Remodelling of the translatome controls diet and its impact on tumorigenesis

Abstract

Fasting is associated with a range of health benefits^1-6. How fasting signals elicit changes in the proteome to establish metabolic programmes remains poorly understood. Here we show that hepatocytes selectively remodel the translatome while global translation is paradoxically downregulated during fasting^7,8. We discover that phosphorylation of eukaryotic translation initiation factor 4E (P-eIF4E) is induced during fasting. We show that P-eIF4E is responsible for controlling the translation of genes involved in lipid catabolism and the production of ketone bodies. Inhibiting P-eIF4E impairs ketogenesis in response to fasting and a ketogenic diet. P-eIF4E regulates those messenger RNAs through a specific translation regulatory element within their 5' untranslated regions (5' UTRs). Our findings reveal a new signalling property of fatty acids, which are elevated during fasting. We found that fatty acids bind and induce AMP-activated protein kinase (AMPK) kinase activity that in turn enhances the phosphorylation of MAP kinase-interacting protein kinase (MNK), the kinase that phosphorylates eIF4E. The AMPK-MNK-eIF4E axis controls ketogenesis, revealing a new lipid-mediated kinase signalling pathway that links ketogenesis to translation control. Certain types of cancer use ketone bodies as an energy source^9,10 that may rely on P-eIF4E. Our findings reveal that on a ketogenic diet, treatment with eFT508 (also known as tomivosertib; a P-eIF4E inhibitor) restrains pancreatic tumour growth. Thus, our findings unveil a new fatty acid-induced signalling pathway that activates selective translation, which underlies ketogenesis and provides a tailored diet intervention therapy for cancer.

Extended Figure 1.. Ketogenesis is controlled at the translation level during fasting through P-eIF4E.

a. Quantification for the indicated proteins and phosphorylation states of immunoblot of liver lysates from mice were given chow ad libitum or fasted for 24 (n=3 mice). b. Blood glucose levels (n=5 mice), c. Serum insulin levels (n=8 mice), d. Body weight of WT and eIF4E^S209A mice in fed and 24h fasted conditions (n=5 mice). e. Serum-free fatty acids (FA) levels (n=4 or 5 mice) and f. Serum glycerol levels (n=5 mice) in 24h fasted WT and eIF4E^S209A mice. g. Relative levels of indicated metabolites in 24h fasted WT or eIF4E^S209A liver (n=3 mice). h. Body weight of WT treated with eFT508 or vehicle after 24h fasting (n=4 mice). i. Serum free fatty acids (FA) levels (n=4 mice), and j. Glycerol levels in 24h fasted WT mice treated with vehicle or eFT508 (n=4 or 5 mice). All values represent the mean ± SEM; Two-way ANOVA for a, d, g, h, and two-sided Student’s t test for b, c, e, g, f, i, j.

Extended Figure 2.. P-eIF4E is required for fasting-induced translation through the PRTE motif on 5’UTR.

a. Representative polysome profiles of 24h fasted WT and eIF4E^S209A liver lysates separated on a sucrose gradient; Inset highlights the free/RNP/80S/light polysome (2-7) and heavy polysomal fractions (8–13) used for translational profiling by polysomal RNAseq. b. Top pathways enriched in the 445 genes that are significantly translationally upregulated upon fasting but fail to increase in eIF4E^S209A liver based on the wiki pathway database. c. Percentages of the house keeping gene, Beta-2 microglobulin (B2M) mRNA distributed in sucrose gradient fractions against its total mRNA, of the 24h fasted WT and eIF4E^S209A livers (n=3 mice). d. Representative images of immunohistochemistry (IHC) of Hmgcs2 on sections from 24h fasted WT or eIF4E^S209A livers. e. Representative images, and f. Quantification of IHC of PPARα on liver sections from 24h fasted WT, PPARα knockout (PPARα^−/−), or eIF4E^S209A mice (n=3 mice). g. Relative mRNA levels of the downstream targets of PPARα, Cpt2 and Hsd17b10 in fasted WT and eIF4E^S209A liver normalized to B2M (n=3 mice). h. Representative image of luciferase expression of WT and eIF4E^S209A mice injected with PPARα 5’UTR linked luciferase in 24h fasted condition. i. Relative expression of Fluc with no 5’UTR normalized to the expression of the internal control Rluc at full medium (FM) or 24h induction of fasting mimicking medium (Fast: serum free with FA, n=3 biological independent samples). j. Relative expression of Fluc with Hmgcs2 5’UTR, k. eIF5A 5’UTR, l. HNRNPC 5’UTR, m. RPL26 5’UTR, and n. ABCD3 5’UTR or C to G transversion mutated PRTE motif in corresponding 5’UTR at full medium (FM) or 24h induction of fasting mimicking medium (n=3 biological independent samples). All values represent the mean ± SEM; Two-way ANOVA for g, One-way ANOVA for e, and two-sided Student’s t test for f, i, j, k, l, m, n.

