Prolonged Intake of Dietary Lipids Alters Membrane Structure and T Cell Responses in LDLr^−/− Mice

It is now widely accepted that nutrition has the ability to both heighten and suppress immune responses leading to alterations in immune function (1). Malnutrition is the leading cause of immunodeficiency globally, and relates not only to undernourishment but also to specific nutrient deficiencies and to the imbalance of dietary macro- and micronutrients (2, 3). Thus, malnutrition is no longer solely the concern of developing economies but also affects Western society, where diets rich in saturated fats, and often lacking in essential nutrients, has resulted in an obesity epidemic (4).

There is increasing evidence of immune dysfunction in obese patients, of which reduced Ab titers to vaccination is a prime example (5). Moreover, decreased immune responsiveness in these patients may contribute to a greater incidence of secondary and surgical site infections, reduced wound healing, and increased time needed for antibiotic therapy (6, 7). All of this information suggests that the overconsumption of saturated fatty acids and cholesterol is a form of malnutrition that can lead to impaired immunity.

T lymphocytes are both the modulators and effector cells of the adaptive immune response and, as such, play an essential role in the fight against both viral and bacterial infections (8). Considerable research has been carried out to investigate the effects of dietary lipids on T cell function, with both clinical and animal studies demonstrating alterations to the frequency and functional capacity of various T cell subsets in response to obesity (9–13); modifications to T cell eicosanoid synthesis, lipoprotein metabolism/peroxidation, gene and receptor expression, or membrane fluidity are all potential explanations (14–16).

Although membrane phospholipids are synthesized de novo (17), a number of the fatty acids that compose their hydrophobic tails are obtained from either the diet or derivatives of these dietary fatty acids (18). Thus, because lipids are constantly removed and replaced at the plasma membrane, the membrane fatty acid profile is thought to be highly adaptive to diet (18). Membrane lipids contribute to cell signaling directly, by acting as precursors to signaling molecules (19), and indirectly, by controlling the spatiotemporal organization of signaling proteins through the formation of ordered membrane domains rich in saturated sphingolipids, glycerophospholipids, and cholesterol (20–23). Often referred to as lipid rafts, these microdomains are less fluid, condensed regions within the membrane that affect protein diffusion and compartmentalize signaling processes (20–23). Thus, the lipid composition and degree of fatty acid saturation present within the cell membrane are thought to influence membrane fluidity and, in turn, T cell signaling (24).

We have previously shown that disrupting membrane condensation at the T cell activation site inhibits T cell signal transduction and downstream effector functions in vitro (25). In this study we examined the impact of dietary lipids on T cell lipid composition and membrane order and how this correlates to in vivo and ex vivo T cell responses, specifically in the absence of obesity and obesity-induced inflammation. Elicitation of contact hypersensitivity (CHS) in low-density lipoprotein receptor knockout (LDLr^−/−) mice fed a Western high-fat (WHF) diet for 9 wk resulted in augmented CD4⁺ and CD8⁺ T cell proliferation in the draining lymph nodes as well as an increase in CD4⁺ central-memory T cells. The enhanced in vivo T cell response correlated with elevated total phospholipid and decreased cholesterol levels within these cells, as well as a decrease in phospholipid fatty acid saturation of phosphatidylcholine (PC) and sphingomyelin (SM). T cells isolated from these high-fat–fed mice exhibited enhanced membrane order and IL-2 mRNA expression in response to ex vivo TCR activation.

Age- and sex-matched LDLr^−/− (B6.129S7-LDLRtm1Her/J) and C57BL/6J wild-type (WT) mice were bred at Australian BioResources Pty (Mossvale, NSW, Australia). The 6- to 8-wk-old mice were first maintained on a standard chow diet (Gordon's Specialty Stock Feeds, Yanderra, NSW, Australia) for 1 wk and then randomly assigned to either a WHF (Specialty Feeds, Glen Forrest, WA, Australia) or standard chow diet for the duration of the study period. Mice were housed five to a cage under specific pathogen–free conditions with a 12-h light/12-h dark cycle. Animals were given free access to food and water and were weighed weekly throughout the study period. All animal work was approved by the University of New South Wales Animal Care and Ethics Committee.

During the final week of feeding, the shaved abdominal surface of the mouse was painted with 25 μl 0.5% (v/v) 2,4-dinitrofluorobenzene (DNFB; Sigma-Aldrich, St. Louis, MO) in a vehicle containing 4:1 acetone/olive oil (v/v) and 1 μg/ml endotoxin-free BSA (Sigma-Aldrich). The following day the abdomen was repainted with 0.25% DNFB in the same vehicle. Control mice were painted on the shaved abdomen with vehicle alone on both days.

