All experiments were approved by local authorities (Regierungspräsidium Karlsruhe, date of approval 01/03/2018) and were conducted in accordance with national legislation. Animals were maintained under controlled, pathogen-free conditions (22 ± 2°C, 50–60% humidity, 12 h light/dark cycles), fed ad libitum and were handled according to the Society of Laboratory Animal Science recommendations. In total, 131 apolipoprotein E-deficient (ApoE^−/−; 62 male, 69 female) mice, on a C57BL/6 background and bred in-house (Interfacultary Biomedical Faculty, University of Heidelberg, Heidelberg, Germany), were used in this study. Genotyping was performed by Transnetyx (Cordova, TN, US). Starting at 8 weeks of age, mice were set on a high-cholesterol Western diet (WD) containing 1.25% cholesterol (Altromin, Lage, Germany). WD was maintained for the duration of the experiment, such that at the time of sacrifice, mice had been on a WD for 9.5 weeks.

To determine the most suitable time point ensuring maximum Treg modulation at the day of surgery and thereafter, Tregs were quantified at three different time points after initiating antibody treatment. Thus, in a preliminary experiment, mice were randomly assigned to one of the five treatment groups: anti-CD25, IgG1, IL-2/anti-IL-2 complex, IL-2/IgG2, and PBS as control (all from BioLegend, San Diego, CA, US). Treg reduction was initiated by a single intraperitoneal (i.p.) injection of 250 μg anti-CD25 mAb (clone PC61, #102040); an isotype-matched IgG1 (clone G0114F7, #401916) was used as control. IL-2/anti-IL-2 (IL-2C), IL-2/IgG2 and PBS were i.p. administered daily on three consecutive days. IL-2 complexes were formed of 1 μg recombinant IL-2 and 5 μg anti-IL-2 mAb (clone JES-1A12, #503706) or IgG2 (clone RTK2758, #400544), respectively, and incubated for 30 min at 37°C in PBS (28). Blood was collected from the facial vein 6, 8, and 10 days after anti-CD25 mAb and IgG1 administration, and at day 3, 4, and 5 after the first injection of IL-2C, IL-2/IgG2, and PBS, respectively. In the final model, based on the kinetic of Treg modulation, anti-CD25-mediated Treg decrease was induced 6 days prior to surgery whereas IL-2C-mediated Treg expansion was initiated 3 days preoperatively. Controls were treated accordingly.

After 9 weeks of WD, mice were subjected to a perioperative stress model as previously described (6). Briefly, mice were anesthetized using isoflurane inhalation followed by longitudinal laparotomy (approx. length of 1.5 cm) and 400 μl blood withdrawal from the facial vein, which corresponds to an intraoperative blood loss of ~20%. After 30 min, the abdomen was closed using single-knot sutures. Controls underwent 30-minute general anesthesia (sham). To investigate the effect of surgery on Treg counts, sham mice additionally received a 100 μl blood draw from the facial vein during anesthesia. Three days postoperatively, mice were euthanized, exsanguinated, and perfused through cardiac puncture using 0.9% saline at physiological pressure. Brachiocephalic arteries were harvested, embedded in OCT medium and stored at −80°C until further processed for histologic analyses.

Serial cross-sections were prepared at 5 μm thickness through the entire length of the brachiocephalic artery using a cryomicrotome (Leica Microsystems, Wetzlar, Germany). For the detection of Foxp3⁺ Tregs in atherosclerotic plaques, every 15th paraformaldehyde-fixed tissue section was treated with citrate buffer for antigen retrieval. After cell permeabilization in 0.3% Triton X-100 solution, sections were blocked in 2.5% normal goat serum (Vector Laboratories, #MP-5444-15) followed by overnight incubation with the primary rat anti-mouse Foxp3 antibody (1:100; Thermo Fisher Scientific, #14-5773). Secondary staining was performed using the ImmPRESS®-AP anti-rat IgG polymer detection kit (Vector Laboratories, #MP-5444-15) following the manufacturer's instructions. Sections were subsequently treated with hematoxylin solution (Carl Roth, Karlsruhe, Germany) to counterstain nuclei. Thymic tissue was used as positive control. The number of Foxp3⁺ cells was counted manually and normalized to plaque area.

