Voriconazole Enhances the Osteogenic Activity of Human Osteoblasts In Vitro through a Fluoride-Independent Mechanism

Abstract

Periostitis, which is characterized by bony pain and diffuse periosteal ossification, has been increasingly reported with prolonged clinical use of voriconazole. While resolution of clinical symptoms following discontinuation of therapy suggests a causative role for voriconazole, the biological mechanisms contributing to voriconazole-induced periostitis are unknown. To elucidate potential mechanisms, we exposed human osteoblasts in vitro to voriconazole or fluconazole at 15 or 200 μg/ml (reflecting systemic or local administration, respectively), under nonosteogenic or osteogenic conditions, for 1, 3, or 7 days and evaluated the effects on cell proliferation (reflected by total cellular DNA) and osteogenic differentiation (reflected by alkaline phosphatase activity, calcium accumulation, and expression of genes involved in osteogenic differentiation). Release of free fluoride, vascular endothelial growth factor (VEGF), and platelet-derived growth factor (PDGF) was also measured in cell supernatants of osteoblasts exposed to triazoles, with an ion-selective electrode (for free fluoride) and enzyme-linked immunosorbent assays (ELISAs) (for VEGF and PDGF). Voriconazole but not fluconazole significantly enhanced the proliferation and differentiation of osteoblasts. In contrast to clinical observations, no increases in free fluoride levels were detected following exposure to either voriconazole or fluconazole; however, significant increases in the expression of VEGF and PDGF by osteoblasts were observed following exposure to voriconazole. Our results demonstrate that voriconazole can induce osteoblast proliferation and enhance osteogenic activity in vitro. Importantly, and in contrast to the previously proposed mechanism of fluoride-stimulated osteogenesis, our findings suggest that voriconazole-induced periostitis may also occur through fluoride-independent mechanisms that enhance the expression of cytokines that can augment osteoblastic activity.

INTRODUCTION

Invasive fungal infections (IFIs) rarely occur in immunocompetent individuals but are potentially devastating in certain patient populations, including patients undergoing intensive chemotherapy or receiving autologous or allogeneic hematopoietic stem cell transplants for the treatment of hematological malignancies (1, 2). In recent years, there have been reports of cutaneous IFIs among immunocompetent individuals, with traumatic injuries, including motor vehicle accidents, natural disasters, and combat-related trauma, being cited as significant risk factors for infection with environmental molds, of which Aspergillus spp. are among the most commonly isolated (3–5). The recommended therapy for treatment of cutaneous IFIs involves extensive surgical debridement combined with empirical systemic antifungal therapy, for which fluorinated triazoles, including voriconazole, are important options for susceptible fungi. As the incidence rates of IFIs have been reported to be as high as 12% among immunocompromised patients (6) and up to 3.5% among U.S. military personnel admitted to Landstuhl Regional Medical Center, Germany, for combat trauma occurring in Afghanistan from June 2009 through December 2010 (7, 8), the use of triazoles for prophylactic and/or prolonged systemic therapy is likely to increase.

Voriconazole use has been associated with a number of side effects, including visual disturbances, hallucinations, edema, hepatotoxicity, phototoxicity, and cutaneous carcinogenesis (9, 10). More recently, reports that document painful periostitis among patients receiving voriconazole for prolonged periods are accumulating (11–15). Periostitis is characterized by multiple areas of periosteal ossification of bones in the axial and appendicular skeleton, which typically abate following discontinuation of therapy, suggesting a direct causative role for voriconazole. In patients with voriconazole-related periostitis, elevations in serum fluoride and bone alkaline phosphatase (ALP) levels have been observed (13, 16–18). These findings are similar to those for individuals with skeletal fluorosis, a pathological condition characterized by disturbances of bone homeostasis resulting from chronic fluoride intake, typically through endogenous water sources (19, 20), and they suggest a role for free fluoride in the underlying pathology.

