Multisite phosphorylation of S6K1 directs a kinase phospho-code that determines substrate selection

SUMMARY

Multisite phosphorylation of kinases can induce on-off or graded regulation of catalytic activity; however, its influence on substrate specificity remains unclear. Here we show multisite phosphorylation of ribosomal protein S6 kinase 1 (S6K1) alters target selection. Agonist-inducible phosphorylation of glutamyl-prolyl tRNA synthetase (EPRS) by S6K1 in monocytes and adipocytes requires not only canonical phosphorylation at Thr³⁸⁹ by mTORC1, but also phosphorylation at Ser⁴²⁴ and Ser⁴²⁹ in the C-terminus by cyclin-dependent kinase 5 (Cdk5). S6K1 phosphorylation at these additional sites induces a conformational switch, and is essential for high-affinity binding and phosphorylation of EPRS, but not canonical S6K1 targets, e.g., ribosomal protein S6. Unbiased proteomic analysis identified additional targets phosphorylated by multisite phosphorylated S6K1 in insulin-stimulated adipocytes, namely, coenzyme A synthase, lipocalin 2, and cortactin. Thus, embedded within S6K1 is a target-selective kinase phospho-code that integrates signals from mTORC1 and Cdk5 to direct an insulin-stimulated, post-translational metabolon determining adipocyte lipid metabolism.

Graphical Abstract

Arif et al report that phosphorylation of the important metabolism-controlling kinase, S6K1, at two sites near the protein terminus induces its phosphorylation of multiple targets related to lipid metabolism. These insulin-stimulated phosphorylation events in adipocytes (fat cells) might contribute to the known influence of S6K1 on obesity.

INTRODUCTION

Protein phosphorylation is a widespread, post-translational event determining metabolic and immunologic responses to a host of stimuli including normal and pathological stresses (Pawson and Scott, 2005). The pervasive significance of phosphorylation in humans is evidenced by the identification of over 500 kinases that drive phosphorylation, 150 phosphatases that reverse phosphorylation, and 250,000 non-redundant phosphorylation sites in over 20,000 proteins (Alonso et al., 2004; Hornbeck et al., 2012; Khoury et al., 2011; Manning et al., 2002). Protein phosphorylation effectively increases proteome complexity to an extent far exceeding the diversity imparted by the genome (Walsh et al., 2005). Despite intensive investigation of the catalytic mechanisms and consequences of protein phosphorylation, the determinants that define kinase-substrate relationships, and direct target specificity are poorly understood (Miller and Turk, 2018; Ubersax and Ferrell, 2007). Critical determinants, with notable exceptions, include (1) consensus sequences on the target protein in close proximity to the phospho-acceptor site and recognized by the kinase catalytic domain, and (2) distal docking sites on the kinase and substrate that facilitate their transient interaction (Goldsmith et al., 2007; Holland and Cooper, 1999; Mok et al., 2010). Phosphorylation of a kinase, generally regulates its activity, usually inducing activation, but occasionally causing inactivation, and can direct cellular localization (Johnson and Lewis, 2001; Ubersax and Ferrell, 2007). Kinases are often modified by both auto- and trans-phosphorylation, and occasionally at multiple nearby sites (Beenstock et al., 2016; Cohen, 2000; Endicott et al., 2012).

We have investigated stimulus-dependent phosphorylation of glutamyl-prolyl tRNA synthetase (EPRS) in myeloid cells and in adipocytes (Arif et al., 2018). In mouse and human monocytes and macrophages, interferon (IFN)-γ induces phosphorylation of EPRS at Ser⁹⁹⁹ in the linker domain that joins the catalytic domains in this unique bifunctional tRNA synthetase (Arif et al., 2009). Phosphorylated EPRS is released from its usual residence in the multi-tRNA synthetase complex (MSC) to join with other proteins to form the GAIT (IFN-γ-activated inhibitor of translation) complex that directs translational silencing of select inflammation-related transcripts (Arif et al., 2018). In adipocytes, insulin induces EPRS phosphorylation at the same Ser residue, and upon release from the MSC, P-EPRS binds and transports fatty acid transport-1 (FATP1) to the plasma membrane for long-chain fatty acid (LCFA) uptake and enhanced triglyceride synthesis (Arif et al., 2017). The physiological significance of Ser⁹⁹⁹ phosphorylation was evidenced by low fat mass and extended life-span in homozygous mice carrying phospho-deficient Ser⁹⁹⁹-to-Ala mutations (Arif et al., 2017).

Investigation of the upstream events responsible for EPRS Ser⁹⁹⁹ phosphorylation in macrophages revealed cytosolic p70 ribosomal protein S6 kinase polypeptide 1 (RPS6KB1, also known as S6K1) as the proximal kinase. The evidence included: (i) failure to phosphorylate EPRS following siRNA-mediated knockdown of S6K1 and in bone marrow-derived macrophages from S6K1^−/− mice, (ii) in vitro phosphorylation of EPRS by recombinant S6K1 or by S6K1 immunoprecipitated from IFN-γ-activated monocytic cells, and (iii) stimulus-dependent interaction of EPRS with S6K1 (Arif et al., 2017). S6K1 is activated by signature phosphorylation at Thr³⁸⁹ by mammalian target of rapamycin complex 1 (mTORC1); the mTORC1-S6K1 kinase axis is central to anabolic pathways (Dibble and Cantley, 2015; Ma and Blenis, 2009; Ruvinsky et al., 2005; Saxton and Sabatini, 2017). Unexpectedly, IFN-γ-stimulated phosphorylation of EPRS at Ser⁹⁹⁹ required not only mTORC1 and S6K1, but also cyclin-dependent kinase 5 (Cdk5) and its obligate activator, Cdk5R1 (p35) (Arif et al., 2011; Arif et al., 2017). Metabolic or genetic interactions between Cdk5 and mTOR pathways have not been reported. Cdk5/p35 was once thought to be neuron-specific, but recent studies show its presence and activity in macrophages and adipocytes, and its dysregulation contributes to tumorigenesis and age-related pathologies (Dhavan and Tsai, 2001; Pozo and Bibb, 2016).

Here, we elucidate a previously unrecognized mechanism of kinase activation and target selection. We show that IFN-γ treatment of myeloid cells and insulin treatment of adipocytes induce Cdk5-dependent phosphorylation of two Ser residues in the carboxyl terminus domain (CTD) of S6K1, in addition to “classical” mTORC1-dependent phosphorylation of Thr³⁸⁹. Multisite phosphorylated S6K1 exhibits an altered conformation, and binds and phosphorylates EPRS. Global, proteomic analysis identified additional non-classical targets, namely, coenzyme A synthase (COASY), cortactin (CTTN) and lipocalin-2 (LCN2). Thus, S6K1 exhibits a target-selective phospho-code that controls a post-translational metabolon influencing adipocyte lipid metabolism.

RESULTS

Cdk5 is Required for S6K1-mediated Phosphorylation of EPRS but not Canonical Targets

We investigated the unexpected requirement of Cdk5, in addition to mTORCI and S6K1, for EPRS Ser⁹⁹⁹ phosphorylation (Arif et al., 2011; Arif et al., 2017). We considered the possibility that Cdk5 was specifically required for activation of mTORC1 or S6K1 in IFN-γ-activated monocytic cells. Cdk5 was inhibited by siRNA-mediated knockdown or by the inhibitor roscovitine in U937 cells. Following stimulation with IFN-γ, phosphorylation of mTORC1 and S6K1, and their downstream targets, was determined by immunoblot. Cdk5 inhibition did not suppress phosphorylation of mTOR at activation sites Ser²⁴⁴⁸ and Ser²⁴⁸¹, nor did it inhibit phosphorylation of the mTORC1 target and S6K1 activation site, Thr³⁸⁹ (Figure 1A and Supplementary Figure S1A). Moreover, phosphorylation of several S6K1 targets, i.e., ribosomal protein S6 (RPS6), eukaryotic translation initiation factor 4B (eIF4B), and eukaryotic elongation factor 2 kinase (eEF2K), were unaffected by Cdk5 inhibition. In contrast, Cdk5 knockdown or inhibition completely blocked EPRS phosphorylation at Ser⁹⁹⁹. Virtually identical results were observed in IFN-γ-stimulated human peripheral blood monocytes (PBM) and in insulin-stimulated 3T3-L1 adipocytes (Supplementary Figure S1B,C). To establish the specific role of S6K1 in the observed differential target phosphorylation, S6K1 was immunoprecipitated from IFN-γ-activated monocytic cells subjected to Cdk5 inhibition, and incubated with recombinant Ser”⁹-containing EPRS linker and RPS6. S6K1 isolated from IFN-γ-activated cells phosphorylated both EPRS linker and RPS6 as detected by [γ-³²P]ATP labeling; however, S6K1 isolated from Cdk5-inhibited cells phosphorylated RPS6, but not EPRS linker (Figure 1B). We investigated the role of Cdk5 in determining differential target specificity under multiple agonist-stimulated cell conditions. U937 cells, differentiated 3T3-L1 adipocytes, and HepG2 cells were treated with an array of mTORC1 agonists, and phosphorylation of EPRS and RPS6 was determined. In the same cell lysates, Cdk5 activity was measured by immunoprecipitation and in vitro phosphorylation of a Cdk5 substrate peptide. Only those agonists that induced Cdk5 activity, i.e., IFN-γ and lipopolysaccharide (LPS) in U937 cells, insulin and LPS in 3T3-L1 adipocytes, also induced EPRS Ser⁹⁹⁹ phosphorylation (Supplementary Figure S1C). In contrast, induction of S6 phosphorylation was seen in all cells with nearly all agonists. Together, these results show that activation of S6K1 by mTORC1 alone is not sufficient to phosphorylate EPRS in IFN-γ-stimulated myeloid cells or insulin-stimulated adipocytes, and suggest that an additional Cdk5-dependent mechanism of S6K1 activation is essential for differential phosphorylation of targets by S6K1.

