Combined KRAS-MAPK pathway inhibitors and HER2-directed drug conjugate is efficacious in pancreatic cancer

Abstract

Targeting the mitogen-activated protein kinase (MAPK) cascade in pancreatic ductal adenocarcinoma (PDAC) remains clinically unsuccessful. We aim to develop a MAPK inhibitor-based therapeutic combination with strong preclinical efficacy. Utilizing a reverse-phase protein array, we observe rapid phospho-activation of human epidermal growth factor receptor 2 (HER2) in PDAC cells upon pharmacological MAPK inhibition. Mechanistically, MAPK inhibitors lead to swift proteasomal degradation of dual-specificity phosphatase 6 (DUSP6). The carboxy terminus of HER2, containing a TEY motif also present in extracellular signal-regulated kinase 1/2 (ERK1/2), facilitates binding with DUSP6, enhancing its phosphatase activity to dephosphorylate HER2. In the presence of MAPK inhibitors, DUSP6 dissociates from the protective effect of the RING E3 ligase tripartite motif containing 21, resulting in its degradation. In PDAC patient-derived xenograft (PDX) models, combining ERK and HER inhibitors slows tumour growth and requires cytotoxic chemotherapy to achieve tumour regression. Alternatively, MAPK inhibitors with trastuzumab deruxtecan, an anti-HER2 antibody conjugated with cytotoxic chemotherapy, lead to sustained tumour regression in most tested PDXs without causing noticeable toxicity. Additionally, KRAS inhibitors also activate HER2, supporting testing the combination of KRAS inhibitors and trastuzumab deruxtecan in PDAC. This study identifies a rational and promising therapeutic combination for clinical testing in PDAC patients.

Fig. 1. Targeting the MAPK pathway is inadequate and compromised by HER2 activation.

a Western blots showing dose-dependent increases in phosphorylated ERK1/2, p90RSK, and cleaved caspase-3 levels in PDAC cells treated with the indicated concentrations of gemcitabine for 24 h. b Western blots showing changes in p-ERK1/2, p-AKT(S473), and cleaved caspase-3 levels after treatment with the indicated agents for 24 h. c Correlation plots with Pearson coefficients (R) showing strong positive correlation between MAPK and two independent gemcitabine resistance signatures in PDAC samples from TCGA PanCancer database. d Growth kinetics of subcutaneous Pa01c and Pa02c xenograft tumours treated as indicated when the tumour volume reached ~100 mm³. Data are presented as mean ± SEM. P-values were calculated using two-way ANOVA with Tukey’s multiple comparison test. e Heatmap of RPPA data showing significantly upregulated and downregulated markers in Pa01c and HPAC cells treated with ulixertinib or trametinib for 24 h. f Venn diagram showing the shared changes in both cell lines. Only markers showing a Log2 fold change or <-1 or >1 are illustrated. Data are presented as the mean ± SEM of two biological samples. g Western blots showing changes in p-ERK1/2, DUSP4, and DUSP6 levels in 293 T cells transfected with HER1, HER2, or HER3 for 36 h and then treated with DMSO or afatinib for 16 h. h Western blots showing changes in different phosphorylated-HER2 signals, p-HER3, p-ERK1/2, p-AKT, DUSP4, and DUSP6, in Pa01c and HPAC cells treated with ulixertinib or trametinib for 24 h. a, b, g, h were conducted two times, and one set of data was presented. Source data are provided in Source Data file.

Fig. 2. Proteasomal degradation of DUSP6 sustains HER2 phospho-activation.

a Western blots showing changes in the indicated DUSPs following treatment with trametinib or ulixertinib overnight (~16 h) in the two different PDAC lines. b Immunoprecipitation (IP) experiment showing polyubiquitination of stably expressed FLAG-tagged DUSP4 and DUSP6 in 293 T cells following 16 h treatment with ulixertinib followed by co-treatment with DMSO or bortezomib for 6 h. c Western blots showing changes in p-HER2 (Y1248), p-ERK1/2, DUSP4, and DUSP6 levels in Pa01c and HPAC cells treated with trametinib or ulixertinib for 16 h, followed by DMSO or bortezomib for 6 h. d Western blots showing changes in p-HER2 and p-ERK1/2 in Pa01c and HPAC cells stably expressing scramble control of two different shRNAs against DUSP4 or DUSP6. e Western blots showing changes in p-HER2 and p-ERK1/2 in Pa01c and HPAC cells stably overexpressing an empty vector, DUSP6, DUSP4, or DUSP7. f Western blots showing changes in p-HER2 and p-ERK1/2 levels in Pa01c and HPAC cells stably expressing wild-type (WT) or enzymatically inactive (C293S) DUSP6. All experiments were conducted two times, and one set of data for each was presented. Source data are provided in Source Data file.

