Axl expression is increased in early stages of left ventricular remodeling in an animal model with pressure-overload

Background AXL is a receptor tyrosine kinase that has been related to kidney and vascular disorders. Heart failure patients with reduced ejection fraction have higher AXL in serum than controls. No information about Axl expression with HF progression is available. Methods Thoracic transverse aortic constriction (TAC) was successfully performed on male Wistar rats (n = 25) with different constriction levels. Controls underwent sham surgery (n = 12). Echocardiography measurements were performed 4–8 weeks after surgery. Collagen deposition was measured with picrosirius red staining. Axl mRNA levels in left ventricle (LV), left kidney (LK) and ascending aorta (aAo) and the LV expression of cardiac remodeling and fibrogenic factors were quantified with real-time PCR. AXL LV protein levels were quantified with western blot and localization was analyzed by immunohistochemistry. Soluble AXL levels in plasma were assayed with ELISA. Results Successful TAC rats were classified into LV hypertrophy (LVH) or heart failure (HF), modeling the progressive cardiac changes after pressure overload. Collagen deposition was increased only in the HF group. LV Axl mRNA levels were higher in LVH and HF than in Sham rats, and correlated with LVHi, and hypertrophic and fibrogenic mediators. However, no association was found with LV systolic function. AXL was expressed in LV myocytes and other cell types. Concentration of circulating sAXL in plasma was increased in the LVH group compared to Sham and HF rats. Axl mRNA levels were similar in all groups in the LK and aAo. Conclusions Axl expression pattern suggests a role in the early progression of LV remodeling in HF but not in the later systolic dysfunction. The higher levels of circulating AXL found in HF patients most probably shed from the heart.


Methods
Thoracic transverse aortic constriction (TAC) was successfully performed on male Wistar rats (n = 25) with different constriction levels. Controls underwent sham surgery (n = 12). Echocardiography measurements were performed 4-8 weeks after surgery. Collagen deposition was measured with picrosirius red staining. Axl mRNA levels in left ventricle (LV), left kidney (LK) and ascending aorta (aAo) and the LV expression of cardiac remodeling and fibrogenic factors were quantified with real-time PCR. AXL LV protein levels were quantified with western blot and localization was analyzed by immunohistochemistry. Soluble AXL levels in plasma were assayed with ELISA.

Results
Successful TAC rats were classified into LV hypertrophy (LVH) or heart failure (HF), modeling the progressive cardiac changes after pressure overload. Collagen deposition was increased only in the HF group. LV Axl mRNA levels were higher in LVH and HF than in Sham rats, and correlated with LVHi, and hypertrophic and fibrogenic mediators. However, no association was found with LV systolic function. AXL was expressed in LV myocytes and other cell types. Concentration of circulating sAXL in plasma was increased in the LVH group compared to Sham and HF rats. Axl mRNA levels were similar in all groups in the LK and aAo. PLOS

