CARDIOMYOPATHY AND WORSENED ISCHEMIC HEART FAILURE IN SM22-α CRE-MEDIATED NEUROPILIN-1 NULL MICE: DYSREGULATION OF PGC1α AND MITOCHONDRIAL HOMEOSTASIS

Abstract

Objective

Neuropilin-1 (NRP-1) is a multi-domain membrane receptor involved in angiogenesis and development of neuronal circuits, however, the role of NRP-1 in cardiovascular pathophysiology remains elusive.

Approach and Results

In this study, we first observed that deletion of NRP-1 induced peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α) in cardiomyocytes (CMs) and vascular smooth muscle cells (VSMCs), which was accompanied by dysregulated cardiac mitochondrial accumulation and induction of cardiac hypertrophy- and stress-related markers. To investigate the role of NRP-1 in vivo, we generated mice lacking Nrp-1 in CMs and VSMCs (SM22-α-Nrp-1 KO), which exhibited decreased survival rates, developed cardiomyopathy and aggravated ischemia-induced heart failure. Mechanistically, we found that NRP-1 specifically controls PGC1α and PPARγ in CMs through crosstalk with Notch1 and Smad2 signaling pathways respectively. Moreover, SM22-α-Nrp-1 KO mice exhibited impaired physical activities and altered metabolite levels in serum, liver, and adipose tissues, as demonstrated by global metabolic profiling analysis.

Conclusions

Our findings provide new insights into the cardio-protective role of NRP-1 and its influence on global metabolism.

Introduction

Neuropilin-1 (NRP-1) is a multi-domain receptor with functional roles in the cardiovascular system and involvement in neuronal development and the pathogenesis of cancer^1–3. Currently, it is known that NRP-1 binds several structurally diverse ligands and mediates different cellular functions, including endothelial chemotaxis, cell adhesion, axon guidance, and participation in multiple signal transduction pathways^{4, 5}. Global Nrp-1 null mice die midway through gestation (E12.5– E13.5) and demonstrate abnormalities in the cardiovascular and nervous systems^{6, 7}. Additionally, mice lacking the genes encoding various NRP-1 ligands, including vascular endothelial growth factor-A (VEGF-A), VEGF-B, placenta growth factor-2, Semaphorin-3A, Semaphorin-3C, platelet-derived growth factor (PDGF), and transforming growth factor β (TGFβ), showed diverse cardiovascular defects^{6, 8–10}.

In the cardiovascular system, NRP-1 is highly expressed in endothelial cells, VSMCs, and CMs^11–13. Conditional deletion of NRP-1 in endothelial cells through the use of Tie2-Cre transgenic mice resulted in mid-to-late embryonic lethality and replicated several of the cardiovascular defects seen in the global Nrp-1 knockout. These defects include a general peripheral vascular branching deficiency and a failure to septate the outflow tract of the heart¹⁴. To date, the precise function of NRP-1 in CMs and VSMCs and its role in mediating cardiovascular pathophysiology have not been characterized. In this study, our results showed that deletion of NRP-1 elevated PGC1α in CMs and VSMCs, induced expression of cardiac hypertrophy, stress-related markers, and abnormal cardiac mitochondrial accumulation, suggesting NRP-1 is involved in the pathogenesis of cardiac diseases. We generated a cell-type specific knockout of NRP-1 in CMs and VSMCs by crossing Nrp-1^flox/flox mice² with the SM22-α-Cre transgenic mice¹⁵. The SM22-α promoter is highly efficient in mediating recombination in CMs and VSMCs in contrast to the smooth muscle myosin heavy chain promoter¹⁶. For these studies, we used two genotypes: SM22-α-Cre negative Nrp-1^flox/flox designated as wild-type (WT) and SM22-α-Cre positive Nrp-1^flox/flox designated as the SM22-α-Nrp-1 KO. Knockout of NRP-1 in CMs and VSMCs decreased survival rates, led to cardiomyopathy and caused the SM22-α-Nrp-1 KO mice to be prone to heart failure after myocardial infarction. Upon exploring the molecular mechanism of this deleterious effect, we found that regulation of PGC1α and PPARγ occurs through crosstalk of NRP-1 with Notch1 and Smad2 pathways. Moreover, knockout of NRP-1 in CMs and VSMCs caused impaired physical activities and aberrant global metabolism. Our findings suggest that expression of NRP-1 in CMs and VSMCs provides a protective effect under basal conditions and during ischemic stress and contributes to global metabolism.

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.

Results

NRP-1 regulates PGC1α in CMs and VSMCs

To investigate the role of NRP-1 in CMs and VSMCs, we first introduced adenovirus Cre to CMs and VSMCs isolated from NRP-1^f/f mice to generate NRP-1 null CMs and VSMCs in vitro. As shown in Figs.1A–C, we observed that deletion of NRP-1 significantly elevated expression of PGC1α, the key regulator of mitochondrial biogenesis¹⁷, in CMs and VSMCs. Accompanied with increased expression of PGC1α, its downstream effectors, including medium-chain acyl-CoA dehydrogenase (MCAD), citrate synthase, ATP synthase C, cytochrome C and mitochondrial transcription factor B 2M (TFB2M), were also significantly increased in NRP-1 KO CMs (Figs.1D&E). Meanwhile, cardiac hypertrophy and stress markers, such as brain natriuretic peptide (BNP), atrial natriuretic factor (ANF), β-myosin heavy chain (βMHC), and c-myc, were induced in NRP-1 null CMs (Fig. 1F). These results suggest that NRP-1 is involved in regulation of PGC1α in CMs and VSMCs and a basal level of NRP-1 is essential to maintain homeostasis of CMs.

