|Home | About | Journals | Submit | Contact Us | Français|
To evaluate whether bisphosphonates modulate vascular calcification by a modification in endothelial progenitor cells (EPCs) co-expressing osteoblastic surface markers and genes.
Double blind, randomized study including 20 healthy early postmenopausal women (from February 1, 2008 through July 31, 2008) treated either with placebo or risedronate (35 mg/week) for 4 months. CD34+/KDR+ cells were isolated and gene expression was studied. Peripheral blood was collected at baseline and at 4 months to determine serum inflammatory markers, osteoprotegerin (OPG) and RANKL levels and bone turnover markers. Peripheral blood mononuclear cells were stained for EPC surface markers (CD34, CD133, and VEGF receptor [KDR]) as well as osteoblast markers (osteocalcin [OCN], alkaline phosphatase [AP], and Stro-1).
Risedronate treatment resulted in a significant downregulation of gene sets for osteoblast differentiation and proliferation in EPCs with a trend of decreasing EPCs coexpressing OCN.
Our findings indicate that bisphosphonate treatment downregulates the expression of osteogenic genes in EPCs and suggest a possible mechanism by which bisphosphonates may inhibit vascular calcification.
There has recently been considerable interest in the mechanisms of vascular calcification and the possible role of endothelial or endothelial progenitor cells (EPCs) in this process. In previous studies, we found that a higher percentage of EPCs (identified by the surface expression of CD34, CD133, and the vascular endothelial growth factor 2/kinase insert domain receptor [KDR]) from patients with coronary atherosclerosis expressed the bone-related protein, osteocalcin (OCN) compared to control subjects1. More recently, simultaneous samples from the proximal aorta and coronary sinus demonstrated that even patients with early coronary atherosclerosis were characterized by retention of OCN+ EPCs within the coronary circulation2. These data suggest that EPCs expressing osteogenic proteins may contribute to vascular calcification as opposed to initiating normal vascular repair. These findings in patients with coronary atherosclerosis are of particular interest given the recent demonstration that, under certain conditions (e.g., exposure to TGF-β2 or BMP4), endothelial cells can undergo an endothelial-to-mesenchymal transition3 which may play a critical role in the pathogenesis of a number of conditions, including not only atherosclerosis, but also pulmonary hypertension, wound healing, and cancer progression4.
Of interest, increased bone turnover has been associated with vascular calcification as well as increased cardiovascular mortality5-8, but the underlying mechanism(s) for these associations remain unclear. EPCs, which reside at least in part in the bone marrow, are a potential candidate for providing a link between bone metabolism and the vascular system since they are mobilized in response to vascular injury and contribute to vascular repair9, 10 but, as noted above, they may also contribute to vascular calcification. In addition, many of the same factors that modulate bone turnover, including certain cytokines, hormones and lipids, also modulate the development of atherosclerosis and vascular calcification. Thus, whereas increased production of cytokines such as IL-1β, IL-6, and IL-8 has been associated with bone loss in the skeletal system, in the vascular system these same cytokines have been associated with atherosclerotic plaque formation and vascular calcification8, 11. In addition, IL-8, as well as other chemokines and proteolytic enzymes, may play a major role in the mobilization of progenitor cells from the bone marrow12, with IL-8 also recently being linked to the homing of EPCs to vascular tissue13.
Interestingly, experimental and clinical studies in postmenopausal women have suggested that bisphosphonates, which are commonly used to treat osteoporosis by reducing bone turnover, may also reduce arterial inflammation and calcification8, 14-18. A recent analysis of a longitudinal cohort study further demonstrated that treatment with bisphosphonates resulted in a lower prevalence of cardiovascular calcification in women older than 65 years of age19. Although the exact mechanisms by which bisphosphonates inhibit vascular calcification are not entirely understood, several hypotheses have been suggested, including an indirect effect through inhibition of bone remodeling and a direct effect of these drugs on the vascular wall20, 21. To further evaluate the possible mechanisms by which bisphosphonates may regulate vascular calcification, in this study we tested whether treatment of healthy postmenopausal women with the bisphosphonate, risedronate, resulted not only in a decrease in bone turnover but also in a reduction in EPCs co-expressing osteoblastic cell surface markers and genes. In addition, we analyzed the effect of risedronate on circulating levels of inflammatory cytokines as well as OPG and RANKL levels.
