Maternal Immune Activation during Pregnancy Alters the Behavior Profile of Female Offspring of Sprague Dawley Rats

Abstract Sex differences are documented in psychiatric and neurological disorders, yet most preclinical animal research has been conducted in males only. There is a need to better understand of the nature of sex differences in brain disease in order to meet the needs of psychiatric patients. We present the behavior profile of adult female offspring produced using a maternal immune activation (MIA) model where pregnant rats receive an immune stimulant and the offspring typically show various abnormalities consistent with psychiatric illnesses such as schizophrenia and autism. The results in female offspring were compared to a previously published cohort of their male siblings (Lins et al., 2018). We examined prepulse inhibition (PPI), sociability, MK-801-induced locomotor activity, crossmodal object recognition (CMOR), and oddity discrimination; behaviors relevant to the positive, negative, and cognitive symptoms of schizophrenia. No between-treatment differences in PPI or locomotor activity were noted. Tactile memory was observed in the control and treated female offspring, visual recognition memory was deficient in the polyinosinic:polycytidylic acid (polyI:C) offspring only, and both groups lacked crossmodal recognition. PolyI:C offspring were impaired in oddity preference and had reduced preference for a stranger conspecific in a sociability assay. Systemic maternal CXCL1, IL-6, and TNF-a levels 3 h after polyI:C treatment were determined, but no relationship was found between these cytokines and the behavior seen in the adult female offspring. Overall, female offspring of polyI:C-treated dams display an array of behavior abnormalities relevant to psychiatric illnesses such as schizophrenia similar to those previously reported in male rats.


Introduction
Adverse events in utero and early life are linked to the development of psychiatric illness. Inflammation during pregnancy is associated with increased risk of psychiatric illnesses including autism, schizophrenia, and major depression in the offspring (Patterson, 2011;Jiang et al., 2016;Brown and Meyer, 2018;Gustafsson et al., 2018). The relationship between inflammation and psychopathology is often studied with models of maternal immune activation (MIA) where an immune stimulant such as polyinosinic:polycytidylic acid (polyI:C) is administered to pregnant rodents (Piontkewitz et al., 2012;Brown and Meyer, 2018). Offspring of treated dams display behavioral and neuropathological profiles consistent with psychiatric illness in humans (Brown and Meyer, 2018). The majority of MIA studies focus on the male offspring or lack consideration of sex as a biological variable despite policies by the National Institute of Health and other grant funding agencies that require the examination of sex as a factor in biomedical research (Clayton and Collins, 2014;Coiro and Pollak, 2019). This is particularly concerning for studies of MIA given the sex differences noted in the human psychiatric disorders associated with MIA as a risk factor (Klein and Corwin, 2002;Arad et al., 2017;Brown and Meyer, 2018;Coiro and Pollak, 2019).
Previous studies of the effects of MIA during pregnancy in rats has resulted in extensive knowledge of behavior effects in male offspring. Maternal treatment with polyI:C during gestation results in male offspring with reduced working memory span capacity, dysregulated fear responses, and impaired associative (object-in-place) and crossmodal memory while simple object recognition and object-location memory are largely unaffected, although impaired novel context recognition has also been reported Wolff et al., 2011;Sangha et al., 2014;Ballendine et al., 2015;Murray et al., 2017). Other studies on adult male offspring from polyI:C-treated pregnancies have shown reduced levels of GAD67 in the dorsal hippocampus which appears to coincide with a loss of hippocampal-frontal coherence and correlate with prepulse inhibition (PPI) deficits (Dickerson et al., 2010(Dickerson et al., , 2013(Dickerson et al., , 2014Wolff and Bilkey, 2010). PPI has been studied extensively in MIA rat models, yet the results are mixed with several studies showing deficits, no effect, or indicating the timing and type of inflammatory agent determines PPI outcomes (Fortier et al., 2004(Fortier et al., , 2007Bilkey, 2008, 2010;Ballendine et al., 2015;Hadar et al., 2015). Other studies with offspring of both sexes or sex unspecified report mixed results on PPI as well Klein et al., 2013;Van den Eynde et al., 2014;Vorhees et al., 2015). Further MIA studies including male and female rat offspring report altered behaviors such as spontaneous hypolocomotion , latent inhibition Weiner, 2003, 2005;, and reduced startle . Other studies show no change in spontaneous locomotion but reduced sensitivity to the hyperlocomotive effects of amphetamine treatment (Bronson et al., 2011) or hyperlocomotion following amphetamine Vorhees et al., 2012) or MK-801 (Zuckerman and Weiner, 2005; but see also Howland et al., 2012). In some tasks, male rat offspring of MIA dams show greater impairments in tasks such as operant conditioning-based set shifting and earlier onset of latent inhibition deficits, reflecting the earlier onset of abnormal developmental trajectories (Piontkewitz et al., 2011a;Zhang et al., 2012;Patrich et al., 2016). While male MIA rats have impaired object-in-place memory, neither control nor MIA females perform this task, possibly reflecting sex-specific differences in baseline task performance Ballendine et al., 2015). In a related rat model of MIA, inflammation induced in lactating dams resulted in the development of distinct sexdependent phenotypes in the suckling offspring, where the females offspring displayed a depressive phenotype and male offspring displayed a psychiatric phenotype (Arad et al., 2017). Taken together, the complicated and often conflicting results from these studies demonstrate the need for sex by treatment analyses in future MIA research (Coiro and Pollak, 2019;Kentner et al., 2019).
The present study aims to contribute to the necessary evaluation of sex effects in the MIA model by highlighting female offspring behavior outcomes in tasks related to the symptoms of schizophrenia and analyzing these results in conjunction with previously published results in their male littermates. The male MIA offspring from the same cohort display hyperlocomotion following MK-801 administration, reduced sociability, impaired visual recognition memory, impaired oddity preference, altered set shifting, and facilitated reversal learning. Using a prospective design, we showed that these behavioral changes did not correlate with acute elevations in a selection of maternal serum cytokines collected 3 h after polyI:C treatment (Lins et al., 2018). We hypothesized that behavior abnormalities would be less severe or absent in early adulthood in accordance with previous literature (Piontkewitz et al., 2011a;Zhang et al., 2012;Patrich et al., 2016). We also correlated behavior of the female offspring with the acute cytokine concentrations from prospectively collected maternal blood and other measurements related to polyI:C treatment to assess relationships between acute maternal cytokine levels, other treatment effects, and offspring behavior, including an oddity discrimination task not previously examined in female rats (Lins et al., 2018).

