Osteopontin alters endothelial and valvular interstitial cell behaviour in calcific aortic valve stenosis through HMGB1 regulation

Abstract

OBJECTIVES

Calcific aortic valve stenosis (CAVS) is an important clinical problem predominantly affecting elderly individuals. Studies suggest that the progression of CAVS is actively regulated with valve endothelial injury leading to inflammation, fibrosis and calcification. The aim of this study was to delineate the possible regulatory role of osteopontin (OPN) on high-mobility group box 1 (HMGB1) function and the associated inflammatory and fibrotic response in CAVS.

METHODS

Aortic valve leaflets were collected from CAVS patients undergoing aortic valve replacement (n = 40), and control aortic valve leaflets were obtained from heart transplant recipients (n = 15). Valves and plasma were analysed by quantitative real-time polymerase chain reaction (PCR), immunohistochemical staining and Western blot. Recombinant OPN or neutralizing OPN antibody was added to cultured endothelial and valvular interstitial cells (VICs), and cell proliferation scores and HMGB1 expression were assessed.

RESULTS

CAVS valves had a decreased total percentage of VICs but increased numbers of infiltrating macrophages relative to control valves. RT-PCR studies showed higher expression of OPN, the inflammatory cytokine tumour necrosis factor-alpha as well as markers of fibrosis, tissue inhibitor of matrix metalloproteinase 1 and matrix metalloproteinase 2 in CAVS valves. Elevated expression of OPN was also observed in plasma of CAVS patients compared with controls. HMGB1 was detected in the secretory granules of cultured valve endothelial and VICs derived from CAVS valves. The addition of exogenous OPN inhibited the proliferation of cultured endothelial and VICs from CAVS valves and was associated with the extracellular expression of HMGB1, whereas neutralizing OPN had the opposite effect.

CONCLUSIONS

We conclude that altered OPN expression in CAVS affects cellular HMGB1 function inducing cytoplasmic translocation and secretion of HMGB1 in endothelial cells and VICs, thus indicating a regulatory role for OPN in the progression of CAVS through alteration of HMGB1 function.

Osteopontin, HMGB1, Calcific aortic valve stenosis, Endothelial cells, Valvular interstitial cells

INTRODUCTION

Calcific aortic valve stenosis (CAVS) is an important clinical problem affecting 2.8% of adults over 75 years of age in the developed world [1]. The most effective treatment remains aortic valve replacement, yet this does not address the mechanism(s) underlying the pathological remodelling that occurs during CAVS progression. Multiple biological pathways are responsible for aortic valve degeneration, with studies suggesting that the progression of CAVS is triggered by injury to the valve's endothelial cell lining and is actively regulated [2]. In addition to endothelial cell loss, injury leads to infiltration of circulating inflammatory cells, extracellular matrix deposition and concomitant activation of proinflammatory cytokines. Identifying markers that play a role in the pathogenesis of CAVS is therefore crucial in understanding the progression of this disease and may reveal potential therapeutic targets for its treatment.

Osteopontin (OPN) was originally linked to bone mineralization, but has since been shown to be a multifunctional proinflammatory cytokine with important roles in promoting inflammation and tissue remodelling including fibrosis and angiogenesis [3]. It has been implicated in both acute and chronic inflammatory processes including wound healing, autoimmune disease and atherosclerosis [4]. OPN negatively regulates mineral deposition and has been shown to be necessary for the differentiation and activity of myofibroblasts formed in response to the profibrotic cytokine transforming growth factor-1 [5]. Furthermore, high levels of OPN have been demonstrated in both the tissue and plasma of patients with aortic valve sclerosis and stenosis [6, 7]. Elevated plasma OPN levels and dephosphorylation of circulating OPN [8] have been associated with both the presence and severity of CAVS.

