To examine the immune effects of chronic exposure to diets enriched in SFAs on lipopolysaccharide (LPS)-induced endotoxemia, we fed age matched (6–8 weeks), female BALB/c mice either with a WD (enriched in SFAs and sucrose), a KD (enriched in SFAs and low-carbohydrate), or standard chow (SC; low in SFAs and sucrose), for 2 weeks (Supplementary file 1). We defined 2 weeks of feeding as chronic exposure, because this is correlated with WD- or KD-dependent microbiome changes and confers metaflammation in WD mice (Napier et al., 2019), sustained altered blood glucose levels in WD mice (Figure 1—figure supplement 1A), and elevated levels of ketones in the urine and blood in KD mice (Figure 1—figure supplement 1B-C). We then induced endotoxemia by a single intraperitoneal (i.p.) injection of LPS. We measured hypothermia as a measure of disease severity and survival to determine outcome (Napier et al., 2019; Napier et al., 2016; Saito et al., 2003). WD- and KD-fed mice showed significant and prolonged hypothermia, starting at 10 hr post-injection (p.i.), compared to the SC-fed mice (Figure 1A). In accordance with these findings, WD- and KD-fed mice displayed 100% mortality by 26 hr p.i. compared to 100% survival of SC-fed mice (Figure 1B). Hypoglycemia is a known driver of endotoxemia, and each of these diets has varying levels of sugars and carbohydrates (Supplementary file 1; Raetzsch et al., 2009; Filkins and Cornell, 1974). However, mice in all diet groups displayed similar levels of LPS-induced hypoglycemia during disease (Figure 1—figure supplement 1D), indicating that potential effects of diet on blood glucose were not a driver of enhanced disease severity.

Considering mice fed KD experience a shift toward nutritional ketosis, we wanted to understand if our phenotype was dependent on nutritional ketosis. 1,3-butanediol (BD) is a compound that induces ketosis by enhancing levels of the ketone β-hydroxybutyrate in the blood (Goldberg et al., 2019). Age matched (6–8 weeks), female BALB/c mice were fed for 2 weeks with KD, SC supplemented with saccharine and 1,3-butanediol (SC + BD), or SC-fed with the saccharine vehicle solution (SC + Veh). BD supplementation was sufficient to increase blood ketones (Figure 1—figure supplement 1C). We next injected LPS i.p. and found KD-fed mice showed significantly greater hypothermia, and increased mortality, compared to SC + BD and SC + Veh (Figure 1—figure supplement 1E, F). Though short-lived, when compared to SC + Veh, the SC + BD mice did confer an increase in hypothermia, suggesting that nutritional ketosis may play a minor role in KD-dependent susceptibility to endotoxemia (Figure 1—figure supplement 1E, F). Together these data suggest that diets enriched in SFAs promote enhanced acute endotoxemia severity, and this is independent of diet-dependent hypoglycemic shock or nutritional ketosis.

Endotoxemia mortality results exclusively from a systemic inflammatory response characterized by an acute increase in circulating inflammatory cytokine levels (e.g. TNF, IL-6, and IL-1β) from splenocytes and myeloid-derived innate immune cells (Lewis et al., 2016; Wang et al., 2016; Radzyukevich et al., 2021; Zhang et al., 2012). Additionally, pre-treatment of myeloid-derived cells with dietary SFAs has been shown to enhance inflammatory pathways in response to microbial ligands (Schwartz et al., 2010; Fang et al., 2022). Considering this, we hypothesized that exposure to enriched systemic dietary SFAs in WD- and KD-fed mice would enhance the inflammatory response to systemic LPS during the acute inflammatory response. 5 hr p.i., age matched (6–8 weeks), female BALB/c mice fed all diets showed induction of Tnf, Il6, and Il1b expression in the blood (Figure 1C–E). However, at 5 hr p.i., WD- and KD-fed mice experienced significantly higher expression of Tnf and Il6 in the blood, compared with SC-fed mice, and WD-fed mice also showed significantly higher Il1b expression (Figure 1C–E), indicating that diets enriched in SFAs are associated with a hyper-inflammatory response to LPS.

Importantly, septic patients often present with two immune phases: an initial amplification of inflammation, followed by or concurrent with an induction of immune suppression (immunoparalysis), that can be measured by a systemic increase in the anti-inflammatory cytokine IL-10 (Cheng et al., 2016; Nedeva et al., 2019). Furthermore, in septic patients, a high IL-10:TNF ratio equates with the clinical immunoparalytic phase and correlates with poorer sepsis outcomes (Gogos et al., 2000; van Dissel et al., 1998). Interestingly, we found there was significantly increased Il10 expression in WD- and KD-fed mice, compared to SC-fed mice (Figure 1F), and WD- and KD-fed mice had significantly higher Il10:Tnf ratios at 10–20 hr and 15–20 hr, respectively, compared to SC-fed mice (Figure 1G). These data conclude that mice exposed to diets enriched in SFAs show an initial hyper-inflammatory response to LPS, followed by an increased immunoparalytic phenotype, which correlates with enhanced disease severity, similar to what is seen in the clinic.

