The first century Roman doctor Cornelius Celsus described the four cardinal signs of inflammation,
rubor et tumor cum
calore et dolore (redness and swelling with heat and pain) [
16]. Only in 1858, the fifth cardinal sign,
functio laesa (disturbance of function), was added by Rudolph Virchow [
17]. In contrast to the four cardinal signs, which only apply to acute inflammation accompanying wounds and infections,
functio laesa is the only universal sign of inflammation [
16]. A typical inflammatory response consists of four components: (1) the inflammatory inducers, classified in exogenous (microbial inducers including pathogen-associated molecular patterns (PAMPs), virulence factors, and non-microbial inducers: allergens, toxic compounds, irritants) and endogenous inducers (danger-associated molecular patterns (DAMPs): cell-, tissue-, plasma-, extracellular matrix (ECM)-derived products); (2) the sensors that detect them including pattern recognition receptors (PRRs) or other sensors like the nucleotide-binding oligomerization domain-like receptor with a pyrin domain 3 (NLRP3) inflammasome; (3) the inflammatory mediators induced by the sensors (vasoactive amines and peptides, fragments of complement components, lipid mediators, proteolytic enzymes, chemokines, and cytokines); and (4) the target tissues that are affected by the inflammatory mediators [
16]. With respect to mediators, this review particularly discusses the relevance of cytokines in HF.
Para-inflammation
This response is characterized by no massive tissue injury and a limited inflammatory activation. Therefore, it is termed para-inflammation derived from the Greek prefix
παρα/para for near [
20]. This response relies mainly on tissue-resident MΦs. If tissue malfunction is present for a sustained period, para-inflammation can become chronic [
20]. This form of inflammation often accompanies obesity, the metabolic syndrome, type 2 diabetes, atherosclerosis, aging, and other chronic inflammatory conditions that are associated with modern human diseases. Para-inflammation is consequently also called “low-grade” chronic inflammation and in case of metabolism-triggered inflammation, “meta-inflammation” [
21]. Environmental factors including caloric excess, intake of processed foods, use of antibiotics, and physical inactivity, common to Western lifestyle [
22], as well as endocrine disruptors and early life influences (maternal nutrition, placental function) underlie para-inflammation.
Inflammation Causes Heart Failure
Inflammation triggers HF in its different aspects, ranging from its impact on the pathogenesis of HF including HF-underlying comorbidities like diabetes and obesity [
26,
27], and on pathological substrates underlying heart disease like endothelial dysfunction [
28‐
31] and atherosclerosis [
32], to its influence on the progression and outcome of acute coronary syndrome (ACS) [
33] and HF [
34]. Blood monocyte levels [
35] and splenic activity [
36] can predict cardiovascular events in patients, C-reactive protein levels are higher in patients with recurrent events [
37], and cardiac inflammation is a predictor for a negative outcome in patients with dilated cardiomyopathy [
34]. Inflammatory cytokine (TNF-α, IL-1ß, IL-6) levels are increased in HF patients [
38]. There is a correlation between serum levels of TNF-α and the severity of the disease [
38], and cytokines and cytokine receptors are independent predictors of mortality in patients with advanced HF [
39]. The relevance of inflammation in HF follows from experimental studies in animal models of MI, diabetic cardiomyopathy, pressure overload, and myocarditis using knockout [
40‐
42], or transgenic [
43] mice, or mice treated with anti-inflammatory or immunomodulatory strategies, including antibodies (e.g., TNF-α antibody [
44], IL-6R antibody [
45]), inhibitors (IL-1 converting enzyme inhibitor [
46]), agonists/antagonists of cytokines/chemokines (IL-2 agonist [
47], CCR2 siRNA [
48]), statins [
49], HDL-raising strategies [
29‐
31,
50,
51], cell therapies including mesenchymal stromal cells (MSC) [
52‐
54], and cardiac-derived stromal cells [
55,
56]. The inflammation-induced cardiac pathophysiological mechanisms underlying HF will next shortly be discussed followed by evidence of high-grade and low-grade systemic inflammation affecting HF.