Although the exact mechanisms vary, the pathogenesis of a number of human disorders characterized by abnormal bone formation, including heterotopic ossification (HO), hypertrophic osteoarthropathy (HOA), and fluorosis, is attributed to enhanced activity of osteoblast precursor cells (19, 21–23). Osteoblasts facilitate new bone formation by differentiating into mature osteocytes capable of producing and depositing large amounts of bone extracellular matrix (24). Differentiation of osteoblasts into mature osteocytes is regulated primarily by the Wnt/β-catenin signaling pathway, in which Wnt ligands, including bone morphogenic proteins (BMPs), bind to osteoblast cell surface receptor proteins, activating the transcription factor β-catenin and increasing the downstream expression of genes involved in the regulation of osteoblast differentiation (e.g., runt-related transcription factor 2 [runX2]) and bone matrix formation (e.g., alkaline phosphatase [alp], osteocalcin [ocn], and collagen I [colI]) (25, 26). In addition to Wnt signaling and as part of normal osteoblast differentiation, activation of an endoplasmic reticulum (ER) stress response and subsequent unfolded protein response (UPR) occurs to maintain cell homeostasis during the production of the large quantities of extracellular bone matrix proteins required for new bone formation (27, 28). In mammals, UPR signaling involves three ER-localized signal transducers, namely, inositol-requiring enzyme 1 (IRE1), double-stranded RNA-activated Prk-like ER kinase (PERK), and activating transcription factor 6 (ATF6) (28, 29); roles for IRE1 and PERK during osteoblast differentiation have been described (27, 30). Fluoride, similar to other Wnt agonists, was documented previously to have anabolic effects on bone metabolism and to stimulate osteoblastic bone formation in vitro through activation of the canonical Wnt/β-catenin signaling pathway (31–33) and the endoplasmic reticulum (ER) stress response and subsequent unfolded protein response (UPR) (34, 35).

Given that voriconazole contains three fluorine atoms and clinical findings for voriconazole-induced periostitis are similar to those for skeletal fluorosis, a mechanism in which excess fluoride, liberated by voriconazole metabolism, enhances osteoblastic activity, producing clinical symptoms, has been proposed (11, 13, 14, 16). Indeed, this would fit with the previously described effects of fluoride on osteoblast proliferation and differentiation. However, this hypothesis has not been confirmed, and the specific mechanisms by which voriconazole may promote the development of periostitis are unknown. To explore potential mechanisms of voriconazole-induced periostitis, we evaluated in vitro osteoblastic activity in response to triazole exposure, by assessing effects on cell proliferation, osteogenic differentiation, and expression of soluble factors involved in bone remodeling and repair.

MATERIALS AND METHODS

Reagents.

Voriconazole (Selleck Chemicals, Houston, TX) and fluconazole (Sigma-Aldrich, St. Louis, MO) were diluted to final concentrations of 10 mg/ml in sterile water containing sulfobutylether-cyclodextrin (SBECD) (Captisol; Ligand Technology, La Jolla, CA) at a final concentration of 160 mg/ml. Fluconazole, a difluorinated azole, was selected for comparison with voriconazole because it has been in long-term use for antifungal treatment but cases of periostitis have not been reported in more than 2 decades of human use. Concentrations reflecting the reported interstitial fluid voriconazole concentrations in humans, i.e., ∼15 μg/ml (36), and surgical practices of direct topical application to wounds in IFI patients, i.e., 200 μg/ml (one vial containing 200 mg voriconazole reconstituted in 1 liter of saline solution), were selected for use in this study. The latter concentration (200 μg/ml) reflects an empirical practice of some U.S. military surgeons attempting to control life-threatening soft tissue-invasive mold infections in combat casualties from southern Afghanistan with severe blast injuries (8). As a positive-control group for stimulation of osteoblastic activity, sodium fluoride (Sigma-Aldrich) treatment of cells was also included, using concentrations (10 μM) shown previously to promote osteoblastic activity in vitro (32).

Osteoblast culture conditions and osteogenic differentiation.