Animals
Timed-pregnant Sprague Dawley rats (n ϭ 43, Charles River) arrived at the animal holding facility on gestational day 7 (GD7). Primiparous dams were mated between 8 and 10 weeks of age and the presence of a vaginal plug considered GD1. Sires were a minimum of 10 weeks of age at time of mating and their specific breeding history (number of matings, successful pregnancies, mating design, or time between matings) was not guaranteed by the provider. On arrival, pregnant dams were housed individually in standard ventilated (395 ϫ 346 ϫ 227 mm) polypropylene cages. Food (Purina Rat Chow) and water were available ad libitum. The colony room is temperature, but not humidity, controlled (21°C) and maintained on an automatic 12/12 h light/dark cycle with lights on at 7 A.M. Dams were undisturbed until they received treatment on GD15. behavior testing was conducted on adult female offspring (total n ϭ 71). All procedures were conducted during the light phase and were conducted in accordance with the Canadian Council on Animal Care guidelines for humane animal use and were approved by the University of Saskatchewan Animal Research Ethics board.

Maternal treatments and blood samples
Maternal treatment followed previously established protocols in Long-Evans and Sprague Dawley rats Zhang et al., 2012;Sangha et al., 2014;Lins et al., 2016Lins et al., , 2018Paylor et al., 2016;Fig. 1A). Additional information is included to improve the scientific rigor of the MIA model as discussed by Kentner et al. (2019). These details also apply to our recent companion publication which tested male rat offspring from the same litters as described here (Lins et al., 2018).
Dam baseline weight (mean ϭ 346.5 g) and rectal temperature were recorded on GD15 immediately before anesthesia with isoflurane (5% induction, 2.5% maintenance, for ϳ3 min) to receive a single intravenous tail vein injection of 0.9% saline or polyI:C (4 mg/kg, high molecular weight, InVivoGen; thawed from storage at -20°C).
Dams were anesthetized a second time (as above, ϳ10 min) 3 h following initial treatment and a blood sample (Ͻ1.5 ml total and Ͻ6% total blood volume) was collected using a sterile catheter (BD Insyte Autogaurd, 24 GA 0.75 IN 0.7 ϫ 19 mm, REF 381412) from the opposite tail vein used to inject polyI:C or saline. Warm saline was administered once following initial treatment (3 ml), and a second time after blood collection (equal in volume to the collected blood sample). Blood samples coagulated at room temperature for 1 h then centrifuged at 10,000 ϫ g for 5 min and serum was stored at -80°C until analysis. ELISAs for CXCL1 (GRO␣/KC; R&D Systems), CXCL2 (GRO␤/MIP-2; R&D Systems), IL-6 (PeproTech), and TNF-␣ (PeproTech) were performed according to the manufacturer's instructions and these results are published elsewhere (Lins et al., 2018).
Dams had weight and rectal temperature measured 8, 24, and 48 h after treatment and then were undisturbed for the remainder of their pregnancy. Treatment was administered to n ϭ 43 dams, but eight developed hypothermia and were euthanized within 48 h of treatment. Four additional dams experienced body temperature below 36°C but lacked additional indicators of severe sickness or suffering and these were given access to a warming pad on their home cage until their temperature returned to normal (within 24 h). Two dams did not produce viable litters, and two litters had no female offspring. Ultimately, offspring from a total of n ϭ 31 litters were included (n ϭ 16 polyI:C-treated dams and n ϭ 15 salinetreated dams; Table 1). On postnatal day (PND)1, litters were weighed, sexed, and culled to a maximum of 10 (four females where possible). Standard husbandry included cage changes twice per week with one additional cage change during PND14 -PND21. Before weaning, all cage changes, feeding, and monitoring of pups was performed by a single investigator to minimize disturbances. On PND23, pups were weaned and housed in same-sex sibling groups of two or three in standard housing as previously described with a PVC tube for enrichment.  (Lins et al., 2018). B, Flowchart depicting the order of the behavior test battery.