Extended Figure 3.. Fatty acids activate the MNK-P-eIF4E axis by enhancing AMPK activity.

a. Immunoblot and quantification of indicated proteins in primary hepatocytes isolated from mice liver treated with BSA, 200 μM palmitic acid (PA), oleic acid (OA), or linoleic acid (LA) for 4 hours after overnight serum starvation, and immunoblot quantification of P-eIF4E/eIF4E in the AML12 cells and primary hepatocytes in response to BSA or different fatty acids (n=3 biological independent samples). b. Immunoblot of indicated proteins in AML12 cells or primary hepatocytes treated with 100μM fatty acids mix (FA) for 0, 1, 2, 3 hours after overnight serum starvation (n=3 biological independent samples). c. Percentages of indicated mRNA distributed in sucrose gradient fractions of AML12 cells treated with BSA or 150 μM linoleic acid for 4h in overnight serum-free medium (n=3 biological independent samples). d. Percentages of indicated mRNA distributed in sucrose gradient fractions of of livers from WT or fat specific adipose triglyceride lipase knockout mice (ATGL^−/−) upon 24h fasting (n=3 mice). e. Relative expression of Fluc with PPARα 5’UTR or empty vector after BSA, 150 μM linoleic acid or linoleic acid plus preincubation of 1μM eFT508 in serum free conditions (n=3 biological independent samples). f. Immunoblot of indicated proteins in AML12 cells treated with different inhibitors before and during 4h of linoleic acid stimuli in serum free conditions. g Immunoblot and quantification of indicated proteins in AML12 cells pretreated with vehicle or AMPK inhibitor (Bay-3827) in serum free conditions upon 4h fatty acids stimuli (n=3 biological independent samples). h. Immunoblot of indicated proteins in primary hepatocytes treated with vehicle or AMPK inhibitor (Compound C) in serum free conditions upon fatty acids (150μM mix of PA, OA and LA) stimuli (n=3 biological independent samples). i. Percentages of PPARα, Hmgcs2 and B2M mRNAs distributed in sucrose gradient fractions against their total mRNA levels of AML12 cells preincubated with vehicle or 10 μM AMPK inhibitor Compound C in serum free conditions upon 4h linoleic acid stimuli (n=3 biological independent samples). j. Immunoblot of indicated proteins in fed and 24h fasted livers (n=3 mice). k. Blood BHB levels of 24h fasted eIF4E^S209A mice treated with vehicle or AMPK inhibitor Compound C before fasting and 12h after fasting (n=4 mice). All values represent the mean ± SEM; Two-way ANOVA for a, c, d, e, i, One-way ANOVA for g, and two-sided Student’s t test for k.

