Effects of acute pro-inflammatory stimulation and 25-hydroxycholesterol on hippocampal plasticity and learning involve NLRP3 inflammasome and cellular stress responses

Abstract

Neuroinflammation is an increasingly important target for therapeutics in neuropsychiatry and contributes to cognitive dysfunction, disability and death across a range of illnesses. We previously found that acute effects of pro-inflammatory stimulation with lipopolysaccharide (LPS) on hippocampal long-term potentiation (LTP), a form of synaptic plasticity involved in learning and memory, requires synthesis of the oxysterol, 25-hydroxycholesterol (25HC) and exogenous 25HC mimics effects of LPS. However, downstream mechanisms engaged by LPS and 25HC remain uncertain. Here we use rat hippocampal slices and in vivo behavioral studies to provide evidence that acute modulation of synaptic plasticity by both LPS and 25HC requires activation of the NLRP3 inflammasome, caspase-1 and interleukin-1 receptor. Furthermore, both LPS and 25HC engage cellular stress responses including synthesis of 5α-reduced neurosteroids and effects on plasticity are prevented by modulators of these responses. In studies of acute learning using a one-trial inhibitory avoidance task, inhibition of learning by LPS and 25HC are prevented by pre-treatment with an inhibitor of NLRP3. The present studies provide strong support for the role of 25HC as a mediator of pro-inflammatory stimulation on hippocampal synaptic plasticity and for the importance of NLRP3 inflammasome and caspase-1 activation in the deleterious effects of acute inflammation.

Introduction

Neuroinflammation contributes to the pathophysiology of multiple neuropsychiatric illnesses and has become a therapeutic target of considerable interest. In particular, pro-inflammatory signaling may contribute to changes in cognition, emotion and motivation across illnesses including infections, neurodegenerative disorders, epilepsy and primary psychiatric illnesses such as mood, anxiety, substance use and psychotic disorders^1–4. These diverse effects make it important to understand how pro-inflammatory signaling alters the function of brain networks and drives changes in neuronal excitability and synaptic function.

In recent studies, we have examined the role of microglia and microglial-derived messengers in mediating the effects of acute pro-inflammatory stimulation on hippocampal function. For these studies, we focused on effects of lipopolysaccharide (LPS), an endotoxin derived from the cell wall of gram-negative bacteria^5,6, on synaptic function and synaptic plasticity in the CA1 region of the rodent hippocampus based on the key role that this brain region plays in memory and emotional processing. We found that acute, brief (10–15 min) exposure of rodent hippocampal slices to 1–10 µg/ml concentrations of LPS has minimal effects on basal synaptic transmission but disrupts the ability to induce long-term potentiation (LTP), a synaptic mechanism that contributes to learning and memory, via activation of microglia, synthesis of the endogenous oxysterol 25-hydroxycholesterol (25HC) and induction of a form of N-methyl-D-aspartate receptor (NMDAR)-dependent metaplasticity⁷. Effects of LPS on LTP are mimicked by exogenous 25HC and are absent in mice deficient in its key synthetic enzyme, cholesterol 25-hydroxylase (Ch25H). Furthermore, LPS causes deficits in learning and memory that are eliminated in Ch25H knockout mice.

LPS stimulates pro-inflammatory signaling through both canonical (toll-like receptor 4 (TLR4)-dependent) and non-canonical (TLR4-independent) pathways^8–11, resulting in intracellular production and release of specific cytokines, including interleukin-1β (IL-1β), IL-6 and tumor necrosis factor-α (TNF-α) as well as Type I interferons^11,12. The acute effects we observed with µg/ml concentrations of LPS did not involve activation of TLR4, a canonical receptor for LPS, but rather appeared to result from non-canonical TLR4-independent intracellular signaling, based on the effects of selective TLR4 antagonists and an antagonist that also inhibits non-canonical signaling. Acute effects of high (µg/ml) LPS concentrations were mimicked by longer (2–4 h) exposures to low (10–100 ng/ml) concentrations of LPS that are TLR4-dependent and share mechanisms with high LPS^7,13.

