Inhibition of phosphodiesterase 4D suppresses mTORC1 signaling and pancreatic cancer growth

The mammalian target of rapamycin complex 1 (mTORC1) senses multiple upstream stimuli to orchestrate anabolic and catabolic events that regulate cell growth and metabolism. Hyperactivation of mTORC1 signaling is observed in multiple human diseases; thus, pathways that suppress mTORC1 signaling may help to identify new therapeutic targets. Here, we report that phosphodiesterase 4D (PDE4D) promotes pancreatic cancer tumor growth by increasing mTORC1 signaling. GPCRs paired to Gαs proteins activate adenylyl cyclase, which in turn elevates levels of 3′,5′-cyclic adenosine monophosphate (cAMP), whereas PDEs catalyze the hydrolysis of cAMP to 5′-AMP. PDE4D forms a complex with mTORC1 and is required for mTORC1 lysosomal localization and activation. Inhibition of PDE4D and the elevation of cAMP levels block mTORC1 signaling via Raptor phosphorylation. Moreover, pancreatic cancer exhibits an upregulation of PDE4D expression, and high PDE4D levels predict the poor overall survival of patients with pancreatic cancer. Importantly, FDA-approved PDE4 inhibitors repress pancreatic cancer cell tumor growth in vivo by suppressing mTORC1 signaling. Our results identify PDE4D as an important activator of mTORC1 and suggest that targeting PDE4 with FDA-approved inhibitors may be beneficial for the treatment of human diseases with hyperactivated mTORC1 signaling.

Ras homolog enriched in brain (Rheb) and the Rag GTPases are crucial for mTORC1 lysosomal translocation and activation (13)(14)(15). In response to amino acids, the Rag GTPases recruit mTORC1 to the lysosome where it encounters Rheb. Growth factors signal through Rheb, and GTP-bound Rheb binds to and activates mTORC1 (16,17). Tuberous sclerosis complex (TSC) is a GTPase-activating protein (GAP) for Rheb, which converts GTP-bound Rheb (active form) to GDP-bound Rheb (inactive form) (18,19). The Rag GTPases form active heterodimers in which GTP-bound RagA or RagB is in complex with GDP-bound RagC or RagD at the lysosome (20). Amino acids (such as leucine, arginine, and methionine) recruit mTORC1 to the lysosomal surface through the Rag GTPase (21,22). Other components involved in the Rag GTPase signaling pathway include the Ragulator complex (23,24), vacuolar H + triphosphate (v-ATPase) (25), the GATOR The mammalian target of rapamycin complex 1 (mTORC1) senses multiple upstream stimuli to orchestrate anabolic and catabolic events that regulate cell growth and metabolism. Hyperactivation of mTORC1 signaling is observed in multiple human diseases; thus, pathways that suppress mTORC1 signaling may help to identify new therapeutic targets. Here, we report that phosphodiesterase 4D (PDE4D) promotes pancreatic cancer tumor growth by increasing mTORC1 signaling. GPCRs paired to Gα s proteins activate adenylyl cyclase, which in turn elevates levels of 3′,5′-cyclic adenosine monophosphate (cAMP), whereas PDEs catalyze the hydrolysis of cAMP to 5′-AMP. PDE4D forms a complex with mTORC1 and is required for mTORC1 lysosomal localization and activation. Inhibition of PDE4D and the elevation of cAMP levels block mTORC1 signaling via Raptor phosphorylation. Moreover, pancreatic cancer exhibits an upregulation of PDE4D expression, and high PDE4D levels predict the poor overall survival of patients with pancreatic cancer. Importantly, FDA-approved PDE4 inhibitors repress pancreatic cancer cell tumor growth in vivo by suppressing mTORC1 signaling. Our results identify PDE4D as an important activator of mTORC1 and suggest that targeting PDE4 with FDA-approved inhibitors may be beneficial for the treatment of human diseases with hyperactivated mTORC1 signaling.
