Peripheral blood mononuclear cell hyperresponsiveness in patients with premature myocardial infarction without traditional risk factors

Summary An increasing number of patients develop an atherothrombotic myocardial infarction (MI) in the absence of standard modifiable risk factors (SMuRFs). Monocytes and macrophages regulate the development of atherosclerosis, and monocytes can adopt a long-term hyperinflammatory phenotype by epigenetic reprogramming, which can contribute to atherogenesis (called “trained immunity”). We assessed circulating monocyte phenotype and function and specific histone marks associated with trained immunity in SMuRFless patients with MI and matched healthy controls. Even in the absence of systemic inflammation, monocytes from SMuRFless patients with MI had an increased overall cytokine production capacity, with the strongest difference for LPS-induced interleukin-10 production, which was associated with an enrichment of the permissive histone marker H3K4me3 at the promoter region. Considering the lack of intervenable risk factors in these patients, trained immunity could be a promising target for future therapy.


INTRODUCTION
Atherosclerosis is a major cause of worldwide morbidity and mortality, with myocardial infarction (MI) being one of the most severe manifestations of this process. 1 The development of atherosclerosis is associated with modifiable risk factors such as dyslipoproteinemia, hypertension, diabetes mellitus, smoking, and obesity. Over the past decades, improved control of these risk factors, e.g. with lipid-lowering therapy, has led to significant improvements in cardiovascular outcome.
There is however a significant proportion of patients that develops MI in the absence of these risk factors. 2 Strikingly, these patients have an increased risk of in-hospital mortality compared to patients with at least one standard modifiable risk factor (SMuRF). 3 These so called ''SMuRFless'' patients with MI lack the traditional targets for pharmacological intervention, even though atherosclerosis is clearly present. Therefore, unraveling the pathophysiological processes that drive the increased cardiovascular risk in this specific group of patients is imperative in order to develop more effective treatment strategies.
The concept that atherosclerosis is merely caused by the passive accumulation of cholesterol in the vessel wall has dominated the field of cardiovascular disease for many years. Nowadays the picture is much more complex with atherosclerosis being established as a chronic, low-grade inflammatory disorder of the arterial wall. The CANTOS trial was one of the first randomized controlled trials targeting inflammation in patients who suffered a previous MI and had elevated C-reactive protein (CRP) concentrations. 4 In this landmark trial, treatment with the interleukin (IL)-1b neutralizing antibody canakinumab reduced the risk of subsequent major cardiovascular events independent of circulating lipid concentrations. More recently, the nonspecific anti-inflammatory drug colchicine has also been shown to reduce cardiovascular event rate in high-risk patients. 5,6 Cells of the innate immune system, especially monocyte-derived macrophages, are the most abundant cells in the atherosclerotic plaque and play a pivotal role in different stages of atherosclerosis. 7 Recently, we and others showed that monocytes can build an immunological memory, which could contribute to persistent atherosclerotic inflammation. 8 This mechanism, termed ''trained immunity'', is mediated through distinct epigenetic and metabolic reprogramming. 9 Trained immunity can be induced by microorganisms, but also by endogenous atherogenic stimuli, such as lipoproteins, glucose, and catecholamines. [10][11][12][13] Trained monocytes are characterized by enhanced cytokine production capacity and enrichment of activating epigenetic histone markers such as histone 3 lysine 4 trimethylation (H3K4me3), or a reduction in repressive H3K9me3 and H3K27me3 on the corresponding cytokine genes. 14 We have recently demonstrated the presence of these markers in patients with increased cardiovascular risk due to familial hypercholesterolemia (FH) or pheochromocytoma. 13,15 We hypothesize that in SMuRFless patients who suffered an MI, in particular those at young age, immune cell activation and more specifically trained immunity contributes to the pathophysiology of cardiovascular disease. Therefore, we assessed circulating monocytes in SMuRFless patients with MI and matched controls without coronary artery disease. We analyzed monocyte phenotype and function, and specific histone marks associated with trained immunity.

