Differential integrated stress response and asparagine production drive symbiosis and therapy resistance of pancreatic adenocarcinoma cells

Abstract

The pancreatic tumor microenvironment drives deregulated nutrient availability. Accordingly, pancreatic cancer cells require metabolic adaptations to survive and proliferate. Pancreatic cancer subtypes have been characterized by transcriptional and functional differences, with subtypes reported to exist within the same tumor. However, it remains unclear if this diversity extends to metabolic programming. Here, using metabolomic profiling and functional interrogation of metabolic dependencies, we identify two distinct metabolic subclasses among neoplastic populations within individual human and mouse tumors. Furthermore, these populations are poised for metabolic cross-talk, and in examining this, we find an unexpected role for asparagine supporting proliferation during limited respiration. Constitutive GCN2 activation permits ATF4 signaling in one subtype, driving excess asparagine production. Asparagine release provides resistance during impaired respiration, enabling symbiosis. Functionally, availability of exogenous asparagine during limited respiration indirectly supports maintenance of aspartate pools, a rate-limiting biosynthetic precursor. Conversely, depletion of extracellular asparagine with PEG-asparaginase sensitizes tumors to mitochondrial targeting with phenformin.

Fig. 1. Metabolic characterization reveals two distinct cellular populations from PDA tumors.

a, A polyclonal cell line established from a mouse pancreatic tumor was subcloned into seven clonal lines and subjected to LC–MS/MS metabolomics analysis. b, Heat map representation of significantly different intracellular metabolites pooled from a positive and negative ionization mode analysis among the clonal cell lines grown under the same media conditions; fold change of ±2; P = 0.01 between group 1 clones (E, V and H) and group 2 clones (K, M, N and T). Rows are clonal cell lines in triplicate, and columns represent metabolites (n = 3 biological replicates per cell line). c–f, Sensitivity of clonal cell lines to 33 µM FX11 (c) and IC₅₀ values for aminooxyacetic acid (AOA) (d), oligomycin (e) and phenformin (f); n = 3 biological replicates per cell line. g, Group 1 clones (H and V) and group 2 clones (N and T) were subjected to NAD(P)H FLIM. Data are presented as a ratio of free to protein-bound NAD(P)H; n = 104 clone H cells, n = 108 clone V cells, n = 136 clone N cells and n = 105 clone T cells; OXPHOS, oxidative phosphorylation. h, Histograms of the indicated clones stained with TMRM and analyzed via flow cytometry. i, Ratio of TMRM to MitoTracker Green staining of the indicated clonal lines (n = 2 biological replicates for clone V and n = 3 biological replicates for clones H, N and T). Error bars represent mean ± s.d.; **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001 by two-tailed Mann–Whitney test (c–f and i) or one-way ANOVA with a Tukey post hoc test; P < 0.0001 (c); P < 0.0001 (d); P = 0.0018 (e); P < 0.0001 (f); H versus T P < 0.0001, H versus N P < 0.0001, V versus T P < 0.0001 and V versus N P = 0.0003 (g); P = 0.0043 (i). Source data

Fig. 2. Cocultures reveal cross-talk interactions between clonal groups.