Figure 1. Cdk5 suppression inhibits phosphorylation of EPRS Ser⁹⁹⁹, but not canonical S6K1 targets.

(A) Immunoblot analysis of S6K1 target phospho-sites in Cdk5-suppressed and control (cont.) human U937 monocytes. Cells were transfected with control or siRNA targeting Cdk5 and allowed to recover for 24 hr. Cells were serum-depleted for 2 hr and treated with interferon- γ (IFN-γ; 500 units/mL/10⁷ cells) for 8 or 24 hr. For pharmacologic inhibition, serum-depleted cells were treated with IFN-γ for 30 min and then with roscovitine (10 μM) for an additional 7.5 and 23.5 hr. Lysates were analyzed by immunoblot.

(B) Immunocomplex kinase analysis shows Cdk5 suppression inhibits phosphorylation of EPRS, but not RPS6. S6K1 was immunoprecipitated from Cdk5-suppressed, IFN-γ-treated U937 cells. The immunoprecipitates were incubated with purified recombinant, His-tagged Ser⁸⁸⁶-to-Ala (S886A) EPRS linker (i.e., bearing only Ser⁹⁹⁹ phospho-acceptor site) and RPS6 in presence of [γ-³²P]ATP, and phosphorylation determined by PAGE and autoradiography.

Cdk5-mediated Phosphorylation of S6K1 at Ser⁴²⁴ and Ser⁴²⁹ in the C-terminus is Required for Differential Target Phosphorylation

We considered the possibility that Cdk5 alters S6K1 target selection by phosphorylation at sites distinct from Thr³⁸⁹. Consistent with this mechanism, Cdk5 knockdown reduced IFN-γ-stimulated S6K1 phosphorylation as shown by ³²P-labeling of cells, followed by immunoprecipitation and autoradiography (Figure 2A). Cdk5, is a Pro-directed Ser/Thr kinase (Arif et al., 2011), and the major effect of knockdown was reduced Ser phosphorylation as detected by immunoblot with Ser-, Thr-, and Tyr-specific antibodies. S6K1 primarily exists in two isoforms, a nuclear p85S6K1 isoform (α1, bearing 23-amino acid N-terminus nuclear localization sequence), and a shorter, cytosolic p70S6K1 isoform (α2, termed here as S6K1) generated by utilization of an alternate ATG initiation site (Figure 2B) (Magnuson et al., 2012). S6K1 consists of an N-terminal domain (NTD) containing the TOR signaling (TOS) motif, kinase domain (KD), linker containing sites critical for catalytic activity including the signature mTORC1 phospho-site, Thr³⁸⁹, and C-terminal domain (CTD). The CTD contains a cluster of phospho-sites bearing the Ser/Thr-Pro motif, and is thought to contribute to S6K1 catalytic activity (Hou et al., 2007; Magnuson et al., 2012; Mukhopadhyay et al., 1992). However, S6 phosphorylation by S6K1 is only minimally influenced by removal of the CTD, or by phospho-defective or -mimetic mutations (Dennis et al., 1998; Ferrari et al., 1992; Weng et al., 1998). We therefore considered the possibility that Cdk5 phosphorylates CTD sites to re-direct substrate specificity of S6K1. Recombinant EPRS linker and RPS6 were subjected to in vitro phosphorylation by insect cell-derived, full-length S6K1 (including the upstream nuclear localization sequence, S6K1^−23–502) and S6K1 lacking the CTD (S6K1^−23–398). Deletion of the CTD from S6K1 specifically decreased phosphorylation of EPRS, but not RPS6 (Figure 2B). The ability of full-length, insect cell-derived S6K1 to phosphorylate EPRS linker is likely due to phosphorylation of sites in the CTD (Kozma et al., 1993), however, it was less effective towards EPRS substrate than endogenous S6K1 immunoprecipitated from IFN-γ-activated monocytic cells.

Figure 2. Cdk5-mediated Ser phosphorylation of S6K1 in the C-terminus directs EPRS phosphorylation.

(A) Cdk5 knockdown reduces S6K1 Ser phosphorylation. Cells transfected with siRNA targeting Cdk5 (or non-targeting control) were labeled with ³²P-orthophosphate in phosphate-free DMEM medium in absence or presence of IFN-γ for 4 hr. ³²P-incorporation in S6K1 was determined by immunoprecipitation with anti-S6K1 antibody and autoradiography. Unlabeled S6K1 immunopreciptates also were analyzed by immunoblot using Ser, Thr, and Tyr phospho-specific antibodies.

(B) Role of C-terminal domain in EPRS phosphorylation. (Top) Domain map and critical phospho-sites in p85 and p70 S6K1. Phospho-site numbering is based on p70 S6K1. Arrows indicate alternate ATG initiation site for generation of S6K1 isoforms. CTD, C-terminal domain; KD, kinase domain; NLS, nuclear localization signal; NTD, N-terminal domain. (Bottom) S6K1 immunoprecipitate from IFN-γ-treated U937 cells (S6K1-IP), recombinant, active full-length (rS6K1^FL, 1-502 aa plus NLS), and an S6K1 truncate lacking the CTD (rS6K1^1-398, 1-398 aa plus NLS) were used for in vitro phosphorylation of S886A EPRS linker and RPS6. Amount of each S6K1 form was added to give equivalent activity determined by in vitro phosphorylation of S6K1-specific peptide substrate KRRRLASLR. Aliquots were spotted on phosphocellulose P-81 filters, and ³²P incorporation determined by scintillation counting (mean ± SEM; n = 3 experiments).

We investigated whether Cdk5-mediated phosphorylation of specific CTD sites contributed to altered substrate selection. Cdk5-dependent phosphorylation of sites that conform to the Ser/Thr-Pro recognition motif was determined in lysates from IFN-γ-treated cells using available site-specific phospho-antibodies against S6K1. Cdk5 knockdown significantly reduced Ser⁴¹¹ and Ser⁴²⁴ phosphorylation, whereas phosphorylation of canonical S6K1 activation sites, i.e., Thr²²⁹, Ser³⁷¹, and Thr³⁸⁹ (as well as Ser⁴¹⁸ and Thr⁴²¹), were unaffected (Figure 3A and Supplementary Figure S2A). Motif-specific antibodies further validated Cdk5-dependent phosphorylation of Ser sites in S6K1 conforming to the Ser-Pro motif (Supplementary Figure S2B). Ser⁴²⁹ was considered a candidate Cdk5-dependent phosphorylation site based on sequence, but a phospho-antibody was not available. Thus, Cdk5-dependent phosphorylation of Ser⁴¹¹, Ser⁴²⁴, or Ser⁴²⁹, or a combination of these sites, might serve as a potential signal dictating altered S6K1 substrate specificity. To investigate specific phospho-site(s) required for EPRS phosphorylation, the three candidate Ser residues, alone and in combination, were mutated to Ala in Myc-tagged-pcDNA3 constructs containing S6K1 cDNA lacking the 3’UTR. U937 cells were co-transfected with mutant or wild-type S6K1 constructs, and with 3’UTR-specific siRNA to knock down endogenous S6K1, and then treated with IFN-γ. S6K1 constructs bearing Ala mutations at either Ser⁴²⁴ or Ser⁴²⁹ markedly reduced phosphorylation of EPRS at Ser⁹⁹⁹, but not RPS6 at Ser^235/236 (Figure 3B). S6K1 bearing a Thr³⁸⁹-to-Ala mutation did not phosphorylate either target, consistent with its essential role in mediating S6K1 activation (Dennis et al., 1998). To further investigate the requirement for combinatorial phosphorylation of both Ser⁴²⁴ and Ser⁴²⁹ sites, a gain-of-function experiment was performed. Phospho-mimetic Ser-to-Asp S6K1 mutants were constructed at the three candidate sites, alone and in combination, in the context of a phospho-mimetic mutation at Thr³⁸⁹ to constitutively activate the canonical mTORC1 target site, even in the absence of agonist. U937 cells were co-transfected with Myc-tagged wild-type and mutated S6K1 cDNA, and with 3’UTR-specific siRNA to knock down endogenous S6K1. Phosphorylation of EPRS Ser⁹⁹⁹ was observed only when S6K1 contained phospho-mimetic mutations at both Ser⁴²⁴ and Ser⁴²⁹; RPS6 phosphorylation at Ser^235/236 was independent of CTD phospho-mimetic mutations (Figure 3C). Essentially identical effects of S6K1 phospho-deficient and - mimetic mutations on differential target phosphorylation were observed in insulin-treated 3T3-L1 adipocytes. (Supplementary Figure S2C,D). To determine if Cdk5 directly phosphorylates Ser⁴²⁴ and Ser⁴²⁹, wild-type and mutated recombinant S6K1 were subjected to in vitro phosphorylation by Cdk5 immunoprecipitated from lysates from IFN-γ-treated U937 cells. In vitro phosphorylation of Cdk5-specific target peptide (PKTPKKAKKL) confirmed kinase activity of the immunocomplex (Figure 3D, right). Phosphorylation of S6K1 bearing single Ala mutants at Ser⁴²⁴ or Ser⁴²⁹ was decreased by about 50%, and by about 80-90% in double Ser⁴²⁴/Ser⁴²⁹ mutants (Figure 2D, left). In vitro phosphorylation of S6K1 mutants by recombinant Cdk5/p35 (i.e., Cdk5 activated by its binding partner, p35) gave similar results, establishing direct phosphorylation of S6K1 by Cdk5 at critical target specificity-determining sites. For brevity we designate the S6K1 proteoform phosphorylated at Thr³⁸⁹, Ser⁴²⁴, and Ser⁴²⁹ as S6K1*.

