PKM2 Confers Purine Metabolic Advantage and Enhances Lenvatinib Resistance in Hepatocellular Carcinoma

doi:10.21203/rs.3.rs-3228233/v1

Download PDF

Article

PKM2 Confers Purine Metabolic Advantage and Enhances Lenvatinib Resistance in Hepatocellular Carcinoma

https://doi.org/10.21203/rs.3.rs-3228233/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Acquired resistance largely impedes the clinical efficacy of lenvatinib in hepatocellular carcinoma (HCC). Despite the identification of several underlying drivers, the comprehensive mechanism underlying lenvatinib resistance remains largely unknown. In the present study, we establish a lenvatinib-resistant model (Len-R) in HCC patients-derived organoids to simulate the emergence of resistant tumors. We demonstrate that upregulation of PKM2 is a hallmark of Len-R and PKM2 directly contributes to lenvatinib resistance. We also identify that purine metabolic flux enriched in Len-R confers the resistance property to tumor cells in a PKM2-dependent manner. For the molecular mechanism, PKM2 interacts directly with SMAD4 and enhances its occupancy on the SMAD4 binding element (SBE) at DHFR promoter region. Consequently, upregulated DHFR enhances purine nucleotide pool to resist the effectiveness of lenvatinib on purine metabolism pathway. Clinically, upregulation of PKM2 together with DHFR is associated with lenvatinib resistance and worse outcome in HCC patients. In summary, our data suggest that PKM2-SMAD4-DHFR axis drives lenvatinib resistance in HCC by provoking de novo purine biosynthesis.

Biological sciences/Cell biology/Mechanisms of disease

Biological sciences/Cancer/Tumour biomarkers

PKM2

DHFR

Purine metabolism

Lenvatinib resistance

HCC

Acquired resistance is the major cause of the failure of clinical effectiveness of targeted therapy. Lenvatinib, a multi-target tyrosine kinase inhibitor, is a first-line approved drug against advanced HCC. REFLECT trial indicated that lenvatinib was non-inferior to sorafenib in overall survival, and even showed superiority regarding the secondary efficacy endpoints [1]. However, the success has been limited by the emergence of drug resistance. Recent studies on this issue revealed that expression levels of the targets of lenvatinib (VEGFR2, FGFR1) may contribute to drug resistance [2, 3]. Activation of canonical driver pathways in HCC such as EGFR, HGF/c-MET axis and CD73, as compensable pathways, may also sustain cell survival under the lenvatinib treatment pressure [4, 5, 6]. Considering the discrepancy between immortalized cell models and in vivo tumor characteristics [7], there is a pressing need to explore lenvatinib resistance utilizing model that can largely recapture the clinical characteristics of tumor, such as patients-derived organoids model [8].

Pyruvate kinase M2 (PKM2), an alternative splicing isoform of PK gene, which catalyzes Phosphoenolpyruvate and ADP to Pyruvate and ATP in the final step of glycolysis, is usually upregulated in tumor tissues. The lower catalysis activity of PKM2 comparing to PKM1 allows accumulation of upstream metabolic intermediates and promotes metabolic flux step into nucleic acid biosynthesis. Apart from its well-known role in glycolysis, the non-metabolic function of PKM2 as a coactivator involves in regulating tumor metabolic reprogramming [9, 10]. PKM2 can translocate into nucleus dependent on EGFR activation and interact with β-catenin to transactivate c-MYC expression, thereby promote the Warburg effect by upregulating glycolytic gene expression (GLUT1, PKM2, LDHA) [9, 11]. Upregulated expression level of PKM2 by c-MYC offers a positive feedback loop in the regulatory pathway [12]. Moreover, PKM2 is a coactivator of HIF-1α to promote its transactivation and reprogram glucose metabolism in cancer cells. Conversely, HIF-1α directly binds to hypoxia response element at PKM2 promoter region to enhance the transcriptional regulation of the latter [10]. Tumor metabolic reprogramming regulated by PKM2 confers rapid growth advantage to cancer cells and promotes tumorigenesis and drug resistance. Previous studies have reported the close association between PKM2 and chemoresistance of tyrosine kinase inhibitors (sorafenib [13], gefitinib [14], erlotinib [15]) as well as nucleic acid synthesis inhibitor (gemcitabine [16], 5-Fu [17], cisplatin [18]) in multiple types of cancer.

High throughput purine nucleotides are necessary to support cell proliferation in cancer cells [19]. Purine metabolism includes de novo and salvage pathways, and liver is the main organ for de novo purine metabolism biosynthesis. The conserved de novo biosynthesis pathway consists of 10 steps transforming 5-phosphoribosyl 1-pyrophosphate to IMP, while the salvage pathway requires 5-phosphoribosyl 1-pyrophosphate and free bases (hypoxanthine, adenine, and guanine) to generate IMP, AMP and GMP, respectively. Purine nucleotides, products of purine metabolism, exert a key role in building blocks of DNA and RNA, energy carriers, and cell signaling molecules [20]. Various of metabolic precursors participate in purine metabolism biosynthesis, including amino acids (glutamine, aspartic acid, and glycine), Nicotinamide Adenine Dinucleotide, and small molecule cofactors N10-formyl Tertrahydrofolate [21]. Particularly, N10-formyl Tertrahydrofolate is a key donor of one-carbon unit for de novo purine metabolism biosynthesis, which is catalyzed by dihydrofolate reductase (DHFR) from dihydrofolate, and contributes to the production of purine nucleotides in cancer cells.

In the present study, we delineate that PKM2 mediates purine metabolism via transcriptional upregulation of DHFR, and facilitates the resistance to lenvatinib in HCC. PKM2 and purine metabolism flux are highly enriched in Len-R organoids, and the alteration of PKM2 expression or supplementation of purine nucleotides changes the sensitivity to lenvatinib. Interaction of PKM2 with SMAD4 enhances the transactivation of the latter. Consequently, upregulated DHFR increases purine metabolic flux in Len-R. Our results suggest a novel molecular mechanism of lenvatinib resistance mediated by the coactivator function of PKM2, which might be serves as a potential target for combinational therapy.

