Targeting LOXL2 for cardiac interstitial fibrosis

Full text

Turn on search term navigation

ARTICLE

Received 11 Aug 2016 | Accepted 26 Oct 2016 | Published 14 Dec 2016

Jin Yang1,2,3,*, Konstantinos Savvatis4,5,*, Jong Seok Kang6,*, Peidong Fan6, Hongyan Zhong6, Karen Schwartz6, Vivian Barry6, Amanda Mikels-Vigdal6, Serge Karpinski6, Dmytro Kornyeyev6, Joanne Adamkewicz6,Xuhui Feng1,2,3, Qiong Zhou1,2,3, Ching Shang1,7, Praveen Kumar6, Dillon Phan6, Mario Kasner4, Begona Lpez8,

Javier Diez8, Keith C. Wright1, Roxanne L. Kovacs1, Peng-Sheng Chen1, Thomas Quertermous7, Victoria Smith6, Lina Yao6,*, Carsten Tschpe4,5,9,* & Ching-Pin Chang1,2,3,6

Interstitial brosis plays a key role in the development and progression of heart failure. Here, we show that an enzyme that crosslinks collagenLysyl oxidase-like 2 (Loxl2)is essential for interstitial brosis and mechanical dysfunction of pathologically stressed hearts. In mice, cardiac stress activates broblasts to express and secrete Loxl2 into the interstitium, triggering brosis, systolic and diastolic dysfunction of stressed hearts. Antibody-mediated inhibition or genetic disruption of Loxl2 greatly reduces stress-induced cardiac brosis and chamber dilatation, improving systolic and diastolic functions. Loxl2 stimulates cardiac broblasts through PI3K/AKT to produce TGF-b2, promoting broblast-to-myobroblast transformation; Loxl2 also acts downstream of TGF-b2 to stimulate myobroblast migration.

In diseased human hearts, LOXL2 is upregulated in cardiac interstitium; its levels correlate with collagen crosslinking and cardiac dysfunction. LOXL2 is also elevated in the serum of heart failure (HF) patients, correlating with other HF biomarkers, suggesting a conserved LOXL2-mediated mechanism of human HF.

1 Krannert Institute of Cardiology and Division of Cardiology, Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana 46202, USA. 2 Department of Biochemistry and Molecular Biology, Indiana University School of Medicine, Indianapolis, Indiana 46202, USA. 3 Department of Medical and Molecular Genetics, Indiana University School of Medicine, Indianapolis, Indiana 46202, USA. 4 Department of Cardiology, Campus Virchow-Klinikum, Charit University Medicine Berlin, 13353 Berlin, Germany. 5 Berlin-Brandenburg Centre for Regenerative Therapies, Charit University Medicine Berlin, 10117 Berlin, Germany. 6 Gilead Sciences Inc., Foster City, California 94404, USA. 7 Division of Cardiovascular Medicine, Stanford University, Stanford, California 94305, USA. 8 Program of Cardiovascular Diseases, Centre for Applied Medical Research, Department of Cardiology and Cardiac Surgery, University Clinic, University of Navarra, 31008 Pamplona, Spain. 9 DZHK, German Centre for Cardiovascular Research, Partner Site Berlin Charit, 13347 Berlin, Germany. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to L.Y.(email: mailto:[email protected]

Web End [email protected] ) or to C.-P.C. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

DOI: 10.1038/ncomms13710 OPEN

Targeting LOXL2 for cardiac interstitial brosis and heart failure treatment

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

Heart failure (HF) is a leading cause of death, with a mortality rate of B50% within ve years of diagnosis1. This high mortality rate reects inadequacy of modern

therapy and calls for a new mechanistic paradigm for treatment. A major cause of cardiac dysfunction is the adverse tissue remodelling with interstitial brosis26, caused by various pathological insults that include hypertension and myocardial infarction, with the extent of interstitial brosis prognosticating clinical outcomes of the diseased heart6. Current therapies that improve HF survival primarily target the pathogenic mechanisms that occur within cardiomyocytes but not those mechanisms that take place outside cardiomyocytes in the interstitial space that houses collagen and broblasts. This lack of direct therapy against interstitial mechanism of HF may contribute to the persistently high morbidity and mortality of HF.

The major causes of death in HF are ventricular arrhythmias and pump failure. Although the arrhythmic death can be effectively prevented by implantable cardiac debrillators, death from pump failure has become an even more pressing issue. Cardiac systolic and diastolic pump functions are both compromised in patients suffering from HF with reduced ejection fraction (HFrEF). However, current therapies only have moderate efcacy in systolic dysfunction and are ineffective for diastolic pump failure. For patients having HF with preserved ejection fraction (HFpEF), who have primarily diastolic dysfunction, there is no proven therapy that can improve diastolic function and long-term outcome. Therefore, an urgent need remains for novel, complementary medical therapies, which target maladaptive left ventricular tissue remodelling/brosis to improve systolic and diastolic functions of failing hearts2.

Interstitial brosis plays a key role in the development and progression of HF by causing adverse mechanical and electrical disturbances in the diseased hearts2. Interstitial brosis impairs the electromechanical coordination between cardiomyocytes to compromise systolic contractility. Fibrosis also increases ventricular stiffness and impairs diastolic relaxation and lling.

We found that Lysyl oxidase-like 2 (LOXL2) is upregulated in the interstitium of diseased mouse and human hearts. Increased LOXL2 expression leads to increased TGF-b2 production, triggering the formation and migration of myobroblasts with enhanced collagen deposition and crosslinking in the hyper-trophic regions of stressed hearts. These effects result in interstitial brosis and cardiac dysfunction. The concept of targeting LOXL2 in HF is supported by our preclinical data from cell-based assays, pharmacological studies and mouse genetic models, as well as by the clinical evidence of elevated cardiac tissue and serum LOXL2 in HF patients and correlation of tissue LOXL2 level with brosis and cardiac physiological changes in HF patients. The efcacy of anti-LOXL2 antibody and genetic LOXL2 disruption in relieving cardiac interstitial brosis and diastolic abnormalities is particularly salient, given the lack of direct therapy against cardiac brosis, and the increasing emphasis on diastolic dysfunction as an integral part of the heart failure syndrome.

ResultsLoxl2 proteins are elevated in the stressed mouse hearts. We used transaortic constriction (TAC) to cause pressure overload and hypertrophy of the hearts of male mice at 68 weeks of age7,8. Quantitative PCR (qPCR), western blot analysis and immunostaining of left ventricles showed that Loxl2 was minimally expressed in sham-operated hearts, but it was highly upregulated within the rst week of TAC, and the Loxl2 upregulation persisted for 410 weeks after TAC (Fig. 1ac,

Supplementary Fig. 1a). Genes that encode other LOX isoforms (Lox, Loxl1, Loxl3 and Loxl4) were also upregulated in the

stressed hearts but to a lesser extent and consistency than Loxl2 (Supplementary Fig. 1be). In TAC-stressed hearts, Loxl2 proteins were present in the interstitial space (Fig. 1c), which contains broblasts and collagen bres. The upregulation of Loxl2 in the interstitium was accompanied by transdifferentiation of broblasts into myobroblasts that are highly migratory (marked by expression of a-smooth muscle actin (a-SMA)) and synthetic of collagen (marked by expression of collagen, type I, alpha 1 (Col1A); Fig. 1d). The interstitium of TAC-stressed hearts, therefore, contains increased Loxl2 and collagen-producing myobroblasts, providing the enzyme (Loxl2) and substrates (collagen) required for collagen crosslinking. Indeed, TAC-stressed hearts showed increased interstitial collagen, which was primarily insoluble (crosslinked; Fig. 1eg), and such TAC-induced progressive increase of crosslinked collagen amount correlated strongly and linearly with the decline of left ventricular fractional shortening (LVFS) over time (Fig. 1h, r 0.95). Collectively, these ndings suggest a role of Loxl2

in stress-induced interstitial collagen deposits and cardiac dysfunction.

LOXL2 proteins are increased in the diseased human hearts. We next tested whether LOXL2 was upregulated in diseased human hearts. RTqPCR of heart tissues showed that the expression of LOXL2 and LOXL4 was low in healthy human hearts, whereas the other isoforms LOX, LOXL1 and LOXL3 were more abundant (Fig. 1i). However, in human hearts with ischaemic or non-ischaemic dilated cardiomyopathy, only LOXL2, but not other LOX isoforms (LOX, LOXL1, LOXL3 and LOXL4), was highly upregulated (Fig. 1j,k). Immunostaining veried the upregulation of LOXL2 protein in the interstitium of diseased human hearts (Fig. 1l). Quantication of messenger RNA (mRNA) revealed that LOXL2 expression correlated well with the expression of brillar collagen genes (COL1A and COL3A) in ventricular tissues of healthy donor hearts (n 8) and

of hearts from patients with idiopathic dilated cardiomyopathy (IDCM, n 10; Fig. 1m,n and Table 1). The clear separation of

control and cardiomyopathy samples in the LOXL2COLLAGEN correlation graphs indicates a functional role of LOXL2 elevation in the pathogenesis of human brotic hearts (Fig. 1m,n). Collectively, these ndings show a specic elevation of the LOXL2 isoform in the interstitial space of diseased human hearts.

LOXL2 is elevated in the serum of patients with heart failure. To explore the potential use of LOXL2 as a HF biomarker, we tested whether LOXL2 proteins could be released from cardiac interstitium into the circulation. We rst examined control subjects and patients with HFrEF (EFr35%), the demographics of which was in Supplementary Table 1. We found that LOXL2 levels were elevated in the serum of HFrEF patients (1299 pg ml 1, n 31) compared with control subjects

(714 pg ml 1, n 24; Students t-test, Po0.0001; Fig. 2a).

A cutoff of LOXL2 at 90100 pg ml 1 showed 88% specicity, 74% sensitivity and 80% accuracy in distinguishing HFrEF from the control subjects (Fig. 2a, red-dashed line). This high sensitivity/specicity was comparable to that of NT-proBNP, an established HF biomarker measured in the same serum samples (225 pg ml 1 cutoff measured by a research kit had 83%

specicity, 87% sensitivity and 85% accuracy, Fig. 2b). These numbers of sensitivity and specicity were also comparable to the published results of NT-proBNP as a diagnostic HF biomarker.

Serum LOXL2 level correlated with that of TIMP-1, a tissue brosis marker9 (Pearson correlation, r 0.5, Po0.001; Fig. 2c)

and ST-2, a cardiac remodelling and brosis marker (Pearson correlation, r 0.5, Po0.001; Fig. 2d). Of note, ST-2 plays a

2 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

Sham n =45

TAC n =45

c d

1 week 2 weeks 7 weeks

Loxl2 Loxl2 Loxl2

-SMA Col1A

Col1A

Loxl2mRNA expression

(normalized to -actin)

P <0.01 P <0.01

Weeks after TAC

Sham TAC 1 week

3 P <0.01 P =0.02 P =0.03

Sham 1 2 4 7

Marker (kDa)

Loxl2

Gapdh

Sham

TAC

-SMA

1 2 4 7 10

Weeks after TAC

e h

Sham TAC 1 wk Sham TAC 2 wk

Sham TAC 10 wk

1.5

P =0.0014

P =0.31

P =0.0015

P =0.24

2.5

P =0.0018

Control n =8

Ischemic cardiomyopathy n =8

P =0.0009

Collagen

(g mg1 of heart weight)

Collagen

(g mg1 of heart weight)

2.0

Collagen

(g mg1 of heart weight)

2.0

r=0.95

LOXs mRNA expression

(normalized to -actin)

1.5

P =0.0067

1.0

LOXs expression

(fold of control)

1.5

n =8

1.0

LVFS (%)

0.5

1.0

0.5

0 1 2 10 Time after TAC (wk)

0.0

0.0 0.5 1.0 2.0

1.5

n=5

Insoluble collagen content (g mg1 of heart weight, n =5)

Total Soluble Insoluble

LOX

LOXL1

LOXL2

LOXL3

LOXL4

LOX

LOXL1

LOXL2

LOXL3

LOXL4

Ventricular tissue Ventricular tissue

LOXL2

P=0.006

0.015 P<0.0001

r=0.82

COL1A/-actin

0.0020 P=0.0003 r=0.75

LOXs expression

(fold of control)

0.0015

COL3A/-actin

0.010

Control n =8

IDCM n =10

Control

ISCH

0.0010

LOXL2 LOXL2

0.005

0.0005

0.000

0.0000 0.0005 0.001

0.0000 0.0000 0.0005 0.0010

LOX

LOXL1

LOXL2

LOXL3

LOXL4

Control IDCM

LOXL2/ -actin

Figure 1 | LOXL2 correlates with cardiac dysfunction in both mouse model and human patients. (a) Quantication of Loxl2 mRNA expression in the heart 110 weeks after sham/TAC operation. n 45 mice per group. P value: Students t-test. Error bar: standard error of the mean (s.e.m.). (b) Western blot

analysis of Loxl2 protein in the mice heart ventricles 17 weeks after sham/TAC operation. P value: Students t-test. Error bar: s.e.m. (c) Representative immunostaining of Loxl2 in the heart 1, 2 and 7 weeks after sham/TAC operation. Scale bars, 100 mm. Blue: haematoxylin. Brown: Loxl2. (d) Representative immunostaining of Col1A and a-SMA in the heart 1 week after sham/TAC operation. Scale bars, 100 mm. Blue: haematoxylin. Brown: Col1A/a-SMA.

