Regulation of cAMP levels in osteocytes by mechano-sensitive focal adhesion kinase and phosphodiesterase 8A Authors: Garyfallia Papaioannou, Tadatoshi Sato, Caroline Houghton, Chuanqing Yan, Parthena E Kotsalidis, Katelyn E Strauss, Thomas Dean, Courtney M Mazur, Alissa J Nelson, Matthew P Stokes, Thomas J Gardella, Marc N Wein Published: 2026 ABSTRACT SummaryOsteocytes are the primary mechano-sensitive cell type in bone. Mechanical loading is sensed across the dendritic projections of osteocytes leading to transient reductions in focal adhesion kinase (FAK) activity. We performed tyrosine-focused phospho-proteomic profiling in osteocyte-like Ocy454 cells to identify FAK substrates. Gsα, parathyroid hormone receptor (PTH1R), and phosphodiesterase 8A (PDE8A), three proteins associated with cAMP signaling, were found as potential FAK targets. FAK pharmacologic inhibition or gene deletion increased basal and G-protein-coupled-receptor (GPCR) ligand-stimulated cAMP levels and downstream signaling events. PDE8A inhibition mimicked FAK inhibition at the levels of increased cAMP, protein kinase A (PKA) activity, and expression of cAMP-regulated target genes. Co-immunoprecipitation assays revealed an intracellular association between FAK and PDE8A. Thus, mechanically regulated FAK can modulate intracellular cAMP levels via effects on PDE8A. These data suggest a signal transduction mechanism that mediates crosstalk between mechanical and cAMP-linked hormonal signaling in osteocytes.Subject areas: molecular biology, endocrinology, cell biology Introduction Mimicking the skeletal benefits of weight-bearing exercise through molecular signals is a major goal in bone biology. Exercise boosts bone formation and bone mass1 while immobilization or lack of gravity, through a reduction of mechanical loading, lead to a reduction of bone mass.2 Osteocytes play a central role in sensing mechanical and hormonal signals.3,4 Osteocytes detect mechanical loading as changes in fluid flow shear stress (FFSS) across their dendritic projections.5 Mechanical loading stimulates bone anabolism through multiple mechanisms, including by downregulating sclerostin, an osteoblast inhibitor.6 Hormonal cues also reduce sclerostin expression by osteocytes. Parathyroid hormone (PTH), a bone anabolic agent when delivered via once daily subcutaneous injections,7 binds to its Gsα-linked G-protein-coupled-receptor (PTH1R) in osteocytes, increases cAMP levels and inhibits sclerostin expression.3 Therefore, both GPCR signaling and mechanical cues converge to regulate sclerostin expression. Furthermore, through multiple proposed mechanisms,8,9,10 mechanical disuse impairs the bone anabolic effects of intermittent PTH injections. On the contrary, combining exercise and PTH may maximize the effects of either treatment.11 For example, combined intermittent PTH treatment with whole body vibration led to enhanced gains in spine bone density compared to PTH alone.12 Despite these observations, little is known about how mechanical and hormonal signals are integrated within osteocytes. Molecular mechanisms that mimic mechanical signals could be used as drug targets to increase bone mass and enhance the effects of PTH in bone anabolism. We previously described a key role of the tyrosine kinase focal adhesion kinase (FAK) in the intracellular signaling pathway downstream of FFSS.13 FAK is an integral part of focal adhesions, complex structures that assemble at the plasma membrane where integrins interact with the extracellular matrix (ECM).14 FAK is activated downstream of integrin/ECM interaction.15 Relative to other cells, osteocytes show constitutively high FAK signals due to “outside-in” signaling from their surrounding bone matrix via integrins.13 FFSS rapidly causes integrin-extracellular matrix dissociation, thereby reducing FAK activity. When active, FAK phosphorylates tyrosine residues on class IIa histone deacetylase proteins (HDAC4 and HDAC5) and promotes their cytoplasmic retention. When FAK kinase activity is suppressed transiently by FFSS, HDAC4/HDAC5 tyrosine phosphorylation is reduced, leading to their nuclear translocation. In the nucleus, class IIa HDACs suppress myocyte enhancer factor 2 (MEF2)-driven expression of sclerostin (encoded by the SOST gene).13 Since sclerostin normally suppresses bone formation by osteoblasts,6 FAK inhibition could lead to increased bone formation by reducing sclerostin levels. Notably, a dual FAK/PYK2 (proline-rich tyrosine kinase 2) inhibitor increases bone anabolism and bone mass in female rats with bone loss due to surgical hypogonadism16 while administration of a bone-targeted FAK inhibitor prevents glucocorticoid-induced bone loss in mice.17 Despite these advances, further development of FAK inhibitors to treat osteoporosis is limited by lack of knowledge regarding the full suite of FAK substrates in osteocytes. In this study, we demonstrate that FAK regulates signaling downstream of Gsα-coupled GPCRs including PTH1R. Tyrosine-focused phospho-proteomic profiling demonstrated that several proteins involved in cAMP signaling showed reduced phosphorylation in response to FAK inhibition or gene deletion. Cells lacking FAK or with inhibited FAK activity have elevated basal and ligand-stimulated cAMP levels and increased signaling downstream of cAMP. FAK interacts with phosphodiesterase 8A (PDE8A) to control cAMP levels in bone cells. Therefore, mechanically regulated FAK inhibition may play a role in increasing the bone anabolic effects of hormonal signaling. Our data suggest a mechanism of crosstalk between mechanical and cAMP-linked hormonal signaling in osteocytes. Results FAK controls phosphorylation of proteins associated with cAMP signaling To understand how FAK regulates bone anabolism, we sought to identify targets of this tyrosine kinase in osteocytes. Phospho-tyrosine-focused phosphoproteomic profiling was performed to identify substrates of FAK in an osteocyte-like cell line (Ocy454 cells). Tryptic peptides were isolated from cells treated ± the FAK inhibitor VS606318 or ± FAK gene deletion (Figures 1A and 1B). 432 proteins showed reduced tyrosine phosphorylation in FAK knockout (KO) and FAK inhibitor-treated cells compared to control (Figure 1C; Table S1). FAK itself and paxillin are known FAK substrates19,20; thus, our ability to detect reduced phosphorylation of these proteins validates this approach. Among proteins whose tyrosine phosphorylation was reduced by both VS6063 and FAK deletion, Gsα and PTH1R (Figure S1A) were of particular interest due to their known roles in bone biology.21,22,23 In addition, other proteins related to cAMP signaling were identified, including phosphodiesterase 8A (PDE8A) (Figure 1D). Reduced tyrosine phosphorylation of PDE8A was not accompanied by reduced total PDE8A protein levels (Figure S1B). Identification of focal adhesion kinase substrates in Ocy454 cells (A) Experimental design of phosphoproteomic profiling in Ocy454 cells. (B) Immunoblotting with antibodies for phosphorylated tyrosine residues, pFAK, total FAK and β-tubulin. Protein lysates were isolated from control, FAK inhibitor (VS6063, 10 μM, 30 min) treated, and FAK KO Ocy454 cells. (C) 432 proteins showed reduced abundance of tyrosine phosphorylated residues in FAK KO and FAK inhibitor-treated cells compared to the control. (D) Dot plot showing the abundance of tyrosine phosphorylation sites in FAK KO cells and FAK inhibitor treated cells versus control. Left lower quadrant: proteins with reduced abundance of tyrosine phosphorylated residues in both comparisons. The phosphorylation sites of interest are shown in parentheses. FAK and Pxn (known Fak targets) are positive controls. FAK inhibition increases basal and ligand-stimulated cAMP levels Having identified cAMP signaling related proteins regulated by FAK, we asked if FAK could control Gsα signaling. We first used HEK293 cells harboring a GloSensor system24 which generates luminescence proportionate to the amount of intracellular cAMP. In this system, VS6063 treatment rapidly increased intracellular cAMP levels in a dose-dependent manner in the absence of exogenously-added GPCR ligand (Figure 2A). Furthermore, VS6063 pre-treatment augmented agonist (isoproterenol, prostaglandin E2 (PGE2, or PTH)-stimulated cAMP levels (Figures 2B and 2C). HEK293 cells express endogenous receptors for isoproterenol and PGE2. To detect PTH-induced cAMP increases in these cells, we used GloSensor HEK293 cells stably expressing PTH1R.25,26 Similar findings were observed using four distinct pharmacologic FAK inhibitors (PF-431396, PF-562271, GSK2256098, and