This study constitutes an extension of the MicroEx study (2), where we characterized the overall metabo-inflammatory response to acute exercise in HFrEF. We now specifically examine how natural killer (NK) cells relate to acute exercise-induced alterations in tryptophan–kynurenine metabolism in this population.

HFrEF patients and controls (CON) were recruited at the University Medicine Greifswald. All study visits were scheduled at the same time of day to minimize potential circadian effects. The study was conducted in accordance with the Declaration of Helsinki (2013 revision) and approved by the Institutional Ethics Committee of the University Medicine Greifswald (application no. BB153/17). Written informed consent was obtained from all participants prior to study enrolment. Baseline assessments and blood collection (T1) were performed before participants completed a standardized CPET. A second blood sample was collected immediately post-exercise (T2), followed by a third sample after a two-hour recovery period (T3). Participant characteristics are presented in (Table 1).

For ex vivo NK cell studies, peripheral blood of 3 male and 3 female healthy donors (age: median: 38 years, min: 24 years, max: 53 years; BMI: median: 22.0 kg/m², min: 19.2 kg/m², max: 32.1 kg/m²) was obtained at the Charité – Universitätsmedizin Berlin under a separate protocol approved by the local Ethics Committee (application no. EA2/059/15).

During screening, every participant underwent transthoracic echocardiography for assessment of the left and right ventricular function, heart morphology, and valve function. Participant’s general medical history was collected via questionnaire, covering sex, age, socioeconomic status, pre-existing medical conditions, medications, and daily physical activity. Additionally, weight, height, waist circumference, heart rate, electrocardiogram, and blood pressure were measured (Table 1).

CPET was performed on a stationary bicycle (MasterScreen CPX system [CareFusion, Höchberg, Germany]) according to the modified Jones protocol (14, 15) under the supervision of trained and certified examiners with on-call doctors, as described before (2).

None of the participants had to terminate exercise prematurely, as no clinically significant abnormalities were observed, including echocardiographic signs of ischemia, arrhythmias, a drop in systolic blood pressure >10 mmHg from baseline, circulatory or respiratory disorders, dyscoordination, confusion, severe dizziness, or hypertensive blood pressure responses. Criteria for achieving peak effort included reaching 90% of the age-predicted maximum heart rate (220 − age), a plateau in the VO₂ curve, respiratory quotient (RQ) >1.1, or blood lactate >8 mmol/L.

Within 30 min of collection, 100 μL EDTA-anticoagulated blood from each time point were stained with directly fluorescence labelled antibodies CD3-PE/Cy7 (cat-no. 300316), CD4-AF488 (cat-no. 317420), CD8-BV510 (cat-no. 301048), CD14-BV421 (cat-no. 301830), CD16-AF647 (cat-no. 302020), CD25-AF700 (cat-no. 302622), CD41-APC/Cy7 (cat-no. 303716), CD45-BV711 (cat-no. 304050) and CD127-BV605 (cat-no. 351334), all BioLegend, San Diego, CA, USA). After 20 minutes of incubation, samples were fixed/diluted with 1% paraformaldehyde in phosphate-buffered saline (PBS). Samples were shipped at 4 °C and measured in an Attune NxT Acoustic Focusing Cytometer (Thermo Fisher Scientific) within 36h. The data compensation was performed according to standard procedures prior to analysis. Acquired.fcs files were analyzed in Kaluza (Analysis version 2.1) as described before (16).

A total of twenty cytokines were determined using the Inflammation 20-Plex Human ProcartaPlex™ Panel (Thermo Fisher Scientific) in cleared EDTA plasma according to the manufacturer’s instructions.

Acid Citrate Dextrose (ACD)-anticoagulated blood samples were centrifuged immediately after collection and stored at -80°C. Targeted metabolomics profiling of the plasma samples was performed using the MxP^® Quant 500 Kit (BIOCRATES LifeSciences AG, Innsbruck, Austria) as recommended by the manufacturer and TRP metabolites were assessed using the method described by Fuertig et al. (17) with modification as delineated in the online supplement.

NK cells were isolated from peripheral blood mononuclear cells (PBMC) of healthy donors by negative depletion using the human NK cell isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Seven to eight x 10^4 purified NK cells of each donor were cultured per mL per well of a 12-well plate for 2 days in RPMI medium supplemented with glutamine, 10% fetal bovine serum (FBS) and penicillin (100 U/mL)-streptomycin (100 μg/mL).

In order to study NK cell phenotype shift, freshly isolated NK cells from the same donor were seeded into 2 parallel wells and interleukin (IL)-12 (final concentration 10 ng/mL) was added to the medium of the IL-12 group, while the control group was cultured without additional additives, as described above. At the end of day 2, the cells were harvested and the cell-free supernatants were collected for further experiments.