The second messenger 3′,5′-cyclic adenosine monophosphate (cAMP) is produced by adenylyl cyclase after the activation of specific G protein-coupled receptors (GPCRs) (32). GPCRs are coupled to the G proteins Gα, Gβ, and Gγ. GPCRs coupled to Gα s subunits promote the cAMP signaling pathway. The GTP-bound Gα s subunit dissociates from Gβ and Gγ subunits in response to their corresponding ligands (33,34). cAMP signaling cascades control a wide range of physiological processes like activating protein kinase A (PKA) (35). PKA is a Ser/Thr kinase consisting of 2 regulatory subunits and 2 catalytic subunits. cAMP binds to the PKA regulatory subunits, resulting in the regulatory subunits dissociating away from the catalytic subunits and PKA activation. PKA phosphorylates a diverse range of substrates, including cAMP response element-binding protein (CREB), which contains the well-characterized PKA phosphorylation motif, Arg-Arg-X-Ser/Thr (RRXS/T) (X represents any amino acid) (36,37). A-kinase anchoring proteins (AKAPs) bind to regulatory subunits of PKA, localizing the PKA holoenzyme throughout the cell to maintain cell signaling cascades. Increased cAMP levels inhibit cell growth and proliferation through mTORC1 in multiple cancers (38)(39)(40). Thus, understanding the crosstalk between cAMP and mTORC1 is important in treating human diseases where mTORC1 is hyperactivated.
A previous study showed that resveratrol, a polyphenol in red wine, inhibits PDE4 (56). Moreover, resveratrol inhibits mTORC1 signaling. Thus, to better understand the mechanistic details by which mTORC1 is inhibited by GPCRs coupled to Gα s , we investigated the role of PDE4D in this pathway. We specifically focused on PDE4D because the depletion of PDE4A, PDE4B, and PDE4C did not alter mTORC1 activity. PDE4D controls cAMP levels, which regulate mTORC1 activity via Raptor Ser791 phosphorylation by PKA. PDE4D interacts with mTORC1 and regulates mTORC1 lysosomal localization. Importantly, the use of US Food and Drug Administration-approved (FDA-approved) PDE4D inhibitors restrains pancreatic cell tumor growth in vivo through mTORC1. Moreover, PDE4D levels play a key role in the overall survival of patients with pancreatic cancer.

Results
PDE4D interacts with mTORC1 and regulates its activity. Increased GPCR/Gα s signaling elevates cAMP levels, resulting in the potent inhibition of mTORC1 (57). Since PDEs regulate cAMP levels and there are many FDA-approved PDE inhibitors (45,58,59), we have been actively looking for the PDE involved in this pathway. A polyphenol in red wine called resveratrol was previously shown to inhibit PDE4 (56), and resveratrol inhibits mTORC1 in a dose-dependent manner in HEK293A cells ( Figure 1A). mTORC1 activity was analyzed by the phosphorylation of its substrates (S6K1 at Thr389, 4EBP1 at Thr37 and Thr46, and ULK1 at Ser758). S6K1 and 4EBP1 phosphorylation promote protein synthesis and ULK1 phosphorylation inhibits autophagy (5,60). PDE4 is encoded by 4 distinct genes, PDE4A-D, with more than 20 different isoforms in mammals (61). Interestingly, the depletion of PDE4D significantly decreased mTORC1 activity ( Figure 1B), and the overexpression of FLAG-tagged PDE4D4 rescued PDE4D-depleted mTORC1 inhibition ( Figure 1C). In addition, a catalytically inactive mutant, PDE4D4 D620A (Asp620 mutated to Ala620; the Asp620 residue is conserved in all PDE4 isoforms, refs. 62, 63), fails to rescue PDE4D-knockdown-mediated inhibition of mTORC1. In contrast to PDE4D, the depletion of PDE4A-C did not alter the activation of mTORC1 (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.158098DS1). Consistently, overexpression of FLAG-tagged PDE4D4 or FLAG-tagged PDE4D6 increased mTORC1 activity ( Figure 1D and Supplemental Figure 2A). PDE4D4 (809 amino acids) is the longest PDE4D isoform containing the UCR1/ UCR2 and catalytic domains, whereas PDE4D6 (518 amino acids) is a shorter isoform that contains a truncated UCR2 and catalytic domain (Supplemental Figure 2B). We previously found that GPCRs paired to Gα s proteins increase cAMP levels to inhibit the activity of mTORC1 (57). Specifically, cAMP levels activate PKA to phosphorylate Raptor on Ser791, resulting in mTORC1 inhibition. As expected, cells treated with forskolin, a pharmacological activator of adenyl cyclase, resulted in decreased mTORC1 activity ( Figure 1E). Overexpression of FLAG-tagged PDE4D4 relieved cAMP-induced mTORC1 inhibition. Moreover, FLAG-tagged PDE4D6 and PDE4D4 formed a complex with mTORC1 and PKA catalytic subunit α (PKA Catα) (Figure 1, F and G, and Supplemental Figure 2, C-E). Endogenous mTORC1 also interacted with PDE4D (Supplemental Figure 2E). We recently found out that A-kinase anchoring protein 13 (AKAP13) regulates mTORC1 activity by scaffolding PKA near mTORC1 (64). Interestingly, AKAP13 only associated with PDE4D4, containing the UCR1 domain at the N-terminus and not PDE4D6 ( Figure 1F and Supplemental Figure 2D). Thus, PDE4D interacts with and regulates mTORC1 activity.