We have previously shown that WD-fed mice experience increased endotoxemia severity and mortality, independent of diet-associated microbiome (Napier et al., 2019). In order to confirm the increases in disease severity that correlated with KD were also independent of KD-associated microbiome changes, we used a germ-free (GF) mouse model. 19–23-week-old female and 14–23-week-old male and female GF C57BL/6 mice were fed SC, WD, and KD for 2 weeks followed by injection with 50 mg/kg of LPS, our previously established LD₅₀ in GF C57BL/6 mice (Napier et al., 2019). As we saw in the conventional mice, at 10 hr p.i. WD- and KD-fed GF mice showed enhanced hypothermia and mortality, compared to SC-fed GF mice (Figure 1H, I). These data show that, similar to WD-fed mice, the KD-associated increase in endotoxemia severity and mortality is independent of diet-associated microbiome.

Our previous studies (Figure 1A–G) in conventional mice were carried out in 6–8-week female mice on a BALB/c background. Importantly, genetic background and age differences can have large effects on LPS treatment outcome. The GF mice used in this study (Figure 1H–N) were on a C57BL/6 background, between the ages of 14 and 23 weeks. Thus, we confirmed WD- and KD-fed conventional C57BL/6 mice aged 20–21 weeks old show enhanced disease severity and mortality in an LPS-induced endotoxemia model (4.5 mg/kg), compared to mice fed SC, similar to what is seen in younger BALB/c mice (Figure 1—figure supplement 1G, H).

Additionally, to confirm that the hyper-inflammatory response to systemic LPS was independent of the WD- and KD-dependent microbiome, we measured systemic inflammation during endotoxemia via the expression of Tnf, Il6, and Il1b in the blood at 0–10 hr p.i. We found, WD- and KD-fed GF mice displayed enhanced expression of Tnf and Il1b at 5–10 hr, and significantly enhanced expression of il-6 at 5 hr, compared to SC-fed GF mice (Figure 1J–L). Interestingly, Il10 expression and the Il10:Tnf ratio were not significantly different throughout all diets, suggesting the SFA-dependent enhanced immunoparalytic phenotype is dependent on the diet-associated microbiomes in WD- and KD-fed mice (Figure 1M–N). These data demonstrate that the early hyper-inflammatory response, but not the late immunoparalytic response, to LPS associated with enriched dietary SFAs is independent of the diet-dependent microbiota.

Thus far we find feeding diets enriched in SFAs (WD and KD) leads to enhanced expression of inflammatory cytokines in the blood after treatment with systemic LPS, suggesting that the SFAs may be inducing an innate immune memory response that leads to a hyper-inflammatory response to secondary challenge. Specifically, trained immunity is an innate immune memory response characterized by reprogramming of myeloid cells by a primary inflammatory stimulus, that then responds more robustly to secondary inflammatory challenge. Trained immunity has been shown to mediate cell sub-types within the HSC compartment that gives rise to “trained” myeloid progeny for weeks to years (De Zuani and Frič, 2022). A previous study in Ldlr^−/− mice has shown 4 weeks of WD feeding significantly enhances multipotent progenitors (MPPs) and granulocyte and monocyte precursors (GMPs) and skews development of GMPs toward a monocyte lineage that is primed to respond with a hyper-inflammatory response to LPS (Christ et al., 2018). Currently, it is unknown if diets enriched in SFAs fed to WT mice can induce changes within the HSC compartment or long-lasting trained immunity.

In order to determine the impact of dietary SFAs on bone marrow reprogramming in vivo, we next evaluated HSCs and progenitor cells via flow cytometry from age-matched (6–8 weeks) female WT BALB/c mice fed SC, WD, and KD for 2 weeks. Using previously published panels for analyzing HSC populations in the bone marrow (Kaufmann et al., 2018; Vazquez et al., 2015; Nowlan et al., 2020), we collected bone marrow and measured relative proportions of long-term HSCs (LT-HSCs; CD201⁺CD27⁺CD150⁺CD48⁻), short-term HSCs (ST-HSCs; CD201⁺CD27⁺CD150⁺CD48⁺), and MPPs (CD201⁺CD27⁺CD150⁻CD48⁺) (Figure 2A and B). Strikingly, we find that KD-fed mice showed significantly enhanced ST- and LT-HSCs, and MPPs compared to SC-fed mice (Figure 2C). Unlike previously reported in Ldlr^−/− mice, there was no significant change in ST-HSCs, LT-HSCs, or MPPs within WD-fed WT mice (Figure 2C). Furthermore, we did not see a significant increase in MPP3s for WD-fed mice, as previously published for Ldlr^−/− mice (Christ et al., 2018), or KD-fed mice; however, this may be due to the difference in genetic backgrounds, or length of diet administration (Figure 2—figure supplement 1A). These data are the first to show that the KD, a diet solely enriched in SFAs, alters hematopoiesis by enhancing expansion and differentiation of HSCs, similar to previously described inducers of trained immunity.