Every 15th section was stained with hematoxylin and eosin (HE) for quantification of atherosclerotic lesion volume and average stenosis. For the analysis of plaque morphology, mice with a maximum stenosis below 10% were excluded, as these were considered preliminary foam cell accumulations not allowing to reliably evaluate lesion stability. For this exploratory analysis, adjacent sections were stained for collagen and necrosis, for macrophages or smooth muscle cells (SMC).

Collagen content and necrotic core (NC) area were quantified on four cross-sections showing the maximum stenosis and stained with Masson's trichrome (Carl Roth, Karlsruhe, Germany) according to the manufacturer's instructions. Collagen was defined as the area that had stained positive for lightgreen above a set threshold and was expressed relative to whole plaque size. Necrosis was defined as anuclear lesion area with total or almost complete loss of collagen and was expressed as mean necrotic area per plaque.

Immunostaining of total CD68-positive macrophages and M2 macrophages was performed as described elsewhere (29). In brief, after cross-sections were fixed in ice-cold acetone, M2 macrophages were identified by dual staining of CD68 (1:300; #MCA1957GA, Bio-Rad, Hercules, CA, USA) and CD206 (1:100; #ab64693, Abcam, Cambridge, UK) for 3 h at room temperature. Sections were subsequently incubated with anti-rabbit IgG Alexa Fluor-555 (#4413S) and anti-rat IgG Alexa Fluor-488 (#4416S), both from Cell Signaling Technology (Danvers, MA, US) and used at dilutions of 1:1,000. Slides were mounted and cover-slipped using fluorescence mounting medium (Dako North America, Carpinteria, CA, US). The number of total macrophages and M2 macrophages was calculated per plaque area.

For fluorescent staining of SMCs an antibody against α smooth muscle actin (1:400; #A2547, Merck, Darmstadt, Germany) and a polyclonal goat anti-mouse IgG secondary antibody conjugated to Alexa Fluor 488 (1:100; #A-11017, Thermo Fisher Scientific, Waltham, MA, US) were used. Cryosections were fixed in ice-cold acetone before treated with M.O.M blocking reagent (Vector Laboratories, Burlingame, CA, US) to reduce endogenous mouse Ig staining. DAPI was used for nuclei counterstaining. Buried fibrous caps (FC) were identified as SMC-rich layers within lesions, that were covered by newly formed plaque material (30).

Following the Stary criteria (31), plaque morphology was further evaluated by scoring features, that indicate lesion complexity, such as necrosis, intraplaque hemorrhage, and buried FCs at the site of maximum stenosis. Four cross-sections stained for collagen and α smooth muscle actin were analyzed for the presence of necrotic area and buried FCs, respectively. Likewise, four HE and Masson's trichrome stainings were used to detect red blood cells indicating intraplaque bleeding. The presence of each individual feature was graded one scoring point, so that complicated plaques containing all three features were rated three points, whereas absence of either criterion was assigned zero points (6).

Images were captured on an Olympus BX63 microscope (Olympus Life Science Solutions, Waltham, MA, US) and analyzed using the CellSens (Olympus Life Science Solutions, Waltham, MA, US) and Fiji imaging platform (32) in an observer-blinded fashion.

To assess the effect of surgery on Tregs as well as the efficacy and specificity of Treg-modulating treatment, Tregs and leukocyte subpopulations were quantified using flow cytometry. EDTA-anticoagulated whole blood or single cell suspensions prepared from lymphoid organs were treated with Fc-block (mouse TruStain FcX™, BioLegend, San Diego, CA, US) to inhibit unspecific binding. Cells were subsequently stained with fixable viability dye (Thermo Fisher Scientific, Waltham, MA, USA) and fluorescently labeled surface antibodies: anti-CD3 FITC, anti-CD4 PerCP-Cy5.5, anti-CD8 APC-Cy7, anti-CD11b PE, anti-CD25 APC, anti-Ly6-C PerCP-Cy5.5, anti-NK1.1 APC-Cy7 (all from BioLegend, San Diego, CA, US), anti-CD19 APC, anti-Ly6-G FITC (both from BD Biosciences, Heidelberg, Germany), and anti-CD45 eFluor® 450 (Thermo Fisher Scientific, Waltham, MA, USA). For Treg identification, intracellular staining of Foxp3 was performed using an anti-Foxp3 mAb (clone 3G3, #A18690) and the Foxp3 transcription buffer staining set (both from Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Isotype controls were included for CD25 staining (IgG1, clone RT4530, BioLegend, San Diego, CA, US). Flow cytometric analyses were performed on a BD FACSVerse™ using the FACSuite Software (BD Biosciences, Heidelberg, Germany) and gating strategies for Treg and leukocyte subpopulation identification are depicted in Supplementary Figures 1, 2, respectively.

