placenta, leading to local immune cell infiltration and a disruption in growth hormone levels during gestation, and culminating in altered behaviors in adult offspring [36]. More recently, Wu et al. (2016) revealed that placental IL-6 signaling is indeed required for maternal-fetal transmission of inflammatory signals that induce neurodevelopmental and behavioral pathologies seen in these MIA models, and that restricted deletion of the IL-6 receptor on placental trophoblasts prevented these pathologies [27]. Additionally, intracellular downstream IL-6 signaling, through phosphorylation of signal transducer and activator of transcription (STAT3), is increased in fetal brains exposed to maternal LPS treatment, but blunted by maternal administration of IL-6 blocking antibody [28]. In vitro work also suggests that inflammatory cytokines like IL-6 can directly impact neuron dendrite development such that the complexity of

developing neurons is significantly reduced, similar to what is seen in disorders like schizophrenia [37]. Meyer et al. (2007) demonstrated that poly I:C-induced MIA in pregnant mice that genetically overexpress anti-inflammatory cytokine IL-10 produced offspring lacking the behavioral and cognitive deficits of their wild-type counterparts [24]. Interestingly, recent evidence shows that IL-17a is also sufficient to produce abnormal cortical development and ASD-like phenotypes [25, 38, 39], and that administration of IL-17a-blocking antibody to dams treated with IL-6 or poly I:C prevented the abnormal cortical development and behavioral phenotypes. IL-17a is primarily produced by TH17 cells, which are most abundant in gut-associated tissues. These T helper cells rely on commensal gut microbes for their development [40], specifically segmented filamentous bacteria [41]. Elegant work completed by the same group demonstrates that pretreatment of dams with vancomycin to eliminate gut bacteria reduces TH17 cells and IL-17a production, and prevents the development of aberrant behaviors during MIA [39]. Furthermore, they showed that the presence of segmented filamentous bacteria in the gut was necessary for MIA-induced symptoms in offspring, and that resident dendritic cells aided in the stimulation and differentiation of TH17 cells by releasing IL-1β, IL- 23, and IL-6 in response to poly I:C. IL-6 in particular may be integral to orchestrating the differentiation of these cells, in part by engaging the IL-21 and IL-23 pathways [42]. Taken together, these data suggest that maternal IL-6 signaling likely follows two major pathways to induce fetal neurodevelopmental alterations: (1) direct signaling through placental trophoblasts to increase IL-6 and activated pSTAT3 within the fetal brain, and (2) indirect mechanisms through induction of gut TH17 cells and IL-17a, which can access the fetal brain and alter cortical development. Thus, further work is required to fully delineate the independent and/or cumulative actions of maternal IL-6 and IL-17a on fetal brain development during MIA, specifically under varying gut microbiome compositions and pathogen exposures. Tissues outside the maternal gut and fetal brain should also be taken into consideration, specifically fetal tissues like the placenta (which contains IL-6 receptors utilized in direct IL- 6 signaling) and amniotic fluid. Though this maternal-fetal interface has been largely neglected in the literature thus far, this boundary likely plays a large role in the etiology of brain pathologies induced by MIA [27, 36, 43-47] and should not be ignored. Studies stressing the importance of the gestational timing of MIA challenge [38, 48, 49], the severity of individual maternal responses [33], and the type of maternal immune activator [50] also exist, emphasizing that the different methods used in MIA models should always be taken into consideration when reviewing this literature. For example, MIA induced by poly I:C injection at E12.5 results in specific cortical malformations and aberrant behavioral phenotypes in offspring that are not present if dams

received the same poly I:C injection at E15.5 or E18.5 [38]. Gestational timing (middle vs. late) of MIA induction also differentially modulates offspring neurogenesis, cell apoptosis, and neuroinflammation [48]. As stated above, Kim et al. (2017) have established the importance of particular microbiota and dendritic cell subsets in the guts of dams prior to immune stimulation, which directs both the type and the amount of maternal cytokines that are released during MIA [39]. Others have shown that even within identical murine strains there are high and low maternal immune responders (primarily predicated on production of TNFα), which determines the presence or absence of certain behavioral abnormalities in offspring [33]. Lastly, the type and dose of the immune stimulator understandably dictates the nature and severity of the maternal response and, consequently, the fetal response. There are multiple ways of inducing maternal immune activation, all of which exist in the literature and have varying pros and cons in relation to the practicalities of working with the model (i.e. biosafety) and its translatability. These include live viral or bacterial challenges (such as influenza virus or bacteria involved in periodontal disease), viral or bacterial mimetics (poly I:C and LPS, which have primarily been discussed thus-far), stimulators of local inflammation (such as turpentine), teratogens (like valproic acid), or immune molecules themselves, which include maternal antibodies reactive to fetal brain proteins (immunoglobulin G; IgG) and recombinant cytokines (IL-6 and IL-17a, for example) [50-53]. Though not technically a model of maternal immune activation, researchers also employ intrauterine injections of immune stimulators (like LPS) to induce intrauterine inflammation directly. As a whole, the literature on MIA spans a wide and complex trajectory, requiring vigilant comparisons across models and species in order to parse out characteristic symptomologies and mechanisms that are ubiquitous throughout the literature. One of these mechanisms, as was previously described, is the direct mediation of offspring outcomes by maternal cytokines.

2.2 Microglia in the Context of Maternal Immune Activation Though evidence linking the etiology of MIA-induced abnormalities with maternal cytokines continues to accumulate, less is known about the involvement of microglial cells; nonetheless, these resident CNS immune cells have been implicated in the pathogenesis of multiple neurodevelopmental disorders [16, 54-57]. Microglia, which make up anywhere from 5% to 15% of the population of adult brain cells, derive from erythro-myeloid precursor cells in the yolk sac [58, 59] and colonize the CNS during the first trimester of prenatal development, beginning at the ventricular zone and moving out to the cortical plate [60, 61]. As they progress tangentially across the cortical plate and then radially through the layers,

immature microglia maintain an amoeboid phenotype, which aids in their migration [62]. Though microglia can take on both ramified (resting) and amoeboid (activated or proliferating) morphologies as they mature [63], the majority of these cells retain their amoeboid shape and display markers of activation during prenatal and early postnatal neurodevelopment [60, 64]. This activated status is necessary not only for their own proliferation, but also for phagocytosis of neural precursor cells, neurons, and neuronal synapses, integral functions of healthy neurodevelopmental processes [60, 65, 66]. Indeed, genome-wide profiling of microglia throughout development reveals that they undergo temporal stages of individual expression profiles, governed by distinctive regulatory circuits, in synchrony with, and in regulation of, brain development and homeostasis [67]. Appreciation for the function of microglial cells during healthy CNS development has mounted during the last decade, as evidence demonstrating their integral involvement in neurogenesis, synaptogenesis, myelination, and angiogenesis greatly expanded (see [61, 63, 68] and [69] for review). In addition, though categorization of microglia activation states originally adopted the canonical macrophage polarization hypothesis (i.e. M1 classical or M2 alternative activation states, reviewed in [70]), more recent evidence suggests that this view is much too narrow [71]. We now know that microglial cells could fall anywhere along a spectrum of activation and morphological states [72, 73], and that even resting or “quiescent” microglia are constantly active in surveying their environment [74]. For example, in the adult hippocampus, continuous neurogenesis occurs within the subgranular zone of the dentate gyrus, yet only a small percentage of these newborn cells are ultimately incorporated into the local circuitry, leading to a disproportionate density of apoptotic cells. In this environment, where there is elevated turnover of newborn neurons under normal physiological conditions, “unchallenged” ramified microglia actively and efficiently phagocytose apoptotic neuroprogenitors, while maintaining expression patterns indicative of ‘inactive’ quiescent microglia [75]. Using two-photon imaging, Kato et al. (2016) demonstrate that resting microglia migrate to experimentally stimulated hyperactive axons; ramified microglial processes wrap affected axons, swiftly returning membrane potentials to resting states, and clearing axonal debris. Blocking microglial migration caused neuronal cell death [76], indicating that the actions of ramified ‘resting’ microglia carry out integral neuroprotective and phagocytic functions despite their quiescent phenotype. Indeed, methods for defining microglia functionality based on phenotype and morphology have been adapted in the last decade, such that the classical microglial markers that were previously used are now recognized as insufficient for drawing broad conclusions on cell function [71]. Gene profiling studies have made it possible to describe microglia cells in selective regions of the brain [77], throughout development [61, 67, 78], revealing that these cells have distinct transcriptional and

phenotypic profiles that allow them to sense endogenous ligands and pathogens and take on a more or less “immune-vigilant” state based on brain region [77]. Recent examination of the impacts of eliminating microglia in the developing brain revealed that these cells are essential for establishing brain circuitry, such that removing microglia caused excessive growth and distribution of neurons [79]. Other evidence supports the idea that microglia communicate bi-directionally with neurons, allowing for microglia-regulated neuronal elimination and survival [61]. Cunningham et al. (2013) have shown that induction of MIA in rats causes increased phagocytic activity of microglia, leading to a reduction of neural progenitor cells in fetal brains [60]. Thus, there is sufficient evidence to suggest that a perturbation of microglia cells during development can have widespread consequences on neuronal development and circuitry. Indeed, MIA caused alterations in genome-wide methylation patterns in adult mouse offspring (including distinct modifications in patterns relevant to brain development), suggesting that MIA-induced changes in offspring neurodevelopment are not only widespread but also prolonged [80]. Perhaps one of the best examples of microglial regulation in neurodevelopment is their role as “synaptic strippers” [66]. In 2007, Beth Stevens and colleagues published the first set of comprehensive experiments demonstrating that synapses are eliminated through the classical complement cascade, leading to deposition of C3 opsonizing protein, which tags the synapse for phagocytosis [81]. A later study by Allison Bialas and Beth Stevens (2013) demonstrated that astrocyte-derived TGF-� is necessary and sufficient for the upregulation of C1q in neighboring neurons; importantly, the authors showed that upregulation of synaptic C1q was required for complement-driven phagocytosis by microglia [82]. A few years earlier, Paolicelli et al. (2011) had shown that synaptic remodeling may be in some part dependent on fractalkine signaling, as fractalkine receptor knock-out mice had reduced synaptic pruning and displayed a phenotype of “exuberant immature synapses” [83]. Microglia, the sole CNS cells expressing

CR3 (CD11b/CD18; the receptor for opsonizing protein C3) and fractalkine receptor CX3CR1, are therefore integrally involved in regulating both synaptic pruning and remodeling during development [61, 69]. Recently, it was discovered that astrocyte production of cytokine IL-33, a member of the IL-1 family, signals to microglia in part through IL-1 receptor-like 1 (IL1RL1), encouraging synapse engulfment, and thereby orchestrating circuit function in the developing thalamus and spinal cord [84]. The potential involvement of excessive synaptic pruning by microglia in ASD and schizophrenia pathologies has been suggested more than once [55, 85-88]. Experimental evidence indicates that CX3CR1 knock-out mice display, in addition to disrupted synaptic pruning, decreased functional brain connectivity, social deficits, and increased restricted repetitive behaviors indicative of both disorders [89]. Disruption

of microglial autophagy also leads to excess synapse accumulation, causing decreased sociability in mice [90]. Strikingly, MIA models have also shown that the integrity of synapses and connectivity networks is compromised [79, 91-94]. Most recently, a more modern concept of the neuronal synapse has emerged, referred to as the ‘tetrapartite synapse’, which encompasses all four synaptic components: pre- and postsynaptic elements, glial cells (microglia and astrocytes), and extracellular matrix proteins [95]. Evidence for dysregulation in the function and plasticity of the tetrapartite synapse in neurodevelopmental disorders, specifically schizophrenia, has recently emerged, and new data indicate that these tetrapartite synapse abnormalities occur in discrete cortical and amygdalar regions [96]. Disrupted neurodevelopmental circuitries and synaptic pruning like that presented above can manifest in various ways in the context of neuropathology. In particular, ASD brains are often characterized by altered cortical thickness/layering [97] and cell patterning [98], altered neurogenesis [97, 99], and both over- and under-connectivity (depending on the region [100-102]). Evidence of similar alterations in schizophrenic patients also exists [35, 103, 104]. Specifically, it appears that insufficiencies in interneuron development, which lead to an altered balance of excitation and inhibition, are hallmark to both MIA animal models [105, 106] and human neurodevelopmental disorders [107, 108]. With new research emerging on the action of microglial cells during development, sexual dimorphisms in microglia across developing male and female brains have also become apparent [64, 109]. In particular, males have a higher density of microglia in specific brain regions (like the cortex, hippocampus, and amygdala) and a higher expression of chemokines CCL4 and CCL20, which may make them more vulnerable to prenatal or early postnatal insults [109]. This hypothesis aligns with the male biases seen in the prevalence of both ASD and schizophrenia, and is bolstered by a recent preclinical MIA model showing sex-dependent abnormalities in microglia in the adult cortex and hippocampus [110]. Interestingly, environmental factors such as the gut microbiome have recently been shown to have significant impacts on microglial development [111] and transcriptional phenotype, both of which are regulated in a sex-specific fashion [112]. As this review has alluded to thus far, there is mounting evidence to suggest that microglial cells are likely involved in pathologies relevant to both ASD [56, 57, 113-116] and schizophrenia [55, 103, 117, 118]. The popular theory describing the involvement of microglia in psychiatric illness surrounds the concept of “microglia priming” or “psychological immune memory”, wherein a psychological or immune stressor early in life results in an over-activation of microglia, leading to prolonged hyper-activation in adolescent and adult brains, even after the stressor is removed [16, 119, 120]. However, conclusions drawn from MIA models examining persistent hyper-activation of microglial cells in juvenile and adult

animals are broadly inconclusive, primarily because most studies use different rodent species or strains and/or different MIA induction time points [33, 110, 118, 121-124]. Transcriptional analysis of microglia from MIA-challenged mice revealed that differential expression in developmental genes was much more evident at the early microglia stage compared to the adult stage, suggesting that microglia phenotype might realign to normal by adulthood [67]. Several studies indicate that cellular and inflammatory CNS markers [123], and microglia density and amoeboid morphology [125], are transiently elevated due to MIA but resolve shortly after birth. This is in agreement with clinical data on patients with neuropsychiatric disorders, who do not display heightened inflammatory markers at birth or in early life [126]. Here, it is important to note that the phenomenon of microglia priming relies on the concept of microglial cells being long-lived; early reports suggested that population turnover was slow, though even at the time it was acknowledged that methodological limitations likely resulted in an underestimation of microglia proliferation [127]. Askew et al. (2017) revisited this concept in the last year, discovering that microglial turnover rates in the adult brain are actually much higher than expected, and result in a complete renewal and comprehensive restructuring of the microglial landscape approximately every 96 days [128]. Furthermore, new evidence indicates that microglial renewal varies by location [129], suggesting that there is functional significance in the disparate turnover rates between specific brain regions. Therefore, reevaluation of the theory of microglial-specific “memory” in the context of this new backdrop is required [62]. Consequently, microglia that are influenced by gestational insults are likely to show the most robust responses during the prenatal period. However, few researchers have investigated the activity and phenotypes of microglia immediately following exposure to MIA, and the present literature on this topic is conflicting. Cunningham et al. (2013) found that two i.p. injections of bacterial endotoxin LPS into pregnant rats at E15 and E16 induced heightened expression of inducible nitric oxide synthase (iNOS) and the proinflammatory cytokine IL-1β by fetal microglia, indicative of a phenotypic shift to cytotoxic activation. Notably, this shift towards a more inflammatory state coincided with a significant reduction in neural precursor cells in the proliferative zones of the cerebral cortex, both four and ten days after LPS injections, though there was no change in cell death [60]. A similar LPS MIA paradigm was recently administered in mice (three consecutive maternal LPS injections on E15, 16, and 17), and fetal microglia isolated 3 hr after the last LPS injection displayed increased gene expression of proinflammatory cytokines IL-1β, TNF-α, and IL-6, though no other parameter of microglia activation was assessed [130]. In contrast, Smolders et al. (2015) administered single (E11.5) or double (E11.5 and E15.5) i.p. injections of viral mimetic poly I:C to CX3CR1-eGFP reporter mice and found no differences in microglia density or activation

state (as measured by Mac-2, IL-1β, and iNOS) in the cortex and hippocampus of fetal brains [131]. Pratt et al. (2013), however, demonstrate that poly I:C administered at E12.5 results in increased production of several pro-inflammatory cytokines and chemokines by fetal microglia, though microglia density was unaffected [132]. Though O’Loughlin et al. (2017) showed an increase in amoeboid versus ramified Iba1+ microglia in pre- and postnatal brains after exposure to maternal LPS on E12, only two phenotypes were classified and no other indications of microglia activation were examined [133]. Interestingly, Smolders et al. (2015) also conducted ex vivo studies on embryonic brain slices, treating with IL-6, poly I:C, and LPS; only LPS could directly activate microglia ex vivo, indicating that different modes of maternal immune activation might produce dissimilar effects in vivo [131]. Indeed, when Arsenault et al. (2014) compared i.p. and i.v. injections of LPS to i.v. injections of poly I:C in pregnant mice, only poly I:C delayed reflex development in pups, whereas only LPS reduced expression of GFAP (glial fibrillary protein, a marker for astrocytes) and NeuN (a marker of mature neurons) in fetal brains; notably, this reduction was corrected by postnatal day 10. Intriguingly, both LPS and poly I:C significantly increased protein expression of proinflammatory cytokine TNFα and CD68 (a marker of activated macrophages and microglia) in fetal brains, and again these levels returned to baseline by postnatal day 10 [123]. Thus, it is apparent that the current understanding of pre- and neonatal microglia in the context of MIA is in the nascent stage, and defining the activity of microglial cells during and immediately following induction of maternal infection is therefore essential to delineating the complete mechanisms by which MIA disrupts neurodevelopment and induces psychiatric disorders.

2.3 Social Behavior in the Context of Maternal Immune Activation Of the core behavioral deficits induced by MIA, dysregulated social behavior is one of the most discernable, and most reproducible, across preclinical models [16]. Though clinical evidence of the correlation between maternal infections and neurodevelopmental disorders has existed for some time, the first epidemiological study specifically linking social impairments in ASD patients to chronic immune activation in mothers was published in August of 2017 [134]. Interestingly, the investigators found that only chronic inflammation (i.e. allergies or asthma; primarily TH2-driven immune responses) in mothers, and not autoimmune conditions (primarily TH1/TH17-driven), influenced severity of ASD symptoms. This is intriguing given the recent work in rodent models implicating maternal activation of TH17 cells, and release of IL-17a in particular, in abnormal neurodevelopment and behavioral dysfunction in offspring [25, 38, 39] (discussed below), which does not immediately align with these clinical data. Thus, exploration of the types and origins of social deficits in neuropsychiatric disorders is of particular interest to researchers,

and preclinical and clinical evidence exists for the involvement of specific brain regions like the amygdala [135], hippocampus [136], and prefrontal cortex [137]. The development of the social neural circuit, and the molecular and cellular mechanisms involved in its formation, remain to be fully elucidated, yet recent evidence suggests that microglia activity is directly related to social behavior. Mice deficient in CX3CR1 (causing depletion of microglia) exhibited decreased functional connectivity between the medial prefrontal cortex and the hippocampus, which was significantly correlated with reduced social interest or motivation [89]; this correlation aligns with the literature on patients with autism [100, 101]. Dickerson and Bilkey (2013) reported that a reduction in synchrony along this same connectivity pathway can be induced in rodents by exposure to MIA, mimicking schizophrenia, where synchronization is also thought to be disrupted [93]. As Cunningham et al. (2013) demonstrated, fetal mice exposed to an LPS model of MIA display a pro-inflammatory microglia phenotype that significantly and lastingly reduces the number of cortical neurons [60], indicating that microglia activation, however transient, may have permanent effects on brain development and subsequent behavior, perhaps through altering connectivity and synchronization pathways. Though microglia were not examined by Shin Yim et al. (2017), this group demonstrated that MIA resulted in cortical abnormalities (called cortical patches, defined by a loss of inhibitory neurons) in specific regions of the brain—the primary somatosensory cortex (S1) and the temporal association cortex (TeA)—which were induced by the activation of IL-17 receptor subunit A (IL-17Ra). Stimulating activity in S1 neurons that project to the TeA in wild-type animals was sufficient to recapitulate the deficits in social behavior induced by MIA alone; conversely, inhibition of neural activity along this circuit rescued social deficits in MIA animals [38]. Intriguingly, other behavioral abnormalities (specifically marble burying, a proxy for restricted, repetitive behaviors) were not impacted by stimulating or inhibiting input to the TeA. These data indicate that an imbalance in excitation and inhibition in this discrete neural circuit is likely involved specifically in the disruption of social behaviors in some MIA models, though how IL-17Ra activation leads to cortical patches, and the cell type involved in this mechanism, has not yet been defined. Interestingly, microglia, which express IL-17Ra [138] and are responsive to IL-17 in vitro [139], have been shown to increase their proliferation, trafficking, and activation status in response to sustained production of IL-17a during pathological states [140]; however, further research is needed to determine whether MIA-induced IL-17a would produce a similar response in microglia, and whether or not microglia cells are involved in the observed loss of inhibitory neurons in the S1 or other brain regions. Other compelling evidence of immune regulation in social behavior comes from manipulation of the newly characterized meningeal/glymphatic immune system [141, 142]. As general immune

dysregulation is common in neurological disorders like ASD and schizophrenia (summarized above), meningeal immunity was investigated as a potential driver of social deficits. Interestingly, depleting meningeal CD4+ T cells that produce IFNγ, a pro-inflammatory cytokine that typically stimulates classical macrophage/microglia polarization [143], resulted in social deficits and hyper-connectivity in mice, both of which could be rescued through lymphocyte repopulation. Depletion and administration of IFNγ alone reproduced this phenotype, suggesting that cells expressing IFNγ receptors are driving these responses. Though gene expression and protein analyses revealed that both neurons and microglia express R1 and R2 IFNγ-receptor subunits, it was inhibitory neurons, not microglia, that were responsive to IFNγ from meningeal T cells [142], emphasizing that a disruption in excitatory and inhibitory tone can be a primary motivator of social deficits, as is seen in MIA models [38].

2.4 Summary Microglia colonize the embryonic brain during early neurodevelopment, establishing an environment conducive to proper neural and glial network organization, of which they are integral mediators. Inhibition, depletion, or ablation of these cells during development is detrimental to neurogenesis and to the formation of proper neural networks, as well as to synaptogenesis and pruning. Animal models of maternal immune activation suggest that embryonic microglia are activated in response to maternal inflammation, which alters their ability to properly perform integral neurotrophic and phagocytic tasks. This is a likely mechanism through which MIA leads to behavioral abnormalities reminiscent of psychiatric disorders like ASD and schizophrenia. Aberrant regulation of immune responses within the brain, including depletion of microglia, has been shown to alter social behaviors in preclinical models, suggesting that altered fetal immune responses during MIA is primary to the etiology of idiosyncrasies in social behavior. Thus, defining the actions of microglia in the context of MIA is vital to developing therapeutic interventions that may be applied prenatally. Herein, we investigate how MIA impacts microglia, neurodevelopment, and behavior in a highly-translatable gyrencephalic swine model.

CHAPTER 3: MATERNAL VIRAL INFECTION DURING PREGNANCY ELICITS ANTI-SOCIAL BEHAVIOR IN NEONATAL PIGLET OFFSPRING INDEPENDENT OF POSTNATAL MICROGLIAL CELL ACTIVATION1

3.1 Abstract Maternal infection during pregnancy increases risk for neurodevelopmental disorders and reduced stress resilience in offspring, but the mechanisms are not fully understood. We hypothesized that piglets born from gilts infected with a respiratory virus during late gestation would exhibit aberrant microglia activity, cognitive deficits and reduced sociability. Pregnant gilts were inoculated with porcine

5 reproductive and respiratory syndrome virus (PRRSV; 5 x 10 TCID50 of live PRRSV) or saline at gestational day 76. Gilts infected with PRRSV exhibited fever (p < 0.01) and reduced appetite (p < 0.001) for 2 weeks post-inoculation and were PRRSV-positive at parturition. Piglets born from infected and control gilts were weaned at postnatal day (PD) 1 and assigned to two groups. Group 1 was challenged with lipopolysaccharide (LPS, 5 µg/kg body weight i.p.) or saline on PD 14 and tissues were collected. Group 2 was tested in a T-maze task to assess spatial learning and in a 3-chamber arena with unfamiliar conspecifics to assess social behavior from PD 14-27. Microglia (CD11b+ CD45low) isolated from Group 2 piglets at PD 28 were challenged ex vivo with LPS; a subset of cells was analyzed for MHCII expression. Maternal infection did not affect offspring circulating TNFα, IL-10, or cortisol levels basally or 4 h post-LPS challenge. While performance in the T-maze task was not affected by maternal infection, both sociability and preference for social novelty were decreased in piglets from infected gilts. There was no effect of maternal infection on microglial MHCII expression or LPS-induced cytokine production. Taken together, these results suggest the reduced social behavior elicited by maternal infection is not due to aberrant microglia activity postnatally.

Keywords: prenatal, maternal immune activation, microglia, cytokines, social behavior, piglet

1Reprinted from Brain, Behavior, and Immunity journal: Antonson, A. M., Radlowski, E. C., Lawson, M. A., Rytych, J. L., & Johnson, R. W. (2017). Maternal viral infection during pregnancy elicits anti-social behavior in neonatal piglet offspring independent of postnatal microglial cell activation. Brain Behav Immun, 59, 300-312. doi:10.1016/j.bbi.2016.09.019

3.2 Introduction Maternal infection during pregnancy is associated with increased risk for development of neuropsychiatric disorders, such as autism and schizophrenia, in offspring [1, 144]. Studies in animal models suggest that during maternal immune activation (MIA), maternally-derived cytokines cross the placenta and affect fetal brain development [24, 29, 30]. Pregnant mice administered the viral mimetic polyinosinic:polycytidylic acid (poly I:C) at mid gestation give birth to offspring that display reduced social behavior, disruptions in vocalizations, and stereotypic behavior [31]. Evidence of mediation by maternal cytokines comes from studies where pregnant dams were injected with recombinant IL-6 at E12.5, resulting in offspring with behavioral abnormalities similar to what is observed in viral infection, poly I:C or LPS models [26]. Furthermore, immunoneutralization of IL-6 in pregnant mice administered poly I:C at E12.5 normalized behavior of offspring [26]. Other investigations revealed that direct injection of IL-17a into the fetal brain was sufficient to produce abnormal cortical development and autism-like phenotypes, while IL-6 was not. Additionally, ASD-like phenotypes produced either from maternal IL-6 or poly I:C administration could be prevented by maternal pretreatment with IL-17a-blocking antibody, suggesting that MIA or maternal IL-6-induced abnormal development is dependent on IL-17a [25]. Exactly how maternal cytokines affect fetal brain development is still unclear but microglia, the resident immune cells of the brain, have been implicated in the neuroimmune pathogenesis of neurodevelopmental disorders (Onore, Careaga, and Ashwood, 2012). Microglia are derived from progenitor cells in the yolk sac [58] and colonize proliferative zones of the neocortex during the first trimester of prenatal development [60]. Fetal microglia display both ramified (resting) and amoeboid (activated or proliferating) morphologies and participate in brain development through phagocytosis of neural precursor cells, neurons, and neuronal synapses [60, 65, 66]. Fetal microglia express many of the typical macrophage markers and express inflammatory genes associated with both classical pro- inflammatory and alternative anti-inflammatory phenotypes [60]. Fetal mice exposed to the LPS model of MIA display a pro-inflammatory microglia phenotype that results in a significant reduction in cortical neurons that persists postnatally [60]. However, fetal mice exposed to the poly I:C model of MIA do not display an activated microglia phenotype or increased microglia density in the cortex and hippocampus [131]. Thus, though it has been suggested that MIA activates fetal microglia and alters their phenotype long term, resulting in prolonged neuroinflammation [16, 60], consistent evidence in support of this hypothesis is lacking. Proliferation and priming of microglia in response to pro-inflammatory stimuli can also be modulated by glucocorticoids (GCs), which can act as endogenous alarmins, or danger signals [145-147].

Heightened hippocampal GC levels and GC receptor activation can lead to chronically sensitized microglia, like that seen in aging, a phenomenon that is also linked to learning and memory deficits [147]. Rodent models of MIA have revealed alterations in hypothalamic-pituitary-adrenal (HPA) function [148, 149] and increased anxiety-like behavior in adult offspring [150-152]. The long-term negative effects of heightened maternal GCs on offspring HPA development and function, immune function, and behavior have garnered significant research interest [153-155]. As maternal GCs can cross the placenta [156] and blood-brain- barrier [157], exposure to heightened levels of maternal GCs poses a direct threat to the developing fetus [158, 159], and likely contributes to the fetal microglia response to MIA. We developed a prenatal MIA paradigm in swine utilizing porcine reproductive and respiratory syndrome virus (PRRSV) during late gestation at a time when the fetal pig brain undergoes a dramatic growth spurt, similar to human neurodevelopment [13-15]. The domestic pig is a precocious, gyrencephalic species whose brain anatomy, neurochemistry, and growth and development trajectories correspond closely to humans in prenatal and early postnatal life [12, 23]. Thus, we sought to extend current findings on MIA in a highly translatable pig model. As data demonstrating prolonged over- activation of microglia in offspring due to MIA are lacking, and recent evidence indicates that prolonged microglia anomalies may not occur in some MIA models [121, 122], we aimed to assess microglia activation status in prenatally challenged neonatal piglets. We hypothesized that maternal infection with PRRSV would lead to aberrant microglia activity in offspring, resulting in altered cognitive and social behaviors in early life. We further postulated that maternal infection would alter fetal HPA development and lead to GC resistance in microglia. Here, we show that piglets born from infected mothers display anti-social behaviors and a decreased preference for social novelty, in the absence of overt microglia activation and GC desensitization.