At the latest, since the cytokine hypothesis from the 1990s [
4], it is well established that cytokines exert detrimental effects on the heart. Cytokines like TNF-α and IL-1ß downregulate the expression of Ca
2+-regulating genes including sarcoplasmic reticulum Ca
2+ ATPase [
57] and Ca
2+-release channel [
58], leading to a direct negative inotropic effect as a direct result of alterations in intracellular Ca
2+ homeostasis in the adult cardiac myocyte [
59]. Abnormalities in sarcoplasmic reticulum Ca
2+ release promote on their turn eccentric myocardial remodeling (eccentric hypertrophy, substantial fibrosis, ventricular dilation) and pump failure, ultimately resulting in overt HF, in response to pressure overload [
60]. This points out that inflammation-triggered Ca
2+ dysbalance can contribute to cardiac remodeling, leading to a vicious circle [
61]. TNF-α and IL-1ß further promote cardiomyocyte hypertrophy [
62] and the cytokine IL-6 has been demonstrated to increase cardiomyocyte stiffness via reducing the phosphorylation of titin [
45]. TNF-α also triggers cardiomyocyte apoptosis [
63] and IL-1ß cardiomyocyte pyroptosis [
64].
On cardiac fibroblasts, TNF-α and IL-1ß upregulate angiotensin II type 1 receptors (AT1R) and they induce AT1R density in the post-MI heart [
65]. The upregulation in AT1 receptor expression enhances the angiotensin (Ang) II-mediated cardiac fibroblast responses that favor fibrosis [
66]. TNF-α and IL-1ß neutralization ameliorates Ang II-induced cardiac damage, further supporting synergistic actions of Ang II and TNF-α/IL-1ß [
67]. TNF-α also induces TGF-ß [
68] and increases the expression of cardiac fibroblast lysyl oxidase (LOX) expression through TGF-β and PI3Kinase signaling pathways [
69]. LOX belongs to a family of enzymes [
70], including LOX-like 2, responsible for the crosslinking of ECM proteins, including collagen types I and III. The relevance of LOX-like 2 as therapeutic target of cardiac fibrosis and as biomarker for HF has recently been demonstrated [
71]. TGF-ß induces fibroblasts to transdifferentiate into active myofibroblasts. Those cells are not only active in producing collagens but they also act as inflammatory support cells via their capacity to express chemokines, to release factors inducing adhesion molecules on endothelial cells, and via their ability to stimulate monocytes to express gelatinases facilitating degradation of the basolateral membrane and subsequent infiltration of immune cells in the heart [
23,
24] and their capacity to modulate the MΦ M1/M2 balance [
72]. Furthermore, activated fibroblasts promote cardiomyocyte hypertrophy and dysfunction via the release of pro-fibrotic factors, such as TGF-β1, Ang II, and fibroblast growth factor [
73,
74].
On (cardiac) endothelial cells, pro-inflammatory cytokines induce adhesion molecule expression [
75] and promote subsequent adhesion of immune cells to the endothelium [
76] and transendothelial migration [
77]. They induce apoptosis in cardiac endothelial cells [
78] and oxygen-centered free radicals, which stimulate the elaboration of plasminogen activator inhibitor-1 and collagen by cardiac microvascular endothelial cells. Accordingly, microvascularly mediated inhibition of fibrinolysis may predispose to the persistence of microvascular thrombi, thereby contributing to impaired microcirculatory function, the no-reflow phenomenon, and cardiac dysfunction after ischemia and reperfusion [
79]. TGF-ß and Ang II induce endothelial-to-mesenchymal transition, the transition from an endothelial to a fibroblast phenotype [
29], a phenomenon, which has been shown to contribute to cardiac fibrosis in a landmark study by Zeisberg et al. [
80]. Recently, it has been shown that TNF-α-induced endothelial natriuretic peptide/guanylate cyclase A/cGMP/phosphodiesterase 2 signaling impairs endothelial barrier functions and enhances myocardial inflammatory infiltration in the early phase after an acute infarction [
81].