Introduction
AXL is a receptor tyrosine kinase that belongs to the TAM family (Tyro3, Axl, MerTK receptors). Gas6 is the only recognized Axl receptor ligand, and its binding triggers Axl oligomerization and activation of downstream signal cascades that are involved in cancer, chronic immune disorders and other diseases [1,2]. The Gas6/AXL axis regulates collagen deposition in different organs. Angiotensin II (AngII), a main pro-fibrotic factor, triggers Axl expression in VSMCs in vitro [3]. In the liver, Gas6/Axl promotes collagen deposition through modulation of the myofibroblast phenotype of hepatic stellate cells [4,5]. In Axl(-/-) mice, CCl4-induced liver fibrogenesis was blunted in comparison with wild type mice [5]. Soluble AXL (sAXL) increases in patients with liver cirrhosis [6,7], evolving as a potential biomarker for hepatopathies. Tissue-wise, AXL expression is increased in keloid fibroblasts [8] and in lung fibroblasts in idiophatic pulmonary fibrosis [9].
In the kidney, the receptor AXL and its ligand Gas6 have been reported to play an important role in several types of kidney disease animal models [10]. Furthermore, increased Axl expression has been detected in kidney samples from humans with diverse inflammatory renal diseases while barely detectable in normal kidney [11].
Evolving evidence supports that AXL plays a role in cardiovascular diseases through modulation of vascular smooth muscle cell (VSMC) phenotype [1]. AXL mediates the vascular response to mechanical and chemical insults and hypertension [12][13][14] and contributes to vascular calcification [15].
More recently, Axl expression in murine cardiac tissue has been demonstrated [16] and a putative role of Axl in heart failure (HF) is under study on grounds of experimental and clinical data. Gas6(-/-) mice lacking the Axl-ligand Gas6 are protected from pressure-overload left ventricle (LV) remodeling and fibrosis, while LV remodeling was remarkably increased in mice with Gas6 cardiac overexpression [17]. In humans, sAXL evolves as an attractive biomarker for HF and other cardiac conditions [18,19].
Our group previously reported higher cardiac and circulating AXL levels in HF patients with reduced ejection fraction (HFREF) than in controls, and sAXL levels predicted worse HF outcomes [20]. Nevertheless, it remains unknown the timeline of Axl expression in HF progression. In the present work, we aimed to analyze Axl expression in LV along with cardiac remodeling progression triggered by pressure-overload in a HF animal model. We also aimed to map the putative origin of the increased sAXL in HF patients by quantifying Axl mRNA expression in other potential origins.

Thoracic transverse aortic constriction (TAC)
The experimental protocols followed the European Community Directive (2010/63/UE) and Spanish guidelines (RD 53/2013) for the use of experimental animals and were approved by Male wistar rats (300-350g) were purchased at Charles River (Germany) and housed at the University of Barcelona animal facility in a controlled environment (12h of light/ 12h of dark) and with food and water ad libitum.
Thoracic transverse aortic constriction (TAC) surgery was performed in 33 rats under surgical lens. Rats were anesthesized with 2-3% inhaled isofluorane, intubated and ventilated (CWE, Ardmore, PA, USA), and kept at 37.0±0.3˚C during the whole experiment with a homeothermic pad (Kent Scientific, Torrington, CT, USA). Intramuscular buprenorphine (0.03-0.05 mg/Kg) was used as analgesic. A portion of the aorta between the brachiocephalic trunk and the left carotid artery was isolated through medial suprasternal partial thoracotomy. A bended and blunted 18G or 20G needle was placed next to the aorta. Afterwards a piece of 5/0 nylon suture was placed and knot around both the vessel and needle to restrict blood flow. The needle was subsequently carefully removed. Sham rats (n = 12) were littermate controls that underwent the same surgery, but no needle and no constriction were applied. The chest cavity and the skin were carefully closed with sutures. After recovery from anesthesia, rats were placed back to their cages with food and water access ad libitum.

Electrocardiogram (ECG)
Four to eight weeks after surgery, a one-lead (DII) surface ECG was recorded. Rats were maintained at 37.0±0.3˚C with a heating pad (KentScientific, Torrington, CT, USA) under inhaled isofluorane 1.5%. Recordings were obtained with PowerLab and Labchart v8.0 (AD Instruments, Oxford, UK). ECGs were analyzed offline (Labchart v8.0, AD Instruments, Oxford, UK) by an investigator blind to group assignment. During a stable 2-minute interval, every two-hundred beats were averaged and semi-automatically measured. Afterwards, the ECG parameters were manually corrected, and average results were obtained for each animal. For each rat, heart rate (HR), RR interval, P duration, P amplitude, PR interval, QRS duration, R amplitude, T amplitude, QT interval and Mitchell-corrected QT interval [21] were measured.