Figure 3. Ser⁴²⁴ and Ser⁴²⁹ phosphorylation in the S6K1 CTD, in addition to Thr³⁸⁹ phosphorylation, is required for EPRS Ser⁹⁹⁹ phosphorylation.

(A) Cdk5 knockdown inhibits phosphorylation of specific Ser sites in S6K1 CTD. Schematic of S6K1 phospho-sites including critical mTORC1 site Thr³⁸⁹ and putative Cdk5 targets in CTD (top). Phosphorylation in cell lysates from IFN-γ-treated control and Cdk5-knockdown U937 cells was analyzed by immunoblot using phospho-specific antibodies (bottom).

(B) Ectopic expression of phospho-defective Ser⁴²⁴- and Ser⁴²⁹-to-Ala S6K1 mutants blocks EPRS Ser⁹⁹⁹ phosphorylation. Endogenous S6K1 expression was suppressed by transfection of 3’UTR-specific siRNA in U937 cells. Cells were co-transfected with C-terminus Myc-tagged wild-type (WT) S6K1 ORF (1-502 minus 23-aa NLS), phospho-mimetic (Thr³⁸⁹-to-Glu, T389E) and phospho-defective Thr³⁸⁹-to-Ala (T389A) mutant S6K1, and single, double, and triple Ser-to-Ala point mutants as shown. Cells were treated with IFN-γ, and phosphorylation of EPRS and RPS6 analyzed by immunoblot.

(C) Ectopic co-expression of Ser⁴²⁴-to-Asp (S424D) and Ser⁴²⁹-to-Asp (S429D) phospho-mimetic S6K1 mutants phosphorylate EPRS Ser⁹⁹⁹ in absence of IFN-γ. U937 cells were transfected as in (B) but with phospho-mimetic, instead of phospho-defective mutants.

(D) Cdk5 phosphorylates Ser⁴²⁴ and Ser⁴²⁹ in the S6K1 CTD. Cdk5 was isolated by Immunoprecipitation of IFN-γ-treated U937 cell lysates with anti-Cdk5 antibody. Activity was determined by in vitro phosphorylation of Cdk5-specific peptide substrate, PKTPKKAKKL, in the presence of [γ-³²P]ATP (right). Recombinant wild-type and mutant S6K1 bearing N-terminus His- and C-terminus Myc-tag were generated by in vitro translation in wheat germ extract and purified by Ni-affinity chromatography utilizing the His-tag (Myc-S6K1). Immunoprecipitated Cdk5 was used for in vitro phosphorylation of Myc-S6K1 in the presence of [γ³²P]ATP. (left). Phosphorylation was quantitated densitometrically and expressed as percentage of γ³²-P incorporation compared to wild-type control. Purified Myc-S6K1 was detected by silver stain. Likewise, recombinant Cdk5 activated by p35 (Cdk5/p35) also was used to phosphorylate Myc-tagged S6K1 (bottom).

Cellular Functions of EPRS Phosphorylated by S6K1*

We investigated the cellular functions of EPRS phosphorylated by S6K1* by determining its ability to reconstitute known monocyte and adipocyte functions. We previously showed that IFN-γ-mediated phosphorylation of EPRS in monocytic cells induces binding to NS1-associated protein (NSAP1), phosphorylated ribosomal protein L13a (RPL13a), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH), to form the GAIT complex that inhibits translation of multiple inflammation-related transcripts (Arif et al., 2018; Ray and Fox, 2007; Sampath et al., 2004). Moreover, the phosphorylated linker domain of EPRS is sufficient for formation of a functional complex with the other GAIT constituents (Arif et al., 2009; Jia et al., 2008). Myc-tagged S6K1 constructs bearing phospho-mimetic mutations were transfected into U937 cells, and collected by immunoprecipitation with anti-Myc tag antibody. The immunocomplexed kinases exhibited the expected substrate specificity as shown by EPRS Ser⁹⁹⁹ phosphorylation by S6K1 bearing T389E mutation plus S424D and S429D mutations, and RPS6 Ser^235/236 phosphorylation by all S6K1 forms bearing a T389E mutation (Figure 4A). Purified EPRS linker, after in vitro phosphorylation by the S6K1 immunocomplexes, was incubated with other GAIT complex constituents, i.e., phospho-RPL13a, NSAP1, and GAPDH, in a wheat germ extract. GAIT element-bearing reporter mRNA, i.e., luciferase (luc) upstream of the ceruloplasmin (Cp) GAIT mRNA element, was subjected to in vitro translation in the extract. Translation was markedly inhibited by S6K1 forms bearing T389E, S424D, and S429D mutations; moderate inhibition was observed by S6K1 forms bearing the T389E mutation (Figure 4B).

Figure 4. Isolated S6K1 triple-phosphomimetic at Thr³⁸⁹, Ser⁴²⁴, and Ser⁴²⁹ (S6K1*) exhibits EPRS-selective specificity, and dictates EPRS-related cell functions.

(A) Immunocomplexed triple phospho-mimetic S6K1* bearing T389E, S424D and S429D mutations exhibits selectivity for EPRS Ser⁹⁹⁹ phosphorylation. Myc-tagged S6K1 bearing phospho-mimetic mutations was transfected into U937 cells, and immunoprecipitated with anti-Myc antibody. Immunoprecipitates were incubated with purified recombinant His-tagged S886A EPRS linker, or RPS6, in presence of ATP, and phosphorylation determined by immunoblot with phospho-specific antibodies.

(B) EPRS linker phosphorylated by S6K1* reconstitutes GAIT complex activity. His-tagged S886A EPRS linker was in vitro phosphorylated by immunocomplexed Myc-S6K1, purified, and pre-incubated with other GAIT complex constituents, i.e., phospho-RPL13a, NSAP1, and GAPDH. Reconstituted complexes were added to a wheat germ extract for in vitro translation of luciferase (Luc)-ceruloplasmin (Cp) GAIT element reporter transcript in the presence of [³⁵S]Met. T7 gene 10 RNA was co-translated as specificity control. Luc translation was quantified by densitometry, normalized by T7 gene 10, and expressed as percent of control (mean ± SEM, n = 3 experiments).

(C) Triple phospho-mimetic S6K1* activates EPRS linker to stimulate fatty acid uptake in adipocytes. siRNA targeting S6K1 3’UTR and Myc-tagged S6K1 ORFs were co-transfected into 3T3-L1 adipocytes. Non-transfected cells treated with insulin (100 nM) for 4 hr served as positive control. Lysates were analyzed by immunoblot. 5 × 10⁴ cells were incubated with bodipy-dodecanoate (C₁₂) for 4 h, and fatty acid uptake assayed by relative fluorescence (485 nm excitation and 515 nm emission) after 30-min. Uptake was expressed as fold-change relative to untreated, non-transfected control cells (mean ± SEM, n = 3 experiments in duplicate).

We previously reported that following insulin stimulation of adipocytes, EPRS phosphorylated by mTORC1/S6K1 binds FATP1 and transports it to the plasma membrane for enhanced fatty acid uptake (Arif et al., 2017). To investigate the specific phosphorylation requirement for this S6K1 activity, differentiated 3T3-L1 adipocytes were co-transfected with Myc-tagged wild-type and mutated S6K1 and with Flag-tagged EPRS linker. As in monocytic cells, Ser⁹⁹⁹ phosphorylation of EPRS linker was observed only when S6K1 contained phospho-mimetic mutations at both Ser⁴²⁴ and Ser⁴²⁹, but RPS6 phosphorylation only required the T389E mutation (Figure 4C). Transfection of S6K1 containing both S424D and S429D mutations at both Ser⁴²⁴ and Ser⁴²⁹, nearly doubled uptake of bodipy-C₁₂, a fluorescently-labeled LCFA reporter. In a positive control experiment, insulin treatment of adipocytes induced phosphorylation of both EPRS linker and RPS6, and also doubled fatty acid uptake.

Altered Binding Properties and Conformation of S6K1*

A reciprocal co-immunoprecipitation experiment showed that IFN-γ induces S6K1 binding to EPRS in monocytic cells (Arif et al., 2017). Following co-transfection of 3T3-L1 adipocytes with Flag-tagged EPRS linker and Myc-tagged S6K1 bearing phospho-mimetic mutations, an interaction between S6K1 and EPRS linker was observed only when S6K1 contained phospho-mimetic mutations at both Ser⁴²⁴ and Ser⁴²⁹ (Figure 5A); a similar result was seen in transfected U937 monocytes (Supplementary Figure S3A). To investigate whether this interaction is direct, binding of recombinant S6K1 forms with EPRS linker was determined by surface plasmon resonance (SPR) spectrometry. Recombinant, His-tagged wild-type S6K1 and S6K1 bearing phospho-mimetic mutations were generated by in vitro translation in wheat germ extract. Wild-type S6K1 was also phosphorylated by incubation with a lysate from insulin-treated adipocytes, repurified, and used as phosphorylated wild-type (P-WT) S6K1 control. EPRS linker was immobilized on a CM5 SPR sensor chip, and S6K1 proteins flowed over the chip for 5 min to permit binding, and then stopped to allow dissociation (Figure 5B). The association and dissociation curves were used to calculate the equilibrium dissociation constants, KD. Importantly, the binding of S6K1 mutant T389E/S424D/S429D (i.e., S6K1*) to EPRS linker exhibited a dissociation constant (56 nM) nearly two orders of magnitude lower than wild-type S6K1 (5.0 μM). The binding affinity of EPRS to S6K1* and to P-WT S6K1 control (40.0 nM) were similar, verifying that the putative phospho-mimetic mutations are highly mimetic and that other phosphorylation sites in S6K1 do not contribute substantially to binding. Finally, the T389E mutant did not exhibit high-affinity binding to EPRS indicating essential requirement for phosphorylation of the CTD sites.

