- Email: [email protected]
RANKL promotes osteoblastic activity in vascular smooth muscle cells by upregulating endothelial BMP-2 release
The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
Contents lists available at ScienceDirect
The International Journal of Biochemistry & Cell Biology journal homepage: www.elsevier.com/locate/biocel
RANKL promotes osteoblastic activity in vascular smooth muscle cells by upregulating endothelial BMP-2 release Colin Davenport a,∗ , Emma Harper a , Hannah Forde a , Keith D. Rochfort b , Ronan P. Murphy c,d , Diarmuid Smith e , Philip M. Cummins a,d a
School of Biotechnology, Dublin City University, Dublin, Ireland Conway Institute, University College Dublin, Dublin, Ireland c School of Health and Human Performance, Dublin City University, Dublin, Ireland d Centre for Preventive Medicine, Dublin City University, Dublin, Ireland e Department of Academic Endocrinology, Beaumont Hospital, Dublin, Ireland b
a r t i c l e
i n f o
Article history: Received 20 January 2016 Received in revised form 18 May 2016 Accepted 19 June 2016 Available online 23 June 2016 Keywords: RANKL Osteoprotegerin Calcification BMP-2 Endothelial
a b s t r a c t Introduction: Receptor activator of nuclear factor kappa beta-ligand (RANKL) is thought to promote vascular calcification (VC) by inducing osteoblastic behaviour in vascular smooth muscle cells (VSMC) in an ill-defined process. The present study assessed whether RANKL affects pro-osteoblastic paracrine signalling between human aortic endothelial cells (HAEC) and human aortic smooth muscle cells (HASMC) using both conditioned media transfer and cell co-culture experimental approaches. Methods and results: For initial experiments (6-well format), HAEC-conditioned media was harvested following 72 h exposure to RANKL, and transferred to reporter HASMCs with/without noggin, an inhibitor of pro-osteoblastic bone morphogenetic protein (BMP) paracrine signalling. In further experiments, HAECs and HASMCs were co-cultured within the CellMax® Duo, a perfusing bioreactor unit that mimics the flow-mediated co-interaction of these cells within the arterial wall, and RANKL was added to the perfusing media for 72 h. At the conclusion of each experiment markers of osteoblastic activity were measured in HASMCs, including alkaline phosphatase (ALP) activity, mRNA levels of ALP and Runx2, as well as BMP2 and BMP-4 concentrations. RANKL increased BMP-2 release from HAECs, while exposure of HASMCs to RANKL-treated HAEC-conditioned media induced osteoblastic behaviour in HASMCs, an effect prevented by noggin. Within the CellMax® Duo bioreactor, the addition of RANKL to the intraluminal HAECs also produced an increase in BMP-2 and increased osteoblastic behaviour within the co-cultured HASMC population. Conclusions: RANKL promotes VC by inducing BMP-2 release from HAECs, which in turn appears to act in a paracrine fashion on the adjacent HASMC population to increase osteoblastic activity. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Vascular calcification (VC) exerts a number of adverse effects on the cardiovascular (CV) system and exhibits a strong positive
Abbreviations: ALP, alkaline phosphatase; BMP-2/4, bone morphogenetic protein 2/4; CV, cardiovascular; ELISA, enzyme-linked immunosorbent assay; ELS, extraluminal space; HAEC, human aortic endothelial cell; HASMC, human aortic smooth muscle cell; ILS, intraluminal space; mRNA, messenger ribonucleic acid; OPG, osteoprotegerin; RT-qPCR, quantitative real-time polymerase chain reaction; RANKL, receptor activator of nuclear factor kappa-beta ligand; Runx2, runt-related transcription factor 2; SEM, standard error of the mean; SMC, smooth muscle cell; TNF-␣, tumour necrosis factor alpha; VC, vascular calcification. ∗ Corresponding author at: School of Biotechnology, Dublin City University, Glasnevin, Dublin 9, Ireland. E-mail address: [email protected] (C. Davenport). http://dx.doi.org/10.1016/j.biocel.2016.06.009 1357-2725/© 2016 Elsevier Ltd. All rights reserved.
correlation with overall rates of CV morbidity and mortality in multiple patient populations (Abedin et al., 2004; Wexler et al., 1996). Depending on its location within the vasculature, VC can inhibit cardiac valve function, decrease arterial compliance and increase the risk of dissection following angioplasty (Demer and Tintut, 2008). Efforts to inhibit or even reverse this process have, to date, been largely unsuccessful, a failure attributed primarily to a lack of understanding of the VC process at the cellular level (Johnson et al., 2006). Whilst the pathophysiology of VC likely differs depending upon its location within the vasculature, as well as the effects of patient factors such as renal failure or diabetes, one unifying feature of VC appears to be the emergence of osteoblastic behaviour in vascular cell populations such as vascular smooth muscle cells (VSMCs) (Byon and Chen, 2015). The emergence of osteoblastic behaviour within vascular cells represents a potential target for treatments designed to prevent VC, and in recent years
172
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
the promoters and inhibitors of this transformation have been the subject of a significant body of research, with increasing interest focused on two interlinked proteins, namely osteoprotegerin (OPG) and receptor activator of nuclear factor kappa-beta ligand (RANKL), with the former acting as a decoy receptor that blocks the biological actions of the latter (Vitovski et al., 2007). Although the actions of OPG and RANKL have been described in considerable detail within the skeletal system where they control osteoclast activation and bone turnover, the role of these regulatory proteins in VC has yet to be properly elucidated (Wu et al., 2013). Both proteins are known to be present within the vascular wall, with RANKL expression particularly associated with areas of calcification (Higgins et al., 2015). In animal models, OPG knockout mice have been shown to develop early and severe VC, atherogenic mice treated with OPG demonstrated a reduction in plaque calcification burden, and OPG was shown to prevent vitamin D-induced VC in rats, findings that collectively suggest that unopposed RANKL activity (due to lack of OPG) promotes VC (Bucay et al., 1998; Morony et al., 2008; Price et al., 2001). In this regard, both Panizo et al. (2009) and Kaden et al. (2004) have reported that RANKL may act directly on VSMCs to induce osteoblastic behaviour. Their findings, however, were subsequently contradicted by Olesen et al. (2012) and Byon et al. (2011), who reported no direct effect from RANKL on VSMCs. More recently, preliminary findings have suggested that whilst RANKL does affect VSMC calcification, it may possibly do so by inducing the endothelial monolayer to produce and release pro-osteoblastic paracrine signals such as members of the bone morphogenetic protein family (BMP) (Osako et al., 2010). It is noteworthy, however, that many of the studies in this field have suffered from significant limitations in their methodology. These include the use of pro-calcifying culture media (a potential confounder that directly induces osteoblastic transformation in VSMCs), as well as the use of calcium deposition as a primary endpoint of their studies (an unreliable endpoint with poor reproducibility) (Olesen et al., 2012). With regards to model oversights, the in vitro mechanistic studies to date on VC have been mainly performed using “static” cell monocultures, which differ significantly from the arterial wall microenvironment in vivo, incorporating as it does constant crosstalk between vascular endothelial and smooth muscle cells (via both direct contact and paracrine signalling), in addition to the hemodynamic effects of blood flow on these cell populations (Chiu et al., 2009; Eddahibi et al., 2006). Taking into account these limitations, the aim of the present study was to characterize more robustly the effects of RANKL on the emergence of osteoblastic behaviour in human aortic smooth muscle cells (HASMC) in vitro. This included an assessment of both the direct effects of RANKL on HASMCs, as well as an assessment of its effects on the production and release of pro-osteoblastic BMP-2 and BMP-4 from human aortic endothelial cells (HAEC). Moreover, in order to enhance the relevance of our monoculture studies to the in vivo state, our VC model was extended to include a HAEC:HASMC co-culture within a perfused artificial capillary system (CellMax® Duo) that reproduces the 3D architecture and physiological hemodynamic shear flow of an arterial vessel (Janke et al., 2013; Redmond et al., 1995).
