Phosphorylation of DEPDC5, a component of the GATOR1 complex, releases inhibition of mTORC1 and promotes tumor growth

Significance

The mTORC1 (mammalian target of rapamycin complex 1) pathway plays a critical role in driving cancer growth. We have identified a phosphorylation-dependent mechanism that controls mTORC1 activity in which Pim and AKT kinases, 2 enzymes with increased activity in cancer phosphorylate DEPDC5, a member of the GATOR1 complex that senses cellular amino acid levels. The critical nature of this substrate to the activity of these protein kinases is demonstrated by the fact that deletion or mutation of DEPDC5 partially blocks the ability of Pim and Pim plus AKT inhibitors to suppress tumor cell growth. Thus, protein kinases regulate the amino acid sensing cascade to control mTORC1 activity and tumor cell growth.

Abstract

The Pim and AKT serine/threonine protein kinases are implicated as drivers of cancer. Their regulation of tumor growth is closely tied to the ability of these enzymes to mainly stimulate protein synthesis by activating mTORC1 (mammalian target of rapamycin complex 1) signaling, although the exact mechanism is not completely understood. mTORC1 activity is normally suppressed by amino acid starvation through a cascade of multiple regulatory protein complexes, e.g., GATOR1, GATOR2, and KICSTOR, that reduce the activity of Rag GTPases. Bioinformatic analysis revealed that DEPDC5 (DEP domain containing protein 5), a component of GATOR1 complex, contains Pim and AKT protein kinase phosphorylation consensus sequences. DEPDC5 phosphorylation by Pim and AKT kinases was confirmed in cancer cells through the use of phospho-specific antibodies and transfection of phospho-inactive DEPDC5 mutants. Consistent with these findings, during amino acid starvation the elevated expression of Pim1 overcame the amino acid inhibitory protein cascade and activated mTORC1. In contrast, the knockout of DEPDC5 partially blocked the ability of small molecule inhibitors against Pim and AKT kinases both singly and in combination to suppress tumor growth and mTORC1 activity in vitro and in vivo. In animal experiments knocking in a glutamic acid (S1530E) in DEPDC5, a phospho mimic, in tumor cells induced a significant level of resistance to Pim and the combination of Pim and AKT inhibitors. Our results indicate a phosphorylation-dependent regulatory mechanism targeting DEPDC5 through which Pim1 and AKT act as upstream effectors of mTORC1 to facilitate proliferation and survival of cancer cells.

The Pim (proviral integration site for Moloney murine leukemia virus) serine/threonine protein kinases have been implicated as a driver of both triple negative breast cancer (TNBC) and advanced prostate cancer (1–3). Regulation of tumor growth by Pim has been closely tied with the ability of this kinase to stimulate protein synthesis by activating mTORC1 (mammalian target of rapamycin complex 1) signaling (4, 5), but the mechanism by which Pim regulates mTORC1 signaling is unknown. The interaction between Pim and AKT kinase pathways has been well established and plays an important role in tumorigenesis, as demonstrated by the observation that PI3K/AKT inhibition increases Pim kinase levels thus sustaining mTORC1 activity (6, 7). It has also been shown that tumor resistance to a PI3K/AKT inhibitor treatment in human breast cancer can be overcome by Pim inhibitor therapy (8), suggesting that Pim and AKT have an overlapping mechanisms of action.

Moreover, mTORC1 controls tumor cell growth, and its activity is often dysregulated in cancer. This enzyme integrates diverse inputs, including amino acids, growth factors, and stress signals to regulate protein synthesis, autophagy, and nutrient metabolism (9–11). mTORC1 activity is controlled by 2 major small GTPases, Rheb and Rag. Repression of mTORC1 by the depletion of specific amino acids is regulated by a cascade of protein complexes that function by modulating the activity of the Rag GTPases, RagA/B and RagC/D (10, 12, 13). The activity of the Rag GTPases is repressed by the GATOR1 complex proteins, DEPDC5, NPRL2, and NPRL3, based on GATOR1’s GTPase-activating protein (GAP) activity (14, 15). Since the Rheb GTPase activity is highly regulated in tumors by TSC2 phosphorylation (16, 17), we hypothesized that a similar mechanism could control the GATOR1 complex and thus Rag GTPase and mTORC1 activity.