In the present studies, we sought to understand how LPS and 25HC alter hippocampal synaptic function, focusing on the roles of the Nucleotide-binding oligomerization domain, Leucine-rich Repeat and Pyrin domain containing 3 (NLRP3) inflammasome, cytokine production, and cellular stress. LPS promotes activation of the NLRP3 inflammasome, which in turn stimulates caspase-1 and release of pro-inflammatory cytokines, including IL-1β. These pro-inflammatory signaling components can adversely affect cognition at least in part via dampening hippocampal function and synaptic plasticity^3,14–16. In the present studies, we found that acute inhibition of LTP in the CA1 hippocampal region by both LPS and 25HC involves the NLRP3 inflammasome, caspase-1 and interleukin-1 receptor, as well as activation of broader but non-stressor specific cellular stress responses including the integrated stress response and neurosteroid synthesis. Inhibition of the NLRP3 inflammasome in vivo also prevented learning defects in a one-trial inhibitory avoidance task.

Methods

Ethics statement

Sprague–Dawley albino rats were obtained from Harlan Laboratories (Indianapolis IN) and were housed under the care of the Washington University School of Medicine Division of Comparative Medicine. All animal use was performed in accordance with NIH and ARRIVE guidelines, and all experimental protocols were approved by the Washington University Institutional Animal Care and Use Committee (IACUC: protocols 22-0220, 22-0228 and 22-0344). Pain and suffering were minimized by use of anesthesia with isoflurane during hippocampal slice preparation and euthanasia at the end of behavioral studies in accordance with NIH and ARRIVE guidelines. Animals had unrestricted access to food and water. The reporting in this manuscript is in accordance with ARRIVE guidelines.

Hippocampal slice preparation

Hippocampal slices were prepared from postnatal day (P) 28–32 Harlan Sprague–Dawley male albino rats (Harlan Indianapolis IN) using published methods^17,18. For slice preparation, rats were anesthetized with isoflurane and dissected hippocampi were pinned on an agar base in ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM): 124 NaCl, 5 KCl, 2 MgSO₄, 2 CaCl₂, 1.25 NaH₂PO₄, 22 NaHCO₃, 10 glucose, gassed with 95% O₂-5% CO₂. The dorsal two-thirds of the hippocampus was cut into 500 µm slices using a rotary tissue slicer and maintained in ACSF at 30 °C for at least 1 h before experiments.

Hippocampal slice physiology

Single slices were transferred to a submersion-recording chamber and perfused with 30 °C ACSF at 2 ml/min. Extracellular recordings were obtained from the apical dendritic region of CA1 to monitor field excitatory postsynaptic potentials (fEPSPs). fEPSPs were evoked using a bipolar stimulating electrode once per minute with 0.1 ms constant current pulses to the Schaffer collateral (SC) pathway. Stimulus intensity was set at half-maximal based on control input–output (IO) curves. After establishing stable fEPSPs, a single 100 Hz × 1 s high frequency stimulus (HFS) at 50% maximal intensity was delivered to the SC pathway. IO curves were repeated 60 min following tetanic stimulation and were used as the primary measure of synaptic change in comparison to baseline. For display in figures, responses are typically shown at 5 min intervals. Drug concentrations used in these experiments were based on published reports and our experience with the agents and were selected for having no significant effects on baseline transmission¹⁹.

Behavioral studies

We tested memory acquisition in P28-31 rats using a one-trial inhibitory avoidance task that has been associated with CA1 hippocampal LTP^7,18,20,21. The testing apparatus has two chambers. One chamber is constantly lit and the other is kept dark. Both compartments have a floor of stainless-steel rods (4 mm diameter, spaced 10 mm apart) through which an electrical shock can be delivered in the dark chamber (12 × 20 × 16 cm). The lit compartment (30 × 20 × 16 cm) was illuminated with four 13 W lights at a light intensity of 1000 lx; light intensity in the dark chamber was < 10 lx. On the first day of study, rats were placed in the lit chamber and allowed to explore the apparatus for 5 min. They were then administered a single injection of MCC950 sodium (50 mg/kg or saline, ip). One hour later, rats were administered a single injection of either LPS (1 mg/kg or saline, ip) or 25HC (10 mg/kg or saline with 10% DMSO, ip). One hour after LPS, 25HC or vehicle, rats were placed in the lit chamber and allowed to explore for up to 300 s. When rats completely entered the dark chamber, they received an electrical shock and were immediately evacuated. After each 300 s session, rats were removed from the apparatus and returned to their home cages. On the next day, rats were placed in the lit chamber without any drug treatment and the latency to enter the dark compartment was recorded over a 300 s trial. The doses of drugs were determined in preliminary experiments and were based on lack of sedation over the time course of the behavioral studies.

