Involvement of Upstream Stimulatory Factors 1 and 2 in RANKL-Induced Transcription of Tartrate-Resistant Acid Phosphatase ( TRAP ) Gene during Osteoclast Differentiation

Tartrate-resistant acid phosphatase (TRAP) plays an important role in bone resorption. TRAP expression in osteoclasts is regulated by receptor activator of NF-kappaB (RANKL), a potent activator of osteoclast differentiation. However, the molecular mechanism underlying the RANKL-induced TRAP expression remains unknown. Here we show that two regions in the mouse TRAP promoter (one at -1858 to -1239 and the other at -1239 to -1039, relative to the translation start site) are implicated in RANKL-induced TRAP transcription in RAW264.7 cells. A detailed characterization of the region at -1239 to -1039 identifies a 12-bp sequence, AGCCACGTGGTG, that specifically binds nuclear proteins from RAW264.7 cells and primary bone marrow macrophages (BMMs) in an electrophoretic mobility shift assay (EMSA). Moreover, the binding is significantly enhanced in EMSA with nuclear extracts from RANKL-treated RAW264.7 cells and BMMs, suggesting that the 12-bp sequence may be involved in RANKL-induced TRAP transcription. Various assays reveal that nuclear proteins binding to the 12-bp sequence are upstream stimulatory factors (USF) 1 and 2. Importantly, mutation of the USF-binding site partially blocks RANKL-induced TRAP transcription in RAW264.7 cells, confirming that USF1 and USF2 are functionally involved in RANKL-induced TRAP transcription. In summary, our data show that USF1 and USF2 play a functional role in RANKL-dependent TRAP expression during osteoclast differentiation.


INTRODUCTION
Osteoclasts, the principal bone-resorbing cells, play a pivotal role in skeletal development and maintenance (1). Osteoclasts are derived from mononuclear precursors of monocyte/macrophages lineage upon stimulation of two critical factors: M-CSF and RANKL (2).
Osteoclastic bone resorption involves various events including attachment of osteoclasts on bone matrix through the sealing zone to form a resorption compartment, formation of the ruffled membrane where protons and proteinases are secreted into the resorption compartment to degrade bone, and removal of degraded products via transcytosis (3;4).
Tartrate-resistant Acid Phosphatase (TRAP) is an iron-binding protein that is abundantly expressed in mature osteoclasts (5)(6)(7). TRAP has been shown to play an important role in bone resorption. In vitro studies demonstrated that a neutralizing antibody against TRAP inhibited bone resorption (8). Confirming the in vitro data, mice lacking TRAP exhibited a defect in endochondral ossification and a mild osteopetrosis (9). Conversely transgenic mice overexpressing TRAP resulted in a decrease in trabecular bone density with a characteristic of a mild osteoporosis (10). Recent studies have suggested that TRAP regulates bone resorption by mediating the degradation of endocytosed matrix products during transcytosis in activated osteoclasts (5;11;12).
TRAP expression is often undetectable in osteoclast precursors but its expression is dramatically up-regulated during osteoclast differentiation (7). As a result, TRAP has been widely used as a marker for osteoclasts (2). Since the discovery of RANKL, it has been established that RANKL plays a key role in TRAP expression during osteoclast differentiation (13;14). However, the molecular mechanism by which RANKL regulates TRAP expression during osteoclast differentiation still remains unknown. 6 lifting the cells by scraping. Cells were transiently transfected using LipofectAmine Plus transfection reagents from Invitrogen. For transfections, cells were plated in 6-well cell culture plates at the concentration of 8 × 10 5 cells/well one day before transfections. For each well, 2 ug reporter plasmid plus 0.05 ug internal-control plasmid phRL-SV40 (Promega) were used.
Transfected cells were treated with or without 200ng/ml GST-RANKL for various times after transfection and lysed for luciferase assays using Dul-Luciferase Reporter Assay System from Promega.