Statistical analyses were performed using the GraphPad Prism software. Data were tested for normality using the Kolmogorov-Smirnov test. As part of the data did not follow normal distribution, non-parametric Mann-Whitney U or Wilcoxon signed-rank test were used as appropriate. Kruskal-Wallis test was performed to test for statistical differences between multiple treatment groups. If the Kruskal-Wallis test revealed p < 0.05, Dunn's post-hoc test was performed for comparison of anti-CD25 mAb and IL-2C to PBS treated mice, respectively. Differences were considered statistically significant, if p < 0.05. Outliers identified based on the ROUT method were excluded from the histomorphometric analyses (33). Data are presented as median (interquartile range); boxes mark interquartile ranges and whiskers represent 5th to 95th percentiles. Median and average values are shown as vertical lines and “+,” respectively. Sex-stratified data for each analysis are included in the Supplementary Material to allow the assessment of sex differences. However, as the study was not adequately powered to separately assess gender-specific effects, descriptive data are provided only.

To analyze the effect of surgical stress on Treg counts, Tregs were quantified in blood, spleen, lymph nodes, thymus and atherosclerotic plaques in the brachiocephalic artery of mice that either underwent surgery or the sham intervention. Surgery induced a 50% increase in CD4⁺CD25⁺Foxp3⁺ Tregs in blood [5.3 (4.6; 5.5) vs. 7.9 (6.2; 8.2) %CD4⁺ for pre-OP vs. POD3, p = 0.031], an effect that was absent in sham mice [5.0 (4.3; 6.8) vs. 6.4 (5.5; 6.9) %CD4⁺ for pre-OP vs. POD3, p = 0.406] (Figure 1A). There was no difference in postoperative Treg levels in spleen [7.1 (5.8; 12) vs. 7.1 (5.0; 12) %CD4⁺ for sham vs. surgery, p = 0.937] or lymph nodes [11 (10, 13) vs. 13 (12, 19) %CD4⁺ for sham vs. surgery, p = 0.093] (Figure 1B). In the thymus, mice with surgery showed higher levels in double (DP) and single positive (SP) Tregs compared to sham animals [DP: 0.12 (0.09; 0.12) vs. 0.25 (0.13; 0.4) %CD4⁺ for sham vs. surgery, p = 0.037; SP: 1.5 (1.4; 1.7) vs. 1.9 (1.7; 2.4) %CD4⁺ for sham vs. surgery, p = 0.009] (Figure 1C). The number of Tregs within atherosclerotic plaques was 3.7 (0.0; 10) Foxp3⁺ cells per μm² plaque x10⁻⁴ in mice that underwent surgery vs. 0.0 (0.0; 3.4) in sham animals (p = 0.275) (Figures 1D–F). In an exploratory approach, we stratified data for male vs. female. Respective data for Treg alterations in response to surgery for male and female mice are reported in Supplementary Figure 3.

As the rapid expansion of Tregs could be interpreted as a protective effect to counterbalance surgery-induced inflammation, we analyzed whether modulation of preoperative Treg levels may affect inflammation-driven atherosclerotic plaque progression during surgery.