Figure 5. S6K1* exhibits altered conformation compared to S6K1, facilitating high-affinity binding to EPRS for site-selective Ser⁹⁹⁹ phosphorylation.

(A) Triple phospho-mimetic S6K1* binds and phosphorylates EPRS linker in absence of insulin. 3’UTR-specific S6K1 siRNA, C-terminus Myc-tagged S6K1 ORFs, and Flag-tagged EPRS linker were co-transfected into 3T3-L1 adipocytes. Lysates were subjected to co-immunoprecipitation with anti-Myc antibody to probe Flag-EPRS linker/Myc-S6K1 interaction. EPRS linker Ser⁹⁹⁹ phosphorylation was determined by immunoblot.

(B) SPR analysis of binding of S6K1 forms to EPRS linker. Wild-type (WT) and mutant S6K1 proteins (1.25 nmol) were flowed over EPRS linker immobilized on a Biacore CM5 sensor chip. Binding was measured as relative response units. Smoothed association and dissociation curves determined by BIAevaluation software was used to calculate binding affinities expressed as dissociation constants, Kd.

(C) Susceptibility of S6K1 forms to limited proteolysis. To establish appropriate limited proteolysis condition, purified Myc-S6K1* (75 ng) bearing T389E, S424D, and S429D mutations was treated with prolyl endopeptidase for 20 min, and probed with anti-Myc antibody (right). Myc-S6K1 bearing phospho-defective and phospho-mimetic mutations were cleaved with prolyl endopeptidase (40 ng, 20 min), and probed with anti-Myc antibody. Myc-S6K1 mutants generated by in vitro translation in wheat germ extract, and purified by Ni-affinity, were detected by silver staining.

(D) Conformational alteration of S6K1* determined by antibody affinity. WT and S6K1 mutants (1.25 nmol) were flowed over anti-S6K1 antibody raised against human S6K1 C-terminus (amino acids 490-502) immobilized on a CM5 Biacore sensor chip. Binding was determined as in (B).

The specific high-affinity binding of EPRS to S6K1* suggests that the phosphorylated CTD sites are an essential part of the EPRS binding site, or that phosphorylation of the sites induces a conformational shift that exposes the binding site. To investigate the latter possibility, we subjected recombinant forms of C-terminus Myc-tagged S6K1 to limited proteolysis with prolyl endopeptidase after determining appropriate conditions for controlled cleavage (Figure 5C, right). Wild-type S6K1 was almost completely degraded by the treatment, however, forms containing both S424D and S429D mutations (in the context of T389E mutation) exhibited a unique, protected ~37 kDa fragment (Figure 5C, left). This result suggests that CTD phosphorylation alters protein conformation as evidenced by decreased susceptibility of the C-terminus half of S6K1* to degradation. In an independent assessment of S6K1 conformation we determined the binding affinity of an antibody specific for the C-terminal 13 amino acids of S6K1 (490-502). The binding results nearly paralleled those observed for binding of S6K1 forms to EPRS linker. The antibody binds with higher affinity to S6K1* or wild-type, endogenously phosphorylated S6K1 compared to wild-type, unmodified protein or the Thr³⁸⁹ phospho-mimetic form (Figure 5D). Together these experiments support a mechanism in which phosphorylation of S6K1 Ser⁴²⁴ and Ser⁴²⁹ induces a conformational shift that increases surface exposure of a region near the C-terminus that contains the EPRS linker docking site.

Additional Targets Uniquely Phosphorylated by S6K1*

A proteomic approach was taken to identify additional S6K1* targets based on the premise that these targets, like EPRS, will exhibit selective, high-affinity binding to S6K1*. HEK293T cells were transfected with Myc-tagged, wild-type or triple phospho-mimetic S6K1*, or pcDNA3 vector control (Figure 6A). Following pull-down with anti-Myc antibody and PAGE, the lanes were segregated into 8 regions subjected to elution, trypsinization, and LC-mass spectrometry. Peptides derived from 605, 999, and 846 proteins (combined total of 958 proteins) were detected in the control, S6K1 and S6K1* lanes, respectively. To select the protein sub-set that specifically binds S6K*, the data were filtered using a cut-off of at least 14 spectral counts in the S6K1* sample, and a S6K1*-to-S6K1 ratio of at least 4. The screen identified 10 candidate S6K1*-binding proteins including EPRS (Figure 6A, Supplementary Table 1). Binding of the candidates to S6K1 was investigated by immunoprecipitation with anti-S6K1 antibody in lysates from insulin-treated 3T3-L1 adipocytes; specific binding to S6K1* was supported by inhibition following knockdown with Cdk5 siRNA (and with S6K1 knockdown as control). Of the nine newly identified candidates, three were shown to bind S6K1 in a Cdk5-dependent way, namely, bifunctional coenzyme A (CoA) synthase (COASY), neutrophil gelatinase-associated lipocalin (LCN2), and Src substrate cortactin (CTTN) (Figure 6B). RPS6, a canonical S6K1 phosphorylation target, did not stably bind either S6K1 or S6K1*, as shown by comparison to vector control, consistent with the transient nature of typical interactions between kinases and their substrates (de Oliveira et al., 2016). Inability to validate several candidates, e.g., BPIFB1 and WDR77, is inconclusive as it might be due to low adipocyte expression, low antibody sensitivity, or both. Phosphorylation of the candidates was determined by ³²P-labeling followed by immunoprecipitation. EPRS, COASY, LCN2, and CTTN all exhibited, insulin-stimulated phosphorylation that was blocked by siRNA targeting S6K1 and Cdk5, consistent with S6K1* targets (Figure 6C). RPS6 was inducibly phosphorylated, but Cdk5-independence indicates S6K1* activity is not required. siRNA-mediated knockdown of the S6K1* targets in insulin-treated 3T3-L1 adipocytes reduced their expression and phosphorylation verifying the specificity of the antibodies (Supplementary Figure S4). Immunoprecipitation followed by immunoblot with phospho-specific antibodies revealed insulin induced phosphorylation primarily on Ser on all four S6K1* targets (Figure 6D).

Figure 6. Proteomic identification of S6K1* targets.

(A) Proteomic analysis identified candidate proteins that preferentially bind S6K1*. Myc-tagged wild-type S6K1, S6K1* bearing phospho-mimetic T389E, S424D, and S429D mutations, and pcDNA3 vector control (Cont.) were transfected in HEK293T cells, and lysates probed by immunoblot (left). Lysates were subjected to co-immunoprecipitation with anti-Myc antibody, and resolved on 4-20% Tris-MOPS-SDS-PAGE followed by imperial blue stain (center). Stained gels were divided into eight areas, excised, and subjected to in-gel tryptic digest. Digests were subjected to LC-MS/MS, and data searched using all high-energy, collision-induced dissociation (HCD) spectra against the human SwissProtKB database with Sequest and Mascot programs. Proteins that specifically bind S6K* with a cut-off of ≥14 spectral counts in the S6K1* lane, and a S6K1*/S6K1 ratio ≥5 (right). RPS6 is included to show lack of specific interaction with S6K1*. #, undefined (division by 0).

(B) Validation of Cdk5-dependent interaction of S6K1* with candidate targets. 3T3-L1 adipocytes transfected with siRNA targeting S6K1 or Cdk5 were treated with insulin for 4 hr. Lysates were probed by immunoblot before (right) and after (left) co-immunoprecipitation with anti-S6K1 antibody.

(C) Validation of Cdk5- and S6K1-dependent phosphorylation of candidate targets of S6K1*. 3T3-L1 adipocytes were labeled with ³²P-orthophosphate in phosphate-free DMEM medium ± insulin for 4 hr. ³²P-incorporation was determined by immunoprecipitation and autoradiography (left). Immunoprecipitated protein determined by immunoblot (right).

(D) Identification of Ser phosphorylation of S6K1* targets. 3T3-L1 adipocytes were immunoprecipitated with antibodies targeting COASY, CTTN, and LCN2, and probed with phospho-Ser/Thr/Tyr antibodies. EPRS Ser⁹⁹⁹ phosphorylation is shown as control.

In addition to determining stimulus- and kinase-dependent requirements, we directly assessed the requirement of multisite-phosphorylated S6K1* for target protein interaction and phosphorylation. 3T3-L1 adipocytes were transfected with Myc-tagged wild-type, triple phospho-mimetic (S6K1*), and T389E forms of S6K1, and binding to targets was determined by immunoprecipitation with anti-Myc antibody. EPRS, COASY, CTTN, and LCN2 all bound S6K1* but not wild-type or T389E forms of S6K1 (Figure 7A). Neither RPS6 nor isoleucyl tRNA synthetase (IARS) bound any S6K1 forms. Phosphorylation of all candidates by S6K1*, but not by wild-type or T389E S6K1, was shown by metabolic labeling with ³²P followed by immunoprecipitation and autoradiography (Figure 7B). Together, these experiments verify COASY, CTTN, and LCN2, in addition to the founding family member EPRS, as authentic S6K1* targets in insulin-stimulated adipocytes, possibly defining a coordinately regulated, post-translational metabolon influencing lipid synthesis (Figure 7C).