3.3 Materials and Methods

3.3.1 Animals and Experimental Design The experimental design is illustrated in Fig. 3.1. Crossbred pregnant gilts (PRRSV-free and not vaccinated), artificially inseminated with semen from the same boar (PIC 359 SS 6278, Birchwood Genetics, Inc., West Manchester, OH), were brought from the University of Illinois swine herd into the biomedical animal facility at gestational day (GD) 69. Gilts were individually housed in identical disease containment chambers kept at 22°C and maintained on a 12 h light/dark cycle. Gilts were provided 2.3 kg of a standard corn-soybean meal-based gestation diet daily, with ad libitum access to water. Rectal temperature and food intake were monitored daily. If food intake decreased, gilts were provided

supplemental corn syrup and/or tap water mixed in the diet until food intake returned to normal. Three identical trials were conducted, each consisting of one control gilt and two MIA gilts. One gilt in the MIA treatment group spontaneously aborted. Thus, the total number of gilts across the three trials was eight, three of which were control, five of which were MIA. On GD 76, gilts in the MIA treatment group were inoculated intranasal with 5mL of 1 � 10 50% tissue culture infectious dose (TCID 50) of live PRRSV (strain P-129-BV), obtained from the School of Veterinary Medicine at Purdue University (West Lafayette, Indiana). Control gilts received the same volume of sterile DMEM. PRRSV is an enveloped single-stranded RNA virus which causes interstitial pneumonia by infecting alveolar macrophages [160], and results in increased secretion of IL-1β, IL-6, IL- 10, and TNFα [161-163]. Gilts were inoculated on GD 76 to coincide with the onset of accelerated fetal brain growth; this gestational day corresponds to approximately the beginning of the third trimester in humans [14, 15, 164, 165]. In a separate but identical study with this MIA model, blood was collected from the marginal ear vein of control and PRRSV gilts once weekly from GD 76 to GD 104 (Control: n = 9, PRRSV: n = 6). Heparinized blood was centrifuged (1,300 x g at 4° C for 15 min) and plasma was collected and stored at -80° C until analyses of TNFα and IL-6. To avoid stress during pregnancy, for the present study saliva was collected from gilts once per week by allowing them to chew on cotton ropes, similar to what is described by Cook et al. [166]. Saliva was collected from the rope and frozen at -80° C until analysis for PRRSV. On GD 104 gilts were moved into standard farrowing crates. On GD113, 10 mg of lutalyse (Pfizer, New York, NY) was given intramuscularly to induce parturition. Approximately 24 h after birth [age postnatal day (PD) 1], all piglets were processed as follows: body weight and rectal temperature were recorded; blood was collected from the external jugular vein to confirm presence/absence of PRRSV; and intramuscular injections of iron dextran (100 mg/pig, Butler Schein Animal Health, Dublin, OH) and penicillin (60 kU/pig, Butler Schein Animal Health, Dublin, OH) were administered. Within 12 h of processing, piglets were moved to separate biomedical containment chambers and into individual cages. Housing, handling, and feeding of piglets was performed as previously described [167], with some modifications. Briefly, piglets born from PRRSV-infected and control gilts were housed in separate containment chambers as a precautionary measure to prevent unintended infections. Piglets from each litter were divided (based on sex and body weight) to form two groups, one designated to be sacrificed at PD 14 (Group 1), and another at PD 28 (Group 2; Supplemental Table S3.1). Each piglet was provided a plastic ball for enrichment (Jingle Ball, Bio-Serv, Flemington, NJ) and an electric heating

pad (K&H Lectro-Kennel Heat Pad, K&H Manufacturing, Colorado Springs, CO). Room temperature was maintained at 27° C and chambers were maintained on a 13 h light/11 h dark cycle (07:00 h lights on, 20:00 h lights off). Piglets were supplied a nutritionally complete commercial sow milk replacer (Advance Liqui- Wean, Milk Specialties, Eden Prairie, MN). Powdered milk replacer was reconstituted fresh each morning (206 g/L tap water) and provided at a rate of 300 mL/kg body weight. Milk replacer was delivered using an automated feeding system, like the one described previously [168], which provided the daily allotment of milk as 18 individual meals (one per hour) from 08:00 h to 02:00 h. Water, aside from that used in the milk replacer, was not provided. Daily body weights and rectal temperatures were recorded. Six age-matched piglets (3 male, 3 female) were obtained from the University of Illinois swine herd at PD 2 to serve as novel piglets for social behavior testing (methods described below). All housing, handling, and feeding was conducted as described above. These piglets were kept separate and no rectal temperatures were recorded. On PD 14, 5 µg/kg body weight LPS (E. coli K-235, Sigma-Aldrich, St. Louis, MO) or an equivalent volume of sterile PBS was administered intraperitoneally to piglets in Group 1. Previous work in our lab has shown this dose of LPS induces an increase in plasma TNFα and cortisol in pigs that peaks 4 h post- injection [169]. Rectal temperatures at time of injection and 4 h post-injection were recorded. Blood was collected from the external jugular vein within 2 min to minimize elevation of cortisol due to sampling. Blood was allowed to clot at room temperature for ~30 min, centrifuged (1,300 x g at 4° C for 15 min) and serum was collected and stored at -80° C until analysis. . Immediately following, Group 1 piglets were sacrificed. Additional cardiac blood was collected into heparinized tubes during sacrifice for plasma cytokine analysis. Blood was placed immediately on ice, centrifuged (1,300 x g at 4° C for 15 min) and plasma was collected and stored at -80° C until analysis. On PD 14 (±1 d), Group 2 piglets began 11 days of cognitive testing using a spatial T-maze task (habituation starting D11 ±1 d). Piglets were then given a 1 d rest period before being assessed in a social behavior test on PD 27 (±1 d). On PD 28 (±1 d) between 12:00 h and 13:00 h, blood was collected by jugular venipuncture for baseline cortisol detection. Piglets were removed from their cages and blood was drawn as before within 2 min to minimize elevation of cortisol due to sampling. Serum was collected and stored at -80° C until analysis. Immediately following blood collection, Group 2 piglets were sacrificed. Hippocampal and hypothalamic tissue were placed directly into GentleMACS C-tubes with neural tissue

dissociation enzymes (Neural Tissue Dissociation Kit (P), Miltenyi Biotec, San Diego, CA) for isolation of CD11b+ cells, flow cytometry, and stimulation with LPS and dexamethasone (DEX). For euthanasia, piglets were anesthetized using a telazol:ketamine:xylazine drug cocktail (50 mg of tiletamine plus 50 mg of zolazepam reconstituted with 2.5 mL ketamine (100 g/L) and 2.5 mL xylazine (100 g/L); Fort Dodge Animal Health, Fort Dodge, IA) administered intramuscularly at 0.03mL/kg body weight. Piglets were then euthanized via intracardiac injection of sodium pentobarbital (86 mg/kg body weight; Fatal Plus, Vortech Pharmaceuticals, Dearborn, MI). All animal care and experimental procedures were in accordance with the National Research Council Guide for the Care and Use of Laboratory Animals and approved by the University of Illinois Institutional Animal Care and Use Committee.

3.3.2 Spatial Learning and Memory The spatial learning and memory T-maze task was performed as previously described [170]. The clear Plexiglas plus-shaped maze consisted of four arms, two start arms (north and south) and two reward arms (east and west), with four different extra-maze visual cues. Start arms were pseudorandomly alternated so that piglets had to use an allocentric (hippocampus-dependent) mechanism to solve the task. Testing was conducted on 11 consecutive days from 9:00 h to 13:00 h by the same trained experimenter. Piglets performed 10 trials per day, each 60 s long. The first 6 d of testing represented the acquisition phase wherein piglets learned the direction and location of a chocolate milk reward (the same milk replacer used for regular feedings with the addition of Nesquik cocoa powder) using the visual cues. To balance olfactory stimuli, chocolate milk was placed in both the correct and incorrect reward bowls but was only accessible from the correct bowl. To prevent satiation during testing, a small amount (3 mL) of chocolate milk was provided at each trial. Once the acquisition criterion of 80% (8 out of 10 trials) correct was reached, the location of the reward bowl was switched (the reversal phase). During reversal (5 d), piglets performed the task as before until 80% criteria was reached once more. Throughout testing, piglet movement was recorded by a video camera mounted above the maze and tracked live using commercially available software (EthoVision 8.5; Noldus Information Technology). Three control piglets were excluded from analyses due to non-compliance in the task (defined as either failure to make a choice within the 60 s time limit for multiple trials or complete lack of motivation to consume the milk reward, across three consecutive days of testing).

3.3.3 Social Behavior Testing The sociability test was similar to that used with rodents to identify social approach deficits [171, 172]. The testing arena was made of Plexiglas with two dividing walls, each with a removable sliding door

that created three equal size chambers measuring 122 cm L x 81 cm W x 91 cm H with doors 30 cm W x 61 cm H. One trial consisted of a 5 min habituation period followed by two consecutive 10 min test periods. The first period was a standard “sociability” test, and the second period was a “preference for social novelty” test. Each piglet went through one trial and was exposed to one male and one female novel piglet (in random order). The testing arena was cleaned with 10% bleach between each trial. During the 5 min habituation period, the test piglet was placed in the middle compartment and the doors were raised, allowing access to all three chambers. After 5 min, the test piglet was returned to the middle compartment and the doors were closed. Two clean pet crates (61 cm, MidWest iCrate Pet Crate, Mid-West® Homes for Pets, Muncie, Indiana), which allowed for visual, olfactory, auditory, and tactile communication between piglets, were placed in the far upper corner of each outside compartment. A novel piglet (Stranger Pig 1) was placed in one crate, then the compartment doors were opened. This first phase comprised the Sociability test, with one novel object (an empty crate) and one novel piglet. After 10 min, the test piglet was placed back into the center compartment and the doors were closed. A second novel piglet was placed into the previously empty crate, then the doors were opened. This second 10 min phase comprised the test of Preference for Social Novelty, with a previously novel piglet (previously Stranger Pig 1, now Pig 1) and a new novel piglet (Stranger Pig 2). Six age-matched piglets were used as novel piglets (see above) and underwent extensive habituation to the pet crates during the weeks leading up to testing. The compartment containing Stranger Pig 1, in addition to the sex of that novel piglet, was randomly alternated throughout testing. The movement of the test piglet was recorded by a video camera mounted above the arena and tracked live using commercially available software (EthoVision 8.5, Noldus Information Technology). One control and one PRRSV piglet had to be excluded from analyses due to improper automated tracking. Exploratory/social behaviors—defined as piglet’s nose in contact with the crate—were scored by an individual blinded to treatment using commercially available software (The Observer XT 11.5, Noldus Information Technology).

3.3.4 Microglial Cell Isolation Microglia were isolated from the hippocampus and hypothalamus based on positive expression of CD11b. Cell isolation was performed using Miltenyi Biotec GentleMACS C-tubes, Neural Tissue Dissociation Kits (P), GentleMACS Octo Dissociator with Heaters, CD11b (microglia) MicroBeads, MACS MS columns, and OctoMACS separator (Miltenyi Biotec, San Diego, CA) according to manufacturer’s instructions, with some modifications. Briefly, tissue was placed directly into C-tubes containing Enzyme

Mix 1 & 2 [NTDK (P)]; volumes of enzyme mix were doubled for each hemisphere of hippocampal tissue. C-tubes were incubated at room temperature for ~20 min before being placed directly onto GentleMACS Octo Dissociator with Heaters. Falcon cell strainers (40 µm) were used in place of the 70 µm strainers listed. Myelin removal was achieved by centrifugation with a 30% Percoll PLUS (GE Healthcare Life Sciences, Pittsburgh, PA) solution in PBS. Sample was incubated with CD11b (microglia) MicroBeads, then passed through MS columns and collected. The final CD11b+ fraction was centrifuged at 300 x g for 10 min at 4° C. All hippocampus samples and half of the hypothalamic samples were used in cell culture, while the other half of the hypothalamic samples were used for flow cytometry.

3.3.5 Primary Microglial Cells Stimulated with LPS and DEX CD11b+ cells were resuspended in media and warmed to 37° C. Cells from the hypothalamus and from the left hemisphere of the hippocampus were resuspended in standard cell culture medium [DMEM with 100 U/mL penicillin/streptomycin, 10% FBS, and 1 µL/mL porcine rpGM-CSF (R&D Systems, Minneapolis, MN)]. Cells from the right hemisphere of the hippocampus were resuspended in standard medium plus dexamethasone (DEX, 3.18 µM, Sigma-Aldrich, St. Louis, MO). DEX, a synthetic GC, binds glucocorticoid receptors [173] and modulates inflammatory gene expression by inhibiting NF-κB. The dose of DEX was chosen based on a prior study showing it achieves 50% inhibition of in vitro LPS- stimulated cytokine production in the whole blood of patients with glucocorticoid resistance [174]. Samples were immediately plated in 6-well culture plates and stimulated with 10 ng LPS (E. coli, 0127:B8, Sigma-Aldrich, St. Louis, MO) or sterile PBS. Cell supernatant was collected 4 h later for cytokine analysis and cells were placed directly into 1 mL TRIzol® Reagent (ThermoFisher Scientific, Grand Island, NY) to collect total RNA.

3.3.6 Flow Cytometry Hypothalamic CD11b+ cells were resuspended in flow buffer [PBS with 1% BSA (ThermoFisher Scientific), 0.1% sodium azide (Sigma-Aldrich), and 20 mM glucose (Sigma-Aldrich)]. Competitive binding of the Fc receptor was blocked with purified CD16/CD32 antibodies (eBioscience, San Diego, CA). Cells were incubated with CD45 antibodies (AbD Serotec, Raleigh, NC) to distinguish microglial cells from monocytes, and with MHCII (Antibodies Online, Atlanta, GA) as a marker of phagocytic cell activation. Cells were flowed through the FACS Aria II flow cytometer (BD Biosciences, San Jose, CA) and gated based on the forward- and side-scatter and autofluorescence properties of a massed unstained control sample.

Cells were confirmed as CD11b+ porcine microglia cells by intermediate expression of CD45 (as previously described, Elmore et al. 2014).

3.3.7 Real-time PCR Total RNA from CD11b+ cells was isolated following the Tri Reagent protocol (Sigma-Aldrich). A High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Grand Island, NY) was used for cDNA synthesis according to manufacturer’s protocol. The Applied Biosystems TaqMan Gene Expression Assay protocol was used for performing quantitative real-time PCR. Genes of interest (Supplemental Table S3.2) were compared against, and expressed as fold change relative to, the endogenous control (GAPDH, Ss0337854_g1).

3.3.8 PRRSV, Cortisol, and Cytokine Detection For PRRSV detection, RT-PCR was performed by the Veterinary Diagnostic Laboratory (University of Illinois, Urbana, Illinois). Serum cortisol was analyzed using a commercial competitive binding ELISA kit according to manufacturer’s instructions (Cayman Chemical, Ann Arbor, MI). Plasma cytokines (TNFα, IL- 6, IL-1B, IL-8, and IL-10) and cell culture media TNFα were analyzed using porcine-specific DuoSet or Quantikine ELISA kits (R&D systems, Minneapolis, MN) according to manufacturer’s instructions.

3.3.9 Statistics Statistical analysis was performed using Statistical Analysis Software (SAS Institute, Cary, NC) or GraphPad Prism (GraphPad Software, Inc., La Jolla, CA) with α set at 0.05. Individual piglets were treated as an experimental unit as each piglet has an individual fetal system that allows for significantly different responses to maternal infection within a litter [175]. All data were subjected to analysis of variance (ANOVA) to reveal main effects and interactions, except the flow cytometry data and the litter characteristics data (Supplemental Table S3.3), which were analyzed using a Student’s t test (two-tailed). Where significant main effects or interactions were found, Tukey or Bonferroni post-hoc test results are displayed in each figure, using letters indicating significant means separations (significantly different means do not share the same letter) or asterisks representing significant group comparisons, respectively. The current study was not designed to reveal significant sex effects and is therefore underpowered for these analyses. Even so, all initial ANOVAs were run with sex included in the model. We found that there was no significant effect of sex in any of the analyses, so this variable was removed from all models. Gilt data were analyzed using 2-way (treatment x time) ANOVAs. Data from the in vivo LPS challenge were analyzed using 2-way ANOVAs (prenatal treatment x LPS). Three-way factorial ANOVAs [prenatal

treatment x in vitro LPS treatment x in vitro DEX treatment] were used to analyze all gene expression and protein data from primary hippocampal CD11b+ cell cultures, whereas 2-way (prenatal treatment x in vitro treatment) ANOVAs were used to analyze primary hypothalamic CD11b+ cell culture data. The spatial T-maze task data were analyzed using a 2-way ANOVA (prenatal treatment x time). Within-group 1-way ANOVAs with Bonferroni post-hoc tests were used to determine chamber preferences in both social behavior tests, whereas time spent actively investigating was analyzed using 2-way (prenatal treatment x chamber) ANOVAs. Two-way ANOVAs (prenatal treatment x chamber) with a Tukey post-hoc analysis were also used, to determine chamber preference differences between treatment groups in both social behavior tests. All data points meeting outlier criteria (more than three standard deviations from the mean) were excluded from analyses.

3.4 Results

3.4.1 Maternal PRRSV infection Saliva samples collected immediately prior to inoculation and tested for presence of PRRSV were negative. Samples collected 7 d post-inoculation from infected gilts tested positive for PRRSV (control gilts were negative). In a separate but similar study, pregnant gilts infected with PRRSV had increased plasma TNFα on GD 83, 90, and 97 (7, 14, and 21 d post-inoculation, respectively; treatment x time, p < 0.0001; Fig. 3.2A). Plasma IL-6 was below detectable levels for most samples, thus these data could not be analyzed or presented. To avoid restraint stress in the current study, weekly blood samples were not collected. Nonetheless, pregnant gilts had increased core body temperature from GD 82-84 (6-8 d post- inoculation; treatment x time, p < 0.01; Fig. 3.2B), and decreased food intake from GD 83-87 (7-11 d post- inoculation; treatment x time, p < 0.0001; Fig. 3.2C). Litters born from PRRSV infected mothers were similar in size, average piglet birth weight and number of stillborn piglets, as litters born from non-infected gilts (p > 0.10; Supplemental Table S3.3). Except for two piglets born from the same PRRSV-infected gilt, piglets from both infected and un-infected gilts tested negative for PRRSV postnatally. The two piglets that tested positive for PRRSV were excluded from all analyses. Piglet brain weight to body weight ratios did not differ between treatment groups at PD 14 (± 1 d; p > 0.10) or PD 28 (± 1 d; p > 0.10; data not shown).

3.4.2 Piglets born from PRRSV-infected mothers had similar circulating cytokines and cortisol in response to LPS treatment.

Baseline serum cortisol determined at PD 28 (± 1 d) was not different between treatment groups (data not shown). Intraperitoneal injection of LPS at PD 14 (± 1 d) resulted in higher body temperature (p < 0.01) and serum cortisol (p < 0.05) compared to saline injection regardless of maternal treatment group (Figs. 3.3A and 3.3B). Plasma concentrations of TNFα, IL-10, IL-1β, and IL-8 were higher in LPS-treated animals compared to saline treated controls (p < 0.0001, p < 0.05, p < 0.01, and p < 0.05 respectively), with no effect of maternal treatment (Fig. 3.3C). Plasma IL-6 was below the detectable limit.

3.4.3 Maternal infection did not alter MHC II in hypothalamic microglia in piglets. Microglia isolated from the hypothalamus of PD 28 (± 1 d) piglets were stained for CD45, to differentiate microglial cells from infiltrating monocytes, and MHC II, to indicate phagocytic activation. Enriched CD11b+ microglia expressed CD45 at low levels indicating that they were microglia and not monocytes [145, 176, 177]. Flow cytometric analysis revealed that microglia expressed MHC II similarly in control piglets (Fig. 3.4A) and piglets from PRRSV-infected gilts (Fig. 3.4B). Representative scatter plots are shown. Percent of microglia with cell surface expression of MHC II was not different between maternal treatment groups (p > 0.10; Fig. 3.4C). In a separate but identical study, similar results were found with hippocampal microglia (data not shown).

3.4 Hippocampal and hypothalamic microglia were responsive to LPS and DEX, and were not impacted by maternal treatment. To investigate whether maternal infection affected microglia sensitivity in the offspring and if this was associated with glucocorticoid resistance, microglia were isolated from the hippocampus and hypothalamus of PD 28 (± 1 d) piglets and stimulated with LPS (10 ng/mL) ± DEX (3.18 µM). Expectedly, LPS treatment increased steady-state mRNA levels of TNFα, IL-6, IL-10, and NR3C1 (glucocorticoid receptor; GR) in hippocampal microglia and these increases were partially attenuated by DEX treatment (TNFα: LPS x DEX, p < 0.05; IL-6: main effect of LPS, p < 0.001, main effect of DEX, p < 0.05, LPS x DEX, p = 0.085; IL-10: LPS x DEX, p < 0.05; NR3C1: LPS x DEX, p < 0.05; Figs. 3.5A-D). LPS treatment also increased steady-state mRNA levels of TNFα, IL-6, and NR3C1, and tended to increase mRNA levels of IL-10, in hypothalamic microglia (TNFα: main effect of LPS, p < 0.05; IL-6: main effect of LPS, p < 0.05; NR3C1: main effect of LPS, p < 0.05; IL-10: trending main effect of LPS, p = 0.079; Fig. 3.6A). Maternal treatment did not affect TNFα, IL-6, IL-10, or NR3C1 relative mRNA expression (p > 0.10 for all genes) for either hippocampal or hypothalamic microglia. The same LPS x DEX interaction (p < 0.001) was evident in TNFα protein concentrations measured in cell culture media from hippocampal microglia, such that the increase in TNFα concentration due to LPS was attenuated by DEX (Fig. 3.5E); LPS also increased TNFα protein

concentrations in media from hypothalamic microglia (p < 0.001, Fig. 3.6B). DEX was not used as a treatment for hypothalamic microglia because GR regulation in the pig hippocampus has been proven to be more sensitive to stressors than the hypothalamus, potentially because GR level in the pig hypothalamus is low [173].

3.4.5 Maternal infection did not alter spatial learning but did decrease social behavior. 3.4.5.1 Learning and Memory Piglets from control and PRRSV-infected gilts were tested in a spatial T-maze task to assess learning and memory function beginning at PD 14 (± 1 d). Piglets from both maternal treatment groups displayed similar learning as both groups of piglets displayed similar performance during the acquisition phase of the testing. Similarly, both groups of piglets had similar cognitive flexibility during the reversal phase of the testing paradigm as both groups reached learning criteria (80% correct trials) for the new reward location on day 5 of reversal testing (main effect of time, p < 0.0001; treatment x time, p > 0.10; Fig. 3.7). 3.4.5.2 Sociability Piglets from control and PRRSV-infected gilts were placed in a 3-chamber social behavior testing paradigm on PD 27 (± 1 d). Total time spent in each compartment throughout the 10 min test is represented graphically (Fig. 3.8B). Piglets born from PRRSV-infected gilts displayed abnormal social behaviors compared to control piglets (treatment x chamber, p < 0.01). Control piglets spent more time within the compartment with the novel piglet (“Stranger Pig 1”) compared to the center compartment (“center”) or the compartment with the empty crate (“empty”; stranger pig 1 vs center, p < 0.001, stranger pig 1 vs empty, p < 0.05; Fig. 3.8B). In contrast, piglets born from PRRSV-infected gilts spent similar time in each of the compartments (p > 0.10; Fig. 3.8B). PRRSV piglets spent numerically less total time in the compartment with Stranger Pig 1 compared to controls and tended to spend more time alone in the empty compartment (p = 0.07, Fig. 3.8B). The total time piglets spent with their nose in contact with either crate while in the corresponding compartment was also calculated to estimate “active investigation”. Duration of active investigation of both the stranger pig and the empty crate was similar between treatment groups (p > 0.10; Fig. 3.8C). 3.4.5.3 Social Novelty To test piglets’ preference for social novelty, a second novel piglet was added to the previously empty crate and the same piglets were again exposed to the social testing paradigm. Piglets born from PRRSV-infected gilts failed to display a preference for social novelty (treatment x chamber, p < 0.01).

Control piglets spent more time in the compartment with the novel piglet (“Stranger Pig 2”) than the compartment with the familiar piglet (“Pig 1”; p < 0.001; Fig. 3.9A), demonstrating a preference for the novel piglet. However, piglets born from PRRSV-infected gilts spent approximately equal time in all three compartments (p > 0.10; Fig. 3.9A). Piglets born from PRRSV-infected gilts also tended to spend less time actively investigating Stranger Pig 2 (p = 0.09) while in that compartment compared to control piglets (Fig. 3.9B).

3.5 Discussion We developed a novel paradigm in swine to investigate the effects of maternal viral infection on neonatal piglet cognition and microglia activity. The present study demonstrated that maternal infection with PRRSV in late gestation altered social behavior in the offspring, absent of deficits in learning and memory, aberrant microglial cell activity, and GC desensitization in microglia. To our best knowledge, this is the first study to demonstrate aberrant social behaviors in neonatal piglets due to maternal infection. Importantly, the change in social behavior caused by MIA was not related to aberrant microglia activity postnatally. Inoculation of late-gestation pregnant gilts with live PRRSV virus resulted in transient elevations in plasma TNFα and body temperature and decreases in food intake. Though our model induces a classic inflammatory response in gilts during a critical window of fetal neurodevelopment, piglets born from PRRSV-infected gilts did not display a pro-inflammatory phenotype at PD 14 or PD 28. Our results are in agreement with recent data from a rodent model of MIA [121, 122]. However, this does not exclude the possibility that the fetal neuroimmune system undergoes transient activation during peak maternal infection, and a secondary insult during adolescence or early adulthood might be necessary for revealing latent consequences of prenatal challenge. When Giovanoli et al. (2013) administered a secondary insult during puberty, previously undetectable consequences of MIA were revealed [178]. Straley et al. reported differences in fetal dopamine neuron numbers two days after maternal LPS treatment at E16, but these differences were no longer apparent postnatally, suggesting that there are direct, yet transient, effects of MIA on dopamine neurons that appear to be resolved by P9 [179]. These authors also suggested that food deprivation, administered as part of a reward paradigm test during adolescence, may have acted as a stressor that altered AMPH response in adulthood [179]. Recent clinical data on patients with neuropsychiatric disorders support the absence of heightened inflammatory markers at birth or in early life [126]. Additionally, the transition to more severe psychiatric symptoms later in life results in measurable increases in both pro- and anti-inflammatory factors [180, 181] and circulating GCs [182-185].

Garay et al. (2013) found that although cytokine levels in the frontal and cingulate cortex of mice exposed to MIA are transiently elevated at birth, these levels decrease postnatally, and then elevate again during adulthood [186]. However, we were unable to detect an exaggerated immune or GC response to a secondary challenge of LPS in vivo (Group 1) or in vitro (Group 2 microglia) in prenatally challenged neonatal piglets. It should also be noted, however, that neonatal piglets challenged with LPS in vivo (Group 1) were not tested for altered behaviors post-challenge. It is possible that maturation of the immune and endocrine systems is necessary for shifting the immune phenotype of prenatally challenged offspring to a primed state [187-190]. Indeed, mounting evidence suggests that there is a critical window of neuroimmune development during adolescence, consisting of potential changes in blood-brain-barrier permeability, continued maturation of immature amoeboid microglia, changes in astrocyte morphology and proliferation, and substantial synaptogenesis and pruning, making the adolescent brain particularly vulnerable to both central and peripheral insults (for review, see [191]). A recent investigation of the synaptic proteome of adolescent prenatally challenged rats revealed more than 100 significantly altered pre- and post-synaptic proteins, of which more than 50% correlated with known alterations in schizophrenia, a disorder where symptoms often emerge in adolescence [91]. Giovanoli et al. (2016) also report emergence of postsynaptic deficits in pubescence, while presynaptic deficits emerged in adulthood in prenatally challenged mice [122]. Using a single injection of LPS in dams at E16, Straley et al. demonstrated impaired motor performance in offspring tested during juvenile and adult stages but not during adolescence [179]. This is consistent with multiple poly I:C MIA models, which do not induce noticeable behavioral, cognitive, or pharmacological abnormalities in offspring until adulthood, perhaps due to post-pubertal maturational processes necessary for the development of behavioral abnormalities [192]. Thus evidence of aberrant microglia activity and cognitive deficits may be absent in prenatally challenged neonatal piglets, but present in adults. Future studies using adolescent and post-pubertal pigs born from PRRSV-infected gilts are needed to test this hypothesis. Though we suspected that prolonged over-activation of fetal microglial cells would be evident in this model, our study is not the first to show that microglia activity is unchanged in the postnatal period [33, 118, 121-123]. We believe that induction of fetal microglia activation by maternal infection may instead be transient, allowing microglia activity to return to baseline before or shortly after birth. This has been demonstrated by Arsenault et al. (2014), who compared cellular and inflammatory markers in fetal and postnatal mouse brains using both LPS and poly I:C MIA models; significant increases in TNFα and CD68 protein expression in prenatally challenged fetal brains were almost all resolved by PD 10 in both

models [123]. Ratnayake et al. (2012) compared Iba1 staining in PD 1 and PD 100 mice born to dams exposed to poly I:C to show that the number of microglial cells in the hippocampus was significantly increased at PD 1, and that these microglia demonstrated a more amoeboid phenotype, compared to controls; these differences were no longer apparent at PD 100 [125]. It is possible that an undefined microglia phenotype exists due to prenatal challenge but this was not detectable in our model due to the classical phenotypic markers used. Microglia serve widely varying functions that differ based on neurodevelopmental stage and pathological states [63], and could fall anywhere along a spectrum of activation states, some of which may not be classically defined [72]. The most robust effect of maternal PRRSV infection was clear abnormalities in piglet social behaviors. Aberrant social behaviors are also observed in both rodent and non-human primate MIA models [31, 32, 193-195]. A deficit in social communication and interactions is one of the two major criteria for diagnosis of Autism Spectrum Disorder (DSM-IV). Social deficits are also indicative of negative symptoms of schizophrenia, namely social withdrawal and impairments in social functioning (DSM-IV). Pigs are a social species [196, 197] and display a preference for novelty exploration [198], thus our tests of sociability and preference for social novelty are ideal for this species. Oro-nasal fixation is well defined in swine [199, 200] and serves many purposes, including social recognition and affiliation [201]. Pig- directed nosing behavior in swine has been shown to indicate social behavior [201]. Here, as the dog crate allowed for limited tactile communication, nose-to-crate contact was used to define social behavior; this also allowed for direct comparison with the empty crate. Though differences in nose-to-crate contact between treatment groups did not reach significance in either test, Petersen et al. (1989) report that proximity to other pigs is a reliable indicator of the social bond between those pigs [197]. Thus, we believe that proximity to a stranger pig (i.e. time spent in that compartment), is the best indicator of overall pro- social behaviors in this context. During sociability testing, control piglets spent more time in the company of novel conspecifics. In contrast, piglets born from PRRSV-infected gilts spent more time alone in the empty compartment, a behavior that could be indicative of social withdrawal. In both the test of sociability and preference for social novelty, PRRSV piglets displayed time-in-compartment distributions similar to autistic-like rodent strains [202]. The development of the social neural circuit, and the molecular and cellular mechanisms involved in its formation, remain to be fully elucidated. However, recent work using mice mutant for Disheveled (scaffold proteins involved in the Wnt signaling pathway) reveals transient fetal brain growth abnormalities linked to neuronal progenitor cell proliferation and differentiation, resulting in abnormal social interactions in adult mice [203]. Evidence from mice deficient in CX3CR1 (a chemokine receptor

extensively expressed by microglia) indicates that microglia are integral in establishing functional connectivity between the medial prefrontal cortex and the hippocampus. The decreased functional connectivity in CX3CR1 KO mice is significantly correlated with reduced social interest or motivation [89], a correlation that aligns with the literature on patients with autism [101]. Dickerson and Bilkey (2013) reported that a reduction in synchrony along this same connectivity pathway can be induced in rodents by exposure to MIA [93]. As Choi et al. (2016) reported, this induction may in part be dependent on the

IL-6-to-IL-17a maternal pathway, such that maternal IL-6 augments TH17 cell differentiation, inducing heightened levels of maternal IL-17a, which is sufficient to produce social deficits (along with several other ASD-like behaviors) in offspring. Pre-treating pregnant dams with IL-17a-blocking antibody fully rescued these phenotypes [25]. Evidence from studies investigating the interaction of maternal infection with offspring epigenetics (reviewed in [204]) suggests that there is likely also involvement of epigenetic regulation (potentially through maternal IL-6 and IL-17a pathways) in MIA-induced ASD-like behaviors. Though the exact mechanism has yet to be identified, mounting evidence indicates that MIA significantly alters social behaviors in offspring [25, 31, 52, 205, 206]. Glia other than microglia may also be involved in shaping the response of the fetal brain to maternal insult. Recently, astrocytes have been implicated in MIA models utilizing poly I:C [207, 208]. Astrocytes, originally thought of as non-neuronal supportive cells, play a key role in normal brain development and in mediating synapse activity; they also express many of the same pattern recognition receptors found on microglia and are able to secrete many of the same anti- and pro-inflammatory cytokines when stimulated by classic TLR ligands or by maternal cytokines [208]. Though astrocytes may have a critical role in mediating neurodevelopment during prenatal infection, there is still ample work to be done in understanding the mechanisms involving both microglia and astrocytes during this critical neurodevelopmental window. Here, we have utilized a novel translational swine model to demonstrate that maternal infection in late gestation disrupts social behaviors of the offspring in the absence of overt inflammation. These altered social behaviors were detectable four weeks after birth, which indicates that this model might be useful for studying onset of antisocial behavior due to MIA. We believe that fetal microglia may become activated at peak maternal infection in this model, leading to transient neuroinflammation that is resolved before or shortly after birth. Future studies are needed to characterize fetal microglia activity during peak maternal infection and identify mechanisms that may lead to altered neurodevelopment of social neural circuits. Overall, the present data indicate MIA can elicit behavioral abnormalities in offspring, independent of postnatal microglia activation.

