The Institutional Animal Care and Use Committee at Michigan State University (MSU) approved all experimental protocols (AUF #201800113) in accordance with guidelines established by the National Institutes of Health. Six-week-old female lupus-prone NZBWF1 mice were procured from the Jackson Laboratory (Bar Harbor, ME) and randomized into experimental groups for each study ( Supplementary Tables 1 , 2 ). Only female NZBWF1 mice were used in this study because female mice of this strain exhibit greater severity and prevalence of lupus-related symptoms (e.g., elevated antinuclear antibody titers, formation of immune complexes, glomerulonephritis) compared to male NZBWF1 mice (61). Mice were housed 2 or 4 per cage in Study 1 and 4 per cage in Study 2, and all mice were given free access to food and water. Consistent lighting (12 h light/dark cycle), temperature (21-24°C), and humidity (40-55%) were maintained in animal housing facilities.

Four defined diet formulations were prepared as described in Supplementary Table 3 . All formulations used purified American Institute of Nutrition (AIN)-93G diet (70 g/kg fat) as a base to provide optimal nutrition to experimental rodents (62). All diets contained 10 g/kg corn oil as a source of essential ω-6 fatty acids. The basal diet for Study 1 and control (CON) diet for Study 2 contained 60 g/kg high-oleic safflower oil (Hain Pure Food, Boulder, CO). For DHA-enriched diets, human equivalent caloric consumption of 5 g DHA per day was achieved by adding 25 g/kg microalgal oil containing 40% DHA (DHASCO; DSM Nutritional Products, Columbia, MD) in place of high-oleic safflower oil, resulting in 10 g DHA/kg diet (63). For TPPU-amended diets, 22.5 mg TPPU (95% purity based on H-NMR analysis), synthesized and purified as described previously (34), was added to 1 kg of CON or DHA diet, resulting in the TPPU and TPPU+DHA diets. Fatty acid ( Table 1 ) and TPPU ( Supplementary Table 4 ) concentrations in each diet were confirmed as described below.

Fatty acid composition in each experimental diet was determined by modifying a previously described protocol (64). Briefly, 400 mg of each diet sample was reconstituted in a 4:1 (v/v) ethanol/methanol solution + 0.1% (v/v) butylated hydroxytoluene (BHT) and heated 15 min at 55°C in a CEM Mars 6 Xpress microwave digestion system (CEM Corporation, Matthews, NC). Then, 2 mg of extracted fatty acids from each diet sample were converted to fatty acid methyl esters (FAMEs) by treating with 500 µl of toluene and 20 µg of internal standard (methyl-12-tridecenoate), incubating with 2 ml of KOH (0.5 N) at 50°C for 10 min, then subsequently incubating with 3 ml of methanolic HCl (5% [v/v] at 80°C for 10 min to allow base-catalyzed methylation and acid-catalyzed methylation, respectively. Following methylation, 2 ml of HPLC-grade water was added to the samples, and FAMEs were extracted by adding 2 ml of hexane to the samples twice. Extracted FAMEs were dried under nitrogen with an Organomation Multivap Nitrogen Evaporator (Organomation Associates, Berlin, MA). Dried FAMEs were then resuspended in 1 ml of isooctane and kept at -20°C until further analysis.

FAMEs were analyzed by GC-MS as previously described (64). Briefly, FAMEs in each sample were separated on a Perkin Elmer 680/600 GC-MS (Waltham, MA) outfitted with a HP-88 capillary column (100 m × 0.25 mm inner diameter × 0.2 µm film thickness; Agilent Technologies, Santa Clara, CA). MassLynx^TM (4.1 SCN 714; Waters Corporation, Milford, MA) was used to compare analyte retention time and electron ionization (EI) mass fragmentation to those in the reference standard, which consisted of Supelco 37 Component FAME Mix (Sigma-Aldrich, St. Louis, MO), mead acid, docosatetraenoic acid, ω-3 docosapentaenoic acid (DPA), ω-6 DPA, and palmitelaidic acid (Cayman Chemical, Ann Arbor, MI). FAME analyte peak areas were converted to individual FAME concentrations using a standard curve based on the reference standard and internal standard. For fatty acids with a detected chain length of 10 to 24 carbon atoms, fatty acid content in the diet is reported as percentage (w/w) of total fatty acids quantified ( Table 1 ).