High-mobility group box 1 (HMGB1) is a highly conserved, ubiquitous protein found in the nucleus and cytoplasm of virtually all cell types [9]. It is passively released from necrotic cells and actively secreted by inflammatory cells, such as macrophages, T-cells, neutrophils and dendritic cells [10], whereas it functions as a proinflammatory cytokine through induction of proinflammatory mediators. There is evidence that HMGB1 contributes to the chronic inflammation and progression of vascular atherosclerosis, which has a similar pathophysiology to atherosclerotic CAVS [11]. In addition, HMGB1 has been shown to directly promote mesenchymal cell differentiation into an osteoblast-like phenotype to induce bone formation in vitro [12].

HMGB1 and OPN have similar regulatory processes and are both able to shuttle between the nucleus and cytoplasm [13, 14]. The function of HMGB1 necessitates its mobilization from the nucleus to the cytoplasm where it is released into the extracellular space, subsequently mediating inflammation. Lenga et al. [5] have suggested a functional association between HMGB1 and OPN with HMGB1 being present in the focal adhesions of wild-type fibroblasts but not detectable in OPN-null cells [15], indicating that OPN is required for the recruitment of HMGB1 following its cytoplasmic translocation. The goal of this study was to delineate the possible regulatory role of OPN on HMGB1 function in CAVS and, in particular, their interactions and involvement in valve pathological remodelling.

METHODS

Patients and sample collection

Calcified human aortic valve specimens were collected from patients undergoing aortic valve replacement surgery (n = 40; median age 74.5; 14 females and 26 males). Medical records were reviewed to exclude patients with a history of endocarditis, rheumatic heart disease or bicuspid valves. Additionally, peripheral blood was withdrawn before anaesthesia/surgery and 24 h following intervention. Control plasma was obtained from healthy volunteers who gave written and informed consent to participate in the study. Control aortic valves were obtained from the explanted hearts of patients undergoing transplantation (n = 15; median age 54; 6 females and 9 males) who exhibited no evidence of aortic valve pathology. Control aortic valves were assessed preoperatively by echocardiography and were included only if the valve area was normal and the mean pressure gradient across the valve was nil. The tri-leaflets were separated and collected into three solutions: 10% buffered formalin, RNAlater (Ambion, Austin, USA) and Dulbecco's Modified Eagle Media (DMEM; Lonza, USA) containing 10% foetal calf serum (FCS) (Invitrogen, Mt Waverley, Australia). Studies were conducted with the approval of the Institutional Research and Ethics Committee (EC28111) at The Prince Charles Hospital and written informed consent was obtained from all patients.

Immunohistochemistry

Aortic valve sections (5 μm) were dewaxed, rehydrated in a series of ethanol and water, and stained overnight at 4°C for primary antibodies against OPN (rabbit polyclonal; Abcam, Cambridge, MA, USA; 1 : 250) and HMGB1 (rabbit polyclonal; Abcam; 1 : 250). Staining was also performed for CD68 (mouse monoclonal; Santa Cruz Biotechnology, Dallas, USA; 1 : 100) to identify cells of macrophage lineage. The Vector universal ABC secondary antibody kit (Vector Laboratories) was used to link the primary antibody to the chromogen for 1 h. A positive reaction was detected with 3,3′-diaminobenzidine tetrahydrochloride (DAB; Sigma-Aldrich, St Louis, USA) which produces a brown colour at the site of the reaction. Sections were counterstained with haematoxylin, dehydrated in ethanol and xylene, and mounted in DePex (ThermoFisher Scientific, Scoresby, Australia). Negative controls were processed as described but without application of primary antibody. Endothelial cells (ECs) and valvular interstitial cells (VICs) isolated from aortic valves were plated on coverslips and grown to confluence. Cells were then fixed in methanol, and a primary antibody against HMGB1 (rabbit polyclonal; Abcam; 1 : 250) was applied and incubated overnight at 4°C. Cell phenotype was confirmed by staining ECs with von Willebrand factor (rabbit polyclonal; DakoCytomation; 1 : 50) and myofibroblasts with smooth muscle α-actin (mouse monoclonal; Sigma-Aldrich; 1 : 500).