Human HCC tumor specimens

Fresh resected tumors were collected with informed consent from patients who were enrolled at Eastern Hepatobiliary Surgery Hospital (Shanghai, China) without preoperative treatment. Human specimen collection was approved by the Ethics Committee of Eastern Hepatobiliary Surgery Hospital. Clinical information is available in Supplementary Table.1. Each tumor specimen was divided into three pieces for organoids culture, histology analysis and WGS analysis, separately.

Tumor dissociation

Fresh resected tissue was washed with PBS (Gibco), cut into pieces in collagenase D (Roche; final concentration was 4 mg/mL), and incubated at 37°C with rotation. Incubation time was ranged from 30 to 90 min. The solvent for collagenase D consisted of Advanced DMEM/F-12 (Gibco), GlutaMAX (Gibco), HEPES (Gibco), Penicillin-Streptomycin (Gibco), 100 µg/mL primocin (Invivogen), 0.1 mg/mL DNase I (Sigma-Aldrich) and 10µM Y-27632 (Sigma-Aldrich). After the tissue digestion, DMEM-HG (Gibco) medium containing 10% FBS (Gibco), primocin, Y-27632 and Penicillin-Streptomycin was added to inactivate collagenase D. Cell suspension was filtered using a 70 µm Nylon cell strainer and centrifugated for 5 min at 300–400×g. Cell pellet was washed for twice with PBS, and the number of living cells were detected using Trypan Blue.

HCC patients-derived organoids culture

Cell pellet from tumor tissue dissociation was resuspended with optimized hepatobiliary tumor organoids culture medium, consisting of Advanced DMEM/F-12, Penicillin-Streptomycin, GlutaMAX, HEPES, 100µg/mL Primocin, B-27 supplement (Gibco), 1.56 mM N-Acetylcysteine (Sigma-Aldrich), 50 ng/mL mouse recombinant EGF (Gibco), 100 ng/mL recombinant human FGF10 (Peprotech), 1ng/mL recombinant human FGF-basic (Peprotech), 25 ng/mL recombinant human HGF (Peprotech), 10 µM forskolin (Selleck), 500nM A8301 (Tocris), 10 µM Y-27632, 10mM Nicotinamide (Sigma-Aldrich), 3nM Dexamethasone (Sigma-Aldrich), 2% (vol/vol) Rspondin-conditioned medium, 5% (vol/vol) Noggin-conditioned medium and 30% (vol/vol) Wnt3a-conditioned medium. The same volume of cold Matrigel Basement Membrane Matrix (CORNING) was added, and each 50µL of organoids-matrigel was pipetted to form a single dome onto suspended cell plate. After solidification at 37°C for 45 min, culture medium was added, and the cell plate was transferred to humidified 37°C/5% CO2 incubator. Fresh medium was replaced each 3 days during organoids culture.

For cell passage, organoids were harvested into a 15mL tube, and centrifugated for 5 min at 1300–1500 rpm. Cell pellet was washed using cold PBS to remove the residual matrigel. Organoids were digested with TrypLE Express (Gibco) at 37°C, then the same volume of DMEM-HG medium (containing 10% FBS) was added to terminate the digestion. Organoids were centrifugated for 5 min at 1300–1500 rpm, and the cell pellet was suspended with culture medium. The same volume of matrigel was mixed and the cell suspension was seeded onto the suspended cell plate.

Cell viability assay

Cells were pre-seeded (1000 cells/well) and cultured in 384-well microplates (CORNING) for 24h. For organoids, 10 µL of matrigel was dispensed into 384-well microplates and allowed to polymerize before seeding. Then, chemicals were added to the culture medium at the indicated concentration. The same volume of solvent was used as the control. After 2–3 days of treatment, cell viability was assayed using CellTiter-Glo® 3D Reagent (Promega) in accordance with the manufacturer’s instructions.

Cell transfection

For lentiviral transfection in organoids, cells were digested with TrypLE Express and suspended in culture medium. The lentivirus and polybrene (Beyotime) were added into the cell suspension, and was centrifugated for 55 min at 500×g. Cell pellet was suspended with culture medium, mixed with the same volume of matrigel, and seeded onto the suspended cell plate. For lentiviral transfection in cell lines, 5×105 cells were pre-seeded onto 60mm dish for 24h, and incubated with DMEM-HG conditional medium containing lentivirus (1–2×107 TU) and polybrene. After incubation for 48h, cells were passage onto 100mm dish and selected with puromycin (Beyotime).

For siRNA transfection, 1×105 cells were pre-seeded onto 6-well plate with 2mL medium for 24h. Transfection mixture was prepared containing 100pmol siRNA, 5µL jetPRIME (Polyplus) and 200µL jetPRIME buffer (Polyplus), and incubated for 10 to 15 min at room temperature. The transfection mixture was added into the cell plate containing medium (2mL) directly, and was incubated at 37°C. Expression levels of the target genes were examined by qPCR and immunoblotting after 24-48h.

Quantitive real-time PCR (qPCR)

Total RNA was isolated by RNAsimple Kit (TIANGEN) as per manufactures protocol. Total RNA (1 µg) was reverse transcribed by ReverTra Ace qPCR RT Kit (TOYOBO) according to the manufacturer’s guidelines. qPCR was performed with gene specific primers at 95°C for 10 sec, 60°C for 60 sec (40 cycles) in 10 µL reaction mix containing 5 µL SYBR Green master mix (Qiagen), 1µL cDNA, 0.5 µL forward primer, 0.5 µL reverse primer and 3µL RNase-Free ddH2O using a Roche thermocycler. β-actin was used as an internal control. Quantification was performed with the ΔΔCt method. See Supplementary Materials for the primer sequences used for qPCR analysis.

Immunoblotting

Cell lysates were prepared by scraping (adherent cells) or centrifugation (suspended organoids) into lysis buffer (Beyotime, containing 1 mM PMSF). Cell lysates were incubated at 4°C for 1h with rotation and centrifuged at 12000 rpm, 4°C for 15 min. The supernatant was transferred into a new tube, and the protein concentration was examined using BCA kit (Beyotime). The antigens transferred to nitrocellulose filter membranes were probed with the primary antibodies against PKM2 (Cell Signaling Technology), DHFR (Cell Signaling Technology), LMNB1 (Proteintech), GAPDH (Proteintech) and β-actin (Sigma-Aldrich).

