The experiments conducted in this study complied with animal welfare regulations and were approved by the local governmental authorities, specifically the North Rhine-Westphalian Ministry for Environment, Health and Consumer Protection, Department for Animal Protection (81-02.04.2020.A326). The animals were provided by Charles River GmbH (Germany). Adult male mice from the wild type inbred mouse strain C57BL/6J, aged 12 weeks, were kept under standard laboratory conditions in the Central Animal Experimental Facility (ZTE), University Hospital Muenster. Animals were randomized into two groups receiving the following dietary interventions: The standard chow of the control group (n = 19) was C1000 (Altromin Spezialfutter GmbH & Co. KG) which interalia contained sunflower oil. For the treatment group (n = 18), the 11.1 g sunflower oil per kilogram of the C1000 diet was replaced by Tonalin^® TG80 (with CLA isomers: cis-9,trans-11 and trans-10,cis-12 at a share of 40% each, BASF/BTC-Europe GmbH, manufactured by Altromin Spezialfutter GmbH & Co. KG). The modified CLA-enriched chow (containing 0.8888% CLA) was fed from day of the photothrombotic stroke until the end of the experiment. Additional sham experiments were performed in which except for laser irradiation the exact same procedure as outlined in the following with injection of the same dose of Bengal Rose was performed. Each sham group consisted of 5 animals, sham procedure was performed according to Li et al. (12). The experimental design was fully randomized (lottery procedure).

To induce photothrombotic stroke, the mice received Meloxicam (0.2 mg/kg bodyweight) as an analgesic 30 min prior to the procedure. Thereafter, mice were anesthetized using intraperitoneal ketamine hydrochloride (100 mg/kg body weight) and xylazine (10 mg/kg body weight) and kept at a constant body temperature of 37°C ± 0.5°C throughout the procedure. The skull was exposed following a dorsal scalp incision, and a laser beam was positioned 2 mm right of the bregma. The skull was irradiated by a continuous-wave laser (CoboltJiveTM75, Cobolt AB, Solna; Sweden) with a wavelength of 561 nm, an intensity of 30–34 mW, and an aperture of 4 mm for 20 min after intraperitoneal injection of 0.2 mL Rose Bengal (10 mg/mL).

Neurological deficit scores were assessed using a previously described modified version of Menzies’ neuroscore (13), which ranges from 0 (no deficit) to 5 (death).

To assess sensorimotor deficits, the adhesive tape removal and the foot fault test were performed as previously described (14). The adhesive tape removal test was conducted as follows: For bilateral sensory stimulation, two adhesive dots (25 mm²) were placed on both palmar forepaws, and the time taken by the animals to remove the adhesive dots was recorded in three trials per day for each forepaw. The course over the experiment is shown by the latency in seconds to remove the adhesive dot from the paretic (left) paw and the healthy (right) paw. For analysis, the course of the latency of the left paw was respected.

To compare sensorimotor deficits, the foot fault test was employed. Animals were individually placed on an elevated 10 mm square wire mesh with a total grid area of 40 cm × 40 cm and were allowed to walk freely for 2 min while being videotaped. An investigator counted the number of foot faults and the total number of steps and calculated the ratio using the following formula: (number of foot faults/number of total steps) × 100. Both tests were performed at baseline, day 1 and then once per week.

The adhesive tape test was practised (once per day for three passes) during the acclimatisation period of 1 week before induction of the photothrombotic stroke. Thereby the animals got used to the laboratory facility and the test procedure and potential confounders due to stress was reduced. No animals were excluded at this stage.