Echocardiography
Subsequent to the ECG recording, a transthoracic echocardiography was obtained under inhaled isofluorane (2-3%). Rats were placed in a heating pad and a phased-array probe 10S (4.5-11.5 MHz) was used attached to the Vivid system (GE Vingmed Ultrasound, Horten, Norway). Images were analyzed using commercially available software (EchoPac v. 108.1.6, GE Healthcare, Madrid, Spain). Left atrial and ventricle diameters, and LV functional measurements were quantified in a longitudinal parasternal plane. LV diameters at both end-diastole (LVEDD) and end-systole (LVESD) were measured at the level of the papillary muscles. The posterior wall (PW) and interventricular septum (IVS) thickness were measured at enddiastole and were indexed by the rat body weight at the sacrifice (BW). LV ejection fraction (LVEF), fractional shortening (FS), and LV mass were estimated using previously validated formulas in rodents [22,23].
Left ventricle mass was body weight-indexed to calculate a LV hypertrophy index (LVHi). Two different stages of pressure-overload were identified according to echocardiographic parameters: compensated LV hypertrophy (LVH), and overt LV systolic dysfunction and heart failure (HF).
Because of the strain and age variability of normality values in rats, and the lack of robust reference values in the literature, we established internally valid thresholds by characterizing sham rats. The fractional shortening (FS) lower limit of normality was established at the lowest value in the sham group, and the LVHi upper limit of normality was established at the highest value in the sham group. Sham-based thresholds are shown in the S1 Fig. Accordingly, those rats in the TAC group with a FS<35.2% were considered to have systolic function impairment, and those TAC rats with LVHi>3.32 g/kg were considered to have LV hypertrophy (LVH).

Euthanasia and sample collection
Rats were sacrificed with an overdose of 4-5% inhaled isoflurane. Blood samples were collected with EDTA tubes and centrifuged (2000 g, 10 min at 4˚C). The supernatant was collected and aliquoted. The ventricular apex was cut and fixed with formol and latter embedded with paraffin. Left ventricle (LV) free wall, left kidney (LK) and ascending Aorta (aAo) tissues were extracted, minced and immediately frozen with liquid nitrogen. All samples were kept at -80˚C until RNA or protein extraction.

Fibrosis quantification
Four-micron thick sections from the apex were stained with Picrosirius-red as described in Sanz-de la Garza et al. [24]. Six representative pictures were taken from each sample with an Olympus BX41TF light microscope and a DP73 camera (10X objective). Relative intramyocardial fibrosis was quantified blinded to group assignment, using Fiji for Windows [25]. Care was taken to exclude perivascular and endocardial and epicardial fibrosis.

RNA isolation and real time PCR assays
Messenger RNA extraction from LV, LK and aAo was performed with Trizol Reagent (Invitrogen, Thermo Fisher) tissue disruption followed with silica column purification following the mirVana miRNA Isolation Kit protocol (Invitrogen). Total RNA (1 μg for LV and LK; 0.5 μg for aAo) was retrotranscripted with the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher). Axl, LV remodelling and fibrosis markers mRNA levels were measured with Real time PCR with the enzyme SYBR Select Master Mix (4472908, Thermo Fisher). Real time PCR results were normalized to the cyclophilin B (Ppib) housekeeping gene, and presented with the 2 -ΔΔCt method.
To further explore the hypertrophic remodeling processes involved in this model, we used the R384 "Cardiac hypertrophy" (Bio-Rad) PCR array including 96-well predesigned assays (assayed genes in S1 Methods) for n = 10 Sham, n = 7 LVH and n = 11 HF rats. Five μg of total RNA were retrotranscribed with the iScript cDNA Synthesis Kit (1708891, Bio-Rad). The cDNA was diluted to a final 25 ng/μl concentration and gene expression levels were assayed with the SsoAdvanced Universal SYBR Green Supermix (1725272, Bio-Rad) and the plates. Cardiac hypertrophy mRNA levels were normalized to the geometric mean of Gapdh and Hprt1, as suggested by the manufacturer. Results are shown with the 2 -ΔΔCt method.
A detailed protocol is provided in the S1 Methods.