(eg) Determination of total, soluble and insoluble collagen content of left ventricles from sham- or TAC-operated mice 1, 2 and 10 weeks after the procedure (n 5 in each group). The amounts of total and soluble collagen were determined by measuring hydroxyproline content of total and pepsin-acid

solubilized left ventricles, respectively. The amount of insoluble collagen was calculated by subtracting the amount of soluble collagen from total collagen. (h) Correlation of left ventricular fractional shortening (LVFS) with the insoluble collagen content of left ventricles after sham or TAC operation (n 5 per

group). r: Pearson coefcient. (ik) mRNA quantication of LOX genes in ventricular tissues of healthy human donor hearts (n 8; i) or from patients with

ischaemic (n 8; j) or idiopathic dilated (n 10; k) cardiomyopathy. IDCM, idiopathic dilated cardiomyopathy. P value: Students t-test. Error bar: s.e.m.

(l) Immunostaining of LOXL2 (brown) in human heart tissues obtained from control subjects and patients with ischaemic (ISCH) or idiopathic dilated cardiomyopathy (IDCM). Scale bars, 20 mm. Blue: haematoxylin. Brown: Loxl2. (m,n) Correlation of the mRNA level of LOXL2 with COL1A (m) and COL3A (n) in the ventricular tissues of patients with idiopathic dilated cardiomyopathy (IDCM). r: Pearson coefcient. Black circles: cardiomyopathy tissues(n 10). Open circles: control heart tissues (n 8).

causal role in HF pathogenesis in animal models10, and ST-2 is a FDA-approved clinical serum HF biomarker that tracks the activity of cardiac broblasts and cardiomyocytes11,12. The correlations between LOXL2 and ST-2 and TIMP-1 suggest involvement of LOXL2 in the pathogenesis of human HFrEF.

Additional evidence implying a crucial role of LOXL2 in human HF came from our studies of patients with severe HFrEF receiving left ventricular assist device (LVAD; n 15;

Supplementary Table 1). Those patients that had persistently low cardiac EF (r35%), despite the LVAD therapy, exhibited elevated serum LOXL2 (146 pg ml 1), whereas those who showed signicant EF recovery (Z40%) had serum LOXL2 comparable to control levels (76 pg ml 1) (Fig. 2e). Serum

LOXL2 levels above 100 pg ml 1 had 90% sensitivity and 100% specicity of separating EF non-responders from responders following LVAD therapy (Fig. 2e). Moreover, EF recovery (DEF)

before and after LVAD therapy correlated well with serum LOXL2 lowering (Pearson correlation, r 0.8, P 0.001;

Fig. 2f). A cutoff of LOXL2 above 100 pg ml 1 had 100%

sensitivity and specicity in separating EF non-responder (DEF o10%) from responders (DEF Z10%; Fig. 2f). These results not only suggest a potential use of LOXL2 as a biomarker in this clinical setting, but also implicate a causal role of LOXL2 in human HF.

Serum LOXL2 levels were also elevated in patients with heart failure with preserved ejection fraction (HFpEF; 12713 pg ml 1, n 25) compared with the control subjects

(734 pg ml 1, n 24; Students t-test, Po0.0001; Fig. 2g).

A cutoff of LOXL2 at 90 pg ml 1 displayed 83% specicity, 68% sensitivity and 76% accuracy in distinguishing HFpEF from control subjects (Fig. 2g, red-dashed line). This high sensitivity/ specicity was again comparable to that of NT-proBNP, measured in the same serum samples (225 pg ml 1 cutoff had 74% specicity, 76% sensitivity and 75% accuracy, Fig. 2h).

Moreover, signicant correlation between sLOXL2 and tissue brotic marker TIMP-1 was observed in HFpEF (Fig. 2i). The serum samples of HFrEF and HFpEF patients were further validated by elevation of serum Troponin I levels measured by

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

Table 1 | Demographics and diagnosis of control subjects and patients.

Age Gender Ethnicity Diagnosis a Control subjects

49w M White NF* 54w M White NF 52w M White NF68 M White NF52 M White NF64 M White NF41 M White NF41 M White NF

b Patients with cardiomyopathy49w M White ISCH 54w M White ISCH 58w M White ISCH56 M White ISCH61 M White ISCH65 M White ISCH62 F White ISCH57 F White ISCH62 M White ISCH 34w M White IDCM 41w M White IDCM 47w M White IDCM61 M White IDCM55 M White IDCM62 M White IDCM47 M White IDCM66 M White IDCM65 M White IDCM53 M White IDCM63 M White IDCM55 M White IDCM46 M White IDCM

(a) Demographics and diagnosis of control subjects of heart organ donors and (b) patients with ischaemic or idiopathic dilated cardiomyopathy. IDCM, idiopathic dilated cardiomyopathy; ISCH, ischaemic cardiomyopathy.*NF: non-failure, normal heart function.

wSample for tissue immunostaining.

high-sensitivity troponin assays (Fig. 2j,k), suggesting continued cardiomyocyte injury in those patients. Given its accuracy in separating HFrEF or HFpEF patients from control subjects, serum LOXL2 may provide a new biomarker to track cardiac tissue remodelling and function in HF patients.

LOXL2 correlates with diastolic dysfunction in HFpEF. To address the diastolic aspect of human heart failure, we studied control subjects (n 15) and HFpEF patients (n 24), whose

demographic, echocardiographic and hemodynamic characteristics are listed in Table 2. The heart tissues from these patients showed intense interstitial brosis, coinciding with LOXL2 expression (Fig. 3a,b). In HFpEF patients, the extent of collagen crosslinking in the heart correlated well with diastolic relaxation abnormalities (measured by echocardiographic E/E0 ratio13) and with left ventricular end-diastolic pressure (LVEDP, measured by cardiac catheterization; Fig. 3c,d). These ndings indicate a crucial role of crosslinked collagen in determining diastolic relaxation and lling pressure of the left ventricle. More importantly, cardiac LOXL2 levels correlated with the amount of crosslinked collagen, the severity of diastolic relaxation abnormalities (by E/E0 ratio), and the elevation of LVEDP (Fig. 3eh). Because LOXL2 catalyses collagen crosslinking, these correlations support a LOXL2-mediated collagen mechanism that contributes to the diastolic dysfunction of HFpEF. Furthermore,

the consistent ndings of mouse and human heart tissue studies suggest an evolutionary conservation of LOXL2-based interstitial mechanism of cardiomyopathy.

Loxl2 inhibition reduces stress-induced cardiac dysfunction. To determine a causative role of Loxl2 in interstitial brosis and heart failure, we used a LOXL2-specic neutralizing monoclonal antibody (AB0023, a-LOXL2; refs 6,14) to inhibit Loxl2 activity in TAC-stressed mouse hearts. TAC and sham operations were performed on four groups of CD1 male micesham/ IgG1, sham/a-LOXL2, TAC/IgG1 and TAC/a-LOXL2 (Fig. 4a).

The a-LOXL2 or control IgG1 treatment (intraperitoneally, 30 mg kg 1, twice per week) was initiated 2 weeks after TAC, when the heart already displayed 35% hypertrophy (by ventricle

body weight ratio) and 25% reduction of left ventricular fractional shortening (LVFS by echocardiography) (Fig. 4a). Remarkably, within 2 weeks of treatment, a-LOXL2 stabilized LVFS of

TAC-stressed hearts, preventing further LVFS decline with a trend toward normalizing LVFS after 8 weeks of treatment (Fig. 4a). Ten weeks after TAC (8 weeks after treatment), the IgG1-treated group developed severe cardiac hypertrophy and brosis with 80.9% increase of ventriclebody weight ratio,87.9% increase of end-systolic LV internal diameter (LVIDs),39.0% increase of end-diastolic LV internal diameter (LVIDd), and 49.1% reduction of LVFS (Fig. 4bg). Conversely, a-LOXL2-treated mice displayed much less TAC-induced cardiac dysfunction. The interstitial brosis was essentially eliminated, with LVFS increased by 50.9% (P 0.01) and LV dilatation

reduced by 25.0% (LVIDd, Students t-test, P 0.03; Fig. 4bf).

Notably, these major improvements of left ventricular function occurred without signicant reduction of cardiomyocyte hyper-trophy (measured by ventriclebody weight ratio; Fig. 4g) and in the absence of signicant changes of immune cell inltration into the myocardium (Supplementary Fig. 2a,b). This suggests that the benecial effects of a-LOXL2 are caused not by attenuation of hypertrophy or immune-mediated cardiac damage but rather by a-LOXL2s reversal of interstitial brosis. Such anti-brosis effects of a-LOXL2 are likely the results of reduced collagen production and enhanced degradation of un-crosslinked collagen15. Collectively, the Loxl2 expression and inhibition data indicate that Loxl2 expression activated by cardiac stress can trigger interstitial brosis and cardiac dysfunction.

To assess how Loxl2 controlled ventricular function of stressed hearts, we used a micro-admittance catheter to measure left ventricular pressurevolume (PV) relationships of anaesthetized mice in vivo. Analyses of mice after 10 weeks of TAC with 8 weeks of a-LOXL2 treatment showed that LV pressure overload was comparable between the control IgG and a-LOXL2-treated hearts (Supplementary Fig. 2c), but a-LOXL2 dramatically reduced the left ventricular size and end-diastolic pressure (EDP) of stressed hearts (Fig. 4h). a-LOXL2 improved ejection fraction by 107% (Students t-test, Po0.01), stroke volume by 73% (Students t-test, P 0.01), stroke work by 48% (Students

t-test, P 0.01), and preload-adjusted maximal power (plPwr) by

78% (Students t-test, P 0.01; Fig. 4il). Strikingly, the stroke

work and plPwr were normalized (Fig. 4k,l), indicating that Loxl2 inhibition prevents stress-induced systolic abnormalities and restores cardiac contractile function to the sham control level.

Consistent with the recovery of cardiac contractile work, a-LOXL2 reduced the left ventricular end-systolic volume (ESV)

by 43% (Students t-test, Po0.001) and end-diastolic volume (EDV) by 19% (Students t-test, Po0.01) (Fig. 4m,n). This led to the elimination of TAC-induced enlargement of heart chamber at end diastole, reducing the heart size to the sham control level (Fig. 4n).