Fig. 3. The TEY motif of HER2 promotes binding to DUSP6.

a Alignment of amino acid sequences of the C-termini of ERK1, ERK2, HER1, HER2, and HER3, indicating the presence of the TEY motif in ERK1/2 and HER2. Schematic and results of IP experiments using HA-DUSP6 produced in human Pa01c cells and ERK2/HER2 variants using BL21 bacteria. IP was performed using HA beads (b) or His beads (c) to delineate the in vitro interaction between DUSP6 and HER2. d In vitro assay showing decrease phospho-HER2(Y877) of purified bead-bound FLAG-HER2 incubated with recombinant DUSP6 for 1 h. e In vitro pNPP phosphatase assay using recombinant DUSP6 and purified bead-bound FLAG-HER2 done in quadruplicates for 30 min at 37 °C in cell-free condition. The liberated p-nitrophenol which reflects DUSP6 phosphatase activity was measured at 405 nm using Biotek Synergy 2 spectrophotometer. f Representative immunofluorescence images and quantification of Duolink® proximity ligase assay identifying the interaction of endogenous HER2 and DUSP6 in Pa01c and HPAC cells treated with DMSO, ulixertinib 2 μM, or trametinib 1 μM for 4 h prior to degradation of DUSP6. Scale bar = 50 μm. For (e, f) data presented as mean ± SEM, P values were calculated by one-way ANOVA with Dunnet’s multiple comparisons test or (e) two-tailed unpaired t-test (f). g IP experiment in 293 T cells showing polyubiquitination of ectopically expressed HA-tagged DUSP6 in the absence or presence of co-transfected FLAG-tagged HER2 as well as trametinib or ulixertinib treatment for 16 h. All cells were co-treated with bortezomib for at least 4 h prior to treatment with trametinib or ulixertinib to prevent the marked downregulation of DUSP6. Experiments in (b–d, g) were conducted two times, and one set of data for each was presented. Source data are provided in Source Data file.

Fig. 4. KRAS regulates DUSP6 stability.

a Western blots showing changes in p-ERK1/2, DUSP4, and DUSP6 levels in 293 T cells stably expressing HA-KRAS^G12V treated with trametinib or ulixertinib for 16 h. The bar graph includes quantification of three independent experiments. Data was presented as mean ± SEM. P-values were calculated using one-way ANOVA followed by Dunnet’s multiple comparisons test, HA-KRAS vs control (p = 0.0031), HA-KRAS vs 2 μM ulixertinib (p = 0.0474), HA-KRAS vs 0.5 μM trametinib (p = 0.0468), HA-KRAS vs 1 μM trametinib (p = 0.0045). Western blots showing serial changes in endogenous DUSP6 protein levels in 293 T cells stably expressing an empty vector or HA-KRAS^G12V (b) or in Pa01c cells stably expressing a scramble sequence or an shRNA against KRAS (c), after treatment with 10 μg/mL cycloheximide (CHX) for the indicated durations. d Western blots, and quantification of serial endogenous DUSP6 protein levels in Pa01c cells stably expressing a scramble sequence or an shRNA against KRAS (c), after treatment with 2 μM of ulixertinib for the indicated durations. e Western blots, and quantification of serial endogenous DUSP6 in Pa01c cells pre-treated with DMSO or MRTX1133 (0.5 μM) for 1 h followed by 2 μM of ulixertinib at the indicated durations. f Western blots, and quantification of serial endogenous DUSP6 in MIA Paca-2 cells pre-treated with DMSO or AMG-510 (0.5 μM) for 1 h, followed by 2 μM of ulixertinib for the indicated durations. For (b–f), half-lives (t_1/2) of DUSP6 were calculated by measuring the DUSP6 band intensities, normalizing to t₀ and performing one-phase exponential decay analysis, as shown in the graph below. a was conducted three times, and (b–f) were conducted two times, and one set of data for each was shown. Source data are provided in Source Data file.

