All procedures involving animals were performed according to the German Law on the Protection of Animals/European Communities Council Directive (86/609/EEC) and approved by the Thuringia State Office for Food Safety and Consumer Protection (UKJ-17/036). Mice were maintained under appropriate barrier conditions in a 14/10hr light-dark cycle and received food and water ad libitum according to ARRIVE guidelines. Animals were housed in groups of three in IVC cages with standardized enrichment. At least two animals were housed in one cage and all cages were kept in the same room. To reduce stress, scoring, weighing, and inoculations were performed at the same time whenever possible.

After one initial week of acclimatization, 4-week-old male C57BL/6 wildtype mice were randomly divided into groups (n=12/group) and put on either PA or OA-enriched diets (PA-ED, OA-ED) (21% calories from fat) (Ssniff, Soest, Germany) or a standard diet (ND) (11% calories from fat) for a total of 16 weeks as described before ( Table S1 ) (27). Animals were assigned blinded number codes, which were used throughout the experiment. The PA-OA substitution group received PA-ED for the first 8 weeks and was placed on OA-ED afterward. Weight was monitored weekly, accompanied by scoring procedures. Thereby, we checked for weight, general condition, and general behavior. The health status of the animals was judged according to the appropriate assessment criteria defined prior to the experiments. Humane endpoints were defined based on a scoring system. The number of animals was calculated prior to the outlined experiments using power analysis (see below).

P. gingivalis W50 (33) (American Type Culture Collection (ATCC53978) (Manassas, VA)) was grown in defined medium as described before (33) and administered via oral gavage at week 10 of FA-enriched diet feeding (n=12/group). Oral inoculation was performed as described previously (27). Briefly, animals received 0.1 mg/ml enrofloxacin over their drinking water for 4 days and were infected with 109 cfu P. gingivalis in 50 µl of 2% methylcellulose in PBS three times a week over a period of 5 weeks then compared to the methylcellulose-treated placebo group (‘control’). Animals were inspected daily, with scorings three times a week, paralleling infections, and sacrificed one week after the final infection.

Blood samples were taken via heart puncture. Samples were left clotting for approximately 2 hours at RT before being centrifuged at 5,000 x g for 10 min. Serum was collected from the supernatant and stored at -20°C. ELISA-based quantification of inflammatory proteins (Mouse IL-6 Quantikine ELISA Kit and Mouse TNF,α Quantikine ELISA Kit (both R&D Systems) or bone remodeling markers (Mouse Osteocalcin (OC) and Mouse C Terminal Telopeptide of Type I Collagen (CTX) (all supplied from MyBioSource/Biozol) in serum was performed according to manufacturer’s protocols. Serum levels were determined at an absorbance of 450 nm (Infinite M Nano plate reader, Tecan Life Science) and calculated by comparing values to a standard curve. Each sample was measured in duplicates. Protein concentrations are shown in picograms per milliliter. Each data set contained six individuals.

Manually defleshed femurs were immersion fixed in 4% PFA overnight. Bones were decalcified in 25% EDTA for a total of 2 weeks. Paraffin slices of 4 µm were stained for tartrate-resistant acid phosphatase (TRAP) and counterstained with hematoxylin. Microscopic pictures were taken using an Axio Vert.A1 (Zeiss) using Zen software. Trabecular bone volume was measured below the growth cone using ImageJ and normalized to the tissue area (indicated by green lines in Figure 1 ) as BV/TV. TRAP⁺ cells were counted along the growth cone or a defined part of the corticalis in the diaphysis and normalized to the length. Histomorphometric data were normalized to uninfected ND-controls and shown as fold change. Each data set contained four individuals.

PIs, DAGs, and TAGs were extracted from serum or pulverized bone (34) by the successive addition of PBS pH 7.4, methanol, chloroform, and saline to a final ratio of 14:34:35:17 (35). After evaporation of the organic layer, the lipid film was dissolved in methanol and subjected to UPLC-MS/MS analysis. Lipids were separated on an Acquity UPLC BEH C8 column (1.7 μm, 2.1 × 100 mm) using an Acquity UPLC system (Waters) as described (36). The UPLC system was coupled to a QTRAP 5500 mass spectrometer (Sciex) equipped with an electrospray ionization source. DAGs and TAGs were detected based on transitions of [M + NH₄]⁺ to [M – fatty acid anion]⁺ fragments by multiple reaction monitoring in the positive ion mode (36). PI species were analyzed in the negative ion mode by monitoring the transitions of [M-H]^- to both fatty acid anions (36). Relative intensities are defined as proportions of individual lipid signals (e.g., DAG (16:0/16:0)) relative to total signal intensity for a given lipid class (e.g., total DAG). Mass spectra were processed using Analyst 1.6 software (Sciex). Analysis was performed with six animals per group.