Chemicals

Trans-ISRIB (CAS 1597403-47-8, Cat# 5284) and VX765 (CAS 273404-37-8, Cat# 7143) were from Tocris (Ellisville MO). MCC950 sodium (CAS 256373-96-3, Cat# S7809) and BAL-0028 (CAS 2842012-69-3, Cat# E1970) were purchased from Selleck Chemicals (Houston TX). Other chemicals, including 25HC (CAS 2140-46-7, Cat#H1015), LPS (Cat# L2630), IL1-Ra (Cat# SRP 3084), and salts were obtained from Millipore Sigma Chemical Company (St. Louis MO).

Data collection and analysis

Experiments were performed and analyzed using pClamp software (Molecular Devices, Union City CA). Results in the text are expressed as mean ± SEM and physiological results are based on analysis of IO curves obtained at baseline and 60 min following HFS. fEPSPs were normalized to baseline recordings (taken as 100%). Statistical comparisons were based on IO curves at baseline and 60 min following HFS based on changes in the maximal rising slope of fEPSPs evoked by 50% maximal stimuli, with p < 0.05 considered significant. A two-tailed unpaired Student’s t-test was used for most comparisons between groups. When appropriate, paired t-tests were used. For non-normally distributed data, the Mann–Whitney U-test was used for independent samples and the Wilcoxan signed-rank test was used for matched/dependent samples. Numbers reported in the text are the number (N) of animals studied in a condition. Statistics were performed using commercial software (SigmaStat, Systat Software, Inc., Richmond City, CA). Data in figures display continuous monitoring of responses at low frequency and thus may differ from numerical results described in the text, which are based on analysis of IO curves.

For behavioral studies, data were analyzed by two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test, using commercial software (GraphPad Prism 9.2.0, GraphPad Software, La Jolla California).

Results

Effects of LPS on hippocampal LTP involve NLRP3, caspase 1 and IL-1β

To probe mechanisms contributing to the effects of LPS on LTP induction, we used 15 min applications of 1 µg/ml LPS prior to delivery of HFS. This relatively brief application of LPS facilitates studies of pharmacological agents, avoiding potential complications with long pre-incubations (2–4 h) required for lower concentrations of LPS^7,13. Because we found similar effects of 1 and 10 µg/ml LPS in our prior study⁷, we used the lower concentration for these studies to avoid more complex and potentially off-target effects of the higher concentration. In control slices, a single HFS reliably induces robust LTP (fEPSP change: 142.1 ± 6.9% of baseline 60 min following HFS, N = 5; Fig. 1A). This LTP was completely inhibited by acute pro-inflammatory stimulation with LPS (99.5 ± 5.4% of baseline, N = 5, p = 0.0013 vs. control LTP, Fig. 1A). When administered alone in the absence of tetanic stimulation, LPS had no acute or persisting effect on CA1 synaptic responses (101.9 ± 4.3% of baseline 60 min following LPS, N = 3; Supplemental Fig. 1).

Fig. 1.

Effects of LPS on hippocampal plasticity involve the NLRP3 inflammasome. A. The graph shows control LTP in the CA1 region (white circles) and dense block of LTP induction by LPS (black circles). (A) single high frequency stimulus (HFS) was delivered at the time marked by the arrow. LPS was administered during the period denoted by the horizontal black bar. Traces to the right show raw fEPSPs at baseline (black dashed traces) and 60 min following HFS (solid red traces). (B) The NLRP3 inhibitor, MCC950, prevented the effects of LPS on LTP. (C) A second NPRP3 inhibitor, BAL-0028, also prevented the effects of LPS. (D) LTP inhibition by LPS was prevented by the caspase-1 inhibitor VX765. Calibration bar: 1 mV, 5 ms.

