Leukocyte iNOS is required for inflammation and pathological remodeling in ischemic heart failure

Abstract

In the failing heart, iNOS is expressed by both macrophages and cardiomyocytes. We hypothesized that inflammatory cell-localized iNOS exacerbates left ventricular (LV) remodeling. Wild-type (WT) C57BL/6 mice underwent total body irradiation and reconstitution with bone marrow from iNOS^−/− mice (iNOS^−/−c) or WT mice (WTc). Chi-meric mice underwent coronary ligation to induce large infarction and ischemic heart failure (HF), or sham surgery. After 28 days, as compared with WTc sham mice, WTc HF mice exhibited significant (p < 0.05) mortality, LV dysfunction, hypertrophy, fibrosis, oxidative/nitrative stress, inflammatory activation, and iNOS upregulation. These mice also exhibited a ∼twofold increase in circulating Ly6C^hi pro-inflammatory monocytes, and ∼sevenfold higher cardiac M1 macrophages, which were primarily CCR2^– cells. In contrast, as compared with WTc HF mice, iNOS^−/− c HF mice exhibited significantly improved survival, LV function, hypertrophy, fibrosis, oxidative/nitrative stress, and inflammatory activation, without differences in overall cardiac iNOS expression. Moreover, iNOS^−/−c HF mice exhibited lower circulating Ly6C^hi monocytes, and augmented cardiac M2 macrophages, but with greater infiltrating monocyte-derived CCR2⁺ macrophages vs. WTc HF mice. Lastly, upon cell-to-cell contact with naïve cardiomyocytes, peritoneal macrophages from WT HF mice depressed contraction, and augmented cardiomy-ocyte oxygen free radicals and peroxynitrite. These effects were not observed upon contact with macrophages from iNOS^−/− HF mice. We conclude that leukocyte iNOS is obligatory for local and systemic inflammatory activation and cardiac remodeling in ischemic HF. Activated macrophages in HF may directly induce cardiomyocyte contractile dysfunction and oxidant stress upon cell-to-cell contact; this juxtacrine response requires macrophage-localized iNOS.

Introduction

Inducible nitric oxide (NO) synthase (iNOS) is upregulated in the failing human heart [9]; however, its functional role is controversial. iNOS overexpression has been considered detrimental in heart failure (HF) based on observations of NO-mediated depression of cardiomyocyte contraction and the β-adrenergic response [9, 17, 20, 21], and studies indicating that genetic iNOS deficiency in mice improves left ventricular (LV) remodeling, dysfunction and mortality in HF [8, 10, 13, 41]. However, other studies are at odds with this view [37]. Specifically, iNOS deficiency abolishes ischemic preconditioning [14], and exacerbates reperfusion injury in the heart [49], whereas myocyte-specific iNOS overexpression [47] or iNOS cardiac gene therapy [25] is infarct-sparing. Moreover, mice with constitutive cardiac-specific iNOS overexpression and ∼40-fold increase in iNOS activity do not exhibit a detrimental cardiac pheno-type [18], whereas intramyocardial iNOS gene transfer in pigs with chronic ischemic HF actually improved regional fibrosis and contractility [1], suggesting that iNOS upregulation per se is not deleterious, and may be beneficial, in HF.

One potential explanation for these incongruous findings is that the effects of iNOS may be determined by its cellular source, rather than by magnitude of expression. During ischemia and preconditioning, and iNOS transgenesis and gene transfer, iNOS expression occurs primarily in cardiomyocytes [3, 20, 25, 47]. In the failing heart, however, iNOS is also derived from infiltrating macrophages [11, 45], where (unlike in cardiomyocytes) NO production is accompanied by superoxide generation, favoring the formation of toxic peroxynitrite [30]. Indeed, macrophage expansion is a hallmark of post-infarction remodeling and HF [22, 24, 33, 40]. Moreover, monocytes in human HF exhibit robust iNOS expression [5] that correlates with disease activity [43], and iNOS expression is an important aspect of classical pro-inflammatory M1 macrophage polarity vs. alternative anti-inflammatory M2 cells [32, 42]. Accordingly, we tested the hypothesis that inflammatory cell iNOS imparts deleterious effects that accelerate LV remodeling and HF progression after myocardial infarction (MI). Our results indicate that leukocyte iNOS is indispensable for chronic inflammation and remodeling in ischemic HF.

Methods

All studies were performed in compliance with the NIH Guide for the Care and Use of Laboratory Animals [DHHS publication (NIH) 85-23, revised 1996], and were approved by the Institutional Animal Care and Use Committees at the University of Louisville and University of Alabama at Birmingham. A total of 183 mice were used for these studies.

Generation of chimeric mice

WT C57BL/6 and iNOS^−/− male mice, 8–10 weeks of age and weighing 20–25 g, were obtained from Jackson Laboratory (Bar Harbor; Stock #000664 and #002609, respectively). Chimeric mice were generated adapting previously published methods (Supplemental Figure 1) [4]. CD45.1 C57BL/6 mice received 950-cGy lethal dose total body irradiation (TBI) from a ¹³⁷Cesium source (Nordion) for bone marrow (BM) ablation, followed by reconstitution with syngeneic donor BM 4–6 h later under sterile conditions. Donor CD45.2 C57BL/6 and iNOS^−/− mice were euthanized via CO₂ inhalation and BM from the tibias and femurs were flushed with cold Media 199 (Gibco) containing 50 μL/mL gentamicin (MEM) using a 22-gauge needle. BM was then resuspended with an 18-gauge needle and filtered through sterile nylon mesh. Cells were cen-trifuged for 10 min at 1000 rpm at 4 °C, resuspended in MEM, and counted using a hemocytometer. Recipient animals were reconstituted by infusing the marrow inoculum (15 × 10 whole BM cells, 1 mL total) via the lateral tail vein using a 27-gauge needle. Mice were quarantined for 30 days to ensure adequate engraftment and tissue resident cell turnover, and then characterized for donor chimerism using flow cytometric analysis of peripheral blood lymphocytes with fluorophore-labeled mAbs (BD Biosciences) CD45.1-PE (recipient cells) and CD45.2-FITC (donor cells) (Supplemental Figure 1). All study mice exhibited ≥85% chimerism. With this approach, we generated iNOS^−/− → WT recipient chimeras (iNOS c) as well as control WT → WT recipient chimeras (WTc). In iNOS^−/−c mice, the iNOS deficiency was, therefore, localized primarily to bone marrow-derived leukocytes.