Collagen staining

Aortic valve sections (7 μm) were dewaxed and rehydrated in a series of ethanol and water before being stained with 0.1% Sirius Red (BDH Laboratory Supplies) in saturated picric acid for 1 h at room temperature to identify collagen types 1 and 3. After washing in water, they were placed in 0.1 N HCl for 2 min followed by a second wash in water. The sections were then dehydrated using ethanol and xylene before being permanently mounted in Depex (Lomb Scientific). Collagen type 1 was identified morphologically as thick fibre located in the area of cell loss typical of fibrosis. Collagen type 3 was identified as fine fibre interstitially where no pathological remodelling occurred.

Quantitative analysis of macrophages

Images for immunohistochemistry were photographed at a magnification of ×250 and visualized using the AxioVision 4.7 Image Analysis system (Carl Zeiss, North Ryde, NSW, Australia). Semi-quantitative analysis was performed by an independent investigator blinded to the slide identity with 20 image fields photographed for each sample. The number of CD68⁺ cells and total number of cells, including VICs, was counted, and the percentage of CD68⁺ cells calculated using AxioVision.

RNA extraction and real-time RT-PCR

Total RNA was isolated from aortic valves using TRIzol (Ambion) and samples were purified using the RNeasy Mini Kit (Qiagen, VIC, Australia). Each sample was treated with DNase (Ambion) and subsequently analysed on the Nanodrop 1000 to determine the RNA concentration and quality spectrophotometrically at 260 nm (A260/280 nm, ThermoFisher Scientific). First-strand cDNA was synthesized from 500 ng RNA using random primer and AMV Reverse Transcriptase (Bio-Rad, Hercules, USA). Real-time quantitative PCR was performed for the genes OPN and HMGB1, as well as for extracellular matrix markers matrix metalloproteinase 2 (MMP2), MMP9, tissue inhibitor of matrix metalloproteinase 1 (TIMP1) and TIMP2 and the proinflammatory cytokine tumour necrosis factor (TNF)-α using the Taqman Viia7 Real-Time PCR system with Taqman Gene Expression PCR Master Mix (Life Technologies, Grand Island, USA). Reactions were performed in triplicate and a no-template and reverse transcription negative control was included for each primer set. The Qbase PLUS program was used to identify the most stably expressed housekeeping genes, and all data were subsequently normalized to a geomean of two genes PGK1 and RPLP0.

Western blotting

Albumin and IgG were depleted from each plasma sample using the Albumin and IgG Depletion SpinTrap, according to the manufacturer's instructions (GE Healthcare, NSW, Australia). Samples were electrophoresed on a 4–12% Bis Tris gel, transferred onto a PVDF membrane and blocked with 5% skim milk powder. The membranes were then washed in TBS/Tween and incubated overnight at 4°C with primary antibodies to OPN (rabbit polyclonal; Abcam1 : 1000), HMGB1 (rabbit polyclonal; Abcam 1 : 500) and myosin IIa (rabbit polyclonal; Cell Signaling Technology, Boston, MA, USA; 1 : 2000) as a loading control. The blot was then developed and imaged using the ImageQuant 350 phosphorimager (GE Healthcare). Densitometric quantification of band intensity was analysed using the ImageJ software and normalized to myosin IIa expression. Immunoblotting was also performed on conditioned media from ECs and VICs and the membranes were incubated with primary antibodies against HMGB1 (rabbit polyclonal; Abcam; 1 : 500) as described above. In the absence of a specific and well-accepted protein loading control for secreted protein, we normalized expression to the non-treated group.

Tissue culture

Aortic valve leaflets were digested in 0.5% trypsin (Invitrogen) for 15 min to detach the ECs lining the valve. The valve leaflets were washed in DMEM/10% FCS, and the supernatants centrifuged at 1000 rpm, 5 min, 4°C. Cells were then plated on cell culture dishes in an EGM-2 endothelial growth media bullet kit (Lonza, USA) at 37°C with 5% CO_2. To establish cultured VICs, the leaflet was then cut into ∼1 mm³ pieces and incubated in DMEM/10% FCS at 37°C with 5% CO₂. VICs and ECs were grown to confluence, and cells plated on coverslips for immunohistochemistry. Additional ECs and VICs were plated from CAVS and controls in 24-well plates at a density of 10 000 cells/well in triplicate and left for 48 h. Recombinant OPN (1 and 5 µg/ml) or neutralizing OPN antibody (R&D Systems, Minneapolis, USA; 5 µg/ml) was added and cells were left for 24 h before being harvested. Cell proliferation scores were assessed and conditioned medium was obtained for immunoblotting. ECs and VICs plated on coverslips were used for immunohistochemical staining.

