C57BL/6 N (N) albino wild-type (WT) mice (Charles River, Germany), RKIP^−/− C57BL/6 N albino mice (Brown University, Rhode Island, USA), C57BL/6 J (J) wild-type mice (The Jackson Laboratory, USA) and RKIP^−/−C57BL/6 J mice (provided by Kristina Lorenz, Leibniz-Institut für Analytische Wissenschaften, Dortmund, Germany) were used [38, 41]. For QTL analysis C57BL/6 J, DBA/2 J, B6D2 F1 hybrids and BxD lines were obtained from The Jackson Laboratory or from Oak Ridge Laboratory, USA, and bred in the facility of the Neurobsik consortium at the VU University Amsterdam [13].

The studies were approved by the Animal Ethics Committee of the Universities of Saarland and Würzburg and conformed to the Guide for the Care and Use of Laboratory Animals published by the Association for Assessment and Accreditation of Laboratory Animal Care (NRC 2011). Transverse aortic constriction (TAC) and treatment with carbon tetrachloride (CCl₄) were performed to induce replacement and interstitial fibrosis as described [13, 17]. For surgery and left ventricle (LV)-pressure measurements animals were intraperitoneally anaesthetized with 100 mg/kg body weight ketaminehydrochloride (Ketanest^®, Pfizer, Germany) and 10 mg/kg body weight xylazinehydrochloride (Rompun^® 2%, Bayer, Germany). Anaesthetic monitoring was performed by testing of rear foot reflexes before and during procedures, observation of respiratory pattern, mucous membrane color, and responsiveness to manipulations throughout the procedures. Experimental and control mice from our work group were sacrificed by i.p. injection of ketamine (1 g/kg body weight) and xylazine (100 mg/kg) and hearts were rapidly excised.

All fibrosis trait data were uploaded into the GeneNetwork database. Pearson’s correlation was used to correlate fibrosis data among themselves and to BxD phenotypes. The identification and mapping of phenotypic QTLs was performed by linking trait data to genotypes at known genetic marker loci. For the identification of single QTLs, interval mapping analyses were performed across all chromosomes [13].

The samples of failing human hearts were obtained from the left ventricle (LV) of patients undergoing heart transplantation due to end-stage heart failure (NYHA IV) (n = 14) and of patients with aortic stenosis undergoing aortic valve replacement (NYHA II-III) (n = 7). Samples from eight non-failing donor hearts that could not be transplanted for technical reasons served as controls. All samples were obtained with written and informed consent from patients or families of organ donors. The analysis was approved by the Ethics Committee of the Universität des Saarlandes and conformed to the Declaration of Helsinki.

Circulating fibroblasts (fibrocytes) in the peripheral blood and bone marrow were detected by flow cytometry [17]. Cells from blood and bone marrow (BM) were double labelled with the following antibodies: biotin-conjugated anti-collagen I (Rockland, Germany) with streptavidin–fluorescein isothiocyanate (FITC) (Vector Laboratories, USA) and CD45-allophycocyanin (APC) (Pharmingen, Germany). The viable lymphocyte population was examined by flow cytometry (BD FACS Calibur™ instrument and BD CellQuest™ software) [17].

Adult mouse cardiac fibroblasts were isolated from the mouse hearts and migration assay was performed by modified Boyden chambers filled with the medium containing stromal derived factor 1 (SDF-1) [17]. The number of migrated fibroblast immunostained for intracellular fibronectin was evaluated by fluorescence microscopy [17].

To detect fibroblasts, cardiomyocytes, cycling cells, CXCR4, platelet-derived growth factor receptor alpha (PDGFRα) and level of oxidative stress immunostaining on 3 µm paraffin sections of the LV were performed using heat-mediated antigen retrieval with citraconic anhydride solution followed by overnight incubation at 4 °C with the first antibody and incubation with the appropriate secondary antibody at 37 °C for 1 h [17, 18].

Apoptosis was quantitated using 3 µm thin sections of formalin-fixed heart sections and the ApopTag Peroxidase in situ Oligo Ligation Kit (Millipore, Germany). To evaluate apoptosis rate in cardiomyocytes, non-cardiomyocytes, and fibroblasts specific immunostainings were performed after the apoptosis assay [17, 18].

