Buffy coats from healthy donors were obtained from the Department of Transfusion Medicine (University Hospital Regensburg) in form of remnants from routine platelet donations. The donations were approved by the Institutional Ethics Committee of the University of Regensburg (vote number 13-101-0240; 13-101-0238) and are in accordance with the Declaration of Helsinki.

CD4⁺ and CD8⁺ T-cells were isolated using MACS cell separation kits (Miltenyi Biotec, 130-096-533 (CD4), 130-096-495 (CD8)). After isolation, cells were stored over night at 37°C at a concentration of 10⁷ cells per ml in RPMI 1640 medium supplemented with 2% human AB Serum, 1% stabilized glutamine and 0,5% Penicillin/Streptomycin. Cells were washed, counted (CASY System) and the cell concentration was adjusted to 5x10⁵ cells per ml using the aforementioned medium plus 100 IU recombinant human (rh) IL-2 (PeproTech, 200-02) per ml. Either PBS (control) or LDL (Kalen Biomedical LLC, 770200-1) was added at the indicated concentrations and cells were seeded in 96 well U-Bottom plates. For some of the experiments as indicated, an anti-human PD-1 blocking antibody (InVivoMAb anti-human PD-1, Bio X Cell, BE0188) was added at a concentration of 10 µg/ml. T cells were stimulated using CD3/28 Dynabeads (ThermoFisher Scientific, cell to bead ratio 1:1, 25 µl bead suspension per 10⁶ cells), incubating at 37°C for 96h. The T cells were harvested after 48h and 96h, Dynabeads were removed and the cells were counted using the CASY System.

Spheroids were generated for 4 days on ultra-low attaching plates, which were coated with a solution of 12 mg/ml Poly-(methacrylsäure-2-hydroxyethylester) (poly-HEMA; Sigma Aldrich) dissolved in 95% ethanol. 50 µl of sterile poly-HEMA solution per well was allowed to evaporate in a sterile biosafety cabinet and dried plates were stored at 4°C before use. Sub-confluent cultures of the highly microsatellite instable colon carcinoma cell line HCT116 (35) were dissociated into single-cell suspensions and subsequently, 10,000 cells in 100 µl RPMI 1640 (GIBCO, 31870-025) with 10% fetal calf serum (Sigma, F7524) and 2 mM glutamine (PAN Biotech, P04-80100)) were seeded per well and incubated in a humidified atmosphere (5% CO2, 95% air) at 37°C (Heraeus Incubator). For co-culture, peripheral blood mononuclear cells were isolated by density gradient centrifugation over Ficoll/Hypaque as described before (Andreesen et al., 1990 (36)). T cells were isolated by magnetic bead separation (human Pan T Cell Isolation Kit, Miltenyi Biotec, 130-096-535). T cells were cultured in RPMI 1640 (GIBCO, 31870-025), supplemented with 10% human AB serum (BRK, Bavarian Red Cross), 2 mM L-glutamine (PAN-Biotech, P04-80100), essential vitamins (GIBCO, 1112037) and non-essential amino acids (GIBCO, 11140035), 1 mM pyruvate (GIBCO, 11360039), b-mercapthoethanol (GIBCO, 31350010), 0.5% penicillin and streptomycin (both GIBCO, 15140122) and 25 IU/mL rhIL-2 (PeproTech, 200-02) in a humidified atmosphere (5% CO2, 95% air) at 37°C in a Heraeus incubator. 1x10⁶ cells T cells were seeded in 24-well plates with indicated treatments and stimulated with anti-CD3/CD28 Dynabeads (Thermo Fisher Scientific, 11132D) at a cell to bead ratio of 1:1. For co-culture with tumor spheroids, beads were removed after 48h of stimulation, T cells were washed and 0.1x10⁶ T cells in 100 µl tumor medium supplemented with 25 IU/mL Il-2 were added to each spheroid with indicated treatments. After 24h co-culture, they were washed and seeded with fresh medium for live cell imaging.