Ceramide content in pulverized bone and bone marrow cells was determined using liquid chromatography coupled with triple-quadrupole mass spectrometry, as previously described (37). Pulverized bone was supplied with 1 mL H₂O, 300 µL 6 M HCl, 1 mL methanol, 2 mL CHCl_3, and 10 µL of 30 µM of the internal standard N-pentadecanoyl-D-erythro-sphingosine (C15-Cer, Merck KGaA Darmstadt, Germany) in a glass centrifuge tube. The suspension was vigorously vortexed for 60 min. After centrifugation at 1900 x g for 5 min, the lower CHCl₃ phase was transferred into a new glass centrifuge tube. After the addition of another 2 mL CHCl₃ to the original tube, the suspension was vortexed again for 5 min and centrifuged at 1900 x g for 5 min. The CHCl₃ phases were combined, evaporated for 50 min at 50°C in a vacuum concentrator (Christ, Osterode, Germany), and resuspended in 100 µL of a mixture of CHCl₃ and methanol (1:4 v/v). Pelleted bone marrow cells were resuspended in 200 µL methanol after the addition of 10 µL of 30 µM of the internal standard N-pentadecanoyl-D-erythro-sphingosine (C15-Cer, Merck), vigorously vortexed for 10 min, stored at -80°C overnight, and centrifuged at 17000 x g for 5 min. The upper methanol phase was collected. Ceramides were separated using the Prominence high-performance liquid chromatography (HPLC) system consisting of the controller CBM-20A, autosampler SIL-20AC, column oven CTO-20A, pump LC-20AD, and degasser LC-20A5R (Shimadzu, Duisburg, Germany). Then, 30 µL of the samples were applied on a MultoHigh 100 RP18-3 60x2 mm column (CS Chromatographie Service GmbH, Langerwehe, Germany) maintained at 35°C under the conditions detailed in Table S2 . Mass spectrometric determination of ceramides was performed using the QTrap triple quadrupole mass spectrometer (AB Sciex, Darmstadt, Germany) run in the multiple reaction monitoring (MRM) modus using positive atmospheric pressure chemical ionization (APCI) at 450°C. Detected single charged masses and their fragments are shown in Table S3 .

Primary osteoblasts were isolated from long bones of the fore and hind limbs of 4-week-old C57Bl6 mice. After the removal of bone marrow, the bones were washed in PBS and cut into small pieces prior to collagenase digestion (500 U/ml Collagenase II in DMEM) for 2 h at 37°C. Bone fragments were washed three times in proliferation medium (DMEM low glucose (Thermo-Fisher Scientific), 100 U/ml penicillin, 100 µg/ml streptomycin (Thermo-Fisher Scientific), 50 µg/ml Gentamycin (Sigma-Aldrich), and 1.25 µg/ml Fungizone (Sigma-Aldrich)) and transferred to a T25-flask with 5 ml proliferation medium. The medium was changed for the first time after 5 days of incubation. Subsequent media changes were performed three times a week until cells reached confluency.

Cells were seeded in a density of 2.8*10⁴ cells per well on 24-well plates. After 2 days in the culture, the cells reached approximately 80% confluency and the proliferation medium was exchanged for differentiation medium (proliferation medium supplemented with 100 µg/ml L-ascorbic acid, 10 nM β-glycerol phosphate, and 100 nM dexamethasone). The following day, fatty acids (FA) were added to the differentiation medium as described elsewhere (26, 38). Briefly, OA or PA was solubilized in sterile water supplemented with 50 mM NaOH at 70°C and added to prewarmed BSA (H2B) (37°C) in a 3:1 molar ratio before being transferred to the culture medium. The final FA concentration was adjusted to 0.2 mM. Medium supplemented with BSA only was used as a control. For P. gingivalis infection experiments, cells were cultured for a total of 6 days in differentiation medium, with one medium change after 3 days. At that time, the PA-OA group was switched from PA to OA. Prior to infection, osteoblasts were washed twice with prewarmed D-PBS (PBS without calcium, magnesium, Thermo-Fisher Scientific) and covered with 1 ml antibiotics-free culture medium (OB – DMEM low glucose with pyruvate supplemented with 10% FBS (Thermo-Fisher Scientific)). Cells were infected with a multiplicity of infection (moi) of 100 cfu in 25 µl PBS. Control cells were supplemented with 25 µl PBS. Soluble IL-6 antibody (sIL-6ab; Ultra-LEAF™ Purified anti-mouse IL-6; BioLegend) was added in a final concentration of 10 µg/ml). Application of IL-6 receptor (IL-6R; Recombinant Mouse IL-6R alpha (aa20-357) Protein, R&D Systems) was done in a concentration of 10 ng/ml (39). Cells were incubated for a total of 6 h, as described before (26). Supernatants were collected, centrifuged, and stored at -20°C until further ELISA analysis. To analyze osteoblast differentiation, cells were incubated in FA-supplemented differentiation media for a total of 14 days, with one group being changed from PA to OA treatment after 7 days. The cell culture medium was changed every 2 to 3 days.