4 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

a b

3,000

NT-proBNP (pg ml1)

400 r

250 P <0.0001

P <0.0001

=0.5P =0.0002

LOXL2 (pg ml1)

TIMP-1 (ng ml1)

200

300

2,000

150

200

100

1,000

100

0 Control n=24

HFrEF n=31

0 Control n=23

HFrEF n=31

0 0 50 100 150 200 250

LOXL2 (pg ml1) n=59

250 P

=0.0002

60 r =0.5P =0.0002

0 Control n=24

ST-2 (ng ml1)

200

LOXL2 (pg ml1)

= 0.0007

150

EF (%)

100

0 50 100 150 200 250

0 LVAD (EF40) n=5

LVAD (EF35%) n=10

10 50 100 150 200

LOXL2 (pg ml1) n=60

LOXL2 (pg ml1) n=13

2,000

P =0.0002

300

TIMP-1 (ng ml1)

300 P =0.0004

LOXL2 (pg ml1)

NT-proBNP (pg ml1)

1,500

200

1,000

100

500

r =0.5P =0.0008

HFpEF n=25

0 Control n=23

HFpEF n=25

0 0 100 200 300 400

LOXL2 (pg ml1) n=49

80 P <0.0001

P =0.0003

cTnI (pg ml1)

0 Control n=22

HFrEF n=29

0 Control n=22

HFpEF n=23

Figure 2 | Serum LOXL2 as a biomarker for HFrEF and HFpEF. (a) Serum LOXL2 was measured by a customized ELISA-based assay. The red-dashed line represents a cutoff level of LOXL2 at 90 pg ml 1. The mean of each group is indicated by a horizontal line in the graph. *Po0.05 using unpaired Students t-test. (b) Plasma NT-proBNP level was measured by Luminex. The red-dashed line represents a cutoff level of NT-proBNP at 225 pg ml 1. The mean of each group is indicated by a horizontal line in the graph. *Po0.05 using unpaired Students t-test. (c) Correlation between serum LOXL2 and TIMP-1 measured by ELISA. The two-tailed P value for Pearson correlation was calculated using GraphPad Prism. (d) Correlation between serum LOXL2 and ST-2 measured by ELISA. The two-tailed P value for Pearson correlation was calculated using GraphPad Prism. (e) Serum LOXL2 versus post-LVAD EF r35%

and Z40%. LVAD: left ventricular assist device. The mean of each group is indicated by a horizontal line in the graph. *Po0.05 using unpaired Students t-test. (f) Serum LOXL2 versus the degree of EF recovery of patients following LVAD therapy. The red-dashed line represents a cutoff level of LOXL2 at 100 pg ml 1. The two-tailed P value for Pearson correlation was calculated using GraphPad Prism. (g) Serum LOXL2 measured in HFpEF patients. The red-dashed line represents a cutoff level of LOXL2 at 90 pg ml 1. The mean of each group is indicated by a horizontal line in the graph. *Po0.05 using unpaired Students t-test. (h) Plasma NT-proBNP level was measured by Luminex in HFpEF patients. The red-dashed line represents a cutoff level of

NT-proBNP at 225 pg ml 1. The mean of each group is indicated by a horizontal line in the graph. *Po0.05 using unpaired Students t-test. (i) Correlation between serum LOXL2 and TIMP-1 measured by ELISA. (j,k) Serum troponin I level in HFrEF (j) and HFpEF (k). The mean of each group is indicated by a horizontal line in the graph. *Po0.05 using unpaired Students t-test.

Of note, the pressure overload by itselfwithout invoking cardiomyopathyis sufcient to increase ESV (but not EDV) of a healthy heart, thus leading to a myopathy-independent, mathematical reduction of EF, which equals (1 minus ESV/EDV). Given this pressure effect, we concluded that the residual ESV and EF changes of a-LOXL2-treated hearts are primarily the result of persistent pressure load. This conclusion is supported by the normalization of EDV, stroke work and contractile power of a-LOXL2-treated hearts. Therefore, a-LOXL2 is capable of decreasing stress-related cardiac dilatation and cardiac contractile dysfunction.

In addition, a-LOXL2 normalized the diastolic function of stressed hearts. In TAC-stressed hearts, myocardial relaxation was impaired, and the left ventricles became stiff. The myocardial relaxation abnormalities were evident by an increase of isovolumic relaxation time constant (Tau, 2.32-fold increase) and decline of diastolic pressure relaxation (maximal diastolic dp/dt, 24% reduction; Fig. 4o,p). Besides abnormal relaxation, the TAC-stressed left ventricles were stiff, as shown by the steep slopes of diastolic pressurevolume curves of stressed hearts (Fig. 4q). The combination of slow relaxation and stiff ventricles resulted in elevated left ventricular lling pressure or

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

Table 2 | Demographics, echocardiographic and hemodynamic parameters in control subjects and patients diagnosed with HFpEF.

Control (n 15) HFpEF (n 24) P value

DemographicsAge 47.36.3 52.57.4 0.07 NT-proBNP (pg ml 1) 61.4454.26 198.6191 0.047

Weight (kg) 81.921 78.115.7 0.833 6-min walk test (m) 540.847 339.5958.69 o0.001

NYHA II/III 0/0 17/7 Women/men 4/11 13/11 0.11 Hypertension 1 19 o0.001

Diabetes mellitus 0 4 0.146 Obesity 2 8 0.263 Dyslipidaemia 4 13 0.116 Smoker 4 6 1.0

Echocardiographic measurementsLA parasternal (mm) 35.64.9 37.56.5 0.319 LVEDD (mm) 51.32.63 49.56.74 0.491 IVSd (mm) 10.81.66 123.7 0.472 LVPWd (mm) 10.11.34 11.343.8 0.35 LVEF (%) 5912 58.38.6 1.0

Mitral inowE (m s 1) 0.810.19 0.830.2 0.831 A (m s 1) 0.680.42 0.820.18 0.038

E/A 1.250.42 1.10.37 0.138 DT (ms) 197.459.9 215.747.9 0.261 IVRT (ms) 103.628.9 96.217.7 0.465

Pulmonary vein owSystolic, S (m s 1) 0.530.07 0.620.1 0.432 Diastolic, D (m s 1) 0.660.12 0.560.13 0.4

Ar (m s 1) 0.320.32 0.350.5 0.629 S/D ratio 0.831.9 1.10.27 0.393

Tissue velocity imagingS0 mean (cm s 1) 9.133.6 7.72.4 0.73 E0 mean (cm s 1) 10.93.9 6.52.1 o0.001

A0 mean (cm s 1) 8.42.4 7.52.9 0.262 E0/A0 mean 1.330.5 10.38 0.061 E/E0 mean 8.42.17 14.98.04 o0.001

Hemodynamic measurementsLVEDP (mm Hg) 11.13.8 15.45.8 0.031

AR: atrial reversal; DT, deceleration time; IVRT, isovolumic relaxation time; IVSd, interventricular septum thickness (measured in diastole); LA, left atrium; LVEDP, left ventricular end-diastolic pressure; LVEF, left ventricular ejection fraction; LVEDD, left ventricular end-diastolic diameter; LVPWd, left ventricular posterior wall thickness (measured in diastole); NT-proBNP, N-terminal pro-brain natriuretic peptide; NYHA, New York Heart Association.

end-diastolic pressure (EDP), 3.83-fold over the control level (Fig. 4r). Conversely, in a-LOXL2-treated, TAC-stressed hearts,

Tau was reduced by 81% (Students t-test, P 0.01) to levels not

signicantly different from the controls (Fig. 4o). Also, EDP was reduced by 74% (Students t-test, Po0.01), with near normalization to the control level (Fig. 4r). Therefore, the TAC-stressed hearts treated with a-LOXL2 had minimal brosis (Fig. 4b,c) and compliant ventricles with essentially normal lling pressure. The data indicate that Loxl2 is essential for the development of interstitial brosis, ventricular stiffness and diastolic dysfunction in pathologically stressed hearts.

The combined effects of a-LOXL2 on the systolic and diastolic abnormalities resulted in 70% enhancement of cardiac output (Students t-test, P 0.01; Fig. 4s) and normalization of

serum levels of cardiac stress marker (BNP) and brosis marker (TIMP-1; Fig. 4t,u). Moreover, a-LOXL2-treated mice showed a mortality rate of 20% over 30 weeks of observation, whereas IgG-treated TAC mice had 70% mortality (Fig. 4v). Therefore,

a-LOXL2 is effective in protecting the hearts from stress-induced failure.

Loxl2 knockout protects stress-induced cardiac dysfunction. Next we asked whether Loxl2, among all Lox isoforms, played a specic role in HF pathogenesis. To address that, we used CRISPR/Cas9 and two-cut strategy to generate LoxP-anked (oxed) alleles of the mouse Loxl2 gene (Loxl2/) and then used the Cre-Lox and tamoxifen-induction methods to exert genetic knockout of Loxl2 at the desired time window (Supplementary Fig. 3a). The LoxP sequences that anked exon 8 and 9 of Loxl2, once recognized and recombined by Cre recombinase, would enable the deletion of the two exons, resulting in disruption of gene regions encoding the SRCR and downstream catalytic domains. The SRCR domain is the Loxl2 region targeted by anti-LOXL2 (ref. 16; Supplementary Fig. 3a). The oxed alleles were genotyped by PCR and conrmed by sequencing (Supplementary Fig. 3b,c). To delete Loxl2, we crossed Loxl2/

6 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

LOXL2

r =0.64P =0.015

Echocardiographic E/E ratio

Cross-linked collagen

LOXL2

COL1A COL1A

Control

HFpEF

Control HFpEF

0 0 10 20 30 40 50

r =0.67 P =0.02

r =0.67P =0.009

r =0.49P =0.033

r =0.55P =0.018

r =0.59P =0.019

Cross-linked collagen

COL1A (area fraction %)

0.0

LVEDP (mmHg)

Echocardiographic

E/Eratio

0.3

0.2

0.1

0 0 1 2 3LOXL2 (area fraction %) LOXL2 (area fraction %)

0 0 5 10 15 20 25

0 1 2 3 4

0 0 2 4 6 8 10

0 0 1 2 3 4

LOXL2 (area fraction %)

Figure 3 | LOXL2 correlates with collagen crosslinking and diastolic abnormalities in HFpEF. (a,b) Immunostaining of LOXL2 (a) and COL1A (b) in heart tissues of control subjects and HfpEF patients. Scale bars, 100 mm. Blue: haematoxylin. Brown: LOXL2 (a) or COL1A (b). (ce) Correlation of crosslinked collagen with echocardiographic E/E0 ratio (c), left ventricular end-diastolic pressure (LVEDP, d) and LOXL2 (e). r: Pearson coefcient.

(fh) Correlation of LOXL2 level with COL1A (f), echocardiographic E/E0 ratio (g) and left ventricular end-diastolic pressure (LVEDP, h) in human studies. r: Pearson coefcient.

LVEDP (mmHg)

LOXL2 (area fraction %)

mice to a driver mouse line that carried the transgene Actin-CreERT. The expression of ActinCreERT is driven by the chicken beta-actin promoter/enhancer coupled with the cytomegalovirus immediate-early enhancer17, and its Cre recombinase (CreERT) requires tamoxifen to activate the enzymatic function. The ActinCreERT; Loxl2/ genetic combination, therefore, allowed us to use tamoxifen to induce global Loxl2 knockout in adult mice, mimicking the systemic administration of anti-LOXL2. We treated the mice with tamoxifen for 5 days to delete Loxl2 exon 8 and 9 from the genome (Supplementary Fig. 3d). Interestingly, the Loxl2 mRNA that lacked exon 8 and 9 became destabilized and could not be detected by qPCR primers that targeted either deleted or non-deleted regions of the mRNA (Supplementary Fig. 3e). Neither could fragments of Loxl2 proteins be detected by western blotting using polyclonal antibodies (Supplementary Fig. 3f). These ndings indicate a complete loss of Loxl2 protein in the knockout mice.

