IntroductionPsychiatric disorders arise from the complex interplay of genetic and environmental factors which shape neurobiological processes. Limited understanding of this multifactorial aetiology has long posed a barrier in the development of effective strategies for prevention, early intervention, and treatment. Epigenetic inheritance, a facet of gene-environment interactions, offers increasing insights into the risk of metabolic diseases [1,2,3], cancers [4,5,6], psychiatric disorders [7,8,9], and other diseases in humans. Supporting this notion, the parental environment has been shown to alter offspring physiology and behavior in rodent models; however, most studies have focused on the maternal line [9]. Research now highlights a parallel role for paternal epigenetic inheritance, with influences such as paternal toxins [10,11,12], stress [13,14,15,16], diet [17,18,19,20], exercise [21, 22], gut microbiome [23, 24], and immune activation [25,26,27,28,29] altering offspring phenotypes. This can occur via modifications to the sperm epigenome including deoxyribonucleic acid (DNA) methylation, histone modifications, and altered small and long non-coding RNA (ncRNA), which are delivered to the oocyte upon fertilization and exert effects on early development [30,31,32,33].Epigenetic modifications in the context of immune activation have gained special attention in recent years, driven by the Coronavirus Disease 2019 (COVID-19) pandemic and heightened awareness of the threat of future infectious disease outbreaks [34]. Increasing evidence suggests that paternal immune activation (PIA) prior to conception alters offspring phenotypes via epigenetic mechanisms [25,26,27,28,29, 35, 36]. Preconceptual infection with Toxoplasma gondii in mice has been shown to induce sex-dependent cognitive, anxiety-like, and depression-like behavioral changes in offspring (intergenerational effects) and grand offspring (transgenerational effects) [25]. More recently it has been demonstrated that paternal infection with Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) can alter offspring anxiety-like behavior and fasting responses, with limited impacts on grand offspring [28]. In these studies, offspring phenotypic changes have been attributed to epigenetic mechanisms, evidenced by alterations to paternal sperm small ncRNAs and the partial recapitulation of offspring phenotypes following microinjection of sperm RNA into fertilized oocytes. The small ncRNA profile of mature sperm is generally comprised of high proportions of transfer RNA (tRNA)-derived small RNAs (tsRNAs) and ribosomal RNA-derived small RNAs (rsRNAs) along with microRNA (miRNA) and P-element-induced wimpy testis (PIWI)-interacting RNA (piRNA) [37,38,39]. The composition of the small ncRNA population shifts dramatically during epididymal transit when epididymosomes deliver small RNA to the maturing sperm [38, 40, 41]. Notably, the RNA cargo of epididymosomes can be altered in response to paternal environment, offering a potential mechanism through which offspring phenotypes may be altered [42]. Although the exact mechanisms through which parasitic and viral infection alter sperm small ncRNA and subsequent offspring phenotypes are unclear, studies have suggested this phenomenon is not due to the infection itself, but rather a consequence of the immune response.Even in the absence of infection, PIA via pathogen mimetics can alter offspring phenotypes. Intraperitoneal (IP) administration of the viral mimetic polyinosinic:polycytidylic acid (poly I:C) in fathers prior to conception has been shown to alter sperm small ncRNA profiles and increase offspring depression-like behavior and alter offspring physiology including brain mass, immune response, and fasting responses [26]. Similarly, paternal pre-conceptual exposure to the bacterial mimetic lipopolysaccharide (LPS) can alter offspring behavior via alterations to sperm small ncRNA, however there are differential effects, with decreased anxiety, heightened sociability, and increased cognitive function [27]. In both pathogen infection mimetic models, limited impacts are observed in grand offspring, mirroring the effects of paternal SARS-CoV-2 [28]. These studies suggest that PIA without infection is sufficient to alter paternal sperm small ncRNA and offspring phenotypes, particularly anxiety-like and depression-like behaviors. As the directionality of effects differ with the source of PIA, the precise nature and dynamics of the immune challenge, and subsequent immune response, may determine offspring phenotypic changes.While poly I:C engages