a, Clonal line N was encoded with a fluorescent label, plated in direct coculture with unlabeled clones and treated with oligomycin or vehicle. b, Representative images (one of three biological replicates) of a labeled oligomycin-sensitive clone N cocultured with unlabeled resistant clones (E and V) or with unlabeled sensitive clones (K and N) and treated with oligomycin or vehicle; scale bar, 400 µm. c, Fluorescent plate area of oligomycin-treated (0.75 nM) versus vehicle-treated N-labeled cocultures (n = 3 samples). d, Fluorescent plate area of oligomycin-treated (1 nM) versus vehicle-treated K-labeled cocultures with unlabeled resistant clones (H and V) or with unlabeled sensitive clones (K and M; n = 3 samples). e, Fluorescent plate area of 1 nM oligomycin (Oligo)- or 25 µM phenformin (Phen)-treated versus vehicle-treated cocultures for labeled sensitive clone N cocultured with unlabeled clone N or insensitive clone V (n = 3 samples). f, Fluorescent plate area of the 1.5 nM oligomycin-treated cocultures relative to vehicle-treated cocultures (n = 3 samples). g, Group 2 clone N was plated with no transwell or with transwells containing group 1 clone V or group 2 clones N and M. Cultures were treated with 0.25 nM oligomycin with fresh drug added every 48 h for 10 d and were fixed and stained with crystal violet. The image is representative of three experimental repeats. h, Quantitation of colony area of g (n = 3 samples). Error bars represent mean ± s.d.; * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001 and **** P ≤ 0.0001 by one-way ANOVA with a Tukey post hoc test (c, d and f) or by two-tailed Student’s t test (e and h); NS, not significant; +E versus +K P = 0.0007, +E versus +N P = 0.0009, +V versus +K P = 0.0157 and +V versus +N P = 0.0207 (c); +H versus +K P = 0.0047, +H versus +M P = 0.0266, +V versus +K P = 0.0006 and +V versus +M P = 0.0024 (d); oligomycin P = 0.0002 and phenformin P = 0.0012 (e); +V versus +N P = 0.0114, +V versus +M P = 0.0108, +H versus +N P = 0.0137 and +H versus +M P = 0.013 (f); no top P = 0.000585, V top P = 0.076023, N top P = 0.0002277 and M top P = 0.000004 (g). Source data

Fig. 3. Media profiling reveals that asparagine rescues inhibition of respiration.

a, Differentially consumed/released metabolites present in the media from group 1 and group 2 clones (n = 3 replicates per cell line) after 48 h of culture. b, Relative abundance of NEAAs present in conditioned medium at higher levels than in basal medium (n = 3 replicates per cell line). c, Doubling time of clone N treated with 1 nM oligomycin in the presence or absence of medium containing a 100 µM cocktail of all NEAAs (n = 3 samples per condition). d, Doubling times of clone N treated with 1 nM oligomycin in the presence or absence of individual NEAAs (n = 3 samples per condition). e, Doubling times of clone N treated with 1 nM oligomycin in the presence or absence of 100 µM or 20 mM aspartate (Asp; n = 3 samples per condition). f, Doubling times of clonal line N treated with 1 nM oligomycin, 25 µM phenformin or 25 nM IACS-10759 ± asparagine (Asn). g, Clonal line N was encoded with a fluorescent label and plated in direct coculture with unlabeled clones transfected with siRNA targeting Asns or non-targeted (NT) control. h, Fluorescent plate area of 1 nM oligomycin-treated cocultures of labeled clone N with unlabeled clone N or insensitive clone V (n = 3 samples) transfected with the indicated siRNA with or without the addition of exogenous asparagine. Error bars represent mean ± s.d.; * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001 and **** P ≤ 0.0001 by two-tailed Mann–Whitney test (b), two-tailed Student’s t test (c and f) or one-way ANOVA with a Tukey post hoc test (d, e and h); alanine P < 0.0001, asparagine P = 0.0043, aspartate P = 0.0043, glutamate P = 0.0184, glycine P = 0.0013 and proline P < 0.0001 (b); P = 0.0008 (c); DMEM versus +asparagine P < 0.0001, +asparagine versus +glycine P = 0.0001), +asparagine versus +serine P = 0.0419, +asparagine versus +alanine P = 0.0008, +asparagine versus +proline P = 0.0001, +asparagine versus +aspartate P = 0.0016 and +asparagine versus +glutamate P = 0.0122 (d); DMEM versus aspartate 20 mM P = 0.0002 and aspartate 100 µM versus aspartate 20 mM P = 0.0004 (e); mock versus +asparagine for oligomycin P = 0.0011, phenformin P < 0.0001 and IACS-10759 P = 0.0006 (f); V siNT versus V siAsns P < 0.0001, V siAsns versus V siAsns + asparagine P = 0.0572 (h). Source data

Fig. 4. PDA clones engage different models of activation of the ISR.

