Co‐culture of human fibroblasts, smooth muscle and endothelial cells promotes osteopontin induction in hypoxia

Abstract Arteriovenous fistulas (AVFs) are the preferred vascular access for haemodialysis of patients suffering from end‐stage renal disease, a worldwide public health problem. However, they are prone to a high rate of failure due to neointimal hyperplasia and stenosis. This study aimed to determine if osteopontin (OPN) was induced in hypoxia and if OPN could be responsible for driving AVF failure. Identification of new factors that participate in remodelling of AVFs is a challenge. Three cell lines representing the cells of the three layers of the walls of arteries and veins, fibroblasts, smooth muscle cells and endothelial cells, were tested in mono‐ and co‐culture in vitro for OPN expression and secretion in normoxia compared to hypoxia after silencing the hypoxia‐inducible factors (HIF‐1α, HIF‐2α and HIF‐1/2α) with siRNA or after treatment with an inhibitor of NF‐kB. None of the cells in mono‐culture showed OPN induction in hypoxia, whereas cells in co‐culture secreted OPN in hypoxia. The changes in oxygenation that occur during AVF maturation up‐regulate secretion of OPN through cell‐cell interactions between the different cell layers that form AVF, and in turn, these promote endothelial cell proliferation and could participate in neointimal hyperplasia.


| INTRODUC TI ON
Renal failure is a major public health problem with increasing incidence every year. Patients with end-stage renal disease require either kidney transplantation or haemodialysis to sustain life.
Arteriovenous fistula (AVF) is the preferred vascular access for haemodialysis. 1 However, only 50% of AVF remain functional six months after creation, which increases the morbidity-mortality of patients. 2 Several clinical factors seem to play a role in the dysfunction of AVFs, including female gender, age, or diabetes. Histologically, dysfunction most frequently results from neointimal hyperplasia (NH), which is responsible for stenosis. 3  Rupture of the vasa vasorum during surgical dissection may contribute to the induction of hypoxia, thereby stabilizing HIF. 5 HIFs are dimeric protein complexes that consist of an α-subunit (HIF-1α or HIF-2α) and a β-subunit (HIF-1β or HIF-2β), 6 and are major regulators of cellular adaptation to hypoxia. HIF-1α is expressed ubiquitously, whereas HIF-2α is primarily detected in endothelial cells but is also selectively highly expressed in a limited number of tissues. 7 There is increasing evidence supporting the contribution of the HIF pathway, both the protective and destructive effects, to the pathogenesis of diseases affecting the vascular wall including atherosclerosis, 8,9 arterial aneurysms, [10][11][12] pulmonary hypertension, [13][14][15] vascular graft failure, 4,16-18 chronic venous diseases 19,20 and vascular malformation. 21,22 Furthermore, increased expression of VEGF-A, a target HIF-1α gene, may contribute to NH, through increased proliferation of smooth muscle cells. A recent study has shown that reducing VEGF-A gene expression during AVF formation reduces NH. 23 Although HIFs are involved in the regulation of the oxygen homeostasis, NF-κB, a major transcription factor that responds to cellular stress, is also activated by hypoxia. 24 The most abundant cytoplasmic form of the NF-κB complex is an inactive heterotrimeric form composed of p50 and p65 subunits, and the inhibitor IKB-α.
Stimulus-induced degradation of IKB-α is critical for nuclear translocation of NF-κB and induction of transcription of target genes. 25 In rat models, overexpression of NF-κB was found following vascular injury and correlated to thickening of the intima compared to that of control vessels. 26 Osteopontin (OPN) is a SIBLING protein (Small Integrin Binding Ligand N-linked Glycoproteins), which was initially identified as a bone matrix protein that links bone cells to the extracellular matrix. 27 OPN exists in two isoforms, a secreted (sOPN) and an intracellular form (iOPN), that have distinct biological functions. 28 At the protein level, OPN has a molecular weight of about 60 kD. This protein undergoes multiple post-translational modifications by phosphorylation and glycosylation variables that can explain the previously described variability in the apparent molecular weights (from 25 to 75 kD). 29 OPN is involved in multiple processes including tissue remodelling, regulation of cellular immunity, pathological chronic inflammatory processes, carcinogenesis as well as cardiovascular diseases. 30 OPN is also involved in several vascular diseases promoting angiogenesis, in parallel with vascular endothelial growth factor (VEGF), through enhanced endothelial cell migration, proliferation and subsequent formation of capillaries, which are essential requirements for the process of angiogenesis.
In particular, it has been found to be expressed in vascular smooth muscle cells of human restenotic lesions and stenotic vascular lesions. 31 Significant association between the level of plasma circulating OPN and atheroma plaque formation has been reported. 32 Moreover, high OPN levels in patients with stenosis have been described after coronary angioplasty compared to patients without stenosis. 33 Interestingly, Hall et al 34 have shown a 40-fold increase in OPN expression in the early stages of AVF maturation in a mouse model of AVF. In addition, constitutive overexpression of OPN in mice was found to result in increased neointima formation after cuffing of the femoral artery. 35 Finally, structural changes were noted on the OPN-null background including disorganized collagen and increased vessel wall compliance. 36 Altogether, these data suggest that OPN may play a role in the development of vascular stenosis associated with excessive intimal proliferation. Reactive oxygen species (ROS) 37 and hyperglycaemia 38 induce in vivo expression of OPN in pancreatic epithelial cells, and also in vascular smooth muscle cells. Even though up-regulation of OPN in hypoxia has been shown to be dependent 39,40 or independent 41 of HIF-1 regulation, OPN expression is enhanced under hypoxia through different mechanisms, 42 leading to co-expression with VEGF. 43 Moreover, we reported that the oxidative stress generated during early maturation of an AVF stabilized the HIF-1α protein and thus activated HIF-target genes such as Vegf-A, Nox-2 and Ho- 1. 44 Given that changes in the oxygen concentration occur during surgical creation of an AVF in parallel to OPN overexpression, we suggested that OPN is induced by hypoxia and thus could contribute to the failure of AVF maturation due to NH and juxtaanastomotic stenosis. We postulated that regulation of OPN through silencing of HIFs or NF-kB could promote AVF maturation. Finally, we describe an all-human in vitro cell model of co-culture that induces OPN in hypoxia but only under conditions of cell-cell interaction.