Nuclear Extract Preparation -Bone marrow macrophages (BMMs) were isolated from long bones of 4-8 week old C3H mice from Harlan Industries as described (28)  min on ice, at which time 32 ul of 10% Nonidet P-40 was added to the suspension, followed by vortexing the tube for 15 seconds and incubating on ice for 10 min. Nuclei were spun down and resuspended in 100 ul of Nuclear Extraction Buffer (20 mM Hepes-KOH, pH 7.9, 420 mM NaCl, 1.2 mM MgCl 2 , 0.2 mM EDTA, 25% glycerol, 0.5 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM BAESF, 5ug/ml pepstatin, and 5 ug/ml leupeptin. DTT, phenylmethylsulfonyl fluoride, BAESF, pepstatin, and leupeptin were added freshly to the buffer). The nuclei were incubated with the extraction buffer on ice for 20 min and spun down in a microcentrifuge. The supernatant (nuclear extract) was aliquoted, quickly frozen in dry ice/ethanol bath, and stored at -7 80 o C. Protein concentration of nuclear extracts was determined using Bio-Rad protein assay kit (Bio-Rad, Hercules, CA).

Electrophoretic Mobility Shift Assays (EMSAs) -Oligonucleotides (oligos) used for EMSAs
were end-labeled with 32 P by T4 polynucleotide kinase (Invitrogen). 2-5X10 4 cpm probe was incubated with 3 ug of nuclear extracts in a 20-ul volume of binding reaction (10mM Tris-Cl, pH7.5, 100mM NaCl, 10% glycerol, 50ng/ml poly(dI/dC) on ice for 20 min. In competition experiments, a 50X and/or 100X excess amount of unlabeled competitors was premixed with labeled probe before being added to the binding mixture. The binding reaction was then allowed to proceed for 20 min on ice. In supershift experiments, probe was incubated with 3 ug/ml nuclear extracts in a 20-ul volume of binding reaction for 20 min on ice, at which time 4 ug control IgG or 4 ug specific antibodies were added, followed by incubation on ice for an additional 30 min.
All binding mixtures were separated, using 0.5XTBE buffer as the running buffer, at 4 o C and 100 V for 3.5 h by 4-20% gradient TBE gels (Invitrogen) in Novex Xcell II minicell electrophoresis system. The gels were transferred to 3M blotting paper, dried and exposed to film.
In vitro translations -In vitro translated USF1 and USF2 were prepared by using PROTEINscript TM II Linked Transcription/Translation Kit (Ambion, Austin, TX) and expression plasmids for USF1 (psvUSF1) and USF2 (psvUSF2) described in (29). These expression plasmids were constructed using vector pSG5 (Stratagene, La Jolla, CA) that contains T7 promoter suitable for in vitro translations. Thus, PROTEINscript II T7 was used for in vitro translation assays with these expression vectors. Briefly, 0.5 ug of each plasmid (pSG5, psvUSF1 or psvUSF2) was used to set up transcription reactions (10 ul reaction volume for each experiment) following the protocol provided by the manufacturer. Subsequently, 2 ul of the transcription reaction was used for the translation reaction following the protocol provided by the manufacturer (50 ul reaction volume for each experiment). 5 ul from each translation reaction was then used to perform EMSA as described above. Sequence Analysis -Sequence analysis was performed using the Genetic Computer Group (Madison, WI) sequence analysis software.  (24)(25)(26). The nucleotide sequence of our clone is compared with those published and differences exist among these three clones (Fig. 1A). To confirm that our sequence is not altered by possible mutations associated with PCR-based amplifications, we sequenced and compared two clones from two independent PCR amplifications. Thus, the differences between our clone and two published ones likely result from variations in mouse strains and/or individual animals.