At 5 d following initial sensitization, both control and sensitized mice were painted with 20 μl 0.2% (v/v) DNFB in the vehicle containing 0.8 μg/ml endotoxin-free BSA on both sides of one ear pinna. The contralateral ear pinna was painted with 20 μl vehicle alone. The degree of ear swelling was measured 24 h following ear painting, using an engineer’s micrometer (Mitutoyo, Tokyo, Japan). Measurements were made twice and averaged. The increment of ear swelling was calculated by subtracting the thickness of the control ear from the thickness of the challenged ear, as described previously (26), that is, (challenged ear − unchallenged ear)/unchallenged ear × 100.

Peripheral blood was collected by cardiac puncture. Blood was then transferred into untreated microcentrifuge tubes and left to clot for 1 h at room temperature before centrifuging at 2000 × g at 4°C for 15 min. The serum fraction was isolated and stored at −80°C until analysis.

Serum fractions were thawed and IL-6 levels determined using a commercial mouse IL-6 DuoSet sandwich ELISA kit, following all manufacturer’s instructions (R&D Systems, Minneapolis, MN).

Serum fractions were thawed and total triglyceride and cholesterol levels determined using commercial enzymatic assay kits (Wako Chemicals, Fuggerstrasse, Neuss, Germany). To monitor the accuracy of each assay, an abnormal control serum (Fisher Diagnostics, Middletown, VA) was tested alongside collected serum samples.

Both right and left superficial inguinal lymph nodes were removed by careful dissection, pooled for each animal, and collected in 2 ml FACS buffer (1% BSA, 0.1% sodium azide in PBS). Lymph nodes were immediately weighed using a fine balance and a cell suspension from both lymph nodes prepared. In brief, the tissue was placed into a six-well plate and minced to complete dissociation through a 70-μm cell strainer (BD Biosciences, San Jose, CA) in 2 ml FACS buffer using the end of a 1-ml plastic syringe plunger. The cells were then transferred to a 15-ml centrifuge tube, topped up with 10 ml FACS buffer, centrifuged at 300 × g at 4°C for 5 min, and resuspended in 2 ml FACS buffer for cell counting. Cell counts and viability were enumerated using an Innovatis CASY Cell Counter (Roche Applied Science, Penzberg, Upper Bavaria, Germany).

The mAbs used for surface staining include the following: anti-CD3–Pe-cy7, anti-CD4–V500, anti-CD8–Pacific Blue, anti-CD62L–FITC, and anti-CD44–allophycocyanin, all purchased from BD Biosciences. Relevant isotype controls were used for all experiments (BD Biosciences). For each sample, 10⁶ lymph node–derived single cells were blocked with 50 μl Miltenyi Biotec FcR Blocking Reagent (Bergisch Gladbach, Rheinisch-Bergische Kreis, Germany) for 10 min at 4°C. Cells were then incubated with the above mAbs in the dark at 4°C for 30 min. Finally, cells were washed twice with 1 ml FACS buffer and resuspended in 300 μl FACS buffer for analysis. Cell sample acquisition was done on the day of collection using either a BD Biosciences FACSCanto II or FACSVerse flow cytometer, and data were analyzed using FlowJo software (TreeStar, Ashland, OR). A total of 50,000 events were acquired. The proportions of CD4⁺ and CD8⁺ T cell subsets within a lymphocyte gate (70% T lymphocytes, >90% viable per 7-aminoactinomycin D exclusion) were quantified. These T cell subsets were further analyzed for coexpression of CD62L and CD44 to differentiate naive from memory cells, respectively. T cells displaying a CD62L^highCD44^low phenotype were considered naive, whereas CD62L^lowCD44^high cells were designated effector-memory T cells. CD62L^highCD44^high were nominated central-memory cells (Fig. 3A) (27).

To determine the proportion of proliferating lymphocytes in response to in vivo DNFB challenge, a mAb against proliferating cell nuclear Ag (PCNA) conjugated to PE and the relevant isotype control were purchased from eBioscience (San Diego, CA). PCNA is expressed more abundantly during the S phase of the cell cycle (28). A total of 10⁶ lymph node–derived single cells were first stained with the Abs, anti-CD3–Pe-cy7, anti-CD4–V500, and anti-CD8–Pacific Blue, as described above. Cells were then fixed for 2 min in 1 ml of 2% paraformaldehyde, followed by washing with 1 ml of cold PBS. Cells were then incubated in 1 ml of 100% methanol at -20°C for 10 min, centrifuged again, and washed with 1 ml PBS containing 0.1% Triton ×100 (Thermo Fisher Scientific, Waltham, MA). Cells were then incubated with anti-PCNA Ab (0.125 μg) or the relevant isotype control in PBS containing 0.1% BSA for 30 min at room temperature. Cells were finally washed twice with 1 ml FACS buffer and resuspended in 500 μl FACS buffer for analysis on either a BD Biosciences FACSCanto II or FACSVerse flow cytometer. Data were again analyzed using FlowJo software. As a positive control for this analysis, lymph node–derived single cells were prestimulated in vitro with 5 μg/ml PHA (Sigma-Aldrich) for 48 h, showing >80% PCNA-positive cells.

