Parathyroid hormone(1–34) and its analogs differentially modulate osteoblastic Rankl expression via PKA/SIK2/SIK3 and PP1/PP2A–CRTC3 signaling

Osteoporosis can result from the loss of sex hormones and/or aging. Abaloparatide (ABL), an analog of parathyroid hormone–related protein (PTHrP(1–36)), is the second osteoanabolic therapy approved by the United States Food and Drug Administration after teriparatide (PTH(1–34)). All three peptides bind PTH/PTHrP receptor type 1 (PTHR1), but the effects of PTHrP(1–36) or ABL in the osteoblast remain unclear. We show that, in primary calvarial osteoblasts, PTH(1–34) promotes a more robust cAMP response than PTHrP(1–36) and ABL and causes a greater activation of protein kinase A (PKA) and cAMP response element–binding protein (CREB). All three peptides similarly inhibited sclerostin (Sost). Interestingly, the three peptides differentially modulated two other PKA target genes, c-Fos and receptor activator of NF-κB ligand (Rankl), and the latter both in vitro and in vivo. Knockdown of salt-inducible kinases (SIKs) 2 and 3 and CREB-regulated transcription coactivator 3 (CRTC3), indicated that all three are part of the pathway that regulates osteoblastic Rankl expression. We also show that the peptides differentially regulate the nuclear localization of CRTC2 and CRTC3, and that this correlates with PKA activation. Moreover, inhibition of protein phosphatases 1 and 2A (PP1/PP2A) activity revealed that they play a major role in both PTH-induced Rankl expression and the effects of PTH(1–34) on CRTC3 localization. In summary, in the osteoblast, the effects of PTH(1–34), PTHrP(1–36), and ABL on Rankl are mediated by differential stimulation of cAMP/PKA signaling and by their downstream effects on SIK2 and -3, PP1/PP2A, and CRTC3.

deacetylases, HDACs 4 and 5, which inhibit the transcription factor, MEF2c, and therefore decrease Sost expression (23). Previous studies have also reported differences in downstream PTHR1 effects between PTH(1-34), PTHrP , and ABL (25). Based on this, we hypothesized that in the osteoblast, PTH , PTHrP , and ABL would differentially modulate this particular signaling axis and ultimately result in differing effects on the regulation of Rankl mRNA. Additionally, because this cascade was reported to involve an unknown serine/threonine phosphatase and PTHrP is able to activate protein phosphatase 2A (PP2A) in the chondrocyte (26), we hypothesized that this activation would also be important in the osteoblast.
Here, we report that PTH , PTHrP , and ABL differentially induce cAMP/PKA signaling. Accordingly, different levels of PKA activation lead to differences in the up-regulation of Rankl mRNA by all three peptides, and we found that this phenomenon is a direct result of PKA activation. We also found that osteoblastic Rankl expression requires SIKs 2 and 3, CRTC3, and PP1/PP2A and that the three peptides, through PP1/PP2A, differentially regulate the nuclear translocation of CRTCs 2 and 3. Interestingly, PTH , PTHrP , and ABL equally inhibit osteoblastic Sost expression, which illustrates a demarcation between SIK/HDAC/MEF2c and SIK/ CRTC signaling. Taken together, our data show that these peptides differentially regulate a particular arm of the cAMP/PKA/ SIK signaling axis and ultimately result in lower expression of Rankl by ABL, which may provide an explanation for the decreased resorptive effects of ABL observed in vivo.

PTH(1-34), PTHrP(1-36), and ABL similarly bind PTHR1
Previous studies aimed at comparing the binding efficacies between PTH  or PTHrP  with their cognate receptor, PTHR1, have shown that although both are able to bind with similar affinity, PTH  binds PTHR1 in both high affinity receptor conformations, R 0 (G-protein uncoupled) and R G (G protein-coupled), whereas PTHrP(1-36) favors the latter (27). These differences were also reflected in the ability of PTH  to elicit a more sustained cAMP response upon receptor internalization, whereas PTHrP(1-36) could not (28). More recently, comparative studies between PTH(1-34), PTHrP , and ABL in GP-2.3 cells have shown that ABL binds more selectively to the R G conformation of PTHR1, thus resulting in a more transient cAMP response compared with PTH(1-34) (25). To determine whether these differences were observable in osteoblastic cells, we measured the capacities of PTH , PTHrP , and ABL to compete for binding to PTHR1 in the osteosarcoma cell line, UMR 106, using [ 125 I-Nle 8,21 ,Tyr 34 ]rPTH(1-34) as a tracer radioligand (Table 1). We found no significant difference in pIC 50 between the three peptides under these conditions.