Y15) (Figure 2C), all of which boosted cAMP levels albeit with different dose-response relationships due to their different FAK IC50 values.27 FAK inhibitors increase cAMP levels in HEK293 cells at baseline and upon GPCR-ligand stimulation GloSensor cAMP luminescence over time is shown (A) Dose response of VS6063 on cAMP in GloSensor-expressing HEK cells. Areas under curve (AUCs) were compared with one-way ANOVA, n = 8, p < 0.0001 for 5 μM and 10 μM vs. all other groups, p = 0.006 for 2.5 μM vs. vehicle, p = 0.0313 for 2.5 μM vs. 0.3 μM, p > 0.05 for other comparisons. (B) GloSensor-expressing HEK293 cells treated with PGE2 (10−3 M) or PTH (200 pM, a submaximal dose) with or without FAK inhibitor (10 μM). GloSensor HEK293 cells stably expressing PTH1R were used for PTH treatment studies. AUCs were compared with one-way ANOVA. For PGE2 experiment: n = 4 and p ≤ 0.0002 for all comparisons. For PTH experiment: n = 12 and p < 0.0001 for all comparisons. (C) Dose response of different FAK inhibitors in GloSensor-expressing HEK cells treated with isoproterenol [10−6 M]. AUCs were compared with one-way ANOVA. For P562271 experiment: n = 8 and p < 0.0001 for all comparisons except p = 0.0435 between 2.5 μM and 1.25 μM AUCs. For P431396 experiment: n = 8 and p < 0.0001 for all comparisons except p = 0.0006 between 5 μM and 2.5 μM and p = 0.0011 between 2.5 μM and 1.25 μM AUCs. For VS6063 experiment: n = 8, p < 0.0001 for all comparisons except p = 0.0674 between 1.25 μM and 2.5 μM. For GSK22: n ≥ 4 and p = 0.0104 between 10 μM and vehicle and p = 0.0376 between 10 μM and 2.5 μM while all other comparisons had p > 0.1. For Y15: n ≥ 12, p < 0.0001 between 10 μM, 5 μM, 2.5 μM, 0.6 μM, and vehicle and p = 0.0005 between 1.25 μM and vehicle while the comparisons of AUCs between different Y15 doses had p > 0.7. Data are represented as mean ± SEM. To study the relevance of these findings to bone cells, we used GloSensor-expressing osteoblast-like Saos2 cells which express endogenous PTH receptors.28 Similar to HEK293 cells, VS6063 treatment increased cAMP levels in Saos2 cells (Figure 3A). Cilengitide is an RGD peptide29 functioning upstream of outside-in integrin signaling which activates FAK. Disrupting ECM/integrins interactions using cilengitide reduces FAK activity in matrix-producing Saos2 cells and in osteocyte-like cells.13 As predicted by our model, cilengitide-induced FAK suppression increased cAMP levels similarly to FAK inhibitors, confirming the effect of the integrin/FAK signaling pathway on cAMP (Figure 3A). Similar to results obtained in PTH1R-expressing HEK293 cells, VS6063 potentiated PTH-induced increases in cAMP levels in Saos2 cells (Figure 3B). Ocy454 cells are a conditionally-immortalized murine osteocyte-like cell line.30 Ocy454 cells were transiently transfected with the GloSensor cDNA and then treated with isoproterenol ± VS6063. Similar to results in HEK293 and Saos2 cells, FAK inhibition potentiated ligand-induced cAMP increases in Ocy454 cells (Figure 3C). FAK inhibition increases cAMP levels in osteoblast- and osteocyte-like cells at baseline and upon PTH stimulation (A) Saos2 cells treated with FAK inhibitor VS6063 (10 μM) or cilengitide (100 μM). AUCs were compared with one-way ANOVA, n ≥ 24, p < 0.0001 for cilengitide or VS6063 vs. vehicle and p = 0.1679 for VS6063 vs. cilengitide. (B) Saos2 cells treated with PTH with or without FAK inhibitor VS6063 (10 μM). AUCs were compared with one-way ANOVA, n = 8, p ≤ 0.008 for all comparisons. (C) Ocy454 cells transiently expressing GloSensor treated with isoproterenol with or without FAK inhibitor VS6063 (10 μM). cAMP luminescence over time is shown in GloSensor assay. Vehicle is DMSO. AUCs were compared with one-way ANOVA, n = 4, p < 0.0001 between VS6063 and vehicle or isoproterenol and p = 0.0224 between vehicle and isoproterenol. Data are represented as mean ± SEM. (D) Immunoblotting for phosphorylated PKA substrates in Ocy454 cells treated with VS6063 (dose response). Cells were treated for 30 min followed by immunoblotting. Protein kinase A (PKA) is a major downstream effector of cAMP.31 The effect of FAK inhibition on PKA activation was demonstrated by measuring the signal of phosphorylated PKA (pPKA) substrates (using a degenerate antibody which detects all proteins bearing the RRxS∗/T∗ PKA consensus phosphorylation motif) in Ocy454 cells treated with FAK inhibitor. This orthologous approach demonstrated increased pPKA substrate signal following VS6063 treatment (Figure 3D). To confirm findings seen with pharmacologic FAK inhibition, we next used genetic approaches to delete or reduce FAK expression. First, FAK-deficient GloSensor HEK293 cells were generated by CRISPR-Cas9-mediated gene deletion (Figure S2). As predicted by our studies with VS6063, FAK-deficient cells showed increased isoproterenol-stimulated cAMP levels compared to control cells (Figure 4A). In addition, FAK knockdown GloSensor HEK293 cells were generated via lentiviral-mediated shRNA knockdown. Like the effects of CRISPR-mediated FAK gene deletion, shRNA-induced FAK knockdown led to increased isoproterenol-stimulated cAMP levels compared to cells infected with lentivirus containing empty vector (Figure 4B). Therefore, genetic and pharmacologic loss of function data indicates that FAK inhibition increases cAMP levels both at baseline and in response to Gsα-coupled GPCR agonists. FAK genetic perturbation increases cAMP levels in HEK293 cells upon stimulation with GPCR ligands (A) Left: Immunoblotting for FAK in protein lysates from FAK KO cells generated via CRISPR-Cas9 (clones 1–3). Right: GloSensor assay revealed increased cAMP luminescence in FAK KO cells compared to control cells upon treatment with isoproterenol (10−6 M). Control cells were transfected with the empty vector PX458. AUCs were compared with one-way ANOVA, n = 24, p < 0.0001 between empty vector and FAK KO1 or FAK KO2 or FAK KO3 (B) Left: immunoblotting for FAK in protein lysates from cells transduced with the indicated lentiviruses. Right: GloSensor assay revealed increased cAMP luminescence in FAK knockdown cells (shRNAs 1–2) compared to control cells upon treatment with isoproterenol (10−6 M). Control cells were infected with the empty pLKO.1 vector. AUCs were compared with one-way ANOVA, n = 9, p < 0.001 between empty vector and shFAK-1 or shFAK-2. Data are represented as mean ± SEM. FAK regulates cAMP signaling downstream of Gsα Next, we sought to understand the mechanism through which FAK controls cAMP levels. First, we focused on Gsα itself since pharmacologic and genetic FAK inhibition augments cAMP levels at “baseline” and in response to several Gsα-coupled receptors. FAK-regulated phosphorylation of two Gsα tyrosine residues (Y358 and Y360) was detected in our phospho-proteomic profiling study. Notably, these residues are located near the Gsα/PTH1R interface (Figure 5A)32 thus their regulated tyrosine phosphorylation may affect GPCR/G protein coupling. However, a tyrosine-to-phenylalanine mutated Gsα (2 YF) construct showed similar effects on basal and ligand-stimulated cAMP levels as a wild type Gsα expression construct when introduced into Gsα deficient HEK293 cells (Figure 5B). The regulation of cAMP by FAK likely occurs via a mechanism downstream of adenyl-cyclase (A) PTH1R-Gsα/β/g cryo structure. Y358 and Y360 residues of Gsα are located at the interface with the IC-3 loop of the receptor and may be contacting residues in that loop or the flanking TM5 and TM6. The receptor is green with TM6 darker green and kinked in the middle for the active state. Gsα is blue and β/g is orange. Y358 and Y360 of Gsα are yellow for carbons and red for Oxygens. The sidechain hydroxyl (OH) of Y358 is exposed to the IC3 loop from the outside while the OH of Y360 is accessed best from the other side of the complex and can be seen in the 180° rotated view on the right. This model that was generated in PyMOL software using PDB datafile 6NBF. (B) cAMP luminescence over time in Gsα KO cells transfected with wild type (Gsα WT + vehicle) or 2 YF (Y358F/Y360F) mutant (Gsα 2 YF + vehicle) Gsα plasmids with or without FAK inhibitor. AUCs were compared with one-way ANOVA, n = 4, with significant comparisons: p < 0.0001 for EV + Vehicle vs. WT + VS6063 or 2 YF + VS6063, EV + VS6063 vs. WT + VS6063 or 2 YF + VS6063, WT + vehicle vs. 2 YF + VS6063, 2 YF + vehicle vs. 2 YF + VS6063, p = 0.0002 for WT vehicle vs. WT + VS6063 and p = 0.0007 for WT + VS6063 vs. 2 YF + vehicle. (C) cAMP luminescence over time in HEK293 cells transfected with Wt PTH1R (PTH1R WT + vehicle) or YF (Y494F/Y496F) mutant (PTH1R 2 YF + vehicle) PTH1R plasmids ± FAK inhibitor (10 μM), AUCs were compared with one-way ANOVA, n = 6, with significant comparisons: p < 0.0001 for EV + vehicle vs. WT + VS6063 or 2 YF + VS6063, EV + VS6063 vs. WT + VS6063 or 2 YF + VS6063, WT + vehicle vs. WT + VS6063 or 2 YF + VS6063, WT + VS6063 vs. 2 YF + vehicle, 2 YF + vehicle vs. 2 YF + VS6063 and p = 0.0227 for EV + vehicle vs. Wt+ vehicle. (D) cAMP luminescence over time in Gsα KO cells treated with vehicle or FAK inhibitor (10 μM). The AUCs of the 2 groups were compared by t test, n = 8, p = 0.0147 (E) cAMP luminescence over time in HEK293 cells treated with forskolin (10 μM) ± FAK inhibitor (10 μM). AUCs were compared with one-way ANOVA, n = 4, p < 0.001 for forskolin+VS6063 vs. vehicle or vehicle+forskolin and p = 0.0141 for vehicle vs. vehicle + forskolin. Data are represented as mean ± SEM. We next interrogated the effects of PTH1R C-terminal tail tyrosine phosphorylation (Y494 and Y496) sites using a similar mutagenesis strategy. PTH1R is phosphorylated on its cytoplasmic tail at Y494 by insulin-like growth factor type I receptor, a modification that may regulate PTH1R localization and function.33 Although PTH1R tyrosine phosphorylation likely does not explain changes in basal or isoproterenol/PGE2-stimulated cAMP levels, these studies were performed based on the central role of PTH in bone metabolism and osteocyte biology.34 Similar to Gsα, tyrosine to phenylalanine mutagenesis of PTH1R (Y494 and Y496) did not affect PTH-induced cAMP induction (Figure 5C). Furthermore, since GPCR C-terminal tails play important roles in β-arrestin recruitment and receptor internalization,35 we tested this “2 YF” PTH1R construct in PTH1R internalization and β-arrestin recruitment and saw no obvious differences between the mutated and the wild type receptor (Figure S3A). In addition, FAK inhibitors did not cause changes in PTH1R internalization (Figure S3B) or Gqα/PKC-dependent increases in intracellular calcium (Figure S3C). Therefore, FAK-mediated changes in cAMP levels are unlikely to occur via direct effects on Gsα and PTH1R at the specific tyrosine sites detected in our phosphoproteomic dataset. At present, we cannot rule out lack of sensitivity to detect effects of point mutations due to limitations associated with overexpression systems. To pinpoint the level(s) in the cAMP signaling cascade at which FAK inhibitors exert their effects, we treated Gsα KO cells with VS6063 and, unlike the case in control cells (Figure 2A), observed minimal increases in cAMP levels (Figure 5D). Therefore, it is unlikely that FAK inhibition directly activates adenylate cyclase in the absence of Gsα. In contrast, VS6063 treatment boosted cAMP levels generated by forskolin, a direct adenylate cyclase activator (Figure 5E). To address the possibility that VS6063 might directly inhibit phosphodiesterases as an “off-target” effect, this compound was tested in vitro against a panel of recombinant phosphodiesterases. VS6063 did not significantly inhibit PDE activity at doses similar to our cell culture studies, with the exception of weak PDE2A inhibition (Table S2). Therefore, it is unlikely that “off-target” direct PDE inhibition by VS6063 is responsible for the cAMP-increasing effects of this compound. Taken together, these findings suggest that the molecular action of FAK in cAMP signaling occurs downstream of adenylate cyclase. Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes We next turned our attention to phosphodiesterase 8A (PDE8A), a selective cAMP- phosphodiesterase36 that was also identified by phosphoproteomics as a possible FAK substrate at position Y315 (Figure 1D). GloSensor Saos2 cells were treated with the selective PDE8A inhibitor PF0495732537 ± PTH. Both basal and ligand-stimulated cAMP levels were increased by PDE8A inhibitor treatment (Figure 6A). To confirm these pharmacologic findings, PDE8A-deficient GloSensor HEK293 cells were generated. These cells demonstrated increased isoproterenol-stimulated cAMP levels compared to control cells (Figure 6B). PDE8A inhibition regulates cAMP levels (A) Left: GloSensor cAMP assay in Saos2 cells following PDE8A inhibitor treatment (PF04957325, 10 μM). The AUCs of the 2 groups were compared by t test, n ≥ 24, p < 0.0001. Right: Dose response of the PDE8A inhibitor PF04957325 in cAMP in GloSensor-expressing Saos2 cells. cAMP luminescence over time is shown. One-way ANOVA, n ≥ 4, p < 0.0001 for vehicle vs. 10 μM and 10 μM vs. 1 μM, p = 0.0014 for vehicle vs. 3 μM, p = 0.1136 for vehicle vs. 1 μM, p = 0.0036 for 10 μM vs. 3 μM, p = 0.2536 for 3 μM vs. 1 μM. (B) Left: PDE8A immunoblotting in PDE8A KO cells. Right: GloSensor assay revealed increased cAMP luminescence in PDE8A KO cells compared to control cells upon treatment with isoproterenol (10−6 M). The AUCs of the 2 groups were compared by t test, n = 30, p = 0.0007. Data are represented as mean ± SEM. (C) RT-qPCR showing relative Sost expression upon ± treatment with PDE8A inhibitor 10 μM (left) and dose response of PDE8A inhibitor (right). Cells were treated for 4 h prior to RNA isolation. For each group in control vs. PDE8A inhibitor treated cells n = 6 biologic replicates for RNA and two-sided unpaired t test were used. N = 2 biologic replicates were used for the dose response experiment. (D) RT-qPCR showing relative expression of Sost and Rankl after treatment with PDE8A inhibitor (1 μM) in a time course. (E) RT-qPCR showing relative expression of Rankl, Wnt4, and Sost after treatment with PTH (100 nM) ± PDE8A inhibitor (1 μM). p values are shown on the graphs. PTH signaling in osteocyte-like cells leads to numerous gene expression changes, including increased Rankl (formal gene name, Tnfsf11) and reduced Sost expression.38 Therefore, we monitored these target genes to measure the effects of PDE8A on PTH-stimulated gene regulation. PDE8A inhibitor treatment led to reduced Sost expression in Ocy454 (Figure 6C) in a manner like that of FAK inhibitor treatment.13 To better define the effects of PDE8A inhibitor on gene expression we performed dose response (Figure 6C) and time course (Figure 6D) experiments in Ocy454 cells which demonstrated responses at 4 h and 1 μM, a dose that is specific for PDE8 compared to other phosphodiesterases.39 Finally, the PDE8A inhibitor showed combined effects with PTH at the level of gene expression (Figure 6E). Specifically, Rankl and Wnt4 upregulation was enhanced when PDE8A inhibitor was added to PTH compared to PTH alone. Sost suppression was maximized with PTH; for this reason, no additional suppression was seen when PDE8A inhibitor was added (Figure 6E). Next, we deleted PDE8A via CRISPR-Cas9-mediated gene deletion in Ocy454 cells (Figure S4A). Due to suboptimal specificity of the existing PDE8A antibodies to detect endogenous mouse PDE8A protein we screened single cell clones of PDE8A KO cells using Sanger sequencing (Figure S4B) which identified a clone bearing a biallelic stop codon just downstream of the CRISPR editing site. This clone showed lower PDE8A mRNA expression compared to control cells (Figure S4C). PDE8A KO cells showed increased basal and PTH-stimulated Rankl expression (Figure S4D) and reduced Sost expression (Figure S4E). This suggests that the deletion of PDE8A in osteocytes mimics or enhances the molecular mechanisms activated by PTH. Next, we sought to assess the global transcriptomic relationship between PDE8A gene deletion and PTH treatment by performing RNA-sequencing in Ocy454 cells. Strikingly, PDE8A knockout cells (versus wild type control cells) show dysregulated expression (|logFC| >1, padj < 0.05) of 1,452 genes (Figure 7A). When this group of PDE8A-dependent differentially expressed genes (DEGs) was compared to the suite of genes regulated by PTH treatment (672 DEGs, Figure 7B), significant transcriptomic overlap was noted (204 commonly regulated genes by both PTH and PDE8A gene deletion, p < 5.524 E-37 for overlap) (Figure 7C; Table S4 lists all DEGs from individual comparisons and overlap analyses; Tables S5, S6, and S7 list all genes detected in RNA-sequencing analysis with read counts from individual libraries and differential expression statistics). PDE8A and FAK regulate similar signaling pathways RNA-sequencing was performed on control, PDE8A KO, PTH (100 nM), and FAKinh (PF431396 10 μM) treated Ocy454 cells. (A, B, and D) Volcano plots demonstrate similarly and differentially expressed genes (blue dots) in each comparison with control cells. (C, E, and F) Venn diagrams reveal number of genes regulated in each cell group and in comparison, (groups: PTH [100 nM], PF431396 [10 μM], PDE8AKO). (G, H, and I) Enrichr gene ontology analysis with WikiPathways 2024 mouse showing the top 10 annotations for the genes that are commonly regulated between the different groups. Our functional studies suggest that FAK inhibition works via blocking PDE8A activity to increase cellular cAMP levels. Thus, we also performed RNA-sequencing in Ocy454 cells treated ± FAK inhibitor (PF431396) and assessed the relationship between the transcriptomic effects of this pharmacologic perturbation and PDE8A gene deletion. Similar to previous studies in Ocy454 cells treated with FAK inhibitors,13 PF431396 treatment led to identification of 3,532 DEGs (Figure 7D). PF431496 treatment also had significant gene expression overlap with PTH: a core set of 226 genes were regulated by both PTH and FAK inhibitor treatment (p < 0.036 for overlap, Figure 7E). Remarkably, concordant transcriptomic effects were also observed between PDE8A gene deletion and PF431396 treatment (521 overlapping DEGs, p < 1.171E-06 for overlap) (Figure 7F). Gene ontology analysis of the groups of genes commonly regulated by PTH, FAK inhibition, and PDE8A gene deletion suggested their involvement in similar pathways including their participation in Wnt signaling and endochondral ossification (Figures 7G–7I). Phosphodiesterase 8A associates with FAK within cells Co-immunoprecipitation studies demonstrated that FAK and PDE8A associate in cells (Figure 8). Interestingly, in a cell-free system we did not detect changes in PDE8A phosphodiesterase activity following its association with FAK (Figures S5A and S5B). These findings raise the possibility that intracellular factors may be necessary for FAK to regulate PDE8A-driven cAMP hydrolysis in cells. In addition, the recombinant FAK used in these experiments may not be fully active and maximal FAK activity requires Src (or Src family member) phosphorylation40 which is absent in the in vitro conditions we used. Treatment of cells with an FAK inhibitor leads to reduced PDE8A expression (Figure S5C), suggesting potential regulation of PDE8A by FAK at both the transcriptional and post-translational levels. FAK and PDE8A associate in cells (A) Immunoblotting for FLAG and V5 in lysates from cells transfected with plasmids expressing FAK and PDE8A tagged with FLAG and V5 respectively after immunoprecipitation with anti-FLAG affinity gel. (B) Immunoblotting for FLAG and V5 in protein lysates from cells transfected with plasmids expressing FAK and PDE8A tagged with FLAG and V5 respectively after immunoprecipitation with anti-V5 affinity gel. Taken together, our findings that PDE8A inhibition phenocopies the effects of FAK inhibition in cells and that FAK can associate with PDE8A suggest a model (Figure 9) in which PDE8A functions downstream of FAK to participate in crosstalk between FAK and cAMP signaling in osteocytes. Model schematizing how FAK controls cAMP levels via PDE8A downstream of mechanical and hormonal cues In osteocytes, constitutive ECM/integrin interactions leads to a high baseline FAK activation. Normally, active FAK interacts with PDE8A and maintains high basal PDE8A-mediated cAMP hydrolysis. FFSS disrupts ECM/integrin association causing transiently reduced FAK activity, reduced FAK-PDE8A interaction, and, in turn, reduced PDE8A-mediated cAMP catabolism. Thus, FAK inhibition leads to suppressed PDE8A function causing an accumulation of cAMP that is generated by upstream GPCR signaling (e.g., isoproterenol, PGE2, PTH). Abbreviations used: ECM, extracellular matrix; FAK, focal adhesion kinase; PDE8A, phosphodiesterase 8A; FFSS, fluid flow shear stress; iso, isoproterenol; PGE2, prostaglandin E2; PTH, parathyroid hormone. Discussion These findings suggest that the mechanosensitive FAK/PDE8A pathway enhances the effects of PTH signaling in osteocytes. FAK is a non-receptor tyrosine kinase that is expressed broadly in human tissues and can regulate many cellular functions including cell survival, proliferation, and migration.41 Extracellular signals including those from extracellular matrix/integrins lead to FAK recruitment to focal adhesion complexes.41,42 There, FAK undergoes activation through a multi-step process that involves multiple phosphorylation events and association with other proteins.43 Several active site (ATP-competitive) FAK inhibitors have been developed, mainly for targeting FAK in cancer.27 Most FAK inhibitors used in this study are ATP-competitive inhibitors that target the kinase domain (VS6063, GSK2256098, PF431396, and PF 562271) while Y15 is a FERM (four-point-one, ezrin, radixin, and moesin) domain inhibitor.27 VS6063, PF431396, and PF562271 are dual FAK/PYK2 inhibitors while GSK2256098 and Y15 are more specific for FAK.27 PYK2 is a closely related kinase that is highly expressed in osteoclasts and controls bone resorption by regulating the osteoclast cytoskeleton.44 FAK has an important role in cellular mechanotransduction, the process in which cells sense extracellular mechanical stimuli and translate them into intracellular biochemical signals.45 In osteocytes, a mechanotransduction-independent role of FAK that regulates cell survival has been reported.46 Osteoblasts that lack FAK failed to induce fluid flow early responsive genes in vitro.47 We previously demonstrated that osteocytes’ high basal FAK activity is inhibited by mechanical cues leading to reduction of its function. In many cell types, FAK activity is constitutively low and is activated by cell-matrix interactions through “outside-in” integrin signaling.48 The dendritic projections of osteocytes are associated with bone matrix which is replete with integrin ligands such as fibrillar collagens, osteopontin, and other bone matrix proteins, thus making them unique regarding constitutive cell/ECM interactions.49 Bone loading causes mechanically induced changes in FFSS across osteocyte dendrites and subsequent transient disruptions in cell/matrix association.50 Ocy454 cells produce high levels of extracellular matrix.51 Constitutive outside-in ECM/integrin signaling leads to basal FAK activity in Ocy454 cells which is disrupted by FFSS.13 The RGD peptide cilengitide disrupts integrin/ECM interactions, reduces FAK activity, and regulates the expression of FFSS-responsive genes similarly to FAK inhibitors.13 The inhibition of FAK activity caused by FFSS leads to reduced tyrosine phosphorylation of class IIa HDACs, their nuclear translocation and suppression of MEF2-driven sclerostin expression.13 Here, we find another link between mechanosensitive FAK and osteocyte function. Normally, active FAK associates with the phosphodiesterase PDE8A and maintains high basal PDE8A-mediated cAMP hydrolysis. Upon mechanical cues, FAK activity is temporarily suppressed, leading to reduced cAMP catabolism. Thus, FAK inhibition, either via mechanical signaling or by the direct FAK inhibition strategies used here leads to an accumulation of cAMP (Figure 9). FAK is associated with the phosphodiesterase PDE8A and PDE8A inhibition leads to increased cAMP in osteocytes. Therefore, it is likely that PDE8A inhibition mediates the FAK-induced increase in cAMP levels. This is the first study in osteocytes to identify a direct link between mechanical stress, via FAK, and cellular phosphodiesterase function. In addition, this is the first study to identify a role of PDE8A in osteocytes. cAMP is a key second messenger downstream of PTH during bone remodeling.21,22,23 PKA activation in human mesenchymal stromal cells increases osteogenesis.52 The duration of cAMP signal is important in promoting or inhibiting osteogenesis53 with transient cAMP signaling leading to osteogenesis whereas continuous cAMP blocks osteoblast maturation.54 PTH generates cAMP through GPCR signaling in plasma membrane or within endosomes by PTH/PTH1R.55 The cAMP signal strength and duration depend on the PTH concentration but also on the PTH1R conformation and the internalization of the receptor/ligand complex.21,56 Here, we report that cAMP levels are regulated by FAK-dependent PDE8A modulation as an additional signaling checkpoint in osteocytes. The role of phosphodiesterases and the concept of using phosphodiesterase inhibitors to treat osteoporosis is not entirely novel.54 Inhibition of phosphodiesterases PDE2, PDE3, and PDE4 increased osteoblast differentiation in vitro.57 PDE7 gene silencing increased osteoblast differentiation and mineralization in vitro.58 Administration of the non-specific PDE inhibitor pentoxifylline increased bone mass in mice through production of Wnt10b from CD8+ T cells.59 