PDE4D regulates mTORC1 activity through Raptor Ser791 phosphorylation. PKA is a cAMP-dependent protein kinase and phosphorylates Ser/Thr residues on substrates, preferably on the Arg-Arg-X-Ser/Thr (RRXS/T, where X tends to be a hydrophobic residue and Ser/Thr are the phosphorylatable residues) recognition motif (65). Raptor Ser791 is the only Ser/Thr residue that resides within the recognition motif. We previously demonstrated through site-directed mutagenesis and in vitro kinase assays that PKA directly phosphorylates Raptor at Ser791 (57). Because PDE4D controls cAMP levels and regulates mTORC1 activity ( Figure 1, B-E, and Supplemental Figure 2A), we investigated whether PDE4D inhibited mTORC1 through Raptor Ser791 phosphorylation. HA-tagged Raptor was coexpressed with either FLAG-tagged PDE4D4 or FLAG-tagged PDE4D6 in HEK293A cells that were treated with or without forskolin ( Figure  2, A and B). HA-tagged Raptor was immunoprecipitated and Raptor on Ser791 phosphorylation was analyzed using a phospho-PKA substrate antibody (pPKA sub [RRXS/T]). This antibody recognizes Raptor when it becomes phosphorylated on Ser791 (57). The expression of FLAG-tagged PDE4D4 or FLAGtagged PDE4D6 decreased Raptor Ser791 phosphorylation, consistent with elevated levels of PDE4D activating mTORC1 ( Figure 1D and Supplemental Figure 2A). However, overexpression of FLAG-tagged PDE4D4 D620A failed to decrease Raptor phosphorylation at Ser791 ( Figure 2C). In contrast, the depletion of PDE4D enhanced Raptor Ser791 phosphorylation after forskolin treatment ( Figure 2D). Because the endogenous PDE4D band runs more slowly (shifted up) in response to forskolin treatment ( Figure 2D) and the UCR1 domain of PDE4 contains a PKA phosphorylation site (47-50), we investigated whether PDE4D phosphorylation by PKA altered Raptor Ser791 phosphorylation. PDE4D has 2 predicted PKA substrate motifs, RRXS/T on Ser190 and KKXS/T (36, 66, 67) on Thr595. There was an increase in PDE4D phosphorylation in response to forskolin treatment as determined by phospho-PKA substrate antibody (Supplemental Figure 3A). A phospho-defective Ser190 (Ser190 to Ala190, S190A) mutation on PDE4D was resistant to RRXS/T phosphorylation in response to forskolin treatment (Supplemental Figure 3B). In addition, we made a phospho-defective Thr595 (Thr595 to Ala595, T595A) mutation on PDE4D4. Overexpression of FLAG-tagged PDE4D4 was still able to decrease Raptor Ser791 phosphorylation even when FLAG-tagged PDE4D was mutated (S190A and T595A) (Supplemental Figure 3C). The phosphorylation of PDE4D4 by PKA does not alter Raptor Ser791. Taken together, these results show that PDE4D regulates mTORC1 activity through Raptor Ser791 phosphorylation.