Extended Figure 4.. AMPK phosphorylates MNK.

a. Relative AMPK activity on MNK1 as substrate in the presence of vehicle or 10 μM AMPK specific inhibitor compound C in vitro with ADP Glo Assay (n=3 biological independent samples). b. MSMS spectra belonging to tryptic peptides spanning amino acids A36 to K49 (panel A), D165 to R210 (B), and N392 to R410 (c) of human MNK1, obtained by HCD fragmentation of precursor ions 841.3626⁺², 1157.8012⁺⁴, and 1067.4977⁺² respectively. An excess mass of 79.9663 compared to the unmodified sequence, corresponding to the phosphate group, is observed in the precursor and in sequence ions than contain serine 39, serine 168 and serine 394. Experimental masses of the most representative sequence ion peaks are labelled in the spectra, also indicating the fragment type (Roepstorff-Fohlmann-Biemann ions nomenclature). The position of the fragmentation events generating these ions are indicated in the sequences over the spectra. A table indicating the theoretical masses of the sequence ions according to the proposed modified sequences is shown in the right side, with red font indicating observed fragments. S^Phospho, phosphorylated serina; C^CM, carboxymethylated cysteine. c. Relative AMPK activity of recombinant AMPK on the same amount of purified wild type MNK1, S39A, S168A, S394A and triple mutation of S39A, S168A and S394A MNK1, and AMPK’s known substrate SAMStide; all readings were normalized by subtracting the values obtained with the AMPK inhibitor (n=3 biological independent samples). d. Silver staining of purified human eIF4E-GST and human wild type or mutated MNK1-GST. e. Stacked ¹H NMR (300 MHz, DMSO-d₆) spectra of linoleic acid (green), CoASH (red), and linoleoyl-CoA (blue). A characteristic shift of the triplet at 2.2 ppm in linoleic acid to 2.8 ppm is seen in linoleoyl-CoA, corresponding to the protons alpha to the thioester carbonyl (indicated by arrows on the chemical structure and in the spectrum). All values represent the mean ± SEM; One-way ANOVA for c, and two-sided Student’s t test for a.

Extended Figure 5.. Linoleic acid activates AMPK through directly binding to AMPKγ.

a. Relative AMPKα1β1γ1 activity on its substrate SAMStide in the presence of vehicle (Veh), BSA, 100 μM linoleic acid (LA), 10 μM linoleoyl-CoA (LA-CoA) or 100 μM AMP (n=5 biological independent samples) b. Relative AMPKα1β1γ1 activity on its substrate peptide SAMStide in the presence of vehicle (Veh), or the indicated linoleic acids and AMP combination (n=3 biological independent samples). c. Silver staining of blue native gel with recombinant AMPK complex with bodipy or C12-bodipy. d. Immunoblot of indicated proteins in whole cell lysates from fed and 24h fasted livers, and LA conjugated or butyric acid conjugated beads pull-downed proteins; the total proteins of each sample on membrane were shown using ponceau S staining. e. Immunoblot of residual AMPKγ in the purified AMPKα1β1γ1 incubated linoleic acid conjugated beads or butyric acid beads after competing with different concentrations of free linoleic acid or butyric acid. f. Relative AMPKα1β1γ1 activity on its substrate SAMStide, in the presence of vehicle (Veh), BSA, 100 μM linoleic acid (LA), 10 μM linoleoyl-CoA (LA-CoA), 10 μM palmitoyl-CoA (PA-CoA) or 100 μM AMP (n=3 biological independent samples). g. Relative AMPKα1β1γ1 activity on its substrate SAMStide, with different concentrations of linoleic acid (n=5 biological independent samples). h. The isolated AMPKγ1-subunit remains stable. Backbone RMSD values of the γ-subunit are report without (blue line) and with (orange line) AMP bound at CBS sites −1, −3, and −4. The relatively low and stable RMSDs over these simulations confirm that simulations of γ can be run in the absence of the rest of the AMPK complex. i. A heat map of linoleic acid contacts with AMPKγ1 reveals primary sites of interaction. Structure of the γ subunit (PDB 4RER) with residues colored by contact time with linoleic acid aggregated across all lipid binding simulations, highlighting identified lipid binding Sites 1 and 2. The color spectrum runs from white to red, with white being no/minimal fatty acid contact and red being high contact. Site 1 is a hydrophobic pocket buried among the main and side chains of L173, K177, L285, I289, V293, L315, I318, and Y29; Site 2 is buried within F91, I94, L95, A250, Y255, K58, F62, V65, and L109. A third site, Site 3, at the bottom of the figure also sampled frequent contact with linoleic acid, but was not tested as a candidate for mutation given the site’s proximity to the binding interface with the β-subunit, reduced contact times compared to Sites 1 and 2, and the diffuse nature of the interaction surface. j. Proposed mutations to Sites 1 (left column) and 2 (right column) for disrupting linoleic acid binding. The images highlight the wild type residues (top row) and proposed mutations (bottom row). Mutations were selected based on fatty acid binding simulations, aiming to crowd out the hydrophobic pockets that linoleic acid was seen to enter in order to prevent binding. k. Scheme of expression of different AMPKγ1 mutants in Hek293T cells, and purification of AMPKα1β1γ1 complex with different AMPKγ mutants using flag beads; AMPKα1 and AMPKγ1 levels in the pulled-down samples were shown using immunoblot; relative AMPK activity of AMPK complex with different AMPKγ1 mutants when used SAMStide as substrate and reaction with AMPK inhibitor compound C as negative control (n=3 biological independent samples). l. Immunoblotting of different AMPKγ1 mutants in the whole cell lysates before (input) and incubated with linoleic acid conjugated beads and bound to the beads. m. Silver staining of purified human wild type or mutated AMPKα1β1γ1 complex. All values represent the mean ± SEM; One-way ANOVA for a, b, f, linear regression analysis was used for g, and two-sided Student’s t test for g, i, j, k. Schematics in d, e, k, m is created with BioRender.com