the toll-like receptor 3, retinoic acid inducible gene 1 protein, and melanoma differentiation-associated protein 5, LPS activates toll-like receptor 4. This leads to the activation of an overlapping but distinct combination of adaptor proteins, transcription factors, and pro-inflammatory genes [43,44,45]. This includes the upregulation of several key pro-inflammatory cytokines such as interleukin-6 (IL-6), interleukin-1β (IL-1β), and tumor necrosis factor-α (TNF-α also known as TNF) [43,44,45] which are produced by a variety of cell types, particularly macrophages, monocytes, and dendritic cells [46,47,48]. These cytokines play crucial roles in driving sickness responses such as fever, weight loss, and diminished food and water intake [49,50,51,52,53] which are shared features of the immune response against pathogens and their mimetics [44, 45]. Consequently, it is possible that similarities in cytokine profiles are responsible for overlapping offspring phenotypes in PIA models, while variations in cytokine profiles drive distinct offspring phenotypes.Cytokines also play important roles in male reproduction, modulating testicular function and spermatogenesis [54, 55]. Previous research has indicated that the blood-testis barrier (BTB) is disrupted during viral and bacterial infections, with cytokines being implicated in this process [56,57,58,59,60,61,62]. Although the role of cytokines has not yet been studied in the context of PIA, several studies have highlighted a role for cytokines in maternal immune activation (MIA), where female mice experience systemic inflammation via pathogenic or non-pathogenic means during pregnancy. In rodent models, administration of a single dose of cytokines in pregnant mothers alters offspring neurodevelopment and behavior [63,64,65,66,67,68]. For instance, when administered IL-6, offspring exhibit deficits in pre-pulse inhibition, latent inhibition, and socialisation [68]. Interestingly, coadministration of anti-IL-6 antibody alongside poly I:C can rescue these offspring phenotypes [68]. These findings suggest that IL-6 may be driving the effects of offspring phenotypic changes in MIA models, however it is unclear whether these changes occur via epigenetic mechanisms, placental mechanisms, or other indirect or direct effects on the developing fetal brain [69].Cytokines are a key component of the innate immune response and, given their involvement in male reproductive processes, BTB integrity and altering offspring health in MIA models, they represent a mechanism worthy of further exploration in the context of PIA. Elevated blood cytokine levels during PIA could alter the sperm epigenome and subsequent offspring phenotypes and, by elucidating the role of cytokines in this process, we can better understand the mechanisms driving paternal epigenetic inheritance in PIA models. We hypothesized that exogenous cytokine administration would elicit paternal sickness responses, modify the sperm epigenome, and alter offspring phenotype in a manner which recapitulates outcomes from other PIA models. Here, we show that paternal cytokine administration of IL-1β and TNF-α elicits a robust sickness response and is sufficient to alter sperm epigenetics and offspring phenotypes in a way which partially recapitulates PIA via SARS-CoV-2 infection and poly I:C immune activation. This study provides the first evidence of how paternal cytokines alter offspring phenotypes and builds understanding of how PIA during pre-conception is contributing to offspring health and risk of neuropsychiatric disorders.Materials and methodsAnimalsC57BL/6J mice were acquired from Ozgene Animal Research Centre (Ozgene ARC; Perth, Australia) between 6–8 weeks of age. Upon arrival, F0 male mice were housed in pairs while F0 female mice were housed in groups of 4–6 per box. F1 offspring were generated from F0 mice in-house, as described below. At weaning, F1 offspring litters were housed in groups of 4–6 per box, separated by sex and paternal treatment. All mice were kept in open-top housing with standard wood-chip bedding (Tapvei Estonia Oü; Harjumaa, Estonia), soft paper tissues, and red houses with ad libitum access to standard chow (Ridley; Melbourne, Australia) and water. Rooms were temperature and humidity-controlled (22 °C ± 2, 45%) on a 12 h light/dark cycle with lights on at 0800 h. All experiments were conducted according to the National Health and Medical Research Council (NHMRC) research guidelines and regulations and with approval from the Florey Institute of Neuroscience and Mental Health Animal Ethics Committee (#23-013-FINMH).Cytokine