a, Immunoblot comparison of phospho-p44/42 MAPK Thr 202/Tyr 204 (pERK), total p44/42 MAPK (ERK), c-Myc and vinculin across group 1 (E, H and V) and group 2 (K, M, N and T) clonal populations. The image is representative of three independent experiments. b, Quantification of Myc expression across clonal groups normalized to vinculin loading controls (n = 3 blots per clone). c, Immunoblot comparison of ATF4, PSPH, SHMT2, ASNS and vinculin across group 1 (E, H and V) and group 2 (K, M, N and T) clonal populations. The image is representative of three independent experiments. d, Quantification of ASNS expression across clonal groups normalized to vinculin loading controls (n = 3 blots per clone). e, Expression of phospho-GCN2 T899 (pGCN2) and vinculin between group 1 (E, H and V) and group 2 (K, M, N and T) clonal populations treated with either vehicle or 1 nM oligomycin. f, Schematic representation of clonal cross-talk between ISR-high or ISR-low clones and key nodes to inhibit cross-talk. The image is representative of three replicates. g, Clonal line N was encoded with a fluorescent label, plated in direct coculture with unlabeled sensitive clone N or insensitive clone V and treated with oligomycin, ISRIB, GCN2iB, oligomycin and ISRIB, oligomycin and GCN2iB or vehicle (n = 3 replicates). Cells were counted, and endpoint data were plotted relative to vehicle; P ≤ 0.01, *** P ≤ 0.001 and **** P ≤ 0.0001 by two-tailed Mann–Whitney test (b and d) and one-way ANOVA with a Tukey post hoc test (g); P = 0.0043 (b), P = 0.0003 (d), +N versus +N + oligomycin P < 0.0001, +N + oligomycin versus +N + oligomycin P < 0.0001, +N + oligomycin + GCN2iB versus +V + oligomycin + GCN2iB P = 0.0844, +N + oligomycin + ISRIB versus +V + oligomycin + ISRIB P > 0.9999, +V + oligomycin versus +V + oligomycin + GCN2iB P = 0.0004 and +V + oligomycin versus +V + oligomycin + ISRIB P = 0.0038 (g). Source data

Fig. 5. Human PDA tumors exhibit differential ISR activation.

a, Immunostaining for ASNS in human PDA tissues; dashed boxes are magnified in insets; scale bars, 100 µm. b, UMAP representations of KRT19 expression delineating the epithelial cell population identified from single-cell RNA analysis of a tumor biopsy of PDA individual 1229. c, UMAP representation of the three cell clusters within the epithelial population in b. d, Non-hierarchical clustering analysis of epithelial clusters across a set of ISR genes. e, Immunoblotting analysis of c-Myc, ATF4, ASNS and vinculin of clonal cells lines derived from the PATC53 human PDA tumor model; red, ISR low; blue, ISR high; purple, does not fit either category. The image is representative of three individual experiments. f, Dose–response of PATC53 clones to oligomycin; blue, insensitive clones; red, sensitive clones; n = 3 biological replicates per clone. g, Clonal line PATC53-23 was encoded with a fluorescent label and plated in direct coculture with unlabeled clones PATC53-23 or PATC53-34. h, Cell numbers of labeled PATC53-23 cocultures treated with 0.5 nM oligomycin or 125 µM phenformin relative to vehicle control (n = 3 biological replicates). Error bars are representative of mean ± s.d.; *P ≤ 0.05 and ***P ≤ 0.001 by two-tailed Student’s t-test; oligomycin P = 0.012 and phenformin P = 0.0002 (h). Source data

Fig. 6. Asparagine rescues respiration inhibition by supporting aspartate pools.