At 90 days post stroke, ex vivo diffusion-weighted magnetic resonance imaging (dMRI) allowed for whole-brain fiber tracking and tissue structure analysis through diffusion metrics. Brains in skulls were fixed in 4% para-formaldehyde (PFA) for 2–3 days, then transferred to 1% PFA doped with 2 mM Magnevist (Gd-DTPA, Bayer, Leverkusen, Germany) for additional 2–3 days with daily medium change. Whole skulls were embedded in 1% low-melting agarose (Sigma Aldrich, St. Louis, United States) doped with 2 mM Magnevist in 5 mL syringes. Samples were stored at 4°C before and after MRI. MR images were acquired at room temperature using a 9.4 T horizontal animal scanner with 20 cm bore (Bruker 94/20 USR BioSpec, Avance III, Bruker BioSpin, Ettlingen, Germany). The system is equipped with a 700 mT/m gradient system (BGA12S, BrukerBioSpin) and a 2-element cryogenic transceive coil (Bruker BioSpin) for image acquisition. The system is operated by ParaVision 6.0.1 (Bruker BioSpin). Images were acquired using a multishot 3D spin-echo EPI sequence with the following scan parameters: echo time = 24.5 ms, repetition time = 300 ms, 8 segments, bandwidth = 250 kHz, 100 × 100 × 100 μm³ resolution, gradient duration = 4.5 ms, gradient separation = 11 ms, 80 diffusion directions, b-values = 1,500 and 3,000 s/mm².

Restricted diffusion was quantified using restricted diffusion imaging (15). Diffusion data were reconstructed using generalized q-sampling imaging (16) with a diffusion sampling length ratio of 1.4. A deterministic fiber tracking algorithm was used (17). Interhemispheric connectivity was assessed by manually placing a seed region at the midline corpus callosum (delineated by dominant left-right diffusion anisotropy, 50,000 seeds). The anisotropy threshold was 0.03, angular threshold 60°. The step size was 0.03 mm. Tracks with length shorter than 1 or longer than 20 mm were discarded. Lesion volumetry was performed on the dMRI scans from a manual outline of cortical voxels with signal voids or aberrant anisotropy vectors.

Fitting of diffusion metrics and fiber tracking were performed using DSI Studio.¹

Ninety days after photothrombotic stroke, mice were perfused through the left ventricle with phosphate buffered saline for 5 min and 4% paraformaldehyde for 10 min under deep xylazine/ketamine anaesthesia. For immunohistochemistry, brains were removed, fixed in 4% paraformaldehyde overnight, immersed in 30% sucrose for 3 days, frozen and stored at −80°C.

Mounted coronal cryosections were incubated in blocking reagent (Roche Diagnostics) for 15 min to prevent unspecific protein binding. Subsequently, we used the following primary antibodies: anti-Laminin (Abcam, ab11575) with Alexa Fluor 549 donkey anti-rabbit (Life, A-21207) and Alexa Fluor 488 goat anti-mouse-IgG (heavy and light chain) (Life, A-11001). To reduce lipofuscin autofluorescence, we applied True Black^® Plus Lipofuscin (1:40 in DMSO, Biotium, 23014) for 10 min. Images were taken with a fluorescence microscope (Nikon Eclipse 80i, Nikon GmbH, Düsseldorf, Germany). For quantification of extravasated IgG, 8-bit fluorescence images were captured using a Nikon Eclipse 80i fluorescence microscope. Pixels between grey values 8–255 were considered positive and summed for each animal.

After induction of deep narcosis by overdosing anaesthesia, blood was taken from the right ventricle and transferred into a tube with 500 μL PBS (phosphate buffered saline)-heparine (4°C). The lymph nodes were taken up onto a plate with HBSS (Hank’s balanced salt solution) (4°C), the small intestine (without Peyer-patches) as well as the colon on a plate with HBSS (at room temperature). The brains were transferred into a 15 mL falcon filled with PBS (4°C).

Isolation of brain lymphocytes was performed in a sterile manner by means of Percoll^™ density gradient centrifugation. Briefly, fresh collagenase (1:125) was added first to disintegrate the tissue/extracellular matrix. Subsequently, Percoll^™ was added stepwise under repetitive centrifugation steps to obtain a continuous density gradient and the respective phase containing the brain lymphocytes were carefully pipetted off.