Baseline characteristics
The clinical characteristics of patients with MI and controls are listed in Table 1. Patients and controls were matched for age, sex, BMI, and smoking status. Additionally, there were no significant differences between subjects in blood pressure or family history of cardiovascular disease. In patients, the MI had occurred at a median of 25 (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38) months prior to inclusion. In controls, there was a median of 28  months between the CT-scan and inclusion. The proportion of active smokers was low in both groups, with 3 (15%) active smokers among patients and 2 (11%) among controls. Evidently, in the patients, there was significantly more use of aspirin, statins, angiotensin-converting enzyme inhibitors/ angiotensin receptor blockers, and b-blockers compared to controls because of their previous MI. Statin therapy was successfully interrupted in all 20 patients with MI, and aspirin therapy was successfully completed in all 18 control participants during the 14 days prior to blood sampling. Laboratory characteristics for both groups are listed in Table 1. There were no significant differences between groups in total cholesterol, low-density lipoprotein cholesterol (LDLc), or triglycerides after 2 weeks of statin cessation. High-density lipoprotein cholesterol was slightly lower in the MI group compared to the control group (1.2 (1.0-1.4) vs. 1.4 (1.2-1.6) mmol/L (p = 0.03)). The median levels of both fasting glucose and lipoprotein(a) (Lp(a)) were not elevated, and did not significantly differ between groups. The clinical characteristics of patients with MI at the time of the index event and lipid levels before interruption of statin therapy are included in Tables S1 and S2, respectively.

Circulating markers of systemic inflammation
A targeted proteome platform was used to measure 96 inflammatory markers in plasma samples of all subjects. Overall, there was no clear separation of the patients and controls based on circulating inflammatory markers ( Figure 1B). Moreover, assessing each protein separately, there was no specific marker that was significantly higher in patients or controls ( Figure 1A). Since CRP is the most widely used marker for (low grade) systemic inflammation, we performed separate ELISA measurements to determine circulating CRP levels. Notably, CRP levels did not significantly differ between patients (0.91 (0.38-1.83) mg/L) and controls (1.54 (0.77-3.49) mg/L (p = 0.11)) ( Figure 1C).

Circulating monocyte counts and subsets are similar in SMuRFless patients and controls
Circulating leukocyte counts in patients and controls are shown in Table 1. No differences were seen in absolute or relative counts of different leukocytes between patients and controls. Neutrophil-lymphocyte and monocyte-lymphocyte ratios did not differ either. Using flow cytometry, we performed a more detailed analysis of blood cell phenotype. The unsupervised uniform manifold approximation projection (uMAP) analysis of whole blood in patients and controls is shown in Figures 2A and 2B. The uMAP did not show any differences in the percentage of monocytes or monocyte subsets between patients and controls based on the cell surface expression markers. Monocytes and monocyte subset populations were also studied in more detail using manual gating. No differences were seen in the percentage of classical, intermediate, or nonclassical monocytes ( Figure 2C). Moreover, there was no difference in the percentage of monocytes iScience Article that expressed CD11b or CD41. The median fluorescent intensity of CD11b, CD41, and HLA-DR did not differ either ( Figure 2C).
Within other leukocyte populations, uMAP analysis demonstrated differences within the granulocytes, natural killer (NK) cells, and NK-T cells between patients and controls ( Figure 2B). Further preliminary assessment of the granulocyte population revealed an upregulation of the activation marker CD16 (p = 0.013) on neutrophils from patients compared to controls. Although HLA-DR expression was also numerically higher in patients, this difference was not statistically significant (p = 0.081). CD11b expression was similar in both groups ( Figure S3).