Spleens were removed immediately after death and placed in either RPMI 1640 culture medium (Invitrogen, Carlsbad, CA) or PBS containing 0.5% BSA. Spleens were placed in a 10-ml Petri dish containing the collection buffer, and cells were released by teasing apart the spleens through a 70-μm cell strainer with the end of a 1-ml plastic syringe plunger. For each dietary group, cells from individual mice were then pooled in a 50-ml centrifuge tube, resulting in separate cell suspensions based on diet. Cell preparations were then filtered through a 40-μM cell strainer (BD Biosciences) before T cell isolation, using nylon wool fiber (Polysciences, Warrington, PA) or autoMACS negative T cell selection (Miltenyi Biotec). The purity of the isolated T cell population ranged from 85 to 97% (average ∼92%).

After isolation, splenic T cells were first mechanically homogenized on ice in buffer (0.25 M sucrose, 20 mM HEPES, 0.5 mM EDTA, and two protease inhibitor tablets in double distilled water) using an in-house cell homogenizer. The protein concentration of the homogenate was then determined using the Pierce BCA Protein Assay Kit (Thermo Scientific, Waltham, MA). Lipid extracts of the whole-cell homogenate were extracted at least twice, and each time prepared in triplicate, using a modification of the Folch method (29) and methyl-tert-butyl ether extraction (30) for comparison. Analysis of these duplicate extracts was then performed independently. Phospholipid analysis of samples extracted using the Folch method were analyzed by nanoelectrospray ionization quadrupole time-of-flight tandem mass spectrometry, with positive ion PC/SM precursor ion scanning performed on a Q-STAR Pulsar i (AB Sciex, Framingham, MA). All spectra were measured using Analyst QS 1.1 (Applied Biosystems, Foster City, CA) and spectral assignment and quantification against internal standards achieved using Lipid Profiler software (AB Sciex) (31). Phospholipid analysis of samples extracted using the methyl-tert-butyl ether method was performed on a hybrid triple quadrupole linear ion trap mass spectrometer (QTRAP 5500; AB Sciex) equipped with an automated chip-based nanoelectrospray ionization source (TriVersa Nanomate; Advion Biosciences, Ithaca, NY). Data were analyzed with Analyst QS 1.5.1 (AB Sciex) and spectral assignment and quantification against internal standards achieved using LipidView 1.3 β software (AB Sciex). The cholesterol content of splenic T cells was analyzed using liquid chromatography–mass spectrometry on an Accela LC and auto-sampler system (Thermo Fisher Scientific) with an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) (32). All data were corrected against sample protein concentration.

A total of 1–2 × 10⁶ isolated splenic T cells were incubated in a microcentrifuge tube with 150 μl RPMI 1640 containing 10% FCS and 20 μM Laurdan (Invitrogen) for 30 min at 37°C in a 5% CO₂ atmosphere. The cells were then mixed with 5 × 10⁵ CD3ɛ and CD28 Ab-coated beads (Spherotech, Lake Forest, IL), centrifuged for 10 s at 250 × g, gently resuspended, and incubated for an additional 10 min at 37°C/5% CO₂ atmosphere. Cell activation was terminated, and cells were fixed by adding an equal volume of 8% paraformaldehyde in PBS. Cells were then washed twice in 500 μl PBS and resuspended in 80–90 μl PBS for mounting onto glass cover slides using a Cytospin Centrifuge (Thermo Fisher Scientific).

Imaging was performed at room temperature on an inverted confocal laser scanning fluorescence microscope (SP5; Leica Microsystems, Wetzlar, Germany) with two-photon excitation at 800 nm, using a femtosecond-pulsed Ti:Sapphire laser (Mai-Tai; Spectra-Physics, Santa Clara, CA) and Leica Microsystems LAS software—×63 oil-immersion objective, numerical aperture = 1.4. Fluorescence was simultaneously detected in the ranges of 400–460 nm and 470–530 nm. Laurdan intensity images were then used to construct generalized polarization (GP) images, as previously described (33) using ImageJ software (National Institutes of Health, Bethesda, MD). Pseudo-colored GP images were merged with mean fluorescence intensity images to preserve structural information. The cell membrane region in contact with the bead was measured as the contact zone (activation site), whereas the noncontact zone was used to determine membrane order at the nonactivation site.