After 48h and 96h, T cells subsets were harvested, Dynabeads were removed and the cell suspensions were partitioned into FACS tubes. Cells were stained with Zombie Aqua Fixable Viability Kit (BioLegend, 423102) or Zombie NIR™ Fixable Viability Kit (BioLegend, 423106) and different surface antibodies PE anti-CD45RO (BD, 555493), PE anti-CD154 (BD 555700), PE anti-CD226 (BioLegend, 337106), PE anti-CD154 (BioLegend, 310806), PE anti-BTLA (BioLegend, 344505), APC anti-TIGIT (BioLegend, 372706), APC anti-CD25 (BD, 340907), APC anti-CD69 (BD, 555533), APC anti-CD62L (BD, 559772), APC anti-CD279 (PD-1) Antibody (BioLegend, 329908), APC anti-GITR (BioLegend, 371206), V450 anti-CD27 (BD, 561408), FITC anti-CD28 (BD, 555728), PE anti-CD39 (BD, 555464), FITC anti-CD44 (BioLegend, 397518), FITC anti-CD95 (BD, 561975), FITC anti-CD134 (OX40) (BioLegend, 350006), BV421 anti-CD137 (BioLegend, 309820), PE-Cy7 anti-CD223 (LAG-3) (eBioscience, 25-2239-42), PerCP-Cy5.5 anti-CD244 (BioLegend, 329515), PE-Cy7 anti-PD-L1 (BioLegend, 329717), PE-Cy7 anti-ICOS (BioLegend, 313519), PE-Cy7 anti-CD366 (Tim-3) (BioLegend, 345013), PE-Cy7 anti-Granzyme B (BioLegend, 396410), BV421 anti-CTLA-4 (Biolegend, 369606), BV421 anti-Perforin (BioLegend, 353307), APC anti-LDLR (R&D Systems, FAB2148A), APC anti-FOXP3 (eBioscience, 17-4776-42), Pacific Blue anti-CD8 (BD, 558207), BV510 anti-PD-1 (BioLegend, 329932), BV510 anti-CD3 (BioLegend, 317332), BV605 anti-CD3 (BioLegend, 317322), BV711 anti-CD4 (BD, 563028) following the manufacturer instructions. Flow cytometry data were acquired using the Fortessa System (BD) or Celesta System (BD). Data were analyzed using FlowJo (v10.8.1).

Cytosolic reactive oxygen species (ROS) were determined after surface marker staining by applying 10 µM 2′,7′-dichlorofluorescin diacetate (Sigma Aldrich, D6883) for 20 minutes in a cell culture incubator at 38°C in FACS wash buffer in air tight tubes. Cells were washed with 3 ml cold PBS, resuspended in FACS wash buffer and measured immediately.

Mitochondrial content was assessed by staining with MitoTracker Green FM (Thermo Fisher Scientific, M7514). Cells were incubated with 15 nM MitoTracker and 1.3 µM cyclosporine A in RPMI1640 supplemented with 2 mM L-Glutamine for 1 hour at 37°C in a cell culture incubator. Surface staining was performed afterwards.

Tetramethylrhodamine methyl ester (TMRM) (Thermo Fisher Scientific, T668) is a membrane-permeable, cationic, red-orange fluorescent dye that is enriched in active mitochondria. Cells were incubated with 10 µM TMRM and 1.3 µM cyclosporine A in RPMI1640 supplemented with 2 mM L-Glutamine for 30 minutes at 37°C in a cell culture incubator. Surface staining was performed afterwards.

Cholesterol was determined by Filipin staining after surface marker staining by adding 50 µg/ml Filipin in 500 µl PBS for 45 minutes at room temperature. Cells were washed with FACS wash buffer, resuspended, and measured immediately.