Coronary ligation

After establishing stable chimerism (≥30 days after BM transplant), left coronary artery ligation was performed in WTc and iNOS^−/− c mice to induce a large MI as previously described [15, 16, 22, 46], and the mice were then followed for 28 days. Sham-operated mice were used as controls. For measurement of late scar size at 28 days, after tissue harvest (see below), the LV was sectioned into transverse slices and imaged with a digital camera. The epicardial circumference for well-demarcated scar in each slice was determined by videoplanimetry, normalized to total LV circumference for all slices, and expressed as a percentage. A subgroup of ligated mice (n = 4/group) was euthanized after 48 h to determine acute infarct size. Briefly, hearts were excised, perfused with PBS, sliced into 4–5 short-axis sections, and placed in fresh 1% 2,3,5-triphenyltetrazolium chloride (TTC) solution at room temperature for 3 min and then 37 °C for 15 min. Infarct zones, delineated as TTC-free white areas, were planimetered on digital images using MetaMorph software, and expressed as % total LV area. In separate studies, coronary ligation or sham operation was performed in non-chimeric WT and iNOS^−/− mice, and peritoneal macrophages were isolated after 28 days to examine cardiomyocyte-macrophage interactions.

Echocardiography

Mouse echocardiography was performed under tribromoethanol (0.25 mg/g IP) sedation, using either a Philips Sonos 5500 equipped with a 15 MHz linear array transducer (120 Hz frame rate), or a VisualSonics Vevo770 High-Resolution System and 30 MHz RMV707B scan head, as previously described [15, 16, 22, 34, 46]. Measurements were taken at baseline, 72 h (for ligated mice), and 28 days after surgery.

LV pressure measurement

Closed-chest LV catheterization with a Millar 1.4 Fr high-fidelity pressure catheter (Model SPR-835) was performed under 1% isoflurane general anesthesia 28 days after ligation or sham surgery as previously described [12, 15, 46]. LV systolic function was indexed by dP/dt_max, and dP/ dt_max normalized for instantaneous LVP (IP) as a relatively load-independent index, and LV diastolic function was assessed using dP/dt_min and tau, the time constant of LV relaxation [12, 15].

Tissue harvest

Mice were deeply anesthetized with sodium pentobarbital (50 mg/kg IP), and arrested in diastole with IV KCl. The heart was rapidly excised, rinsed in ice-cold PBS, and the ventricles and atria were dissected and weighed. The LV was sectioned into five transverse slices from apex to base. A mid short-axis LV section was formalin-fixed and paraffin-embedded for histological studies. The remaining LV was separated into infarcted (scar) and non-infarcted tissue, snap-frozen in liquid nitrogen, and stored at −80 °C for biochemical/molecular studies.

(Immuno)histological studies

LV sections (5 μm thickness) were deparaffinized, rehydrated, and stained with Masson trichrome and Alexa Fluor 488-conjugated wheat-germ agglutinin (WGA; Invitrogen) for determination of fibrosis and myocyte cross-sectional area, respectively, as previously described [15, 16, 22, 34, 46]. Immunofluorescent staining for MOMA-2, F4/80, CD206, iNOS, CD68, CCR2, and nitrotyrosine were performed using standard protocols [22, 46]. Tissue sections were blocked and exposed to primary antibodies against MOMA-2 (Abcam), CD206 (AbD Serotech), F4/80 (eBiosciences), iNOS (Santa Cruz), CD68 (Bio-Rad), CCR2 (Abgent), or nitrotyrosine (Santa Cruz), and secondary antibodies conjugated with Alexa 555, Alexa 488, or FITC. Nuclei were counterstained with DAPI. Optical sections were obtained with a Zeiss LSM510 inverted confocal scanning laser microscope with excitation wavelengths appropriate for multi-channel scanning to allow co-localization. Images were recorded within 24 h and analyzed with Zeiss LSM image browser software (version 4.2).

Quantitative real-time PCR

Total RNA extraction from LV tissue, cDNA synthesis, and quantitative real-time PCR were performed as before [15, 16]. mRNA transcript levels for atrial natriuretic factor (ANF), connective tissue growth factor (CTGF), tumor necrosis factor-α (TNF), interleukin (IL)-1β, IL-6, IL-10, CD206, iNOS, and GAPDH were determined using the forward and reverse primer pairs listed in Supplemental Table 1. mRNA expression for each gene was normalized to GAPDH mRNA expression using the ΔΔC_T comparative method [15].

Western immunoblotting

Total LV protein extraction, SDS-PAGE Western blotting, and electro-chemiluminescence (ECL) immunodetection were performed as described [15, 16, 46]. Anti-iNOS primary antibody and horseradish peroxidase (HRP)-conjugated secondary antibody were obtained from Santa Cruz. Primary antibody specificity was determined upon concomitant incubation with a fivefold excess of blocking peptide (Santa Cruz, sc-651P). Protein-nitrotyrosine was also quantitated using slot blots. LV protein (1 μg) was loaded in the wells of a Bio-Dot apparatus (Bio-Rad), microfiltered through nitrocellulose membranes under vacuum, and probed with anti-nitrotyrosine antibody. Intensity of the immunoreactive bands was quantitated by ImageQuant TL software.

Electrophoretic mobility shift assay (EMSA)

Nuclear protein extraction from frozen LV tissue, EMSA to assess nuclear factor(NF)-κB DNA binding activity, autoradiography, and densitometry were performed as previously described [12, 15, 16].