Here we describe the regulation of GATOR1 complex by the phosphorylation of DEPDC5 mediated by the Pim and AKT protein kinases. Deleting DEPDC5 or mutating specific phosphorylation sites within the protein partially blocks the antitumor activity of small molecule inhibitors used clinically to inhibit Pim and AKT protein kinases (18, 19). These findings shed light on a phosphorylation-dependent regulatory mechanism targeting the Pim1/AKT-GATOR1-mTORC1 signaling cascade that is a driver of cancer cell proliferation.

Results

Pim Kinases Regulate the Amino Acid-Sensitive mTORC1 Pathway.

Upon amino acid starvation, the GATOR1 protein complex is recruited to the lysosome by the interaction of KICSTOR components after the dissociation of the GATOR2, and this complex induces Rag dimers to switch to an inactive conformation containing GDP-bound RagA/B, thereby inactivating mTORC1 (20). To study whether specific protein kinases might also play a role in controlling this pathway, we have chosen to use 3 different tumor cell types: 1) prostate cancer (PC3-LN4); 2) breast cancer (TNBC; MDA-MB 231 and BT549 and a non-TNBC; HCC1954), both tumor types have elevated Pim protein kinase and increased AKT activity (2, 3, 7, 21); and 3) the T-ALL (HSB-2), whose growth is driven by the Pim kinases without significant AKT input (22). Our results demonstrate that under conditions of leucine starvation of prostate cancer cells, PC3-LN4, Pim1 overexpression using a doxycycline (Dox)-inducible Pim1 vector sustains mTORC1 activation as measured by phosphorylation levels of p70S6 kinase on threonine 389 (P-S6K) and ribosomal S6 protein on serine 240/244 (P-S6) (Fig. 1A). This indicates that the Pim protein kinase could be involved in the regulation of mTORC1 under amino acid-restricted conditions. Using constitutive active forms of Rag B and D (Rag CA; Rag B 99L and Rag D 77L) GTPases (23) stably expressed in PC3-LN4 cells, mTORC1 activity, as measured by P-S6K and P-S6, is shown to be resistant to both a Pim kinase inhibitor, Pim447, and an AKT inhibitor, GSK690693 (Fig. 1B), suggesting the ability of Pim or AKT to regulate mTORC1 is upstream of the Rag GTPases. In contrast to P-S6K, the phosphorylation of IRS1 S1101, a substrate of Pim kinases (22), is inhibited by Pim447 treatment, indicating that Pim kinase activity is suppressed by this treatment. Similarly, phosphorylation of GSK3β and Foxo3a, substrates of AKT kinase, is inhibited by GSK690693 addition; these data suggest that the Pim and AKT kinases act as potential upstream regulators of Rag GTPase and play a role in modulating the signaling mechanism regulated by leucine levels.

Fig. 1.

Pim kinases regulate amino acid mediated mTORC1 activation. (A) Time course of leucine deprivation (Leu−) with and without Pim1 overexpression performed in PC3-LN4 cell line containing Dox-inducible Pim1 (Tripz Pim1) vector, treated with and without 100 ng/mL of Dox for 18 h. (B) PC3-LN4 control, overexpressing Rag WT or constitutively active Rag (CA) (see SI Appendix, SI Materials and Methods) cells were cultured with DMSO (C), 3 µM Pim447 (P), 5 µM GSK690693 (A), and the combination (AP) for 6 h. Cells were lysed and analyzed by Western blotting (WB). See SI Appendix, SI Materials and Methods for WB quantification. Numerical values shown under the blot are calculated relative to the DMSO (C) treatment.

