Lysyl oxidase‐like 2 is a regulator of angiogenesis through modulation of endothelial‐to‐mesenchymal transition

Abstract Lysyl oxidase‐like 2 (LOXL2) belongs to the family of lysyl oxidases, and as such promotes crosslinking of collagens and elastin by oxidative deamination of lysine residues. In endothelial cells (ECs), LOXL2 is involved in crosslinking and scaffolding of collagen IV. Additionally, several reports have shown a role for LOXL2 in other processes, including regulation of gene expression, tumor metastasis, and epithelial‐to‐mesenchymal transition (EMT). Here, we demonstrate an additional role for LOXL2 in the regulation of angiogenesis by modulation of endothelial‐to‐mesenchymal transition (EndMT). LOXL2 knockdown in ECs results in decreased migration and sprouting, and concordantly, LOXL2 overexpression leads to an increase in migration and sprouting, independent of its catalytic activity. Furthermore, LOXL2 knockdown resulted in a reduced expression of EndMT markers, and inhibition of transforming growth factor‐β (TGF‐β)‐mediated induction of EndMT. Interestingly, unlike in EMT, overexpression of LOXL2 alone is insufficient to induce EndMT. Further investigation revealed that LOXL2 expression regulates protein kinase B (PKB)/Akt and focal adhesion kinase (FAK) signaling, both pathways that have been implicated in the regulation of EMT. Altogether, our studies reveal a role for LOXL2 in angiogenesis through the modulation of EndMT in ECs, independent of its enzymatic crosslinking activity.

Furthermore, BAPN reduced sprouting in human umbilical vein endothelial cells (HUVEC), but not to the same extent as a LOXL2 knockdown, strongly suggesting that LOXL2 regulates angiogenesis in part through processes independent of extracellular enzymatic activity.
Endothelial-to-mesenchymal transition (EndMT) is a process in which endothelial cells differentiate toward a mesenchymal phenotype. This process is characterized by a loss of apical-basolateral polarity, increased ECM production, cell migration, sprouting, and an increase in mesenchymal marker expression, including α-smooth muscle actin (α-SMA). Furthermore, EndMT results in decreased expression of endothelial markers such as vascular endothelial cadherin (VE-cadherin) and platelet endothelial cell adhesion molecule (PECAM-1;van Meeteren & ten Dijke, 2012;Welch-Reardon, Ehsan, & Wang, 2014). EndMT is a process that plays an important role in cardiovascular development and repair, fibrosis, and angiogenesis (Lin, Wang, & Zhang, 2012;Muylaert, de Jong, & Slaats, 2015;Welch-Reardon et al., 2014), and is remarkably similar to EMT, a process in which epithelial cells undergo a differentiation toward a more mesenchymal, fibroblast-like phenotype. Both processes are important in embryonic development and show a large overlap in regulatory pathways (Kovacic, Mercader, & Torres, 2012). Interestingly LOXL2 has been shown to play a regulatory role in EMT through a variety of mechanisms, including the inhibition of transcription factors that regulate both EMT and EndMT (Medici, Potenta, & Kalluri, 2011;. We thus hypothesize that, alongside its described role in ECM remodeling, LOXL2 regulates angiogenesis through modulation of EndMT.
ECs were used up to maximum 10 passages to avoid loss of lentiviral constructs. Furthermore, cells of the same passage numbers, as well as passage numbers since lentiviral transduction, were used within experiments to correct for potential effects of cellular passage number and construct expression.

| Immunoblotting
Cell samples were resuspended in RIPA buffer (Santa Cruz Biotechnology, Dallas, TX), and protein concentrations were determined by BCA protein assay (Thermo Fisher Scientific). Equal protein amounts were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) using NuPAGE 4-12% Bis-Tris gradient gels or 12% Bis-Tris gels (Thermo Fisher Scientific) alongside a prestained EZ-Run protein ladder (Thermo Fisher Scientific), and subsequently transferred to polyvinylidene difluoride (PVDF) membranes (Thermo Fisher Scientific).
Depending on the antibody, PVDF membranes were either blocked in 5% DE JONG ET AL.