Primary human osteoblasts (Promocell, Heidelberg, Germany) were maintained at 37°C in 5% CO₂ in α-minimal essential medium (α-MEM) supplemented with fetal bovine serum (10% [vol/vol]), penicillin (10 U/ml), and streptomycin (10 μg/ml). Osteoblasts were seeded into 48-well tissue culture wells and allowed to grow to ∼80% confluence. For differentiation, osteoblasts at 80% confluence were then grown in osteoinductive medium (α-MEM supplemented with 50 μM ascorbic acid, 5 mM β-glycerolphosphate, and 0.1 μM dexamethasone) for up to 7 days. Osteoblasts were treated with either 15 or 200 μg/ml voriconazole or fluconazole, diluted in cell medium, for up to 7 days during culture with or without osteoinductive agents. As experimental controls for each assay, culture medium containing the equivalent concentrations of SBECD without triazoles diluted in medium without osteoinductive agents was used a negative control, whereas sodium fluoride (10 μM) in medium without osteoinductive agents was used as a positive control.

Osteoblast proliferation.

Osteoblasts (5 × 10⁴ cells/well) were seeded into 48-well tissue culture wells and treated with either voriconazole or fluconazole for up to 7 days. At days 1, 3, and 7, the medium was removed, the cells were washed and lysed, and total cellular DNA was quantified using the CyQUANT cell proliferation assay (Molecular Probes, Grand Island, NY). Total cellular DNA was determined against a standard curve prepared using bacteriophage λ DNA.

Measurement of alkaline phosphatase.

Osteoblasts (2 × 10⁴ cells/well) were seeded into 48-well tissue culture plates and cultured with cell medium containing osteoinductive agents, as described above. Quantification of ALP activity was carried out using the SensoLyte pNPP alkaline phosphatase assay (AnaSpec, Fremont, CA). In brief, at study time points, cells were washed and lysed using 0.1% Triton X-100, and ALP activity was determined by measuring the absorbance of metabolized p-nitrophenyl phosphate at 405 nm. Total ALP concentrations were determined by using a standard curve and were normalized to total protein amounts in cell lysates.

Alizarin red staining.

Calcium deposition after 7 days of culture of osteoblasts in the presence of osteoinductive agents, with or without triazole treatment, was determined by staining osteoblasts with 2% (wt/vol) alizarin red S in deionized water (pH 7), followed by extraction of alizarin red from cells using 0.5 N HCl in 5% SDS solution. Quantification of intracellular calcium was performed by measuring absorbance at 415 nm. Calcium deposits prior to extraction were visualized by light microscopy (Olympus IX71 microscope; Olympus, Center Valley, PA).

Osteoblast gene expression.

Gene expression at the specified time points was evaluated by quantitative real-time PCR (qRT-PCR). Cellular RNA was extracted from osteoblasts cultured for up 7 days with the different treatments by using an RNAeasy minikit (Qiagen, Valencia, CA). First-strand synthesis was achieved with SuperScript III SuperMix with oligo(dT) primers (Invitrogen, Carlsbad, CA) for each RNA sample, using a PTC-100 thermal cycler (GMI Inc., Ramsey, MN). qRT-PCR was performed using a Bio-Rad C1000 system, and data were analyzed using iQ5 software (Bio-Rad, Hercules, CA). Primer sets, including reverse and forward primers, were purchased from SABiosciences, and amplification reactions were performed using SYBR Green MasterMix (Bio-Rad, Hercules, CA), following the manufacturer's recommended preoptimized cycling conditions. Transcript levels were normalized to the mRNA expression of the internal control gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Changes in relative gene expression were calculated using the 2^−ΔΔCT method. Values were expressed as fold increases relative to values for untreated cells grown under osteoinductive conditions.

Fluoride ion detection.

Fluoride levels in the supernatants collected from osteoblasts exposed to fluconazole or voriconazole were measured using a fluoride combination electrode (Hanna Instruments, Carrollton, TX) attached to a benchtop pH/electrochemical meter with an ion-selective setting (Beckman-Coulter pHi 570; Beckman-Coulter, Brea, CA). All standards and samples were diluted 1:1 with total ionic strength adjustment buffer (TISAB) II ion suppression buffer (Hanna Instruments), and fluoride concentrations were determined against calibration standards at concentrations of 1.0, 0.5, 0.1, 0.05, 0.025, and 0.01 ppm. The limit of quantification was determined to be 0.025 ppm.