Pim and AKT Kinases Phosphorylate DEPDC5 to Regulate mTORC1 Signaling.

Bioinformatic analysis demonstrates that the GATOR1 component DEPDC5 contains the Pim consensus phosphorylation site at S1002 RxRHx[S/T] and contains a potential AKT and Pim phosphorylation site RxRxx[S/T] at S1530 (15, 24, 25). When Pim1 is overexpressed in 293T cells, results using an antibody that recognizes the motif “RXRXXS*/T*,” demonstrate that Pim1 is capable of phosphorylating DEPDC5 (Fig. 2A). To validate these phosphorylation sites, we generated a phospho-specific antibody against S1002 and purchased a commercially available S1530 antibody. The specificity of the antibodies employed for these experiments was validated using lysates of 293T cells expressing wild type (WT) and site-directed DEPDC5 mutants containing S1002A and S1530A (Fig. 2 B and C). Both mutant and WT DEPDC5 were transfected into 293T cells with and without Pim inhibitor. Pim inhibition reduced the phosphorylation of both S1002 and S1530 sites, suggesting that both sites are Pim targets (Fig. 2D). To examine AKT activity, we cotransfected the plasmids encoding a constitutively active form of AKT, myristoylated (myr) AKT and DEPDC5 into 293T cells. Results demonstrate that AKT phosphorylates only the S1530 site on DEPDC5 and not S1002 (Fig. 2E).

Fig. 2.

Pim and AKT kinases phosphorylate DEPDC5. (A) Pim1 and Flag-tagged DEPDC5 (Flag DEPDC5) plasmids were cotransfected into 293T cells. Flag immunoprecipitated (IP) samples and whole cell lysates (WCL) were analyzed by WB using the indicated antibodies. (B and C) Plasmids expressing DEPDC5 WT, S1002A (SA1), and S1530A (SA2) were transfected into 293T cells which were treated with and without 3 µM Pim447 for 16 h and then cells were lysed. WCL and Flag-IP samples were analyzed by WB using the indicated antibodies. (D and E) Plasmids expressing Flag DEPDC5, Pim1, or myr-AKT were cotransfected into 293T cells and then treated with 3 µM Pim447 or the AKT inhibitor AZD5363, 3 µM for 18 h. The Flag-IP samples and WCL were analyzed by WB. (F) BT549 cells stably expressing EGFP as a control or DEPDC5 were incubated with DMSO (C), 3 µM Pim447 (P), 3 µM AZD5363 (A), and the combination (AP) for 6 h and cell lysates were analyzed by WB using the indicated antibodies. (G) BT549 cells stably expressing DEPDC5 were starved for 24 h and pretreated with DMSO (C), 3 µM Pim447 (P), 5 µM GSK690693 (A), and the combination (AP) for 1.5 h before stimulation with insulin (0.5 µg/mL) for 30 min, and cell lysates were analyzed by WB using the indicated antibodies. (H) BT549 cells stably expressing DEPDC5 were transfected with siRNA targeting Pim1 or all 3 Pim kinases, and after 48 h, cell lysates were analyzed by WB. See SI Appendix, SI Materials and Methods for WB quantification. Numerical values shown are calculated relative to the DMSO (C) treatment for F or no insulin treatment for G.