2. Materials and methods 2.1. Materials Unless otherwise stated, all reagents were purchased from Sigma-Aldrich (Dublin, IRL). Recombinant human RANKL, antiOPG neutralizing antibody (OPG-Ab), and noggin (BMP inhibitor) were purchased from R&D Systems (Mineapolis, MN, USA). Tumor necrosis factor-alpha (TNF-␣) was purchased from Merck Milli-
pore (Danvers, MA, USA). Primers were purchased from Eurofins Genomics (Ebersburg, DEU). 2.2. Cell culture HAECs and HASMCs were purchased from Promocell Gmbh (Heidelberg, Germany), as were the culture media for both cell types. The HAECs (positive for von Willebrand factor and CD31) were received from two donors, a 28 year old Caucasian male (utilized in the majority of the experiments) and a second batch from a 23 year old Caucasian female (utilized for dose-response curves for RANKL and OPG release secondary to RANKL exposure). The HASMCs (positive for ␣-actin) were received from two donors, one batch from a 23 year old Caucasian male (utilized in the majority of the experiments) and a second batch from a 32 year old Caucasian male (utilized in 2 RANKL-treated bioreactors and 1 control bioreactor experiment in the co-culture experiments). HAECs were cultured in endothelial cell growth medium with the following supplements; fetal calf serum (0.05 ml/ml), endothelial cell growth supplement (0.004 ml/ml), epidermal growth factor (10 ng/ml), heparin (90 g/ml) and hydrocortisone (1 g/ml). This media was also supplemented with penicillin (100 IU/ml) and streptomycin (100 g/ml). HASMCs were cultured in smooth muscle cell growth medium supplemented with antibiotics in the same manner as the HAEC media, in addition to fetal calf serum (0.05 ml/ml), epidermal growth factor (0.5 ng/ml), basic fibroblast growth factor (2 ng/ml) and insulin at 5 g/ml. As insulin exerts a potent inhibitory effect on OPG production (Davenport et al., 2015), this supplement was omitted for a minimum of 2 passages prior to the commencement of experiments). Basic fibroblast growth factor (bFGF), which has been reported as promoting OPG production (albeit at higher concentrations), was retained in the media to maintain acceptable levels of growth and viability for experimental purposes (Zhang et al., 2002). Cells were maintained in a humidified incubator at 37 ◦ C and 5% CO2 . Passages 5–9 were routinely used for experimental purposes. Cell number and viability were routinely measured using the advanced detection and accurate measurement (ADAMTM ) cell counter (Digital Bio, Seoul, KOR), both for seeding density purposes and to allow normalization of results where necessary. This process involved the digital analysis of cells on custom Accuchip slides following the addition of propidium iodide with or without a membrane permeabilising solution, providing total numbers of both viable and non-viable cells. Cell culture experiments were conducted in three formats: (i) monoculture; (ii) conditioned media; and (iii) perfused co-culture. 2.3. Monoculture experiments The effects of RANKL on osteoblastic activity in HASMCs was initially investigated. HASMCs were grown to confluency in standard 6-well dishes and subsequently treated for 72 h with RANKL (0 and 50 ng/ml) in the absence or presence of OPG-neutralizing antisera (optimized dose of 2.5 g/ml). The concentration of 50 ng/ml RANKL was utilized to overcome endogenous OPG production from HASMCs, which we measured as reaching 33 ng/ml after 72 h in culture. OPG-neutralizing antisera has previously been reported as exerting no effects on HASMCs in isolation (Osako et al., 2010). Posttreatment, media and cells (mRNA) were harvested for analysis (alkaline phosphatase/ALP activity and osteoblastic gene expression, namely ALP and runt-related transcription factor 2 (Runx2). Harvested media was centrifuged at 1000 rpm (4 ◦ C) for 10 min to remove any cellular debris. All samples were stored at −80 ◦ C with all relevant assays being performed within one month as part of a single freeze-thaw cycle. In further experiments, the effects of RANKL on BMP-2 and −4 production and/or secretion by HASMCs and HAECs were also investigated. Both cell types were grown to
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
confluency in 6-well dishes. Cells were treated with RANKL as follows: HAECs −0, 5 and 25 ng/ml, 0–72 h; HASMCs −0 and 50 ng/ml, 0–72 h. HAECs were also exposed to TNF-␣ (100 ng/ml) as a positive control for BMP production. Post-treatment, media and cells (mRNA, protein) were harvested for analysis (BMP-2/4 mRNA and protein levels). 2.4. Conditioned media experiments The effect of HAEC paracrine signalling on HASMC osteoblastic activity was next investigated. HAECs were seeded into permeable (0.4 m pore) transwell culture inserts (Merck Millipore, MA, USA) and grown to confluency. The HAEC-seeded inserts were then placed into standard 6-well dishes containing fresh HAEC culture media to create separate abluminal (apical) and subluminal (basolateral) compartments. Cells were abluminally treated for 72 h with RANKL (5 ng/ml) in the absence and presence of the BMP inhibitor, noggin (2.5 g/ml), after which conditioned media was harvested from the subluminal compartment. The concentration of 5 ng/ml RANKL was utilized to overcome endogenous OPG production from HAECs, which we measured as reaching 0.8 ng/ml after 72 h in culture. Conditioned media (containing HAEC-derived paracrine factors) was then transferred directly onto confluent HASMC reporter cell cultures for a further 72 h. Post-treatment, HASMC media and mRNA were harvested for analysis (alkaline phosphatase activity and osteoblastic gene expression). 2.5. Perfused co-culture experiments The effects of RANKL on HAEC-HASMC paracrine signalling leading to the emergence of osteoblastic activity in HASMCs was investigated in a perfused CellMax® Duo co-culture system (Spectrum Laboratories, Rancho Dominguez, CA, USA). Each perfused bioreactor comprised a parallel bundle of ProNectin® -coated semipermeable capillaries, with the intraluminal space (ILS) representing the vessel interior, and the extraluminal space (ELS) representing the medial layer of the vessel. Within this system, HAECs were seeded and grown upon the intraluminal surface of the semipermeable capillaries, while HASMCs were seeded and grown on the outside of the semi-permeable capillaries, in the ELS. Media flowed through the intraluminal aspect of the capillaries. Upon connecting the media reservoir, gas permeable tubing and bioreactor unit, the various components of the CellMax® Duo became a single sealed, perfusing circuit maintained under shear flow within a dedicated humidified incubator at 37 ◦ C and 5% CO2 . As both HAECs and HASMCs were grown within the CellMax® Duo for these experiments, and only one growth media could be used within the system at any given time, a 1:1 ratio of HAEC and HASMC media was utilized. HASMCs, with their slower growth rate, were inoculated (1 × 107 cells) into the ELS first as previously described by Redmond et al. (1995) and allowed to grow for a minimum of 3 weeks at an intraluminal media flow rate of 20 ml/min [19]. Next, HAECs were inoculated into the ILS (2.5 × 107 cells) and allowed to grow and form a monolayer over the intraluminal capillary walls for a further 2 weeks at an intraluminal media flow rate of 5 ml/min. As these were relatively long periods of incubation, the proliferation and metabolic parameters of the inoculated HASMCs and HAECs within each bioreactor unit were ascertained by daily measurements of glucose consumption and lactate production using an YSI 2300 STAT Plus Analyzer (YSI Incorporated Life Sciences, Ohio, USA). During co-culture establishment, media in the reservoir was replaced whenever glucose concentrations dropped by 50%. Experimentation commenced once a plateau of glucose consumption and lactate production rates had been reached. At this point, the intraluminal media flow rate through the capillaries was adjusted to generate physiological levels of laminar shear stress equiva-
173
lent to 10 dyn/cm2 . During the experiments, CELLMAX® units being exposed to RANKL had their capillary perfusion halted for 20 min every 12 h while RANKL was added to the ILS at 50 ng/ml (sufficient RANKL to overcome endogenous HAEC and HASMC OPG production), after which capillary perfusion was recommenced. CELLMAX® systems that were not exposed to RANKL were treated in the same manner but without RANKL being added to the ILS. The experiments continued for 72 h in total with a final circulating RANKL concentration of 5.1 ng/ml at the end of the experiment. Post-treatment, culture media was harvested from both the ILS and ELS compartments for measurement of ALP activity and BMP2 concentrations, whilst HASMCs were harvested from the ELS for analysis of osteoblastic gene expression. 2.6. Quantitative real-time RT-PCR (RT-qPCR) Post-treatment, HAECs and HASMCs were harvested for extraction of total RNA and analysis of gene expression as previously described by Rochfort et al. (2014) using an ABI 7900 HT Fast RT-PCR System (Applied Biosystems/Life Technologies, Paisley, UK). Each cDNA sample was assayed in triplicate and results analyzed by the comparative CT method. S18 was routinely used for normalization purposes. Primer pairs were screened for correct product size by 1% agarose gel electrophoresis and underwent melt-curve analysis for primer-dimers. S18 (250 bp): Forward 5 -cagccacccgagattgagca3 ; Reverse 5 -tagtagcgacgggcggtgtg-3 ; ALP (312 bp): Forward 5 -gcctggctacaaggtggtg-3 ; Reverse 5 -ggccagagcgagcagc-3 ; Runx2 (297 bp): Forward 5 -ggtaccagatgggactgtgg-3 ; Reverse 5 -gaggcggtcagagaacaaac-3 ; BMP-2 (199 bp): Forward 5 caagccaaacacaaacagcg-3 ; Reverse 5 -ccaacgtctgaacaatggca-3 . Annealing temperatures for S18, ALP, Runx2, and BMP-2 were 62 ◦ C, 58 ◦ C, 59 ◦ C, and 57.3 ◦ C, respectively. 2.7. Enzyme-linked immunosorbent assay (ELISA) The concentrations of BMP-2 and BMP-4 in media samples were measured using commercially available Duoset ELISA kits (R&D Systems, Minneapolis, USA) in conjunction with a protocol previously described by our group (Martin et al., 2014). Both kits exhibited intra-assay coefficients of variation of <5%. BMP-2/4 concentrations were normalized for viable HAEC and HASMC numbers as appropriate. 2.8. Alkaline phosphatase activity assay ALP activity, a critical marker of osteoblastic activity, was measured in media samples using the QuantichromTM Kit (BioAssay Systems, California, USA) according to manufacturer’s protocol. This was a colorimetric kinetic assay based on the principle that ALP (if present in the sample) will convert p-nitrophenyl phosphate (a colourless solution) to p-nitrophenyl (a yellow coloured solution) and phosphate, with the colour change measured via microplate reader at baseline and again at 4 mins. The inter-assay coefficient of variation for this kit was <5%. 2.9. Statistical analysis Data were analyzed using GraphPad Prism version 4.03 (Graphpad Software, California, USA), and are expressed in mean ± SEM (normality was tested for via the Kolmogorov-Smirnov test). Statistical comparisons between controls versus treatment groups were made by analysis of variance in conjunction with a Dunnett’s posthoc test for multiple comparisons. A Student’s unpaired t-test was also employed for pair-wise comparisons. A value of p ≤ 0.05 was considered significant.
174
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
3. Results 3.1. RANKL does not directly induce osteoblastic activity in HASMCs The ability of RANKL to directly induce osteoblastic activity in HASMCs was initially monitored. Following treatment of HASMCs for 72 h with 50 ng/ml RANKL, we observed no significant changes in mRNA levels for either ALP (Fig. 1B) or Runx2 (Fig. 1C), nor was cell viability affected. We also observed no significant change in ALP enzymatic activity in HASMC-conditioned media following RANKL treatment (Fig. 1D). To control for the possibility that the lack of effect of RANKL may be due to it becoming bound to OPG constitutively released from HASMCs, an excess of OPG neutralizing antibody (OPG-Ab, 2.5 g/ml) was included with the RANKL treatment. In this context, the OPG-Ab had no influence on the lack of effect of RANKL (Fig. 1B–D). 3.2. RANKL induces BMP-2 release from HAECs (but not HASMCs) In our next series of experiments, the effect of RANKL on BMP release from HAECs and HASMCs was investigated. RANKL was incubated for 72 h with HAECs (5 ng/ml) and HASMCs (50 ng/ml), after which the conditioned media was harvested for analysis of BMP-2 and BMP-4 release by ELISA. In response to RANKL treatment, we observed a significant increase in release of BMP-2 (but not BMP-4) from HAECs (Fig. 2A and B, Supplementary Fig. S1 in the online version at DOI: 10.1016/j.biocel.2016.06.009). The release of BMP-2 from HAECs appeared to occur due to the release of pre-formed BMP-2 from these cells, as exposure to RANKL did not significantly affect BMP-2 mRNA levels or intracellular protein levels (Supplementary Fig. S2 in the online version at DOI: 10.1016/ j.biocel.2016.06.009), whereas the positive control TNF-␣ induced both an increase in the release and production of BMP-2 (Supplementary Fig. S3 in the online version at DOI: 10.1016/j.biocel.2016. 06.009). By contrast, RANKL had no significant effect on release of either BMP from HASMCs (although a statistically insignificant upward trend in BMP-4 release was noted) (Fig. 2C and D). 3.3. RANKL treatment of HAECs can elicit osteoblastic activity in HASMCs—conditioned media model The ability of RANKL to induce release of pro-osteogenic factors from HAECs was next examined. HAECs cultured in transwell inserts were treated abluminally with 5 ng/ml of RANKL for 72 h, and HAEC-conditioned media subsequently harvested from the subluminal compartment. HASMCs were subsequently treated for 72 h with either unconditioned media (control) or with HAEC-conditioned media harvested under basal conditions (HCMB) or RANKL conditions (HCMR). Relative to basal conditions, RANKL treatment of HAECs significantly increased HASMC mRNA levels for ALP (Fig. 3A) and Runx2 (Fig. 3B), as well as ALP activity levels in HASMC media (Fig. 3C). HAEC-conditioned media experiments were subsequently repeated with RANKL in the absence and presence of 2.5 g/ml noggin, a BMP antagonist. In these experiments, noggin was found to completely block the RANKL-dependent increase in HASMC osteogenic gene expression (ALP, Runx2—Fig. 4A) and osteoblastic activity (ALP activity—Fig. 4B). 3.4. RANKL treatment of HAECs can elicit osteoblastic activity in HASMCs - perfused co-culture model The effects of RANKL on HAEC:HASMC paracrine signalling leading to the emergence of osteogenic activity in HASMCs was next investigated in a perfused CellMax® Duo co-culture sys-
tem that reproduces arterial 3D architecture and physiological hemodynamic shear flow (Fig. 5). Stable, confluent HAEC:HASMC co-cultures were established within the perfused capillary format under physiological laminar shear flow (10 dyn/cm2 ). Paired glucose and lactate consumption of a typical bioreactor unit is illustrated in Supplementary Fig. S4 in the online version at DOI: 10. 1016/j.biocel.2016.06.009. Intraluminal HAECs were maintained under untreated control conditions or were exposed to RANKL (50 ng/ml, twice a day for 72 h). In response to RANKL treatment of HAECs, mRNA levels of ALP and Runx2 were significantly elevated in HASMCs (Fig. 6B). In response to RANKL treatment of HAECs, elevated BMP-2 (but not BMP-4) levels, as well as elevated ALP activity levels, were also observed in the ECS media surrounding the HASMCs (located subluminally to the HAEC monolayer) (Fig. 6C).