PTH(1-34) elicits a stronger cAMP response and downstream effects on PKA/CREB compared with PTHrP(1-36) and ABL
Because we did not observe differences in PTHR1 binding, we hypothesized that in primary murine calvarial osteoblasts, a system that more closely resembles the physiological setting, more pronounced differences that will integrate downstream signaling would be observed. Upon treating primary osteoblasts with PTH(1-34), PTHrP , or ABL, we found that all three peptides achieved maximal cAMP responses at 30 min, but PTH(1-34) resulted in higher cAMP concentrations (240 Ϯ 23 pmol/l), compared with PTHrP(1-36) (175 Ϯ 4 pmol/l) or ABL (45 Ϯ 5 pmol/l; Fig. 1a). Because cAMP levels are the rate-limiting step to PKA activation, we therefore sought to observe how these differing cAMP responses affected PKA activity. Indeed, we found that whereas all three peptides achieve peak PKA activation between 1 and 2 min, PTH  reached 72% PKA activation, whereas PTHrP(1-36) and ABL resulted in 50 and 30%, respectively (Fig. 1b). The differences in magnitude of each peptide's cAMP response were proportional to their effect on PKA activation. To assess the dose-dependent effects of these peptides, we performed a dose-response at 1 min and found that at 0.75 nM, PTH(1-34) resulted in 40% PKA activation, whereas PTHrP(1-36) and ABL elicited less than half of that at that dose (16 and 18%, respectively). Additionally, the relative differences in PKA activation between all three peptides grew smaller with larger doses, but ABL did not achieve the same levels of PKA activation (53%) as PTH(1-34) (73%) or PTHrP(1-36) (68%) (Fig. 1c). Previously, we reported that cAMP/PKA signaling leads to activation of the cAMP response element-binding protein (CREB) and the stimulation of c-Fos expression in osteoblastic cells (29,30). We found that all three peptides resulted in maximal CREB activation (assayed by phosphorylation of CREB at Ser-133) at 5 min, and again, in proportions similar to their effects on cAMP/PKA (PTH(1-34) ϭ 77%, PTHrP(1-36) ϭ 55%, ABL ϭ 20%; Fig. 1d). Dose-response of CREB activation at 5 min shows a similar curve as seen with PKA activation (Fig. 1e), thus confirming that the initial differences in cAMP response exerted by PTH(1-34), PTHrP , and ABL on cAMP are directly reflected in their downstream effectors, PKA and CREB.

PTH(1-34), PTHrP(1-36), and ABL differentially regulate osteoblastic genes c-Fos and Rankl
To determine the consequences, if any, of the initial signaling differences between PTH(1-34), PTHrP , and ABL on the regulation of osteoblastic genes, we performed time course and dose-response analyses, followed by qRT-PCR on a set of osteoblastic genes. After treating primary osteoblasts and/or osteoblastic UMR 106-01 cells with peptide concentrations ranging from 0.001 to 100 nM at 1, 2, and 4 h, we found no significant differences with almost all of the genes assayed ( Table 2). PTH is a known regulator of sclerostin (Sost) (31) and  2a) and a complete suppression of Sost mRNA in UMR 106-01 cells (Fig. 2b), with no significant differences in the dose-response between the three peptides. Interestingly, both genes have been reported to be regulated by PTH in a cAMP/PKA/HDAC4/5-dependent mechanism (33,34), suggesting that this arm of the cAMP/PKA signaling axis is similarly regulated by all three peptides, and likely is sensitive to low PKA activation. Because the inhibition of Sost represents one of the major anabolic effects of PTH, its identical inhibition by PTH(1-34), PTHrP , and ABL may account for the similar osteoanabolic effects observed in vivo. Importantly, our analyses revealed two genes that were differentially regulated by the peptides: the transcription factor, c-Fos, and the pro-osteoclastogenic factor, Rankl. We have shown that c-Fos is an immediate-early downstream effector of PTH by way of cAMP/PKA signaling (29,30), and here, we confirm peak c-Fos mRNA abundance at 1 h, where PTH(1-34) resulted in a 10-fold increase, whereas PTHrP(1-36) and ABL resulted in 5-and 4-fold increases, respectively, and all returned to basal levels by 4 h (Fig. 2c). Dose-response analysis of c-Fos mRNA abundance (Fig. 2d) mirrored the differential effects of all peptides on cAMP/PKA/CREB signaling and confirmed that in osteoblasts, PTHrP(1-36) and ABL are weaker effectors of this arm of cAMP/PKA signaling compared with PTH .
Rankl is a well-studied PTH-responsive gene in osteoblasts and osteocytes (16 -18, 21, 35, 36) and here, we found peak fold-induction by all three peptides at 4 h and a significant difference in the effect of the peptides at 1 nM (PTH(1-34) ϭ 8-fold, PTHrP(1-36) ϭ 3-fold, ABL ϭ 3-fold; Fig. 2, e and f). Due to the central role of RANKL as the driver of osteoclastogenesis and a mediator between osteoblasts/osteocytes and osteoclasts, we sought to investigate the differential effects of PTH(1-34), PTHrP , and ABL on osteoblastic Rankl expression.