After 48 hours and 96 hours, cell cultures were harvested, centrifuged (7 min, 1300 PRM (300g) at 4°C) and the supernatants were stored at -20°C. Supernatants were thawed and subsequently TNFα, INFg, IL-4, IL-7, IL-10, IL-15 and IL-17 concentrations were analyzed using the equivalent ELISA-kits (Human DuoSet ELISA R&D Systems, IFNg DY285B, TNF-alpha DY210, IL-4 DY204, IL-7 DY207, IL-10 DY217B, IL-15 DY247, IL-17 DY317). ELISAs were performed following the manufacturer instructions.

Cellular oxygen consumption and pH levels in culture medium were determined non-invasively by the PreSens technology (PreSens Precision Sensing GmbH). 0.8x10⁶ T cells with anti-CD3/CD28 Dynabeads (with a cell to bead ratio of 1:1, Thermo Fisher Scientific) were seeded in 24-well OxoDish^® OD24 plates without fixation in 1 mL medium under cell culture conditions for the indicated period of time. Data were analyzed using PreSens SDR_v38 software.

After 24h co-culture with pre-activated T cells, spheroids were washed and transferred to a fresh 96-Well Poly-HEMA plate with 200 µl RPMI 1640 (GIBCO, 31870-025), 10% fetal calf serum (Sigma, F7524), 2 mM glutamine (PAN Biotech, P04-80100)) and 25 IU/ml IL-2, and labelled with 20 µl/ml Cyto3D™ Live-Dead Assay Kit dye (TheWell Bioscience, BM01). Plates were incubated in the Incucyte ZOOM live-cell imager (Essen Bioscience, Welwyn Garden City, UK) at 37°C and 5% CO2 and images were acquired (4x or 10x objective) at the indicated time points. Data were analyzed with the Incucyte ZOOM 2020B software (Essen Bioscience) by creating a threshold-based mask for the calculation of the green object total area (GOTA) of viable cells (green = viable, red = dead).

Depending on normal or non-normal distribution, RM one-way ANOVA with Geisser-Greenhouse correction and Dunnett`s multiple comparison test or Friedman test with Dunn`s multiple comparison test was performed. Significance was indicated as p < 0.05 *, p < 0.01 **, p < 0.001 *** referring to control. Data were corrected for multiple testing according to Benjamini and Hochberg as indicated (retrieved from https://statistikguru.de/rechner/adjustierung-des-alphaniveaus.html).

Kishton et al. acknowledged effector T cells to exhibit an enhanced metabolic activity in-vitro, characterized by a high glycolytic activity and reactive oxygen species (ROS) production, resulting in a strong proliferation and cytokine production during expansion. However, upon in-vivo transfer, these cells showed a poor persistence and anti-tumor activity.

On the contrary, T cells exhibiting a balanced metabolic activity and a memory phenotype in-vitro were associated with a high anti-tumor activity and increased persistence in-vivo (37, 38). Furthermore, Gicobi et al. recently demonstrated that resilient T cells, which were resistant in a harsh tumor microenvironment and responsive to immunotherapy, compensated for excessive ROS to maintain metabolic fitness and preserve high cytotoxic capacity (39).

Therefore, we were intrigued to see, if the presence of LDL^high was linked to a balanced T cell metabolism.

T cells were freshly isolated and stimulated with anti-CD3/CD28 beads, IL-2 and treated with different LDL dosages (LDL^low = 50 µg/ml and LDL^high = 200 µg/ml) versus medium control. We analyzed T cell proliferation, mitochondrial metabolism, intracellular cholesterol, acid production, respiration, and ROS accumulation ( Figure 1 , Supplemental Data S3 , 6 , 8 , 9 statistics and data points).

T cell oxygen consumption was significantly reduced in the presence of LDL^high, indicating a reduced turnover by oxidative phosphorylation. Furthermore, the CD4⁺ T cell subset showed less acidification by trend, however glycolytic activity was preserved ( Figures 1A–H ).