Malondialdehyde (MDA) assay

Malondialdehyde in LV tissue homogenates was determined colorimetrically using the OxiSelect™ MDA adduct ELISA kit (Cell Biolabs, STA-332) following the manufacturer's instructions, with MDA adducts measured at 450 nm. MDA-BSA concentrations ranging from 0–120 pmol/mg were used to construct standard curves.

Nitrate/nitrite (NO_x) assay

Tissue NO_x levels were determined with the Nitrate/Nitrite Colorimetric Assay Kit (Cayman Chemical), with photometric measurement of absorbance at 540 nm [38].

Isolation of mouse cardiomyocytes

Calcium tolerant mouse ventricular myocytes were isolated using modified Langendorff perfusion and collagenase digestion of isolated WT hearts as previously described [16, 28, 46]. Cardiomyocytes were plated at a density of 10 rod-shaped cells/cm² in modified serum-free DMEM medium (37 °C, 5% CO₂) until experimentation.

Isolation of peritoneal macrophages

Mice were injected i.p. with 3 mL of Brewer thioglycollate (TG) medium (Sigma, B2551). Five to seven days later, mice were euthanized with CO₂ inhalation followed by cervical dislocation, and peritoneal lavage was performed with PBS and fluid collected on ice. Macrophages were then isolated from lavage fluid by Ficoll density gradient purification, followed by washing with PBS and resuspension in serum-free DMEM to a final concentration of 2.5 × 10⁶ cells/mL. Macrophage purity was confirmed in a small aliquot (in PBS) by staining with FITC-labeled F4/80 antibody and epifluorescence microscopy (Nikon TE200). Cell suspensions with <90% positive F4/80 staining were discarded. Cells were used for experimentation within 1 h of isolation.

Cardiomyocyte-macrophage interaction studies

We assessed the impact of macrophages on cardiomyocyte contraction and free radical generation. Cardiomyocytes from WT naïve mice and macrophages from WT and iNOS^−/− naïve, sham, and HF mice were used. Sarcomere shortening was determined using an IonOptix StepperSwitch system and video-based digitized sarcomere spacing (SarcLen Acquisition, IonOptix) [28]. After a post-isolation rest period of ∼4h, 10³ cardiomyocytes (10 -cells/mL) were placed in a cell chamber (Warner) on an inverted stage microscope (Nikon TE200) and superfused at 1 mL/min with modified Tyrode solution (in mM: 137 NaCl, 4.9 KCl, 1.2 MgSO₄, 15 d-glucose, 20 HEPES, 1.2 NaH₂PO₄, 1.8 CaCl₂, pH 7.35). Baseline shortening was measured during digital field stimulation (1 Hz) for 5 min. Macrophages (100 μL of 2.5 × 10⁶ cells/mL) were then introduced into the chamber and cells observed over a 15-min period, at which time shortening was measured in cells with and without macrophage attachment over 5 min. For some experiments, macrophages were pre-stimulated with 5 μg/mL lipopolysaccharide (LPS, Sigma) for 24 h, and then washed with PBS several times (to remove the LPS) prior to use.

To measure intracellular reactive oxygen species (ROS), cardiomyocytes were incubated (in the dark for 2 h at 37 °C) with 2 μM 2′,7′-dichlorofluorescin diacetate (DCFH-DA, Molecular Probes), a fluorescent indicator of ROS levels [29]. DCFH-DA-loaded myocytes were washed several times with modified Tyrode solution, loaded into a borosilicate-chambered cover glass system, co-incubated with macrophages, and monitored with epifluorescence microscopy (Nikon Eclipse TE2000-U). Immediately upon macrophage-myocyte interaction, baseline cardiomyocyte fluorescence (530 nm) was recorded and repeated every 2 min for 14 min. Phase-contrast images were also obtained at 15 min to quantify the number of attached macrophages. Intracellular peroxynitrite was measured in an analogous manner using cardiomyocytes loaded with the specific peroxynitrite fluorescent indicator dye HK-Green2 (10 μM for 1 h) [48], kindly provided by Dr. Lu Cai, University of Louisville.

Flow cytometry

Peripheral blood (∼100 μL) was obtained via facial vein puncture and collected into BD Microtainer EDTA tubes (BD Biosciences). Erythrocytes were lysed with 2 mL of RBC lysis buffer (eBiosciences) for 5 min on ice, followed by quenching with 10 mL of cold PBS. Leukocytes were collected by centrifugation (380g for 10 min at 4 °C) and resuspended in ice-cold flow cytometry staining buffer (eBioscience). Cell suspensions were incubated for 30 min on ice with fluorophore-labeled cell surface antibodies CD45-605NC, F4/80-eFluor 450, CD11b-Alexa Fluor 700 (eBioscience), and Ly6C-FITC (BD Biosciences). After staining, cells were centrifuged and resuspended in PBS and analyzed on a BD LSRII flow cytometer. Non-debris gates were established on SSC/FSC plots and surface marker positivity was determined from histograms and dot plots. Isotype control samples were used to determine negative fluorescence thresholds. From the lymphocytemonocyte gate, pro-inflammatory (CD45⁺CD11b⁺F4/ 80^lowLy6C^hi) and patrolling (CD45⁺CD11b⁺F4/80^low-Ly6C^low) monocytes were identified [19, 26, 44]. Final analysis was performed using FlowJo v.7.6 software.

Cytometric bead array (CBA) immunoassay

Concentrations of IL-6, IL-10, IL-12, TNF, monocyte chemoattractant protein (MCP)-1, and interferon (IFN)-γ in mouse serum were determined simultaneously using a CBA Mouse Inflammation Kit (BD Biosciences) as previously described [22].