administrationRecombinant murine IL-1β (Cat. #211-11B-1MG), TNF-α (Cat. #315-01A-1MG), and IL-6 (Cat. #216-16-1MG) were purchased from PeproTech (New Jersey, United States) and prepared as per the manufacturer’s instructions. Cytokines were diluted in a sterile solution of phosphate-buffered saline (PBS) and 0.1% bovine serum albumin (BSA) (CAS #9048-46-8) from Sigma-Aldrich (Missouri, United States). Following a 2-week habituation period, 8–9-week-old C57BL/6J male mice received an intraperitoneal (IP) injection of key pro-inflammatory cytokines. When determining the contribution of individual cytokines to the sickness response, mice were administered cytokines in three separate cohorts: IL-1β (50 μg/kg, n = 8 per group) or TNF-α (250 μg/kg, n = 8 per group) alone; IL-6 (250 μg/kg, n = 8 per group) alone; or IL-6 (250 μg/kg) + IL-1β (0.5 μg/kg) in combination (n = 10 per group). For these initial studies, rectal body temperature was measured 2 h post-injection (PI) and every 24 h thereafter for 6 days. The addition of 0.1% BSA had no significant impacts on sickness responses (Supplementary Figure 1a–d) and was therefore used in all further studies. For F0 male mice, IL-1β (50 μg/kg) or vehicle, or TNF-α (250 μg/kg) or vehicle (n = 10 per group) was administered IP in two separate cohorts. In all experiments, food, water, and body weight were measured 24 h and 2 h before cytokine administration and every 24 h thereafter for 6 days. Locomotor activity was measured 4 h PI in a 15 min open field test. The arena consisted of a transparent Plexiglas box (ENV-510, 27 cm × 27 cm × 20.5 cm, Med Associates Inc.; Vermont, United States) and infrared beam arrays (ENV-256, Med Associates Inc.). Automated activity monitoring software was used to determine distance travelled. Mice were excluded from the analysis if they escaped the testing arena (n = 1 mouse excluded from the initial TNF-α cohort).BreedingPrior to a full spermatogenesis cycle [70], approximately 4 weeks after injection, cytokine-treated F0 male mice were mated with age-matched naïve F0 female mice. After a mating period of 5 days, male mice were sacrificed for tissue collection. Female mice were single-housed and red plastic enclosures removed to allow for unobstructed maternal care observations. Following the birth of a litter, on postnatal day (PND) 0, litter size was recorded. From PND 1–7 maternal care was recorded once every morning and afternoon as described below. Following the conclusion of maternal care evaluation, the sex of pups was determined between PND 8–10. F1 mice were weaned between PND 24–27, and litters were mixed using random number generators, separated by sex and paternal treatment. F1 mice were group housed with 4–6 per box.Maternal careMaternal care was scored daily from PND 1–7 over two sessions per day (0830 – 1000 and 1630 – 1800) for each litter in the IL-1β control (n = 4), IL-1β treatment (n = 6), TNF-α control (n = 9), and TNF-α treatment (n = 9). Behavior was observed via instantaneous scan sampling to assess the quality and quantity of maternal care as previously described [23, 26]. In each session, observations were made 30 times per litter, approximately once every 90 s. The presence and absence of self-maintenance (self-grooming, eating/drinking), nurturing (nest building, arch-backed nursing, passive nursing, grooming/licking pups) and neglectful (climbing, digging, pup outside nest) behavior were recorded. Observations were conducted by a single scorer blinded to group assignment. The relative frequency of these behaviors was calculated by averaging the proportion of each behavior across the two sessions.BehaviorF1 male control (IL-1β: n = 13, TNF-α: n = 14), male treatment (IL-1β: n = 13, TNF-α: n = 14), female control (IL-1β: n = 13, TNF-α: n = 14), and female treatment (IL-1β: n = 14, TNF-α: n = 15) mice began behavioral testing upon reaching adulthood, at 8 weeks of age. Behavioral tests were carried out from the least to most stressful in the following order: large open field (LOF); elevated plus maze (EPM); novelty-supressed feeding test (NSFT); and Porsolt swim test (PST). The saccharin preference test (SPT) was conducted last due to the requirement of individual housing. Mice were given at least 24 h of rest between individual behavioral tests with a 48 h resting period for more stressful tests such as NSFT and PST. All testing was carried out between 0800 and 1700 under 30–50 lx unless specified otherwise. Testing for male and female mice were conducted on different days where possible, with experimenters blinded to experimental