3.6 Figures and Tables Figure 3.1

PD11 PD14

LPS challenge; euthanized

GD76 GD113 PD1 Begin weekly saliva collection PRRSV Inoculation; Labor induced Groups 1 & 2 and moved Processing; Piglets Gilts Spatial T-maze testing

arrive Acquisition Reversal PD28 at

facility PD11 PD14 PD20 PD24 PD27 split into

Begin T-maze testing Sociability testing Begin habituation to testing Cortisol blood collection; euthanized

Fig. 3.1. Study Design. Gilts were brought into the animal facility at GD69 and inoculated 1 wk later. Newborn piglets were moved to separate cages after 24 h with the gilt (in order to obtain adequate intake of colostrum). One group of piglets was challenged with LPS at PD 14 (± 1 d) then sacrificed, the other group began behavioral testing at PD 14 (± 1 d); this group was sacrificed at PD 28 (± 1 d).

Figure 3.2 A

Fig. 3.2. Maternal PRRSV infection induced fever and reduced food intake in gilts. PRRSV infection caused (A) a transient increase in gilt plasma TNFα (treatment x time, p < 0.0001; Control: n = 9, PRRSV: n = 6), (B) a transient increase in gilt body temperature (treatment x time, p < 0.01) and (C) a transient decrease in gilt food intake (treatment x time, p < 0.0001). * = p < 0.05, ** = p < 0.01, *** = p < 0.001. Control: n = 3, PRRSV: n = 5 unless otherwise specified.

Figure 3.3

A B

Fig. 3.3. Piglets born from PRRSV-infected mothers had similar circulating cytokines and cortisol. Intraperitoneal injection of 5 µg/kg BW LPS resulted in increased rectal temperature, increased serum cortisol, and increased plasma cytokines 4hrs post-injection. Main effect of LPS treatment on (A) rectal temperature (p < 0.01; n = 5 – 10 per group); (B) serum cortisol (p < 0.01; n = 3 – 10 per group); and (C) plasma TNFα (p < 0.0001; n = 5 – 10 per group), IL-10 (p < 0.05; n = 5 – 10 per group), IL-1β (p < 0.01; n = 3 – 10 per group), and IL-8 (p < 0.05; n = 4 – 8 per group). Tukey post-hoc: significantly different means do not share the same letter.

Figure 3.4 A

Fig. 3.4. Maternal infection did not alter MHC II in hypothalamic microglia in piglets. CD11b+ cells isolated from the hypothalamus of control and PRRSV piglets did not differ in expression of MHCII (a marker of phagocytic cell activation). Representative scatter plot of CD45 and MHCII expression of hypothalamic CD11b+ cells from (A) a control piglet and (B) a piglet born from a PRRSV-infected mother; (C) percent of CD11b+ cells co-expressing MHCII for each treatment group (p > 0.10). Cells were isolated from the hypothalamus of one animal and were analyzed as single individual samples; percent of MHCII positive cells in each sample was averaged across treatment groups; control: n = 8, PRRSV: n = 11.

Figure 3.5 A

Fig. 3.5. Hippocampal microglia were responsive to LPS and DEX, and were not impacted by maternal treatment. CD11b+ cells isolated from the hippocampus and stimulated with 10 ng/mL LPS displayed increased gene expression of (A) TNFα (LPS x DEX, p < 0.05), IL-6 (main effect of LPS, p < 0.001, main effect of DEX, p < 0.05), IL-10 (LPS x DEX, p < 0.05), and NR3C1 (LPS x DEX, p < 0.05), which was partially attenuated by treatment with 3.18 µM Dexamethasone (DEX) 4 h post-treatment. These results were recapitulated in (B) TNFα protein expression (LPS x DEX, p < 0.001). Cells were isolated from the right and left hemisphere of the hippocampus separately and then plated and stimulated. Thus, one piglet provided two individual samples, resulting in 8-10 samples per treatment group. Tukey post-hoc: significantly different means do not share the same letter.

Figure 3.6

Fig. 3.6. Hypothalamic microglia were responsive to LPS and were not impacted by maternal treatment. CD11b+ cells isolated from the hypothalamus and stimulated with 10 ng/mL LPS displayed increased gene expression of (A) TNFα (p < 0.05), IL-6 (p < 0.05), and NR3C1 (p < 0.05), but not IL-10 (p = 0.079) 4 h post-treatment. (B) TNFα protein expression was elevated 4 h post-treatment (p < 0.001). Cells were isolated from the hypothalamus of one animal and were analyzed as single individual samples; control saline: n = 2, control LPS: n = 3, PRRSV saline: n=3, PRRSV LPS: n=3. Tukey post-hoc: significantly different means do not share the same letter.

Figure 3.7

Fig. 3.7. Piglets in both groups performed similarly in the learning and memory spatial T-maze task. Displayed is the proportion of correct choices out of total trials per day (A1 – A6: acquisition days 1-6; R1 – R5: reversal days 1- 5). The dotted line represents the goal criteria (8/10 correct choices) set for acquiring the task. Main effect of time (p < 0.0001), non-significant treatment x time (p > 0.10). Control: n = 14, PRRSV: n = 17.

Figure 3.8

B C

Fig. 3.8. Piglets born from PRRSV-infected gilts displayed abnormal social behaviors compared to control piglets in the test of sociability. Between-group ANOVA revealed different chamber preferences in the sociability test (treatment x chamber, p < 0.01). (A) Schematic representation of the Social Behavior testing arena. Subject piglets were placed in the center compartment and then the doors were lifted, signifying the start of the test. (B) Duration (s) that piglets spent in each chamber (Stranger Pig 1, center, or empty; within-group ANOVA, * = p < 0.05, *** = p < 0.001). (C) Duration (s) that piglets spent actively investigating Stranger Pig 1 (i.e. nose in contact with the crate containing Stranger Pig 1) or the empty crate while in the corresponding chambers (no significant difference between maternal treatment groups, p > 0.10). Control: n = 16, PRRSV: n = 16. Tukey post-hoc: significantly different means do not share the same letter.

Figure 3.9

Fig. 3.9. Piglets born from PRRSV-infected gilts displayed abnormal social behaviors compared to control piglets in the preference for social novelty test. Between-group ANOVA revealed different chamber preferences in the preference for social novelty test (treatment x chamber, p < 0.01). (A) Duration (s) that piglets spent in each chamber (Pig 1, center, or Stranger Pig 2; within-group ANOVA, ** = p < 0.01, *** = p < 0.001). (B) Duration (s) that piglets spent actively investigating Stranger Pig 2 and Pig 1 (i.e. nose in contact with the crate containing stranger pigs; trending treatment x stranger pig, p = 0.094). Control: n = 16, PRRSV: n = 16. Tukey post-hoc: significantly different means do not share the same letter.

Supplemental Table S3.1. Number and sex of piglets designated to Group 1 and Group 2 from each litter. Control 1 Control 2 Control 3 PRRSV 1 PRRSV 2 PRRSV 3 PRRSV 4 PRRSV 5

Group 1 4 6 0 9 5 0 5 0

Female 3 5 - 5 2 - 4 -

Male 1 1 - 4 3 - 1 -

Group 2 6 5 6 3 3 1 3 7

Female 4 5 3 2 1 - 1 2

Male 2 - 3 1 2 1 2 5 Total piglets per group: Group 1, Control: n = 10, PRRSV: n = 19; Group 2, Control: n = 17, PRRSV: n = 17.

Supplemental Table S3.2. Quantitative real-time PCR primer information. Gene Classification Accession numbera Assay identificationb GAPDH Reference NM_001206359 Ss0337854_g1 TNFα Pro-inflammatory NM_214022 Ss03391318_g1 IL-6 Pro-inflammatory NM_214399 Ss03384604_u1 IL-10 Anti-inflammatory NM_214041 Ss03382372_u1 NR3C1 Glucocorticoid Receptor NM_001008481 Ss03378868_u1 aNCBI GenBank accession number. bApplied Biosystems TaqMan Gene Expression Assay identification number.

Supplemental Table S3.3. Average litter characteristics by treatment group. Piglet Treatment Litter Size Stillborn Live Born Piglet Birth Wt. (kg) Mortality

Control (n = 3) 14.7 ± 2.6 4.3 ± 2.6 10.3 ± 2.6 1.7 ± 0.2 0.14

PRRSV (n = 5) 11.2 ± 1.1 3.0 ± 1.3 8.2 ± 1.1 1.5 ± 0.1 0.08

Data are means ± S.E.M. No significant differences between treatment groups. However, the authors would like to stress that the current study was not designed to detect significant litter differences due to maternal PRRSV infection and thus is underpowered for those analyses. These data are merely meant to summarize the average characteristics of the litters used in the study.

CHAPTER 4: ALTERED HIPPOCAMPAL GENE EXPRESSION AND MORPHOLOGY IN FETAL PIGLETS FOLLOWING MATERNAL RESPIRATORY VIRAL INFECTION2

4.1 Abstract Maternal infection during pregnancy increases the risk of neurobehavioral problems in offspring. Evidence from rodent models indicates that the maternal immune response to infection can alter fetal brain development, particularly in the hippocampus. However, information on the effects of maternal viral infection on fetal brain development in gyrencephalic species is limited. Thus, the objective of this study was to assess several effects of maternal viral infection in the last one-third of gestation on hippocampal gene expression and development in fetal piglets. Pregnant gilts were inoculated with porcine reproductive and respiratory syndrome virus (PRRSV) at gestational day (GD) 76 and fetuses were removed by cesarean section at GD 111 (3 d before anticipated parturition). Gilts infected with PRRSV had elevated plasma interleukin-6 and developed transient febrile and anorectic responses lasting approximately 21 days. Despite having a similar overall body weight, fetuses from PRRSV-infected gilts had decreased brain weight and altered hippocampal gene expression compared to fetuses from control gilts. Notably, maternal infection caused a reduction in estimated neuronal numbers in the fetal dentate gyrus and subiculum. The number of proliferative Ki-67+ cells was not altered, but relative integrated density of GFAP+ staining was increased, in addition to an increase in GFAP gene expression, indicating astrocyte-specific gliosis. Maternal viral infection caused an increase in fetal hippocampal gene expression of inflammatory cytokines TNFα and IFNγ, and myelination marker MBP. MHCII protein, a classical monocyte activation marker, was reduced in microglia, while expression of the MHCII gene was decreased in hippocampal tissue of fetuses from PRRSV-infected gilts. Together, these data suggest that maternal viral infection at the beginning of the last trimester results in a reduction in fetal hippocampal neurons that is evident 5 weeks post infection, when fetal piglets are near full term. The neuronal reduction was not accompanied by pronounced neuroinflammation at GD 111, indicating that activation of classical neuroinflammatory pathways by maternal viral infection, if present, are mostly resolved by parturition.

Keywords: maternal immune activation, prenatal insult, microglia, hippocampus, neurogenesis, fetal pig, neuroinflammation

2Reprinted with permission from Developmental Neuroscience journal: Antonson, A. M., Balakrishnan, B., Radlowski, E. C., Petr, G., & Johnson, R. W. (2018). Altered Hippocampal Gene Expression and Morphology in Fetal Piglets following Maternal Respiratory Viral Infection. Dev Neurosci. doi:10.1159/000486850. Publisher: S. Karger AG, Basel.

4.2 Introduction Epidemiological studies have linked prenatal maternal infections, both viral and bacterial, to altered structural and functional brain abnormalities postnatally [1, 209]. It is speculated that activation of the maternal immune system can lead to changes in the maternal-fetal cytokine balance, ultimately altering fetal brain development [29, 30]. The third trimester in human pregnancy represents a critical window for both fetal brain and immune system development, and is characterized by neural migration, apoptosis, synaptogenesis, myelination, and immune cell colonization [16, 210]. Neurons and glia in the developing brain have numerous cytokine receptors and are sensitive to inflammatory conditions [211, 212]. Indeed, animal models of maternal immune activation (MIA) have demonstrated that offspring develop symptoms reminiscent of neuropsychiatric or neurological disorders, such as autism, schizophrenia, and epilepsy [31, 213-215]. Numerous studies support the concept that maternal cytokines signal across the maternal-fetal interface and drive MIA outcomes. Smith et al. (2007) demonstrated, through inhibition or injection of recombinant IL-6 to pregnant dams, the critical involvement of this particular cytokine in mediating offspring behavioral abnormalities [26]. A potential mechanism through which IL-6 accesses the fetal compartment was revealed by Wu et al. (2016), who performed a restricted deletion of the IL-6 receptor on placental trophoblasts to prevent the neurodevelopmental pathologies caused by MIA [27]. In the brain of fetuses exposed to maternal lipopolysaccharide (LPS) treatment, intracellular IL-6 signaling, through signal transducer and activator of transcription (STAT3), is increased and can be blunted by IL-6 blocking antibody [27, 28]. A series of elegant studies performed in a rodent maternal polyinosinic:polycytidylic acid (poly I:C) model reveal that IL-6 may also be driving T helper cell development and activation, augmenting IL-17a release [216]. This group has demonstrated that IL-17a is sufficient to produce abnormal cortical development [38] and autism-like phenotypes [39] in offspring, and that these phenotypes can be prevented by blocking T helper 17 cells or by administration of IL-17a- blocking antibody in pregnant dams [25]. Investigations into the involvement of microglia in MIA mechanisms are of interest [16] due to the fact that these resident CNS immune cells are the primary responders to cytokine signaling in the brain and are involved in neurodevelopmental pathologies [54-57]. Microglia begin colonizing the brain during early gestation [60, 62] and aid in healthy neurodevelopment by producing neurotrophic factors and by phagocytosing excess cells and synapses [60, 61, 63, 65-67]. Currently, studies determining the role of fetal microglia in MIA are conflicting; though Cunningham et al. (2013) demonstrate that activated microglia in a maternal LPS model deplete the fetal neural progenitor pool by excessive phagocytosis [60],

Smolders et al. (2015) failed to find any activation of fetal microglia due to maternal poly I:C treatment at E11.5 or double injection at E11.5 and E15.5 [131]. Pratt et al. (2013), however, demonstrate that the same dose of poly I:C administered at E12.5 resulted in increased production of several pro-inflammatory cytokines and chemokines by fetal microglia, while total microglia number was unchanged [132]. These studies indicate that microglial cell activation in MIA is dependent on the timing and nature of the maternal insult, and that the answer to whether fetal microglia are involved in MIA etiology is still not clear. Evidence does suggest, however, that the hippocampus, a region of the brain important for learning and memory, may be particularly vulnerable to maternal insults. Zhang and van Praag (2015) have demonstrated that maternal treatment with poly I:C results in reduced dentate gyrus volume and differential firing properties of neonatal offspring hippocampal neurons, resulting in impaired spatial navigational ability [217]. Pineda et al. (2013) have demonstrated that altered maternal cytokine levels during pregnancy can change the intrinsic property of offspring hippocampal neurons. A rise in maternal pro-inflammatory cytokines, such as IL-1β and IL-6, increases offspring hippocampal excitability, in turn increasing susceptibility to seizures [214]. Maternal poly I:C treatment has also been shown to decrease offspring presynaptic proteins and impair long term potentiation (LTP) within the CA1-CA3 pathway [218]. Taken together, these data indicate that prenatal immune activation has the potential to permanently alter neuronal circuits within the hippocampus. We sought to determine the impact of maternal viral infection on the neonatal hippocampus using a pig MIA model. The domestic piglet serves as an excellent model of early brain development due to neurodevelopmental trajectories and anatomy, which align closely to that of humans [12-14, 23]. Additionally, the gyrencephalic pig brain undergoes a major growth spurt from the late prenatal to the early postnatal period, closely mimicking the accelerated brain growth and development that occurs in human infants [13, 15]. Here, we inoculated pregnant gilts with porcine reproductive and respiratory syndrome virus (PRRSV) at gestational day (GD) 76 and fetuses were removed by cesarean section at GD 111 (3 d before parturition). PRRSV infects swine alveolar macrophages, leading to interstitial pneumonia [160]; symptoms include anorexia, fever and sometimes cyanosis [219], similar to what is observed with human influenza infection [220]. Previous studies in our lab have shown that early postnatal PRRSV infection results in reduced neurogenesis and enhanced microglia activity in the neonatal hippocampus [21, 163], and that postnatal PRRSV infection impairs neonatal piglet performance in a hippocampal-dependent learning and memory task [167]. More recent evidence from our swine MIA model demonstrates that

prenatal exposure to maternal PRRSV infection causes altered social behaviors postnatally, while prolonged microglia activation is not evident [221]. Thus, we aimed to characterize the impacts of maternal immune activation on hippocampal neurodevelopment and glial reactivity prenatally. We hypothesized that maternal PRRSV would alter fetal hippocampal development (decreasing neuron numbers) and that fetal microglia cells would display an activated phenotype. However, our results indicate that pro-inflammatory MHCII+ microglia are not more abundant at GD 111, though estimated neuronal cell density was indeed decreased in the dentate gyrus and subiculum, and piglets exposed to maternal infection demonstrated GFAP-specific gliosis.

4.3 Materials and Methods 4.3.1 Animals Eight pregnant crossbred (Large White/Landrace) gilts, PRRSV-free and not vaccinated, were brought from the University of Illinois swine herd into the biomedical animal facility at GD 69. Gilts were housed individually in standard farrowing crates (1.83 m x 1.83 m) in identical disease containment chambers kept at 22°C and maintained on a 12 h light/dark cycle. All gilts were provided with ad libitum access to water and 2.3 kg standard corn-soybean meal-based gestational diet daily (University of Illinois feed mill, Urbana-Champaign). On GD 76, gilts were inoculated intranasal with either 5 mL of 1 × 105 50% tissue culture infected dose (TCID 50) of live PRRSV (strain P129-BV, obtained from the School of Veterinary Medicine at Purdue University, West Lafayette, IN) or sterile phosphate buffered saline (PBS; PRRSV: n = 4, control: n = 4). PRRSV is an enveloped single-stranded RNA virus which causes interstitial pneumonia by infecting alveolar macrophages [160], and results in increased secretion of IL-1β, IL-6, and TNFα [161, 162]. Rectal temperature and feed intake were monitored daily. Blood was collected from the marginal ear vein once weekly, on GD 76 (immediately before PRRSV inoculation), GD 83, GD 90, GD 97, and GD 104 (0, 7, 14, 21, and 28 days post infection [dpi], respectively). On GD 111 (35 dpi), gilts were anesthetized using a telazol:ketamine:xylazine drug cocktail (50 mg of tiletamine plus 50 mg of zolazepam reconstituted with 2.5 mL ketamine [100 g/L] and 2.5 mL xylazine [100 g/L]; Fort Dodge Animal Health, Fort Dodge, IA) administered intravenously through the marginal ear vein at 0.005 mL/kg body weight. Gilts were then euthanized via intravenous injection of sodium pentobarbital (86 mg/kg body weight; Fatal Plus, Vortech Pharmaceuticals, Dearborn, MI) and placed in a lateral recumbent position on an operating table once euthanasia was confirmed. A 3-4 inch midline incision was made at the lower abdomen and the uterine horns were exposed. Individual fetuses were carefully removed and the preservation for each fetus was

recorded as viable (healthy before euthanasia) or non-viable (evidence of decomposition; these fetuses were counted but not included in the analyses). The umbilical artery of each viable fetus was severed and blood was collected in blood collection tubes containing clot activator or EDTA. Blood tubes were allowed to clot at room temperature for 30 min and then centrifuged (1,300 x g at 4°C for 15 min) and serum and plasma were collected and stored at −80°C. The body and brain weights of each fetus were recorded. The right hippocampus was immediately dissected out and placed in Hanks’ Balanced Salt Solution (without calcium and magnesium; Mediatech, Inc., Manassas, VA) for microglia cell isolation. The left hippocampus was divided into two halves; the dorsal half was frozen on dry ice for analysis of gene expression and the ventral half was placed in 4% paraformaldehyde for immunohistochemistry. All animal care and experimental procedures were in accordance with the National Research Council Guide for the Care and Use of Laboratory Animals and approved by the University of Illinois Institutional Animal Care and Use Committee.

4.3.2 Microglia cell isolation and flow cytometry Microglia cells were isolated from hippocampal tissue based on positive expression of CD11b using the Miltenyi Biotec neural cell isolation procedure, as described previously [221]. Briefly, cell isolation was performed using Miltenyi Biotec Neural Tissue Dissociation Kits (P), CD11b (microglia) MicroBeads, and equipment (Miltenyi Biotec, San Diego, CA) according to manufacturer’s instructions, with some modifications. Falcon cell strainers (40 µm) were used in place of the 70 µm strainers listed. Myelin removal was achieved by centrifugation with a 30% Percoll PLUS (GE Healthcare Life Sciences, Pittsburgh, PA) solution in PBS. Sample was incubated with CD11b (microglia) MicroBeads, then passed through MS columns and collected. The final CD11b+ fraction was centrifuged at 300 x g for 10 min at 4° C. Isolated cells are then resuspended in flow buffer [PBS with 1% BSA (ThermoFisher Scientific), 0.1% sodium azide (Sigma-Aldrich), and 20 mM glucose (Sigma-Aldrich)]. Competitive binding of the Fc receptor was blocked with purified CD16/CD32 antibodies (eBioscience, San Diego, CA). Cells were incubated with CD11b antibodies (Biolegend, San Diego, CA) to confirm CD11b+ selection, CD45 antibodies (AbD Serotec, Raleigh, NC) to distinguish microglial cells from monocytes, and MHCII (Antibodies Online, Atlanta, GA) as a marker of phagocytic cell activation. Cells were flowed through the FACS Aria II flow cytometer (BD Biosciences, San Jose, CA) and gated based on the forward- and side-scatter and autofluorescence properties of a massed unstained control sample. Cells were confirmed as CD11b+ microglia cells by intermediate expression of CD45 (as previously described, [167, 221]). Thus, cells that were

CD11b+CD45low-int and MHCII+ were considered to be activated microglia. FACSDiva software (BD Biosciences, San Jose, CA) was used to analyze flow cytometry data.

4.3.3 PRRSV and cytokine detection The presence of PRRS virus in the serum of all gilts and fetuses was analyzed by the Veterinary Diagnostic Laboratory (University of Illinois, Urbana, IL) using RT-PCR. Porcine Duo-set ELISA kits (R&D Systems, Minneapolis, MN) for IL-6, IL-10, and TNF-α were used, according to manufacturer’s instructions, to quantify cytokines in gilt and fetal plasma.

4.3.4 Quantitative real-time PCR Total RNA was extracted from the hippocampus of each fetus using E.Z.N.A Total RNA kit 11 (Omega, Norcross, GA, USA) and cDNA was synthesized using a QuantiTect Reverse Transcription Kit (Qiagen, Valencia, CA). Quantitative real-time PCR was performed using the Applied Biosystems TaqMan Gene Expression Assay protocol. The custom TaqMan Low Density Array card consisted of 384 wells and is preloaded with Taqman gene expression assays that will detect real time amplifications and user defined targets. Two reference genes (18S ribosomal RNA [18S rRNA]; ribosomal protein L19 [RPL19]) and 22 genes of interest (Table 4.1) were used. Simultaneous amplification of target cDNA and reference cDNA was obtained using an oligonucleotide probe with a 5ʹ fluorescent reporter dye (6-FAM) and a 3ʹ quencher dye (NFQ). The PCR conditions included 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, and 1 cycle at 60°C for 1 min. Relative gene expression was determined from the fluorescent data using the ABI PRISM 7900HT-sequence detection system (PerkinElmer, Waltham, MA). Thirty-one representative fetuses, from two to three litters per group, were analyzed, and each sample was run in duplicate. Data were analyzed using the comparative threshold (Ct) cycle method [222], and results are expressed as fold change to the standardized relative quantification baseline (relative quantification = 1), whereby a value > 1 would represent a fold increase in mRNA expression compared to the control, and a value between 0 and 1 would represent a fold decrease in mRNA expression. Data were normalized to internal control RPL19. The data exclusion criteria included (1) both wells were below detection limit (undetermined Ct), or (2) values were above or below the upper or lower fences of the distribution, respectively (see outlier exclusion criteria in “Statistical analysis” section below). There was a significant (or trending) effect of sex, but not maternal treatment, on three genes; in this case, as there was no sex x treatment interaction, maternal treatment was removed from the model.

4.3.5 Cresyl violet staining

Ten μm hippocampal sections, 100 µm apart, were stained with 1% cresyl violet. Paraffin embedded brain sections were dewaxed in xylene, rehydrated with 100% alcohol and stained with 1% cresyl violet (Sigma Aldrich) for 4-5 min. Excess stain was rinsed in tap water, then slides were washed in 70% alcohol and dehydrated through absolute alcohol. The sections were cleared in xylene, mounted with DPX Mounting medium (Sigma Aldrich), coverslipped, and allowed to air dry under a fume hood [223]. Exclusion criteria included (1) variable staining intensity, i.e. either too dark or too light; or (2) slide or tissue section damage.

4.3.5.1 Estimated neuronal cell counts The cresyl violet stained sections were scanned on a Nanozoomer (Hamamatsu Photonics, Hamamatsu, Japan) slide scanner at 40X magnification. The images were exported to 20X, and the CA1, CA3, and subiculum regions were manually traced using the free hand tool in Image J, then the number of cells within those regions were counted. A representative image of the traced regions of the hippocampus is shown in Figure 4.1. The CA1, CA3, and subiculum regions were identified by the location and size of cells. Pyramidal cells in the CA3 region were in the angle of the “C” shape of the tissue and were large and tightly arranged; CA1 cells were located at the top peak region of the “C” and were smaller in size. This CA1 region was selected to avoid overlapping with the CA2 region. The cells in the subiculum were loosely arranged and were located at the rear end of the “C” region. The hilar region was identified as the region immediately below the dentate gyrus. Neurons within each traced region were counted and values are expressed as number of cells/mm2. All cell counting was performed by an experimenter who was blinded to the treatments. Due to the large number of cells in the dentate gyrus (DG) and their compact arrangement, the relative integrated density of cells in this region was estimated using Image J. Here, the images were converted to 16 bit and the region of the DG was traced using the free hand tool. Five different regions along the DG that were devoid of any cells were selected to subtract the background. Relative integrated density was calculated as mean integrated density of positive staining in the region of interest minus the non-stained background integrated intensity. Integrated density is the sum of the values of pixels in a region of interest and can be used to measure the density of positively stained cells [224-227]. Representative piglets from each group (n = 11 – 17) were selected for cresyl violet analyses, and each group was balanced for sex and body weight. Only piglets whose body weight was within the range of 1.0 - 1.6 kg were included in analyses to prevent the confounding effect of small-for-gestational-age body weight. Our group has previously reported that being small-for-gestational-age can increase the risk

of impaired cognitive processing in piglets [228]. Four to six separate tissue sections were analyzed per piglet and then averaged.

4.3.6 GFAP and Ki-67 immunostaining Ten µm paraffin embedded sections were mounted onto slides. Sections were dewaxed, hydrated using different grades of alcohol, and washed with phosphate buffered saline. Following antigen retrieval in citrate buffer, the sections were blocked with 3% hydrogen peroxide to remove endogenous peroxidase activity. Non-specific staining was blocked using 2% normal goat serum containing 0.3% Triton X-100 and incubated in primary antibody anti-GFAP (1:100, Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C or anti Ki-67 (1:100 BD Pharmingen, San Jose, CA) for 48 hrs at 4°C. The next day, sections were washed and incubated in biotinylated secondary antibody (1:200, goat anti-mouse antibody, Jackson Immunoresearch Laboratories, West Grove, PA) followed by signal amplification using Avidin Biotin complex (ABC reagent, Vector laboratories, Burlingame, CA). Color was developed using diaminobenzidine kit (Vector laboratories, Burlingame, CA). Exclusion criteria was identical to that described above for cresyl violet staining.

4.3.6.1 Quantification of GFAP+ and Ki-67+ cells GFAP and Ki-67 stained sections were scanned using a nanozoomer (Hamamatsu Photonics, Hamamatsu, Japan) under 20x magnification. The images were exported to Image J. Each image was converted to 16 bit and the hilar region was traced using the free hand tool. Relative integrated density of GFAP staining was calculated as described above. For Ki-67+ cell counts, two regions of interest (ROI) were selected: the dentate gyrus (consisting of the suprapyramidal blade, infrapyramidal blade and subgranular zone), and the hilar region. An automated cell analysis plugin in Image J was used to count the number of positive cells in these two ROIs. The total number of Ki-67+ cells is expressed as number of cells/mm2. For each stain, four to six separate tissue sections were analyzed per piglet and then averaged. Seven to 11 representative piglets per group were included, and each group was balanced for sex and body weight.

4.3.7 Statistical analysis Statistical analysis was carried out using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA), or Statistical Analysis Software (SAS Institute, Cary, NC). For all comparisons, alpha was set at 0.05 and data are expressed as means ± SEM. The current study was not designed to reveal significant sex effects and is therefore underpowered for these analyses. Even so, all initial tests were run with sex included in

the model. We found that there was no significant effect of sex in most analyses, so this variable was removed from those models. Outliers were identified using SAS code “proc boxplot, boxstyle=schematicid”, which identifies variables beyond the fences (beyond the third or first quartile plus or minus 1.5 times the interquartile range, respectively); all outliers were removed before statistical testing was performed. Two-way repeated measures analysis of variance (ANOVA; treatment x day) was used for rectal temperature and feed intake in gilts. Bonferroni post hoc tests were performed for two- way ANOVA comparisons with significant main effects or interactions. One-way ANOVAs were used for analysis of piglet brain weight, body weight, gene expression, hippocampal estimated neuronal cell counts, and flow cytometry data. A student’s t-test was used to analyze IL-6 concentration in gilts. In the current study, each piglet was considered an experimental unit as pigs are multiparous species, where each fetus has an individual fetal system, and it is plausible that piglets within the same litter would present with varied responses to maternal infection [175]. Positive PRRSV infection was included in all statistical tests but did not impact any measures, and thus PRRSV+ fetuses are equally represented in all analyses.

4.4. Results 4.4.1 Clinical signs and infection status of PRRSV-inoculated pregnant gilts PRRSV infection caused an increase in gilt rectal temperature (treatment x day, p < 0.01; Fig. 4.2A), and a decrease in gilt food intake (treatment x day, p < 0.05, Fig. 4.2B). PRRSV-inoculated gilts had increased levels of plasma IL-6 at 7 dpi (p < 0.05, Fig. 4.2C). RT-PCR analysis of maternal serum at 7, 14, 21, and 28 dpi confirmed that PRRSV-inoculated gilts were positive for the virus one week after inoculation and throughout the remainder of the study, except for one PRRSV-inoculated gilt who was positive at 7, 14, and 21 dpi but negative at 28 dpi. Serum from all four control gilts tested negative for PRRSV throughout the study, and all eight gilts tested negative at 0 dpi (data not shown).

4.4.2 Maternal PRRSV infection decreased fetal brain weight but did not impact fetal body weight Fifty-one fetuses were obtained from 4 saline control gilts and 45 fetuses from 4 PRRSV- inoculated gilts (litter characteristics summarized in Table 4.2). A total number of 1 and 12 non-viable fetuses were obtained from control and PRRSV-inoculated gilts, respectively, and were excluded from the study (thus, 50 control fetuses and 33 fetuses from PRRSV gilts were included in the study). There was a main effect of maternal PRRSV infection on the number of viable and non-viable piglets obtained from each group (p < 0.05; Table 4.2). Fetal body weights were similar between treatment groups (control: 1.19

± 0.05 kg; MIA: 1.16 ± 0.05 kg; p = 0.619), indicating that PRRSV-induced anorexia did not affect overall fetal growth. Unsurprisingly, there was a trending effect of sex (p = 0.0693) and a main effect of litter size (p < 0.0001) on fetal body weight. Fetuses from PRRSV-inoculated gilts had decreased brain weights compared to controls (p < 0.05, Fig. 4.2D), which may indicate an underlying neurodevelopmental pathology due to maternal viral infection. Litter size and sex did not affect brain weights.

4.4.3 Cytokine concentrations in cord blood were below detectable levels Plasma IL-6, IL-10, and TNF-α concentrations were measured in cord blood collected from fetal piglets immediately following extraction from the uterus. Concentrations were below detectable levels for all three cytokines.