Statistical analysis

For two-group comparisons, we used the unpaired two-sample t test. To compare more than two groups, we used one-way ANOVA if there was one independent variable, and two-way ANOVA if two independent variables (e.g., chimera type and ligation status), with Bonferroni or Tukey's post-test to adjust for multiple comparisons. A p value <0.05 was considered significant. Animal survival was evaluated by Kaplan-Meier curves, and the log-rank test was used to compare survival between WTc sham and HF, iNOS^−/− c sham and HF, and WTc and iNOS^−/− c HF. Data are shown as mean ± SEM (or SD, where indicated).

Results

Leukocyte iNOS deficiency improves survival and post-infarct LV remodeling

Surgically naïve mice that underwent transplantation with either WT or iNOS^−/− BM exhibited normal cardiac structure and function 28 days later (Supplemental Table 2). The mice subsequently underwent coronary ligation or sham operation. Echocardiography 72 h post-surgery indicated similar extent of akinetic LV myocardium at the mid-papillary level in ligated WTc and iNOS^−/−c mice. Moreover, acute infarct size (by TTC staining, % total LV area) was equivalent in a subgroup of ligated mice assessed at 48 h (42.5 ± 3.7% WTc vs. 43.0 ± 3.5% iNOS^−/− c, p = NS). Kaplan–Meier survival curves (Fig. 1a) revealed significantly increased mortality for WTc HF mice vs. WTc sham (p = 0.018) and iNOS^−/−c HF (p = 0.047) at 28 days post-MI, but no differences in mortality between iNOS^−/−c HF and iNOS^−/−c sham mice. Late scar size post-mortem (at 28 days post-MI) in iNOS^−/−c and WTc HF mice (% total LV circumference) was also similar (Fig. 1b). These data suggested that improved survival in iNOS^−/−c HF mice was not related to differences in initial injury, but rather to beneficial effects on post-MI wound healing and long-term LV remodeling.

Fig. 1.

a. Kaplan–Meier survival curves over 28 days following coronary ligation [to induce heart failure (HF)] or sham operation in wild-type chimeric (WTc) and iNOS^−/− chimeric (iNOS^−/−c) mice. *p = 0.0184 vs. WTc sham; ^#p = 0.0469 vs. WTc HF (log-rank test); n = 8–19/group. b Group data for scar size 4 weeks after coronary ligation in WTc and iNOS^−/−c mice (n = 9/group, unpaired t test). c Representative short-axis LV sections, end-diastolic 2D images, and M-mode echocardiograms from WTc and iNOS^−/−c sham and HF mice 4 weeks after operation. d Example LV pressure and dP/dt tracings from WTc and iNOS^−/−c sham and HF mice 4 weeks after operation

Figure 1c depicts representative short-axis LV sections, enddiastolic 2D parasternal echocardiographic images, and M-mode echocardiograms from WTc and iNOS^−/−c sham and HF mice (28 days post-surgery). While both WTc and iNOS^−/−c HF mice exhibited LV dilatation and dysfunction compared with sham, it was less pronounced in the iNOS^−/−c HF mouse. Group data (Table 1) confirmed that iNOS^−/−c HF mice exhibited less chamber enlargement, better LV ejection fraction (EF), and less wall thinning compared with WTc HF mice. Indeed, iNOS^−/−c HF mice exhibited wall thickness ratios comparable to iNOS^−/−c sham mice. Gravimetry revealed increased normalized LV, lung, and liver weights in WTc HF compared with sham, indicative of LV hypertrophy and pulmonary/systemic congestion. In contrast, iNOS^−/−c HF mice exhibited lower LV and lung weights as compared with WTc HF, and, no change in lung and liver weights as compared to iNOS^−/−c sham, suggesting less pathological hypertrophy and fluid overload. Hemodynamic data (Fig. 1d; Table 1) showed increased filling pressure (LVEDP), depressed contractility (dP/dt_max/IP), and impaired relaxation (tau) in WTc HF mice, with improvement of contractility and LVEDP in iNOS^−/−c HF mice. These data indicate that leukocyte iNOS exacerbates LV chamber remodeling, hypertrophy, LV dysfunction, and fluid overload in ischemic HF.

Table 1. Echocardiographic, gravimetric, and hemodynamic data.

	WTc		iNOS−/−c
	Sham	Heart failure	Sham	Heart failure
HR (bpm)	481 ± 55	499 ± 85	467 ± 41	508 ± 64
LVEDV (μL)	27 ± 8	97 ± 14*	19 ± 4	50 ± 12*,#
LVESV (μL)	8 ± 3	72 ± 15*	6 ± 1	29 ± 7*,#
LVEF (%)	70 ± 4	26 ± 8*	68 ± 5	40 ± 3*,#
RWT (mm)	0.42 ± 013	0.27 ± 0.07*	0.40 ± 0.06	0.37 ± 0.11#
LV weight/TL (mg/mm)	4.3 ± 0.6	5.3 ± 0.9*	4.3 ± 0.8	4.9 ± 1.4*,#
Lung weight /TL (mg/mm)	11.3 ± 0.5	12.4 ± 1.1*	11.2 ± 1.3	10.8 ± 1.7#
Liver weight/TL (mg/mm)	65.5 ± 4.1	69.1 ± 2.7*	65.6 ± 6.0	67.9 ± 3.8
LVPSP (mmHg)	91 ± 17	62 ± 9*	84 ± 19	70 ± 4
LVEDP (mmHg)	5 ± 2	12 ± 5*	3 ± 2	6 ± 2*,#
dP/dtmax (mmHg/s)	8902 ± 1355	3807 ± 694*	6514 ± 3006	4994 ± 853#
dP/dtmax/IP (/s)	190 ± 47	121 ± 20*	166 ± 25	146 ± 12#
dP/dtmin (mmHg/s)	−6884 ± 896	−3180 ± 694*	−6318 ± 3801	−3900 ± 423#
Tau (ms)	7.1 ± 1.2	11.0 ± 1.2*	8.5 ± 2.7	11.6 ± 3.6