Pentoxifylline also increased skeletal angiogenesis and reversed osteopenia in ovariectomized rats.60 Pentoxifylline, when combined with PTH, enhanced BMP2- induced ectopic bone formation, and an additive effect of PTH and phosphodiesterases in bone formation was suggested.61 Both cAMP and cGMP phosphodiesterases may play roles in bone biology. Inhibition of the cGMP phosphodiesterase PDE5 increased angiogenesis and bone formation in a murine fracture model62 and restored bone mass in ovariectomized mice.63 The cAMP phosphodiesterase PDE4 inhibitor rolipram increased bone mass after 5 weeks of daily injections.64 An association between bone mass and PDE4D gene polymorphisms was found in an English population sample.65 Rolipram has an inhibitory effect on osteoclastogenesis in vivo and in vitro66,67 though other in vitro studies suggested the opposite.68 Future work will be needed to ask if cAMP-specific phosphodiesterases other than PDE8A are regulated by FAK. Interestingly, FAK indirectly associates with PDE4D in squamous cancer cells via an intermediate molecule, RACK1, to control cell polarization and adhesion.69 This FAK/PDE4D interaction controls melanoma cell invasion.70 Current models report that cells respond to extracellular cues by modulating the FAK/PDE4D interaction.71 Functional and physical associations between FAK and PDE8A have not been reported. We considered the possibility that the FAK inhibitor VS6063 compound might directly inhibit phosphodiesterases as an off-target effect. Therefore, we tested VS6063 against a panel of recombinant phosphodiesterases in vitro. VS6063 did not significantly inhibit the activity of phosphodiesterases at doses similar to our cell culture studies, with the exception of weak PDE2A inhibition (IC50 3.8 μM). Therefore, an “off-target” effect that causes cAMP elevation via direct inhibition of phosphodiesterases by VS6063 is unlikely. FAK-PDE8A association may affect its interaction with cellular binding partners. Raf1 binds PDE8A via PDE8A residues 454–461. PDE8A protects Raf1 from inhibitory PKA-mediated phosphorylation.72 PDE8A also associates with the PKA regulatory subunit (Riα isoform)73 and 14-3-3 proteins.74 This is the first study to report an interaction between PDE8A and FAK. The predicted PDE8A tyrosine phosphorylation site that is regulated in response to FAK inhibition (Y315) in our phosphoproteomics assay (Figure 1) is located in the PAS domain of the protein, a site of interaction with other proteins or molecules.75,76 Further study is needed to understand the details of this protein-protein interaction and how FAK phosphorylation controls PDE8A activity in cells. Generating a PDE8A PAS domain inhibitor77,78 could be used as another approach to inhibit FAK-PDE8A interaction and, potentially, modulate PDE8A function. Phosphodiesterase assays did not reveal obvious effects of FAK phosphorylation on PDE8A enzymatic activity in vitro. Since the predicted FAK phosphorylation site in PDE8A is outside of its phosphodiesterase domain (Y315 in the phosphoproteomic dataset), it is possible that FAK may promote PDE8A activity in cells via an allosteric mechanism requiring cofactors not present in our in vitro assays using recombinant proteins. PDE8A activity was not modified by FAK in an in vitro activity assay (Figures S5A and S5B). One possibility to explain this observation is that the recombinant FAK used in these experiments is not fully active. The recombinant human FAK we used consists of amino acids 393–698, bearing part of the linker between the FERM and kinase domain, the kinase domain, and major autophosphorylation site (Y397)79,80 as well as additional tyrosine phosphorylation sites including Y576 and Y577. However, maximal FAK activity requires Src (or Src family member) Y576 and 577 phosphorylation40 which is absent in the in vitro conditions we used. Future studies are needed to define if PDE8A is a direct FAK target and to understand how FAK modifies the cellular functions of PDE8A. Future research will investigate the role of this pathway in bone. The effects of conditional FAK deletion in vivo appear to be cell stage-specific in the osteoblast lineage. When FAK was deleted in mice in osteoblast progenitors using Osx-Cre, decreased osteoblast number and low bone mass were observed.81 A similar phenotype was seen by FAK deletion in osteoprogenitor cells using Dermo1-Cre.82 Constitutive FAK deletion in mature osteoblasts via Col1a1-Cre yielded no obvious skeletal phenotype at baseline or in response to mechanical loading83,84; however, adequate deletion of the FAK protein in bone cells was not shown. In addition, these mouse models demonstrate the results of continuous FAK inhibition in bone cells. In contrast, daily injections of a dual FAK-PYK2 inhibitor caused net bone gain in an osteoporosis rat model16 while a bone-targeted PYK2 inhibitor given orally, prevented glucocorticoid-induced bone loss.17 Intermittent PTH causes net gains in bone mass, while continuous PTH elevations cause net bone loss.7 Since, like PTH, FAK inhibition upregulates intracellular cAMP levels, this treatment may have discordant effects when it occurs continuously versus intermittently. The phosphodiesterase inhibitor PF-04957325 has been used in mice in vivo and can alleviate inflammation, autoimmune encephalomyelitis,37 and Alzheimer-like nervous system changes.85 Future animal studies will assess the effect of PDE8A inhibitors in bone in vivo. In summary, here, we demonstrate a link between mechanical signals and phosphodiesterase function that regulates the cAMP pathway in osteocytes and results in enhanced parathyroid hormone signaling. These results provide a mechanistic connection between mechanical and PTH signaling in bone cells and suggest potential avenues for future development of osteoporosis therapeutics that simultaneously engage multiple bone anabolic pathways. Limitations of the study This study has certain limitations. First, experiments here are performed entirely in vitro. Though our previous work confirmed the important role of FAK inhibition in suppressing sclerostin in vivo13 the present study does not include experiments in mice. Second, we do not directly demonstrate an effect of mechanical stress on PDE8A or cAMP. Our previous study13 showed that FFSS reduces FAK phosphorylation in osteocytes. Here, we report that FAK inhibition increases cAMP levels (Figure 2A), thereby indirectly connecting FFSS with cAMP. Moreover, the effect of integrin antagonism via the RGD peptide cilengitide was measured. Integrins respond to environmental cues including mechanical stress.14 Cilengitide reduces FAK phosphorylation in Ocy454 cells13 and the present study demonstrates that cilengitide treatment also leads to increased cAMP signaling, further suggesting a relationship between mechanical signals and cAMP. Currently, technical hurdles preclude real time cAMP measurements using the GloSensor system in cells exposed to FFSS. Future studies are needed to directly assess the effects of FFSS on cAMP in osteocytes in real time. Third, future studies using targeted tyrosine-to-phenylalanine mutagenesis are needed to test if PDE8A is a direct FAK target and the functional consequences of a potential FAK-mediated PDE8A Y315 phosphorylation (predicted by phosphoproteomics assay) on PDE8A function in cells. Fourth, the FAK inhibitors VS6063, PF562271, and PF431396 are dual FAK/PYK2 inhibitors. However, PYK2 is expressed 200-fold less in osteocytes compared to FAK (RNA-seq data from38) and its role is known to be significant for osteoclast function.44 Therefore, the effect in cAMP signaling that is seen with these compounds is osteocytes is most likely mediated by the inhibition of FAK. Resource availability Lead contact Dr. Garyfallia Papaioannou (gpapaioannou@mgh.harvard.edu). Materials availability For all materials related questions please contact Drs. Marc N Wein (mnwein@mgh.harvard.edu) and Garyfallia Papaioannou (gpapaioannou@mgh.harvard.edu). Data and code availability •Data: Primary RNA sequencing data are available in the databases: BioProject: PRJDB37988 and NCBI GEO: GSE326861. Data: Primary RNA sequencing data are available in the databases: BioProject: PRJDB37988 and NCBI GEO: GSE326861. •Code: This article does not report original code. Code: This article does not report original code. •Other: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Other: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments G.P. acknowledges support from the National Institute Of Arthritis And Musculoskeletal And Skin Diseases of the National Institutes of Health (NIAMS, Clinical Investigator Award K08AR084618), the MGH Endocrine Division T32 NIH training grant T32DK007028, the Endocrine Fellows Foundation, and the Foundation of Women’s Wellness (Peggy Ogden Women’s Health Fellowship Award). M.N.W. acknowledges support from NIDDK (R01DK116716) and NIAMS (P50AR080596), the Smith Family Foundation Odyssey Award, and a Chen Institute MGH Research Scholar award. T.J.G. acknowledges support from DK11479, DK113039, and AR066261. T.S. acknowledges support from NIH/NIAMS (R21AR079633, R21AR084644) and Center for Skeletal Research P30 core P&F award (P30 AR075042). Author contributions G.P., T.S., C.H., C.Y., P.E.K., K.E.S., T.D., and A.J.N. performed experiments. G.P., T.S., C.M.M., A.J.N., M.P.S., T.J.G., and M.N.W. analyzed data. G.P. and M.N.W. wrote the manuscript. All authors reviewed and approved the manuscript. Declaration of interests M.N.W. previously received research funding from Radius Health and is a coinventor on a pending patent (US patent application 16/333,546) regarding the use of SIK inhibitors for osteoporosis. M.N.W. holds equity in and is a scientific advisory board member for Relation Therapeutics. M.S. and A.J.N. are employees of Cell Signaling Technology. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesphospho-PKA substrate (RRxS∗/T∗)Cell Signaling TechnologyCat#9624S; RRID: AB_331817FAKCell Signaling TechnologyCat#13009; AB_2798086p-FAK(Y397)Cell Signaling TechnologyCat#8556; AB_10891442beta-tubulinCell Signaling TechnologyCat#5346; AB_1950376FLAGCell Signaling TechnologyCat#2368S; AB_2217020V5Cell Signaling TechnologyCat#13202S; AB_2687461PhosphotyrosineCell Signaling TechnologyCat#8954S; AB_2687925PDE8AProteintechCat#13956-1-AP; AB_2161728Chemicals, peptides, and recombinant proteinsVS6063SelleckchemCat#S7654PF431396SelleckchemCat#S7644PF562271SelleckchemCat#S2890GSK2256098MedChemExpressCat#HY-100498Y15MedChemExpressCat#HY-12444cilengitideAPExBIOCat#A8660PF04957325MedChemExpressCat#HY-15426ForskolinSigmaCat#F6886PTH (1–34)Synthesized by the Massachusetts General Hospital (MGH) peptide core facility–PolyJet transfection reagentSignaGenCat#SL100688calcium-sensitive fluorophore Fura2-AMInvitrogenCat#Ab176766Recombinant human PDE8ASigmaCat#SRP0273Recombinant FAKPromegaCat#V1971anti-FLAG affinity gelSelleckchemCat#B23101V5 affinity gelMilliporeCat#A7345V5 peptideSigma AldrichCat#V7754Critical commercial assaysPTMScan phosphotyrosine pY-1000 KitCell Signaling TechnologyCat#8803QIAshredderQIAGENCat#79656PureLink RNA mini kitInvitrogen12183018APrimescript RT kitTakara Inc.RR037ALANCE Ultra cAMP kitPerkin ElmerCat#TRF0262Deposited dataRNA Sequencing dataBioProject databasePRJDB37988NCBI GEO databaseGSE326861Experimental models: Cell linesGS22 cellsPreviously generated in our lab28–GP2.3 cellsPreviously generated in our lab28–GBR-24 cellsPreviously generated in our lab86–GPG10 cellsPreviously generated in our lab86,87–GSG-5 cellsPreviously generated in our lab88–SGS-72 cellsPreviously generated in our lab26–Ocy454 cellsPreviously generated in our lab30–OligonucleotidesSee Table S3––Plasmid psPAX2psPAX2 was a gift from Didier Trono http://n2t.net/addgene:12260Addgene 12260Plasmid MD2.GpMD2.G was a gift from Didier Trono http://n2t.net/addgene:12259Addgene 12259Plasmid pLKO.1pLKO.1 puro was a gift from Bob Weinberg http://n2t.net/addgene:8453.89Addgene 8453pLV[Exp]-Puro-EF1A>{rGs-WT-FLAG}VectorBuilder–pLV[Exp]-Puro-EF1A>{rGs-WT-FLAG-YFYF}VectorBuilder–pRP-hP1R, pRP[Exp]-CMV>{hP1R-HA-Y494F,Y496F}VectorBuilder–pRP- EF1a> V5/hPDE8AVectorBuilder–pRP-EF1aVectorBuilder–pRP EF1A > FLAG/hFAKVectorBuilder–Software and algorithmsGraphPad Prism 10GraphPad softwareRRID:SCR_00278ImageJSchneider et al.90https://imagej.net/ij/DeepVennhttps://www.deepvenn.com/–Venni 2.1https://bioinfogp.cnb.csic.es/tools/venny/–Enrichrhttps://amp.pharm.mssm.edu/Enrichr/–SalmonPatro et al.91– Experimental model and study participant details Cell lines GS22 cells, HEK293 cells stably expressing the GloSensor cAMP reporter, GP2.3 cells, HEK293/GloSensor (GS22) cells stably expressing human PTH1R28; GBR-24 cells, HEK293/GloSensor (GS22) cells stably expressing bArrestin2yfp86; GPG10 cells, HEK293/Glosensor (GS22) cells stably expressing a pH-sensitive PTH1R (PTH1R-GFPpHs),86,87 and GSG-5 cells, HEK293-Gαs-KO/GloSensor cells,88 were all maintained in DMEM (10% Fetal Bovine Serum (FBS), 1% penicillin/streptomycin).92 SGS-72 cells, human osteosarcoma SaOS-2 cells stably expressing the GloSensor cAMP reporter, were maintained in modified McCoy’s 5A medium (15% FBS, 1% penicillin/ streptomycin, and 1% nonessential amino acids).26 All the above cells were maintained at 37 degrees in 5% CO2. Ocy454 cells30 were maintained in alpha-MEM supplemented with 5% FBS and 1% penicillin/streptomycin at 33 °C with 5% CO. A single-cell sub-clone of Ocy454 cells was used for all experiments. Cells were seeded at 50,000 cells/mL and allowed to reach confluency at 33°C in 2–3 days. Then, cells were transferred from 33°C to 37°C to inactivate the temperature-sensitive T antigen and facilitate osteocytic differentiation. For protein and gene expression analyses, cells were analyzed after culture at 37°C for 14–21 days. Cells were routinely assessed for Sost and Dmp1 expression at 37°C and examined for osteocytic morphology. Cells were routinely screened for Mycoplasma contamination. HEK293T cells (and their derivative cell lines), Saos2 cells and Ocy454 cells are female in origin. Method details Phosphoproteomic analysis Cell extracts from wild type Ocy454 cells, FAK KO Ocy454 cells and Ocy454 cells treated with FAK inhibitor (VS6063, at 10 μM for 30 minutes) were prepared in Urea Lysis Buffer (9M Urea, 20mM HEPES pH 8.0 + phosphatase inhibitor cocktail) and shipped to Cell Signaling Technology for analysis. Cells were sonicated and centrifuged to remove insoluble material. Protein content was determined by Bradford assay and equal protein quantities from all samples were used for analysis. Samples were reduced with DTT and alkylated with iodoacetamide. Samples were digested with trypsin (CST #56296), purified over C18 columns (Waters #WAT051910) and enriched using the PTMScan Phosphotyrosine pY-1000 Kit (CST #8803) as previously described.93 Samples were desalted over C18 tips prior to LC-MS/MS analysis. LC-MS/MS analysis was performed using a Thermo Orbitrap Fusion™ Lumos™ Tribrid™ mass spectrometer as previously described93,94 with replicate injections of each sample. Briefly, peptides were separated using a 50cm x 100μM PicoFrit capillary column packed with C18 reversed-phase resin and eluted with a 90-minute linear gradient of acetonitrile in 0.125% formic acid delivered at 280 nl/min. Tandem mass spectra were collected in a data-dependent manner using a 3 sec cycle time MS/MS method, a dynamic repeat count of one, and a repeat duration of 30 sec. Real time recalibration of mass error was performed using lock mass95 with a singly charged polysiloxane ion m/z = 371.101237. MS spectra were evaluated by Cell Signaling Technology using Comet and the GFY-Core platform (Harvard University).96,97,98 Searches were performed against the most recent update of the Uniprot Mus musculus database with a mass accuracy of +/-20 ppm for precursor ions and 0.02 Da for product ions. Cysteine carbamidomethylation was specified as a static modification, oxidation of methionine and serine, threonine, or tyrosine phosphorylation were allowed as variable modifications. Results were filtered to a 1% peptide-level FDR with mass accuracy +/-5ppm on precursor ions and presence of a phosphorylated residue for enriched samples. Site localization confidence was determined using AScore.99 All quantitative results were generated using Skyline100 to extract the integrated peak area of the corresponding peptide assignments. Accuracy of quantitative data was ensured by manual review in Skyline or in the ion chromatogram files. The relative abundance of each phosphorylated tyrosine residue was plotted in GraphPad prism. These results are shown in Table S1. Chemicals Chemical reagents VS6063 (Selleckchem, S7654), PF431396 (Selleckchem, S7644), PF562271 (Selleckchem, S2890), GSK2256098 (MedChemExpress HY-100498), Y15 (MedChemExpress HY-12444), cilengitide (APExBIO A8660), PF04957325 (MedChemExpress HY-15426) were used at doses and times indicated in figure legends. All compounds were dissolved in DMSO as 10 mM stocks and stored at -20°C. Forskolin (Sigma F6886) was diluted in water with 2% ethanol as per manufacturer’s instructions. PTH (1-34) was synthesized by the Massachusetts General Hospital (MGH) peptide core facility. GloSensor assay Cells stably expressing the GloSensor cAMP reporter were seeded into 96-well white plates (40,000 cells/100μL/well); 24 to 48 h after confluency, wells were rinsed with 100 μL carbon dioxide-independent culture medium (CIDB; Life Technologies) containing 0.1% bovine serum albumin (BSA), which was then replaced with 90 μL of luciferin (0.5 mM in CIDB) with FAK inhibitor or vehicle for 30 min before adding 10 μL of GPCR ligands (PTH 1-34, PGE2, Isoproterenol) or forskolin diluted in 10 μL of CIDB. cAMP-dependent luminescence was then measured at 2-min intervals using a PerkinElmer Envision 2104 plate reader (PerkinElmer), as described.28 Ligand dose–response curves were generated as described.28 CRISPR/Cas9-mediated gene deletion and lentiviral knockdown To design sgRNA target sequences, we used “Design sgRNA for CRISPRko” web tool (https://portals.broadinstitute.org/gpp/public/analysis-tools/sgrna-design) and selected the top 2-3 sgRNA sequences for each gene. Target sequences were cloned into the PX458 vector which was subsequently transfected into GloSensor HEK293 cells using PolyJet transfection reagent (SignaGen SL100688). GFP high cells were recovered by flow cytometry based sorting at a density of one cell/well into 96 well plates. Single cell clones were selected and screened for FAK deletion by immunoblotting. FAK-KO clones 1–3 were used for subsequent experiments. Single cell clones were selected and screened for PDE8A deletion by Sanger sequencing. DNA was isolated from the clones, PCR was performed using primers specific for the deletion site (Table S3), and subsequently DNA containing the PDE8A sequence was isolated via gel electrophoresis and DNA extraction for Sanger sequencing. One clone was identified to have an indel mutation that created a stop codon. In addition, this clone demonstrated lower PDE8A gene expression (Figure S4C). For shRNA experiments, HEK293 cells were transduced with lentiviruses harboring human PTK2 shRNAs that were cloned into the pLKO.1 backbone. To design the shRNA sequences we used https://portals.broadinstitute.org/gpp/public/gene/search and selected the top three sequences (clone IDs TRCN0000121318, TRCN0000196310, TRCN0000196667 – DNA sequences shown in Table S3). The 3rd sequence did not suppress FAK expression and was not used for subsequent experiments. To produce lentiviruses, HEK293T cells transfected with pLKO.1, psPAX2 (Addgene; plasmid #12260), and MD2.G (Addgene; plasmid #12259) using PolyJet transfection reagent (SignaGen SL100688). Medium was changed the next day and then collected 48 h later. Cells were exposed to lentiviral particles overnight in the presence of polybrene (2.5 μg/mL). Media was then changed with puromycin (4 μg/mL). Cells were maintained in selection medium throughout the duration of the experiment (1 week). Control cells were transduced with an empty shRNA- expressing lentivirus. Immunoblotting Whole-cell lysates for phosphorylated PKA substrate assessment were prepared using RIPA buffer (Boston BioProducts BP-115). Whole-cell lysates for immunoprecipitation assays were prepared using TNT (Tris-NaCl-Tween buffer, 20 mM Tris-HCl pH 8, 250 mM NaCl, 0.5% Triton X-100 containing protease inhibitor (PI), 1 mM NaF, 1 mM DTT, 1 mM vanadate). Adherent cells were washed with ice cold PBS, then scraped into RIPA or TNT buffer on ice. Material was then transferred into Eppendorf tubes kept on ice, vortexed at top speed for 30 s, then centrifuged at 14,000 g for 6 min at 4°C. The supernatants were collected, 2x SDS sample buffer was added and heated at 95°C for 5 min for immunoblotting. Lysates or immunoprecipitates were separated by SDS-PAGE, and proteins were transferred to the nitrocellulose (Amersham #10600008). Membranes were blocked with 5% milk in tris-buffered saline plus 0.05% Tween-20 (TBST) and incubated with primary antibody overnight at 4°C. The next day, membranes were washed, incubated with appropriate HRP-coupled secondary antibodies, and signals detected with ECL Western Blotting Substrate (Pierce), ECL Plus Western Blotting Substrate (Pierce), or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo scientific). Antibodies purchased from Cell Signalling Technology: phospho-PKA substrate (RRxS∗/T∗) (1:1000, Cell Signaling Technology, 9624S), FAK (1:1000, Cell Signaling Technology, 13009), p-FAK(Y397) (1:500, Cell Signaling Technology, 8556), beta-tubulin (1:250, Cell Signaling Technology, 5346), FLAG (1:1000, Cell Signaling Technology, 2368S), V5 (1:1000, Cell Signaling Technology, 13202S), phospho-tyrosine (1:1000, Cell Signaling Technology, 8954S), PDE8A (1:1000, Proteintech, 13956-1-AP). RNA isolation and RT-qPCR Total RNA was collected from cultured cells using QIAshredder (QIAGEN) and PureLink RNA mini kit (Invitrogen) following the manufacturer’s instructions. In brief, lysis buffer with 2-mercaptoethanol was added to cold PBS-washed cells and collected into QIAshredder, then centrifuged at 15,000 g for 3 min. The flow-through was collected into a new tube, and RNA isolation was carried out with PureLink RNA mini kit following the manufacturer’s instructions. For qRT-PCR, cDNA was prepared with 750 ng of RNA using the Primescript RT kit (Takara Inc.) and analyzed with PerfeCa® SYBR® Green FAstMix® ROX (Quanta bio) in the StepOnePlusTM Real-time PCR System (Applied Biosystems) using specific primers designed for each targeted gene (Table S3). Relative expression was calculated using the 2−ΔΔCT method101 by normalizing with β-actin housekeeping gene expression, and presented as relative abundance to β-actin. Plasmids pLV[Exp]-Puro-EF1A>{rGs-WT-FLAG}, pLV[Exp]-Puro-EF1A>{rGs-WT-FLAG-YFYF}, pRP-hP1R, pRP[Exp]-CMV>{hP1R-HA-Y494F,Y496F}, pRP-EF1a, pRP- EF1a> V5/hPDE8A and pRP EF1A> FLAG/hFAK were synthesized by VectorBuilder. PTH1R β-arrestin recruitment assay Recruitment of β-arrestin was assessed in GBR-24 cells, HEK293/GloSensor (GS22) cells stably expressing βArrestin2yfp 86,87 and transiently transfected to express PTH1R-WT or PTH1R-Y494F,Y496F.102 Cells were treated with red fluorescent PTH(1-34)tmr and imaged by fluorescent microscopy. PTH1R internalization assay Internalization was measured in GPG10 cells (HEK293/glosensor/PTH1R-GFPpH stable expression) as the change in the ratio of fluorescence derived from PTH1R-GFPpHs upon sequential excitation at wavelengths (λex) of 485 and 405 nm and fluorescence emission at a wavelength (λem) of 535 nm. Peptides were added to cells in black-walled 96-well plates and fluorescence was recorded at 1-minute plate intervals over the course of 90 minutes.102 Intracellular calcium flux assay Signaling via the intracellular Ca++ pathway was assessed in stably transfected HEK-HEK293/hPTH1R/glosensor (GP-2.3) cells using the calcium-sensitive fluorophore Fura2-AM (Invitrogen, Life Tech. Grand Island, NY, USA).102,103 Confluent cells in a black, 96-well plate were preloaded with Fura2-AM (5 μM) for 45 min and then unloaded in buffer for 30 min. The plate was then processed using a Perkin Elmer Envision plate reader (Perkin Elmer, Waltham, MA, USA) to monitor fluorescence emission at a wavelength (λem) of 515 nm, upon sequential excitation at wavelengths (λex) of 340 nm and 380 nm. Data were recorded in each single well at 2-s intervals for 10 s prior to, and for 150 s after ligand addition. The data at each time point were calculated as the ratio of the fluorescence signal obtained with excitation at 340 nm to that obtained with excitation at 380 nm. Co-immunoprecipitation (coIP) assays For the co-IP experiments in 293T cells, a total of 5 μg of the following plasmids were transfected using PolyJet transfection reagent (SignaGen SL100688): EF1A empty vector, EF1A-FLAG/hFAK, EF1A-V5/hPDE8A plasmid. Forty-eight hours later, protein lysates were collected with 1 mL of TNT (Tris-NaCl-Tween buffer, 20 mM Tris-HCl pH 8, 200 mM NaCl, 0.5% Triton X-100 containing protease inhibitors (PI), 1 mM NaF, 1 mM DTT, 1 mM vanadate). 