Hematoxylin-eosin (HE) and immunohistochemistry staining

For HE, tumor tissues or organoids were fixed with 4% paraformaldehyde, dehydrated and embedded into paraffin blocks. Sections (4–5 µm) were deparaffinized and stained with hematoxylin-eosin for histological analysis.

For Immunohistochemistry, sections were deparaffinized and incubated in 3% H2O2 for 20 min. Sections were dipped in boiling citrate buffer solution (0.01M, pH = 6.0) for 5 min and repeated 2 times. After cooling to room temperature, sections were blocked with 1%BSA/PBS for 1h and incubated with the primary antibody against PKM2 (Cell Signaling Technology) and DHFR (Cell Signaling Technology) at 4°C overnight. Sections were incubated with the secondary antibody for 1h and stained with DAB. Image collecting was performed using the Leica Biosystems. IHC scores of specimens were in accordance to the percentage of positive cells and staining intensity.

Tumor growth studies

For the tumor growth of organoids, 5 weeks of male NSG mice were purchased from the Model Organisms of Shanghai. We subcutaneously implanted 2×106 parental or Len-R organoids in the dorsal side of mice. Tumor volumes were monitored for a period of 42 days.

For the tumor growth of cell lines, 5 weeks of female nude mice were purchased from the Model Organisms of Shanghai. We subcutaneously implanted 1×106 WT or PKM2-shRNA expressing cells. Mice were subjected to lenvatinib (10 mg/kg, 3 times a week) or solvent after tumor volumes reached 100mm3. Tumor volumes and body weights were monitored 2–3 times a week.

LC-MS/MS based metabolomics

For the sample preparation, cells were harvested by centrifugation and washed by PBS. Cell pellet was quenched rapidly by liquid nitrogen. 2-chloro-1-phenylalanine (0.3mg/mL) dissolved in methanol as internal standard and methanol: acetonitrile: water (2/2/1, vol/vol/vol) were added to each sample. Samples were placed at -20°C for 2 min, grinded at 60 Hz for 2 min, and ultrasonicated at ambient temperature for 5 min. Samples were centrifuged at 12000 rpm, 4°C for 10 min. Supernatant was transferred into a new 1.5 mL tube, while the cell pellet was added the mixture of methanol: acetonitrile: water (2:2:1, vol/vol/vol). Samples were vortexed for 2 min, ultrasonicated at ambient temperature for 5 min, and centrifuged at 12000rpm, 4°C for 10 min. Supernatant was transferred into the above 1.5 mL tube and added the same volume of chloroform. Samples were placed at -20°C for 30 min and centrifuged at 12000 rpm, 4°C for 5 min. Supernatants were collected using crystal syringes, filtered through 0.22 µm microfilters and transferred to LC vials. The vials were stored at -80°C until LC-MS analysis.

For the data preprocessing and statistical analysis, the acquired LC-MS raw data were analyzed by the Compound Discoverer software. The positive and negative data were combined to import into R ropls package. Principle component analysis and orthogonal partial least-squares-discriminant analysis were carried out to visualize the metabolic alterations among experimental groups, after mean centering and Pareto variance scaling, respectively. In this study, the default 7 round cross validation was applied with 1/seventh of the samples being excluded from the mathematical model in each round, in order to guard against overfitting. The differential metabolites were selected on the basis of the combination of a statistically significant threshold of variable influence on projection values obtained from the orthogonal partial least-squares-discriminant analysis model and p-Values from a two-tailed Student's t-test on the normalized peak areas, where metabolites with variable influence on projection values larger than 1.0 and p-Values less than 0.05 were considered as differential metabolites.

Extraction of nuclear and cytoplasmic components

Organoids were harvested by centrifuging at 500×g for 5 min and washed by suspending the cell pellet with cold PBS. The nuclei and cytosolic fractions were collected using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo scientific) according to the manufacturer’s instructions.

CO-IP

PKM2-overexpressed cells were lysed in lysis buffer (20mM Tris (pH = 7.5), 150mM NaCl, 1% Triton X-100, sodium pyrophosphate, β-glycerophosphate, EDTA, Na3VO4, leupeptin) and rotated at 4°C for 1h. Cell lysates were centrifuged at 12,000 rpm for 15 min at 4°C to remove the cell debris, and supernatants were incubated with PKM2 antibody (Cell Signaling Technology) at 4°C overnight with rotation. Equivalent mass of IgG was as a control. Cell lysates were incubated with Protein A/G Magnetic Beads (MedChemExpress) for 2 hours at 4°C with rotation. Beads were collected by performing magnetic separation, and the protein complex was eluted from the Protein A/G Magnetic Beads at 95°C for 5min. Immunoblotting was performed with primary antibody against PKM2 and SMAD4 (Cell Signaling Technology).

CUT&Tag

The assay was performed by using CUT&Tag Kit (Yeasen Biotechnology, Shanghai) in accordance with the manufacturer’s guidelines. Cells were incubated with the primary antibody against SMAD4. Equivalent mass of IgG was as a control. Goat Anti-Rabbit IgG H&L (abcam) was performed as the secondary antibody. PCR or qPCR analysis was performed with specific primers to SBE motif at DHFR promoter region. See Supplementary Materials for the primer sequences used for the amplification of SBE motif.

Luciferase reporter assay

5×104 cells were pre-seeded onto 24-well plates containing 500µL culture medium for 24h. 50µL jetPRIME buffer was prepared containing 0.5µg reporter plasmid (backbone was pGL3-Basic), 0.05µg internal control plasmid pSV-Renilla and 1.5µL jetPRIME. The mixture was incubated at room temperature for 10min and added into 24-well plates containing culture medium. Cell plate was incubated at 37°C for 48 h. FLuc and RLuc activities were determined using the Dual-Luciferase Assay System (Promega).

scRNA-Seq data analysis

Gene count matrices for each single-cell sample were generated by aligning reads to the hg38 genome using 10X Genomics CellRanger software v.4.0.0. The generated cell-by-gene unique molecular identifier count matrix was analyzed using the Seurat R package v.4.1.0. We kept cells with number of expressed genes between 200 and 7000. The cells were further filtered by removing cells with more than 15% mitochondrial genes. We retained genes with expression in at least 3 cells. The unique molecular identifier counts were then normalized for each cell by the total expression, multiplied by 10,000 and log-transformed. We used Seurat’s default method to identify highly variable genes and scale data for regressing out variation from unique molecular identifier and mitochondrial genes. The scaled data with variable genes were used to principal component analysis. The top 20 principal components were chosen for further analysis, including clustering to identify cell populations. UMAPs were calculated in the Seurat R package using the top 20 PCs and min_dist = 0.75. CCA was used to perform batch-effect correction for each untreated and lenvatinib treated paired samples. Intra-tumor heterogeneity was measured by calculating the average distance between the normalized expression profiles of each cell and all other cells in the same sample. The cell stemness of indicated samples and subgroups were characterized by calculating the average expression level of CSCs markers. We performed pseudo-time analysis in each integrated sample using Monocle 3 R package v.1.2.7.

