The extracellular matrix glycoprotein ADAMTSL2 is increased in heart failure and inhibits TGFβ signalling in cardiac fibroblasts

Fibrosis accompanies most heart diseases and is associated with adverse patient outcomes. Transforming growth factor (TGF)β drives extracellular matrix remodelling and fibrosis in the failing heart. Some members of the ADAMTSL (a disintegrin-like and metalloproteinase domain with thrombospondin type 1 motifs-like) family of secreted glycoproteins bind to matrix microfibrils, and although their function in the heart remains largely unknown, they are suggested to regulate TGFβ activity. The aims of this study were to determine ADAMTSL2 levels in failing hearts, and to elucidate the role of ADAMTSL2 in fibrosis using cultured human cardiac fibroblasts (CFBs). Cardiac ADAMTSL2 mRNA was robustly increased in human and experimental heart failure, and mainly expressed by fibroblasts. Over-expression and treatment with extracellular ADAMTSL2 in human CFBs led to reduced TGFβ production and signalling. Increased ADAMTSL2 attenuated myofibroblast differentiation, with reduced expression of the signature molecules α-smooth muscle actin and osteopontin. Finally, ADAMTSL2 mitigated the pro-fibrotic CFB phenotypes, proliferation, migration and contractility. In conclusion, the extracellular matrix-localized glycoprotein ADAMTSL2 was upregulated in fibrotic and failing hearts of patients and mice. We identified ADAMTSL2 as a negative regulator of TGFβ in human cardiac fibroblasts, inhibiting myofibroblast differentiation and pro-fibrotic properties.


Supplementary Figures
Purity of primary heart cell cultures prepared from neonatal rats. Primary heart cell cultures from neonatal rats were prepared to separate cardiomyocytes (rCM) and cardiac fibroblasts (rCFB) according to a standard protocol 1 . (a) The purity of the cultures was confirmed by gene expression analysis of the cardiomyocyte-specific marker troponin-I (Tnni3). (b) Presence of endothelial cells was determined by expression of von willebrand factor (Vwf). (c) Fibroblast cultures were confirmed by expression of type I collagen (Col1a2), periostin (Postn), and α-smooth muscle actin (Acta2) (myofibroblasts). Gene expression was normalized to Rpl4. Data represent n=3 isolations with n=1-3 technical replicates from each isolation. Data are presented as minimum, mean and maximum values, and statistical analysis was performed using the Student t-test. Human foetal cardiac fibroblasts (hfCFBs) were cultured for two days in growth medium and serum starved for one day before harvest. Data represent experiments from ≥3 different cell passages. (a) CT values from qPCR of ADAMTSL2 mRNA with 10 ng and 60 ng input, with RPL4 expression as reference. The ADAMTSL2 signal is detected at CT values >30 even with maximum input, in line with negligible expression in untreated hfCFBs. (b) Immunoblot of ADAMTSL2 protein in cell lysates (16 g loaded) and (c) cell medium (maximum loading 68 g), n=9, showing that ADAMTSL2 protein was not detected in, or secreted from, hfCFBs. ADAMTSL2 over-expressing cells (L2) were used as positive control for ADAMTSL2 immunoblot signal. (d) RNA sequencing data from the Genotype-Tissue Expression (GTEx) project portal. In GTEx, a total of 17382 samples are collected from 54 different non-diseased tissues considered normal relative to age, across 948 deceased human organ donors, and thus, a control population. Donor age is 20-70 years, 33% female, 85% White and 13% African American. ADAMTSL2 expression (dbGaP Acession phs000424.v8.p2) shown as transcripts per kilobase million (TPM), in cardiac tissue from the left ventricle (LV, n=432, median TPM=3.55), atrial appendage (AA, n=429, median TPM=54.51) and cultured dermal fibroblasts (FBs, n=504, median TPM=0.05).