4. Discussion The present study demonstrates a number of novel findings regarding the role of OPG and RANKL in the VC process. In addition to identifying and characterizing a paracrine relationship between ECs and SMCs that may help to explain the method by which OPG opposes VC in vivo, we have further confirmed this relationship in a perfusing circuit incorporating a co-culture of both vascular cell populations designed to simulate the vascular wall microenvironment. To the best of our knowledge, this paper is the first to examine the complex process of VC in an in vitro sheared co-culture model of this nature. With regards to the initial findings of this study, we demonstrated that RANKL, when applied directly to HASMCs, did not promote osteoblastic behaviour in terms of osteogenic gene upregulation (Runx2, ALP) or the secretion of ALP, an enzyme that is critical for the crystallization aspect of VC. While this relationship has been examined previously, the methodological approaches utilized in prior studies were limited in a number of important areas and conflicting data have been reported (Byon et al., 2011; Kaden et al., 2004; Olesen et al., 2012; Panizo et al., 2009). The vast majority of previous studies of this nature supplemented the cell media with phosphate in an effort to maximize observable differences in VC with RANKL. Phosphate, however, is known to be a potent inducer of VC and thus a potential confounder of these experiments. In addition, prior studies made routine use of calcium deposition as a primary endpoint, as opposed to gene and protein markers of osteoblastic activity. As illustrated by Olesen et al. (2012), calcium deposition is an unreliable endpoint with a high degree of variability between vascular cell donors, and furthermore typically requires the use of phosphate-containing media in the experiments. We also note that some authors did not take the endogenous production of OPG (with its ability to block the actions of RANKL) from HASMCs (and, to a lesser extent HAECs) into account (Candido et al., 2010; Moran et al., 2005). Finally, it is interesting to compare our own results with those of Yuan et al. (2011). This group reported that insulin induced osteoblastic behaviour in VSMCs via RANKL signalling, indicating that RANKL may directly promote osteoblastic change in these cells. These experiments, however, were performed on a specific subpopulation of VSMCs termed calcifying vascular cells, which are prone to adopting an osteoblastic phenotype spontaneously in culture, and this methodological difference in terms of cell populations studied may explain why we observed no apparent effect with RANKL when applied to a general heterogeneous HASMC population. Ultimately, our experimental approach was designed to test osteoblastic behaviour in HASMCs with a minimisation of potential confounders. As such, we did not utilize calcifying phosphate-rich media, and for our endpoints we measured established gene and protein markers of osteoblastic activity. Endogenous OPG production was overcome
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
175
Fig. 1. Direct effects of RANKL on osteoblastic activity in HASMCs. HASMCs were treated for 72 h with RANKL (50 ng/ml) ± OPG Ab (2.5 g/ml). Representative agarose gel insets verify HASMC mRNA expression and correct fragment sizes generated using the S18 (250 bp), Runx2 (297 bp), and ALP (312 bp) primers (A). After 72 h HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (B) and Runx2 (C). Histograms represent fold change in mRNA levels relative to untreated control HASMCs. HASMCs were also monitored for changes in ALP activity (D). Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
through the use of either neutralizing antibodies to OPG and/or the use of RANKL concentrations significantly in excess of those of endogenous OPG. It is within this setting that we report no evidence for a direct effect from RANKL on osteoblastic activity and VC in HASMCs. In the second part of this study we assessed the effects of RANKL on the production of BMP-2 and BMP-4 from both HAECs and HASMCs. Both of these pro-osteogenic proteins have been identified within the vascular wall in vivo and both are recognised promoters of osteoblastic activity in HASMCs (Abedin et al., 2004; Higgins et al., 2015). We noted that HAECs had a 10-fold higher constitutive production of both BMP-2 and BMP-4 in comparison to HASMCs, and that RANKL significantly increased the release of BMP-2 from HAECs. Interestingly, the increased release of BMP-2 was not accompanied by increased expression of BMP-2 mRNA or intracellular protein levels, as was previously reported by Osako et al. (2010). The contrasting results between our data and that of Osako et al. may have arisen as a result of differing methodologies in terms of cell culture conditions, as well as possible phenotypic heterogeneity between HAEC populations. Our results suggest that pre-formed BMP-2 may be released and secreted by HASMCs without an accompanying increase in intracellular production, although further experiments will be required to clarify this response. Using a conditioned media model by
exposing HASMCs to HAEC-conditioned media, we noted significant increases in osteoblastic activity. Stimulating the HAECs with RANKL enhanced this effect further. We then added noggin to the conditioned media derived from RANKL-treated HAECs, a protein which binds to BMP-2 and blocks its pro-osteoblastic activity (Krause et al., 2011). Upon doing so, we noted that the addition of this BMP-2 blocking agent reduced the subsequent development of osteoblastic behaviour in HASMCs to baseline levels. Ultimately, we conclude that BMP activity is necessary for HAEC-conditioned media to induce osteoblastic activity in HASMCs, and that induction of BMP-2 (but not BMP-4) release from HAECs likely represents the process by which RANKL promotes VC. In the final step of the present study, we utilized a perfused artificial capillary co-culture system (CellMax® Duo) to test our initial findings in an experimental model designed to approximate the conditions of the vascular wall, and in doing so overcome a number of important limitations of traditional monoculture and conditioned media experiments (Chiu et al., 2009; Eddahibi et al., 2006; Redmond et al., 1995). Using CellMax® Duo perfused bioreactors, we grew HAECs on the inner luminal surface of a semi-permeable artificial capillary within a capillary bundle format, with HASMCs growing on the outer surface of the porous capillary walls. In doing so, we more closely replicated the 3D vessel architecture and close proximity of these two cell populations in vivo, and facilitated
176
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
Fig. 2. Effect of RANKL on BMP-2/4 release from HAECs and HASMCs. RANKL was incubated for 72 h with HAECs (5 ng/ml) and HASMCs (50 ng/ml), after which media was harvested for analysis by ELISA. Histograms represent pg/104 cells of (A) BMP-2 and (B) BMP-4 released from HAECs, as well as (C) BMP-2 and (D) BMP-4 released from HASMCs, under untreated control and RANKL-treated conditions. Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
Fig. 3. Effects of HAEC-conditioned media on osteoblastic activity in HASMCs. HASMCs were treated for 72 h with HAEC-conditioned media harvested under basal conditions (HCMB) and under RANKL-treated conditions (HCMR, 5 ng/ml RANKL). HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (A, upper) and Runx2 (A, lower). Histograms represent fold change in mRNA levels relative to untreated control HASMCs (i.e. standard unconditioned media, normalized to 1.0). HASMCs were also monitored for changes in ALP activity (B). Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
177
Fig. 4. Effect of noggin on osteoblastic activity in HASMCs. HAEC-conditioned media harvested under basal conditions (HCMB) and under RANKL-treated conditions (HCMR, 5 ng/ml RANKL) were used to treat HASMCs for 72 h in the absence and presence of 2.5 g/ml noggin, a BMP antagonist. HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (A, upper) and Runx2 (A, lower). Histograms represent fold change in mRNA levels relative to untreated control HASMCs (i.e. standard unconditioned media, normalized to 1.0). HASMCs were also monitored for changes in ALP activity (B). Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
Fig. 5. Schematic view of the CellMax® Duo perfused capillary system. The pores in the capillary wall were 0.2 m. The available luminal surface area in one bioreactor was 100 cm2 and the ELS volume was 1.5 mls. Key: EC, endothelial cell monolayer; ECS, extra-capillary space; ILS, intraluminal space; VSMC, vascular smooth muscle cell.
close paracrine signalling between the sub-luminal aspect of the HAECs and the overlying HASMCs. Furthermore, the CellMax® Duo allowed pumping of growth media through the lumen of the capillaries at flow rates designed to reproduce the laminar shear stress
that HAECs are exposed to in healthy conditions in vivo. As the viability and responsiveness of HAECs are affected in various ways by the level/pattern of shear stress they are exposed to, the addition of physiological levels of laminar shear stress to this model allowed
178
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
Fig. 6. CellMax® Duo perfused co-culture studies. Stable, confluent HAEC:HASMC co-cultures were established within the perfused capillary format under physiological laminar shear flow (10 dyn/cm2 ). Intraluminal HAECs were maintained under untreated control conditions or were exposed to RANKL (50 ng/ml, every 12 h for 72 h). (A) Post-treatment, ILS media reservoir levels of OPG (RANKL decoy receptor) and RANKL were monitored in both control (LHS) and RANKL-treated (RHS) CellMax® experiments. (B) Post-treatment, extracapillary HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (LHS) and Runx2 (RHS). Histograms represent fold change in mRNA levels relative to untreated controls. (C) ECS media was also harvested and monitored for changes in BMP-2 (upper), BMP-4 (middle), and ALP (lower) levels. Data are presented as mean ± SEM (n = 3 bioreactors for untreated controls, n = 4 bioreactors for RANKL-treatments). Key: ECS, extracapillary space; ILS, intraluminal space; LHS, left hand side; RHS, right hand side.
us to include an important aspect of systemic hemodynamics, the absence of which constitutes a potential confounder in previous methodological approaches. Ultimately, in using this system we noted that the addition of RANKL to the intraluminal circulating media flowing over HAECs led to a significant increase in circulating BMP-2 concentrations, and, when the extra-capillary media and HASMCs were harvested from the bioreactor and assayed, increased osteoblastic activity in terms of osteogenic gene expression and extracellular ALP activity was recorded. To the best of our knowledge, only one other group has examined the importance of HAEC-HASMC signalling to VC in the context of OPG and RANKL, namely Osako et al. (2010). This group, while examining the effects of oestrogen on HAECs, also demonstrated a promotion of BMP release by RANKL that was associated with increased osteoblastic activity in SMCs. Our own findings add significantly to these data in a number of aspects. With regards to the conditioned media experiments, our approach incorporated an avoidance of multiple potential confounders of VC research, and we report that it is BMP-2, and not BMP-4, that is the specific signalling protein responsible for promoting osteoblastic behaviour in HASMCs. With regards to our co-culture experiments, it should also be noted that to the best of our knowledge no other studies to date have examined OPG/RANKL and VC with HAECs and HASMCs co-cultured in close proximity as part of an artificial capillary incor-
porating physiological flow. Given the ongoing conflicting findings within this field of research in the literature, we submit that data from models approximating the in vivo state as closely as possible are especially relevant to unravelling this complex pathological process. With regards to potential limitations of the present study, we note that while the CellMax® Duo circuit was designed to approximate elements of the in vivo setting, the generation and use of an animal model of VC was an alternative approach that we considered when designing our experimental protocol. VC is a highly complex integrative process, however, with multiple competing and interacting regulatory systems, and in this context the CellMax® Duo allowed us to achieve a compromise between incorporating aspects of the in vivo arterial environment while also allowing us to focus on and characterize a single defined regulatory system. We also note it is possible that RANKL may also exert effects on inhibitors of calcification as well as acting on inducers such as BMP2, and this relationship, while outside the focus of the present study, represents an important and interesting target for future research. Finally, we note that while a number of previous studies in this area have tested for calcium deposition within cell cultures as an endpoint, we did not adopt this approach as calcium deposition has been shown to demonstrate considerable heterogeneity across different donors, and the calcifying media needed to test for cal-
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
179
Fig. 7. Schematic view of RANKL-induced paracrine signalling and osteoblastic activity within HAEC:HASMC co-culture. RANKL treatment of HAECs induces release of BMP-2, which in turn initiates a pro-calcification signalling cascade within HASMCs (i.e. increased ALP and Runx2 expression, increased levels of released ALP activity).