Differential regulation of Rankl relies on the PKA/PP1/PP2A signaling axis in vitro
Rankl gene expression has been shown to be regulated by cAMP/PKA activity in vascular cells (37), mesenchymal stem cells (38), and osteocytes (21,23), and due to the differential effects of PTH(1-34), PTHrP , and ABL on this signaling axis, we hypothesized that modulating PKA activity would abolish any differences they elicit in the osteoblast with respect to Rankl mRNA expression. Indeed, pre-treatment with the PKA agonist, 8-bromo-cAMP (8-Br-cAMP) alone increased Rankl mRNA levels by 38-fold and abolished the effects of all three peptides on RANKL, suggesting Rankl mRNA cannot be enhanced above a maximal level (Fig. 3a). Conversely, pretreatment with the PKA-specific antagonist, myristoylated PKA inhibitor (myr-PKI), did not affect basal Rankl mRNA levels, but completely abolished the effects of all three peptides (Fig. 3a). To confirm these findings in a different system, we performed similar experiments with primary osteoblasts derived from bone chips of adult C57Bl/6 WT femurs and obtained identical results (Fig. S1).
Protein phosphatase 1 (PP1) and PP2A are two major Ser/ Thr phosphatases that account for the majority of phosphatase activity in eukaryotes. PP1 activity is modulated through the formation of heterotrimeric complexes with numerous regulatory subunits. Targeting to specific substrates involves interaction of PP1 catalytic subunit C with targeting subunits, of which 30 have been identified (39). In skeletal muscle, PP1 activity has been shown to be indirectly regulated by PKA, because it directly phosphorylates a PP1 targeting subunit, G M (40). PP2A has been implicated in the regulation of Rankl in MC3T3-E1 cells (41), in addition to its involvement in PTHrP signaling in the chondrocyte (26). PP2A holoenzymes consist of a dimer formed by a catalytic C subunit (PP2A-C) and scaffolding A subunit (PP2A-A), which is targeted to the substrate by a regulatory B subunit (42). We found that MC3T3-E1 cells are not PTH-responsive (data not shown), and therefore sought to determine phosphatase involvement in primary calvarial osteoblasts. Okadaic acid is a commonly used inhibitor of PP1/PP2A, and inhibition of either or both phosphatases can be determined by using two different concentrations (PP2A: IC 50 0.02-0.5 nM PP1: IC 50   Relative PKA activity is expressed as a ratio of phosphorylated PKA substrate over total substrate. d and e, CREB phosphorylation. Primary calvarial osteoblasts were treated with (d) 10 nM peptides from 1 to 60 min or (e) with the indicated doses for 5 min prior to whole cell lysis and immunoblotting. Samples were quantified using Bio-Rad ImageLab software and calculated against the total detectable protein per lane. All data represent the mean Ϯ S.D. of n ϭ 3 independent experiments and images are representative of mean results.

Table 2 Comparative effects of PTH(1-34), PTHrP(1-36), and ABL on osteoblastic genes
Primary calvarial osteoblasts were treated with peptides for 1, 2, or 4 h with peptide doses ranging from 0.001 to 100 nM, followed by qRT-PCR. Messenger RNA abundance of each gene was analyzed relative to the housekeeping gene, ␤-actin, and represent the mean Ϯ S.D. of n ϭ 3 independent experiments.

Gene ID Gene name PTH PTHrP ABL
Receptor activator of nuclear factor B ligand ϩϩϩ ϩ ϩ a Ϫ, decrease; ϩ, increase; and NC, no change.

Regulation of Rankl by PTH(1-34), PTHrP(1-36), and ABL
levels of Rankl by 64% and decreased the effects of PTH(1-34), PTHrP , and ABL by 93, 59, and 75%, respectively (Fig.  3b). Adenoviral-mediated expression of the PP2A-specific inhibitor, SV40 small T-antigen, attenuated the effects of all peptides by 60 -70% (Fig. 3c), and abolished any differential effects between the peptides, implicating PP2A as a phosphatase that plays a major role in the regulation of Rankl mRNA expression. Nevertheless, PP1 may also be involved. These data show that cAMP/PKA and PP1/PP2A signaling is required for maximal osteoblastic Rankl expression.

