Maternal gastrointestinal nematode infection enhances spatial memory of uninfected juvenile mouse pups

The developing brain is particularly vulnerable to factors including maternal infection during pregnancy. Establishment of neural networks critical for memory and cognition begins during the perinatal period, when Heligmosomoides bakeri, a gastrointestinal (GI) nematode restricted to the maternal mouse intestine, has been shown to upregulate expression of long-term potentiation genes in the young rodent pup brain. We explored the impact of maternal infection during pregnancy and early lactation on the spatial behavior of uninfected male and female juvenile mice. Pre-weaned pups of H. bakeri infected dams exhibited less exploratory behaviour compared to pups of uninfected dams on postnatal day (PD) 16 but not PD 17, possibly reflecting a transient fear of an unfamiliar environment and/or a brief neurodevelopmental delay. Our two spatial memory tests show for the first time an enhancement of spatial memory in response to maternal nematode infection regardless of pup sex. At PD 17, pups of infected dams expressed object location memories after 3 h in the Object Location Test whereas offspring of uninfected mothers did not. In addition, at PD 34, juveniles of infected mothers retained their ability to find the escape hole in the Barnes Maze Test for one week whereas offspring from uninfected mothers did not. This finding is even more striking given that spatial memory was positively associated with pup length, yet this maternal infection impaired linear growth of pups. Thus, the positive impact of maternal infection on spatial memory countered any impairment associated with the shorter length of the pups. Overall, these novel findings indicate that a maternal GI nematode infection during pregnancy and lactation positively influences the spatial memory of uninfected juvenile offspring with potential fitness implications for the next generation.

Impact of maternal infection on offspring spatial behavior. Early exploratory behavior. In the OFT, pups of infected dams exhibited less spatial exploration at PD 16 than pups of uninfected dams as evidenced by lower total path traveled (P = 0.01), mean velocity (P = 0.01), and time in center zone (P = 0.015) and a greater time without movement (P = 0.006) ( Table 1). There were no sex effects (all P values > 0. 15 www.nature.com/scientificreports/ Object location memory. There was no bias in exploration of Objects 1 and 2 during the training phase (P = 0.40). Furthermore, maternal infection did not affect the total path traveled, mean velocity, time without movement, object investigation time (Table 2) or % investigation time of mobile object ( Fig. 1) during the training trial or during the test trial. Additionally, there were no sex effects (all P values > 0.27, data not shown). However, when using our derivative variables to compare object location memory between training and test trials, we found that maternal infection improved pup memory. Comparing the % investigation time of mobile object between training and test trials, pups born to infected dams remembered object locations and explored the moved object significantly more during the test trial compared to the training trial (P = 0.019) (Fig. 1) whereas pups of uninfected dams spent a similar % investigation time of mobile object in both the training and test trial (P = 0.74) indicating that they had not retained object location memory after a 3 h period (Fig. 1). This was also reflected in the change in % investigation time of mobile object (Fig. 2). Pups of infected dams increased their investigation of the mobile object during the test trial whereas pups of uninfected dams did not, and the difference between these two groups was significant (P = 0.027), providing further evidence that pups of infected dams were able to recall object location memory for 3 h. These findings were not affected by offspring sex (all P values > 0.28, data not shown), but change in % investigation time of mobile object was positively associated with pup length (P = 0.041, data not shown), independent of maternal infection.
Spatial learning. In the BMT, regardless of maternal infection or offspring sex, pups learned the location of the escape hole on the first training day as indicated by a decrease in the average latency (P < 0.0001; Fig. 3a), path length (P < 0.0001; Fig. 3b), and errors (P < 0.0001; Fig. 3c) between training days 1 and 2. Thereafter, values remained low. Neither maternal infection nor offspring sex influenced mean velocity (all P values > 0.27, data not shown).