cium deposition is in itself pro-calcifying, and therefore a potential confounder in our assessment of the effects of RANKL. In conclusion, therefore, with this series of experiments we describe a regulatory system for VC whereby RANKL, when present in concentrations greater than that of OPG, induces the release of BMP-2 from vascular endothelial cells, which in turn acts on vascular smooth muscle cells in a paracrine fashion to induce both gene and protein changes consistent with the emergence of osteoblastic activity in the latter cell type (Fig. 7). Furthermore, while we did not attempt to include the effects of prescribed medications on VC in this series of experiments, we note that the findings of this study are timely in that the medication denosumab (a monoclonal antibody that mimics the action of OPG) is now being utilized in the general population for the treatment of osteoporosis, with extremely limited information available on its potential interactions with VC in humans (Cummings et al., 2009). Ultimately, and based on these
observations, we submit that a determination of whether RANKL blockade inhibits the emergence of VC in humans is now merited.
Fundings This work was supported by a research training fellowship grant supplied by the Health Research Board of Ireland to CD (Grant reference HPF/2010/30). Additional financial support was provided by a DCU O’Hare Scholarship and a Government of Ireland—Irish Research Council Postgraduate Scholarship to EH (Grant reference GOIPG/2015/3758).
Conflict of interest None declared.
180
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
References Abedin, M., Tintut, Y., Demer, L.L., 2004. Vascular calcification: mechanisms and clinical ramifications. Arterioscler. Thromb. Vasc. Biol. 24, 1161–1170. Bucay, N., Sarosi, I., Dunstan, C.R., Morony, S., Tarpley, J., Capparelli, C., Scully, S., Tan, H.L., Xu, W., Lacey, D.L., Boyle, W.J., Simonet, W.S., 1998. Osteoprotegerin-deficient mice develop early onset osteoporosis and arterial calcification. Genes Dev. 12 (9), 1260–1268. Byon, C.H., Chen, Y., 2015. Molecular mechanisms of vascular calcification in chronic kidney disease: the link between bone and the vasculature. Curr. Osteoporos. Rep. 13 (4), 206–215. Byon, C.H., Sun, Y., Chen, J., Yuan, K., Mao, X., Heath, J.M., Anderson, P.G., Tintut, Y., Demer, L.L., Wang, D., Chen, Y., 2011. Runx2-upregulated receptor activator of nuclear factor (B ligand in calcifying smooth muscle cells promotes migration and osteoclastic differentiation of macrophages. Arterioscler. Thromb. Vasc. Biol. 31 (6), 1387–1396. Candido, R., Toffoli, B., Corallini, F., Bernardi, S., Zella, D., Voltan, R., Grill, V., Celeghini, C., Fabris, B., 2010. Human full-length osteoprotegerin induces the proliferation of rodent vascular smooth muscle cells both in vitro and in vivo. J. Vasc. Res. 47 (3), 252–261. Chiu, J.J., Usami, S., Chien, S., 2009. Vascular endothelial responses to altered shear stress: pathologic implications for atherosclerosis. Ann. Med. 41, 19–28. Cummings, S.R., San Martin, J., McClung, M.R., Siris, E.S., Eastell, R., Reid, I.R., Delmas, P., Zoog, H.B., Austin, M., Wang, A., Kutilek, S., Adami, S., Zanchetta, J., Libanati, C., Siddhanti, S., Christiansen, C., 2009. FREEDOM Trial. Denosumab for prevention of fractures in postmenopausal women with osteoporosis. N. Engl. J. Med. 361 (8), 756–765. Davenport, C., Mahmood, W.A., Forde, H., Ashley, D.T., Agha, A., McDermott, J., Sreenan, S., Thompson, C.J., McGrath, F., McAdam, B., Cummins, P.M., Smith, D., 2015. The effects of insulin and liraglutide on osteoprotegerin and vascular calcification in vitro and in patients with type 2 diabetes. Eur. J. Endocrinol. 173 (1), 55–61. Demer, L.L., Tintut, Y., 2008. Vascular calcification: pathobiology of a multifaceted disease. Circulation 117, 2938–2948. Eddahibi, S., Guignabert, C., Barlier-Mur, A.M., Dewachter, L., Fadel, E., Dartevelle, P., Humbert, M., Simonneau, G., Hanoun, N., Saurini, F., Hamon, M., Adnot, S., 2006. Cross talk between endothelial and smooth muscle cells in pulmonary hypertension: critical role for serotonin-induced smooth muscle hyperplasia. Circulation 113 (15), 1857–1864. Higgins, C.L., Isbilir, S., Basto, P., Chen, I.Y., Vaduganathan, M., Vaduganathan, P., Reardon, M.J., Lawrie, G., Peterson, L., Morrisett, J.D., 2015. Distribution of alkaline phosphatase, osteopontin, RANK ligand and osteoprotegerin in calcified human carotid atheroma. Protein J. 34 (5), 315–328. Janke, D., Jankowski, J., Rüth, M., Buschmann, I., Lemke, H.D., Jacobi, D., Knaus, P., Spindler, E., Zidek, W., Lehmann, K., Jankowski, V., 2013. The artificial artery as in vitro perfusion model. PLoS One 8 (3), e57227. Johnson, R.C., Leopold, J.A., Loscalzo, J., 2006. Vascular calcification: pathobiological mechanisms and clinical implications. Circ. Res. 99, 1044–1059. Kaden, J.J., Bickelhaupt, S., Grobholz, R., Haase, K.K., Sarikoc¸, A., Kilic¸, R., Brueckmann, M., Lang, S., Zahn, I., Vahl, C., Hagl, S., Dempfle, C.E., Borggrefe, M., 2004. Receptor activator of nuclear factor kappaB ligand and osteoprotegerin regulate aortic valve calcification. J. Mol. Cell. Cardiol. 36 (1), 57–66. Krause, C., Guzman, A., Knaus, P., 2011. Noggin. Int. J. Biochem. Cell Biol. 43, 478–481.
Martin, F.A., McLoughlin, A., Rochfort, K.D., Davenport, C., Murphy, R.P., Cummins, P.M., 2014. Regulation of thrombomodulin expression and release in human aortic endothelial cells by cyclic strain. PLoS One 9 (9), e108254. Moran, C.S., McCann, M., Karan, M., Norman, P., Ketheesan, N., Golledge, J., 2005. Association of osteoprotegerin with human abdominal aortic aneurysm progression. Circulation 111, 3119–3125. Morony, S., Tintut, Y., Zhang, Z., Cattley, R.C., Van, G., Dwyer, D., Stolina, M., Kostenuik, P.J., Demer, L.L., 2008. Osteoprotegerin inhibits vascular calcification without affecting atherosclerosis in ldlr(-/-) mice. Circulation 117 (3), 411–420. Olesen, M., Skov, V., Mechta, M., Mumm, B.H., Rasmussen, L.M., 2012. No influence of OPG and its ligands RANKL and TRAIL, on proliferation and regulation of the calcification process in primary human vascular smooth muscle cells. Mol. Cell. Endocrinol. 362, 149–156. Osako, M.K., Nakagami, H., Koibuchi, N., Shimizu, H., Nakagami, F., Koriyama, H., Shimamura, M., Miyake, T., Rakugi, H., Morishita, R., 2010. Estrogen inhibits vascular calcification via vascular RANKL system: common mechanism of osteoporosis and vascular calcification. Circ. Res. 107 (4), 466–475. Panizo, S., Cardus, A., Encinas, M., Parisi, E., Valcheva, P., López-Ongil, S., Coll, B., Fernandez, E., Valdivielso, J.M., 2009. RANKL increases vascular smooth muscle cell calcification through a RANK-BMP4-dependent pathway. Circ. Res. 104 (9), 1041–1048. Price, P.A., June, H.H., Buckley, J.R., Williamson, M.K., 2001. Osteoprotegerin inhibits artery calcification induced by warfarin and by vitamin D. Arterioscler. Thromb. Vasc. Biol. 21, 1610–1616. Redmond, E.M., Cahill, P.A., Sitzmann, J.V., 1995. Perfused transcapillary smooth muscle and endothelial cell co-culture—a novel in vitro model. In vitro Cell. Dev. Biol. Anim. 31, 601–609. Rochfort, K.D., Collins, L.E., Murphy, R.P., Cummins, P.M., 2014. Downregulation of blood-brain barrier phenotype by proinflammatory cytokines involves NADPH oxidase-dependent ROS generation: consequences for interendothelial adherens and tight junctions. PLoS One 9 (7), e101815. Vitovski, S., Phillips, J.S., Sayers, J., Croucher, P.I., 2007. Investigating the interaction between osteoprotegerin and receptor activator of NF-kappaB or tumor necrosis factor-related apoptosis-inducing ligand: evidence for a pivotal role for osteoprotegerin in regulating two distinct pathways. J. Biol. Chem. 282, 31601–31609. Wexler, L., Brundage, B., Crouse, J., Detrano, R., Fuster, V., Maddahi, J., Rumberger, J., Stanford, W., White, R., Taubert, K., 1996. Coronary artery calcification: pathophysiology, epidemiology, imaging methods, and clinical implications. A statement for health professionals from the American Heart Association. Writing Group. Circulation 94, 1175–1192. Wu, M., Rementer, C., Giachelli, C.M., 2013. Vascular calcification: an update on mechanisms and challenges in treatment. Calcif. Tissue Int. 93 (4), 365–373. Yuan, L.Q., Zhu, J.H., Wang, H.W., Liang, Q.H., Xie, H., Wu, X.P., Zhou, H., Cui, R.R., Sheng, Z.F., Zhou, H.D., Zhu, X., Liu, G.Y., Liu, Y.S., Liao, E.Y., 2011. RANKL is a downstream mediator for insulin-induced osteoblastic differentiation of vascular smooth muscle cells. PLoS One 6 (12), e29037. Zhang, J., Fu, M., Myles, D., Zhu, X., Du, J., Cao, X., Chen, Y.E., 2002. PDGF induces osteoprotegerin expression in vascular smooth muscle cells by multiple signal pathways. FEBS Lett. 521 (1–3), 180–184.