Extended Figure 6.. Ketogenic diet activates P-eIF4E dependent translation regulation of ketogenic mRNAs.

a. Quantification of indicated protein in the immunoblot for regular chow and ketogenic diet (n=3 mice). b. Quantification of indicated protein in the immunoblot for ketogenic diet treated with vehicle and AMPK inhibitor (Comp. C) (n=3 mice). c. Body weight of WT and eIF4E^S209A mice fed with regular chow and ketogenic diet (n=5 or 7 mice). d. Serum glycerol (n=7 mice), e. Serum-free fatty acids (FA) (n=7 mice) in WT and eIF4E^S209A mice fed with ketogenic diet. f. Percentages of indicated mRNA distributed in sucrose gradient fractions of ketogenic diet fed WT and eIF4E^S209A livers (n=3 mice). g. Immunoblot of indicated proteins, h. Representative images of immunohistochemistry (IHC) or Myc in the liver, and i. Triglyceride levels (n=3 or 4 mice) in the liver from eIF4E^S209A mice injected with vehicle or Hmgcs2 recombinant cDNA with Myc tag. j. Immunoblot of indicated proteins and k. Representative images of IHC of Myc in the liver from eIF4E^S209A mice injected with vehicle or PPARα recombinant cDNA with Myc tag. l. qPCR analysis of PPARα and Hmgcs2 mRNA levels in the liver from WT mice, and eIF4E^S209A mice with hydrodynamic injection of vehicle or PPARα cDNA (n=3 mice). All values represent the mean ± SEM; Two-way ANOVA for a, b, c, l, and two-sided Student’s t test for d, e, i.

Extended Figure 7.. Combination of eFT508 and ketogenic diet systemically inhibits pancreatic tumor growth.

a. Tumor size, b. Tumor weight, c. Blood BHB levels, and d. Body weight of xenograft mouse models with the human pancreatic cancer cell line AsPC1, treated with chow, eFT508, ketogenic diet or eFT508 and ketogenic diet after the tumors reach 150 mm³ (n=4 mice). e. Tumor weight (n=9 or 10 biological independent samples) f. Mouse body weight (n=5 mice), g. Food intake (n=5 days), h. Blood glucose levels (n=5 mice), i. Blood glycerol levels (n=5 mice), j. Serum-free fatty acids levels (n=5 mice), and k. Serum insulin levels (n=4 or 5 mice) of xenograft mouse models with the human pancreatic cancer cell line T3M4 with different treatment after the tumors reach 150 mm³. l. Relative cell survival of T3M4 cells treated without or with 1mM BHB in fasting mimicking medium (no glucose serum-free medium with 100 μM linoleic acid) for 48h (n=3 biological independent samples. m. Mouse body weight (n=5 mice) and n. Represent tumor images of xenograft mouse models with the human pancreatic cancer cell line T3M4, treated with ketogenic diet or eFT508 and ketogenic diet or eFT508 and ketogenic diet with 1% BHB in the drinking water after the tumors reach 150 mm³. All values represent the mean ± SEM; Two-way ANOVA for a, b, c, d, f, h, l, m, One-way ANOVA for e, and two-sided Student’s t test for g, i, j, k.

