The risk of developing neuropsychiatric conditions, such as anxiety and mood disorders is increased following viral infection, and the recent COVID-19 pandemic has highlighted this phenomenon; one in eight COVID-19 patients received a neuropsychiatric diagnosis within 6 months of infection (Taquet et al., 2021). Many aspects of the reported post-viral syndromes are consistent with the adaptive sickness behaviours that are exhibited by both humans and rodents in the aftermath of a bacterial or viral infection, which are associated with an alteration in mood, anxiety, and cognition (Dantzer et al., 2008). They are typically transient and designed to assist in the recovery process and are clearly evident in individuals with SARS-CoV-2 infection (Davis et al., 2021). Once an infection has been successfully cleared, most individuals and animals gradually return to their normal behaviours and levels of activity. However, the specific duration and persistence of sickness behaviours can vary depending on the circumstances and the nature of the infection or illness and we, as have others, have shown that changes in behaviour in rodents persist beyond the acute phase in a manner that might be relevant to long COVID (Couch et al., 2015).

To date, the activation of Toll-Like Receptor (TLR)-4 by the endotoxin lipopolysaccharide (LPS) has been a prominent focus in the study of sickness behaviour in rodents owing to its central role in gram-negative bacterial recognition, but other TLRs also have the capacity to induce behavioural changes when activated (Cunningham et al., 2007; Yates et al., 2022). RNA viruses, such as SARS-CoV-2, are detected by TLR-7, a pattern recognition receptor (PRR) that recognises single-stranded RNA (ssRNA) (Lund et al., 2004; Salvi et al., 2021). The inflammatory response to TLR-7 stimulation induces a type-1 interferon response (Lund et al., 2004) and is initiated by the transcription factor IRF7. Activation of TLR-7 triggers a MyD88-dependent cascade, involving interleukin-1 receptor-associated kinase 1 (IRAK-1), IRAK-4, and TNF receptor-associated factor 6 (TRAF6), which leads to NFkB and IRF7 activation (Cao et al., 1996; Deng et al., 2000; Suzuki et al., 2002; Kawai et al., 2004), which, on the face of it, ought to generate a similar cytokine and, downstream, metabolic response to TLR-4 activation that also activates a MyD88-dependent pathway (Hemmi et al., 2002). However, the differences in the underlying mechanisms are large enough to justify exploration of the consequence of a TLR7-mediated event.

A further, often overlooked issue is that most of the LPS-based rodent studies are performed in male mice. However, sex is known to have a profound effect on immunological outcomes (Klein and Flanagan, 2016). It is widely reported that the females generally respond more strongly to an inflammatory challenge, leading to overall better vaccine efficacies and faster clearance of pathogens with less severe resulting disease, for example (Cook, 2008). Concordantly, 80% of autoimmune disorders occur in females (Klein and Flanagan, 2016). These differences are largely attributed to the actions of sex steroids, the receptors of which are found on most leuocytes (Kadel and Kovats, 2018). However, germline and other factors must contribute to these outcomes as differences are found throughout different stages of life and can be resilient to gonadectomy (Smith-Bouvier et al., 2008). Despite the striking influence of sex on immunity, many studies fail to take this factor into consideration (Klein and Flanagan, 2016). This issue is especially important in the current study, as TLR-7 is an X-linked gene as is IRAK, which is also pertinent to this study (Fish, 2008). Differences in expression and the downstream responses are known to arise owing to incomplete X chromosome inactivation in females in leuocyte populations especially (Souyris et al., 2018). In certain contexts, this has been reported to result in differences in susceptibility to infections, autoimmune diseases, or responses to immunomodulatory therapies that target TLR-7 (Shen et al., 2010; Jenks et al., 2018; Spiering and de Vries, 2021).

Immunity and metabolism are inextricably linked, and the energetic demands following immune activation are heightened as a result of cell differentiation, proliferation, and the execution of effector functions (Jung et al., 2019). Shifts in central metabolism following peripheral immune activation have been proposed to contribute to the behavioural effects of LPS challenge and can be detected using 1H NMR metabolomics (Chan et al., 2020). The metabolome is defined as the collection of small molecules (<1.5 kDa) in a biofluid such as plasma or urine, or extracted from a tissue, such as liver or brain. Influenced by internal (e.g., genetics) and external (e.g., lifestyle) factors, it is the closest of the “omics” platforms to the phenotype and is therefore, not only useful in determining pathological mechanisms, but can also be manipulated to resolve any differences (Patti et al., 2012). In the current study, it is used to explore the behavioural outcomes of immunological challenge.

Here, we sought to discover how TLR-7 activation results in sickness-associated behaviours in comparison to TLR-4 activation. We included males and females in the study design to investigate a sex-specific relationship between inflammation, metabolism, and behaviour.