Short and long-term spatial reference memory. Neither maternal infection (all P values > 0. 34, Fig. 4a-c) nor offspring sex (all P values > 0.47, data not shown) altered short-term spatial reference memory in probe trial 1. However, regardless of sex, offspring born to infected dams had enhanced long-term spatial reference memory (Fig. 4) as assessed in probe trial 2. Offspring of infected dams had lower latency (P = 0.0044; Fig. 4a), path length Table 1. Effect of maternal H. bakeri infection on displacement variables measured during the 10 min Open Field Test. Pups were nested within dam and pup crown-rump length was included as a covariate. Since no significant sex differences were evident, pooled female and male data are shown. Values are LSmeans ± SEM of each outcome variable, n = 32 pups from 8 uninfected dams and n = 32 pups from 8 infected dams.  Fig. 4b), and fewer errors (P = 0.0031; Fig. 4c) in finding the escape hole than offspring of uninfected dams. Furthermore, when controlling for individual performance in probe trial 1, offspring of infected dams retained their memory over the one-week interval whereas the performance of offspring of uninfected dams declined strongly as shown by the positive change in latency (P = 0.0067; Fig. 4d), path length (P = 0.015; Fig. 4e), and errors (P = 0.01; Fig. 4f

Discussion
Using a nematode parasite that remains in the maternal intestine, we tested our hypothesis that maternal infection during pregnancy and lactation would positively influence the spatial behavior of pre-weaned and juvenile uninfected male and female offspring. We report for the first time that PD 16 offspring of H. bakeri infected dams exhibit less exploratory behaviour compared to pups of uninfected dams, possibly reflecting transient fear of an unfamiliar environment and/or a brief developmental delay. However, in response to maternal infection, PD 17 offspring exhibited better retention of object location memory and at PD 34 they had enhanced long-term spatial reference memory. These novel findings indicate that a maternal GI nematode infection during pregnancy and lactation positively influences the spatial memory of uninfected juvenile mice, despite the negative impact of maternal infection on the linear growth of the pup. Findings from the OFT indicate that on first introduction to an open arena, offspring of H. bakeri infected mothers explored less compared to offspring of uninfected mothers, raising the possibility of a developmental delay and/or heightened fear or anxiety 17,26 . As spontaneous exploration in an open field is first detectable between PD 16-19 17,26 , it is possible that some component of neurodevelopment is delayed at PD 16 in response to maternal infection, which may have negative consequences for the offspring. However, this lower exploration was not observed one day later when these pups were placed in the open field with two novel objects, suggesting that if a developmental delay did occur, it was brief, and may not have had consequences for the growing pup. These findings from the OFT are similar to other maternal infection studies as exposure of pregnant mice to E. coli 53 or influenza virus 25 resulted in heightened anxiety-like behaviours albeit in adolescent (5 week old) 53 and adult (9 month old) 25 offspring, indicated by less exploratory behavior in an OFT. While a heightened fear/ anxiety response can be considered a negative attribute, under some circumstances it can be advantageous to the host 3 . Fear and anxiety act as a response to danger or threat 54 , thus when exposed to an unknown environment, mammals typically freeze as it is more difficult for a predator to observe a non-moving animal 26 . Considering that wild rodents are exposed to a number of natural predators, the lower exploration in the open field arena in response to maternal infection may indicate more caution when placed in an unknown and potentially dangerous environment which could actually be beneficial to survival.
The ability to recognize and remember the spatial characteristics of the environment, such as the location of objects, is an important component of spatial cognition 9,19 . This typically begins in 16-17 day old rodents with memory lasting only for a few minutes 18,29 , but for a few hours in 21-24 day old subjects 18,30 . Thus, our observation that PD 17 pups of uninfected mothers were unable to detect object rearrangement after a three hour period was consistent with the literature and suggests a normal immaturity in recalling spatial information at PD 17 55 . However, despite their young age we found that PD 17 pups of H. bakeri infected mothers were able to retain object location memories for three hours, as evidenced by a significant increase in investigation of an object after it had been moved. This finding is in contrast with reports that exposure of pregnant rodents to viral mimics had no influence on offspring object location memory, although the studies were done using adult offspring 56,57 . Our findings indicate that the maturational process needed to recall object location memories for three hours occurred earlier as a result of maternal H. bakeri infection. This is consistent with recent findings that maternal H. bakeri infection up-regulated expression of genes associated with LTP in brains of perinatal uninfected offspring 48 and thus may promote cognitive development. www.nature.com/scientificreports/ The ability to learn the route to an escape location is detectable at PD 21 in rodents 17,19,32,33 , however, longterm reference memories for an escape location in the Morris water maze do not