S-LPS from Salmonella enterica serotype minnesota (cat. #L6261) and R-LPS from Salmonella enterica serotype minnesota Re 595 (cat. #L9724) were purchased from Sigma Aldrich (St. Louis, MO). Immediately prior to all intraperitoneal (i.p.) injections, stock suspensions of LPS were prepared in sterile phosphate buffered saline (PBS), sonicated for 15 min, and vortexed for 1 min.

Experimental designs for Study 1 and Study 2 are shown in Figures 1A, B , respectively. In both studies, female lupus-prone mice were administered experimental diets beginning at age 6 wk and maintained on the same diets throughout the entire experiment. To prevent lipid oxidation, all experimental diets were prepared every other week and stored at -20°C until administered to mice. Mice received fresh diet every day. Starting at age 8 wk, all groups of mice were injected intraperitoneally with S-LPS, S-LPS, or PBS vehicle twice per wk for 5 wk, for 10 total injections. On a weekly basis, body weights were measured and urine sampled for development of proteinuria and hematuria using clinical protein dipsticks (Cortez Diagnostics, Calabasas, CA) and blood dipsticks (Teco Diagnostics, Anaheim, CA), respectively. To compare the inflammatory and autoimmune responses triggered by S-LPS and R-LPS (Study 1), groups of female mice (n = 2-4/gp) were given control (CON) AIN-93G diet and intraperitoneally injected with S-LPS (0.8 µg/g body weight [BW]) or R-LPS (0.8 µg/g BW) in 500 µl of PBS or PBS vehicle as previously described (52). To assess effects of separate and concurrent DHA and TPPU administration on lupus GN induced by R-LPS (Study 2), female mice (n = 8/gp) were fed one of four experimental diets: 1) CON, 2) DHA, 3) TPPU, or 4) TPPU+DHA. Mice were intraperitoneally injected with R-LPS (0.6 µg/g BW) in 500 µl of PBS or PBS vehicle as previously described (53). After 5 R-LPS injections, blood samples were collected from the lateral saphenous vein to assess TPPU plasma concentration (Study 2). Mice for both Study 1 and Study 2 were terminated at age 13 wk (5 wk after the first LPS injection). This timepoint was selected for termination because it corresponds with development of accelerated, severe lupus GN previously reported in female NZBWF1 mice (48, 52, 53).

Primary euthanasia for all mice occurred by intraperitoneal injection of 56 mg/kg BW sodium pentobarbital, followed by abdominal aortic exsanguination as a means of secondary euthanasia. Blood was obtained with heparin-coated syringes and plasma isolated by centrifugation at 3500 x g for 10 min under cold conditions (4°C). An antioxidant cocktail (0.2 mg/ml butylated hydroxytoluene, 0.2 mg/ml triphenylphosphine, 0.6 mg/ml EDTA) (65) was prepared and added at a 5% (v/v) concentration to all plasma aliquots designated for LC-MS/MS analysis. All plasma samples were stored at -80°C as single-use aliquots for LC-MS/MS, blood urea nitrogen (BUN) and creatinine quantification, and AAb microarray profiling. The left kidney was removed and fixed in 10% (v/v) neutral-buffered formalin (Fisher Scientific, Pittsburgh, PA) for 24 h. The right kidney was cut longitudinally, with one half immersed in RNAlater (Thermo Fisher Scientific, Waltham, MA) overnight at 4°C then stored at –80°C for RNA analysis. The spleen was transversely cut in half, with one half fixed in 10% formalin and the other half immersed in RNAlater as described above. The left lateral lobe of the liver was cut transversely, with one half of the lobe fixed in 10% formalin fixative and the other half immersed in RNAlater as described above. All fixed tissues were transferred to 30% (v/v) ethanol for additional routine processing, for light microscopic examination, and for long-term storage.

Red blood cell samples were sent to OmegaQuant Inc. (Sioux Falls, SD) for determination of membrane fatty acid concentrations by gas-liquid chromatography (GLC) as previously described (63).