Furthermore, it is unknown if enriched dietary SFAs lead to long-lasting functional reprogramming associated with trained immunity that leads to a hyper-inflammatory response. Thus, we fed age-matched (6–8 weeks) female BALB/c mice SC, WD, or KD for 2 weeks, isolated bone marrow, differentiated into BMDMs for 7 days, and analyzed baseline inflammation and response to LPS. We found that untreated BMDMs isolated from mice fed SC and WD showed no significant differences in TNF or IL-6, and those from KD-fed mice showed a modest increase only in IL-6 compared to BMDMs from SC-fed mice (Figure 2—figure supplement 1B). However, when BMDMs were stimulated with LPS for 24 hr ex vivo, BMDMs from WD- and KD-fed mice showed significantly higher secretion of TNF, and only those from KD-fed mice showed significantly enhanced IL-6 secretion (Figure 2D–E). These data show that diets enriched in SFAs are inducing long-lasting inflammatory reprogramming of myeloid cells in vivo, and that reprogramming takes place within the bone marrow.

Importantly, monocytes and splenocytes are necessary for induction of systemic inflammatory cytokines during endotoxemia (Radzyukevich et al., 2021; Zhang et al., 2012). Thus, we wanted to assess if enriched dietary SFA induces in vivo reprogramming of monocytes and splenocytes, leading to an enhanced response to LPS ex vivo. First, we fed age-matched (6–8 weeks) female BALB/c mice SC, WD, or KD for 2 weeks, isolated bone marrow monocytes (BMMs) via magnetic negative selection using bone marrow extracted from femurs and tibias, and determined baseline expression of inflammatory cytokines. We found that prior to ex vivo LPS stimulation, BMMs isolated from mice fed SC, WD, or KD showed no significant difference in Tnf expression, and Il6 expression was significantly decreased in BMMs from KD-fed mice (Figure 2—figure supplement 1C). However, when BMMs were stimulated with LPS for 2 hr ex vivo, those from KD-fed mice showed significantly higher expression of Tnf and Il6, while those from WD-fed mice exhibited no significance in expression compared to SC-fed mice (Figure 2—figure supplement 1D). Similarly, we isolated splenocytes from SC-, WD-, and KD-fed mice and found no difference between homeostatic inflammation of splenocytes between diets, but a significantly enhanced expression of Tnf in the splenocytes of KD-fed mice, and not WD-fed mice, challenged with LPS (2 hr) compared to splenocytes from SC-fed mice (Figure 2—figure supplement 1E, F).

These data show the KD stimulates expansion of HSC populations and skew differentiation of myeloid progenitors that then give rise to macrophages with enhanced inflammatory potency (Figure 2A–E; Figure 2—figure supplement 2). Furthermore, these data suggest that BMDMs, BMMs, and splenocytes from WD- and KD-fed mice are not more inflammatory at homeostasis; however, when challenged with LPS, KD feeding confers a hyper-inflammatory response. Together, our results suggest the KD, a diet that comprised 90.5% SFAs, leads to reprogramming of the HSC compartment and long-lasting trained immunity.

It is known that the SFAs consumed in the diet determine the SFA profiles in the blood (Dougherty et al., 1987; Skeaff et al., 2006; Zöllner and Tatò, 1992), and that these SFAs have the potential to be immunomodulatory. Thus, we next wanted to identify target SFAs enriched in the blood of mice fed a diet exclusively enriched in SFAs that may be altering the systemic inflammatory response to LPS. Considering that the KD is enriched in SFAs and not sucrose, and that KD-fed mice showed distinct HSC alterations and LPS-induced hyper-inflammation in BMDMs, BMMs, and splenocytes treated ex vivo, the subsequent studies were performed exclusively on KD-fed mice. We used mass spectrometry lipidomics to create diet-dependent profiles of circulating fatty acids in SC- and KD-fed mice (Choi et al., 2018). Age matched (6–8 weeks), female BALB/c mice were fed SC or KD for 2 weeks, then serum samples were collected and analyzed using qualitative tandem liquid chromatography quadrupole time of flight mass spectrometry. We used principal component analysis (PCA) to visualize how samples within each data set clustered together according to diet, and how those clusters varied relative to one another in abundance levels of free fatty acids (FFA), triacylglycerols (TAG), and phosphatidylcholines (PC). For all three groups of FAs, individual mice grouped with members of the same diet represented by a 95% confidence ellipse with no overlap between SC- and KD-fed groups (Figure 3A–C). These data indicate that 2 weeks of KD feeding are sufficient to alter circulating FFAs, TAGs, and PCs, and that SC- and KD-fed mice display unique lipid blood profiles. Similarly, the relative abundance of SGs in SC- and KD-fed mice displayed unique diet-dependent profiles with no overlapping clusters, and abundance of specific SGs was significantly higher in the serum of KD-fed mice compared to SC-fed mice (Figure 3—figure supplement 1A, B). Though the independent role of each FFA, TAG, PC, and SG species has not been clinically defined, each are classes of lipids that when accumulated is associated with metabolic diseases, which have been shown to enhance susceptibility to sepsis and exacerbate inflammatory disease (Meikle and Summers, 2017; I S Sobczak et al., 2019; Sokolowska and Blachnio-Zabielska, 2019; Papadimitriou-Olivgeris et al., 2016).