Because prior studies indicate that LPS can increase microglial expression and activation of the NLRP3 inflammasome as a mechanism to promote cytokine production²², we examined a role for NLRP3 using the antagonist MCC950^23,24. When administered for 30 min prior to and during 15 min LPS exposure, 0.1 µM MCC950 prevented LTP inhibition (157.6 ± 20.0% of baseline, N = 5, p = 0.0230 vs. LPS alone, Fig. 1B). A longer preincubation of MCC950 (1–2 h) also prevented the effects of LPS on LTP (145.3 ± 11.7%, N = 5, p = 0.0075 vs. LPS alone, Supplemental Fig. 2). We also examined BAL-0028, an agent that inhibits NLRP3 via a site that is distinct from MCC950^25–27. At 0.1 µM, BAL-0028, like MCC950, prevented the effects of LPS when administered for 15 min prior to and during LPS (142 ± 12.9%, N = 5, p = 0.0151 vs. LPS alone, Fig. 1C). Caspase-1 is known to contribute to pro-inflammatory responses downstream of NLRP3²⁸. At 1 µM, the cell-permeable and relatively selective inhibitor of caspase-1, VX765²⁹, completely prevented the effects of LPS (138.1 ± 10.3% of baseline, N = 5; p = 0.0106 vs. LPS alone, Fig. 1D).

Because NLRP3 inflammasome activation results in pro-inflammatory signaling, we next examined whether interleukin-1β (IL-1β), a cytokine stimulated by LPS²² and a product of NLRP3, modulates LTP induction based on prior studies showing effects of this mediator on synaptic plasticity^30–32. We found that 1 ng/ml IL-1β dampened LTP induction (108.8 ± 11.1%, N = 10, p = 0.03 vs. control LTP, Fig. 2A). These effects were blocked by the IL-1β receptor antagonist, IL1RA at 100 ng/ml (144.4 ± 14.0%, N = 5, p < 0.05, Fig. 2A)^33–35. Furthermore, IL1RA blocked the effects of LPS on LTP (171.0 ± 8.4%, N = 5, p = 0.0001 vs. LPS alone, Fig. 2B). As expected based on the upstream role of NLRP3, MCC950, had no effect on LTP inhibition by 1L1β (99.9 ± 6.9%, N = 5, p = 0.4336 vs. IL1β alone, Fig. 2C).

Fig. 2.

The cytokine, IL-1β, blocks LTP and the IL-1 receptor contributes to LTP inhibition by LPS. (A) The graph shows the block of LTP by exogenous IL-1β (black bar and black circles) and block of the IL-1β effect by an IL1 receptor antagonist (IL1RA; white bar and white circles). (B) The IL-1 receptor antagonist also prevents the effects of LPS on LTP. (C) The NLRP3 antagonist, MCC950 (black bar), does not prevent the effects of exogenous IL-1β even when applied for an extended duration (white bar). Traces to the right show representative fEPSPs as in Fig. 1.

Effects of LPS are prevented by inhibitors of cellular stress

We previously found that effects of LPS on LTP are inhibited by an NMDAR antagonist administered during LPS exposure⁷ and involve a form of NMDAR-dependent metaplasticity that activates cellular stress responses³⁶. Consistent with this, we observed that 1 µM ISRIB, an agent that inhibits the cellular integrated stress response (ISR)^37,38, prevented the adverse effects of LPS on LTP induction (161.9 ± 13.8%, N = 5, p = 0.0030 vs. LPS alone; Fig. 3A). Under conditions of cellular and synaptic stress, excitatory neurons also produce 5α-reduced neurosteroids, predominantly allopregnanolone (alloP) and, in turn, these neurosteroids play a role in LTP inhibition as a homeostatic protective mechanism^36,39. Consistent with this, we previously found that finasteride, an inhibitor of 5-alpha reductase (5AR) prevented the effects of LPS on LTP¹³. To confirm this result, we used a second and broader spectrum 5AR antagonist, dutasteride, and found that 1 µM dutasteride also completely prevented the effects of LPS (148.5 ± 14.2%, N = 5, p = 0.0121 vs. LPS alone; Fig. 3B).