Differential regulation of Rankl is observed in vivo and in a PKA-dependent fashion
To corroborate our data in vivo, we injected 4-month-old male C57Bl/6 WT mice with 80 g/kg/day of vehicle or pep-tides for 6 weeks and harvested tibial shafts to enrich for RNA derived from osteoblast/osteocyte-rich cortical bone. We found that PTH(1-34) injections led to a 3-fold increase in Rankl, whereas PTHrP(1-36) and ABL only resulted in a 1.5fold increase (Fig. 3d). To examine the effects of PKA up-regulation in vivo, we deleted the regulatory subunit of PKA (PRKAR1a) in osteoblasts by crossing Col1-Cre-ER T mice with Prkar1a fl/fl mice and induced deletion by tamoxifen injections. Four-week-old male and female mice were injected for 3 weeks and 5-month-old mice were injected once a week for 4 weeks prior to tissue harvest. We observed an increase in Rankl mRNA of 2.8-fold in 7-week-old female mice and 4-fold in male mice (Fig. 3e). We also observed a 5-fold increase of Rankl mRNA in 6-month-old mice (Fig. 3f). Taken together, the in vivo findings corroborate the results seen in primary cultured

Osteoblastic Rankl expression depends on SIK2/SIK3 and CRTC3
Previous studies had reported that PTH-induced osteoblastic Rankl expression depended on the PKA-dependent phosphorylation of HDAC4 and its dissociation from the transcription factor, MEF2c (43). Surprisingly, we found that siRNA-mediated knockdown of either HDAC4 or MEF2c did not abrogate the effect of PTH(1-34) on Rankl mRNA expression (Fig. S2). Recently, SIKs have been shown to be prominent effectors of osteocytic Rankl and Sost expression. PTH has been shown to promote the PKA-dependent inhibitory phosphorylation of SIK2 and SIK3, which in turn, decreases the tonic phosphorylation of CREB-regulated transcription coactivator, CRTC2. CRTC2 dephosphorylation leads to its dissociation from 14-3-3 cytoplasmic proteins and allows its nuclear translocation and subsequent association with a bZIP transcription factor at the Ϫ75 kb enhancer region of the Rankl promoter (23). In primary osteoblasts, knockdown of SIK2 and SIK3 results in a 5-fold increase in basal Rankl mRNA levels, and a 2-3-fold increase in the effects of all peptides (Fig. 4a). Interestingly, knockdown of CRTC2 or CREB did not significantly alter the basal levels of Rankl mRNA or the effects of PTH(1-34), PTHrP(1-36), or ABL (Fig. 4b), implying that PKA signal-ing modulates osteoblastic Rankl expression through other effectors. However, knockdown of CRTC3 decreased basal RANKL mRNA levels by 50%, and reduced the effects of all peptides by 30% (Fig. 4c). This suggests that SIK2, SIK3, and CRTC3 play an important role in osteoblastic Rankl expression, and that CRTC2 and/or CRTC3 do not require CREB, but presumably bind other transcription factors containing bZIP domains on the Rankl promoter.

PTH-induced translocation of CRTC2 and CRTC3 requires SIK2 and PP1/PP2A
Because we found that PTH(1-34) and its analogs require SIKs 2 and 3, CRTC3, and PP1/PP2A to modulate Rankl mRNA expression, and all three peptides affect the nuclear translocation of CRTC2/CRTC3, we assessed the potential of PTH , the strongest of the three cAMP/PKA agonists, to regulate the localization of CRTC2 and CRTC3 in the absence of SIK2 and PP1/PP2A activity. siRNA-mediated SIK2 knockdown resulted in a 36% increase in basal levels of nuclear CRTC2, and enhanced the PTH(1-34)-induced nuclear translocation by 10% (Fig. 6a), thus confirming its inhibitory effect on CRTC2. Surprisingly, knockdown of SIK2 did not have any significant effects on basal levels of nuclear CRTC3, nor did it alter the   (1-34), PTHrP(1-36), and ABL effects of PTH(1-34) (Fig. 6b). Inhibition of PP1/PP2A by 200 nM OA decreased basal levels of nuclear CRTC2 by 17%, and decreased the effects of PTH(1-34) by 32% (Fig. 6c). Additionally, PP1/PP2A inhibition did not alter basal levels of nuclear CRTC3, but dramatically decreased the effects of PTH(1-34) on CRTC3 localization by 76% (Fig. 6d). These data suggest that whereas both PP1/PP2A and SIK2 are responsible for CRTC2 localization, PP1/PP2A seem to play a more prominent role in controlling nuclear CRTC3 localization in PTH signaling.

Discussion
We have shown that whereas PTH(1-34), PTHrP(1-36), and ABL exert similar effects on a subset of genes (Table 2), PTHrP(1-36) and ABL appear to be weaker agonists of the cAMP/PKA/SIK2-SIK3/CRTC signaling axis, and this is par-  (1-34), PTHrP(1-36), and ABL ticularly clear in the differential effects on Rankl mRNA, as well as in CRTC2/CRTC3 localization. Our in vitro binding data suggest that in osteosarcoma cells, ABL binding to PTHR1 is similar to PTH(1-34). Because the assay was performed on intact cells and without GTP␥S, which, in assays performed in membranes, enriches for the G protein-uncoupled conformation (R 0 ) through inducing receptor-G protein dissociation (45), we were unable to distinguish between R 0 and R G and further studies must be conducted to delineate this. Nevertheless, the binding of ABL was slightly weaker (ϳ2-fold) than that of human PTH(1-34) and significantly weaker (ϳ10-fold) than that of a rat PTH(1-34) analog, which suggests a relatively weaker overall affinity of ABL as compared with PTH ligands for the rat PTHR1 in the intact cell line. Although the differences we report with respect to the cAMP responses elicited by the peptides vary slightly from other reports (25), we believe that this is due to the nature of the cell types being studied. Our system employs primary calvarial osteoblasts, which express native levels of PTHR1. Nonetheless, we observed time and dose-dependent differences between all three peptides in their ability to activate PKA (Fig. 1, b and c), in addition to downstream activation of CREB (Fig. 1, c and d), which are both sequentially downstream of cAMP stimulation. Our data show that this pathway is titratable, as observed with PKA agonists and antagonists (Fig. 3a), and with the deletion of the PKA regulatory subunit in vivo (Fig. 3, d and e).