Cervical and mesenterial lymphocytes were isolated under sterile conditions by utilizing different cell strainer sizes to dissociate the lymphocytes from cells and tissue fragments of larger size. Cervical lymph nodes were transferred to a 70 μm plastic cell strainer and mesenteric lymph nodes were carefully homogenised by a metal strainer. Thereafter, strainers were washed with MACS buffer and after centrifugation, the pellet was resuspended in TexMACS^™ medium (Milteny Biotec). Isolation of the lymphocytes of the lamina propria was performed by using the Lamina Propria Dissociation kit (mouse) from Milteny Biotec via gentleMACS^™ dissociator following manufacturer’s instructions. Briefly, enzymatic digestion and solution is performed by combining different enzymatic and mechanical dissociation steps. Immediately, lymphocytes were isolated by MACS^® MicroBead separation. Blood lymphocytes were isolated by differential centrifugation and treatment with hypotonic eBioscience^™ 1× RBC (red blood cell) lysis buffer (Thermo Fisher). Isolated lymphocytes were resuspended in TexMACS^™ medium (Milteny Biotec).

Restimulation of all cells for cytokine analysis was performed by adding a mix of phorbol myristate acetate (PMA) and ionomycine as well as Brefeldin A (Golgi-Plug^™) to block intracellular transportation processes for analysis. The following antibodies were used in a volume of 50 μL in the referred dilutions (if not stated otherwise, antibodies were manufactured by BioLegend): mLy-6G FITC (1:300, clone: 1A8), mCD64-PE (1:200, clone: X54-5/7.1), mLy-6C PC5.5 (1:300, clone: HK1.4, eBioscience), mCX3CR1 PECy7 (1:200, clone SA011F11), F4/80 APC (1:100, BM8), mCD11c AF700 (1:200, N418), m/hCD11b BV510 (1:300, M1/70), mCD45 APC/Cy7 (1:300, clone: 30-F11), m/hCD11b BV510 (1:300, clone M1/70), mCD3 APC-Cy7 (1:200, 17A2), mCD25 BV510 (1:200, PC61), mCD4 BV605 (1:200, clone: GK1.5), mCD8a BV650 (1:200, 53-6.7). Intracellular stainings were performed using the Cytofix/Cytoperm Plus Kit^™ (BD) following the manufacturer’s protocol. The following antibodies were used: mIL-17A BV421 (1:100, clone: TC11-18H10.1), mIFN-y APC (1:100, XMG1.2), mIL-6 PE (1:100, MP5-20F3), mTNF-α AF700 (1:100, MP6-XT22), mFoxP3 AF700 (1:200, FJK-16s, eBioscience^™), mGM-CSF APC-Fire750 (1:100, clone MP1-22E9). The measurements were performed on a CytoFLEX^™ flow cytometer (both Beckman Coulter Inc.) and data analysis was conducted with FlowJo v10.8.1 Software (Becton Dickinson GmbH). Gating strategies are depicted in Supplementary Figures S1–S3.

Eighteen millilitres blood were obtained from six stroke patients and six healthy controls. Stroke patients had a median age of 62.5 (IQR_25–75 = 53.75–86.25) years and a median NIHSS of 3 (IQR_25–75 = 0.75–12.75). The median age of the healthy controls was 32.5 (IQR_25–75 = 29.5–47.5) years. One patient suffered from infection, but as his cytokine levels did not differ significantly from the other patients as tested via robust outlier (ROUT) method based on the false discovery rate (18), we did not exclude this patient from the analysis. Half of the stroke patients received intravenous thrombolysis, two others received mechanical thrombectomy. peripheral blood mononuclear cells (PBMC) were isolated applying density gradient centrifugation. After measuring the cell count (cell counter CASY, OLS omni life science), cells were resuspended in a calculated volume of freezing medium to obtain a concentration of 1 × 10⁷ cells/mL. The samples were stored under liquid nitrogen until further processing.