Gene set enrichment analysis

SsGSEA was performed using GSVA in R v.1.42.0 using single-cell expression matrix with integrated normalized values. We applied hallmark gene sets and canonical pathways from KEGG gene sets of the C2 collection of Molecular Signatures Database (MSigDB) to each single cell to obtain enrichment score of each signature. Progeny python package was utilized to get activity score of canonical cancer relating pathways.

Differential expression analysis

DEseq2 was applied to identify the differentially expressed genes with a p.adjust value < 0.01 and a |log2FC| >1. For the single cell dataset, differential marker genes were identified using FindMarkers function with Wilcox test (only.pos = TRUE, min.pct = 0.25, log2FC.threshold = 0.25).

Whole genome sequencing analysis

The sequencing data was analyzed using a custom pipeline. In brief, data was preprocessed by Trim Galore v.0.6.6 which is a Perl wrapper around two tools: Cutadapt and FastQC. Sequencing reads were aligned to the human genome (hg38) using Burrows-Wheeler Aligner Mem v.0.7.17. Duplications were marked using Picard Tools v.2.22.2. Insertion–deletion realignment and base recalibration were achieved using GATK v.4.2.0. Somatic variants were detected using an ensemble approach with one variant caller: MuTect. Variant annotation was performed using SnpEff v.4.3 to obtain the final set of mutation calls. We used a two-step approach: (1) to reduce false-positive calls, the mutant variant site contains at least one allele that pass filters; (2) to get significant variant calls, the “HIGH” and “MODERATE” impact mutant variants were kept. Only variants with the following functional classification were considered in this study: nonsynonymous single nucleotide variants, stop gain mutations, and frameshift mutations.

Increased chemoresistance signatures in Len-R organoids of HCC

To investigate the molecular basis of lenvatinib resistance in HCC, we generated lenvatinib resistant models in HCC organoids with increasing concentrations of lenvatinib for 3–4 months (Fig. 1A). The parental HCC organoids (HCC-277, HCC-297) were established from tumor tissue surgically removed from HCC patients. Tissue morphology and genetic alterations were greatly retained in organoids comparing with the matched tumor tissues (Fig. 1B,1C). During the period of acquiring chemoresistance, parental organoids were first stimulated with lenvatinib for two days at the indicated concentration, and replaced with fresh medium without lenvatinib for cell recovery. The persistent cells which are tolerant to lenvatinib were accumulated by removing dead cells via centrifugation. To generate the stable resistant cells under the indicated concentration, the accumulated persistent cells were re-cultured and incubated with lenvatinib for 7–10 days, replacing with fresh medium with lenvatinib for each three days. At the end of the treatment period, Len-R organoids (HCC-277R and HCC-297R) were more resistant as compared to parental organoids (Fig. 1D). Additionally, the subcutaneous tumorigenic ability was substantially enhanced in lenvatinib resistant organoid than that in parental organoid (Fig. 1E).

To gain a deep insight into tumor resistance at the single-cell resolution, we performed single-cell RNA sequencing (scRNA-Seq) analyses in parental and Len-R organoids. Intra-tumoral heterogeneity score (ITHs) and stemness of cancer stem cells (CSCs), which are canonical cancer hallmarks, were distinctly increased in Len-R organoids (Fig. 1F,1G). Moreover, gene set enrichment analysis (GSEA) revealed a higher enrichment of chemoresistance-relevant pathways in Len-R organoids comparing with the parental organoids, suggesting various drug resistant mechanisms can coexist forming a complex heterogeneity (Fig. 1H). In conclusion, Len-R organoids acquired the resistance property and simulated the emergence of resistant tumors.

Lenvatinib resistance property is dispersed in each subpopulation by scRNA-Seq analyses

To visualize the transcriptional expression profile, we performed uniform manifold approximation and projection (UMAP) analysis in two groups of samples (HCC-277/277R, HCC-297/297R) (Fig. 2A). We observed that cell clusters were indiscernible in parental and resistant organoids. Principal component analysis revealed that organoids from the same individual showed an overlap even after a long-term treatment culture, and the main differences primarily caused by heterogeneity among different individuals.

To determine whether the resistance property of Len-R was conferred by specific clusters, we first estimated the degree of ITH and stemness of CSCs in each cluster of our resistant models. We found higher ITH scores in each cell subgroup of Len-R (Fig. 2C). Similarly, we also observed higher stemness of CSCs in clusters of Len-R compared with the parental organoids (Fig. 2D). In addition, we identified the canonical resistance-relevant pathways in the two groups of samples. In HCC-277 group, the resistant pattern of the examined pathways has no apparent differences between parental and resistant organoids. Cluster1 of HCC277R has a more extensive resistant signatures, which accounted for most cell proportion in HCC277R, while the rest of clusters also have distinct activation of resistance-relevant pathways (Fig. 2B,2E). In HCC-297 group, HCC-297R displayed a largely different resistant pattern compared to HCC-297. Specifically, cluster1, cluster2 and cluster3 of HCC-297R have multiple activated pathways and consistently upregulated MAPK, EGFR, JAK-STAT and Androgen pathways, while cluster4 also enriched resistance-relevant pathways in HCC-297R comparing with HCC-297, such as PI3K and EGFR pathways, and PI3K was specifically enriched in cluster4 (Fig. 2E). Collectively, these results suggest that each cluster of Len-R could contribute to the resistance property, instead of relay on a subset of clusters.