To examine whether endogenous Pim phosphorylates DEPDC5 in breast tumor cells, BT549 and MDA-MB231 TNBC cell lines stably expressing Flag-DEPDC5 were developed. These specific TNBC cell lines were chosen as they have been shown to express high levels of the Pim1 kinase (SI Appendix, Fig. S1A) (2, 3). Importantly, the combination of a Pim and an AKT inhibitor synergistically decreased the DEPDC5 phosphorylation at the S1530 site, whereas the S1002 site is specifically decreased by Pim inhibitor treatment (Fig. 2F). Both inhibitors were capable of inhibiting mTORC1, as evidenced by a decrease in P-S6, as well as specific targets IRS1 for Pim and GSK3β for AKT. Since both AKT and Pim are known to be activated by mitogens (26, 27), we examined whether a specific growth factor could activate DEPDC5 phosphorylation. The addition of insulin to serum-starved BT-549 cells stably expressing DEPDC5 activates both Pim and AKT activity, as shown by IRS1 and GSK3β phosphorylation, and stimulates a 1.8-fold increase in phosphorylation of DEPDC5 S1530 and a 1.5-fold increase in S1002 (Fig. 2G). To examine whether a specific Pim isoform drives phosphorylation of these sites, Pim1, Pim2, Pim3, and all 3 Pims were knocked down (KD) with siRNA in BT549 cells overexpressing DEPDC5. KD of Pim1 alone or all 3 Pim isoforms decreased the DEPDC5 phosphorylation (Fig. 2H), while in both BT549 and MDA-MB 231 KD of Pim2 or Pim3 had no effect on the DEPDC5 phosphorylation (SI Appendix, Fig. S1 B and C). To exclude off-target effects of siRNA, we performed rescue experiments using PC3-LN4 cells stably expressing DEPDC5 and Dox-inducible Pim1. Pim1 KD was capable of inhibiting mTORC1, as evidenced by a decrease in P-S6 and P-S6K, as well as decreasing the DEPDC5 phosphorylation and the known Pim1 substrate IRS1, although Pim1 KD was not effective in cells with Dox-induced Pim1 expression (SI Appendix, Fig. S1D). These results indicate that the mTORC1 suppression caused by Pim1 KD could be rescued by Pim1 overexpression. Thus, Pim1 and AKT regulate the phosphorylation of DEPDC5 as a potential control mechanism in modulating mTORC1 activity.

DEPDC5 Is Essential for the Pim Kinase-Mediated mTORC1 Regulation.

Knock out (KO) of DEPDC5 is purported to increase mTORC1 activity (25). However, we find that in fresh medium for 6 h the DEPDC5 KO effect is minimal compared to parental cells, while in nutrient-deprived media either secondary to amino acid depletion or prolonged culture for 72 h, the KO of DEPDC5 has a much more dramatic effect on mTORC1 activity (SI Appendix, Fig. S2 A and B). Importantly, based on our hypothesis that Pim phosphorylates DEPDC5 and regulates mTORC1 activity, KO of DEPDC5 in PC3-LN4, MDA-MB231, and HSB-2 cell lines develops resistance to genetic or pharmacological inhibition of Pim kinases, as shown by continued growth and mTORC1 signaling (Fig. 3 A–D and SI Appendix, Fig. S2A). Additionally, when MRKNU1, a breast cancer cell line that does not contain DEPDC5 (SI Appendix, Fig. S1A), is transduced with this cDNA, the cell line then becomes sensitive to Pim dependent inhibition of mTORC1 activity. Because inhibitors in this experiment are added in fresh media, major changes in P-S6 levels do not occur with DEPDC5 expression (Fig. 3E). Similarly, mTORC1 signaling was not suppressed when PC3-LN4 KO cells are complemented with DEPDC5 (Fig. 3A). These data demonstrate that the ability of Pim inhibition to block mTORC1 activity is dependent on DEPDC5 levels. DEPDC5 is necessary and sufficient to maintain Pim1-dependent mTORC1 activation, and Pim1 is capable of controlling the amino acid-sensing machinery by modifying the GATOR1 complex.

Fig. 3.