Quantification of soluble cytokines.

Enzyme-linked immunosorbent assays (ELISAs) were performed to quantify cytokines in culture supernatants from osteoblasts exposed to triazoles. ELISA kits for human vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF) (Abcam, Cambridge, MA) were all used according to the manufacturer's recommendations. In brief, cell supernatants were collected on days 1 and 5 from individual wells containing osteoblasts, following exposure to either voriconazole or fluconazole. Cytokine levels in supernatants were quantified by extrapolation from standard curves generated using recombinant protein for each cytokine, and the concentrations determined were normalized to the total protein content in each supernatant sample, as determined with the bicinchoninic acid (BCA) assay.

Statistical analysis.

All experiments were performed in triplicate, as three independent experiments performed at different times. For in vitro comparisons of groups, statistical analyses were performed using an unpaired Student's t test or one-way analysis of variance (ANOVA) with a Bonferroni post hoc test to determine statistical differences between groups, by using GraphPad InStat version 3.0 (GraphPad Software, San Diego, CA). P values of <0.05 were considered to be statistically significant.

RESULTS

Voriconazole enhances cell proliferation in human osteoblasts.

Exposure of human osteoblasts to voriconazole or fluconazole at 15 and 200 μg/ml (diluted in water containing SBECD) under nonosteogenic conditions did not impair osteoblast viability over the 7-day treatment period (Fig. 1A and B). Exposure to voriconazole at 200 μg/ml enhanced proliferation after 3 days (257 ± 21 ng/ml DNA; P < 0.01) and 7 days (503 ± 53 ng/ml DNA; P < 0.01) of treatment, compared to untreated controls at the same study time points (182 ± 66 and 268 ± 27 ng/ml DNA, respectively; P < 0.01), whereas significant effects on proliferation were observed only after 7 days of treatment with 15 μg/ml voriconazole, compared to untreated groups (380 ± 66 versus 268 ± 27 ng/ml DNA; P < 0.05) (Fig. 1A). Notably, the proliferative effects of voriconazole on osteoblasts were observed to be similar to the effects of fluoride, which was shown recently to enhance osteoblast growth in vitro (Fig. 1). In contrast, exposure of osteoblasts to fluconazole at 15 and 200 μg/ml had moderate but nonsignificant effects on cell proliferation after 7 days (322 ± 51 ng/ml DNA [P = 0.09] and 337 ± 65 ng/ml DNA [P = 0.07], respectively), compared to untreated control and voriconazole-treated groups (Fig. 1B).

FIG 1.

Voriconazole enhances cell proliferation of human osteoblasts. Proliferation of human osteoblasts exposed to voriconazole (VCZ) (A) or fluconazole (FCZ) (B) at 15 or 200 μg/ml, under nonosteogenic culture conditions, for up to 7 days was determined by measuring total DNA using the CyQUANT cell proliferation assay. As positive and negative controls, cells were exposed to either sodium fluoride (NaF) at 10 μM or medium containing equivalent Captisol (sulfobutylether-cyclodextrin) levels (10 mg/ml), respectively. Values represent the means ± standard deviations (SD). *, P < 0.05, compared to the untreated control group at each time point.

Exposure of osteoblasts to voriconazole, but not fluconazole, enhances osteoblastic activity in vitro.