Fig. 3.

Interfering with cellular stress counteracts LTP inhibition by LPS. (A) In the presence of ISRIB, an inhibitor of the integrated stress response (white bar), LPS (black bar) failed to inhibit CA1 LTP. (B) Dutasteride (white bar), an inhibitor of endogenous neurosteroid synthesis, prevents the effects of LPS on LTP. Traces show representative fEPSPs as in Fig. 1.

25-hydroxycholesterol mimics the effects of LPS on LTP

LPS-mediated LTP inhibition involves endogenous synthesis of the oxysterol, 25HC and exogenous 25HC mimics the effects of LPS on LTP⁷. These latter observations prompted us to examine whether NLRP3 activation and cytokine production are involved in the adverse synaptic effects of 25HC. Consistent with our prior report⁷, 1 µM 25HC administered for 15 min before and during HFS completely inhibited induction of LTP (101.9 ± 3.9% of baseline 60 min after HFS, N = 6; p = 0.0005 vs. control LTP, Fig. 4A). Complete LTP inhibition was also observed with 10 µM 25HC (102.2 ± 3.7%, p = 0.9499 vs. 1 µM 25HC, N = 5, Supplemental Fig. 3). When administered alone, 25HC had no effect on CA1 transmission (104.1 ± 2.3% of baseline 60 min following 25HC, N = 3; Supplemental Fig. 1). Similar to what we observed with LPS, we found that pre-treatment of slices with 0.1 µM MCC950 to inhibit the NLRP3 inflammasome prevented the effects of 1 µM 25HC on LTP (131.3 ± 3.2%, N = 5; p = 0.0003 vs. 25HC alone; Fig. 4A). Similarly, BAL-0028 prevented the effects of 25HC (134.8 ± 3.3% of baseline, N = 5; p = 0.0001 vs. 25HC alone; Fig. 4B). The caspase-1 inhibitor, VX765²⁹, also blocked the effects of 25HC on LTP (132.1 ± 5.7%, N = 5; p = 0.0015 vs. 25HC alone, Fig. 4C) as did the IL1 receptor antagonist (163.4 ± 10.7%, N = 5; p = 0.0003 vs. 25HC alone; Fig. 4D).

Fig. 4.

LTP inhibition by exogenous 25HC involves the NLRP3 inflammasome, caspase-1 and IL-1 receptor. (A) When administered alone, 1 µM 25HC (black bar) completely inhibits induction of CA1 LTP. This LTP inhibition is prevented by MCC950 (white bar). (B–D) Akin to LPS, LTP inhibition by 25HC is prevented by BAL-0028 (B), VX765 (C) and IL1RA (D). Traces depict representative fEPSPs as in Fig. 1.

We also examined the effects of cellular stress modulators on acute LTP inhibition by 25HC. Consistent with what we observed with LPS, the ISR inhibitor, ISRIB, completely prevented the effects of 25HC (141.8 ± 11.3%, N = 5; p = 0.0104 vs. 25HC; Fig. 5A). Similarly, the 5AR antagonist, dutasteride, which inhibits synthesis of 5α-reduced neurosteroids, also prevented acute LTP inhibition by 25HC (152.5 ± 11.2%, N = 5; p = 0.0027 vs 25HC alone; Fig. 5B).

Fig. 5.

LTP inhibition by exogenous 25HC engages cell stress mechanisms and synthesis of endogenous neurosteroids. (A) The ISR inhibitor, ISRIB (white bar), completely prevents the effects of 25HC (black bar) on LTP. (B) LTP inhibition by 25HC (black bar) is also prevented by inhibition of 5α-reductase with dutasteride (white bar). Traces show representative fEPSPs.