Pim kinase regulation of mTORC1 activity is dependent on DEPDC5. (A) PC3-LN4 CRISPR-control (CTR) and DEPDC5 KO cells transduced with or without Flag-DEPDC5 were cultured in fresh medium with DMSO (C) or 3 µM Pim447 (P) for 6 h. The cell lysates were analyzed by WB. (B) MDA-MB231 CTR and DEPDC5 KO cells were transfected with siRNA targeting Pim1 or all 3 Pims, and after 48 h, cells were lysed and analyzed by WB. (C) HSB-2 CTR and DEPDC5 KO cells were incubated with the indicated doses of Pim447 for 72 h. The percentage of viable cells was measured by an XTT assay. The growth of DMSO control cells was considered 100% and percent growth after individual treatments is reported relative to the DMSO. XTT data shown are the average ± SD of 3 independent experiments. (D) HSB-2 CTR and DEPDC5 KO cells were cultured in fresh medium with DMSO (C) or 3 µM Pim447 (P) for 8 h, and cell lysates were analyzed by WB. See SI Appendix, SI Materials and Methods for WB quantification. Numerical values shown are calculated relative to the Si negative control (siNC) treatment for B or control DMSO (C) for C. (E) MRKNU1 cells stably expressing DEPDC5 were incubated with DMSO or 3 µM Pim447 for 6 h and cells were lysed and analyzed by WB. WB data shown are representative of 3 or more independent experiments with similar results.

Inhibition of Pim and AKT Kinases Cooperatively Down-Regulates Cell Growth and the mTORC1 Pathway in a DEPDC5-Dependent Manner.

mTORC1 activity in both DEPDC5 KO prostate and breast tumor cells in vitro is resistant to Pim inhibitor (Pimi) and the combination of Pim and AKT inhibitor treatment (Fig. 4 A and B and SI Appendix, Fig. S3 A–C). In comparison, the doses of these inhibitors were sufficient to block the phosphorylation of other known targets of Pim and AKT: IRS1, GSK3β, TSC2, or Foxo3a, respectively. Additionally, as seen in both cell viability (XTT) and growth (crystal violet staining or IncuCyte real-time imaging) assays, the knockout of DEPDC5 also blocked the ability of Pim and AKT inhibitors to decrease the growth of multiple cancer cells lines, PC3-LN4, BT459, MDA-MB231, and HCC1954 (Fig. 4 C–F and SI Appendix, Fig. S3 D and E). As shown by cell colony formation (Fig. 4E) and mTORC1 activity (SI Appendix, Fig. S3F) expression of wild-type DEPDC5 in PC3-LN4 and MDA-MB231 KO cells restores the sensitivity to a Pim inhibitor both alone and in combination with an AKT inhibitor. Thus, these observations indicate that the ability of Pim and AKT kinase inhibitors to control cell growth is partially regulated by DEPDC5.

Fig. 4.

DEPDC5 deficiency contributes resistance to Pimi (i, inhibitor) and AKTi. (A) PC3-LN4 CTR and DEPDC5 KO cells were cultured in fresh medium with DMSO (C), 3 µM Pim447 (P), 5 µM GSK690693 (A), and the combination (AP) for 6 h, and cell extracts were subjected to WB. (B) MDA-MB231 CTR and DEPDC5 KO cells were cultured in fresh medium with DMSO (C), 3 µM Pim447 (P), 3 µM AZD5363 (A), and the combination (AP) for 6 h, and cell lysates were analyzed by WB. (C) PC3-LN4 CTR and DEPDC5 KO cells were cultured with DMSO, 3 µM AZD1208 (Pimi), 5 µM GSK690693 (AKTi), and the combination (AKTi+Pimi) for 72 h. The percentage of viable cells was measured by an XTT assay. (D) BT549 CTR and DEPDC5 KO cells were cultured with DMSO, 3 µM Pim447 (Pimi), 3 µM AZD5363 (AKTi), and the combination (AKTi+Pimi) for 72 h. The percentage of viable cells was quantified by an XTT assay. The growth of DMSO control cells was considered 100% and percent growth after individual treatments is reported relative to the DMSO. XTT data shown are reported as the average ± SD of 3 independent experiments. (E) PC3-LN4 (100 cells/well) CTR, DEPDC5 KO, and KO cells with DEPDC5 overexpression were seeded into 12-well plates and then incubated with either DMSO (C), 3 µM Pim447 (P), 5 µM GSK690693 (A), or the combination (AP) for 10 d; every 3 d fresh medium with drugs was added. Colony formation was visualized by crystal violet staining. (F) BT549 (4,000 cells/well) CTR and DEPDC5 KO cells were seeded into 12-well plates and then incubated with either DMSO (C), 3 µM Pim447 (P), 3 µM AZD5363 (A), or the combination (AP) for 6 d; culture medium with fresh drugs was replaced every 3 d. Cell growth was visualized by crystal violet staining and representative images are shown. A statistical comparison with Pim inhibitor versus the combination is shown and statistical significance is evaluated using a Student’s t test.