Here we sought to compare the acute phase response and changes to the metabolome downstream of systemic TLR-4 and TLR-7 activation in male and female mice. Analysis of the sickness behaviours induced by TLR challenges revealed that the female mice, compared to their male counterparts, were more resilient but cytokine production in the brain and liver in the females was either the same or elevated compared to the males. Given that the central proinflammatory cytokine response is viewed as a dose-dependent regulator of sickness behaviour this was an unexpected result. However, perhaps more striking, were the distinct metabolomic and cytokine responses to each challenge. TLR7-induced sickness behaviour was independent of increased central IL-1β expression. The metabolic shifts induced by immune activation were better correlated with the sex-dependent behavioural dimorphisms; increased levels of antioxidant potential in the female brain are likely to govern the intrinsic male/female metabolome differences. Among the distinct challenge-specific changes observed, a notable common feature of both TLR4 and TLR7 activation were marked changes in NAA in the PFC that were inversely correlated with the sickness behaviours. These observations will be discussed in turn below.

Female mice have been shown in a few cases to be more resilient, in terms of behaviour, to an LPS challenge (Millett et al., 2019). Previous studies of LPS challenge in CD1 mice have similarly found females to demonstrate more resilient behaviours in terms of recovery time (Meneses et al., 2018). These studies also identified more profound cytokine responses in females, further demonstrating an inconsistency in the classical paradigm of inflammation-induced negative affect (Cai et al., 2016; Murray et al., 2019). In the current study, we found that female resilience was accompanied by a more marked hepatic acute phase response (SAA2/3 and CXCL10) following the administration of LPS especially. The increase in basal TLR7 and IL-1β expression in the females is likely to be a consequence of the location of TLR-7 on the X-chromosome. Biallelic B lymphocytes from women display greater TLR7 transcriptional expression than the monoallelic cells, and this was correlated with higher TLR7 protein expression in female than in male leuocyte populations (Souyris et al., 2018). Thus, TLR7 escape from X inactivation, which endows the B cell compartment with added responsiveness to TLR7 ligands, might have been expected to generate greater responses in the female vs. the male mice. While this was true to some extent, it should be noted that hepatic IL-lβ expression was suppressed in the male and female mice in response to the TLR7 agonist, and, in the prefrontal cortex, despite the resilient behaviour exhibited by the female mice in response to TLR4 and TLR7 activation, the induction of each cytokine was comparable between sexes.

In addition to the observed sex-specific cytokine response, the type of TLR ligand gave rise to distinct cytokine profiles. Despite both TLR-4 and TLR-7 activation inducing sickness behaviour, the expression of prefrontal cortex (PFC) CXCL-1 and IL-1β was only elevated by the TLR-4 ligand, LPS. Conversely, PFC CXCL-10 was only altered by TLR-7 activation. Central IL-1β has long been recognised as a mediator of inflammation-associated behaviours, as direct injection of IL-1β into the brain generates sickness behaviour (reduced activity and exploratory behaviour, fever, and reduced food intake), while administration of the IL-1β receptor antagonist (IL- 1RA) reverses these behaviours (Kent et al., 1992). Thus, the induction of CNS IL-1β by LPS was not unexpected; we and many others have used a peripheral injection of LPS to induce central IL-1β expression over many years (Van Dam et al., 1992; Layd et al., 1994; Radford-Smith et al., 2023). Other studies have found a correlation between elevated levels of IL-1β in the blood or cerebrospinal fluid and the severity of sickness behaviour in animals (Dantzer, 2001). This further supports the idea that IL-1β is a key mediator of these behaviours. In the PFC mRNA encoding IL-1R1 and IL-1AcP has been reported to be present in pyramidal cells and interneurons and IL-1β was shown to induce excitatory/inhibitory imbalances at the network level (Mittli et al., 2023). In human studies, elevated levels of IL-1β have been observed in individuals with major depressive disorder and other mood disorders. These findings suggest a link between IL-1β and symptoms of depression, which are often part of sickness behaviour. Thus, it was a surprise that that TLR-7 activation, similar to TLR-4, induces stereotypical sickness behaviours but without the induction of PFC IL-1β. Together, these sex- and TLR-specific effects on behaviour suggest that the relationship between inflammation and behaviour is more complex and that other factors contribute to the behavioural effects of TLR challenge. Here CXCL10, rather than IL-1β, was strongly induced by R848, which we previously reported 6 h after an R848 challenge in male mice (Yates et al., 2022). Others have shown that at 24 h post-LPS injection, using a comparable dose of LPS, CCL2 and CXCL10 were elevated in the brain and CXCL10 expression in the brain was most positively and significantly correlated with sickness behaviour (Davis et al., 2017). More relevant to TLR7 activation, in mice challenged with vesicular stomatitis virus (VSV)-M2, CXCL10 from brain endothelia was shown to disrupt synaptic plasticity and generate virus-induced sickness behaviour (Blank et al., 2016). In their compelling paper, mice lacking CXCR3 or CXCL10 and subjected to IFN-β treatment or infection with the virus were protected from depressive-like behaviour. Thus, brain CXCL10, as for IL-1β, is clearly able to initiate sickness behaviours, but it remains unclear whether the action is directly on neurons or via an intermediate cell population such as microglia or astrocytes (Van Weering et al., 2011).