4.4.4 Maternal infection did not increase fetal microglial MHCII expression Enriched CD11b+ microglia expressed CD45 at low levels indicating that they were microglia and not monocytes [145, 176, 177]. GD111 fetal microglia expressed MHCII at very low levels (control: 0.73 ± 0.06%; MIA: 0.36 ± 0.08%). There was a decrease in the percent of microglia expressing MHCII in the prenatal infection group (p < 0.001), though it is uncertain whether this is physiologically relevant.

4.4.5 Fetuses from PRRSV-infected gilts displayed altered hippocampal gene expression Maternal immune activation upregulated relative TNFα and IFNγ gene expression in the hippocampus (p < 0.001 and p < 0.0001, respectively) and tended to upregulate expression of IL-1β (p = 0.056) and STAT3 (p = 0.079). Expression of glial fibrillary acidic protein (GFAP; p < 0.001) and myelin basic protein (MBP; p < 0.05) were upregulated in MIA fetuses, while MHCII expression was downregulated (p < 0.001), indicative of gliosis that is specific to GFAP+ cells. Maternal infection tended to upregulate CD200 expression (p = 0.077), while CD200R expression was unchanged. The results for all 22 genes are graphically represented in Fig. 3A. The expression of nerve growth factor (NGF; p < 0.01; Fig. 4.3B) and heat-shock protein 70 (HSP70; p < 0.01; Fig. 4.3C) was decreased in male fetuses compared to female, but there was no effect of maternal treatment. Males also tended to express less CCL2 compared to females (p = 0.055; Fig. 4.3D).

4.4.6 Maternal infection decreased neuronal cell density in the fetal dentate gyrus and subiculum While no significant difference was found between the number of neurons in the CA1 (Control: 1890 ± 235.6 cells/mm2; MIA: 1940 ± 206.5 cells/mm2), CA3 (Control: 1025 ± 52.3 cells/mm2; MIA: 1051 ± 63.3 cells/mm2), and hilar (Control: 423.2 ± 26.9 cells/mm2; MIA: 441.3 ± 31.9 cells/mm2) regions of

fetuses from PRRSV-inoculated gilts and fetuses from control gilts (p > 0.10), a reduction in the relative integrated density of cells located in the dentate gyrus of MIA fetuses was observed (p < 0.001; Fig. 4.4Aa & Ab, B). A similar reduction was observed in the number of neurons in the subiculum of fetuses from PRRSV-inoculated gilts (p < 0.001; Fig. 4.4Ac & Ad, C).

4.4.7 Relative integrated density of GFAP+ cells, but not number of Ki-67+ cells, was altered by maternal infection To understand if the reduction in neuron number in the dentate gyrus and subiculum was due to decreased neuronal proliferation, we counted the number of proliferative cells (Ki-67+) in the dentate gyrus and hilus. However, the number of Ki-67+ cells was similar between groups (dentate gyrus: Control 207 ± 28.8 cells/mm2, MIA 171.1 ± 24.3 cells/mm2, p = 0.354; hilus: Control 52.1 ± 7.4 cells/mm2, MIA 59.6 ± 4.7 cells/mm2, p = 0.376; Control: n = 9, MIA: n = 14 - 15). Given that there was no change in the total number of proliferative cells in the hippocampus, we hypothesized that maternal infection was causing a greater number of differentiating cells to preferentially develop into glial cells over neurons. Thus, we quantified the relative integrated density of GFAP+ cells in the hilar region of the hippocampus. Indeed, fetuses from PRRSV-infected gilts presented with an increased relative integrated density of GFAP+ cells compared to control fetuses (p < 0.05; Fig. 4.5).

4.5 Discussion In the present study, we demonstrate that late-gestation maternal PRRSV infection causes a reduction in fetal viability and brain weight, which is accompanied by decreases in relative density of neurons, upregulation of GFAP-specific gliosis, and altered gene expression in the hippocampus. Hippocampal microglia did not appear to be activated and there were no differences in cell proliferation at GD 111, which leads us to suspect that fetal microglia are instead acutely activated by maternal viral infection, causing the observed alterations in hippocampal development, and that this activation is resolved by parturition. It is important to note that most of the existing literature on MIA is from rodents, which differ from both pigs and humans in their anatomy and physiology and their innate immune responses [17, 18, 229]. It should be acknowledged that rodents are not always the ideal animal model for human diseases [230], and that certain immune parameters, as well as the neurodevelopmental timeline, are dissimilar between mice and pigs. To our knowledge, this is the first study to investigate the impacts of maternal viral infection on fetal hippocampal neurons and glia in a gyrencephalic animal.

Maternal PRRSV infection resulted in febrile, anorectic, and inflammatory cytokine (IL-6) responses in pregnant gilts post-inoculation, and the presence of the virus in maternal serum was confirmed through rtPCR. We have previously shown that plasma TNFα is also elevated in PRRSV- inoculated gilts on GD 83, 90, and 97 (7, 14, and 21 dpi, respectively; treatment x time, p < 0.0001) [221]. PRRSV infection during gestation results in increased incidences of abortion, premature delivery, and increased neonate mortality [219]. Here, maternal PRRSV infection increased piglet mortality, in agreement with reports of reproductive failure due to late-gestation PRRSV infection [231-233]. While transplacental fetal infection does occur, there was no significant effect of positive viral infection on any analyses in our model. Epidemiological studies assessing the risks of maternal infection on offspring psychiatric disorders indicate that both pathogens that are transmissible to the fetus (rubella virus, cytomegalovirus, herpes simplex virus, toxoplasma gondii) and those that are non-transmissible (influenza virus) confer risk [1, 2]; animal models strongly suggest that the maternal inflammatory response (i.e. cytokines such as IL-6 and IL-17a), and not the pathogen, drive the detrimental effects of MIA on offspring neurodevelopment and behavior [25-27]. Thus, in our model, PRRSV infection serves to activate a comparable maternal immune response to those observed during human viral infections, such as influenza. Brain weights of viable piglets obtained from PRRSV-inoculated gilts at GD 111 were significantly reduced compared to controls, indicating a targeted impact of maternal viral infection on offspring brain development. It is important to note that piglets obtained from infected gilts had comparable body weights to those obtained from control gilts, indicating that maternal infection did not cause intrauterine growth restriction (IUGR) or low birth weight (LBW) fetuses. In most cases of IUGR, fetuses undergo an adaptive brain-sparing response that results in asymmetrical growth, wherein only body size, and not brain size, is reduced [234]. In our MIA model, there is no increase in the incidences of IUGR or LBW piglets [seen here and previously [221]]; to the contrary, it appears that MIA fetuses undergo inverse asymmetrical growth, wherein body weight is spared but brain weight is reduced. Estimated neuronal cell density in the dentate gyrus and neuronal cell counts in the subiculum were significantly decreased in GD111 fetal brains exposed to maternal infection, which could partly explain the reduction in brain weight. One possibility that could account for this decrease would be a reduction in neurotrophic factors. Yet, NGF, BDNF, NTRK3, and NTF3 gene expression was unaltered by maternal infection at GD111. Expression of markers involved in synaptic activity (synaptophysin) and pruning [TGF-β and CX3CL1; [82, 83]] were also unchanged. Interestingly, NGF expression was reduced in male fetuses compared to female fetuses, though there was no effect of MIA. We have previously

observed sex differences in the hippocampus of neonatal piglets challenged with PRRSV, such that viral infection decreased the number of newly-divided cells in males but not females [162], though expression of NGF was not investigated. In the hippocampus specifically, NGF has been implicated as a key player in plasticity and long-term potentiation throughout adult life [235]. Previous studies using two different rodent MIA models report increased or unchanged expression of both NGF and BDNF in the fetal brain due to MIA. In a model using LPS, BDNF was only transiently increased 2 - 24 hours after maternal LPS and returned to baseline levels by PD 6; in contrast, NGF was unaltered in the hours following LPS injection but was increased at PD 6 [45]. When these markers were again examined by the same group, using a poly I:C MIA model, differences in NGF and BDNF expression in fetal brain were not present at any time point, pre- or postnatally [47]. Poly I:C is a viral mimetic immunogen that binds TLR3, so the maternal inflammatory mechanisms initiated during poly I:C challenge mimic what is seen during active viral infection, like PRRSV. This could explain why no differences were observed here in the expression of either gene. As both NGF and BDNF are important regulators of neuronal survival and differentiation, future research designed to map the response of these markers over time, and across sex, to different MIA challenges, is needed. A two-fold increase in expression of MBP was also observed in the fetal hippocampus at GD 111. This is in contrast with the current literature demonstrating impaired myelination in prenatal immune activation models, including studies focused on the hippocampus [91, 236, 237]; and summarized in [238]). However, as MBP mRNA was only examined at one time point in the current study, we surmise that MBP gene expression may follow the same pattern as previous studies if examined during acute maternal infection, and that the increase in expression observed three weeks post-inoculation may be indicative of a compensatory mechanism, though further studies will need to be conducted to verify this hypothesis. Evidence of astrocyte-specific gliosis was prominent at GD 111. GFAP gene expression was enriched in whole hippocampal tissue and the relative integrated density of GFAP+ cells was significantly increased in the hilar region. Studies linking astrocytic activity to the etiology of poly I:C-specific MIA [207, 208] highlights the ability of these glial cells, which express many of the same innate immune receptors as microglia and produce the same cytokines, to respond to MIA insults and participate in neurodevelopmental alterations. MIA models utilizing poly I:C are arguably the most similar to the live viral infection used here, and it is plausible that virally-mediated maternal infections follow similar etiologies in the fetal brain. Further research is needed to parse out the specific involvement of fetal astrocytes in MIA and to separate their effects from those of microglia.

As established by Cunningham et al. (2013), activation of microglial cells by MIA can result in excessive phagocytosis of neural precursor cells, which causes a decrease in neuronal numbers that endures postnatally [60]. In our model, an indication of excessive neural progenitor cell phagocytosis was evident in the decreased estimated hippocampal neuronal cell density of MIA piglets, while microglia over-activation (measured by MHCII expression) was not apparent. A lack of microglia activation aligns with previous studies [33, 121, 122, 131], though we acknowledge that lasting microglia activation in this model cannot be ruled out solely by MHCII expression. Though MHCII is a widely used marker of classical microglia activation, microglia continuously take on varying states of activation that do not necessarily fall into classical or alternative categorizations [71, 72, 239] and resting or quiescent microglia still actively survey their environment and can engulf and clear parenchymal debris [74]. In agreement with the significant decrease in MHCII expression measured using flow cytometry, MHCII gene expression was significantly downregulated in whole hippocampal tissue as well. We have previously demonstrated that MHCII expression in isolated microglia from piglets of PRRSV-inoculated gilts is unchanged at postnatal day (PD) 28 in our model, while lasting effects of MIA were evident in the emergence of altered social behaviors postnatally [221]. Gene expression of CD200 (the CD200R ligand expressed by neurons, astrocytes or oligodendrocytes to signal microglia inhibition) tended to be upregulated, indicating a potential attenuation of microglial activation [240]; expression of CD200R was unchanged. Differences in hippocampal inflammatory gene expression indicated enduring impacts of maternal viral infection on the innate immune responses of the fetal CNS. In agreement with the altered fetal brain cytokine balance hypothesis [30], gene expression of TNFα and IFNγ were significantly upregulated, along with a trend in upregulation of both IL-1β and STAT3, due to maternal immune activation. These findings further indicate that any immediate markers of peripheral fetal inflammation are resolved before parturition in this model, but specific central markers of inflammation remain. IFNγ, which induces the classical inflammatory microglia phenotype, can initiate the release of nitric oxide (NO) from activated microglia, leading to cellular damage of bystander cells like neurons [241]. TNFα, which is preferentially derived from glia during neuroinflammatory states, is a powerful inhibitor of neurite outgrowth and branching [242]. Further in vitro work suggests that TNFα, IL-1β, and IL-6 can directly impact neuron dendrite development such that the overall complexity of developing neurons is significantly reduced, similar to what is seen in disorders like schizophrenia [37]. Expression of STAT3, activated by IL-6 signaling, tended to be increased. IL-6 signaling has been shown to be necessary and sufficient for the behavioral and transcriptional abnormalities observed in models of MIA [26-28], and

likely potentiates peripheral IL-17a production and over-expression of IL-17Ra in fetal brain, which was not measured here but results in robust alterations in cortical development and patterning [25].

Though there were no differences in the expression of NF-�B, IL-6, IL-10, and CX3CL1 between treatment groups at GD 111, this does not exclude the possibility of transient altered expression of these genes during peak maternal infection, which is no longer evident 35 days post-inoculation. Indeed, Wu et al. (2016) and Mouihate & Mehdawi (2015) demonstrate increased expression of IL-6, STAT3 and pSTAT3 in the fetal brain in the immediate hours following MIA, though these studies did not extend past 24 h post-MIA induction [27, 28]. In fact, the most compelling and comprehensive evidence of the impacts of MIA on fetal brain gene and protein expression to date focus on the immediate time points following immune activation [25, 243], and several studies support the idea that neuroinflammation is resolved by the neonatal and postnatal periods [33, 121, 123]. Interestingly, expression of heat shock protein 70 (HSP70) and CCL2 was reduced in male fetuses, but was not impacted by maternal infection. HSP70, which has been linked to the pathophysiology of schizophrenia [244], is involved in protein folding and protects against both thermal and oxidative cellular stress. A reduction in HSP70 in the brains of schizophrenics appears to contribute to reduced cognitive function [245], and the sexual dimorphism observed in the incidence of psychiatric disorders like schizophrenia and autism in human populations [higher in males [5, 246]] indicates that male piglets may be more at risk of developing behavioral abnormalities in our model. Chemokine (C-C motif) ligand 2 (CCL2), also known as monocyte chemoattractant protein 1 (MCP-1), recruits monocytes and microglia [240] and induces microglia proliferation [247]. Although microglia proliferation in particular was not investigated here, Ki-67 staining indicated that overall cell proliferation did not differ between treatment groups. Work by Schwarz et al. (2012) indicates that some chemokines, including CCL2, are up-regulated in the developing hippocampus and cortex and that they may be integral for recruiting and maintaining primitive and premature macrophages and microglia; however, no sex differences in CCL2 were reported [64]. Thus, further investigation into the potential impacts of sexual dimorphism in CCL2 expression, specifically within the context of MIA, is needed. In conclusion, we demonstrate that maternal viral infection in swine increases fetal mortality, decreases fetal brain weight, and alters hippocampal development by decreasing estimated neuronal cell density and increasing GFAP expression immediately prior to parturition. Hippocampal gene expression revealed significant increases in TNFα and IFNγ, and a trending increase in IL-1β and STAT3, while expression of other pro- and anti-inflammatory cytokines and downstream signaling molecules were unchanged, indicative of neuroinflammation that is partially resolved. However, an overall reduction in

brain weight signifies that the lasting effects of MIA extend well past the hippocampus. Investigation of MHCII expression by microglia revealed a decrease in both protein and gene expression at GD 111, indicating that classical microglia activation, if present, may be resolved by this time. Future studies will aim to characterize fetal microglia activation during peak maternal inflammation and the concurring impacts on fetal neurodevelopment across multiple brain regions.

4.6 Figures and Tables

Figure 4.1

Fig. 4.1. Representative image of cresyl violet-stained cells in the fetal hippocampus, demonstrating the regions of interest selected for analyses (subiculum, CA1, CA3, dentate gyrus, and hilus).

Figure 4.2

Figure 4.2 Cont. Fig. 4.2. PRRSV inoculation resulted in transient febrile and anorectic responses in gilts, and increased IL-6 in plasma. Daily recorded rectal temperature (A) and food intake (B) in control (n = 4) and PRRSV-inoculated gilts (n = 4). Arrow indicates inoculation with sterile saline or PRRSV. A significant treatment*day interaction revealed increased body temperature (p < 0.01) and decreased food intake (p < 0.05) in gilts inoculated with PRRSV. (C) Plasma IL-6 concentration at 7 dpi was increased in PRRSV-inoculated gilts (GD 83; p < 0.05; Control: n = 3, PRRSV: n = 4). Data are represented as means ± SEM. (A, B) Bonferroni post-hoc tests: * = p < 0.05, (C) Student’s t-test: * = p < 0.05. (D) Maternal viral infection resulted in decreased fetal brain weight compared to controls. Fetuses from PRRSV-infected gilts (n = 32) had decreased brain weight compared to fetuses from control gilts (n = 47; p < 0.05). Data are represented as means ± SEM; * = p < 0.05.

Figure 4.3

Fig. 4.3. Prenatal exposure to (A) maternal viral infection differentially regulates the expression of several inflammatory and neuronal- or glial-associated genes in the fetal hippocampus. Male fetuses expressed less NGF (B; p < 0.01), HSP70 (C; p < 0.01) and tended to express less CCL2 (D; p = 0.055) compared to females. Data are expressed as fold change ± SEM; *** = p < 0.001, ** = p < 0.01, # = p ≤ 0.08; Control: n = 7-9, MIA: n = 15-22 (except IFNγ: n = 7 per group).

Figure 4.4

Figure 4.4 Cont. Fig. 4.4. Prenatal exposure to maternal viral infection resulted in decreased neuronal number in the hippocampus. (A) Cresyl violet+ cells in the dentate gyrus (a, b) and subiculum (c, d) of fetuses from control and PRRSV-inoculated gilts, respectively. A decrease in the number of cresyl violet-stained neurons was observed in the (B) dentate gyrus (p < 0.001) and (C) subiculum (p < 0.001) regions of fetuses from PRRSV-infected gilts. Data are represented as means ± SEM; *** p < 0.001; Control: n = 11, MIA: n = 14 - 17.

Figure 4.5

Fig. 4.5. Maternal viral infection resulted in an increased relative integrated density of GFAP staining in the hilar region of the hippocampus. (A) Representative image of GFAP stained cells in the hilus of fetuses born to control and PRRSV-inoculated gilts at 5X and 10X magnifications. Arrows identify representative positively-stained cells in each section. (B) An increased relative integrated density of GFAP staining was observed in fetuses from PRRSV- infected gilts (p < 0.05). Data are represented as means ± SEM; * = p < 0.05; Control: n = 7, MIA: n = 11.

Table 4.1. Quantitative real-time PCR primer information. Gene Classification Assay identificationa

18S rRNA, 18S ribosomal RNA Reference Hs99999901_s1 RPL19, ribosomal protein L19 Reference Ss03375624_g1 TNF-α Pro-inflammatory Ss03391318_g1 TLR3 Viral Ss03388861_m1 TGF-β Anti-inflammatory Ss03382325_u1 SYP, synaptophysinb Neuronal Custom STAT3, signal transducer and activator of Pro-inflammatory Ss03388426_m1 transcription NTRK3, neurotrophic receptor tyrosine kinase Neuronal Ss03394533_m1 NTF3, neurotrophin 3 Neuronal Ss03387837_u1 NGF, nerve growth factorc Neuronal Custom NF-κB Pro-inflammatory Ss03388575_m1 MHCII, major histocompatibility complex class II Pro-inflammatory Ss03389942_m1 MBP, myelin basic protein Neuronal Ss03385047_u1 IL-10 Anti-inflammatory Ss03382372_u1 IL-6 Pro-inflammatory Ss03384604_u1 IL-1β Pro-inflammatory Ss03393804_m1 IFNγ Pro-inflammatory Ss03391054_m1 HSP70, heat shock protein Oxidative stress Ss03387784_u1 GFAP, glial fibrillary acidic protein Glial Ss03373547_m1

CX3CL1, fraktalkine Neuronal Ss03377157_u1 CD200, cluster of differentiation 200 Neuronal Ss03375826_u1 CD200R, receptor for CD200d Microglial Custom CCL2, chemokine (C-C motif) ligand 2 Pro-inflammatory Ss03394377_m1 BDNF, brain-derived neurotrophic factor Neuronal Ss03822335_s1 aApplied Biosystems TaqMan Gene Expression Assay identification number. bSynaptophysin custom probe: forward primer, GGCCAAGGACGGCTCAT; reverse primer, TTTTCCGCCCTTAGCATGTAG; probe, CAAGATTAAATGGTACGTAGGAC. cNerve growth factor custom probe: forward primer, TCAACAGGACTCACAGGAGCAA; reverse primer, ACTCCCCCCGGTGGAAA; probe, CGGTCGTCATCCC. dCD200R custom probe: forward primer, TGTTCCAAGTTACTAATCAGGCTGAA; reverse primer, AGCCCATTAGCAACATGATACTCTTT; probe, ACATAGAATTGAAGGAAGGG.

Table 4.2. Average litter characteristics. Treatment Group Litter Size Viable Piglets Non-viable Piglets

Control (n = 4) 12.8 ± 1.5 12.5 ± 1.3 0.3 ± 0.3

PRRSV (n = 4) 11.3 ± 1.3 8.3 ± 0.5* 3.0 ± 1.6*

p-value ns p = 0.023 p = 0.048 Data are means ± SEM. The authors would like to note that the current study was not designed to detect significant litter differences due to maternal PRRSV infection and thus is underpowered for those analyses. These data are merely meant to summarize the average characteristics of the litters used in the study; ns = non-significant; * = p < 0.05.

Chapter 5: MATERNAL VIRAL INFECTION CAUSES GLOBAL CHANGES IN FETAL MICROGLIA AND ALTERS GENE EXPRESSION AND MICROGLIA DENSITY IN THE FETAL AMYGDALA

5.1 Abstract Epidemiological studies demonstrate that prenatal exposure to maternal infection increases the risk of psychiatric disorders like Autism Spectrum Disorder, though the mechanisms remain to be elucidated. We have previously established a maternal immune activation (MIA) swine model, which results in altered piglet social behaviors postnatally, in the absence of microglia activation. Thus, we sought to identify microglia activity prenatally, immediately following maternal infection, and hypothesized that MIA would elicit transient fetal microglia activation concomitant to maternal symptoms of infection. Pregnant gilts were inoculated with porcine reproductive and respiratory syndrome virus (PRRSV) during late gestation and cesarean sections were performed 7 and 21 days post-inoculation (dpi). Notably, MIA fetuses had reduced brain weights at 21 dpi compared to controls. Primary microglia were isolated from fetal brains and assessed for activation status through flow cytometry, in vitro assays, and gene expression; and microglia density and morphology in the hippocampus and amygdala was examined through Iba1 immunohistochemical staining. At 7 and 21 dpi, MIA fetal microglia expressed more of the classical activation marker MHCII and displayed reduced chemotactic activity compared to controls. Phagocytosis was also reduced in MIA microglia at 7 dpi, but not 21 dpi. High-throughput gene expression analysis of microglial-enriched genes involved in neurodevelopment, the microglia sensome, and inflammation, revealed differential regulation in primary microglia and in whole amygdala tissue across both time points. Microglia density was increased in the fetal amygdala at 7 dpi, though few changes in microglia number and morphology were evident in the fetal hippocampus. There was an overall increase in Iba+ cells in the fetal amygdala from 7 to 21 dpi, regardless of treatment. Our preliminary data also reveal sexual dimorphisms in gene expression patterns and microglia number and morphology across all time points and tissues, indicating that MIA likely induces dissimilar effects between males and females prenatally. Overall, these data suggest that fetal microglia, particularly in the amygdala, are transiently but significantly altered by maternal viral infection, indicating a potential mechanism through which MIA could negatively impact prenatal neurodevelopment and cause altered behaviors postnatally.

5.2 Introduction In the recent decade, characterization of the complex involvement of microglia, the innate CNS macrophage, in healthy neurodevelopmental processes has greatly expanded. We now know that microglia are required for integral processes like neurogenesis [60, 248], circuit wiring [79], and clearing excess neuroprogenitors [75] and synapses [83], and that depleting microglia populations or inhibiting microglial activity in early development leads to detrimental accumulation of excess synapses and neural connections [90]. Altered microglial activity during development, therefore, can have long-lasting injurious consequences. Specifically, altered microglial activity has been linked to pathological forebrain wiring and disrupted functional connectivity across limbic and cortical neural networks, as well as to behavioral deficits, all characteristic of neuropsychiatric illnesses like Autism Spectrum Disorder (ASD) [79, 89, 90]. Maternal infections during pregnancy have specifically been linked to an augmented risk of ASD [1, 134], and animal models of maternal immune activation (MIA) indicate that increased maternal cytokines could perturb embryonic microglia, shifting their phenotype to a more pro-inflammatory state [60, 130, 132]. Elegant work thus far advocates for maternal pro-inflammatory cytokines, namely IL-6 [26, 27, 36] and IL-17a [25, 38, 39], as the principal drivers of adverse offspring outcomes in MIA models. Injection of recombinant IL-6 or IL-17a are each sufficient for reproducing the behavioral and neurological abnormalities seen in offspring exposed to MIA, and blocking the action of either cytokine during MIA induction rescues these aberrant phenotypes [25-27, 38, 39]. Though the effects of MIA and maternal cytokines on postnatal offspring behavior have been well-characterized [31-33, 38], less is known about the impacts of maternal infection on microglial activation during the prenatal period. Stimulating MIA with viral mimetic polyinosinic:polycytidylic acid (poly I:C) at embryonic day (E) 12.5 or E14.5 has been shown to shift gene expression profiles of newborn mouse microglia towards a more advanced developmental gene profile [67]. Examination of cytokine and chemokine production by embryonic microglia after maternal poly I:C injection at E12.5 indicated that microglia are more pro-inflammatory [132]. However, microglial activation in the embryonic mouse brain was unchanged if the maternal poly I:C injection was given at E11.5 or two injections at E11.5 and E15.5 [131], indicating that both dose and timing of MIA dictate embryonic microglial sensitivity. Bacterial mimetic lipopolysaccharide (LPS) is also widely used to induce MIA in rodents; it was recently demonstrated that three consecutive maternal LPS injections on E15, 16, and 17 in mice increases embryonic microglial gene expression of pro-inflammatory cytokines IL-1β, TNFα, and IL-6 [130]. Interestingly, LPS injections in pregnant rats at E15 and E16 increased expression of inducible nitric oxide

synthase (iNOS; required for production of nitric oxide) and cytokine IL-1β, both markers of classical cytotoxic activation. Embryonic microglial activation coincided with a decrease in neural precursor cells and reduced thickness of proliferative zones in the fetal rat brain [60]. Overall, these observations indicate that MIA in rodents has the potential to shift embryonic microglia towards a more activated state, which may contribute to pathological alterations in neural function associated with psychiatric illness. However, this has never been explored in a gyrencephalic species. We have established a translatable MIA model using a live viral infection in conventional swine during late gestation. Previous data from our lab confirms the manifestation of altered behaviors indicative of neurodevelopmental dysregulation in neonatal piglets, namely reduced sociability and preference for social novelty [221]. We have also shown that fetal piglets exposed to MIA have reduced neuronal density in the hippocampus and evidence of astrogliosis before birth [249]. Using this model, we sought to characterize the function and phenotypes of fetal microglia both during peak maternal viral infection and after maternal symptoms had resolved. We found that fetal porcine microglia increase the expression of classical antigen presenting marker MHCII in response to MIA, but not surface glycoprotein CD68. Unexpectedly, we show that isolated primary fetal microglia from MIA piglets have downregulated phagocytic and chemotactic activity. High-throughput quantitative PCR revealed distinct changes in the expression of genes involved in neurodevelopment, the sensome, and the inflammatory response; expression patterns in these same genes were perturbed in the fetal amygdala, a region involved in the regulation of social behaviors. MIA increased microglial density only in the fetal amygdala, while microglia morphology was relatively unchanged in both the amygdala and the dentate gyrus and hilar region of the hippocampus. Gene expression patterns and microglia number and morphology were differentially altered between male and female fetuses, providing preliminary evidence for sexual dimorphisms across healthy neurodevelopment and in response to MIA. Our data show for the first time that fetal microglia within a gyrencephalic species are transiently but significantly altered by MIA, providing a possible mechanism for the disrupted neurodevelopment and social behavior previously observed in our model. More broadly, our data suggest that disruption of prenatal microglial homeostasis by MIA may be a contributing factor to the neuropathologies characteristic of disorders like ASD.

5.3 Materials and Methods 5.3.1 Animals

Twenty-seven pregnant Large White/Landrace crossbred gilts were obtained from the PRRSV-free University of Illinois swine herd at gestational day (GD) 69 and maintained in disease containment chambers within the biomedical animal facility as previously described [221, 249]. In brief, all gilts were littermate pairs and artificially inseminated with semen from the same boar (PIC 359 SS 6278, Birchwood Genetics, Inc., West Manchester, OH). One gilt from each littermate pair was inoculated intranasally with

5 5 mL of 1 x 10 50% tissue culture infectious dose (TCID50) of live PRRSV (P-129-BV strain, obtained from Purdue University, West Lafayette, IN) at GD 76, while the alternate littermate received sterile DMEM. Six total replicates were completed, for a total of 13 gilts in the control group and 14 gilts in the PRRSV group. To confirm viral presence, oral fluids were collected from each gilt immediately prior to inoculation (GD 76) and at 7 days post inoculation (dpi; GD 83). Oral fluid collection consisted of allowing gilts to chew on cotton rope until saturated; extracted oral fluids were stored at -20℃ until evaluation. The study design is outlined in Figure 5.1A. At GD 83 ± 1 d (7 dpi; peak maternal infection), half of the gilts were euthanized (control: n = 6; PRRSV: n = 7) for immediate collection of fetal tissues and microglia isolation. The other half (control: n = 7; PRRSV: n = 7) was euthanized at GD 97 ± 1 d (21 dpi; resolution of maternal symptoms). First, gilts were anesthetized through intramuscular administration of 0.04 mL/kg body weight telazol:ketamine:xylazine (TKX) drug cocktail (50 mg tiletamine plus 50 mg zolazepam reconstituted with 2.5 mL ketamine [100 g/L] and 2.5 mL xylazine [100 g/L]; Henry Schein Animal Health, Dublin, OH). Once anesthesia was confirmed, blood was collected from the marginal ear vein into blood collection tubes containing EDTA (for plasma) or clot activator (for serum). Blood tubes containing clot activator were allowed to clot at room temperature for ≥ 30 min; tubes containing EDTA were preserved on ice. Gilts were then euthanized through intravenous administration of a lethal dose (86 mg/kg) of sodium pentobarbital (Fatal Plus, Vortech Pharmaceutical, Dearborn, MI). Upon confirmation of euthanasia, cesarean sections were performed as previously described [249]. In brief, the uterus and both ovaries were removed and litter size was determined by gently palpating the uterus. Fetuses were given unique IDs, beginning closest to the ovary, for each left and right horn. Amniotic fluid was collected from each fetus by passing a needle into the amniotic cavity, then individual fetuses were extracted and categorized as viable or non-viable [250]. Umbilical cord blood was collected from each fetus using vacutainer needles and blood collection tubes containing EDTA or clot activator. Blood tubes containing clot activator were allowed to clot at room temperature for ≥ 30 min; tubes containing EDTA were preserved on ice. All blood tubes were then centrifuged at 1,300 x g at 4℃ for 15 min and serum and plasma were aliquoted and stored at -80℃.

The distal end of each umbilical cord was severed leaving minimal placental tissue attached. Fetal body weights (including exsanguinated umbilical cord) and brain weights were obtained. The placental tissue attached to the cord, and endometrial tissue immediately adjacent to the umbilical stump of each individual fetus, was collected and snap frozen on dry ice. The fetal brain was bisected, separating the left hemisphere from the right. The entire left hemisphere was immediately placed in 10% neutral buffered formalin and preserved for immunohistochemistry. Specific regions of interest (including the hypothalamus, hippocampus, amygdala, frontal cortex, and striatum) were dissected from the right hemisphere and snap frozen. The remaining brain tissue of the right hemisphere was minced with a scalpel blade to obtain a homogenous sample, then submerged in Hanks’ balanced salt solution (without calcium and magnesium; HBSS; Mediatech, Inc., Manassas, VA) and placed on ice for microglial cell isolation. A subset of seven separate litters (eight representative fetuses from each, balanced for sex, weight, and uterine horn location) were reserved for microglia in vitro analyses; these animals were excluded from all other analyses. All animal husbandry and experimentation was in accordance with the National Research Council Guide for the Care and Use of Laboratory Animals and was approved by the University of Illinois Institutional Animal Care and Use Committee.