Statistical analysis

Clinical parameters are expressed as median with ranges given. Gene expression data are expressed as mean ± SEM with statistical differences between the groups assessed using a non-parametric Mann–Whitney U-test. A P-value of ≤0.05 was considered statistically significant. Statistical analyses were performed using the GraphPad Prism 5 software.

RESULTS

Patient demographics

The characteristics of CAVS and control heart transplantation patients are summarized in Table 1. CAVS patients had a higher median age when compared with control heart transplantation patients. The mean pressure gradient across the aortic valve and left ventricular ejection fraction was also higher in CAVS patients versus controls.

Table 1:

Demographics of calcific aortic valve stenosis (CAVS) and control heart transplantation patients

	CAVS	Heart transplantation
Age	60–90 (median 74.5) years	19–67 (median 54) years
Gender	14 females	6 females
	26 males	9 males
Surgical indication	Severe aortic stenosis	Dilated cardiomyopathy (n = 9)
		Ischaemic cardiomyopathy (n = 3)
		Restrictive cardiomyopathy (n = 1)
		Hypertrophic obstructive cardiomyopathy (n = 1)
		Chemotherapy-induced cardiomyopathy (n = 1)
Mean pressure gradient across valvea	49.7 ± 13 mmHg	Nil
LVEF (%)a	61.7 ± 10.5	25.6 ± 11.6

LVEF: left ventricular ejection fraction.

^aData are shown as mean ± SD.

Expression of cellular infiltrate in calcific aortic valves

Macrophages (CD68⁺ cells) were observed to be distributed diffusely throughout the entire layer of the valves from both CAVS patients and controls. The result of quantitative analysis showed that there were increased numbers of macrophages in CAVS compared with control valves (22.5 ± 3.9 vs 14.7 ± 4.3% per field; P < 0.05; n = 12 CAVS and n = 5 controls). The number of VICs in CAVS valves, as identified by negative CD68 staining, was lowered by 8% compared with controls (85.3 ± 4.3 vs 77.2 ± 3.8% per field; Fig. 1A and B).

Figure 1:

CD68, OPN and HMGB1 staining of aortic valve leaflets in CAVS patients (A, C and E) and controls (B, D and F). The percentage of macrophages (CD68⁺ cells) was increased in CAVS versus control valves (P < 0.05). There was reduced OPN expression in CAVS compared with control valves, whereas the expression of HMGB1 was detected in the cytoplasmic vesicles of VICs and extracellular HMGB1 was observed in areas of fibrosis. OPN: osteopontin; HMGB1: high-mobility group box 1; CAVS: calcific aortic valve stenosis; VICs: valvular interstitial cells.

Picrosirius red staining in CAVS valves showed accumulation of type 1 collagen below the endothelial cell layer on both the ventricular and aortic sides of the valve. This was associated with a considerable increase in the total valve layer thickness when compared with control valves. Various degrees of calcification were found underneath the type 1 collagen at the centre of the CAVS valve (data not shown).

Tissue expression of osteopontin and high-mobility group box 1 in calcific aortic valves

OPN expression was observed predominantly in extracellular areas of the valves. Strong OPN expression was found diffusely distributed in the extracellular areas in control valves and was associated with layers of type 3 collagen but not with type 1 collagen.

Overall, OPN expression was reduced in CAVS compared with control valves (Fig. 1C and D) and was lacking in the areas of fibrosis associated with collagen type 1 accumulation and calcification.