All values mean ± SD

HR heart rate, LV left ventricular, EDV and ESV end-diastolic and end-systolic volume, EF ejection fraction, RWT relative wall thickness at end-diastole, TL tibia length, PSP peak systolic pressure, EDP end-diastolic pressure, dP/dt_max maximal rate of increase in LV pressure, dP/dt_max/IP dP/dt_max normalized for instantaneous LV pressure, dP/dt_min maximal rate of decrease in LV pressure, tau time constant of LV relaxation

^*p<0.05 vs. respective sham,

^#p<0.05 vs. WTc HF (two-way ANOVA with Bonferroni post-test). n = 7–12/group

Leukocyte iNOS deficiency reduces fibrosis, myocyte hypertrophy, and oxidative and nitrative stress in the failing heart

Trichrome staining revealed a ∼ fivefold increase in border and remote zone fibrosis in WTc HF hearts, with much less pronounced border zone fibrosis in iNOS^−/−c HF hearts (Fig. 2a, b). Similarly, cardiac gene expression of the pro-fibrotic matrix-associated protein CTGF was increased 2.5fold in WTc HF hearts vs. sham, but this upregulation was moderated in iNOS^−/−c HF hearts. WGA staining revealed equivalent myocyte size in WTc and iNOS^−/−c sham mice and significant myocyte hypertrophy in both HF groups as compared with sham (Fig. 2c); however, the degree of hypertrophy was less pronounced in iNOS^−/−c HF hearts as compared with WTc HF hearts. Similarly, WTc HF hearts exhibited significantly greater expression of the hypertrophic gene marker ANF as compared to iNOS^−/−c HF hearts.

Fig. 2.

Representative Masson's trichrome stains (a) and quantitation of interstitial fibrosis and connective tissue growth factor (CTGF) gene expression by real-time PCR (b) in myocardium of WTc and iNOS^−/−c sham and HF hearts. c Representative wheat-germ agglutinin staining (green) of myocyte cell membranes and DAPI staining of nuclei (blue) in WTc and iNOS^−/−c sham and HF hearts, and quantitation of myocyte cross-sectional area and atrial natriuretic factor (ANF) gene expression by real-time PCR. *p < 0.05 vs. respective sham, ^#p < 0.05 vs. WTc HF (two-way ANOVA with Bonferroni post-test); n = 5–7/group

As shown in Fig. 3a, western blotting revealed a ∼threefold increase in cardiac iNOS protein expression in WTc HF over WTc sham. Cardiac iNOS expression was also significantly increased in iNOS^−/−c HF compared with iNOS^−/−c sham, and was not statistically different from WTc HF expression. As seen in Fig. 3b, c, myocardial NO_x levels decreased significantly, and myocardial protein-MDA adducts increased significantly, in WTc HF, consistent with increased oxidant stress and reduced NO bioavailability. In contrast, both NO_x and MDA levels in iNOS^−/−c HF hearts were comparable to sham. Proteinnitrotyrosine levels, by either immunofluorescence (Fig. 3d) or slot blots (Fig. 3e), were augmented ∼2.5-fold in WTc HF hearts as compared with WTc sham, but significantly suppressed in both iNOS^−/−c sham and HF hearts. Hence, in the failing heart, iNOS-expressing leukocytes augment oxidant and nitrative stress, and reduce NO bioavailability. These effects are independent of iNOS expression by resident cardiac cells, which was similar in WTc and iNOS^−/−c HF mice.

Fig. 3.

a Western immunoblotting for iNOS protein in WTc and iNOS^−/−c sham and HF hearts. +C, recombinant iNOS-positive control. Myocardial nitrate + nitrite (NO_x) levels (b) and proteinmalondialdehyde (protein-MDA) adducts (c) corresponding to the same experimental groups. d Representative protein-nitrotyrosine immunostaining (green) of WTc and iNOS^−/−c sham and HF hearts and corresponding quantitative group data for fluorescence intensity. e Protein-nitrotyrosine slot blots from the same experimental groups. *p < 0.05 vs. respective sham, ^#p < 0.05 vs. WTc HF, ^@p < 0.05 vs. WTc sham (two-way ANOVA with Bonferroni post-test); n = 3–6/group

Leukocyte iNOS deficiency reduces cardiac and systemic inflammation in HF

We next examined the effects of leukocyte iNOS deficiency on inflammatory activation in HF. As seen in Fig. 4a, WTc HF hearts exhibited significant upregulation of pro-inflammatory TNF and IL-6, but no change in anti-inflammatory IL-10 as compared with WTc sham hearts. In contrast, iNOS^−/−c HF hearts exhibited markedly diminished TNF and IL-6 expression, together with augmented IL-10 expression, as compared with WTc HF. Activation of NF-κB, a transcriptional regulator of inflammatory cytokines, was markedly increased (∼3.5-fold) in WTc HF hearts as compared with sham (Fig. 4b), consistent with our prior studies [15, 16]. In contrast, NF-κB binding to DNA was markedly diminished in iNOS^−/−c HF hearts, which in turn was comparable to iNOS^−/−c sham hearts. Serum levels of TNF, IL-12, and IFN-γ were also significantly increased in WTc HF, with moderation of levels in iNOS^−/−c HF as compared with WTc HF (Fig. 4c). Similar trends were also seen for serum IL-6. Moreover, iNOS^−/−c HF mice exhibited a ∼threefold increase in serum IL-10 as compared with iNOS^−/−c sham, and lower serum levels of MCP-1 as compared with WTc HF. Hence, both local and systemic inflammation in ischemic HF requires leukocyte iNOS, which helps regulate pro- vs. anti-inflammatory balance.