Waters Oasis-HLB cartridges (part WAT094226, lot 176A30323A) were used for sample preparation and clean-up purposes. Solid-phase extraction (SPE) cartridges were prepared for solid phase extraction by washing once with 2 ml of ethyl acetate, twice with 2 ml of methanol, and twice with 2 ml of 95:5 (v/v) water/methanol + 0.1% (v/v) acetic acid. Plasma was then loaded onto the cartridges, and samples were spiked with 10 μl of deuterated internal standard solution (16 nM BGB2-d4, 10 nM LTB4-d4, 16 nM 8,9-DiHETrE-d11, 16 nM 9-HODE-d4, 20 nM 15(S)-HETE-d8, 40 nM 5(S)-HETE-d8, 40 nM 8,9-EpETrE-d11) and 10 μl of antioxidant cocktail (0.2 mg/ml butylated hydroxytoluene, 0.2 mg/ml triphenylphosphine, 0.6 mg/ml EDTA). After loading samples, cartridges were washed with 1.5 ml of 95:5 (v/v) water/methanol + 0.1% (v/v) acetic acid then dried with a low vacuum for 20 min to remove water and other unwanted residues. For elution, 6 µl of trap solution (30% [v/v] glycerol in methanol) was added to separate 2-ml Eppendorf tubes, then SPE cartridges were washed with 0.5 ml of methanol followed by 1 ml of ethyl acetate. The eluents were then concentrated under a high vacuum, and residues were reconstituted in 100 μl of 75% ethanol (v/v) containing 10 nM 12-[[(cyclohexylamino)carbonyl]amino]-dodecanoic acid (CUDA) as an internal standard. The samples then vortexed for 5 min followed by filtration through a 0.45-μm filter, then the filtrates were transferred to LC-MS/MS vials for analysis.

A XBridge BEH C18 2.1x150 mm, 5 µm, HPLC column, (ser. #01723829118314) was used for ultra-performance liquid chromatography (UPLC). The column was connected to a Waters TQ-XS tandem quadrupole UPLC/MS/MS instrument outfitted with a Waters ACQUITY SDS pump and Waters ACQUITY CM detector (Milford, MA). For UPLC, the chromatographic method was optimized to separate all analytes in 20 min using a sample volume of 10 µl and flow rate of 250 µl/min ( Supplementary Table 5 ). Gradient elution was performed by using 0.1% (v/v) acetic acid in water for mobile phase A and 84:16 (v/v) acetonitrile/methanol + 0.1% glacial acetic acid for mobile phase B. During sample injection, the Waters ACQUITY FTN autosampler (Milford, MA) was held at a consistent temperature of 10°C.

The ionization source for multiple reaction monitoring (MRM) modes was electrospray. MRM transitions and source parameters were optimized for each standard compound by individually infusing each compound separately into the mass spectrometer, ultimately to achieve the most optimal selectivity and sensitivity. For each experimental sample, Waters MassLynx™ MS software v4 (Milford, MA) was used to quantify analyte area, internal standard (IS) area, raw concentration (in nM), and signal-to-noise (S/N) ratio based on an 8spots-calibration linear standard curve. Dilution factors were calculated for each sample by dividing the original sample volume (in µl) by 100 µl. Normalized analyte concentrations in each sample were then quantified by dividing raw analyte concentrations by the sample’s corresponding dilution factor.

Plasma levels of BUN and creatinine were quantified using a Urea Nitrogen Colorimetric Detection Kit (Thermo Fisher Scientific, Waltham, MA; cat. #EIABUN) and Creatinine Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI; cat. #700460), respectively, according to the manufacturers’ instructions.

Formalin-fixed kidneys were embedded in paraffin, sectioned at a thickness of 5 µm, and stained with hematoxylin and eosin (H&E) or Periodic acid Schiff and hematoxylin (PASH). A board-certified veterinary pathologist semi-quantitatively scored sectioned tissues in a blinded manner (i.e., without knowledge of individual animal treatments) using a modification of the International Society of Nephrology/Renal Pathology Society (ISN/RPS) classification system for lupus GN (66). Each tissue section was assigned one of the following grades (0): normal glomeruli and no tubular proteinosis (1); multifocal segmental proliferative GN with mild tubular proteinosis and occasional early glomerular sclerosis and crescent formation (2); diffuse segmental proliferative GN with moderate tubular proteinosis, early glomerular sclerosis, and crescent formation; or (3) pervasive global proliferative and sclerosing GN with marked tubular proteinosis.