Rolipram (68) and roflumilast (69) are PDE4 inhibitors and approved by the FDA. Moreover, GEBR-7b has been shown to be a selective inhibitor for PDE4D (70). Importantly, rolipram, roflumilast, and GEBR-7b inhibited mTORC1 in a dose-dependent manner ( Figure 2E). A phospho-defective Ser791 (Ser791 to Ala791, S791A) on Raptor relieved mTORC1 inhibition by the PDE4D inhibitor (GEBR-7b) ( Figure 2F PDE4D regulates mTORC1 trafficking to the lysosome. mTORC1 translocates to the lysosomal surface in response to amino acids, where it is activated by the small GTPase Rheb (71). mTORC1 is not at the lysosome when cells are depleted of amino acids and is thought to be dispersed throughout the cell at an unknown Resveratrol inhibits mTORC1 activity. HEK293A cells were treated for 2 hours with resveratrol (0-100 μM). mTORC1 activity was analyzed by immunoblotting for pULK1 (Ser758), pS6K1 (Thr389), and p4EBP1 (Thr37/46). (B) Depletion of PDE4D inhibits mTORC1 activity. HEK293A cells were stably generated with 3 shRNAs targeting PDE4D. mTORC1 activity was analyzed as in A. (C) mTORC1 activity is rescued in PDE4D-depleted cells by reintroducing FLAG-tagged PDE4D. Wild-type or an inactive PDE4D mutant (D620A) was expressed in shPDE4D cells. mTORC1 activity was analyzed as in A. s.e., short exposure; l.e., long exposure. (D) Elevated PDE4D levels increase mTORC1 activity. FLAG-tagged PDE4D4 was overexpressed in cells for 48 hours. mTORC1 activity was analyzed as in A. (E) Increased PDE4D levels promote mTORC1 activity. FLAG-tagged PDE4D4 (0.5-2 μg) was transfected into HEK293A cells for 48 hours, and then stimulated with forskolin (10 μM) for 1 hour. mTORC1 activity was analyzed as in A. The quantification of pS6K1 was analyzed using ImageJ and normalized to S6K1. The data represent mean ± SD of triplicate experiments. NS, not significant. **P < 0.01, ****P < 0.0001 by 1-way ANOVA with Dunnett's test for multiple comparisons. (F) PDE4D binds to mTORC1 and AKAP13. Cells were transfected with FLAG-tagged PDE4D4 for 48 hours and then lysates immunoprecipitated (IP) with anti-FLAG antibody (PDE4D4). mTORC1 (mTOR, Raptor, mLST8), AKAP13, and FLAG (PDE4D4) were probed for. WCL, whole-cell lysates. (G) PDE4D interacts with PKA catalytic subunits. FLAG-tagged PDE4D4 was overexpressed in cells and cells were treated with forskolin (10 μM) for 1 hour. Lysates were immunoprecipitated with anti-FLAG antibody (PDE4D4). Raptor, PKA RIα, PKA RIIα, PKA Catα, pCREB (Ser133), CREB, FLAG (PDE4D4), and β-actin were probed for. Immunoblots probed for ULK1, S6K1, 4EBP1, β-actin, pCREB (PKA activation), CREB, PDE4D, and FLAG are controls.  location (9). We previously reported that increased intracellular cAMP levels can inhibit amino acid signaling to mTORC1 (57). However, altering intracellular cAMP levels did not impact mTORC1 lysosomal localization, but still blocked amino acid-induced mTORC1 activation through an unknown mechanism. Because PDE4D interacts with and regulates mTORC1 activity (Figure 1, B-G, and Supplemental Figure 2, C-E) and PDE4D depletion inhibits amino acid-induced mTORC1 activation ( Figure 3A), we investigated whether PDE4D played a role in mTORC1 lysosomal localization (Figure 3, B-D, and Supplemental Figure  4, A and B). As expected, amino acids promoted mTORC1 lysosomal localization in control cells, whereas mTORC1 was not present at the lysosome under amino acid starvation conditions. And as previously seen, increasing intracellular cAMP levels with forskolin did not alter mTORC1 lysosomal localization in response to amino acids. Interestingly, the depletion of PDE4D by knockdown or pharmacological inhibition significantly impaired mTORC1 lysosomal localization in response to amino acid stimulation with or without forskolin treatment. This result suggests that PDE4D somehow plays a role in mTORC1 lysosomal localization, possibly through the PDE4D-mTORC1 interaction. It has been reported that PDE4D is localized to cytosolic compartments in mouse embryonic fibroblasts (72). Consistent with this, PDE4D localized in the cytoplasm in HEK293A cells (Supplemental Figure 4C), and subcellular fractionation experiments revealed that PDE4D appeared to reside in the cytoplasm (not at the lysosome) regardless of cAMP stimulation ( Figure 3, E and F). Taken together, these results show that PDE4D promotes mTORC1 recruitment to lysosomes in response to amino acids by a mechanism independent of cAMP levels.