NLRP3 mediates effects of LPS and 25HC on learning and memory

To determine whether activation of NLRP3 inflammasome signaling by LPS contributes to in vivo effects on learning and memory, we examined whether MCC950 alters LPS induced changes in one-trial inhibitory avoidance learning, a form of learning dependent upon hippocampal LTP²⁰. Rats treated with 1 mg/kg LPS one hour prior to training show markedly impaired memory when tested 24 h later compared to control rats administered vehicle alone. Control rats remained in the lit chamber of the test apparatus for the entire 300 s testing period 24 h after receiving a shock in the dark chamber (N = 5). In contrast, animals treated with LPS (1 mg/kg ip) one hour prior to training remained in the lit chamber for only 86 ± 37 s (N = 8; p < 0.01 by Tukey’s multiple comparison test). This learning impairment was prevented by pretreatment with a single injection of MCC950 (50 mg/kg ip) two hours prior to LPS (time in lit chamber = 261 ± 33 s (N = 8); p < 0.01 vs. LPS, Fig. 6A). MCC950 had no effect on LPS-induced changes in body weight over the 24-h experimental period (weight change: -3.8 ± 2.6 g for LPS alone (N = 8) vs. -5.7 ± 2.3 g (N = 8) for MCC950 + LPS). The two LPS-treated groups showed a significant decrease in weight compared to control rats who gained 7.7 ± 0.8 g (N = 5) of weight over the 24-h period (Fig. 6A). Similar to LPS, 10 mg/kg 25HC impaired learning when administered one hour prior to training, and MCC950 prevented this memory impairment when administered two hours prior to 25HC (25HC alone: 115 ± 30 s (N = 8) vs. MCC950 + 25HC: 282 ± 17 s (N = 8); p < 0.001 by Tukey’s multiple comparison test, Fig. 6B). 25HC, either with or without MCC950 pretreatment, had no effect on body weight over the 24-h testing period (Fig. 6B).

Fig. 6.

LPS and 25HC inhibit one-trial inhibitory avoidance learning via activation of the NLRP3 inflammasome. (A) The upper bar graph shows that control rats exposed to a single foot shock 24 h prior to testing remain in the lit chamber of the test apparatus during a 5-min trial (gray bar). In contrast, animals pretreated with LPS prior to conditioning exit the lit chamber indicating a failure to learn the task (red bar). MCC950 prevents the effects of LPS on learning (purple bar). The bottom graph shows changes of body weight in controls and animals treated with LPS in the absence or presence of MCC950. (B) Akin to LPS, exogenous 25HC dampened learning in the one-trial inhibitory avoidance task (upper graph, blue bar). Effects of 25HC on learning were prevented by pretreatment with MCC950 (purple bar). The bottom graph shows that unlike LPS, 25HC had no effect on body weight compared to controls over the 24-h experimental period, either alone or in the presence of MCC950. *p = 0.0121; **p < 0.01; ***p = 0.001; ****p < 0.001.

Discussion

In prior studies we found that pro-inflammatory stimulation with the bacterial endotoxin, LPS, activates microglia and triggers production of the endogenous oxysterol, 25HC, which drives adverse effects on the induction of LTP and learning via a form of NMDAR-mediated metaplasticity⁷. Here we examined downstream inflammatory signaling activated by LPS and 25HC, and the role of this signaling in modulating hippocampal function. The present experiments indicate that LPS and 25HC engage similar mechanisms involving activation of the NLRP3 inflammasome, caspase-1, IL-1 receptors and cellular stress signaling to evoke changes in hippocampal function.

25HC is an important mediator and modulator of inflammatory signaling in the body and brain^40,41. In monocytes, 25HC is the primary oxysterol produced in response to pro-inflammatory stimulation and is the only oxysterol secreted from activated monocytes⁴². In brain, the key enzyme for 25HC synthesis, Ch25H, is primarily expressed in microglia^43,44 and is highly sensitive to activation by pro-inflammatory stimulation^44,45. 25HC, in turn, promotes and amplifies production of cytokines including IL-1β²². Our results indicate that exogenously administered 25HC shares mechanisms with LPS to promote pro-inflammatory signaling and previously we showed that elimination of Ch25H prevents adverse effects of LPS on LTP and memory^7,13. Furthermore, our results support the hypothesis that 25HC is an upstream mediator in the cascade of events leading to altered synaptic plasticity and learning based on the effects of inhibitors of NLRP3, caspase-1, cytokine activity and cell stress signaling. 25HC also has complex effects on the synthesis and cellular effects of cholesterol. 25HC inhibits cholesterol biosynthesis via effects on sterol regulatory binding protein (SREBP). It also promotes cholesterol efflux via activation of liver X receptor (LXR) signaling, while mobilizing and depleting accessible cholesterol at the cell membrane^46–48. Both glutamate receptor signaling⁴⁹ and induction of LTP⁵⁰ depend on membrane cholesterol, providing another means by which 25HC could alter hippocampal function.

