LOXL4 is induced by transforming growth factor β1 through Smad and JunB/Fra2 and contributes to vascular matrix remodeling

Abstract

Transforming growth factor β1 (TGF-β1) is a pleiotropic factor involved in the regulation of extracellular matrix (ECM) synthesis and remodeling. In search for novel genes mediating the action of TGF-β1 on vascular ECM, we identified the member of the lysyl oxidase family of matrix-remodeling enzymes, lysyl oxidase-like 4 (LOXL4), as a direct target of TGF-β1 in aortic endothelial cells, and we dissected the molecular mechanism of its induction. Deletion mapping and mutagenesis analysis of the LOXL4 promoter demonstrated the absolute requirement of a distal enhancer containing an activator protein 1 (AP-1) site and a Smad binding element for TGF-β1 to induce LOXL4 expression. Functional cooperation between Smad proteins and the AP-1 complex composed of JunB/Fra2 accounted for the action of TGF-β1, which involved the extracellular signal-regulated kinase (ERK)-dependent phosphorylation of Fra2. We furthermore provide evidence that LOXL4 was extracellularly secreted and significantly contributed to ECM deposition and assembly. These results suggest that TGF-β1-dependent expression of LOXL4 plays a role in vascular ECM homeostasis, contributing to vascular processes associated with ECM remodeling and fibrosis.

Fig 1

LOXL4 is a TGF-β1 target gene in aortic endothelial cells. (A) Panel of ECM-related genes showing significant expression changes (fold change, >2) upon 24 h of TGF-β1 (5 ng/ml) treatment, identified by genome-wide expression. (B) Quantitative PCR of LOX gene family members (LOX and LOXL1 to -4) in TGF-β1-stimulated BAEC. Values above the bars in parentheses indicate the induction rates with respect to basal levels. (C) LOXL4 mRNA expression as a function of TGF-β1 incubation time as assessed by quantitative PCR. (D) Time-dependent effect of TGF-β1 on LOXL4 protein (representative Western blotting and densitometric analysis of the LOXL4/β-actin signal). +, positive control of cells infected with LOXL4 adenovirus. Values are represented as fold induction with respect to basal values (means ± standard errors of the means [SEM]; n = 3).

Fig 2

Mapping of TGF-β1-responsive elements within the LOXL4 promoter. (A) TGF-β1-dependent activity of luciferase constructs driven by 5′ deletion fragments of the LOXL4 promoter from bp −3870 to −304 transfected into BAEC. A schematic representation of the bp −3900/−2300 promoter sequence shows the position of putative AP-1 and Smad sites (white and black boxes, respectively). (B and C) The −3870/−3756 promoter fragment is sufficient to confer TGF-β1 responsiveness to its cognate proximal promoter (−304/+87) (B) and to a viral SV40 promoter (pGL3-Promoter) (C) in an orientation-independent manner. (D) Specific point mutations in the bp −3870/+87 LOXL4 promoter construct that alter Smad and AP-1 elements as indicated were introduced and their TGF-β1 inducibility measured by luminometry. TGF-β1 induction rates are expressed as mean ± SEM (n = 3). *, P < 0.05 versus promoterless construct (A) versus −304/+87 proximal LOXL4 promoter (B), versus pGL3-Promoter (C), or versus the wild type (D), in the absence of TGF-β1.

Fig 3

Analysis of conserved regions in the LOXL4 promoter across mammalian genomes. (A) Comparative genomic analysis of the promoter regions of several mammalian LOXL4 orthologs (Mulan multiple-alignment engine). The graphical representation shows stacked-pairwise conservation profiles for the 4-kb promoter region of the LOXL4 gene between the indicated species and the reference bovine sequence. The color intensity of a conserved region depends on the number of different species that contain the region (the darker the color, the more conserved species). Only evolutionarily conserved regions (ECR) present in at least six out of eight total secondary species are highlighted in the alignment. Note that sequence corresponding to proximal promoter from cat was not available in the ENSEMBL database. (B) Multiple-sequence alignment of the most upstream ECR, showing the locations of the AP-1 site (blue) and Smad binding elements (red). (C) TGF-β1 inducibility of bovine, human, and mouse 4-kb and proximal LOXL4 promoter luciferase constructs transfected into BAEC (Btau) and MAEC (Mmus). (D) Effect of TGF-β1 on LOXL4 mRNA expression in BAEC and MAEC as assessed by quantitative PCR. Values are represented as fold induction with respect to basal levels in the absence of TGF-β1 (mean ± SEM; n = 3). *, P < 0.05 versus the corresponding proximal promoter for luciferase activity (C) or versus the basal value for mRNA expression (D).

Fig 4

Involvement of Smad and AP-1 transcription factors in the effect of TGF-β1 on LOXL4 expression. (A) Luciferase activity of LOXL4 promoter luciferase constructs cotransfected with increasing amounts of Smad3 or Smad7 overexpression plasmids. (B) Wild-type and Smad3 knockout (KO) mouse embryonic fibroblasts (MEF) were transfected with the LOXL4 promoter together with Smad3 plasmid. (C and D) Effect of adenovirus-mediated overexpression of the dominant negative form of AP-1, TAM67 (10⁸ to 10⁹ PFU/ml) on TGF-β1-mediated induction of LOXL4 mRNA (C) and promoter activity (D). Values are represented as normalized units for luciferase activity and as fold induction with respect to basal values for mRNA expression (mean ± SEM; n = 3). *, P < 0.05 versus pcDNA3 empty vector (−) (A and B, right panel), versus basal levels in the absence of TGF-β1 stimulation (B, left panel), or versus empty vector null adenovirus (C and D).