Pharmacologic PDE4D inhibition suppresses mTORC1 signaling and pancreatic cancer growth in vivo. mTORC1 modulates fundamental physiological processes like cell growth, and dysregulation of mTORC1 signaling can result in human cancer (5,10,73). Analysis of The Cancer Genome Atlas (TCGA) data indicates that PDE4D expression is high in patients with pancreatic adenocarcinoma ( Figure 4A). Moreover, data obtained from gene expression profiling interactive analysis (GEPIA) show that high PDE4D expression is associated with poor survival of patients with pancreatic adenocarcinoma ( Figure 4B) (74). We confirmed that PDE4D expression is higher in tissues of patients with pancreatic ductal adenocarcinoma (PDAC) compared with paired benign tissues ( Figure 4C and Supplemental Table 1). Moreover, the level of PDE4D correlates with mTORC1 activity in pancreatic cancer cells ( Figure 4D). Like HEK293A cells ( Figure 2E), the PDE4 inhibitors roflumilast and GEBR-7b inhibited mTORC1 in pancreatic cancer MIA PaCa-2 cells ( Figure 4E). Moreover, depletion of PDE4D inhibited mTORC1 signaling in MIA PaCa-2 cells ( Figure 4F). MIA PaCa-2 size and proliferation were significantly reduced, similar to that of mTORC1 signaling, when PDE4D was either depleted or pharmacologically inhibited (roflumilast or GEBR-7b) (Figure 4, G and H, and Supplemental Figure 5). Put-back experiments overexpressing FLAG-tagged PDE4D4 rescued mTORC1 activity and cell growth (cell size and proliferation) when PDE4D was depleted, whereas inactive FLAG-tagged PDE4D4 D620A was unable to rescue (Supplemental Figure 6, A-C). A phospho-defective S791 (Ser791 mutated to Ala791) in Raptor relieved mTORC1 inhibition and decreased cell growth (cell size and proliferation) in PDE4D-depleted MIA PaCa-2 cells (Supplemental Figure 7, A-C). PDE4D depletion or inhibition also inhibited colony formation (Figure 4, I and J). Interestingly, it appears that PDE4D regulates mTORC1 activity, cell size, and cell proliferation in KRAS-mutant (AsPC-1) compared with non-KRAS-mutant (BxPC-3) pancreatic cancer cells (Supplemental Figure 8, A-C). Collectively, these results demonstrate that PDE4D inhibition prevents pancreatic cell growth, cell proliferation, and colony formation.
Next we investigated the role of PDE4D in vivo. MIA PaCa-2 control or PDE4D-depleted cells were injected subcutaneously into mice. Tumor volume and weight were significantly decreased in PDE4D-depleted cells. Consistently, tumor volume and weight were lower in mice injected with PDE4D inhibitors (Figure 5, A and B). Tumors isolated from mice showed a reduction in mTORC1 signaling when PDE4D was depleted or inhibited ( Figure 5C). Immunohistochemical experiments also revealed a decrease in mTORC1 activity when PDE4D was knocked down or inhibited using an anti-pS6 antibody (measure of mTORC1 activity) ( Figure 5D). This anti-pS6 antibody has been validated in mice using rapamycin and other inhibitors upstream of mTORC1 (75)(76)(77). Taken together, these results show that the pharmacological inhibition of PDE4D results in a decreased mTORC1 signaling and tumor growth in vivo.