To activate isolated splenic T cells, 2.5 × 10⁵ cells were incubated in RPMI 1640 containing l-glutamine, 50 mM HEPES, and 2% BSA in a 96-well flat-bottom cell culture plate with 10 μg/ml plate-bound anti-mouse CD3ɛ and CD28 Abs (eBioscience) overnight at 37°C in a 5% CO₂ atmosphere. Four wells per triplicate were used to give a total of 10⁶ cells. Following overnight incubation, cells were washed twice with PBS and the four wells per triplicate were pooled in 1 ml TRIsure (Bioline, London, U.K.) before storing at −80°C until extraction. A total of 10⁶ nonactivated control cells per triplicate were stored in 1 ml TRIsure without overnight incubation. Total RNA was then extracted and quantified using the NanoDrop spectrophotometer (Thermo Fisher Scientific) and RNA integrity was checked using the Agilent Technologies 2100 Bioanalyzer (Santa Clara, CA). Equal amounts of RNA were then reverse transcribed to cDNA using the SuperScript III First-Strand Synthesis System (Invitrogen), following all manufacturer’s guidelines. Finally, real-time PCR was performed using the SensiMix SYBR Kit (BioRad, Hercules, CA) and an RG-3000 Rotor-Gene thermal cycler (QIAGEN, Venlo, the Netherlands). Briefly, a 20-μl reaction volume was used that contained 10 μl SensiMix SYBR, 8 μl nuclease-free water, 0.5 μl forward primer (10 μM), 0.5 μl reverse primer (10 μM), and 1 μl cDNA. The cycling conditions comprised 1 cycle at 95°C for 10 min for polymerase activation, 40 cycles at 95°C for 15 s, 56°C for 30 s, and 72°C for 30 s with a melting curve ramp of 60–95°C. The threshold cycle was calculated using Rotor-Gene 6 software (QIAGEN), and relative changes in mRNA expression between treatment groups were determined using the 2^−∆∆Ct method (34). The results were normalized to the geometric mean of the reference genes GAPDH and G6PDX and were calculated relative to the expression of chow nonactivated samples (35). All primers used are listed in Table I. For each RNA extraction, cDNA was prepared in duplicate and gene expression was analyzed in triplicate. Negative controls included amplification of a sham reverse transcription reaction (incubated without enzyme) and amplification of a reaction mixture with no added cDNA (data not shown).

The statistical significance of two data sets was assessed with an unpaired t test assuming equal variances. Multiple comparisons were assessed with the one-way ANOVA or two-way ANOVA and Sidak’s posttesting. All data analysis was performed using GraphPad Prism software (GraphPad Software, La Jolla, CA). Differences were considered significant at p < 0.05.

To distinguish the effects of weight gain/obesity from those of circulating lipids on T cell function, we sought a mouse model that mimics the human situation and was highly susceptible to dietary change but not prone to weight gain. Feeding C57BL/6J WT mice a WHF diet for a sustained period resulted in significant weight gain but only modest elevation of circulating lipids (Supplemental Fig. 1A, 1B). In contrast, LDLr^−/− mice on the same diet and background had no significant elevation of body weight (Fig. 1A) but significantly increased levels of circulating cholesterol and triglycerides (Fig. 1B). LDLr^−/− mice are a well-established model to delineate the role of the immune system in a hyperlipidemic environment (36) because the loss of LDL receptor expression has no (direct or indirect) impact, for example, on Ag processing and presentation (37), immunization, or Ab production (38). Thus, we used LDLr^−/− mice and carried out two dietary challenges: an acute intervention in which mice were fed a WHF diet for a period of 1 wk, the other a long-term study of 9 wk. The standard chow diet was used as a control, with the WHF diet containing >10 times the amount of saturated fatty acids compared with this formulation. Even after 9 wk, 19 LDLr^−/− mice fed the high-fat diet did not show significant weight gain relative to 18 chow-fed mice over the same period (Fig. 1A), with mice gaining an average of 5 g throughout the study, indicating that the mice did not develop an obese phenotype.