emerge until much later 32,33 . The Morris water maze is similar to the BMT as it assesses spatial learning and reference memory by testing the ability of a subject to locate a hidden underwater platform in order to escape from water in a circular water tank 9 . When PD 20, 34 and 60 subjects were tested in a Morris water maze, all age groups were capable of learning the route to the escape location, and remembering this location for one-day 33 . However, PD 20 and 34 rodents were not yet capable of retaining long-term reference memories for a one-week period whereas PD 60 subjects were 33 . We assessed spatial learning over four days from PD 23-26, followed by short-term reference memory one day later at PD 27 and finally long-term reference memory one-week later at PD 34. Maternal infection had no impact on spatial learning or short-term reference memory, but long-term reference memory was enhanced as a result of this maternal infection. The ability of the juvenile control pups to learn the location of the escape box and recall this location after one day but not one week was consistent with studies using the Morris water maze 32,33 . Unlike control pups, offspring from infected mothers were capable of retaining long-term reference memories for a week as they performed equally well after the one-week delay, compared with the one-day delay. These findings are in the opposite direction to reports from maternal E. coli infection models where exposure of pregnant rodents impaired offspring spatial learning and short and long-term reference memory in the Morris water maze 23,24,35 . Overall, our findings reinforce our observation from the OLT that the maturational processes required for the retention of spatial memories occur earlier as a result of this maternal infection and lead us to speculate that maternal H. bakeri infection may increase the fitness of the next generation. Effect of maternal H. bakeri infection on offspring short-term (probe trial 1) and long-term (probe trial 2) spatial reference memory and change in reference memory between probe trials 1 and 2 in the Barnes Maze Test. Probe trial 1 was conducted 24 h after the last training day and probe trial 2 was conducted one-week later. Graphs A, B and C compare juveniles of uninfected vs H. bakeri infected dams within probe trial 1 and 2. Graphs D, E and F show the change in performance between probe trials 1 and 2 calculated by subtracting the value in probe trial 1 from the value in probe trial 2 for each mouse. A value of zero or a negative value would indicate that the subject performed as well or better during probe trial 2 compared to probe trial 1, suggesting strong memory retention. (a) latency, (b) path length, (c) errors to reach the escape hole, (d) change in latency, (e) change in path length and (f) change in errors. Pups were nested within dam, and pup crown-rump length was included as a covariate. Since no significant sex differences were evident, pooled female and male data are shown. Values are LSmeans ± SEM, n = 32 pups from 8 uninfected dams and n = 32 pups from 8 infected dams (ns not significant, *P < 0.05, **P < 0.01). www.nature.com/scientificreports/ Some evidence of sex dependent differences in spatial learning and memory of offspring has been reported in response to prenatal infection mimics whereby molecules of pathogens are injected into the pregnant dam [34][35][36] . For example, exposure of pregnant rats to E. coli lipopolysaccharide (LPS) impaired spatial learning and reference memory in the Morris water maze in 28-day-old male but not female offspring 35 . The underlying mechanisms are unknown, although sex hormones might play a role 35,37,42 . Other studies have shown no impact of offspring sex on spatial behaviour in response to prenatal infection mimics 58,59 . Our results using a direct nematode infection of pregnant mice are consistent with these latter studies in that offspring sex did not affect spatial exploration by offspring in the OFT, their ability to retain object location memories in the OLT, or to learn or remember the escape location in the BMT. Similarly, in the absence of maternal stress, no difference in the spatial behaviour and memory performance was observed between male and female pre-weaned (PD 17-18) CD-1 mice and rats in an OFT and OLT 18,55 , nor between juvenile (PD 22) male and female mice in a Morris water maze test 60 .
The observed impact of maternal infection on spatial learning and memory of their pups could be an indirect consequence of infection-induced nutrient deficiencies but evidence from the literature suggests that this is unlikely. Unlike hookworms that feed on blood and can lead to iron deficiency anemia and protein deficiency when in high numbers 61 , adult H. bakeri feed on the epithelial cell layer of the small intestine and are not typically associated with blood loss 62 . Despite lower maternal food intake in response to H. bakeri infection during pregnancy 63 , we found no impact of infection on maternal body mass during pregnancy or lactation or on the date of delivery or litter size, all of which would be expected consequences of maternal malnutrition 46,64 . Furthermore, total serum protein concentrations have been shown to be higher in H. bakeri infected dams at day 20 of lactation 65 . In the absence of evidence of nutrient deficiencies in pups of H. bakeri infected dams and knowing that nutrient deficiencies would be expected to impair not improve spatial memory [66][67][68][69][70] , it is unlikely that nutrient deficiencies account for the improved memory of pups in response to maternal infection.