Behavioral testing
Behavior tests were conducted according to published protocols. One or two female offspring per litter were included in each test, except PPI where all available females were included (Table 2). To control for the inherent relationships between siblings, effects from littermates were averaged and one value per litter was used (Zorrilla, 1997;Lazic, 2013). Estrous phase was determined daily between the hours of 7 and 8:30 A.M. before behavior testing. A single investigator used lavage with a p200 pipette and 20 l of sterile physiologic saline to collect cells from the vaginal wall for immediate visual examination with a light microscope. Proestrus was defined by the presence of uniform nucleated cells, while unnucleated cornified squamous cells were characteristic of estrus, densely packed leukocytes indicated metestrus, and scattered leukocytes alongside nucleated cells indicated diestrus (Hubscher et al., 2005). Estrous determination began 5 d before behavior testing and continued throughout experimentation. All rats displayed a typical 4-to 5-d cycle. Additional handling included exposure to investigators and emphasized picking up and moving the rats until these motions could be conducted with ease, as well as habituation to transport between the housing and testing locations. All animal work occurred during the light phase (7 A.M. to 7 P.M.) with the majority of behavior testing performed between 8:30 A.M. to 5 P.M. Testing began at eight weeks of age (young adulthood) and was completed by 15 weeks of age. The order of testing was PPI, crossmodal object recognition (CMOR), sociability, oddity discrimination, and finally MK-801-induced locomotor activity (Fig. 1B). Ethanol (40%) was used to clean all behavior testing equipment between rats.

PPI
PPI measures the percentage attenuation of motor response to a startling tone when the tone is preceded by a brief prepulse ( Fig. 2A; Lins et al., 2018). Two SR-LAb startle boxes (San Diego Instruments) were used. Each session had constant background noise (70 dB) and began with 5 min of acclimatization, followed by six pulse-alone trials (120 dB, 40 ms). Pulse-alone (6), prepulse ϩ pulse (36), and no stimulus (6) trials were then presented in a pseudorandom order, followed by six additional pulse-alone trials. Prepulse ϩ pulse trials began with a 20 ms prepulse of 3, 6, or 12 dB above background (70 dB). Prepulse-pulse intervals (time between the onset of the prepulse and the 120 dB pulse) were short (30 ms) or long (80 ms). The intertrial interval varied randomly from 3 to 14 s (Meyer et al., 2009;Howland et al., 2012;Ballendine et al., 2015).

Sociability task
The testing apparatus was a rectangular arena (150 ϫ 40 cm) of black corrugated plastic divided into three compartments, one middle start compartment (30 ϫ 40 cm) and two 'stranger' compartments on either side (60 ϫ 40 cm, see Fig. 3A; Henbid et al., 2017;Lins et al., 2018). Eight dams were killed within 48 h of polyI:C administration because they developed low body temperature and showed sickness behaviors beyond what is acceptable as outlined in our humane intervention protocol. One litter per treatment included male, but no female offspring, resulting in exclusion from the final count of litters included in this manuscript. The two saline-treated control dams that did not produce litters showed no evidence of pregnancy. Viable offspring count per dam includes all surviving offspring of both sexes present on PND1 before culling to a maximum of 10, excludes dams that did not give birth, and is presented as mean Ϯ SEM. All additional data presented on the dams only includes those that produced viable offspring. Number of litters with n ϭ 5 or fewer offspring included in PPI Treatment n ϭ 5 n ϭ 4 n ϭ 3 n ϭ 2 n ϭ 1 Saline 1 5 1 0 8 PolyI:C 1 0 7 1 7 Number of litters with n ϭ 1 or n ϭ 2 offspring in CMOR, sociability, and oddity Treatment n ϭ 2 n ϭ 1 Saline 7 8 PolyI:C 9 6 Number of litters with n ϭ 1 or n ϭ 2 offspring tested in MK-801-induced locomotor activity Treatment n ϭ 2 n ϭ 1 Saline 3 8 PolyI:C 7 6 All female offspring from n ϭ 31 litters completed PPI, while all other tasks included one or two offspring per litter. The unexpected death of one polyI:C rat with no female littermates reduced the number of litters tested to 15. Locomotor activity n was reduced due to some rats being diverted to concurrent research. The number of offspring per litter in each task is further summarized below. behavior scores for littermates were averaged for a single value per litter.
The walls dividing the middle compartment from the stranger compartments were clear Plexiglas (extend 12 cm from each wall leaving a 16-cm opening allowing travel between compartments) and removable black opaque barriers which, when inserted, prevented entry into the stranger compartments. Each stranger compartment contained a circular mesh cage (18 cm in diameter, 20 cm in height) with hinged lid (3/4" plywood, painted matte black). The height of the cage was extended 20 cm with vertical metal rods to discourage climbing. The task began with 10 min habituation with the barriers removed. The test rat was then contained in the middle section with the barriers in place and a stranger rat was placed in one of the mesh cages. The barriers were removed, and the test rat explored for an additional 10 min. Video recording and locomotor activity tracking was done with EthoVision software, and videos were manually scored with a stopwatch by a trained investigator blind to treatment status, and the opaque cage roof obscured the location of the stranger rat. Stimulus exploration was scored when the test rat directly approached (watching, contacting, sniffing, or circling) each of the cages, with the face of the rat oriented toward the cage at a maximum distance of 2 cm. All stranger rats were sex, age, and treatment matched to the test rat (Bitanihirwe et al., 2010;Henbid et al., 2017).