AXL protein levels quantification in plasma and cardiac tissue
Soluble AXL (sAXL) was quantified in plasma samples with the commercial ELISA Kit for AXL Receptor Tyrosine Kinase in Rat (EKU02655, Biomatik). Total protein extract was obtained from frozen LV samples of the Sham (n = 5), LVH (n = 8) and HF (n = 10) groups. Thirty micrograms of total protein were run in an electrophoresis setup, transferred onto a nitrocellulose membrane, checked by Ponceau staining, and incubated with a primary polyclonal anti-Axl antibody (ab-72069, abcam) and a goat anti-rabbit secondary antibody. Bands were scanned in Image Quant LAS4000 (GE Healthcare, OH, USA), quantified by Image Lab (Biorad) and normalized with the Ponceau staining.
A detailed protocol is provided in the S2 Methods.

Immunohistochemistry
Immunohistochemistry staining for AXL was performed in paraffin embedded apex slides by using the polyclonal rabbit Axl antibody (PA5-23254, ThermoFisher). A detailed protocol is provided in the S3 Methods.

Statistical analysis
Results are presented as mean±SEM or as percentages. Gaussian distribution was checked with Q-Q plots. For mRNA, the 2 -ΔΔCt results did not follow a normal distribution and a logtransformation was applied. Comparisons between the three groups (Sham, LVH and HF) were performed with a one-way ANOVA, followed by LSD tests in the case of a significant omnibus test. For the PCR array, those genes in which amplification was not achieved in >30% of samples, were excluded. Values were back-transformed to 2 -ΔΔCt for visual representation. Correlations were analyzed with a Pearson test; a log-transformation was performed if any variable was non-normally distributed.
To identify correlated (co-expressed) genes, a principal component analysis was performed. The ΔCt value of all genes in the cardiac hypertrophy R384 array and Axl mRNA levels were jointly analyzed. Missing values (n = 49 for the whole panel) were imputated with the pca function in R. A principal component analysis was conducted (prcomp function in R), and genes were plotted on the basis of their two main components.

TAC as a progressive HF model
From the 33 rats that underwent TAC surgery, eight (24%) presented a normal systolic function and normal cardiac mass. These rats were classified as non-successful TAC and excluded from further analyses. Fourteen rats (42%) had systolic function impairment (FS<35.2%; lowest value of sham rats) and were classified as HF rats. Eleven rats (33%) had no FS impairment but had a LVHi>3.32 g/kg (largest value of sham rats) and were classified as LVH rats. Overall, 25 rats were classified as successful TAC into two different groups: a first stage involving LV hypertrophy and normal systolic function, and a second stage characterized by FS impairment. Representative LV parasternal, M-mode echocardiographic images of all groups are shown in S2 Fig. LVHi and FS values are represented in Fig 1A and 1B.
Echocardiographic and ECG studies reflected a progressive remodeling in rats submitted to TAC. A thicker body weight-indexed IVS was found in the LVH group compared with Sham, in contrast to no differences for indexed PW (S1 Table). HF rats had a dilated LV (larger LVEDDi and LVESDi, S1 Table) compared with both Sham and LVH rats. All rats had an ECG performed before sacrifice (S2 Table). LVH and HF rats presented prolonged QT and HR-corrected QT intervals in comparison to Sham rats. No significant differences were found amongst groups for P wave, PR-interval, QRS duration and HR.
Collagen deposition in the ventricular apex was assessed with picrosirius red staining. Fibrosis was similar in LVH and Sham rats, while HF rats had a remarkable~2.5-fold increase ( Fig 1C and 1D). The magnitude of collagen deposition correlated through an apparently asymptotic curve with the degree of structural remodeling: fibrosis negatively correlated with FS and positively with LVHi, LVESDi and LVEDDi (S3A Fig).