Figure 7. S6K1* binds and phosphorylates EPRS, COASY, CTTN, and LCN2.

(A) Ectopically expressed S6K1* binds targets in cells. cDNA encoding Myc-tagged wild-type, triple phospho-mimetic (S6K1*), and T389E S6K1 were transfected into 3T3-L1 adipocytes. Lysates were subjected to co-immunoprecipitation and analyzed by immunoblot. Detection with anti-IARS and -RPS6 antibodies served as negative controls.

(B) Ectopically expressed S6K1* phosphorylates targets in cells. S6K1 forms were transfected into 3T3-L1 adipocytes as above, and labeled with ³²P-orthophosphate in phosphate-free DMEM medium for 4 hr. Lysates were immunoprecipitated with target-specific antibodies and phosphorylation determined by autoradiography.

(C) Schematic of insulin-induced noncanonical activation of S6K1, the kinase phospho-code, and downstream influence on lipid metabolism.

DISCUSSION

Our studies establish an unprecedented molecular mechanism in which disparate upstream signaling cascades are integrated by combinatorial, multisite phosphorylation of S6K1, which in turn dictates substrate selection. Inducible phosphorylation of EPRS, CTTN, COASY, LCN2 by differentially phosphorylated S6K1 is evidence for a target-selective phospho-code embedded in the kinase. Canonical mTORC1 agonists phosphorylate S6K1 at Thr³⁸⁹, and together with phosphorylation events in the kinase and linker domains (i.e., Thr²²⁹ and Ser³⁷¹), activate S6K1 to induce phosphorylation of canonical targets that in turn drive global stimulation of translation (Figure 7C) (Ma and Blenis, 2009; Ruvinsky et al., 2005). In contrast, insulin stimulation of adipocytes and IFN-γ stimulation of myeloid cells, activate both the mTORC1 and Cdk5/p35 pathways, inducing phosphorylation of S6K1 at Thr³⁸⁹, and additional phosphorylation events at CTD sites Ser⁴²⁴ and Ser⁴²⁹. Multisite phosphorylated S6K1* phosphorylates canonical S6K1 targets, but also additional targets including COASY, CTTN, LCN2, and EPRS.

The mTORC1-S6K1 axis integrates intracellular and extracellular signals and controls nutrient-driven metabolic pathways with a well-established link to inflammation (Fontana et al., 2010; Hotamisligil and Erbay, 2008). Our results show an unusual contribution of Cdk5 in addition to mTORC1 in integrating inputs from IFN-γ and insulin for substrate-selective activation of S6K1*. Cdk5 is a ubiquitously expressed atypical member of the Cdk family of kinases with no known role in cell cycle progression. Cdk5 was considered to be a neuron-specific kinase due to abundant expression of its obligate activators, p35 and p39, in neurons (Dhavan and Tsai, 2001; Pozo and Bibb, 2016). However, context- and stimulus- dependent activation of Cdk5 has been detected in non-neuronal cells, including monocytes and adipocytes (Arif et al., 2011; Chen and Studzinski, 2001; Lalioti et al., 2009; Pareek et al., 2010). In adipocytes, insulin- and obesity-induced activation of Cdk5 is critical for glucose uptake and stimulation of a diabetogenic expression program (Banks et al., 2015; Lalioti et al., 2009; Li et al., 2011). Cdk5/p35 is also critical for insulin secretion by pancreatic islets in response to glucose challenge (Ubeda et al., 2004; Wei et al., 2005). Likewise, Cdk5 is linked to multiple inflammation-related responses such as differentiation of leukemic cells to monocytes, induction of T-cell-mediated inflammatory disorders, and anti-tumor immunity via regulation of the cell death ligand, PD-L1 (Chen et al., 2004; Dorand et al., 2016; Pareek et al., 2010). Our results revealing Cdk5- and mTORC1-dependent S6K1 activation link the kinase trio to the regulation of pro-inflammatory gene expression and metabolic activities in myeloid and fat cells, respectively.

Cdk5- and mTORC1-directed multisite phosphorylation drives activation and substrate selectivity of S6K1*. Multisite phosphorylation, defined as phosphorylation of two or more sites in a single polypeptide chain, can modulate or fine-tune protein activity, localization, stability, and interaction partners (Cohen, 2000; Holmberg et al., 2002). In the case of a single kinase inducing phosphorylation of multiple sites, then a simple input can be amplified into a complex response. As an example, temporally regulated multisite phosphorylation of ELK-1 by ERK directs rapid target activation and subsequent inactivation (Mylona et al., 2016). In contrast, if multiple sites are phosphorylated by distinct kinases, then integration of upstream signaling pathways is a key system feature (Ferrell, 1996; Nash et al., 2001; Park et al., 2006). For example, the tumor suppressor proteins p53 and retinoblastoma have multiple phospho-sites that combinatorially generate a range of responses by stimulus- and context-dependent interactions with multiple downstream effectors (Murray-Zmijewski et al., 2008; Rubin, 2013).

Multisite phosphorylation of multiple kinases, including S6K1, can control enzymatic activity via orthosteric or allosteric regulation (Cohen, 2000; Magnuson et al., 2012; Nussinov et al., 2012). However, multisite phosphorylation has not been reported to influence target selection, which is generally determined by the intrinsic property of the substrates (Kang et al., 2013; Miller and Turk, 2018). In response to diverse inputs, S6K1 can be phosphorylated at eight or more well-defined sites (Magnuson et al., 2012). In addition to the signature mTORC1-mediated phosphorylation of Thr³⁸⁹, phosphorylation of Thr²²⁹ and Ser³⁷¹ is considered essential for canonical S6K1 activation (Ma and Blenis, 2009; Moser et al., 1997; Pullen et al., 1998). Stepwise phosphorylation of multiple CTD sites in S6K1 by several proline-directed Ser/Thr kinases including Cdk5 has been suggested to contribute to S6K1 activation, possibly by relieving an inhibitory interaction between N- and C-termini (Ferrari et al., 1992; Hou et al., 2007; Lai et al., 2015; Mukhopadhyay et al., 1992; Papst et al., 1998; Sarker and Lee, 2004; Shah et al., 2003). However, the minimal effect of CTD truncation or site-specific CTD mutations on activation suggests a complex relationship between proline-directed Ser/Thr kinases and S6K1 (Cheatham et al., 1995; Dennis et al., 1998; Edelmann et al., 1996; Magnuson et al., 2012; Mahalingam and Templeton, 1996; Weng et al., 1998). Furthermore, stimulus-dependent activation of Cdk5 in a limited set of non-neuronal cells suggests a specialized role for proline-directed Ser/Thr kinases in mTORC1-S6K1 signaling, and possibly explains the cell-type specificity observed for EPRS phosphorylation (Supplementary Figure S1B)(Arif et al., 2017).

Our findings reveal a new function for multisite phosphorylation of a kinase, endowing S6K1 with a “barcode” that, depending on phosphorylation state of key sites, directs differential target recognition emblematic of kinase plasticity. Phospho-codes have been described for several key regulatory proteins, including, retinoblastoma protein, histones, and p53 tumor suppressor (Murray-Zmijewski et al., 2008; Rubin, 2013; Sims and Reinberg, 2008). Phospho-codes in kinases have been reported, but they generally determine binding partners and signaling pathways, not substrate selection and target post-translational modification as observed for S6K1* (Nussinov et al., 2012). Receptor tyrosine kinases exhibit phospho-codes that determine binding to adaptors without directly influencing target selection or phosphorylation (Olsson et al., 2006). Likewise, insulin-induced multisite phosphorylation of the kinase Akt2 generates distinct temporal signaling profiles but the mechanism, downstream effector, and output cellular response remain undefined (Humphrey et al., 2015). Importantly, the elucidation of a Cdk5- and mTORC1-directed phospho-code embedded in S6K1 establishes a foundation and incentive for discovery of analogous mechanisms in other kinases to specify differential, context-dependent target selection. Moreover, the S6K1 phospho-code provides a conceptual foundation for exploration of kinase inhibitors that inhibit phosphorylation of a selected cluster of targets.

S6K1*-directed EPRS Ser⁹⁹⁹ phosphorylation is defined not by consensus sequence recognition, but by high-affinity binding to a conformational variant of the kinase. An unbiased, proteomic approach identified additional candidate proteins that stably interact with S6K1*. As for EPRS, three additional proteins, COASY, CTTN, and LCN2, were experimentally validated to be phosphorylated by S6K1* in adipocytes by an insulin- and Cdk5-dependent mechanism. All of these proteins are previously unrecognized targets of S6K1, with the exception of COASY that was observed to bind S6K1, but was not phosphorylated by the kinase (Nemazanyy et al., 2004). Each S6K1* target exhibits activities actually or arguably related to lipid metabolism; however, the specific effects of phosphorylation on their activities are unknown. Particularly significant is the targeting of bifunctional enzyme COASY, which catalyzes the final two reactions in the 5-step pathway converting pantothenate to coenzyme A (Zhyvoloup et al., 2002). For triacylglycerol synthesis, LCFA imported into adipocytes require initial activation by coenzyme A, catalyzed primarily by long-chain coenzyme A synthetase 1 (ACSL1) (Lobo et al., 2009). Phosphorylation of COASY might increase the catalytic rate of one or both of its rate-limiting enzymatic activities, thereby increasing available coenzyme A (Zhyvoloup et al., 2002). Alternatively, phosphorylation could also induce translocation of COASY from its principal residence at the outer mitochondrial membrane to the plasma membrane for improved accessibility to ACSL1 (Zhyvoloup et al., 2003).

