Chemoresistance mechanisms to 5-Fluorouracil and reversal strategies in lung and breast cancer

Abstract

Chemotherapy drug 5-Fluorouracil (5-FU) is a major treatment for many cancers; however, its efficacy is limited by chemoresistance. Here, we investigate the resistance mechanisms to 5-FU and reversal strategies in lung and breast cancer cells. Using multiple 5-FU-resistant lung cancer and breast cancer cell models, we reveal differential cellular and molecular features of 5-FU resistance between different cancer types. We further unravel the implications of immune-related processes, NOTCH and WNT signaling with 5-FU resistance. In lung cancer, the activation of WNT/β-catenin signaling promotes the resistance and blocking this signaling re-sensitizes resistant cells to 5-FU treatment. Our study not only reveals differential features and mechanisms underlying 5-FU resistance across different cancers, but also suggests potential strategies against such resistance.

Introduction

5-Fluorouracil (5-FU) is a widely used chemotherapy drug in the treatment of various solid tumors such as colorectal cancer (CRC), head and neck, pancreas, breast, and ovarian cancers¹. Some of the oral 5-FU pro-drugs (e.g., S‑1, tegafur and capecitabine) were also approved to treat lung, liver, prostate, and cervical cancers^2,3. As an antimetabolite and an analogue of uracil, the fluoropyrimidine 5-FU exerts cytotoxic function by inhibition of the nucleotide synthetic enzyme thymidylate synthase (TS) and misincorporation of its metabolites into DNA and RNA, thereby resulting in aberrant nucleotide metabolism in the cells¹.

5-FU has been proved as one of the most successful chemotherapy drugs for cancer treatment since its development in1950s. Nevertheless, drug resistance often happens for patients administrated with 5-FU or 5-FU-containing combo regimen. Several lines of mechanisms were proposed to mediate chemoresistance to 5-FU and other chemo drugs, including but not limited to increased drug efflux or decreased influx into the cells, altered drug metabolism and target, aberrant DNA repair machinery, changes of the cell cycle and survival pathways, epithelial-to-mesenchymal transition, cancer stem cell involvement, imbalanced redox status, and abnormal tumor microenvironment^{4, 5, 6, 7–8}. Specifically for 5-FU, the expression levels of nucleotide metabolizing enzymes such as TS and dihydropyrimidine dehydrogenase can determine the resistance status of tumors to 5-FU⁹. Moreover, several signaling pathways such as NOTCH and WNT pathways were reported to be involved in developing 5-FU resistance possibly via their regulation on cancer stem cells^{10, 11, 12, 13–14}. In addition, a bunch of microRNAs and long noncoding RNAs were identified to underlie 5-FU resistance through modulation of the above resistance-associated signaling pathways or cellular processes⁵. However, most of these mechanisms of 5-FU resistance were learned from studies conducted in CRC. Whether other types of cancer employ similar routes to develop 5-FU resistance remains largely unknown, and how to tackle such resistance in clinics is still challenging.

Here we aim to investigate how lung cancer and breast cancer cells develop 5-FU resistance and explore potential strategies to overcome such resistance. Multiple 5-FU resistance cell lines models were established via either traditional drug pulsing method or specific gene knockout approach. Comparative analysis of these diverse and orthogonal resistance models from multiple angles reveals both consensus and interestingly different features associated with 5-FU resistance across different types of cancers and models. We further provide compelling evidence to show that WNT/β-catenin signaling is a primary driver to 5-FU resistance in lung cancer but not in breast cancer cells, and targeting WNT pathways might be a promising strategy to reverse 5-FU resistance in lung cancer. These results highlight the varied mechanisms adopted by different types of cancer to acquire 5-FU resistance and pave the way to developing effective solutions against such resistance during clinical practice.

Results

Establishment of multiple 5-FU-resistant cells

We firstly derived 5-FU-resistant cell line models using lung cancer NCI-H1568 cells and breast cancer MCF7 cells by long-term stepwise drug adaptation (Materials and methods). Compared to parental cells, 5-FU-resistant (5-FUR) NCI-H1568^5-FUR lines were more viable upon 5-FU treatment (Fig. 1a,b). The IC₅₀ value of NCI-H1568^5-FUR to 5-FU was more than ten folds higher than that of parental control cells (14.5 μM vs. 1.2 μM) (Fig. 1c,d). Likewise, the MCF7^5-FUR cells also displayed significant resistance to 5-FU than WT parental cells (Fig. 1e-h), suggesting that the establishment of 5-FU-resistant cell lines was successful. Furthermore, during our previous efforts to systematically identify chemoresistance-related genes by genome-wide CRISPR knockout screens¹⁵, we revealed a set of genes (highlighted in red) whose loss-of-function might drive the 5-FU resistance in NCI-H1568 or MCF7 cells (Fig. 1i,j). Metabolism-related genes were enriched among top hits in NCI-H1568 cells but not in MCF7 cells (Fig. 1i,j). Using two independent sgRNAs to knockout two top hits MMACHC (a gene relating to cobalamin or vitamin B12 metabolism) or MED19 (encoding a subunit of the Mediator complex) in NCI-H1568 or MCF7 cells, respectively, we found that these gene-specific knockout cells also exhibited comparable resistance to 5-FU (5 μM for NCI-H1568; 20 μM for MCF7) as NCI-H1568^5-FUR or MCF7^5-FUR cells did (Fig. 1k-n; ¹⁵). Such specific gene knockout approach significantly expedited the derivation of 5-FU-resistance cell lines and provided diverse and orthogonal cell models to investigate 5-FU resistance.