groups. Before beginning any behavioral test, mice were acclimatized to the room for at least 1 h, unless specified otherwise. Between trials, arenas were cleaned with 20% ethanol for all behavioral tests.Large open fieldThe LOF was performed to evaluate anxiety-like behavior and locomotion as previously described [71]. Mice were placed in the corner of the arena (opaque PVC box, 100 cm x 100 cm x 20 cm), which was illuminated at 1000 lx (using two flood lights placed on the corners of the arena) in the centre, and allowed to explore for a 10 min period. For analysis, the arena was divided into a centre zone (50% of total arena) and outer zone (50% of total arena). The total distance travelled over 10 min, latency to centre, and the percentage of total time spent in the centre zone were automatically calculated using an overhead camera and TopScan Lite Software (CleverSys Inc.; Virginia, United States).Elevated-plus mazeThe EPM was used to evaluate anxiety-like behavior as described previously [26]. Mice were placed in an arena elevated approximately 40 cm off the ground. The arena consisted of two open arms (5 cm x 30 cm), two closed arms (5 cm x 30 cm x 15 cm opaque walls), and a central zone (5 cm x 5 cm) which intersect at 90 degrees to form a plus sign. Mice were placed in the centre zone facing the open arms and allowed to explore all arms for 5 min. Total distance travelled, and time spent (%) in the open and closed arms was automatically calculated using an overhead camera and TopScan Lite Software (CleverSys Inc.).Social interaction testThe SIT was used to assess social interaction abilities as previously described [26]. Sociability was assessed over three trials in a transparent Plexiglas box (San Diego Instruments, Inc.; California, United States; 40 cm x 37.5 cm x 10 cm) divided into 3-chambers consisting of a central chamber connected to two outer chambers (16 cm x 37.5 cm x 10 cm), each with a metal cage at the end. In the first trial, the test mouse was placed in the central chamber and habituated to the arena for 5 min. The mouse was then removed from the arena. In the second trial a “guest” mouse, sex and age matched, was placed in a cage in either the left or right chamber with the test mouse being placed in the central chamber. The test mouse was allowed to explore for 10 min. Between each trial, the arena was cleaned with 20% ethanol. The experimental mouse was tracked automatically using an overhead camera and TopScan Lite Software (CleverSys Inc.).Novelty-suppressed feeding testThe NSFT was used to assess the anxiety-like behavior hyponeophagia as previously described [26]. Mice were fasted for 24 h with water available ad libitum. Mice were weighed pre- and post-fast to determine bodyweight loss after 24 h. During testing, mice were placed in a brightly lit (750 lx) novel arena (60 cm x 60 cm x 60 cm) filled with 3 cm of bedding and a standard chow pellet in the centre affixed to a circular piece of filter paper. The latency to take the first bite from a chow pellet was recorded, with the trial ending after 10 min. Following testing, the mouse was placed alone in a new cage containing a pre-weighed food pellet and allowed to eat for 5 min. The amount of food consumed was recorded. Bedding was changed on each testing day and between sexes, with fresh bedding mixed with bedding from each cage to be tested.Porsolt swim testThe PST was used to evaluate stress-coping behavior [72, 73] as previously described [23]. Mice were individually placed in a 2000 mL glass beaker partially filled with approximately 1400 mL water at 24 ± 1 °C. Mice were recorded for 6 min using digital cameras. Mice were dried and warmed with towels and heat lamps prior to being returned to a holding cage or home cage. The video footage was analyzed manually to determine latency to first immobility episode by a blinded experimenter, and the final 4 min of the test were used to automatically determine total time immobile via the DepressionScan software suite (CleverSys Inc.).Saccharin preference testThe SPT was used as a measure of anhedonia, a depression-like behavior, and was conducted as previously described [23]. Mice were single housed with ad libitum access to food and water. In the first 24 h, mice were given water in two identical 50 mL falcon tubes with sippers to allow for habituation to the two-bottle setup. For the first testing night, mice were given one tube containing approximately 40 mL of 0.1% saccharin solution (#S1002, Sigma-Aldrich) and the other containing 40 mL of standard drinking water with placement on the left