Maternal H. bakeri infection is known to impair fetal 12 and offspring 46 linear growth as observed in our study and this impaired growth could have impacted spatial memory. In humans, height, brain size and general cognitive ability are positively correlated [38][39][40][41] . Rodent length is correlated with brain size 44 and brain size has been reported to be a strong indicator of cognitive ability, including the ability to find an escape location in laboratory mice 45 . Our novel finding that mouse length is directly correlated with spatial memory would lead to the expectation that the shorter pups of infected dams would have had impaired spatial memory. However, we found the opposite. Despite their smaller size, it is noteworthy that pups of infected dams were able to recall object locations for 3 h in the OLT and to recall the location of the escape box in the BMT for 1 week, whereas the larger pups of uninfected dams could not.
Formation and retention of spatial memories are controlled in the hippocampus and promoted by LTP and neurogenesis 20,71 . Our observation that offspring of H. bakeri infected mothers have enhanced spatial memory is consistent with previous evidence that the brains of PD 7 pups of infected dams have increased expression of LTP genes as well as the ITGA3 gene 48 , which may promote neurogenesis 72 . Further evidence for this hypothesis is found in physical exercise models, where exposure of mice to running enhances hippocampal neurogenesis and LTP which results in enhanced spatial memory performance in the Morris water maze 73,74 . Thus, we speculate that maternal H. bakeri infection is capable of enhancing hippocampal LTP and/or neurogenesis in the uninfected pup which promotes the enhanced spatial memory we observed. Further studies would be needed to explore this hypothesis.
The mechanism whereby a nematode living in the lumen of the maternal intestine could influence brain gene expression and alter cognitive processes which promote the spatial memory ability of offspring is unknown. One possibility is that the Th2/Treg immune response in the infected dam 75 induces a similar systemic response in the uninfected pup that extends to and alters the immune profile in the pup brain. Consistent with this, maternal H. bakeri infection up-regulated expression of Th2/Treg pathways and their associated cytokines including interleukin (IL)-4 and transforming growth factor-β (TGF-β) in the PD 7 pup brain, while down-regulating Th1 pathways and the inflammatory cytokine IL-1β 48,76 . Elevated IL-1β has been shown to impair spatial memory 77 , and knock-out studies have highlighted the beneficial and critical importance of IL-4 for the formation and retention of spatial memories 78,79 . Performance of spatial tasks leads to the accumulation of IL-4 producing Th2 cells in the meninges, and deficiency of IL-4 results in severely impaired performance of spatial memory tasks 78 . IL-4 stimulates astrocytes to produce brain-derived neurotrophic factor (BDNF) 78 , a key molecule for regulating cognitive processes, including LTP and neurogenesis 80,81 . Of note, in addition to up-regulating IL-4 expression 48,76 , maternal H. bakeri infection also up-regulated BDNF expression in the brains of PD 7 neonates (unpublished data). Therefore we hypothesize that the enhanced spatial memory in the pups of infected dams is associated with a regulatory Th2/Treg neuroimmune environment which promotes LTP and neurogenesis via the production of BDNF by astrocytes. Consistent with our hypothesis that a maternal helminth infection is capable of altering the neuroimmune environment of offspring, Williamson et al. 82 found that maternal infection with H. diminuta blunted the normal increase in hippocampal IL-1β mRNA response to LPS injection in PD 4 offspring. Similar to H. bakeri, H. diminuta infects the small intestine and induces a Th2/Treg immune response 83 . Further research is needed to determine whether the Th2/Treg bias is reflected in the neuroimmune environment of the uninfected pup.