| 10261
BSA (Roche, Basel, Switzerland) or 5% low-fat dry milk powder (Campina, Woerden, The Netherlands) in tris-buffered saline (TBS) with 0.1% Tween-20 (TBST) for 1 hr at room temperature. PVDF membranes were incubated with primary antibodies in their respective blocking buffers overnight at 4°C. PVDF membranes were then washed in TBST, and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies in 5% low-fat dry milk powder in TBST for 1 hr at room temperature. PVDF membranes were subsequently washed in TBS and proteins were visualized using Chemiluminescent Peroxidase Substrate Secondary antibodies used were HRP-conjugated swine anti-rabbit, rabbit anti-mouse, and rabbit anti-goat (all from Dako). Statistical analysis was performed using a one-tail Student's t test, after densitometric analysis using the ImageJ software (Laboratory for Optical and Computational Instrumentation, University of Wisconsin at Madison, WI).

| Quantitative polymerase chain reaction (qPCR)
Total RNA was isolated from cells using TRIzol isolation (Thermo Fisher Scientific), according to the manufacturer's protocol, followed by cDNA synthesis using the iScript cDNA Synthesis Kit (Bio-Rad). qPCR was performed using IQ Sybr Green Super Mix (Bio-Rad) in a CFX96 Real-Time PCR Detection System (Bio-Rad). Primer sequences were taken from the PrimerBank PCR primer public resource (Wang, 2003) and synthesized by Sigma-Aldrich. The following primers were used: ACCAGATGCACATTGAT-3′, 5′-TCTTGCGACTCACGCTTGAC-3′), and fibronectin (5′-GAAGCTCTCTCTCAGACAACCA-3′, 5′-GCCCACGGTA ACAACCTCTT-3′). Cycle threshold (C t) values were normalized per experiment and per gene. ΔΔC t was calculated using housekeeping gene RPLP0. Statistical analysis was performed using a two-tail Student's t test.

| Scratch migration assay
HMEC-1 cells were plated in 24-well plates, in a density that ensured full confluency during the assay. EC migration was measured by making a single, straight scratch in the cell monolayer, using a 200 µl pipet tip.
Wells were then washed once with phosphate-buffered saline (PBS), and 0.5 ml MCDB 131 with 100 U/ml penicillin and 100 µg/ml streptomycin (basal HMEC-1 medium) was added to each well. Cells were incubated at 37°C for 6 hr. Images were recorded at t = 0 and t = 6 hr at the same location, and scratch closure was determined by measuring the scratch surface at both time points using the ImageJ software. Statistical analysis was performed using a two-tail Student's t test, or a two-way analysis of variance (ANOVA) with Tukey's multiple comparisons post hoc analysis to compare multiple groups.

| Angiogenic sprouting assay
Angiogenic sprouting assays were performed as described in Balkom (2013); sprouting was assessed by seeding HMEC-1 cells onto Cytodex 3 microcarrier beads (Sigma-Aldrich), and embedding them in a mixture of basal HMEC-1 medium, complete HMEC-1 medium, and growth factor reduced Matrigel (Becton Dickinson, Franklin Lakes, NJ) in a 1:1:4 ratio, supplemented with doxycycline or a vehicle control where indicated, in a 100 µl volume in 48-well plate wells. After the gels had solidified, 0.5 ml basal HMEC-1 medium was added on top of the gels, and images were taken after a 72 hr incubation at 37°C, 5% CO 2 . Sprout lengths were measured using the ImageJ software. Ind LOXL2 EC and WT EC were stimulated with doxycycline or a vehicle control in HMEC-1 medium with 0.2% FCS supplemented for 96 hr before the assay. Statistical analysis was performed using a two-tail Student's t test, or a two-way ANOVA with Tukey's multiple comparisons post hoc analysis to compare multiple groups.

| Statistics
Data were normalized to means of each experiment with the reference condition set as 1. All statistical analyses were performed using Graphpad Prism 6.05. All values are expressed as the mean ± standard deviation (SD). Differences were considered statistically significant at p < 0.05.  Figure S1a).

| 10263
to LOXL2 overexpression, which significantly increased angiogenic sprouting in EC (Figure 2c,d). LOXL2 knockdown reduced the number of angiogenic sprouts per bead, but not the average sprouting length (Supporting Information Figure S2a,b), whereas overexpression of LOXL2 resulted in a significant increase of the number of angiogenic sprouts per bead, as well as the average sprouting length (Supporting Information Figure S2c,d). In a control experiment, stimulation of wildtype EC with doxycycline showed no significant effect on angiogenic sprouting (Supporting Information Figure S1b).