Fixed spleens and livers were processed and semi-quantitatively scored for histopathological development in a similar manner as the kidneys in this study. Scored liver lesions included (1) hepatocellular small and large vacuoles resembling lipid droplets and (2) periportal cellular inflammation (consisting primarily of inflammatory lymphocytes, plasma cells, and occasional neutrophils). Severity scores for these hepatic lesions were based on the percentage of the liver tissue section affected: (0) no treatment-induced lesions, (1) minimal (<10% affected), (2) mild (11-25% affected), (3) moderate (26-50% affected), (4) marked (51-75% affected), or (5) severe (76-100% affected).

Kidney immunohistochemistry was performed as previously described (67). Briefly, formalin-fixed kidney sections were stained with polyclonal goat anti-mouse IgG antibody (Bethyl Labs, Montgomery, TX; cat. #A-90-100A), polyclonal rabbit anti-mouse CD3 antibody (Abcam, Cambridge, MA; cat. #ab5690), or monoclonal rat anti-mouse CD45R antibody (Becton Dickinson, Franklin Lakes, NJ; cat. #550286) at the MSU Investigative Histopathology Laboratory to detect total IgG, CD3⁺ T lymphocytes, and CD45R⁺ B lymphocytes, respectively. Slides were scanned with a Slideview VS200 research slide scanner (Olympus, Hicksville, NY). Semi-quantitative scores for IgG deposition in kidneys were assigned using the following scale: (0) no changes compared to VEH/CON mice, (1) minimal (<10% affected), (2) mild (11-25% affected), (3) moderate (26-50% affected), (4) marked (51-75% affected), or (5) severe (76-100% affected).

IgG and IgM AAbs were profiled in plasma (Study 2) by OmicsArray™ Systemic Autoimmune-associated Antigen Array (Genecopoiea Inc., Rockville, MD; cat. #PA001). All plasma samples within experimental groups were pooled prior to analysis. Briefly, plasma samples were incubated on microscope slides with 120 purified antigens adhered to nitrocellulose filters. One identical OmicsArray panel was reserved for a PBS negative sample control. After incubation, all slides were washed and incubated with Cy3-labeled anti-mouse IgG and Cy5-labeled anti-mouse IgM secondary antibodies. Slides were washed and fluorescent signals (532 nm for Cy3/IgG, 635 nm for Cy5/IgM) were detected using a GenePix^® 4400B microarray scanner (Molecular Devices, San Jose, CA), and GenePix^® 7.0 software (Molecular Devices) was used to determine fluorescent signal intensity values. Antibody scores (Ab-scores) for all AAbs were calculated using normalized signal intensity (NSI) and signal-to-noise ratio (SNR) values using the following formula:

Total RNA from kidneys was extracted using TissueLyser II (Qiagen, Germantown, MD) and a RNeasy Mini Kit (Qiagen; cat. #74104) according to the manufacturer’s instructions. Isolated RNA was reconstituted in RNase-free water and quantified using a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA). cDNA was prepared from isolated RNA at a concentration of 100 ng/µl using a High-Capacity cDNA Reverse Transcriptase Kit (Thermo Fisher Scientific, Waltham, MA). Taqman assays were run with technical triplicates using a Smart Chip Real-Time PCR System at the MSU Genomics Core to assess interleukin (Il1a, Il1b, Il2, Il6, Il17a, Il18), chemokine (Ccl2, Ccl7, Ccl12, Cxcl9, Cxcl10, Cxcl13), inflammation/autoimmunity (C1qa, C3, Casp1, Casp4, Icam1, Ifng, Lbp, Nfkb1, Nlrp3, Nos2, Pparg, Tlr4, Tlr9, Tnfa, Tnfsf13b), type I interferon (IFN)-related (Ifi44, Irf7, Isg15, Nlrc5, Oas2), eicosanoid-related (Alox15, Cyp2c44, Cyp2j6, Cyp2j9, Cyp2j11, Ephx1, Ephx2, Pla2g4c, Ptgs2), kidney injury (Ankrd1, Cd14, Havcr1, Tgfbr1), oxidative stress-related (Hmox, Ncf1, Nqo1, Sod2), and housekeeping (Actb, Gusb) gene expression. Raw Ct values for each gene were converted to ΔCt values by subtracting the average Ct of the housekeeping genes from the Ct of the specified gene, and ΔΔCt values for each gene were calculated relative to the VEH/CON group by subtracting the average VEH/CON ΔCt value from individual ΔCt values within all experimental groups. The ΔΔCt values for each gene were then converted to relative copy number (RCN) values using the following equation (68):

