Materials and methods
Isolation and culture of human and bovine disc cells
Human degenerated IVD fragments were obtained from patients undergoing surgical intervention (Additional file 1: Table S1), after informed consent and approval by ethics committee of North and Central Switzerland (EKNZ). Degeneration grade was determined by magnetic resonance imaging (MRI) and based on surgeons expertise, according to the Pfirrmann’s modification of the Thompson classification. Five-score grades from healthy (grade I) to the most advanced degenerated (grade V) disc are defined in terms of sequential changes to MRI features: disc height, distinction between AF and NP, brightness and uniformity of the NP [15].
Human degenerated IVD fragments were digested with 0.05% Collagenase Type-2 (Worthington—Bioconcept, Allschwil, Switzerland) in DMEM/F12 + GlutaMAX with 5% fetal bovine serum (FBS) (Gibco—Paisley, UK) for 6 h at 37 °C. As healthy human IVD (control) were inaccessible for ethical reason, we isolated discs from bovine tail obtained from a local slaughterhouse. Healthy bovine disc fragments were digested with 0.3% pronase (Sigma, Buchs, Switzerland) in DMEM/F12 + GlutaMAX with 5% FBS, for 1 h at 37 °C, then further digested with Collagenase overnight at 37 °C. Disc cells were expanded in monolayer culture in DMEM/F12 + GlutaMAX supplemented with 10% FBS (Gibco) and 5 ng/ml recombinant bFGF (Peprotech—LuBioScience, Lucerne, Switzerland) at 37 °C in a humid atmosphere containing 5% CO2.
RNA extraction and gene expression
IVD tissues stored at − 80 °C were disintegrated mechanically while still frozen and lysed in RNA Lysis Buffer (Aurum Total Mini Kit, Bio-Rad—Cressier, Switzerland). Disc cells in monolayer were trypsinized and the pellet was dissolved in RNA Lysis Buffer. Total RNA was extracted (Aurum Total Mini Kit, Bio-Rad) and used for synthesis of cDNA (SuperScript VILO cDNA Synthesis Kit, Invitrogen—LuBioScience). Template cDNA was mixed with PCR reaction solution (IQ SYBR Green Supermix, Bio-Rad) containing 0.25 μM specific primers (Additional file 2: Table S2). Quantitative PCR (qRT-PCR) reactions were carried out in duplicate in 96-well plates (Bio Rad) for 40 amplification cycles, followed by melting curve analysis. Relative quantification was calculated based on the 2−ΔΔCt method and normalized to GAPDH.
Immunoblotting
Flow-through washes from the RNA extraction procedure were stored overnight at − 20 °C to allow protein precipitation, and then centrifuged at 10,000g for 15 min at 4 °C. Protein pellets were washed three times with 70% ethanol and resuspended in CelLytic M Buffer (Sigma) with Protease Inhibitor Cocktail (Sigma). In Western blot assays, proteins were separated by SDS-PAGE using Tris–Glycine 5–15% gradient gels (Bio-Rad) with a Protein Standard 10–250 kDa (Bio-Rad) and transferred onto a nitrocellulose membrane using the semi-dry Trans-Blot Turbo system (Bio-Rad). After blocking of non-specific sites with 5% milk (Rapilait, Migros, Switzerland) in PBS, membranes were incubated overnight at 4 °C with primary antibodies diluted in 5% milk: anti-VEGFR1 1:1000 (AF321 R&D System—Abingdon, UK); anti-soluble VEGFR1 1:50 (36–1100 Thermo Scientific—LuBioScience); anti β-Actin 1:10,000 (AC-15 Novus—LuBioScience). Secondary antibody HRP-conjugated (anti-mouse and anti-rabbit 1:10,000, anti-goat 1:20,000—Bethyl, LuBioScience) was incubated 1 h at room temperature in 5% milk-PBS and detected with chemiluminescence substrate (LumiGlo Reserve, KPL—Bio Concept, Allschwil, Switzerland). Acquisition was performed with digital SLR camera (Nikon D600—Nikon, Zürich, Switzerland) [16].
Western blot quantification was performed with ImageJ 1.49v.
Microarray data analysis
Two microarray data sets (GSE15227 and GSE23130) [17, 18] were downloaded from the Gene Expression Omnibus (GEO) using GEOquery, an R bioconductor package. These datasets have been generated from AF tissues of human degenerated IVD classified according to Thompson grading system. A quality check was performed using ArrayQualityMetrics to identify the arrays with poor quality. It was found that there were no notable deviations amongst 26 chips out of 38 chips. The intensities were re-calculated and normalized using Robust Multichip Average (RMA), Guanine Cytosine Robust Multi-Array (GCRMA) and Variance Stabilizing (VSN) methods. Having observed that the normalization could bring the intensity distribution of the selected 26 chips to similar characteristics, we decided to use these chips in our analysis (Additional file 3: Table S3). We then filtered out unwanted information (e.g. genes without entrez information, duplicated entrez gene identifiers etc.) together with genes having low variance, which would not pass the statistical tests for differential expression. Subsequently the filtered data sets were processed using limma R package to identify deferentially expressed genes. The results presented are based on VSN normalization.
Statistical analysis
For statistical analysis, we used non-parametric Mann–Whitney–Wilcoxon U test for independent variables. Data analysis was performed with SPSS version 24.0 for Windows (SPSS Inc.). Significance was indicated as *p < 0.05. After publication the data will be shared on Research Gate.