Figure 1. Knockout of NRP-1 increased PGC1α in CMs and VSMCs.

(A–B). Neonatal CMs were isolated from NRP-1^f/f mice and then infected with adenovirus Cre or adenovirus GFP as control. Expressions of NRP-1 and PGC1α were validated with qPCR (A) and Western blotting (B). (C). VSMCs were isolated from NRP-1^f/f mice and then infected with adenovirus Cre or adenovirus GFP as control. Expressions of NRP-1 and PGC1α were validated with qPCR. (D–F). qPCR was performed in neonatal CMs isolated from NRP-1^f/f mice and infected with adenovirus Cre or adenovirus GFP as control. (G). Frozen heart sections of WT and SM22-α-Nrp-1 KO mice were stained with antibodies against NRP-1 (red) and α-MHC (green). Nuclei were counterstained with DAPI. Results are representative of three independent experiments. (H). qPCR was performed in neonatal CMs of WT mice and SM22-α-Nrp-1 KO mice. (I). Total cellular DNA was extracted from neonatal CMs of control mice and SM22-α-Nrp-1 KO mice. Mitochondrial cytochrome oxidase 1 (CO1) and lipoprotein lipase (LPL) were analyzed with qPCR and expressed as a ratio of mtDNA and nuclear DNA. (J). Oxygen consumption of control and Nrp-1 KO MEFs were measured using O2k oxygraph. ROUNTINE was measured with normal serum-free culture medium. LEAK respiration was measured after oligomycin-induced inhibition of ATP synthesis. ETS, electron transfer system, was measured with the addition of FCCP. N=3 for each group. (K). Heart tissues of 1 year old WT mice and SM22-α-Nrp-1 KO mice were subjected to transmission electron microscopy. Mitochondrial area was measured and analyzed. Results are representative from 10 sections of one heart each group. (L). qPCR was performed with heart tissues of WT mice and SM22-α-Nrp-1 KO mice. N=3 for each group. (M). Kaplan-Meier survival curves of WT mice and SM22-α-Nrp-1 KO mice. In A, C–F, the ad GFP group was normalized to 1, and the ad Cre group was expressed as relative fold of control group. *, p<0.05, **, p<0.01, ***, p<0.001, compared with control group.

To further investigate the role of NRP-1 in CMs and VSMCs in vivo, we generated conditional knockout mice by crossing Nrp-1^flox/flox mice with SM22-α-Cre mice. Unlike the endothelial cell specific Nrp-1 knockout mice, which die soon after birth, the SM22-α-Nrp-1 KO mice are viable. The efficiency of Cre-mediated recombination was confirmed with robust and specific X-gal staining in the hearts and aortas of SM22-α-ROSA26 mice (Supplemental Fig. IA). By crossing with SM22-α-Cre mice, NRP-1 was specifically deleted in CMs (α-MHC positive cells) but unaffected in endothelial cells (CD31 positive cells) in the heart of SM22-α-Nrp-1 KO mice (Figs.1G& Supplemental IB). NRP-1 expression was observed in VSMCs of WT mice but was completely undetectable in VSMCs from SM22-α-Nrp-1 KO mice (Supplemental Fig. IC).

Elevated levels of PPARγ together with its downstream target carnitine palmitoyltransferase 1B (CPT1b), were also observed in NRP-1 deleted CMs (Figs.1H, Supplemental IIA and IID). Immunofluorescent and immunohistochemical staining further confirmed increased PGC1α and PPARγ in adult heart tissues from SM22-α-Nrp-1 KO mice (Supplemental Figs. IIB&C). Accompanied with increased expression of PGC1α, elevated mtDNA copy number was present in neonatal CMs isolated from SM22-α-Nrp-1 KO mice (Fig. 1I). Correspondingly, NRP-1 knockout fibroblasts exhibited significantly increased mitochondrial oxygen consumption under different respiratory conditions (Fig. 1J). Meanwhile, metabolites analysis showed increased trends of TCA metabolites in cardiac tissues from SM22-α-Nrp-1 KO mice (Supplemental Fig. III). Electron microscopy further confirmed the presences of increased deposition of mitochondria and disorganized sarcomeres in hearts of 1 year old SM22-α-Nrp-1 KO mice (Figs. 1K& Supplemental IIE), which also showed significantly increased expression of BNP together with increased trends of ANF and c-Myc (Fig. 1L). Furthermore, a prospective study for mortality was conducted to investigate the overall effect of deletion of NRP-1 in CMs and SMCs. As shown in Fig. 1M, whereas no significant mortality was observed in the WT group until 600 days old, the survival rate of SM22-α-Nrp-1 KO mice progressively declined. By 663 days of age, one third of the SM22-α-Nrp-1 KO mice died whereas more than 85% of the WT mice remained alive.

Taken together, our results demonstrate that knockout of NRP-1 increased PGC1α and PPARγ expression in CMs and SMCs and led to decreased survival rate accompanied with abnormal mitochondrial accumulation and induction of hypertrophy and stress related markers in CMs.