Regulation of Rankl by PTH
Because we observed weaker cAMP/PKA signaling by PTHrP(1-36) and ABL compared with PTH(1-34), and others have reported that this signaling accounts for the inhibitory effects of PTH on osteocyte Sost expression through SIK2/ HDAC4/5/MEF2c (23), we expected to find differential inhibition of osteoblast Sost expression by all three peptides. Surprisingly, their inhibitory effects were identical and relatively sensitive in UMR 106-01 cells (Fig. 2b). We attempted to replicate this in primary calvarial osteoblasts, but basal Sost levels were much lower compared with UMR 106-01 cells and we were unable to observe any effects with PTH(1-34) (data not shown). Nonetheless, this finding suggests clear differences in the effects of all three peptides on the SIK/HDAC4/5/MEF2c and SIK/CRTC signaling axes, and that the former is more sensitive to changes in cAMP/PKA activity. Also, the equipotent effects of PTH(1-34), PTHrP , and ABL on Sost suppression may account for the equipotent effects observed in vivo (7). It is also interesting to note the differences between the effects of these peptides on Sost mRNA versus Rankl mRNA. The minimum dose required for an observable change in Sost mRNA was 0.1 nM (1/2 max ϭ 0.085 nM), whereas the effects of all three peptides on RANKL mRNA were observable at 1 nM, and both effects are maximal at 4 h. Because the regulation of each gene requires transcriptional termination and activation for Sost and Rankl, respectively, it is conceivable that this indicates that transcriptional termination may require a weaker stimulus compared with transcriptional activation.
SIKs are direct downstream effectors of cAMP/PKA signaling and respond to metabolic changes in different systems (47)(48)(49). There appears to be redundancy between SIK2 and SIK3 (50), and we observed this in our system, because knockdown of either protein results in the modulation of Rankl mRNA (Fig.   4). Although knockdown of SIK2 yielded greater effects on Rankl mRNA compared with SIK3, SIK3 knockdown was still effective in increasing osteoblastic Rankl mRNA. Although SIK3 is not a reported downstream target of PTH signaling, SIK3 deficiency phenocopies PTHrP overexpression in chondrocytes, suggesting that it is somehow involved in the pathway (51,52).
Our results show that the effects of PTH(1-34) and its analogs differ in the primary osteoblast versus the osteocyte with respect to the regulation of Rankl (Fig. 7). Although others have reported that CRTC2 is the main cofactor involved in osteocytic Rankl expression (23), we report that PTH(1-34) and its analogs indeed affect CRTC2 localization, but CRTC3 appears to be the main effector in the osteoblast. Kim et al. (53) observed redundancy between CRTC2 and CRTC3 in bone marrow, because CRTC2/CRTC3 double knockout mice exhibited embryonic lethality, but only one allele of either CRTC2 or -3 was sufficient for viability. It is possible that there is redundancy between CRTC2 and CRTC3 in the osteoblast, and perhaps there may even be events in late osteoblast differentiation/maturation that generate a switch from CRTC3 to CRTC2 with respect to Rankl signaling. Simultaneous knockdown of CRTC2 and CRTC3 in the osteoblast would shed light on these open issues. However, this was attempted and the knockdown was not effective in primary calvarial osteoblasts (data not shown).
Several groups have implicated CREB in Rankl regulation through its upstream enhancers (17,36,54,55). Interestingly, we have shown that the CREB coactivator, CRTC3, plays a role  , and ABL have differential effects on cAMP stimulation and PKA activation. Nonetheless, PKA phosphorylates salt-inducible kinases SIK2 and SIK3, which lessens the inhibitory phosphorylation of CRTC3 by SIK2/SIK3. Concurrently, PP1/PP2A dephosphorylates CRTC3 and both events dissociate CRTC3 from 14-3-3 cytoplasmic proteins. CRTC3 enters the nucleus and binds an unknown bZIP transcription factor to enhance Rankl gene transcription.
in enhancing Rankl up-regulation, but knockdown of CREB did not hinder the effects of PTH  or the other peptides on Rankl mRNA (Fig. 4, b and c). We postulate that other transcription factors with bZIP domains are able to bind the CRTCs at the Ϫ75 kb enhancer region, but further studies must be conducted to determine this.