Contents lists available at ScienceDirect
The International Journal of Biochemistry & Cell Biology journal homepage: www.elsevier.com/locate/biocel
RANKL promotes osteoblastic activity in vascular smooth muscle cells by upregulating endothelial BMP-2 release Colin Davenport a,∗ , Emma Harper a , Hannah Forde a , Keith D. Rochfort b , Ronan P. Murphy c,d , Diarmuid Smith e , Philip M. Cummins a,d a
School of Biotechnology, Dublin City University, Dublin, Ireland Conway Institute, University College Dublin, Dublin, Ireland c School of Health and Human Performance, Dublin City University, Dublin, Ireland d Centre for Preventive Medicine, Dublin City University, Dublin, Ireland e Department of Academic Endocrinology, Beaumont Hospital, Dublin, Ireland b
a r t i c l e
i n f o
Article history: Received 20 January 2016 Received in revised form 18 May 2016 Accepted 19 June 2016 Available online 23 June 2016 Keywords: RANKL Osteoprotegerin Calcification BMP-2 Endothelial
a b s t r a c t Introduction: Receptor activator of nuclear factor kappa beta-ligand (RANKL) is thought to promote vascular calcification (VC) by inducing osteoblastic behaviour in vascular smooth muscle cells (VSMC) in an ill-defined process. The present study assessed whether RANKL affects pro-osteoblastic paracrine signalling between human aortic endothelial cells (HAEC) and human aortic smooth muscle cells (HASMC) using both conditioned media transfer and cell co-culture experimental approaches. Methods and results: For initial experiments (6-well format), HAEC-conditioned media was harvested following 72 h exposure to RANKL, and transferred to reporter HASMCs with/without noggin, an inhibitor of pro-osteoblastic bone morphogenetic protein (BMP) paracrine signalling. In further experiments, HAECs and HASMCs were co-cultured within the CellMax® Duo, a perfusing bioreactor unit that mimics the flow-mediated co-interaction of these cells within the arterial wall, and RANKL was added to the perfusing media for 72 h. At the conclusion of each experiment markers of osteoblastic activity were measured in HASMCs, including alkaline phosphatase (ALP) activity, mRNA levels of ALP and Runx2, as well as BMP2 and BMP-4 concentrations. RANKL increased BMP-2 release from HAECs, while exposure of HASMCs to RANKL-treated HAEC-conditioned media induced osteoblastic behaviour in HASMCs, an effect prevented by noggin. Within the CellMax® Duo bioreactor, the addition of RANKL to the intraluminal HAECs also produced an increase in BMP-2 and increased osteoblastic behaviour within the co-cultured HASMC population. Conclusions: RANKL promotes VC by inducing BMP-2 release from HAECs, which in turn appears to act in a paracrine fashion on the adjacent HASMC population to increase osteoblastic activity. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Vascular calcification (VC) exerts a number of adverse effects on the cardiovascular (CV) system and exhibits a strong positive
Abbreviations: ALP, alkaline phosphatase; BMP-2/4, bone morphogenetic protein 2/4; CV, cardiovascular; ELISA, enzyme-linked immunosorbent assay; ELS, extraluminal space; HAEC, human aortic endothelial cell; HASMC, human aortic smooth muscle cell; ILS, intraluminal space; mRNA, messenger ribonucleic acid; OPG, osteoprotegerin; RT-qPCR, quantitative real-time polymerase chain reaction; RANKL, receptor activator of nuclear factor kappa-beta ligand; Runx2, runt-related transcription factor 2; SEM, standard error of the mean; SMC, smooth muscle cell; TNF-␣, tumour necrosis factor alpha; VC, vascular calcification. ∗ Corresponding author at: School of Biotechnology, Dublin City University, Glasnevin, Dublin 9, Ireland. E-mail address: [email protected] (C. Davenport). http://dx.doi.org/10.1016/j.biocel.2016.06.009 1357-2725/© 2016 Elsevier Ltd. All rights reserved.
correlation with overall rates of CV morbidity and mortality in multiple patient populations (Abedin et al., 2004; Wexler et al., 1996). Depending on its location within the vasculature, VC can inhibit cardiac valve function, decrease arterial compliance and increase the risk of dissection following angioplasty (Demer and Tintut, 2008). Efforts to inhibit or even reverse this process have, to date, been largely unsuccessful, a failure attributed primarily to a lack of understanding of the VC process at the cellular level (Johnson et al., 2006). Whilst the pathophysiology of VC likely differs depending upon its location within the vasculature, as well as the effects of patient factors such as renal failure or diabetes, one unifying feature of VC appears to be the emergence of osteoblastic behaviour in vascular cell populations such as vascular smooth muscle cells (VSMCs) (Byon and Chen, 2015). The emergence of osteoblastic behaviour within vascular cells represents a potential target for treatments designed to prevent VC, and in recent years
172
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
the promoters and inhibitors of this transformation have been the subject of a significant body of research, with increasing interest focused on two interlinked proteins, namely osteoprotegerin (OPG) and receptor activator of nuclear factor kappa-beta ligand (RANKL), with the former acting as a decoy receptor that blocks the biological actions of the latter (Vitovski et al., 2007). Although the actions of OPG and RANKL have been described in considerable detail within the skeletal system where they control osteoclast activation and bone turnover, the role of these regulatory proteins in VC has yet to be properly elucidated (Wu et al., 2013). Both proteins are known to be present within the vascular wall, with RANKL expression particularly associated with areas of calcification (Higgins et al., 2015). In animal models, OPG knockout mice have been shown to develop early and severe VC, atherogenic mice treated with OPG demonstrated a reduction in plaque calcification burden, and OPG was shown to prevent vitamin D-induced VC in rats, findings that collectively suggest that unopposed RANKL activity (due to lack of OPG) promotes VC (Bucay et al., 1998; Morony et al., 2008; Price et al., 2001). In this regard, both Panizo et al. (2009) and Kaden et al. (2004) have reported that RANKL may act directly on VSMCs to induce osteoblastic behaviour. Their findings, however, were subsequently contradicted by Olesen et al. (2012) and Byon et al. (2011), who reported no direct effect from RANKL on VSMCs. More recently, preliminary findings have suggested that whilst RANKL does affect VSMC calcification, it may possibly do so by inducing the endothelial monolayer to produce and release pro-osteoblastic paracrine signals such as members of the bone morphogenetic protein family (BMP) (Osako et al., 2010). It is noteworthy, however, that many of the studies in this field have suffered from significant limitations in their methodology. These include the use of pro-calcifying culture media (a potential confounder that directly induces osteoblastic transformation in VSMCs), as well as the use of calcium deposition as a primary endpoint of their studies (an unreliable endpoint with poor reproducibility) (Olesen et al., 2012). With regards to model oversights, the in vitro mechanistic studies to date on VC have been mainly performed using “static” cell monocultures, which differ significantly from the arterial wall microenvironment in vivo, incorporating as it does constant crosstalk between vascular endothelial and smooth muscle cells (via both direct contact and paracrine signalling), in addition to the hemodynamic effects of blood flow on these cell populations (Chiu et al., 2009; Eddahibi et al., 2006). Taking into account these limitations, the aim of the present study was to characterize more robustly the effects of RANKL on the emergence of osteoblastic behaviour in human aortic smooth muscle cells (HASMC) in vitro. This included an assessment of both the direct effects of RANKL on HASMCs, as well as an assessment of its effects on the production and release of pro-osteoblastic BMP-2 and BMP-4 from human aortic endothelial cells (HAEC). Moreover, in order to enhance the relevance of our monoculture studies to the in vivo state, our VC model was extended to include a HAEC:HASMC co-culture within a perfused artificial capillary system (CellMax® Duo) that reproduces the 3D architecture and physiological hemodynamic shear flow of an arterial vessel (Janke et al., 2013; Redmond et al., 1995).
2. Materials and methods 2.1. Materials Unless otherwise stated, all reagents were purchased from Sigma-Aldrich (Dublin, IRL). Recombinant human RANKL, antiOPG neutralizing antibody (OPG-Ab), and noggin (BMP inhibitor) were purchased from R&D Systems (Mineapolis, MN, USA). Tumor necrosis factor-alpha (TNF-␣) was purchased from Merck Milli-
pore (Danvers, MA, USA). Primers were purchased from Eurofins Genomics (Ebersburg, DEU). 2.2. Cell culture HAECs and HASMCs were purchased from Promocell Gmbh (Heidelberg, Germany), as were the culture media for both cell types. The HAECs (positive for von Willebrand factor and CD31) were received from two donors, a 28 year old Caucasian male (utilized in the majority of the experiments) and a second batch from a 23 year old Caucasian female (utilized for dose-response curves for RANKL and OPG release secondary to RANKL exposure). The HASMCs (positive for ␣-actin) were received from two donors, one batch from a 23 year old Caucasian male (utilized in the majority of the experiments) and a second batch from a 32 year old Caucasian male (utilized in 2 RANKL-treated bioreactors and 1 control bioreactor experiment in the co-culture experiments). HAECs were cultured in endothelial cell growth medium with the following supplements; fetal calf serum (0.05 ml/ml), endothelial cell growth supplement (0.004 ml/ml), epidermal growth factor (10 ng/ml), heparin (90 g/ml) and hydrocortisone (1 g/ml). This media was also supplemented with penicillin (100 IU/ml) and streptomycin (100 g/ml). HASMCs were cultured in smooth muscle cell growth medium supplemented with antibiotics in the same manner as the HAEC media, in addition to fetal calf serum (0.05 ml/ml), epidermal growth factor (0.5 ng/ml), basic fibroblast growth factor (2 ng/ml) and insulin at 5 g/ml. As insulin exerts a potent inhibitory effect on OPG production (Davenport et al., 2015), this supplement was omitted for a minimum of 2 passages prior to the commencement of experiments). Basic fibroblast growth factor (bFGF), which has been reported as promoting OPG production (albeit at higher concentrations), was retained in the media to maintain acceptable levels of growth and viability for experimental purposes (Zhang et al., 2002). Cells were maintained in a humidified incubator at 37 ◦ C and 5% CO2 . Passages 5–9 were routinely used for experimental purposes. Cell number and viability were routinely measured using the advanced detection and accurate measurement (ADAMTM ) cell counter (Digital Bio, Seoul, KOR), both for seeding density purposes and to allow normalization of results where necessary. This process involved the digital analysis of cells on custom Accuchip slides following the addition of propidium iodide with or without a membrane permeabilising solution, providing total numbers of both viable and non-viable cells. Cell culture experiments were conducted in three formats: (i) monoculture; (ii) conditioned media; and (iii) perfused co-culture. 2.3. Monoculture experiments The effects of RANKL on osteoblastic activity in HASMCs was initially investigated. HASMCs were grown to confluency in standard 6-well dishes and subsequently treated for 72 h with RANKL (0 and 50 ng/ml) in the absence or presence of OPG-neutralizing antisera (optimized dose of 2.5 g/ml). The concentration of 50 ng/ml RANKL was utilized to overcome endogenous OPG production from HASMCs, which we measured as reaching 33 ng/ml after 72 h in culture. OPG-neutralizing antisera has previously been reported as exerting no effects on HASMCs in isolation (Osako et al., 2010). Posttreatment, media and cells (mRNA) were harvested for analysis (alkaline phosphatase/ALP activity and osteoblastic gene expression, namely ALP and runt-related transcription factor 2 (Runx2). Harvested media was centrifuged at 1000 rpm (4 ◦ C) for 10 min to remove any cellular debris. All samples were stored at −80 ◦ C with all relevant assays being performed within one month as part of a single freeze-thaw cycle. In further experiments, the effects of RANKL on BMP-2 and −4 production and/or secretion by HASMCs and HAECs were also investigated. Both cell types were grown to