Cardiomyopathy in SM22-α-Nrp-1 KO mice

Accompanied with the induction of cardiac hypertrophic markers, the SM22-α-Nrp-1 KO mice exhibited progressive hypertrophy of right ventricles (Figs. 2A–C and Supplemental IVA). Specifically, at 3 weeks old and 12 weeks old, SM22-α-Nrp-1 KO mice showed a hypertrophic right ventricle, while at 8 months old and 1 year old, SM22-α-Nrp-1 KO mice displayed a dilated right ventricle (Figs. 2A–C). The mean cross-sectional area of CMs was increased in the right ventricle of SM22-α-Nrp-1 KO mice at the age of 3 weeks old (Fig. 2D). More apoptotic cells were also noted in SM22-α-Nrp-1 KO mice at the age of 3 weeks old (Fig. 2E). However, no histological signs of fibrosis were observed in SM22-α-Nrp-1 KO mice as shown by Masson's Trichrome staining in Fig. 2B.

Figure 2. Cardiomyopathy in SM22-α-Nrp-1 KO mice.

(A–B). Representative morphology and Masson Trichrome’s staining of hearts of WT mice and SM22-α-Nrp-1 KO mice at 3 weeks, 12 weeks, 8 months and 1 year old. N=5 for each group. (C). Thicknesses of right ventricles and left ventricles were analyzed from the images of Masson Trichrome’s staining and expressed as a ratio of RV/LV. N=5 for each group. (D). Averaged CM cross-sectional areas of WT and SM22-α-Nrp-1 KO mice at 3 weeks old were measured with Alexa Fluor® 594 labeled wheat germ agglutinin (WGA). Nuclei were counterstained with DAPI. Two hundred to three hundred CMs were blindly counted per heart. N=7 for each group. (E) Representative micrographs of TUNEL staining of Heart sections of WT mice and SM22-α-Nrp-1 KO mice at 3 weeks old. Nuclei were counterstained with DAPI. Total TUNEL positive cells were counted blindly per heart section. N=7 for each group. *, p<0.05, ***, p<0.001, compared with WT group.

M-mode/2-dimensional (2-D)/3-D echocardiography was performed to examine cardiac function. As shown in Fig. 3, hearts of 1-year old SM22-α-Nrp-1 KO mice demonstrated significant right ventricular dilation, compared to wild-type, in dimension by M-mode (Fig.3A top, Figs.3B–D), area by 2-D (Fig.3A middle, Figs.3E–G), and volume by 3-D (Fig.3A bottom). Although LV contractility (Fig.3I) and mass (Fig.3J) appeared equivalent between the two cohorts, the left ventricle of SM22-α-Nrp-1 KO mice was characterized by a D-shaped cavity in the short-axis view (Fig.3A middle right) likely due to the right ventricular overload and exhibited reduced dimension, volume and area at systolic stage (Figs.3C, F and H), suggesting that even though the ejection fraction was normal, the total volume of blood ejected by left ventricle at the systolic stage was decreased in SM22-α-Nrp-1 KO mice. This result was further supported by the decreased blood pressure observed in SM22-α-Nrp-1 KO mice, which showed normal structures of mesenteric arteries but had notably decreased systolic/diastolic blood pressures at the ages of both 3 weeks and 8 months (Figs.3K & Supplemental IVB).These results confirm that hypertrophic right ventricles and reduced cardiac output were present in SM22-α-Nrp-1 KO mice.

Figure 3. Evaluation of cardiac function of SM22-α-Nrp-1 KO mice.

(A). Representative M-mode/2-dimenational (D)/3-D echocardiography in 1-year-old wild-type and SM22-α-Nrp-1 KO (NRP-1 KO) animals. Dimension by M-mode (B–D), area by 2-D (E–G), volume by 3-D (H), LV contractility (I) and mass (J) were measured. BW, body weight; LV, left ventricle; RV, right ventricle; 2-D/3-D, 2-/3- dimensions. N=5 for each group. (I). Blood pressure was measured by tail cuffed method in male WT mice and SM22-α-Nrp-1 KO mice at 3 weeks old and 8 months old. N=8 for WT mice at 3 weeks old, n=7 for SM22-α-Nrp-1 KO mice at 3 weeks old. N=7 for each group at 8 months old. (J). Masson's trichrome staining was performed in lungs of WT mice and SM22-α-Nrp-1 KO mice at 4 months and 8 months old. Results are representative from experiments with 3 mice of each group. *, p<0.05.

Due to the predominant change of the right ventricle in SM22-α-Nrp-1 KO mice, we also examined the lung structures of both groups. As shown in Fig. 3L, neither significant fibrosis nor abnormal vascular remodeling was observed in either group at 4 months and 8 months of age. However, perivascular inflammatory cell infiltration was observed in the lungs of 8-month old SM22-α-Nrp-1 KO mice. Meanwhile, hemosiderin-laden macrophages were noticed in some 1 year old SM22-α-Nrp-1 KO mice, suggesting the presence of cardiac dysfunction in 1 year old NRP-1 KO mice (Supplemental Fig. IVC).