After isolation of PBMCs, CD4⁺ T-lymphocytes were isolated using the CD4⁺ T Cell Isolation Kit^™, human and the MidiMACS^™ cell separation system (Miltenyi Biotec, today BD Bioscience) following the manufacturer’s recommendations. The cells were adjusted to a concentration of 1 × 10⁶ cells/mL TheraPEAK^™ X-VIVO^™-15 Serum-free Hematopoietic Cell Medium (Lonza). A 96 well U bottom plate was coated with 5 mg/mL anti-CD3 (1:395, clone: OKT3, BioLegend) and anti-CD28 (1:250, clone, CD28.2, BioLegend) for 1 h at 37°C. After washing the plates 3 times with PBS, 6 wells per donor were seeded with 200,000 CD4⁺ T cells (in a volume of 200 μL X-VIVO medium). Three wells were additionally incubated with 50 μL CLA-mix (in pure form dissolved in ethanol to avoid contamination of oils) containing 50 mM cis-9,trans-11 CLA-isomer and 50 mM trans-10,cis-12 CLA-isomer (Sigma-Aldrich) for 72 h at 37°C. After 72 h, the supernatant was collected and analyzed for IL-17A, IFN-γ, TNF-α and IL-6 via Luminex xMAP technology (Thermo Fisher Scientific) following manufacturer’s protocol.

The study on human blood samples was approved by the ethical committee of the Medical Association Westphalia Lippe and the University of Münster (AZ 2018-193-f-S).

Statistical analyses were performed using GraphPad Prism version 10 (GraphPad Software, La Jolla, CA, United States) and R version 4.1.0 (R Foundation for Statistical Computing, Vienna, Austria). Automatic outlier analysis was performed by ROUT analysis in Prism as outlined above. All tests were performed at an α = 0.05.

Statistical tests were selected based on the results of normality testing (Shapiro–Wilk test) and equality of variances testing (F-test): specifically, independent samples t-tests were used when data met assumptions of normality and equal variances, whereas Mann–Whitney U tests were employed in cases where these assumptions were not met. p-value <0.05 was considered significant.

For the functional outcome assessed by the adhesive tape test, a one-sided t-test was calculated to investigate whether the difference between day 1 and 84 in the left paw was greater for CLA than for Ctrl (the difference was normally distributed). For the functional outcome assessed by the foot fault test between the CLA-and the standard diet were compared by repeated measures two-way-ANOVA with post hoc Tukey’s correction. Weight course was compared by repeated measures two-way-ANOVA with post hoc Śídák’s correction. Effects of CLA versus standard diet on corpus callosum metrics were investigated separately for animals with infarct sizes smaller and larger than the median infarct size using Mann–Whitney U tests. Statistical analysis of the flow cytometry data was performed with R version 4.2.2. Different cell types were compared between the two groups (CLA and control) across multiple tissues (blood, brain, LNc, LNm, siLP, and colP). To account for multiple comparisons in the flow cytometry analysis, the Benjamini–Hochberg (BH) procedure was applied to adjust the p-values. Analysis human blood samples were performed with paired mixed ANOVA and post hoc Śídák’s multiple comparison test.

Group sizes were calculated using the Sample Size Calculator software (Department of Statistics, University of British Columbia, http://www.stat.ubc.ca/~rollin/stats/ssize/n2.html). Behavioural data and histological examinations from previously conducted experiments were considered. In preliminary work, a functional, neurological deficit in the foot fault test of 16 ± 4% missteps was determined. With an expected reduction of the deficit to 13% following training therapy (approx. 20% reduction) and assuming a significance level of α = 0.05, this results in a test strength (1 − β) of 0.8 (80%) with a number of animals of n = 28 per group examined.