Macrophages had variable intracellular expression of HMGB1 in both controls and CAVS valves. Intracellular HMGB1 was expressed in the cytoplasmic vesicles of VICs in CAVS valves, whereas extracellular HMGB1 expression was detected in fibrotic areas associated with cell remnants (Fig. 1E and F). Negligible HMGB1 expression was found in VICs of control valves. There was intracellular expression of HMGB1 in a small subset of ECs from CAVS valves, whereas there was negligible expression in controls.

Plasma expression of osteopontin and high-mobility group box 1

Western blot analysis showed that there was a trend for higher plasma levels of OPN in CAVS patients compared with controls (Fig. 2; P = 0.05) with both pre- and post-surgery at similar levels. The levels of HMGB1 pre-surgery were lower when compared with control plasma, although this was not statistically significant, and were further reduced 24 h post-surgery (P < 0.05).

Figure 2:

Representative Western blot images of OPN and HMGB1 expression in plasma (A) (n = 6 CAVS and n = 6 controls) with densitometric quantitation normalized to myosin IIa expression (B). Data show that the levels of OPN were higher (P = 0.05) compared with that of controls, whereas pre-surgery HMGB1 levels were reduced in CAVS patients compared with controls. Post-surgery levels of HMGB1 were further lowered when compared with controls (P < 0.05). OPN: osteopontin; HMGB1: high-mobility group box 1; CAVS: calcific aortic valve stenosis. *P < 0.05 vs control.

Gene expression of proinflammatory and fibrosis markers

In addition to measuring the gene expression levels of HMGB1 and OPN, we also measured the proinflammatory cytokine TNF-α and extracellular matrix proteins and their inhibitors. A summary of the assays-on-demand used for real-time RT-PCR can be seen in Supplementary Table 1. Real-time RT-PCR showed that the levels of OPN (P < 0.05), TNF-α (P < 0.05), MMP2 (P < 0.05) and TIMP1 (P < 0.05) were higher in CAVS than in control valves (Fig. 3). Although there was a trend towards higher expression levels of HMGB1, MMP9 and TIMP2 in CAVS versus control valves, these were not statistically significant (Supplementary Fig. 1).

Figure 3:

Relative gene expression data for OPN (A), MMP2 (B), TIMP1 (C) and TNF-α (D) in aortic valve tissue. Data are presented as mean ± SEM (n = 40 CAVS and n = 15 controls). *P < 0.05 vs control. OPN: osteopontin; MMP: matrix metalloproteinase; TIMP: tissue inhibitor of matrix metalloproteinase; TNF-α: tumour necrosis factor-alpha; SEM: standard error of the mean; CAVS: calcific aortic valve stenosis.

HMGB1 and OPN expression in cultured endothelial cells and valvular interstitial cells

Immunohistochemical analysis showed positive HMGB1 staining in the secretory granules of both cultured ECs and VICs derived from CAVS valves (Fig. 4A). Negligible expression of HMGB1 was observed in VICs and ECs derived from control valves. OPN expression was reduced in ECs and VICs derived from CAVS compared with control valves (Fig. 4B).

Figure 4:

Representative immunohistochemical staining of ECs and VICs, showing HMGB1-positive staining in the secretory granules of CAVS valves (A). Negligible HGMB1 staining was detected in control valves. OPN expression was reduced in CAVS valves (B) compared with controls. OPN: osteopontin; ECs: endothelial cells; VICs: valvular interstitial cells; HMGB1: high-mobility group box 1; CAVS: calcific aortic valve stenosis.

Cell proliferation and translocation of HMGB1 with addition of recombinant OPN and OPN-neutralizing antibody

Addition of recombinant OPN at 1 (P < 0.05) and 5 µg/ml (P = 0.05) inhibited proliferation of cultured VICs from CAVS valves compared with non-treated cells, although this was not significant in ECs (P = 0.0713; Table 2). There was no dose-dependent difference between the 1- and 5-µg/ml cell proliferation numbers. Immunohistochemical staining showed HMGB1 translocation from the nucleus to the cytoplasm in ECs (Fig. 5A and B) and VICs with recombinant OPN (Fig. 5G and H). In controls, addition of recombinant OPN led to increased cell proliferation of ECs compared with non-treated cells (P < 0.05), whereas there was no change in VICs. There was negligible expression of HMGB1 in non-treated ECs, whereas the addition of recombinant OPN led to slightly increased cytoplasmic expression (Fig. 5D and E). There was no nuclear expression in VICs, and this was unchanged with the addition of OPN (Fig. 5J and K).