Cells were stained with fluorochrome-conjugated monoclonal antibodies targeting surface markers to distinguish NK cell subsets: CD3-AF488 (cat-no. 300320), CD14-Pacific Blue (cat-no. 301828), CD16-AF647 (cat-no. 302020), CD45-BV711 (cat-no. 304050) and CD56-BV510 (cat-no.318340), all from BioLegend, San Diego, CA, USA and Sytox (cat-no. S34860, Thermo fisher Scientific). For surface staining, 100 µL of cell suspension was incubated with antibodies for 30 minutes at room temperature in the dark. Cells were then washed with PBS and fixed/diluted with 0.5% paraformaldehyde in PBS. Flow cytometric acquisition was performed using Attune NxT Acoustic Focusing Cytometer (Thermo Fisher Scientific). Flow cytometry data was compensated using standard methodologies prior to analysis to correct for spectral overlap between fluorochromes. Acquired.fcs files were analyzed in Kaluza (Analysis version 2.1). At least 50,000 events were recorded per sample.

PBMCs were isolated from 30 mL of EDTA-anticoagulated blood from six healthy volunteers, after which NK cells were purified by negative selection using the Miltenyi Biotec human NK Cell Isolation Kit. Labeled cells (i.e. non-NK cells) were retained in the magnetic column, while the flow-through contained unlabeled cells (enriched NK cells) and was collected as the “NK cell fraction”. The non-NK cell fraction, consisting predominantly of monocytes and other lymphocytes, was collected separately by subsequential elution after separating the column from the magnet. A total of 5 × 10^5 cells from each fraction (NK cells or non-NK cells) were plated per well in two separate 48-well plates, allowing duplicates for each condition, donor, and cell fraction. NK cells were exposed to IL-12 (10 ng/ml) or left untreated and non-NK cell fraction was exposed to IFN-γ (20 ng/ml). Samples were collected after 4 h from plate 1 and after 24 h of incubation from plate 2. Throughout all experiments, cells were cultured in RPMI as before, but with the addition of L-tryptophan at a final concentration of 5 µM. Unconditioned medium and freshly isolated cells at 0 h were frozen as baseline controls. A graphical overview of the experimental setup is provided in the online supplement.

At each harvest time point, cells and medium were collected and separated by centrifugation (300 RCF, 8 minutes, 4°C). Cells were sonicated, incubated with protease inhibitor (cOmplete™, Merck, cat-no. 04693132001) and centrifuged at 16000 RCF for 15 minutes at 4°C to pellet debris. Cleared supernatant as well as cell-conditioned medium were aliquoted separately and stored at −80°C until measurement. Supernatants and sonicated cells were used for assessment of tryptophan metabolites by mass spectrometry as delineated in the online supplement.

Total RNA was extracted from purified NK cells using the miRNeasy Micro Kit (cat-no. 1071023, Qiagen) according to the manufacturer’s instructions. RNA concentration and purity were assessed using a Nanovue plus spectrophotometer (GE healthcare). Complementary DNA (cDNA) was synthesized from 200 ng of total RNA using the High-capacity cDNA reverse transcription Kit (cat-no. 4368814, Applied biosystems, Thermo Fisher Scientific) following the manufacturer’s protocol. Quantitative real-time PCR (qPCR) was performed using the SYBR Select Master Mix (cat-no. 4472908, Applied biosystems, Thermo Fisher Scientific) on a Viia7 Real-Time PCR Detection System (Applied Biosciences). PCR reactions were run in a final volume of 10 μL containing 5 μL SYBR Select Master Mix, 0.5 μM of each primer, and 1 μL cDNA. Relative IDO1 mRNA expression was calculated using the 2^−ΔΔCt method, normalized to GAPDH, and expressed as fold change relative to the unstimulated control condition. All samples were measured in technical triplicates.

Data were compiled in Excel 2019 and analyzed using R version 4.5.2 (R Foundation for Statistical Computing, Vienna, Austria). Parameters were eliminated for all participants when they were biologically or biochemically redundant, or when more than 10% of values were missing. For all the replicate measurements, median was used for analysis.

Several parameters showed non-normal distributions of the within-subject differences, and sample sizes were small. Therefore, nonparametric methods were chosen throughout. HFrEF and control groups were assessed separately.

Paired Wilcoxon Signed-Rank test was used to compare between two groups of stimulated and unstimulated cells originating from the same donor and isolation batch.

For all comparisons between time points T1 and T2 (assessing changes upon acute exercise/CPET) and between T1 and T3 (assessing recovery), values of the same individuals were compared and therefore paired analyses were employed.