MK-801-induced locomotor activity
The apparatus was a square arena (40 ϫ 40 ϫ 60 cm) made of black corrugated plastic ( Fig. 3D; Lins et al., 2018). A camera mounted to the ceiling recorded all activity and EthoVision software was used to track activity. Rats were tested four at a time, with each rat placed in one of four separate arenas for 30 min of habituation. Immediately following, rats were administered MK-801 (0.1 mg/kg, i.p.) and placed back into the arena for an additional 120 min. Activity was recoded with Noldus EthoVision XT 11.5 software.

Visual, tactile, and CMOR
This task uses spontaneous exploratory behavior to assess visual memory, tactile memory, and visual-tactile sensory integration (Winters and Reid, 2010;Jacklin et al., 2012). The testing apparatus was a Y-shaped maze with one start arm and two object arms (10 ϫ 27 cm) made of white corrugated plastic ( Fig. 4; Lins et al., 2018;Paylor et al., 2018). A white plastic guillotine-style door separated the start arm from the object arms, and Velcro at the distal end of the object arms fixed objects in place. A removable, clear Plexiglas barrier could be inserted in front of the objects. A tripod positioned above the apparatus held a video camera that recorded the task activity. Rats were habituated to the apparatus twice for 10 min. Lighting alternated during habituation between white light (during visual phases) and red light (during tactile phases) for 5 min each with the order counterbalanced, and the clear barriers were in place for 1 d of habituation and removed for the other with order counterbalanced. Test days consisted of a 3 min sample phase with two identical Figure 2. A, Schematic illustrating a startle response to a 120 dB tone (top panel) versus the typical reduction in startle reactivity when a prepulse of 3, 6, or 12 dB precedes the startling tone (bottom panel). Schematic has been published previously (Lins et al., 2018). B, Startle reactivity decreased over the course of the PPI testing protocol and polyI:C offspring had significantly higher reactivity at the "after" time point (‫ء‬p Ͻ 0.05). C, There were no differences between groups in %PPI for the short (30 ms) prepulse-pulse interval but %PPI increased with increasing prepulse intensity where the 12 dB prepulse had higher PPI than 2 or 6 dB prepulses. D, There were no differences between groups in %PPI for the long (80 ms) prepulse-pulse interval but %PPI increased with increasing prepulse intensity (3 dB Ͻ 6 dB Ͻ 12 dB; ‫ء‬p Ͻ 0.05).
copies of an object attached with Velcro to the maze, a 60-min delay, and then a 2-min test phase with a third copy of the original object and a novel object placed in the maze. Rats began each phase in the start arm; the guillotine door was opened and closed once the rat entered the object arms. This task consisted of three distinct tests performed on three separate days in the following sequence: tactile memory (day 1; Fig. 4A), visual memory (day 2; Fig. 4C) and crossmodal memory (day 3; Fig. 4E). Red light illuminated the tactile phases allowing the rats' behavior to be recorded while preventing the rats' visual assessment of the objects and the removal of the clear barriers allowed for tactile exploration. White light was used during visual phases, but clear Plexiglas barriers in front of the objects prevented tactile exploration. CMOR had a tactile sample phase (red light, no barriers) and a Figure 3. A, Schematic representing the black, three-chambered arena used to conduct the sociability task. The chambers on either side of the center start chamber contain identical holding cages, one of which would contain a social stimulus (an age, sex, and treatment matched stranger rat), while the test rat was free to explore. Schematic has been published previously (Henbid et al., 2017;Lins et al., 2018). B, When the exploration data are presented as a DR, both groups show significant preference for the stranger rat; however, polyI:C offspring show significantly less preference when compared to saline offspring. C, There was no significant difference between groups in total exploration or exploration of the social stimulus, although polyI:C rats spent more time exploring the non-social stimulus than saline rats. D, Schematic of the black, square arena where rats' activity was monitored before and after administration of MK-801. Schematic has been published previously (Lins et al., 2018). E, Graph displaying locomotor activity as distance traveled per 10 min time bin. Both groups had elevated locomotor activity following MK-801 administration, but there was no effect of maternal treatment. ‫ء‬p Ͻ 0.05. visual test phase (white light, clear barriers). Recognition memory was defined as significantly greater exploration of the novel object than the familiar object. Behavior recordings were manually scored with a stopwatch by investigators blind to the treatment status of the rats and identity of the objects (Winters and Reid, 2010;Ballendine et al., 2015).