Fig. 4.

a Myocardial gene expression of tumor necrosis factor-α (TNF), interleukin (IL)-6, and IL-10 in WTc and iNOS^−/−c hearts as indicated (n = 5–6/group). b EMSA for nuclear factor(NF)-κB DNA binding in WTc and iNOS^−/−c sham and HF hearts and corresponding quantitation of binding intensity (n = 3/group). c Serum levels of various cytokines, measured using cytometric bead array (n = 4–6/group). IFN-γ interferon-γ, MCP-1 monocyte chemoattrac-tant protein-1. *p < 0.05 vs. respective sham, ^#p < 0.05 vs. WTc HF, ^@p < 0.05 vs. WTc sham (two-way ANOVA with Bonferroni post-test)

Leukocyte iNOS deficiency promotes an anti-inflammatory monocyte/macrophage profile in HF

As shown in the representative flow cytometry plots and group data in Fig. 5, circulating pro-inflammatory Ly6C^hi monocytes and patrolling Ly6C^low monocytes (and total CD11b⁺F4/80^low cells) were all significantly increased in WTc HF vs. sham mice. In contrast, blood Ly6C^hi monocytes were diminished in iNOS^−/−c HF mice as compared to WTc HF, and unchanged from iNOS^−/−c sham. Circulating Ly6C^low monocytes exhibited similar trends between iNOS^−/−c HF and WTc HF, but did not reach statistical significance.

Fig. 5.

a Flow cytometry gating strategy used to identify circulating CD45⁺CD11b⁺F4/80^lowLy6C^hi and CD45⁺CD11b⁺F4/80^lowLy6C^low monocytes using the lymphocyte–monocyte gate from initial side vs. forward scatter plots (SSC vs. FSC, respectively). b Quantitative group data for total blood CD11b⁺F4/80⁺ cells, and Ly6C^hi and Ly6^low subsets, in WTc and iNOS^−/−c sham and HF mice. *p < 0.05, **p < 0.005, ***p < 0.0005 (two-way ANOVA with Bonferroni post-test); n = 4–6/group

In the myocardium, there were significantly increased numbers of iNOS⁺ macrophages in WTc HF hearts (border and remote zone) as compared with WTc sham hearts (Supplemental Figure 2). Immunostaining for F4/80 and CD206 revealed robust expansion of F4/80⁺CD206⁻ M1-type macrophages in WTc failing hearts vs. WTc sham hearts, without differences in the number of F4/ 80⁺CD206⁺ M2-type macrophages (Fig. 6). In contrast, iNOS^−/−c HF hearts exhibited no augmentation of M1 macrophages as compared with sham, but instead demonstrated a marked (∼2.5-fold) increase in M2 macrophages vs. all other experimental groups. Moreover, iNOS^−/−c HF hearts exhibited a robust increase in CD206 mRNA expression as compared with either iNOS^−/−c sham or WTc HF. We further performed immunostaining for CD68, a pan macrophage marker, and CCR2, which labels monocyte-derived infiltrating macrophages (CCR2⁺), as opposed to locally sourced cells (CCR2⁻) [40]. As shown in Fig. 7, the expanded macrophage population in WTc HF hearts was predominantly (∼88%) CCR2⁻ locally sourced cells. However, in iNOS^−/−c HF hearts, the macrophage distribution was reversed, and was mainly (∼67%) CCR2⁺ infiltrating cells. The CD68⁺CCR2⁺ cells in iNOS^−/−c HF hearts also appeared larger morphologically as compared to those in WTc HF hearts. Taken together, these data suggest that leukocyte iNOS deficiency promotes the accumulation of monocyte-derived macrophages in the remodeling heart post-MI, but that these cells nonetheless exhibit an anti-inflammatory and reparative M2 phenotype that imparts beneficial effects on pathological LV remodeling.

Fig. 6.

a Confocal microscopic images of immunofluorescent staining for F4/80 (red), CD206 (green), and nuclei (DAPI, blue) from representative WTc and iNOS^−/−c sham and HF hearts (border zone), along with magnified images from the failing heart sections and quantitative data for F4/80⁺CD206⁻ (M1) and F4/80⁺CD206⁺ M2 macrophages (n = 3–7/group). Red arrows M1 cells, and yellow arrows M2 cells. *p < 0.05 vs. WTc sham; **p <0.005 vs. WTc and iNOS^−/−c sham, ^#p < 0.05 vs. WTc HF (two-way ANOVA with Bonferroni post-test). b Myocardial CD206 mRNA expression by real-time PCR for the same experimental groups. *p < 0.05 vs. iNOS^−/−c sham and WTc HF. n = 4–6/group

Fig. 7.

a High-power confocal microscopic images of immunofluo-rescent staining for CD68 (red), CCR2 (green), and nuclei (DAPI, blue) from representative WTc and iNOS^−/−c sham and HF hearts (near scar area), and b quantitative group data for CD68⁺CCR2⁻ (locally sourced) and CD68⁺CCR2⁺ infiltrating macrophages (n = 3–5/group). *p < 0.05, **p < 0.005 (two-way ANOVA with Tukey's post-test)

HF-derived macrophages induce cardiomyocyte dysfunction and oxidant stress in a contact-and iNOS-dependent manner

Adult cardiomyocytes (from WT naïve mice) and peritoneal macrophages (from WT and/or iNOS^−/− naïve, sham, and HF mice) were isolated and cell interaction studies were performed. In pilot studies, macrophages derived from naïve WT mice were activated with LPS, introduced into a cell chamber containing cardiomyocytes stimulated at 1 Hz, and shortening measurements were performed after 15–20 min. As seen in Supplemental Figure 3A, sarcomere shortening in myocytes in close proximity, but without attachment to macrophages, was unchanged from baseline. Conversely, myocytes exhibiting physical attachment to macrophages exhibited robust ∼50% reduction in contraction and prolongation of contraction kinetics. In subsequent studies, peritoneal macrophages derived from mice with HF, and not otherwise stimulated, were used in the same assay. As shown in Fig. 8a, macrophages derived from WT HF mice induced contractile depression in cardiomyocytes, but only upon cell-to-cell attachment, analogous to LPS-stimulated macrophages, indicating a potent juxtacrine effect and contact-dependent signaling. In contrast, macrophages derived from iNOS^−/− HF mice did not induce contact-dependent cardiomyocyte dysfunction, even after pre-stimulation with LPS. These differential effects between WT and iNOS^−/−HF-derived macrophages were not due to differences in the number of cells attached to cardiomyocytes (Fig. 8b), and hence were related to differences in macrophage phenotype.