Fig. S3
Culture protocols used for human cardiac fibroblasts, creating cultures with an immature or mature extracellular matrix. (a) Schematic overview of the human foetal cardiac fibroblasts culture protocol. Cells were cultured for four or seven days, forming an immature or mature extracellular matrix (ECM), and transduced with ADAMTSL2 or control virus on day one or four, respectively. Data represent experiments from three different cell passages with the respective culture conditions. (b) mRNA levels of collagen I (COL1A2), collagen III (COL3A1), lysyl oxidase (LOX), fibronectin (FN1), fibrillin-1, -2 (FBN1,2) and elastin (ELN) in non-treated mature vs. immature ECM cultures (n=12-15). (c) mRNA levels of TGFB1-3, LTBP1, 3 and 4, in nontreated mature vs immature ECM cultures (n=12-15). (d) Representative immunocytochemistry images from three different cell passages of non-transduced mature vs immature ECM cultures, stained for EDA-fibronectin, latent TGFβ binding protein 1 (LTBP1), fibrillin-1, collagen type I and α-smooth muscle actin (α-SMA). Scale bar = 20µM. (e) Representative immunoblot of LTBP1 in mature vs immature ECM cultures, showing LTBP1 and the large latent complex (LLC) consisting of LTBP1 and TGFβ, and quantification of intensity from three independent experiments (n=9). (f) mRNA levels of proliferating cell nuclear antigen (PCNA) in non-treated mature vs immature ECM cultures (n=12-15). (g) mRNA levels of α-SMA (encoded by ACTA2) and osteopontin (SPP1), and representative immunoblots of α-SMA in mature vs immature ECM cultures (n=12-18). (h) Representative immunoblots of phosphorylated Smad2 (pSMAD2) and total Smad2/3, and mRNA levels of connective tissue growth factor (CTGF) and periostin (POSTN), in mature vs immature ECM cultures (n=12-18). All full-size, uncropped blots are available in Supplementary figure VI. Gene expression was normalized to RPL4. GAPDH was used for protein loading control. Data are presented as mean ± min/max. Statistical differences were tested using the Student t-test. Human foetal cardiac fibroblasts were cultured for four or seven days (see Supplementary  Fig. S3a), forming an immature or mature extracellular matrix (ECM), and transduced with ADAMTSL2 (L2) or control (vehicle, Veh) adenoviruses on day one or four, respectively. Data represent experiments from three different cell passages. mRNA levels of (a) latent TGFβ-binding protein (LTBP1), (b) LTBP3 and (c) LTBP4 in L2 compared to Veh (n=12-15). Gene expression was normalized RPL4. Data are mean ± min/max. Statistical differences were tested using the Student t-test. Human foetal cardiac fibroblasts were cultured for four or seven days (see Supplementary  Fig. S3a), forming an immature or mature extracellular matrix (ECM), and transduced with ADAMTSL2 (L2) or control (vehicle, Veh) adenoviruses on day one or four, respectively. Samples were pooled from three experiments in different cell passages, each with 3-6 technical replicates (n=15). Gene expression analysis was performed using a pre-designed microarray plate containing probes for 84 extracellular matrix and adhesion-related genes and 12 housekeeping genes. The table shows differentially expressed genes in L2 vs. Veh with colour coding according to fold-change (green, upregulated; red, downregulated; white, unchanged). Gene expression was normalized to GAPDH using the 2 -ΔΔCT algorithm. Immunoblot for ADAMTSL2 in left ventricular biopsies from dilated cardiomyopathy (DCM, n=2) and hypertrophic obstructive cardiomyopathy (HOCM, n=2) patients showing a band of the expected size (approx. 140 kDa 2 ), representing the full-length, glycosylated ADAMTSL2 protein (gL2). Treatment with PNGase F resulted in de-glycosylated ADAMTSL2 (cL2). Foetal human cardiac fibroblasts over-expressing ADAMTSL2 (L2) or control virus (Veh) were used as positive and negative control, respectively.   RNA sequencing data from the Genotype-Tissue Expression (GTEx) project portal. In GTEx, a total of 17382 samples are collected from 54 different non-diseased tissues considered normal relative to age, across 948 deceased human organ donors, and thus, a control population. Donor age is 20-70 years, 33% female, 85% White and 13% African American.

Human heart tissue samples
Left IVS biopsies from HOCM patients (n=15) referred for septal reduction therapy, were obtained during septal myectomy, and patient characteristics were described previously 4 . In brief, the diagnostic criteria were LVPWd >1.5 cm with other causes of hypertrophy excluded, non-dilated LV, and EF>50%.
LV biopsies from DCM patients (n=20) were obtained from beating hearts during heart transplantation. LV tissue from non-diseased hearts considered for transplantation, but deemed unsuitable due to surgical reasons, served as controls for HOCM and DCM (n=3 for RNA samples and n=8 for protein samples). DCM patient and donor characteristics were described previously 5 . In brief, hearts were dilated (LVIDd 7.41 ± 0.22 cm), had reduced systolic function (LVEF of 19.2 ± 1.6%) and walls were not hypertrophic (IVSd 0.81 ± 0.05 cm and LVPWd 0.71 ± 0.02 cm, respectively). Tissue samples were snap-frozen in liquid nitrogen and stored at -70˚C until molecular analysis.