All statistical analyses were conducted using GraphPad Prism Version 9 (GraphPad Software, San Diego, CA, www.graphpad.com). Outliers were identified using Grubb’s outlier test (Q = 1%), and normality was assessed using the Shapiro-Wilk test (p < 0.01). Quantitative data that failed to meet the assumption of normality and semi-quantitative data were analyzed by the Kruskal-Wallis nonparametric test followed by Dunn’s post-hoc test. The Brown-Forsythe test (p < 0.01) was used to test the assumption of equal variances across treatment groups. Normal data with unequal variances were analyzed using the Brown-Forsythe/Welch analysis of variance (ANOVA) followed by Dunnett’s T3 post-hoc test. Normal data that met the assumption of equal variance were analyzed by standard one-way ANOVA followed by Tukey’s post-hoc test. Data are presented as mean ± standard error of the mean (SEM), with a p-value < 0.05 considered statistically significant.

In Study 1 ( Figure 1A ), no significant differences in weight change among experimental groups were observed from 8 to 10 wk of age ( Figure 2A ). Beginning at age 10 wk, mice in the R-LPS group began losing weight while the weights of animals in the VEH and S-LPS groups steadily increased. The average combined kidney weight (sum of left kidney and right kidney) approximated to 0.45 g and 0.40 g within the VEH and S-LPS groups, respectively, at experiment termination (age 13 wk), whereas combined kidney weight the in R-LPS group increased to 0.58 g ( Figure 2B ). In line with these findings, mice in the R-LPS group alone began exhibiting proteinuria ( Figure 2C ) and hematuria ( Figure 2D ) after age 10 wk, whereas animals in the VEH and S-LPS groups displayed neither proteinuria nor hematuria at any point during the study. At age 13 wk, trends toward increased blood urea nitrogen (BUN) ( Figure 2E ) and plasma creatinine ( Figure 2F ) were observed in the R-LPS group compared to the VEH and S-LPS groups.

Examination of periodic acid Schiff and hematoxylin (PASH)-stained renal sections and subsequent semi-quantitative scoring revealed minimal to no PAS+ medullary membrane thickening in glomeruli of VEH- and S-LPS-treated mice ( Figures 3A, E, G ). In contrast, kidneys of R-LPS-treated mice contained markedly hypertrophic glomeruli with thickened periodic acid fast-stained medullary membranes, hyalinized proteinaceous material in renal tubular lumens, and mild lymphoplasmacytic infiltrate in cortical interstitial tissue, all of which were indicative of GN ( Figures 3C, G ). Consistent with these findings, immunohistochemical staining indicated that R-LPS but not S-LPS induced glomerular deposition of IgG ( Figures 3B, D, F, H ). In further congruence with histopathology findings, renal tissue from VEH-injected mice exhibited no significant influx of CD45R⁺ B lymphocytes ( Supplementary Figure 1A ) and minimal influx of CD3⁺ T lymphocytes ( Supplementary Figure 1B ). On the other hand, R-LPS-injected mice demonstrated a moderate increase in renal CD45R⁺ lymphoid cell infiltration ( Supplementary Figure 1C ) and a marked increase in CD3⁺ lymphoid cell infiltration ( Supplementary Figure 1D ), while kidney tissues from S-LPS- injected mice resembled those from VEH-injected mice ( Supplementary Figures 1E, F ). CD45R⁺ and CD3⁺ lymphocytes did not localize to any specific region in the kidney. Altogether, blood and urine analyses, histopathology, and immunohistochemistry indicated R-LPS but not S-LPS induced robust GN.

