The goal of this study was to determine whether in vitro treatment of primary osteoclasts with TDF, the prodrug of tenofovir (), would alter gene expression as determined by microarray analysis, and provide insights into how drug exposure may influence osteoclast function. We chose primary murine osteoclasts for analysis as they provide a readily tractable model system to assess the mechanism(s) involved in tenofovir-mediated bone abnormalities that can be readily translated into mouse models of HIV infection. Such models would allow for the in vivo analysis of TDF administration on bone morphology in the context of HIV infection.
Figure 1 Tenofovir structure and primary osteoclast cell viability following drug exposure. The structure of tenofovir (A) and tenofovir disoproxil fumarate (TDF, the prodrug of tenofovir) (B) is shown. C. Viability of primary osteoclasts following exposure to (more ...)
We first investigated the effect of TDF exposure on cell viability. We sought to determine whether physiologically relevant TDF concentrations corresponding to the dosing of TDF in antiretroviral therapy had any effects on cell viability. A range of TDF concentrations (50 nM to 500 uM) was analyzed for their effects on osteoclast viability. The highest TDF concentrations analyzed, i.e., 50 uM, 250 uM and 500 uM, were found to significantly reduce cell viability (). The lower TDF concentrations, i.e., 50 nM, 500 nM and 5 uM, had no effect on cell viability. Tenofovir has been shown to cause very little cytotoxicity in human HepG2 and skeletal muscle cell lines, with average CC50
of 399 and 870 uM, respectively [1
]. It is to be expected that primary culture is more sensitive to drug treatment in comparison to immortalized cell lines. As shown in our cytotoxicity assay, primary osteoclast culture reacts to TDF in a much lower concentration (). Tartrate-resistant acidic phosphatase (TRAP) is a marker commonly used in the identification of primary osteoclasts in culture. Therefore, TRAP staining was next used to further verify osteoclast identity and viability at lower TDF concentrations. shows the results of TRAP staining on osteoclasts treated at TDF concentrations of 500 nM, 1 uM and 2 uM. Compared to the no drug control, TDF treatment at the tested concentrations had no effect on osteoclast viability, as well as osteoclast size and numbers (). Based upon these results, we chose the TDF of 500 nM for analyzing the impact TDF exposure on osteoclast gene expression. This TDF concentration is physiologically relevant based upon the serum concentrations and dosing regimens used in antiretroviral therapy of HIV infected individuals [17
]. Therefore, the TDF concentration used in microarray analysis of primary osteoclast gene expression is relevant to the concentrations used in anti-HIV therapy.
To investigate the impact of TDF treatment on osteoclast gene expression, primary osteoclasts were treated with 500 nM TDF, and then total RNA was extracted used to generate cDNAs that were then prepared for use in microarray analysis. The microarray analysis was intensive and allowed for the thorough analysis of osteoclast gene expression profiles. As described below, this intensive approach helped to focus the gene expression profile on a relatively limited number of genes. The replicate-to-replicate variability of TDF treatment of primary osteoclasts was likely another variable that led to the resulting focused gene expression profile. The focused gene expression profile could also be a reflection of the limited amount of TDF uptake by osteoclasts in tissue culture. Since TDF is a phosphonate, this could enhance its uptake into cells when associated with bone tissue. Bisphosphonates (ie, diphosphonates) are drugs used clinically to prevent loss of bone density, particularly in diseases such as osteoporosis, bone metastasis, and multiple myeloma. Bisphosphonates target bone and inhibit osteoclast function after their cellular uptake by inducing apoptosis [19
]. Since TDF is phosphonate, it is possible that their association with bone could enhance their uptake by osteoclasts by a mechanism similar to that of bisphosphonates.