Our data implicate PP1 and PP2A as important mediators of PKA signaling in PTH-stimulated osteoblasts. Indeed, inhibition of PP1/PP2A by okadaic acid significantly altered the effects of PTH(1-34) on both Rankl mRNA (Fig. 3b) and CRTC2 and -3 localization (Fig. 6, c and d), but inhibition of PP2A by SV40 small T-antigen was sufficient to attenuate the effects of all peptides on Rankl mRNA (Fig. 3c). Thus, we hypothesize that PP2A is the main Ser/Thr phosphatase involved in PTH signaling. Additionally, pre-treatment with the PP2A agonist, FTY720, was able to increase Rankl mRNA 2-fold (data not shown). It has previously been shown that PKA phosphorylates three Ser residues on the PP2A subunit, BЉ, and that this promotes PP2A activity (56). Further studies in vitro using HEK293 cells and in vivo, in striatal neurons, revealed that PKA phosphorylates the PP2A subunit, B56␦, and this leads to the up-regulation of PP2A activity (57). In another study using NRK-52E and L6 cells, cAMP up-regulation by forskolin and isobutylmethylxanthine (IBMX) led to decreased phosphorylation of PP2A substrate, elongation factor-2 and these dephosphorylation events were blocked by PP2A inhibition with okadaic acid. However, inhibition of PKA with H-89 did not block cAMP-dependent dephosphorylation, suggesting that the pathway of PP2A activation in these cells is cAMP-dependent, but PKA-independent (58). The complexity of its regulation leaves many open questions about the mechanism of osteoblastic PP2A activation and substrate (CRTC3) recognition, which require further investigation. Although calcineurin (or protein phosphatase 3) was implicated in CRTC dephosphorylation (59), our experiments with OA and SV40 small T-antigen support a role for PP2A in the PTH-induced nuclear translocation of CRTC2 and CRTC3, in addition to the regulation of Rankl mRNA expression (Figs. 3, b and c, and 6, c and d). However, inhibition of calcineurin with cyclosporine in primary calvarial osteoblasts would definitively rule out this possibility. Whether or not PP2A directly dephosphorylates CRTC2 and CRTC3 or another substrate (perhaps SIK2 and SIK3)remainstobeseen.WehypothesizethatPP1/PP2Adephosphorylates CRTCs 2 and 3, because PP1/PP2A inhibition decreased the effects of PTH(1-34) on CRTC2 nuclear localization (Fig. 6c) and potently inhibited its effects on CRTC3 localization (Fig. 6d), whereas the effects of SIK2 knockdown were milder. Interestingly, SIK2 knockdown had greater effects on CRTC2 and no effects on CRTC3, which may suggest one of two possibilities: there is an alternative mode of signaling outside of the SIK axis that governs PTH-induced CRTC2 and CRTC3 nuclear localization, or the dephosphorylation of CRTCs plays a more pivotal role in modulating its subcellular localization. SIK2 and PP2A have been shown to interact with each other, and increases in cAMP/PKA activity disrupt this association (47). Others have also shown that this SIK2-PP2A binding acts to preserve the activity of each enzyme to regulate calcium/calmodulin-dependent Protein Kinase I (60). How-ever, we have shown that in the osteoblast, inhibition of PP1/ PP2A did not weaken SIK2 activity, as this would have resulted in increased CRTC localization and Rankl mRNA expression, whereas we observed the opposite (Fig. 6, c and d). Others have shown that in MC3T3-E1 cells, inhibition of PP2A-C␣ by short hairpin RNA decreased Rankl mRNA, but also increased osteoprotegerin mRNA (41). The same group also observed an increase in osteoblastic transcription factors, Osterix, bone sialoprotein, and Osteocalcin, upon inhibition of PP2A by 1 nM okadaic acid (46). Concurrent with our findings, these data suggest a major role for PP2A in osteoblastic signaling and, particularly, in the regulation of Rankl expression. In this study, we have shown titratable differences between PTH(1-34), PTHrP , and ABL on cAMP/PKA signaling in the osteoblast. Our study provides greater insight into how these peptides exert their effects and may provide a deeper understanding as to how their differences are achieved in vivo.