Spleen and liver tissue sections were also histologically evaluated after Study 1 termination ( Figures 4 , 5 ). No histopathology was evident in spleens of VEH-treated control mice ( Figures 4A, B ) that was histologically similar to S-LPS-treated mice ( Figures 4E, F ). Splenic tissue from R-LPS mice ( Figures 4C, D ) had lymphoid cell hyperplasia in white pulp with correspondingly lesser red pulp. Consistent with the expansion of white pulp, the R-LPS group showed a marked average weight increase at 0.30 g compared to the VEH and S-LPS groups at 0.08 g and 0.11g, respectively ( Figure 4G ). Histologic assessment of liver showed periportal large and small hepatocellular vacuoles resembling fatty liver histopathology (steatosis) in VEH-treated control mice ( Figure 5A ). There was marked periportal interstitial lymphoid cell accumulation in R-LPS-treated mice without hepatocellular vacuolization ( Figure 5B ). Histology of liver tissue from S-LPS mice ( Figure 5C ) resembled that of VEH/CON mice. Average liver weights did not significantly change with either R-LPS (1.58 g) or S-LPS (1.41 g) compared to the VEH group (1.28 g) ( Figure 5D ). Accordingly, R-LPS but not S-LPS caused enlargement and lymphoid cell expansion in the spleen as well as modest lymphoid cell recruitment in the liver.

In Study 2, we compared the effects of i.p. injection of R-LPS on GN and related endpoints in mice fed control CON, DHA, TPPU, and TPPU+DHA diets ( Figure 1B ). When total red blood cell fatty acids including saturated fatty acids (SFAs), monounsaturated fatty acids (MUFAs), ω-6 PUFAs, and ω-3 PUFAs were determined by GLC, the seven most abundant fatty acids were palmitic acid (PA, C16:0), stearic acid (SA, C18:0), oleic acid (OA, C18:1ω9), linoleic acid (LA, C18:2ω6), arachidonic acid (ARA, C20:4ω6), EPA (C20:5ω3), and DHA (C22:6ω3) ( Table 2 and Figure 6A ). LPS treatment had no effect on fatty acid profiles of CON-fed mice. Consistent with prior findings (63), we found that substituting high-oleic safflower oil with DHA-rich algal oil in the AIN-93G diet increased incorporation of DHA and EPA into the red blood cell membrane, at the expense of ARA and OA. There was also a slight increase in membrane LA while SA slightly decreased with dietary DHA incorporation. The ω-3 index, or measure of EPA and DHA in relation to total red blood cell fatty acids (69), was elevated in mice that received either DHA or TPPU+DHA diet. TPPU administration alone had no significant effect on total membrane SFAs, MUFA, and PUFAs. Overall, feeding DHA elevated ω-3 PUFAs and decreased total MUFAs and ω-6 PUFAs.

Omega-6 and ω-3 PUFAs act as substrates for CYP450 monooxygenases, which convert the parent PUFA into epoxy-fatty acids (EpFAs). In turn, EpFAs act as substrates for sEH, which converts EpFAs into their corresponding vicinal diols, dihydroxy-fatty acids (DiHFAs). Inclusion of TPPU in experimental diets resulted in presence of 5 to 6 µM of the drug in plasma ( Supplementary Figure 2 ), which is consistent with the TPPU blood concentration obtained from efficacious doses (3 mg/kg/day) in other preclinical studies without reported side effects (41, 70–72). We assessed the impacts of DHA and TPPU on plasma levels of EpFAs, DiHFAs, and other PUFA-derived oxylipins using a comprehensive LC-MS/MS oxylipin panel ( Supplementary Table 6 ). Prominent metabolites included ones derived from LA (i.e., 12,13-EpOME and 12,13-DiHOME) ( Figure 6B ), ARA (i.e., 14,15-EpETrE and 14,15-DiHETrE) ( Figure 6C ), EPA (i.e., 17,18-EpETE and 17,18-DiHETE) ( Figure 6D ), and DHA (i.e., 19,20-EpDPE and 19,20-DiHDPE) ( Figure 6E ). No significant changes were observed between VEH/CON and LPS/CON mice. Consistent with our total red blood cell fatty acid data ( Figure 6A ) and prior reports (73–75), we found that substituting high-oleic safflower oil with DHA-rich algal oil elicited decreases in plasma LA- and ARA-derived EpFAs and DiHFAs and corresponding increases in plasma EPA- and DHA-derived EpFAs and DiHFAs ( Figure 6F ). Increases in DHA-derived metabolites were much more pronounced than those of EPA-derived metabolites. In addition, mice in the LPS/TPPU group exhibited modest increases in LA-, ARA-, EPA- and DHA-derived EpFAs compared to the LPS/CON group, whereas LPS/TPPU displayed a modest decrease in 14,15-DiHETrE but not 17,18-DiHETE and 19,20-DiHDPE. Furthermore, the LPS/TPPU+DHA group displayed modest increases in 14,15-EpETrE and 17,18-EpETE compared to the LPS/TPPU group, although these changes were not statistically significant with the LPS/CON and LPS/DHA groups; 19,20-EpDPE levels were not significantly affected by TPPU. Furthermore, TPPU+DHA co-treatment increased 17,18-DiHETE and 19,20-DiHDPE relative to the TPPU group and caused modest, but not significant, decreases in 14,15-DiHETrE, 17,18-DiHETrE, and 19,20-DiHDPE compared to CON- or DHA-fed mice. In summary, the LPS/TPPU group exhibited significant increases in epoxide/diol ratios for LA-, ARA-, EPA-, and DHA-derived metabolites compared to the LPS/CON group ( Figure 7A and Table 3 ), and the LPS/TPPU+DHA group exhibited significant increases epoxide/diol ratios in EPA- and DHA-derived metabolites compared to the LPS/DHA group ( Figure 7B and Table 3 ).