T cell proliferation was strongly impaired by LDL^high after 96 h, cell counts x 10⁵/ml hardly differed from pre-proliferation cell counts after 48 h ( Figures 1I, J ).

ROS accumulation was significantly reduced at both time points in both subsets ( Figures 1K, L ). However, we did not find any significant differences in the entire CD4⁺ and CD8⁺ T cell subset, respectively, concerning the mitochondrial mass, the mitochondrial membrane potential and intracellular cholesterol after 48 h and 96 h ( Supplemental Data 3 and 6 ).

The synopsis of the findings provided strong indications for a balanced metabolism in the presence of LDL^high in both T cell subsets. As a balanced metabolic activity has been associated with a central memory phenotype, we investigated for a phenotypical shift in the T cell subsets (37).

Central memory T cells (T_CM) have been shown to exhibit a superior persistence and anti-tumor immunity compared to effector memory T cells (T_EM) (40). T_CM were associated with a favorable prognosis in oral squamous cell carcinoma and gastric cancer. Furthermore, a predominance of T_CM predicted response to checkpoint therapy in Merkel cell carcinoma (41–43).

CD4⁺ and CD8⁺ T cells were stimulated with activating beads and IL-2 in the presence of medium control, LDL 50µg/ml (LDL^low) or LDL 200µg/ml (LDL^high), respectively. Memory markers were analyzed after 48 h and 96 h.

CD4⁺ T cells shifted toward CD45RO⁺ CD62L⁺ central memory phenotype in the presence of LDL^high after 96 h ( Figure 2A ).

We also observed a trend towards a T_CM phenotype in the CD8⁺ subset after 96 h, however the data were not significant. No effects were seen after 48 h.

A preliminary experiment also revealed the up-regulation of CD45RO⁺ CCR7⁺ cells in the CD4⁺ T cell subset by LDL ( Figure 2C ), however there was only a limited fraction of CD62L/CCR7 double positive cells ( Figure 2D ), which may be traced back to a mixed memory phenotype and alternatively shedding of CD62L (44, 45).

Furthermore, completing the phenotype, the expression of CD44 was significantly enhanced by LDL^high in the CD4⁺ subset after 96 h ( Figure 2E ). Besides its function as an activation marker and high expression on memory cells, CD44 can promote survival and memory cell development in Th1 cells (46). No effects were seen regarding the expression of CD27, CD28 and FOXP3. T cells were mostly positive for CD27 and CD28, and FOXP3 was barely expressed under all conditions ( Supplemental Data S1 – 5 : FACS gating, statistics and data points).

In a further step, we investigated the impact of LDL on memory cell-modulating cytokines. IL-21 has been associated with the induction of a central memory phenotype (47, 48) and IL-7 and IL-15 have been linked to maintenance of a long-term memory survival (49). Autocrine production of IL-7 and IL-15 has been reported (50, 51). Moreover, IL-10 has been linked to suppression of memory development and memory cell responses (52, 53). We could not find any evidence for memory cell induction and maintenance by IL-21, IL-7 and IL-15. Secretion of IL-21 was downregulated and IL-7 and IL-15 could not be detected in the presence of LDL^high.

However, secretion of IL-10 was significantly impaired by LDL^high, possibly thereby enabling memory formation ( Figure 2F , Supplemental Data S6 , 7 ).

Besides downregulation of ROS and shift towards of a central memory phenotype, the mitochondrial membrane potential has been acknowledged to identify cells with a balanced metabolism and an enhanced stemness for cellular therapy (54). Subtyping for T_CM phenotype after 96 h revealed a significant reduction of the mitochondrial membrane potential in the CD8⁺ subset ( Figure 2B ), perhaps indicating resilient T cells with a high cytotoxic capacity (39).

In a further step we also investigated phenotypical changes concerning exhaustion, activation, and co-stimulatory surface markers ( Table 1 ).