Data are presented as median and interquartile range for continuous-and percentages for categorical variables. See also Tables S1 and S2.  Figure 3A). We subsequently performed PCA with only those genes identified as being related to inflammation according to GO term GO:0006954 and also this did not separate patients from controls ( Figure S5). Differential expression analysis identified only three genes that were significant following multiple testing correction (FDR %0.05): MXRA7 (p adj = 0.012), and RAB20 (p adj = 0.035) were significantly lower expressed in patients, whereas PTK2 (p adj = 0.02) was higher expressed ( Figure 3B). Pathway analysis of the 100 most up-or downregulated genes using Reactome 16 or WikiPathways 17 did not identify any significantly enriched pathways after multiple testing correction.

PBMCs isolated from SMuRFLess patients with MI show enhanced cytokine production capacity
To investigate whether peripheral blood mononuclear cells (PBMCs) isolated from SMuRFless patients with MI are functionally different from controls, we stimulated PBMCs ex vivo with Toll-like receptor 4 and 2 ligands (lipopolysaccharide (LPS) and Pam3Cys) for 24 h and measured cytokine production of TNF-a, IL-6, IL-1b, IL-1Ra, and IL-10. The concentrations of all individual cytokines we measured were consistently numerically higher in patients versus controls ( Figure 4A). However, most cytokines did not reach statistical iScience Article significance on their own ( Figure S4). We summarized these cytokine values in one cytokine score, that was previously established, 18 by using principal component analysis as dimensional reduction. We extracted the loadings of PC1 and PC2 which captured 72% of the variance in the data (inclusion of more components did not change the results) for logistic regression. This model then established that there is a statistically significant difference across all cytokine production experiments that distinguishes these patients from matched controls (p = 0.046, Figure 4B). For most study participants, TNF-a concentrations upon P3C stimulation were below the detection limit and were therefore not included in the cytokine score. Since the difference in the cytokine score appeared to be mainly driven by an enhanced production of TNF-a, IL-6, and IL-10 after LPS stimulation ( Figures 4A and 4C), these cytokine genes were selected for further examination of specific histone marks associated with trained immunity. 14 In a previous study in patients with very high LDLc concentrations due to familial hypercholesterolemia, we have reported increased cytokine production capacity compared to normocholesterolemic controls. 15 To explore whether the cytokine production capacity in the study groups was affected by plasma cholesterol levels, we calculated correlations between LDLc concentrations (two weeks after stopping statins) and the cytokine production after stimulation with LPS or Pam3Cys using Spearman correlation. The results are shown in Table S3. A significant positive correlation in the patient group was only found for IL-6 after Pam3Cys stimulation (Spearman r 0.56, p = 0.02) and in the control group for IL-1b after Pam3Cys stimulation (Spearman r 0.56, p = 0.02). Therefore, the circulating LDLc concentrations do not seem to contribute to the differences we observed in cytokine production capacity between the patients and controls.
Hyperresponsiveness in monocytes from SMuRFless patients with MI is associated with more transcriptionally permissive H3K4me3 on cytokine gene promoter In order to investigate whether the hyperresponsive phenotype of the patients is due to trained immunity, we explored the histone markers most often involved; the transcriptionally permissive H3K4me3 and the transcriptionally repressive H3K9me3 and H3K27me3. We performed ChIP-PCR for these epigenetic markers on the promoter regions of the genes encoding TNF-a, IL-6, and IL-10 (the cytokines most markedly upregulated in patients after ex vivo stimulation) ( Figure 5). Levels of H3K4me3 on IL-10 were significantly higher in patients versus controls (p = 0.04). For TNF-a and IL-6, H3K4me3 levels did not reach statistical significance. There were no differences in the repressive markers H3K9me3 and H3K27me3 on the promoters of any of the cytokines between groups.