Spearman rho correlations were performed in HFrEF and control groups separately, using the calculated ratios between T2 and T1 to assess acute response to CPET. The Benjamini-Hochberg method was applied to adjust for multiple testing.

An alpha level of 0.05 was applied throughout.

Plasma TRP concentrations were significantly reduced following acute exercise in both control and HFrEF groups (Figure 1A). TRP serves as a substrate in multiple metabolic pathways, including those related to serotonin and melatonin biosynthesis, as well as the kynurenine pathway (Figure 1E- adapted from KEGG: map00380, Tryptophan metabolism) (18, 19). Mass spectrometry–based analysis of additional TRP metabolites indicates elevated flux down the serotonin (5-HT) and KYN (Figure 1B) routes in response to acute exercise, with a greater flux down the KYNA (Figure 1C) route as compared to the serotonin route and differences between controls and HFrEF patients in the utilization of the KYNA versus 3-hydroxy kynurenine (3-HKYN) to QUIN routes (Figures 1D, F). TRP to KYN and subsequently to kynurenate acutely increased at T2 in both patient groups, with accumulation of 3-hydroxykynurenine and QUIN observed at T3 in the HFrEF group (Figures 1B, D, F; Supplementary Figure 1A). Furthermore, the melatonin-to-TRP ratio increased at T3 in both groups (Supplementary Figure 1B). Additionally, indoxyl sulphate, indole-3-acetic acid in the indole pathway increased at T2 in both groups (Supplementary Figures 1C–E).

Shifts in the counts and activation of circulating NK cells and other leukocyte subsets following acute exercise have been proposed to drive the observed shift in TRP downstream metabolic flow (20), but mechanistic validation has not been performed. We performed correlation analyses between the plasma Kyn/Trp ratio and the circulating number of all major leukocyte populations (Supplementary Table 1) and a panel of 19 cytokines (<10% missing values/below detection limit were present for IL-13 and the parameter was therefore not used.) (Supplementary Table 2), as well as between the increase of circulating NK cell counts and the cytokine panel (Supplementary Table 3). Results highlight associated increases in circulating NK and NK-T cell counts and KYN-to-TRP ratio between T1 and T2 in HFrEF patients but not in controls (Figure 2A, Supplementary Figure 3A, Supplementary Table 1). In HFrEF plasma only, change of both, KYN-to-TRP ratio as well as circulating NK cell count, between T1 and T2 correlated with IL-12p70 and IL-1β (Figures 2B, C, Supplementary Tables 2, 3). Due to the stronger association, IL-12p70 was selected for subsequent ex vivo experiments.

Exposure of NK cells isolated from healthy human donors to recombinant IL-12 led to an increased relative abundance of CD56 bright (CD56^bri) NK cell subsets (Figures 3A, B, E, F) and no significant change in the CD56^dim NK cells (Figures 3A–D). IL-12 treatment did not affect cell death rates across NK cell phenotypes (Supplementary Figures 4A–D).

NK cells exposed to IL-12 tended to express higher levels of indoleamine 2,3-dioxygenase (IDO) mRNA (p = 0.074; Figure 3G) and IDO mRNA levels showed a positive association with the relative abundance of CD56^bri (Figures 3J, K) and a negative association with CD16^negCD56^dim NK cells (Figure 3I), while no significant association was observed for CD16_posCD56_dim NK cells (Figure 3H).

Concomitantly, NK cells exposed to IL-12 metabolized TRP to KYN at a higher rate than unstimulated NK cells and release it into the supernatant at 24 h, but not at 4h, resulting in net KYN accumulation over time (Figures 4A, D, E, F). Consistently, KYNA/TRP and QUIN/TRP ratios in the supernatant are increased in the IL-12 group versus the unstimulated NK group at 24 h, but not at 4 h (Figures 4B, C).

Beyond their cytotoxic effector functions, NK cells, particularly CD56^bri subsets, also exert regulatory roles, including cytokine release such as interferon (IFN)-γ (21). Upon exposure to IFN-γ, KYN/TRP ratio in the supernatant of the non-NK cell fraction of peripheral blood mononuclear cells (PBMCs), predominantly lymphocytes and monocytes, increased at 24 h (Figure 4A). At 24 h, the QUIN/TRP ratio within non-NK cell supernatant tended to increase with IFN-γ stimulation (p = 0.059; Figure 4C). The unstimulated non-NK cell fraction converted TRP to xanthurenate (XANA) and melatonin at a higher rate than unstimulated NK cells of the same donor (Figure 4F, Supplementary Figures 7C, E).