Fig. 5. KRAS recruits TRIM21 to regulate DUSP6 stability.

a Schematic of the immunoprecipitation followed by mass spectrometry (IP-MS) experiment employed to identify binding partners of FLAG-KRAS^G12V. 3 biological replicates from each condition were pooled for mass spectrometry. Linear plot shows the enriched proteins. b IP western blots showing the interaction between HA-KRAS^G12V and endogenous TRIM21 and DUSP6 in 293 T cells stably expressing vector or HA-KRAS^G12V treated with DMSO, trametinib, or ulixertinib for 16 h. The bar graph includes quantification of three independent experiments. Data was presented as mean ± SEM. P-values were calculated using one-way ANOVA followed by Dunnet’s multiple comparisons test. c Western blots, and quantification of serial endogenous DUSP6 protein levels in 293 T cells expressing empty vector or GFP-tagged TRIM21 following treatment with CHX (10 µg/mL) for the indicated durations. d IP experiment showing differences in K48-polyubiquitination of FLAG-tagged DUSP6 in 293 T cells transfected with the vector, wild-type, or enzymatically inactive (C54Y) TRIM21. e IP experiment showing differences in K48-polyubiquitination of FLAG-tagged DUSP6 with/without TRIM21 co-expression following 16 h treatment with DMSO, ulixertinib, or trametinib in 293 T cells. f Western blots showing changes in p-HER2, p-ERK1/2, and DUSP6 in Pa01c and HPAC cells stably expressing a scramble shRNA or two different shRNAs against TRIM21. b was conducted three times, (c–f) were conducted two times, and one set of data for each was shown. g 2-dimensional colony formation assay showing the colony-forming ability of the indicated Pa01c and HPAC cells co-treated with DMSO or afatinib for 2 weeks. h GSEA plots showing the enrichment of ERBB2 and KRAS signatures in TRIM21-silenced HPAC cells subjected to bulk RNA sequencing. i proposed model of the mechanism by which KRAS/MAPK inhibitors dissociate TRIM21 from DUSP6 and causes the latter to be destabilized, leading to sustained phospho-activation of HER2. Phospho-activated HER2 may in turn promotes DUSP6 degradation while simultaneously activate other survival cascades such as the PI3K-AKT pathway. Source data are provided in Source Data file.

Fig. 6. Targeted MAPK-based combinations require cytotoxic chemotherapy to achieve meaningful therapeutic efficacy.

a Growth kinetics of the indicated PDCL subcutaneous xenograft tumours treated as indicated when the tumours reached 50–100 mm³ in NOD-SCIDγ mice. Dosages of each agent were: ulixertinib 100 mg/kg BID orally 5 days/week, afatinib 12.5 mg/kg orally daily, copanlisib 10 mg/kg by tail vein injection two times/week, gemcitabine 75 mg/kg by intraperitoneal injection once per week. N = 4 tumours per treatment. Data was presented as mean ± SEM. P-values were calculated using one-way ANOVA with Dunnet’s multiple comparisons test. b Clinical and genomic profiles and ERBB2 mRNA expression in PDAC PDX models used in this study. c Waterfall plot summarizing changes of tumour volume from baseline for all 16 PDX models and determination of treatment response using clinical RECIST 1.1. P-values were calculated using Brown-Forsythe and Welch ANOVA test followed by Dunnett’s T3 multiple comparisons test. d Paired data plots showing changes in the body weight of mice treated with different regimens for 4 weeks. P-values were calculated by two-tailed paired t-test. Loss in body weight after two weeks of treatment as indicated. All data are provided in the Source Data file.