Mutations of the DEPDC5 Phosphorylation Site Alters Sensitivity to Pim and AKT Inhibitor Treatments In Vitro and In Vivo.

To test the ability of DEPDC5 phosphorylation to regulate mTORC1 signaling and tumor growth, PC3-LN4 WT, DEPDC5 KO, and CRISPR/Cas9 knockin (KI) of DEPDC5 S1530A (A-MUT, phospho inhibitory), and DEPDC5 S1530E (E-MUT, phospho mimic) tumor cells were treated with Pim and AKT inhibitors. DEPDC5 KO cells were completely resistant to these treatments (SI Appendix, Fig. S4 A and B). The S1530E-MUT cells were resistant to Pim and AKT inhibitor treatment with quite similar cell growth and activation of mTORC1 to DEPDC5 KO PC3-LN4 cells (Fig. 5 A and B). In terms of mTORC1 activity, the PC3-LN4 DEPDC5 S1002E cells were moderately resistant to both Pim inhibitor treatment and leucine deprivation (SI Appendix, Fig. S4C), suggesting possibly that phosphorylation at S1002 might affect DEPDC5 along with phosphorylation of S1530 or function independently to regulate other interactions of the DEPDC5 protein. Conversely, in PC3-LN4 expressing S1530 A-MUT mTORC1 signaling is significantly suppressed as detected by P-S6K and P-S6 (Fig. 5C), and growth of these tumor cells is ∼50% less both in vitro and in animal experiments when compared to the wild type (Fig. 5D and SI Appendix, Fig. S4D). However, these mutant cells were more sensitive to both the AKT inhibitor and the combination treatment. Thus, the phospho status of S1530 plays a role in regulating DEPDC5 and mTORC1 activity.

Fig. 5.

DEPDC5 phosphorylation by Pim and AKT kinases regulates mTORC1 activity both in vitro and in vivo. (A–C) PC3-LN4 WT, DEPDC5 KO, and PC3-LN4 cells with CRISPR knockin mutants DEPDC5 S1530A and S1530E were seeded into 12-well plates and incubated with DMSO (C), 3 µM Pim447 (P), 5 µM GSK-AKTi (A), and the combination (AP) for 6 d. Culture medium with fresh drugs was replaced every 3 d. Cell growth was visualized by crystal violet staining and representative images are shown. For WB, cells were cultured in fresh medium with drugs and treated for 6 h, and cell lysates were analyzed by WB using the indicated antibodies. (D) Male SCID mice were injected s.c. (5 × 10⁶ cells) with PC3-LN4 WT or S1530A MUT (SA2) cells. Mice were monitored for 3 wk and the percentage of tumor growth in S1530A mice (±SEM, n = 5) as compared to WT mice is plotted. (E) SCID mice were injected s.c. with 1 × 10⁶ PC3-LN4 CRISPR-CTR or DEPDC5 KO cells in groups of 5. After tumors reached 200 mm³, mice were treated with vehicle or the combination of Pim447 (30 mg/kg) and AZD5363 (40 mg/kg) daily by oral gavage for 2 wk. The percentage of tumor growth in the treated mice (±SEM) as compared to vehicle control is plotted. (F and G) SCID mice were injected s.c. with 1 × 10⁶ PC3-LN4 WT (n = 3) and CRISPR knockin DEPDC5 S1530E mutant (n = 5) cells. After tumors reached 200 mm³, mice were treated with vehicle or Pim447 (30 mg/kg), AZD5363 (40 mg/kg), and the combination of Pim447 and AZD5363 daily by oral gavage for about 4 wk. The average tumor volume ± SEM was plotted. A statistical comparison with vehicle versus single drug and single drug versus the combination treated tumors is shown and statistical significance is evaluated using a Student’s t test. N.S., not significant.