To explore potential unknown mediators of sickness behaviour, we turned to the metabolome as it is influenced by both intrinsic and extrinsic factors and is the closest of the omics to the phenotype. Surprisingly, TLR-4 and TLR-7 activation gave rise to distinct metabolic profiles in the liver, plasma, and brain at 24 h. Most notably, the decrease in plasma glucose following LPS administration was not observed following R848. Hypoglycaemia resulting from endotoxaemia is often recognised for its sickness-inducing properties (Chan et al., 2020), yet this shows this is not the cause of all post-infection related fatigue and mood alterations. Pertinently, following a more viral-like challenge by R848 compared to LPS, glucose remained unchanged and reveals the metabolic response is sensitive to the nature of the inflammatory challenge. Despite the accuracy of the OPLSDA models, indicating a clear separation between TLR-4 and TLR-7 groups, brain NAA increased consistently across all groups at 24 h post-injection and was the most discriminatory metabolite when comparing the TLR challenge (LPS and R848 groups combined) to controls. The role of NAA in the brain remains elusive; however, it is the precursor of the most abundant neuropeptide, NAAG, and may be involved in myelin synthesis and neuronal metabolism. It has been demonstrated to reach high concentrations in the brain following acute inflammatory insult and has been associated with white matter destruction in such cases (Moffett et al., 2008). Here, we show NAA is a non-specific marker of peripheral PRR activation, which associates with the sickness behaviour displayed by male mice.

Next, we sought to determine whether the brain metabolic response to TLR challenge depended on sex. After accounting for baseline sex differences, OPLS-DA analysis revealed the male “inflammatory” metabolome was distinct from that of the females. This shows that the males and females presented a divergent metabolic response to the challenge. While sex differences in immune responses have clearly been established in both rodents and humans, reports on sexual dimorphisms in immunometabolism are comparatively scarce. When testing the difference between the male and female brain metabolites after TLR challenge, we found comparatively lower brain glutathione (GSH) and respective raised cysteine (Cys) in the males. Depleted brain GSH has been used to model neuronal loss and cognitive decline in rodents (González-Fraguela et al., 2018), is associated with cognitive performance in humans and has been demonstrated, via magnetic resonance spectroscopy, to be higher in female than male brains (Mandal et al., 2012). The association between brain GSH and its precursor, Cys, with rearing and distance travelled in the OF could underly the sex-specific behavioural phenotypes in the current study. In humans, depression, anxiety, and cognitive decline have been associated with oxidative stress (Correia et al., 2023). Moreover, lower antioxidant potential in the brain following COVID-19 has been identified in those convalescing, illustrating that this may also be a phenomenon of post-viral mood disorders, and may explain the more resilient behaviours of the female mice in the current study (Al-Hakeim et al., 2023).

Males and females respond differently to an immune challenge. The current study demonstrates that the relationship between cytokines and behaviour is not dose-dependent but is sex-specific. This was demonstrated by the higher levels of cytokines induced by TLR-4 and TLR-7 activation in females relative to males, despite the more resilient behaviour demonstrated by females. Additionally, the cytokine and corresponding metabolic response is sensitive to the nature of the inflammatory challenge. Specifically, the LPS-induced IL-1β and glucose responses may not be extrapolated to other TLR challenges. The metabolic shift following TLR challenge was also dependent on the agonist, as well as sex. A striking commonality between the TLR challenges was the response of brain NAA, which was increased in both groups relative to saline controls and correlated strongly with the common sickness behaviours in the TLR male mice. Finally, most disparate changes in the brain metabolomes of male and female mice included a higher antioxidant response in females, which may underly behavioural resilience.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The animal study was approved by the LERP and ACER, University of Oxford. The study was conducted in accordance with the local legislation and institutional requirements.

ID: Data curation, Formal Analysis, Investigation, Methodology, Project administration, Writing – original draft. RM: Data curation, Project administration, Writing – review and editing. DR-S: Supervision, Writing – review and editing. WX: Data curation, Writing – review and editing. TP: Data curation, Writing – review and editing. FP: Supervision, Writing – review and editing. DA: Conceptualization, Supervision, Writing – review and editing.