Inflammatory cytokines further promote structural and electrical atrial remodeling via impairment of gap junctions by changes in connexins and via inducing intracellular Ca
2+-handling abnormalities and atrial fibroblast activation, leading to impaired atrial conduction [
82].
Sources of Inflammation
The cytokines inducing cardiac remodeling and dysfunction can originate from the heart itself (cardiokines) [
83], produced by cardiomyocytes [
84], cardiac endothelial cells [
85], cardiac fibroblasts [
25], cardiac tissue MΦs [
86], and cardiac infiltrated immune cells, or can be of extra-cardiac tissues including adipose tissue, gut, and lymphoid organs. Failing human myocardium expresses abundant quantities of TNF-α [
11]. Cardiomyocytes have TNF-α receptors on their surfaces [
87] and these receptors appear to be released into the circulation during HF [
11]. The importance of TNF-α in HF has experimentally been shown in transgenic mice where chronic cardiomyocyte overexpression of TNF-α resulted in the development of dilated cardiomyopathy with ventricular hypertrophy, ventricular dilatation, interstitial infiltrates, interstitial fibrosis, rare myocyte apoptosis, diminished ejection fraction, attenuation of β
1-adrenergic responsiveness, and expression of atrial natriuretic peptide (ANP) in the ventricle [
43].
NLRP3 is considered necessary for initiating a profound sterile inflammatory response. Cardiac endothelial cells [
85] and cardiac fibroblasts [
25] are both important sources of IL-1ß, one of the endproducts of NLRP3 inflammasome activity. By ischemia/reperfusion injury, the NLRP3 inflammasome is activated as indicated by increased NLRP3 expression, caspase-1 activity, and increased IL-1β and IL-18 production. Simulated ischemia/reperfusion-stimulated NLRP3 inflammasome activation in cardiac microvascular endothelial cells, but not in cardiomyocytes [
85]. In another study, a marked increase in NLRP3, IL-1ß, and IL-18 mRNA expression was found in the left ventricle after MI, primarily located to myocardial fibroblasts [
25]. The relevance of NLRP3 inflammasome activity in HF follows from studies demonstrating that when hearts were isolated from NLRP3-deficient mice, perfused and subjected to global ischemia and reperfusion, a marked improvement of cardiac function and reduction of hypoxic damage was found compared with wild-type hearts [
25], whereas Toldo et al. [
88] showed that the formation of the inflammasome in acute myocarditis is predictive for the NYHA class and outcome.
In the healthy mouse heart, ≈6 to 8% of non-cardiomyocytes are resident MΦs [
86], a number comparable to the frequency of resident MΦs in other tissues. Humans may have comparable numbers, after MI, the MΦ numbers increase in the heart through the combined effects of massive recruitment of circulating monocytes (that become macrophages in tissues) and local self-renewal of tissue-resident MΦs [
89]. MΦs are traditionally classified in inflammatory MΦs, often referred to as classical or M1 MΦs, secreting pro-inflammatory cytokines as IL-6, TNF-α, IL-1β, IL-12, and IL-23, and heal/growth-promoting MΦs, commonly called alternatively activated or M2 MΦs, expressing anti-inflammatory IL-10 and TGF-ß [
90]. M1 and M2 MΦs usually appear in sequence upon MI, i.e., in the inflammatory versus the wound healing phase, respectively, whereas also mixed M1/M2 phenotypes can be found [
91]. The relevance of cardiac M1 toward M2 MΦ phenotype transition for the resolution of inflammation and tissue repair post MI has recently been shown by Courties et al. (2014) [
92] who demonstrated that in vivo silencing of the transcription factor IRF5, which is involved in inflammatory M1 MΦ polarization, supported resolution of inflammation, accelerated infarct healing, and attenuated development of post-MI HF.