To clarify why the S1530A cells still responded to Pim and AKT inhibitors, PC3-LN4 WT and DEPDC5 KO cells were treated with the mTORC1 inhibitor, rapamycin (100 nM). Rapamycin inhibited cell growth in WT and KO cells by ∼60%, suggesting that inhibition of mTORC1 was not sufficient to totally abrogate cell growth. However, the addition of Pim and AKT inhibitors along with rapamycin further inhibited cell growth (90%), thus indicating that these kinase inhibitors target other growth pathways in addition to mTORC1 (SI Appendix, Fig. S4E), suggesting that these inhibitors are functioning similarly in DEPDC5 S1530A KI cells.

To test the response of growing tumors to dual kinase inhibition, PC3-LN4 WT, DEPDC5 KO, and S1530E-MUT KI cells were injected s.c. into male severe combined immunodeficient (SCID) mice. Once the tumors reached a specific size (200 mm³), mice were treated with Pim447 (30 mg/kg) (3, 28) or AKT inhibitor AZD5363 (40 mg/kg) (29, 30), the combination of drugs, or vehicle once daily by oral gavage without any significant change to the body weight (SI Appendix, Fig. S5 A and B). PC3-LN4 WT xenografts responded to the combination kinase inhibitor treatment with significantly decreased tumor growth (P < 0.05), while mice with DEPDC5-KO (Fig. 5E) were resistant to these agents, and DEPDC5 E-MUT xenografts were relatively resistant to the combined Pim and AKT inhibitor treatment (Fig. 5 F and G). These results demonstrate that DEPDC5 levels and the phosphorylation of DEPDC5 on S1530 both play an important role in the tumor growth inhibitory activity of these anticancer agents. These experiments demonstrate that Pim and AKT protein kinases regulate the activity of GATOR1. Thus, these protein kinases, which are overexpressed and/or activated in multiple cancers, enhance tumor growth in part by modulating this regulatory mechanism that controls mTORC1 activity.

Discussion

Regulation of mTORC1 plays a key role in controlling normal and tumorigenic cell growth. The Pim and AKT kinases (24, 31–33) have a dual role in regulating protein synthesis with both kinases phosphorylating and modifying the activity of various substrates, including eIF4B, TSC2, and PRAS40 (5, 17, 34), that control critical growth pathways. We demonstrate that by modifying DEPDC5, these protein kinases stimulate mTORC1 activity and control tumor growth. Experiments using overexpression of Pim1 and AKT protein kinase and a constitutively active Rag mutant demonstrate that the identical cascade of protein complexes normally regulated by amino acid availability that controls mTORC1 activity is modified by these protein kinases. DEPDC5 contains a Pim phosphorylation site which was validated by using phospho-specific antibodies. The Pim kinase can phosphorylate S1002 and S1530 while AKT phosphorylates S1530. The overlapping ability of these 2 protein kinases to phosphorylate DEPDC5 and the importance of this protein in controlling mTORC1 activity and tumor growth (35–37) may explain the necessity of inhibiting both Pim and AKT activity to block tumor cell growth (6, 24, 32, 38). Our results demonstrate that DEPDC5 is phosphorylated by Pim1 and not Pim2 or Pim3. However, KD of Pim1 alone using siRNAs while inhibiting DEPDC5 phosphorylation had only a modest effect on blocking cell growth, while inhibiting all Pim kinases with siRNAs mimicked the effects of the small molecule Pim inhibitor. The effect of Pim1 KD alone may be compromised by the presence of highly active AKT or the phosphorylation by Pim 2, 3 of other targets which control cell growth. This observation is a biochemical example of an isoform-specific substrate for the Pim kinase family of enzymes. Previous results from this laboratory demonstrated that all 3 isoforms can phosphorylate another Pim substrate IRS1 (39).