Oddity discrimination
The testing apparatus was a square arena (60 ϫ 60 ϫ 60 cm) constructed of white corrugated plastic with Velcro in each of the four corners. Two days of habituation to the arena (10 min sessions) preceded the test day. On test day, three identical objects made of glass or plastic and one different, or "odd" object were fixed to the Velcro locations ( Fig. 5A; Lins et al., 2018) and the rats' activity were recorded for 5 min using a video camera mounted to the ceiling. Object exploration times were manually scored using a stopwatch by an investigator blind to the treatment status of the rats (Bartko et al., 2007a). Object examination was counted when a rat's face was oriented toward the object at a maximum distance of 2 cm.

Statistical analyses
A between-subjects design was used, and analyses were conducted with independent samples t tests, one sample t tests, and ANOVAs using Statistical Package for the Social Science version 22 (IBM). Outliers were defined as having a performance metric falling Ͼ2 SDs from the mean and were removed from analysis on a case by case basis. Outliers were identified and removed before calculating litter averages to prevent excessive exclusion of data points. One saline and two polyI:C rats were re- Figure 4. A, The Y-maze assembled to conduct the tactile phase in red light conditions where the rat is able to explore objects via touch. B, both groups of offspring display robust novelty preference in the tactile phase with novel object exploration significantly greater than chance levels. C, The Y-maze assembled for the visual phase which is conducted in white light conditions with the addition of a clear, Plexiglas window to prevent tactile exploration of the objects, limiting the rats to visual observation. D, Saline offspring demonstrated visual memory with novel object exploration significantly greater than chance but polyI:C offspring did not perform above chance levels. E, The Y-maze assembled for the crossmodal phase which has a tactile sample phase and visual test phase. F, both groups failed to display crossmodal recognition memory as novel object exploration was equal to chance; # indicates significant difference from chance exploration (DR ϭ 0, p Ͻ 0.05) in a single sample t test. Schematic has been published previously (Lins et al., 2018;Paylor et al., 2018). moved from visual recognition memory, but all litters remained represented. Two polyI:C litters were excluded from analysis for startle to the P120 tone, one polyI:C litter was removed from the analysis for PPI 30 ms interval and PPI 80 ms interval. No litters or individuals were removed from the Oddity, Sociability, or Locomotor task analyses. Estrous phase was incorporated into analysis as a covariate, however no consistent patterns were observed, potentially due to a low n of rats in each phase. Sphericity violations were accounted for using the Greenhouse-Geisser adjustment and degrees of freedom were adjusted when Levene's Test was violated. The use of onetailed and two-tailed tests is specified for each task. Relationships between maternal serum cytokine concentrations and long-term offspring outcomes were determined using bivariate correlations followed by a Benjamini-Hochburg (B-H) adjustment to control for multiple comparisons (Benjamini and Hochberg, 1995). All data are presented as group mean Ϯ SEM, and asterisks indicate a significant difference between groups with a 95% confidence interval (p Ͻ 0.05). The pound symbol (#) is used to indicate a significant difference from a chance result.

Results
The acute effects of saline and polyI:C treatment on this cohort of pregnant dams and neonatal pups have been published previously (Lins et al., 2018). Briefly, dams treated with polyI:C had reduced body weight compared to saline when followed up at 8, 24, and 48 h after treatment, but no significant change in body temperature at the same timepoints. Maternal serum collected 3 h after treatment was analyzed with ELISA for concentrations of cytokines CXCL1, CXCL2, IL-6, and TNF-␣. CXCL1 and IL-6 were significantly elevated in polyI:C-treated dams. On PND1, the average offspring mass (males and females pooled) from polyI:C-treated litters was significantly less than controls but there was no difference in litter size between the groups (Lins et al., 2018).

Maternal polyI:C treatment failed to significantly affect startle or PPI
Startle responses to acoustic stimuli were assessed by measuring startle alone and PPI in saline (n ϭ 15 litters) and polyI:C female offspring (n ϭ 15 litters). Startle to the 120 dB pulses alone decreased during the session (main effect of time: F (1.35,36.45) ϭ 26.26, p Ͻ 0.001; Fig. 2B) but no treatment or interaction was present. For prepulse trials with a 30 ms (short) interval, a main effect of prepulse intensity on PPI (F (2,56) ϭ 40.33, p Ͻ 0.001; Fig. 2C) was found with no effect of treatment. Overall, PPI was greater at 12 dB compared to 3 and 6 dB (p Ͻ 0.001). For trials with an 80 ms (long) prepulse-pulse interval, a main effect for prepulse intensity was found (F (2,56) ϭ 89.37, p Ͻ 0.001; Fig. 2D) for PPI, but no effect of treatment and no interaction. Overall, PPI increased with louder prepulses.

PolyI:C offspring have sociability deficits
Both groups of female offspring (n ϭ 15 saline litters, n ϭ 15 polyI:C litters) displayed a significant preference for the cage containing an unfamiliar rat compared to the empty cage when analyzed using a within-subjects design (saline: t (14) ϭ 14.18, p Ͻ 0.001; polyI:C: t (14) ϭ 8.11, p Ͻ 0.001). When the results were compared between treatment groups, female polyI:C rats had a significantly lower discrimination ratio (DR; calculated as exploration stranger -exploration empty /exploration total ; t (28) ϭ 2.61, p Ͻ 0.05; Fig. 3B) and spent significantly more time exploring the empty cage compared to the saline controls (t (28) ϭ -2.59, p Ͻ 0.05; Fig. 3C). There was no difference in total exploration times (p Ͼ 0.05).