Male mice of Loxl2-null and their littermate control (ActinCreERT; Loxl2/ or Loxl2/, Loxl2/ ) at 810 weeks of age were then subjected to TAC and followed for the development of HF for 10 weeks (Fig. 5a). During the course, there were no observable adverse effects of Loxl2 mutations on the cardiac or gross functions of the mice, suggesting that Loxl2 is dispensable for the baseline function of mice. However, Loxl2 knockout prevented TAC-induced cardiac interstitial brosis and dysfunction (Fig. 5be), but no effects on myocyte hypertrophy (Supplementary Fig. 4a). Interstitial brosis was essentially eliminated (Fig. 5b), with 81% increase of cardiac FS (Fig. 5c) and 2436% reduction of LV dimensions measured by echo-cardiography (Fig. 5d,e). Cardiac catheterization and PV loop analysis further conrmed that Loxl2 deletion protected the heart from TAC-induced chamber dilatation and functional decline. After 10 weeks of TAC, the Loxl2-null mice (ActinCreERT; Loxl2/) exhibited great improvement of cardiac mechanical function (Fig. 5f). In Loxl2-null mice, the systolic function was greatly improved: Cardiac EF was improved by 99%, plPwr increased by 55% (normalized), ESV reduced by 42% and EDV reduced by 63% (normalized; Fig. 5gj). The diastolic function was also greatly improved. EDP was reduced by 57%, Tau reduced by 32% and the slope of diastolic PV curve reduced, approaching the diastolic slope of compliant hearts (Fig. 5km). Overall, the cardiac performance was greatly improved by

anti-LOXL2: SV increased by 63%, CO by 94% and SW by 61% (normalized; Fig. 5np). Consistently, cardiac stress and brosis gene markersb-MHC, b/a-MHC ratio, Bnp, Timp-1, a-SMA,

Col1A, Col3A, Opn and Osf2were much reduced and essentially normalized in Loxl2-null hearts (Supplementary Fig. 4b,c). Taken together, the genetic studies thus demonstrate a requirement of Loxl2 in the development of HF.

To map the cellular site where Loxl2 functioned to trigger HF, we performed tissue-specic knockout of Loxl2 in cardiac broblasts, using the promoter of Tcf21a transcription factor specically expressed in broblasts18to drive the expression of Cre recombinase (CreERT) whose enzymatic activity is inducible by tamoxifen (iTcf21CreERT; Supplementary Fig.5a,b). Fibroblast-specic deletion of Loxl2 can therefore be executed in iTcf21CreERT;Loxl2/ mice treated with tamoxifen (Supplementary Fig. 5c). By the end of 10 weeks after TAC, mice without Loxl2 in cardiac broblasts exhibited fractional shortening (FS) 67% higher than that of mice expressing Loxl2 in broblasts (Fig. 5q,r). The magnitude of FS improvement in mice lacking broblast Loxl2 is comparable to that of anti-LOXL2 therapy and to that of global Loxl2 knockout at the corresponding time window (Figs 4a and 5a). These observations indicate a crucial role of Loxl2 secreted by broblasts in HF pathogenesis.

LOXL2 promotes TGF-b2 production through the PI3K path. Although a-LOXL2 and LOXL2 specic gene deletion essentially eliminated interstitial brosis and ventricular stiffness, it had only marginal effect on cardiomyocyte hypertrophy (Fig. 4g and Supplementary Fig. 4a). The uncoupling of myocyte hypertrophy and interstitial brosis suggests that LOXL2 provides a myocyte-independent, interstitial mechanism to control stress-induced matrix modications. In the interstitium of stressed hearts, TGF-b provides the primary signal to trigger the transdifferentiation of broblasts into myobroblasts, which then migrate to and deposit collagen in stressed areas of the heart4,5,15,1926. To test whether LOXL2 was required for TGF-b-induced cardiac myobroblast transformation, we knocked down LOXL2 and examined TGF-b activity in the primary human cardiac ventricular broblasts. Human cardiac broblasts contained abundant LOXL2, and LOXL2-targeted siRNA effectively removed LOXL2 from these cells (Fig. 6a).

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

In LOXL2-decient cells, the phosphorylation of SMAD2 was nearly absent without signicant changes of SMAD2 protein amount (Fig. 6a), suggesting inactive TGF-b signalling.

Quantication of TGF-b ligands revealed that the human cardiac broblasts secreted primarily TGF-b2 and that LOXL2 was required for TGF-b2 production. Without LOXL2, the

TGF-b2 ligand secreted by broblasts was reduced by B50% (Students t-test, Po0.0001; Fig. 6b), accompanied by reduced expression of TGF-b target genes, COL1A, FN1 and a-SMA (Fig. 6c) that are characteristic markers of myobroblasts27.

Consistent with the results from LOXL2 knockdown experiments, overexpression of LOXL2 in cardiac broblasts increased TGF-b2,

a b

IgG1 sham n =10 -LOXL2 sham n =10

IgG1

-LOXL2

P =0.002

Fractional shortening

(LVFS) (%)

2.5

2.0

1.5

1.0

0.5

0.0

Sham

Fibrosis (% of area)

P<0.01

TAC 10 weeks

20 IgG1 or -LOXL2(30 mg kg1, i.p., twice/week)

IgG IgG -LOXL2

-LOXL2

0 0 1 2 3 4 5 6 7 8 9 10 11

n =8 n =9 n =8 n =9

Weeks after TAC

Sham TAC 10W

P =0.03

6 P =0.06

P <0.01

Fractional shortening

(LVFS) (%)

Ventricle-body weight ratio

(mg g1 )

P =0.02

LVIDs (mm)

LVIDd (mm)

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

n =10n =10n =10n =10

n =10 n =10 n =10 n =10

n =10n =10n =10n =10

Sham TAC 10W

h j

IgG sham 10W

-LOXL2 TAC 10W

P =0.06

-LOXL2 sham 10W

Ejection fraction (%)

Stroke volume (l)

Stroke work (pwr103)

IgG TAC 10W

P =0.01

160

P <0.01

P =0.01

Left ventricular

pressure (mmHg)

120

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

0 10 20 30 40 50 60

n =8 n =9 n =8 n =9

Left ventricular volume (l)

Sham TAC 10W

m n

P =0.45

P =0.20

P =0.37

End diastolic volume (l)

0.05

P <0.001

80 P <0.01

plPwr (mWatt/(ll))

End systolic volume (l)

P =0.01

0.04

Tau (mS)

0.03

0.02

0.01

0.00

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2

Sham TAC 10W

n =8 n =9 n =8 n =9

Sham TAC 10W

P =0.11 P =0.07

Maximal diastolic dp/dt

(mmHg s1 )

IgG1 sham 10W

10,000

EDP (mmHg)

Left ventricular

diastolic pressure

(mmHg)

P <0.01

8,000

-LOXL2 sham 10W IgG1 TAC 10W -LOXL2 TAC 10W

IgG1, TAC

6,000

4,000

Sham

2,000

IgG IgG -LOXL2

-LOXL2

10 20 30 40 50

IgG IgG -LOXL2

-LOXL2

n=8 n=9 n=8 n=9

n =8 n =9 n =8 n =9

Left ventricular diastolic volume (l)

Sham TAC 10W

s v

P =0.76

P =0.08

Start

150

120

Percent survival (%)

P =0.017

P =0.01

P =0.019

CO (ml min1 )

BNP (pg ml1 )

TIMP1 (ng ml1 )

100

P<0.01

IgG1 sham n =10 -LOXL2 sham n =10 IgG1 TAC n =20 -LOXL2 TAC n =20

IgG IgG -LOXL2

-LOXL2

IgG IgG -LOXL2

-LOXL2 IgG IgG -LOXL2

-LOXL2

n =8 n =9 n =8 n =9

n =9 n =9 n =8n =10

Sham TAC 10W

n =9 n =10 n =9 n =10

0 0 4 8 20

12 16 24 28 32

Sham TAC 10W

Weeks after TAC

8 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

myobroblasts, which are critical for the development of ventricular brosis2,4,5. This conclusion is further supported by the observation that LOXL2 inhibition nearly eradicated myobroblast-dependent Col1A deposition and brosis in TAC-stressed mouse hearts (Figs 4b and 6j).

In the stressed hearts, the newly formed myobroblasts migrate to the stressed or hypertrophic region, where the myobroblasts produce collagen-rich matrix, causing ventricular brosis2,4,5.Given that myobroblast migration requires TGF-b signals27,3133, we tested whether LOXL2 was essential for controlling myobroblast migration using a gap closure assay.Human ventricular broblasts were seeded onto collagen I-coated plates, and the cell monolayer was scratched to create a cell-free zone or gap on the plate (Fig. 6k). The cells were then treated with TGF-b2 to stimulate cell migration into the gap (Fig. 6k). In this assay, TGF-b2 enhanced the number of cells migrating into the gap by B3-fold, whereas a-LOXL2 completely blocked such effect of TGF-b2 (Fig. 6l). Because TGF-b2 was added exogenously in the assay, the inhibition of cell migration by a-LOXL2 was not a result of inadequate TGF-b2 secretion (Fig. 6b). LOXL2 inhibition likely changed collagen crosslinking and the microstructure of the extracellular matrix to inhibit broblast migration. Collectively, the in vitro and in vivo studies indicate that LOXL2 is a crucial molecule activated by cardiac stress to transform and mobilize broblasts, triggering collagen synthesis, crosslinking and myobroblast migration, thereby causing diffuse interstitial brosis and ventricular dysfunction (Fig. 7).

DiscussionCardiac interstitial brosis is a major cause of systolic and diastolic abnormalities26 and a strong predictor of the clinical outcomes of patients with HF across a wide spectrum of disease severity6. However, the brotic process has not been a direct therapeutic target for HF. In patients with HF, LOXL2 levels are elevated in heart tissues and serum, its levels correlating with cardiac dysfunction and HF biomarker levels. In mice, LOXL2 activation is essential for cardiac brosis and HF development.The genetic or pharmacological inhibition of LOXL2 greatly reduces cardiac brosis and halts HF progression. The human sample and animal efcacy studies, in combination, suggest a pathogenic role of LOXL2 in cardiac brosis and human HF.Our studies therefore delineate a novel LOXL2-mediated HF mechanism and provide new insights into HF therapy.

In mechanically stressed hearts, LOXL2 expression is activated in the broblasts, which then release LOXL2 proteins into the interstitial space, particularly in the hypertrophic area where the mechanical stress is highest (Fig. 7). This suggests that LOXL2 activation is a response of cardiac broblasts to enhanced mechanical stress to maintain structural integrity of the heart.Indeed, LOXL2 elevation has multiple biological effects that

Figure 4 | LOXL2 inhibition reduces interstitial brosis and reverses heart abnormalities. (a) Time course of left ventricular fractional shortening changes after sham/TAC operation with IgG1 or a-LOXL2 antibody treatment. i.p.: intraperitoneal injection. P value: Students t-test. Error bar: s.e.m.

(b,c) Trichrome staining (b) and quantication (c) of cardiac interstitial brosis in IgG1 or a-LOXL2-treated mice 10 weeks (10W) after sham or TAC operation. Scale bars, 100 mm. Red: cardiomyocytes. Blue: brosis. (df) Echocardiographic measurement of left ventricular fractional shortening (d), left ventricular internal diameter at systole (LVIDs, e) and at diastole (LVIDd, f) after 10 weeks of TAC. P-value: Students t-test. Error bar: s.e.m.

(g) Ventriclebody weight ratio of hearts harvested 10 weeks after sham or TAC operation. P value: Students t-test. Error bar: s.e.m. (h) Representative cardiac pressurevolume relationships of IgG1- or a-LOXL2-treated mice 10 weeks after sham or TAC operation. (ip) Quantication of ejection fraction (i), stroke volume (j), stroke work (k) and preload-adjusted maximal power (plPwr, l), end-systolic volume (m), end-diastolic volume (n), isovolumic relaxation time constant Tau (o) and maximal diastolic dp/dt (p) of IgG1- or a-LOXL2-treated mice 10 weeks after sham or TAC operation. P value: Students t-test.