Interestingly, ACSL1 is complexed to FATP1, the major insulin-stimulated LCFA importer, which in turn is transported to the plasma membrane by another S6K1* target, phospho-EPRS (Arif et al., 2017; Gargiulo et al., 1999; Richards et al., 2006). These transport process might be facilitated by CTTN, an actin-binding protein required for insulin-stimulated translocation of vesicles containing the Glut4 glucose transporter to the plasma membrane (Nazari et al., 2011; Tunduguru et al., 2017). This mechanism is supported by evidence that Cdk5 is stimulated by insulin and plays a key role in Glut4-mediated glucose uptake and FATP1-mediated long-chain fatty acid uptake in adipocytes (Arif et al., 2017; Lalioti et al., 2009). Possibly, S6K1*-mediated phosphorylation of CTTN enables actin-mediated transport of the LCFA import machinery, i.e., CoASY or EPRS/FATP1, to the plasma membrane. Lastly, LCN2 is a secreted adipokine elevated during obesity and promotes insulin resistance (Yan et al., 2007). Exogenous application of LCN2 can stimulate both LCFA uptake and β-oxidation in adipocytes (Law et al., 2010; Paton et al., 2013). Possibly, insulin-stimulated phosphorylation of LCN2 represses its β-oxidation activity, thereby favoring anabolic pathways promoting triglyceride synthesis. These results suggest that S6K1* likely represents a critical control point of an insulin-stimulated, post-translational metabolon determining lipid metabolism in adipocytes. Future studies will delineate the specific lipid-related functions of the phospho-forms of these S6K1* target proteins, identify additional targets, and determine the respective roles of S6K1 and S6K1* in health and disease.

STAR★METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to, and will be fulfilled by, the Lead Contact, Paul L. Fox (foxp@ccf.org).

Plasmids and proteins

Recombinant, His-tagged, wild-type and Ser⁸⁸⁶-to-Ala (S886A) mutant linker proteins spanning Pro⁶⁸³ to Asn¹⁰²³ of human EPRS were expressed and purified as described (Arif et al., 2011; Arif et al., 2009). Recombinant, active, full-length S6K1 (rS6K1^FL, 1502 aa plus NLS) was purchased from R&D Systems and Cell Signaling (Kozma et al., 1993). Full-length human S6K1 cDNA in pCMV6-Entry vector was purchased from Origene and recloned, deleting the 23-aa N-terminus nuclear localization sequence (NLS) and adding an in-frame upstream 6-His tag and downstream C-terminus Myc-tag in pcDNA3. Specific mutations were introduced using appropriate mutation-bearing primers and GENEART Site-Directed Mutagenesis System (Invitrogen). Recombinant, wild-type and mutant NLS-deleted S6K1 with N-terminus His- and C-terminus Myc-tag were generated by in vitro translation using wheat germ extract system (Promega), and purified by Ni-affinity chromatography (Thermo Scientific Pierce).

Cell lines and cell culture

Human U937 monocytic cells (CRL 1593.2) were obtained from American Type Culture Collection (ATCC), and cultured in RPMI 1640 medium and 10% fetal bovine serum (FBS) with penicillin and streptomycin at 37 °C in 5% CO₂. Peripheral blood monocytes (PBM) from human blood were isolated by leukapheresis and elutriation under an Institutional Review Board-approved protocol adhering to the guidelines of American Association of Blood Banks. PBMs were cultured in RPMI 1640 medium and 10% FBS with penicillin and streptomycin at 37 °C in 5% CO₂. U937 cells and PBMs (10⁷ cells) were incubated with 500 U/ml IFN-γ (R&D Systems) for up to 24 hr as described (Arif et al., 2011; Arif et al., 2017). Human HepG2 hepatoma (HB-8065; ATCC) and mouse 3T3-L1 fibroblasts (CL-173; ATCC) were cultured in high-glucose containing Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies), 10% FBS, 4 mM L-glutamine, and 1× antibiotic/antimycotic solution (Life Technologies) at 37 °C in 5 and 10% CO₂, respectively. For differentiation into adipocytes, fibroblasts at 75% confluence were cultured in medium containing DMEM and 10% FBS supplemented with 1× solutions of insulin:dexamethasone:3-isobutyl-1-methylxanthine (Cayman) as described (Arif et al., 2017). After 72 h, the medium was replaced with 10% FBS and DMEM containing insulin and maintained for a week with three medium changes. Adipocytes were maintained in DMEM medium with 10% calf serum and 1× antibiotic/antimycotic for at least 3 days before use. Differentiated adipocytes were serum-deprived for 4 hr followed by treatment with 100 nM insulin (Sigma-Aldrich) for 4 hr. HEK293T cells (CRL-3216; ATCC) were cultured in the high-glucose DMEM medium at 10% FBS, 4 mM L-glutamine and 1× antibiotic/antimycotic at 37 °C in 5% CO₂. Cell lysates were prepared using Phosphosafe extraction buffer (Novagen) supplemented with protease and phosphatase inhibitor cocktails.

Transfection procedures

Cells were transfected with endotoxin-free plasmid DNAs or siRNAs (target-specific and scrambled control) using 100 μl of nucleofector solution V (U937 cells) or solution L (differentiated 3T3-L1 adipocytes) from Amaxa nucleofection kit (Lonza) following manufacturer’s protocol. Transfected cells were immediately transferred to pre-warmed Opti-MEM media for 6 hr and then to RPMI 1640 (for U937 cells) and DMEM (for 3T3-L1 adipocytes) containing 10% FBS supplemented with 4 mM L-glutamine, penicillin, streptomycin, and geneticin (G418; 20 μg/ml) for 18 to 24 hr before treatment with insulin and inhibitors.

In vitro kinase and phosphorylation assays

Purified active, recombinant kinases or immunocomplexed kinases were pre-incubated with EPRS S886A linker or RPS6 for 5 min in kinase assay buffer (50 mM Tris-HCl, pH 7.6, 1 mM dithiotheitol, 10 mM MgCl₂, 1 mM CaCl₂, and phosphatase inhibitor cocktail) (Arif et al., 2011; Arif et al., 2009). Phosphorylation was initiated by addition of 5 μCi [γ-³²P]ATP (Perkin-Elmer), and after 15 min was terminated using denaturing Laemmli gel-loading dye and heat denaturation. ³²P incorporation was determined by autoradiography after resolution on Tris-glycine SDS-PAGE and fixation in 40% methanol plus 10% acetic acid. For kinase activity assays, 50 μM of synthetic oligopeptide substrates were phosphorylated with 1 μCi [γ-³²P]ATP in kinase assay buffer. Equal volumes were spotted onto P81-phosphocellulose squares, washed in 0.5% H₃PO₄, and ³²P incorporation determined by scintillation counting.

Immunocomplex kinase assay

Pre-cleared cell lysates were incubated with specific antibodies for 4 hr in kinase assay buffer, and immunocomplexes captured by incubating with protein A-Sepharose beads for an additional 4 hr. (Arif et al., 2011; Arif et al., 2009). Immunocomplexes were washed three times with kinase assay buffer supplemented with 0.1% Triton X-100. Finally, the immunocomplexes were resuspended in kinase assay buffer and used to phosphorylate EPRS S886 linker, RPS6, or peptide substrate. ³²P incorporation into substrates was determined by SDS-PAGE coupled with autoradiography and scintillation counting (Arif et al., 2009; Arif et al., 2017).

Immunoblot analysis

Ccell lysates or immunoprecipitates were denatured in Laemmli sample buffer (Bio-Rad) and resolved on Tris-glycine SDS-PAGE (10, 12, or 15% polyacrylamide) prepared using 37.5:1 acrylamide:bis-acrylamide stock solution (National Diagnostics), or on precast Express Plus Tris-MOPS-SDS PAGE (8, 10, 12, 4-12%, 4-20%; GenScript). After transfer to polyvinyl difluoride membranes, proteins were probed with target-specific antibody, followed by incubation with horseradish peroxidase-conjugated secondary antibody (GE Healthcare), and detected with ECL prime western blotting detection reagent (GE Healthcare). Immunoblots shown are typical of experiments independently done at least three times.

Phosphorylation assay by ³²P-metabolic labeling

24 hr post-transfection with siRNAs, U937 and 3T3-L1 cells were incubated with IFN-g and insulin, respectively, in presence of 250 μCi of ³²P-orthophosphate (MP Biomedical) in phosphate-free RPMI 1640 medium for 4 hr. ³²P-labeled cells were lysed in Phosphosafe extraction buffer (Novagen) supplemented with protease and phosphatase inhibitor cocktails, and subjected to immunoprecipitation. Target ³²P-labeled proteins were immunoprecipitated by overnight incubation of pre-cleared lysates (1 mg) with specific antibodies in buffer containing 50 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1% Triton X-100, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease/phosphatase inhibitor cocktails. Immunoprecipitates were captured with Protein A-Sepharose beads (Sigma) for an additional 4 hr. Immunoprecipitated beads were washed three times in detergent-containing buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, and 0.1% Triton X-100) followed by three washes in detergent-free buffer. Finally, immunoprecipitated proteins were suspended in denaturing Laemmli gel-loading dye, resolved on SDS-PAGE, and phosphorylation determined by autoradiography.