a, Oxygen consumption of clonal line N treated with vehicle, oligomycin or phenformin, as measured by a Seahorse metabolic flux analyzer (n = 4 samples per condition). b–e, Relative abundance of TCA cycle metabolites (b), aspartate (c), nucleotides (d) and asparagine (e) present in clone N cells treated with asparagine, oligomycin, asparagine + oligomycin or vehicle 4 h after treatment (n = 3 samples per condition). f, Labeled aspartate pools of clone N prelabeled with universally labeled ¹³C-glutamine and switched to medium containing unlabeled glutamine and treated with asparagine, oligomycin, asparagine + oligomycin or vehicle for 4 h (n = 3 samples per condition). g,h, Relative aspartate (g) and nucleotide levels (h) of siNT- or siAsns-transfected clone N cells treated with 1 nM oligomycin for 4 h (n = 3 samples per condition). i, Aspartate levels of oligomycin-sensitive clone N and oligomycin-insensitive clone V relative to vehicle 4 h after treatment (n = 3 samples per condition). j, Ratio of NADH/NAD⁺ of clone N and clone V treated for 4 h with oligomycin or vehicle (n = 3 samples per condition). k, Labeled aspartate pools of sensitive clone N and insensitive clone V were treated for 4 h with oligomycin or vehicle and switched to medium containing universally labeled ¹³C-glutamine and treated with oligomycin or vehicle for an additional 4 h (n = 3 samples per condition); *P ≤ 0.05, **P ≤ 0.01 and ****P ≤ 0.0001 by one-way ANOVA with a Tukey post hoc test (a–f and j) or two-tailed Student’s t-test (g–i and k); vehicle versus oligomycin P < 0.0001, vehicle versus phenformin P < 0.0001 (a); pyruvate: vehicle versus oligomycin P = 0.0017, vehicle versus oligomycin + asparagine P = 0.0019, asparagine versus oligomycin P = 0.0021 and asparagine versus oligomycin + asparagine P = 0.0023; (iso)citrate: vehicle versus oligomycin P < 0.0001, vehicle versus oligomycin + asparagine P < 0.0001, asparagine versus oligomycin P < 0.0001, asparagine versus oligomycin + asparagine P < 0.0001; aconitate: vehicle versus oligomycin P < 0.0001, vehicle versus oligomycin + asparagine P < 0.0001, asparagine versus oligomycin P < 0.0001, asparagine versus oligomycin + asparagine P < 0.0001; α-ketoglutarate (AKG): vehicle versus oligomycin P < 0.0001, vehicle versus oligomycin + asparagine P = 0.0002, asparagine versus oligomycin P < 0.0001, asparagine versus oligomycin + asparagine P < 0.0001; succinate: vehicle versus oligomycin P < 0.0001, vehicle versus oligomycin + asparagine P < 0.0001, asparagine versus oligomycin P < 0.0001, asparagine versus oligomycin + asparagine P = 0.0004; malate: vehicle versus oligomycin P = 0.0208 (b); oligomycin versus oligomycin + asparagine P = 0.0072 (c); oligomycin versus oligomycin + asparagine: CTP P = 0.0295, AMP P = 0.0301, ADP P = 0.0082, ATP P = 0.0449, UMP P = 0.0348, UDP P = 0.0035, UTP P = 0.0429, inosine P = 0.0276, IDP P = 0.0101, guanosine P = 0.0015, GTP P = 0.0147 (d); vehicle versus asparagine P < 0.0001, oligomycin versus oligomycin + asparagine P < 0.0001 (e); P = 0.0106 (f); P = 0.0003 (g); siNT + oligomycin versus siAsns + oligomycin inosine P = 0.0007, IMP P = 0.0045, IDP P = 0.006, CMP P = 0.0058, cAMP P = 0.0022, AMP P = 0.0171, ADP P = 0.013, UMP P = 0.0017, dU P = 0.0308 (h); N versus V P < 0.0001 (i); N vehicle versus V vehicle P = 0.0002, N oligomycin versus V oligomycin P < 0.0001 (j); P = 0.027 (k). Source data

Fig. 7. Asparagine rescues the inhibition of respiration in diverse systems and is an exploitable vulnerability in pancreatic cancer.