Osteoclasts were generated from the bone marrow of 4-week-old C57Bl6 mice. As described previously (40), bone marrow cells (BMC) were centrifuged out of the long bones of the front and hind limbs (10,000 x g for 20s). BMC were resuspended in proliferation medium (αMEM containing 10% FBS (Gibco) and 1% penicillin/streptomycin) and incubated overnight in 10 ml proliferation medium on a 10 cm dish (one per animal). For treatment with conditioned osteoblast media, non-adherent cells were seeded in a density of 3*10⁵ cells per well in 600 µl proliferation medium 50 ng/ml M-CSF (PeproTech) on 24-well plates. The following day, 80% of the medium was replaced by prewarmed cell culture supernatant from murine osteoblasts (FA- and PenStrep-free DMEM including 10% FBS after a 6 h secretion period) supplemented with 50 ng/ml M-CSF. sIL-6ab and/or IL-6R were added for the complete duration of supernatant incubation in a final concentration of 10 µg/ml (sIL-6ab) or 10 ng/ml (IL-6R), respectively. The cells were incubated for 2 more days prior to RNA extraction. At least three independent experiments were performed and each sample was measured in duplicate.

RNA was isolated with TRIzol Reagent (Thermo Fisher Scientific, Carlsbad, CA, USA)/1-bromo-3-chloropropane. Purification was performed using RNA Clean and Concentrator-5 kit (Zymo Research, Freiburg, Germany) according to the manufacturer’s instructions. The concentration and purity of RNA were measured with Nanodrop 2000 (Avantor, Radnor, PA, USA). RNA was transcribed to cDNA with SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific, Carlsbad, CA, USA) according to the manufacturer’s protocol. For quantitative RT-PCR gene expression levels, Luminaris HiGreen Master Mix (Thermo Fisher Scientific) was used and monitored by qTOWER3 (Analytic JENA). Gapdh (glyceraldehyde 3-phosphate dehydrogenase) and Rps29 (40S ribosomal protein S29) served as respective reference genes. Sequences of all primers are depicted in Table S4 . Each primer pair was tested with template dilution series for the calculation of efficiency followed by the analysis of melting curve and agarose gel electrophoresis to exclude primer dimers. Analysis of data was performed using the efficiency corrected ΔΔCT-method.

Protein detection was performed using Mouse IL-6 Quantikine ELISA Kit (R&D Systems) according to the manufacturer’s protocol. Secreted IL-6 levels were calculated by comparison to a standard curve at an absorbance of 450 nm (Infinite M Nano plate reader, Tecan Life Science). At least four independent experiments were performed and each sample was measured in duplicate. Values were normalized to BSA-treated controls.

Osteoblast mineralization activity was determined using an Alizarin Red Assay (Sigma-Aldrich) according to the manufacturer’s recommendations. Briefly, at day 14, cells were fixed for 15 minutes in 10% PFA and incubated with 40 mM Alizarin Red for 20 minutes. Free Alizarin Red was washed away with water prior to cell lysis using 10% acetic acid for 30 minutes. Afterward, the suspension was heated to 85°C for 10 min and allowed to cool down for 5 min on ice. Thereafter, samples were centrifuged for 15 min at 20,000 x g and the pH of the supernatant was neutralized with 10% ammonium hydroxide. Values were determined at OD405 (Infinite M Nano plate reader, Tecan Life Science). Concentrations were determined by comparing values to a standard curve and normalized to BSA-treated controls. The experiment was repeated three times with two wells per condition.

The primary objective was to study the presence of specific lipids and bone loss in mice. Based on our previous results (26, 38), we calculated a sample size of less than six animals in each group to have a statistical power of more than 80% to detect a difference in bone loss with a significance level of 0.05 (two-tailed, STAT MAT 2.0). If not stated differently, six animals per group were analyzed.