Peptides and chemicals
Rat parathyroid hormone (PTH(1-34)) was purchased from Sigma. Parathyroid hormone-related protein (PTHrP(1-36)) and ABL were custom synthesized by the Peptide/Protein Core Facility at the Massachusetts General Hospital. Human PTH(1-34) was purchased from Bachem. Additionally, a second preparation of PTHrP  and ABL were custom synthesized by Bachem. All PTH(1-34), PTHrP , and ABL peptide sequences were confirmed and analyzed for purity and degradation by the New York University Langone Mass Spectrometry Core Facility. RIPA buffer, 8-bromo-cAMP, and ascorbic acid were purchased from Sigma. Collagenase A was purchased from Worthington Biochemical Corp. Myr-PKI was purchased from Tocris. IBMX was purchased from MP Biomedicals. Paraformaldehyde was purchased from EM Sciences. For binding assays, C-terminal amides were synthesized by the Peptide Core Facility at the Massachusetts General Hospital using conventional Fmoc (N-(9-fluorenyl)methoxycarbonyl)-based solid-phase chemistry and reverse phase-HPLC on a prep-C18 column for purification. Purity (Ͼ95) and molecular mass were established by analytical RP-HPLC and MALDI-TOF-MS.

Antibodies
See Table S2 for a full list of primary antibodies and concentrations. Goat anti-rabbit horseradish peroxidase (HRP)-linked secondary antibody was purchased from Cell Signaling Technologies. Donkey anti-rabbit Alexa Fluor 488 secondary antibody was purchased from Life Technologies.

Cell culture
Rat osteoblastic, osteosarcoma cell line, UMR 106-01, was cultured using Eagle's minimal essential medium, supplemented with 5% fetal bovine serum (FBS), 25 mM HEPES, 1% nonessential amino acids, 100 units/ml of penicillin, and 100 g/ml of streptomycin. Primary calvarial osteoblasts were harvested from C57Bl/6 WT mice aged 2-5 days postnatal. Mice were euthanized with ketamine and xylazine (0.5 mg and 0.025 mg/pup, respectively). Calvariae were digested in 1 mg/ml of (1-34), PTHrP(1-36), and ABL collagenase A at 37°C by five sequential digestions, and cells from digests 3-5 were collected and plated at a density of 6.4 ϫ 10 3 cells/cm 2 in ␣MEM supplemented with 10% FBS and 100 units/ml of penicillin and 100 g/ml of streptomycin. After reaching confluence, osteogenic medium (50 g/ml of ascorbic acid) was added for 3 days to allow osteoblastic differentiation. Primary bone cells were harvested using femoral chips from 3-month-old male C57BL/6 WT mice. Cells were cultured to confluence and after the first plating, were treated similarly to primary calvarial osteoblasts. The osteoblastic phenotype of all primary cells was confirmed by measuring the relative mRNA levels of osteoblastic genes, Col1a1 and Alp. For all experiments, cells were serum-starved with 0.1% FBS 24 h prior to peptide treatments.

In vivo study design
All experiments using mice were performed following protocols approved by the New York University Institutional Animal Care and Use Committee (IACUC). Intermittent subcutaneous peptide injections of 80 g/kg/day were administered to 4-month-old C57Bl/6 male mice for 6 weeks (n ϭ 10/group). Eighteen h after the final injection, tibial shafts were harvested and processed for qRT-PCR analyses. Additionally, 4-week-old Col1-Cre-ER T /Prkar1a fl/fl mice (male: n ϭ 9, female: n ϭ 8) were injected with 1 mg of tamoxifen/mouse once a week for 3 weeks and 5-month-old Col1-Cre-ER T /Prkar1a fl/fl mice (n ϭ 7) were injected with 2.5 mg of tamoxifen once a week for 4 weeks prior to tibial mRNA harvest. Prkar1a fl/fl mice injected with tamoxifen were used as controls.

In vitro binding assays
Ligand binding to the rat PTHR1 in intact UMR 106 cells was assessed by competition methods using [ 125 I-Nle 8,21 ,Tyr 34 ]rat PTH(1-34) as a tracer radioligand (2.2 Ci/mmol). Assays were performed in 96-well plates containing confluent cell monolayers in a buffer of Hanks' balanced salts solution supplemented with 10 mM HEPES, pH 7.4, and 0.1% BSA. Reactions contained radioligand (30,000 cpm/well) and varying concentrations of an unlabeled ligand, and were incubated at 4°C for 17 h. The wells were then rinsed three times with cold buffer, lysed with 1 N NaOH, and the lysate was counted for ␥-irradiation. Data from 4 independent experiments, each with either 3 replicates per dosepoint (1 plate, n ϭ 2) or 6 replicates per dose-point (2 plates, n ϭ 2) were combined and plotted by fitting to a sigmoidal dose-response model with variable slope, such that each data point is an average In vitro cAMP and PKA assays cAMP concentrations were measured using a cAMP XP assay kit and performed as described by the manufacturer (Cell Signaling). Briefly, cells were plated in triplicate in 96-well plates and cultured as described above. Following peptide treatments using ␣MEM supplemented with 0.1% FBS, the cells were washed in warm 1ϫ PBS and lysed with 100 l of lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM Na 2 EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 g/ml of leupeptin) supplemented with 2 mM IBMX. Fifty l were transferred into microtiter plates coated with anti-cAMP rabbit mAb. HRP substrate TMB was added to develop color, spectrophotometry measurements were taken at 405 nM, and readings were calculated against a cAMP standard curve. For PKA assays, cells were plated in triplicate in 6-well dishes and cultured as described above. PKA activity was measured using a PepTag nonradioactive protein kinase A assay and performed as described by the manufacturer (Promega). Following peptide treatments using ␣MEM supplemented with 0.1% FBS, the cells were washed in warm 1ϫ PBS, scraped, and lysed in a Dounce homogenizer with 200 l of PKA extraction buffer (25 mM Tris-HCl, pH 7.4, 0.5 mM EDTA, 0.5 mM EGTA, 10 mM ␤-mercaptoethanol, 1 g/ml of leupeptin, 1 g/ml of aprotinin) and 10 l were incubated with PKA reaction buffer and Peptag A1 Peptide PKA substrate (P-L-S-R-T-L-S-V-A-A-K). Samples were processed in 1% agarose gels and imaged with Bio-Rad gel documentation system for densitometric detection of PKA substrates. Samples were analyzed by measuring the ratio between phosphorylated PKA substrates over total PKA substrates.