To ensure maximum Treg modulation at the day of surgery and 3 days thereafter, we determined peripheral Treg levels at three different time points after treatment initiation. Compared to IgG1 controls, a single anti-CD25 mAb administration resulted in 60% decrease of peripheral Treg levels 6 [5.7 (5.0; 6.4) vs. 2.5 (1.7; 3.4) %CD4⁺, p = 0.002] and 10 days [7.8 (6.2; 8.7) vs. 3.1 (2.3; 4.6) %CD4⁺, p = 0.004] after injection. At 8 days, Tregs were reduced from 4.9 (4.4; 6.0) to 3.2 (2.4; 3.5) %CD4⁺ (p = 0.002) (Supplementary Figure 4A). In PBS-treated mice, Tregs accounted for 4.4 (3.3; 5.6) % of the CD4⁺ T cell pool on average. Treg expansion was induced by three daily injections of IL-2C. Treg counts increased to 21 (16, 32) %CD4⁺ (p = 0.006) and 21 (11, 29) %CD4⁺ (p = 0.002) 3 and 5 days after treatment start, respectively, and reached its peak on day four after treatment initiation [28 (18, 34) %CD4⁺, p < 0.001]. Animals treated with IL-2/IgG2 control did not develop increased Treg numbers. The maximum value was seen 4 days after the first injection [6.6 (5.6; 7.5) %CD4⁺, p = 0.272] (Supplementary Figure 4B). Based on these results, we implemented the final model with Treg downregulation 6 days before surgery (Figure 2A), while upregulation was induced 3 days preoperatively (Figure 2B). Accordingly, at the day of surgery, IL-2C-treated mice presented with 8-fold elevated peripheral Tregs compared to mice receiving anti-CD25 mAb (Figure 2C) (Kruskal-Wallis p = 0.0004). Other peripheral leukocyte subpopulations were not significantly affected by our Treg-modulating treatment (Supplementary Figure 5).

To investigate the effect of preoperative Treg levels on perioperative atherosclerotic plaque progression, 17-week-old atherosclerotic ApoE^−/− mice with modulated Treg levels underwent a surgical stress model. Morphological characterization of the atherosclerotic lesion phenotype was assessed 3 days post-surgery in cross-sections of the brachiocephalic artery.

Independent of the preoperative Treg level, surgical stress inflicted by laparotomy and moderate blood loss had no effect on perioperative plaque volume growth as compared with sham animals, which received general anesthesia only (PBS: p = 0.780; anti-CD25 mAb: p = 0.579; IL-2C: p = 0.400) (Figure 3A). Moreover, statistical comparison of surgical intervention groups revealed no difference in postoperative plaque volume (Kruskal-Wallis p = 0.960). Similar results were obtained for the analysis of mean percentage stenosis (PBS: p = 0.481; anti-CD25 mAb: p = 0.842; IL-2C: p = 0.436) (Figures 3B,C) as well as for plaque area at the site of maximal stenosis (PBS: p = 0.905; anti-CD25 mAb: p = 0.684; IL-2C: p = 0.401) (Supplementary Figure 6A). Likewise, we did neither observe any quantitative difference in the extent of postoperative stenosis (Kruskal-Wallis p = 0.693) nor in plaque size at the site of maximum stenosis among surgical intervention groups (Kruskal-Wallis p = 0.439). These observations were comparable between male and female mice under study. However, overall plaque size was larger in male compared to female mice (Supplementary Figures 6B,C, 7).

To further assess the effect of preoperative Treg levels on perioperative atherosclerotic lesion progression, multiple markers of plaque instability were analyzed using histochemical stainings. In total, 12 mice had to be excluded from the analysis (PBS: n = 6; anti-CD25 mAb: n = 4; IL-2C: n = 2) as the maximum stenosis was below 10%, thereby not allowing a reliable evaluation of plaque morphology.

Tregs are known to promote collagen synthesis (28), which largely contributes to plaque mechanical strength and resilience to shear stress (34). Collagen content was reduced by 20% in response to surgical stress in mice with preoperatively high Treg level only [IL-2C; 27 (23, 30) vs. 21 (18, 25) % for sham vs. surgery, p = 0.034]. However, no differences were observed among surgical intervention groups (Kruskal-Wallis p = 0.258). The comparison of sham groups showed the amount of collagen to be significantly increased by short-term IL-2C treatment when compared to PBS-injected animals [Kruskal-Wallis p = 0.001; 27 (23, 30) vs. 17 (15, 22) % collagen for IL-2C vs. PBS, p = 0.002] (Figure 4A). There were no sex-specific differences (Supplementary Figures 8A,B).