SM22-α-Nrp-1 KO mice showed deleterious effect in ischemic cardiomyopathy

To further investigate the role of NRP-1 in cardiac function, we produced myocardial infarction (MI) in WT mice and SM22-α-Nrp-1 KO mice by permanent left anterior descending (LAD) coronary artery ligation at 12 weeks of age. As shown in the survival graph (Fig. 4A), it was observed that survival for WT-MI mice dropped from 100% to 60% on post-operative day 3, and remained at 46.66% up to post-operative day 30, the endpoint of the experiment. In contrast, the SM22-α-Nrp-1 KO-MI group showed a dramatic and abrupt drop in survival after LAD ligation, from 100.00% to 38.46% on post-operative day 3, with no survivors past post-operative day 21.

Figure 4. SM22-α-Nrp-1 KO mice exhibited deleterious effect in ischemic cardiomyopathy.

(A). Kaplan-Meier survival curves in WT and SM22-α-Nrp-1 KO mice up to 4 weeks after myocardial infarction (N = 15 for WT mice, n=13 for SM22-α-Nrp-1 KO mice). (B). Bar graphs represent quantitative data of ejection fraction and fractional shortening from WT and SM22-α-Nrp-1 KO mice at 3 days after LAD occlusion (N=6 for WT mice, n=9 for SM22-α-Nrp-1 KO mice). (C). Histological assessment on day 4 post-LAD ligation in ventricle infarct areas. Panels represent standard hematoxylin and eosin (H&E) as well as immunohistochemistry for capillary density (Cav-1 staining) in the central infarct myocardium of WT mice and SM22-α-Nrp-1 KO hearts. Results are representative from experiments with 3 mice of each group. Areas of vascular dilation and interstitial hemorrhage are indicated (>). Magnification of all images is 200×. ***, p<0.001.

Echocardiographic analysis performed preoperatively to assess the baseline function did not show any significant differences in systolic function of left ventricles between WT and SM22-α-Nrp-1 KO groups (Supplemental Fig. VA). However, the ejection fraction (EF) and fractional shortening (FS) were significantly lower in SM22-α-Nrp-1 KO mice from as early as 3 days following MI (Fig. 4B). As expected, no significant differences in EF and FS were noted in the sham groups (Supplemental Fig. VB).

To evaluate the potential cause of the increased mortality rate and heart failure following permanent LAD ligation in the SM22-α-Nrp-1 KO mice, we evaluated three different heart tissues (unaffected myocardium, myocardium at the ischemic interface, and myocardium within the infarcted muscle) at 5 days post-LAD ligation (Figs. 4C& Supplemental VC). The unaffected myocardium appeared phenotypically similar with a minor increase in microvessel staining with the caveolin-1 antibody (Cav-1) (Supplemental Fig. VC). At the interface of the affected ischemic tissues, there was inflammatory cell infiltration and muscle fiber dissociation in both the WT and SM22-α-Nrp-1 KO hearts (Supplemental Fig. VC), suggesting LAD ligation with respect to the inflammatory response appeared to be intact in the SM22-α-Nrp-1 KO mice. The central infarcted region of the ventricular apex showed significant depletion of cardiac muscle and a global infiltration of inflammatory cells in each genotype (Fig. 4C). Hearts harvested from SM22-α-Nrp-1 KO mice had severely dilated vessels and extravascular hemorrhage in the inflammatory zone (Fig. 4C). These findings demonstrate that knockout of NRP-1 in CMs and VSMCs aggravated ischemia-induced heart failure.

Involvement of Smad2 and Notch1 in NRP-1-mediated regulation of PPARγ and PGC1α

To further delineate the upstream signaling of NRP-1 to PPARγ and PGC1α, the effect of several NRP-1 ligands on PPARγ were examined. As shown in Supplemental Fig. VIA, TGF-β, but not other ligands of NRP-1, such as VEGF-A or PDGF-BB, downregulated PPARγ mRNA levels in mouse embryonic fibroblasts (MEFs). Employing NRP-1 KO MEFs, we showed that knockout of NRP-1 partially restored the TGFβ-induced decrease of PPARγ and PGC1α expression (Figs. 5A&B). Our results suggest that NRP-1-dependent signaling is involved in the TGF-β-mediated downregulation of PPARγ and PGC1α mRNA expression.

Figure 5. Involvement of Smad2 and Notch1 in NRP-1-mediated regulation of PPARγ and PGC1α.

(A–B). Control MEFs and Nrp-1 KO MEFs were stimulated with TGF-β (10 ng/mL) for 24 h. PPARγ and PGC1α mRNA levels were analyzed with qPCR and expressed as relative fold of control group. (C) Heart lysates from WT and SM22-α-Nrp-1 KO mice were subjected to Western blotting with the indicated antibodies. (D–E). WT MEFs, Smad2 KO MEFs, and Smad3 KO MEFs were serum-starved overnight and then stimulated with TGFβ for 24 h. Total lysates (D), nuclear protein and cytoplasmic protein (E) were subjected to Western blotting with the indicated antibodies. (F–H). CMs isolated from neonatal NRP-1^f/f mice were infected with adenovirus Cre or adenovirus GFP as control for 48 h, and then transfected with either Smad2 expressing plasmids (F) or Nothc1-FLAG expressing plasmid (H) for 24 h. Total lysates were collected and subjected to Western blotting. *, p<0.05, **, p<0.01, ***, p<0.001.