For our primary endpoint, the functional neurological deficit assessed by the foot fault test and the adhesive tape test, we used a total number of 37 animals (CLA group with photothrombotic stroke, n = 19; control group with photothrombotic stroke, n = 18). The secondary endpoints consisted of structural changes (MRI analysis; CLA group with photothrombotic stroke, n = 6; control group with photothrombotic stroke, n = 6) and immunological alterations (CLA group with photothrombotic stroke, n = 8; control group with photothrombotic stroke, n = 8). For our exploratory analysis, we did not exhaust the total number of animals per group precalculated which represents a potential limitation of our study.

This study followed the ARRIVE guidelines. The graphics were created using CorelDRAW Home & Student X8.

To assess functional recovery, we performed the adhesive tape removal test reflecting sensorimotor deficits by asymmetric forelimb use, as detailed in the methods section. The course of removal of the adhesive dot from the paretic paw was not different from the course in the standard diet group after 90 days (Figure 1A). Additionally, we analyzed the difference between day 1 and day 84 in the left paw. The difference was non-significantly higher with an average effect size based on Cohen’s d in the CLA-group, indicating greater improvement (55.30 vs. 46.71 s; p = 0.266, d = 0.21). The foot fault course between the CLA and the control group subjected to photothrombotic stroke did not differ significantly (Figure 1B). The increasing course of weight of the animals in the group fed a CLA-rich diet was significantly lower than the one in the standard diet group over the 90 days (two-way ANOVA for animals subjected to photothrombotic stroke, p adj. = 0.031; Figure 1C). Livers were significantly enlarged in the CLA-group (1.82 vs. 3.54 g, p < 0.0001, Figure 1D left) and consequently accounted for a higher percentage of body weight (5.64 vs. 12.32%, p < 0.0001; Figure 1D right). We observed this in the sham animals as well, so liver weight was independently increased from photothrombotic stroke (1.63 vs. 2.68 g, p = 0.005, Figure 1D left) and accounted for a higher percentage of body weight as well (4.86 vs. 8.86%, p = 0.004, Figure 1D left) (Supplemental Table 1).

To investigate potential structural correlates of functional improvement after CLA supplementation, we performed deterministic fiber tracking via diffusion MRI measurements. We chose the corpus callosum (CC) as region of interest as it represents a brain region tightly packed with myelinated axons and thus is especially vulnerable to ischemia (11, 19, 20). The white matter repair is one critical element of long-term stroke recovery (21). The stroke lesion in the left somatosensory cortex shows deformed corpus callosum and the lateral ventricle in all images (Figures 2A-J). The diffusion sum (Figure 2B) is reduced in the area of the lesion, corresponding to low anisotropy values (Figures 2C,D) albeit at relatively high mean diffusivity (Figure 2E). High mean diffusivity is most likely caused by the inclusion of embedding agar or residual fluid, as this area also shows high axial (Figure 2F) and radial diffusivities (Figure 2G). The corpus callosum is characterized by low radial diffusivity and medium-to-high axial diffusivity, leading to prominent diffusion anisotropy. Consequently, direction-coded colour maps clearly highlight the corpus callosum as a subcortical structure with dominant left-right diffusion anisotropy (red in panel H; directions were derived from nQA, Figure 2H) ventral to the cingulum and other supra-cortical white matter that run anterior-posterior in each respective hemisphere (green) (22).

As we visually had the impression of differential effects of CLA depending on infarct size, we compared animals with small infarcts (<median = 1.53 mm²) and large infarcts (>median = 1.53 mm²). For small infarcts, we found significantly improved microstructural regeneration among animals that received a CLA-enriched diet. We observed a trend towards increased fractional anisotropy (FA) of the corpus callosum (0.43 vs. 0.46; p = 0.100, Figure 2K), indicating improved microstructural properties of the white matter tracts. In line with the trend for FA, axial diffusivity (AD) and radial diffusivity (RD) of the corpus callosum were significantly lower (AD: 0.63 vs. 0.58, p = 0.050; RD: 0.32 vs. 0.29, p = 0.050, Figure 2K), indicating enhanced myelination. Finally, the mean diffusivity (MD) of the corpus callosum was also significantly lower (0.42 vs. 0.38, p = 0.050, Figure 2K), indicating reduced vasogenic edema (15). We observed no significant microstructural changes for large infarcts (Figure 2K). Raw data is provided in the supplement (Supplemental Table 2).