Importantly, we identified a significant increase in multiple circulating FFAs within the KD-fed mice, compared to the SC-fed mice, many of which were SFAs (Figure 3D). Interestingly, in KD-fed mice we found a significant increase in free PA (C16:0), an immunomodulatory SFA that is found naturally in animal fats, vegetable oils, and human breast milk (Mancini et al., 2015) and is eightfold enriched in KD (Figure 3D, Supplementary file 1). Additionally, PA-containing TAGs and PCs were significantly elevated in KD-fed mice serum, compared to SC-fed mice (Figure 3—figure supplement 1C, D). These data indicate that KD feeding not only enhances levels of freely circulating PA, but also enhances the frequency PA is incorporated into other lipid species in the blood.

Many groups have shown that PA alone induces a modest, but highly reproducible increase in the expression and release of inflammatory cytokines in macrophages and monocytes (Lancaster et al., 2018; Korbecki and Bajdak-Rusinek, 2019). However, it remains unknown if PA can act as a primary inflammatory stimulus to induce a hyper-inflammatory response to a secondary heterologous stimulus in primary cells. Thus, we next wanted to determine if pre-exposure to physiologically relevant concentrations of PA altered the macrophage response to a secondary challenge with LPS. Current literature indicates a wide range of serum PA levels, between 0.3 and 4.1 mM, reflects a high-fat diet in humans (Abdelmagid et al., 2015; Liu et al., 2015; Gallego et al., 2018; Buchanan et al., 2021; Perreault et al., 2014). We aimed to use a physiologically relevant concentration of PA reflecting a human host for our in vitro studies, thus we treated primary bone marrow-derived macrophages (BMDMs) with and without 1 mM of PA containing 2% bovine serum albumin (BSA) for 12 hr, removed the media, subsequently treated with LPS (10 ng/mL) for an additional 24 hr, and measured expression and release of TNF, IL-6, and IL-1β. Importantly, the BSA dissolved in the media used for PA treatment solutions was endotoxin- and FA-free to ensure aberrant TLR signaling would not occur via BSA contamination, and fresh PA was conjugated to BSA-containing media immediately prior to use. We found that BMDMs pre-treated with PA (1 mM) for 12 hr expressed significantly higher levels of Tnf and Il6 in response to secondary treatment with LPS, compared to naïve BMDMs (Figure 3E and F). Il1b expression was significantly lower in cells pre-treated with PA (Figure 3G); however, secretion of TNF, IL-6, and IL-1β was enhanced in BMDMs pre-treated with PA (1 mM) for 12 hr and challenged with LPS (Figure 3H–J). We found a similar enhanced Il6 and Tnf expression in response to LPS in BMDMs treated with PA (1 mM) for twice the length of exposure (24 hr), and Il-1b expression was decreased (Figure 3—figure supplement 2A-C).

Furthermore, we pre-treated BMDMs with a concentration of PA that reflects the lower range of physiologically relevant serum levels and found 0.5 mM of PA induced significantly higher expression of Tnf, Il6, and Il1b after 12-hr challenge with LPS; however, only Tnf and Il6 were significantly enhanced after 24-hr LPS treatment, compared to naive BMDMs treated with LPS (Figure 3—figure supplement 2D-I).

Importantly, PA treatment can induce apoptosis and pyroptosis in various cell types (Borradaile et al., 2006; Li et al., 2018; Ly et al., 2017; Tao et al., 2021); however, we found only an average of 3.4 and 4.4% of cell death after a 12-hr or 24-hr incubation, respectively, with PA (1 mM) and subsequent 24 hr of LPS treatment or control media (Figure 3—figure supplement 3A, B). These data demonstrate PA pre-treatment of macrophages induces a hyper-inflammatory response to LPS independent of cell death, suggesting PA is sensitizing macrophages to secondary inflammatory challenge.

Thus, we conclude that both 12- and 24-hr pre-treatments with 0.5 mM or 1 mM of PA conjugated to 2% BSA are sufficient to induce reprogramming of macrophages and alter the response to stimulation with a heterologous ligand. Additionally, these data demonstrate that even serum concentrations of PA that are at the lower end of the spectrum for humans consuming a high-fat diet pose a risk for inflammatory dysfunction.