Table 2:

Cell proliferation counts in ECs and VICs following addition of recombinant OPN or OPN-neutralizing antibody

Treatment	CAVS		Controls
	ECs	VICs	ECs	VICs
Non-treated	356 667 ± 75 719	315 000 ± 77 782	396 667 ± 28 868	180 000 ± 81 854
1 µg OPN	213 333 ± 68 069	176 667 ± 11 547*	683 333 ± 80 208*	130 000 ± 17 321
5 µg OPN	383 333 ± 111 505	163 333 ± 40 415*	500 000 ± 81 854*	133 333 ± 58 595
Non-treated	150 000 ± 36 056	256 667 ± 58 595	196 667 ± 37 859	160 000 ± 20 000
OPN antibody	303 333 ± 26 166*	156 667 ± 5 774*	233 333 ± 87 369	320 000 ± 52 915*

Values are mean ± SD. n = 3 CAVS and controls each.

ECs: endothelial cells; VICs: valvular interstitial cells; OPN: osteopontin; CAVS: calcific aortic valve stenosis.

*P < 0.05 vs non-treated.

Figure 5:

Immunohistochemical staining of ECs in CAVS (A–C) and controls (D–F) and VICs in CAVS (G–I) and controls (J–L). Translocation of HMGB1 from the nucleus (A and G) to the cytoplasm (B and H) can be seen with the addition of 1 µg/ml of recombinant OPN. When OPN is neutralized, HMGB1 expression remains in the nucleus in ECs (C), whereas VICs had both nuclear and cytoplasmic expression (I). In control cells, there was negligible expression of HMGB1 in non-treated ECs, whereas the addition of recombinant OPN led to slightly increased cytoplasmic expression (D and E). There was no nuclear expression in VICs and this was unchanged with the addition of OPN (J and K). When OPN was neutralized, the nuclear and cytoplasmic expression of HMGB1 was increased in ECs (F) while remaining unchanged in VICs (L). ECs: endothelial cells; CAVS: calcific aortic valve stenosis; VICs: valvular interstitial cells; HMGB1: high-mobility group box 1; OPN: osteopontin.

In CAVS valves, neutralizing OPN led to increased EC numbers compared with non-treated cells (P < 0.05), whereas cell numbers were reduced in VICs (P < 0.05). To confirm that our cell count results were due to changes in the level of cell proliferation and not due to apoptosis, we conducted immunohistochemical studies using Ki67 and caspase-3. The results support the cell counts, suggesting that reduced proliferation was not due to increased apoptosis but inhibition of mitosis and vice versa. A greater nuclear expression of HMGB1 was observed in ECs (Fig. 5C) when OPN was neutralized, whereas VICs had both nuclear and cytoplasmic expression (Fig. 5I). Neutralizing OPN in controls led to VIC proliferation versus non-treated cells (P < 0.05), whereas ECs were unchanged. The nuclear and cytoplasmic expression of HMGB1 was increased in ECs (Fig. 5F) while remaining unchanged in VICs (Fig. 5L). A summary of the cell proliferation results is given in Table 2.

High-mobility group box 1 expression in conditioned media

Western blot analysis of conditioned media obtained from tissue culture experiments showed that the addition of recombinant OPN to ECs from CAVS valves led to an increase in HMGB1 expression at the doses used in this experiment (Fig. 6). The addition of an OPN-neutralizing antibody lowered the expression of HMGB1. The levels of HMGB1 in VICs did not change with the addition of recombinant OPN or neutralizing antibody, although the expression levels were reduced compared with controls.