or right side being randomized. The placement of the two bottles was reversed for the second night of testing to control for side preferences. The weight of solution consumed was recorded on the habituation day and the first and second testing nights by a blinded experimenter. Saccharin preference index (%) was determined by calculating the amount of saccharin consumed in comparison to total fluid consumed over a 24 h period.Tissue collectionF1 male and female mice were administered a lethal dose (80 mg/kg) of pentobarbitone sodium (Lethabarb) (Virbac; NSW, Australia) via IP injection followed by cervical dislocation. For collection of fresh whole brains, the head was decapitated and the scalp removed to expose the skull. Small dissection scissors were used to make a midsagittal incision in the skull, taking caution to avoid pressure on the brain. Forceps were used to remove pieces of the skull. After inverting the head, the trigeminal and optic nerves were severed using a spatula, allowing the brain to be released and extracted. Whole brain mass was recorded prior to dissections of the prefrontal cortex, striatum, hippocampus, and hypothalamus. This whole brain collection was carried out on a subset of F1 offspring with experimenters blinded to group condition.Epididymal sperm collectionF0 male mice were administered a lethal dose (80 mg/kg) of pentobarbitone sodium (Virbac) via IP administration followed by cervical dislocation. Caudae epididymides and vasa deferentia were separated and cleaned from the epididymal fat pad as previously described [26]. Caudae epididymides contain mature sperm which are responsible for fertilizing the ova and therefore are more relevant to any alterations in offspring phenotype. Consequently, caudae epididymides were bisected and immersed in tubes containing 1 mL of Modified Tyrode’s Medium 6 (MT6) [26] (solution contains (in mM): 124 NaCl, 2.68 KCl, 17.14 CaCl2, 3.2 NaH2PO4·H2O, 0.49 MgCl2·6H2O, 25 NaHCO3, 5.6 D-glucose, 4 mg/mL BSA, 28.2 Phenol Red) kept at 37 °C in and saturated with CO2 overnight. Tubes containing the epididymides in MT6 were incubated in a heating block at 37 °C for a minimum of 30 min. Tissues were removed and sperm was pelleted by centrifugation at 400 g for 10 min. Supernatant was removed, and sperm pellets were stored at −80 °C.Small RNA extractionPrior to total RNA extraction, all samples underwent somatic cell lysis to remove contamination of samples by somatic cells, which may skew sequencing results. 1 mL of somatic cell lysis buffer [41] (solution contains 0.01% sodium dodecyl sulfate, 0.005% Triton X-100 in nuclease-free water) was added to each frozen sample and homogenized. Samples were incubated on ice for 15 min followed by centrifugation at 3000 g for 4 min at 4 °C. The supernatant was removed and washed twice with PBS. Sperm lysis was carried out in QIAzol (Cat. #79306, QIAGEN; Hilden, Germany). Total RNA was extracted and purified from sperm samples as previously described [26]. The QIAGEN miRNeasy Mini Kit (Cat. #217004) and RNase-Free DNase Set (Cat. #79254) was used according to the manufacturer’s instructions to extract total RNA and carry out DNase treatment respectively. The RWT buffer was prepared with isopropanol. For the IL-1β cohort, each condition contained 4 biological replicates and for the TNF-α cohort there were 5 biological replicates, each containing sperm RNA pooled from 2 male mice to ensure adequate RNA concentration for sequencing (following the addition of ethanol for precipitation). The final elution step was repeated three times with 20 μl of RNase-free water. RNA was quantified using the Qubit RNA Broad Range Assay Kit (Cat. # Q10210; Thermo Fisher Scientific; Massachusetts, United States) in a Qubit 4 Fluorometer (Thermo Fisher Scientific). Quality control was carried out by the 4200 TapeStation System from Agilent (California, United States) using the RNA ScreenTape and Buffer (Cat. #5067-5576, 5067-5577) and RNA ScreenTape Ladder (Cat. #5067-5578) to determine RNA quality and potential somatic cell contamination.Small RNA sequencingThe standard NEXTflex Small RNA v4 library was prepared with bead size selection for Illumina Next-Generation Sequencing using the NovaSeq X 10B lane, 300-cycle at the Australian Genome Research Facility. All samples were processed on a single flow cell lane with 150-bp-paired-end reads. An average of 24 100 840 reads per biological replicate were obtained for the IL-1β cohort and 24 602 294 for the TNF-α cohort.Small ncRNA annotation and analysesSmall RNA sequence annotation was carried out using SPORTS