In vitro reconstitution of GAIT complex and translation assays

Capped, poly(A)-tailed luciferase (Luc) reporter upstream of the ceruloplasmin (Cp) 3’UTR GAIT element (Luc-Cp GAIT) and T7 gene 10 reporter transcripts were prepared using mMessage mMachine SP6 and T7 kits (Ambion), respectively (Jia et al., 2008). For in vitro reconstitution of GAIT complex, His-tagged S886A EPRS linker (permitting site-specific Ser⁹⁹⁹ phosphorylation) was phosphorylated by immunocomplexed Myc-S6K1 (wild-type and mutant) in presence of ATP (10 μM), and re-purified using Ni-NTA purification kit (Thermo-Scientific). Purified NSAP1, phospho-L13a, and GAPDH were generated as described (Jia et al., 2008). The GAIT complex was reconstituted by incubating 5 pmol each of purified EPRS linker, NSAP1, GAPDH, and phospho-L13a. The reconstituted complex was incubated with Luc-Cp-GAIT (200 ng) and T7 gene 10 (200 ng) template transcripts, and added to wheat germ extract system (Promega) containing Met-free amino acid mixture and ³⁵S-Met (Perkin-Elmer). Translation of Luc-Cp-GAIT and T7 gene 10 transcripts was visualized by resolution on 10% SDS-PAGE and autoradiography.

Fatty acid uptake assay

Fatty acid uptake was determined using the long-chain fatty acid analog, bodipy-dodecanoic acid and a QBT fatty acid uptake assay (Molecular Devices) (Arif et al., 2017; Liao et al., 2005). Untransfected and Myc-tagged S6K1 transfected 3T3-L1 adipocytes (5 × 10⁴ cells per well in a 96-well plate) were incubated in serum-free Hanks balanced salt solution for 4 h, and then with the fatty acid analog for an additional 4 hr. As a positive control, untransfected adipocytes were treated with the fatty acid analog in the presence of 100 nM insulin. After 30 min, the relative fluorescence was determined at 485 nm excitation and 515 nm emission wavelengths in bottom-read mode (SpectraMax GeminiEM, Molecular Devices).

Limited proteolysis assay

Purified, recombinant, Myc-tagged wild-type and mutant S6K1 (75 ng) were subjected to proteolysis by prolyl endopeptidase (40 to 150 ng; Abcam) for 20 min at 30 °C in 50 mM Tris HCl (pH 7.5) buffer. The reaction was terminated using non-denaturing Laemmli gel-loading dye supplemented with 0.1 M EDTA. Proteolyzed S6K1 was analyzed by SDS PAGE followed by immunoblot with anti-Myc antibody.

Protein-protein interaction analysis by co-immunoprecipitation

Cell lysates were pre-cleared by incubation with protein A-Sepharose beads (Sigma) for 1 hr. Pre-cleared lysates (1 mg) were incubated for 12-16 hr with antibody conjugated to protein A-Sepharose beads in detergent-free buffer containing 50 mM Tris-HCl (pH 7.6), 150 mM NaCl, and EDTA-free protease/phosphatase inhibitor cocktail. Immunoprecipitated beads were washed three times in the same buffer followed by elution with 0.2 M glycine-HCl (pH 2.6), immediately neutralized with 50 mM Tris-HCl (pH 8.5), and suspended in denaturing Laemmli gel-loading dye. Co-immunoprecipitation was also performed by directly incubating the pre-cleared cell lysates (1 mg) with antibody overnight in detergent-free buffer, followed by a 4 hr incubation with protein A-Sepharose beads. The immunoprecipitates were washed as above, suspended in denaturing Laemmli gel-loading dye, and analyzed by Tris-glycine SDS-PAGE and immunoblot.

Protein-protein interaction analysis by SPR

Recombinant N-terminus, His-tagged human EPRS linker protein was purified using Ni-NTA purification kit (Thermo-Scientific). Subsequently, the His-tag was removed using enterokinase cleavage capture kit (Novagen), and re-purified by Ni-NTA column. N-terminus 6-His tagged wild-type and mutant S6K1 were generated by in vitro translation in wheat germ extract and purified by Ni-NTA column. Phosphorylated S6K1 was generated by in vitro phosphorylation of purified S6K1 by lysates from insulin-treated 3T3-L1 adipocytes in presence of ATP (10 μM) as above, followed by re-purification by Ni-NTA chromatography. Purified EPRS linker or anti-S6K1 antibody were immobilized on Biacore sensor chip CM5 using amine-coupling procedures per manufacturer’s instructions. Purified wild-type and mutant His-tagged S6K1 proteins were prepared in HBS-EP (0.01 M HEPES, pH 7.4, 0.15 M NaCl, 3 mM EDTA and 0.005% v/v surfactant P20) buffer (Biacore). Purified S6K1 proteins (1.25 nmol) were flowed over the sensor chip at 10 μl/min for 5 min. After a 2-min dissociation period, the bound proteins were removed with regeneration buffer containing 1.5 M NaCl and 10 mM glycine-HCl, pH 3.0 (30 μl/min, 30 s). Finally the sensor chip was equilibrated by washing with HBS-EP buffer (2.5 min). Kinetics of S6K1 binding (dissociation constant, Kd) to EPRS linker and S6K1 antibody was calculated using association and dissociation curves generated by Biaevaluation software (Biacore).

Proteomic analysis of S6K1*-interacting proteins

HEK293T cells (3 × 10⁵ cells) in triplicate were transfected with 1 μg endotoxin-free DNA encoding Myc-tagged ORFs of S6K1, S6K1* with phospho-mimetic T389E, S424D, and S429D mutations, and pcDNA3 vector, using Fugene 6 transfection reagent (Promega) following manufacturer’s instruction. 36 hr post-transfection, the replicates were pooled and lysed in 50 mM Tris HCl, pH 7.6 by four freeze-thaw cycles, followed by passage several times through a 26-gauge needle, and centrifugation at 5,000 rpm for 5 min at 4 °C. Supernatant (1.5 mg protein) was subjected to co-immunoprecipitation by incubation with anti-Myc tag Sepharose-conjugated mouse mAb (Cell Signaling) for 16 hr. The precipitate was washed 4 times with 50 mM Tris HCl, pH 7.6, 150 mM NaCl containing protease and phosphatase inhibitor cocktail, and denatured in Laemmli sample buffer (Bio-Rad). The samples were resolved by 4-20% GenScript Express Plus Tris-MOPS-SDS PAGE, and protein detected by Imperial blue stain (ThermoFisher). Each lane was divided into eight regions, excised, washed/destained in 50% ethanol, 5% acetic acid, and dehydrated in acetonitrile. Excised regions were reduced with 1,4-dithiotreitol and alkylated with iodoacetamide prior to overnight in-gel digestion by trypsin (5 μl of 10 ng/μl in 50 mM NH₄HCO₃). Peptides were extracted using 30 μl 50% acetonitrile with 5% formic acid, solvent was evaporated, and peptides resuspended in 1% acetic acid for liquid chromatography-mass spectrometry (LC-MS) analysis using a Fusion Lumos mass spectrometer system (ThermoScientific). Extracts were injected into a reversed-phase capillary chromatography HPLC column (Dionex Acclaim Pepmap 100 C18 LC), and peptides eluted by acetonitrile/0.1% formic acid gradient at a flow rate of 0.3 μl/min with detection by a 2.5 kV microelectrospray ion source. Data-dependent multitask capability of the instrument was used to acquire full-scan mass spectra and generate high-energy, collision-induced dissociation (HCD) spectra to determine peptide molecular weights and amino acid sequences. The HCD spectra were searched against the human SwissProtKB database with Mascot and Sequest programs and peptide and protein validation was performed using the program Scaffold (Proteome Software, Portland, OR) to determine the specific S6K1* interactome.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data obtained from in vitro phosphorylation of peptide substrate, autoradiography, and in vitro translation were analyzed using Student’s t test (GraphPad Prism 5) and results shown as mean ± SEM.

DATA AVAILABILITY

Comprehensive data from the proteomic analysis of S6K1*-interacting proteins is included as a single Excel spreadsheet.

Supplementary Material

3

Table S1. Proteomic analysis of S6K1*-interacting proteins, Related to Figure 6.