Fig. 1 [Images not available. See PDF.]

Derivation of multiple 5-FU-resistant cell lines. (a) Cell growth evaluation of parental and 5-FU-resistant (5-FUR) lung cancer NCI-H1568 cells upon 5-FU treatment. Mean ± SD with n = 3. Unpaired t test, ***p < 0.001. (b) Crystal violet staining of parental and NCI-H1568^5-FUR cells. (c) MTT-based cell viability assay for parental and NCI-H1568^5-FUR cells. Mean ± SD with n = 3. (d) IC₅₀ values of parental and NCI-H1568^5-FUR cells in response to 5-FU. (e–h) Evaluation results of cell growth (e), crystal violet staining (f), MTT assay (g) and IC₅₀ values (h) for parental and breast cancer MCF7^5-FUR cells. (i-j) Ranked genes from genome-wide CRISPR knockout screen in NCI-H1568 (i) or MCF7 (j) cells against 5-FU. Red highlighted gene indicates a resistance phenotype upon gene knockout while blue indicates lethal upon 5-FU treatment. (k-l) Crystal violet staining showing a resistance phenotype to 5-FU when MMACHC or MED19 was knocked out using two independent sgRNAs (-1 and -2) in NCI-H1568 (k, 5 μM for 9 days) or MCF7 (l, 20 μM for 12 days) cells, respectively. (m–n) Cell growth evaluation of control (vector, sgAAVS1) and 5-FU-resistant (5-FUR) NCI-H1568 (m, 5 μM for 8 days) or MCF7 (n, 20 μM for 11 days) cells derived from MMACHC or MED19 knockout, respectively. Mean ± SD with n = 3. Unpaired t test (5-FU vs. DMSO), ***p < 0.001.

Cellular and molecular features of 5-FU-resistant cells

Next, we tried to characterize these 5-FU-resistant cells from multiple aspects. Compared to sensitive control cells, multiple resistance lines showed no apparent morphology change and had significantly fewer dead cells upon 5-FU treatment (Supplementary Fig. 1a,b). Cell cycle analysis of lung cancer NCI-H1568 cells showed that 5-FU (5 μM) significantly reduced cell fraction of G1 and G2/M phase while increased the cell population of S phase for sensitive control cells. In contrast, three resistant lines established via different routes all successfully counteracted such effect (Fig. 2a). Meanwhile, 5-FU treatment caused slightly increased “late apoptosis and necrosis” rather than “early apoptosis” for parental NCI-H1568 cells at given drug concentration (5 μM), which effect was also dismissed in multiple lines of resistant cells (Fig. 2b,c; Supplementary Figs. 2a,b, 3a,b). When looking into breast cancer MCF7 cells, the cellular effects of 5-FU was different. Sensitive control MCF7 cells were significantly arrested at G1 phase with reduced cell fraction of S phase upon 20 μM 5-FU challenge, while multiple resistant MCF7 cells did not exhibit such change (Fig. 2d). In addition, neither late apoptosis nor necrosis was significantly induced by 5-FU at given dose (20 μM) (Fig. 2e,f; Supplementary Fig. 3c,d). However, at a higher dose (40 μM) or longer treatment duration (96 h rather than 48 h), 5-FU still induced significant apoptosis and necrosis in sensitive control cells but not or to a lesser extent in multiple resistant lines (Supplementary Fig. 4a–d). In addition, the concentration of 5-FU also matters in determining its cytostatic effect on cancer cells, as revealed by cell cycle analysis using reciprocally interchanged 5-FU concentration in both NCI-H1568 and MCF7 cells (Supplementary Fig. 5a,b). Similar trends were observed for both sensitive parental cell lines at either 5 μM or 20 μM, but the cell cycle effects were different between 5 μM and 20 μM doses (Fig. 2a,d; Supplementary Fig. 5a,b). These results suggest that cell cycle control might be one of the major modes of action for 5-FU to regulate tumor growth in vitro before inducing apparent cell death at higher concentration.

Fig. 2 [Images not available. See PDF.]

Cellular and molecular features of 5-FU-resistant cells. (a) Cell cycle analysis by propidium iodide (PI) staining of indicated NCI-H1568 cells in the absence (DMSO) or presence of 5 μM 5-FU for 48 h. Mean ± SD with n = 3. Unpaired t test, **p < 0.01 and ***p < 0.001. (b-c) Early apoptosis (b) or late apoptosis and necrosis (c) analysis of indicated cells in the absence (DMSO) or presence of 5 μM 5-FU for 48 h by Hoechst 33,342 and PI staining method. Mean ± SD with n = 3. (d-f) Cell cycle (d), early apoptosis (e) and late apoptosis and necrosis (f) analysis of indicated MCF7 cells in the absence (DMSO) or presence of 20 μM 5-FU for 48 h. Mean ± SD with n = 3. Unpaired t test, **p < 0.01 and ***p < 0.001. (g-h) Immunoblotting analysis of indicated proteins for multiple 5-FU-sensitive or -resistant NCI-H1568 cells lines in the absence (DMSO) or presence of 5 μM 5-FU for 48 h (left). Quantification of relative protein expression from the immunoblot image was shown on the right. Mean ± SD with n = 3. Unpaired t test, *p < 0.05, **p < 0.01 and ***p < 0.001.