Consistent with previous reports, exposure of osteoblasts to fluoride in medium lacking osteoinductive agents was observed to enhance osteogenic activity over the 7-day period (Fig. 2). In contrast, exposure of osteoblasts to voriconazole or fluconazole, at either tested concentration, under nonosteoinductive conditions had no significant effect on osteogenic activity, compared to the control group grown under osteoinductive conditions (data not shown). However, exposure of osteoblasts cultured in the presence of osteoinductive agents to voriconazole, but not fluconazole, resulted in significant increases in osteoblastic activity, in a dose-dependent manner (Fig. 2). Voriconazole enhanced osteoblast ALP activity (Fig. 2A) and increased intracellular calcium levels within osteoblasts, as detected by alizarin red staining (Fig. 2B and C). Consistent with these findings, the expression of genes involved in osteoblast differentiation (runt-related transcription factor 2 [runx2]) and bone matrix release by osteoblasts (alkaline phosphatase [alp], osteocalcin [ocn], bone morphogenic protein 2 [bmp2], and collagen I [colI]) was upregulated following exposure to voriconazole after 7 days, as determined by qRT-PCR (Fig. 3). Expression of bmp4 was highest in cells treated with 200 μg/ml voriconazole. In agreement with these results and recent reports describing activation of the UPR during osteoblast differentiation (27, 37), exposure to voriconazole also differentially enhanced the expression of two of the three ER-localized signal-transducing components of the UPR (Fig. 4). Specifically, voriconazole (at 200 μg/ml) increased expression of the Prk-like ER kinase (perk) and, to a lesser extent, X-box-binding protein 1 (xbp1). As confirmation of the activity of PERK signaling, gene expression of downstream targets of PERK, including activation transcription factors 3 and 4 (atf3 and atf4), DNA damage-inducible transcript 3 (chop), and growth arrest and DNA damage-inducible transcript 34 (gadd34), was increased following exposure to voriconazole, compared to fluconazole (Fig. 5). Increases in the expression of proteins involved in osteogenic differentiation or the UPR pathway were also observed in cells exposed to fluoride, although the effects of fluoride on osteoblasts were not dependent on the presence of osteogenesis-inducing agents (Fig. 2 and 3).

FIG 2.

Voriconazole increases the osteogenic activity of human osteoblasts in vitro. Osteoblasts were exposed to voriconazole (VCZ) or fluconazole (FCZ) at 15 or 200 μg/ml in osteoinductive medium (OS+) for up to 7 days. As a positive control, cells were also exposed to sodium fluoride (NaF) at 10 μM in nonosteoinductive medium (OS−). Osteoblastic differentiation was evaluated by measuring alkaline phosphatase (ALP) activity (A) and staining for calcium deposition with alazarin red S stain (B and C); calcium deposition was quantified by measuring the absorbance (A₄₁₅) of extracted alazarin red S (B) and was assessed by the light microscopic appearance (white arrows) of stained osteoblast cultures (C). Values represent the means ± SD. *, P < 0.05, compared to osteoblasts grown under osteogenic conditions without treatment.

FIG 3.

Voriconazole enhances the expression of genes involved in osteogenic differentiation. The relative expression of genes involved in the regulation and differentiation of osteoblasts was determined by qRT-PCR analysis following 7 days of exposure to 15 or 200 μg/ml voriconazole (VCZ) or fluconazole (FCZ) or 10 μM sodium fluoride (NaF). Gene expression levels were normalized to levels of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene, and changes in relative expression were calculated using the 2^−ΔΔCT method. Values are expressed as relative fold increases, compared to osteogenic control groups. Values represent the means ± SD from three independent experiments. *, P < 0.05, relative to osteoblasts grown under osteogenic conditions without treatment.

FIG 4.

Voriconazole activates the unfolded protein response (UPR) pathway in human osteoblasts. The relative gene expression of UPR signal transducers in osteoblasts was evaluated by qRT-PCR on day 3 (D3) or day 7 (D7) of exposure to 15 or 200 μg/ml voriconazole (VCZ) or fluconazole (FCZ) or 10 μM sodium fluoride (NaF). Gene expression levels were normalized to levels of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene, and changes in relative expression were calculated using the 2^−ΔΔCT method. Values are expressed as relative fold increases, compared to the osteogenic control group. Values represent the means ± SD from three independent experiments. *, P < 0.05, relative to osteoblasts grown under osteogenic conditions without treatment. xbp1, X-box-binding protein 1; atf6, activating transcription factor 6; perk, Prk-like endoplasmic reticulum kinase.