Our previous studies have shown that knockdown of NRP-1 attenuated TGFβ-induced phosphorylation of Smad2/3¹⁸, so we examined activation of Smad2/3 in NRP-1 deleted CMs and cardiac tissues. As shown in Figs.5C and Supplemental VIB, both phosphorylation of Smad2 and Smad3 were decreased in NRP-1 deleted CMs as well as cardiac tissues of SM22-α-Nrp-1 KO mice. Then we examined expressions of PGC1α and PPARγ in Smad2 KO MEFs and Smad3 KO MEFs. As shown in Supplemental Figs. VIC&D, the level of PGC1α mRNA was not significantly increased in either Smad2 KO or Smad3 KO MEFs. Expression of PPARγ transcript was significantly increased in MEFs derived from Smad2 KO but not Smad3 KO mice relative to WT MEFs. Immunoblotting (Figs.5E, Supplemental VII C–D) showed that both PPARγ and PGC1α protein levels were increased in nuclear fractions of lysates from Smad2 KO MEFs. Upon stimulation with TGF-β, PGC1α protein levels were decreased in WT MEFs and Smad3 KO MEFs but not Smad2 KO MEFs (Figs. 5D, Supplemental VII A–B). Interestingly, although nuclear PPARγ was increased in Smad2 KO MEFs (Figs.5E & Supplemental VII B), total PPARγ protein level was reduced in Smad2 KO MEFs (Fig.5D), suggesting multiple mechanisms, including nuclear translocation of PPARγ, are involved in Smad2-mediated regulation of PPARγ regulation. Furthermore, overexpression of Smad2 partially rescued the expression of PPARγ but not PGC1α in NRP-1 deleted CMs (Fig.5F), suggesting a different mechanism was involved in the regulation of PGC1α. Then we examined the effect of NRP-1 on Notch1 in CMs, which was reported recently to control PGC1α expression in adipocytes¹⁹. As shown in Fig.5G, intracellular form of Notch1 (Notch NICD) was significantly reduced in NRP-1 knockout CMs. Moreover, overexpression of Notch NICD partially recovered the PGC1α protein level in NRP-1 deleted CMs (Fig. 5H). Taken together, our results suggest that multiple downstream signaling pathways of NRP-1 synergize to regulate PGC1α and PPARγ.

Decreased cardiorespiratory fitness in SM22-α-Nrp-1 KO mice

To evaluate the effect of NRP-1 knockout in CMs and SMCs on cardiorespiratory fitness, we performed a high-precision indirect calorimeter in SM22-α-Nrp-1 KO mice. Our results showed that the SM22-α-Nrp-1 KO and WT mice had similar oxygen consumption (VO2) and energy expenditure (EE) (Table 1). In contrast, the WT mice exhibited greater baseline physical activity (horizontal, vertical, ambulatory, and total activity counts) compared to the SM22-α-Nrp-1 KO mice (Table 2). In fact, the SM22-α-Nrp-1 KO mice appeared to have a severe impairment of all types of activity during both day and night light cycles. Considering that total energy expenditure and heat production were similar between the two phenotypes, the findings indicate that SM22-α-Nrp-1 KO mice have either a higher basal metabolic rate or reduced physical activity. In this regard, we proposed that the latter may result from cardiovascular limitations, and NRP-1 deficiency in CMs and VSMCs leads to a significant deconditioning and impairment in ambulation.

Table 1.

Similar oxygen consumption and energy expenditure activity in SM22-α-Nrp-1 KO mice.

Day	VO2	Heat	VCO2	Food intake	Adj Food intake	Body weight
WT	3606±72.51	0.35±0.01	3066±74.19	1.37±0.11	69.27±6.42	19.9±0.40
KO	3516±181.20	0.36±0.01	3059±159.9	1.14±0.10	54.37±6.15	21.26±1.09
Night
WT	4291±196.30	0.42±0.02	3859±216.70	2.89±0.34	145.2±17.10
KO	4094±223	0.43±0.01	3846±238.40	3.19±0.15	151.1±9.35

Animal was placed in chamber for measurement of 24-hour activity. Oxygen consumption and energy expenditure were analyzed as described in materials and methods. N=6 for each group.

Table 2.

Decreased ambulatory activity in SM22-α-Nrp-1 KO mice.

Day	Met Rate	RER	Activity	Ambulation	Rearing
WT	17.54±0.37	0.84±0.01	22502±3955	10616.4±2486	4594±1055
KO	17.18±0.93	0.87±0.018	12742.8±614.2	3381.4±280.2	317.2±129.5**
Night
WT	21.12±1.02	0.89±0.01	43839±4818	23608±3186	12358.8±942.1
KO	20.36±1.14	0.94±0.02*	22801.8±858.30***	7511.6±643.3***	2145.4±572.7***

Animal was placed in chamber for measurement of 24-hour activity. Activities were monitored as described in materials and methods. Two sided T test was performed. N=6 for each group.

^*p<0.05,

^**p<0.01,

^***p<0.001.