RNA and proteins were isolated from the left ventricle. Gene expression was assessed by the semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR), the real-time quantitative RT-PCR and western blot [17, 18].

Lipid peroxidation was performed using the ALDetect Lipid Peroxidation Assay Kit (Enzo Life Science, Germany) to detect the concentrations of malondialdehyde (MDA) according to the manufacturer´s instructions [29].

The genetically mosaic CCl₄-treated BXD inbred mouse lines displayed significant variation of quantitative fibrosis phenotypes, consistent with polygenic inheritance of fibrosis (Fig. 1a) [13]. Cardiac fibrosis was quantified morphometrically as fractional area of collagen content using picrosirius red staining. QTL analyses identified Raf Kinase Inhibitor Protein (RKIP) as potential genetic marker of fibrosis both in the heart and in the liver (Fig. 1b, c) [13]. Collagen content in the LV myocardium of the parental strains, the F1 hybrids and recombinant inbred BxD mouse lines correlated positively with the myocardial mRNA-expression of RKIP (r = 0.4, p = 0.04) (Fig. 1c, d). The increased cardiac fibrosis in CCl₄-treated DBA/2 J was associated with the enhanced myocardial mRNA-expression of RKIP (mRNA RKIP CCl₄ DBA/2 J 120 ± 7% vs.CCl₄ C57BL/6 J 100 ± 2%, p = 0.02) and increased ROS production in cardiac fibroblasts (32 ± 7% 8-dOHG⁺ fibroblasts in CCl₄ DBA/2 J vs. 5 ± 2%, in CCl₄ C57BL/6 J, p = 0.01) (Fig. 1a, c).

To induce interstitial cardiac fibrosis, mice were treated with carbon tetrachloride (CCl₄), an established inductor of systemic fibrosis. CCl₄ treatment for 6 weeks did not elicit cardiac hypertrophy and did not increase peripheral blood pressure in WT and RKIP^−/− animals (Table 1). Cardiac fibrosis was quantified morphometrically as fractional area of collagen content using picrosirius red staining. Systemic RKIP-knockout diminished CCl₄-induced interstitial fibrosis (Fig. 2a, c), prevented an increase of fibroblast density (Fig. 2b, c) and reduced myocardial collagen Iα2 (COL1A2) mRNA expression (Suppl. Figure 1a).

Replacement cardiac fibrosis was elicited by chronic cardiac afterload induced by TAC. 5 weeks post TAC left ventricular systolic pressure (LVSP) and left ventricular end-diastolic pressure (LVEDP) were increased as expected (Table 2). TAC induced myocardial and cardiomyocyte hypertrophy both of which were reduced in RKIP^−/− mice (Table 2). Systemic RKIP-deficiency decreased TAC-induced replacement fibrosis (Fig. 2a, c), diminished fibroblast density (Fig. 2b, c), reduced myocardial collagen Iα2 (COL1A2) and connective tissue growth factor (CTGF) mRNA expression (Suppl. Figure 1a, b) in RKIP^−/− N TAC. Isolated adult cardiac fibroblasts (ACF) of RKIP^−/− N mice demonstrated reduced migration capacity in a modified Boyden chamber, untreated and after treatment with CCl₄ or angiotensin II (Ang II) (Fig. 2d, e). ACF of RKIP^−/− N mice did not increase the production of intracellular fibronectin in cell culture after treatment with Ang II (Fig. 2f).

The percentage of cycling Ki67⁺ fibroblasts in the LV myocardium was diminished in RKIP^−/− N mice 1 week and 5 weeks post TAC, compared with corresponding WT N TAC (Fig. 2g–j).

Both CCl₄ treatment and TAC increased apoptosis in cardiomyocytes, non-cardiomyocyte cells and fibroblasts (Suppl. Figure 2, 3) accompanied by increased number of cycling cardiomyocytes and cardiac fibroblasts identified by expression of Ki67 (Suppl. Figure 4). Systemic RKIP-deficiency ameliorated cardiac cell turnover in both models (Suppl. Figure 2–4).