PA treatment of various cell types diverts cellular metabolism toward the synthesis of the toxic metabolic byproducts: diacylglycerols and ceramide (Palomer et al., 2018). PA-induced ceramide synthesis has specifically been demonstrated to enhance inflammation (Schilling et al., 2013; Zhang et al., 2017; Schwartz et al., 2010; Jin et al., 2013). Considering this, we wanted to determine the role of enhanced macrophage ceramide production in driving PA-induced hyper-inflammatory response to LPS. Thus, we treated BMDMs simultaneously with PA (0.5 mM) and a ceramide synthase inhibitor Fumonisin B1 (FB1; 10 μM), for 12 hr, removed the media, subsequently treated with LPS (10 ng/mL) for an additional 24 hr, and measured release of TNF, IL-6, and IL-1β. We found that BMDMs pre-treated simultaneously with PA and FB1 for 12 hr expressed significantly lower levels of TNF, IL-6, and IL-1β secretion in response to LPS, compared to BMDMs pre-treated with only PA (Figure 3K–M). We conclude that ceramide synthesis induced by PA is required for the macrophage hyper-inflammatory response to secondary challenge with LPS.

Considering the drastic effect of PA on macrophage response to secondary challenge with LPS, we next wanted to understand if exposure to PA alone is sufficient to induce a hyper-inflammatory response during endotoxemia in vivo. We answered this question using age-matched (6–8 weeks) female BALB/c mice fed SC for 2 weeks, by mimicking systemic PA levels found in serum of humans on high-fat diet via a single i.p. injection of ethyl palmitate (750 mM), and then after 12 hr, challenging with LPS i.p. (Eguchi et al., 2012). Similar to previous publications, we find that a 750-mM i.p. injection of ethyl palmitate enhances free PA levels in the serum to 173–425 μM compared to Veh-treated mice with 110–250 μM (Figure 4—figure supplement 1A). Important to note, free PA is only transiently enhanced by systemic application and is quickly (<1 hr) taken up by peripheral tissues; thus, our detected free serum levels are most likely an underestimation of transient systemic PA (Black, 2007; Mansbach and Gorelick, 2007; Karavolos et al., 2008).

Interestingly, after LPS challenge, PA-treated mice experienced increased disease severity as indicated by their significant decline in temperature compared to Veh mice (Figure 4A). Similar to WD- and KD-fed mice, PA-treated mice also exhibited enhanced mortality, compared to Veh mice (Figure 4B). Importantly, mice injected with PA for shorter time periods (0, 3, and 6 hr) and then challenged with LPS did not exhibit increased disease severity or poor survival outcome (Figure 4—figure supplement 1B, C), concluding that a 12-hr pre-treatment with PA is required for an increase in disease severity.

Next, we measured systemic inflammatory status during disease and found similar to KD-fed mice, the 12-hr PA-pre-treated mice showed significantly enhanced expression of Tnf (5 hr and 10 hr) and Il6 (5 hr) post-LPS challenge, compared to Veh control (Figure 4C and D). Expression of Il-1b trended upward but was not significantly upregulated in 12-hr PA-pre-treated mice, compared to Veh-treated mice (Figure 4E). Importantly, as a control we looked at LPS-induced hypoglycemia in PA-treated mice, and 12-hr pre-treatment with PA did not alter LPS-induced hypoglycemia (Figure 4—figure supplement 1D), indicating that diet-dependent hypoglycemic shock was not a driver of endotoxemia severity in PA-treated mice. Thus, exposure to PA to mimic systemic levels found in humans eating high-fat diets is sufficient to drive enhanced inflammation and disease severity in mice stimulated with endotoxin, and this effect is dependent on length of PA exposure.

Our data show that pre-treatment with systemic PA alone enhances endotoxemia severity in vivo and inflammatory responses of macrophages to a secondary and heterologous stimulus in vitro. This form of regulation resembles trained immunity; however, it remains unclear if PA is inducing trained immunity in vivo. We first evaluated the basal level expression of Tnf, Il6, and Il1b in mice treated with 750 mM of PA or Veh i.p. for 12 hr, before stimulation with LPS. Interestingly, we did not see significant differences in Tnf, Il6, or Il1b expression at 12 hr of exposure with PA (Figure 4F), which suggests that circulating immune cells of these mice is not in a primed state at these time points prior to LPS injection. These data suggest PA induces trained immunity, and not priming; however, the time point of initial inflammation induced by PA remains unknown.

As mentioned previously, canonical inducers of trained immunity (e.g. BCG or β-glucan) induce long-lived enhanced innate immune responses to secondary inflammatory stimuli (Kaufmann et al., 2018, Netea et al., 2020). Thus, we hypothesized that exposure to a PA bolus would enhance disease severity and mortality in mice, and that this phenotype would persist even after mice were rested from PA injections for 1 week. We injected age matched (6–8 weeks), female BALB/c mice fed SC with a vehicle solution (Veh→SC) or PA (750 mM; PA→SC) i.p. once a day for 9 days and then rested the mice for 1 week. When challenged with systemic LPS, PA→SC showed an increase in disease severity and mortality compared to Veh→SC mice (Figure 4G and H), indicating that PA alone can induce long-lived trained immunity that increases susceptibility to inflammatory disease. Importantly, the difference between Veh→SC and PA→SC survival was not significant (Figure 4H), suggesting PA is not the sole driver of the enhanced mortality we see in KD.