Comparisons within treatment groups (infected and non-infected) were analyzed using ANOVA and a post hoc test (Tukey) followed by a two-tailed Student’s t-test for pairwise comparisons. Analysis for differences was performed between nutrition groups (ND vs PA vs OA vs PA-OA) and infection groups vs their respective mock-infected controls. All values are depicted as mean ± SE. For the exclusion of single data points, we performed Grubb’s outlier test. We did not exclude any data points from animal experiments. Differences between groups were considered statistically significant at P < 0.05. Analyses were performed in GraphPad/InStat3 (GraphPad Software Inc., San Diego, CA, USA).

To examine whether the nutritional intake of specific FAs mirrors previous oral findings, showing that PA compromises PDL-lining bone, mice were fed normal chow (ND), palmitic (PA)- or oleic acid (OA)-enriched diets (PA-ED, OA-ED). Additionally, a PA-to-OA dietary substitution group (PA/OA-ED) was included. All final weights were independent of dietary intake within the normal-weight range ( Figure 1A ). Histomorphometric analysis of TRAP-stained femur sections revealed a 38% reduced trabecular bone volume over total volume (BV/TV) in PA-ED groups compared to OA-ED groups ( Figures 1B, C ). Further, PA-ED reduced BV/TV compared to ND by approximately 30% at a p=0.08 trend. OA-ED substitution of PA-ED resulted in trabecular BV/TV not different from OA-ED and comparable to ND. Although serological IL-6- and TNFα concentrations were below inflammatory levels (41) ( Figures 1D, E ), PA-ED-associated trabecular deterioration was paralleled by significantly reduced levels of the stress-limiting lipokine PI(18:1/18:1) compared to OA-ED at the time of analysis ( Figure 1F ). Moreover, dietary substitution of PA with OA-ED significantly increased PI(18:1(18:1) compared to ND and PA-ED.

To investigate whether the PA-dependent reduction and substitutional-dependent rescue in BV/TV correlate with global changes in the lipid fatty acid profile, we performed lipidomics studies on TAGs, DAGs, and ceramides from the femoral bone. PA-ED strongly elevated the relative proportion of DAG species carrying 16:0 as well as the ratio of TAG(16:0/16:0/16:0) ( Figure 2A ). The latter remained at a high level, even after nutritional OA-replacement ( Figure 2C ), whereas the accumulation of 16:0-containing DAGs decreased ( Figures 2A, B ). In consequence, substitution of PA with OA adopts a DAG/TAG profile that is more similar to OA-ED than to PA-ED, except for the fully saturated TAG(16:0/16:0/16:0).

Next, hard bone tissue and bone marrow were analyzed for PA-derived lipotoxic ceramides ( Figure 2D ). Specifically, incorporations of saturated versus unsaturated ceramides and Cer16 levels were quantified. Whereas ceramide subgroups in hard tissue were comparable, significantly more saturated ceramides and especially Cer16-ceramides (Cer18:1/16:0) were incorporated in the bone marrow of PA-ED compared to ND ( Figures 2D, E ). Overall, unsaturated ceramide levels were comparable in all FA groups. However, OA-ED by itself and the replacement of PA-ED with nutritional OA significantly lowered saturated ceramides and Cer16-content compared to PA-ED in bone marrow ( Figure 2F ).

We next investigated how specific FAs impact bone cell interaction to underline morphological findings. Osteoblasts from long bones were differentiated in the presence of FAs for 14 days. PA significantly reduced mineralization activity, whereas OA did not. Substitution of PA with OA after 7 days alleviated the impact of PA by restoring mineralization activity ( Figure 3A ).

Expression levels of osteoblastic differentiation marker (Runx2) or osteoclastic differentiation factors (Rankl, Tnfα) during differentiation were comparable in all groups ( Figure 3B ). However, summative IL-6 secretion during 6 hours in FA-free media after 6 days of FA-influenced differentiation was significantly enhanced in PA-cultures compared to BSA- and OA-treated cells. Change from PA- to OA-charged medium reduced IL-6 secretion significantly, resulting in levels not different from BSA- or OA-treated cells ( Figure 3C ). Neither the addition of soluble IL-6-receptor (sIL-6R) nor IL-6 scavenger antibody (IL-6ab) impacted the expression of osteoblastic and osteoclastic differentiation factors during this time frame ( Figure 3D ).