Extended Figure 8.. Combination of eFT508 and ketogenic diet intrinsically affects pancreatic tumor growth.

a. AsPC1 cells were treated with 100 μM linoleic acid or BSA for 4h or 24h in low glucose serum-free (SF) medium and cell lysates were analyzed by immunoblotting for the indicated proteins and phosphorylation states (n=3 biological independent samples). b. Percentages of indicated mRNA distributed in sucrose gradient fractions of tumors from xenograft mouse models with the human pancreatic cancer cell line T3M4, treated with ketogenic diet or eFT508 and ketogenic diet (n=3 biological independent samples). c. Immunoblot of GFP in T3M4 cells overexpressing PPARα-GFP cDNA and relative cell numbers of control T3M4 and PPARα overexpressing T3M4 in full medium. d. Tumor weight and representative images of end point tumors from xenograft mouse models with the control T3M4 or T3M4 cells overexpressing PPARα cDNA treated with ketogenic diet (Keto) and vehicle (Veh) or a combination of eFT508 and ketogenic diet (n=9 or 10 biological independent samples). e. Immunoblot of indicated proteins from the xenograft tumors with the control T3M4 or PPARα cDNA overexpressing T3M4 cells treated vehicle (Veh) eFT508 upon a ketogenic diet (n=3 mice). f. Blood BHB levels of xenograft mouse models with the control T3M4 or PPARα cDNA overexpressing T3M4 cells treated with vehicle (Veh) or eFT508 upon a ketogenic diet (n=5 mice). g. Relative cell survival levels of T3M4 or AsPC1 pancreatic cancer cells treated with eFT508, PPARα inhibitor, eFT508 and PPARα inhibitor, and β-hydroxybutyrate in the presence of eFT508 in fasting mimicking medium (FFM: no glucose serum-free medium with 100 μM linoleic acid) for 2 days (n=3 biological independent samples). h. Scheme of T3M4 stably expressing luciferase reporter with wild type PPARα 5’UTR (WT) or PRTE mutated PPARα 5’UTR (Mut). i. The quantification of luciferase signal and representative images of the tumors from the xenograft mouse model in response to regular chow or ketogenic diet (n=7 or 8 mice). j. The quantification of luciferase signal and representative images of the tumors from the xenograft mouse model in response to vehicle or eFT508 upon ketogenic diet (n=6 mice). All values represent the mean ± SEM; Two-way ANOVA for b, d, f, and One-way ANOVA for g, and two-sided Student’s t test for i, j; Schematics in a, h is created with BioRender.com

Figure 1.. Ketogenesis is controlled at the translation level during fasting through P-eIF4E.

a and b Immunoblot of indicated proteins in the livers from Ad libitum (fed), 24h fasted, fasted and 2h refed mice. c. Representative polysome profiles of fed and 24h fasted WT liver lysates separated on a sucrose gradient. No/low translation fractions (2-7) and polysomal fractions (8–13) used for PolyRibo-seq. d. Volcano plot of log2 fold change of translation efficiency (TE) of transcripts upon fasting in the liver. Red points indicate significantly translationally upregulated genes. e. Top pathways enriched in the significantly translationally upregulated genes based on the wiki pathway. f. Percentages of indicated mRNA distributed in sucrose gradient fractions of fed and fasted WT livers (n=3). g. Blood BHB levels in 24h fasted wild type (WT) and eIF4E^S209A mice (n=9). h. Volcano plot of metabolites changes in the liver from fasted eIF4E^S209A and WT mice (n=3). i. Relative butyryl-CoA levels in 24h fasted WT and eIF4E^S209A livers (n=3). j. Triglyceride levels in 24h fasted WT and eIF4E^S209A liver (n=4). k. Relative HMG-CoA levels in fasted WT and eIF4E^S209A livers (n=3). l. Blood BHB levels in 24h fasted PPARα^−/− mice (n=6), or WT mice pretreated with vehicle or eFT508 (n=4). All values represent the mean ± SEM. Statistical analysis was performed with Two-way ANOVA analysis for f, l, two-sided Student’s t test for g, i, j, k; Limma package analysis and adjusted P value for d, e. Each n in g, i, j, k, l is an individual mouse; n in f is biological independent sample. Schematics in a, c, g, h are created with BioRender.com