| Tissue collection
Studies on human samples received authorization from regulatory boards and the local ethics committee. Informed consent was obtained from patients regarding the collection of samples and data.
Deidentified matured patent human AVF samples were provided by Yale Vascular Surgery. Samples were obtained after ≥6 months of haemodialysis during surgical revision of the fistula resulting from severe anastomotic stenosis. 45 The samples analysed are of a segment of normally remodelling patent vein from the fistulae. Control veins were obtained from renal disease patients at the time of initial AVF creation. Sample procurement with informed consent was approved by Human Investigation Committee of Yale University Institutional Review Board HIC No. 1005006865. Tissues were fixed in formalin and embedded in paraffin.

| Pharmacological inhibitors and chemicals
Recombinant human OPN was from R&D system and was used at 1 µg/ mL. Inhibition of the NF-κB activity was conducted with a specific inhibitor (inhibitor of IKK2 (AS602868), iNF-κB), a gift from Dr JF Peyron.

| RNAi transfection
Cells were plated at 70% confluence and transfected the following day using Lipofectamine RNAiMAx TM (Thermo Fisher Scientific) according to the manufacturer's protocol, at a 50 nmol/L final concentration of RNAi. The set of RNAi sequences (Eurogentec) targeting human were as follows: SIMA 5′-GCCACAAGCAGUCCAGAUU-3′.

| Immunoblotting
Protein immunodetection was performed as previously ///described. 49 Briefly, cells were lysed in 1.5X SDS buffer, and if exposed to hypoxia, they were lysed while inside the hypoxic chamber. Proteins (40 μg) were separated on 7.5% SDSpolyacrylamide gels and transferred onto polyvinylidene difluoride membranes (Millipore). Blots were blocked in 5% milk in Tris-HCl/NaCl buffer and incubated with antibodies at 4°C overnight. Immunoreactive signals were revealed with horseradish peroxidase (HRP)-conjugated antibodies (Promega) using the Pierce TM ECL Western blotting system (Thermo Fisher Scientific).
The antibody against HIF-1α was produced in our laboratory and used at 1:1000. The antibody against HIF-2α (NB100-122) was purchased from Novus Biologicals (Littleton) and used at 1:1000.
The antibody against OPN (AF 14-33OP) was purchased from R&D System and used at 1:500. The antibody against β-tubulin (T4026; Sigma-Aldrich) was used at a 1:2000 dilution as a loading control.