Fig. 8.

a Group data for sarcomere shortening from cardiomyocyte– macrophage cell interaction studies in the presence or absence of attachment to macrophages derived from WT or iNOS^−/− HF mice. In some experiments, iNOS^−/− HF-derived macrophages were also pre-stimulated with lipopolysaccharide (LPS). *p < 0.05 vs. baseline and WT unattached, ^#p < 0.05 vs. iNOS^−/− attached, without and with LPS; n = 4–6/group. b Group data for number of WT and iNOS^−/− HF-derived macrophages attached to cardiomyocytes in these studies. NS not significant; n = 5/group. c Top representative fluorescent images of DCFH-loaded cardiomyocytes (green) exposed for 15 min to macrophages derived from either WT or iNOS^−/− HF mice, with or without cell attachment as indicated. Corresponding phase-contrast images are also shown. Bottom quantitative group data for cardiomyocyte fluorescent intensity after 15 min of exposure to macrophages derived from WT and iNOS^−/− (sham and HF) mice with or without attachment. *p < 0.05; n = 4–6/group. d Top representative fluorescent images of cardiomyocytes loaded with peroxynitrite indicator HK-Green2 exposed for 15 min to either positive control SIN-1 (peroxynitrite generator) or attached macro-phages derived from WT and iNOS^−/− HF mice. Bottom quantitative group data for cardiomyocyte fluorescence intensity. *p < 0.05; n = 3–4/group. (a–c two-way ANOVA, d one-way ANOVA; Bon-ferroni post-test)

Given that HF is a pro-oxidant milieu, we evaluated whether macrophage contact with cardiomyocytes generates ROS that depress contractility [29]. Cardiomyocytes were preloaded with DCFH, and intracellular fluorescence was monitored for 15 min after macrophage–cardiomy-ocyte attachment, paralleling the time course for the contraction studies. As seen in Supplemental Figure 3C, cardiomyocyte attachment to LPS-stimulated WT naïve macrophages increased intracellular DCF fluorescence after 15 min, indicating augmented ROS production. Similarly, sustained contact of macrophages derived from WT HF mice robustly increased (∼30-fold) cardiomyocyte DCF fluorescence; this did not occur either in the absence of HF macrophage attachment or upon attachment of macrophages derived from WT sham mice (Fig. 8c). Hence, this juxtacrine oxidant stress response was dependent on both cell contact and macrophage activation in the HF milieu. Interestingly, contact-dependent augmentation of DCF fluorescence did not occur upon attachment of macrophages derived from iNOS^−/− HF mice. Parallel studies with cardiomyocytes loaded with the peroxynitrite indicator HK-Green2 further demonstrated that cardiomyocyte attachment to iNOS^−/− HF-derived macrophages induced less peroxynitrite generation as compared with WT HF-derived macrophages (Fig. 7d), corroborating the differences in nitrotyrosine levels in WT and iNOS^−/− HF hearts (Fig. 3).

Discussion

We demonstrate several key and novel findings in this study. First, inflammatory cell iNOS is indispensable for the genesis of pathological inflammation in the post-MI failing heart, and helps regulate adverse cardiac remodeling. Alleviation of remodeling in iNOS^−/−c HF mice occurred despite maintained iNOS expression in resident cardiac cells, underscoring the pathophysiological importance of the cellular source of iNOS. Our results also establish that cytokine expression in the failing heart originates primarily from infiltrating inflammatory cells rather than myocytes, and that leukocytes expressing iNOS are key arbiters of tissue fibrosis, oxidative stress, and peroxynitrite formation. Second, leukocyte iNOS deficiency improved survival after MI, suggesting better wound healing and less cardiac rupture [15, 31]. Moreover, leukocyte iNOS deficiency modified the macrophage phenotype in HF toward an anti-inflammatory and reparative profile, with tissue persistence of monocyte-derived M2 macrophages in the heart, but suppression of chronic circulating Ly6C^hi monocytosis. These effects would serve to resolve inflammation, promote repair, and improve remodeling [33, 39, 40]. Third, macrophages activated in HF exert significant juxtacrine effects on cardiomyocytes that promote cellular oxidant stress and contractile dysfunction. These effects were prevented by macrophage iNOS deficiency and an accompanying shift toward an M2 phenotype, suggesting that macrophage polarity drives macrophage–myocyte juxtacrine responses in HF. Taken together, we conclude that the genesis of inflammation, tissue oxidative stress, and LV remodeling and dysfunction in ischemic HF requires leukocyte iNOS.

iNOS and LV remodeling in HF

iNOS has traditionally been considered to promote LV dysfunction and remodeling in chronic HF [9, 10, 13, 17, 41]. However, this concept is at odds with observations that iNOS limits infarct size following ischemia–reperfusion injury [25, 47, 49] does not induce significant cardiac dysfunction [18, 47], and actually improves regional fibrosis and contraction in chronic HF [1]. Also, the level of iNOS upregulation reported in the failing heart has been variable [7, 38], suggesting that iNOS expression in HF is time and etiology dependent. Lastly, two studies reported no significant differences in post-infarction LV remodeling between wild-type and iNOS^−/− mice [23, 27], sharply contrasting with prior work [10, 13, 41] and raising uncertainty as to the pathophysi-ological importance of iNOS in HF.

iNOS expression in the post-ischemic failing heart localizes to both infiltrating macrophages and cardiomy-ocytes, particularly myocytes near the infarct border zone [11, 13, 45]. However, iNOS activity in failing human hearts correlates better with the abundance of infiltrating macrophages than with cardiomyocyte levels of expression [11], hinting at a primary role for inflammatory cell iNOS in HF. Our findings definitively establish this to be the case, and show that leukocyte-specific iNOS, rather than global cardiac iNOS expression, drives pathophysiological responses in HF. Interestingly, the effects of iNOS were also source dependent in a sepsis model, whereas neu-trophil-derived iNOS exerted damaging effects on endotoxemic myocytes, myocyte-restricted iNOS expression was protective and preserved β-adrenergic responsiveness [35]. Our current results indicate similar dichotomous effects of leukocyte vs. cardiomyocyte iNOS in ischemic HF; this novel observation may help to reconcile some of the conflicting reports in the literature.