Besides endogenous cardiac cells, infiltrated inflammatory cells are responsible for local cardiac cytokine expression. Those immune cells originate from lymphoid organs as the spleen and the bone marrow [
93]. Pre-clinical studies have demonstrated that after MI in mice, monocyte progenitor cells depart bone marrow niches, which results in amplified extramedullary monocytopoiesis [
36,
94]. The observation of the activation of splenic monocytes and the migration of pro-inflammatory monocytes from the spleen to the heart in animal models of MI [
95] and chronic HF [
96] have given rise to the concept of a cardiosplenic axis. This recruitment from the spleen depends in part on Ang II, an observation that may underlie the beneficial effect upon angiotensin converting enzyme inhibition on remodeling in the infarcted myocardium [
97]. In accordance with the cardiosplenic axis and the immunomodulatory properties of MSC [
52,
53], we recently demonstrated that intravenous MSC application in CVB3-induced myocarditis modulates monocytes trafficking to the heart. They reduced blood and cardiac pro-inflammatory monocytes and retained those in the spleen, whereas MSC increased anti-inflammatory monocytes in the spleen, blood, and heart [
54].
Evidence for the existence of a cardiosplenic axis is further supported by observations in human post-mortem tissue specimens of the heart, spleen, and bone marrow demonstrating a unique spatio-temporal pattern of monocyte accumulation in the human myocardium following acute MI that coincides with a marked depletion of monocytes from the spleen, suggesting that the human spleen contains an important reservoir function for monocytes [
98]. Patients with acute MI exhibit an increased inflammatory status/metabolic activity of the spleen, bone marrow, and carotid artery. This has been demonstrated via
18F-fluorodeoxyglucose (
18F-FDG) positron emission tomography, which evaluates the metabolic activity based on the finding that activated inflammatory cells express high levels of glucose transporters and accumulate
18F-FDG [
93]. Emami et al. [
36] further demonstrated that after ACS, the gene expression of circulating pro-inflammatory monocytes (i.e., CD36, S100A9, IL-1ß, and TLR4) was more closely associated with the metabolic activity of the spleen than it was for the bone marrow. They further observed that the metabolic activity of the spleen independently predicted the risk of subsequent cardiovascular disease events. In patients with acute MI, high monocyte blood levels, which are a strong predictor of mortality, correlate inversely with the ejection fraction [
99]. Collectively, the abovementioned findings provide evidence of a cardiosplenic axis in humans similar to that shown in pre-clinical studies [
36].
High-Grade Systemic Inflammation
Evidence from chronic immune-mediated diseases like rheumatoid arthritis associated with persistent high-grade systemic inflammation demonstrates the impact of systemic inflammation on HF. Patients with rheumatoid arthritis have a 1.5–2.0 times higher prevalence of ischemic heart disease and congestive HF compared to the general population [
100]. Furthermore, atherosclerosis progresses most rapidly during the first 6 years after rheumatoid arthritis diagnosis [
101], indicating how enduring systemic inflammation plays a major role in accelerating heart disease development in these patients. Systemic inflammation can induce autonomic nervous system dysfunction. Inflammatory cytokines increase the sympathetic outflow by targeting the autonomic centers in the brain, which in turn inhibits cytokine production and immune-inflammatory activation by stimulating the ß2 adrenoreceptors in circulating lympho-monocytes [
102]. This self-controlling loop, so-called inflammatory reflex [
103], and in this context, sympathetic activation, consequently damps excessive immune-inflammatory activation, but also affects the heart, potentially favoring the onset of arrhythmias [
104] and HF. In extreme cases of inflammation as systemic inflammatory response syndrome (SIRS), or sepsis, the hemodynamic changes due to hypotension may directly underlie the induction of the neuroendocrine system, independent of the inflammatory response.