To evaluate the potential proinflammatory effect of high-fat feeding on LDLr^−/− mice, we tested serum for IL-6, one of the key proinflammatory cytokines secreted by adipocytes (39). IL-6 was not detected in the serum of chow- or WHF-fed control mice following either 1 wk or 9 wk of feeding (n > 6 per dietary group and time point), suggesting that the high-fat diet was unlikely to trigger systemic inflammation in these mice (data not shown). Serum analysis showed elevated total triglyceride and cholesterol in LDLr^−/− mice, compared with WT C57BL/6J mice. Further, the WHF diet caused rapid enrichment of circulating lipids, with fat-fed mice displaying elevated total triglyceride and cholesterol serum levels after only 1 wk of feeding (Fig. 1B, p < 0.0001). Following the 9-wk feeding period, triglyceride levels further increased, resulting in a 3-fold enhancement, whereas cholesterol remained ∼4-fold higher than in chow-fed mice (Fig. 1B, p < 0.0001).

Together, our findings illustrate that feeding LDLr^−/− mice a WHF diet for ≤9 wk did not cause obesity or systemic inflammation but did caused rapid enrichment of circulating lipids.

During the final week of the 9-wk dietary intervention, we used a DNFB model of CHS to assess the T cell response in the draining lymph nodes of high-fat–fed and control LDLr^−/− mice (Fig. 2A). CHS reaction can be classified into two phases, termed sensitization and elicitation. The sensitization (afferent) phase occurs during the initial skin contact with the DNFB hapten that is not allergenic on its own but on topical application forms a hapten–carrier complex that acts as a neoantigen and stimulates naive T cells within the draining lymph nodes. The elicitation (efferent) phase of CHS occurs through subsequent contact with the same hapten, stimulating hapten-specific memory T cells to migrate to the area of secondary skin contact, resulting in localized inflammation. For the acute dietary challenge, CHS induction was carried out during the 1-wk feeding period. For the immune challenge, each dietary group was split in two, resulting in a control and a CHS group for each diet (Fig. 2B).

Successful induction of hypersensitivity reaction was illustrated by measuring the ear-swelling response of control and challenged mice 24 h after ear painting with DNFB; following the acute and long-term dietary challenges a clear CHS response was observed in both dietary groups (Fig. 2C). Reduced swelling was observed in high-fat–fed challenged mice, compared with their chow-fed counterparts, following 1 wk of feeding (p < 0.05); however, no significant difference between challenged chow and WHF-fed mice was observed at the 9-wk time point (Fig. 2C), indicating that long-term feeding had no impact on CHS responsiveness. Further, no significant difference in ear pinna thickness was observed between control groups at either time point, confirming that no nonspecific irritation or inflammation occurred as a result of the WHF diet (Fig. 2C). Induction of CHS was also confirmed by H&E staining of the ear pinna, which revealed prominent dermal edema and cell infiltration in the DNFB-treated ear of challenged mice.

To further investigate the magnitude of the immune response following elicitation, we weighed the draining inguinal lymph nodes directly following excision (Fig. 2D) and carried out a total cell count to measure cellularity (Fig. 2E). No significant differences were found in control or challenged WHF-fed mice compared with chow-fed equivalents.

Finally, we measured IL-6 serum levels in challenged mice following induction of CHS at both the 1- and 9-wk time points. IL-6 was detected in all challenged groups, indicating that IL-6 production had occurred as a result of DNFB exposure in presensitized animals. Following 1 wk of feeding there was a significant increase in IL-6 production in high-fat–fed mice, with IL-6 levels of 14 ± 7.3 pg/ml and 47 ± 8.9 pg/ml (mean ± SEM from here on) in chow- and WHF-fed mice, respectively (Fig. 2F, p < 0.05). This suggests that the DNFB-specific inflammatory response may have been enhanced by an acute period of high-fat feeding. In contrast, no difference in IL-6 production occurred between dietary groups following the 9-wk dietary intervention, with chow- and WHF-fed mice producing an average of 31 ± 12.5 pg/ml and 17 ± 5.0 pg/ml, respectively (Fig. 2F).

These results indicate that although 1 wk of high-fat feeding may have had an impact on the inflammatory response, at 9 wk, the magnitude of the irritant and allergic response to DNFB was similar between dietary groups.

We used the CHS model to specifically examine the effect of a high-fat diet and its resultant elevated levels of circulating lipids on the T cell response in vivo, using flow cytometry to quantitate CD4⁺ and CD8⁺ T cells within the draining lymph nodes following elicitation of CHS (Fig. 3A). For both the 1- and 9-wk time points, the effector-memory and central-memory CD8⁺ T cell response was varied over duplicate experiments (Supplemental Fig. 2); thus, as the best representative, the pooled data from duplicate experiments are shown and effector-memory/central-memory CD8⁺ T cell populations are pooled (E+C-Memory).