10% of the protein lysate was saved as a “pre-IP” sample. The remaining protein lysate was incubated with 25 μl of anti-FLAG affinity gel (Selleckchem B23101), rotated at 4°C overnight after which it was washed twice with TNT buffer containing PI, NaF, DTT, and vanadate. Precipitated samples were eluted using FLAG peptide 1μg/μl in elution buffer (50mM Tris PH 7.5, 150mM NaCl) supplemented with protease inhibitors (PI), 1 mM NaF, 1 mM DTT, 1 mM vanadate. After centrifugation at 5000 rpm for 1 min, the supernatants were collected for western blot analysis (IP samples). In both pre-IP and IP samples 50 μL SDS sample buffer was added and heated at 95°C for 5 min. In separate experiments, protein lysates were incubated with V5 affinity gel (Millipore A7345) and eluted with V5 peptide (Sigma Aldrich V7754) and with similar remaining steps. PDE8A Activity assay PDE8A activity assay was performed with recombinant PDE8A and cAMP detection using the LANCE Ultra cAMP kit (Perkin Elmer). To determine the most sensitive range for the assay, the concentration of the cAMP standard supplied in the kit was titrated. The TR-FRET signal at 665 nm was normalized to the signal of the donor-channel at 620 nm to correct for well-to-well variability of the signal. A linear working cAMP range was obtained between 0.5 and 10 nM. Therefore, a cAMP concentration of 6 nM was used for subsequent assays. To optimize assay conditions, the effect of enzyme concentration on cAMP levels was first examined. Different concentrations (between 0.003 and 1 ng/μL) PDE8A (Sigma SRP0273) were tested in the presence of 6 nM cAMP in buffer containing 1X HBSS, 5 mM HEPES, 0.1% BSA, 3mM MgCl2, 50μM ATP. The reaction was stopped after 1 hour of incubation at RT by the addition of stop/detection buffer containing 1 mM IBMX and additional incubation for 1 hour at RT. The signal was read on the Envision plate reader 2104 at 620 and 665 nm. The assay was performed in white, opaque 384-well microplates. We noted submaximal but detectable PDE8A-stimulated cAMP hydrolysis when using PDE8A concentrations between 0.01-0.1 ng/μL for 1h. Consequently, 0.1 ng/μL PDE8A and an incubation time of one hour were selected for further experiments. Recombinant PDE8A and FAK (1 ng) proteins were added together in the reaction buffer and 665/620 ratios were measured after the above procedure. PDE profiling PDE profiling of VS6063 against 13 PDEs was performed by reaction biology VS6063 was provided at a concentration of 10mM in DMSO. VS6063 was tested in a 10-dose IC50, singlet, with 3-fold serial dilutions starting at 10 μM. Control compounds were tested in a 10-dose IC50, singlet, with 3-fold serial dilution starting at different concentrations. For PDE3A, PDE3B, PDE4A, PDE4B, PDE4C, PDE4D, PDE7A, PDE8A and PDE10A the reaction product, AMP and for PDE1A, PDE1B, PDE1C, and PDE2A the reaction product, GMP were detected by Transcreener fluorescent polarization assay. Substrate concentrations used: 1 μM cAMP for PDE3A, PDE3B, PDE4A, PDE4B, PDE4C, PDE4D, and PDE10A, 0.2 μM cAMP for PDE7A and PDE8A, 1 μM cGMP for PDE1A, PDE1B, PDE1C, and PDE2A. The reaction was performed at room temperature for 1hr. The assay buffer contained 10 mM Tris, pH 7.5, 5 mM MgCl2, 0.01% Brij 35, 1 mM DTT, and 1% DMSO and for PDE1A, PDE1B and PDE1C, 0.2 mM CaCl2 and 0.36 uM CaM were also added. Compounds in DMSO were added into enzyme using Acoustic Technology. Substrate was added and incubated for 1 hour. The reaction was stopped with the addition of Stop/detection mixture (40 mM Tris-HCL, pH 7.5, containing 80 mM EDTA and 0.04% Brij35). Fluorescent polarization was measured after 90 minute incubation at room temperature and calculated mP (Ex = 620 nm FP, Em = 688 nm P and S). Data was analyzed based on AMP or GMP standard curve to obtain product, AMP or GMP, amount. RNA sequencing Analysis Ocy454 RNA samples were measured with High Sensitivity RNA ScreenTape analysis (Agilent). High quality RNA samples with RIN >7 (n = 6) were processed for RNA library construction and sequencing using the Novoseq6000 platform (Novogene). Quality of raw sequencing data was confirmed with quality control using FastQC/0.11.5. Transcripts were quantified using Salmon.91 Salmon index for mouse was generated based on the Ensembl fasta file for GRCm38. Differential expression analysis was performed using DESeq2104 package based on the criteria of p adjusted value (padj) < 0.05 and log2|FC | ≥ 1. The volcano plots were made using GraphPad Prism 10. The overlap diagrams were made using DeepVenn https://www.deepvenn.com/ and Venni 2.1https://bioinfogp.cnb.csic.es/tools/venny/. Statistical significance of the overlap between lists of DEGs was assessed by hypergeometric test based on 11,600 expressed genes (genes with >10 normalized counts in all samples) using an online tool (nemates.org/MA/progs/overlap_stats.html). Analysis of gene ontology enrichment was performed using Enrichr (https://amp.pharm.mssm.edu/Enrichr/) on the Wikipathways 2024 mouse set. The combined EnrichR score based on Z scores and P-values was used in graphic representation. The RNA-Seq data were deposited at DNA Data Bank of Japan (DDBJ) under an accession number PRJDB37988. Quantification and statistical analysis All statistical analyses were performed by GraphPad Prism 10 for Mac (GraphPad Software Inc, USA). For GloSensor cAMP assays, area under the curve (AUC) was calculated for each biologic replicate within a given experiment and AUCs for each condition were compared. For comparison between two experimental groups, two-tailed t tests were used. When more than two groups were present, ANOVA with posthoc Tukey test was used. Values were expressed as mean ± SEM unless otherwise stated. A p-value < 0.05 was considered significant. Statistical details can be found in the figure legends. n represents number of biologic replicates. Footnotes Supplementary data related to this article can be found online at https://doi.org/10.1016/j.isci.2026.116049. Contributor Information Garyfallia Papaioannou, Email: gpapaioannou@mgh.harvard.edu. Marc N. Wein, Email: mnwein@mgh.harvard.edu. Supplemental information Document S1. Figures S1–S5 Table S1. (related to Results - FAK controls phosphorylation of proteins associated with cAMP signaling): Full data report from Phosphoproteomic assay Table S2. (related to Results FAK regulates cAMP signaling downstream of Gsα): VS6063 tested in vitro against a panel of recombinant phosphodiesterases Up to 10 μM VS6063 concentrations did not significantly inhibit PDE activity with the exception of a weak effect (IC50 3.8 μM) on PDE2A. Table S3. (related to Methods): List of nucleotide sequences of primers, CRISPR guides and shRNA sequences Table S4. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): List of differentially regulated genes of each group (compared to vehicle) and commonly regulated genes between groups Table S5. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): Comparison of all genes identified in the RNA Sequencing between control and PDE8AKO Ocy454 cells Table S6. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): Comparison of all genes identified in the RNA Sequencing between vehicle and PF431396 treated Ocy454 cells Table S7. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): Comparison of all genes identified in the RNA Sequencing between vehicle and PTH treated Ocy454 cells References Associated Data Supplementary Materials Document S1. Figures S1–S5 Table S1. (related to Results - FAK controls phosphorylation of proteins associated with cAMP signaling): Full data report from Phosphoproteomic assay Table S2. (related to Results FAK regulates cAMP signaling downstream of Gsα): VS6063 tested in vitro against a panel of recombinant phosphodiesterases Up to 10 μM VS6063 concentrations did not significantly inhibit PDE activity with the exception of a weak effect (IC50 3.8 μM) on PDE2A. Table S3. (related to Methods): List of nucleotide sequences of primers, CRISPR guides and shRNA sequences Table S4. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): List of differentially regulated genes of each group (compared to vehicle) and commonly regulated genes between groups Table S5. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): Comparison of all genes identified in the RNA Sequencing between control and PDE8AKO Ocy454 cells Table S6. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): Comparison of all genes identified in the RNA Sequencing between vehicle and PF431396 treated Ocy454 cells Table S7. (related to Results Phosphodiesterase 8A regulates cAMP and gene expression in osteocytes): Comparison of all genes identified in the RNA Sequencing between vehicle and PTH treated Ocy454 cells Data Availability Statement •Data: Primary RNA sequencing data are available in the databases: BioProject: PRJDB37988 and NCBI GEO: GSE326861. Data: Primary RNA sequencing data are available in the databases: BioProject: PRJDB37988 and NCBI GEO: GSE326861. •Code: This article does not report original code. Code: This article does not report original code. •Other: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Other: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.