Overall, MRI measurements thus indicate that a CLA-enriched diet enhances the microstructural reorganization of connecting white matter tracts after stroke in small infarcts.

As we interpreted the lower MD-values as reduced vasogenic edema, we additionally performed immunohistochemical experiments of IgG presence after 90 days in the corpus callosum in another cohort of animals. IgG should not be present in the brain given an intact selectivity of the blood-brain-barrier. As expected, no IgG was found in the sham animals, while it was still detected in the experimental animals after 90 days. However, the visual difference between control animals and those fed with CLA was not significant (Figure 2L, the depicted stained infarcts would theoretically be categorized into small infarcts).

As we observed an improved microstructural reorganization following CLA-rich diet, we next aimed to assess immunological correlates. First, we analyzed immune cell differentiation of myeloid cells and T cells along the different compartments of the gut-brain-axis, including the lamina propria of the small intestine and colon as well as mesenteric lymph nodes (gut), the peripheral blood and the draining cervical lymph nodes.

We did not obtain significant alterations (Figures 3A-D). Albeit, total neutrophil count tended to increase in the blood upon CLA-treatment (1625.46 vs. 998.46; p adj. = 0.852, Figure 3A, left) whereas the CD11b⁺ Ly6G-populations consisting of monocytes/macrophages, dendritic cells and eosinophils visually seemed to decrease in the blood (13.60 vs. 11.35; p adj. = 0.935, Figure 3A, right). Regulatory T cells visually appeared to increase in the brain, without significancy (11.72 vs. 5.51 per CD4⁺; p adj. = 0.936; 88.20 vs. 58.55 per CD25⁺; p adj. = 0.936) (Figure 3D).

Furthermore, we investigate possible cytokine alterations in myeloid and T cells upon CLA diet. We were not able to find significant alterations potentially correlating with improved microstructural reorganization on this level (Figures 4A, B).

To assess a potential translational relevance of CLA-mediated effects for stroke patients, we next investigated the in vitro effect of CLA on human lymphocytes. We thus isolated CD4⁺ T lymphocytes from peripheral mononuclear cells of healthy donors and stroke patients.

As the stroke represents a strong inflammatory event, we performed anti-CD3/anti-CD28 (αCD3/αCD28) treatment of all samples to imitate the inflammatory stimulus. Half of the cells were concomitantly treated with CLA (Figures 5A-F).

Most importantly, the lymphocytes of stroke patients and healthy donors were responsive to the CLA treatment: Lymphocytes of stroke patients showed a significantly reduced production of pro-inflammatory cytokines IFN-γ (2775.0 vs. 1915.0 pg/mL, p adj. =0.047, Figure 5A), and TNF-α (5018.0 vs. 4005.0 pg/mL, p adj. = 0.016, Figure 5D). Similarly, the concentrations of IFN-γ (5989.0 vs. 1571.0 pg/mL, p adj. <0.0001, Figure 5A) and TNF-α (7841.0 vs. 4761.0 pg/mL, p adj. <0.0001, Figure 5D) of healthy controls were decreased after incubation with CLA for 72 h.

Comparing the secretion of pro-inflammatory cytokines between healthy donors and stroke patients upon stimulation with αCD3/αCD28, stroke patients were less responsive as indicated by reduced secretion of IFN-γ (5989.0 vs. 2775.0 pg/mL, p adj. = 0.004, Figure 5E) and TFN-α (7841.0 vs. 5018.0 pg/mL, p adj. = 0.013, Figure 5E). Upon combined αCD3/αCD28-and CLA-treatment, the cytokine levels responded similarly.

Altogether, these findings demonstrate CLA has the capacity to partially suppress pro-inflammatory cytokine production in human lymphocytes.