The presence and extent of NC area strongly correlates with atherosclerotic plaque vulnerability (35, 36). Assessment of lesional NC area showed an increase from 0.0 (0.0; 3.5) to 11 (5.2; 14) % in reponse to surgery in mice with normal preoperative Treg level (PBS; p < 0.001). Similarly, mice with preoperatively low Tregs showed a 7-fold increase of plaque necrosis after surgery [anti-CD25 mAb; 1.9 (0.2; 5.2) vs. 7.8 (2.7; 12.8) % for sham vs. surgery, p = 0.037]. This effect was absent in mice with preoperatively elevated Treg counts [IL-2C; 3.3 (0.1; 7.4) % vs. 5.0 (1.4; 7.7) %, p = 0.556]. Comparison of surgical treatment groups confirmed a statistical differenc among postoperative NC area of PBS and IL-2C-treated mice (Kruskal-Wallis p = 0.029; Dunn's post-hoc PBS vs. IL-2C: p = 0.025) (Figures 4B,C). Increase of NC area in PBS and anti-CD25 mAb-adminstered mice was present in both, male and female animals. Lack of perioperative NC formation in the IL-2C intervention group could mainly be attributed to female mice. However, compared to other treatment groups, NC area increase was rather small in male mice with preoperative high Treg count (Supplementary Figures 8C,D).

As NC formation is closely related to macrophage apoptosis and efferocytosis, we next quantified the number of total and M2 subtype macrophages per plaque area. The number of lesional CD68⁺ total macrophages remained unchanged independent of preoperative Treg level and surgical stress (PBS: p = 0.943; anti-CD25 mAb: p = 0.879; IL-2C: p = 0.497) (Figure 5A). No statistical differences were observed in the number of alternatively activated, anti-inflammatory M2 macrophages (PBS: p = 0.524; IL-2C: p = 0.777). Though, mice with preoperatively low Tregs tended to develop decreased lesional M2 content in response to surgical stress [anti-CD25 mAb; 2.1 (0.9; 3.1) vs. 0.7 (0.3; 1.8) μm² plaque area x10⁻⁴ for sham vs. surgery, p = 0.072] (Figures 5B,C). There was no difference between surgical intervention groups (Kruskal-Wallis p = 0.282). Observations were similar in both, male and female mice under study (Supplementary Figure 9).

SMCs represent the major source of extracellular matrix significantly promoting plaque stability in advanced atherosclerosis (37). Relative SMC content did not differ between groups (PBS: p = 0.943; anti-CD25 mAb: p = 0.959; IL-2C: p = 0.661) (Figure 6A) and independent of sex (Supplementary Figures 10A,B). Likewise, the number of buried FCs, possibly indicating previous episodes of plaque rupture (38), did not change in response to surgical stress and independent of the preoperative Treg level (PBS: p = 0.730; IL-2C: p = 0.827). Conversly, a trend toward an increase of buried FCs was seen in anti-CD25 mAb-treated mice with low Treg count with some plaques showing signs of up to four preceding rupture events (p = 0.118) (Figures 6B,C). Postoperative SMC content (Kruskal-Wallis p = 0.178) and the number of buried FCs (Kruskal-Wallis p = 0.575) did not differ between mice that underwent surgery. Overall, atherosclerotic plaques of female mice tended to display a lower number of buried FCs males (Supplementary Figures 10C,D).

Atherosclerotic plaque complexity was further classified using a gradual score based on the presence or absence of necrosis, buried FCs, and intraplaque hemorrhage. There was no difference in postoperative lesion complexity between PBS- and anti-CD25 mAb-treated mice (p > 0.999). However, postoperative plaques of mice with high preoperative Treg levels presented with a more stable phenotype compared to controls (p = 0.036) (Figures 7A,B). This trend was present in both, male and female mice under study. However, male mice generally presented with a more complex plaque phenotype than females (Supplementary Figure 11).