Figure 6:

Representative Western blot images of HMGB1 expression in conditioned media from ECs and VICs (A) (n = 3 CAVS and n = 3 controls) with levels normalized to non-treated (NT) groups (B). Data show that the addition of recombinant OPN to ECs leads to an increase in the secretion of HMGB1, although this is not dose-dependent. Addition of an OPN-neutralizing antibody decreases the secretion of HMGB1. The levels of HMGB1 in VICs do not change with the addition of recombinant OPN or neutralizing antibody, although the expression levels are decreased compared with controls. HMGB1: high-mobility group box 1; ECs: endothelial cells; VICs: valvular interstitial cells; OPN: osteopontin; NT: non-treated; CAVS: calcific aortic valve stenosis.

DISCUSSION

We investigated the expression of HMGB1 and OPN in CAVS. This study is the first to report that OPN regulates HMGB1 activation and its subsequent release. Immunohistochemical studies found reduced extracellular expression of OPN in CAVS valves as well as diminished expression in cultured ECs and VICs isolated from calcific aortic valves compared with controls. Previous studies have shown intense OPN staining in aortic valve sclerosis and stenosis, and that this was localized to the proximity of the biomineralized lesions in the fibrous layer of the leaflets [16]. In our study, however, OPN was transiently expressed in CAVS and was mostly expressed extracellularly in the surrounding interstitial cell layer and associated with type 3 collagen in the resident cell layer of control valves. These discrepancies are potentially due to differences in the sampling area as well as different stages of disease progression with our study utilizing samples from advanced stages of biomineralization with a valve area of <1 cm². The previous study had a broader inclusion criterion including patients with an aortic valve area of <2 cm². Given the reduced OPN expression in ECs and VICs in CAVS valves together with the observed loss of VICs, this suggests that these resident cells may be the source of OPN. Previous investigators have reported the presence of phenotypic changes in resident cells in CAVS [17]. It is likely that, in this study, phenotypic changes in the resident cells might also have led to altered expression of OPN and potentially affected its extracellular release. HMGB1 was found to be associated with macrophage infiltration and areas of collagen accumulation and calcification. Cultured VICs and ECs isolated from CAVS valves showed HMGB1 expression in the secretory granules compared with mainly nuclear localization in controls. Cell counts revealed an increase in the number of macrophages, suggesting an important role of macrophages as a source of HMGB1. HMGB1 is known to be actively secreted by inflammatory cells [9] and to signal through the receptor for advanced glycation end-products or via Toll-like receptors, TLR2 and TLR4, expressed on inflammatory cells, ECs and fibroblasts [18]. Previous studies have shown upregulation of TLR2 and TLR4 expression in human aortic VICs [19, 20].

Gene expression studies revealed higher expression of OPN as well as TNF-α, MMP2 and TIMP1 in CAVS compared with control valves, indicating that these markers may be involved in tissue remodelling and CAVS development. The levels of OPN have been shown to be higher in patients with aortic valve sclerosis and stenosis with OPN splicing variants shown to be differentially expressed during CAVS development [16]. Higher levels of TIMP1 and MMP2 in CAVS valves indicate possible dysregulation of these genes in aortic stenosis and subsequent matrix remodelling. MMPs contribute to the pathophysiological changes associated with aortic valve stenosis [21]. Previous studies have shown upregulated synthesis of MMP2, MMP9 and TIMP1 in degenerative valve disease [22]. Significantly higher levels of TNF-α were observed in the calcified aortic valve. TNF-α has been shown to induce the synthesis and secretion of HMGB1 by monocytes and macrophages [23]. This, in turn, might be associated with the induction of degenerative enzymes including MMP2 and TIMP1, resulting in changes in proliferative activity, accumulation of collagen type 1 and ultimately fibrosis and calcification.