DISCUSSION
An increasing number of patients suffer from myocardial infarction in the absence of traditional cardiovascular risk factors. 2 Identifying the underlying mechanisms of coronary artery disease in these patients is paramount in order to develop more targeted treatment strategies. In the current study, we show that SMuRFless patients suffering an MI at a young age are characterized by a hyperresponsive state of innate immune cells. This is likely mediated via increased chromatin accessibility at the promoters of inflammatory genes. Notably, this is present without further indications of systemic inflammation, such as increased CRP levels, or widespread baseline transcriptional differences in monocytes. These data suggest that inflammatory reprogramming of circulating monocytes in a process called trained immunity contributes to MI in young SMuRFless patients and offers potential targets for pharmacological prevention.
Monocyte-derived macrophages are the most important immune cells in atherosclerotic plaques and are key in the pathophysiology of MI. There is accumulating evidence that long-term epigenetic monocyte reprogramming, termed trained immunity, contributes to atherogenesis. 19 After transient exposure to pathogens or damage-associated molecular patterns, monocytes and macrophages mount an enhanced pro-inflammatory response upon subsequent encounters with similar or unrelated stimuli. This trained iScience Article immune phenotype is mediated via distinct epigenetic and metabolic reprogramming of these cells. 9 In the context of cardiovascular disease (CVD), it has previously been shown that pro-atherogenic particles such as oxidized low-density lipoprotein (oxLDL) can induce trained immunity via enrichment of H3K4me3 at the level of pro-inflammatory cytokine genes. 10 This epigenetic phenotype is also seen in patients with an increased CVD risk due to FH or pheochromocytoma. 13,15 Christ et al. showed that in Ldlr À/À mice, a Western-type diet induced this functional immune cell reprogramming along with systemic inflammation. 20 Interestingly, after switching back to chow diet, the systemic inflammation subsided whereas the myeloid cell responses toward innate immune stimuli remained augmented. This persistent change was mediated by transcriptional and epigenetic reprogramming at the level of hematopoietic stem and progenitor cells.
In the current study, ex vivo stimulated monocytes from SMuRFless patients with MI mounted an exaggerated overall cytokine response compared to their matched controls. Analysis of the individual cytokines revealed a higher IL-10 production after LPS stimulation, and a trend for TNF and IL-6 (p = 0.066). The enhanced cytokine production capacity of these cells was associated with a difference in the histone modification at the level of the promoter of cytokine genes, which is characteristic of trained immunity. Patients showed an enrichment of the activating histone methylation marker H3K4me3 on the promoter of IL-10, whereas no differences were found in the repressive markers H3K9me3 and H3K27me3. Although H3K4me3 is the most classical histone marker implicated in trained immunity, many other histone markers and epigenetic processes are involved in shaping overall chromatin accessibility in trained immunity. 21 Consistent with our findings, patients with FH and patients with pheochromocytoma with increased risk of CVD are also characterized by a hyperresponsive monocyte phenotype, which coincides with H3K4me3 enrichment. 13,15 However, in a previous study in patients with traditional risk factors and established coronary atherosclerosis, the augmented cytokine production was accompanied by a lower H3K4me3 presence, but also a lower presence of the repressive histone modification H3K27me3. 22 Furthermore, mirroring the findings in the study by Christ et al., this trained immune phenotype was not accompanied by signs of systemic inflammation such as increased levels of CRP or other circulating inflammatory markers.
Although monocytes from SMuRFless patients with MI were hyperresponsive upon stimulation, unstimulated monocytes showed no differences in surface activation markers or transcriptome apart from three genes. This is consistent with the phenotype of trained monocytes, that have returned to their baseline homeostatic state, but are primed to behave hyperresponsive after restimulation. Three genes were differentially expressed in the unstimulated condition. Protein tyrosine kinase 2, a protein involved in cellular adhesion and previously described as possible therapeutic target against atherosclerosis, 23 was iScience Article upregulated in SMuRFless patients with MI, whereas Ras-related protein RAB20 and Matrix Remodeling associated 7 (MXRA7) were downregulated. RAB20 is known to be involved in brain inflammation 24 and a lower expression has been described in patients with obstructive sleep apnea. 25 The MXRA family of genes has been suggested to be involved in the pathophysiology of psoriasis and inflammatory liver disease. 26 Although both genes evidently play a role in inflammatory diseases, to date neither of these genes have been described as being involved in CVD.