A 1,858-bp mouse TRAP promoter confers responsiveness to RANKL -
The amplified TRAP promoter was subcloned into pGL3-basic in sense orientation to luciferase gene to generate a reporter construct TP(-1858). Since primary BMMs are extremely difficult to transfect, we used RAW264.7 cells for our transfection assays. TP(-1858) was transiently transfected into RAW264.7 cells, and the transfection result shows that RANKL activated the TRAP promoter in a time-dependent manner with a maximal induction at day 4 (about 2.5-fold) (Fig. 1B), thus indicating that the 1,858-bp TRAP promoter contains a sequence(s) mediating RANKL-induced activation of TRAP gene transcription. In addition, since the data show that a 4-day RANKL treatment gave rise to the highest TRAP promoter activation, all of our following transfection studies will be performed with a 4-day RANKL treatment. This band was significantly enhanced when RANKL-treated RAW264.7 nuclear extracts were used (lane 3), suggesting that nuclear proteins corresponding to band A may be implicated in RANKL-induced TRAP transcription. In addition to band A, a minor band with a higher mobility (band B) was also observed (lanes 2 and 3). However, in our subsequent EMSAs, band B appears to be very unstable. Thus, we only focused on band A in the present study. Interestingly, oligo IV also binds nuclear proteins induced by RANKL in our EMSA (lanes 11 and 12). These data suggest that oligo I and oligo IV may contain cis-elements regulating RANKL-induced TRAP transcription. However, our computer analysis found no consensus sequences for NF-κB or AP-1 in this 40-bp region, suggesting that the nuclear proteins binding to this 40-bp sequence are not NF-κB or AP1.

Localization of a 40-bp TRAP promoter region (-1239 to -1199) that binds RANKL-induced
To experimentally confirm this, we performed a competition assay using excess cold oligo I and oligos containing NF-κB/AP-1 consensus sequences as competitors (Fig. 4A). While excess cold oligo I were able to compete for the nuclear protein binding, both NF-κB and AP-1 oligos failed to do so, further suggesting that this 40-bp does not bind NF-κB or AP-1. Subsequently, our supershift assays with antibodies against p50, P65, c-fos and c-jun confirmed that the 40-bp sequence binds transcription factors other than NF-κB and AP-1 (Fig. 4B).

Identification of a 12-bp sequence AGCCACGTGGTG that specifically binds RANKL-
induced USF1 and USF2 in oligo I -Upon the exclusion of NF-κB and AP-1 as the nuclear proteins binding to oligo I, we proceeded to elucidate the identity of the nuclear proteins. Our computer analysis revealed that this 40-bp region contains putative binding sequences for many transcription factors, including ELP/SF1/FTZ (33), USF (34;35), PPAR (36), CP1/2 (37), PHO4 (38;39) and c-Myc (40). Since this information is not specific enough, we decided to identify the specific nuclear protein-binding sequence as our first step in elucidating the nuclear proteins. In doing so, we performed a series of competition assays using shortened oligos derived from oligo I as competitors (Fig. 5A). As shown in Fig. 5B, while SOI 1 competed efficiently for the nuclear protein binding (lane 2), SOIs 2-5 failed to do so (lanes 3-6), revealing the 5' end of the specific nuclear protein-binding sequence (Fig. 5A). Similarly, the experiment in Fig. 5C showed that SOIs 6-11 could compete efficiently (lanes 2-7) but not SOI 12 (lane 8), elucidating the 3' end of the sequence (Fig. 5A). Together, these data identified a 12-bp sequence within the 40-bp region that binds specifically the nuclear proteins induced by RANKL (Fig. 5A).
The 12-bp sequence contains a sequence CACGTG, which is recognized as a core sequence present in binding sites for both the Myc family members (40) and USF proteins (34;35). This suggests that the nuclear proteins binding to the 12-bp sequence might be members of the Myc family or USF proteins. As a result, we performed supershif assays with antibodies against c-Myc, c-Max, USF1 and USF2 (Fig. 6A). While antibodies against c-Myc and c-Max had no effect on band A (lanes 3 and 4), those against USF1 and USF2 supershifted the band (lanes 5 and 6), confirming that the nuclear proteins binding to the 12-bp sequence are USF1 and USF2.