v1.1.2, a non-coding RNA annotation and analysis software optimized for rsRNA- and tRNA-derived small RNAs [74]. Adapter removal and trimming (15–45 bp inclusive) was conducted in SPORTS v1.1.2. Read quality was assessed using FastQC [75] to verify clean reads. Following quality control, there were 14 960 005 and 14 282 763 clean reads per biological replicate across the IL-1β and TNF-α cohorts, respectively. SPORTS v1.1.2 was used to map reads to the following databases sequentially: (1) miRbase v22.1 for miRNA; (2) National Center for Biotechnology Information (NCBI) for ribosomal RNA (3) GtRNAdb for tRNA; (4) ensemble and Rfam for other ncRNA; and (6) piRBase and piRNABank for piRNA. SPORTS v1.1.2 default databases were used except for miRbase v22.1, where the database was compiled from mature.fa files pre-processed according to SPORTS v1.1.2 documentation. Approximately 91.39% of clean reads in the IL-1β cohort and 93.11% in the TNF-α cohort to aligned to these databases and/or mouse reference genome (GRCm38). Output was used to calculate reads per million and restructured into a matrix formatted for differential expression analysis in DEseq2 [76]. Here, piRNAs were removed due to inclusion in a separate pipeline for piRNA cluster analysis. In DEseq2 the false discovery rate (FDR) was set to 5% using the Benjamini and Hochberg method with independent filtering in DEseq2 removing low count genes. Graphs were generated using ggplot2 [77] package (v3.5.1), and heatmaps created using pheatmap [78] package (v1.0.12). Gene ontology (GO) analysis of differentially expressed tsRNAs was carried out by taking the seed sequence (positions 2–8 inclusive of the 5’ end of the molecule) and searching in TargetScan [79] (https://www.targetscan.org/vert_50/seedmatch.html) to identify messenger RNA (mRNA) targets. Target gene names were entered into the Gene Expression Database [80] (https://www.informatics.jax.org/gxd/batchSearch) from Mouse Genome Informatics to filter for genes that are only expressed during early embryonic development up until day 5 (Theiler Stage 1–7 inclusive). Following filtering, 1 141 genes remained from the initial 1 236 target genes. Functional annotation was carried out using Enrichr to identify cellular, molecular, biological, and knock-out phenotype ontology. Knock-out phenotypes are determined from data collected by the Knockout Mouse Phenotyping Program (KOMP2) which draws on the results from 120 peer-reviewed publications to characterize the function of protein coding genes.For piRNA cluster analysis, reads were collapsed into unique sequences to remove redundancies and allow for efficient mapping. The bowtie 2 [81] alignment tool was used to map collapsed reads to the mouse reference genome [82] (GRCm38, GENCODE https://www.gencodegenes.org/mouse/release_M25.htmL). On average 75.57% of reads per sample were successfully mapped in the IL-1β cohort and 76.79% of reads per sample in the TNF-α cohort. The SAM files generated by bowtie 2 were used in proTRAC (v2.4.3), a piRNA cluster prediction tool. Default settings were used except for 1TorA which was set to 0. This parameter normally adjusts for 1U and 10 A biases caused during the ping-pong cycle, however male gametes are mostly comprised of RapiRNAs, which rarely undergo ping-pong amplification mechanisms during spermiogenesis [83]. piRNA length was set to 21–33 [83]. From the generated list of piRNA clusters, the intersect gene annotation file (gencode.vM25.annotation, GENCODE) was used for identification of nearby genes. Bedtools [84] was used to generate a count matrix of the merged piRNA clusters which was later used for differential expression analysis. For differential expression analysis, read counts corresponding to each predicted piRNA cluster was generated. The DEseq2 package was used to identify statistically significant differentially expressed piRNA clusters. The false discovery rate (FDR) was set to 5% using the Benjamini and Hochberg method with independent filtering in DEseq2 removing low count genes. Circlize [85] was used to visualise piRNA cluster expression across samples.Statistical analysesStatistical analyses and graphing were carried out in R (v4.4.2) using R Studio (v2023.06.01 build 764) and GraphPad Prism (v10.4.2) with data presented as mean ± standard error of the mean (SEM). For experiments investigating the contribution of cytokines to the sickness response, ANOVAs were used followed by Dunnet’s, Šídák’s, or Tukey’s test for multiple comparisons where applicable. For analysis of F1 behavioral data, linear mixed models (LMM) were used