For this double blind, randomized study we recruited 20 healthy postmenopausal women who had cessation of menses for more than a year but who were within 5 years of their last menstrual period. Patients were recruited through an institutional classified advertisement seeking research participants at the Mayo Clinic in Rochester , MN, and were included during a 6-month enrollment period (from February 1, 2008 through July 31, 2008). Screening laboratory studies included a complete blood count, serum levels of 25-hydroxyvitamin D (25OHD), follicle stimulating hormone (FSH), parathyroid hormone (PTH), creatinine, calcium, and phosphorus. Exclusion criteria were: use of bisphosphonates or other bone-active drugs in the previous 3 years; history of metabolic bone disease, diabetes, or significant cardiac, renal, or liver disease; history of fracture within the last 5 years; hysterectomy; history of esophageal reflux/stricture; abnormalities in the screening laboratory studies. The study was approved by the Mayo Institutional Review Board and all subjects provided written, informed consent to participate.
The study subjects received placebo or 35 mg weekly risedronate for 4 months (n=10 per group). All patients were instructed to take the drug with water on an empty stomach at least 30 minutes before breakfast. Patients complied to the treatment as assessed by interview of the patients at the end of the study. Peripheral blood was collected to determine serum bone turnover markers, levels of IL-8, hsCRP, OPG, RANKL and to obtain peripheral blood mononuclear cells (PBMNCs) for flow cytometry. After 4 months the measurements were repeated and CD34+/KDR+ cells isolated for gene expression analysis.
PBMNCs were stained with fluorescent conjugated antibodies: CD34 (Beckton-Dickinson), CD133 (Miltenyi Biotec GmbH) and KDR (R&D Systems). Co-staining for osteoblast markers (OCN, Stro-1, and alkaline phosphatase [AP]) was performed using anti-human OCN (Santa Cruz), anti-human Stro-1 (R&D Systems), biotinylated anti-human AP (R&D Systems) antibodies. Cell fluorescence was measured immediately after staining (Becton Dickinson, FACS Calibur) and data were analyzed using the CellQuest software (Becton Dickinson). Based on the surface antibody expression and in order to subclassify the cells EPCs were divided into four different populations 1) CD34+/KDR+; 2) CD34-/CD133+/KDR+; 3) CD34+/CD133+/KDR+; and 4) CD34+/CD133-/KDR+1, 9.
WT-OvationTM Pico RNA linear amplification (NuGEN, Technologies, Inc) was used to synthesize cDNA from total RNA of CD34+/KDR+ cells. Primers for bone and stem cell-related genes (apoptosis genes: Bax, Bcl-2, Bcl-XL, Caspase 3, Caspase 8, Fas, P53; BMP targets: Id2, Smad 1, Smad 5, Sox 4, TIEG; osteoblast differentiation: BSP, Col1α2, OCN, Osteonectin, Runx 2; proliferation genes: Cyclin B1, Cyclin C, Lef 1; Wnt signaling: Axin, b-Catenin, Tcf-7, Veriscan, Wnt 4; others: OPG, RANKL) were used in QPCR. Sample normalization was performed using the ribosomal protein, L13. Individual gene expression was determined by 2−δCT
Venous blood was drawn at 8 am at baseline and after 4 months following overnight fasting. Bone formation was assessed by measuring serum amino-terminal propeptide of type I procollagen (P1NP) by radioimmunoassay (Immunodiagnostic Systems [IDS] Ltd), interassay CV<9%). Bone resorption was assessed using serum carboxy-terminal telopeptide of type I collagen (CTx) (interassay CV<10%) and tartrate-resistant acid phosphatase 5b (TRAP5b) (interassay CV<4%), both measured by ELISA (IDS). Serum OPG and RANKL were measured using quantitative immunoassays (ALPCO Diagnostics, Windham, NH) (interassay CV8% and 9%, respectively). IL-8 was measured with the Human Ultrasensitive Cytokine 10-plex Assay using the Luminex® xMAP® platform and ELISA assays (Invitrogen), interassay CV<10%. Hs-CRP levels were measured using a high-sensitivity solid phase direct sandwich ELISA (Calbiotech, Spring Valley, CA, interassay CV<8.5%).
Screening laboratory tests (including complete blood count, creatinine, calcium, and phosphorus) were analyzed using standard procedures.
The primary aims of the study were to evaluate the effect of risedronate on EPCs co-expressing osteoblast surface markers and genes. The secondary aims were to evaluate the effect of risedronate in inflammatory and bone turnover markers. Based on our experience with qPCR22, 10 subjects per group has been sufficient for detecting differences of between 1.5-3 fold in the expression of most genes.