Lastly, the most commonly studied models for inducing trained immunity are immunization with BCG or stimulation with β-glucan, and they have been shown to protect mice from systemic Candida albicans infection via lymphocyte-independent immunological reprogramming that leads to decreased kidney fungal burden (Kleinnijenhuis et al., 2012). Therefore, we next tested if PA treatment induces lymphocyte-independent clearance of C. albicans infection. For these experiments, Rag1 knockout (Rag1^−/−) mice were treated with a vehicle or PA solution for 12 hr and subsequently infected intravenously (i.v.) with 2×10⁶ C. albicans. In accordance with canonical trained immunity models, mice treated with PA for 12 hr showed a significant decrease in kidney fungal burden compared to Veh mice, 24 hr post-infection (Figure 4I). These are the first data to suggest PA enhances innate immune clearance of C. albicans in vivo.

We have reported here that diversion of ceramide synthesis reverses the PA-dependent hyper-inflammatory response to LPS in macrophages in vitro (Figure 3K–M). Interestingly, OA (C18:1) is a mono-unsaturated fatty acid naturally found in animal fats and vegetable oils, and in the presence of PA, diverts lipid metabolism away from ceramide production (Palomer et al., 2018; Listenberger et al., 2003). Considering OA and PA are the most prevalent fatty acids found in the human diet and in human serum (Palomer et al., 2018), we wanted to test if OA diversion of ceramide synthesis could reverse the PA-dependent hyper-inflammatory response to LPS in macrophages. Thus, we treated BMDMs with OA (0.2 mM), PA (0.5 mM), or OA and PA together for 12 hr and then with LPS. We found that macrophages simultaneously pre-treated with PA and OA produced significantly lower levels of TNF, IL-6, and IL-1β following subsequent LPS exposure, compared to BMDMs pre-treated with only PA prior to LPS stimulation (Figure 5A–C). These data reveal OA-dependent depletion of intracellular ceramides neutralizes the PA-dependent hyper-inflammatory response to LPS in macrophages.

Considering this, we next wanted to know if i.p. injections of OA in KD-fed mice would mitigate enriched dietary SFA-associated disease severity and mortality. Thus, we fed age-matched (6–8 weeks) female BALB/c mice SC or KD for 2 weeks and injected them i.p. with 300 mM OA or Veh once per day for the final 3 days of feeding. We then injected LPS i.p. and measured hypothermia and survival. Veh-injected KD-fed mice showed significant and prolonged hypothermia starting at 8 hr p.i. compared to SC-fed mice (Figure 5D). In accordance with these findings, KD-fed mice displayed significantly enhanced mortality by 24 hr p.i. compared to 100% survival of SC-fed mice (Figure 5E). Strikingly, for KD-fed mice injected with 300 mM OA prior to LPS treatment, there was minimal temperature loss comparable to SC-fed mice, and 100% survival (Figure 5D and E). Together, these data show systemic OA can abrogate KD-dependent hypothermia and survival defect in response to LPS in mice-fed diets enriched solely in SFA and highlight the fascinating plasticity of dietary fatty acid reprogramming of innate immune cell populations and disease dynamics.

RAW 264.7 macrophages (from ATCC), CASP-1KO BMDMs, BMDMs, and BMMs were maintained in Dulbecco's Modified Eagle Medium (DMEM; Gibco) containing L-glutamine, sodium pyruvate, and high glucose supplemented with 10% heat-inactivated fetal bovine serum (FBS; GE Healthcare, SH3039603). BMDMs were also supplemented with 10% macrophage colony-stimulating factor (M-CSF; M-CSF-conditioned media was collected from National Institutes of Health (NIH) 3T3 cells expressing M-CSF, generously provided by Denise Monack at Stanford University).

BMDMs and BMMs were harvested from the femurs and tibias of age-matched (6–8 weeks) CO₂-euthanized female BALB/c mice or male and female C57BL/6 J mice. BMDM media was supplemented with 10% M-CSF for differentiation, cells were seeded at 5×10⁶ in petri dishes and cultured for 6 days, collected with cold PBS, and frozen in 90% FBS and 10% dimethyl sulfoxide (DMSO) in liquid nitrogen for later use. BMMs were isolated from BMDM fraction using EasySep Mouse Monocyte Isolation Kit (STEMCELL). Spleens were harvested from age-matched (6–8 weeks) CO₂-euthanized female BALB/c mice, tissue was disrupted using the end of a syringe plunger on a 70-μm cell strainer and rinsed with FACS buffer (PBS + 2 mM EDTA). Cells were subjected to red blood cell lysis with RBC lysing buffer (Sigma) followed by neutralization in FACS buffer.