We then examined whether some typical cell growth related signaling cascades were altered in resistant cell models. Interestingly, we found a reduced level of phosphorylated ERK1/2 (p-EKR1/2) in both NCI-H1568^5-FUR and MCF7^5-FUR resistant cells (Fig. 2g; Supplementary Fig. 6a). Furthermore, GSK3β phosphorylation was apparently elevated in NCI-H1568^5-FUR cells but not in MCF7^5-FUR cells. We did not observe consistent change for phosphorylated AKT (p-AKT) in MCF7 cells and p-AKT was barely detected in NCI-H1568 cells (Fig. 2g; Supplementary Fig. 6a). Similar results were obtained using specific gene knockout 5-FU resistance cell models (Fig. 2h; Supplementary Fig. 6b). These data further highlight the differences of resistance features between different cancer types. On the other hand, multiple resistance models derived via either traditional way or specific gene knockout on the same genetic background tended to behave similarly in terms of cellular processes (Fig. 2a-f) and downstream signaling activities (Fig. 2g,h; Supplementary Fig. 6a,b).

Gene expression signatures underlying 5-FU resistance

To further unravel the molecular mechanisms of 5-FU resistance in various cell models, we performed transcriptome analyses of each paired sensitive and resistant cell lines by RNA sequencing (RNA-Seq) (Supplementary Table 1). To obtain 5-FU resistance-associated gene signatures, gene expression profiles of each 5-FU-resistant cell line were compared to their parental control sensitive cells for differentially expressed genes (DEGs). Varied number of DEGs were identified in different resistance models. A significant portion of DEGs were shared between NCI-H1568^5-FUR and ΔMMACHC^5-FUR-1 resistance lines (Fig. 3a), further supporting a consensus of molecular feature and underlying mechanism across different 5-FU resistance models of lung cancer. In contrast, very few common DEGs were identified in breast cancer MCF7^5-FUR and ΔMED19^5-FUR-1 resistance models (Fig. 3b), suggesting that divergent routes and mechanisms might exist toward 5-FU resistance even for the same parental cell background. Functional enrichment analysis of common DEGs in the two NCI-H1568 resistance models showed that several signaling pathways such as “Interferon signaling”, “NOTCH signaling” and “WNT signaling” were highly enriched especially in those up-regulated DEGs (Fig. 3c). The implications of “immunity” or “cell adhesion” processes here highlight the potential importance of cell–cell interactions between cancer cells in these cell-autonomous 5-FU resistance models without a real tumor microenvironment. We further checked the individual genes enriched for NOTCH and WNT pathways, and found that several WNT ligands (e.g., WNT6, WNT10A, WNT5A, WNT5B and WNT7B) as well as other WNT signaling proteins were all present in both NOTCH and WNT signaling terms (Fig. 3d,e). Given such significant enrichment of WNT pathway components, we surmised that WNT signaling activation might be a primary driver for lung cancer resistance to 5-FU. Furthermore, higher expression of WNT signaling pathway genes (GO:0016055) is significantly associated with bad patient survival shown in three independent lung cancer cohorts (GEO accession: GSE30219, GSE31210 and GSE157011)^{16, 17, 18–19} (Fig. 3f-h). In addition, to establish a clinical relevance of our resistance gene signatures here, we examined the association status between those commonly up-regulated DEGs in NCI-H1568 resistance models and patient survival in a lung cancer cohort (GEO accession: GSE31210)^18,19. Indeed, we observed that high expression of these gene signatures correlated with bad patient survival (Fig. 3i), suggesting a potential driver function toward drug resistance for these up-regulated DEGs.

Fig. 3 [Images not available. See PDF.]

The gene expression profiles of 5-FU-sensitive and -resistant cells. (a) Venn diagrams of overlapped up-regulated or down-regulated genes between NCI-H1568^5-FUR vs. untreated Mock and ΔMMACHC^5-FUR-1 vs. Vector control cells determined by RNA-seq. (b) Venn diagrams of overlapped up-regulated or down-regulated genes between MCF7^5-FUR vs. untreated Mock and ΔMED19^5-FUR-1 vs. Vector control cells determined by RNA-seq. (c) Functional enrichment analysis showing the top enriched terms among NCI-H1568^5-FUR and ΔMMACHC^5-FUR-1 commonly up- or down-regulated genes. (d-e) Highlight of individual gene expression within NOTCH signaling (d) or WNT signaling (e) across the indicated 5-FU-resistant NCI-H1568 cell lines. (f–h) Survival analysis of lung cancer patient cohorts (GSE30219 for f; GSE31210 for g; GSE157011 for h) with gene expression levels of WNT signaling pathway (GO:0,016,055). Unpaired two-sided t test. (i) Survival analysis of commonly up-regulated genes in NCI-H1568^5-FUR and ΔMMACHC^5-FUR-1 resistance models using a lung cancer patient cohort (GSE1210). (j) Expression change of typical ABC transporter genes in four 5-FU-resistant cell models.