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
confluency in 6-well dishes. Cells were treated with RANKL as follows: HAECs −0, 5 and 25 ng/ml, 0–72 h; HASMCs −0 and 50 ng/ml, 0–72 h. HAECs were also exposed to TNF-␣ (100 ng/ml) as a positive control for BMP production. Post-treatment, media and cells (mRNA, protein) were harvested for analysis (BMP-2/4 mRNA and protein levels). 2.4. Conditioned media experiments The effect of HAEC paracrine signalling on HASMC osteoblastic activity was next investigated. HAECs were seeded into permeable (0.4 m pore) transwell culture inserts (Merck Millipore, MA, USA) and grown to confluency. The HAEC-seeded inserts were then placed into standard 6-well dishes containing fresh HAEC culture media to create separate abluminal (apical) and subluminal (basolateral) compartments. Cells were abluminally treated for 72 h with RANKL (5 ng/ml) in the absence and presence of the BMP inhibitor, noggin (2.5 g/ml), after which conditioned media was harvested from the subluminal compartment. The concentration of 5 ng/ml RANKL was utilized to overcome endogenous OPG production from HAECs, which we measured as reaching 0.8 ng/ml after 72 h in culture. Conditioned media (containing HAEC-derived paracrine factors) was then transferred directly onto confluent HASMC reporter cell cultures for a further 72 h. Post-treatment, HASMC media and mRNA were harvested for analysis (alkaline phosphatase activity and osteoblastic gene expression). 2.5. Perfused co-culture experiments The effects of RANKL on HAEC-HASMC paracrine signalling leading to the emergence of osteoblastic activity in HASMCs was investigated in a perfused CellMax® Duo co-culture system (Spectrum Laboratories, Rancho Dominguez, CA, USA). Each perfused bioreactor comprised a parallel bundle of ProNectin® -coated semipermeable capillaries, with the intraluminal space (ILS) representing the vessel interior, and the extraluminal space (ELS) representing the medial layer of the vessel. Within this system, HAECs were seeded and grown upon the intraluminal surface of the semipermeable capillaries, while HASMCs were seeded and grown on the outside of the semi-permeable capillaries, in the ELS. Media flowed through the intraluminal aspect of the capillaries. Upon connecting the media reservoir, gas permeable tubing and bioreactor unit, the various components of the CellMax® Duo became a single sealed, perfusing circuit maintained under shear flow within a dedicated humidified incubator at 37 ◦ C and 5% CO2 . As both HAECs and HASMCs were grown within the CellMax® Duo for these experiments, and only one growth media could be used within the system at any given time, a 1:1 ratio of HAEC and HASMC media was utilized. HASMCs, with their slower growth rate, were inoculated (1 × 107 cells) into the ELS first as previously described by Redmond et al. (1995) and allowed to grow for a minimum of 3 weeks at an intraluminal media flow rate of 20 ml/min [19]. Next, HAECs were inoculated into the ILS (2.5 × 107 cells) and allowed to grow and form a monolayer over the intraluminal capillary walls for a further 2 weeks at an intraluminal media flow rate of 5 ml/min. As these were relatively long periods of incubation, the proliferation and metabolic parameters of the inoculated HASMCs and HAECs within each bioreactor unit were ascertained by daily measurements of glucose consumption and lactate production using an YSI 2300 STAT Plus Analyzer (YSI Incorporated Life Sciences, Ohio, USA). During co-culture establishment, media in the reservoir was replaced whenever glucose concentrations dropped by 50%. Experimentation commenced once a plateau of glucose consumption and lactate production rates had been reached. At this point, the intraluminal media flow rate through the capillaries was adjusted to generate physiological levels of laminar shear stress equiva-
173
lent to 10 dyn/cm2 . During the experiments, CELLMAX® units being exposed to RANKL had their capillary perfusion halted for 20 min every 12 h while RANKL was added to the ILS at 50 ng/ml (sufficient RANKL to overcome endogenous HAEC and HASMC OPG production), after which capillary perfusion was recommenced. CELLMAX® systems that were not exposed to RANKL were treated in the same manner but without RANKL being added to the ILS. The experiments continued for 72 h in total with a final circulating RANKL concentration of 5.1 ng/ml at the end of the experiment. Post-treatment, culture media was harvested from both the ILS and ELS compartments for measurement of ALP activity and BMP2 concentrations, whilst HASMCs were harvested from the ELS for analysis of osteoblastic gene expression. 2.6. Quantitative real-time RT-PCR (RT-qPCR) Post-treatment, HAECs and HASMCs were harvested for extraction of total RNA and analysis of gene expression as previously described by Rochfort et al. (2014) using an ABI 7900 HT Fast RT-PCR System (Applied Biosystems/Life Technologies, Paisley, UK). Each cDNA sample was assayed in triplicate and results analyzed by the comparative CT method. S18 was routinely used for normalization purposes. Primer pairs were screened for correct product size by 1% agarose gel electrophoresis and underwent melt-curve analysis for primer-dimers. S18 (250 bp): Forward 5 -cagccacccgagattgagca3 ; Reverse 5 -tagtagcgacgggcggtgtg-3 ; ALP (312 bp): Forward 5 -gcctggctacaaggtggtg-3 ; Reverse 5 -ggccagagcgagcagc-3 ; Runx2 (297 bp): Forward 5 -ggtaccagatgggactgtgg-3 ; Reverse 5 -gaggcggtcagagaacaaac-3 ; BMP-2 (199 bp): Forward 5 caagccaaacacaaacagcg-3 ; Reverse 5 -ccaacgtctgaacaatggca-3 . Annealing temperatures for S18, ALP, Runx2, and BMP-2 were 62 ◦ C, 58 ◦ C, 59 ◦ C, and 57.3 ◦ C, respectively. 2.7. Enzyme-linked immunosorbent assay (ELISA) The concentrations of BMP-2 and BMP-4 in media samples were measured using commercially available Duoset ELISA kits (R&D Systems, Minneapolis, USA) in conjunction with a protocol previously described by our group (Martin et al., 2014). Both kits exhibited intra-assay coefficients of variation of <5%. BMP-2/4 concentrations were normalized for viable HAEC and HASMC numbers as appropriate. 2.8. Alkaline phosphatase activity assay ALP activity, a critical marker of osteoblastic activity, was measured in media samples using the QuantichromTM Kit (BioAssay Systems, California, USA) according to manufacturer’s protocol. This was a colorimetric kinetic assay based on the principle that ALP (if present in the sample) will convert p-nitrophenyl phosphate (a colourless solution) to p-nitrophenyl (a yellow coloured solution) and phosphate, with the colour change measured via microplate reader at baseline and again at 4 mins. The inter-assay coefficient of variation for this kit was <5%. 2.9. Statistical analysis Data were analyzed using GraphPad Prism version 4.03 (Graphpad Software, California, USA), and are expressed in mean ± SEM (normality was tested for via the Kolmogorov-Smirnov test). Statistical comparisons between controls versus treatment groups were made by analysis of variance in conjunction with a Dunnett’s posthoc test for multiple comparisons. A Student’s unpaired t-test was also employed for pair-wise comparisons. A value of p ≤ 0.05 was considered significant.
174
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
3. Results 3.1. RANKL does not directly induce osteoblastic activity in HASMCs The ability of RANKL to directly induce osteoblastic activity in HASMCs was initially monitored. Following treatment of HASMCs for 72 h with 50 ng/ml RANKL, we observed no significant changes in mRNA levels for either ALP (Fig. 1B) or Runx2 (Fig. 1C), nor was cell viability affected. We also observed no significant change in ALP enzymatic activity in HASMC-conditioned media following RANKL treatment (Fig. 1D). To control for the possibility that the lack of effect of RANKL may be due to it becoming bound to OPG constitutively released from HASMCs, an excess of OPG neutralizing antibody (OPG-Ab, 2.5 g/ml) was included with the RANKL treatment. In this context, the OPG-Ab had no influence on the lack of effect of RANKL (Fig. 1B–D). 3.2. RANKL induces BMP-2 release from HAECs (but not HASMCs) In our next series of experiments, the effect of RANKL on BMP release from HAECs and HASMCs was investigated. RANKL was incubated for 72 h with HAECs (5 ng/ml) and HASMCs (50 ng/ml), after which the conditioned media was harvested for analysis of BMP-2 and BMP-4 release by ELISA. In response to RANKL treatment, we observed a significant increase in release of BMP-2 (but not BMP-4) from HAECs (Fig. 2A and B, Supplementary Fig. S1 in the online version at DOI: 10.1016/j.biocel.2016.06.009). The release of BMP-2 from HAECs appeared to occur due to the release of pre-formed BMP-2 from these cells, as exposure to RANKL did not significantly affect BMP-2 mRNA levels or intracellular protein levels (Supplementary Fig. S2 in the online version at DOI: 10.1016/ j.biocel.2016.06.009), whereas the positive control TNF-␣ induced both an increase in the release and production of BMP-2 (Supplementary Fig. S3 in the online version at DOI: 10.1016/j.biocel.2016. 06.009). By contrast, RANKL had no significant effect on release of either BMP from HASMCs (although a statistically insignificant upward trend in BMP-4 release was noted) (Fig. 2C and D). 3.3. RANKL treatment of HAECs can elicit osteoblastic activity in HASMCs—conditioned media model The ability of RANKL to induce release of pro-osteogenic factors from HAECs was next examined. HAECs cultured in transwell inserts were treated abluminally with 5 ng/ml of RANKL for 72 h, and HAEC-conditioned media subsequently harvested from the subluminal compartment. HASMCs were subsequently treated for 72 h with either unconditioned media (control) or with HAEC-conditioned media harvested under basal conditions (HCMB) or RANKL conditions (HCMR). Relative to basal conditions, RANKL treatment of HAECs significantly increased HASMC mRNA levels for ALP (Fig. 3A) and Runx2 (Fig. 3B), as well as ALP activity levels in HASMC media (Fig. 3C). HAEC-conditioned media experiments were subsequently repeated with RANKL in the absence and presence of 2.5 g/ml noggin, a BMP antagonist. In these experiments, noggin was found to completely block the RANKL-dependent increase in HASMC osteogenic gene expression (ALP, Runx2—Fig. 4A) and osteoblastic activity (ALP activity—Fig. 4B). 3.4. RANKL treatment of HAECs can elicit osteoblastic activity in HASMCs - perfused co-culture model The effects of RANKL on HAEC:HASMC paracrine signalling leading to the emergence of osteogenic activity in HASMCs was next investigated in a perfused CellMax® Duo co-culture sys-
tem that reproduces arterial 3D architecture and physiological hemodynamic shear flow (Fig. 5). Stable, confluent HAEC:HASMC co-cultures were established within the perfused capillary format under physiological laminar shear flow (10 dyn/cm2 ). Paired glucose and lactate consumption of a typical bioreactor unit is illustrated in Supplementary Fig. S4 in the online version at DOI: 10. 1016/j.biocel.2016.06.009. Intraluminal HAECs were maintained under untreated control conditions or were exposed to RANKL (50 ng/ml, twice a day for 72 h). In response to RANKL treatment of HAECs, mRNA levels of ALP and Runx2 were significantly elevated in HASMCs (Fig. 6B). In response to RANKL treatment of HAECs, elevated BMP-2 (but not BMP-4) levels, as well as elevated ALP activity levels, were also observed in the ECS media surrounding the HASMCs (located subluminally to the HAEC monolayer) (Fig. 6C).