For the duration of the study, mice in all experimental groups gained weight at similar rates, regardless of dietary intervention ( Supplementary Figure 3 ). During the 5 wk of LPS injections, mice were assessed weekly for development of hematuria and proteinuria as indicators of GN ( Figures 8A, B ). Individuals in the VEH/CON group did not display proteinuria or hematuria. At 10 wk of age (after 6 injections), mice in the LPS/CON group began developing hematuria ( Figure 8A ). At age 11 wk (after 8 injections), mice in the LPS/DHA, LPS/TPPU, and LPS/TPPU+DHA experimental groups began developing hematuria. After the final injection, 75% of animals in the LPS/CON and LPS/TPPU+DHA groups displayed hematuria, 50% of animals in the LPS/TPPU group displayed hematuria, and 38% of animals in the LPS/DHA group displayed hematuria. In similar fashion, mice in the LPS/CON, LPS/DHA, and LPS/TPPU groups began developing proteinuria at age 10 wk (after 6 injections), and mice in the LPS/TPPU+DHA group began developing proteinuria at 11 wk of age (after 8 injections) ( Figure 8B ). After the final injection, proteinuria was evident in 87.5% of the LPS/CON group, 75% of the LPS/TPPU+DHA group, 50% of the LPS/TPPU group, and 38% of the LPS/DHA group. Consistent with GN, the average combined kidney weight was significantly elevated in the LPS/CON group compared to the VEH/CON group ( Figure 8C ). The only group that demonstrated a significant decrease in kidney weight compared to the LPS/CON group was the LPS/TPPU group, while no significant differences were observed for the other groups.

Histologic evaluation and scoring of PASH-stained kidney sections showed no evidence of GN in VEH/CON-treated mice ( Figures 9A, F ). Markedly hypertrophic and hypercellular glomeruli with thickened medullary membranes consistent with GN were observed in kidneys of LPS/CON and LPS/TPPU+DHA mice ( Figures 9B, E, F ), while less glomerular histopathology was evident in LPS/DHA and LPS/TPPU mice ( Figures 9C, D, F ). Consistent with histopathological findings, immunohistochemical evaluation of kidney sections stained with IgG-specific antibody similarly revealed that DHA alone and TPPU alone suppressed R-LPS-induced IgG deposition in the kidney but are antagonistic when delivered together ( Figures 10A–E ).