The SDF-1/CXCR4 axis plays pivotal role in the regulation of fibroblast activity driving proliferation, migration and collagen production as well as in the mobilisation of circulating fibroblasts from the bone marrow and their recruitment into the myocardium [16, 17]. Co-immunostaining for intracellular fibronectin and CXCR4 was used for the histological evaluation of the influence of TAC and RKIP-knockout on myocardial fibroblasts. The percentage of CXCR4⁺ fibroblasts was reduced in RKIP^−/− N mice 1 week and 5 weeks post TAC (Fig. 3a–c). Systemic RKIP-deficiency significantly reduced the number of CD45⁺ collagen I⁺ fibrocytes in the peripheral blood during acute and chronic pressure overload (Fig. 3d). TAC for 5 weeks significantly increased the number of fibrocytes in the BM which was reduced in RKIP-knockout (Fig. 3e).

Systemic RKIP-deficiency significantly reduced oxidative stress elicited by chronic pressure overload in RKIP^−/− N TAC mice evaluated by co-immunostaining for 8-dOHG and intracellular fibronectin (Fig. 3f, g).

At the first day and 1 week after TAC protein expression and phosphorylation of RKIP were increased compared with SHAM, while this difference vanished after 5 weeks (Fig. 4a–c). Both RKIP-knockout and TAC increased phosphorylation of extracellular signal-regulated kinase 1/2 (ERK1/2) which remained unchanged during the first week after TAC surgery (Suppl. Figure 5a, b) but was significantly reduced in C57BL/6 N 5 weeks post TAC (Suppl. Figure 5c).

The nuclear localization of the main transcriptional activator of antioxidant proteins, nuclear factor erythroid 2-related factor 2 (Nrf2), was significantly reduced during the first week of TAC (Fig. 4d). Systemic RKIP-deficiency drastically increased the nuclear accumulation of Nrf2 accompanied by an enhanced protein expression of its negative regulator, Kelch-like ECH-associated protein 1 (Keap1), during chronic pressure overload (Fig. 4d, f).

Furthermore, cultured adult cardiac fibroblasts from RKIP^−/− N mice treated with Ang II demonstrated increased nuclear accumulation of Nrf2 (Fig. 4e). Co-immunostainings for Nrf2 and myocytic α-sarcomeric actin and intracellular fibronectin revealed an increased nuclear localization of Nrf2 in pressure-overloaded cardiomyocytes and fibroblasts of RKIP^−/− N mice (Suppl. Figure 6a, b).

The non-receptor protein tyrosine kinase Fyn phosphorylates Nrf2 inducing its export from the nucleus and proteasomal degradation [36]. Systemic RKIP-deficiency significantly reduced the nuclear accumulation of the active phosphorylated form Thr12 phospho-Fyn in both groups 5 weeks after surgery and in cultured adult cardiac fibroblasts treated with Ang II (Fig. 5a, b).

In response to Ang II, Janus kinase 2 (Jak 2) activation is required for Fyn activation [37]. Both Western blot analyses of Jak2 protein and co-immunoprecipitation with anti-Jak2 antibody demonstrated an increased myocardial expression of the active form of Tyr1007/1008 phospho-Jak2 in RKIP^−/− N SHAM mice, which was significantly reduced after 5 weeks of TAC (Fig. 5c).

Phospho-RKIP inhibits G-protein coupled receptor kinase 2 (GRK2), which subsequently induces desensitization and internalization of angiotensin II type 1 (AT1)-receptors [3, 34]. Systemic RKIP-deficiency significantly enhanced myocardial expression of GRK2 after 5 weeks of pressure overload (Fig. 5d). To further confirm that the observed effects are caused by RKIP-inhibition in cardiac fibroblasts, ACF of C57BL/6N mice were treated with siRNA for RKIP (siR) and AngII. Untreated ACF and scrambled control (sc) RNA (con RNA) were used for control experiments (n = 5–6 per group) (Fig. 5e–h). The treatment with Ang II significantly increased protein expression of RKIP, intracellular fibronectin, and Keap1 (Fig. 5e, f, h). siRKIP markedly reduced both basal and AngII-stimulated protein expression of RKIP accompanied by the increased nuclear accumulation of Nrf2 in both groups and a reduction of intracellular fibronectin in AngII-treated fibroblasts (Fig. 5e–g).