Both groups of offspring increase locomotor activity following MK-801 administration
Locomotor data comparing female polyI:C (n ϭ 13 litters) and saline (n ϭ 11 litters) offspring were analyzed with a mixed repeated measures ANOVA (Fig. 3E). Results revealed a main effect of Time (F (2.47,62.24) ϭ 62.24, p Ͻ 0.001; Fig. 3E) but no treatment effect and both groups displayed increased locomotion after MK-801 administration.

PolyI:C offspring perform tactile, but not visual, object recognition memory and neither group display crossmodal recognition
All CMOR data are presented as a DR (exploration novel -exploration familiar /exploration total ) for the first minute of the test phase. One-tailed single sample t tests compared each group's exploration to chance (DR of 0). Both groups Figure 5. A, Schematic of the white square arena used to conduct the oddity discrimination task showing the arrangement of three identical objects and one different, or odd object. Schematic has been published previously (Lins et al., 2018). B, bar graph displaying the percentage of total object exploration spent examining the odd object. PolyI:C offspring displayed significantly less oddity preference than saline offspring (‫ء‬p Ͻ 0.05).

Correlations between measurements taken of the dams during pregnancy and offspring outcomes
Measurements taken during pregnancy were correlated with long-term behavior outcomes in the female offspring.
Serum cytokine concentrations of CXCL1, CXCL2, IL-6 and TNF␣ were determined from blood samples collected from the dams 3 h after treatment and analyzed with ELISA. Additional effects of treatment were determined through monitoring with weight and rectal temperature measurements taken 8, 24, and 48 h after treatment (Lins et al., 2018). These data were correlated to behavior of the offspring using bivariate correlations. Maternal weight changes following treatment (anesthesia with saline or polyI:C administration and blood sampling) was the only variable associated with offspring behavior. Greater weight loss in polyIC-treated dams 8 h after treatment was associated with reduced startle response in their female offspring during the initial tone-alone trials (r ϭ 0.623, p Ͻ 0.05, B-H p Ͻ 0.05), and this was not seen in the saline group (r ϭ 0.44, p Ͼ 0.05). Weight loss in the saline dams at 24 h after treatment was correlated to lower %PPI at the 80 ms interval (r ϭ -0.555, p Ͻ 0.05, B-H p Ͻ 0.05). Saline dam weight loss 8 h after treatment was related to the DR in the sociability task (r ϭ 0.716, p Ͻ 0.01, B-H p Ͻ 0.05). The importance of these relationships is difficult to gauge due to no significant effects of saline treatment on weight, and the lack of treatment effects in startle or PPI. Previous studies have also shown mixed results regarding maternal weight changes and offspring behavior outcomes Vorhees et al., 2012). Dedicated studies will be necessary to determine the reliability and potential importance of these results.

Male and female offspring show similar behavioral profiles in response to MIA
The data presented in this paper were further analyzed in conjunction with the male littermates from Lins et al. (2018) with sex and treatment as factors using 2 ϫ 2 factorial ANOVAs. It should be noted that this analysis necessitates including two values per litter (for each sex) which violates the assumption that subjects are independent because littermates are inherently related. No main effect of treatment (F (1,60) ϭ 0.59, p Ͼ 0.05) or sex (F (1,60) ϭ 0.25, p Ͼ 0.05), and no sex by treatment interaction (F (1,60) ϭ 0.11, p Ͼ 0.05) was found for 30 ms interval PPI. For 80 ms interval PPI trials, no main effect of treatment (F (1,56) ϭ 0.10, p Ͼ 0.05) or sex (F (1,56) ϭ 0.82, p Ͼ 0.05) was found, but a sex by treatment interaction was shown (F (1,56) ϭ 4.07, p Ͻ 0.05). A Tukey HSD post hoc test revealed saline females had lower PPI than saline males (p Ͻ 0.05). Main effects of treatment (polyI:C offspring have a lower DR than saline offspring; F (1,58) ϭ 9.55, p Ͻ 0.01) and sex (female offspring have a lower DR than male offspring; F (1,58) ϭ 4.87, p Ͻ 0.05) were found for sociability, with no significant interaction (F (1,58) ϭ 0.13, p Ͼ 0.05). Tactile object recognition memory did not differ by treatment (F (1,59) ϭ 0.30, p Ͼ 0.05) or sex (F (1,59) ϭ 1.06, p Ͼ 0.05); however, but a significant sex by treatment interaction was revealed (F (1,59) ϭ 4.25, p Ͻ 0.05). Tukey HSD post hoc testing failed to reveal any significant differences between individual groups. Visual object recognition memory was not affected by treatment (F (1,56) ϭ 3.60, p Ͼ 0.05) or sex (F (1,56) ϭ 0.21, p Ͼ 0.05), and no 1, heightened response or facilitation; 2, diminished response or impaired performance in comparison to a control group; -, no significant change compared to controls; n.d., not determined. Percentage change was calculated as a comparison to the equivalent control group [(saline -polyI:C/saline) ϫ 100]. Male locomotor data were calculated from the total distance traveled after MK-801 administration. For consistency, both male and female sociability percentage change was calculated using the DR data, although it should be noted the DR was a non-significant (n.s.) effect in the males and they instead spent significantly less time (s) exploring the social stimulus than controls.