Metabolic changes in SM22-α-Nrp-1 KO mice

Previous studies have shown that metabolism in heart contributes to systemic metabolic homeostasis²⁰. To evaluate the effect of NRP-1 deficiency in CMs and VSMCs on whole body metabolism, we examined the metabolic phenotype of SM22-α-Nrp-1 KO mice. Global metabolite profiles in serum, adipose tissue, and livers were analyzed in mice fed a normal diet. Metabolomic profiles obtained from SM22-α-Nrp-1 KO mice and controls are summarized in Table 3. Based on a targeted metabolomics approach, a total of 305 compounds of known identity (named biochemicals) were detected and quantified in serum, livers and adipose tissues (Supplemental Table I). Thirty metabolites differed significantly upon comparison of the serum of WT and SM22-α-Nrp-1 KO (7 increased and 23 decreased), 18 metabolites in adipose tissue (18 increased and 1 decreased), and 24 metabolites in liver (14 increased and 10 decreased). Interestingly, in serum, up-regulated compounds mainly consisted of carbohydrate metabolites (4 in 7) such as 1,5-anhydroglucitol (1,5-AG), glycerate, ribose, and xylitol (Fig. 6A) while down-regulated biochemicals were mainly lipid metabolites, such as stearate (18:0), nonadecanoate (19:0), arachidonate (20:4n6), which are involved in the sub-pathway of long-chain fatty acid metabolism (Fig. 6B). Unique and shared metabolites and 33 metabolic pathways associated with metabolites characterized from the experiments (Supplemental Fig. VIII) are provided. Notably, two dihydroxy fatty acids, which serve as natural PPARγ ligands²¹, including 12,13-hydroxyoctadec-9(Z)-enoate (12,13-DHOME) and 9,10-hydroxyoctadec-12(Z)-enoic acid (9,10-DHOME), were significantly elevated in adipose tissues of SM22-α-Nrp-1 KO mice (Fig. 6C), Meanwhile, although expression of NRP-1 was not changed significantly, epoxide hydrolase 1, microsomal, but not soluble epoxide hydrolase, which mediates the production of 12,13-DHOME and 9,10-DHOME, was unexpected increased in adipose tissues of SM22-α-Nrp-1 KO mice (Supplemental Fig. IX), suggesting that indirect regulation of NRP-1 or its downstream mediators in CMs and VSMCs may contribute to the global metabolism.

Table 3.

Global metabolic changes in the serum, liver and adipose tissue of WT and SM22-α-Nrp-1 KO mice. Serum, adipose tissues, and liver tissues from 6 WT mice and 6 SM22-α-Nrp-1 KO mice were analyzed for global biochemical profile as described in the materials and methods. N=6 in each group.

Statistical Comparisons Welch's Two-Sample t-Test
KO WT	Serum	Liver	Adipose
Total biochemicals p≤0.05	30	24	19
Biochemicals (↑↓)	7|23	14|10	18|1
Total biochemicals 0.05<p<0.10	15	16	38
Biochemicals (↑↓)	5|10	14|2	30|8

Figure 6. Metabolic profile change in SM22-α-Nrp-1 KO mice.

(A–B). Relative levels of 1,5-AG, glycerate, ribose, xylitol (A), and stearate, nonadecanoate, arachidonate (B) in the serum of WT and SM22-α-Nrp-1 KO mice. (C). Relative levels of 12,13-DHOME and 9,10-DHOME in the adipose tissues of WT and SM22-α-Nrp-1 KO mice. N=6 for each group. *, p<0.05, **, p<0.01, ***, p<0.001.

Discussion

Our results established, for the first time, the protective role of NRP-1 signaling in mitochondrial homeostasis of CMs. Previous studies have shown that mitochondrial dysfunction and reduced ATP production are associated and accelerate the progressive nature of heart failure^{22, 23}. Modulation of mitochondrial function reduces the infarct size of the ischemic heart in mice and represents a promising therapeutic strategy for heart failure^24–26. In this context, our findings suggest that loss of NRP-1 in CMs results in significant alterations in mitochondrial homeostasis, which may be due to altered PGC1α and PPARγ expression and eventually lead to cardiomyopathy and increased heart failure after ischemic stress.

Previous studies examining global Nrp-1 null mice^{6, 7} and an endothelial-specific knockout of Nrp-1 showed extensive cardiovascular pathology¹⁴. However, neither of these studies examined cardiac NRP-1 function due to the developmental lethality in the null mice nor addressed any underlining mechanisms. Interestingly, two recent studies have shown that VEGF-B, a ligand of NRP-1, can control fatty acid transport through endothelial cells of heart and muscle²⁷ and inhibition of VEGF-B significantly improved the ectopic lipid deposition and insulin resistance in obese mice²⁸. Although these studies suggest a potential role for NRP-1 in endothelial cells with respect to fatty acid transport, we did not observe significant downregulation of fatty acid transport protein (data not shown) or CD36 (Supplemental Fig. II) upon NRP-1 knockout in CMs or other non-endothelial cells. Therefore, the most likely role for NRP-1 in CMs and in the VSMC compartments may not be consistent with endothelial cells as VEGF may be the main functional ligand in endothelium but in CMs and VSMCs, TGFβ may be the key NRP-1 ligand of interest.

Consistent with previous studies that reported NRP-1 acts as a co-receptor for TGFβ and mediates activation of the TGFβ-Smad2 pathway in fibroblasts¹⁸, cancer cells²⁹ and T cells⁵, our results showed decreased phosphorylation of Smad2 in NRP-1 KO CMs (Figs.5& Supplemental VI). Meanwhile, previous studies also demonstrated the TGFβ pathway crosstalks with PPARγ in such a way that TGFβ downregulates PPARγ expression and its downstream effectors^30–32, while PPARγ ligands counteracted TGFβ-mediated fibrosis³³; however, the crosstalk between TGFβ-Smad2 pathway and PPARγ remain unclear in CMs. Our results showed that overexpression of Smad2 decreased the expression of PPARγ in CMs. Notably, TGFβ significantly reduced the PPARγ mRNA level but not total protein level in CMs (Fig.5), suggesting that there were other downstream mediators of TGFβ present to mediate inhibitive regulation on PPARγ.