lists the resulting gene expression profile identified from intensive microarray analysis of TDF-treated primary osteoclasts. A total of nine genes were identified. Notably, no significantly up-regulated genes were observed from our microarray analysis. Among those candidates, most were non-verified, relatively smaller RNAs (six out of nine). There were only three previously identified genes, Snord32a, Gnas and Got2, two of which encode for proteins. Gnas (guanine nucleotide binding protein, alpha stimulating – also known as GalphaS) is a membrane-bound G-protein receptor that is involved in signaling to MAPK/ERK (). GalphaS activates the cAMP-dependent pathway by stimulating the production of cAMP from ATP by direct stimulation of the membrane-associated enzyme adenylate cyclase. The cAMP can then act as a second messenger that interacts with and activates protein kinase A, which can phosphorylate a wide array of downstream targets. The MAPK/ERK pathway is important for osteoclast proliferation and for the formation of actin filaments to form the sealing zone on bone tissue prior to bone resorption. Gnas has a highly complex imprinted expression pattern that can result in maternally, paternally, as well as biallelically expressed transcripts derived from four alternative promoters and the 5′ exons. There is also an antisense transcript. One of these transcripts and the antisense transcript are both paternally expressed noncoding RNAs, and may regulate imprinting in this region. Due to the complex nature of the gene locus, qPCR analysis of the observed downregulation in the microarray analysis was not done. The downregulation of Gnas gene expression would be expected to result in less MAPK/ERK signaling and ultimately a reduction in osteoclast proliferation and actin filament formation – these reductions would result in less bone resorption. This observation, in the context of TDF-mediated bone loss, suggests that the bone formation activity of osteoblasts may also be downregulated so that an overall loss of bone density is observed. Reports in the literature have also linked Gnas to bone abnormalities. A recent report described a transgenic mouse model with locally restricted inactivation of Gnas to extend studies of the role of Gnas function in vivo [21
]. In this study, a hRen-Cre transgenic mouse line was crossed with a mouse strain in which exon 1 of the Gnas gene was flanked by loxP sites. The most notable phenotype involved marked skeletal malformations of the forelimbs with shortened and fused extremity bones. This phenotype is similar to syndromes in humans associated with Gnas loss-of-function, including Albright hereditary osteodystrophy and progressive osseous heteroplasia. Interestingly, loss of Gnas has been found in osteoblasts to result in a defect in the formation of the primary spongiosa with reduced immature osteoid, overall length, and trabecular bone volume [22
]. The cortical bone was thickened, with narrowing of the bone marrow cavity, and was likely due to decreased cortical bone resorption since osteoclasts were observed to be markedly reduced on the endosteal surface of cortical bone. To date, no specific reports on the effects of reduced or loss of Gnas expression on osteoclast function have been reported.
Intensive microarray analysis identifies a distinct gene expression profile from TDF-treated primary osteoclasts.
Figure 2 Target genes, Gnas, Got2 and Snord32a, following tenofovir exposure of primary osteoclasts. A. Gnas or stimulatory G-protein alpha subunit (GαS) is a key part of the classic signal transduction pathway linking receptor-ligand interactions with (more ...)
Got2 (glutamate oxaloacetate transaminase 2, also known as aspartate aminotransferase or AST) is a mitochondrial enzyme involved in energy transduction, specifically amino acid metabolism as well as the urea and tricarboxylic acid cycles (). AST is commonly detected in the blood as a clinical indicator of organ injury and cellular stress – particularly for the liver, with alanine transaminase (ALT). Organ injury typically results in higher serum AST levels, and the AST/ALT ratio is commonly analyzed. Interestingly, a reduction of Got2 expression was found to correlate to a loss in bone mineral density in a knockout mouse model for Canavan’s Disease, a neurodegenerative disorder involving a deficiency in aspartocyclase [23
]. Although these findings are correlative, this data suggests an association between the general reduction in Got2 expression and loss of bone mineral density. Previous studies of the effects of NRTIs on mitochondrial gene expression in the mouse liver have been studied by microarray analysis and led to the identification of many genes influenced by NRTI exposure [24
]. Although Got2 was not one of the genes characterized, liver cells and non-osteoclasts were analyzed and it is formally possible that tissue-specific differences exist. We followed up the microarray data analysis with qPCR analysis. We chose to detect the expression level of several genes (i.e., cathepsin K, NFAT, and TRAP) that are specifically involved in osteoclast differentiation by qPCR, and their gene expression was not significantly effected by TDF, and in agreement with the microarray analysis (). The general trends from qPCR suggest downregulation of Got2 expression, though our results were not statistically significant (). This observed discrepancy with the microarray data may be due to the SYBR green method used for qPCR, which is a relatively non-quantitative method for qPCR compared to other methodologies (e.g, TaqMan). Also, more than 1 or 2 housekeeping genes is likely required in order to establish accurate normalization for the verification of the small fold differences observed in the microarray analysis.