To further assess the role of oxidative stress for the observed effects of RKIP, C57BL/6 N and C57BL/6 J mice were compared. We have previously demonstrated that the nicotinamide nucleotide transhydrogenase (Nnt) plays a very important role for the ROS-production and enhanced cardiac fibrosis in the pressure-overloaded myocardium [29]. Due to the mutation of the Nnt gene, the inbred mouse strain C57BL/6 J is protected from oxidative stress and fibrosis in response to pressure overload [29]. Indeed, systemic RKIP deficiency caused differential effects on cardiac fibrogenesis in pressure-overloaded myocardium of the two mouse strains: cardiac fibrosis and COL1A2 mRNA expression were reduced in RKIP^−/− N after TAC but increased in RKIP^−/− J TAC compared with the corresponding wild types (Fig. 6a–e).

Oxidative stress in cardiomyocytes and cardiac fibroblasts was evaluated by co-immunostaining for 8-dOHG and α–sarcomeric actin and intracellular fibronectin respectively. TAC significantly increased ROS production both in cardiomyocytes and fibroblasts of WT N but not WT J mice (Fig. 6f–j). RKIP^−/−N mice demonstrated decreased percentage of 8-dOHG⁺ cardiomyocytes in both groups (Fig. 6f). In contrast, the percentage of 8-dOHG⁺ cardiomyocytes in RKIP^−/−J mice was lower in the control group but was significantly enhanced by TAC (Fig. 6g). The percentage of 8-dOHG⁺ fibroblasts was diminished by RKIP-knockout and not affected by TAC in both mouse strains (Fig. 6h, i). Representative western blot demonstrates different expression of intracellular fibronectin in the experimental groups (Fig. 6k).The observations were further confirmed by the co-immunostaining for 8-dOHG and the fibroblast marker platelet-derived growth factor receptor alpha (PDGFRα) (Fig. 6l, m, p). Pressure overload during 5 weeks significantly increased the myocardial concentration of malondialdehyde (MDA) in WT N TAC but not in WT J TAC (Fig. 6n, o). Systemic RKIP-deficiency reduced MDA concentration in RKIP^−/− N TAC and did not change it in RKIP^−/− J TAC (Fig. 6n, o).

Nuclear accumulation of Nrf2 and protein expression of Keap1 were differentially regulated by RKIP-knockout in the two mouse strains. RKIP^−/− N TAC mice revealed a two-fold increased nuclear accumulation of Nrf2 and protein expression of Keap1, compared with WT N TAC (Fig. 7a, c). In contrast, RKIP^−/− J Control mice demonstrated increased basal nuclear accumulation of Nrf2 and protein expression of Keap1 that were not further enhanced by TAC (Fig. 7b, d). The increased nuclear accumulation of Nrf2 was accompanied by enhanced protein expression of catalase and mitochondrial superoxide dismutase in RKIP^−/−N TAC and RKIP^−/− J Control mice (Fig. 7e–h).

The expression of the gene coding for RKIP, PEBP-I, assessed by RT-PCR was differentially influenced by TAC in the two mouse strains: PEBP-I was increased in J and was not changed in N (Fig. 7i). Furthermore, both basal and TAC-induced expression of PEBP-I was significantly lower in J than in N mice (Fig. 7i). The differential effects of RKIP-knockout in pressure-overloaded N and J mice are summarized in Fig. 8 [10, 38].

The RKIP protein expression correlated negatively with the nuclear accumulation of Nrf2 in myocardial samples from human non-failing and failing left ventricles (r = − 0.38, p = 0.04). There were no differences in protein expression of phospho-RKIP, RKIP and nuclear protein content of Nrf2 (Fig. 7j–l).