Clinically relevant drug-resistant cancer cells evolve from treatment-naive cells [22]. To verify this assertion in our study, we performed pseudo-time analysis and found a distinct cellular evolutionary trajectory from parental to resistant organoids during the treatment process (Fig. 2F), further proving the success of simulating the emergence of lenvatinib resistance in HCC organoid models.

PKM2 contributes to lenvatinib chemoresistance

To investigate the molecular basis sustaining cancer cell survival under lenvatinib treatment pressure, we first performed KEGG analysis and found that multiple signaling pathways were greatly enriched in Len-R organoids, including metabolic relevant pathways, such as glycolysis, pentose phosphate pathway (PPP) and nucleotides biosynthesis pathway (Fig. 3A). Previous studies have reported a strong correlation between metabolic flux alteration and drug resistance [23, 24]. Differential expression genes (DEGs) analysis showed candidates with upregulated expression levels (fold change ≥ 2) in Len-R organoids comparing with the parental samples (Fig. 3B). Notably, PKM, a regulator of metabolic reprogramming by its alternative splicing of PKM2, was upregulated in Len-R organoids with chemoresistance property (SupFig.1A,1B). Moreover, the expression level of PKM was gradually increased during the evolution process under lenvatinib treatment (Fig. 3C). To further clarify the dominant isoform of PKM expressed in Len-R, we identified its expression levels and found an upregulated expression of PKM2 (Fig. 3D, SupFig.1C), which were accumulated in cytoplasm and nucleus (Fig. 3E), whereas PKM1 has a slightly increase with lower basal expression (SupFig.1C).

Next, we demonstrated that overexpression of PKM2 showed chemoresistance to lenvatinib in parental organoids (Fig. 3F,3G), as well as in HCC cell lines (LM3 and PLC) (Fig. 3H,3I). Moreover, we subcutaneously implanted WT and PKM2-shRNA expressing cells in nude mice (SupFig.1D). Treatment with solvent or lenvatinib started at the point when tumor volume reached 100mm³. Lenvatinib could significantly reduce tumor burden without noticeable body weight changes (Fig. 3J-M), indicating the effectively pharmacological inhibition with accepted toxicity. Importantly, PKM2 knocking down synergized with lenvatinib treatment showed a significant reduction of tumor volumes compared to the WT cells plus lenvatinib treatment (Fig. 3J,3K). These results indicate that lenvatinib resistance is at least in part dependent on PKM2 expression.

Resistance to lenvatinib might also occur due to bypassed target inhibition through impaired binding of the drug [25]. Thus, we focused on the alteration of the targets (VEGFR1–3, FGFR1–4, PDGFRα, as well as proto-oncogenes RET and KIT [26]) by whole genome sequencing (WGS) analysis. Reassuringly, none of the targets mutated in Len-R organoids in comparison to parental organoids (SupFig.1E). Collectively, lenvatinib resistance was not attribute to the mutation of its targets, which could affect the binding with lenvatinib.

Enrichment of purine metabolic throughput maintains lenvatinib resistance

Due to the enrichment of metabolism pathways in Len-R, we performed Liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS)-based metabolomics analysis to further identify the precise altered metabolism pathways. As a result, pyrimidine and purine metabolism were the most significant enriched pathways in Len-R organoids compared with the parental organoids (Fig. 4A), and most of the related metabolite levels were highly increased (Fig. 4B). Additionally, scRNA-Seq analyses further supported that pyrimidine and purine metabolism gene sets, as well as glycolysis and PPP were significantly enriched in Len-R (Fig. 4C).

To identify whether pyrimidine and purine metabolism pathways associated with lenvatinib resistance, we first treated HCC-277 and HCC-297 with indicated concentration of the mixture of deoxyadenosine, deoxyguanosine, deoxycytidine, and thymidine (AGCT). Supplementation with AGCT significantly mitigated the effectiveness of lenvatinib (Fig. 4D). This effect was likely caused by stimulating DNA repair [27]. To further clarify which metabolism pathway plays a major role in lenvatinib resistance, we treated HCC-277 with deoxyadenosine and deoxyguanosine (AG), or deoxycytidine and thymidine (CT), separately. Unexpectedly, increased levels of AG diminished the effectiveness of lenvatinib, while CT had no such distinct effect (Fig. 4E).

We assumed that the resistance property of AG may attribute to the compensable effect for the extensive inhibition of lenvatinib on purine metabolic flux. To verify this hypothesis, metabolomic analysis was performed in HCC cell lines with or without lenvatinib treatment. Expectedly, purine metabolism pathway was more affected by lenvatinib than pyrimidine metabolism (Fig. 4F). Consistent results have been reported that tyrosine kinase inhibitors preferentially targets purine metabolism, rather than pyrimidine metabolism [28]. Additionally, glycolysis and PPP, which were enriched in Len-R organoids, were also largely affected by lenvatinib treatment (Fig. 4F). In conclusion, purine metabolic flux imparts lenvatinib resistance to cancer cells.

PKM2 mediates purine metabolism through transcriptional upregulation of DHFR

Previous studies suggest that PKM2 promotes tumor metabolic reprogramming dependent on its coactivator function [9, 10]. Thus, we made a possible hypothesis that increased purine metabolic flux may mediated by PKM2. To verify this idea, we first performed metabolomic analysis in PKM2-overexpressed cells and found a significant enrichment of purine metabolic flux (Fig. 5A, Fig. 5B). Additionally, knocking down PKM2 correspondingly decreased the purine metabolic throughput (SupFig.2A-C). The nucleotide requirement is met by upregulation of the de novo biosynthetic pathway under high cellular purine demands [29]. Thus, we examined genes involved in de novo purine metabolism as well as pyrimidine metabolism in PKM2-overexpressed organoids. Surprisingly, we found that DHFR expression levels were highly upregulated by PKM2 (Fig. 5C, SupFig.2D). Importantly, we demonstrated an upregulated expression of DHFR in Len-R organoids by IHC staining (Fig. 5D). Due to the regulation of PKM2 to purine metabolism via DHFR, we transfected DHFR-shRNA lentivirus into PKM2-overexpressed cells. Advantage of purine metabolism in PKM2 overexpressed cells was reversed by knocking down DHFR (Fig. 5E, SupFig.2E,2F), and the resistance property was also diminished (Fig. 5F).