Error bar: s.e.m. (q) Diastolic pressurevolume relationships of IgG1- or a-LOXL2-treated mice 10 weeks after sham or TAC operation. (ru) End-diastolic pressure (EDP, r), cardiac output (s), serum brain natriuretic peptide (BNP, t) and serum tissue inhibitor of metalloproteinase 1 (TIMP1, u) of IgG1- or a-LOXL2-treated mice 10 weeks after sham or TAC operation. P value: Students t-test. Error bar: s.e.m. (v) LOXL2 antibody signicantly reduces mortality in TAC-operated mice. Data were analysed with KaplanMeier tests.

but not TGF-b1 or TGF-b3 production in the culture media (Supplementary Fig. 6a). Interestingly, LOXL2 increased TGF-b2 protein production without signicant effect on TGF-b2 mRNA expression (Supplementary Fig. 6b), suggesting that LOXL2 regulates TGF-b2 at the translation level. Given that the

PI3K/AKT/mTORC1 signalling pathway is known to regulate translation and promote heart failure28,29, we tested whether LOXL2 functioned through the PI3K path to enhance TGF-b2 translation. We found that LOXL2 overexpression in human cardiac broblasts increased AKT phosphorylation at serine 473 and threonine 308, without changing total AKT amount (Fig. 6d). LOXL2 also increased phosphorylation of the mTORC1 target protein S6K and 4E-BP1 without changing their total protein levels (Fig. 6d). Notably, PI3Ka (but not PI3b, PI3Kg or PI3Kd)

knockdown or inhibition decreased LOXL2-induced AKT/ mTORC1 signalling and reduced TGF-b2 protein level in the culture media (Supplementary Fig. 6cf). These results suggested that LOXL2 activates PI3K/AKT/mTORC1 signalling in cardiac broblasts. Such PI3K activation by LOXL2 was required for TGF-b2 production. PI3K and mTORC1 inhibition not only blocked TGF-b2 signalling activation (Fig. 6d), but also prevented the LOXL2-induced TGF-b2 production (Fig. 6e).

We also measured TGF-b2 level in heart tissues by western blotting after 6 weeks of TAC with or without antibody treatment. We found that cardiac TGF-b2 level increased by B2.5-folds in

TAC-stressed hearts, whereas in a-LOXL2-treated TAC hearts, such stress-induced change of TGF-b2 was reduced by 55% (Students t-test, P 0.03) to levels not signicantly different

from the controls (Fig. 6f,g). Consistent with the antibody treatment data, in TAC-stressed Loxl2-null hearts, the target genes of TGF-ba-SMA, Col1A and Col3Awere reduced to normal level (Supplementary Fig. 4c). These in vivo data are consistent with the ability of Loxl2 to stimulate production of TGF-b2 in broblasts. Collectively, these results indicate a signalling cascade from LOXL2 to PI3Ka/AKT/mTORC1 and then to TGF-b2 to stimulate cardiac brosis (Fig. 6h).

LOXL2 controls myobroblast transformation and migration. The regulation of myobroblast marker expression by LOXL2 suggested that LOXL2 was required for broblasts to transform into myobroblasts. To test that, we examined broblast cell morphology in cell culture. Myobroblasts are morphologically large and polygonal, whereas broblasts are slender and spindle-shaped30. Under cell culture conditions, human cardiac broblasts produced TGF-b2 and displayed morphological features of the large, polygonal myobroblasts (Fig. 6i). In contrast, LOXL2-decient broblasts, with reduced TGF-b signalling and a-SMA expression (Fig. 6b,c), were slender and spindle-shaped (Fig. 6i), consistent with a lack of myobroblast transformation. Therefore, both the marker and cell morphology studies indicate that LOXL2 promotes the formation of cardiac

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

a b c d

Global Loxl2 KO

P =0.02

Ctrl

Fractional shortening (%)

P =0.02

24%

FS (%)

LVIDd (mm)

ActinCre; Loxl2 fl/fl

10 * Ctrl TAC vs Mut TAC, P <0.05 0

0 Ctrl n =4

Mut n =4

Ctrl n =4

4 7 10 Weeks after TAC

Mut n =4

Sham TAC 10W

0 Ctrl Mut Ctrl Mut

Sham TAC 10W

n =4 n =4

n =4

e f g

P =0.02

Ctrl Sham n =4

ActinCre; Loxl2 TAC 10W n =4

Ctrl TAC 10W n =4

ActinCre; Loxl2 Sham n =4

160

LV pressure (mmHg)

36%

140

P <0.01

LVIDs (mm)

EF (%)

120

100

+99%

Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

0 10 20 30 40 50 60

Sham TAC 10W

LV volume (l)

Sham TAC 10W

h i j k

0.04

pIPwr (mWatt/ll)

P =0.01

P <0.01

42%

P =0.01

20%

EDP (mmHg)

P =0.03

57%

0.03

0.02

ESV (l)

EDV (l)

0.01

0.00

Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

0 Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

0 Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4 Sham TAC 10W

0 Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

Sham TAC 10W

l m n

P =0.04

32%

P =0.02

+63%

Ctrl Sham

ActinCre; Loxl2 fl/fl TAC 10W

ActinCre; Loxl2 fl/fl Sham

Ctrl TAC 10W

Tau (mS)

Left ventricular

diastolic pressure

(mmHg)

SV (l)

Ctrl, TAC

Sham

0 Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

0 10 20 30 40 50

0 Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

Sham TAC 10W

Left ventricular diastolic volume (l)

Sham TAC 10W

q r

Fibroblast Loxl2 KO

CO (ml min1 )

0.4

Stroke work (mJ)

P =0.03

P =0.02

Fractional shortening (%)

<0.01

0.3

0.2

FS (%)

0.1

0.0

0 Ctrl n =4

Mut n =4

Ctrl n =4

10 * Ctrl TAC vs Mut TAC, P <0.05

Mut n =4 Sham TAC 10W

Ctrl n =4

Mut n =4

Ctrl n =4

Mut n =4

0 Ctrl n =4

Mut n =4

Ctrl n =5

Mut n =4

Sham TAC 10W

0 4 7 10

Weeks after TAC

Sham TAC 10W

Figure 5 | LOXL2 gene deletion improves cardiac function of stressed hearts in mice. (a) Cardiac fractional shortening changes of ActinCreERT;Loxl2/ KO mice over 10 weeks (10W) after TAC. Ctrl: ActinCreERT; Loxl2/ , Loxl2/ or Loxl2/ . KO: ActinCreERT; Loxl2/. (b) Cardiac interstitial brosis detected by Massons trichrome staining in Ctrl or Loxl2-null mice 10 weeks after sham or TAC operation. Scale bars, 100 mm. Red: cardiomyocytes. Blue: brosis. (blue: collagen staining). (ce) Echocardiographic measurement of left ventricular fractional shortening (c), left ventricular internal diameter at diastole (LVIDd, d) and at systole (LVIDs, e) after 10 weeks of TAC. P value: Students t-test. Error bar: s.e.m. (f) Representative cardiac PV-Loop of Ctrl or Loxl2-null mice 10 weeks after sham or TAC operation. (gp) Quantication of ejection fraction (g), preload-adjusted maximal power (plPwr, h), end-systolic volume (i), end-diastolic volume (j), end-diastolic pressure (k), isovolumic relaxation time constant Tau (l), diastolic pressurevolume relationships (m), stroke volume (n), cardiac output (o) and stroke work (p) of Ctrl or Loxl2-null mice 10 weeks after sham or TAC operation. P value: Students t-test. Error bar: s.e.m. (q,r) Cardiac fractional shortening changes of iTcf21CreERT; Loxl2/ mutant mice over 10 weeks after TAC (q), and quantication of fractional shortening at 10 weeks after TAC (r). Ctrl: iTcf21CreERT; Loxl2/ , Loxl / or Loxl / . Mut: iTcf21CreERT; Loxl2/. P value: Students t-test. Error bar: s.e.m.

10 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

a b

Ctrl siRNA, n =4

LOXL2 siRNA, n =4

2.5

TGF-in the media

(ng ml1 )

TGF-2 in the media

(pg ml1 )

Ctrl siRNA LOXL2 siRNA

Control+IgG1 TGF-2+IgG1

1.0

P <0.0001

LOXL2 siRNA

2.0

0.8

1.5

0.6

Marker (kDa)

0.4

P =0.239

1.0

0.20.0

0.5

LOXL2

p-SMAD2

SMAD2

GAPDH

IgG+TAC10W -LOXL2+TAC10W

TGF-1 TGF-2 TGF-3

DMSO

A66

BYL719

LY294002

PI828

Rapamycin

TGF- target genes

Fold expression vs. control

COL1A1 3.9

-SMA 8.8 FN1 2.3

PI3Kpan PI3K Adeno-LOXL2

Inhibitors

mTORC1

Adeno-GFP

d f

Inhibitors

Sham TAC 6W

Marker (kDa)

TGF-2+-LOXL2

mTORC1

IgG -LOXL2

-LOXL2

IgG

DMSO

PI3Kpan PI3K PI3K

DMSO

LY294002 10 M

PI828 10 M

TGF-2

TFIIb

A66 2 M

A66 0.5 M

A66 0.1 M

BYL719 2 M

BYL719 0.5 M

BYL719 0.1 M

Rapamycin 0.1 M

PI3K

AKT

mTORC1

4EBP1 S6K1

Translation of TGF-2

TGF- Signalling

Col1A Col1A

Marker (kDa)

Ctrl siRNA

Ad-GFP

Ad-LOXL2

100

LOXL2

Phospho-AKT(S473)

Phospho-AKT(T308)

Total AKT

Phospho-S6K1

Total-S6K1

Phospho-4EBP1

GAPDH

P =0.52

TGF-2 level normalized to TFIIb

Number of migrated cells

(per field of the gap)

Control+IgG1

n=10

P =0.03

400

P =0.0003

300

200

100

0 IgG

IgG -LOXL2

-LOXL2

0 TGF-2+ IgG1

n=7

TGF-2+-LOXL2

n=5

n =4

n =4 n =4 n =4

Figure 6 | LOXL2 acts with TGF-b through PI3K/ATK/mTORC1 signalling to control myobroblast transformation and migration. (a) Western blot analysis of LOXL2, SMAD2, p-SMAD2 and GAPDH in human primary cardiac broblasts transfected with control (Ctrl) or LOXL2 siRNA. (b) Quantication of TGF-b isoforms in the culture media of human primary cardiac broblasts transfected with control or LOXL2 siRNA (n 4). P value: Students t-test. Error

bar: s.e.m. (c) Changes of COLA1, a-SMA and FN1 mRNA in human primary cardiac broblasts transfected with control (Ctrl) or LOXL2 siRNA. (d) Western blot of LOXL2, p-AKT, AKT, p-S6K, S6K and p-4E-BP1 with or without PI3K inhibitors LY294002/PI828 (10 mM each), PI3Ka inhibitors A66/BYL719 and mTORC1 inhibitor Rapamycine (0.1 mM) in cells infected with Ad_GFP/Ad_LOXL2. (e) TGF-b2 protein in the culture media (n 4). P value: Students t-test.

Error bar: s.e.m. (f,g) Western blot (f) and quantication (g) of TGF-b2 protein in the mice heart ventricles 6 weeks after sham/TAC operation. n 4 mice

per group. P value: Students t-test. Error bar: s.e.m. (h) A signalling cascade from LOXL2 to PI3K/AKT/mTORC1 to TGF-b2 translation. (i) Representative phase-contrast image of human primary cardiac broblasts 72 h after transfection with control (Ctrl) or LOXL2 siRNA. Scale bars, 50 mm.

(j) Immunostaining of Col1A in IgG1- or a-LOXL2-treated mice 10 weeks (10W) after sham or TAC operation. Scale bars, 100 mm. Blue: haematoxylin. Brown: Col1A. (k,l) Gap closure assay of broblast migration in control, TGF-b2, a-LOXL2 (k) groups and quantication of cells migrating into the gap (l).

Scale bars, 200 mm. n 510, P value: Students t-test. Error bar: s.e.m.

promote interstitial collagen formation. LOXL2 not only triggers myobroblast transformation to enhance collagen production (through PI3K-AKT-mTOR and TGF-b), but also augments collagen strength (by crosslinking collagen bres). LOXL2 also stimulates myobroblasts migration to large areas of the heart, where these cells produce collagen bres. Collagen bres, once released into the interstitial space, are crosslinked by LOXL2 to form bundles of collagen that are much stiffer than isolated collagen bres. Furthermore, activated broblasts secrete more LOXL2 and collagen, creating a positive feedback loop to sustain the brotic process. All these factors, in combination, trigger diffuse interstitial brosis of pathologically stressed hearts.