Following 1 wk of feeding (Fig. 3B, 3C), no change to the DNFB-specific CD4⁺ or CD8⁺ T cell response occurred owing to diet, except in the percentage of CD8⁺ T cells, which showed an ∼3% decrease (26 ± 0.5–23 ± 0.9%) in WHF-fed challenged mice. However, it should also be noted that a similar decrease in WHF-fed control mice also occurred (∼3%; 34 ± 0.7–31 ± 0.6%). This finding indicates that the increase in circulating lipid levels at that time point did not alter T cell responsiveness. In contrast, 9 wk of high-fat feeding caused an amplification of the T cell response following immune challenge (Fig. 3D, 3E), with WHF-fed mice showing an ∼5% increase in proliferating CD4⁺ T cells, from 13 ± 0.7% to 18 ± 1.0%, and an ∼4% increase in proliferating CD8⁺ T cells, from 6 ± 0.4% to 10 ± 0.8% (Fig. 3D, p < 0.001; Fig. 3E, p < 0.0001, respectively). Augmented cell proliferation resulted in an increase in the proportion of central-memory CD4⁺ T cells, from ∼10 + 0.6% to ∼4 + 0.8% (Fig. 3D, p < 0.01), whereas no significant increase in memory CD8⁺ T cells occurred (Fig. 3E). However, although no significant increase in memory CD8⁺ T cells was observed in the pooled data, this is most likely due to variation in the CD8⁺ T cell response, as an ∼8% increase in central-memory (9 ± 1.2–17 ± 3.0%) and an ∼3% increase in effector-memory (9 ± 0.4 to 12 ± 1.0%) CD8⁺ T cells was observed within experiments 1 and 2, respectively (Supplemental Fig. 2C, 2D, p < 0.05).

Despite an increase in the proportion of central-memory CD4⁺ T cells following 9 wk of high-fat feeding, the total number of CD4⁺ T cells within the draining lymph nodes did not increase (Fig. 3D). We observed an ∼5% decrease (36 ± 0.4–31 ± 0.8%) in the proportion of CD4⁺ T cells within the draining lymph nodes of WHF-fed control mice, which may account for this finding (Fig. 3D, p < 0.001).

Our data indicate that 9 wk of high-fat feeding caused an elevated T cell response in the draining lymph nodes, which was not due to increased body weight or a systemic inflammatory state. In support of this observation, induction of CHS (Supplemental Fig. 1C) in C57BL/6J WT mice following 16 wk of high-fat feeding had no impact on T cell responsiveness (Supplemental Fig. 3), despite significant weight gain. Thus, the data suggest that the observed elevated T cell response is related to the exposure of these cells to a modified availability of circulating lipids resulting from the WHF diet.

To correlate the amplified T cell response we observed in vivo to diet-induced changes in the lipid composition and fatty acid profile of the T cells, we isolated splenic T cells and analyzed their lipid content by mass spectrometry. For the 9-wk dietary intervention, we analyzed the total amount, and calculated the molar percentage, of the major phospholipid classes PC, SM, phosphatidic acid (PA), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylserine (PS), and phosphatidylinositol (PI), as well as cholesterol. For the 1-wk dietary challenge, we analyzed the total amount, and calculated the molar percentage, of the most abundant phospholipid, PC, as well as SM and cholesterol.

After 1 wk on the WHF diet, splenic T cells in these mice had no increase in the total amount (Fig. 4A) or percentage (Fig. 4B) of phospholipids or cholesterol, compared with T cells from mice fed a chow diet containing 80 ± 1.5% phospholipids and 20 ± 1.5% cholesterol. In contrast, 9 wk of high-fat feeding significantly increased the absolute amount of phospholipids within the T cells of WHF-fed mice relative to T cells of chow-fed mice (Fig. 4A, p < 0.001), causing a change in the ratio of phospholipid to cholesterol (Fig. 4B). Further, an ∼6% increase (∼10 + 1.6–16 + 1.1%) in the percentage of PE was observed following 9 wk of high-fat feeding (Fig. 4B inset, p < 0.05), which correlated with an ∼4% increase (∼80 + 1.2–84 + 0.7%, Fig. 4B, p < 0.05) in total phospholipids and an ∼4% reduction in cellular cholesterol (∼20 + 1.2–16 + 0.7%, Fig. 4B, p < 0.05).