The serum bone markers and flow cytometr data measured at four months were compared using a linear model, testing for a treatment group difference after adjusting for the baseline measurement. Based on model diagnostics, the log transformation was used for the flow cytometry variables. The data is summarized as mean ± SEM. Percent changes of each variable between baseline and 4 months was calculated, and comparisons of these values were made using the Spearman Rank correlation ignoring treatment status. Gene cluster analysis was performed using the O’Brien Umbrella method23,with data presented as medians and 25th-75th percentiles (interquartile range [IQR]). The data was analysed using an intent-to-treat approach. P <0.05 was considered statistically significant.
The relevant clinical and biochemical data of the study subjects are shown in Table 1. At entry, all subjects had normal serum calcium, phosphorus, creatinine, 25OHD and PTH values. As expected, FSH serum levels were elevated (84 ± 19 U/L) in all patients, consistent with the postmenopausal status of the study subjects. Even though baseline serum phosphorus and creatinine values were slightly higher in the risedronate group, they were within the normal range. All patients reported compliance with the treatment during the study as assessed by an interview at the end of the study.
Figure 1 shows the mean values of bone formation (P1NP) and bone resorption (TRAP5b and CTx) markers in patients treated with placebo and risedronate at baseline and after 4 months of treatment. Serum P1NP and TRAP5b levels decreased significantly after risedronate treatment, with mean decreases of 38% and 22%, respectively. Patients treated with placebo also showed a significant decrease in both markers but of a smaller magnitude: serum P1NP decreased by 14% and TRAP5b levels by 9.7%, a change attributed to the seasonal variation in bone markers24. There was a significant difference in the change of P1NP from baseline to 4 months between the risedronate and the placebo group (p=0.01) and similarly for TRAP5b (p=0.03). CTx levels decreased with risedronate treatment, albeit, not significantly (Figure 1), a finding that was attributed to the higher variability of this marker24.
Serum IL-8, OPG, RANKL and hsCRP values were not significantly different either at baseline or after risedronate treatment (Table 2). Spearman rank correlation was used to assess the association between percent change from baseline to 4 months in serum OPG and percent change from baseline to 4 months in hsCRP ignoring treatment status (r=.4, p=0.01). Likewise r=.3, p=0.03 for the association between percent change from baseline to 4 months in RANKL and percent change to 4 months in IL-8. Rho=.5, p=0.04 for the association between percent change from baseline to 4 months in RAKL and percent change from baseline to 4 months in hsCRP.
On average we isolated 8630±995 CD34+/KDR+ cells in the groups. The gene set analysis for osteoblastic differentiation (bone sialoprotein [BSP], collagen 1 alpha 2 [Col 1α2], OCN, osteonectin [ON], runx 2) and proliferation (cyclin C, cyclin B1, Lef 1) demonstrated a significant downregulation of all these genes after risedronate treatment (Figure 2).
CD34-/CD133+/KDR+ decreased after risedronate therapy (119 [IQR 25th: 59; 75th: 142] to 51 [IQR 25th: 37; 75th: 78] but this decrease was not significantly different from the placebo group decrease (109 [IQR 25th: 49; 75th: 159] to 71 [IQR 25th: 53; 75th: 95]). Neither treatment resulted in significant changes in any of the other EPC subpopulations.
Figure 3 (A-D) shows that all of the EPC subpopulations co-staining for OCN tended to decrease in the risedronate group and either increase or remain unchanged in the placebo group, but this pattern was not significantly different between the two groups. Analysis of the additional cell surface markers, AP and Stro1 was performed in all EPC populations in both groups of patients at baseline and at 4 months. Overall, risedronate treatment resulted in a ~59% reduction in CD34+/KDR+ EPCs co-expressing osteoblastic surface markers (−63% in EPCs co-stained with Stro-1, −54% with AP), whereas placebo treatment showed an average reduction of 4%. These changes, however, were not significantly different between the two groups.
EPCs co-staining for AP and OCN were significantly correlated in CD34/KDR/OCN versus CD34/KDR/AP populations (r=0.61, P=0.005) and CD34+/CD133-/KDR+ populations (r=0.47, P=0.03). Conversely, EPCs co-expressing Stro-1 did not correlate with either AP or OCN expressing EPCs
We observed a direct correlation between the bone resorption markers (TRAP5b and CTx) and CD34+/CD133+/KDR+ EPCs co-expressing AP (r = 0.54, P = 0.01 for TRAP5b; r = 0.62, P = 0.003 for CTx). Moreover, on analyzing the correlations between the percentage changes of EPCs co-expressing osteoblastic markers and those of OPG, RANKL and the inflammatory markers, we observed a direct correlation between changes in hsCRP levels and changes in CD34/KDR EPCs co-expressing OCN (r = 0.5, P = 0.04).