The analysis of the inbred BxD mouse lines revealed the association of RKIP with increased cardiac fibrosis and ROS production. Therefore we used RKIP^−/− N mice to investigate the influence of RKIP on myocardial fibrosis and oxidative stress. Systemic RKIP-deficiency in the Nnt-positive N-strain significantly diminished both CCl₄- and TAC-induced interstitial and replacement cardiac fibrosis, fibroblast proliferation as well as the expression of fibrotic mediators such as CTGF and collagen. The activation of SDF-1/CXCR4 axis drastically increases fibroblast activity and migration during cardiac fibrogenesis [16, 17]. Enhanced expression of SDF-1 in inflammatory cardiomyopathy is associated with increased fibrosis and mortality [46]. Systemic RKIP-deficiency significantly reduces the number and the percentage of CXCR4⁺ fibroblasts ameliorating LV remodeling during pressure overload. Adult RKIP^−/− N cardiac fibroblasts show reduced migration capacity basal and after treatment with CCl₄ and angiotensin II, decreased angiotensin II-stimulated production of intracellular fibronectin. In agreement with recent reports in the literature, [3, 11] TAC and CCl₄-treated RKIP^−/− N mice reveal decreased apoptosis and decreased proliferation of cardiomyocytes and non-cardiomyocytes. Thus, systemic RKIP-deficiency in C57BL/6 N mice ameliorates interstitial and replacement cardiac fibrosis and cardiac cell turnover.

In addition to the resident cardiac fibroblasts, circulating bone marrow-derived fibroblasts (fibrocytes) contribute to cardiac fibrosis [17, 42, 43]. Although recent experimental findings question the participation of circulating fibrocytes in pressure-overload induced fibrosis, clinical studies demonstrate an association between their elevated number in the peripheral blood and adverse clinical outcomes [1, 19, 27]. Thus, the reduction of the fibrocyte numbers in the peripheral blood and bone marrow of pressure-overloaded C57BL/6 N is consistent with an ameliorative effect of systemic RKIP-knockout.

Acute pressure overload increases ERK1/2 phosphorylation in WT which remains unchanged during the first week but is reduced below the basal level 5 weeks post TAC [24]. Basal phosphorylation of ERK1/2 is enhanced by RKIP-knockout and not changed by TAC. Recent reports demonstrate that the Raf-MEK-ERK signaling pathway plays a differential role during the time course of cardiac remodeling: increased phosphorylation and activation of ERK1/2 in early stage promotes cardiomyocyte hypertrophy and cardiac remodelling but in later stages decreased phosphorylation and downregulation of ERK1/2 increases myocyte apoptosis leading to LV dilatation and heart failure [14, 24‐26]. Thus, systemic RKIP-deficiency ameliorates cardiac remodeling activating Raf-MEK-ERK signaling pathway during 5 weeks of pressure-overload.

TAC induced myocardial oxidative stress is a pivotal and well-characterized trigger of cardiac hypertrophy and fibrosis [18, 23, 45]. Nrf2 is the main transcriptional activator of antioxidant proteins and enzymes protecting against cardiac hypertrophy and fibrosis during pressure overload [7, 23, 45]. Immunostaining for 8-hydroxyguanosine revealed drastically increased oxidative stress both in pressure-overloaded cardiomyocytes and fibroblasts of WT N mice after TAC. Despite technical limitations of 8-hydroxyguanosine as oxidative damage marker [32] clinical studies demonstrate an increased level in patients with cardiovascular disease [8]. The finding of increased oxidative stress was confirmed by MDA concentration and decreased myocardial expressions of catalase and mitochondrial superoxide dismutase.

RKIP-knockout in N-mice markedly reduced oxidative stress and increased nuclear accumulation of Nrf2 [2]. This increase was accompanied by an enhanced myocardial protein expression of Keap1. Keap1 functions as substrate adaptor protein for Cullin 3(CUL3)/Ring box protein 1 (RBX1)-dependent E3 ubiquitin ligase complex which promotes rapid degradation of Nrf2 in the absence of oxidative stress [33]. Nrf2 regulates its protein level through an autoregulatory loop activating the expression of Keap1, CUL3 and RBX1 [33]. Keap1 decreases apoptosis and inflammation by reduction of NF-κB activation through Keap1E3 ligase-mediated degradation of IKKβ [6, 20]. The increased nuclear accumulation of Nrf2 is caused by down-regulated Jak2/Fyn myocardial signaling inhibiting export of Nrf2 from the nucleus (pathway depicted in the schematic (Fig. 8). RKIP^−/−N mice demonstrate increased myocardial expression of GRK2 in pressure-overload and abrogated angiotensin II response in isolated adult cardiac fibroblasts [34]. We speculate that the ameliorative effect of Nrf2 in RKIP^−/− N TAC exceeds the deleterious one of GRK2 [5, 9, 39]. Moreover, ERK1 phosphorylates GRK2 reducing its pro-fibrotic activity [35]. Since RKIP is implicated in the regulation of different cell types we performed cell culture experiments to further confirm the influence of RKIP-knockout on the nuclear protein content of Nrf2 in adult cardiac fibroblasts. The fibroblasts from RKIP^−/− N mice demonstrate enhanced nuclear accumulation of Nrf2 after treatment with angiotensin II. Furthermore, small interfering RNA-mediated silencing of RKIP expression in adult cardiac fibroblasts of C57BL/6 N mice significantly reduced angiotensin II-induced expression of intracellular fibronectin and increased the nuclear accumulation of Nrf2. Our findings are in agreement with recent data from cell culture experiments demonstrating ameliorative effects of RKIP silencing: resistance to oxidative stress caused by increased nuclear accumulation of Nrf2 [2] and retardation of cellular senescence elicited by enhanced activity of ERK [22]. The retardation of aging can ameliorate cardiac fibrosis decreasing the number of cardiac myeloid and mesenchymal fibroblasts in age-dependent cardiac fibrosis [43]. These findings suggest that reduction of Ang II signaling leading to nuclear accumulation of Nrf2 represents an important mechanism of the observed ameliorative effects of systemic RKIP-deficiency in pressure-overloaded myocardium of C57BL/6 N mice.