To determine the FA-specific influence on osteoblast-osteoclast interaction, we challenged pre-osteoclasts for 2 days with M-CSF-supplied FA-incubated osteoblastic culture supernatant (OB-conditioned DMEM with 10% FBS; OB/BSA, OB/PA, OB/OA or OB/PA/OA) ( Figure 3E ) and quantified the expression of the osteoclastic differentiation marker Trap. Whereas Trap expression of OB/OA-treated BMCs was comparable to controls (OB/BSA), OB/PA-treated cells exhibited a significant ~13-fold increase compared to controls. PA replacement with OA (OB/PA/OA) directly reduced Trap-expression levels that were now comparable to controls and OB/OA ( Figure 3F ). To verify IL-6-dependency in this context, BMCs were incubated with OB/PA-cell culture supernatant in the presence of IL-6ab, sIL-6R, or both, which significantly decreased Trap-expression levels in all combinations ( Figure 3G ).

We next evaluated how osteoblasts respond to P. gingivalis infection under the influence of FAs. P. gingivalis incubation of FA-pretreated OBs significantly enhanced IL-6-secretion ( Figure 4A ), especially PA-stimulated secretion, compared to pretreatment conditions. After 6 days of differentiation in FA-supplied media, differentiation marker Runx2 levels were similar within FA groups and not modified by P. gingivalis infection compared to uninfected controls ( Figure 4B , compare with Figure 3B for uninfected controls). Although not modified by different FA incubation, P. gingivalis infection significantly increased the expression of osteoclastic stimulators Tnfα and Rankl in almost all groups compared to the respective controls (Tnfα: BSA/ctrl vs P.g.: ^##; PA/ctrl vs P.g.: ^####; OA/ctrl vs P.g.: ^####; PA/OA/ctrl vs P.g.: ^##; Rankl: BSA/ctrl vs P.g.: ^##; PA/ctrl vs P.g.:^#; OA/ctrl vs P.g.: ^####; PA/OA/ctrl vs P.g.: ns);. When M-CSF-stimulated BMCs were incubated with osteoblastic supernatant of OB/BSA- and OB/OA-treated infected and placebo-infected cells, Trap-expression increased significantly in infection groups ( Figure 4C ). Notably, the expression pattern of OB/PA/OA/P.g.-incubated cells mirrored BSA and OA-conditions ( Figure 4C ). Trap-expression of OB/PA-supernatant challenged cells was already elevated and did not increase further in response to infection. Furthermore, sIL-6R alone or in combination with IL-6ab did not change Trap-expression in PA cultures, indicating that at this time point, the process was P. gingivalis-stimulated IL-6-independent, opposite to sole PA-incubation ( Figure 4D ).

To determine whether dietary FAs modify the systemic impact of an oral P. gingivalis inoculation, mice received ND, PA-ED, or OA-ED for 16 weeks. One additional PA-ED group was placed on OA-ED after 8 weeks. Animals received oral P. gingivalis or placebo inoculation three times a week for 5 weeks. One week after final inoculation, circulating serum IL-6 and TNFα levels were low in all groups ( Figures 5A, B ). Furthermore, the serological bone resorption marker CTX and the formation marker osteocalcin (Ocn) were not substantially altered across treatment groups ( Figure 5C ). In line with this, inoculation groups did not exhibit additional bone loss compared to the respective control groups ( Figure 5D ). However, P. gingivalis infection significantly elevated the proportion of 16:0-containing DAGs in the bone of ND/P.g. animals compared to ND/Ctrl. Moreover, we detected significantly increased ratios of 16:0 in DAG for PA-ED/P.g. compared to OA-ED/P.g.-treated animals ( Figures 5E, F ). Notably, serum levels of the stress-protective lipokine PI(18:1/18:1) significantly decreased upon treatment with PA-ED/P.g. but not by OA-ED/P.g. ( Figure 5G ). Switching from PA to OA substitution effectively restored the proportion of PI(18:1/18:1) within 8 weeks ( Figure 5G ). Based on these findings, we investigated the cellular presence of osteoblasts and osteoclasts in femoral bone. P. gingivalis inoculation did not modulate the abundance of osteocalcin-positive (Ocn⁺) cells in dietary groups (data not shown), whereas the presence of osteoclastic TRAP⁺-cells was FA-specifically modulated ( Figure 5H ). While oral periodontal inoculation in ND animals did not cause any differences between ND-control (ctrl) and inoculation (P.g.)groups ( Figure 5H ; panel 1; ND), animals on PA-ED/P.g. ( Figure 5H ; panel 2; PA) presented with significantly more TRAP⁺ cells compared to PA-ED/ctrl. In contrast, OA-ED/P.g. animals ( Figure 5H ; panel 3; OA-ED) showed lower amounts of osteoclasts than OA/ctrl. Further, dietary replacement of PA-ED to OA-ED reversed PA-lipotoxic impact, leading to TRAP-staining patterns that were comparable to controls and independent of oral inoculation ( Figure 5H ; panel 4; PA/OA).