Fig. 7. MAPK inhibitors upregulate HER2 expression.

a Representative IHC images showing changes in HER2 protein expression in the indicated PDAC xenografts using a CLIA-certified HER2 staining kit routinely used for the analysis of gastric cancer samples, Scale bars = 20 μM. b Representative FACS plots and quantification showing changes in surface (without cell permeabilization) and total (with cell permeabilization) HER2 abundance following 16 h treatment as indicated (ulixertinib 2 μM, trametinib 0.5 μM, DS-8201a 0.1 μg/ml). Data represents one of three independent experiments each done in triplicates. Data are presented as the mean ± SEM. P-values were calculated using unpaired t-test with Welch’s correction or ANOVA followed by Tukey’s test. c Representative IF images showing increased surface and total HER2 expression (red) in Pa01c cells treated for 16 h with the indicated agents as (b). LysoTracker Green DND-26 was used to stain the endolysosomes, Scale bars = 50 μM. d Western blots showing suppression of MAPK activity and upregulation of DUSP6 in Pa01c cells after treatment with vehicle or DS-8201a 0.1 μg/ml for 16 h. d was conducted two times, and one set of data was presented. All data are provided in Source Data file.

Fig. 8. MAPK inhibitors plus antibody-drug conjugate trastuzumab deruxtecan (DS-8201a) lead to deep and durable treatment response.

a Growth kinetics of the subcutaneous PDAC tumours treated as indicated when tumours reached ~100 mm³ in NOD-SCIDγ mice. WU0002, WU0006, WU0022 Capan-1, Pa02c, Pa14c & Pa16c (DS-8201a + ulixertinib vs ulixertinib or DS-8201a + trametinib vs trametinib: p < 0.0001), for WU0009 (trametinib vs vehicle: p = 0.9998, DS-8201a vs vehicle: p = 0.0043 & DS-8201a + trametinib vs trametinib: p = 0.0043), for Pa01c (DS-8201a + ulixertinib vs ulixertinib: p = 0.0027 & DS-8201a + trametinib vs trametinib ulixertinib: p < 0.0001). b Waterfall plot summarizing changes of tumour volume for different PDAC models treated as indicated. Comparison to the baseline was made using the best response for combo-treated mice or when mice were euthanized. The treatment response was determined using the clinical RECIST 1.1 criteria. Each bar represents the average change in volume from replicates of a different model. P-values calculated using one-way ANOVA with Tukey’s multiple comparison test. c Growth kinetics of the subcutaneous PDAC models treated as indicated when tumours reached between 50–100 mm³ or above in NOD-SCIDg mice. a–c data are presented as the mean ± SEM. P-values were calculated using ANOVA followed by Tukey’s test. Source data are provided in the Source Data file.

Fig. 9. KRASG12C inhibitor plus trastuzumab deruxtecan showed promising preclinical efficacy.

a Representative FACS plots and quantification showing changes in surface and total HER2 abundance following 16 h treatment, as indicated in three different KRAS^G12C-mutant cell lines. (AMG-510 0.5 μM, DS-8201a 0.1 μg/ml). Data represents one of three independent experiments each done in triplicates. Data are presented as the mean ± SEM. P-values were calculated using one way ANOVA followed by Tukey’s multiple comparisons test. NCL-H2122 Surface HER2 (AMG-510 vs vehicle: p < 0.0001, DS-8201a vs vehicle: p < 0.0001, DS-8201a + AMG-510 vs AMG-510: p < 0.0001 & DS-8201a + AMG-510 vs DS-8201a: p = 0.0120), and Total HER2 (AMG-510 vs vehicle: p = 0004, DS-8201a vs vehicle: p < 0.0001, DS-8201a + AMG-510 vs AMG-510: p < 0.0001 and DS-8201a + AMG-510 vs DS-8201a: p = 0.8964); NCL-H2030 Surface HER2 (AMG-510 vs vehicle: p = 0.0006, DS-8201a vs vehicle: p < 0.0001, DS-8201a + AMG-510 vs AMG-510: p < 0.0001 & DS-8201a + AMG-510 vs DS-8201a: p = 0041), and Total HER2 (AMG-510 vs vehicle: p = 0.3401, DS-8201a vs vehicle: p < 0.0001, DS-8201a + AMG-510 vs AMG-510: p < 0.0001 and DS-8201a + AMG-510 vs DS-8201a: p = 0.9033; MIA Paca-2 Surface HER2 (AMG-510 vs vehicle: p = 0.0047, DS-8201a vs vehicle: p < 0.0001, DS-8201a + AMG-510 vs AMG-510: p < 0.0001 & DS-8201a + AMG-510 vs DS-8201a: p = 0.2871), and Total HER2 (AMG-510 vs vehicle: p = 0.0030, DS-8201a vs vehicle: p < 0.0001, DS-8201a + AMG-510 vs AMG-510: p < 0.0001 and DS-8201a + AMG-510 vs DS-8201a: p = 0.9985). b Median effect analyses of AMG-510 in combination with DS-8201a in three KRAS^G12C-mutant cell lines, as represented by combination indices (CI) calculated using Compusyn software. Cells were cultured in triplicate at six fixed-ratio concentrations (1:1, 0.5:0.5, 0.25:0.25, 0.125:0.125, 0.063:0.063, and 0.031:0.031) for 3 days, and viability was measured using the Alamar Blue assay. c Waterfall plot summarizing changes in tumour volume for each MIA Paca-2 tumour. All mice were euthanized simultaneously when vehicle-treated mice reached the humane endpoints, and comparison to the baseline was made. The treatment response was determined using the clinical RECIST 1.1 criteria. Each bar represents an individual tumour. N = 7/arm. P-values were calculated using one-way ANOVA followed by Tukey’s multiple comparison test. AMG-510 vs vehicle (p = 0.0007), DS-8201a vs vehicle (p = 0.0002), DS-8201a + AMG-510 vs vehicle (p < 0.0001), DS-8201a + AMG-510 vs DS-8201a (p = 0084) and DS-8201a + AMG-510 vs DS-8201a (p = 0022).