For breast cancer MCF7 resistance models, we also explored their enriched functional terms using the resistance-associated DEGs. Although the DEGs varied significantly between the two MCF7^5-FUR and ΔMED19^5-FUR-1 resistance models, we unexpectedly found a consensus at the functional term levels (Supplementary Fig. 7a,b). Immune-related “Th17-derived cytokines” stood as the top first term for both resistance models. Furthermore, other “immune-related” functions or “cell adhesion” processes also appeared among the top 10 enriched terms, which were not only shared between the two MCF7 resistance models (Supplementary Fig. 7a,b), but also consistent with lung cancer NCI-H1568 resistance models (Fig. 3c). These results indicate that cancer cells might employ different genes with convergent functions to drive 5-FU chemoresistance. Notably, we did not find the implication of WNT signaling in MCF7 resistance models, suggesting a specific role of WNT pathway only in lung cancer resistance models here.

ABC (ATP-binding cassette) drug transporter proteins were usually involved in multidrug resistance⁷. We examined the expression status of several typical ABC transporter genes in our 5-FU resistance models. Indeed, we found several of them were significantly up- or down-regulated in 5-FU-resistant cells (Fig. 3j). For example, ABCB1 and ABCC4 showed highly increased expression in all of the four 5-FU-resistant lines, consistent with their roles in drug transport and promoting resistance to multiple chemotherapeutic agents⁷. These results support drug transport as a general mechanism during multidrug resistance including 5-FU chemoresistance.

WNT/β-catenin signaling as a driver and vulnerability for 5-FU chemoresistance in lung cancer cells

Given the significant implication of WNT signaling in the above analysis, we next tried to explore whether WNT signaling indeed drives 5-FU resistance in lung cancer cells and the possibility of targeting this pathway to antagonize drug resistance. Using multiple 5-FU-resistant NCI-H1568 cell lines, we firstly confirmed enhanced WNT signaling activities as evidenced by increased GSK3β phosphorylation and cytoplasmic β-catenin protein accumulation in 5-FU-resistant cells with or without 5-FU addition (Fig. 4a). We then employed Wnt3a conditioned media (Wnt3aCM) to activate WNT signaling activity in normal NCI-H1568 cells (Fig. 4b). Wnt3aCM-treated cells displayed significantly better cell survival in the presence of 5-FU compared to control cells with lower WNT signaling activity (Fig. 4c), suggesting that WNT/β-catenin signaling was indeed a driver for lung cancer cells to develop 5-FU resistance. Next, we used IWP-2, an inhibitor of WNT processing and secretion²⁰, to antagonize WNT/β-catenin signaling activity in normal NCI-H1568 cells (Fig. 4d and Supplementary Fig. 8a). The inhibition of WNT signaling not only decreased cell growth in the vehicle condition, but also caused more severe cytotoxic effect in the presence of 5-FU (Fig. 4e). More importantly, even in 5-FU-resistant NCI-H1568^5-FUR cells, targeted inhibition of WNT signaling with IWP-2 could effectively kill resistant tumor cells especially when in combination with 5-FU (Fig. 4f,g). Furthermore, we also employed another WNT signaling inhibitor XAV-939 (Supplementary Fig. 8b), which enhances AXIN stability through tankyrase inhibition²⁰, and obtained similar results as IWP-2 (Fig. 4h-k). To exclude the cell line-specific effect, we further established another 5-FU-resistant cell line on A549 cell background aided by TP53 gene knockout according to our previous CRISPR screen data¹⁵ (Supplementary Fig. 9a-c). Consistently, Wnt3aCM treatment conferred resistance to 5-FU on normal A549 cells (Supplementary Fig. 9d-e), and inactivation of WNT signaling by IWP-2 sensitized both normal and resistant cells to 5-FU (Supplementary Fig. 9f-i). These results demonstrate that WNT/β-catenin signaling serves as both a driver and therapeutic vulnerability for 5-FU resistance in lung cancer cells.

Fig. 4 [Images not available. See PDF.]

Functions of WNT/β-catenin signaling during 5-FU resistance in lung cancer cells. (a) Immunoblotting analysis of indicated proteins for multiple 5-FU-sensitive or -resistant NCI-H1568 cells in the absence (DMSO) or presence of 5 μM 5-FU for 48 h. (b) Immunoblotting analysis of cytoplasmic β-catenin in NCI-H1568 cells treated with Wnt3aCM and/or 5-Fluorouracil (5 μM) for 10 days. (c) Cell growth evaluation of indicated NCI-H1568 cells after 10 days treatment of Wnt3aCM and/or 5-FU (5 μM). Mean ± SD with n = 3. Unpaired t test, *p < 0.05. (d-g) Immunoblotting analysis of cytoplasmic β-catenin in NCI-H1568 cells (d) or NCI-H1568^5-FUR cells (f) and cell growth evaluation of NCI-H1568 cells (e) or NCI-H1568^5-FUR cells (g) treated with IWP-2 (1 μM) and/or 5-FU (5 μM) for 7 days. Mean ± SD with n = 3. Unpaired t test, *p < 0.05 and **p < 0.01. (h–k) Immunoblotting analysis of cytoplasmic β-catenin in NCI-H1568 cells (h) or NCI-H1568^5-FUR cells (j) and cell growth evaluation of NCI-H1568 cells (i) or NCI-H1568^5-FUR cells (k) treated with XAV-939 (1 μM) and/or 5-FU (5 μM) for 5 days (h, i) or 8 days (j, k). Mean ± SD with n = 3. Unpaired t test, *p < 0.05.