Real-time PCR analysis of selected genes from TDF-treated primary osteoclasts.
Snord32a (small nucleolar RNA, C/D box 32A) is a member of the small nucleolar RNA (snoRNA) group of highly abundant, non-polyadenylated, non-coding transcripts that localize to the nucleus but are thought to have more broad functions (). Snord32a is located in the second intron of Rpl13a, whose gene product is part of the large ribosomal subunit [25
]. The snoRNAs are essential for regulating biological functions such as RNA processing and modifications, gene expression, protein trafficking, and genome stability. snoRNAs are highly conserved throughout evolution, and are found in Archaea as well as in eukaryotes [26
]. The two main classes of snoRNAs include the C/D box snoRNAs, which are associated with methylation, and the H/ACA box snoRNAs, which are associated with pseudouridylation. After transcription, pre-rRNAs undergo processing in order to generate the mature rRNA. A complex pattern of nucleoside modifications occurs prior to cleavage by exo- and endonucleases, which includes methylation and pseudouridylation, that are directed by snoRNAs. The C/D box snoRNAs specifically guide site-specific 2′-O-methylation of RNAs. 2′-O-methylation of rRNA is required for ribosome function, likely by affecting proper rRNA folding, maturation and stability. It is thought that rRNA modifications act to optimize rRNA structure for the production of accurate and efficient ribosomes. Overall, snoRNA binds to target RNA via base pairing and acts as a scaffold for its protein partners to be recruited and therefore triggers the enzymatic reactions. The qPCR analysis suggested downregulation of gene expression, but the results were not statistically significant (). The lack of microarray analysis verification of Snord32a is likely due to the limitations in the qPCR analysis (as described above). The reduction in Snord32a expression could result in a decrease of fully functional ribosomes, which would lead to a reduction in gene expression. Three previously uncharacterized snoRNAs were identified in our analysis that had significantly altered expression from the TDF-treated primary osteoclasts as well as two small nuclear RNAs (snRNAs) and one unknown transcript (). Two of these snoRNAs had higher levels of gene expression, while there was a reduction in expression for the other four snoRNA genes. Many studies have revealed some snRNAs do not possess sequence motifs, nor are complementary to known target RNAs [29
]. It is likely that there are many putative targets of snoRNAs that are yet to be identified. The function of snoRNA (and many downstream reactions affected by them) is therefore more complex than anticipated. It is plausible that these other snoRNAs may play some roles in altering gene expression due to TDF treatment.
In summary, we report that TDF-exposure of primary osteoclasts results in a striking alteration of a distinct gene expression profile. Our findings suggest that the clinical observations of TDF-mediated bone density loss observed in children and adolescents could be due, in part, to changes in osteoclast gene expression that result in osteoclast dysfunction. Future studies should focus on extending our observations to in vivo analysis of gene expression alterations due to TDF treatment. In addition, it would be of interest to the analyze the specific alteration of gene expression for the genes identified in this study in osteoclasts in vivo to assess their impact on bone formation and resorption. Finally, extending these studies to osteoblasts would provide greater insight into the effects of TDF exposure on bone mineral density.