According to the coactivator function of PKM2 in transcriptional regulation, we first predicted candidates of transcriptional factors at DHFR promoter region (-2000bp to + 500bp). Intriguingly, SMAD4 with higher score affected DHFR expression at the transcriptional level (SupFig.2G). DHFR promoter region encompassed a proximal (+ 4 to + 11) and a distal (-1534 to -1541) SBE motif (5’-GTCTG-3’) (Fig. 5G). CUT&Tag assay demonstrated the occupancy of SMAD4 at the proximal and distal SBE (Fig. 5I). Then, we performed Co-IP assay and found a distinct interaction between PKM2 and SMAD4 immunoprecipitated by PKM2 antibody (Fig. 5H). Importantly, PKM2 displayed the coactivator function with enhanced occupancy of SMAD4 at DHFR promoter region and transcriptional regulation in comparison with the control cells (Fig. 5J,5K). In contrast, mutation of the proximal SBE apparently decreased the transcriptional regulation than that at the distal SBE, suggesting the high transcriptional activation of the proximal SBE region.

Next, we treated Len-R and parental organoids with DHFR inhibitor pyrimethamine in combination with lenvatinib. Pyrimethamine increased the efficacy of lenvatinib and inhibited cell survival in Len-R organoids (Fig. 5L). Collectively, these results suggest PKM2-SMAD4-DHFR axis as a novel molecular mechanism in lenvatinib resistant tumors.

PKM2 and DHFR expression levels negatively correlate with lenvatinib response in HCC specimens

Clinically, we performed IHC staining in tissue microarrays of human HCC specimens. As a result, we observed a high correlation between PKM2 and DHFR expression performed in 113 HCC specimens (Fig. 6A,6B). To identify the association of PKM2 with the clinical response to lenvatinib, we examined PKM2 expression level in several human HCC specimens with subsequent therapy of lenvatinib. Of note, patients who suffered progression of disease (PD) were examined with high expressional levels of PKM2, whereas those who were benefited with the chemo had a lower expression (Fig. 6C,6D). Additionally, nuclear PKM2 expression accumulated in a part of percentage of tumor cells of lenvatinib resistant group (Fig. 6C). Similar expression levels of DHFR were observed in the corresponding matched specimens (Fig. 6C,6D). Groups were divided in accordance with the clinical evaluation of therapeutic efficacy, and lenvatinib treatment was utilized for adjuvant chemotherapy after surgery or treating recurrent patients. In HCC available datasets from The Cancer Genome Atlas (TCGA), PKM2^high samples were predicted with a poor survival in comparison with PKM2^low samples. Similar results were also observed between DHFR^high and DHFR^low groups (Fig. 6E). These observations suggest that PKM2-DHFR regulatory mechanism may exists in HCC inherently and was utilized by tumor cells to acquire the adaptation to lenvatinib, indicating PKM2 and DHFR as underlying biomarkers of lenvatinib resistant tumors in HCC.

Lenvatinib application has been changed the status of sorafenib as the sole first-line tyrosine kinase inhibitor treatment for advanced HCC. Evidence indicated that lenvatinib has antitumor proliferation and immunomodulatory activity in preclinical studies [30]. Positive outcomes have been achieved in mono- and combination therapy with immune checkpoint inhibitors, specifically lenvatinib plus pembrolizumab [1, 31]. However, the emergence of drug resistance limits its clinical efficacy and accelerates tumor progression. According to the immortal cell lines could not phenocopy the clinical characters, we adopted patients-derived organoids to generate Len-R to mimic the occurrence of chemo-resistant tumors. Here, we demonstrated PKM2 as a coactivator in reprogramming purine metabolism flux by transcriptionally upregulated DHFR in weakening lenvatinib effectiveness. Targeting PKM2, DHFR, or de novo purine metabolism biosynthesis may improve the efficacy of lenvatinib in chemo-resistant tumors (Fig. 6F).

According to the aberrant expression of PKM2 in many cancers, PKM2 inhibitors display a promising therapy strategy in tumor. Compound 3k, a selective inhibitor of PKM2, has antitumor activity in nanomole dosage [32]. Shikonin from natural sources exerts cytotoxicity activity by multiple mechanisms, including the effect on glycolysis by inhibiting PKM2 expression [33]. PKM2 agonists could promote PKM2 tetramer formation and process of oxidative phosphorylation, restoring the normal metabolism in cancer cells [34]. Previous studies have shown that small molecule agonists bind at the subunit interaction interface to activate PKM2, thereby interfering with anabolic metabolism [35]. Additionally, PKM2 agonists could reduce nucleus translocation of PKM2 to inhibit tumor proliferation which is mediated by PKM2 non-metabolic function [36]. DHFR inhibitors are an important class of drugs for anticancer agents. Methotrexate is well known for its widely application in clinical for cancer therapy. Numerous effects have been made to identify the anticancer potential of pyrimethamine in vitro and in vivo [37]. Therefore, PKM2 and DHFR are available targets and altering its expression levels or activity could provide promising outcomes in cancer therapy.

Purine nucleotides are basic components of DNA and RNA in cellular biological process, thus impaired purine metabolism is associated with the regression of cancer [37]. In our present study, we provide a new insight into lenvatinib resistance contributed by purine metabolism pathway. In our data, pyrimidine metabolism pathway was also enriched in Len-R organoids, while supplement of pyrimidine nucleotides was identified with no apparent lenvatinib-resistant property, which is partly attribute to the limitation of experimental system, for the effect of lenvatinib on pyrimidine metabolism was not as extensive as on purine metabolism after 2 days of lenvatinib treatment (Fig. 4F).

Multiple molecular mechanisms of regulating PKM2 expression have been proposed, including activation of PI3K-AKT pathway [38], transcriptional regulation by HIF-1α [10], and mediation of alternative splice isoforms by c-MYC-PTB axis [12]. Additionally, translocation of PKM2 could mediated by EGFR pathway activation [9] and enter into nucleus interacted with importin α5 [11]. In our present study, we mainly examined PI3K-AKT activation and c-MYC-PTB axis in the regulation of PKM2. However, PKM2 expression has no apparent changes after the activation of p-AKT and p-S6 upon IGF-1 stimulation or silencing the expression level of c-MYC (Data was not shown).