Molecular validation of different stages of pressure-overload
Consistent with HF progression, BNP mRNA levels in LV were significantly different among the three groups largely due to a 2.5-fold increase in both TAC groups (LVH and HF) compared to Sham (ANOVA, p<0.001, Fig 2A). The βMHC/αMHC ratio also differed with increasing expression from Sham to LVH rats (4-fold) to HF rats (6-fold increase in comparison with Sham) (ANOVA, p<0.001, Fig 2B).
Analysis of cardiac hypertrophy related genes revealed a characteristic, progressive hypertrophic remodeling pattern from Sham to LVH to HF rats. The complete report of results for all genes is provided in S3 Table. Increased LV mRNA levels of proteins involved in the IGF1 (Insulin-like growth factor 1, Fig 2C) pathway such as IGF1 receptor (p<0.05) and the Akt (RAC-alpha serine/threonine-protein kinase) kinases (Akt1, p<0.05 and Akt2, p<0.01) were found in LVH and HF rats compared to Sham, while Akt3 did only increased in HF rats (p<0.05). We also found an apparent activation of downstream G protein-coupled receptors (Fig 2D and S3 Table) pathways such as Mapk3 (Mitogen-activated protein kinase 3, p<0.01), Camk2d and Camk2g (Calcium/calmodulin-dependent protein kinase type II subunits delta and gamma, p<0.01 in both cases), Mef2c and Mef2d (Myocyte-specific enhancer factor 2C and 2D, p<0.01 in both cases) and Hdac5 (Histone deacetylase 5, p<0.05).
Left ventricular BNP mRNA expression correlated with all the echocardiographic measures except for FS and LVEF (Table 1).
Fibrotic and extracellular remodeling markers Tgfβ1, Mmp2, Timp1, Col1a1, Col3a1, fibronectin and vimentin demonstrated a complex balance underlying myocardial fibrosis at different overload stages. Expression of the profibrotic Tgfβ1 cytokine and the extracellular matrix remodeling factors Mmp2 and Timp1 were significantly increased in both LVH and HF groups compared to Sham (S3B Fig). Col1a1 and Col3a1 mRNA levels were also different between groups (ANOVA, p<0.01 and p<0.05, respectively), with higher values in LVH rats than in Sham rats (S3C Fig). Differences between groups were also found with the profibrotic factors fibronectin and vimentin (S3C Fig) with higher values in both LVH and HF than in Sham rats.

Axl expression is increased in LV in TAC animals
Left ventricular Axl mRNA levels significantly differed among the three groups (Fig 3A). Axl expression was higher in LVH and HF rats compared with Sham animals; expression was similar in LVH and HF animals. A similar pattern was found for AXL protein levels in Westernblot experiments, although this did not reach statistical significance (p = 0.067 for Sham vs  (Table 1). At the molecular level, Axl mRNA levels showed a moderate, positive correlation with BNP (Fig 4A), βMHC/αMHC ratio (Fig 4B), Tgfβ1 (Fig 4C), Mmp2 ( Fig  4D) and Timp1 (Fig 4E)     AXL protein expression in the LV localized in myocytes, inflammatory and mesothelial cells, as well as in vascular endothelial and smooth muscle cells (Fig 3C).
Plasma concentration of soluble Axl was higher in LVH compared with Sham and HF rats (Fig 3D). Soluble AXL yielded a moderate to good estimate of LV Axl mRNA levels (Fig 3E).
In view of the association between AXL and cardiac hypertrophy, we aimed at further exploring Axl co-expression patterns with hypertrophic factors. Axl mRNA levels were analyzed in the context of hypertrophy mediators. A joint principal component analysis showed that Axl expression clustered with other well-known hypertrophy factors such as Edn (Endothelin-1), Tlr4 (Toll-like receptor 4 precursor), Cxcr4 (C-X-C chemokine receptor type 4) and Igf1 (Insulin-like growth factor I) (Fig 5).

Axl expression is not increased in LK or aAo in TAC animals
To test whether HF prompted remote changes in Axl expression, expression in the LK and aAo was quantified. We found no differences between Sham, LVH and HF rats in either LK or aAo (Fig 6A and 6B).