FIG 5.

Voriconazole enhances the expression of PERK target genes in human osteoblasts. The relative gene expression of signal transducers downstream of PERK in osteoblasts was evaluated by qRT-PCR on day 3 (D3) or day 7 (D7) of exposure to 15 or 200 μg/ml voriconazole (VCZ) or fluconazole (FCZ) or 10 μM sodium fluoride (NaF). Gene expression levels were normalized to levels of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene, and changes in relative expression were calculated using the 2^−ΔΔCT method. Values are expressed as relative fold increases, compared to osteogenic control groups. Values represent the means ± SD from three independent experiments. *, P < 0.05, relative to osteoblasts grown under osteogenic conditions without treatment. atf3, activating transcription factor 3; atf4, activating transcription factor 4; gadd34, growth arrest and DNA damage-inducible transcript 34; chop, DNA damage-inducible transcript 3.

Free fluoride levels in culture supernatants of osteoblasts exposed to voriconazole and fluconazole.

To explore the relationships between voriconazole, fluoride ions, and osteoblast stimulation, we examined the levels of free fluoride ions (Fl⁻) in culture supernatants from osteoblasts exposed to voriconazole or fluconazole. No significant increases in free fluoride ion levels in the supernatants of osteoblasts exposed to either voriconazole or fluconazole were detected at day 3 or day 7 (data not shown). For the groups treated with 200 μg/ml voriconazole or 200 μg/ml fluconazole, at days 3 and 7 we observed a nonsignificant trend of decreasing fluoride concentrations, with generally lower mean concentrations (0.042 ± 0.004 ppm for fluconazole and 0.048 ± 0.008 ppm for voriconazole on day 7) than for the 15-μg/ml exposures (0.048 ± 0.010 ppm for fluconazole and 0.053 ± 0.014 ppm for voriconazole) or the untreated control (0.057 ± 0.010 ppm). Collectively, these findings suggest that the increased osteoblastic activity observed following voriconazole exposure occurred in the absence of increased free fluoride ion levels.

Voriconazole exposure is associated with increased expression of cytokines involved in bone remodeling by osteoblasts.

Having observed no significant increases in free fluoride ion levels, we assessed the effects of voriconazole exposure on expression by osteoblasts of the cytokines VEGF and PDGF, which are known to augment osteoblastic activity and to play critical roles in bone development (38, 39). Osteoblasts exposed to voriconazole at 15 or 200 μg/ml had higher levels of VEGF and PDGF in the supernatants than did the untreated control group (Fig. 6). Consistent with the previous data, the effects of voriconazole on cytokine expression were much greater in the presence of osteogenic supplements. In contrast, fluconazole at either of the concentrations tested had little to no effect on the expression of either of the cytokines evaluated. Importantly, exposure of osteoblasts to fluoride only slightly increased the levels of VEGF and, to a lesser extent, PDGF, compared to groups of osteoblasts exposed to voriconazole (see Fig. S1 in the supplemental material).

FIG 6.

Voriconazole affects the expression of cytokines by osteoblasts. Quantification of the cytokines VEGF (A) and PDGF (B) in supernatants of osteoblasts exposed to 15 or 200 μg/ml voriconazole (VCZ) or fluconazole (FCZ), under nonosteoinductive conditions (OS−) or osteoinductive conditions (OS+), for 1 or 5 days was performed by ELISA. Cytokine levels in supernatants were quantified by extrapolation from standard curves generated using recombinant protein for each cytokine, and results were normalized to total protein levels. Values represent the means ± SD from three independent experiments. *, P < 0.05, relative to untreated control groups for each time point.