| RNA extraction and quantitative real-time PCR
For patient samples, total RNA was extracted with the RNeasy

| Immunofluorescence
Human vein and AVF samples were sectioned at 5 μm and mounted onto glass slides. After heating overnight at 55°C, deparaffinization was accomplished through a series of xylene and graded ethanol soaks.
Sections were heated in citric acid buffer (pH 6.0) at 100°C for 10 min-

| Statistics
All values are the means ± SEM. Statistical analyses were performed using Student's t test in Microsoft Excel. The P values are indicated.
All categorical data used numbers and percentages. Quantitative data were presented using the median and range or mean.
Differences between groups were evaluated using the chi-square test for categorical variables and Student's t test for continuous variables. Analyses were performed using SPSS 16.0 statistical software (SPSS Inc.). All statistical tests were two-sided, and P-values < .05 indicated statistical significance whereas P-values between .05 and .10 indicated a statistical tendency.

| Intracellular OPN (iOPN) is not induced in hypoxic cells
First, we evaluated the level of expression of OPN in AVF patients.
Both expressions of OPN mRNA ( Figure 1A) and OPN immunoreactivity ( Figure 1B) in the AVF were significantly increased, compared to control veins, confirming our previous results and the potential role of OPN in AVF maturation. 34 Knowing that OPN has two isoforms, intracellular OPN (iOPN) and soluble OPN (sOPN), we first characterized iOPN by immunoblot in different cellular models. Recombinant OPN (recOPN) was used to quantitate the expression ( Figure 1C).
Human bone marrow mesenchymal cells osteoblasts ( Figure 1D) and LNCap cells ( Figure 1E) were examined as controls, since these two models are known to express OPN. Moreover, OPN is known as a biomarker for prostate cancer and for its role in tumour progression. 51 The antibody against OPN detected recOPN at the concentra- These results suggest that iOPN expression is below the detectable level of 16.6 ng, ( Figure 1C) in NHF, and weakly expressed in HUVSMC and HUVEC.

| Secreted OPN (sOPN) is not induced in hypoxic cells
We then tested for sOPN in these cells using an ELISA. Two different kits were used, one that measured sOPN in nanograms ( Figure 2A) and one in picograms ( Figure 2B)

| mRNA expression of OPN is not induced in hypoxic cells
As we were unable to detect hypoxic induction of either iOPN or sOPN, we tested an even more sensitive technique to measure OPN.
Using real-time qPCR, we quantified the endogenous expression of OPN in normoxia and hypoxia ( Figure 3). As a control for hypoxia, the glucose transporter 1 (GLUT1) mRNA expression was evaluated ( Figure 3A,3,3). Two-to threefold induction at the mRNA level of GLUT1 was observed in NHF ( Figure 3A) and HUVEC ( Figure 3E). Similarly, a threefold induction was observed at the mRNA level for OPN in hypoxia in HUVEC with siHIF-1α, but also with siHIF-2α and siHIF-1/2α. However, siRNA to OPN did not affect the mRNA level of OPN under these conditions; therefore, these results ( Figure 3F) are not specific to OPN expression and reflect non-specific amplification due to undetectable OPN expression in HUVEC ( Figure 2D).
We thus confirmed that mRNA expression of OPN is not induced in hypoxia in the three cell lines tested.

| Co-culture favours OPN induction in hypoxia
Finally, given that a potential link between OPN and AVF has been

| D ISCUSS I ON
The main objective of this study was to determine whether OPN was induced in hypoxia and thus could be responsible for driving the failure of AVF maturation due to NH and juxtaanastomotic stenosis. We first used cell mono-layers grown on tissue culture plastic to examine hypoxic production of OPN observed in AVF, which is the approach used by the majority of research groups. Mono-cultures are less complex to perform but certainly less adapted to, and not very representative of, the physiological extracellular microenvironment present in humans.
While the expression of OPN has been shown to be increased during early AVF maturation 54

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicting interests.