| LOXL2 induces angiogenesis independent of its enzymatic activity
To determine whether the positive effect of LOXL2 overexpression on EC migration and sprouting is due to enzymatic crosslinking of collagens and elastins by LOXL2, a catalytically inactive mutant of LOXL2 was generated by introducing two-point mutations, replacing the histidines at positions H626 and H628 with glutamine residues (Ind LOXL2 H626/ 628Q; Figure 3a). These mutations have been reported to completely abolish the oxidative deamination activity of LOXL2, the process through which lysyl oxidases crosslink collagens and elastins, and not affect the induction of EMT in epithelial cells (Cuevas, Moreno-Bueno, & Canesin, 2014). Inducible overexpression of LOXL2 H626/628Q in EC was confirmed by immunoblot analysis (Figure 3b). Interestingly, overexpression of this catalytically inactive LOXL2 mutant resulted in a similar increase in EC migration as seen in EC with wildtype LOXL2 overexpression (Figure 3c,d). Additionally, a similar increase in angiogenic sprouting as a result of overexpression of wildtype LOXL2 and LOXL2 H626/628Q was observed (Figure 3e,f). These data show that the effects of LOXL2 on EC migration and sprouting do not depend on LOXL2mediated enzymatic crosslinking of ECM components, but rather point toward a mechanism involving LOXL2-mediated intracellular signaling.

| LOXL2 is required but not sufficient to induce EndMT
Given the known stimulatory effects of EndMT on angiogenesis (Muylaert et al., 2015;van Meeteren & ten Dijke, 2012) and the reported dominant role of LOXL2 in the regulation of EMT , we hypothesized that the observed results may be explained by a similar regulatory function of LOXL2 in EndMT.
To assess whether LOXL2 indeed plays a role in the regulation of EndMT, shLOXL2 EC were analyzed for expression of the endothelial markers VE-cadherin and PECAM-1, and mesenchymal markers α-SMA, calponin 1, and fibronectin. qPCR analysis revealed an increased expression of endothelial markers and a decreased expression of mesenchymal markers (Figure 4a). Upregulation of VE-cadherin and PECAM-1, as well as downregulation of α-SMA and fibronectin, were also confirmed by immunoblot analysis (Figure 4b). These observed changes indicate a balance toward a more endothelial phenotype, in line with a decrease in EndMT. To determine whether LOXL2 plays a role in the TGF-β-mediated regulation of EndMT, LOXL2 expression was measured after stimulation of wildtype EC with TGF-β for 24 hr. Indeed, a substantial increase of LOXL2 expression was observed (Figure 4c). To further assess the role of LOXL2 in EndMT, we exposed shLOXL2 EC to F I G U R E 4 LOXL2 knockdown in EC reduces EndMT, but LOXL2 overexpression does not induce EndMT. (a) LOXL2 knockdown in EC reduces mRNA expression of mesenchymal markers α-SMA, calponin 1, and fibronectin, and increases mRNA expression of endothelial markers PECAM-1 and VE-cadherin (n = 3, + SD, Student's t test). (b) LOXL2 knockdown in EC increases protein expression of endothelial markers PECAM-1 and VE-cadherin, and reduces protein expression of mesenchymal markers α-SMA and fibronectin. (c) The LOXL2 expression is increased in EC after stimulation with TGF-β for 24 hr. (d) LOXL2 knockdown results in delayed TGF-β-mediated upregulation of α-SMA in EC. (e) LOXL2 overexpression does not affect mRNA expression of mesenchymal markers α-SMA, calponin 1, and fibronectin, or expression of endothelial markers PECAM-1 and VE-cadherin (n = 3, + SD, Student's t test). (f) LOXL2 overexpression does not affect protein expression of endothelial markers PECAM-1 and VE-cadherin, or expression of mesenchymal markers α-SMA and fibronectin; *p < 0.05, **p < 0.01, and ***p < 0.001. EC: endothelial cells; EndMT: endothelial-to-mesenchymal transition; GAPDH: glyceraldehyde 3-phosphate dehydrogenase; LOXL2: lysyl oxidase-like 2; mRNA: messenger RNA; PBS: phosphate-buffered saline; PECAM-1: platelet endothelial cell adhesion molecule 1; shCtrl: shRNA control; α-SMA: α-smooth muscle actin; shLOXL2: short hairpin-mediated LOXL2 knockdown; TGF-β: transforming growth factor-β; VE-cadherin: vascular endothelial cadherin TGF-β for 7 days (168 hr) to induce EndMT, and measured expression of EndMT marker α-SMA (Figure 4d). Immunoblot analysis showed a significantly lower expression of α-SMA in shLOXL2 EC compared with control EC (shCtrl) after 24, 48, and 96 hr of TGF-β stimulation (p = 0.04, 0.02, and 0.04, respectively; n = 4), but not after 168 hr, indicating a delayed, but not blocked, response to TGF-β. This demonstrates a LOXL2-dependent regulation of EndMT. To investigate whether LOXL2 upregulation is sufficient to induce EndMT, Ind LOXL2 EC were stimulated with doxycycline for 7 days to establish continuous LOXL2 overexpression. Interestingly, no significant changes in EndMT marker expression could be observed by both qPCR (Figure 4e) and immunoblot analysis (Figure 4f). As a control, stimulation of wildtype EC with doxycycline showed no significant effect on EndMT marker mRNA levels (Supporting Information Figure S1c). These data show that LOXL2 plays a regulatory role in EndMT, but overexpression alone is not sufficient to induce EndMT.  (Deng, Yang, & Liu, 2010;Larue & Bellacosa, 2005;Meadows, Iyer, & Stevens, 2009;Shen, Park, & Alcaraz, 2005), we studied the effect of LOXL2 knockdown and overexpression on these regulatory pathways. We found that knockdown of LOXL2 resulted in decreased PKB/Akt and FAK signaling: Phosphorylation of FAK was decreased, and expression of PKB/Akt-as a result presence of phosphorylated PKB/Akt-as well was substantially decreased (Figure 5a). Concordantly, overexpression of LOXL2 for 24 hr resulted in an increase in phosphorylation of PKB/ Akt and FAK (Figure 5b). Similarly, overexpression of the catalytically inactive LOXL2 H626/628Q resulted in increased phosphorylation of PKB/Akt and FAK as well, demonstrating that activation of these signaling pathways is independent of the enzymatic activity of LOXL2 ( Figure 5c). Unlike in EMT , knockdown or overexpression of LOXL2 did not affect the expression of Snail (Supporting Information Figure S3).