Discussion

The resistance mechanisms to 5-FU were mostly studied in CRC and how other solid tumors develop 5-FU resistance remains elusive. To fill this gap, this study employed multiple lines of lung cancer and breast cancer cell models and explored several cellular and molecular features associated with 5-FU resistance. Both shared and differential mechanisms were found between different cancers, and potential reversal strategies were proposed against 5-FU resistance in lung cancer cells.

From the genetic basis, a differential set of genes implicated in either metabolism or other processes were identified to drive 5-FU resistance in lung or breast cancer cells by CRISPR screening, manifesting diversified routes to chemoresistance. In both lung and breast cancer cells, we found aberrant cell cycle control upon 5-FU treatment. Interestingly, compared to cell cycle arrest, cell death was not an apparent effect in our in vitro assays using given 5-FU concentrations. Nevertheless, cell death effect inevitably becomes apparent as 5-FU concentration increases. These results add more complexity to the cytotoxic effect of 5-FU and suggest that multiple modes of action might be at work partially depending on the target cell type and accessible drug concentration. Notably, all the 5-FU-resistant cells could counteract these abnormal cellular effects elicited by 5-FU. We found a consensus p-ERK decrease in all the 5-FU-resistant cell models. This was probably in agreement with the findings that ERK activation can mediate cell cycle arrest and apoptosis after DNA damage^{21, 22–23}. The implication of immune-related processes such as “interferon signaling” in both lung and breast cancer resistance models was intriguing. Moreover, “Th17-derived cytokines” was significantly associated with 5-FU resistance only in breast cancer model but not in that of lung cancer. Previous studies indicated that interferon could enhance the efficacy of 5-FU to induce cell death and cancer-cell-intrinsic STING signaling was required for effective 5-FU responsiveness in vivo^{24, 25–26}. Furthermore, 5-FU treatment led to increased intratumoral T cells which were necessary for the full efficacy of 5-FU in vivo²⁴. Here we further linked these immune-related processes to the development of 5-FU resistance in vitro, suggesting that, even without real tumor microenvironment and interactions with immune or other cell types, the cancer-cell-intrinsic immunity might also play a role in driving 5-FU resistance. Further work is needed to validate the implication of such immunity during 5-FU resistance.

Finally, we pinpointed that WNT/β-catenin signaling not only promoted 5-FU resistance in lung cancer cell models, but also served as a therapeutic target to overcome such resistance (Fig. 5). The association of WNT signaling activity with poor survival further supports the clinical significance of WNT pathway in lung cancer (Fig. 3f-h). Similar findings were also reported in CRC^10,11,14. One study suggested that WNT signal activation confers 5-FU resistance in CRC via suppressing the DNA damage checkpoint CHK1 pathway¹⁴, while another study claimed that 5-FU can activate cancer stem cells in CRC via p53-induced WNT3 transcription and resulting WNT/β-catenin pathway activation¹⁰. However, in lung cancer, the link between WNT signaling and 5-FU resistance are not reported before. Among the top up-regulated genes in 5-FU-resistant lung cancer cells, several WNT signal genes were present but not for WNT3 (Fig. 3e), suggesting that differential mechanisms might exist between lung cancer and CRC. In addition, we did not identify WNT as a key player in breast cancer 5-FU resistance models, suggesting a regulatory specificity for lung cancer and a necessity to study each relevant cancer type to fully unleash the potency of 5-FU in more solid tumors. More in vivo evidence is also needed in future work to further prove the efficacy of targeting WNT pathway to overcome 5-FU resistance in lung cancer. Given that many WNT targeted drugs are currently under investigation or clinical trials^27,28, it is promising to use these therapies for managing 5-FU-resistant lung cancer as well as CRC clinically.

Fig. 5 [Images not available. See PDF.]

A schematic model illustrating the roles and mechanisms of WNT signaling in 5-FU resistance of lung cancer cells. In 5-FU-resistant cells, the gene expression and activities for NOTCH and WNT signaling tend to be high compared to sensitive control cells. Enhanced WNT signaling leads to cytosolic β-catenin activation which then elicits downstream gene expression program in the nucleus. Activation of WNT target genes promotes cell proliferation and resistance to 5-FU in lung cancer cells. Blocking WNT signaling with inhibitors such as IWP-2 or XAV-939 helps to re-sensitize resistant cells to 5-FU and might serve as a therapeutic option to antagonize 5-FU resistance in lung cancer management.

In summary, our study reveals differential and intriguing mechanisms of 5-FU resistance in lung and breast cancer cell models, provides potential strategies to reverse such drug resistance, and emphasizes the need to study 5-FU resistance in a broader range of relevant cancer types.