Dysfunction of the right ventricle has been recognized as an important contributing factor in heart disease^{34, 35}. Right ventricle shortening is an independent predictor of survival in patients with advanced congestive heart failure^36–39. However, the mechanism involved in right ventricle dysfunction is still largely unknown. Our results identified the hypertrophic right ventricle in SM22-α-Nrp-1 KO mice, highlighting the cardiac protective role of NRP-1 in CMs and SMCs for the first time. Meanwhile, the dysfunction of the right ventricle in SM22-α-Nrp-1 KO mice is more predominant than the left ventricle, as shown by the hypertrophic right ventricle in younger mice and the dilated right ventricle in older mice (Fig.2). These results suggest that increased afterload may be present, which leads to the observed hypertrophy and subsequent dilation of the right ventricle. No significant lung vascular remodeling and fibrosis were found; however, perivascular infiltration of inflammatory cells was found in older SM22-α-Nrp-1 KO mice where significant dilation of the right ventricle was observed (Fig.3). Meanwhile, no significant difference in NRP-1 expression and the SM22-α-Cre expression pattern were present between the right ventricle and left ventricle (data not shown). This additionally suggests that a distinct NRP-1 signaling component may be involved between the right ventricle and left ventricle.

Recent studies have highlighted that metabolism in the heart contributes to systemic energy homeostasis. Resistance to high fat-induced obesity and insulin resistance has been achieved through the overexpression of MED13 or the inhibition of mir-208a in CMs²⁰. This increased cardiac energy expenditure was mediated through the regulation of several nuclear hormone receptors including PPARγ, PGC1α, and the thyroid receptor²⁰. We also examined the expression of mir-208a in NRP-1 deleted CMs and did not observe much difference compared to WT CMs (Supplemental Fig. XI). Our results (Fig.6) raised a possibility that signaling through cardiac NRP-1 is also essential for whole body metabolic homeostasis.

PGC1α is highly expressed in CMs and acts as a central regulator of mitochondrial biogenesis and respiration. PGC1α knockout mice exhibit moderate cardiac defects with decreased ATP concentration and diminished cardiac function after hemodynamic and metabolic challenges^{40, 41}. However, forced overexpression of PGC1α in CMs was shown to increase mitochondria biogenesis but lead to death from heart failure with a stage-specific mechanism^{42, 43}. Similarly, induced expression of PGC1α together with increased mitochondrial deposition and disorganized sarcomere structures were also observed in cardiac tissues of SM22-α-Nrp-1 KO mice (Fig.1). Meanwhile, downstream effectors of PGC1α, including TFB2M, MCAD, etc., which mediate proliferation and respiration of mitochondria, were also induced in NRP-1 deleted CMs. Metabolites analysis further confirm the increased TCA activity in cardiac tissues of SM22-α-Nrp-1 KO mice (Supplemental Fig.III). These results suggested that increased level of PGC1α contributed to the abnormal mitochondrial accumulation in SM22-α-Nrp-1 KO hearts. However, we need to note that the mitochondrial accumulation in SM22-α-Nrp-1 KO heart was not as robust as previously reported in cardiac-specific PGC1α transgenic mice^{42, 43}, which is probably due to the varying degree of PGC1α induction in SM22-α-Nrp-1 KO hearts and PGC1α transgenic hearts. Previous studies have shown that PPARγ expression is relatively low in CMs. Cardiomyocyte-specific overexpression of PPARγ can lead to dilated cardiomyopathy with increased lipid and glycogen storage defects in mice⁴⁴. We also evaluated the lipid accumulation in cardiac tissues of SM22-α-Nrp-1 KO mice. Although either total triglyceride or neural lipid was significantly altered (Supplemental Fig. X), expression of CPT1b, which acts as a downstream target of PPARγ and rate-limiting enzyme of fatty acid oxidation, was increased in cardiac tissues of SM22-α-Nrp-1 KO hearts (Supplemental Fig. II). Meanwhile, increased trends of free fatty acids, including linolenic, palmitic fatty acid, etc., were observed in SM22-α-Nrp-1 KO hearts (Supplemental Fig. X). These results suggest that elevated expression of PPARγ was likely related to increased fatty acid oxidation. Although the role of increased fatty acid oxidation in cardiomyopathy is still not clear, many studies suggest a beneficial effect results from inhibiting fatty acid oxidation in heart failure, especially in ischemic heart failure^45–48. Similarly, we observed that SM22-α-Nrp-1KO mice were more vulnerable to ischemic stress (Fig.3). These results suggest that increased PPARγ expression contributed to the cardiac phenotype in SM22-α-Nrp-1 KO mice. Previous studies have shown that PPARγ transgenic mice exhibited increased lipid accumulation at 8 months of age, and a significant increase in expression of PPARγ downstream targets was observed in MHC-PPARγ High but not MHC-PPARγ Low mice at 4 months old⁴⁴. These findings suggest the lipid toxicity effect of PPARγ is associated with its expression level and age, which partially explains the absence of increased lipid accumulation in SM22-α-Nrp-1 KO hearts. Furthermore, cardiomyocyte-specific knockout of PPARγ induced cardiac hypertrophy via increasing cardiac embryonic genes and activation of NFκB in mice⁴⁹. Taken together, the literatures strongly suggest that normal levels of PGC1α and PPARγ are essential for optimal CM function. Our data reveal a novel role for NRP-1 as a negative regulator of PGC1α and PPARγ expression and is critical in maintaining normal cardiovascular health by balancing mitochondrial homeostasis, particularly under ischemia.