Our findings illustrate that while the monocytes from SMuRFless patients with MI seem phenotypically indistinctive from controls in resting state, they are functionally primed to evoke a more aggressive response when triggered. This could, for instance, take place in the circulation or after migration into the atherosclerotic plaque, and thereby contribute to the progression of atherosclerosis. Nevertheless, it remains unknown what stimulus could have induced the trained immune state in these patients in the absence of traditional pro-atherogenic triggers. In vitro, trained immunity can be induced by various endogenous atherogenic molecules such as oxLDL, Lp(a), catecholamines, glucose, and aldosterone. 8 Figure 4. PBMC cytokine production capacity PBMCs isolated from controls and patients were stimulated with TLR agonists LPS and P3C for 24 h before measuring cytokine production. (A) A radar plot shows the mean cytokine production by controls and patients for each stimulation and cytokine. This reveals that the difference between groups is mainly driven by enhanced production of TNF-a, IL-6, and IL-10 upon LPS stimulation. (B) A combined cytokine score reveals an overall elevated cytokine production in patient compared to controls. Data are presented as box and whisker plots with 95% CI. (C) Shows production of TNF-a, IL-6, and IL-10 upon LPS stimulation in more detail. Data are presented as median with interquartile range. N = 18 controls vs. 19 patients. See also Figure S4 and We did not assess stress exposure or perceived stress and related catecholamine exposure in our patients, nor did we measure aldosterone. In addition, training can be induced by previous infections or vaccinations, and it has been proposed as potential mechanism that links the infectious burden with cardiovascular disease prevalence. 27 We did not assess previous infections in our patients. Furthermore, in the patients included in this study, the MI had occurred more than 1 year before blood sampling. This timing was chosen in order to mitigate any acute systemic effects that are known to arise after MI. 28 Although this does not completely rule out any possible long-term effect of the MI itself, it was previously shown that in Bacillus Calmette-Gué rin vaccination-induced trained immunity, the enhanced cytokine production capacity of innate immune cells returned back to baseline after 1 year 29 Finally, there is a large inter-individual variation in the response to stimuli that can induce trained immunity, which is associated with single-nucleotide polymorphisms in various pathways known to be important in this process. 30,31 One may hypothesize that SMuRFless patients with MI are particularly susceptible to stimuli that would normally not be capable of inducing trained immunity. iScience Article The strength of this study is the extensive phenotyping of circulating monocytes in a well characterized and matched group of patients: patients that suffered an MI before the age of 50 without SMuRFs that were carefully matched to control subjects without coronary atherosclerosis. In addition, statins and aspirin use was matched for two weeks. We specifically assessed patients who had undergone an MI at the age of 50 years or younger, as advanced age itself is associated with chronic low-grade inflammation. 32

Limitations of the study
Our study also has a couple of limitations. First, a relatively liberal LDLc cutoff of 4.5 mmol/L (95% percentile of normal distribution) was used to indicate dyslipoproteinemia, but there was no difference in LDLc between patients and controls (3.4 mmol/L in both). Second, we found slightly lower concentrations of HDLc in patients compared to controls, although still in the normal range, which could have potentially contributed to the differences in immune cell function. Third, as a result of the explorative nature of this study, no formal sample size calculation was performed and the sample size was relatively small. However, we have previously shown significant differences in cytokine production capacity and epigenetic markers of trained immunity in studies with comparable group sizes. 