DISCUSSION

Clinical reports of periostitis, a reversible but painful bone condition, are accumulating in association with long-term use of voriconazole in humans. Included in these reports are observations of elevated serum fluoride and alkaline phosphatase levels, as well as radiographic abnormalities consistent with enhanced osteoblastic activity (11, 16–18). As these findings are similar to the clinical syndrome of skeletal fluorosis in humans, a pathogenic mechanism has been suggested whereby chronic daily consumption of trifluorinated voriconazole molecules leads to excess plasma fluoride, which then promotes the development of periostitis by enhancing the activity of osteoblasts (11, 13, 17). Although this is consistent with the known clinical and cellular effects of fluoride (19, 20, 31, 32, 40), this hypothesis has not been directly evaluated for voriconazole-associated periostitis. Therefore, the goal of this study was to evaluate this hypothesis by assessing the effects of voriconazole exposure on osteoblastic activity in vitro.

Exposure of osteoblasts to voriconazole or fluconazole at either of the concentrations tested was not observed to have any significant toxicity for the 7 days of exposure evaluated. This is consistent with the reported cytotoxicity of voriconazole and fluconazole against human endothelial cells and mononuclear cells, as well as mouse fibroblasts. In these cell lines, toxicity has been observed at concentrations exceeding 100 μg/ml. In a recent study evaluating voriconazole toxicity in mouse osteoblasts, reversible toxicity at concentrations of up to 1,000 μg/ml (5 times the highest concentration used in our experiments) was reported (although possibly toxic solvents were omitted from that report) (41). Our findings with human primary osteoblasts are consistent with the previously reported ranges of in vitro toxicity and indicate favorable tolerance of osteoblasts to triazoles at clinically relevant concentrations.

Interestingly, osteoblasts exposed to voriconazole, but not fluconazole, demonstrated significantly increased proliferation, compared to untreated groups. This intriguing finding is consistent with reports of clinical periostitis in patients receiving voriconazole but not in those treated with other triazoles, such as itraconazole, posaconazole, and fluconazole, for prolonged periods (13), and it suggests that chronic voriconazole administration may stimulate osteoblastic activity in vivo to produce this syndrome. To our knowledge, this is one of the first reports examining the effects of triazoles on osteoblastic activity. Given the fluorinated nature of triazoles, it is plausible that free fluoride contributed to these effects. This is indicated by the effect of fluoride on osteoblast proliferation that we observed, which is consistent with previous studies (31–33). However, the finding that fluconazole did not elicit a similar response in osteoblasts suggests that this effect is specific to voriconazole.

In addition to the observed effects on proliferation, exposure of osteoblasts to voriconazole significantly enhanced differentiation in cells grown in osteoinductive medium and, to a lesser extent, those grown in nonosteoinductive medium. Enhanced osteoblast differentiation, as indicated by increases in alkaline phosphatase and intracellular calcium levels and enhanced expression of genes involved in differentiation and the UPR stress pathway, compared to cells grown under osteoinductive conditions, is consistent with previous studies documenting the activation of these pathways during osteoblast differentiation (27, 28). Wnt signaling is known to affect the differentiation of osteoblasts and is generally divided into two major branches, i.e., the canonical and noncanonical Wnt pathways (42, 43). The canonical Wnt pathway, also called the Wnt/β-catenin pathway, is dependent on downstream activation of the transcriptional regulator β-catenin, which in turn upregulates the expression of genes involved in differentiation and the production of bone matrix. In contrast, noncanonical Wnt signaling involves pathways that promote differentiation independent of β-catenin. While our findings provide evidence that Wnt signaling was activated in osteoblasts exposed to voriconazole, the relationship between voriconazole and activation of the different Wnt signaling pathways was not explored directly. Future studies evaluating this further may provide a better understanding of these mechanisms. Notably, the majority of our statistically significant findings with regard to osteogenic activity resulted from exposure to voriconazole at 200 μg/ml, a much higher concentration than the observed steady-state interstitial fluid concentration (approximately 15 μg/ml) resulting from systemic dosing in humans (19). However, given that the average duration of voriconazole administration to patients with periostitis was 167 days, the limited effects of exposure of osteoblasts at the lower concentration might have been due to the short time frame utilized in our study.