We acknowledge four limitations. First, given our hypothesis that spatial memory may emerge earlier due to this maternal infection, we needed to test pre-weaned mice in the OLT, but some of them did not meet our inclusion criterion as they did not explore either object. This was expected as pups would likely have a high level of anxiety and fear due to being separated from their mothers for the test, leading to freezing events and a complete absence of exploration of the arena and objects. Although this lowered our sample size, we had sufficient pups that did explore to be able to detect significant differences between pups of infected and uninfected mothers. Second, we did not determine brain mass of the pups which may have been a more direct covariate for behavioural variables than body length. Third, despite the evidence for improved spatial memory, this maternal GI nematode www.nature.com/scientificreports/ infection may have negative (or positive) implications on other aspects of brain function and behaviour. Fourth, as our study was focused on the development of spatial cognition in young offspring, our findings cannot be extrapolated to adult mice. Future studies are needed to determine if this maternal GI nematode infection has positive long-term influences on brain development and behavior of the next generation.
To the best of our knowledge, this is the first study to assess the impact of a maternal GI nematode infection on the spatial behaviour of offspring, and to demonstrate enhanced spatial memory in pre-weaned and juvenile uninfected offspring. These findings shed light on a possible unappreciated benefit of maternal GI nematode infection and highlight a possible increase in fitness of the next generation. It would be important to determine if this behavioural impact persists as mice mature and how this maternal infection influences other aspects of offspring behaviour.

Methodology
Experimental design. We employed a 2 × 2 factorial design using H. bakeri infected versus uninfected dams, and their male versus female offspring.

Mice and parasites. Of the 19 primiparous 8-week-old timed pregnant (gestation day [GD] 4) outbred
CD-1 mice (Charles River Laboratories, Quebec, Canada), 16 were pregnant (84% pregnancy rate). Each dam and her litter was housed individually in a Nalgene cage (Fisher Scientific, Canada) at 21-23 °C, 40-60% relative humidity and a 12 h light and dark cycle. Mice had ad libitum access to a 2920X Teklad rodent diet (18% crude protein, 5% crude fat, 5% crude fiber). Within each of the seven staggered groups of dams received over 5 months, dams were randomized into uninfected and infected groups, and a total of eight dams per group were used for this study, providing an acceptable sample size based on a minimum of at least six dams per treatment condition 84 . Using standard H. bakeri protocols 85 , infective L 3 were obtained by fecal culture of stock parasites maintained in outbred CD-1 mice. Dams in the infected group were intubated using an oral gavage needle with 100 ± 3 L 3 suspended in 0.1 mL distilled water on GD 7, 12, 17, and PD 3, 8 and 13 (Fig. 5). Uninfected dams were intubated at the same frequency with 0.1 mL distilled water. Given that H. bakeri eggs released into the environment develop into infective larvae after 7 days, all cages were cleaned every 5 days to ensure offspring could not ingest infective larvae. Successful infection of dams was confirmed through faecal egg counts at weaning (PD 21), and worm counts 13-32 days after weaning (235.4 ± 45.4 worms/dam). Dams were then used for a separate experiment.
Pups were born on GD 19 or 20, litter size was recorded on PD 3, 8, 13, 15 and 21, and body mass and length from the top of the head to the base of the tail were recorded on PD 15 and 21. At PD 15, pups were sexed and given a unique identifier with a permanent marker. Pups were randomly selected to provide two male and two female pups per litter for the OFT/OLT and two male and two female pups per litter for the BMT (Fig. 5). At www.nature.com/scientificreports/ weaning, pups were separated by sex and 3-4 littermates were housed per cage. After the OFT/OLT or the BMT, experimental pups were necropsied and intestines were examined for larval and adult H. bakeri 85,86 . This confirmed that the pups had not been accidentally infected. Pups not used for this study were assigned to a separate experiment.
Compliance with guidelines for research with experimental animals. This study (protocol #2000-4601) was approved by the McGill University Animal Care Committee according to the guidelines of the Canadian Council on Animal Care. All methods were carried out in accordance with relevant guidelines and regulations, and the study was carried out in compliance with ARRIVE guidelines (https:// arriv eguid elines. org).
Experimental room and procedures. All spatial tests were conducted in a quiet room (340 cm × 260 cm) with a floor lamp in each corner that provided dim, even illumination to minimize stress of young pups during the OFT and OLT. During the BMT, a bright over-head light was added to provide a mild negative reinforcement. Trials were recorded using an overhead monochromatic video camera (Basler Ace monochrome) connected to a computer that was located in the back corner of the room behind a curtain. The experimenter remained behind the curtain during all recordings. Data was extracted from the videos using the Ethovision XT software (version 15). All equipment remained in the same location in the room, providing visual spatial cues.