13,22 Nonetheless, it is important to validate these results in a larger group of SMuRFless patients with MI. Fourth, since we only included five female participants in both groups, no analysis on sex-differences could be performed. Fifth, we only measured three specific epigenetic markers of trained immunity on the promotor regions of specific cytokine genes, whereas the overall chromatic accessibility is defined by more epigenetic processes, which might be different for different stimuli. Future studies should therefore include a more comprehensive epigenetic evaluation. Last, as we already assessed absolute leukocyte counts using a Sysmex hematology analyzer, we decided not to include count-beads in the flow cytometry measurements, hence limiting interpretation of absolute counts in this specific assay. Moreover, in this study, we mainly focused on characterization of monocytes, as monocyte-derived macrophages are the most abundant immune cells in atherosclerotic plaques. Yet, the contribution of other immune cells should also be investigated in the context of the SMuRFless MI patient group. In the uMAP flow cytometry analysis, we identified a difference in granulocyte subset between patients and controls. Further preliminary analyses indeed showed upregulated expression of CD16 and HLA-DR on the neutrophils of patients compared to controls. Unfortunately, due to the monocyte-focused design of our flow cytometry panel, we could not perform a deeper analysis of these cells. This does offer an interesting new target for further study, as neutrophils can also adopt a trained phenotype, 33 and their involvement in atherosclerosis has recently gained considerable interest. 34,35 In conclusion, our study shows that circulating monocytes in SMuRFless patients with MI are characterized by a hyperresponsive phenotype. This behavior is associated with enrichment of a permissive histone marker at the promoter region of key inflammatory genes. This is indicative of trained immunity and it was observed in the absence of active systemic inflammation. Considering the lack of intervenable risk factors in this specific patient group, this could be a promising target for future therapy.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Patient population
This study was designed as an exploratory study, hence no formal sample size calculation was performed. However, we have previously shown significant differences in cytokine production capacity and epigenetic markers of trained immunity in groups of comparable size. For example, statistically significant differences were found in 10 pheochromocytoma patients versus 14 essential hypertension patients, 13 as well as in 20 patients with symptomatic coronary atherosclerosis versus 20 control subjects. 22 Patients who presented with a type 1 myocardial infarction (MI) caused by atherothrombotic coronary artery disease more than 1 year and less than 4 years prior to the current study were screened for eligibility. This timing was chosen in order to avoid any acute systemic effects of the MI. To be eligible, patients had to be 50 years or younger and have none of the following standard modifiable risk factors (SMuRFs) at the time of MI: diabetes mellitus, hypertension, dyslipoproteinemia (defined as a low-density lipoprotein cholesterol (LDLc) > 4.5 mmol/L), or obesity (defined as BMI > 30 kg/m 2 ). Smoking was not considered a strict exclusion criterium, as it has previously been shown that it does not affect circulating inflammatory markers and peripheral blood mononuclear cell (PBMC) cytokine production capacity, and it was carefully matched in the control group. 41 Additionally, patients had no history of a coagulation-or inflammatory disorder, did not use any anti-inflammatory drugs at the time of blood sampling, and did not have a recent history (within 1 month) of infection or vaccination.
Subjects who had an Agatston score of 0 on a coronary computed tomography (coronary CT-scan) within the last 5 years were screened for eligibility as controls. Control participants were matched to MI patients based on sex assigned at birth, age, BMI, and smoking status.
After screening for eligibility 20 patients and 18 control subjects were included. All of the patients and control subjects completed the study. The baseline characteristics (i.e. age, sex and race) can be found in Table 1.

Blood sampling
Venous blood samples were collected from 20 young SMuRFLess MI patients and 18 matched controls without coronary atherosclerosis. The blood samples from MI patients were obtained after 14 days of interruption of statin treatment, as statins potentially have immunomodulating effects. 42 After 4-5 days, all statins have been completely cleared from the circulation. Circulating markers of inflammation were previously found to increase to pre-statin therapy concentrations within days after interrupting statins. 43,44 We extended this period to two weeks since this would allow also cholesterol levels to rise towards prestatin concentrations. Since aspirin can also affect cytokine production, matched controls received aspirin ll OPEN ACCESS iScience Article