Examining PC, SM, PE, and PS in T cells from the 9-wk dietary intervention in more detail, we observed within cells from high-fat–fed mice a reduction in the percentage of saturated PC species (Fig. 4C, p < 0.05) and a significant increase in monosaturated fatty acids (Fig. 4C, p < 0.01). This trend was mirrored in the SM composition, with a significant reduction in saturated SM species and an increase in monounsaturated SM species (Fig. 4D, p < 0.05). Minor changes to the abundance of individual PC and SM species, and PC and SM fatty acid acyl chain length, were observed following 9 wk of high-fat feeding (data not shown). No differences in PE or PS composition were observed between dietary groups (data not shown). We also examined polyunsaturated fatty acids (e.g., n-3 versus n-6) but found no difference in the levels in T cells following any of the dietary interventions (data not shown).

Our data indicate that dietary intervention, with the consequential changes in circulating lipids, has the ability to alter the lipid composition of T cell membranes. Compared with T cells from chow-fed mice, T cells from WHF-fed mice showed a clear increase in absolute PL, whereas their proportion of cholesterol is reduced. This finding suggests that these cells not only increase their membrane content but also alter their membrane composition. It is possible that the augmented T cell response observed in vivo could be due to changes in T cell lipid composition. This idea was further highlighted by examining the T cell lipid composition of C57BL/6J WT mice following 16 wk of high-fat feeding, which illustrated that modest changes to circulating lipids, as well as having no impact on in vivo T cell responses, did not alter the proportion or absolute amount of PL and cholesterol, compared with their chow-fed counterparts (Supplemental Fig. 4).

To determine how the observed changes in T cell lipid composition potentially translated into alterations in plasma membrane order, we activated splenic T cells using Ab-coated beads and measured membrane order at contact zones using Laurdan microscopy (40). Laurdan is a polarity-sensitive membrane dye whose emission wavelength depends on the degree of water penetration into the membrane. In fluid membranes, water can partially penetrate the lipid bilayer, whereas in highly ordered membranes water is largely excluded. A ratiometric quantification of the emission spectra, called GP, is therefore a measure of membrane order, with GP values of −1 being the most fluid and +1 the most ordered (33).

Laurdan fluorescence demonstrated a significantly higher GP value at the plasma membrane region that was in contact with an anti-CD3/CD28–coated bead, compared with noncontact zones, indicative of membrane condensation at the bead–cell contact site (Fig. 5C, p < 0.01, p < 0.0001, chow and WHF, respectively). More importantly, cells isolated from mice fed a WHF for 9 wk displayed a significantly greater mean GP value at the activation site compared with cells from chow-fed mice, suggesting membrane condensation was increased as a result of the altered cellular lipid composition (Fig. 5C, p < 0.01).

Membrane condensation plays a critical role in T cell signaling (26), and production of the T cell growth factor IL-2 directly depends on the activation of multiple T cell signaling pathways and transcription factors (41). Thus, to evaluate the functional consequences of TCR signaling, in response to changed membrane condensation at the activation site induced by the high-fat diet, we measured IL-2 gene expression from activated splenic T cells by quantitative RT-PCR (Table I). Stimulation via CD3/CD28 resulted in an ∼1.5-fold increase in IL-2 mRNA expression by cells from WHF-fed mice, relative to cells from the chow-fed group, indicative of a possibly potentiated signal transduction in cells from WHF-fed mice (Fig. 5D, p < 0.0001).

Collectively, these results show that the changes we observed to the lipid composition of T cell membranes in WHF-fed mice following 9 wk of feeding resulted in altered membrane organization and IL-2 response after activation, which may lead to enhanced signal transduction and amplified in vivo T cell responses to immune challenge.

Clinical data suggest that obesity affects T cell immunity; however, little is known regarding whether dietary lipids directly influence T cell activation and responsiveness. It is possible that a high intake of saturated dietary fats causes alterations in T cell function by introducing changes to membrane lipid composition and structure that in turn affect TCR signaling and thus T cell activation. Previous studies in which rodents are fed a high-fat diet demonstrate an altered lipid composition of serum (42, 43) and whole splenic T cells (42–44), which corresponds to enhanced in vivo and ex vivo T cell proliferation, as well as IL-2 signaling pathways (44).

We demonstrated that feeding LDLr^−/− mice a high-fat diet for 9 wk causes a rise in CD4⁺ and CD8⁺ T cell proliferation and an increase in the proportion of CD4⁺ central-memory T cells within the draining lymph nodes following induction of CHS. This augmented T cell response occurred in the absence of any excess weight gain or IL-6 production, indicating that a diet-induced proinflammatory environment did not cause the altered T cell response. This is because LDLr^−/− mice gain less weight and fat mass when fed a WHF diet compared with LDLr^+/+ mice on the same background and are thus less prone to obesity and obesity-induced inflammation (45–47). Of interest, an acute dietary intervention of 1 wk did not alter the T cell response in CHS, likely because the lipid composition of T cells had not changed after 1 wk of WHF diet and/or owing to enhanced (polarized) innate inflammatory responses to the acute high-fat feeding (48), at the expense of a lower T cell response.