(A) Summary of results obtained from each lane divided into eight regions. (B) Summary of results by lane type, i.e., control, S6K1 and S6K1*. (C) Comparison of proteins in control, S6K1, and S6K1* lanes.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-phospho-S6K1-Ser418	Abcam	Cat# ab58534; RRID:AB_883110
Anti-phospho-S6K1-Thr229	Abcam	Cat# ab5231; RRID:AB_304788
Anti-phospho-S6K1-Ser411	Novus Biologicals	Cat# NBP1-60941 RRID:AB_11006602
Anti-phospho-S6K1-Ser424	Novus Biologicals	Cat# NBP1-60942; RRID:AB_11019606
Anti-phospho-S6K1-Thr421	Novus Biologicals	Cat# NB100-92608; RRID:AB_1217585
Anti-S6K1	Cell Signaling Technology	Cat# 2708; RRID:AB_390722
Anti-phospho-S6K1-Ser371	Cell Signaling Technology	Cat# 9208; RRID:AB_330990
Anti-phospho-S6K1-Thr421/Ser424	Cell Signaling Technology	Cat# 9204; RRID:AB_2265913
Anti-RPS6	Cell Signaling Technology	Cat# 2217 RRID:AB_331355
Anti-phospho-RPS6-Ser235/236	Cell Signaling Technology	Cat# 2211 RRID:AB_331679
Anti-eIF4B	Cell Signaling Technology	Cat# 3592 RRID:AB_2293388
Anti-phospho-eIF4B-Ser422	Cell Signaling Technology	Cat# 3591 RRID:AB_2097522
Anti-eEF2K	Cell Signaling Technology	Cat# 3692 RRID:AB_2231040
Anti-phospho-eEF2K-Ser366	Cell Signaling Technology	Cat# 3691 RRID:AB_2097313
Anti-mTOR	Cell Signaling Technology	Cat# 2972 RRID:AB_330978
Anti-phospho-mTOR-Ser2448	Cell Signaling Technology	Cat# 2971 RRID:AB_330970
Anti-phospho-mTOR-Ser2481	Cell Signaling Technology	Cat# 2974 RRID:AB_2231885
Anti-P-Ser PXSP or SPXR/K	Cell Signaling Technology	Cat# 2325 RRID:AB_331820
Anti-P-Thr PXTP motif	Cell Signaling Technology	Cat# 4391 RRID:AB_331247
Anti-Myc	Cell Signaling Technology	Cat# 2278 RRID:AB_490778
Anti-Myc tag Sepharose-conjugated	Cell Signaling Technology	Cat# 3400; RRID:AB_1281297
Anti-Flag	Cell Signaling Technology	Cat# 2368; RRID:AB_2217020
Anti-phospho-Thr	Cell Signaling Technology	Cat# 9386; RRID:AB_331239
Anti-phospho-Ser	Abcam	Cat# ab9332 RRID:AB_307184
Anti-phospho-Tyr	Millipore/Calbiochem	Cat# 525322
Anti-Cdk5	Santa Cruz Biotechnology	Cat# sc-173; RRID:AB_631224
Anti-β-Actin	Santa Cruz Biotechnology	Cat# sc-1615; RRID:AB_630835
Anti-His	Santa Cruz Biotechnology	Cat# sc-803; RRID:AB_631655
Anti-COASY	Novus Biologicals	Cat# NBP2-15934
Anti-BPIFB1	GeneTex	Cat# GTX60659
Anti-WDR77	GeneTex	Cat# GTX49156
Anti-MIB1	R&D Systems	Cat# AF7305; RRID:AB_11127218
Anti-HP	R&D Systems	Cat# AF4409; RRID:AB_1964600
Anti-MPO	R&D Systems	Cat# MAB3174; RRID:AB_2250873
Anti-IARS	Bethyl	Cat# A304-748A; RRID:AB_2620943
Anti-LCN2	R&D Systems	Cat# AF1857; RRID:AB_355022
Anti-CTTN	Santa Cruz Biotechnology	Cat# sc-55579; RRID:AB_831187
Anti-S6K1 (490-502)	Abcam	Cat# ab14708; RRID:AB_301427
Anti-EPRS	(Ray and Fox, 2007)	N/A
Anti-phospho-EPRS-Ser886	(Arif et al., 2009)	N/A
Anti-phospho-EPRS-Ser999	(Arif et al., 2009)	N/A
Bacterial and Virus Strains
MAX Efficiency Dh5a Competent Cells	ThermoFisher	Cat# 18258012
One Shot BL21(DE3)pLysS Competent Cells	ThermoFisher	Cat# C606003
Chemicals, Peptides, and Recombinant Proteins
Recombinant, purified His-tagged, wild-type and Ser886-to-Ala (S886A) mutant linker proteins spanning Pro683 to Asn1023 of human EPRS	(Arif et al., 2011) (Arif et al., 2009)	N/A
S6K1 peptide (KRRRLASLR) substrate	SignalChem	Cat# S05-58
Cdk5 peptide (PKTPKKAKKL) substrate	Santa Cruz Biotechnology	Cat# sc-3066
Recombinant, active, full-length S6K1 (rS6K1FL, 1-502 aa plus NLS)	R&D Systems; Cell Signaling Technology (Kozma et al., 1993)	Cat# 896-KS Cat# 7684
Recombinant active S6K1 truncate lacking CTD (rS6K11-398, 1-398 aa plus NLS)	EMD Millipore	Cat# 14-486
Recombinant active Cdk5/p35	EMD Millipore	Cat# 14-477
Recombinant full-length RPS6	Abcam	Cat# ab91724
NSAP1 protein	(Jia et al., 2008)	N/A
Phosphorylated-L13a protein	(Jia et al., 2008)	N/A
GAPDH protein	(Jia et al., 2008)	Cat# G6019
Amaxa cell line nucleofector kit V	Lonza	VCA-1003
Amaxa cell line nucleofector kit L	Lonza	VCA-1005
Roscovitine	Calbiochem	Cat# 557360
Prolyl endopeptidase	Abcam	Cat# ab80376
Protein A-Sepharose	Sigma-Aldrich	Cat# P3391
Insulin	Sigma-Aldrich	Cat# I9278
LPS	Sigma-Aldrich	Cat# L8274
IFN-α	Pestka Biomedical Lab.	Cat# 11200-1
IFN-β	Pestka Biomedical Lab.	Cat# 11415-1
IFN-γ	R&D Systems	Cat# 285IF
Insulin (for 3T3-L1 cell culture)	Cayman Chemical	Cat# 10008979
Dexamethasone (for 3T3-L1 cell culture)	Cayman Chemical	Cat# 10008980
3-isobutyl-1-methylxanthine (for 3T3-L1 cell culture)	Cayman Chemical	Cat# 10008978
Phosphosafe extraction buffer (Novagen)	EMD Millipore	Cat# 71296
Laemmli gel-loading dye	Bio-Rad	Cat# 1610737
Imperial blue stain	ThermoFisher	Cat# 24615
[γ-32P]ATP	Perkin-Elmer	Cat# BLU502A250UC
32P-orthophosphoric acid	Perkin-Elmer MP Biomedicals	Cat# NEX053H005MC Cat# 016401405
ATP	Cell Signaling Technology	Cat# 9804
P81-phosphocellulose squares	EMD Millipore	Cat# 20-134
Protease inhibitor cocktail	Roche Diagnostics	Cat# 11836170001
Phosphatase inhibitor cocktail	ThermoFisher	Cat# 78420
Polyvinyl difluoride membranes	EMD Millipore	Cat# IPVH00010
ECL prime western blotting detection reagent	GE Healthcare	Cat# RPN2232
Biacore Sensor chip CM5	Biacore Life Sciences (GE Healthcare)	Cat# BR100012
ECL-Anti-Rabbit HRP-conjugated secondary antibody	GE Healthcare	NA9340V
ECL-Anti-Mouse HRP-conjugated secondary antibody	GE Healthcare	NA931V
[35S]Met	Perkin-Elmer	NEG709A500UC
Critical Commercial Assays
GENEART Site-Directed Mutagenesis System	Life Technologies	Cat# A1382
SilverQuest Silver Staining Kit	Life Technologies	Cat# LC6070
Fugene 6 transfection reagent	Promega	Cat# E2691
Wheat germ extract system	Promega	Cat# L4380
6xHis Fusion Protein Spin Purification Kit (Ni- affinity chromatography)	ThermoFisher (Pierce)	Cat# 78300
EndoFree Plasmid Maxi Kit	Qiagen	Cat# 12362
Express Plus Tris-MOPS-SDS PAGE	GenScript	Cat# M00138
QBT fatty acid uptake assay kit	Molecular Devices	Cat# R8132
mMESSAGE mMachine T7 Transcription Kit	Ambion (Jia et al., 2008)	Cat# AM1344
mMESSAGE mMachine SP6 Transcription Kit	Ambion (Jia et al., 2008)	Cat# AM1340
Enterokinase cleavage capture kit (Novagen)	EMD Millipore	Cat# 69067-3
Experimental Models: Cell Lines
Human: U937 monocytic cells	ATCC	CRL-1593.2
Human: HepG2 hepatoma	ATCC	HB-8065
Human: peripheral blood monocytes (PBM)	Cleveland Clinic IRB-approved protocol	N/A
Human: HEK293T	ATCC	CRL-3216
Mouse: 3T3-L1	ATCC	CL-173
Oligonucleotides
siRNA targeting mouse EPRS 5’-UGAUACGAAGAUCUUCUCAG-3’ 5’-GCCUAAAUUAACAGUGGAA-3’	Origene Technologies	N/A
siRNA targeting human and mouse S6K1 (3’UTR-specific)	Origene Technologies	Cat# SR304164
siRNA targeting mouse S6K1	Origene Technologies	Cat# SR416706
siRNA targeting mouse CTTN	Origene Technologies	Cat# SR427044
siRNA targeting mouse LCN2	Origene Technologies	Cat# SR405112
siRNA targeting mouse COASY	Origene Technologies	Cat# SR417447
siRNA targeting mouse CDK5	Santa Cruz Biotechnology	Cat# sc-35047
siRNA targeting human Cdk5	Dharmacon Santa Cruz Biotechnology	Cat# L-003239-00-0050 Cat# sc-29263
Control siRNA	Santa Cruz Biotechnology	Cat# sc-37007
Recombinant DNA
Full-length human S6K1 cDNA in pCMV6-Entry	Origene Technologies	Cat# RC217324
Myc-tagged ORFs of wild -type and various mutant S6K1 (N-terminus 6-His, C-terminus Myc-tag S6K1 without 23-aa N-terminus NLS in pcDNA3)	This study	N/A
Flag-EPRS linker in pcDNA3	(Arif et al., 2009)	N/A
Software and Algorithms
GraphPad Prism 5	www.graphpad.com	N/A
BIAevaluation Software	www.biacore.com	N/A
NIH Image J	https://imagej.nih.gov/ij/	N/A

Highlights.

• Insulin-stimulated phosphorylation of EPRS Ser⁹⁹⁹ by S6K1 requires mTORCI and Cdk5

• C-terminus phosphorylation of S6K1 by Cdk5 is required for EPRS phosphorylation

• S6K1 exhibits a phospho-site dependent, target-selective phospho-code

• Multisite phosphorylated S6K1 induces a metabolon driving adipocyte lipid metabolism