Discussion
The GPCR/Gα s signaling pathway decreases cell growth in a variety of tumor types, including skin cancer and medulloblastoma (78)(79)(80). PDEs degrade cAMP to negatively modulate GPCR/Gα s signaling (32,45). We previously reported that elevated intracellular cAMP levels inhibit mTORC1 by activating PKA to phosphorylate  activation. HEK293A cells stably expressing shGFP (control) or shRNA targeting PDE4D (shPDE4D) were generated. Cells were starved in amino acid-free media for 2 hours, pretreated with or without forskolin (10 μM) for 1 hour, and then stimulated with amino acids for 1 hour. mTORC1 activity with pS6K1 (Thr389) was analyzed. S6K1, pCREB (Ser133) (measure of PKA activation), CREB, and β-actin are controls. (B-D) Depletion of PDE4D blocks mTORC1 lysosomal localization. (B) HEK293A cell lines stably expressing shGFP (control) or shPDE4D were starved in amino acid-free media for 2 hours, pretreated with or without forskolin (10 μM) for 1 hour, and then stimulated with amino acids for 1 hour. Immunofluorescence experiments were performed with anti-mTOR (green) and -LAMP2 (lysosome marker, red) antibodies. Representative images were obtained under a Zeiss LSM 900 confocal microscope with 100× objective. (C) Staining intensity profiles across the 3.6-μm distance of the green (mTOR) and red (LAMP2) channels in the magnified pictures. (D) Ten immunofluorescence images per group were quantified using Squassh in ImageJ. The data represent mean ± SD. ****P < 0.0001 by 2-way ANOVA with Tukey's test for multiple comparisons. (E) PDE4D localizes in the cytoplasm. Cells were fractionated after forskolin (10 μM) treatment for 1 hour and then analyzed by immunoblotting for PDE4D. LAMP2 (lysosome marker), lamin A/C (nuclear marker), and α-tubulin (cytoplasm marker) are controls. Arrow indicates PDE4D. (F) PDE4D does not localize to the lysosome. HEK293A were starved in amino acid-free media for 2 hours, pretreated with or without forskolin (10 μM) for 1 hour, and then stimulated with amino acids for 1 hour. Immunofluorescence experiments were performed with anti-PDE4D (green) and -LAMP2 (lysosome marker, red) antibodies. Representative images were obtained under an LSM 900 confocal microscope with 100× objective. AA, amino acid. Raptor at Ser791 (57). Here, we show that PDE4D decreases Raptor Ser791 phosphorylation and promotes mTORC1 signaling. In contrast to PDE4D, other PDE4 family members (PDE4A, PDE4B, and PDE4C) did not alter mTORC1 activity in our studies. However, a previous study reported that PDE4A is involved in autophagy-related 5-and 7-(ATG5-and ATG7-dependent) autophagy induced by yessotoxin, a potential antiallergic and anticancer drug mediated by mTOR in erythroleukemia cells (81). Similarly, PDE4B was shown to modulate colorectal cancer growth through mTOR (82). Thus, it is possible that different PDE4 family members may play a role in cell proliferation and tumor development through mTORC1 in different types of cancers. Among the PDE family members, PDE4, PDE7, and PDE8 hydrolyze cAMP (41,42). PDE1, PDE2, PDE3, PDE10, and PDE11 can hydrolyze cAMP and cGMP. The potential role of additional PDEs that hydrolyze cAMP in mTORC1 regulation remains to be investigated.
In response to amino acids, mTORC1 is localized to the lysosome and activated via a Rag GTPase-dependent (23) or Rag GTPase-independent (31) pathway. We discovered that PDE4D is required for mTORC1 recruitment to the lysosome in the presence of amino acids. Because the elevation of cAMP by forskolin does not alter mTORC1 lysosomal localization, it is possible that PDE4D plays a role in mTORC1 lysosomal localization via the PDE4D-mTORC1 interaction. Perhaps PDE4D promotes the trafficking of mTORC1 to the lysosome. For example, PDE4D is predicted to interact with the lysosome trafficking protein Abelson murine leukemia viral oncogene homolog 1 (ABL1) by BioGRID (83), suggesting that PDE4D may have a role in the compartmentalization of mTORC1 throughout the cell.
GPCRs constitute the largest category of FDA-approved drugs on the market today (84). Similarly, PDE4 inhibitors, including roflumilast, cilomilast, and rolipram, have been approved by the US FDA for treatment of inflammation and chronic obstructive pulmonary disease (55). We believe that targeting GPCR/Gα s signaling in combination with using PDE inhibitors may be beneficial in human diseases with hyperactivated mTORC1.
Plasmids. Relevant PDE4D (PDE4D4 and PDE4D6) constructs encoding wild-type, S190A, S190A/ T595A, and D620A of PDE4D4 were generated by PCR and cloned into FLAG-pcDNA3.1 and lentiCRISPR v2 with a FLAG tag. The generation of FLAG-or HA-tagged Raptor wild-type and S791A constructs was described in previous work (57).