a, Doubling times of respiration inhibition-insensitive clone V treated with 1.5 nM oligomycin in the presence or absence of medium containing all NEAAs or individual NEAAs (n = 3 samples per condition). b, Relative doubling times of cell lines treated with phenformin with or without 100 µM exogenous asparagine; KPC7940B and HEK-293FT 150 µM, PA-TU-8902 and MIA PaCa-2 37.5 µM; n = 3 samples per condition. c, Human-derived PDA lines UM2, UM18, UM19 and UM53 treated with phenformin in the presence or absence of 100 µM exogenous asparagine (n = 3 samples per condition). d–f, Athymic nude mice were implanted subcutaneously with clone N (d), clone V (e) or a combination of clone N and V (f), and tumors were allowed to establish for 9 d. Mice were then treated with asparaginase, phenformin, asparaginase + phenformin or vehicle until collection (n = 10 tumors per treatment.). g, Representative images (one of three) of frozen sections of vehicle- or phenformin-treated tumors established from co-injection of nuclear red (Nuc-Red)-labeled clone N and cytoplasmic GFP (Cyto-GFP)-labeled clone V; scale bar, 90 µm. h, Quantification of the respective fluorescent labels from g (n = 3 images per treatment group). i, C57BL/6J mice were implanted subcutaneously with syngeneic mouse KPC7940B PDA cells, and tumors were allowed to establish for 9 d. Mice were then treated with asparaginase, phenformin, asparaginase + phenformin or vehicle until collection (n = 10 tumors per treatment group). j,k, Final tumor mass from C57BL/6J mice implanted orthotopically with KPC-7490B (j; n = 7 per treatment group) or KPC-MT3 (k; n = 8 vehicle, n = 7 asparaginase, n = 7 phenformin and n = 9 phenformin + asparaginase tumors per treatment group) syngeneic mouse PDA cells and treated 14 d after establishment with asparaginase, phenformin, asparaginase + phenformin or vehicle until collection 10 d later. Error bars represent mean ± s.d. (a–c and h) and mean ± s.e.m. (d–f and i–k); *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001 by one-way ANOVA with a Tukey post hoc test (a, d–f and i–k) or a two-tailed Student’s t-test (b and h); DMEM versus asparagine P < 0.0001, asparagine versus glutamate P < 0.0001, asparagine versus aspartate P < 0.0001, asparagine versus serine P < 0.0001, asparagine versus proline P < 0.0001 (a); mock versus asparagine KPC7940B P = 0.0377, HEK-293FT P = 0.0006, MIA PaCa-2 P = 0.0204, PA-TU-8902 P = 0.008 (b); mock versus asparagine UM2 P = 0.0377, UM18 P = 0.0017, UM19 P < 0.0001, UM53 P = 0.0032 (c); control versus phenformin P < 0.0001, control versus phenformin + asparaginase P < 0.0001, asparaginase versus phenformin P < 0.0001, asparaginase versus phenformin + asparaginase P < 0.0001 (d); control versus phenformin P = 0.0021, control versus phenformin + asparaginase P < 0.0001, asparaginase versus phenformin + asparaginase P < 0.0001, phenformin versus phenformin + asparaginase P = 0.0029 (e); control versus phenformin P = 0.0043, control versus phenformin + asparaginase P < 0.0001, asparaginase versus phenformin + asparaginase P = 0.0028, phenformin versus phenformin + asparaginase P = 0.028 (f); Nuc-Red vehicle versus Nuc-Red phenformin P = 0.5435 (h); control versus phenformin P = 0.0268, control versus phenformin + asparaginase P < 0.0001, phenformin versus phenformin + asparaginase P = 0.002 (i); control versus phenformin + asparaginase P = 0.007, asparaginase versus phenformin + asparaginase P = 0.1573, phenformin versus phenformin + asparaginase P = 0.0647 (j); control versus phenformin + asparaginase P = 0.0036, asparaginase versus phenformin + asparaginase P = 0.0193, phenformin versus phenformin + asparaginase P = 0.0698 (k). Source data

Extended Data Fig. 1. Enriched Metabolic Pathways Present in Clonal Populations.