4. Discussion The present study demonstrates a number of novel findings regarding the role of OPG and RANKL in the VC process. In addition to identifying and characterizing a paracrine relationship between ECs and SMCs that may help to explain the method by which OPG opposes VC in vivo, we have further confirmed this relationship in a perfusing circuit incorporating a co-culture of both vascular cell populations designed to simulate the vascular wall microenvironment. To the best of our knowledge, this paper is the first to examine the complex process of VC in an in vitro sheared co-culture model of this nature. With regards to the initial findings of this study, we demonstrated that RANKL, when applied directly to HASMCs, did not promote osteoblastic behaviour in terms of osteogenic gene upregulation (Runx2, ALP) or the secretion of ALP, an enzyme that is critical for the crystallization aspect of VC. While this relationship has been examined previously, the methodological approaches utilized in prior studies were limited in a number of important areas and conflicting data have been reported (Byon et al., 2011; Kaden et al., 2004; Olesen et al., 2012; Panizo et al., 2009). The vast majority of previous studies of this nature supplemented the cell media with phosphate in an effort to maximize observable differences in VC with RANKL. Phosphate, however, is known to be a potent inducer of VC and thus a potential confounder of these experiments. In addition, prior studies made routine use of calcium deposition as a primary endpoint, as opposed to gene and protein markers of osteoblastic activity. As illustrated by Olesen et al. (2012), calcium deposition is an unreliable endpoint with a high degree of variability between vascular cell donors, and furthermore typically requires the use of phosphate-containing media in the experiments. We also note that some authors did not take the endogenous production of OPG (with its ability to block the actions of RANKL) from HASMCs (and, to a lesser extent HAECs) into account (Candido et al., 2010; Moran et al., 2005). Finally, it is interesting to compare our own results with those of Yuan et al. (2011). This group reported that insulin induced osteoblastic behaviour in VSMCs via RANKL signalling, indicating that RANKL may directly promote osteoblastic change in these cells. These experiments, however, were performed on a specific subpopulation of VSMCs termed calcifying vascular cells, which are prone to adopting an osteoblastic phenotype spontaneously in culture, and this methodological difference in terms of cell populations studied may explain why we observed no apparent effect with RANKL when applied to a general heterogeneous HASMC population. Ultimately, our experimental approach was designed to test osteoblastic behaviour in HASMCs with a minimisation of potential confounders. As such, we did not utilize calcifying phosphate-rich media, and for our endpoints we measured established gene and protein markers of osteoblastic activity. Endogenous OPG production was overcome
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
175
Fig. 1. Direct effects of RANKL on osteoblastic activity in HASMCs. HASMCs were treated for 72 h with RANKL (50 ng/ml) ± OPG Ab (2.5 g/ml). Representative agarose gel insets verify HASMC mRNA expression and correct fragment sizes generated using the S18 (250 bp), Runx2 (297 bp), and ALP (312 bp) primers (A). After 72 h HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (B) and Runx2 (C). Histograms represent fold change in mRNA levels relative to untreated control HASMCs. HASMCs were also monitored for changes in ALP activity (D). Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
through the use of either neutralizing antibodies to OPG and/or the use of RANKL concentrations significantly in excess of those of endogenous OPG. It is within this setting that we report no evidence for a direct effect from RANKL on osteoblastic activity and VC in HASMCs. In the second part of this study we assessed the effects of RANKL on the production of BMP-2 and BMP-4 from both HAECs and HASMCs. Both of these pro-osteogenic proteins have been identified within the vascular wall in vivo and both are recognised promoters of osteoblastic activity in HASMCs (Abedin et al., 2004; Higgins et al., 2015). We noted that HAECs had a 10-fold higher constitutive production of both BMP-2 and BMP-4 in comparison to HASMCs, and that RANKL significantly increased the release of BMP-2 from HAECs. Interestingly, the increased release of BMP-2 was not accompanied by increased expression of BMP-2 mRNA or intracellular protein levels, as was previously reported by Osako et al. (2010). The contrasting results between our data and that of Osako et al. may have arisen as a result of differing methodologies in terms of cell culture conditions, as well as possible phenotypic heterogeneity between HAEC populations. Our results suggest that pre-formed BMP-2 may be released and secreted by HASMCs without an accompanying increase in intracellular production, although further experiments will be required to clarify this response. Using a conditioned media model by
exposing HASMCs to HAEC-conditioned media, we noted significant increases in osteoblastic activity. Stimulating the HAECs with RANKL enhanced this effect further. We then added noggin to the conditioned media derived from RANKL-treated HAECs, a protein which binds to BMP-2 and blocks its pro-osteoblastic activity (Krause et al., 2011). Upon doing so, we noted that the addition of this BMP-2 blocking agent reduced the subsequent development of osteoblastic behaviour in HASMCs to baseline levels. Ultimately, we conclude that BMP activity is necessary for HAEC-conditioned media to induce osteoblastic activity in HASMCs, and that induction of BMP-2 (but not BMP-4) release from HAECs likely represents the process by which RANKL promotes VC. In the final step of the present study, we utilized a perfused artificial capillary co-culture system (CellMax® Duo) to test our initial findings in an experimental model designed to approximate the conditions of the vascular wall, and in doing so overcome a number of important limitations of traditional monoculture and conditioned media experiments (Chiu et al., 2009; Eddahibi et al., 2006; Redmond et al., 1995). Using CellMax® Duo perfused bioreactors, we grew HAECs on the inner luminal surface of a semi-permeable artificial capillary within a capillary bundle format, with HASMCs growing on the outer surface of the porous capillary walls. In doing so, we more closely replicated the 3D vessel architecture and close proximity of these two cell populations in vivo, and facilitated
176
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
Fig. 2. Effect of RANKL on BMP-2/4 release from HAECs and HASMCs. RANKL was incubated for 72 h with HAECs (5 ng/ml) and HASMCs (50 ng/ml), after which media was harvested for analysis by ELISA. Histograms represent pg/104 cells of (A) BMP-2 and (B) BMP-4 released from HAECs, as well as (C) BMP-2 and (D) BMP-4 released from HASMCs, under untreated control and RANKL-treated conditions. Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
Fig. 3. Effects of HAEC-conditioned media on osteoblastic activity in HASMCs. HASMCs were treated for 72 h with HAEC-conditioned media harvested under basal conditions (HCMB) and under RANKL-treated conditions (HCMR, 5 ng/ml RANKL). HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (A, upper) and Runx2 (A, lower). Histograms represent fold change in mRNA levels relative to untreated control HASMCs (i.e. standard unconditioned media, normalized to 1.0). HASMCs were also monitored for changes in ALP activity (B). Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
177
Fig. 4. Effect of noggin on osteoblastic activity in HASMCs. HAEC-conditioned media harvested under basal conditions (HCMB) and under RANKL-treated conditions (HCMR, 5 ng/ml RANKL) were used to treat HASMCs for 72 h in the absence and presence of 2.5 g/ml noggin, a BMP antagonist. HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (A, upper) and Runx2 (A, lower). Histograms represent fold change in mRNA levels relative to untreated control HASMCs (i.e. standard unconditioned media, normalized to 1.0). HASMCs were also monitored for changes in ALP activity (B). Data are presented as mean ± SEM (n = 18 for total number of wells analyzed after experiments were performed in triplicate).
Fig. 5. Schematic view of the CellMax® Duo perfused capillary system. The pores in the capillary wall were 0.2 m. The available luminal surface area in one bioreactor was 100 cm2 and the ELS volume was 1.5 mls. Key: EC, endothelial cell monolayer; ECS, extra-capillary space; ILS, intraluminal space; VSMC, vascular smooth muscle cell.
close paracrine signalling between the sub-luminal aspect of the HAECs and the overlying HASMCs. Furthermore, the CellMax® Duo allowed pumping of growth media through the lumen of the capillaries at flow rates designed to reproduce the laminar shear stress
that HAECs are exposed to in healthy conditions in vivo. As the viability and responsiveness of HAECs are affected in various ways by the level/pattern of shear stress they are exposed to, the addition of physiological levels of laminar shear stress to this model allowed
178
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
Fig. 6. CellMax® Duo perfused co-culture studies. Stable, confluent HAEC:HASMC co-cultures were established within the perfused capillary format under physiological laminar shear flow (10 dyn/cm2 ). Intraluminal HAECs were maintained under untreated control conditions or were exposed to RANKL (50 ng/ml, every 12 h for 72 h). (A) Post-treatment, ILS media reservoir levels of OPG (RANKL decoy receptor) and RANKL were monitored in both control (LHS) and RANKL-treated (RHS) CellMax® experiments. (B) Post-treatment, extracapillary HASMCs were subsequently harvested and monitored for changes in mRNA levels of ALP (LHS) and Runx2 (RHS). Histograms represent fold change in mRNA levels relative to untreated controls. (C) ECS media was also harvested and monitored for changes in BMP-2 (upper), BMP-4 (middle), and ALP (lower) levels. Data are presented as mean ± SEM (n = 3 bioreactors for untreated controls, n = 4 bioreactors for RANKL-treatments). Key: ECS, extracapillary space; ILS, intraluminal space; LHS, left hand side; RHS, right hand side.
us to include an important aspect of systemic hemodynamics, the absence of which constitutes a potential confounder in previous methodological approaches. Ultimately, in using this system we noted that the addition of RANKL to the intraluminal circulating media flowing over HAECs led to a significant increase in circulating BMP-2 concentrations, and, when the extra-capillary media and HASMCs were harvested from the bioreactor and assayed, increased osteoblastic activity in terms of osteogenic gene expression and extracellular ALP activity was recorded. To the best of our knowledge, only one other group has examined the importance of HAEC-HASMC signalling to VC in the context of OPG and RANKL, namely Osako et al. (2010). This group, while examining the effects of oestrogen on HAECs, also demonstrated a promotion of BMP release by RANKL that was associated with increased osteoblastic activity in SMCs. Our own findings add significantly to these data in a number of aspects. With regards to the conditioned media experiments, our approach incorporated an avoidance of multiple potential confounders of VC research, and we report that it is BMP-2, and not BMP-4, that is the specific signalling protein responsible for promoting osteoblastic behaviour in HASMCs. With regards to our co-culture experiments, it should also be noted that to the best of our knowledge no other studies to date have examined OPG/RANKL and VC with HAECs and HASMCs co-cultured in close proximity as part of an artificial capillary incor-
porating physiological flow. Given the ongoing conflicting findings within this field of research in the literature, we submit that data from models approximating the in vivo state as closely as possible are especially relevant to unravelling this complex pathological process. With regards to potential limitations of the present study, we note that while the CellMax® Duo circuit was designed to approximate elements of the in vivo setting, the generation and use of an animal model of VC was an alternative approach that we considered when designing our experimental protocol. VC is a highly complex integrative process, however, with multiple competing and interacting regulatory systems, and in this context the CellMax® Duo allowed us to achieve a compromise between incorporating aspects of the in vivo arterial environment while also allowing us to focus on and characterize a single defined regulatory system. We also note it is possible that RANKL may also exert effects on inhibitors of calcification as well as acting on inducers such as BMP2, and this relationship, while outside the focus of the present study, represents an important and interesting target for future research. Finally, we note that while a number of previous studies in this area have tested for calcium deposition within cell cultures as an endpoint, we did not adopt this approach as calcium deposition has been shown to demonstrate considerable heterogeneity across different donors, and the calcifying media needed to test for cal-
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
179
Fig. 7. Schematic view of RANKL-induced paracrine signalling and osteoblastic activity within HAEC:HASMC co-culture. RANKL treatment of HAECs induces release of BMP-2, which in turn initiates a pro-calcification signalling cascade within HASMCs (i.e. increased ALP and Runx2 expression, increased levels of released ALP activity).