Many studies have documented reduced levels of PPARγ/PGC1α in human and rodent heart failure, which is usually accompanied by a shift of utilization from fatty acid to glucose as an energy substrate and impaired mitochondrial biogenesis^{50, 51}; however, a few studies also have reported unchanged or increased PGC1α is observed in conjunction with human heart failure^{52, 53}. These studies suggested the roles of PPARγ/PGC1α in failing hearts need to be evaluated with respects to severity and subtype of heart failure. Our results hereby showed that knockout of cardiac NRP-1 increased PPARγ/PGC1α expression and led to cardiomyopathy and aggravated ischemia induced heart failure, which will facilitate the elucidation of how endogenous PPARγ/PGC1α are regulated in cardiomyopathy and ischemic heart failure.

From a syndromic perspective, previous studies have implicated NRP1 in several disease modalities of the cardiovascular system including DiGeorge syndrome⁵⁴. Genetic variations, specifically copy number variations (CNV), of NRP1 have been implicated in congenital heart disease⁵⁵. Recently, a genome-wide association study (GWAS) on Tetralogy of Fallot has suggested an association with a non-synonymous single-nucleotide polymorphism (SNP) rs2228638 located on the MAM domain in NRP1⁵⁶. Pharmacogenomic traits (response to angiotensin II receptor blocker therapy)⁵⁷ and inflammatory biomarkers⁵⁸ were shown to be associated with the loci 10p12.1, close to the NRP1 locus (10p12). Childhood obesity-related traits in the Hispanic population were also mapped to a genomic region (10p12.1) near the NRP1 locus⁵⁹. Emerging evidence suggests that apart from common variants, multiple rare variants of a gene in low frequency may impact complex diseases⁶⁰. To examine the contribution of rare variants in NRP1 to heart, blood, or lung diseases, we compiled rare variants of NRP1 from the NHLBI GO Exome Sequencing Project Rare Variants for African-Americans and European-Americans and results are summarized in Supplemental Fig. XII. Deep resequencing of these regions with a large number of patient samples will be required to delineate the association of rare or common NRP1 variants with various cardiovascular, inflammatory, obesity, and pharmacogenomics related traits.

In conclusion, our findings have identified the protective role of NRP-1 in cardiomyopathy and ischemia-induced heart failure and demonstrated that NRP-1 controls PGC1α and PPARγ expression and mitochondrial homeostasis through crosstalk with the Notch1 and Smad2 signaling pathways. Our results also have shown that NRP-1 in CMs and SMCs contributes to systemic metabolism. These results provide potential therapeutic targets for cardiomyopathy and metabolic disorders.

Significance.

Oxidative phosphorylation of fatty acid in mitochondria is the main source of energy production in heart. Aberrant mitochondrial bioenergetics contributes to energy deprivation, cellular stress and pathogenesis of cardiac diseases, such as cardiomyopathy, but it is still unclear how mitochondrial homeostasis is maintained in hearts. We now identified neuropilin-1 (NRP-1) controlled expression of PGC1α, the key regulator of mitochondrial biogenesis, in cardiomyocytes (CMs) and vascular smooth muscle cells (VSMCs). We provide evidences that knockout of NRP-1 in CMs and VSMCs in mice caused cardiomyopathy and aggravated ischemic stress induced-heart failure. Meanwhile, the importance of NRP-1 was supported by altered systemic metabolites in mice deficient of NRP-1 in CMs and VSMCs. Overall, our results suggest that NRP-1 signaling could be promising targets to modulate mitochondrial homeostasis in prevention and treatment of cardiac diseases and provide insights for role of cardiac metabolism in systemic metabolic disorders.

Abbreviations

ANF

Atrial natriuretic factor

BNP

Brain natriuretic peptide

βMHC

β-myosin heavy chain

CMs

Cardiomyocytes

CPT1b

Carnitine palmitoyltransferase 1B

Cav-1

Caveolin-1

EF

Ejection fraction

EE

Energy expenditure

FS

Fractional shortening

Notch NICD

Intracellular domain of Notch

LAD

Left anterior descending

MCAD

Medium-chain acyl-CoA dehydrogenase

SM22-α-Nrp-1 KO

Mice lacking Nrp-1 in CMs and VSMCs

TFAM

Mitochondrial transcription factor A

mtDNA

Mitochondrial DNA

MEF

Mouse embryonic fibroblasts

MI

Myocardial infarction

NRP-1

Neuropilin-1

NRF1 and NRF2

Nuclear respiratory factors 1 and 2

VO2

Oxygen consumption

PGC1α

Peroxisome proliferator-activated receptor γ coactivator 1α

PPARγ

Peroxisome proliferator-activated receptor γ

TFB1M

Transcription factor B1M

TFB2M

Transcription factor B1M

VSMCs

Vascular smooth muscle cells

WT

Wild-type

12, 13-DHOME

12, 13-hydroxyoctadec-9(Z)-enoate

9, 10-DHOME

9, 10-hydroxyoctadec-12(Z)-enoic acid