Materials and methods

Cell culture

HEK293FT (Cat# CRL-1573), MCF7 (Cat# HTB-22), NCI-H1568 (Cat# CRL-5876) and A549 (Cat# CRM-CCL-185) cells were obtained from the American Type Culture Collection (ATCC). All the cells were regularly tested negative for mycoplasma contamination. Cells were maintained in DMEM (for HEK293FT, MCF7 and A549 cells) or RPMI 1640 (for NCI-H1568 cells) media supplemented with 10% fetal bovine serum plus 1% penicillin/streptomycin at 37 °C with 5% CO₂. The Wnt3a conditioned medium (Wnt3aCM) was generated by L-Wnt cells stably expressing Wnt3a.

Lentivirus production

Lentiviruses were generated by co-transfecting indicated lentivectors together with the packaging plasmids pCMVR8.74 and pMD2.G using Opti-MEM (Gibco) and Lipofectamine 2000 reagent (Invitrogen) in HEK293FT cells. The viral supernatants were collected at 48 h (h) or 72 h post transfection and centrifuged at 3000 rpm for 5 min to remove the cell debris. The viral supernatants collected were subsequently aliquoted and stored at − 80 °C before use.

Resistance gene selection from genome-wide CRISPR screens

To create gene-specific resistance cell models, the candidate resistance genes were selected from results of our previous genome-wide CRISPR screens¹⁵. Briefly, we employed a genome-wide CRISPR knockout library (Addgene, #1,000,000,132) which contains 92,817 gRNAs targeting 18,436 genes (5 gRNAs per gene) in the human genome and is constructed under lentiCRISPRv2-puro backbone. Pooled lentiviruses encapsulating Cas9 and conjugated gRNAs in the library were produced in HEK293FT cells. MCF7, A549 and NCI-H1568 cells were infected with lentiviruses at a low MOI (< 0.3). After 48 h, infected cells were selected with puromycin for 3 days followed by recovery for additional two days. Day 0 (start point of screen) samples were collected at seven days post infection, and the rest of cells were treated with DMSO or 5-FU (MCF7-20 μM, A549-5 μM and NCI-H1568-5 μM) for 15 days (MCF7), 21 days (A549), and 21 days (NCI-H1568), respectively before harvesting the end-point samples. Genomic DNA was extracted from these samples to amplify the sgRNA fragment. High-throughput sequencing (PE150) was performed to determine the abundance of the sgRNAs. The MAGeCK algorithm was employed to analyze the data. The top hits whose gene knockout confer strong drug resistance phenotype were considered as candidate genes to generate gene-specific 5-FU-resistant cell models.

Establishment of 5-FU-resistant cell lines

To generate MCF7^5-FUR and NCI-H1568^5-FUR 5-FU-resistant cell lines, wild type (WT) MCF7 and NCI-H1568 cells were treated with a stepwise increase of 5-Fluorouracil for about 2–3 months to continually keep the survived cells until significant resistance appeared. Generally, cells were treated starting at a concentration of 1 µM 5-FU, followed by gradually increasing concentrations until the residual cells displays an IC₅₀ value greater than tens of times than that of parental WT cells. For resistant cells derived from specific gene knockout, WT MCF7, NCI-H1568 and A549 cells were infected by lentiviruses containing Cas9 and specific single guide RNAs (sgRNAs). Two independent sgRNAs were used (denoted as -1 and -2). After 48 h, cells were selected with puromycin (2, 2, and 1 μg/ml for MCF7, NCI-H1568 and A549 cells, respectively) for 3 days followed by recovery for additional two days. Successfully infected cells were further treated with 5-FU for 15 days (ΔMED19^5-FUR-1 and ΔMED19^5-FUR-2) (5-FU 20 µM), 19 days (ΔMMACHC^5-FUR-1 and ΔMMACHC^5-FUR-2) (5-FU 5 µM) or 10 days (ΔTP53^5-FUR-1 and ΔTP53^5-FUR-2) (5-FU 5 µM) to obtain a stable 5-FU-resistance phenotype. Cells expressing empty lentiCRISPRv2-puro vector or sgRNA targeting the AAVS1 locus (sgAAVS1) served as control for gene-specific resistant cells. The primers for cloning sgRNAs plasmids are as follows (5′ to 3′): sgRNA_MED19-1_Forward: CACCGCACCGCCTCCTGCGGGCGG, sgRNA_MED19-1_Reverse: AAACCCGCCCGCAGGAGGCGGTGC, sgRNA_MED19-2_Forward: CACCGGCCACGGCTCCTCCTGGCG, sgRNA_MED19-2_Reverse: AAACCGCCAGGAGGAGCCGTGGCC, sgRNA_MMACHC-1_Forward: CACCGTGTGGAGGCTGACCCATGG, sgRNA_MMACHC-1_Reverse: AAACCCATGGGTCAGCCTCCACAC, sgRNA_MMACHC-2_Forward: CACCGCTAACACGGCCCAGATGGT, sgRNA_MMACHC-2_Reverse: AAACACCATCTGGGCCGTGTTAGC, sgRNA_AAVS1_Forward: CACCGCCTGTCATGGCATCTTCCAG, sgRNA_AAVS1_Reverse: AAACCTGGAAGATGCCATGACAGGC.