Discussion
The adult female offspring of rat dams that received an immune stimulant during pregnancy displayed various behavior abnormalities compared to the offspring of saline-treated dams. The polyI:C-treated offspring had reduced sociability, impaired visual discrimination, and lack of preference for an odd object compared to offspring from control litters. both treatment groups displayed heightened locomotor activity in response to MK-801 administration and tactile recognition memory was intact in both groups. Neither group of offspring demonstrated crossmodal memory, and there were no treatment effects on PPI. These results complement a companion paper that assessed the male offspring (Lins et al., 2018), and by directly analyzing sex by treatment interactions where possible, we shown that MIA during pregnancy had similar effects on both sexes of offspring.
A significant limitation of this study is the use of timedpregnant dams. Several studies show an impact of travel stress on the dams and offspring. For example, Moriyama et al. (2013) examined the effect of transport stress on seizure susceptibility in the offspring and found an increase in variability in those transported during gestation; however, maternal care behavior had a greater impact than transport stress on seizure susceptibility. We have previously reported no observed changes to maternal behavior following polyI:C administration , but we did not assess this directly in this cohort or strain. Shipment stress also increases susceptibility to the valproate-induced developmental toxicity model of autism (Ogawa et al., 2007;Kuwagata et al., 2009). We are unable to confirm if shipment stress had a similar impact in our study. Despite these limitations, many comparable studies on development and gestational adverse events have relied on the use of timed-pregnant dams (Lodge and Grace, 2001;Grace, 2013, 2016a,b;Van den Eynde et al., 2014;Ballendine et al., 2015;Lins et al., 2018). Recently, Kentner et al. (2019) highlighted that consideration of all MIA protocols will enable comprehensive understanding of their impacts on offspring outcomes. Thus, we believe our results are of value.