After thawing and culturing for 5 days, BMDMs were pelleted and resuspended in DMEM containing 5% FBS, 2% endotoxin- and fatty acid-free BSA (Proliant Biologicals) and 10% M-CSF. Cells were seeded at 2.5×10⁵ cells/well in 24-well tissue culture plates, treated with EtOH (1.69%, or 0.83%) 10 ng/mL LPS (Ultrapure LPS, Escherichia coli 0111:B4, Invivogen), 500 μM or 1 mM PA (Sigma-Aldrich, PHR1120), 10 uM FB1 (Sigma-Aldrich, F1147), or 200 μM OA (Sigma-Aldrich, O7501) and incubated at 37°C and 5% CO₂ for 12 or 24 hr. Next, cells were treated with an additional 10 ng/mL LPS and incubated an additional 12 or 24 hr. RAW 264.7 macrophages were thawed and cultured for 3–5 days, pelleted and resuspended in DMEM containing 5% FBS and 2% endotoxin- and fatty acid-free BSA, and treated identical to BMDM treatments. BMMs were seeded immediately after harvesting at 4×10^⁵ cells/well in 96-well V-bottom plates in DMEM containing 10% FBS and treated with LPS for 2 or 24 hr. Splenocytes were seeded immediately after harvesting at 1×10⁵ cells/well in 96-well V-bottom plates in RPMI media with L-glutamine (Cytiva) containing 10% FBS and treated with LPS for 2 or 24 hr. BMDMs for ex vivo treatments were isolated as described above, plated at 2.5×10⁵ cells/well in 24-well plates, and stimulated with 10 ng/mL LPS after 12 hr of adherence. For all treatments, supernatant was removed for ELISA analysis, and cells were lysed with TRIzol (ThermoFisher), flash-frozen in liquid nitrogen, and stored at –80°C until qRT-PCR analysis. For all plates, all treatments were performed in triplicate.

Modified panel using combined methods from Kaufmann et al., Nowlan et al., and Vasquez et al. Red blood cells were lysed in BM cells using RBC lysis buffer (Biolegend). BM cells (3×10⁶ cells) were stained with viability stain Live/Dead Fixable Aqua (ThermoFisher) at the concentration of 1:200 for 30 min at 4°C. Next, cells were washed with FACS buffer (PBS supplemented with 0.5% BSA; Proliant Biologicals, fatty acid free), and incubated with anti-CD16/32 (clone 93, BioLegend) at a concentration of 1:100 in FACS buffer for 10 min at 4°C. The following antibodies were then used for staining HSCs, and MPPs: anti-Ter-110, anti-CD11b (clone M1/70), anti-CD5 (clone 53–7.3), anti-CD4 (clone RM4-5), anti-CD8a (clone 53–6.7), anti-CD45R (clone RA3-6B2), and anti-Ly6G/C (clone RB6-8C5), all biotin-conjugated (all BD Bioscience), were added at a concentration of 1:100 for 30 min at 4°C and washed with FACS buffer. Streptavidin-APC-Cy7 (eBioscience), anti-CD150-eFluor450 (clone Q38-480, eBioscience), anti-CD48-PerCPeFluor710 (BD Bioscience), anti-Flt3-PE (clone A2F10.1, BD Bioscience), anti-CD34-PEDazzle 594 (clone HM34, BioLegend), anti-CD27-PE-Cy7 (eBioscience), and anti-CD201-APC (eBioscience) were added all at a concentration of 1:100 for 20 min at 4°C. All cells were then washed with FACS buffer before and after incubation in 1% paraformaldehyde for 30 min at 4°C. Cells were acquired on BD flow cytometer (FACSymphony A1 Cell Analyzer) with FACSDiva Software. Analyses were performed using FlowJo software v.10.1. The DownSample version 3.3.1 plugin was used to standardize events for each sample after populations were gated.

BMDMs were cultured as stated above with culture media, PA, or ethanol in 96-well tissue-culture plates at a concentration of 5×10⁴ cells/well and incubated for 12 hr. Cells were treated with PBS or 10 ng/mL LPS in a phenol-red-free Optimem media (ThermoFisher) and incubated an additional 12 or 24 hr. Supernatants were collected at the specified time points with LDH release quantified with a CytoTox96 Non-Radioactive Cytotoxicity Assay (Promega). Cytotoxicity was measured per well as a percentage of max LDH release, with background media-only LDH release subtracted. For all plates, all treatments were performed in triplicate.

Cell viability was determined by 0.4% Trypan Blue dye exclusion test executed by a TC20 Automated Cell Counter (Bio-Rad).

Mice were treated with PBS or LPS, and at specified time points 10–20 μL of blood was collected from the tail vein, transferred into 50 μL of RNALater (ThermoFisher Scientific), and frozen at –80°C. RNA extractions were performed using RNeasy Mini Kit (Qiagen), and cDNA was synthesized from RNA samples using SuperScript III First-Strand synthesis system (Invitrogen). Gene-specific primers were used to amplify transcripts using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad). A complete list of all primers used, including the names and sequences, is supplied as Supplementary file 2.