To reduce handling anxiety, each pup in every litter was allowed to explore the palm of the experimenter for two minutes on PD 14 and 15, in their home room. Home cages were moved into the experimental room for 15-20 min acclimation prior to trials and all equipment was cleaned with 70% ethanol between trials.

Open field test (OFT) and object location test (OLT). The OFT/OLT arena (Maze engineers,
412 Wilmette Ave, Glenview, IL 60025, USA) (80 × 80 × 30 cm) had four opaque plexiglass compartments (40 × 40 × 30 cm) that allowed us to test the four pups per litter at the same time (Fig. 6a). An environmental cue (a large cross in colored tape) was placed on an inside wall of each compartment.
The OFT, conducted 24 h prior to the OLT, is an important component of the OLT protocol as it allows the pups to habituate to the novel arena. It also provides information on their exploratory behaviour 28 . On PD 16, pups were introduced to a designated compartment of the arena, and their activity was recorded for 10 min 53 . A preference to stay close to the walls of the field along with freezing behavior (not moving) indicates decreased spatial exploration and increased anxiety-like behavior 26 . For the purpose of data collection, the arena was conceptually partitioned into the peripheral zone (5.86 cm from each wall, totaling 50% of the surface area), and the center zone occupying the remaining area. Four descriptive displacement variables were measured to assess activity: total path traveled (cm), mean velocity (cm/s), time without movement (%), and time in center zone (%).
The OLT tested object location memory based on exploration of an object that had been moved to a novel location 27,28,88 between the 5 min training trial and the 5 min test trial. Pilot testing confirmed that our multicolored metal cylindrical aerosol cans (diameter: 4 cm, height: 15 cm) were suitable objects for the OLT because young pups did not fear them or climb on, sit on top of, or tip them over. The OLT was conducted on PD 17 using the same mice that had been habituated to the arena during the OFT on PD 16. Prior to the OLT training trial, two identical objects were positioned at designated locations within each compartment (Fig. 6a). Each pup was placed in the compartment as far as possible from both objects to avoid any position bias, and behaviour was video-recorded for 5 min after which pups were returned to their home cage. Prior to the test trial, one of the two objects (Object 2) in each compartment was moved to a novel location (Fig. 6a). A 3 h interval between training and test was selected given that 5 week old CD-1 mice have been shown to retain object location memories after 2 h but not after 4 h 88 , and we had hypothesized that spatial memory would be enhanced in response to maternal infection. Pups were returned to the same compartment for a 5 min test trial. Three descriptive displacement variables were measured during both the training and test trials: the total path traveled (cm), mean velocity (cm/s), and time without movement (%). In addition, one exploration variable was recorded for both Object 1 (stationary object) and Object 2 (mobile object): object investigation time (s) which measured how long a subject's nose was within a one-cm radius of the respective Object.
Pups were excluded from analysis of the OLT if they did not explore objects during either the training or test trial (6 males and 2 females from the uninfected group; 6 males and 8 females from the H. bakeri group).
Barnes maze test. The BMT procedure followed a protocol 31 that successfully tested spatial learning and short and long-term reference memory in CD-1 mice 88 . The Barnes Maze (Maze engineers, 412 Wilmette Ave, Glenview, IL 60025, USA) is an opaque circular platform (diameter: 92 cm, height: 70 cm) with 20 equally spaced holes (diameter: 5 cm) located 2 cm from the edge (Fig. 6b). In a brightly lit environment, mice naturally seek the dark enclosed area provided by the black goal box (20 × 10 × 4 cm) which was located under the same escape hole throughout all trials (Fig. 6b). From the surface of the maze, the escape hole, containing the goal box, looks identical to the other 19 holes. Mice learn the location of the goal box based on spatial cues.