Lack of sex-specific effects of polyI:C treatment on behavior of the offspring
We previously reported that male polyI:C offspring from this cohort displayed greater startle to the 120 dB tone at the end of the PPI protocol compared to saline males.
Although the effect in the males was small and limited to a single parameter, the female data presented here shows no effect of MIA on any measure of PPI and acoustic startle response. The effects of MIA on PPI in rodent models are mixed with many studies showing PPI impairments in the offspring of immune challenged rats (Borrell et al., 2002;Romero et al., 2007;Bilkey, 2008, 2010;Dickerson et al., 2010Dickerson et al., , 2013Dickerson et al., , 2014Howland et al., 2012;Klein et al., 2013;Ballendine et al., 2015;Hadar et al., 2015) and mice (Ozawa et al., 2006;Smith et al., 2007), including impairments seen in both sexes (Meyer et al., 2009;Howland et al., 2012;Basta-Kaim et al., 2015). Other studies show no effects of MIA on PPI, similar to our observations Van den Eynde et al., 2014;Vorhees et al., 2015;Lins et al., 2018). Sex effects in PPI in general have been reviewed and the influence of female sex hormone fluctuations have been studied (Kumari, 2011). High estrogen phases of the menstrual cycle have been associated with lower PPI, although this is not consistently observed (Swerdlow et al., 1997;Jovanovic et al., 2004;Kumari et al., 2008). Additionally, PPI disruption by a 5HT 1A agonist can be prevented by administration of exogenous estrogen and progesterone in rats which may imply a protective role of sex hormones against PPI disruption (Gogos and Van den Buuse, 2004). The results from the present study do not support the claim of a strong influence of estrous phase on PPI performance; however, it should be noted that estrus was not controlled for and the ability of the present study to detect an effect may be underpowered.
In the sociability task, polyI:C-treated male offspring spent less time exploring a same-sex, unfamiliar conspecific compared to saline controls (Lins et al., 2018). We observed a different pattern of reduced sociability in the female polyI:C offspring, indicated by a significantly lower DR compared to the saline offspring which was driven by significantly more time exploring the empty cage on the opposite side of the apparatus. The social exploration data are presented as a comparison between the saline and polyI:C offspring (comparing the relative degree of social preference), while others have presented the data as a within-subjects comparison to report either the presence or absence of social preference (Silverman et al., 2010). We believe comparing treatment groups allows the detection of subtle behavior differences that could be missed in instances where stimuli with a substantial difference in salience (such as an unfamiliar rat vs an empty cage) result in high DRs, and this natural preference would need to be abolished to show a treatment effect. Natural preference for social stimuli versus objects is documented in rodents (Lee and Green, 2016). Presenting the data as a between groups comparison also allows direct evaluation alongside previously published sociability data from our lab, including that of the male littermates from this cohort (Henbid et al., 2017;Lins et al., 2018). The use of single sample comparisons may be best suited to tasks that are challenging for control animals to complete, such as the complex visual discrimination tasks including visual and crossmodal recognition memory where visual stimuli are less salient and the resulting DRs tend to be lower (Winters and Reid, 2010). Both strategies of data analysis are common in behavior literature, and factors such as strength or salience of stimuli and task difficulty should be considered when representing data. Overall, both male and female polyI:C offspring display a deficit in sociability compared to controls, but this presents in a subtly different manner depending on sex and may be related to previously observed differences in PFC development (Piontkewitz et al., 2011b;Paylor et al., 2016), The direct significance of MIA-induced developmental trajectory differences to the aberrant social behavior observed here remains to be determined.
The effects of MK-801 administration on locomotor activity have been reported in previous studies with mixed results (Zuckerman and Weiner, 2005;Howland et al., 2012;Vorhees et al., 2012;Missault et al., 2014). The male siblings in this cohort were significantly affected by a dose of 0.2 mg/kg IP indicated by heightened locomotor activity which was not seen in the control males (Lins et al., 2018). The females in this paper were given a lower dose of 0.1 mg/kg comparable to other studies (Andiné et al., 1999;Howland et al., 2012;Zhao et al., 2013). Unfortunately, both the saline and polyI:C females showed increased locomotion which confounds the ability to discern whether prenatal polyI:C treatment affected sensitivity to MK-801.
In the CMOR task, both saline and polyI:C females demonstrated object recognition in the tactile phase of the test, similar to what was seen in males of the same cohort (Lins et al., 2018) and Long Evans males (Ballendine et al., 2015). PolyI:C-treated offspring were impaired in the visual phase, and neither group performed significantly different from chance exploration in the crossmodal phase, a notable distinction from the crossmodal memory exhibited by saline-treated males (Lins et al., 2018). Previous studies on MIA offspring have shown reduced discrimination in object memory behavior tasks in females and lower DRs are common in the crossmodal task, possibly reflecting task difficulty Ballendine et al., 2015;Marks et al., 2016). Visual and crossmodal memory depend on the perirhinal cortex while the posterior parietal cortex is necessary for tactile memory, suggesting there may be regionally specific deficits as a result of MIA (Winters and Reid, 2010;Jacklin et al., 2016).
Oddity preference and perception have been assessed in several tasks using rats and mice (Bussey et al., 2005;Bartko et al., 2007a,b;Cowell et al., 2010;Cloke et al., 2016;Marks et al., 2018;Paylor et al., 2018); however, to our knowledge this is the first study to assess this oddity task in female rats. Improved understanding of the nature of oddity discrimination is relevant for the successful management of cognitive impairment in conditions such as schizophrenia, a symptom domain highly related to patient functional outcomes . Prenatal polyI:C treatment affected females in the same manner as males with a significant reduction in oddity preference compared to saline offspring (Lins et al., 2018). The successful performance of oddity preference depends on multisensory integration similar to CMOR, yet distinct in that visual and tactile associations can be formed simultaneously and there is no mnemonic demand in the oddity task . Multisensory integration is disrupted by NMDA receptor antagonism using ketamine in the orbitofrontal cortex and reversed with ␣ 4 ␤ 2 nicotinic acetylcholine in a GABA A -dependent mechanism  and abnormalities in these brain regions and receptor types may be good candidates to explore in future studies of MIA and oddity preference.
The degree to which these behavioral effects replicate or contradict previous data varies. The MIA literature displays a lack of reproducibility, which may be due to variety of protocols used. Procedural variations in model species and strain, timing of inflammatory insult, inflammatory agent, and dose and route of administration have the potential to alter experimental outcomes. Other details such as rodent housing (bedding type, pathogen-free status, temperature, etc.), parental age, maternal experience, food and water quality, cage companions, and age at weaning, which are not commonly reported, may influence outcomes and reduce reproducibility of the model (Smolders et al., 2018;Kentner et al., 2019). The basic protocol used here is relatively common in MIA literature, yet very similar protocols yield contrasting behavior responses; for example, hypolocomotion versus hyperlocomotion in an open field (Van den Lins et al., 2018). Comparison in this case is complicated by the use of spontaneous versus drug-enhanced locomotor paradigms and the collection of maternal blood samples Lins et al., 2018). Indeed, distinct neuopathological alterations were noted with microglia activation in MIA offspring found in one study , while previous work from our group has found no changes in microglia in offspring generated in our laboratory (Paylor et al., 2016). These results support the notion that enhanced reporting of such variables is warranted, and dedicated future studies should assess the effects of such procedural differences directly (Goldstein et al., 2014;Careaga et al., 2018;Kentner et al., 2019;Mac Giollabhui et al., 2019).

Implications for sex differences in the MIA model
MIA caused by polyI:C administration resulted in altered behavior in female offspring in multiple behavior tasks including sociability, visual and crossmodal memory, and oddity preference. Overall, the present data do not provide strong evidence for sex differences in response to polyI:C treatment. Inflammation in pregnancy relates to the etiology of sexually dimorphic disorders, notably schizophrenia and autism (Davis and Pfaff, 2014;Goldstein et al., 2014;Patel et al., 2018). The lack of overt sex differences observed in the present study suggest the MIA model in rats may be limited in this respect, although others have shown more promising results in this regard (Piontkewitz et al., 2011a;Zhang et al., 2012).