Results
Human degenerated IVD express membrane and soluble VEGFR1
Immunoblotting assay with anti-VEGFR1 antibody against the N-terminus region of the protein showed the expression of the full-length membrane form (mVEGFR1 ~ 200 kDa) in human degenerated (Fig. 1a) and healthy bovine (Fig. 1c) disc cells. An antibody against the C-terminal region of VEGFR1 soluble form showed a band of ~ 130 kDa (sVEGFR1) in human degenerated (Fig. 1a) and healthy bovine (Fig. 1c) disc cells. Both antibodies showed a band at ~ 60 kDa. β-Actin was used as a control. Western blot quantification showed no significant differences between mVEGFR1 and sVEGFR1 in human degenerated IVD (n = 3—Fig. 1b) and healthy bovine discs (n = 2—Fig. 1d). Relative protein levels represent the average of the area (square pixels) normalized on β-Actin. qRT-PCR analysis with specific primers (Additional file 2: Table S2) designed in the unique regions of VEGFR1 gene to discriminate between different isoforms, showed that mVEGFR1 and sVEGFR1 were comparably expressed in human degenerated IVD tissues (n = 11) and disc cells cultures (n = 11) (Additional file 4: Figure S1).
Unchanged VEGF pathways and abnormal vascularization in IVD degeneration process
Comparative analysis of AF tissues (n = 26) expression profiles from two microarray data sets [17, 18] showed that sVEGFR1 and mVEGFR1, along with other members of the classic signalling vascularization pathway [19] (Fig. 2a), were unchanged at the level of transcript abundance through the degeneration grades (Fig. 2b). An exception was VEGFA, which exhibited a fluctuating expression between degeneration grades (Fig. 2b).
On the other hand, the expression levels of genes regulating abnormal alternative vascularization, such as hypoxia-inducible factor-1A (HIF1A) and High-Temperature Requirement A Serine Peptidase 1 (HTRA1) were significantly increased in degenerated AF grade IV and V (Fig. 3a). There were also increased expression levels of some of the components of the interactome of HIF1A and HTRA1 proposed by STRING database (Fig. 3b). Meta-analysis gene expression data are given in Additional file 5: Table S4 and Additional file 6: Table S5. The qRT-PCR validation of six selected genes regulating abnormal vascularization (Fig. 3c) showed significant increased expression levels of Ubiquitin C (UBC) in severe degenerated AF tissues (grade IV and V; n = 5), compared to grade II and III (n = 5). HIF1A, HTRA1, 40S Ribosomal Protein S27A (RPS27A) and Ubiquitin A-52 (UBA52) showed higher expression levels in degeneration grade IV and V, but the results were not significant.
Discussion
In this study, we observed that gene expression in the classic vascular endothelial growth factor (VEGF) vascularization pathway is preserved across human intervertebral disc (IVD) degeneration stages and we propose that alternative vascularization pathways may be involved with the pathological progression of disc degeneration.
The degeneration of IVD is associated with vascularization and innervation [6]. VEGF promotes vasculogenesis binding to the full-length transmembrane vascular endothelial growth factor receptor 1 (mVEGFR1), while shorter soluble forms of this receptor (sVEGFR1) behave as competitive inhibitors of vascularization. Since sVEGFR1 has the essential role to preserve corneal avascularity in diverse mammals [12], we hypothesized that human IVD expresses the decoy sVEGFR1 as an additional way to maintain its avascularity and the expression of sVEGFR1 and mVEGFR1 may change during degeneration process. We expected that healthy avascular discs prevalently express the inhibitor soluble form, while the vascularized degenerated discs express more the membrane form. Contrary to our expectations, we showed that both forms were similarly expressed in human degenerated IVD tissues and disc cell cultures. By immunoblot, we observed the expression of mVEGFR1 (~ 200 kDa) and sVEGFR1 (~ 130 kDa), both isoforms well described in literature [10, 20,21,22]. It has been demonstrated that VEGFR1 can undergo proteolytic fragmentation, resulting in the formation of soluble form and cytoplasmic fragment of ~ 60 kDa [23]. We observed, by microarray meta-analysis, unchanged expression levels of sVEGFR1 and mVEGFR1 through the IVD degeneration grades, as well as of the other components of classic vascularization pathways [19], except for the fluctuating transcript levels expressed by VEGFA. VEGFR1 together with VEGFA are the imperative upstream components in a highly studied signalling pathway that regulates angiogenesis [19, 24,25,26]. The mechanism of neovascularization mediated by VEGF and its receptors has been closely correlated with inflammation, chronic back pain and accelerated IVD degeneration [27,28,29,30,31,32]. However, our meta-analysis rejected the initial hypothesis and showed that the classic vascularization signalling pathway is constitutively expressed across disc degeneration stages.
This leaves the option that other vascularization pathways may be involved in the pathological vascularization progression during IVD degeneration. Interestingly, the meta-analysis and our qRT-PCR validation experiments revealed number of genes-either known or proposed to regulate abnormal vascularization—with increased expression levels in severely degenerated discs. Above all, hypoxia-inducible factor-1A (HIF1A), which has been shown to regulate VEGF [33, 34] and induce angiogenesis [35], High-Temperature Requirement A Serine Peptidase 1 (HTRA1), which has a potential role in IVD degeneration [36, 37] and abnormal vascularization [38, 39], and Ubiquitin C (UBC), which was hypothesized to have a crucial role in inhibiting cell proliferation of annulus fibrosus in IVD degeneration [40].
In conclusion, this study showed that the classic VEGF vascularization pathway is unchanged across disc degeneration, advancing that decoy sVEGFR1 does not have a major role in protecting from disc degeneration process.