A. Agarose gel visualization of PCR products for recombination of Kras^G12D allele in clonal lines vs. Kras^LSL-G12D/+ mouse tail control DNA. B. Top 50 pathways identified by MetaboAnalyst among metabolites differentially represented between group 1 and group 2 clones. C. Relative abundance of metabolites from glycolysis between clonal populations (n = 3 replicates per cell line). F-1,6-BP = fructose 1,6-bisphsophate, GA3P = glyceraldehyde-3-phosphate, DHAP = dihydroxyacetone phosphate. D. Lactate production of clonal cell lines (n = 3 replicates per cell line). E. Relative abundance of glutamine and glutamate between clonal populations (n = 3 replicates per cell line). F. Amniooxyacetic acid (AOA) dose response curves used to generate IC₅₀ values in Fig. 1D. G. Oligomycin dose response curves used to generate IC₅₀ values in Fig. 1E. H. Phenformin dose response curves used to generate IC₅₀ values in Fig. 1F. Each line represents an independent dose response analysis. Error bars are mean ± SD, *** P ≤ 0.001; **** P ≤ 0.0001 by two-tailed Mann–Whitney test. Source data

Extended Data Fig. 2. Clonal Metabolic Phenotypes.

A. Citrate Synthase (CS) activity of clonal lines (n = 3 per clone). B. MitoStress profile of the oxygen consumption rate (OCR) over time, treated with (2 μM), FCCP (0.5-2 μM), and rotenone/antimycin A (1 μM) at the indicated time points. Data are average of technical replicates normalized to independent experiments for each clone (n = 4 clone V, n = 4 clone E, n = 3 clone H, n = 5 clone N, n = 4 clone K, n = 2 clone M, n = 4 clone T). C. Metabolic phenotypes of clonal cell lines defined as the ratio of the OCR to extracellular acidification rate (ECAR) (n = 4 clone V, n = 4 clone E, n = 3 clone H, n = 4 clone K, n = 2 clone M, n = 5 clone N, n = 4 clone T). D. Basal ECAR of clonal lines (n = 4 clone V, n = 4 clone E, n = 3 clone H, n = 4 clone K, n = 2 clone M, n = 5 clone N, n = 4 clone T). E. Basal OCR of clonal lines (n = 4 clone V, n = 4 clone E, n = 3 clone H, n = 4 clone K, n = 2 clone M, n = 5 clone N, n = 4 clone T). F. Metabolic fitness as defined by the spare respiratory capacity of clonal cell lines (n = 4 clone V, n = 4 clone E, n = 3 clone H, n = 4 clone K, n = 2 clone M, n = 5 clone N, n = 4 clone T). G. Histogram representation of the MitoTracker Red intensity of group 1 clones (H,V) and group 2 clones (T,N). H. Representative images of MitoTracker Red staining at the mean fluorescence index (MFI) of I. Scale bar = 7 µm. J. Histogram representation of the MitoTracker Green intensity of group 1 clones (H,V) and group 2 clones (T,N). K. Doubling times of the individual clone lines (n = 3 per clone). Error bars are mean ± SD, * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P ≤ 0.0001 by two-tailed Mann–Whitney test. Source data

Extended Data Fig. 3. Co-Culture Metabolic Assays.

A. Fluorescently labeled clone K co-cultures were treated with oligomycin or vehicle. B. Representative images from Fig. 2D of labeled K co-cultured with unlabeled resistant clones (H,V) or with unlabeled sensitive clones (K,M). C. Fluorescently labeled clone N co-cultures were treated with oligomycin, phenformin or vehicle. D,E. Representative images of 1 nM oligomycin or vehicle treated co-cultures with unlabeled resistant V (D), or unlabeled sensitive clones N (E) to generate data presented in Fig. 2E. F. Fluorescently labeled clone V co-cultures were treated with 1.5 nm oligomycin or vehicle. G. Representative images from Fig. 2F of labeled V co-cultured with unlabeled resistant clones (V,H) or with unlabeled sensitive clones (N,M). Error bars are mean ± SD, Scale Bars = 400 µm, * P ≤ 0.05 by one-way Anova with Tukey post hoc.

Extended Data Fig. 4. Consistent Metabolic Features in a Second PDA Clonal Cell Model and Media Metabolite Profiling.