Moreover, USF1 antibody only partially supershifted band A, resulting in a weak band (lane 5).
In contrast, USF2 antibody completely supershifted the band (lane 6). These data further indicate that this 12-bp sequence is able to bind USF proteins from RAW264.7 cells as either USF1-USF2 heterodimers or USF2 homodimers.
At this point we identified a 12-bp sequence in TRAP promoter that binds USF1 and USF2 from RAW264.7. Since TRAP gene expression is also activated by RANKL in primary BMMs, we examined whether this sequence also binds RANKL-induced USF1 and USF2 from primary BMMs. As shown in Fig. 6B, oligo I binds nuclear proteins from untreated primary BMMs, resulting in two bands (A and B, lane 1). Band A is significantly enhanced when nuclear extracts from RANKL-treated BMMs were used (lane 2). Consistently, the pattern of banding in the EMSA (Fig. 6B) is very similar to that seen in the EMSA with nuclear extracts from RAW264.7 cells (Fig. 2D, lanes 2 and 3). Importantly supershift assays indicate that the nuclear proteins binding to oligo I (band A) from primary BMMs treated with RANKL are also USF1 and USF2 (Fig. 6C). These results further support that the 12-bp sequence in TRAP promoter may play a functional role in RANKL-dependent TRAP transcription by utilizing USF1 and USF2.
As additional evidence supporting that this 12-bp sequence binds USF1 and USF2, we performed EMSA using in vitro translated USF 1 and USF2, which were prepared by using expression vectors psvUSF1 (for USF1) and psvUSF2 (for USF2) described in (29). These expression plasmids were constructed using vector pSG5 (29). As shown in Fig . This assay was independently repeated twice and the same result was obtained. This discrepancy may result from a difference in the affinities of these in vitro translated USF proteins for the 12-bp sequence or from the low translation efficiency associated with psvUSF2. Nevertheless, these data strongly supports that the 12-bp sequence is a USF-binding site.

The 12-bp USF binding sequence is functionally involved in the RANKL-induced TRAP
transcription -Finally we determined whether the 12-bp USF-binding sequence is functionally involved in the RANKL-dependent TRAP transcription. To this end, we need to identify a mutation in the USF-binding site that is capable of blocking the USF binding. We synthesized a mutant oligo I (mOligo I) in which TG, in the core binding CACGTG, were converted to GA 14 (Fig. 8A). Competition assays show that mOligo I failed to compete for the USF binding (lane 3, Fig. 8B), indicating that the chosen mutation is sufficient to eliminate the binding capacity of the sequence. To exclude the possibility that the chosen mutation, although it abolish USF recognition, render the sequence capable of associating with other nuclear proteins, mOligo I was used as probe to perform an EMSA (Fig. 8C). The data confirm that mOligo I did not bind any additional nuclear proteins.

DISCUSSION
TRAP expression is dramatically up-regulated during osteoclast differentiation (7). In mature osteoclasts, TRAP plays an important role in osteoclastic bone resorption (8)(9)(10). As a result, elucidation of the regulatory mechanism controlling TRAP expression in osteoclasts has long been an active part of bone biology research. Prior to the discovery of the RANKL/RANK system in the late 1990's, osteoclasts were generated in vitro virtually by co-culturing primary BMMs with osteoclasts/stromal cells (30)(31)(32), which were believed to provide essential osteoclastogenic factors (30;41). However, the identities of these factors were unknown. Consequently, the identity of the molecule(s) regulating TRAP expression in osteoclasts was also unclear. Since the unraveling of the RANKL/RANK system, it has now been established that osteoblasts/stromal cells support osteoclast differentiation in the co-culture system primarily by producing two critical factors: M-CSF and RANKL (13;15), which play an essential role in osteoclast differentiation (2). Moreover, it has also become clear that RANKL is an essential and potent factor involved in the up-regulation of TRAP expression during osteoclast differentiation (13;14).
These findings prompted us to investigate the molecular mechanism by which RANKL regulates TRAP expression during osteoclast differentiation.