5.3.2 PRRSV and cytokine detection Oral fluid and serum samples from gilts, and serum samples from fetal cord blood, were analyzed by the Veterinary Diagnostic Laboratory (University of Illinois, Urbana, IL) for presence of PRRSV through rRT-PCR. Inflammatory cytokines IL-6 and TNFα were measured in plasma from gilts and fetal cord blood, and in cell culture media, using porcine-specific DuoSet ELISA kits (R&D systems, Minneapolis, MN) as per manufacturer’s instructions. Circulating IL-17a was measured in gilt plasma using a porcine-specific ELISA kit (Thermo Fisher Scientific, Waltham, MA) according to manufacturer’s instructions.

5.3.3 Microglial cell isolation Fetal microglia were isolated using Miltenyi Biotech (San Diego, CA) cell separation techniques, based on positive selection with CD11b beads, as previously described [221], with few changes. Briefly, homogenized tissue from the right hemisphere was removed from HBSS and transferred to GentleMACS c-tubes containing triple the recommended volumes of Enzyme Mix 1 and 2 (NTDK[P]). Isolated CD11b+ microglia from each fetus were suspended in flow buffer (PBS with 1% BSA [Thermo Fisher Scientific, Waltham, MA] and 0.1% sodium azide [Sigma-Aldrich, St. Louis, MO]) and split in half: one half was used for flow cytometric analyses, and the other half was resuspended in 1 mL TRIzolTM Reagent (Invitrogen, Carlsbad, CA) for isolation of RNA. Samples from seven different litters were reserved for in vitro assays,

and thus were instead suspended in 5 mL PEB (PBS with 2 mM EDTA and 0.5% BSA) and counted using a Z-Series Coulter Counter (Beckman Coulter Life Science, Indianapolis, IN).

5.3.4 Flow cytometry Flow cytometric procedures were carried out as previously described [221]. Cells were incubated with the following antibodies: mouse anti-pig FITC-conjugated CD45 (AbD Serotec, Raleigh, NC), mouse anti-pig PE-conjugated MHCII (Antibodies Online, Atlanta, GA), and mouse anti-human APC-conjugated CD68 (BioLegend, San Diego, CA), then fixed using 10% neutral buffered formalin for 10 min. Cells were flowed through the FACS Aria II flow cytometer (BD Biosciences, San Jose, CA) and fluorescence intensity was compared against a massed unstained control sample.

5.3.5 Stimulation of primary microglia cells with LPS and poly I:C Isolated CD11b+ cells were suspended in standard cell culture medium (DMEM containing 100 U/mL penicillin/streptomycin, 10% FBS, and 1 µL/mL porcine rpGM-CSF [R&D Systems, Minneapolis, MN]) and plated in 12-well cell culture plates at ~4 x 106 cells per well. Cells were incubated with LPS (1 ng/mL media; E. coli 0127:B8, Sigma-Aldrich, St. Louis, MO), poly I:C (1 µg/mL media; Sigma-Aldrich, St. Louis, MO) or an equivalent volume of sterile D-PBS for 4 h, then supernatant was collected.

5.3.6 Phagocytosis and Chemotaxis assays Phagocytic and chemotactic activity of CD11b+ cells were assessed as previously described [163], using the Vybrant Phagocytosis Assay Kit (Life Technologies, Carlsbad, CA) and the Neuro Probe ChemoTx system (101-8; Neuro Probe, Inc., Gaithersburg, MD). Samples were plated in replicates of 4 for the phagocytosis assay, and replicates of 5 for the chemotaxis assay.

5.3.7 RNA extraction, cDNA synthesis, and quantitative real-time-PCR Total RNA from gilt and fetal tissues was extracted using the TRI Reagent® protocol (Sigma- Aldrich, St. Louis, MO). One-to-four µg of RNA was used to synthesize cDNA using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Grand Island, NY). Quantitative real-time PCR was performed using the Applied Biosystems TaqMan Gene Expression Assay protocol (Thermo Fisher Scientific, Waltham, MA); samples were assayed in duplicate. Housekeeping gene ribosomal protein L19 (RPL19) was used for calculating relative fold change of target genes (listed in Supplemental Table S5.1) using the delta-delta CT method.

RNA was isolated from the hypothalamus of 3-4 representative fetuses from each litter and cDNA was generated. RT-PCR results for IL-1RA at 7 dpi were used to run a power analysis (using ‘proc power’ for a one-way ANOVA in Statistical Analysis Software 9.4; SAS Institute, Cary, NC) based on a standard deviation of 0.2 and 90% power, which revealed that 7 fetuses from each treatment group were needed. Thus, two-three representative fetuses from each litter were selected for any future analyses. Samples were excluded if RNA yield was below 1 µg, resulting in a total of 9-11 per treatment group at both 7 and 21 dpi.

5.3.8 High-throughput qPCR using the Fluidigm amplification system cDNA from fetal microglia and amygdala tissue was submitted to the W.M. Keck Center for Comparative and Functional Genomics (University of Illinois, Urbana, IL) for quantitative PCR analysis using the Biomark HD Fluidigm® high-throughput amplification system. Forty-eight genes of interest (selected from [78]; TaqMan assays; Supplemental Table S5.2) were assessed on the 96 x 96 platform, and all samples were run in duplicate. Data were analyzed using the Biomark & EP1 Real-Time PCR Analysis Software with quality threshold set at 0.65, baseline correction set at ‘Linear (Derivative)’, and Ct threshold method of ‘Auto (Global)’. Delta delta Ct calculations were used to obtain fold change of target genes compared to housekeeping control RPL19.

5.3.9 Iba1 immunohistochemistry A stereotaxic atlas for the Large White pig brain [251] was used to calculate the correct orientation and slicing of the left hemisphere to obtain coronal slices containing the hippocampus and amygdala. Coronal sections (~5 mm thick) were placed into tissue cassettes and post-fixed for 24-48 h in zinc formalin (Fisher Scientific, Hampton, NH). Paraffin embedding, slicing (10 µm sections), and slide mounting was done by the Veterinary Diagnostic Laboratory (University of Illinois, Urbana, IL). Anti-Iba1 rabbit primary antibody (Wako, Richmond, VA) was used at 1:500 dilution and goat anti-rabbit IgG secondary antibody (Jackson ImmunoResearch Inc., West Grove, PA) at 1:5000 dilution. Signal was amplified using Vector ABC kit PK4000 (Vector Laboratories, Burlingame, CA) and color was developed using 3,3ʹ-Diaminobenzidine tetrahydrochloride (DAB) tablets (Sigma Aldrich, St. Louis, MO). Stained slides were counterstained with hematoxylin and cover-slipped with DPX Mounting Medium (Sigma Aldrich, St. Louis, MO). A total of 10- 11 individual fetuses per treatment group (two-three per litter), per time point, were stained.

5.3.10 Microglia counting and morphology

Digital images of Iba1-stained slides were obtained at 40X magnification using a NanoZoomer (Hamamatsu Photonics, Hamamatsu, Japan). Images were analyzed using the NDP.view2 software (Hamamatsu Photonics, Hamamatsu, Japan) by trained observers blinded to treatment. Regions of interest (the amygdala and the dentate gyrus and hilar region of the hippocampus) were located using the stereotaxic pig brain atlas [251] and traced using the “Freehand Region” annotation tool. Individual Iba1+ cells were marked and totaled for each region of interest, and the number of cells per mm2 was calculated. The morphology of Iba1+ cells were then categorized based on four defined phenotypes: cells with ‘Thin, Ramified’ processes (or quiescent microglia); cells with ‘Long, Thick’ processes; cells with short ‘Stout’ processes; or ‘Amoeboid’ activated cells, which have enlarged somas and few to zero processes (representative images are presented in Fig. 5.8A). All cells within the dentate gyrus/hilar region were assessed, while only cells within a 1 mm2 box (‘Mitotic Density Box’ annotation tool) placed in the center of the amygdala were assessed. The total number of cells within each category was summed and total Iba1+ cells per mm2 per morphological category was calculated. Three slides per fetus were analyzed for the dentate gyrus/hilar region (due to the dense orientation of hematoxylin+ cells), and then averaged. Slide exclusion criteria included, (1) slide or tissue damage during staining; (2) air bubbles within the DPX mounting media that obscured the region of interest; and (3) portions of the region of interest were out of focus.

5.3.11 Statistics GraphPad Prism 7 (GraphPad Software, Inc., La Jolla, CA) and Statistical Analysis Software 9.4 (SAS Institute, Cary, NC) were used for all statistical analyses, with an α of 0.05 and trends reported at p ≤ 0.10. ‘Proc mixed’ or ‘proc glm’ procedures were used in SAS 9.4 for almost all analyses; a ‘repeated’ statement was included in the model for gilt body temperature and food intake. For any significant interactions, the ‘lsmeans’ statement was used with a Bonferroni adjustment to assess means separation. For endometrial gene expression, percent of CD68+ microglia at 7 vs. 21 dpi, and microglia morphologies at 7 vs. 21 dpi, unpaired Welch’s t tests were used in GraphPad 7. All fetal groups were balanced for body weight and left and right uterine horn placement. Though most analyses were underpowered for comparisons across sex, sex was included in most models and main effects of sex or MIA*sex interactions are reported throughout; all fetal groups were balanced for sex when possible. The authors would like to note that the experiments herein were designed to elucidate the effects of MIA, not sex, on the fetal

brain, so effects of sex should be interpreted with caution, as fully-powered tests are needed to wholly examine the effects of sex in the context of MIA.

5.4 Results 5.4.1 Maternal PRRSV infection causes inflammatory responses in pregnant gilts and reduces fetal brain weight All gilts tested negative for PRRSV at the beginning of the study, and all PRRSV-inoculated gilts were positive at 7 dpi (data not shown). Maternal PRRSV infection resulted in transient increases in gilt body temperature (Fig. 5.1B; PRRSV x time, p < 0.0001) and a decrease in food intake (Fig. 5.1C; PRRSV x time, p < 0.0001). By GD 97 ± 1 d (21 dpi), body temperature and food intake returned to control levels. Circulating pro-inflammatory cytokines IL-6 (Fig. 5.1D) and TNFα (Fig. 5.1E) were elevated in infected gilts compared to controls at both time points (p < 0.0001 and p < 0.05, respectively). Maternal PRRSV infection did not impact plasma IL-17a, though there was a main effect of gestational day (p < 0.05; GD 83: Control = 291.2 ± 91 pg/mL, n = 6, PRRSV = 297.9 ± 51 pg/mL, n = 7; GD 97: Control = 127.6 ± 24 pg/mL, n = 5, PRRSV = 194.12 ± 53 pg/mL, n = 6). A total number of 45 viable fetuses were collected from control gilts and 39 viable fetuses from PRRSV-infected gilts at GD 83 ± 1 d (7 dpi). At GD 97 ± 1 d (21 dpi), a total of 50 and 47 viable fetuses were collected from control and PRRSV-infected gilts, respectively. Maternal PRRSV infection did not impact overall litter size nor total numbers of viable and non-viable piglets (average litter characteristics are summarized in Supplemental Table S5.3). Of the 84 viable fetuses collected at 7 dpi, and 97 viable fetuses at 21 dpi, eight were considered small-for-gestational-age (SGA; < 10th percentile for body weight) at each time point; maternal infection did not impact incidence of SGA fetuses. Nine total fetal cord blood samples from three different MIA litters tested positive for PRRS virus by rtPCR analysis; only cord blood collected at 7 dpi tested positive. Fetuses with PRRSV-positive cord blood did not differ from PRRSV-negative on any measures, so these fetuses were included in all analyses. Unsurprisingly, litter size affected both fetal body weight and brain weight at both time points (p < 0.0001; Table 5.1); offspring from larger litters were smaller on average. At 7 dpi, there was no difference in fetal body or brain weights between treatment groups. At 21 dpi, though body weights did not differ, maternal infection resulted in decreased fetal brain weight (p < 0.0001; Table 5.1) compared to controls.

5.4.2 Maternal viral infection induces expression of inflammatory genes at the maternal-fetal interface

RNA isolated from endometrial tissue revealed an upregulation of several inflammatory genes at 7 dpi, including Ifng, Tnfa, and Il-6 (p < 0.01); Il-1b and chemokine receptor Cxcr3 tended to be upregulated (p < 0.10), while expression of the transcription factor Nf-kb did not differ between treatment groups (Fig. 5.2A). At 21 dpi, expression of Ifng, Tnfa, and Cxcr3 remained elevated (p < 0.001, p < 0.01, and p < 0.05, respectively), and Nf-kb tended to be elevated (Fig. 5.2B; p < 0.10). However, expression of Il-6 tended to decrease compared to controls at this time point (p < 0.10), while Il-1b, though numerically elevated, did not statistically differ from controls. Placental tissue collected from the distal end of the umbilical cord was analyzed for changes in gene expression. At 7 dpi, expression of TNFα tended to be elevated in placental tissue of MIA fetuses (Fig. 5.2C; p < 0.10), though expression of Bdnf, Il-10, Il-1b, Il-6, and Il-6r did not differ between treatment groups. At 21 dpi, there was an interaction between MIA and sex on placental expression of both Bdnf and Il-10 (Fig. 5.2D; p < 0.05), while MIA alone elevated expression of pro-inflammatory cytokines Il-1b and Tnfa (p < 0.01). Expression of Il-6 and Il-6r did not differ between treatment groups. Circulating cytokines (TNFα, IL-10, IL-1β, and IL-6) measured in amniotic fluids and fetal cord blood were below detection levels.

5.4.3 Few inflammatory genes are impacted by maternal viral infection in the fetal brain Whole fetal hippocampal and hypothalamic tissue was examined for changes in inflammatory gene expression at 7 and 21 dpi. There were no differences in gene expression in the hippocampus at 7 dpi (Table 5.2); however, at 21 dpi there was an interaction between MIA and sex on Il-1b expression (p < 0.05; Table 5.3). There was a similar trend with hippocampal expression of Cx3cl1 (fractalkine; MIA*sex interaction p = 0.10; Table 5.3) at 21 dpi, with female MIA fetuses expressing more Cx3cl1 mRNA compared to MIA males, though expression did not differ between control males and females. In the fetal hypothalamus, MIA caused a reduction in both Il-6 and IL-1 receptor antagonist (Il- 1ra) mRNA expression (p < 0.01 and p < 0.05, respectively; Table 5.2) at 7 dpi, though Il-1b and Stat3 mRNA were unchanged. No differences in hypothalamic inflammatory gene expression between groups were observed at 21 dpi (Table 5.3).

5.4.4 Fetal microglia transiently alter expression of MHCII and phagocytic and chemotactic activity due to maternal viral infection Primary microglia (CD11b+ CD45low) isolated from fetuses at both 7 and 21 dpi were assessed using flow cytometry for expression of antigen-presenting and phagocytic markers MHCII and CD68, respectively. All primary microglia expressed CD45 at low levels, confirming that they were microglia and

not monocytes/macrophages (representative scatter plots at 7 dpi shown in Supplemental Fig. S5.1 and at 21 dpi shown in Supplemental Fig. S5.2). At 7 dpi, primary microglia from MIA fetuses expressed more MHCII (Fig. 5.3A; p < 0.0001) but not CD68 (Fig. 5.3B; p > 0.10) compared to controls; overall, percent of primary microglia expressing both markers (MHCII+CD68+) increased due to MIA (Fig. 5.3C; p < 0.0001). There was no effect of sex on MHCII or CD68 expression at 7 dpi. Isolated primary microglia were also assessed for phagocytic and chemotactic activity in vitro. Microglia isolated from fetuses at 7 dpi displayed decreased phagocytosis (Fig. 5.3D; p < 0.0001) and chemotaxis (Fig. 5.3E; p < 0.0001), regardless of in vitro stimulation. Microglia isolated from fetuses at the 21 dpi time point still expressed more MHCII due to MIA (p < 0.01, Fig. 5.4A), though CD68 expression (Fig. 5.4B) and co-expression of both markers (Fig. 5.4C) did not differ between treatment groups, indicating a partial resolution of microglial inflammatory phenotype by 21 dpi. There was, however, a trending effect of sex on co-expression of MHCII and CD68 (p = 0.10) such that males tended to have a higher percentage of microglia expressing both markers compared to females, regardless of maternal treatment (control female: 1.7 ± 0.3%, control male: 2.0 ± 0.3%; MIA female: 1.6 ± 0.2%, MIA male: 2.2 ± 0.3%). Interestingly, the percent of CD11b+ CD45low cells expressing CD68 significantly decreased from 7 to 21 dpi (p < 0.0001; 7 dpi: 28.5 ± 0.6%, n = 79; 21 dpi: 11.9 ± 0.8%, n = 97). Microglia isolated at 21 dpi no longer differed in phagocytic activity due to MIA, though pretreatment with either LPS or poly I:C lowered phagocytosis (main effect of in vitro treatment, p < 0.01; Fig. 5.4D). Chemotaxis remained reduced in microglia isolated from MIA fetuses at 21 dpi (p < 0.05; Fig. 5.4E), though to a lesser extent compared to 7 dpi.

5.4.5 Gestational day, and not maternal viral infection, impacts microglial production of TNF� in vitro Microglia isolated from fetuses at GD 83 (± 1 d) and GD 97 (± 1 d) were stimulated with poly I:C or LPS in culture to assess inflammatory response; cell culture media was assayed for TNFα protein 4 hrs post-stimulation. Interestingly, MIA did not impact production and release of TNFα by these cells, but gestational timing did, such that microglia isolated at GD 83 and treated with saline or LPS produced more TNFα at baseline and post-LPS stimulation compared to microglia isolated at GD 97 (in vitro treatment*gestational day interaction, p < 0.0001; Fig. 5). However, no such effect was observed in cells treated with poly I:C, as TNFα concentrations were comparable across time points and maternal treatment group.

5.4.6 Maternal viral infection alters microglial development, sensome, and inflammatory genes at both 7 and 21 dpi

Using high-throughput qPCR, 47 target genes (listed in Supplemental Table S5.2) involved in neurodevelopment, the microglia sensome, and inflammation, which are highly enriched in microglia, were examined to assess the response of fetal microglia to MIA at both 7 and 21 dpi. Il-17a and Mip-2 failed to amplify, leaving data on 45 total genes. Though only 5 genes were altered due to MIA alone at 7 dpi, 15 genes were differentially regulated between females and males, and 11 genes were significantly altered by the interaction of MIA and sex at this time point (Fig. 5.6A). Specifically, MIA caused a reduction in the expression of brain-derived neurotrophic factor (Bdnf), crystallin beta B1 (Crybb1), chemokine IL-8 beta receptor Cxcr2, interferon induced protein with tetratricopeptide repeats 3 (Ifit3), and Il-6 compared to controls (marked with blue asterisks; Fig. 5.6A), indicating that genes primarily involved in neurodevelopment and inflammation are differentially regulated by MIA alone at 7 dpi. While Fc receptor Fcgr2b and anti-inflammatory cytokine Il-8 tended to also be downregulated due to MIA, transforming growth factor beta receptor 1 (Tgfbr1) and solute carrier family 2 member 5 (Slc2a5), which encodes the GLUT5 fructose transporter, tended to be upregulated in MIA microglia, though no comparison reached significance. The most dramatic changes in gene expression were observed when comparing microglia of female fetuses compared to male fetuses, with females demonstrating a high expression, and males a lower expression, of all differentially regulated genes (marked with red asterisks; Fig. 5.6A). Interestingly, MIA caused a shift in this paradigm, where gene expression in MIA females did not differ from that of MIA males in all affected genes (marked with yellow asterisks; Fig. 5.6A; means separations listed in Supplemental Table S5.4). The most striking example of this effect can be seen in the expression of type II interferon Ifng, where MIA caused an upregulation in male fetuses, but a downregulation in female fetuses (Table S5.4). At 21 dpi, only 3 genes were impacted by the interaction between MIA and sex, while 13 genes were altered due to MIA, and 20 genes were altered across females and males; however, there were significant main effects of both MIA and sex on many of these genes, despite the interaction not reaching significance (Fig. 5.6B). Expression of Bdnf, Crybb1, Cxcr2, and Il-6 remained reduced at 21 dpi due to MIA (main effect of MIA: p < 0.05, p < 0.001, p < 0.01, and p < 0.01 respectively), similar to 7 dpi; however, these reductions were considerably more dramatic at 21 dpi compared to 7 dpi. Likewise, the transcription factor early growth response gene 1 (Egr1) and Ifng were reduced in MIA microglia compared to control (main effect of MIA, p < 0.05 and p < 0.01, respectively), though these genes had a notably high expression in control animals. The remaining genes impacted by MIA were increased (transforming growth factor beta [p < 0.05; Tgfb], chemokine C-C motif receptor-like 2 [p < 0.01; Ccrl2], Cd86 antigen [p < 0.01], Fc receptor Fcgr3 [p < 0.05], lymphocyte antigen 86 [p < 0.001; Ly86], and tumor necrosis factor receptor 2

[p < 0.05; Tnfrsf1b]), indicating disparate regulation by MIA across developmental and sensome genes at 21 dpi. A similar differential regulation was observed for sex (red asterisks; Fig. 5.6B). Though most genes affected by sex were increased in females compared to males at 21 dpi (like at 7 dpi), this pattern was reversed for developmental genes Bdnf (main effect of sex, p < 0.05), Crybb1 (main effect of sex, p < 0.05), Cxcr2 (main effect of sex, p < 0.01), and macrophage colony stimulating factor 1 (main effect of sex, p < 0.05; Csf1), as well as for inflammatory cytokine Il-6 (main effect of sex, p < 0.01). There was an interaction between MIA and sex on sensome genes Cd74 and Fcer1g (yellow asterisks; Fig. 5.6B), where MIA normalized the disparity between females and males (means separations listed in Supplemental Table S5.4). This pattern, however, did not continue for cytokine Il-8, where MIA instead caused a reduction in females (Table S5.4). A small subset of microglial genes of interest was also examined using standard qPCR (Tables 5.2 and 5.3). There were no differences in Il-6r mRNA in microglia at 7 dpi, despite the reduction in Il-6 at both 7 and 21 dpi; however, at 21 dpi, expression of Il-6r was increased in MIA microglia (p < 0.01; Table 5.3). This receptor was also differentially regulated by sex at 21 dpi (p < 0.01), with males having lower expression compared to females. Expression of Stat3, downstream of IL-6 signaling, was not impacted at either time point. Interestingly, though the percent of MHCII+ primary microglia increased at both 7 and 21 dpi due to MIA, there was only a trending interaction between MIA and sex on SLA-DRA (MHCII) gene expression, and only at 7 dpi (p = 0.10; Table 5.2). We have previously found that Ccl2 mRNA in whole hippocampal tissue tends to be decreased in male fetuses compared to female [249]; examination of this chemokine in microglia revealed a tendency for greater expression in males compared to females at 21 dpi (main effect sex, p = 0.08; Table 5.3), in contrast to our previous findings. Expression of cytokine- producing transcription factor Nf-kb in microglia was unaffected at 7 dpi, but tended to be downregulated by MIA at 21 dpi (main effect MIA, p = 0.07; Table 5.3), with a tendency for decreased expression mostly in MIA female microglia (MIA*sex interaction, p = 0.09; Table 5.3).

5.4.7 Genes involved in neurodevelopment, the microglial sensome, and inflammation are altered in the fetal amygdala by maternal viral infection The same set of microglial-enriched genes (Table S5.2) were assessed in fetal amygdala tissue to investigate the impacts of MIA on whole brain tissue, and specifically a region involved in social behavior. Previous data from our laboratory using this swine MIA model revealed that early postnatal sociability is decreased in MIA offspring compared to controls [221]. We hypothesized that this gene set would be differentially regulated by MIA during fetal development, and could contribute to changes in social

behavior postnatally. Indeed, at 7 dpi, a total of 21 genes were impacted by the interaction between MIA and sex (yellow asterisks), while 3 genes were altered by MIA alone (blue asterisks), and 3 genes by sex alone (red asterisks; Fig. 5.7A). Ifng was excluded from this data set due to inconsistent amplification. Overall, most differentially expressed genes in the 7 dpi amygdala were downregulated in control females compared to control males, and the interaction between MIA and sex resulted in an upregulation in MIA females and a subsequent downregulation in MIA males (means separations listed in Supplemental Table S5.5). Interestingly, gene expression of the cell surface marker Cd4 (typically expressed by peripheral immune cells like T helper cells, monocytes/macrophages, and dendritic cells, but also induced in microglia in some pathological conditions [252]) tended to be downregulated in males (p = 0.08; Fig. 5.7A), in contrast to all other genes impacted by sex or the interaction between MIA and sex. Only 3 genes were impacted by sex alone at 7 dpi (purinergic receptor P2ry12, cell-adhesion molecule selectin P ligand [Selplg], and IL-6), with females having decreased expression compared to males, regardless of MIA treatment. In comparison, only 3 genes were solely regulated by MIA at 7 dpi (Bdnf, Il-1b, and Tnfa), where MIA caused an increase in expression regardless of sex. At 21 dpi, 5 total genes were impacted by sex (red asterisks) and 17 total genes were differentially regulated by MIA (blue asterisks; Fig. 5.7B); there was no interaction between MIA and sex on amygdala gene expression at 21 dpi. Of the 5 genes differentially regulated by sex (Cdk1, de-novo methyl transferase Dnmt3a, Fc receptor Fcer1g, purinergic receptor P2ry6, and tyrosine kinase binding protein Tyrobp), expression in females was downregulated compared to males (p < 0.05); this held true for complement component C5 and MHCII chaperone Cd74, though differential expression of these genes did not reach significance (p = 0.06 and p = 0.09, respectively; Fig. 5.7B). The majority of genes impacted by MIA alone (11 out of 17) were downregulated (including 9 genes that did not reach significance, marked with a blue hash sign; Fig. 5.7B), though expression of Csf1, Cxcr2, Ccr5, P2ry6, Tlr2, and Il-6 increased due to MIA.

5.4.8 Number and morphology of fetal hippocampal microglia was relatively unchanged by MIA Previous data from our laboratory indicated that MIA reduces total neuron number in the fetal hippocampus, specifically in the dentate gyrus [249]. We hypothesized that altered microglial density and activation in this region could contribute to this loss. To assess this possibility, microglia number and morphology were analyzed in the dentate gyrus and hilar regions of the fetal hippocampus through Iba1 immunohistochemical staining. Though MIA did not impact total microglia number at 7 dpi, there was a trending effect of sex on total microglia in this region (p = 0.10; Fig. 5.8B), where males tended to have more total microglia compared to females. At 21 dpi, there was a trending interaction between MIA and

sex (p = 0.07; Fig. 5.8D). Gestational timing did not impact microglia number in the dentate gyrus and hilar region (p > 0.10; GD 83: 346 ± 20 Iba1+ cells per mm2; GD 97: 310 ± 18 Iba1+ cells per mm2). Though MIA did not significantly impact microglia number, a skewing of microglia towards a more phagocytic phenotype might also contribute to a loss of neurons, thus microglia morphology was assessed by categorizing each cell into one of four activation phenotypes (represented in Fig. 5.8A). However, at 7 and 21 dpi, microglia morphologies in the dentate gyrus and hilar region did not significantly differ between treatment groups (Fig. 5.8C and E), though females tended to have more amoeboid microglia compared to males at 21 dpi (p = 0.054; Fig. 5.8E). As brain development progressed from GD 83 to GD 97, the total number of fetal microglia displaying thick, long processes decreased (p < 0.05; GD 83, 7 dpi: 46 ± 6 Iba1+ cells per mm2; GD 97, 21 dpi: 25 ± 7 Iba1+ cells per mm2); no other microglia morphologies were impacted by gestational timing within the dentate gyrus and hilus.

5.4.9 MIA alters microglia number, but not morphology, in the fetal amygdala As piglets born to PRRSV-infected gilts demonstrate reduced social behavior during early life [221], we hypothesized that fetal microglia may be altered specifically in brain regions involved in social behavior, such as the amygdala. Indeed, examination of microglia number revealed a significant increase in Iba1+ cells per mm2 in the amygdala of MIA fetuses at 7 dpi (p < 0.05, Fig. 5.9C inset), though this did not persist to 21 dpi (Fig. 5.9E inset). However, the tendency for female fetuses to have more total microglia in the amygdala compared to males (regardless of treatment) was evident at both 7 (p < 0.056; Fig. 5.9C) and 21 dpi (p = 0.10; Fig. 5.9E). Advancement of gestational timing significantly increased the total number of microglia within the fetal amygdala (p < 0.0001; GD 83: 158 ± 15 Iba1+ cells per mm2; GD 97: 295 ± 11 Iba1+ cells per mm2; representative images in Supplemental Figure S5.3A and B). Despite these differences in total microglia number, microglia morphology within the amygdala did not differ due to maternal treatment at 7 dpi (Fig. 5.9D) or 21 dpi (Fig. 5.9F). However, in agreement with the increase in total microglia due to gestational timing, there was an increase in the number of amoeboid microglia, microglia with stout processes, and microglia with long, thick processes at GD 97 compared to GD 83 (p < 0.001; Supplemental Figure S5.3); total microglia with thin, ramified processes did not differ across gestation.