Akin to other cellular stressors³⁵, the adverse effects of LPS on hippocampal plasticity involve NMDAR-dependent metaplasticity in which the stressors promote untimely activation of NMDARs to trigger cellular mechanisms that dampen the relative ease with which LTP can be induced. Consistent with a key role of 25HC in the effects of LPS, exogenous 25HC closely mimics the actions of LPS. This form of NMDAR-dependent metaplasticity has previously been observed with other stressors including ethanol^17,21,46, acetaldehyde, low glucose and brief hypoxia^36,51, among others. In the case of neuroinflammation, both LPS and 25HC acutely enhance NMDAR EPSPs, and effects of LPS on NMDARs are eliminated in the absence of 25HC⁷. The latter observations are consistent with 25HC’s ability to act as a partial agonist at a putative oxysterol site on NMDARs where another endogenous oxysterol, 24S-hydroxycholesterol (24S-HC) functions as a full agonist positive allosteric modulator^40,41. In turn, this untimely NMDAR stimulation activates cellular stress mechanisms including serine phosphatases, nitric oxide synthase (NOS) and p38 MAP kinase⁵². Similarly, stress signaling is activated by LPS and this signaling plays an important role in LTP inhibition³⁶. Here we provide evidence that the effects of LPS and 25HC are prevented by ISRIB, an agent that inhibits the cellular integrated stress response (ISR). The role of the ISR in the effects of LPS is not specific for pro-inflammatory stress and is involved in adverse effects of multiple stressors on synaptic plasticity^21,36–38. The current results are also consistent with findings in peripheral immune cells in which 25HC promotes activation of the NLRP3 inflammasome⁵³. Furthermore, 25HC activates the ISR and ER stress responses that negatively modulate synaptic plasticity^37,38. Here, we found that the ISR inhibitor completely prevents the effects of both LPS and 25HC on LTP inhibition. Figure 7 presents a modified scheme depicting events involved in metaplastic LTP inhibition by LPS¹³.

Fig. 7.

The diagram depicts events in the cascade leading to metaplastic LTP inhibition and learning impairment by acute pro-inflammatory stimulation with LPS. Steps highlighted in red indicate results from the present studies; steps identified in prior studies are in black. The process begins with activation of microglia and synthesis of 25HC, which promotes NLRP3 activation and pro-inflammatory signaling. Untimely release of glutamate activates neuronal NMDARs to promote pyramidal neuron stress and production of alloP, which enhance GABAergic inhibition and result in LTP inhibition and memory deficits. This diagram extends mechanisms initially outlined in prior publications^7,13.

Our previous work indicates that neuronal stressors also promote the endogenous synthesis of 5α-reduced neurosteroids, most notably alloP, and these steroids play a key role in acute LTP inhibition. We have found that untimely, low-level activation of NMDARs promotes neurosteroid synthesis primarily in hippocampal excitatory (pyramidal) neurons and that the ability of these neurosteroids to enhance GABA-A receptor-mediated inhibition is important in dampening LTP induction³⁶. In this context, LTP inhibition in response to cellular stressors appears to be a non-stressor-specific homeostatic protective mechanism that helps to sustain neuronal integrity and basal function in part by dampening higher-order and energy demanding synaptic plasticity. Here, we provide evidence that the oxysterol, 25HC, akin to LPS, also involves 5α-reduced neurosteroid generation. Once synthesized or exogenously administered, 25HC is known to rapidly mobilize accessible cholesterol⁴⁸ and stress-induced cholesterol mobilization promotes cholesterol trafficking to mitochondria for synthesis of pregnenolone, the first step in neurosteroidogenesis⁵⁴. Consistent with this, prior studies indicate that cholesterol redistribution within neurons is a sensitive indicator of NMDAR activation, including under conditions of cellular stress⁵⁵ and chemically-induced long-term plasticity⁵⁶. Steroidogenic cells typically store low basal levels of oxysterols and neurosteroids but use rapid synthesis to generate these modulators in response to demands⁵⁷. The present results are consistent with the idea that oxysterols such as 25HC act as upstream modulators to promote neurosteroid synthesis likely via cholesterol mobilization^48,58,59.