All H&E-stained splenic tissues from R-LPS- treated groups were enlarged due to lymphoid hyperplasia ( Figures 11A–E ). Splenic tissue from TPPU-fed mice had slightly less lymphoid hyperplasia then other LPS-treated mice ( Figure 11D ). Mean spleen weights were significantly elevated in the LPS/CON group compared to the VEH/CON group ( Figure 11F ). Mice fed DHA and TPPU diet exhibited trends toward reductions in spleen weight that were not statistically significant. Hematoxylin and eosin-stained liver sections were evaluated for histopathology ( Figure 12 ). Periportal hepatocellular vacuolization was prominent in VEH/CON group possibly reflecting steatosis previously reported in NZBWF1 mice ( Figure 12A ). Less hepatocellular vacuolization with marked lymphoid cell infiltration was observed in periportal interstitial tissue in livers of LPS/CON mouse ( Figure 12B ). R-LPS-treated mice fed DHA, TPPU, and TPPU+DHA diet had less periportal inflammatory cells and absence of hepatocellular vacuolization ( Figures 12C–F ). Inflammatory severity scores were suppressed in DHA-fed mice with similar non-significant trends being observed in mice fed TPPU and TPPU+DHA diets ( Figure 12G ).

Plasma from all mice within each experimental group were pooled and subjected to high-throughput autoantigen array for 120 IgG and IgM AAbs. While R-LPS robustly induced total and group-specific IgG and IgM autoantibodies in CON-fed mice, the magnitude of these responses was unaffected by DHA, TPPU, or TPPU+DHA treatments ( Supplementary Figure 4 ).

We evaluated the impacts of DHA and TPPU on expression levels of inflammatory/autoimmune (i.e., Il1b, Ccl2, Ccl7, Cxcl10, Cxcl13, C1qa, C3, Casp1, Tlr9, Tnfa, Tnfsf13b) and fatty acid metabolism genes (i.e., Alox15, Cyp2c44, Cyp2j6, Cyp2j9, Cyp2j11, Ephx1, Ephx2, Pla2g4a, Ptgs2) genes in the kidney ( Supplementary Figure 5A ; Supplementary Table 7 ). R-LPS significantly induced expression of proinflammatory cytokines (i.e., Il1b, Tnfa), chemokines (i.e., Ccl2, Ccl7, Cxcl13), complement proteins (i.e., C1qa, C3), and other related (i.e., Casp1, Tlr9, and Tnfsf13b) genes relative to VEH/CON mice. Although Cxcl10 mRNA was not significantly elevated by R-LPS, it demonstrated a modest increase compared to VEH/CON mice. Intriguingly, none of the selected inflammatory/autoimmune genes were significantly downregulated with DHA or TPPU, though some treatment groups exhibited therapeutic trends. For instance, LPS/DHA mice showed modest decreases in mRNA for Ccl2, Ccl7, Cxcl13, C1qa, Casp1, Tlr9, and Tnfa relative to LPS/CON mice. In addition, LPS/TPPU mice exhibited modest reductions in mRNA for Ccl2, Ccl7, Casp1, and Tnfa. Upon combining DHA and TPPU, the individual inhibitory effects of DHA and TPPU were diminished, with an exception to Tlr9 expression which might be influenced only by DHA.

We found that R-LPS significantly reduced expression of several selected genes involved in lipid metabolite synthesis, including Cyp2c44, Cyp2j6, Cyp2j9, Cyp2j11, and Ephx2, compared to VEH-injected mice ( Supplemental Figure 5B ; Supplementary Table 7 ). In addition, R-LPS modestly downregulated Ephx1, Pla2g4a, and Ptgs2. In line with our observations for the inflammatory/autoimmune genes, we found that neither DHA nor TPPU significantly restored the expression levels of most fatty acid metabolism genes in our panel except for Ephx2, which was significantly upregulated in the LPS/DHA and LPS/TPPU+DHA groups compared to the LPS/CON group.

Other genes measured in our panel are noted in Supplementary Table 7 . As anticipated, we observed significant upregulation of genes associated with kidney injury (i.e., Ankrd1, Lcn2) and oxidative stress (i.e., Hmox, Nqo1, Sod2) in LPS/CON mice relative to VEH/CON mice but found no significant expression level changes in DHA- and/or TPPU-treated mice. No significant changes in gene expression were noted for some type I IFN-regulated genes (i.e., Irf7, Isg15), but Ifi44 expression was significantly reduced in all DHA-fed mice, regardless of TPPU consumption, relative to VEH/CON and LPS/CON mice.