Cell viability assay

Cell viability for IC₅₀ determination were based on MTT tetrazolium salt colorimetry. Cells (5 × 10³ per well) were plated in 96-well plates with 100 μL medium and treated with 5-FU for 3 days followed by 50 μg MTT addition. Cell viability was determined by the absorbance value at 490 nm using a BioTek microplate spectrophotometer (Gene Company Limited). For other cell growth assay, the cell number was directly counted using a hemocytometer. Cell survival status was visualized by crystal violet staining. Tested cells were washed with cold PBS and fixed with 100% methanol for 15 min at − 20 °C. Subsequently, cells were stained with crystal violet and incubated for 15 min at room temperature. Wash away crystal violet with tap water and then use a camera to acquire images of cell staining.

RNA-seq analysis

Around 10⁷ cells were harvested for each RNA-seq sample in triplicates. RNA was extracted by TRIzol reagent and strand-specific libraries were prepared by Novogene followed by Illumina PE150 sequencing. Raw reads were aligned to the hg38 human reference genome with the UCSC known gene transcript annotation using HISAT2 with default parameters. Gene counts were quantified by HTSeq. Differentially expressed genes were identified by DESeq2 with the cutoff of |log2 (fold change)|> 2 & p.adjust < 0.01 & top 500 genes. MetaCore functional enrichment analysis were performed using the functions of “enricher” in clusterProfiler R package.

Immunoblotting assay

Cells were lysed in RIPA buffer (Beyotime) containing phosphatase inhibitors (Meilunbio, # MB12707) and protease inhibitors (Meilunbio, # MB26780) for 15 min at 4 °C followed by centrifuging at 14,000 rpm for 10 min at 4 °C. For detection of cytoplasmic β-catenin, the lysis buffer consists of 1 × TBS, 10% Glycerol, 0.015% digitonin, 50 mM PMSF, 50 mM Sodium Fluoride, 100 mM Sodium Orthovanadate and 1 M EDTA, pH = 7.4. Mix the protein supernatant with sample loading buffer. Separate proteins by running on a 10% bis–tris polyacrylamide gel and then transfer the proteins to nitrocellulose (NC) membrane (Pall Corporation, #27,574,625). After sequential incubation with primary and secondary antibodies, the signal was detected using a chemiluminescent imaging system (Tanon-5200). The following antibodies were used: GAPDH (Santa Cruz Biotechnology, Cat# sc-25778), p-AKT (Ser473) (Cell Signaling Technology, Cat# D9E), AKT (Proteintech, Cat# 10,176–2-AP), p-GSK3β (Ser 9) (Beyotime, Cat# AF1531), GSK3β (Proteintech, Cat# 22,104–1-AP), p-ERK1/2 (Cell Signaling Technology, Cat# 4370S), ERK1/2 (Cell Signaling Technology, Cat# 137F5), β-catenin (ABclonal, Cat# A19657), Goat anti-Rabbit IgG (Thermo Fisher Scientific, Cat# 31,460) and Rabbit anti-Mouse IgG (Thermo Fisher Scientific, Cat# 31,450).

Cell cycle and apoptosis analysis

For cell cycle analysis, cells (~ 70–80% confluency) were washed once with phosphate buffered saline (PBS) and fixed with 70% pre-cooled ethanol at -20 °C overnight. After washing with PBS, add 500 μL propidium iodide (PI)/RNase staining buffer (BD Pharmingen, # 550,825) and incubate at room temperature for 30 min in the dark. Filter the samples with 200-mesh nylon membrane and proceed to flow cytometry for cell cycle analysis (BD LSRFortessa). For apoptosis assays, cells were stained by two methods (PI + Hoechst or PI + Annexin V) and analyzed by flow cytometry. For PI plus Hoechst staining, ~ 1 million cells were collected by centrifugation and the cell pellet was resuspended with 0.8 mL of cell staining buffer in Apoptosis and Necrosis Assay Kit (Beyotime # C1056). Add 5 μL Hoechst 33,342 staining solution and 5 μL PI staining solution. Mix well and incubate on ice for 20–30 min. Pass the cells through 200-mesh nylon membrane and proceed to flow cytometry analysis (BD LSRFortessa). For PI plus Annexin V (FITC) staining, cells were collected and washed once with PBS. Take about 1 × 10⁵ resuspended cells, add 195 μL Annexin V-FITC binding solution, 5 μL Annexin V-FITC (Beyotime # C1062) and 10 μL of PI staining solution (Beyotime # C1062). Mix gently and incubate in the dark for 10–20 min at room temperature. Cells were passed through 200-mesh nylon membrane and analyzed on a flow cytometer (BD LSRFortessa). The gating strategy for flow cytometry analysis was shown in Supplementary Fig. 10.

Statistical analysis

GraphPad Prism 9 software and the R package were employed for statistical analysis. The replicate data were shown as Mean ± SD. The unpaired two-sided t test was used to compare between two groups. Asterisks indicate *p < 0.05, **p < 0.01, and ***p < 0.001.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (31871344; 32071441), Guangdong Basic and Applied Basic Research Foundation (2023A1515140084), the 111 Project (B16009), the Construction Project of Liaoning Provincial Key Laboratory, China (2022JH13/10200026) and LiaoNing Revitalization Talents Program (XLYC1807212) to T.F.

Author contributions

T.F. conceived and designed the research. C.Z. and S. W. performed most of the experiments. W.-J.J. and Z.L. conducted the bioinformatics analysis. All the authors analyzed the data. T.F. wrote the manuscript with the input from C.Z. T.F. supervised the study.