5.5 Discussion In the current study, we demonstrate for the first time that fetal microglia from a gyrencephalic species are transiently altered by maternal viral infection. As we have reported previously, MIA in swine reduces overall fetal brain weight, but not body weight [249], emphasizing the global impact of MIA on

brain development emerging within 3 weeks post-inoculation. Inflammation at the maternal-fetal interface, a likely pathway through which maternal cytokines signal to the fetal brain [27, 36, 43], was present even after maternal symptoms resolved. Global changes in fetal microglia were evidenced first by increased expression of MHCII and changes in phagocytic and chemotactic capacity, with evidence of normalization by 21 dpi. Second, widespread modifications in microglial gene expression indicated differential regulation of processes vital to neurodevelopment and to sensing of endogenous signals and danger molecules. Finally, microglia density, but not morphology, was impacted specifically in the fetal amygdala, a brain region integral to the control of social behaviors. Our data also indicate that certain microglial characteristics, such as production of TNFα and cell morphology and density in the hippocampus and amygdala are specifically impacted by gestational timing, coinciding with maturation and colonization patterns of newborn microglia [253, 254]. We also observed preliminary sexual dimorphisms among microglial gene expression patterns and cell distribution and morphology in the fetal brain, which align with current concepts in the field [109]. Maternal cytokines, particularly IL-6 [26, 27, 36] and IL-17a [25, 38, 39], have been implicated as the chief mediators of altered neurodevelopment and behavior in MIA models. Here, we show that pregnant gilts infected with PRRSV have increased levels of IL-6 as well as TNFα in circulation, and that these elevated levels persist even after sickness symptoms have resolved. Although swine do have TH17 cells that produce IL-17a [255, 256], maternal plasma IL-17a was not impacted by PRRSV infection, suggesting that translation of maternal inflammatory signaling to the fetal brain may not involve IL-17a in our model as it does in rodents [210]. Additionally, though fetal tissues were examined for IL-17a mRNA, there was poor amplification across all tissues; IL-17R mRNA was not measured. Nonetheless, increases in circulating plasma IL-6 and TNFα were mirrored through increased gene expression of both cytokines in endometrial tissue at 7 dpi. Interestingly, CXCR3, a chemokine receptor expressed on TH1 cells [257], was increased at 21 dpi, suggesting effector T cell infiltration in endometrial tissue. This aligns with the 3 and 15 fold tissue-specific increase in IFNγ at 7 and 21 dpi, respectively, as IFNγ not only induces CXCR3 ligands [257] and directs TH1 cell differentiation, but is also produced by TH1 cells; TNFα is also produced by these cells, which may contribute to the 3 fold increase in Tnfa expression at 21 dpi. Previous studies have found changes in placental gene expression [43, 46], including neurotrophic genes [45, 47], following MIA, though differential regulation of gene expression did not reach significance in porcine placental tissue until 21 dpi. Increases in pro-inflammatory cytokines TNFα and IL-1β were evident, along with differential regulation of BDNF and IL-10 (although expression was low) due to the interaction between MIA and sex. Unlike humans and rodents, swine have a non-invasive epitheliochorial

placenta, where placental trophoblasts do not directly contact maternal blood. This difference in tissue structure at the maternal-fetal interface may delay inflammatory signaling from the maternal endometrium to fetal tissues, which would explain why changes in placental gene expression were not evident within the first week of inoculation, even though inflammatory genes were increased in maternal tissues at this time. Contrary to evidence presented in rodent models [27, 36], Il-6 gene expression was not upregulated in the porcine placenta at either time point; however, it is important to note that placentas from rodent studies were assessed within hours of MIA induction, which likely does not physiologically align with the 7 and 21 dpi time points presented here. Thus, signaling of IL-6 through the placenta cannot be ruled out in our model. Consistent with previous data [27], Il-6r expression was unchanged between treatment groups. Major histocompatibility complex class II (MHCII) is an antigen presenting molecule upregulated by activated microglia during pathological CNS states (usually in response to IFNγ [258, 259]), as well as during postnatal PRRSV infection [163, 167]. CD68 is expressed on lysosomes and the plasma membrane and is used as an indicator of active phagocytosis [260]; though both MHCII and CD68 can be used as markers for microglia, expression patterns fluctuate with changes in cell morphology and activation status [259, 261], and CD68 in particular can be expressed by both ‘alternatively’ or ‘classically’ activated cells. Both CD68 and MHCII are expressed by microglia in the developing brain [60, 253], and are often upregulated during development to enhance phagocytic capacity. Here we show that even though MHCII was expressed at low levels within both treatment groups, microglia from MIA fetuses displayed a significant 2-3% increase in this antigen presenting molecule at both time points, indicating a slight shift in activation status due to MIA. Expression of SLA-DRA (MHCII) mRNA tended to be impacted by the interaction between MIA and sex only at 7 dpi, such that MIA females had the highest expression. On the contrary, CD68 expression in microglia was not impacted by MIA, though there was an effect of gestational day. At 7 dpi, approximately 28.5% of microglia express CD68 compared to only 12% at 21 dpi, which aligns with the phenotypes of embryonic rodent and non-human primate microglia that are primarily amoeboid and phagocytic during development [60], but shift to a more ramified state as they mature. Our data indicate that microglial CD68 expression may therefore be a good marker for cell maturation during brain development, though it should be acknowledged that use of this marker in flow cytometric techniques without cell permeabilization (as was the case here) only represents surface CD68 expression, and not intracellular expression. Only a small percentage of cells expressed both MHCII and CD68 at either time point, in agreement with previous reports [259], though at 21 dpi there was a tendency for male microglia to have a higher expression of both markers. This suggests that male fetuses tend to have more

activated microglia compared to females at this time in swine gestation, as is the case in early postnatal rodent neurodevelopment [86]. Despite similar expression of CD68 across treatment groups, measurement of phagocytosis in culture revealed that MIA microglia had reduced phagocytic capacity compared to control microglia at 7 dpi, though there were no differences at 21 dpi. This is in contrast to rodent data suggesting that MIA activates microglia and leads to a reduction in the neuroprogenitor pool [60], however, direct phagocytic capacity of embryonic rodent microglia was not investigated. The reduction in overall fetal brain weight observed in the current study could be in agreement with a diminution of the neuroprogenitor pool, yet there is also data indicating that inhibition or ablation of microglia (i.e. removal of trophic support) can induce neuron death or inhibit neurogenesis and oligodendrogenesis in the developing cortex [248] or subventricular zone [262], which could logically cause a decrease in overall brain weight as well. We have previously shown that MIA in swine results in a reduction in estimated neuron density in the hippocampus evident several days before birth, with no differences in cell proliferation [249]. Overall, there is evidence to suggest that perturbing microglia during neurodevelopment has significant effects on neuron and oligodendrocyte survival, but whether this can be attributed solely to (1) an increase in phagocytosis of otherwise healthy neuroprogenitors or to (2) a removal of trophic support, is unknown. Thus, further studies are needed to investigate microglia phagocytosis and neuron cell death in MIA models across multiple brain regions in vivo. Interestingly, there was no effect of in vitro treatment on microglia phagocytosis at 7 dpi in the current study, yet at 21 dpi, stimulation with either LPS or poly I:C caused a reduction in the uptake of E. coli by both control and MIA microglia, contrary to previous data from our lab [163]. This could indicate that fetal porcine microglia respond differently to antigen stimulation in vitro (at least in their phagocytic capacity) compared to neonatal porcine microglia, though further studies are needed to confirm this. Intriguingly, production of inflammatory cytokine TNFα by fetal microglia in vitro did not appear to be impaired, as TNFα protein in cell culture media was increased by both poly I:C and LPS, mirroring previous results [163]. Thus, basic inflammatory responses appear to be intact in porcine fetal microglia, and though production of TNFα is not affected by MIA, gestational timing reduced the total protein that was produced. This is again in agreement with the general downregulation of activated or inflammatory phenotypes of embryonic microglia as brain development progresses and the cells mature [253, 262]. Investigation of microglial-enriched genes in isolated primary microglia and in whole amygdala tissue revealed wide-spread changes in gene expression profiles across both time points and tissues. In general, markers that are expressed during quiescent or homeostatic microglial states were

downregulated. These included developmental and sensome genes, like trophic factors and purinergic receptors. Neurotrophin BDNF, which is produced by microglia to promote new synapse formation [263], was reduced by MIA at both 7 and 21 dpi; however, microglial contributions to total brain BDNF concentrations are minimal. Indeed, BDNF was increased in whole amygdala tissue at 7 dpi, and returned to baseline by 21 dpi, which is likely a more suitable indication of brain BDNF responses to MIA, though production by microglia cannot be entirely discounted. Microglial production of TGF-β and activation of its receptor on these cells is integral for proper microglia development [264] and for resolution of neuroinflammation [265]. During embryonic development, TGF-β signaling is vital for microglial cell survival [264] and helps maintain a more quiescent or homeostatic phenotype after cell maturation [266, 267]. Production of TGF-β by microglia is immunomodulatory and is usually released from ‘M1-like’ alternative microglia [265]. This cytokine is also involved in synaptic pruning, as production of TGF-β by astrocytes initiates C1q deposition on neighboring neurons, targeting them for phagocytosis by microglia [82]. At 7 dpi, we observed an interaction between MIA and sex on microglial TGF-β gene expression, with a general reduction in expression by MIA fetuses compared to control females. Decreased expression in males was still apparent at 21 dpi, though MIA fetuses now expressed more TGF-β compared to controls. TGF-β mRNA was not impacted in whole amygdala tissue at 7 dpi, though it was downregulated at 21 dpi. Expression of integrin β5 (Itgb5), involved in the extracellular storage and activation of TGF-β [268], was not impacted by MIA in microglia, though its expression tended to be downregulated at 21 dpi in whole amygdala tissue. Overall, this suggests a downregulation of microglial TGF-β immunomodulatory action. Purinergic receptors (like P2RY12 and P2RY6) allow microglia to sense purines (like ATP and UDP), to which they respond by migrating or extending processes towards the source (usually dead or damaged cells). ATP can act at synapses or surround injured or apoptotic cells and typically causes responding microglia to become hyper-ramified [269]. At the synapse, P2RY12 activation aids in the eventual phagocytosis of inactive synapses during pruning [270]. During experimentally-induced neuronal hyperactivity, P2RY12 is required for extension of ramified microglial processes around swollen axons, initiating neuroprotective membrane repolarization and phagocytic clearance of debris [76]. Thus, P2RY12 can also be used as a marker for assessing cell activation as it is highly expressed during resting/surveying states, but considerably reduced during classical amoeboid activation [261, 269]. Here, we show that P2ry12 mRNA is reduced in MIA microglia at 7 dpi, and that in control conditions, male microglia have lower expression compared to females (in agreement with general sexual dimorphisms in microglial

activation observed in developing rodent brains [109]), though this dichotomy is reversed with MIA. At 21 dpi, expression of microglial P2ry12 returned to baseline, with a tendency again for males to have lower expression compared to females. These data suggest that fetal porcine microglia may be transiently shifted towards a more activated phenotype at 7 dpi, which has resolved by 21 dpi. Activation of P2RY6 on microglia (through UDP) also causes increased phagocytic activity [271]. Though MIA did not impact the expression of this receptor on microglia, P2ry6 was increased in amygdala tissue at 21 dpi, altogether suggesting that there is heterogeneous control of purinergic receptor expression (and thus cell chemotaxis and phagocytosis) in response to prenatal insult across tissues. Microglial chemotaxis is also controlled by binding of chemokines to their cell-surface receptors, such as CXCR2 and CCR5. Chemotactic capacity of primary fetal microglia measured in vitro indicated a significant decline at 7 dpi, though this was less pronounced at 21 dpi. Analysis of microglial gene expression revealed substantial downregulation of Cxcr2, the beta receptor for IL-8, which could partly explain the reduction in chemotaxis observed in vitro (although cell migration in response to saline was similar to chemokines CCR2 and IL-8). Overexpression of neuronal CCR5 has recently been implicated as a potent suppressor of synaptic plasticity and learning and memory [272]; in the porcine fetal amygdala, MIA increases Ccr5 expression at 21 dpi, though further research would be required to determine if this could result in similar pathologies. Recent evidence also shows that CCR5 is upregulated in neonatal microglia in response to MIA [67]. Here, we show that Ccr5 tends to be increased in 21 dpi microglia due to MIA, though most changes in Ccr5 are due to sex. In fact, rodent data indicates that chemokines like CCL4 (which binds CCR5) are overexpressed in embryonic male brains compared to females [109] during normal neurodevelopment. Interestingly, male porcine fetal microglia had lower expression of Ccr5 mRNA compared to females, indicating that fetal microglia may be compensating for an otherwise exaggerated cell recruitment to certain brain areas, though CCL4 was not measured in fetal porcine tissue and further investigations are needed. Altered expression of other developmental and sensome genes, like macrophage colony stimulating factor Csf1, cell-cycle genes Cdk1 and Egr1, and Tyrobp (which encodes DAP12), were also evident. CSF-1 receptor (CSF-1R), expressed constitutively on microglia, is essential for microglial colonization and proliferation during development and for proliferation and survival throughout the lifespan [128, 253]. Although CSF-1R was not measured here, Csf1 is especially enriched in microglia during neurodevelopment [67]. Csf1 tended to be decreased in female microglia at 7 dpi, and was robustly reduced at 21 dpi in MIA microglia, suggesting a decline in autocrine signaling and potentially in cell proliferation. In MIA amygdala tissue, Csf1 was increased at both 7 and 21 dpi, indicating a potential

induction of cell proliferation in whole tissue. This aligns with the increase in Iba1+ cell density observed in the fetal amygdala at 7 dpi. Likewise, continued expression of Csf1 in this tissue may be involved in the overall increase in microglia number in the fetal amygdala from 7 to 21 dpi. Differential regulation of cell- cycle associated genes, such as cyclin dependent kinase 1 (Cdk1 [128]) and transcription factor early growth response protein 1 (Egr1 [67]) was also evident, with the most drastic effect on Egr1 in microglia at both time points. This indicates reduced cell proliferation in female microglia at 7 dpi, while control microglia upregulate Egr1 2-fold compared to MIA at 21 dpi. Tyrobp (or DAP12) has been implicated in CSF-1-to-CSF-1R signaling, and DAP12 deficiency causes latent reductions in microglial density in specific CNS regions [273]. Interestingly, evidence suggests that expression of DAP12 is also essential for maintaining homeostasis during microglial colonization, as loss- of-function DAP12 mutations in mice transiently increase neonatal microglia density during development [117]. Similar to the latter report, a reduction in Tyrobp mRNA in female porcine fetuses coincides with a tendency for increased amygdalar microglia density in females compared to males at both 7 and 21 dpi. Remarkably, MIA in mice was sufficient to cause increased microglial density and a more activated microglial phenotype, resulting in impaired signaling at synapses similar to what is seen in DAP12 mutant mice [117]. Our data suggest that a comparable phenotype may be induced by MIA in swine. Pattern recognition receptors like TREM2, ITGAM (CD11b), and Fc receptors sense endogenous ligands like β- amyloid, phospholipids, complement, or IgG; each of these receptors has been shown to complex with DAP12 to induce microglial phagocytosis [274-276]. Thus, a reduction in porcine microglial Tyrobp likely impacts a wide range of microglial functions [78], and predominantly coincides with reduced phagocytic capacity of primary microglia in vitro. A similar reduction of microglial phagocytosis was observed recently using an MIA mouse model, though adult and not embryonic microglia were assessed [124]. Notably, Mattei et al. (2017) found that genes involved in microglial phagocytosis were disrupted in their model, including those that complexed with and were involved in signaling of Tyrobp [124]. Similar genes were differentially regulated in porcine fetal microglia in the present study, indicating that alterations to microglial phagocytosis may be a primary route by which MIA impacts both fetal and adult brain function. Though expression of pro-inflammatory cytokines (such as IFNγ, IL-1β, IL-6, and TNFα) by activated microglia during healthy neurodevelopment is necessary for enhancing neurogenesis and oligodendrogenesis [262], overproduction of these cytokines also occurs during brain injury [261], and can be neurotoxic at high levels. Interestingly, there were few indications of classical cytokine responses by microglia at either time point, and in general, production of both anti- and pro-inflammatory cytokine mRNA was reduced. Contrary to rodent data suggesting that IL-6 signaling is upregulated in the fetal brain

[27], there were reductions in microglial IL-6 mRNA at both 7 and 21 dpi. While IL-6 mRNA was reduced in whole amygdala tissue at 7 dpi, expression increased at 21 dpi, indicating that gene expression patterns of IL-6 in whole brain tissue do not mirror that of microglia. Likewise, few changes in inflammatory gene expression in whole fetal hippocampal and hypothalamic tissue were evident. However, mRNA of both IL- 6 and IL-1RA was reduced in the hypothalamus at 7 but not at 21 dpi, which is relatively consistent with the pattern observed in the amygdala. Notably, pro-inflammatory Ifng mRNA was drastically reduced by MIA in microglia at 21 dpi, while the interaction between MIA and sex drove a significant decrease in females and an increase in males at 7 dpi. Expression of interferon-induced Ifit3 was also downregulated by MIA in microglia at 7 dpi, and tended to have differential regulation at 21 dpi due to MIA and sex. Amplification of Ifng was inconsistent across amygdala tissue, and was thus excluded from analyses, while expression of Ifit3 was impacted by the interaction of MIA and sex at 7 dpi, though this normalized by 21 dpi. Aside from a tendency for differential expression due to the interaction between MIA and sex at 7 dpi in microglia, expression of IFNγ receptor (Ifngr1) was generally not impacted by MIA in either tissue. Type II interferon IFNγ is a primary driver of MHCII expression and Ifng expression was induced in male microglia at 7 dpi, though there were no effects of sex on SLA-DRA (MHCII) expression at 7 or 21 dpi. Expression of Cd74 (the MHCII invariant chain, involved in antigen presentation) increased in male but not female microglia at 21 dpi. In whole amygdala tissue, Cd74 was 2.5 fold higher in MIA females compared to controls at 7 dpi, altogether suggesting that the control of microglial MHCII expression may involve pathways outside of IFNγ in the context of MIA. Type I interferon genes are upregulated in microglia during postnatal PRRSV infection [163], though these were not examined here and thus require further examination in this model. One of the genes that was consistently altered due to MIA in microglia across both time points (and tended to be altered within the whole amygdala) was crystallin beta B1 (Crybb1). Intriguingly, Crybb1 has a yet undefined role in microglial homeostasis, though it has been repeatedly shown to be enriched in microglial populations [67, 78]. In our model, Crybb1 expression was consistently downregulated in microglia, and tended to be downregulated in the amygdala as well, which is contrary to previous data demonstrating acute induction of fetal brain crystallin expression in rodents using three different models of MIA [277]. Crystallin proteins play a role in controlling cellular homeostasis under conditions of stress, and in the brain they appear to promote a more immunoregulatory and neuroprotective phenotype [277], specifically in microglia [278, 279]. Dysregulation of Crybb1 has been linked to schizophrenia [280], and downregulation in the medial prefrontal cortex can alleviate anxiety-like behavior in mice [281]. Further research is needed to understand how a downregulation in fetal porcine microglia observed here may

affect microglial immunoregulatory phenotype and postnatal behaviors. Ly86 was also dysregulated in microglia and amygdala, due to both sex and MIA. Ly86 is involved in responding to LPS and is enriched in embryonic mouse and human microglia [78, 112]; its expression is dysregulated in germ-free mice, indicating that this particular gene may be involved in priming microglial innate immune responses [112]. Direct RNA sequencing and network analysis of mouse microglia indicates that LY86 directly interacts with DAP12 [78], and Ly86 and Tyrobp indeed follow similar, if not exact, patterns of expression in fetal piglets in both microglia and amygdala tissue across time. However, the role of LY86 expression in MIA models has yet to be determined. Though our analyses were underpowered for examinations of sexual dimorphisms between treatment groups, our preliminary results suggest that porcine microglial development may follow similar patterns to what is currently observed in rodents. In the DG and hilar region of the hippocampus, male fetal pigs tended to have a higher density of microglia compared to females at 7 dpi, and this aligns with data from P4 rats, where males have more microglia in the hippocampus compared to females [86]. However, female rats appear to have more microglia compared to males in the amygdala at P0, and then less from P4 onwards [86], which misaligns with the tendencies observed in the fetal piglet amygdala, where females tended to have more microglia than males at both 7 and 21 dpi. As much of early rodent brain development continues after birth, microglial colonization and proliferation within the fetal porcine brain during late gestation could logically follow similar patterns to the early postnatal male and female mouse or rat brain, though a thorough characterization of microglial distribution within the fetal pig is needed to confirm this. Interestingly, expression of Itgam mRNA, which encodes for CD11b, was significantly increased in female microglia at both 7 and 21 dpi; however, amygdalar expression of Itgam mRNA was significantly downregulated at 21 dpi but not at 7 dpi, contradictory to the changes in Iba1+ cell density in this region. Iba1+ cell density patterns in the amygdala were mirrored in the DG and hilus at 21 dpi; however, MIA males tended to have more microglia compared to females in this region, suggesting that MIA impacts fetal microglial colonization and/or proliferation a sex- and region-specific manner. Overall, our data suggest that porcine fetal microglial cells show distinct gene expression patterns in response to late gestation MIA, which correspond to altered chemotactic and phagocytic functions, as well as altered density in the fetal amygdala. Importantly, these changes appeared to be transient, as almost all metrics of microglia function and phenotype progressed towards resolution by 21 dpi. However, lingering effects of MIA on gene expression patterns were evident in whole amygdala tissue, as well as a significant reduction in overall brain weight, indicating that altered microglial function, while acute, may

have prolonged detrimental effects. We show that microglial phagocytosis may be particularly sensitive to the challenges of MIA, and that increased microglial density and altered gene expression patterns in the fetal amygdala may regulate social behavior deficits in later life. Physiological gene expression patterns in fetal porcine microglial cells were similar to that of rodents [67, 78], and preliminary data on the effects of sex indicate that sexual dimorphisms during neurodevelopment are a species-wide phenomenon. To our knowledge, this is the first study to demonstrate that fetal microglia in a gyrencephalic species are globally altered by MIA, emphasizing that these cells likely play a substantial role in development of neuropsychiatric disorders in humans.

5.6 Figures and Tables

Figure 5.1

Fig. 5.1. Study design (A): pregnant gilts were inoculated with porcine reproductive and respiratory syndrome virus (PRRSV) on gestational day (GD) 76 and cesarean sections were performed 7 and 21 days post-inoculation (dpi). PRRSV infection resulted in (B) increased body temperature (treatment x time, p < 0.0001) and (C) decreased food intake (treatment x time, p < 0.0001). GD 70-83: Control n = 13, PRRSV n = 14; GD 83-97: Control n = 6, PRRSV n = 7. Bonferroni post hoc: *** = adj p < 0.001, ** = adj p < 0.01, * = adj p < 0.05, # = adj p < 0.10. PRRSV-inoculated gilts had elevated plasma (D) IL-6 and (E) TNFα at both 7 dpi and 21 dpi (main effect of PRRSV: IL-6 = p < 0.0001, TNFα = p < 0.05; no main effect of time or PRRSV x time interaction); 7 dpi: Control n = 5-6, PRRSV n = 5-6, *** = main effect of PRRSV, p < 0.0001; 21 dpi: Control n = 7, PRRSV n = 6, * = main effect of PRRSV, p < 0.05.

Figure 5.2

Fig. 5.2. PRRSV infection resulted in an upregulation of inflammatory genes in maternal endometrial tissue at both (A) 7 dpi and (B) 21 dpi. Tissue was collected immediately adjacent to the umbilical stump of representative fetuses. N = 5 gilts per treatment group, per time point (1-2 representative samples per gilt). Maternal infection induced an upregulation of placental TNFα gene expression at (C) 7 dpi (p = 0.054) and an upregulation of TNFα and IL-1β gene expression at (D) 21 dpi (p < 0.01). There was a significant interaction of MIA and sex at 21 dpi for expression of BDNF and IL-10 (p < 0.05). Placental tissue was collected near the distal end of the umbilical cord; n = 6-11 fetuses per treatment group, per time point. *** = p < 0.001, ** = p < 0.01, * = p < 0.05, # = p < 0.10. MIA = maternal immune activation.

Figure 5.3

Phagocytosis Chemotaxis

Fig. 5.3. Percentage of primary fetal microglia (CD11b+ CD45low) from litters of PRRSV-infected gilts had greater expression of MHCII and co-expression of MHCII and CD68 compared to litters of control gilts at 7 dpi (± 1 d). Fetal microglia from PRRSV-infected litters also tended to express more CD68. Percentage of primary microglia expressing (A) MHCII, (B) CD68, or (C) both MHCII and CD68; n = 4-5 litters per group; n = 39-40 fetuses per group. Primary fetal microglia from litters of PRRSV-infected gilts had decreased phagocytic activity (D; p < 0.0001), and decreased chemotactic activity (E; p < 0.0001) at 7 dpi. There was no effect of in vitro treatment; n = 7-16 fetuses per group; *** = main effect of MIA, p < 0.0001. MIA = maternal immune activation.

Figure 5.4

Phagocytosis Chemotaxis

Fig. 5.4. Percentage of primary fetal microglia (CD11b+ CD45low) from litters of PRRSV-infected gilts had greater expression of MHCII, but not CD68, compared to litters of control gilts at 21 dpi (± 1 d). Percentage of primary microglia expressing (A) MHCII (** = main effect MIA, p < 0.01), (B) CD68, or (C) both MHCII and CD68. N = 5 litters per group; n = 47-50 fetuses per group. Primary fetal microglia from litters of PRRSV-infected gilts had phagocytic activity comparable to controls (D; ** = main effect of in vitro treatment, p < 0.01), but presented with reduced chemotactic activity (E; * = main effect of MIA, p < 0.05; no effect of in vitro treatment) at 21 dpi. N = 7-16 fetuses per group. MIA = maternal immune activation.

Figure 5.5

Fig. 5.5. Maternal infection did not impact production of pro-inflammatory cytokine TNFα by primary microglia stimulated with saline, poly I:C (1 μg/mL), or LPS (1 ng/mL). However, TNFα protein concentration in media was reduced in cells isolated at GD 97 (21 dpi) compared to GD 83 (7 dpi; in vitro treatment x gestational day, p < 0.0001). Four million primary cells per piglet, per treatment; n = 7-16 fetuses per group. *** = individual ANOVA for each in vitro treatment group, main effect of gestational day, p < 0.0001. MIA = maternal immune activation; GD = gestational day.

Figure 5.6

Fig. 5.6. Gene expression assessed in primary fetal microglia at (A) 7 and (B) 21 dpi. Target genes are highly enriched in microglia and involved in neurodevelopment, the microglia “sensome”, and inflammation. Results are expressed as average fold change; n = 7-11 fetuses per maternal treatment group; red symbols indicate a main effect of sex, blue symbols indicate a main effect of MIA, yellow symbols indicate an interaction between MIA and sex; *** = p < 0.001, ** = p < 0.01, * = p < 0.05, # = p ≤ 0.10. Dpi = days post inoculation; MIA = maternal immune activation.

Figure 5.7

Fig. 5.7. Gene expression assessed whole amygdala tissue at (A) 7 and (B) 21 dpi. Target genes are highly enriched in microglia and involved in neurodevelopment, the microglia “sensome”, and inflammation. Results are expressed as average fold change; n = 7-11 fetuses per maternal treatment group; red symbols indicate a main effect of sex, blue symbols indicate a main effect of MIA, yellow symbols indicate an interaction between MIA and sex; *** = p < 0.001, ** = p < 0.01, * = p < 0.05, # = p ≤ 0.10. Dpi = days post inoculation; MIA = maternal immune activation.

Figure 5.8

Fig. 5.8. Maternal infection tended to have few impacts on total number and morphology of fetal microglia in the dentate gyrus and hilar regions of the fetal hippocampus. (A) Representative images of microglia morphologies within the fetal pig brain, classified into four categories of activation, from less active at the top to more active at the bottom. At 7 dpi, males tended to have (B) more total microglia compared to females (p ≤ 0.10), regardless of MIA treatment; however, (C) microglia morphologies did not differ between sex or treatment groups. At 21 dpi, there tended to be (D) an interaction of MIA and sex on total microglia number (p = 0.07) and (E) females tended to have a higher number of amoeboid microglia compared to males, though morphology did not differ between treatment groups. N = 6-10 individual fetuses per treatment group; # = p ≤ 0.10. MIA = maternal immune activation.

Figure 5.9

Fig. 5.9. Maternal infection transiently increased the total number of microglia in the fetal amygdala during peak infection, but did not impact microglia morphology. Representative images of microglia densities at (A) 7 dpi, and (B) 21 dpi. At 7 dpi, MIA fetuses displayed more microglia (C, inset; p < 0.05) in the amygdala compared to controls; additionally, female fetuses tended to express more microglia in the amygdala compared to males (C; p = 0.056). Microglia morphology at 7 dpi (D) did not differ between treatment groups, and most microglia displayed ramified morphology. At 21 dpi, microglia number no longer differed between treatment groups (E; inset), though female fetuses still (E) tended to have more microglia compared to males (p = 0.10). At 21 dpi, (F) microglia morphology in the amygdala did not differ between treatment groups. MIA = maternal immune activation.

Table 5.1. Fetal body and brain weight (by treatment group and time point). 7 dpi 21 dpi Treatment Group Body Wt. (g) Brain Wt. (g) N Body Wt. (g) Brain Wt. (g) N Control 530.9 ± 13.9 14.9 ± 0.2 45 854.7 ± 18.9 25.2 ± 0.3 50 Maternal PRRSV 538.3 ± 16.1 15.3 ± 0.2 39 822.4 ± 27.3 23.6 ± 0.3*** 47 Data are means ± S.E.M. Significant main effect of litter size (p < 0.0001) on all parameters. No differences between treatment groups at 7 dpi for body or brain weight; significant main effect of maternal PRRSV at 21 dpi on brain weight (p < 0.0001; no maternal treatment*litter size interaction), but not body weight. N = 5 litters per treatment group for each time point (20 litters total); *** = p < 0.0001 (bold).

Table 5.2. Gene expression in fetal brain tissue (7 d post inoculation).

100

Table 5.2 (continued).

Values are expressed as relative fold change (mean ± SEM). ** = p < 0.01 (bold), * = p < 0.05 (bold), # = p ≤ 0.10 (italicized). MIA = maternal immune activation.

101

Table 5.3. Gene expression in fetal brain tissue (21 d post inoculation).

102

Table 5.3 (continued).

Values are expressed as relative fold change (mean ± SEM). ** = p < 0.01 (bold), * = p < 0.05 (bold), # = p ≤ 0.10 (italicized).

103

Supplemental Figure S5.1

Supp. Fig. S5.1. Representative scatter plots displaying percent of CD11b+ cells that express CD45 and MHCII. CD11b+ cells isolated from fetal brains of control and MIA piglets at 7 dpi.

Supplemental Figure S5.2

Supp. Fig. S5.2. Representative scatter plots displaying percent of CD11b+ cells that express CD45 and MHCII. CD11b+ cells isolated from fetal brains of control and MIA piglets at 21 dpi.

104

Supplemental Figure S5.3

Supp. Fig. S5.3. Total microglia displaying differing morphologies in the amygdala increased across gestational time point, in synchrony with the increase in total microglia number. Representative images of microglial densities and morphologies at 40X magnification at (A) GD 83 and (B) GD 97. There was an increase in (C) the number of Iba1+ cells classified as amoeboid, having stout processes, or having long, thick processes at GD 97 compared to GD 83 (p < 0.001). The number of microglia classified as having thin, ramified processes did not differ across time points. *** = p < 0.001. GD = gestational day.

105

Supplemental Table S5.1. Quantitative rt-PCR primer information. Gene Accession Number Category Assay IDa Ccl2 NM_214214 Immune Ss03394377_m1 Cd200rb custom Immune custom Cx3cl1 DQ991100 Immune Ss03377157_u1 Cxcr3 AJ851240 Immune Ss03375858_u1 Ifng NM_213948 Immune Ss03391054_m1 Il-10 NM_214041 Immune Ss03382372_u1 Il-1b NM_214055 Immune Ss03393804_m1 Il-1ra NM_214262 Immune Ss03383715_u1 Il-6 NM_214399 Immune Ss03384604_u1 Il-6r NM_214403 Immune Ss03394904_g1 SLA-DRA (MHCII) NM_001113706 Immune ss03389942_m1 Nf-kb NM_001048232 Immune Ss03388575_m1 Rpl19 AF435591 Housekeeping ss03375624_g1 Stat3 NM_001044580 Immune Ss03388426_m1 Tnfa NM_214022 Immune Ss03391318_g1 aApplied Biosystems TaqMan Gene Expression assay identification number. b CD200R custom probe: forward primer, TGTTCCAAGTTACTAATCAGGCTGAA; reverse primer, AGCCCATTAGCAACATGATACTCTTT; probe, ACATAGAATTGAAGGAAGGG

106

Supplemental Table S5.2. Fluidigm rt-PCR primer information. Gene Accession Number KEGG Ortholog Category Applied Biosystems Assay IDa Bdnf NM_214259 K04355 Development Ss03822335_s1 C1qa NM_001003924 K03986 Development Ss03378489_u1 C5 NM_001001646 K03994 Development Ss03391586_m1 Ccr5 NM_001001618 K04180 Sensome Ss03378121_u1 Ccrl2 NM_001001617 K08373 Sensome Ss03378109_u1 Cd14 NM_001097445 K04391 Sensome Ss03818718_s1 Cd180 NM_214357 K06555 Sensome Ss03384320_s1 Cd4 NM_001001908 K06454 Sensome Ss03391676_m1 Cd40 NM_214194 K03160 Sensome Ss03394337_m1 Cd74 NM_213774 K06505 Sensome Ss03381367_u1 Cd86 NM_214222 K05413 Sensome Ss03394401_m1 Cdk1 NM_001159304 K02087 Development Ss03372912_g1 Crybb1 NM_001078681 No KO assigned Development Ss03388726_m1 Csf1 AJ583705 K05453 Development/Sensome Ss03373560_g1 Cxcl16 NM_213811 K10035 Sensome Ss03381587_u1 Cxcr2 AK230995 K05050 Development Ss03375929_g1 Dnmt1 NM_001032355 K00558 Development Ss03392016_m1 Dnmt3a NM_001097437 K17398 Development Ss03385484_u1 Egr1 AJ238156 K09203 Development Ss03373483_s1 Fcer1g NM_001001265 K07983 Sensome Ss03391475_m1 Fcgr2b NM_001033013 K12560 Sensome Ss03392060_m1 Fcgr3 NM_214391 K06463 Sensome Ss03384597_u1 Ifit3 NM_001204395 No KO assigned Development Ss04248506_s1 Ifng NM_213948 K04687 Immune Ss03391054_m1 Ifngr1 NM_001177907 K05132 Sensome Ss04246620_m1 Il-10 NM_214041 K05443 Immune Ss03382372_u1 Il-1b NM_214055 K04519 Immune Ss03393804_m1 Il-17a NM_001005729 K05489 Immune Ss03391803_m1 Il-6 NM_214399 K05405 Immune Ss03384604_u1 Il-8 NM_213867 K10030 Immune Ss03392437_m1 Itgam JF709973 K06461 Immune/Sensome Ss03374588_m1 Itgb5 NM_001246669 K06588 Sensome Ss04322936_m1 Ly86 NM_001097415 No KO assigned Sensome Ss03388794_m1 Cxcl2 NM_001001861 K05505 Immune Ss03378360_u1 Nos2 NM_001143690 K13241 Immune Ss03374608_u1 P2ry12 NM_001173518 K04298 Sensome Ss03373817_s1 P2ry6 NM_001244296 K04272 Sensome Ss03376684_u1 RelA NM_001114281 K00951 Immune Ss03253758_m1 Rpl19 AF435591 N/A Housekeeping ss03375624_g1 Selplg NM_001105307 K06544 Sensome Ss04328834_s1 Slc2a5 EU012359 K08143 Sensome Ss03377332_u1 Tgfb1 NM_214015 K13375 Development Ss03382325_u1 Tgfbr1 NM_001038639 K04674 Sensome/Development Ss03392139_m1 Tlr2 NM_213761 K10159 Sensome Ss03381278_u1 Tnfa NM_214022 K03156 Immune Ss03391318_g1 Tnfrsf1b NM_001097441 K05141 Sensome Ss03385518_u1 Tpi1 NM_001037151 K01803 Development Ss03379689_u1 Tyrobp NM_214202 K07992 Sensome Ss03394361_m1

107

Supplemental Table S5.3. Average litter characteristics. Treatment Group Litter Size Viable Piglets Non-viable Piglets

Control (n = 10) 11.1 ± 1.1 9.5 ± 1.3 1.6 ± 0.6 PRRSV (n = 10) 12.4 ± 1.1 8.7 ± 1.4 2.3 ± 0.8 Data are means ± S.E.M. No significant differences between treatment groups. However, the current study was not designed to detect significant litter differences due to maternal PRRSV infection and thus is underpowered for those analyses. These data are merely meant to summarize the average characteristics of the litters used in the study.