The results of our study demonstrate that bisphosphonate therapy in healthy postmenopausal women not only results in a decrease in bone turnover but also lower expression of osteoblast-related genes, with a trend of decreasing the expression of osteoblastic cell surface markers by circulating EPCs. The current results expand our previous findings1, 2 and further generate the hypothesis that bisphosphonates may inhibit vascular calcification by preventing EPCs from developing an osteogenic phenotype. In a broader context, our work also raises the likelihood that bisphosphonates may modulate the process of endothelial-to-mesenchymal transition3, 4 and point to the need for further studies to address these intriguing possibilities.
We observed that the expression of genes related to cell proliferation such as cyclin C, cyclin B1, and lef 1 in CD34+/KDR+ cells was significantly lower in women receiving risedronate as was the expression of genes related to osteoblastic differentiation, such as BSP, collagen type 1, OCN, ON and Runx2, further supporting an anti-osteogenic effect of this therapy on EPCs. Although previous studies have indicated that bisphosphonates can enhance the differentiation and proliferation of osteoblastic cells25, the effect of these agents likely depends on the cell type and the bisphosphonate used26. Indeed, recent studies have found that zoledronic acid and other compounds have an inhibitory effect on the proliferation and differentiation of endothelial cells16, 27, 28. It is likely that bisphosphonate treatment decreases the clonal expansion capacity of EPCs. Furthermore, we also observed a consistent downregulation of genes related to apoptosis as well as Wnt and BMP targets in CD34+/KDR+ cells following risedronate treatment, although without statistical significance. In agreement with our results, recent data suggest that the use of nitrogen-containing bisphosphonates, especially in older women, is associated with decreased prevalence of vascular calcification19, 27.
We also observed that risedronate treatment tended to decrease the number of different EPC populations co-expressing OCN. Furthermore, analysis of additional osteoblastic cell surface markers (AP and Stro-1) showed that following risedronate treatment, fewer cells of the different EPC subpopulations co-expressed, albeit not significantly, these markers. Despite previous studies showing a relationship between circulating EPC levels and cardiovascular outcomes, with higher risk related to lower EPC levels29, recent data indicate that the EPC phenotype may play an important role in endothelial repair10. Our group has shown that EPCs co-expressing an osteogenic phenotype are significantly increased in patients with severe coronary artery disease or endothelial dysfunction and that these cells can mineralize, at least in vitro1. Indeed, although in most studies EPCs are identified by flow cytometric characteristics, specifically by the expression of CD34, CD133, or KDR, the origins and functions of EPCs remain controversial. Moreover, EPCs seem to fulfill varying roles at different stages of their development, e.g. late EPCs seem to have a higher proliferative capacity in vitro, whereas early EPCs may act to secrete angiogenic growth factors10. Furthermore, the concept of “osteogenic” versus “non-osteogenic” EPCs could partially explain, previous discordant results in experimental studies where treatment with EPCs or with bone marrow mononuclear cells may accelerate atherosclerotic plaque formation instead of improving vascular function30, 31. Also relevant to our findings is that bisphosphonates have been associated with anti-angiogenic effects, a finding that has been related to the anticancer activities of these agents and also to the development of jaw osteneocrosis, especially with the most potent amino-bisphosphonates such as zoledronic acid. Therefore, we cannot exclude effects of risedronate on either an inhibition of EPCs migration or increased cell apoptosis 33.
Although the evidence that EPCs co-expressing osteogenic markers are involved in vascular calcification is indirect, the in vitro capacity of these cells to calcify leads to the hypothesis that EPCs may contribute to vascular calcification. Interestingly, a recent study identified a novel type of blood-derived procalcific cell potentially involved in vascular calcification of diabetic patients. These cells had a myeloid origin but also expressed OCN and AP32.