CD4⁺ and CD8⁺ T cells were stimulated with activating beads and IL-2 in the presence of medium control, LDL 50µg/ml (LDL^low) or LDL 200 µg/ml (LDL^high), respectively. Surface checkpoint markers, costimulatory markers and activation markers were analyzed after 48 h and 96 h.

In the group of checkpoint markers, LDL^high induced a significant down-regulation of the fraction (%) of PD-1 positive cells in both T cell subsets after 48 h and in the CD4⁺ subset also after 96 h ( Figure 3A ).

The expression (MFI) of PD-1 was significantly reduced in the CD4⁺ subset after 48 h and in both subsets after 96 h ( Figure 3B ). The fraction (%) of LAG-3⁺ T cells was significantly reduced in both T cell subsets after 96 h ( Figure 3C ), the expression (MFI) of LAG-3 was reduced temporarily in the CD4⁺ subset after 48h. High expression of PD-1 and LAG-3 have been shown to be associated with a loss of T cell function (55–58). Downregulation of these exhaustion and suppression markers, especially LAG-3, may enhance the efficacy of PD-1 blockade (59–61).

Intriguingly, in the presence of LDL^low, but not LDL^high, PD-L1 was significantly up-regulated in the CD4⁺ subset after 48 h and 96 h. Although the expression of PD-L1 on CD4⁺ T cells was associated with an improved PFS in NSCLC patients (62), PD-L1 signaling on human memory CD4⁺ T cells induced a regulatory phenotype (63). The expression of all other checkpoint markers was not significantly affected by LDL ( Supplemental Data S1 – 5 ).

In the group of co-stimulatory markers the fraction and expression of CD154 (CD40L) was significantly up-regulated in the CD4⁺ T cell subset after 96 h ( Figures 3D, E ). Interaction of CD154 with CD40 has been demonstrated to mediate anti-tumoral immune responses by enhancing the immunogenic cell death of tumor cells, activation of antigen presenting cells, production of proinflammatory factors, co-stimulation of CD4⁺ and CD8⁺ T cells, and the tumor cell susceptibility to T cell lysis (64, 65).

After 48 h, OX40⁺ T cells were temporarily reduced in the CD4⁺ subset under both conditions containing LDL, however no differences were seen after 96 h. The expression of all other co-stimulatory markers was not significantly affected by LDL. Furthermore, in the group of activation markers and other markers, expression of CD25 and the LDL receptor were temporarily impaired and CD95 temporarily up-regulated under both conditions containing LDL referring to control after 48 h. No differences were seen after 96 h ( Supplemental Data S1 – 5 ).

As the T cells exhibited a less exhausted phenotype, we were also intrigued to investigate cytokine secretion. In monoculture, we could not reveal any significant differences for IFNg, TNFa, granzyme B, IL-17 and IL-4, however the production of perforin was enhanced in both T cell subsets by trend ( Supplemental Data S3 , 6 , 7 ).

To complete our understanding of the phenotypical und functional alterations induced by LDL, we investigated in a further step functional properties in a spheroid model.

Spheroids were generated for 4 days. In parallel, T cells were pre-activated with anti-CD3/CD28 beads and IL-2. After 2 days of stimulation with anti-PD-1, LDL^high or LDL^high + anti-PD-1 versus medium control, T cells were added to the spheroids and allowed to infiltrate for 24 h. Subsequently, the co-cultured spheroids were stained with a viability dye (red = dead, green = viable). Fluorescence was monitored for further 48 h ( Figure 4 , Supplemental Data S10 , Video S1 ).

We did not find any significant differences concerning viability subsequently to the sole addition of an anti-PD-1 antibody or LDL^high in comparison to medium control. However, combination of LDL200µg/ml with an anti-PD-1 antibody induced a significant reduction of the normalized spheroid green object total area (GOTA, representing viable cells) compared to medium control after 24 h and 48 h ( Figures 4C, D ; n = 5).