108

Supplemental Table S5.4. Differences of Least Squares Means for the interaction between MIA and sex on fetal microglia gene expression.

109

Supplemental Table S5.4 (continued).

Least squares means separation for each MIA*sex comparison, expressed as unadjusted and Bonferroni adjusted p- values. Comparisons are for gene expression from primary microglia at 7 and 21 d post-inoculation; significant adjusted p-values are in bold; trends are italicized; *** = p < 0.001, ** = p < 0.01, * = p < 0.05; # = p ≤ 0.10. DPI = days post inoculation; F = female; M = male; MIA = maternal immune activation.

110

Supplemental Table S5.5. Differences of Least Squares Means for the interaction between MIA and sex on fetal amygdala gene expression.

111

Supplemental Table S5.5 (continued).

112

Supplemental Table S5.5 (continued).

Least squares means separation for each MIA*sex comparison, expressed as unadjusted and Bonferroni adjusted p- values. Comparisons are for gene expression from amygdala tissue at 7 d post-inoculation; significant adjusted p- values are in bold; trends are italicized; *** = p < 0.001, ** = p < 0.01, * = p < 0.05; # = p ≤ 0.10. DPI = days post inoculation; F = female; M = male; MIA = maternal immune activation.

113

CHAPTER 6: CONCLUSIONS AND FUTURE DIRECTIONS The goal of this dissertation was to characterize the impacts of maternal viral infection on fetal and neonatal brain development, microglial activation, and behavior in a gyrencephalic species in order to delineate mechanisms through which MIA may be modulating the risk for neuropsychiatric disorders in human offspring. Data collected over the course of three major experiments indicate that a live viral infection in pregnant gilts during late gestation induces widespread changes in fetal microglial gene expression and activity, as well as specific changes in gene transcription and glial and neuron densities in the fetal hippocampus and amygdala, culminating in altered social behaviors during the first postnatal month. Overall, our data indicate that microglial dysfunction during the prenatal period in response to MIA could be a contributor to altered neurodevelopment and behavioral deficits observed in human neuropsychiatric disorders. Our initial experiments revealed that the lasting impacts of maternal viral infection on neonatal piglets was confined to their ability to invoke appropriate social responses to conspecifics, and did not involve postnatal microglial priming. Unexpected but vital data from this study proved that long-lasting neuroinflammation and microglial hyper-activation were not a characteristic of the postnatal MIA piglet brain, indicating that the mechanisms mediating the observed aberrant behaviors must have occurred prenatally. This built the foundation for our follow-up experiments. Additional characterization of the prenatal piglet hippocampus showed reduced neuron density as well as astrogliosis several days before piglets reached full term. Again, however, brief probing into microglia phenotype at this time point indicated that these cells were, if anything, less pro-inflammatory, again suggesting that the detrimental impacts of MIA on brain development have already occurred at this point. Our final aim was, therefore, to thoroughly investigate the actions of microglial cells at the peak of maternal infection, when they are likely receiving the strongest inflammatory signaling across the maternal-fetal interface. Additionally, we were interested in characterizing microglial actions at the time when these signals were expected to have dissipated and fetal CNS responses could theoretically return to baseline. Data from these final experiments demonstrated that indeed fetal microglia undergo global changes in pro-inflammatory MHCII expression, gene expression, phagocytosis and chemotaxis, and overall cell density in specific brain regions at the peak of maternal infection. These responses tended to return to baseline, as predicted, following the resolution of outward maternal symptoms, which aligns with the lack of microglial activation observed immediately prior to parturition and within the first postnatal month. These data indicate that altered microglial activity has the undeniable potential to alter

114

neurodevelopmental processes and therefore induce postnatal behavioral abnormalities like those observed in the first study. Notably, our data designate fetal microglia as prime candidates for mediating the detrimental impacts of MIA observed across preclinical and clinical studies. Altogether, this dissertation has provided evidence that supports the following theory (depicted in Fig. 6.1): maternal inflammatory responses, communicated to the developing fetus through placental cytokine signaling, initiate altered activity in fetal microglia during a time of otherwise highly-orchestrated neurodevelopmental processes and brain growth. An overall reduction in microglial chemotaxis, phagocytosis, and trophic support significantly impedes the ability of these cells to aid in neuron outgrowth, synapse formation and pruning, and collection and removal of apoptotic or injured cell debris. Colonization patterns of particular brain regions, such as the amygdala, are disrupted, leading to excess microglial cell accumulation, albeit transient. As a whole, this results in altered neural network formation across regions like the hippocampus (reduced neuron density) and amygdala (acute inflation of microglia density), including failure of proper synaptic connections, that persists even after microglial responses have returned to baseline. Following parturition, the detrimental impacts of microglial malfunction are evident in the emergence of altered social behaviors in neonatal piglets. However, there are conceptual leaps in this proposed mechanism that should be addressed in future investigations. First, extensive research has recently been undertaken to eloquently demonstrate the contribution of pro-inflammatory cytokine IL-17a in a mouse model of MIA [25, 38, 39, 210]. Though we were able to measure circulating plasma IL-17a in pregnant gilts, maternal viral infection did not appear to impact these concentrations, at least not at 7 or 21 dpi. Additionally, we were unable to measure IL-17a or IL-17Ra mRNA in either maternal or fetal tissues, leaving this particular pathway relatively unexplored. Given the compelling evidence suggesting that IL-17a is necessary and sufficient to induce offspring abnormalities in a mouse model, the contributions of IL-17a in our particular swine model warrant further investigation. Additionally, we observed increased microglial numbers in the fetal amygdala at 7 dpi, but it was unclear whether this was due to increased local proliferation or to increased trafficking of microglia to that particular brain region. Using markers of proliferation, incorporated during cell cycling (Ki-67 or Brd-U), along with Iba1 staining in situ will reveal whether or not there is an increase in double-positive cells and thus microglial proliferation. Additionally, examination of the direct contact of microglial processes with (and incorporation of) apoptotic cells, neuroprogenitor cells, or synapses should be addressed to understand if fetal microglia are increasing or decreasing phagocytosis in specific brain regions. Compelling evidence of the phagocytic capacity of embryonic microglia has been demonstrated through immunohistochemical staining in

115

rodents and non-human primates [60, 75, 83], and detrimental impacts of inhibited or ablated microglia demonstrate that these processes are essential to proper neuron numbers and network connections [60, 61, 79]. To fully delineate the consequences of altered microglial phagocytosis observed in the present studies, similar techniques should be employed in the fetal piglet brain. Here, we have shown evidence of astrogliosis in the hippocampus of prenatal piglets due to MIA. Astrocytes, which express many of the same sensome receptors as microglia, are also critical for synaptic pruning [82], and are an essential component of the tetrapartite synapse [95]. Involvement of astrocytes in MIA mechanisms has been proposed previously [207, 208], and warrants further investigation across the field as well as in our particular swine model. As has been demonstrated with rodents, specific cytokine inhibition in pregnant dams is successful in ameliorating the behavioral and neurological dysfunction of offspring [25, 26]. Administering cytokine blocking antibodies in pregnant gilts will allow us to address which particular cytokines, or combination thereof, are mitigating most of the adverse effects summarized thus far. If possible, cytokine tracer experiments should also be performed to assess the exact concentration of pro-inflammatory cytokines in fetal circulation that were acquired through maternal blood, if any, as well as to delineate the contribution of signaling across the placenta. Finally, potential routes to ameliorate microglial dysfunction need to be explored. Previous studies have shown beneficial outcomes with the use of antibiotic tetracycline (minocycline [124, 282]), which specifically inhibits microglial activation. However, the use of minocycline to target prenatal inflammatory responses is difficult, specifically due to the route of administration. One preclinical study has demonstrated the benefits of another tetracycline derivative, doxycycline, administered to pregnant dams [60], though this still raises questions of feasibility (it was given on the same day as the prenatal insult and continued for approximately two weeks), and of unintended secondary consequences (altered gut microbiome and innate immune responses, both peripheral [283] and central [111, 112]) when considering translatability to humans. The importance of the maternal microbiome composition during MIA has recently been demonstrated, and is vital for inducing IL-17a during the response to poly I:C insult in dams [39]. Undoubtedly, maternal antibiotic administration would play a substantial role in these mechanisms, and careful dissection of the microbial changes in dams and in offspring should be performed if tetracycline treatments are to be considered viable therapeutic options. Collectively, our results indicate that acute disruptions in fetal microglial activity immediately following maternal inflammation have prolonged impacts on neurodevelopmental and inflammatory gene expression, neuron density, and postnatal social behaviors. Further investigations into the exact activity of microglial cells, such as trophic capacity and phagocytosis of neuroprogenitors and synapses,

116

as well as overall population densities in different brain regions, are needed to better understand the etiology of anomalous neurodevelopmental processes and behavior controlled by microglia in the context of MIA. Additionally, it will be necessary to define the exact contribution of secondary supportive glial cells (astrocytes) to these phenotypes. Illuminating these mechanisms will allow for the development and refinement of therapeutic interventions for the possible prevention or amelioration of specific psychiatric illnesses.

Figure 6.1

Fig. 6.1. Schematic representation of the proposed mechanisms involved in our MIA swine model. Data from Chapter 5 indicate that there are 7 fetal microglial actions that are altered: (1) shift from a homeostatic phenotype; (2) decreased phagocytosis; (3) decreased chemotaxis; differential expression of genes involved in (4) synaptic pruning and (5) synaptogenesis; differential expression of genes involved in (6) trophic support and (7) cell proliferation. Ultimately, there is increased microglial density in the amygdala at 7 dpi, and reduced neuron density in the hippocampus at 35 dpi. Overall, these changes result in altered social behaviors postnatally.

117

CHAPTER 7: REFERENCES 1. Atladottir, H.O., et al., Maternal infection requiring hospitalization during pregnancy and autism spectrum disorders. J Autism Dev Disord, 2010. 40(12): p. 1423-30. 2. Brown, A.S., Prenatal infection as a risk factor for schizophrenia. Schizophrenia bulletin, 2006. 32(2): p. 200-202. 3. Meyer, U., J. Feldon, and O. Dammann, Schizophrenia and autism: both shared and disorder-specific pathogenesis via perinatal inflammation? Pediatric research, 2011. 69: p. 26R-33R. 4. Kulage, K.M., A.M. Smaldone, and E.G. Cohn, How will DSM-5 affect autism diagnosis? A systematic literature review and meta-analysis. Journal of autism and developmental disorders, 2014. 44(8): p. 1918- 1932. 5. Baio, J., Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years — Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2010, in Surveillance Summaries. 2014, CDC: Morbidity and Mortality Weekly Report (MMWR) Centers for Disease Control and Prevention CDC. p. 1-21. 6. Andari, E., et al., Promoting social behavior with oxytocin in high-functioning autism spectrum disorders. Proc Natl Acad Sci U S A, 2010. 107(9): p. 4389-94. 7. Hollander, E., et al., Oxytocin increases retention of social cognition in autism. Biol Psychiatry, 2007. 61(4): p. 498-503. 8. Werling, D.M. and D.H. Geschwind, Sex differences in autism spectrum disorders. Curr Opin Neurol, 2013. 26(2): p. 146-53. 9. Tandon, R., et al., Definition and description of schizophrenia in the DSM-5. Schizophr Res, 2013. 150(1): p. 3-10. 10. Tandon, R., H.A. Nasrallah, and M.S. Keshavan, Schizophrenia, "just the facts" 4. Clinical features and conceptualization. Schizophr Res, 2009. 110(1-3): p. 1-23. 11. McGrath, J., et al., Schizophrenia: a concise overview of incidence, prevalence, and mortality. Epidemiol Rev, 2008. 30(1): p. 67-76. 12. Lind, N.M., et al., The use of pigs in neuroscience: modeling brain disorders. Neurosci Biobehav Rev, 2007. 31(5): p. 728-51. 13. Dickerson, J.W. and J. Dobbing, Prenatal and postnatal growth and development of the central nervous system of the pig. Proc R Soc Lond B Biol Sci, 1967. 166(1005): p. 384-95. 14. Dobbing, J. and J. Sands, Comparative aspects of the brain growth spurt. Early Hum Dev, 1979. 3(1): p. 79- 83. 15. Pond, W.G., et al., Perinatal ontogeny of brain growth in the domestic pig. Proc Soc Exp Biol Med, 2000. 223(1): p. 102-8. 16. Knuesel, I., et al., Maternal immune activation and abnormal brain development across CNS disorders. Nat Rev Neurol, 2014. 10(11): p. 643-60. 17. Dawson, H., A Comparative Assessment of the Pig, Mouse and Human Genomes, in The Minipig in Biomedical Research. 2011, CRC Press. p. 323-342. 18. Dawson, H.D., et al., Structural and functional annotation of the porcine immunome. BMC Genomics, 2013. 14(1): p. 332. 19. Mair, K.H., et al., The porcine innate immune system: an update. Dev Comp Immunol, 2014. 45(2): p. 321- 43. 20. Meurens, F., et al., The pig: a model for human infectious diseases. Trends Microbiol, 2012. 20(1): p. 50-7. 21. Conrad, M.S., R.N. Dilger, and R.W. Johnson, Brain growth of the domestic pig (Sus scrofa) from 2 to 24 weeks of age: a longitudinal MRI study. Dev Neurosci, 2012. 34(4): p. 291-8. 22. Conrad, M.S., et al., Magnetic resonance imaging of the neonatal piglet brain. Pediatr Res, 2012. 71(2): p. 179-84. 23. Conrad, M.S. and R.W. Johnson, The domestic piglet: an important model for investigating the neurodevelopmental consequences of early life insults. Annu Rev Anim Biosci, 2015. 3(1): p. 245-64.

118

24. Meyer, U., et al., Adult behavioral and pharmacological dysfunctions following disruption of the fetal brain balance between pro-inflammatory and IL-10-mediated anti-inflammatory signaling. Mol Psychiatry, 2008. 13(2): p. 208-21. 25. Choi, G.B., et al., The maternal interleukin-17a pathway in mice promotes autism-like phenotypes in offspring. Science, 2016. 351(6276): p. 933-9. 26. Smith, S.E., et al., Maternal immune activation alters fetal brain development through interleukin-6. J Neurosci, 2007. 27(40): p. 10695-702. 27. Wu, W.L., et al., The placental interleukin-6 signaling controls fetal brain development and behavior. Brain Behav Immun, 2017. 62: p. 11-23. 28. Mouihate, A. and H. Mehdawi, Toll-like receptor 4-mediated immune stress in pregnant rats activates STAT3 in the fetal brain: role of interleukin-6. Pediatr Res, 2016. 79(5): p. 781-7. 29. Gayle, D.A., et al., Maternal LPS induces cytokines in the amniotic fluid and corticotropin releasing hormone in the fetal rat brain. Am J Physiol Regul Integr Comp Physiol, 2004. 286(6): p. R1024-9. 30. Meyer, U., J. Feldon, and B.K. Yee, A review of the fetal brain cytokine imbalance hypothesis of schizophrenia. Schizophr Bull, 2009. 35(5): p. 959-72. 31. Malkova, N.V., et al., Maternal immune activation yields offspring displaying mouse versions of the three core symptoms of autism. Brain Behav Immun, 2012. 26(4): p. 607-16. 32. Bauman, M.D., et al., Activation of the maternal immune system during pregnancy alters behavioral development of rhesus monkey offspring. Biol Psychiatry, 2014. 75(4): p. 332-41. 33. Missault, S., et al., The risk for behavioural deficits is determined by the maternal immune response to prenatal immune challenge in a neurodevelopmental model. Brain Behav Immun, 2014. 42(0): p. 138-46. 34. Buka, S.L., et al., Maternal cytokine levels during pregnancy and adult psychosis. Brain Behav Immun, 2001. 15(4): p. 411-20. 35. Miller, B.J., et al., Prenatal inflammation and neurodevelopment in schizophrenia: a review of human studies. Prog Neuropsychopharmacol Biol Psychiatry, 2013. 42: p. 92-100. 36. Hsiao, E.Y. and P.H. Patterson, Activation of the maternal immune system induces endocrine changes in the placenta via IL-6. Brain Behav Immun, 2011. 25(4): p. 604-15. 37. Gilmore, J.H., et al., Prenatal infection and risk for schizophrenia: IL-1b, IL-6, and TNFa inhibit cortical neuron dendrite development. Neuropsychopharmacology, 2004. 29: p. 1221-1229. 38. Shin Yim, Y., et al., Reversing behavioural abnormalities in mice exposed to maternal inflammation. Nature, 2017. 549(7673): p. 482-487. 39. Kim, S., et al., Maternal gut bacteria promote neurodevelopmental abnormalities in mouse offspring. Nature, 2017. 549(7673): p. 528-532. 40. Ivanov, II, et al., Specific microbiota direct the differentiation of IL-17-producing T-helper cells in the mucosa of the small intestine. Cell Host Microbe, 2008. 4(4): p. 337-49. 41. Ivanov, II, et al., Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell, 2009. 139(3): p. 485-98. 42. Zhou, L., et al., IL-6 programs T(H)-17 cell differentiation by promoting sequential engagement of the IL-21 and IL-23 pathways. Nat Immunol, 2007. 8(9): p. 967-74. 43. Fatemi, S.H., et al., The viral theory of schizophrenia revisited: abnormal placental gene expression and structural changes with lack of evidence for H1N1 viral presence in placentae of infected mice or brains of exposed offspring. Neuropharmacology, 2012. 62(3): p. 1290-8. 44. Bergeron, J., et al., Activation of the IL-1beta/CXCL1/MMP-10 axis in chorioamnionitis induced by inactivated Group B Streptococcus. Placenta, 2016. 47: p. 116-123. 45. Gilmore, J.H., L.F. Jarskog, and S. Vadlamudi, Maternal infection regulates BDNF and NGF expression in fetal and neonatal brain and maternal-fetal unit of the rat. J Neuroimmunol, 2003. 138(1-2): p. 49-55. 46. Urakubo, A., et al., Prenatal exposure to maternal infection alters cytokine expression in the placenta, amniotic fluid, and fetal brain. Schizophr Res, 2001. 47(1): p. 27-36. 47. Gilmore, J.H., L.F. Jarskog, and S. Vadlamudi, Maternal poly I: C exposure during pregnancy regulates TNFα, BDNF, and NGF expression in neonatal brain and the maternal–fetal unit of the rat. J Neuroimmunol, 2005. 159(1): p. 106-112.

119

48. Meyer, U., et al., The time of prenatal immune challenge determines the specificity of inflammation- mediated brain and behavioral pathology. J Neurosci, 2006. 26(18): p. 4752-62. 49. Meyer, U., et al., Adult brain and behavioral pathological markers of prenatal immune challenge during early/middle and late fetal development in mice. Brain Behav Immun, 2008. 22(4): p. 469-86. 50. Meyer, U., Prenatal poly(i:C) exposure and other developmental immune activation models in rodent systems. Biol Psychiatry, 2014. 75(4): p. 307-15. 51. Patterson, P.H., Immune involvement in schizophrenia and autism: etiology, pathology and animal models. Behav Brain Res, 2009. 204(2): p. 313-21. 52. Bauman, M.D., et al., Maternal antibodies from mothers of children with autism alter brain growth and social behavior development in the rhesus monkey. Transl Psychiatry, 2013. 3: p. e278. 53. Roullet, F.I., J.K. Lai, and J.A. Foster, In utero exposure to valproic acid and autism--a current review of clinical and animal studies. Neurotoxicol Teratol, 2013. 36: p. 47-56. 54. Onore, C., M. Careaga, and P. Ashwood, The role of immune dysfunction in the pathophysiology of autism. Brain Behav Immun, 2012. 26(3): p. 383-92. 55. Monji, A., et al., Neuroinflammation in schizophrenia especially focused on the role of microglia. Prog Neuropsychopharmacol Biol Psychiatry, 2013. 42(0): p. 115-21. 56. Morgan, J.T., et al., Microglial activation and increased microglial density observed in the dorsolateral prefrontal cortex in autism. Biol Psychiatry, 2010. 68(4): p. 368-76. 57. Morgan, J.T., et al., Abnormal microglial-neuronal spatial organization in the dorsolateral prefrontal cortex in autism. Brain Res, 2012. 1456(0): p. 72-81. 58. Alliot, F., I. Godin, and B. Pessac, Microglia derive from progenitors, originating from the yolk sac, and which proliferate in the brain. Developmental Brain Research, 1999. 117(2): p. 145-152. 59. Ginhoux, F., et al., Fate mapping analysis reveals that adult microglia derive from primitive macrophages. Science, 2010. 330(6005): p. 841-5. 60. Cunningham, C.L., V. Martinez-Cerdeno, and S.C. Noctor, Microglia regulate the number of neural precursor cells in the developing cerebral cortex. J Neurosci, 2013. 33(10): p. 4216-33. 61. Pierre, W.C., et al., Neonatal microglia: The cornerstone of brain fate. Brain Behav Immun, 2017. 59: p. 333-345. 62. Askew, K. and D. Gomez-Nicola, A story of birth and death: Insights into the formation and dynamics of the microglial population. Brain, behavior, and immunity, 2017. 63. Boche, D., V.H. Perry, and J.A. Nicoll, Review: activation patterns of microglia and their identification in the human brain. Neuropathol Appl Neurobiol, 2013. 39(1): p. 3-18. 64. Schwarz, J.M., P.W. Sholar, and S.D. Bilbo, Sex differences in microglial colonization of the developing rat brain. J Neurochem, 2012. 120(6): p. 948-63. 65. Kettenmann, H., et al., Physiology of microglia. Physiol Rev, 2011. 91(2): p. 461-553. 66. Kettenmann, H., F. Kirchhoff, and A. Verkhratsky, Microglia: new roles for the synaptic stripper. Neuron, 2013. 77(1): p. 10-8. 67. Matcovitch-Natan, O., et al., Microglia development follows a stepwise program to regulate brain homeostasis. Science, 2016. 353(6301): p. aad8670. 68. Low, D. and F. Ginhoux, Recent advances in the understanding of microglial development and homeostasis. Cell Immunol, 2018. 69. Eyo, U.B. and M.E. Dailey, Microglia: key elements in neural development, plasticity, and pathology. J Neuroimmune Pharmacol, 2013. 8(3): p. 494-509. 70. Martinez, F.O. and S. Gordon, The M1 and M2 paradigm of macrophage activation: Time for reassessment. F1000Prime Reports, 2014. 6. 71. Ransohoff, R.M., A polarizing question: do M1 and M2 microglia exist? Nat Neurosci, 2016. 19(8): p. 987- 91. 72. Town, T., V. Nikolic, and J. Tan, The microglial "activation" continuum: from innate to adaptive responses. J Neuroinflammation, 2005. 2: p. 24. 73. Karperien, A., H. Ahammer, and H.F. Jelinek, Quantitating the subtleties of microglial morphology with fractal analysis. Front Cell Neurosci, 2013. 7: p. 3.

120

74. Nimmerjahn, A., F. Kirchhoff, and F. Helmchen, Resting microglial cells are highly dynamic surveillants of brain parenchyma in vivo. Science, 2005. 308(5726): p. 1314-8. 75. Sierra, A., et al., Microglia shape adult hippocampal neurogenesis through apoptosis-coupled phagocytosis. Cell Stem Cell, 2010. 7(4): p. 483-95. 76. Kato, G., et al., Microglial Contact Prevents Excess Depolarization and Rescues Neurons from Excitotoxicity. eNeuro, 2016. 3(3): p. ENEURO. 0004-16.2016. 77. Grabert, K., et al., Microglial brain region-dependent diversity and selective regional sensitivities to aging. Nat Neurosci, 2016. 19(3): p. 504-16. 78. Hickman, S.E., et al., The microglial sensome revealed by direct RNA sequencing. Nat Neurosci, 2013. 16(12): p. 1896-905. 79. Squarzoni, P., et al., Microglia modulate wiring of the embryonic forebrain. Cell Rep, 2014. 8(5): p. 1271-9. 80. Richetto, J., et al., Genome-wide DNA Methylation Changes in a Mouse Model of Infection-Mediated Neurodevelopmental Disorders. Biol Psychiatry, 2017. 81(3): p. 265-276. 81. Stevens, B., et al., The classical complement cascade mediates CNS synapse elimination. Cell, 2007. 131(6): p. 1164-78. 82. Bialas, A.R. and B. Stevens, TGF-beta signaling regulates neuronal C1q expression and developmental synaptic refinement. Nat Neurosci, 2013. 16(12): p. 1773-82. 83. Paolicelli, R.C., et al., Synaptic pruning by microglia is necessary for normal brain development. Science, 2011. 333(6048): p. 1456-8. 84. Vainchtein, I.D., et al., Astrocyte-derived interleukin-33 promotes microglial synapse engulfment and neural circuit development. Science, 2018: p. eaal3589. 85. Chung, W.S., et al., Do glia drive synaptic and cognitive impairment in disease? Nat Neurosci, 2015. 18(11): p. 1539-1545. 86. Bilbo, S.D., S.H. Smith, and J.M. Schwarz, A lifespan approach to neuroinflammatory and cognitive disorders: a critical role for glia. J Neuroimmune Pharmacol, 2012. 7(1): p. 24-41. 87. Bourgeron, T., A synaptic trek to autism. Curr Opin Neurobiol, 2009. 19(2): p. 231-4. 88. Rodriguez, J.I. and J.K. Kern, Evidence of microglial activation in autism and its possible role in brain underconnectivity. Neuron Glia Biol, 2011. 7(2-4): p. 205-13. 89. Zhan, Y., et al., Deficient neuron-microglia signaling results in impaired functional brain connectivity and social behavior. Nat Neurosci, 2014. 17(3): p. 400-6. 90. Kim, H.J., et al., Deficient autophagy in microglia impairs synaptic pruning and causes social behavioral defects. Mol Psychiatry, 2017. 22(11): p. 1576-1584. 91. Gyorffy, B.A., et al., Widespread alterations in the synaptic proteome of the adolescent cerebral cortex following prenatal immune activation in rats. Brain Behav Immun, 2016. 56: p. 289-309. 92. Pendyala, G., et al., Maternal Immune Activation Causes Behavioral Impairments and Altered Cerebellar Cytokine and Synaptic Protein Expression. Neuropsychopharmacology, 2017. 42(7): p. 1435-1446. 93. Dickerson, D.D. and D.K. Bilkey, Aberrant neural synchrony in the maternal immune activation model: using translatable measures to explore targeted interventions. Front Behav Neurosci, 2013. 7: p. 217. 94. Fernandez de Cossio, L., et al., Prenatal infection leads to ASD-like behavior and altered synaptic pruning in the mouse offspring. Brain Behav Immun, 2017. 63: p. 88-98. 95. Dityatev, A. and D.A. Rusakov, Molecular signals of plasticity at the tetrapartite synapse. Current opinion in neurobiology, 2011. 21(2): p. 353-359. 96. Chelini, G., et al., The tetrapartite synapse: a key concept in the pathophysiology of schizophrenia. Eur Psychiatry, 2018. 97. Amaral, D.G., C.M. Schumann, and C.W. Nordahl, Neuroanatomy of autism. Trends Neurosci, 2008. 31(3): p. 137-45. 98. Avino, T.A. and J.J. Hutsler, Abnormal cell patterning at the cortical gray-white matter boundary in autism spectrum disorders. Brain Res, 2010. 1360: p. 138-46. 99. Mercadante, M.T., et al., Neurogenesis in the amygdala: a new etiologic hypothesis of autism? Med Hypotheses, 2008. 70(2): p. 352-7. 100. Belmonte, M.K., et al., Autism and abnormal development of brain connectivity. J Neurosci, 2004. 24(42): p. 9228-31.

121

101. Courchesne, E. and K. Pierce, Why the frontal cortex in autism might be talking only to itself: local over- connectivity but long-distance disconnection. Curr Opin Neurobiol, 2005. 15(2): p. 225-30. 102. Just, M.A., et al., Autism as a neural systems disorder: a theory of frontal-posterior underconnectivity. Neurosci Biobehav Rev, 2012. 36(4): p. 1292-313. 103. Na, K.S., H.Y. Jung, and Y.K. Kim, The role of pro-inflammatory cytokines in the neuroinflammation and neurogenesis of schizophrenia. Prog Neuropsychopharmacol Biol Psychiatry, 2014. 48(0): p. 277-86. 104. Ellman, L.M., et al., Structural brain alterations in schizophrenia following fetal exposure to the inflammatory cytokine interleukin-8. Schizophr Res, 2010. 121(1-3): p. 46-54. 105. Canetta, S., et al., Maternal immune activation leads to selective functional deficits in offspring parvalbumin interneurons. Mol Psychiatry, 2016. 21(7): p. 956-68. 106. Wöhr, M., et al., Lack of parvalbumin in mice leads to behavioral deficits relevant to all human autism core symptoms and related neural morphofunctional abnormalities. Translational psychiatry, 2015. 5(3): p. e525. 107. Nelson, S.B. and V. Valakh, Excitatory/Inhibitory Balance and Circuit Homeostasis in Autism Spectrum Disorders. Neuron, 2015. 87(4): p. 684-98. 108. Frankle, W.G., et al., In vivo measurement of GABA transmission in healthy subjects and schizophrenia patients. Am J Psychiatry, 2015. 172(11): p. 1148-59. 109. Schwarz, J.M. and S.D. Bilbo, Sex, glia, and development: interactions in health and disease. Horm Behav, 2012. 62(3): p. 243-53. 110. Hui, C.W., et al., Prenatal Immune Challenge in Mice Leads to Partly Sex-Dependent Behavioral, Microglial, and Molecular Abnormalities Associated with Schizophrenia. Front Mol Neurosci, 2018. 11: p. 13. 111. Erny, D., et al., Host microbiota constantly control maturation and function of microglia in the CNS. Nat Neurosci, 2015. 18(7): p. 965-77. 112. Thion, M.S., et al., Microbiome Influences Prenatal and Adult Microglia in a Sex-Specific Manner. Cell, 2018. 172(3): p. 500-516 e16. 113. Cohly, H.H.P. and A. Panja, Immunological Findings in Autism, in International Review of Neurobiology, M.D. Dirk, Editor. 2005, Academic Press. p. 317-341. 114. Depino, A.M., Peripheral and central inflammation in autism spectrum disorders. Mol Cell Neurosci, 2013. 53: p. 69-76. 115. Goines, P. and J. Van de Water, The immune system's role in the biology of autism. Curr Opin Neurol, 2010. 23(2): p. 111-7. 116. Vargas, D.L., et al., Neuroglial activation and neuroinflammation in the brain of patients with autism. Ann Neurol, 2005. 57(1): p. 67-81. 117. Roumier, A., et al., Prenatal activation of microglia induces delayed impairment of glutamatergic synaptic function. PLoS One, 2008. 3(7): p. e2595. 118. Van den Eynde, K., et al., Hypolocomotive behaviour associated with increased microglia in a prenatal immune activation model with relevance to schizophrenia. Behav Brain Res, 2014. 258(0): p. 179-86. 119. Kato, T.A. and S. Kanba, Are microglia minding us? Digging up the unconscious mind-brain relationship from a neuropsychoanalytic approach. Front Hum Neurosci, 2013. 7(13): p. 13. 120. Bilbo, S.D. and J.M. Schwarz, Early-life programming of later-life brain and behavior: a critical role for the immune system. Front Behav Neurosci, 2009. 3: p. 14. 121. Giovanoli, S., et al., Late prenatal immune activation causes hippocampal deficits in the absence of persistent inflammation across aging. J Neuroinflammation, 2015. 12(1): p. 221. 122. Giovanoli, S., et al., Prenatal immune activation causes hippocampal synaptic deficits in the absence of overt microglia anomalies. Brain Behav Immun, 2016. 55: p. 25-38. 123. Arsenault, D., et al., The different effects of LPS and poly I:C prenatal immune challenges on the behavior, development and inflammatory responses in pregnant mice and their offspring. Brain Behav Immun, 2014. 38(0): p. 77-90. 124. Mattei, D., et al., Maternal immune activation results in complex microglial transcriptome signature in the adult offspring that is reversed by minocycline treatment. Transl Psychiatry, 2017. 7(5): p. e1120.