The LOXL2-mediated stress reaction, although capable of reinforcing tissue strength in the face of increased mechanical stress, is ultimately maladaptive. Excessive amount of collagen bres in cardiac interstitial space impede coronary vasodilation, oxygen diffusion and electromechanical coordination between cardiomyocytes, causing contractile abnormalities (systolic pump dysfunction)34. Meanwhile, the increase of collagen and its crosslinking stiffens the left ventricle, impairing ventricular

relaxation and lling (diastolic pump dysfunction)35,36. Therefore, the LOXL2-mediated interstitial reaction leads to both systolic and diastolic abnormalities, revealing a novel therapeutic avenue for HF. The diastolic aspect of LOXL2 effects is particularly important, given the recognition and increasing prevalence of diastolic dysfunction as part of HF syndrome and the lack of approved therapy for HFpEF with primarily diastolic failure.

Although LOXL2 can interact with TGF-b in models of cancer and bone remodelling14,37, the roles of LOXL2 gene in HF have not been demonstrated in mouse genetic models in vivo. By generating a new LOXL2 genetic model with broblast-specic LOXL2 knockout, we showed crucial roles of LOXL2 in cardiac broblasts for stress-induced interstitial brosis and cardiac dysfunction. What remains unanswered is how LOXL2-mediated changes of extracellular collagen composition affect TGF-b processing and signalling and guide the cellular process of myobroblast migration. Given that LOXL2 is transcriptionally activated in the stressed hearts, it will be essential to know what factors control the expression of LOXL2 gene in the hearts. Gene

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

Cardiac stress

TGF

Fibroblast transformation

Myofibroblast Collagen

Collagen deposition and cross-linking

-SMA

LOXL2

Migration

Diffuse interstitial fibrosis

Systolic dysfunction Diastolic dysfunction

Impaired oxygen diffusion vasodilation myocyte coordination

Ventricular stiffness

Figure 7 | Working model. A working model of how cardiac stress activates LOXL2 to trigger myobroblast transformation, collagen synthesis, collagen crosslinking and myobroblast migration, leading to diffuse ventricular brosis and dysfunction.

expression regulation can occur at the chromatin, transcription or posttranscriptional level, mediated respectively by epigenetic factors (chromatin-regulating factors, long noncoding RNAs), transcription factors and microRNAs. Understanding LOXL2 regulation will provide an opportunity to integrate cardiac brosis with epigenetics and RNA mechanisms of HF and to identify additional new targets for HF therapy.

Methods

Animal sample size and randomization. Littermate CD1 male mice were purchased from Charles River (Strain Code: 022). ActinCreERT (ref. 17) and Rosa-mT/mG (ref. 38) mice were purchased from Jackson Lab. iTcf21CreERT(ref. 18) was provided by Dr Michelle D. Tallquist at the University of Hawaii and Dr Eric Olson at the UT Southwestern. Loxl2/ mice were generated by the

Chang Lab. The number of animals used (n) was denoted in each test in the gures, including technical replicates when applicable. Each subgroup of experiments had n 8 to 10 biological replicates, many of which had technical replicates of three.

The mice were randomly selected from the cage and assigned to different control and experimental subgroups as described in the text. The control and experimental groups are blinded to the operators of echocardiography, catheterization and heart tissue analyses. The use of mice for studies was in compliance with the regulations of Indiana University and National Institute of Health.

Study population and analysis of human heart samples. The patients presenting at the Charit University Medicine Berlin with heart failure symptoms and reduced exercise capacity despite preserved LVEF (450%; heart failurewith preserved ejection fraction, HFpEF) who showed diastolic dysfunction according to echocardiographic analysis and European Society of Cardiology recommendations13 were enrolled in the study (n 24) (Table 2). All the patients

were evaluated by echocardiography, invasive angiography, 6-min walk test and NT-proBNP. The patients without symptoms and signs of congestive heart failure but with atypical angina or intermittent arrhythmias who underwent right ventricular (RV) biopsy for evaluation of cardiomyopathy and showed regular systolic and diastolic LV function were enrolled as controls (n 15; Table 2).

Endomyocardial biopsies were obtained from the RV septum as previously described35. Collagen volume fraction was analysed with collagen-specic Picrosirius red. Crosslinking was calculated as the ratio of insoluble to soluble collagen forms by a colorimetric approach in nine control subjects and 14 HFpEF patients39. Immunohistochemical staining was performed in heart tissues of adequate quality and quantity35. Other heart tissue samples used for LOXL2 immunostaining and for LOXL2, COL1A and COL3A mRNA quantication were derived from patients with ischaemic or idiopathic dilated cardiomyopathy and

from heart transplantation donors that did not have heart failure (Table 1). The use of human subjects or tissue samples was in compliance with the regulations of Charit University, Gilead Sciences and Indiana University.

Biomarker analysis in heart failure patients. Serum LOXL2 was measured using customized RUO LOXL2 kits (bioMrieux) on Vitek Immuno Diagnostic Assay System (VIDAS) platform. The plasma concentrations of ST-2 and tissue inhibitor of metalloproteinase-1 (TIMP-1) were determined using ELISA kits obtained from R&D Systems and NT-proBNP was measured using a Luminex kit from EMD Millipore following manufacturers instructions. The serum level of cardiac troponin I was measured using Simoa cTnI assay (Quanterix Corp).

Antibody production and purication. Hybridoma cells expressing anti-LOXL2 (AB0023) antibody were transferred to Aragen Bioscience (Gilroy, CA, USA) for the production of ascites uid in BALB/c mice. Ascites uid was then puried by batch mode on MabSelect resin and dialysed into phosphate-buffered saline (PBS) with 0.01% Tween 20. Puried AB0023 was then tested in multiple assays for release. The control IgG1 antibody directed against a non-naturally occurring protein was purchased from Antibody Solutions (Sunnyvale, CA, USA).

Transaortic constriction. Surgeries were adapted from ref. 40 and were performed on CD1 male mice of 68 weeks of age, and Loxl2 genetic male mice of 810 weeks of age and between 20 and 25 grams of weight. The mice were anaesthetized with isourane (23%, inhalation) in an induction chamber and then intubated with a 20-gauge intravenous catheter and ventilated with a mouse ventilator (Minivent, Harvard Apparatus, Inc). Anaesthesia was maintained with inhaled isourane (12%). A longitudinal 5-mm incision of the skin was made with scissors at midline of sternum. The chest cavity was opened by a small incision at the level of the second intercostal space 23 mm from the left sternal border. The chest retractor was gently inserted to spread the wound 45 mm in width. The transverse portion of the aorta was bluntly dissected with curved forceps. Then, 6-0 silk was brought underneath the transverse aorta between the left common carotid artery and the brachiocephalic trunk. One 26-gauge needle was placed directly above and parallel to the aorta. A loop was then tied around the aorta and needle, and secured with a second knot. The needle was immediately removed to create a lumen with a xed stenotic diameter. The chest cavity was closed by 6-0 silk suture. The sham-operated mice underwent similar surgical procedures, including isolation of the aorta, looping of aorta, but without tying of the suture. The pressure load caused by TAC was veried by the pressure gradient across the aortic constriction measured by echocardiography. Only mice with a peak pressure gradient430 mm Hg were analysed for cardiac hypertrophy and gene expression.

Echocardiography. The echocardiographer was blinded to the genotypes, surgical or pharmacological treatment of the mice tested. Transthoracic ultrasonography with a GE Vivid 9 ultrasound platform (GE Health Care, Milwaukee, WI, USA) and a 13 MHz transducer was used to measure aortic pressure gradient and left ventricular function. To minimize the confounding inuence of different heart rates on aortic pressure gradient and left ventricular function, the ow of isourane (inhalational) was adjusted to anaesthetize the mice while maintaining their heart rates at 450550 beats per minute. The peak aortic pressure gradient was measured by continuous wave Doppler across the aortic constriction. The left ventricular function was assessed by the M-mode scanning of the left ventricular chamber, standardized by two-dimensional, short-axis views of the left ventricle at the mid papillary muscle level. The fractional shortening (FS) of the left ventricle was dened as 100% (1 end-systolic/end-diastolic diameter), representing the

relative change of left ventricular diameters during the cardiac cycle. The mean FS of the left ventricle was determined by the average of FS measurement of the left ventricular contraction over ve beats. The P values were calculated by the Students-t test. The error bars indicate standard error of mean.

In vivo catheterization. The trachea was exposed by a midline incision from the base of the throat to just above the clavicle. The mice were intubated with a piece of polyethylene-90 tube. After the tube was secured in place by using a 6-0 silk suture, 100% oxygen was gently blown across the opening. The mice receiving ventilation were placed on a warmed (37 C) pad. The right carotid artery was then isolated. Care was taken to prevent damage to the vagal nerve. The mice were lightly anaesthetized with isourane maintaining their heart rates at 450550 beats per minute. A 1.2F Pressure-Volume Catheter (FTE-1212B-4518, Scisense, Inc) was inserted into the right carotid artery and then advanced into the left ventricle. The transducer was securely tied into place, after it was advanced to the ventricular chamber as evidenced by a change in pressure curves. The hemodynamic parameters were then recorded in close-chest mode. The parameters include left ventricular systolic pressure, EF, plPwr, stroke volume, stroke work, ESV, EDV, Tau, EDP and cardiac output.

Generation of Loxl2-oxed mice by CRISPR/Cas9 gene editing. Two sgRNAs (50sgRNA and 30sgRNA) were designed to target the Loxl2 locus (the intron

12 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

between exon 7 and 8 and the intron between exon 9 and 10; illustrated in Supplementary Fig. 3) using online software (http://tools.genome-engineering.org

Web End =http://tools.genome-engineer http://tools.genome-engineering.org

Web End =ing.org )41. T7 promoter was added to the sgRNA templates by PCR amplication, using PX330 as template with primer sets Loxl2-50LoxP-sg-F and T7-sgRNA

Common_R, and Loxl2-30LoxP-sg-F and T7-sgRNA Common_R. SgRNA sequences are underlined in the primers.

Loxl2-50LoxP-sg-F: 50-TTAATACGACTCACTATAGGGGGTCCTAGATG GCCACCCT GTTTTAGAGCTAGAAATAGCAAGTT-30

Loxl2-30LoxP-sg-F: 50-TTAATACGACTCACTATAGGCTCGGGGTGGGG AAGCGCCG GTTTTAGAGCTAGAAATAGCAAGTT-30

T7-sgRNA Common_R: 50-AAAAGCACCGACTCGGTGCC-30The T7-sgRNA PCR product was gel puried and used as the template for in vitro transcription using MEGAshortscript T7 kit (Life Technologies).

We rst tested the cleavage efciency of sgRNAs at the genomic target site. Capped polyadenylated Cas9 mRNA (Sigma-Aldrich; 100 ng ml 1) was co-injected with sgRNAs (50 ng ml 1) into pronuclear (PN) stage one-cell mouse embryos and assessed the frequency of altered alleles (insertions and deletions) at the blastocyst stage using PCR assay with following primer sets.

Loxl2-50sg-test F: 50-ACCTATGAGTGAATGACCCCTG-30Loxl2 50sg-test R: 50-GGTCCAAGAAAGGTATAACGG-30Loxl2 30sg-test F: 50-GTTTCACAGTGTGTGCTCGG-30Loxl2 30sg-test R: 50-ACTAATACACACTCACTCCC-30Both sgRNAs achieved over 50% indel at the genomic target site. Plasmid-based donor repair templates that contain homology arms (980 bp length on each side) and 50 and 30 LoxP sequences at denoted sites (Supplementary Fig. 3) were cloned to the TOPO-Blunt vector (Thermo Fisher) and veried by sequencing. Cas9 mRNA (100 ng ml 1), 50-sgRNA (50 ng ml 1), 30-sgRNA (50 ng ml 1) and donor circular dsDNA (100 ng ml 1) were mixed and injected into the cytoplasm of fertilized eggs with well-recognized pronuclei in the M2 media (Sigma-Aldrich).