The Nnt-deficient C57BL/6 J strain is protected from myocardial oxidative stress in response to pressure overload [29]. Thus, we performed the proof-of concept experiments on RKIP^−/− J mice to elucidate the role of the myocardial redox status in the observed effects of RKIP-knockout. Overexpression RKIP is associated with increased cardiac contractility that is mediated by the β1-adrenoceptor and with anti-apoptotic and anti-fibrotic effects mediated by the β2-adrenoceptor (AR) [38]. The stimulation of β2-AR activates the cytoprotective antifibrotic and antiapoptotic phosphatidylinositol 3´kinase (PI3 K)-Akt signaling pathway [38, 40]. Systemic RKIP-deficiency in C57BL/6 J mice exaggerates pressure overload–induced cardiac failure [38]. Systemic RKIP deficiency has opposite effects on cardiac fibrosis in the pressure-overloaded myocardium of the two mouse strains: cardiac fibrosis and pro-fibrotic signaling were reduced in the Nnt-positive N and increased in the Nnt-negative J mice. Both basal and TAC-stimulated myocardial ROS production is markedly reduced in J mice. The expression of PEBP-I, the gene coding for RKIP, is higher in the N compared to the J mouse strain and TAC does not increase nuclear accumulation of Nrf2 in RKIP^−/− J mice. Thus, the fibrotic signaling of RKIP depends on the myocardial redox milieu. Under conditions of decreased myocardial ROS, detrimental effects of systemic RKIP-deficiency override its antioxidant protection.

To elucidate a potential clinical relevance of our findings, we evaluated the protein expression of RKIP, its phosphorylation level and nuclear accumulation of Nrf2 in the myocardial samples from non-failing and failing human left ventricles. The data suggest that RKIP protein expression correlates negatively with the nuclear protein content of Nrf2 in the human hearts. These findings are consistent with the concept that myocardial ROS production that is typical for maladaptive cardiac remodeling switches the effects of RKIP expression to pro-fibrotic signaling [4, 28]. Clearly, additional future studies are needed to confirm the role of RKIP during cardiac remodeling in humans.

Using immunohistochemistry to demonstrate the nuclear accumulation of Nrf2 has its limitations because of the potential non-specific cross-reactivity of some antibodies against Nrf2 [30, 31]. The association between the nuclear accumulation of Nrf2 and the reduced ROS production was reported previously, e.g. using Nrf2-LacZ mice [15] and GFP-Nrf2 fusion proteins [44]. Furthermore, antibody-based and antibody-independent detection of nuclear localization of Nrf2 revealed shuttling of Nrf2 between the nucleus and the cytoplasm [44].

In conclusion, the data show the important role of RKIP for the regulation of cardiac remodeling and fibrosis. Systemic RKIP deficiency ameliorates cardiac remodeling under conditions of increased myocardial production of reactive-oxygen species by activation of the Nrf2–Keap1 system. These findings may provide interesting perspectives to create novel strategies for the detection and prevention of myocardial fibrosis.