122

125. Ratnayake, U., et al., Behaviour and hippocampus-specific changes in spiny mouse neonates after treatment of the mother with the viral-mimetic Poly I:C at mid-pregnancy. Brain Behav Immun, 2012. 26(8): p. 1288-99. 126. Nielsen, P.R., et al., Neonatal levels of inflammatory markers and later risk of schizophrenia. Biol Psychiatry, 2015. 77(6): p. 548-55. 127. Lawson, L.J., V.H. Perry, and S. Gordon, Turnover of resident microglia in the normal adult mouse brain. Neuroscience, 1992. 48(2): p. 405-15. 128. Askew, K., et al., Coupled Proliferation and Apoptosis Maintain the Rapid Turnover of Microglia in the Adult Brain. Cell Rep, 2017. 18(2): p. 391-405. 129. Tay, T.L., et al., A new fate mapping system reveals context-dependent random or clonal expansion of microglia. Nat Neurosci, 2017. 20(6): p. 793-803. 130. Schaafsma, W., et al., Maternal inflammation induces immune activation of fetal microglia and leads to disrupted microglia immune responses, behavior, and learning performance in adulthood. Neurobiol Dis, 2017. 106: p. 291-300. 131. Smolders, S., et al., Maternal immune activation evoked by polyinosinic:polycytidylic acid does not evoke microglial cell activation in the embryo. Front Cell Neurosci, 2015. 9: p. 301. 132. Pratt, L., et al., Maternal inflammation promotes fetal microglial activation and increased cholinergic expression in the fetal basal forebrain: role of interleukin-6. Pediatr Res, 2013. 74(4): p. 393-401. 133. O'Loughlin, E., et al., Acute in utero exposure to lipopolysaccharide induces inflammation in the pre- and postnatal brain and alters the glial cytoarchitecture in the developing amygdala. J Neuroinflammation, 2017. 14(1): p. 212. 134. Patel, S., et al., Social impairments in autism spectrum disorder are related to maternal immune history profile. Mol Psychiatry, 2017. 135. Baron-Cohen, S., et al., The amygdala theory of autism. Neurosci Biobehav Rev, 2000. 24(3): p. 355-64. 136. Felix-Ortiz, A.C. and K.M. Tye, Amygdala inputs to the ventral hippocampus bidirectionally modulate social behavior. J Neurosci, 2014. 34(2): p. 586-95. 137. Felix-Ortiz, A.C., et al., Bidirectional modulation of anxiety-related and social behaviors by amygdala projections to the medial prefrontal cortex. Neuroscience, 2016. 321: p. 197-209. 138. Das Sarma, J., et al., Functional interleukin-17 receptor A is expressed in central nervous system glia and upregulated in experimental autoimmune encephalomyelitis. J Neuroinflammation, 2009. 6(1): p. 14. 139. Kawanokuchi, J., et al., Production and functions of IL-17 in microglia. J Neuroimmunol, 2008. 194(1-2): p. 54-61. 140. Zimmermann, J., et al., IL-17A Promotes Granulocyte Infiltration, Myelin Loss, Microglia Activation, and Behavioral Deficits During Cuprizone-Induced Demyelination. Mol Neurobiol, 2018. 55(2): p. 946-957. 141. Louveau, A., et al., Understanding the functions and relationships of the glymphatic system and meningeal lymphatics. J Clin Invest, 2017. 127(9): p. 3210-3219. 142. Filiano, A.J., et al., Unexpected role of interferon-gamma in regulating neuronal connectivity and social behaviour. Nature, 2016. 535(7612): p. 425-9. 143. Martinez, F.O. and S. Gordon, The M1 and M2 paradigm of macrophage activation: time for reassessment. F1000Prime Rep, 2014. 6: p. 13. 144. Brown, A.S. and E.J. Derkits, Prenatal infection and schizophrenia: a review of epidemiologic and translational studies. Am J Psychiatry, 2010. 167(3): p. 261-80. 145. Nair, A. and R.H. Bonneau, Stress-induced elevation of glucocorticoids increases microglia proliferation through NMDA receptor activation. J Neuroimmunol, 2006. 171(1-2): p. 72-85. 146. Frank, M.G., et al., Chronic exposure to exogenous glucocorticoids primes microglia to pro-inflammatory stimuli and induces NLRP3 mRNA in the hippocampus. Psychoneuroendocrinology, 2014. 40(0): p. 191- 200. 147. Barrientos, R.M., et al., Greater glucocorticoid receptor activation in hippocampus of aged rats sensitizes microglia. Neurobiol Aging, 2015. 36(3): p. 1483-95. 148. French, S.S., E.M. Chester, and G.E. Demas, Maternal immune activation affects litter success, size and neuroendocrine responses related to behavior in adult offspring. Physiol Behav, 2013. 119(0): p. 175-84.

123

149. Reul, J.M., et al., Prenatal immune challenge alters the hypothalamic-pituitary-adrenocortical axis in adult rats. J Clin Invest, 1994. 93(6): p. 2600-7. 150. Depino, A.M., Early prenatal exposure to LPS results in anxiety- and depression-related behaviors in adulthood. Neuroscience, 2015. 299: p. 56-65. 151. Babri, S., M.H. Doosti, and A.A. Salari, Strain-dependent effects of prenatal maternal immune activation on anxiety- and depression-like behaviors in offspring. Brain Behav Immun, 2014. 37(0): p. 164-76. 152. Hsiao, E.Y., et al., Modeling an autism risk factor in mice leads to permanent immune dysregulation. Proc Natl Acad Sci U S A, 2012. 109(31): p. 12776-81. 153. Moisiadis, V.G. and S.G. Matthews, Glucocorticoids and fetal programming part 1: Outcomes. Nat Rev Endocrinol, 2014. 10(7): p. 391-402. 154. Howerton, C.L. and T.L. Bale, Prenatal programing: at the intersection of maternal stress and immune activation. Horm Behav, 2012. 62(3): p. 237-42. 155. Couret, D., et al., Maternal stress during late gestation has moderate but long-lasting effects on the immune system of the piglets. Vet Immunol Immunopathol, 2009. 131(1-2): p. 17-24. 156. Klemcke, H.G., Placental Metabolism of Cortisol at Mid- and Late Gestation in Swine1. Biology of Reproduction, 1995. 53(6): p. 1293-1301. 157. Bellavance, M.A. and S. Rivest, The HPA - Immune Axis and the Immunomodulatory Actions of Glucocorticoids in the Brain. Front Immunol, 2014. 5: p. 136. 158. Kapoor, A., S. Petropoulos, and S.G. Matthews, Fetal programming of hypothalamic-pituitary-adrenal (HPA) axis function and behavior by synthetic glucocorticoids. Brain Res Rev, 2008. 57(2): p. 586-95. 159. Liu, L., A. Li, and S.G. Matthews, Maternal glucocorticoid treatment programs HPA regulation in adult offspring: sex-specific effects. Am J Physiol Endocrinol Metab, 2001. 280(5): p. E729-39. 160. Kim, T.S., D.A. Benfield, and R.R. Rowland, Porcine reproductive and respiratory syndrome virus-induced cell death exhibits features consistent with a nontypical form of apoptosis. Virus Res, 2002. 85(2): p. 133- 40. 161. Liu, C.H., et al., Expression of Toll-like receptor mRNA and cytokines in pigs infected with porcine reproductive and respiratory syndrome virus. Vet Microbiol, 2009. 136(3-4): p. 266-76. 162. Conrad, M.S., et al., Early postnatal respiratory viral infection alters hippocampal neurogenesis, cell fate, and neuron morphology in the neonatal piglet. Brain Behav Immun, 2015. 44: p. 82-90. 163. Ji, P., et al., Peripheral viral infection induced microglial sensome genes and enhanced microglial cell activity in the hippocampus of neonatal piglets. Brain Behav Immun, 2016. 54: p. 243-251. 164. Dobbing, J. and J. Sands, Quantitative growth and development of human brain. Arch Dis Child, 1973. 48(10): p. 757-67. 165. Andersen, S.L., Trajectories of brain development: point of vulnerability or window of opportunity? Neurosci Biobehav Rev, 2003. 27(1-2): p. 3-18. 166. Cook, N.J., et al., The collection of multiple saliva samples from pigs and the effect on adrenocortical activity. Canadian Journal of Animal Science, 2013. 93(3): p. 329-333. 167. Elmore, M.R., et al., Respiratory viral infection in neonatal piglets causes marked microglia activation in the hippocampus and deficits in spatial learning. J Neurosci, 2014. 34(6): p. 2120-9. 168. Dilger, R.N. and R.W. Johnson, Behavioral assessment of cognitive function using a translational neonatal piglet model. Brain Behav Immun, 2010. 24(7): p. 1156-65. 169. Webel, D.M., et al., Time course of increased plasma cytokines, cortisol, and urea nitrogen in pigs following intraperitoneal injection of lipopolysaccharide. J Anim Sci, 1997. 75(6): p. 1514-20. 170. Elmore, M.R., R.N. Dilger, and R.W. Johnson, Place and direction learning in a spatial T-maze task by neonatal piglets. Anim Cogn, 2012. 15(4): p. 667-76. 171. Moy, S.S., et al., Sociability and preference for social novelty in five inbred strains: an approach to assess autistic-like behavior in mice. Genes, Brain and Behavior, 2004. 3: p. 287-302. 172. Nadler, J.J., et al., Automated apparatus for quantitation of social approach behaviors in mice. Genes, Brain and Behavior, 2004. 3(5): p. 303-314. 173. Kanitz, E., G. Manteuffel, and W. Otten, Effects of weaning and restraint stress on glucocorticoid receptor binding capacity in limbic areas of domestic pigs. Brain Res, 1998. 804(2): p. 311-5.

124

174. DeRijk, R.H., F. Eskandari, and E.M. Sternberg, Corticosteroid resistance in a subpopulation of multiple sclerosis patients as measured by ex vivo dexamethasone inhibition of LPS induced IL-6 production. J Neuroimmunol, 2004. 151(1-2): p. 180-8. 175. Giovanoli, S. and U. Meyer, Response to comment on "Stress in puberty unmasks latent neuropathological consequences of prenatal immune activation in mice". Science, 2013. 340(6134): p. 811. 176. Ford, A.L., et al., Normal adult ramified microglia separated from other central nervous system macrophages by flow cytometric sorting. Phenotypic differences defined and direct ex vivo antigen presentation to myelin basic protein-reactive CD4+ T cells compared. J Immunol, 1995. 154(9): p. 4309-21. 177. Henry, C.J., et al., Peripheral lipopolysaccharide (LPS) challenge promotes microglial hyperactivity in aged mice that is associated with exaggerated induction of both pro-inflammatory IL-1beta and anti- inflammatory IL-10 cytokines. Brain Behav Immun, 2009. 23(3): p. 309-17. 178. Giovanoli, S., et al., Stress in puberty unmasks latent neuropathological consequences of prenatal immune activation in mice. Science, 2013. 339(6123): p. 1095-9. 179. Straley, M.E., et al., Distinct alterations in motor & reward seeking behavior are dependent on the gestational age of exposure to LPS-induced maternal immune activation. Brain Behav Immun, 2017. 63: p. 21-34. 180. Meyer, U., M.J. Schwarz, and N. Muller, Inflammatory processes in schizophrenia: a promising neuroimmunological target for the treatment of negative/cognitive symptoms and beyond. Pharmacol Ther, 2011. 132(1): p. 96-110. 181. Meyer, U., Anti-inflammatory signaling in schizophrenia. Brain Behav Immun, 2011. 25(8): p. 1507-18. 182. Walker, E.F., et al., Longitudinal changes in cortisol secretion and conversion to psychosis in at-risk youth. J Abnorm Psychol, 2010. 119(2): p. 401-8. 183. Corcoran, C.M., et al., HPA axis function and symptoms in adolescents at clinical high risk for schizophrenia. Schizophr Res, 2012. 135(1-3): p. 170-4. 184. Thompson, K.N., et al., Stress and HPA-axis functioning in young people at ultra high risk for psychosis. J Psychiatr Res, 2007. 41(7): p. 561-9. 185. Belvederi Murri, M., et al., Hypothalamic-pituitary-adrenal axis and clinical symptoms in first-episode psychosis. Psychoneuroendocrinology, 2012. 37(5): p. 629-44. 186. Garay, P.A., et al., Maternal immune activation causes age- and region-specific changes in brain cytokines in offspring throughout development. Brain Behav Immun, 2013. 31(0): p. 54-68. 187. Le Pen, G., et al., Peri-pubertal maturation after developmental disturbance: a model for psychosis onset in the rat. Neuroscience, 2006. 143(2): p. 395-405. 188. Lamason, R., et al., Sexual dimorphism in immune response genes as a function of puberty. BMC Immunol, 2006. 7(1): p. 2. 189. McCormick, C.M. and I.Z. Mathews, Adolescent development, hypothalamic-pituitary-adrenal function, and programming of adult learning and memory. Prog Neuropsychopharmacol Biol Psychiatry, 2010. 34(5): p. 756-65. 190. Lupien, S.J., et al., Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nat Rev Neurosci, 2009. 10(6): p. 434-45. 191. Brenhouse, H.C. and J.M. Schwarz, Immunoadolescence: Neuroimmune development and adolescent behavior. Neurosci Biobehav Rev, 2016. 70: p. 288-299. 192. Meyer, U., J. Feldon, and S.H. Fatemi, In-vivo rodent models for the experimental investigation of prenatal immune activation effects in neurodevelopmental brain disorders. Neurosci Biobehav Rev, 2009. 33(7): p. 1061-79. 193. Aavani, T., et al., Maternal immune activation produces cerebellar hyperplasia and alterations in motor and social behaviors in male and female mice. Cerebellum, 2015. 14(5): p. 491-505. 194. Xuan, I.C. and D.R. Hampson, Gender-dependent effects of maternal immune activation on the behavior of mouse offspring. PLoS One, 2014. 9(8): p. e104433. 195. Bitanihirwe, B.K., et al., Late prenatal immune activation in mice leads to behavioral and neurochemical abnormalities relevant to the negative symptoms of schizophrenia. Neuropsychopharmacology, 2010. 35(12): p. 2462-78.

125

196. Newberry, R.C., D.G. Wood-Gush, and J.W. Hall, Playful behaviour of piglets. Behav Processes, 1988. 17(3): p. 205-16. 197. Petersen, H.V., K. Vestergaard, and P. Jensen, Integration of piglets into social groups of free-ranging domestic pigs. Applied Animal Behaviour Science, 1989. 23(3): p. 223-236. 198. Wood-Gush, D.G.M. and K. Vestergaard, The seeking of novelty and its relation to play. Animal Behaviour, 1991. 42(4): p. 599-606. 199. Studnitz, M., M.B. Jensen, and L.J. Pedersen, Why do pigs root and in what will they root? Applied Animal Behaviour Science, 2007. 107(3-4): p. 183-197. 200. Graves, H.B., Behavior and Ecology of Wild and Feral Swine (Sus Scrofa). Journal of Animal Science, 1984. 58(2): p. 482-492. 201. Camerlink, I. and S.P. Turner, The pig's nose and its role in dominance relationships and harmful behaviour. Applied Animal Behaviour Science, 2013. 145(3-4): p. 84-91. 202. Moy, S.S., et al., Mouse behavioral tasks relevant to autism: phenotypes of 10 inbred strains. Behav Brain Res, 2007. 176(1): p. 4-20. 203. Belinson, H., et al., Prenatal beta-catenin/Brn2/Tbr2 transcriptional cascade regulates adult social and stereotypic behaviors. Mol Psychiatry, 2016. 21(10): p. 1417-33. 204. Nardone, S. and E. Elliott, The Interaction between the Immune System and Epigenetics in the Etiology of Autism Spectrum Disorders. Front Neurosci, 2016. 10: p. 329. 205. Harvey, L. and P. Boksa, Additive effects of maternal iron deficiency and prenatal immune activation on adult behaviors in rat offspring. Brain Behav Immun, 2014. 40(0): p. 27-37. 206. Machado, C.J., et al., Maternal immune activation in nonhuman primates alters social attention in juvenile offspring. Biol Psychiatry, 2015. 77(9): p. 823-32. 207. Bernstein, H.G., Y. Piontkewitz, and G. Keilhoff, Commentary: Maternal immune activation evoked by polyinosinic: polycytidylic acid does not evoke microglial cell activation in the embryo. Front Cell Neurosci, 2016. 10: p. 41. 208. Ibi, D. and K. Yamada, Therapeutic Targets for Neurodevelopmental Disorders Emerging from Animal Models with Perinatal Immune Activation. Int J Mol Sci, 2015. 16(12): p. 28218-29. 209. Khandaker, G.M., et al., Prenatal maternal infection, neurodevelopment and adult schizophrenia: a systematic review of population-based studies. Psychol Med, 2013. 43(2): p. 239-57. 210. Estes, M.L. and A.K. McAllister, Maternal immune activation: Implications for neuropsychiatric disorders. Science, 2016. 353(6301): p. 772-7. 211. Dantzer, R. and K.W. Kelley, Twenty years of research on cytokine-induced sickness behavior. Brain Behav Immun, 2007. 21(2): p. 153-60. 212. Mousa, A. and M. Bakhiet, Role of cytokine signaling during nervous system development. Int J Mol Sci, 2013. 14(7): p. 13931-57. 213. Burd, I., B. Balakrishnan, and S. Kannan, Models of fetal brain injury, intrauterine inflammation, and preterm birth. Am J Reprod Immunol, 2012. 67(4): p. 287-94. 214. Pineda, E., et al., Maternal immune activation promotes hippocampal kindling epileptogenesis in mice. Ann Neurol, 2013. 74(1): p. 11-9. 215. Meyer, U., et al., Towards an immuno-precipitated neurodevelopmental animal model of schizophrenia. Neurosci Biobehav Rev, 2005. 29(6): p. 913-47. 216. Estes, M.L. and A.K. McAllister, IMMUNOLOGY. Maternal TH17 cells take a toll on baby's brain. Science, 2016. 351(6276): p. 919-20. 217. Zhang, Z. and H. van Praag, Maternal immune activation differentially impacts mature and adult-born hippocampal neurons in male mice. Brain Behav Immun, 2015. 45: p. 60-70. 218. Oh-Nishi, A., et al., Maternal immune activation by polyriboinosinic-polyribocytidilic acid injection produces synaptic dysfunction but not neuronal loss in the hippocampus of juvenile rat offspring. Brain research, 2010. 1363: p. 170-179. 219. Holtkamp, D.J., et al., Assessment of the economic impact of porcine reproductive and respiratory syndrome virus on United States pork producers. Journal of Swine Health and Production, 2013. 21(2): p. 72-84.

126

220. Monto, A.S., et al., Clinical signs and symptoms predicting influenza infection. Arch Intern Med, 2000. 160(21): p. 3243-7. 221. Antonson, A.M., et al., Maternal viral infection during pregnancy elicits anti-social behavior in neonatal piglet offspring independent of postnatal microglial cell activation. Brain Behav Immun, 2017. 59: p. 300- 312. 222. Livak, K.J. and T.D. Schmittgen, Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods, 2001. 25(4): p. 402-8. 223. Pilati, N., et al., A rapid method combining Golgi and Nissl staining to study neuronal morphology and cytoarchitecture. J Histochem Cytochem, 2008. 56(6): p. 539-50. 224. Mancuso, R., et al., Sigma-1R agonist improves motor function and motoneuron survival in ALS mice. Neurotherapeutics, 2012. 9(4): p. 814-26. 225. Moraga, A., et al., Toll-like receptor 4 modulates cell migration and cortical neurogenesis after focal cerebral ischemia. FASEB J, 2014. 28(11): p. 4710-8. 226. Penas, C., et al., Cytoskeletal and activity-related changes in spinal motoneurons after root avulsion. J Neurotrauma, 2009. 26(5): p. 763-79. 227. Moreno-Igoa, M., et al., Fragment C of tetanus toxin, more than a carrier. Novel perspectives in non-viral ALS gene therapy. J Mol Med (Berl), 2010. 88(3): p. 297-308. 228. Radlowski, E.C., et al., A neonatal piglet model for investigating brain and cognitive development in small for gestational age human infants. PLoS One, 2014. 9(3): p. e91951. 229. Zschaler, J., D. Schlorke, and J. Arnhold, Differences in innate immune response between man and mouse. Critical Reviews in Immunology, 2014. 34(5): p. 433-454. 230. Seok, J., et al., Genomic responses in mouse models poorly mimic human inflammatory diseases. Proc Natl Acad Sci U S A, 2013. 110(9): p. 3507-12. 231. Ladinig, A., et al., Maternal and fetal predictors of fetal viral load and death in third trimester, type 2 porcine reproductive and respiratory syndrome virus infected pregnant gilts. Vet Res, 2015. 46(1): p. 107. 232. Cheon, D.S. and C. Chae, Distribution of porcine reproductive and respiratory syndrome virus in stillborn and liveborn piglets from experimentally infected sows. J Comp Pathol, 2001. 124(4): p. 231-7. 233. Kranker, S., et al., Experimental inoculation of swine at various stages of gestation with a Danish isolate of porcine reproductive and respiratory syndrome virus (PRRSV). Veterinary Microbiology, 1998. 61(1-2): p. 21-31. 234. Brodsky, D. and H. Christou, Current concepts in intrauterine growth restriction. J Intensive Care Med, 2004. 19(6): p. 307-19. 235. Conner, J.M., et al., NGF is essential for hippocampal plasticity and learning. J Neurosci, 2009. 29(35): p. 10883-9. 236. Farrelly, L., et al., Maternal immune activation induces changes in myelin and metabolic proteins, some of which can be prevented with risperidone in adolescence. Dev Neurosci, 2015. 37(1): p. 43-55. 237. Makinodan, M., et al., Maternal immune activation in mice delays myelination and axonal development in the hippocampus of the offspring. J Neurosci Res, 2008. 86(10): p. 2190-200. 238. Boksa, P., Effects of prenatal infection on brain development and behavior: a review of findings from animal models. Brain Behav Immun, 2010. 24(6): p. 881-97. 239. Scheffel, J., et al., Toll-like receptor activation reveals developmental reorganization and unmasks responder subsets of microglia. Glia, 2012. 60(12): p. 1930-43. 240. Perry, V.H. and C. Holmes, Microglial priming in neurodegenerative disease. Nat Rev Neurol, 2014. 10(4): p. 217-24. 241. Moss, D.W. and T.E. Bates, Activation of murine microglial cell lines by lipopolysaccharide and interferon- gamma causes NO-mediated decreases in mitochondrial and cellular function. Eur J Neurosci, 2001. 13(3): p. 529-38. 242. Neumann, H., et al., Tumor necrosis factor inhibits neurite outgrowth and branching of hippocampal neurons by a rho-dependent mechanism. J Neurosci, 2002. 22(3): p. 854-62. 243. Lombardo, M.V., et al., Maternal immune activation dysregulation of the fetal brain transcriptome and relevance to the pathophysiology of autism spectrum disorder. Mol Psychiatry, 2017.

127

244. Patel, S., et al., Crosstalk between endoplasmic reticulum stress and oxidative stress in schizophrenia: The dawn of new therapeutic approaches. Neurosci Biobehav Rev, 2017. 83: p. 589-603. 245. Leonidas, M., The Distribution of the Stress Protein HSP70 in the Cerebellum of Patients with Schizophrenia. Neuroscience and Medicine, 2012. 03(04): p. 368-373. 246. Iacono, W.G. and M. Beiser, Are males more likely than females to develop schizophrenia? Am J Psychiatry, 1992. 149(8): p. 1070-4. 247. Hinojosa, A.E., et al., CCL2/MCP-1 modulation of microglial activation and proliferation. J Neuroinflammation, 2011. 8(1): p. 77. 248. Ueno, M., et al., Layer V cortical neurons require microglial support for survival during postnatal development. Nat Neurosci, 2013. 16(5): p. 543-51. 249. Antonson, A.M., et al., Altered Hippocampal Gene Expression and Morphology in Fetal Piglets following Maternal Respiratory Viral Infection. Dev Neurosci, 2018. 250. Ladinig, A., et al., Variation in fetal outcome, viral load and ORF5 sequence mutations in a large scale study of phenotypic responses to late gestation exposure to type 2 porcine reproductive and respiratory syndrome virus. PLoS One, 2014. 9(4): p. e96104. 251. Felix, B., et al., Stereotaxic atlas of the pig brain. Brain Res Bull, 1999. 49(1-2): p. 1-137. 252. Almolda, B., et al., CD4 microglial expression correlates with spontaneous clinical improvement in the acute Lewis rat EAE model. J Neuroimmunol, 2009. 209(1-2): p. 65-80. 253. Tay, T.L., et al., Microglia across the lifespan: from origin to function in brain development, plasticity and cognition. J Physiol, 2017. 595(6): p. 1929-1945. 254. Mosser, C.A., et al., Microglia in CNS development: Shaping the brain for the future. Prog Neurobiol, 2017. 149-150: p. 1-20. 255. Stepanova, H., et al., CD4+ and gammadeltaTCR+ T lymphocytes are sources of interleukin-17 in swine. Cytokine, 2012. 58(2): p. 152-7. 256. Zhu, Y.H., et al., Dose-dependent effects of Lactobacillus rhamnosus on serum interleukin-17 production and intestinal T-cell responses in pigs challenged with Escherichia coli. Appl Environ Microbiol, 2014. 80(5): p. 1787-98. 257. Groom, J.R. and A.D. Luster, CXCR3 in T cell function. Exp Cell Res, 2011. 317(5): p. 620-31. 258. Kreutzberg, G.W., Microglia: a sensor for pathological events in the CNS. Trends Neurosci, 1996. 19(8): p. 312-8. 259. Hendrickx, D.A.E., et al., Staining of HLA-DR, Iba1 and CD68 in human microglia reveals partially overlapping expression depending on cellular morphology and pathology. J Neuroimmunol, 2017. 309: p. 12-22. 260. Walker, D.G. and L.F. Lue, Immune phenotypes of microglia in human neurodegenerative disease: challenges to detecting microglial polarization in human brains. Alzheimers Res Ther, 2015. 7(1): p. 56. 261. Ahlers, K.E., et al., Transient activation of microglia following acute alcohol exposure in developing mouse neocortex is primarily driven by BAX-dependent neurodegeneration. Glia, 2015. 63(10): p. 1694-1713. 262. Shigemoto-Mogami, Y., et al., Microglia enhance neurogenesis and oligodendrogenesis in the early postnatal subventricular zone. J Neurosci, 2014. 34(6): p. 2231-43. 263. Parkhurst, C.N., et al., Microglia promote learning-dependent synapse formation through brain-derived neurotrophic factor. Cell, 2013. 155(7): p. 1596-1609. 264. Butovsky, O., et al., Identification of a unique TGF-beta-dependent molecular and functional signature in microglia. Nat Neurosci, 2014. 17(1): p. 131-43. 265. Hu, X., et al., Microglial and macrophage polarization-new prospects for brain repair. Nat Rev Neurol, 2015. 11(1): p. 56-64. 266. Buttgereit, A., et al., Sall1 is a transcriptional regulator defining microglia identity and function. Nature Immunology, 2016. 17: p. 1397. 267. Abutbul, S., et al., TGF-β signaling through SMAD2/3 induces the quiescent microglial phenotype within the CNS environment. Glia, 2012. 60(7): p. 1160-1171. 268. Spittau, B., et al., Endogenous transforming growth factor-beta promotes quiescence of primary microglia in vitro. Glia, 2013. 61(2): p. 287-300.

128

269. Haynes, S.E., et al., The P2Y12 receptor regulates microglial activation by extracellular nucleotides. Nat Neurosci, 2006. 9(12): p. 1512-9. 270. Sipe, G.O., et al., Microglial P2Y12 is necessary for synaptic plasticity in mouse visual cortex. Nature Communications, 2016. 7: p. 10905. 271. Koizumi, S., et al., UDP acting at P2Y6 receptors is a mediator of microglial phagocytosis. Nature, 2007. 446: p. 1091. 272. Zhou, M., et al., CCR5 is a suppressor for cortical plasticity and hippocampal learning and memory. Elife, 2016. 5. 273. Otero, K., et al., Macrophage colony-stimulating factor induces the proliferation and survival of macrophages via a pathway involving DAP12 and beta-catenin. Nat Immunol, 2009. 10(7): p. 734-43. 274. Arcuri, C., et al., The Pathophysiological Role of Microglia in Dynamic Surveillance, Phagocytosis and Structural Remodeling of the Developing CNS. Front Mol Neurosci, 2017. 10: p. 191. 275. Kim, S.M., et al., TREM2 promotes Abeta phagocytosis by upregulating C/EBPalpha-dependent CD36 expression in microglia. Sci Rep, 2017. 7(1): p. 11118. 276. Zhang, B., et al., Integrated systems approach identifies genetic nodes and networks in late-onset Alzheimer's disease. Cell, 2013. 153(3): p. 707-20. 277. Garbett, K.A., et al., Effects of maternal immune activation on gene expression patterns in the fetal brain. Transl Psychiatry, 2012. 2: p. e98. 278. Wu, N., et al., alpha-Crystallin downregulates the expression of TNF-alpha and iNOS by activated rat retinal microglia in vitro and in vivo. Ophthalmic Res, 2009. 42(1): p. 21-8. 279. Dulle, J.E. and P.E. Fort, Crystallins and neuroinflammation: The glial side of the story. Biochim Biophys Acta, 2016. 1860(1 Pt B): p. 278-86. 280. Pickard, B., Progress in defining the biological causes of schizophrenia. Expert Rev Mol Med, 2011. 13: p. e25. 281. Spadaro, P.A., et al., Long Noncoding RNA-Directed Epigenetic Regulation of Gene Expression Is Associated With Anxiety-like Behavior in Mice. Biol Psychiatry, 2015. 78(12): p. 848-59. 282. Mattei, D., et al., Minocycline rescues decrease in neurogenesis, increase in microglia cytokines and deficits in sensorimotor gating in an animal model of schizophrenia. Brain Behav Immun, 2014. 38(0): p. 175-84. 283. Gray, J., et al., Intestinal commensal bacteria mediate lung mucosal immunity and promote resistance of newborn mice to infection. Sci Transl Med, 2017. 9(376): p. eaaf9412.

129