Figure 2.. P-eIF4E is required for fasting-induced translation through the PRTE motif on 5’UTR.

a. 2D plot of (log₂) FC of TE of transcripts between fasted and fed conditions in WT liver (X axis), and between fasted WT and eIF4E^S209A livers (Y axis). Significantly upregulated genes upon fasting but fail to increase in fasted eIF4E^S209A liver are labeled in green (445 genes), and b. are shown in the heatmap where the top 2 enriched pathways are highlighted. c. Percentages of indicated mRNA distributed in sucrose gradient fractions of 24h fasted WT and eIF4E^S209A livers (n=3). d. Relative PPARα activity (n=5) and Hmgcs2 levels based on IHC staining (n=3) in fasted WT and eIF4E^S209A livers. e. Scheme of hydrodynamic tail vein injection of luciferase reporter into mice; Relative luciferase expression in each group normalized to fed condition (n=3 or 5). f. Relative expression of luciferase linked with PPARα 5’UTR in different condition (n=3 or 5). g. Scheme of the consensus sequence of CERT and PRTE motifs. h. Pie chart of genes with CERT or PRTE motifs in 445 genes from Fig.2a. i. Expression levels of luciferase with indicated 5’UTR in fed and fasted WT livers (n=3 or 4), and j. in fasted WT and eIF4E^S209A livers (n=3 or 4). k. Scheme of the AML12 cell model with fasting mimicking condition (Fast): serum free medium with 150μM fatty acids, and relative Fluc to Rluc ratio with indicated 5’UTR at full medium (FM) or 24h induction of Fast condition (n=3). All values represent the mean ± SEM; Two-way ANOVA for c, i, and two-sided Student’s t test for d, e, f, j, k. Each n in d, e, f, i, j is an individual mouse; n in k, c is biological independent sample. Schematics in a, e, g, k is created with BioRender.com

Figure 3.. Fatty acids enhance AMPK activity to induce the MNK-P-eIF4E axis.

a. Immunoblot of indicated proteins inAML12 cells treated with BSA, 100 μM linoleic acid (LA), oleic acid (OA), palmitic acid (PA), or all of them for 4h after overnight in serum free medium. b. Scheme and quantification of P-eIF4E/eIF4E levels of FA-stimulated AML12 cells treated with different inhibitors (n=3). c. Percentages of indicated mRNA distributed in sucrose gradient fractions of FA-stimulated AML12 cells pretreated with vehicle or Bay-3827 (n=3). d. Blood BHB levels in 24h fasted mice pretreated with vehicle or Compound C or Bay-3827 (n=5). e. Relative AMPK activity when using vehicle or recombinant MNK1 as a substrate. f. Scheme of phosphorylation sites on MNK1 by AMPK or ERK1/2. g. Immunoblot of indicated proteins of wild-type MNK or S394A MNK treated with or without AMPK for 2 h, then adding vehicle or ERK1 for 30min. h. Relative AMPKα1β1γ1 activity on SAMStide, in the presence of vehicle, 100 μM AMP or LA or both of them (n=3), and i. different concentation of LA. j. Structure of AMPKγ with LA interaction. k. Scheme of purification of AMPK complex; Immunoblot of different AMPKγ1 mutants in the inputs and bound to the LA-beads. l. Relative activity of purified AMPKα1β1γ1 complex with different AMPKγ1 mutants on SAMStide, induced by vehicle, 100 μM LA or AMP, and immunoblot of P-AMPKα in each AMPK complex (n=5 or 7). All values represent the mean ± SEM; Two-way ANOVA for c, l, one-way ANOVA for b, h, two-sided Student’s t test for d, and linear regression analysis was used for e, i. Each n in d is an individual mouse, n in b, c, h, l is biological independent sample. Schematics in a, b, d, e, f, k are created with BioRender.com