TNF, IL-6, and IL-1β concentrations in mice serum were measured and analyzed using TNF, IL-6, and IL-1β Mouse ELISA kits (ThermoFisher Scientific), according to the manufacturer’s instructions. Absorbances were measured at a wavelength of 450 nm using a microplate reader (BioTek Synergy HTX). Values below the limit of detection (LOD) of the ELISA were imputed with LOD divided by 2 (LOD/2) values.

Age-matched (6–8 weeks) female BALB/c mice were anesthetized with isoflurane and injected subcutaneously with ID transponders (Bio Medic Data Systems). 2 weeks post diet change, and 1 week post ID transponder injection, mice were stimulated with a single injection of 6–10 mg/kg LPS reconstituted in endotoxin-free LAL reagent water (Invivogen) and diluted in PBS for a total volume of 200 μL. Control mice received corresponding volumes of PBS. Progression of disease was monitored every 2 hr after LPS injection for clinical signs of endotoxin shock based on weight, coat and eyes appearance, level of consciousness, and locomotor activity. Age-matched (20–21 weeks) female C57BL/6 mice were treated as described above, except for their LPS dose (4.5 mg/kg). Temperature was recorded using a DAS-8007 thermo-transponder wand (Bio Medic Data Systems). For PA injections, a solution of 750 mM ethyl palmitate (Millipore Sigma), 1.6% lecithin (Sigma-Aldrich), and 3.3% glycerol was made in endotoxin-free LAL reagent water (Lonza). The lecithin-glycerol-water solution was used as a vehicle, and mice were injected with 200 μL of the vehicle as a control or ethyl palmitate solution to increase serum PA levels. For OA injections, a solution of 300 mM OA (Sigma-Aldrich) was made using the same solution and vehicle described above. Mice were injected i.p. with 200 μL of the vehicle as a control, or OA solution, between 7 and 9 pm for 3 days prior to LPS exposure.

Six-week-old female mice were fed soft, irradiated chow (PicoLab Mouse Diet 20, product 5058) and allowed to acclimate to research facility undisturbed for 1 week. Chow was replaced by WD (Envigo, TD.88137), KD (Envigo, TD.180423), or SC (Envigo, TD.08485), and mice were fed ad libitum for 2 weeks before induction of endotoxemia. For KD, food was changed daily. For WD, food was changed every 72 hr. Ketones and blood glucose were measured weekly and immediately prior to LPS injections with blood collected from the tail vein using Blood Ketone and Glucose Testing Meter (Keto-Mojo), or with urine collected on ketone indicator strips (One Earth Health, Ketone Test Strips).

Mann Whitney, Mantel-Cox, and student’s t-tests were carried out with GraphPad Prism 9.0 software.

All animal studies were performed in accordance with NIH guidelines, the Animal Welfare Act, and US federal law. All animal experiments were approved by the Oregon Health and Sciences University (OHSU) Department of Comparative Medicine or Oregon State University (OSU) Animal Program Office and were overseen by the Institutional Care and Use Committee (IACUC) under Protocol IDs #IP00002661 and IP00001903 at OHSU and #5091 at OSU. Conventional animals were housed in a centralized research animal facility certified by OHSU. Conventional 6–10-week-aged female BALB/c mice (Jackson Laboratory 000651) were used for the endotoxemia model, and isolation of BMDMs, BMMs, and splenocytes. GF male and female C57BL/6 mice (Oregon State University; bred in house) between 14- and 23-week-old were used for the GF endotoxemia model. BALB/c Rag1^−/− mice between 8 and 24 weeks were infected i.v. with 2×10⁶ CFUs of C. albicans SC5314 (ATCC #MYA-2876), and kidney fungal burden was assessed 24 hr post-infection. Kidneys were harvested 24 hr post-infection, and homogenized organs were plated in serial dilutions on Yeast Peptone Dextrose plates to assess fungal burden.

Mice on specialized diets were sacrificed at the indicated time points after PBS or LPS treatment with 300–600 µL of blood collected via cardiac puncture into heparinized tubes. Blood samples were centrifuged for 15 min at 2500 rpm at 4°C, and serum was transferred to a new tube before storage at –80°C. Serum samples were analyzed via LC-MS/MS. Lipidomic data sets were scaled using the scale function, and PCAs were performed using the prcomp function from the stats package in R Version 3.6.2. Visualization of PCAs and biplots was performed with the fviz_pca_ind and fviz_pca_biplot functions from the factoextra package and with the ggplot2 package (Mundt and Weisweiler, 2017; Wickham, 2016). For each diet group, 95% confidence ellipses were plotted around the group mean using the coord.ellipse function from the FactoMineR package (Lê, 2008). Heatmaps were created using the pheatmap package (Kolde, 2018).