Regorafenib and immunotherapies are the options to treat lenvatinib resistance in clinical treatment of HCC. Some patients could benefit from the treatment, while the others may fail. Accordingly, various clinical trials are ongoing in patients with HCC who have progressed on lenvatinib, including the evaluation of varlitinib (NCT03499626), anlotinib (NCT04080154), gefitinib (NCT04642547) and VIC-1911 (NCT05718882). Similarly, preclinical investigations are also needed to provide more vulnerabilities as candidate targets progressing to clinical therapy. Here, our study demonstrates that increased purine metabolism regulated by PKM2 via DHFR could resist lenvatinib treatment, thus synergizes with PKM2 or DHFR inhibitors may overcome lenvatinib resistance.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (U21A20376, 82273277, 81902412, 81988101, 91859205, 82002921 and 81830054) and the National Science Foundation of Shanghai (21XD1404600, 17ZR143800, 21JC1406600, and 22140901000).

Author Contributions

Y. Z. was responsible for designing the experiment, carrying out the experiment, optimizing the protocol, extracting and analysing data, interpreting results and writing the manuscript. FF.W. was responsible for analysing the sequencing data and writing the manuscript. YJ.Z. was responsible for designing and carrying out the experiment. MS.X., YN.Z. and KT.W. carried out the experiment. S.W., SY.S. and J.H. provided technical assistance. L.C., P.W. and HY.W. supervised the progress of the study, provided feedback on the work report and edited the manuscript.

Competing Interests

The authors have declared that no competing interests exist.

Data Availability Statement

The authors declare that all data that support the findings of this study are available within the paper and supplementary files.

Competing Interests statement

The authors declare no potential conflicts of interest.

Kudo M, Finn RS, Qin S, Han KH, Ikeda K, Piscaglia F et al (2018). Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial. Lancet 391: 1163-1173.
Zhao Z, Zhang D, Wu F, Tu J, Song J, Xu M et al (2021). Sophoridine suppresses lenvatinib-resistant hepatocellular carcinoma growth by inhibiting RAS/MEK/ERK axis via decreasing VEGFR2 expression. J Cell Mol Med 25: 549-560.
Zhao Z, Song J, Zhang D, Wu F, Tu J, Ji J (2021). Oxysophocarpine suppresses FGFR1-overexpressed hepatocellular carcinoma growth and sensitizes the therapeutic effect of lenvatinib. Life Sci 264: 118642.
Jin H, Shi Y, Lv Y, Yuan S, Ramirez C, Lieftink C et al (2021). EGFR activation limits the response of liver cancer to lenvatinib. Nature 595: 730-734.
Fu R, Jiang S, Li J, Chen H, Zhang X (2020). Activation of the HGF/c-MET axis promotes lenvatinib resistance in hepatocellular carcinoma cells with high c-MET expression. Med Oncol 37: 24.
Ma XL, Hu B, Tang WG, Xie SH, Ren N, Guo L et al (2020). CD73 sustained cancer-stem-cell traits by promoting SOX9 expression and stability in hepatocellular carcinoma. J Hematol Oncol 13: 11.
Thomas A, Desai P, Takahashi N (2022). Translational research: A patient-centered approach to bridge the valley of death. Cancer Cell.
Broutier L, Mastrogiovanni G, Verstegen MM, Francies HE, Gavarro LM, Bradshaw CR et al (2017). Human primary liver cancer-derived organoid cultures for disease modeling and drug screening. Nat Med 23: 1424-1435.
Yang W, Xia Y, Ji H, Zheng Y, Liang J, Huang W et al (2011). Nuclear PKM2 regulates beta-catenin transactivation upon EGFR activation. Nature 480: 118-22.
Luo W, Hu H, Chang R, Zhong J, Knabel M, O'Meally R et al (2011). Pyruvate kinase M2 is a PHD3-stimulated coactivator for hypoxia-inducible factor 1. Cell 145: 732-44.
Yang W, Zheng Y, Xia Y, Ji H, Chen X, Guo F et al (2012). ERK1/2-dependent phosphorylation and nuclear translocation of PKM2 promotes the Warburg effect. Nat Cell Biol 14: 1295-304.
David CJ, Chen M, Assanah M, Canoll P, Manley JL (2010). HnRNP proteins controlled by c-Myc deregulate pyruvate kinase mRNA splicing in cancer. Nature 463: 364-8.
Feng J, Dai W, Mao Y, Wu L, Li J, Chen K et al (2020). Simvastatin re-sensitizes hepatocellular carcinoma cells to sorafenib by inhibiting HIF-1alpha/PPAR-gamma/PKM2-mediated glycolysis. J Exp Clin Cancer Res 39: 24.
Li Q, Zhang D, Chen X, He L, Li T, Xu X et al (2015). Nuclear PKM2 contributes to gefitinib resistance via upregulation of STAT3 activation in colorectal cancer. Sci Rep 5: 16082.
Chen C, Liu WR, Zhang B, Zhang LM, Li CG, Liu C et al (2020). LncRNA H19 downregulation confers erlotinib resistance through upregulation of PKM2 and phosphorylation of AKT in EGFR-mutant lung cancers. Cancer Lett 486: 58-70.
Li C, Zhao Z, Zhou Z, Liu R (2016). Linc-ROR confers gemcitabine resistance to pancreatic cancer cells via inducing autophagy and modulating the miR-124/PTBP1/PKM2 axis. Cancer Chemother Pharmacol 78: 1199-1207.
Wu H, Du J, Li C, Li H, Guo H, Li Z (2022). Kaempferol Can Reverse the 5-Fu Resistance of Colorectal Cancer Cells by Inhibiting PKM2-Mediated Glycolysis. Int J Mol Sci 23.
Wang Y, Hao F, Nan Y, Qu L, Na W, Jia C et al (2018). PKM2 Inhibitor Shikonin Overcomes the Cisplatin Resistance in Bladder Cancer by Inducing Necroptosis. Int J Biol Sci 14: 1883-1891.
Yin J, Ren W, Huang X, Deng J, Li T, Yin Y (2018). Potential Mechanisms Connecting Purine Metabolism and Cancer Therapy. Front Immunol 9: 1697.
Zhao H, Chiaro CR, Zhang L, Smith PB, Chan CY, Pedley AM et al (2015). Quantitative analysis of purine nucleotides indicates that purinosomes increase de novo purine biosynthesis. J Biol Chem 290: 6705-13.
De Vitto H, Arachchige DB, Richardson BC, French JB (2021). The Intersection of Purine and Mitochondrial Metabolism in Cancer. Cells 10.
Hata AN, Niederst MJ, Archibald HL, Gomez-Caraballo M, Siddiqui FM, Mulvey HE et al (2016). Tumor cells can follow distinct evolutionary paths to become resistant to epidermal growth factor receptor inhibition. Nat Med 22: 262-9.
Fanciulli M, Bruno T, Giovannelli A, Gentile FP, Di Padova M, Rubiu O et al (2000). Energy metabolism of human LoVo colon carcinoma cells: correlation to drug resistance and influence of lonidamine. Clin Cancer Res 6: 1590-7.
Shukla SK, Purohit V, Mehla K, Gunda V, Chaika NV, Vernucci E et al (2017). MUC1 and HIF-1alpha Signaling Crosstalk Induces Anabolic Glucose Metabolism to Impart Gemcitabine Resistance to Pancreatic Cancer. Cancer Cell 32: 71-87.e7.
Redmond KM, Wilson TR, Johnston PG, Longley DB (2008). Resistance mechanisms to cancer chemotherapy. Front Biosci 13: 5138-54.
Tohyama O, Matsui J, Kodama K, Hata-Sugi N, Kimura T, Okamoto K et al (2014). Antitumor activity of lenvatinib (e7080): an angiogenesis inhibitor that targets multiple receptor tyrosine kinases in preclinical human thyroid cancer models. J Thyroid Res 2014: 638747.
Majsterek I, Slupianek A, Hoser G, Skorski T, Blasiak J (2004). ABL-fusion oncoproteins activate multi-pathway of DNA repair: role in drug resistance? Biochimie 86: 53-65.
Zheng JF, Lu J, Wang XZ, Guo WH, Zhang JX (2015). Comparative Metabolomic Profiling of Hepatocellular Carcinoma Cells Treated with Sorafenib Monotherapy vs. Sorafenib-Everolimus Combination Therapy. Med Sci Monit 21: 1781-91.
Pedley AM, Benkovic SJ (2017). A New View into the Regulation of Purine Metabolism: The Purinosome. Trends Biochem Sci 42: 141-154.
Kimura T, Kato Y, Ozawa Y, Kodama K, Ito J, Ichikawa K et al (2018). Immunomodulatory activity of lenvatinib contributes to antitumor activity in the Hepa1-6 hepatocellular carcinoma model. Cancer Sci 109: 3993-4002.
Finn RS, Ikeda M, Zhu AX, Sung MW, Baron AD, Kudo M et al (2020). Phase Ib Study of Lenvatinib Plus Pembrolizumab in Patients with Unresectable Hepatocellular Carcinoma. J Clin Oncol 38: 2960-2970.
Ning X, Qi H, Li R, Li Y, Jin Y, McNutt MA et al (2017). Discovery of novel naphthoquinone derivatives as inhibitors of the tumor cell specific M2 isoform of pyruvate kinase. Eur J Med Chem 138: 343-352.
Boulos JC, Rahama M, Hegazy MF, Efferth T (2019). Shikonin derivatives for cancer prevention and therapy. Cancer Lett 459: 248-267.
Hsu MC, Hung WC (2018). Pyruvate kinase M2 fuels multiple aspects of cancer cells: from cellular metabolism, transcriptional regulation to extracellular signaling. Mol Cancer 17: 35.
Anastasiou D, Yu Y, Israelsen WJ, Jiang JK, Boxer MB, Hong BS et al (2012). Pyruvate kinase M2 activators promote tetramer formation and suppress tumorigenesis. Nat Chem Biol 8: 839-47.
Ding Y, Xue Q, Liu S, Hu K, Wang D, Wang T et al (2020). Identification of Parthenolide Dimers as Activators of Pyruvate Kinase M2 in Xenografts of Glioblastoma Multiforme in Vivo. J Med Chem 63: 1597-1611.
Ramchandani S, Mohan CD, Mistry JR, Su Q, Naz I, Rangappa KS et al (2022). The multifaceted antineoplastic role of pyrimethamine against human malignancies. IUBMB Life 74: 198-212.
Iqbal MA, Siddiqui FA, Gupta V, Chattopadhyay S, Gopinath P, Kumar B et al (2013). Insulin enhances metabolic capacities of cancer cells by dual regulation of glycolytic enzyme pyruvate kinase M2. Mol Cancer 12: 72.