The BMT was conducted on pups that had not been tested in the OFT/OLT. It involved a habitation phase of 5 min on PD 22 (Day 0), a training phase from PD 23-26 (Day 1-4), and probe trials 1 and 2 to test short-term and long-term spatial reference memory on PD 27 (Day 5) and 34 (Day 12) respectively. Training involved four 3 min trials per day for four training days. Each of the 16 training trials began by placing a pup in an opaque starting cylinder (diameter: 10.5 cm, height: 8 cm) at the center of the platform. After 10 s, the cylinder was removed, recording began, and the animal was allowed to freely explore the apparatus for 3 min. Once the animal entered the goal box, it was allowed to remain there for 60 s. Mice that failed to find the goal box within 3 min were gently guided to its location and placed inside. After each of the four 3 min training trials per day, mice Statistical analyses. Statistical analyses were performed in R statistical software 4.0.2 89 , and figures were produced using the package ggplot2 90 . Maternal treatment condition (H. bakeri infected versus uninfected) and offspring sex (male versus female) were always included as fixed factors. For comparisons over time, trial was included as a fixed factor. To account for pseudoreplication, dam was a random factor in all models, and the identity of the pup was also included as a random factor for comparisons over time where we had repeated measures on pups 91 . Non-significant interactions between fixed effects were excluded from models 92 . Pup length was included as a covariate in all models of behaviour data.
Linear mixed models (LMMs) or Generalized linear mixed models (GLMMs) were built using the lmer or glmer function, respectively (lme4 package 93 ), with significance assessed using the Anova function (car Subjects were assigned to one of four identical plexiglass compartments. Two identical objects (Object 1, stationary object and Object 2, mobile object) were added to each compartment after the Open Field Test for the object location training, and the mobile object was moved to a novel location prior to the test trial. Subjects that recognize that Object 2 was in a different location are expected to increase their investigation of this object during the test trial. (b) Barnes Maze with 20 equally spaced holes, one of which is the escape hole with a goal box beneath it during the training but not the two probe trials. During training, subjects learn the location of the escape hole relative to spatial cues which surround the maze. Short term spatial reference memory was assessed in probe trial 1, 1 day after training, and long-term spatial reference memory was assessed during probe trial 2, 1 week later. www.nature.com/scientificreports/ package 94 ). Where necessary, post hoc pairwise comparisons were performed using the emmeans function (emmeans package 95 ) with a Tukey correction. Normality, independence and homogeneity of variances of mixed models were assessed using fitted residuals from the plotresid function (RVAideMemoire package 96 ), and in the case of GLMMs, also using the DHARMa package 97 . Unless otherwise stated, values are presented as LSmeans ± SEM from the emmeans function. The significance level was set at 0.05. As no pup mortality occurred, the influence of the maternal infection status on litter size was analyzed on PD 21 using a linear model (lm function 89 ). LMMs were used to compare pup length and mass at PD 15 and PD 21 between experimental groups, with litter size as a covariate.
OFT/OLT: LMMs were used to assess object bias, displacement and exploration variables in the OFT and OLT. Two additional derivative variables were calculated from the OLT data to assess object location memory and both were analysed by LMM. The % investigation time of mobile object was used to compare the investigation of the mobile object relative to the total time spent investigating both objects in both the training and test trial, and calculated as: [Object 2 (mobile) investigation time (s) / [Object 1 (stationary) investigation time (s) + Object 2 (mobile) investigation time (s)]] *100. This variable ranged from 0% (only investigated Object 1) to 100% (only investigated Object 2). The change in % investigation time of mobile object allowed us to control results from the test trial with individual performance during the training trial, and was calculated as: % investigation time of mobile object during the test trial-% investigation time of mobile object during the training trial. A positive value indicated that the subject explored Object 2 more during the test trial compared to the training trial, indicating an increased investigation in the object after it had been moved.
BMT: Data in the BMT were positively skewed, and in some instances, heteroscedastic. The best distribution was assessed using the functions descdist and fitdist (package fitdistrplus 98 ) and comparing model residuals for best fit. In the training phase, we used LMM with log transformations for latency, path length and mean velocity. In the probe trials, we used Gamma GLMM, with log link function, for latency and path length. The number of errors was a discrete and overdispersed variable, and a negative binomial GLMM, with log link function, was used for both the training and probe trials.
In addition, a set of derivative variables reflecting change in latency, path length and errors between the two probe trials was calculated by subtracting probe trial 1 values from probe trial 2 values, allowing us to control for individual performance during probe trial 1. These derivative variables were normally distributed and homoscedastic, and LMMs were used without transformation.