Fig. 10. KRASG12D inhibitor plus trastuzumab deruxtecan showed promising preclinical efficacy.

a Representative IF images showing increased surface and total HER2 expression (red) in Pa01c cells treated for 16 h with the indicated agents. LysoTracker Green DND-26 was used to stain the endolysosomes. (MRTX1133 0.5 μM, DS-8201a 0.1 μg/ml). Scale bars = 20 μM. b Representative FACS plots and quantification showing changes in surface (without cell permeabilization) and total (with cell permeabilization) HER2 abundance following 16 h treatment, as indicated in two different KRAS^G12D-mutant cell lines. (MRTX1133 0.5 μM, DS-8201a 0.1 μg/ml). Data represents one of three independent experiments each done in triplicates. Data are presented as the mean ± SEM. P-values were calculated using one way ANOVA followed by Tukey’s multiple comparisons test. Pa01c Surface HER2: MRTX1133 vs vehicle (p < 0.0001), DS-8201a vs vehicle (p < 0.0001), DS-8201a + MRTX1133 vs MRTX1133 (p < 0.0001) & DS-8201a + MRTX1133 vs DS-8201a (p = 0.1598), and Total HER2: MRTX1133 vs vehicle (p = 0.0008), DS-8201a vs vehicle (p < 0.0001), DS-8201a + MRTX1133 vs MRTX1133 (p < 0.0001) and DS-8201a + MRTX1133 vs DS-8201a (p = 0.1215). HPAC Surface HER2: MRTX1133 vs vehicle (p < 0.0001), DS-8201a vs vehicle (p < 0.0001), DS-8201a + MRTX1133 vs MRTX1133 (p < 0.0001) & DS-8201a + MRTX1133 vs DS-8201a (p = 0.1502), and Total HER2: MRTX1133 vs vehicle (p < 0.0001), DS-8201a vs vehicle (p = 0.0055), DS-8201a + MRTX1133 vs MRTX1133 (p < 0.0001) and DS-8201a + MRTX1133 vs DS-8201a (p = 0.1926). c Waterfall plot summarizing changes in tumour volume for each Pa01c tumour, as indicated. All mice were euthanized simultaneously when vehicle-treated mice reached the humane endpoints, and comparison to the baseline was made. The treatment response was determined using the clinical RECIST 1.1 criteria. Each bar represents an individual tumour. N = 7. P-values were calculated using one-way ANOVA followed by Tukey’s multiple comparison test. MRTX1133 vs vehicle (p < 0.0001), DS-8201a vs vehicle (p = 0.0001), DS-8201a + MRTX1133 vs vehicle (p < 0.0001), DS-8201a + MRTX1133 vs MRTX1133 (p = 0.0867) and DS-8201a + MRTX1133 vs DS-8201a (p = 0083). Source data are provided in Source Data file.