LDLr^−/− mice do not possess LDL receptors, and therefore liver and other cell types cannot remove cholesterol-rich intermediate-density lipoproteins and LDLs from plasma, resulting in the buildup of circulating lipids (49, 50). In terms of immune response, genetic deletion of LDL receptors has no impact on Ag processing and presentation (37), immunization and Ab production (38), or virus-induced immunopathology (51). The reduced clearance of lipids from the circulation in LDLr^−/− mice was already evident after 1 wk of the WHF diet and corresponds to the hyperlipidemia reported previously (47, 52). However, such acute changes in the circulating lipid levels were not sufficient to substantially alter T cell lipid composition, or responses to an immune challenge, suggesting that elevated levels of circulating lipids alone did not amplify T cell responses. We also noticed that despite hypercholesterolemia in LDLr^−/− mice on a WHF diet (53–56), the total cholesterol levels in T cells were not increased even after 9 wk. Conversely, the hypertriglyceridemia (54) appears to have altered the phospholipid and fatty acid composition of the T cells in WHF-fed mice observed following 9 wk of feeding. Hence, it is possible that dietary lipids affect the phospholipid metabolism in T cells but may have little impact on T cell cholesterol metabolism. This idea was further illustrated by our C57BL/6J WT data, which clearly showed that hypercholesterolemia did not change T cell cholesterol levels even following 16 wk of feeding.

Our data suggest a link between WHF diet–induced altered T cell lipid composition, efficiency of membrane condensation at TCR activation sites, and activation response ex vivo and in vivo. We have previously reported that engagement and signaling through the TCR reorganizes the membrane at these sites to yield highly ordered domains (40). In addition, we observed a similar relationship as found in this study, in which ex vivo lipid manipulations of Jurkat T cells decrease the membrane order specifically at the T cell activation site, impairing IL-2 production and secretion (25).

The WHF diet contained >14 times more total saturated fats and 3 times more total monosaturates than the chow diet. Despite the large changes in circulating lipids, we observed relatively small changes in lipid composition in T cells after a 9-wk WHF diet. We detected changes in cholesterol-to-phospholipid ratio and a shift toward phospholipid monosaturation, but no changes in polyunsaturated fatty acid levels. The most obvious change in lipids in these cells was the increase in total phospholipids, particularly PE. This observation agrees with previous findings in which changes to the relative abundance and saturation of dietary fats altered the abundance of cellular PE as well as the phospholipid-to-cholesterol ratio (57) of cells and membranes (57–59). It became clear during ex vivo activation that the lipid changes we observed in T cells were functionally important, as we observed significant changes not only in membrane order at the T cell activation site but also in IL-2 mRNA levels.

Alterations to individual phospholipid classes, especially PE, may have a number of functional consequences, as the head group often confers enzymatic activity for phospholipids and influences membrane asymmetry (60). TCR activation domains have been shown to be enriched with plasmenyl PE (61), suggesting an important role for PE in signal transduction and membrane fluidity. Further, it has been shown that PE is involved in cell signaling through its hydrolysis to form a number of second messengers such as diacylglycerol, phosphatidic acid, and lysophosphatidic acid (17). Membrane PE (62) is also known to be one of the main phospholipid classes to play a key role in membrane fusion, fission, and curvature, essential processes in membrane protein assembly and subsequent T cell signaling transduction (63, 64). Hence, even seemingly minor changes in T cell lipid levels and ratios may have significant impact on T cell function.

How specialized membrane domains at T cell activation sites facilitate signaling is not yet fully understood, and it is surprising that dietary lipids can result in a gain-of-function in this respect. However, lipid domains may not be static entities but constantly reshaped in membranes that exhibit lipid phase separation (65). Indeed, it may be the ability to reorganize the plasma membrane upon TCR triggering, and not the basal membrane order and composition per se, that is functionally important for downstream activation. Hence, dietary lipids may induce only subtle change in membrane compositions that nevertheless have large effects on membrane organization, receptor signaling, and T cell responses.

We thank the Biomedical Imaging Facility and Bioanalytical Mass Spectrometry Facility of the Mark Wainwright Analytical Centre at the University of New South Wales.

The authors have no financial conflicts of interest.