Fluoride has the ability to stimulate new bone formation and to augment bone mass significantly, primarily by enhancing osteoblastic activity (31–33). Since voriconazole is a trifluorinated molecule and elevated plasma fluoride concentrations have been observed in patients with voriconazole-associated periostitis, it has been proposed that circulating free fluoride and its effects on osteoblasts may be involved in the mechanisms of voriconazole-induced periostitis. While voriconazole metabolism in humans is extensive and complex (less than 2% of the administered drug is excreted unchanged), free fluoride ion is only minimally liberated (approximately 5% of the dose) by voriconazole metabolism (44). In contrast to clinical studies that repeatedly noted substantial elevations in serum fluoride concentrations in periostitis (11, 16–18), we did not observe significant elevations in fluoride ion concentrations in the culture supernatants of either voriconazole- or fluconazole-exposed osteoblasts. This is not surprising, given that osteoblasts lack the enzymes responsible for the metabolism of triazoles. Although we cannot exclude a possible contribution of free fluoride to periostitis following prolonged voriconazole therapy, only voriconazole, and not fluconazole, stimulated osteoblastic activity, and only in the presence of osteogenesis-inducing agents. This contrasts with the activity of fluoride, which stimulated osteoblasts in non-osteogenesis-inducing medium, and suggests that voriconazole may stimulate osteoblasts through an unidentified, fluoride-independent mechanism. Importantly, our data raise questions regarding the origin of the high serum fluoride concentrations observed in patients with voriconazole-associated periostitis. Given that bones sequester fluoride over the human life span, it is possible that voriconazole stimulates fluoride release from bone. While fluoride has been demonstrated to inhibit osteoclastic activity in proportion to its concentration (40, 45, 46), to our knowledge no studies have examined the effects of voriconazole on osteoclasts.

Interestingly, voriconazole exposure was observed to increase the expression of the paracrine cytokines VEGF and PDGF in osteoblasts. Both of these factors play critical roles in osteogenesis, stimulating angiogenesis and enhancing osteoblast proliferation and differentiation (39, 47). Their increased expression following exposure to voriconazole but not fluconazole suggests that these factors may have contributed to the osteoblast proliferation and differentiation in the absence of significant increases in fluoride ion levels. This is consistent with reports of increased PDGF and VEGF expression being implicated in the pathogenesis of several other human diseases involving excessive bone proliferation. Compared to healthy controls, VEGF and PDGF were found at significantly higher levels in patients with hypertrophic osteoarthropathy (HOA), a syndrome characterized by digital clubbing, localized vascular hyperplasia, and periosteal proliferation of tubular bones (48, 49). Similarly, VEGF and PDGF are among the cytokines reported to exhibit elevated levels in wounds with heterotopic ossification (HO) (21, 50). Thus, there is precedent for VEGF- and PDGF-mediated stimulation of osteoblastic activity leading to human bone disease, and this axis may plausibly constitute a fluoride-independent mechanism linking voriconazole, osteoblast stimulation, and clinical periostitis. Importantly, while fluoride induced osteoblast differentiation, increases in VEGF and PDGF levels were smaller than those produced by voriconazole stimulation, suggesting that osteoblast differentiation may be promoted by two distinct mechanisms. This is supported by studies indicating that the effects of fluoride primarily involve Akt- and glycogen synthase kinase 3β (GSK-3β)-dependent activation of the Wnt/β-catenin signaling pathway, which does not necessarily involve VEGF and PDGF (32).

Although the present study carries the limitations inherent to an in vitro investigation, our findings offer insights into potential mechanisms underlying the development of periostitis after prolonged voriconazole treatment. In contrast to reports of elevated serum fluoride levels, our data indicate that voriconazole may enhance osteoblastic activity through a fluoride-independent mechanism, possibly by increasing the expression of growth factors involved in osteogenesis. Although we were not able to corroborate in vitro the findings of increased free fluoride levels observed in voriconazole-induced periostitis, in vivo studies are warranted to evaluate the relationship between fluoride and other systemic factors that may contribute to increased osteoblastic activity.