cium deposition is in itself pro-calcifying, and therefore a potential confounder in our assessment of the effects of RANKL. In conclusion, therefore, with this series of experiments we describe a regulatory system for VC whereby RANKL, when present in concentrations greater than that of OPG, induces the release of BMP-2 from vascular endothelial cells, which in turn acts on vascular smooth muscle cells in a paracrine fashion to induce both gene and protein changes consistent with the emergence of osteoblastic activity in the latter cell type (Fig. 7). Furthermore, while we did not attempt to include the effects of prescribed medications on VC in this series of experiments, we note that the findings of this study are timely in that the medication denosumab (a monoclonal antibody that mimics the action of OPG) is now being utilized in the general population for the treatment of osteoporosis, with extremely limited information available on its potential interactions with VC in humans (Cummings et al., 2009). Ultimately, and based on these
observations, we submit that a determination of whether RANKL blockade inhibits the emergence of VC in humans is now merited.
Fundings This work was supported by a research training fellowship grant supplied by the Health Research Board of Ireland to CD (Grant reference HPF/2010/30). Additional financial support was provided by a DCU O’Hare Scholarship and a Government of Ireland—Irish Research Council Postgraduate Scholarship to EH (Grant reference GOIPG/2015/3758).
Conflict of interest None declared.
180
C. Davenport et al. / The International Journal of Biochemistry & Cell Biology 77 (2016) 171–180
References Abedin, M., Tintut, Y., Demer, L.L., 2004. Vascular calcification: mechanisms and clinical ramifications. Arterioscler. Thromb. Vasc. Biol. 24, 1161–1170. Bucay, N., Sarosi, I., Dunstan, C.R., Morony, S., Tarpley, J., Capparelli, C., Scully, S., Tan, H.L., Xu, W., Lacey, D.L., Boyle, W.J., Simonet, W.S., 1998. Osteoprotegerin-deficient mice develop early onset osteoporosis and arterial calcification. Genes Dev. 12 (9), 1260–1268. Byon, C.H., Chen, Y., 2015. Molecular mechanisms of vascular calcification in chronic kidney disease: the link between bone and the vasculature. Curr. Osteoporos. Rep. 13 (4), 206–215. Byon, C.H., Sun, Y., Chen, J., Yuan, K., Mao, X., Heath, J.M., Anderson, P.G., Tintut, Y., Demer, L.L., Wang, D., Chen, Y., 2011. Runx2-upregulated receptor activator of nuclear factor (B ligand in calcifying smooth muscle cells promotes migration and osteoclastic differentiation of macrophages. Arterioscler. Thromb. Vasc. Biol. 31 (6), 1387–1396. Candido, R., Toffoli, B., Corallini, F., Bernardi, S., Zella, D., Voltan, R., Grill, V., Celeghini, C., Fabris, B., 2010. Human full-length osteoprotegerin induces the proliferation of rodent vascular smooth muscle cells both in vitro and in vivo. J. Vasc. Res. 47 (3), 252–261. Chiu, J.J., Usami, S., Chien, S., 2009. Vascular endothelial responses to altered shear stress: pathologic implications for atherosclerosis. Ann. Med. 41, 19–28. Cummings, S.R., San Martin, J., McClung, M.R., Siris, E.S., Eastell, R., Reid, I.R., Delmas, P., Zoog, H.B., Austin, M., Wang, A., Kutilek, S., Adami, S., Zanchetta, J., Libanati, C., Siddhanti, S., Christiansen, C., 2009. FREEDOM Trial. Denosumab for prevention of fractures in postmenopausal women with osteoporosis. N. Engl. J. Med. 361 (8), 756–765. Davenport, C., Mahmood, W.A., Forde, H., Ashley, D.T., Agha, A., McDermott, J., Sreenan, S., Thompson, C.J., McGrath, F., McAdam, B., Cummins, P.M., Smith, D., 2015. The effects of insulin and liraglutide on osteoprotegerin and vascular calcification in vitro and in patients with type 2 diabetes. Eur. J. Endocrinol. 173 (1), 55–61. Demer, L.L., Tintut, Y., 2008. Vascular calcification: pathobiology of a multifaceted disease. Circulation 117, 2938–2948. Eddahibi, S., Guignabert, C., Barlier-Mur, A.M., Dewachter, L., Fadel, E., Dartevelle, P., Humbert, M., Simonneau, G., Hanoun, N., Saurini, F., Hamon, M., Adnot, S., 2006. Cross talk between endothelial and smooth muscle cells in pulmonary hypertension: critical role for serotonin-induced smooth muscle hyperplasia. Circulation 113 (15), 1857–1864. Higgins, C.L., Isbilir, S., Basto, P., Chen, I.Y., Vaduganathan, M., Vaduganathan, P., Reardon, M.J., Lawrie, G., Peterson, L., Morrisett, J.D., 2015. Distribution of alkaline phosphatase, osteopontin, RANK ligand and osteoprotegerin in calcified human carotid atheroma. Protein J. 34 (5), 315–328. Janke, D., Jankowski, J., Rüth, M., Buschmann, I., Lemke, H.D., Jacobi, D., Knaus, P., Spindler, E., Zidek, W., Lehmann, K., Jankowski, V., 2013. The artificial artery as in vitro perfusion model. PLoS One 8 (3), e57227. Johnson, R.C., Leopold, J.A., Loscalzo, J., 2006. Vascular calcification: pathobiological mechanisms and clinical implications. Circ. Res. 99, 1044–1059. Kaden, J.J., Bickelhaupt, S., Grobholz, R., Haase, K.K., Sarikoc¸, A., Kilic¸, R., Brueckmann, M., Lang, S., Zahn, I., Vahl, C., Hagl, S., Dempfle, C.E., Borggrefe, M., 2004. Receptor activator of nuclear factor kappaB ligand and osteoprotegerin regulate aortic valve calcification. J. Mol. Cell. Cardiol. 36 (1), 57–66. Krause, C., Guzman, A., Knaus, P., 2011. Noggin. Int. J. Biochem. Cell Biol. 43, 478–481.
Martin, F.A., McLoughlin, A., Rochfort, K.D., Davenport, C., Murphy, R.P., Cummins, P.M., 2014. Regulation of thrombomodulin expression and release in human aortic endothelial cells by cyclic strain. PLoS One 9 (9), e108254. Moran, C.S., McCann, M., Karan, M., Norman, P., Ketheesan, N., Golledge, J., 2005. Association of osteoprotegerin with human abdominal aortic aneurysm progression. Circulation 111, 3119–3125. Morony, S., Tintut, Y., Zhang, Z., Cattley, R.C., Van, G., Dwyer, D., Stolina, M., Kostenuik, P.J., Demer, L.L., 2008. Osteoprotegerin inhibits vascular calcification without affecting atherosclerosis in ldlr(-/-) mice. Circulation 117 (3), 411–420. Olesen, M., Skov, V., Mechta, M., Mumm, B.H., Rasmussen, L.M., 2012. No influence of OPG and its ligands RANKL and TRAIL, on proliferation and regulation of the calcification process in primary human vascular smooth muscle cells. Mol. Cell. Endocrinol. 362, 149–156. Osako, M.K., Nakagami, H., Koibuchi, N., Shimizu, H., Nakagami, F., Koriyama, H., Shimamura, M., Miyake, T., Rakugi, H., Morishita, R., 2010. Estrogen inhibits vascular calcification via vascular RANKL system: common mechanism of osteoporosis and vascular calcification. Circ. Res. 107 (4), 466–475. Panizo, S., Cardus, A., Encinas, M., Parisi, E., Valcheva, P., López-Ongil, S., Coll, B., Fernandez, E., Valdivielso, J.M., 2009. RANKL increases vascular smooth muscle cell calcification through a RANK-BMP4-dependent pathway. Circ. Res. 104 (9), 1041–1048. Price, P.A., June, H.H., Buckley, J.R., Williamson, M.K., 2001. Osteoprotegerin inhibits artery calcification induced by warfarin and by vitamin D. Arterioscler. Thromb. Vasc. Biol. 21, 1610–1616. Redmond, E.M., Cahill, P.A., Sitzmann, J.V., 1995. Perfused transcapillary smooth muscle and endothelial cell co-culture—a novel in vitro model. In vitro Cell. Dev. Biol. Anim. 31, 601–609. Rochfort, K.D., Collins, L.E., Murphy, R.P., Cummins, P.M., 2014. Downregulation of blood-brain barrier phenotype by proinflammatory cytokines involves NADPH oxidase-dependent ROS generation: consequences for interendothelial adherens and tight junctions. PLoS One 9 (7), e101815. Vitovski, S., Phillips, J.S., Sayers, J., Croucher, P.I., 2007. Investigating the interaction between osteoprotegerin and receptor activator of NF-kappaB or tumor necrosis factor-related apoptosis-inducing ligand: evidence for a pivotal role for osteoprotegerin in regulating two distinct pathways. J. Biol. Chem. 282, 31601–31609. Wexler, L., Brundage, B., Crouse, J., Detrano, R., Fuster, V., Maddahi, J., Rumberger, J., Stanford, W., White, R., Taubert, K., 1996. Coronary artery calcification: pathophysiology, epidemiology, imaging methods, and clinical implications. A statement for health professionals from the American Heart Association. Writing Group. Circulation 94, 1175–1192. Wu, M., Rementer, C., Giachelli, C.M., 2013. Vascular calcification: an update on mechanisms and challenges in treatment. Calcif. Tissue Int. 93 (4), 365–373. Yuan, L.Q., Zhu, J.H., Wang, H.W., Liang, Q.H., Xie, H., Wu, X.P., Zhou, H., Cui, R.R., Sheng, Z.F., Zhou, H.D., Zhu, X., Liu, G.Y., Liu, Y.S., Liao, E.Y., 2011. RANKL is a downstream mediator for insulin-induced osteoblastic differentiation of vascular smooth muscle cells. PLoS One 6 (12), e29037. Zhang, J., Fu, M., Myles, D., Zhu, X., Du, J., Cao, X., Chen, Y.E., 2002. PDGF induces osteoprotegerin expression in vascular smooth muscle cells by multiple signal pathways. FEBS Lett. 521 (1–3), 180–184.