A. Dose response curves of oligomycin treated KPC clones 2838C3, 6499c4, 6419c5, and 6694c2 (n = 3 per clone). B. Lactate production by clones 2838c3 and 6499c4 (n = 3 per clone). C. Group 2 clone 2838c3 was labeled with a fluorescent probe and co-cultured with either unlabeled 2838c3 or with unlabeled group 1 clone 6499c4 and then treated with either phenformin, oligomycin, or vehicle. D. Viability of 5 nM oligomycin or 100 µM phenformin treated co-cultures relative vehicle (n = 5). E. Representative images from Extended Data Fig. 4D. F. Heatmap of unsupervised 2-dimensional hierarchical clustering for metabolites from conditioned media after 48 hours of culture with ±2 fold difference and p = 0.01 between groups (n = 3 per clone). Error bars are mean ± SD, Scale Bars = 400 µm, ** P ≤ 0.01 by two-tailed student’s t test. Source data

Extended Data Fig. 5. Asparagine Rescues the Inhibition of Respiration.

A. Relative confluence of clone N treated with 1 nM oligomycin in the presence or absence of media containing a 100 µM cocktail of all NEAAs after 5 days (n = 3). B. Relative confluence of clone N treated with 1 nM oligomycin in the presence or absence of individual NEAAs after 5 days (n = 3). C,D. Dose dependent rescue of doubling time (C) and confluence after 5 days of culture (D) by exogenous asparagine (Asn) following treatment with 1 nM oligomycin. E. Relative confluence of clone N treated with 1 nM oligomycin in the presence or absence of 100 µM or 20 mM aspartate (n = 3). F. Relative confluence of clonal line N treated with 1 nM oligomycin, 25 µM phenformin, or 25 nM IACS-10759 after 5 days + /- Asn (n = 3). G. Immunoblot analysis of siRNA targeting Asns or non-targeted (NT) control in clonal line N or V. H,I. Representative images of vehicle vs 1 nM oligomycin treated co-cultures of labeled clone N with unlabeled clone N (H) or insensitive clone V (I) transfected with the indicated siRNA with or without the addition of exogenous Asn. Error bars are mean ± SD, Scale Bars = 400 µm, * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P ≤ 0.0001 by two-tailed student’s t test (A,F), or one-way Anova with Tukey post hoc (B,E). Source data

Extended Data Fig. 6. Transcriptional Programs and Stress Response in PDA Populations.

A. Gene Set Enrichment Analysis (GSEA) of pathways differentially expressed between group 1 and group 2 clones. Negative NES scores indicate pathways enriched in group 1 and positive NES scores are pathways enriched in group 2. B. Immunoblot analysis of Slug and Hypoxia Inducible Factor 1 Subunit Alpha (HIF-1α) across group 1 clones (E,V,H) vs. group 2 (K,M,N,T). C. Immunoblot analysis of integrated stress response (ISR) activation proteins general control nonderepressible 2 (GCN2), protein kinase RNA-like ER kinase (PERK), protein kinase RNA-activated (PKR), and heme-regulated initiation factor 2 alpha kinase (HRI) across group 1 clones (E,V,H) vs. group 2 (K,M,N,T). D. Immunoblot analysis of ER stress and the unfolded-protein response (UPR) pathway activation proteins inositol-requiring enzyme 1 α (IRE1α) and activation transcription factor 6 (ATF6) across group 1 clones (E,V,H) vs. group 2 (K,M,N,T). E. Immunoblot analysis of ATF4 of insensitive clone V and sensitive clone N treated with DMSO, and insensitive clone V treated with 1, 2, and 5 µM of ISRIB or GCN2iB. F. Representative images from Fig. 4G of clonal line N was encoded with a fluorescent label, plated in direct co-culture with unlabeled sensitive clone N or insensitive clone V, and then treated with oligomycin, ISRIB, GCN2iB, oligomycin and ISRIB, oligomycin and GCN2iB, or vehicle. Scale bar = 1 mm. Source data

Extended Data Fig. 7. Interrogation of Human PDA Clonality.