Our studies took advantage of a macrophage-like cell line RAW264.7, which is not only transfectable but it is also capable of differentiating into osteoclasts in response to RANKL (14).
Significantly, our initial transfection studies showed that two distinct TRAP promoter regions are important for TRAP transcriptional activation in response to RANKL, suggesting that multiple transcriptional events are involved in RANKL-induced TRAP gene activation. A detailed characterization of one such region in this paper revealed that a 12-bp sequence AGCCACGTGGTG is involved in enhancing TRAP gene transcription in response to RANKL.
More significantly, we further showed that this 12-bp sequence does so by binding USF1 and USF2.
Transcription factors USF1 and USF2 were originally identified by their ability to bind to the adenovirus major late promoter (42). Structurally, USF1 and USF2 are related to the Myc family of transcription factors, which are characterized by the presence of a C-terminal basic helix-loophelix-leucine zipper (bHLH-Zip) domain (33;43). USF1, USF2 and the Myc family members were shown to recognize DNA sequences containing a core sequence CACGTG (29). While the RANKL-responsive sequence (AGCCACGTGGTG) identified in TRAP promoter contains such a core sequence (underlined), our data showed that it binds USF1 and USF2 but not c-Myc or Max in response to RANKL (Fig. 6A and Fig. 6C). This indicates that this 12-bp RANKLresponsive sequence regulates RANKL-induced TRAP transcription exclusively by utilizing  One possibility is that RANKL increases the USF gene expression, and the enhanced USF binding directly results from the increase in amounts of USF proteins available. Alternatively, RANKL has no effect on the USF gene expression. Instead, RANKL activates USF proteins by phosphorylating them and phosphorylated proteins have higher affinity for the binding site. The latter represents a more reasonable hypothesis because: 1) RANKL is able to activate the p38 MAPK pathway (50-52) and 2) the p38 MAPK pathway was shown to play a role in phosphorylating USF1 in mediating UV-induced Tyrosinase expression in a mouse melanocyte cell line (53). Nonetheless, the precise mechanism of the RANKL-mediated USF activation is currently under investigation. Elucidation of the signaling pathway involved in RANKL-mediated USF activation will provide more insights into the molecular mechanism by which RANKL regulates osteoclast differentiation.
Our initial transfection data showed that two distinct TRAP promoter regions are involved in regulation of TRAP transcription in response to RANKL ( Figs. 2A and 2B). Consistently, mutation of the USF-binding site in the region at -1239 to -1039 only partially blocked the RANKL-induced TRAP transcriptional activation (Fig. 8E), further supporting that the other region at -1858 to -1239 also contributes to the RANKL-induced TRAP transcriptional activation. To fully elucidate the molecular mechanism underlying RANKL-dependent TRAP gene activation, the cis-elements in the region at -1858 to -1239 need to be characterized.
Furthermore, oligo IV binds a nuclear protein(s) (band C, lane 12, Fig. 2C) in response to RANKL. Given that mutant constructs TP(-1199) and TP(-1159) failed to confer RANKL responsiveness, the nuclear protein(s) binding to oligo IV is not sufficient to activate TRAP promoter in response to RANKL. However, our data by no means exclude the possibility that the nuclear protein(s) are necessary for RANKL-induced TRAP gene activation. Thus, investigation of the potential role of the nuclear protein(s) in RANKL-dependent TRAP transcription represents is warranted.
In conclusions, our data presented here demonstrate that USF1 and USF2 play an important role in RANKL-dependent TRAP transcription. This finding not only defines a role of USF proteins in TRAP expression in osteoclasts, but more importantly it also raises many important questions. First, whether and how USF proteins collaborate with other transcription factors in regulating TRAP transcription is unknown. Furthermore, the precise signaling pathway by which RANKL activates USF transcription factors remains to be elucidated. Future studies aimed at addressing these questions will provide more insights into molecular mechanisms governing osteoclast differentiation and function. Each bar is the mean of three replicates + S.D.