We investigated the circulating plasma levels of OPN by Western blot and found they were higher in CAVS patients compared with controls, although there was no difference in the pre- and post-surgery levels. Our results confirm the findings of previous studies where OPN levels in plasma were found to be raised in CAVS pre-surgery [16, 24]. This is surprising given the reduced expression found both extracellularly in the valve and in cultured ECs and VICs derived from the valves of CAVS patients. This is likely due to a possible difference in the source of cells that produced OPN in circulation compared with tissue origin. Furthermore, these conflicting results for OPN may be dependent on post-translational modification such as phosphorylation and, therefore, the role of OPN might not be solely dependent on circulating levels of OPN but rather modification of the molecule [24]. Grau et al. [16] showed that OPN splicing variants, including OPN-a, OPN-b and OPN-c, are differentially expressed during CAVS progression. We speculate that retaining a certain level of OPN in the valve may be beneficial for tissue homeostasis and therefore, its depletion may contribute to disease progression. The elevated mRNA could also be a negative feedback controlling mechanism due to failure of OPN to be retained in the tissue. Contrary to previous findings reporting that HMGB1 concentrations in the serum were significantly greater at 24 h following cardiopulmonary bypass [25], our study revealed the levels of HMGB1 were lowered pre-surgery compared with controls and were further reduced 24 h post-surgery. If the source of HMGB1 is circulating inflammatory cells, this observation may reflect a decrease in their number pre- and post-surgery.

To further elucidate the effect of changes in OPN availability on HMGB1 function, we conducted cell culture studies to examine the effect of exogenous addition or neutralization of OPN. Addition of recombinant OPN inhibited proliferation of cultured ECs and VICs from CAVS valves compared with non-treated cells. Subsequent immunohistochemical staining showed HMGB1 translocation from the nucleus to the cytoplasm implying active extracellular release. This result was confirmed by immunoblots of conditioned media suggesting increased secretion by ECs but not by VICs. Neutralizing OPN had the opposite effect with proliferation of ECs, retention of HMGB1 in the nucleus and reduced expression in conditioned media, indicative of the inhibition of HMGB1 secretion. These results suggest that OPN regulates HMGB1 secretion by mediating HMGB1 translocation from the nucleus to the cytoplasm and its active release into the extracellular space. This modified cellular response to OPN could be the result of changes to resident cells after disease progression or an inherent characteristic of the cell. Further investigations are needed to confirm this mechanism. Additional in vitro experiments are planned with subcellular fractionation to separate out the nuclear and cytoplasmic lysate.

We believe we have discovered a novel regulatory role for OPN in CAVS whereby it activates HMGB1 and controls its release into circulation. Thus, targeting these molecules as markers in CAVS progression could lead to new discoveries in CAVS treatment or even prevention. Intervention studies in an animal model of CAVS would determine whether normalization of OPN expression is associated with controlled HMGB1 release and prevention of CAVS. Based on the outcome of animal studies, clinical trials can be proposed to determine whether drug intervention reduces or prevents CAVS development.

CONCLUSION

Our data suggest that EC injury in CAVS induces inflammatory and resident valve cells to secrete OPN. In addition, we found that OPN upregulation subsequently leads to HMGB1 relocation into the secretory granules of resident valve cells. The relocation of HMGB1 might successively trigger active and/or passive secretion, lead to a proinflammatory state and contribute to the ensuing tissue remodelling. Gene expression levels of OPN, TNF-α, MMP2 and TIMP1 in CAVS valves were higher, indicating active pathological remodelling in the form of fibrosis. Further studies are needed to determine the specific role of OPN in CAVS and whether there is post-translational modification of the molecule such as phosphorylation. In conclusion, this study has demonstrated the presence of increased OPN expression in CAVS. We found that OPN regulates HMGB1 activation and its subsequent release from inflammatory and valve resident cells, suggesting a regulatory role for OPN in HMGB1 function in pathological remodelling in CAVS. Targeting this molecular pathway may lead to a successful drug for its treatment.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

Funding

This work was supported by The Prince Charles Hospital Foundation (grant numbers NR2010-213 and MS2013-04). John Fraser also acknowledges the funding provided in support of this research by the Office of Health and Medical Research, Queensland Health.

Conflict of interest: none declared.

ACKNOWLEDGEMENTS

The authors acknowledge cardiac surgeons Thomson, Jalali and Kang for their help with collection of human valve specimens.