Mouse pressure overload heart failure model
Mouse heart samples used for this study were derived from a previously published cohort 6

Cultures of neonatal rat cardiac myocytes and fibroblasts
Primary cultures were prepared as previously described 1 . In brief, 1-3 day old, neonatal rats  Unless otherwise stated, cells were plated at 20,000 cells/cm 2 and cultured for four days (generating an immature, developing ECM) or plated at 10,000 cells/cm 2 and cultured for seven days (generating a mature ECM) before harvesting (see Supplementary Fig. S3a).

Human cardiac fibroblast cultures
hfCFBs and haCFBs were transduced using replication-deficient human adenovirus Non-transduced hfCFBs were seeded onto 6-well plates in serum-containing medium.
After 24 h, the medium was replaced with conditioned medium containing ADAMTSL2 protein, harvested from hfCFBs transduced with L2 (4-day protocol), or Veh. Conditioned medium was prepared by diluting harvested medium from cells 1:1 with serum-free medium.
After another 24 h, the medium was replaced with fresh conditioned medium again, and finally, cells were harvested 72 h after seeding.
Non-transduced hfCFBs were seeded onto 6-well plates in serum-containing medium.
For experiments with conditioned medium and TGFβ stimulation, non-transduced hfCFBs were seeded onto 6-well plates in serum-containing medium. After 24 h, the cells were washed with PBS and the medium was replaced with fresh serum-free medium. After 3 h, L2 or Veh conditioned medium, pre-incubated for 1 h with recombinant TGFβ1 (10 µg/µL), was added to the cells. Cells were harvested at T0, T30 and T60 minutes after stimulation. BioMatrix). A cell-collagen gel mixture was made to a final concentration of 36.5 x 10 3 cells/cm 2 , and added to the BSA-coated plates. The gels were allowed to polymerize for 2 h at 37°C, before serum-free medium was added to release the collagen gel from the plastic surface. Contraction was observed over the next 24 h, and images were taken at six and 24

Immunocytochemistry
hours. The circumferences of the collagen gels were measured using ImageJ (NIH), and percent contraction from baseline was calculated.

Gene expression analysis
Total RNA was isolated from LVs or cell cultures using the RNeasy Mini Kit (Cat# 74106, Qiagen Nordic, Oslo, Norway) according to the manufacturer's protocol. Reverse transcription and cDNA generation was performed using iScript cDNA Synthesis Kit (Cat# 1708891, Bio-Rad Laboratories, Inc., Hercules, CA). Gene expression was determined using TaqMan Gene Expression Assays (Table S1)

Protein isolation and immunoblotting
Unless otherwise stated, all reagents were purchased from Merck KGaA. Proteins were extracted from mouse and human LVs as previously described 1,6 , using a PBS-based lysis buffer containing 1% Triton X-100, 0.1% Tween-20, protease inhibitors (cOmplete EDTAfree tablets) and PhosStop phosphatase inhibitors (both from Roche Diagnostics).
Whole-cell protein lysates from cell cultures were extracted using the buffer above, or with lysis buffer containing 1% SDS, 31.5 mM Tris-HCl, pH 6.       Fig. 5a, immunoblotting for anti-ADAMTSL2. The gel was loaded with conditioned medium from cultured human foetal cardiac fibroblasts (hfCFBs)over-expressing ADAMTSL2 (L2-medium) or vehicle control (Veh-medium). (B) Full size images of the blot used to make Fig. 5c, immunoblotting for anti-pSMAD2 (top), anti-SMAD2/3 (middle), and anti-α-smooth muscle actin (α-SMA, 42 kDa) and anti-GAPDH (36 kDa) as loading control (bottom). The gel was loaded with lysates from cultured hfCFBs treated with L2-medium or Veh-medium. (C) Full size images of the blot used to make Fig. 5e, immunoblotting for anti-pSMAD2 (top), anti-SMAD2/3 (middle), and anti-GAPDH (bottom) as loading control. The gel was loaded with lysates from cultured hfCFBs treated for 0 (T0), 30 (T30) or 60 (T60) minutes with L2medium or Veh-medium that was pre-incubated with TGFβ. The membranes were cut to probe for different antibodies. Red boxes indicate the wells that were used to make the manuscript figure.  . Both gels were loaded with lysates from cultured human foetal cardiac fibroblasts that were grown for four or seven days in total. Red boxes indicate the wells that were used to make the Supplementary figure.