Our results demonstrate the central importance of the DEPDC5 protein in regulating tumor sensitivity to Pim and AKT inhibitors. KO of DEPDC5 in breast, prostate, and leukemic cells blocks the ability of Pim and AKT small molecule inhibitors to suppress tumor growth both in culture and in animal experiments. This result suggests DEPDC5 protein as an important control point in regulating the anticancer activity of these inhibitors. Because the S1530 site is modified by both Pim and AKT, its role in controlling the activity of DEPDC5 was further evaluated using PC3-LN4 prostate cancer cells. A KI of DEPDC5 changing S1530 to a glutamic acid (E) in PC3-LN4 cells made these cells partially resistant to growth inhibition and mTORC1 suppression in vitro and in vivo by the combination of AKT and Pim inhibitors. Thus, phosphorylation of this site plays a role in both mTORC1 regulation and cell growth. The mutation of S1002E had a less dramatic biologic effect, suggesting that this site alone may function in concert with S1530 to regulate GATOR1. Similar results have been obtained previously for eIF4B where S406 was phosphorylated by Pim and S422 by AKT. However, phosphorylation of both sites is required to control translation via the interaction with eIF3 (6, 40). Interestingly, we demonstrate that the addition of insulin will increase the phosphorylation of DEPDC5, giving a glimpse into the possibility that Rag activity, like Rheb, is controlled by multiple factors other than amino acids. Thus, regulation of DEPDC5 phosphorylation is critical for the antitumor activity of kinase inhibitors.

Rag GTPases interact with mTORC1, GATOR1 components, and the guanine nucleotide exchange factor (GEF), Ragulator. Rag GTPase interaction with mTORC1 is spatially regulated and the dissociation of Rag can attenuate mTORC1 activity. The inactive form of Rag GTPase can be released from mTORC1 and be reactivated by Ragulator in response to amino acids (41–44). At least 2 binding modes between Rag GTPases and the GATOR1 complex (25) are needed for mTORC1 to respond to amino acid withdrawal. A strong interaction between the RagA and DEPDC5 blocks the ability of GATOR1 to stimulate GTP hydrolysis, and a second weaker interaction with NPRL2 and RagA/C stimulates the GAP activity. In preliminary experiments, phosphorylation of DEPDC5 did not affect the ability of this protein to interact with either NPRL2 in the GATOR1 or SZT2 in the KICSTOR complex. The mechanism by which Rag binding to DEPDC5 leads to interaction with an arginine finger on NPRL2 (45) necessary for the GTPase activity is unknown. It is possible that DEPDC5 phosphorylation plays an important role in regulating GTP hydrolysis, and further experiments will be needed to understand how phosphorylation may control this process.

Together these experiments define a mechanism by which Pim1 and AKT kinases function as upstream regulators of mTORC1 through DEPDC5 phosphorylation. The results highlight an important rationale for the combination treatment for breast, prostate, and other cancers with Pim and AKT inhibitors and the importance of DEPDC5 levels playing a role in their activity.

Materials and Methods

All in vivo studies were approved by and conducted in accordance with the guidelines of the Institutional Animal Care and Use Committees at the University of Arizona Cancer Center. Detailed materials and methods, including cell culture conditions, cell viability (XTT) assay, cell growth assay, transient transfection and DNA plasmids, immunoprecipitation, Western blot analysis, mouse xenografts, CRISPR-Cas9 genome editing, information of antibodies and reagents, and statistics are available in SI Appendix, SI Materials and Methods.