Injected zygotes were cultured in KSOM (media for in vitro embryo culture, MR-121-D, EMD Millipore) with amino acids at 37C and 5% CO2 after the two-cell stage and were then transplanted to pseudopregnant CD1 females. Three out of 23 mice were genotyped positive with 50 and 30 LoxP insertion using the upstream PCR primer set 1 (product size 1,069 bp) and the downstream primer set 2 (product size 1,082 bp).

Primer set 1 (upstream)50 Loxl2-G F: 50-CCTTCTGGAACTGCCAGAAGTATTTTAAG-30 (upstream of template DNA sequence)

50 Loxl2-G R: 50-CGTATAATGTATGCTATACGAAGTTATTTGGATTCAG AACTTCC-30

Primer set 2 (downstream)30 Loxl2-G F: 50-GTATAGCATACATTATACGAAGTTATCCGTGGATTCT

GGG-30

30 Loxl2-G R: 50-CAGAGAGAAGATCCCCATCAG-30 (downstream of template DNA sequence).

These three founder mice were grossly normal and heterozygous as determined by additional PCR primer setsLoxl2-50sg-test F and R (primer set 3), as well as Loxl2-30sg-test F and R (primer set 4; sequences shown above; Supplementary Fig. 3). Founder mice were then crossed to C57BL6 mice. Genotyping of the offspring mice using PCR primers described above showed that two of the founder mice carried the template transgene ectopically in the genome in addition to the LoxP target insertions (primer set 3, 4 positive,but negative for primer set 1, 2). To genetically remove the ectopic insertionsof the template transgenes, we back-crossed the founder mice C57BL6 mice for three to ve generations. This process also helped wash out potential off-targets of sgRNA (off-target effects were rare in mice generated by microinjectionusing CRISPR/Cas9 into mouse zygotes42,43.) According to genotyping and Mendelian ratio, one male heterozygous Loxl2/ was selected to cross with

ActinCreERT and iTcf21CreERT female mice for knockout studies. Throughoutthe observation period of current studies, all homozygous, heterozygous and wild-type offspring mice were grossly normal and non-distinguishable from their littermates.

Western blot analysis. The blots were reacted with antibodies of anti-LOXL2 (1:500, ab96233, Abcam), anti-TGF-b2 (1:1,000, MAB73461, R&D Systems), anti-GAPDH (1:20,000, G9545, Sigma-Aldrich), anti-TFIIb (1:1,000, ab109106, Abcam), anti- phosphor-SMAD2 (1:1,000, 138D4, Cell Signaling), anti-SMAD2 (1:1,000, D43B4, Cell Signaling), anti-GAPDH (1:1,000, 2118, Cell Signaling), aSMA (1:1,000, A2547, Sigma-Aldrich), total AKT (1:1,000, 4685, Cell Signaling), phospho-AKT (Ser473) (1:1,000, 4060, Cell Signaling), phospho-AKT (Thr308; 1:1,000, 4056, Cell Signaling), phospho-S6K1 (1:1,000, 9234, Cell Signaling), total S6K1(1:1,000, 9202, Cell Signaling), phospho-4EBP1 (1:1,000, 2855, Cell Signaling), PI3Ka (1:1,000, C73FB, Cell Signaling), PI3Kb (1:1,000, C33D4, Cell Signaling)

and PI3Kd (1:1,000, D55D5, Cell Signaling). Then followed by HRP-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Chemiluminescence was detected with ECL Western blot detection kits (GE) with LI-COR odyssey image system. Uncropped blot images are shown in Supplementary Fig. 7.

RTqPCR. RTqPCR reactions were performed using SYBR green master mix (Bio-Rad, Hercules, CA, USA) with an Eppendorf realplex, and the primer sets were tested to be quantitative. Threshold cycles and melting curve measurements were performed with software. The P values were calculated by the Students t-test. Error bars indicate standard error of mean.

Histology and trichrome staining. Staining for LOXL2 and COL1A was performed on frozen 5 mm-thick sections of human endomyocardial biopsies. Rabbit anti-LOXL2 (1:250, Gilead Sciences) and anti-COL1A antibody (1:75,BT 21-5000-20, Biotrend Chemikalien, Germany) were used as primary antibodies, followed by secondary EnVision anti-rabbit antibody (Dako, Germany) according to manufacturers instructions. Visualization was performed with carbazol-staining solution and background staining was performed with a Mayers hemalum solution (Sigma-Aldrich, Germany). Digital image analysis was conducted on a Leica DMRB microscope (Leica Microsystems, Germany) at a 200 magnication.

As described previously44, adult mice hearts were xed overnight in 4% paraformaldehyde in phosphate-buffered saline (PBS). Subsequently, they were dehydrated through an ethanol series, treated with xylenes and embedded in parafn wax overnight with several changes. The hearts were oriented for transverse sections and cut in 7 mm sections using a Leica microtome. Trichrome staining was performed following the protocol of trichrome stain (Masson) kit (HT15) from Sigma-Aldrich. Deparafnize the slides to deionized water, then mordant in preheated Bouins Solution (HT10-1) at 56 C for 15 min. Cool the slides in tap water (1826 C) and wash in running tap water to remove yellow colour from sections. Put the slides in Working Weigerts Iron Haematoxylin Solution for 5 min, Biebrich Scarlet-Acid Fucshin (HT151) 5 min, working Phosphotungstic/Phosphomolybdic Acid Solution for 5 min, Aniline Blue Solution (HT154) for 5 min, 1% acetic acid for 2 min. Finally, rinse slides, dehydrate through alcohol, clear in xylene and mount. The imaging was performed using Leica microscope. The following primary antibodies were used for mice tissue immunostaining: anti-LOXL2 (1:100, Rabbit Polyclonal antibody, Gilead Sciences), anti-CD4 (1:100, sc-13573, Santa Cruz), anti-COL1A (Chemicon, MA, USA) and aSMA (1:200, A2547, Sigma-Aldrich). Immunohistochemistry was conducted using biotinylated secondary antibodies (anti-rabbit IgG, 1:250; anti-mice IgG, 1:250; Jackson Immunoresearch), VECTASTAIN Elite ABC Kit (PK-6200, Vector Laboratories) and DAB developing reagents (DAKO). Imaging was performed using a Leica microscope.

Collagen content measurements. To determine total cardiac collagen content, we measured the content of hydroxyproline, a major component of collagen,as described39. Left ventricles were hydrolysed in hydrochloric acid HCl, and hydroxyproline level was measured by a colorimetric method using assay kits from Qickzyme. The total collagen content was calculated from the hydroxyproline content of collagen standards. To determine soluble collagen content, left ventricles were solubilized in pepsin-acid solution, and the supernatants were precipitated by trichloroacetic acid solution. Dried pellets were then hydrolysed in HCl, and hydroxyproline measured. The soluble collagen content was calculated from the hydroxyproline content of collagen standards. The insoluble collagen content was determined by subtracting the amount of soluble collagen from total collagen.

LOXL2 siRNA knockdown in human primary cardiac broblasts. Human primary cardiac broblasts were isolated from the ventricles (Lonza, Walkersville Inc., Walkersville, MD, USA; CC-2904). These cells were cultured in FGM-2 BulletKits media (Lonza, CC-4526) at 37 C and 5% CO2. Human LOXL2 and control siRNA (Qiagen, SI00036134 and SI03650318) were transfected into human broblasts with lipfectmain RNAiMAX. After 48 h, the cells were collected, culture media collected, cells lysed and total RNA extracted for further analyses. The levels of mRNA were measured by RTqPCR using primers for individual genes or TGF-b/BMP signalling arrays (Qiagen). The siRNA for LOXL2 was 50-CCGGAG

TTGCCTGCTCAGAAA-30. Primers for human LOXL2 (QT00019425), TGF-b1 (QT00000728), TGF-b2 (QT00025718) and TGF-b3 (QT00001302) were from Qiagen. The levels of TGF-b in the culture media were measured with Milliplex Map TGF-b 3-Plex (EMD Millipore).

LOXL2 overexpression in human primary cardiac broblasts. Adenoviral vectors expressing human LOXL2 or GFP were purchased from Applied Biological Materials. Human cardiac broblasts were transduced with adenovirus for 6 h, and the cells were cultured in FGM-2 medium with 2% FBS for 48 h and culture medium, cell lysates and total RNA were isolated for measuring mRNA and TGF-b levels.

Gap closure assay. A gap closure assay was used to study the migration of human ventricular broblasts (Lonza Walkersville Inc., Walkersville, MD, USA). The cells were seeded into collagen I-coated 24-well plates (Life Technologies Corporation, Grand Island, NY, USA) at the density of 105 cm 2 cells. The culture was grown to high conuency, and then the cells were serum-starved for 24 h before the treatment. The cell monolayer was scratched with a sterile pipette tip to remove cells and create a gap. The average width of the scratch was 87638 mm.

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710

The culture medium was replaced with fresh media immediately after scratching. The anti-LOXL2 antibody (AB0023) was added at a concentration of 10 mg ml 1, and the cell migration was activated with TGF-b2 (5 ng ml 1). A mouse IgG1 antibody was used as an isotype control. The broblasts were allowed to migrate into the gap for approximately 16 h. After that, the cells were loaded with uorescent dye calcein AM (0.2 mM, Life Technologies Corporation) at 37 C for 30 min. Once the dye loading was completed, uorescence images were acquired using Zeiss LSM 5 Pascal confocal microscope equipped with 4 objective.

Calcein uorescence was excited with an Argon laser at 488 nm and detected at wavelengths 4505 nm using a long pass lter. To count the migrated cells, the images were analysed using Image J (NIH) software.

Statistical analysis. Comparisons were performed with the MannWhitney U-test used for non-parametric continuous variables or the Students t-test for parametric variables and Fishers exact test for categorical variables. Correlation of parameters was performed with Pearson or Spearman coefcients, and regression analysis was used for exact relations. Statistical signicances were calculated using GraphPad Prism 5 and 7 (GraphPad Software, La Jolla, CA, USA). A value of Po0.05 was considered statistically signicant.

Data availability. The authors declare that the data supporting the ndings of this study are available within the article and its Supplementary Information les and from the corresponding author on request.

References

1. Go, A. S. et al. Heart disease and stroke statistics--2014 update: a report from the American Heart Association. Circulation 129, e28e292 (2014).

2. Schelbert, E. B., Fonarow, G. C., Bonow, R. O., Butler, J. & Gheorghiade, M. Therapeutic targets in heart failure: refocusing on the myocardial interstitium.J. Am. Coll. Cardiol. 63, 21882198 (2014).3. Butler, J. et al. Developing therapies for heart failure with preserved ejection fraction: current state and future directions. JACC Heart Fail. 2, 97112 (2014).

4. Davis, J. & Molkentin, J. D. Myobroblasts: trust your heart and let fate decide.J. Mol. Cell. Cardiol. 70, 918 (2014).5. Kong, P., Christia, P. & Frangogiannis, N. G. The pathogenesis of cardiac brosis. Cell. Mol. Life Sci. 71, 549574 (2014).

6. Schelbert, E. B. et al. Myocardial brosis quantied by extracellular volume is associated with subsequent hospitalization for heart failure, death, or both across the spectrum of ejection fraction and heart failure stage. J. Am. Heart Assoc. 4, e002613 (2015).

7. Hang, C. T. et al. Chromatin regulation by Brg1 underlies heart muscle development and disease. Nature 466, 6267 (2010).

8. Han, P. et al. A long noncoding RNA protects the heart from pathological hypertrophy. Nature 514, 102106 (2014).

9. Frantz, S. et al. Tissue inhibitor of metalloproteinases levels in patients with chronic heart failure: an independent predictor of mortality. Eur. J. Heart Fail. 10, 388395 (2008).

10. Sanada, S. et al. IL-33 and ST2 comprise a critical biomechanically induced and cardioprotective signaling system. J. Clin. Invest. 117, 15381549 (2007).

11. Gandhi, P. U., Testani, J. M. & Ahmad, T. The current and potential clinical relevance of heart failure biomarkers. Curr. Heart Fail. Rep. 12, 318327 (2015).