Low-Grade Systemic Inflammation
Obesity is characterized by a low-grade systemic chronic inflammatory state [
26]. The multisystem effects of obesity are linked to an imbalance in homeostatic and pro-inflammatory immune responses. A major player in systemic low-grade chronic inflammation in obesity is the increased numbers of adipose tissue pro-inflammatory MΦs and deregulated production and function of adipose tissue hormones and adipokines including adiponectin [
105], which strongly contributes to the initiation and exacerbation of type 2 diabetes [
106]. Over time, ectopic lipid accumulation in the muscle, liver, and blood vessels activates tissue leukocytes, contributes to organ-specific disease, and exacerbates systemic insulin resistance. Cellular- and cytokine-mediated inflammation in the pancreatic islets accelerates the progression toward diabetes [
26]. The obesity-associated alterations in adipokine expression (adiponectin ↓, TNF-α ↑) also contribute to HFpEF [
106]. Indeed, adiponectin deficiency known to exacerbate the development of obesity-related hypertension [
107], adverse cardiac remodeling [
108] in ischemia-reperfusion injury [
109], and MI [
110], increased the propensity to develop diastolic HF and diastolic dysfunction in a murine model of HFpEF/diastolic HF [
111]. In contrast, adiponectin overexpression in aldosterone-infused mice ameliorated left ventricular (LV) hypertrophy, diastolic dysfunction, lung congestion, and myocardial oxidative stress without affecting the blood pressure and LV ejection fraction [
112].
Diabetes and obesity both induce hematopoiesis and myelopoiesis. Hyperglycemia promotes myelopoiesis via interaction of neutrophil-derived S100A8/A9 with RAGE on hematopoietic stem cells [
113]. S100A8/A9-induced TLR4/MyD88 and NLRP3 inflammasome-dependent IL-1ß production in adipose tissue MΦs interacts with the IL-1 receptor on bone marrow myeloid progenitors to stimulate the production of monocytes and neutrophils. These studies uncover a positive feedback loop between adipose tissue MΦs and bone marrow myeloid progenitors and suggest that inhibition of TLR4 ligands or the NLRP3-IL-1ß signaling axis could reduce adipose tissue inflammation and insulin resistance in obesity [
114].
In line with the HFpEF paradigm postulated by Paulus and Tschöpe [
6•], it has been demonstrated that the systemic, low-grade inflammation of metabolic risk contributes to diastolic LV dysfunction and HFpEF through coronary microvascular endothelial activation, which alters paracrine signaling to cardiomyocytes and predisposes them to hypertrophy and high diastolic stiffness [
115,
116]. In detail, the authors showed upregulated E-selectin and intercellular adhesion molecule-1 expression levels, increased NADPH oxidase (NOX) 2 expression in MΦs and endothelial cells but not in cardiomyocytes, and uncoupling of endothelial nitric oxide synthase, which was associated with reduced myocardial nitrite/nitrate concentration, cGMP content, and protein kinase G activity in the myocardium of HFpEF patients and ZSF1-HFpEF rats. The ZSF1-HFpEF rats are characterized by titin hypophosphorylation and cardiomyocyte stiffness and do not exhibit cardiac fibrosis [
116], the other main contributor to cardiac diastolic dysfunction [
5,
117,
118]. Murdoch and coworkers [
119] demonstrated how Ang II-induced endothelial NOX 2 activation had profound pro-fibrotic effects in the heart in vivo that lead to a diastolic dysfunction phenotype. Endothelial NOX 2 had pro-inflammatory effects and enhanced endothelial-to-mesenchymal transition, which might be an important mechanism underlying cardiac fibrosis and diastolic dysfunction during increased renin-angiotensin activation. A positive correlation between cardiac collagen, the amount of inflammatory cells, and diastolic dysfunction evident in HFpEF patients further suggests a direct influence of inflammation on fibrosis contributing to diastolic dysfunction [
5].
Many studies have indicated that an overactive RAAS, excess oxidative stress, and excess inflammation in the brain cause sympathoexcitation in HF [
120]. Partial silencing of brain TLR4 via intracerebroventricular injection of TLR4 siRNA causes sympathoinhibition with the prevention of left ventricular remodeling in MI-induced HF through the reduction of brain pro-inflammatory cytokines [
120]. Kishi [
121] recently demonstrated that systemic infusion of Ang II directly affects brain AT1R with sympathoexcitation and LV diastolic dysfunction. Furthermore, they demonstrated that targeted deletion of AT1R in astrocytes strikingly improved survival with prevention of LV remodeling and sympathoinhibition in MI-induced HF. Based on these results, the authors propose a novel concept that the brain works as a central processing unit integrating neural and hormonal input, and that the disruption of dynamic circulatory homeostasis mediated by the brain causes HF.