We also found that the bone resorption markers, TRAP5b and CTx, were positively correlated with the number of CD34+/CD133+/KDR+ EPCs co-expressing osteoblast surface markers. These findings indicate a possible relationship between the bone turnover rate and circulating osteogenic EPCs and suggest the possibility that increased bone turnover may contribute to the development of an osteogenic phenotype by circulating EPCs. Indeed, a recent study has reported a positive relationship between bone turnover markers and the number of circulating CD34 cells co-expressing AP or OCN34. All these findings coincide with the previously reported association between increased bone turnover and vascular calcification, linking bone metabolism with the vascular system5, 6, 8, 35, 36. Conversely, Stro-1 expression by EPCs was not correlated with AP or OCN EPC populations or with bone turnover markers. It has previously been reported that, compared to osteoblastic cells expressing AP, those only expressing the Stro-1 antigen represent a less differentiated osteoblast population with reduced capacity for mineralization and lack of expression of various bone-related markers37. Our data thus suggest that EPC subpopulations co-expressing AP and OCN may be more differentiated and consequently, may have a higher capacity for mineralization and vascular calcification. Furthermore, since our previous work had only stained EPC populations with OCN1, 2, the significant correlation we noted between EPCs co-staining for OCN and AP provides greater confidence that staining for OCN identifies EPC populations also likely to express other bone-related proteins.
Several lines of evidence suggest that inflammation plays a major role in the development of vascular calcification11, 38, 39 and that various cytokines and proteolytic enzymes regulate the release, migration and homing of progenitor cells from the bone marrow12. Nonetheless, despite evidence for an immunomodulatory effect of bisphosphonates16, most data are based on animal or in vitro studies40, 41 or on patients with inflammatory or metabolic bone disease16, 42, 43. In the present study, inflammatory markers, such as IL-8 and hsCRP, and bone regulatory factors, such as OPG and RANKL, tended to decrease, albeit not significantly, with bisphosphonate therapy, likely due to the relatively small number of subjects included. Moreover, serum hsCRP levels were directly correlated with OPG concentrations and changes in hsCRP levels were also correlated with changes in RANKL and in CD34+/KDR+ EPCs co-expressing OCN.
This study has several limitations. Our subjects were healthy postmenopausal women. Nonetheless, it remains unknown whether the presence of associated vascular disease or osteoporosis may modify the changes we observed following bisphosphonate therapy. Furthermore, we did not assess parameters of vascular calcification and the number of individuals was low, limiting the strength of our results. The biological effects of bisphosphonates may also vary depending on the bisphosphonate used16; therefore, different response with other bisphosphonates cannot be ruled out. However, despite the small sample size of the study the changes in the osteoblastic potential of the EPC populations were consistent with downregulation of the osteoblastic differentiation genes in EPCs after risedronate treatment.
In conclusion, our data indicate that treatment with a bisphosphonate in healthy postmenopausal women may modulate cellular pathways leading to vascular calcification by downregulating the expression of osteogenic genes in EPCs. However, these preliminary findings need further validation in larger studies and additional evidence to define a role for EPCs in inducing vascular calcification. Nonetheless, our results provide a rationale for further studies examining the possible effects of bisphosphonate therapy on expression of osteogenic markers by EPCs as well as on vascular calcification.
We would like to thank James Peterson and Kelley Hoey for technical support.
Funding Sources: This work was supported by NIH Grants P01 AG004875, RO1 HL 092954 and 1UL1RR02415 (Mayo Center for Translational Science Activities) and an investigator-initiated grant from Procter and Gamble.
This work was supported by NIH Grants P01 AG004875, RO1 HL 092954 and 1UL1RR024150 (Mayo Center for Translational Science Activities) and an investigator-initiated grant from Procter and Gamble.
Disclosures: The authors have no conflicts of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Current department: Rheumatology Department, Hospital Clinic, IDIBAPS, CIBERehd, Barcelona.
Current department: Division of Imaging Sciences and Biomedical Engineering, Rayne Institute, St. Thomas’ Hospital, King’s College London, UK.
Pilar Peris, Endocrine Research Unit, Mayo Clinic, Rochester, MN.
Elizabeth J. Atkinson, Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN.
Mario Gössl, Division of Cardiovascular Diseases, Mayo Clinic, Rochester, MN.
Trevor L. Kane, Endocrine Research Unit, Mayo Clinic, Rochester, MN.
Louise K. McCready, Endocrine Research Unit, Mayo Clinic, Rochester, MN Amir Lerman, MD, Division of Cardiovascular Diseases, Mayo Clinic, Rochester, MN.
Sundeep Khosla, Endocrine Research Unit, Mayo Clinic, Rochester, MN.
Ulrike I. Mödder (McGregor), Endocrine Research Unit, Mayo Clinic, Rochester, MN.