A. Immunostaining for ASNS in human PDA patient tissues, dashed boxes magnified in insets. Scale bars = 100 µm. B. Uniform Manifold Approximation and Projection (UMAP) representation of 3 epithelial cell clusters identified from single cell RNA analysis of a tumor biopsy of PDA Patients 1238 and 1314, with non-hierarchical clustering analysis of epithelial clusters across a set of integrated stress response (ISR) genes. C. Representative images from Fig. 5H. Clonal line PATC53-23 was encoded with a fluorescent label, plated in direct co-culture with unlabeled clones PATC53-23 or PATC53-34, then treated with 0.5 nM oligomycin or 125 µM phenformin relative to vehicle control. Scale bar = 1 mm.

Extended Data Fig. 8. Impacts of Asparagine Rescue on Metabolism.

A. Immunoblot analysis of Vinculin and Cleaved Caspase 3 across 1 nM treated oligomycin group 1 clones (E,V,H) vs. group 2 (K,M,N,T). Staurosporine treated clonal line N is included as a positive control. B. Non-hierarchical clustering heatmap of metabolites that differed between experimental conditions (p = 0.1) from clone N cells treated with asparagine, oligomycin, asparagine + oligomycin, or vehicle 4 hours post treatment (n = 3). C-H. Isotopic labeling patterns in clones N and V using media supplemented with universally-labeled ¹³C-asparagine (Asn), ¹³C-glucose, or ¹³C-glutamine, relative to unlabeled controls, for asparagine (C, n = 3), aspartate (D, n = 3), citrate (E, n = 3), or uridinylates (F-H, n = 3). I. Isotopologue abundance of aspartate from clone N pre-labeled with universally labeled ¹³C-glutamine, then switched to media containing unlabeled-glutamine and treated with asparagine, oligomycin, asparagine + oligomycin, or vehicle for 4 hours (n = 3). Error bars are mean ± SD. Source data

Extended Data Fig. 9. Time Course Metabolomics.

Non-hierarchical clustering heatmap of metabolites that differed between experimental conditions (p = 0.1) from sensitive clone N or insensitive clone V cells treated with asparagine, oligomycin, asparagine + oligomycin, or vehicle and harvested at 1, 2, 4, 8, and 12 hours post treatment (n = 3).

Extended Data Fig. 10. Asparagine Rescue and Tumor Study Controls.

A. Isotopologue abundance of aspartate from sensitive clone N and insensitive clone V treated for 4 hours with oligomycin or vehicle, then switched to media containing with universally labeled ¹³C-glutamine and treated with oligomycin or vehicle for an additional 4 hours (n = 3). B. Relative doubling time of clone N cells treated with 1 nM oligomycin with or without the addition of exogenous asparagine, pyruvate, or alpha-ketobutyrate (AKB) (n = 3). C. Dose response curve of clone N cells treated with oligomycin in the presence of absence of 100 µM exogenous nucleosides (n = 3). D. Plate confluence at endpoint of respiration-inhibition insensitive clone V treated with 1.5 nM oligomycin in the presence of absence of media containing all NEAAs or individual NEAAs (n = 3). E. Plate confluence at endpoint of cell lines treated with phenformin with or without 100 µM exogenous asparagine; KPC7940B and HEK-293FT 150 µM, PA-TU-8902 and MIA PaCa-2 37.5 µM (n = 3). F. Relative metabolite levels of asparagine in the serum and tumor lysate of mice bearing orthotopic KPC-7940B treated with vehicle, phenformin, asparaginase, or phenformin + asparaginase (n = 5 each). G. H&E staining of pancreas from WT C57BL/6 J treated with either vehicle or asparaginase. H-J. Dose response curves across clonal lines of the Stearoyl-CoA desaturase-1 (SCD) inhibitor CAY10566 (H, n = 3), Fatty Acid Synthase (FASN) inhibitor TVB-2640 (I, n = 3), and ATP citrate lyase inhibitor (ACLY) BMS303-141 (J, n = 3). Scale bar = 100 µm. Error bars are mean ± SD, * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P ≤ 0.0001 by two-tail student’s t test (E) or one-way Anova with Tukey post hoc (B,D,F). Source data