In our chimeric model, iNOS deficiency was primarily limited to bone marrow-derived leukocytes (as opposed to resident cells), including circulating monocytes that infiltrate the heart after MI and during subsequent tissue repair and long-term remodeling. The impact of selective leukocyte iNOS deficiency was surprisingly broad, improving survival and alleviating mechanical dysfunction, inflammation, fibrosis, pathological hypertrophy, oxidant stress, and nitrative stress. This scope of effects underscores the wide-ranging importance of bone marrow-derived immune cells, and by extension their gene expression profile and polarity, in the genesis and progression of cardiac remodeling in ischemic HF. The mortality benefit was observed both within the first week post-MI, a time point when cardiac rupture is the primary cause of attrition [15, 31], and beyond, when progressive remodeling and failure also contribute. Early survival benefit suggests improved early post-MI wound healing, potentially through the suppression of excessive inflammation. Bone marrow-derived leukocyte iNOS deficiency also imparted salutary effects on long-term post-MI remodeling and HF. While HF is uniformly considered a state of inflammation with inappropriate cytokine elaboration, the source of cytokines and inflammatory activation has classically been ascribed to injured myocytes and hypoperfused and/or edematous peripheral tissues, with secondary activation of immune cells [2]. In contrast, our results suggest that the immune cells are themselves the primary inducers of the inflammatory response in HF, as selective alteration of the leukocyte phenotype induced profound changes in both local and systemic cytokine production, and cardiac NF-κB activity.

Leukocyte iNOS, macrophage polarity, and cardiac dysfunction in HF

Mononuclear phagocytes play important pathogenetic roles during both early post-infarct tissue repair [24, 33, 36] and long-term chronic remodeling [22, 36, 40]. Following acute MI, there is robust recruitment of monocytes to the heart. These are initially pro-inflammatory and proteolytic acutely (1–4 days), followed by reparative and anti-inflammatory monocytes that promote wound healing in the sub-acute period (4–8 days) [33]. Beyond the sub-acute post-MI period, in chronic ischemic HF, pro-inflammatory macrophage expansion (both locally sourced and infiltrating cells) promotes long-term pathological remodeling [22, 40]. As iNOS represents an integral component of the M1 macrophage pro-inflammatory gene profile [32, 42], loss of leukocyte iNOS would be expected to favor alternative M2 activation of recruited monocytes, thereby hastening tissue reparative responses both early post-MI and during chronic remodeling. Indeed, we uncovered a robust M2 macrophage response with leukocyte iNOS deficiency, despite persistence of infiltrating CCR2⁺ bone marrow-derived macrophages in the failing heart, that coincided with significant attenuation of Th1/pro-inflammatory cytokines and augmentation of the Th2/anti-inflammatory cytokine IL-10. Hence, immune cell iNOS modifies mac-rophage polarity in HF, and is required for the maintenance of classical macrophage activation and inflammation. Of clinical relevance, increased circulating levels of activated, iNOS-expressing monocytes characterize chronic human HF [5, 6]. Our findings raise the consideration that selectively targeting mononuclear cell iNOS may be a fruitful approach to immunomodulation.

M1 macrophages are pro-oxidant and promote tissue digestion and injury. While these local effects are operative in the failing heart—and are alleviated upon loss of leukocyte iNOS—we also demonstrate that direct macro-phage–cardiomyocyte interactions produce cardiomyocyte dysfunction. Specifically, macrophages directly induce negative inotropy and oxidative/nitrative stress in cardiomyocytes, responses that required macrophage activation in the HF milieu (or exogenously with LPS), physical contact between macrophages and cardiomyocytes, and the presence of macrophage iNOS and, by extension, M1 polarity. In our in vitro system, these effects were not related to the secretion of paracrine factors by activated macrophages. However, this does not suggest that paracrine effects do not occur in vivo, as there may be significant differences in both the number and types of accumulated inflammatory cells, the effective concentration and total portfolio of secreted factors over time, and receptor expression for secreted ligands in diseased cardiomyocytes. Nonetheless, our studies do establish a discernible impact of macrophage contact on cardiomy-ocyte contraction that should be considered, independent of extant paracrine effects. Robust M1 macrophage infiltration of the remodeling heart, both early and late after MI [22, 24, 33, 40], favors amplification of juxtacrine responses in the failing heart. Hence, direct cell–cell effects may contribute importantly to LV dysfunction and the oxidative and nitrative stress encountered in HF. Our results further suggest that modulation of mono-cyte/macrophage polarity towards a reparative phenotype would result in beneficial local effects in the failing heart.

In summary, bone marrow-derived leukocyte iNOS plays an obligatory role in the pathogenesis of inflammation in ischemic HF and regulates cardiac remodeling, independent of iNOS expressed elsewhere in the failing heart. The effects of leukocyte iNOS in HF are broad and multi-faceted, influencing both tissue responses—including cardiac hypertrophy, fibrosis, oxidative/nitrative stress, and inflammation—and local and systemic mono-cyte/macrophage profiles. Moreover, macrophages in HF directly induce cardiomyocyte contractile dysfunction and cellular oxidant stress in an iNOS- and contact-dependent manner. Collectively, these data suggest that targeting leukocyte, and especially macrophage, iNOS may represent a viable approach to therapeutic immunomodulation in HF.