Discussion
This work describes for the first time changes in the LV expression of the tyrosine kinase receptor Axl in a pressure-overload murine model. Our main findings are that (a) Axl mRNA levels are increased in the LV of rats subjected to pressure-overload, (b) Axl expression correlates with the degree of LV hypertrophy but not with systolic function impairment, (c) circulating sAXL levels estimate LV expression and are higher in plasma from LVH rats compared to Sham and HF rats, and (d) renal and vascular Axl expression are not altered in this model.

AXL participates in pressure overload cardiac remodeling
The role of Axl in heart disease is still under debate. In transgenic mice selectively overexpressing cardiac TNF-α, the expression of TAM family members Tyro3 and MerTK was up- Axl role in left ventricular remodeling regulated in end-stage HF animals, while Axl and Gas6 remained unaltered; unfortunately, no time-course experiments at earlier time points were performed [26]. Conversely, mice lacking Gas6, the only known Axl ligand, were protected from pressure-overload and developed a blunted hypertrophic response and dysfunction in comparison with wild-type animals [17]. In a different setting, Zhang et al. showed that Gas6/Axl may have a protective role in a rat model of myocardial ischemia/reperfusion injury [16].
In order to better characterize the role of AXL in HF and understand its role as a biomarker, we studied Axl expression in rats subjected to TAC. Transverse aortic constriction is a wellestablished model that mimics cardiac remodeling occurring after the instauration of LV pressure-overload where different degrees of constriction trigger uneven remodeling levels. Thereby, this model replicates the timeline stages that lead to the initial LVH and, eventually, HF [22,27]. In the LV of TAC rats, Axl mRNA expression was increased even in the absence of overt HF, and correlated with both BNP expression and the βMHC/αMHC ratio; the later are standard measures of pressure-overload cardiac remodeling and dysfunction [17,27]. Nevertheless, differences between these are to be considered. BNP showed a better correlation with systolic function than Axl, in keeping with recent clinical data [20]. The switch from a predominantly alpha-MHC isoform to a predominantly beta-MHC isoform progresses from Sham to LVH to HF rats [27], while Axl mRNA expression was similar in both TAC groups.

Axl and cardiac hypertrophy.
In the LV of this pressure-overload model, we found characteristic hypertrophy signatures in LVH and HF rats. Specifically, we found a progressive (Sham to LVH to HF) activation of IGF1 and GPCRs downstream pathways (Fig 2), including an increased expression of both endothelin receptors A and B (S3 Table). Axl mRNA expression in LV correlated with the LVHi and with well-established hypertrophy factors such as Edn (Endothelin-1), Igf1 (Insulin-like growth factor I), Tlr4 (Toll-like receptor 4 precursor) and Cxcr4 (C-X-C chemokine receptor type 4). In plasma, sAXL was increased only in LVH rats but not in HF rats. All these results are concordant with a role of Axl in the early hypertrophic response of pressure-overload triggered cardiac remodeling, possibly substantiating the lack of changes in late HF stages in previous works [26].

Axl and myocardial fibrosis.
The Axl role in cardiac fibrosis is not conclusive from our results as it is in liver [4,5] or lung [9] fibrosis. A correlation between Axl and Tgfβ1, Mmp2 and Timp1 expression suggests that Axl has a role in early cardiac changes, but we did not find correlation with Col1a1, Col3a1, fibronectin and vimentin mRNA levels in LV, nor with the histological fibrosis. Tgfβ1 is a key factor in pathological fibrosis and a tight relationship of Axl with Tgfβ1 can be implied from the reduced expression of Tgfβ1 in Axl knockdown [8].
The progression of heart fibrosis after acute myocardial infarction (AMI) in an early and a late phase parallels our results. In the early phase, Tgfβ1 stimulates fibroblast chemotaxis, proliferation and transformation to myofibroblast and MMPs and TIMPs expression increases. On the contrary, collagen-I and III accumulation occurs in late fibrosis when there is replacement of myocytes and scar formation [28]. Altogether, it cannot be discarded that Axl triggers Tgfβ1 expression in the early phase and consequently the whole fibrotic cascade.