siRNA knockdown
si-control, si-SIK2, si-SIK3, si-CRTC2, si-CRTC3, si-CREB, si-HDAC4, and si-MEF2c Accell siRNA reagents were purchased from Dharmacon and transfection was performed as described by the manufacturer. Briefly, siRNA was diluted to a final concentration of 1 M with ␣MEM supplemented with 2.5% fetal bovine serum (FBS) and incubated at 37°C for 1 h. siRNA media were then added to cells and incubated for 72 h prior to peptide treatments. Confirmation of knockdowns was confirmed by Western blotting or RNA (Fig. S3).

Adenoviral infection
For adenoviral infection, primary calvarial osteoblasts were plated as described above. Cells were exposed to 10 pfu/cell of Ad-SV40 small T antigen overnight at 37°C. The following morning, cells were serum-starved in ␣MEM supplemented with 0.1% FBS for 4 h prior to peptide treatments.

qRT-PCR
Total RNA was extracted using TRIzol (Sigma). Complementary DNA (cDNA) was synthesized from 1 g of total RNA using a TaqMan reverse transcription kit (Applied Biosystems) with hexamer primers following the protocol described by the manufacturer. Gene expression levels were measured using SYBR Green PCR reagents (Applied Biosystems). Primer pairs used for quantitative detection of gene expression are listed in Table S1. The quantity of mRNA was calculated by normalizing the threshold cycle value (C t ) of specific genes to the C t of the housekeeping gene ␤-actin and/or glyceraldehyde-3-phosphate dehydrogenase (Gapdh).

Western blotting
Whole cell lysates were prepared using RIPA buffer (150 mM NaCl, 1% IGEPAL, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0; Sigma) supplemented with 1ϫ HALT protease and phosphatase inhibitor cocktails (Thermo). Total protein (1-34), PTHrP(1-36), and ABL concentration was determined by Bradford reagents (Bio-Rad). Twenty g of lysates were resolved on TGX Stain-free SDSpolyacrylamide gels and transferred onto polyvinylidene fluoride membranes using the Trans-Blot Turbo Transfer system (Bio-Rad). Prior to immunoblotting, membranes were imaged to detect total protein content as described by the manufacturer. Membranes were blocked with 5% nonfat milk in TBST, and incubated overnight with primary antibodies. Subsequently, membranes were washed, incubated with HRP-linked secondary antibodies, and signals were detected by chemiluminescence (Pierce). Blots were imaged and quantified using Bio-Rad ImageLab software and calculated against total detectable protein per lane.

Immunofluorescence
Cells were fixed with 3.7% paraformaldehyde and 0.002% Triton X-100 for 30 min and blocked with 1% BSA in 1ϫ PBS for 1 h. Cells were incubated with 1:250 dilution of primary antibodies for 1 h, followed by washes and incubation with 1:450 dilution of Alexa Fluor 488-conjugated secondary antibodies. Cells were incubated with 1:10,000 DAPI (Life Technologies) for nuclear staining and mounted using Fluoromount-G fluorescence mounting medium (Invitrogen). Images were captured with a Zeiss LSM 700 laser scanning confocal microscope and automated quantitation of nuclear localization was performed on ImageJ and measured by comparing intensities of GFP signal overlaid onto nuclear 4Ј,6-diamidino-2-phenylindole (DAPI) signal. For each condition, n ϭ 500 Ϯ 50 cells were counted for statistical analysis.

Statistics
All experiments were performed at least three times and in triplicate per experiment. Statistical differences were analyzed either by Student's t test, or one-way or two-way analysis of variance using IBM SPSS (v24). Results are expressed as mean Ϯ S.D. or S.E. and a p Ͻ 0.05 was considered significant comparing treatment groups.