Results presented here and our other recent studies^7,13 strongly support a role for 25HC in pro-inflammatory signaling in hippocampus in response to LPS. This conclusion is consistent with prior studies in peripheral cells⁶⁰ and the CNS¹⁹, with 25HC promoting and amplifying inflammatory signals⁶¹. However, 25HC appears to play complex roles in inflammatory signaling, exhibiting anti-inflammatory effects under certain circumstances⁶². As a result, 25HC has been described as a “context-dependent” inflammatory modulator⁶² that helps to balance immune responses⁶⁰. In these complex roles, the nature of the inflammatory stimulus and concentration of oxysterol may be critical in determining observed effects^60,62,63. Similar considerations can be raised about alloP. On the one hand, this neurosteroid contributes endogenously to LTP inhibition under acute stressful conditions, including pro-inflammatory stimulation, at least in part via modulation of GABARs^7,33. Yet, other studies indicate that alloP can prevent the effects of pro-inflammatory stimulation depending on timing of administration and context^13,64,65. Mechanisms contributing to these complex effects of oxysterols and neurosteroids are not certain but likely include effects on ion channels (NMDARs and GABARs) as well as intracellular effects on stress-related pathways, including pregnane X receptors, liver X receptors, mitochondrial function and autophagy⁶⁶.

Our present results also provide insights into downstream mediators that underlie the effects of LPS and 25HC on hippocampal plasticity including the NLRP3 inflammasome, caspase-1 and IL-1Rs. Prior studies indicate that LPS and 25HC promote activation of NLRP3 and synthesis and release of several cytokines, including 1L-1β^19,60. In turn IL-1β is known to modulate hippocampal function and dampen induction of LTP^29,67. Here we provide further evidence that LPS stimulates cytokine production and that NLRP3, caspase-1 and IL-1R activation contribute to LTP inhibition by both LPS and 25HC.

Taken together with our prior studies^7,13, the present results support a critical role of the oxysterol, 25HC, as a mediator of the effects of pro-inflammatory stimulation on hippocampal function. In prior studies, we found that the effects of LPS on CA1 LTP are eliminated in mice deficient in Ch25H, the major enzyme responsible for 25HC synthesis from cholesterol^7,13. Here we show that LPS and 25HC activate NLRP3 signaling, caspase-1 and IL-1Rs to promote cellular stress and adverse effects on LTP induction. Our studies focused on acute effects of LPS and 25HC on hippocampal plasticity and learning but are limited in not providing direct biochemical evidence for changes in cytokine expression and key inflammatory and stress response proteins; these studies will be targets for future investigation.

The present observations indicate that 25HC is a key contributor to cognitive dysfunction accompanying neuroinflammation and a potential target for therapeutic interventions to preserve memory function under neuroinflammatory conditions. LPS is a strong pro-inflammatory stimulus and it remains to be determined whether other forms of inflammation, particularly those involving microglial activation, also engage 25HC as a mediator. Additionally, aseptic activation of inflammatory signaling may contribute to multiple neuropsychiatric illnesses and 25HC also may be a viable therapeutic target in those illnesses as a way to dampen sickness behavior and cognitive dysfunction. Alternatively, treatments that dampen cellular stress responses as a reaction to inflammation could also provide therapeutic avenues for preserving brain function.

Author contributions

YI, SM and CFZ conceived the experiments. YI and CFZ designed the studies and analyzed data. YI and KAO performed experiments. CFZ wrote the first draft and all authors edited and revised the manuscript.

Data availability

All data generated or analyzed during the study are included in the manuscript. Data are available from the corresponding author upon reasonable request.

Declarations

Competing interests

CFZ serves on the Scientific Advisory Board of Sage Therapeutics and has equity in Sage Therapeutics. Sage Therapeutics did not fund and was not involved in this research. Other authors have no conflicts to declare.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-90149-2.