Fig 5

Involvement of Fra2 in the effect of TGF-β1 on LOXL4 expression. (A) Adenovirus-mediated overexpression of Fra2 potentiates the action of TGF-β1 on LOXL4 expression as assessed by luciferase reporter and Western blotting. (B) siRNA-mediated downregulation of Fra2 inhibits TGF-β1-mediated induction of LOXL4 expression. (C) Effect of overexpression of forced AP-1 dimers, including JunB/Fra2, JunD/Fra2, and c-Jun/Fra2, on TGF-β1-induced LOXL4 promoter activity. (D) Effect of siRNA-mediated downregulation of Jun members (JunB, JunD, and c-Jun) on TGF-β1-induced LOXL4 promoter activity. Values are represented either as fold induction with respect to basal values with null adenovirus or siRNA control or as normalized luciferase units (mean ± SEM; n = 3), *, P < 0.05 versus null adenovirus or empty vector (A and C), or versus siRNA control (B and D), in the presence of TGF-β1. Validation of overexpression or siRNA-mediated downregulation is shown in Fig. S7 in the supplemental material.

Fig 6

TGF-β1 promotes the phosphorylation of Fra2. (A) Protein extracts from BAEC infected with null or Fra2 adenoviruses and stimulated with TGF-β1 or basal were either left untreated or preincubated with calf intestinal alkaline phosphatase (CIAP) (1 unit/μg protein) or with λ-phosphatase (λ-PPase) (3 units/μg protein) for 1 h, followed by detection of Fra2 by Western blotting. Note that high-MW bands were eliminated with CIAP and λ-PPase, likely corresponding to hyperphosphorylated forms. (B) Domain structure of mouse Fra2 protein. Sequences from several Fra2 and c-Fos orthologs were aligned to show the conserved motifs for MAPK consensus phosphorylation sites (*). Note that the mouse T274-containing motif is conserved among all species compared in this study, whereas T263 is partially conserved in some Fra2 orthologs and in human and mouse c-Fos. (C) Analysis of electrophoretic mobility of Flag-tagged Fra2 with point mutations in MAPK consensus phosphorylation sites T263 and T274. Single and double inactivating T→A or phosphomimetic T→E mutant constructs were transiently transfected into bovine aortic endothelial cells with and without TGF-β1 incubation, and protein extracts were analyzed by Western blotting using an anti-Flag antibody. (D) Luciferase activity of LOXL4 promoter luciferase construct cotransfected with wild-type, T263/274A, and T263/274E Fra2 overexpression plasmids. Values are represented as normalized units for luciferase activity (mean ± SEM; n = 3). *, P < 0.05 versus pcDNA3 empty vector (−) in the presence of TGF-β1.

Fig 7

TGF-β1 induces LOXL4 expression by ERK-mediated phosphorylation of Fra2. (A) Effects of MAPK inhibitors SB203580 (p38, 10 μM), CI-1040 (ERK, 2 μM), and SP600125 (JNK, 25 μM) on luciferase activity of LOXL4 promoter luciferase construct. Values are represented as normalized luciferase units (mean ± SEM; n = 3). *, P < 0.05 versus vehicle (dimethyl sulfoxide [DMSO] for SB203580 and CI-1040 and ethanol for SP600125) upon TGF-β1 stimulation. (B and C) Dose dependence analysis of the effect of the ERK inhibitor CI-1040 on TGF-β1-induced LOXL4 promoter activity (B) and mRNA expression (C). (D) Time course of the effect of TGF-β1 on ERK activation (phospho-ERK [p-ERK]) and Fra2 phosphorylation (p-Fra2) analyzed by Western blotting in extracts from cells treated with vehicle (control) or with CI-1040 (2 μM). Values above the blots refer to the p-Fra2/β-actin and p-ERK/total ERK ratios after densitometric analysis. Results are representative of three experiments.

Fig 8

Subcellular localization and extracellular secretion of LOXL4 in aortic endothelial cells. (A) Domain structure of full-length (FL) LOXL4-GFP fusion protein and deletion mutants thereof lacking the SRCR domains and the C-terminal catalytic region. (B) Subcellular localization of full-length LOXL4-GFP expressed in bovine aortic endothelial cells by adenoviral infection as assessed by GFP fluorescence microscopy (upper panel) and colocalization of the endoplasmic reticulum marker calnexin by immunofluorescence (lower panel). (C) Analysis of the extracellular secretion of LOXL4-GFP chimeras stably expressed in HEK 293 cells by Western blotting using anti-GFP antibody. Note the presence of secreted GFP-immunoreactive bands in FL, ΔSRCR1, and ΔC-t. (D) Extracellular secretion of endogenous LOXL4 upon TGF-β1 incubation in aortic endothelial cells as assessed by Western blotting with anti-LOXL4 antibody. The secretion of LOXL4-GFP from cells infected with the corresponding adenovirus is also shown (Media, supernatant; Cells, cell extract). Results are representative of three experiments.

Fig 9

LOXL4 expression leads to matrix remodeling. (A) Type I collagen solution was mixed with medium from cells left under basal conditions, exposed to TGF-β1, or infected with LOXL4 or null adenoviruses, and fibrillar collagen formation was monitored by confocal reflection microscopy. Collagen fiber number and size were counted in at least three random fields and represented as number of fibers per range of size. (B) Lysyl oxidase activity as assessed by H₂O₂ Amplex determination in supernatants from BAEC with or without TGF-β1 stimulation or infected with Ad-LOXL4 and in HEK 293 cells treated with or without doxycycline (Dox). A positive control with recombinant human LOXL2 is also shown. (C) Increased collagen IV deposition and assembly by BAEC infected with adeno-LOXL4 compared with adeno-null visualized by immunofluorescence microscopy. Results are representative of at least three experiments.