Figure 4.. Ketogenic diet activates P-eIF4E dependent translation to regulate ketogenesis.

a. Scheme of a ketogenic diet (Keto); Immunoblot of indicated proteins and relative activity (n=3) from AMPK immunoprecipitated from livers of mice fed chow or Keto for 24 h. b. Immunoblot of indicated proteins in livers from mice upon chow or Keto for 24h, and c. or upon Keto and dosed with vehicle or AMPK inhibitor 4 h. d. Blood BHB levels in WT and eIF4E^S209A mice fed with Keto for 1 day or 2 months (n=5). e. 2D plot of (log₂) FC of TE of transcripts between fasted and fed WT livers (X axis), and between Keto and chow fed WT livers (Y axis). f. Heatmap of (log₂) TE of P-eIF4E-dependent 445 genes upon Keto, chow, or fasting. g. Percentages of indicated mRNA distributed in sucrose gradient fractions of WT and eIF4E^S209A livers upon Keto (n=3). h. Triglyceride levels in the WT and eIF4E^S209A livers upon Keto for 24h (n=5 or 7). i. Relative levels of short-chain Acyl-CoA in WT and eIF4E^S209A livers upon Keto for 24h (n=4 or 5). j. Scheme of hydrodynamic tail vein injection of Hmgcs2 cDNA into mice and blood BHB levels in WT mice, or eIF4E^S209A mice injected with vehicle or Hmgcs2 cDNA with Keto for 24h (n=5 or 6). k. BHB levels of WT mice, or eIF4E^S209A mice injected with vehicle or PPARα cDNA with regular chow, or Keto for 24h (n=5 or 6). All values represent the mean ± SEM; Two-sided Student’s t test for a, h, i, two-way ANOVA for d, g, k, and one-way ANOVA for j; Limma package analysis and adjusted P value for e. Each n in a, d, h, I, j, k is an individual mouse, n in g is biological independent sample. Schematics in a, d, j are created with BioRender.com

Figure 5.. Pharmacologically targeting P-eIF4E in combination with a ketogenic diet limits pancreatic tumor growth.

a. Scheme of a xenograft model of human pancreatic cancer cells in Athymic nude mice; Tumor size and b. images of T3M4 xenograft model, treated with indicated diet (chow or Keto) or drug (vehicle or eFT508) (n=9 or 10). c. Percentages of indicated mRNA distributed in sucrose gradient fractions of livers from mice under indicated treatments (n=3). d. Blood BHB levels in the mice under indicated treatments (n=5). e. Scheme and blood BHB in xenograft mouse model upon Keto with vehicle, eFT508, or eFT508 with 1% BHB addition in the drinking water (n=4). f. Tumor size of T3M4 xenograft model under indicated treatment (n=8 or 9). g. Immunoblot of indicated proteins of T3M4 cells treated with 100 μM LA or BSA in low glucose serum free (SF) medium. h. 2D plot of (log₂) FC of TE of transcripts between Keto and chow fed T3M4 xenograft tumors (X axis), and between Keto plus eFT508 and Keto alone treated T3M4 xenograft tumors (Y axis). i. Tumor size of xenograft mouse model with the control or PPARα cDNA overexpressed T3M4 cells under indicated treatment (n=9 or 10). j. Scheme of metabolic and translatomic reprogramming in the pancreatic tumor in response to Keto and eFT508. k. Relative cell survival of control or PPARα overexpressed T3M4 cells upon Fast condition without or with 1mM BHB (n=3). All values represent the mean ± SEM; Two-way ANOVA for a, c, f, i, k, and one-way ANOVA for d, e; Limma package analysis and adjusted P value for h. Each n in a, d, e, f, i is an individual mouse, n in c, k is biological independent sample. Schematics in a, c, e, g-k are created with BioRender.com