12. Pascual-Figal, D. A. & Januzzi, J. L. The biology of ST2: the International ST2 Consensus Panel. Am. J. Cardiol. 115, 3B7B (2015).

13. Paulus, W. J. et al. How to diagnose diastolic heart failure: a consensus statement on the diagnosis of heart failure with normal left ventricular ejection fraction by the Heart Failure and Echocardiography Associations of the European Society of Cardiology. Eur. Heart J. 28, 25392550 (2007).

14. Barry-Hamilton, V. et al. Allosteric inhibition of lysyl oxidase-like-2 impedes the development of a pathologic microenvironment. Nat. Med. 16, 10091017 (2010).

15. Creemers, E. E. & Pinto, Y. M. Molecular mechanisms that control interstitial brosis in the pressure-overloaded heart. Cardiovasc. Res. 89, 265272 (2011).

16. Rodriguez, H. M. et al. Modulation of lysyl oxidase-like 2 enzymatic activity by an allosteric antibody inhibitor. J. Biol. Chem. 285, 2096420974 (2010).

17. Hayashi, S. & McMahon, A. P. Efcient recombination in diverse tissues by a tamoxifen-inducible form of Cre: a tool for temporally regulated gene activation/inactivation in the mouse. Dev. Biol. 244, 305318 (2002).

18. Acharya, A., Baek, S. T., Ban, S., Eskiocak, B. & Tallquist, M. D. Efcient inducible Cre-mediated recombination in Tcf21 cell lineages in the heart and kidney. Genesis 49, 870877 (2011).

19. Sun, Y. & Weber, K. T. Infarct scar: a dynamic tissue. Cardiovasc. Res. 46,

250256 (2000).

20. Willems, I. E., Havenith, M. G., De Mey, J. G. & Daemen, M. J. The alpha-smooth muscle actin-positive cells in healing human myocardial scars. Am. J. Pathol. 145, 868875 (1994).

21. Leask, A. TGFbeta, cardiac broblasts, and the brotic response. Cardiovasc. Res. 74, 207212 (2007).

22. Hinz, B. Formation and function of the myobroblast during tissue repair.J. Invest. Dermatol. 127, 526537 (2007).23. Hinz, B. et al. The myobroblast: one function, multiple origins. Am. J. Pathol. 170, 18071816 (2007).

24. Hinz, B. et al. Recent developments in myobroblast biology: paradigms for connective tissue remodeling. Am. J. Pathol. 180, 13401355 (2012).

25. Bujak, M. & Frangogiannis, N. G. The role of TGF-beta signaling in myocardial infarction and cardiac remodeling. Cardiovasc. Res. 74, 184195 (2007).

26. Davis, J., Burr, A. R., Davis, G. F., Birnbaumer, L. & Molkentin, J. D.

A TRPC6-dependent pathway for myobroblast transdifferentiation and wound healing in vivo. Dev. Cell 23, 705715 (2012).

27. Hinz, B., Celetta, G., Tomasek, J. J., Gabbiani, G. & Chaponnier, C. Alpha-smooth muscle actin expression upregulates broblast contractile activity. Mol. Biol. Cell 12, 27302741 (2001).

28. Zhang, Y. E. Non-Smad pathways in TGF-beta signaling. Cell Res. 19, 128139 (2009).

29. Aoyagi, T. & Matsui, T. Phosphoinositide-3 kinase signaling in cardiac hypertrophy and heart failure. Curr. Pharm. Des. 17, 18181824 (2011).

30. Evans, R. A., Tian, Y. C., Steadman, R. & Phillips, A. O. TGF-beta1-mediated broblast-myobroblast terminal differentiation-the role of Smad proteins. Exp. Cell Res. 282, 90100 (2003).

31. Dobaczewski, M. et al. Smad3 signaling critically regulates broblast phenotype and function in healing myocardial infarction. Circ. Res. 107, 418428 (2010).

32. Verrecchia, F., Chu, M. L. & Mauviel, A. Identication of novel TGF-beta / Smad gene targets in dermal broblasts using a combined cDNA microarray/ promoter transactivation approach. J. Biol. Chem. 276, 1705817062 (2001).

33. Hu, B. et al. Gut-enriched Kruppel-like factor interaction with Smad3 inhibits myobroblast differentiation. Am. J. Resp. Cell Mol. Biol. 36, 7884 (2007).

34. Rohr, S. Myobroblasts in diseased hearts: new players in cardiac arrhythmias? Heart Rhythm 6, 848856 (2009).

35. Kasner, M. et al. Diastolic tissue Doppler indexes correlate with the degree of collagen expression and cross-linking in heart failure and normal ejection fraction. J. Am. Coll. Cardiol. 57, 977985 (2011).

36. Lopez, B., Querejeta, R., Gonzalez, A., Larman, M. & Diez, J. Collagen cross-linking but not collagen amount associates with elevated lling pressures in hypertensive patients with stage C heart failure: potential role of lysyl oxidase. Hypertension 60, 677683 (2012).

37. Atsawasuwan, P. et al. Lysyl oxidase binds transforming growth factor-beta and regulates its signaling via amine oxidase activity. J. Biol. Chem. 283, 3422934240 (2008).

38. Muzumdar, M. D., Tasic, B., Miyamichi, K., Li, L. & Luo, L. A global double-uorescent Cre reporter mouse. Genesis 45, 593605 (2007).

39. Lopez, B. et al. Impact of treatment on myocardial lysyl oxidase expression and collagen cross-linking in patients with heart failure. Hypertension 53, 236242 (2009).

40. Trivedi, C. M. et al. Hdac2 regulates the cardiac hypertrophic response by modulating Gsk3 beta activity. Nat. Med. 13, 324331 (2007).

41. Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. 8, 22812308 (2013).

42. Yen, S. T. et al. Somatic mosaicism and allele complexity induced by CRISPR/Cas9 RNA injections in mouse zygotes. Dev. Biol. 393, 39 (2014).

43. Mashiko, D. et al. Generation of mutant mice by pronuclear injection of circular plasmid expressing Cas9 and single guided RNA. Sci. Rep. 3, 3355 (2013).

44. Stankunas, K. et al. Endocardial Brg1 represses ADAMTS1 to maintain the microenvironment for myocardial morphogenesis. Dev. Cell 14, 298311 (2008).

Acknowledgements

We thank Drs William Lee, Tyler White, J. Shryock and Roman Sakowicz for advice and manuscript reading; Drs Michelle D. Tallquist and Eric Olson for kindly providing iTcf21CreERT mice; Juan Xu for technical support. C.-P.C. was supported by National Institutes of Health (NIH, HL118087, HL121197), American Heart Association(AHA, Established Investigator Award 12EIA8960018), March of Dimes Foundation (#6-FY11-260), California Institute of Regenerative Medicine (CIRM, RN2-00909), Indiana University (IU) School of MedicineIU Health Strategic Research Initiative,IU Physician-Scientist Initiative, endowed by Lilly Endowment, Inc., Charles Fisch Endowed Chair at Krannert Institute of Cardiology (KIC), and Funded Research Program by Gilead Sciences, Inc. J.Y. was supported by Charles Fisch Cardiovascular Research Award endowed by Dr Suzanne B. Knoebel of KIC. P.-S.C. was supported by National Institutes of Health (R42DA043391, RO1HL71140, PO1HL78931). Part of blood samples used in this study were obtained from the Indiana Biobank, which receives government support under a cooperative agreement grant (UL1TR001108) awarded by the National Center for Advancing Translational Research (NCATS) and the Lilly Endowment. We thank contributors who collected samples used in this study, as well as subjects whose help and participation made this work possible.

Author contributions

J.Y., K.S. and J.S.K. contributed equally; L.Y. and C.T. had equal contributions to the work. C.-P.C. and J.Y. were primarily responsible for initial manuscript preparation and

14 NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13710 ARTICLE

responses to reviewers critiques. C.-P.C., J.Y. and L.Y. designed animal and cell-based experiments and prepared the manuscript with contributions from J.S.K., P.F., H.Z., K.S., A.M.-V., J.A., D.K., C.T., J.Y., X.F. and Q.Z., K.S. and C.T. were responsible for the studies of HFpEF patients; C.-P.C., L.Y. and H.Z. for biomarker studies of HFrEF and HFpEF. More specically, J.Y. and C.-P.C. conducted animal experiments, including TAC, echocardiography, in vivo catheterization with PV loop analyses, tissue harvest, histopathology and gene expression analyses with assistance from C.S., J.Y. and C.-P.C. generated oxed alleles of Loxl2 and performed tissue-specic knockout of Loxl2; K.S. and C.T. were responsible for human tissue collection and analyses of patients with HFpEF; M.K. was involved in patient recruitment and echocardiographic studies; D.P. was responsible for human heart tissue collection; K.C.W., R.L.K., P.-S.C. and C.-P.C. were responsible for the blood and tissue sample collections from patients with HF; H.Z., L.Y. and C.-P.C. performed the HF biomarker studies; J.S.K., P.F., B.L. andJ.D. performed the collagen crosslinking analysis; J.S.K., P.F., K.S., H.Z. and V.S. performed the cell-based experiments; S.K. and D.K. performed the gap closure assay; P.K. performed the brosis quantication; V.B. and J.A. performed the immunostaining; A.M.-V. and V.S. were responsible for a-LOXL2 antibody production and experiment design; L.Y., V.S., T.Q. and C.T. contributed to data analysis.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: C.-P.C. consults for Gilead Sciences, Inc. The remaining authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Yang, J. et al. Targeting LOXL2 for cardiac interstitial brosis and heart failure treatment. Nat. Commun. 7, 13710 doi: 10.1038/ncomms13710 (2016).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2016

NATURE COMMUNICATIONS | 7:13710 | DOI: 10.1038/ncomms13710 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 15

Word count: 13895

Show less

Abstract

Translate

Interstitial fibrosis plays a key role in the development and progression of heart failure. Here, we show that an enzyme that crosslinks collagen--Lysyl oxidase-like 2 (Loxl2)--is essential for interstitial fibrosis and mechanical dysfunction of pathologically stressed hearts. In mice, cardiac stress activates fibroblasts to express and secrete Loxl2 into the interstitium, triggering fibrosis, systolic and diastolic dysfunction of stressed hearts. Antibody-mediated inhibition or genetic disruption of Loxl2 greatly reduces stress-induced cardiac fibrosis and chamber dilatation, improving systolic and diastolic functions. Loxl2 stimulates cardiac fibroblasts through PI3K/AKT to produce TGF-β2, promoting fibroblast-to-myofibroblast transformation; Loxl2 also acts downstream of TGF-β2 to stimulate myofibroblast migration. In diseased human hearts, LOXL2 is upregulated in cardiac interstitium; its levels correlate with collagen crosslinking and cardiac dysfunction. LOXL2 is also elevated in the serum of heart failure (HF) patients, correlating with other HF biomarkers, suggesting a conserved LOXL2-mediated mechanism of human HF.

Details

Title

Targeting LOXL2 for cardiac interstitial fibrosis and heart failure treatment

Author

Yang, Jin; Savvatis, Konstantinos; Kang, Jong Seok; Fan, Peidong; Zhong, Hongyan; Schwartz, Karen; Barry, Vivian; Mikels-vigdal, Amanda; Karpinski, Serge; Kornyeyev, Dmytro; Adamkewicz, Joanne; Feng, Xuhui; Zhou, Qiong; Shang, Ching; Kumar, Praveen; Phan, Dillon; Kasner, Mario; López, Begoña; Diez, Javier; Wright, Keith C; Kovacs, Roxanne L; Chen, Peng-sheng; Quertermous, Thomas; Smith, Victoria; Yao, Lina; Tschöpe, Carsten; Chang, Ching-pin

Pages

13710

Publication year

2016

Publication date

Dec 2016

Publisher

Nature Publishing Group

e-ISSN

20411723

Source type

Scholarly Journal

Language of publication

English

DOI

https://doi.org/10.1038/ncomms13710

ProQuest document ID

1848507170

Targeting LOXL2 for cardiac interstitial fibrosis and heart failure treatment

Jump to:

Full text

Abstract

Details

Suggested sources