Soluble AXL as a plasmatic biomarker
Soluble AXL has been used as a plasmatic prognostic biomarker in several heart diseases and settings. In cardiac transplant recipients, sAXL values were higher in those presenting with adverse CV events, an association that remained significant after multivariate adjustment [18]. After an ST-elevation AMI, higher sAXL levels flag those patients who will develop HF and an adverse LV remodeling during follow-up [19]. We have reported that sAXL is a good prognostic marker of the HF major events all-cause mortality, transplantation and HF hospitalizations at 1 year [20] and at 3.6 years follow-up [29]. The higher sAXL levels in plasma from LVH rats, along previous clinical data [18][19][20]29], substantiate the use of sAXL as a plasmatic biomarker in HF patients. In contrast with our observations that Axl may play a role in early remodeling, sAXL was higher in patients with NYHA functional class III than II [20]. Nevertheless, functional class assessment is determined by several factors beyond, and may not fully be explained by, cardiac remodeling. Also, the heterogeneity within HF patients regarding etiology and evolution time could account for the clinic-experimental differences encountered.

Circulating AXL may be of cardiac origin in HF patients.
Axl has been involved in remodeling processes in vessel walls and in the kidney, among others. It is therefore possible that, in HF patients, sAXL does not reflect cardiac remodeling, but rather the co-existence of other comorbidities or the systemic impact of HF. We therefore quantified Axl expression in the kidney and the ascending aorta wall. On one hand, Axl has been shown to contribute to the pathology of several glomerulonephritis and kidney conditions [10,11,30], but our results show that Axl expression is not altered in the kidney. On the other hand, Axl has been demonstrated to mediate VSMC migration and proliferation in vessel walls after vascular injury [31], hypertension [32] and flow-induced vascular injury [15,31]. Due to the increased wall stretch in the proximal side of the aortic constriction, we quantified Axl mRNA levels in aAo, where we found a similar expression of Axl in Sham, LVH and HF rats.
In contrast, we found AXL synthesis in LV myocytes, amongst other cellular types. Moreover, there was a good correlation between cardiac Axl mRNA levels and peripheral sAXL protein. Altogether, these data support that sAXL originates from the LV of HF patients. Nevertheless, we cannot exclude that with worsening of kidney function in later HF stages, an sAXL increase could partially arise from a lack of renal excretion.

Limitations
Some limitations of our work should be acknowledged. First, data from animal models, in which acute pressure overload is imposed to a previously healthy LV might not be comparable to chronic cardiac conditions in humans. Second, up to 24% of rats submitted to TAC did not develop signs of LVH nor of HF. In this regard, a report with mice followed with cardiac magnetic resonance confirmed the high variability of response to TAC; band internalization could account for a lack of cardiac changes in some animals [33].

Conclusions
Axl expression is increased in the LV of compensated and decompensated forms of pressureoverload insults, while its expression in the kidney and ascending aorta remains unaltered. The higher sAXL plasma levels in LVH rats and the relationship between Axl mRNA and LVHi suggest a role of Axl in early cardiac remodeling stages. Nevertheless, further research will be needed to gain more insight about the Axl role in the cardiac remodeling process that leads to HF.  Table. ECG parameters in all groups. LVH: left ventricular hypertrophy rats; HF: heart failure rats; bpm: beats per minute (bpm). �� p<0.01 vs Sham. (DOCX) S3 Table. mRNAs analyzed with the bio-rad rat cardiac hypertrophy plate. Data are provided with the 2 -ΔΔCt method, where fold change in left ventricular hypertrophy rats (LVH) and heart failure rats (HF) are compared to Sham. Whenever amplification was detected in <70% of samples, non-applicable (NA) is reported. � p<0.05 vs Sham, �� p<0.01 vs Sham,