with paternal treatment and sex as fixed factors and litter as a random factor due to possible reductions in phenotypic variability from offspring being sired from the same father and mother. For repeated measures analysis, offspring ID was nested within litter. Data and residuals were visually assessed for linearity, normality, and homogeneity of variance. The Shapiro-Wilk test was used to assess for normal distribution and a general linear mixed model (GLMM) with gamma distribution used for latency data which deviated from the assumptions. Regression models were assessed using the emmeans package and the ‘joint_tests’ function. If paternal treatment or sex had a higher relative contribution, these contrasts were tested using the emmeans ‘pairs’ function. If the interaction term had a relatively equal or greater contribution than paternal treatment or sex, the ‘pairs’ function was used to determine the nature of this interaction with Tukey’s test where applicable. Sample sizes were chosen based on previous behavioral neuroscience research from our laboratory aiming to detect moderate to large effect sizes [23, 26,27,28]. Full statistics can be found in Supplement 1. The significance level used was α = 0.05.ResultsPaternal IL-1β and TNF-α administration in male mice induces a robust sickness responseTo isolate which cytokines contribute to the shared sickness response observed across pathogenic and pathogen mimetic models of PIA, several pro-inflammatory cytokines were exogenously administered. Male mice received a single IP injection of either IL-1β, TNF-α, IL-6, IL-6 + IL-1β (combined cytokines), or vehicle solution, to determine whether they elicited a sickness response characterized by decreased locomotor activity, bodyweight loss, reduced food and water intake, or fever. Cytokine doses were chosen based on previous literature where sickness responses were reported at 50 μg/kg for IL-1β [52, 86], 250 μg/kg for TNF-α [52, 87] and 250 μg/kg for IL-6 [68]. There was a significant treatment effect on locomotor activity (P = 0.0428) with post hoc analysis revealing that only IL-1β administration led to a statistically significant decrease in distance travelled when compared to controls (IL-1β: Padj = 0.0350; TNF-α: Padj = 0.0942, Fig. 1a). Baseline bodyweights did not differ across the groups (P = 0.407 Supplementary Figure 1e) and as expected, there was a statistically significant difference in bodyweight 24 h PI (P = 0.0003) where IL-1β and TNF-α administration resulted in greater bodyweight loss when compared to controls (IL-1β: Padj = 0.0037; TNF-α: Padj = 0.0002, Fig. 1b). The 24 h timepoint reflected the most severe drop in bodyweight with a gradual return to baseline, which was reached at 72 h (Supplementary Figure 1f). Food consumption was also altered across groups (P = 0.0003) and post hoc analysis revealed that both IL-1β- and TNF-α-treated mice consumed less food in the 24 h PI (IL-1β: Padj = 0.0005, TNF-α: Padj = 0.0004, Fig. 1c). Food intake returned to baseline at 48 h PI (Supplementary Figure 1g). At 24 h PI, no statistically significant effects were observed in water intake (P = 0.6227, Fig. 1d) or across the 144 h PI (P = 0.4359, Supplementary Figure 1h). Similarly, there were no differences between control and treatment groups for rectal body temperature at 2 h PI (P = 0.6679, Supplementary Figure 1i) or across the 144 h PI (P = 0.7236, Supplementary Figure 1j).Fig. 1: Sickness responses following a single dose of cytokines.The alternative text for this image may have been generated using AI.Full size imageMale mice received a single intraperitoneal (IP) injection of either IL-1β, TNF-α, IL-6, a combination of IL-6 + IL-1β, or vehicle, at 8-9 weeks of age. Sickness responses were measured. (a) Locomotor activity 4 h post injection (PI) in control, TNF-α, and IL-1β groups. (b) Bodyweight loss 24 h PI in control, TNF-α, and IL-1β groups. (c) Food intake 24 h PI in control, TNF-α, and IL-1β groups. (d) Water intake 24 h PI in control, TNF-α, and IL-1β groups. (e) Locomotor activity 4 h PI in control and IL-6 groups (f) Bodyweight 24 h PI in control and IL-6 groups. (g) Food intake 24 h PI in control and IL-6 groups. (h) Water intake 24 h PI in control and IL-6 groups. (i) Locomotor activity 4 h PI in control and IL-6 + IL-1β groups. (j) Bodyweight 24 h PI in control and IL-6 + IL-1β groups. (k) Food intake 24 h PI in control and IL-6 + IL-1β groups. (l) Water intake 24 h PI in control and IL-6 + IL-1β groups. Individual values and means are presented with error bars representing ± SEM (males n = 6–10 per group), *p