Immunoprecipitation and immunoblot analysis. Cells were lysed with CHAPS buffer (40 mM HEPES pH 7.5, 120 mM NaCl, 1 mM EDTA, 10 mM sodium pyrophosphate, 10 mM glycerol-2-phosphate, 50 mM NaF, 0.5 mM sodium orthovanadate, Complete EDTA-free protease tablet [Roche], and 0.3% CHAPS). The lysates were briefly vortexed and cleared by centrifugation at 13,000 rpm for 10 minutes at 4°C. The cleared supernatants were transferred to fresh tubes and were ready to use for immunoprecipitation with anti-HA magnetic beads (Thermo Fisher Scientific), anti-FLAG-M2 agarose beads (Sigma-Aldrich), and Protein A/G magnetic beads (Thermo Fisher Scientific). Equal quantities of protein extracts and immunoprecipitation products were electrophoresed in SDS-polyacrylamide gels and then transferred to polyvinylidene difluoride membranes (Bio-Rad). The membranes were blocked with 5% (w/v) skim milk in Tris buffer pH 7.4 containing 0.1% (v/v) Tween Subcellular fractionation. Cells were lysed with HNMEK lysis buffer (20 mM HEPES pH 7.4, 50 mM NaCl, 2 mM MgCl 2 , 2 mM EDTA, 10 mM KCl, 50 nM EGTA, protease inhibitors) using a Dounce homogenizer. The nuclei and cell debris were removed from lysates by centrifugation at 750g for 10 minutes. The pellets were collected for organelle fractionation by centrifugation at 12,500g for 10 minutes, and supernatants were used as the cytoplasmic fraction. The pellets were lysed with RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 5 mM EDTA pH 8.0, 1% [v/v] NP-40, 0.5% (w/v) deoxycholate, protease inhibitor).
Cell size. Cells (1 × 10 5 ) were plated in triplicate in 12-well plates and were trypsinized and resuspended with 1× PBS after 48 hours. Cell number and size were measured using a Z2 Coulter Particle Count and Size Analyzer (Beckman Coulter) and processed using Z2 Accucomp software. Final binned-histogram plots were created with GraphPad Prism 9.
Clonogenic assay of cells in vitro. Two hundred cells were seeded in 6-well plates and were incubated to form colonies for 14 days. The colonies were fixed with 20% (v/v) methanol and 0.5% (w/v) crystal violet (Sigma-Aldrich) and then were counted using ImageJ.
Immunohistochemistry. Mouse xenograft pancreatic tumor tissues were formalin-fixed and paraffin-embedded. Sections were sliced at 5 μm thickness and placed on slides, and then deparaffinized and hydrated. Citrate buffer (10 mM citric acid, 0.05% [v/v] Tween 20, pH 6.0) was used for antigen retrieval, and then the slides were incubated in 3% (v/v) hydrogen peroxide, followed by primary antibodies overnight at 4°C. The slides were incubated with biotinylated secondary antibodies and VECTASTAIN ABC reagents (Vector Laboratories). Antigen signals were visualized using DAB (Vector Laboratories), and the slides were counterstained with hematoxylin (Sigma-Aldrich). The images were obtained using an LSM 900 (Carl Zeiss) and portions of the positive-stained areas were semiquantified with ImageJ. Human pancreatic cancer and paired benign tissue slides were obtained from the UT Southwestern Tissue Management Shared Resource. Immunohistochemical analysis was performed on a Dako Autostainer Link 48 system. Briefly, the slides were baked for 20 minutes at 60°C, and then deparaffinized and hydrated before the antigen retrieval step. Heat-induced antigen retrieval was performed at pH 9.0 for 20 minutes in a Dako PT Link. The tissue was incubated with a peroxidase block and then with an antibody (1:200 dilution) for 20 minutes. The staining was visualized using the Envision FLEX visualization system. The percentage and staining intensity of PDE4D-positive PDAC tumor cells or normal pancreatic ducts were recorded, using a grading scale of absent (0), weak (1+), moderate (2+), or strong (3+). An H-score was calculated as described previously (86). Antibodies against pS6 (Cell Signaling Technology, 4858) and PDE4D (Proteintech, 12918-1-AP) were used.
Statistics. The survival plots and expression of pancreatic adenocarcinoma patients were obtained from GEPIA (74) and the log-rank test was used for statistical analysis. The data are presented as mean ± standard deviation (SD). For immunohistochemical H-scores, mouse tumor growth and weight are presented as mean ± standard error of the mean (SEM). Statistical analysis was performed using the 2-tailed Student's t test and 2-way ANOVA to compare 2 groups of independent experiments using GraphPad Prism 9. A P value of less than 0.05 was considered significant: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
Study approval. All mouse experiments were conducted according to approved guidelines at the Institutional Animal Care and Use Committee (IACUC) of UT Southwestern Medical Center.
Patients enrolled in the UT Southwestern study provided written consent, allowing the use of discarded surgical samples for research purposes according to the IRB-approved protocol. Human PDAC tissues were obtained from the UT Southwestern Tissue Repository under IRB-approved protocol 102010-051.