There is NO conflict of interest to disclose.

SupplementaryFigure.docx
Supplementary figure
SupplementaryTable15.xlsx
Supplementary table 1-5
SupplementaryTable610.xlsx
Supplementary table 6-10
SupplementaryTable1115.xlsx
Supplementary table 11-15
SupplementaryTable1622.xlsx
Supplementary table 16-22
SupplementaryTable23.xlsx
Supplementary table 23

Download PDF

Version 1

posted

You are reading this latest preprint version

PKM2 Confers Purine Metabolic Advantage and Enhances Lenvatinib Resistance in Hepatocellular Carcinoma

Status:

Version 1

Abstract

Figures

Introduction

Materials and methods

Human HCC tumor specimens

Tumor dissociation

HCC patients-derived organoids culture

Cell viability assay

Cell transfection

Quantitive real-time PCR (qPCR)

Immunoblotting

Hematoxylin-eosin (HE) and immunohistochemistry staining

Tumor growth studies

LC-MS/MS based metabolomics

Extraction of nuclear and cytoplasmic components

CO-IP

CUT&Tag

Luciferase reporter assay

scRNA-Seq data analysis

Gene set enrichment analysis

Differential expression analysis

Whole genome sequencing analysis

Results

Increased chemoresistance signatures in Len-R organoids of HCC

Lenvatinib resistance property is dispersed in each subpopulation by scRNA-Seq analyses

PKM2 contributes to lenvatinib chemoresistance

Enrichment of purine metabolic throughput maintains lenvatinib resistance

PKM2 mediates purine metabolism through transcriptional upregulation of DHFR

PKM2 and DHFR expression levels negatively correlate with lenvatinib response in HCC specimens

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1