| DISCUSSION
Here we demonstrate that LOXL2 is a regulator of EndMT, and through this process plays a regulatory role in angiogenesis. By studying the effects of LOXL2 knockdown, overexpression, and overexpression of a catalytically inactive mutant of LOXL2 in EC, we determined the respective roles of LOXL2 enzymatic activity and its intracellular signaling in the regulation of angiogenesis. While a role for LOXL2 in angiogenesis and embryonic development through lysyl oxidase-mediated ECM remodeling has previously been reported, we show for the first time that LOXL2 is also involved in additional intracellular signaling pathways in EC. These findings complement the previously reported effects of LOXL2-mediated ECM crosslinking activity on angiogenesis (Bignon et al., 2011).
These findings indicate that EndMT, as compared to EMT, is controlled by an additional layer of regulatory pathways, which might be in place to protect EC function to maintain the integrity of the circulatory system, as inappropriate EndMT can result in fibrosis and atherosclerosis (Chen, Qin, & Baeyens, 2015;Elliott, Gurtu, & McCollum, 2014;Zeisberg, Tarnavski, & Zeisberg, 2007).
Our data indicate that the effects of LOXL2 overexpression on migration and sprouting are due to activation of pathways that are involved in the regulation of EndMT and EMT, but are not sufficient to induce full EndMT in current conditions. Indeed, we find that LOXL2 expression regulates FAK signaling, which has been implicated in the regulation of EMT (Deng et al., 2010), and PKB/Akt signaling, which has been implicated in the regulation of both EndMT and EMT (Larue & Bellacosa, 2005;Meadows et al., 2009).
Additionally, we find that LOXL2 expression is increased by TGF-β stimulation in EC. TGF-β-mediated regulation of LOXL2 had been reported in hepatocellular carcinoma (Wong, Tse, & Huang, 2014) but had not yet been described in endothelial cells. Importantly, LOXL2 knockdown also resulted in delayed TGF-β-mediated EndMT, confirming its crucial role in the regulation of EndMT.
In conclusion, we demonstrate a regulatory role for LOXL2 in adult endothelial cells during angiogenesis and EndMT, through intracellular signaling independent of its ECM remodeling activity.