Data availability

The raw sequence data generated in this study have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA007245) at https://ngdc.cncb.ac.cn/gsa-human.

Declarations

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-025-90532-z.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Longley, DB; Harkin, DP; Johnston, PG. 5-fluorouracil: Mechanisms of action and clinical strategies. Nat. Rev. Cancer.; 2003; 3, pp. 330-338.1:CAS:528:DC%2BD3sXjtlals70%3D [DOI: https://dx.doi.org/10.1038/nrc1074] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12724731]

2. Urushiyama, H et al. Oral fluorouracil vs vinorelbine plus cisplatin as adjuvant chemotherapy for stage II-IIIA non-small cell lung cancer: Propensity score-matched and instrumental variable analyses. Cancer Med.; 2018; 7, pp. 4863-4869.1:CAS:528:DC%2BC1cXhvF2mur7L [DOI: https://dx.doi.org/10.1002/cam4.1725] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30151905][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6198210]

3. Wilson, PM; Danenberg, PV; Johnston, PG; Lenz, HJ; Ladner, RD. Standing the test of time: Targeting thymidylate biosynthesis in cancer therapy. Nat. Rev. Clin. Oncol.; 2014; 11, pp. 282-298.1:CAS:528:DC%2BC2cXmsFagtbc%3D [DOI: https://dx.doi.org/10.1038/nrclinonc.2014.51] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24732946]

4. Blondy, S et al. 5-Fluorouracil resistance mechanisms in colorectal cancer: From classical pathways to promising processes. Cancer Sci.; 2020; 111, pp. 3142-3154.1:CAS:528:DC%2BB3cXhsFOlt73F [DOI: https://dx.doi.org/10.1111/cas.14532] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32536012][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7469786]

5. Ghafouri-Fard, S et al. 5-Fluorouracil: A narrative review on the role of regulatory mechanisms in driving resistance to this chemotherapeutic agent. Front. Oncol.; 2021; 11, 1:CAS:528:DC%2BB3sXht1CntbzP [DOI: https://dx.doi.org/10.3389/fonc.2021.658636] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33954114][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8092118]658636.

6. Longley, DB; Johnston, PG. Molecular mechanisms of drug resistance. J. Pathol.; 2005; 205, pp. 275-292.1:CAS:528:DC%2BD2MXhsFOgsrk%3D [DOI: https://dx.doi.org/10.1002/path.1706] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15641020]

7. Szakacs, G; Paterson, JK; Ludwig, JA; Booth-Genthe, C; Gottesman, MM. Targeting multidrug resistance in cancer. Nat. Rev. Drug. Discov.; 2006; 5, pp. 219-234.1:CAS:528:DC%2BD28XhvFOlsr8%3D [DOI: https://dx.doi.org/10.1038/nrd1984] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16518375]

8. Azwar, S; Seow, HF; Abdullah, M; Faisal Jabar, M; Mohtarrudin, N. Recent updates on mechanisms of resistance to 5-fluorouracil and reversal strategies in colon cancer treatment. Biology (Basel); 2021; [DOI: https://dx.doi.org/10.3390/biology10090854] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34571795]

9. Oguri, T et al. The role of thymidylate synthase and dihydropyrimidine dehydrogenase in resistance to 5-fluorouracil in human lung cancer cells. Lung Cancer; 2005; 49, pp. 345-351. [DOI: https://dx.doi.org/10.1016/j.lungcan.2005.05.003] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15993511]

10. Cho, YH et al. 5-FU promotes stemness of colorectal cancer via p53-mediated WNT/beta-catenin pathway activation. Nat. Commun.; 2020; 11, 5321.2020NatCo.11.5321C1:CAS:528:DC%2BB3cXitF2js7bI [DOI: https://dx.doi.org/10.1038/s41467-020-19173-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33087710][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7578039]

11. Hu, YB et al. Exosomal Wnt-induced dedifferentiation of colorectal cancer cells contributes to chemotherapy resistance. Oncogene; 2019; 38, pp. 1951-1965.1:CAS:528:DC%2BC1cXitV2qu7zO [DOI: https://dx.doi.org/10.1038/s41388-018-0557-9] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30390075]

12. Hemati, H; Kaur, J; Sobti, RC; Trehanpati, N. Inhibition of NOTCH signaling pathway chemosensitizes HCC CD133(+) cells to vincristine and 5-fluorouracil through upregulation of BBC3. Biochem. Biophys. Res. Commun.; 2020; 525, pp. 941-947.1:CAS:528:DC%2BB3cXltV2qs7w%3D [DOI: https://dx.doi.org/10.1016/j.bbrc.2020.03.009] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32173531]

13. Li, DD et al. A novel inhibitor of ADAM17 sensitizes colorectal cancer cells to 5-Fluorouracil by reversing Notch and epithelial-mesenchymal transition in vitro and in vivo. Cell Prolif.; 2018; 51, e12480.1:CAS:528:DC%2BC1cXhvVSms73M [DOI: https://dx.doi.org/10.1111/cpr.12480] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30069943][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6528951]

14. He, L et al. Wnt pathway is involved in 5-FU drug resistance of colorectal cancer cells. Exp. Mol. Med.; 2018; 50, pp. 1-12.1:CAS:528:DC%2BC1cXhsFaht77J [DOI: https://dx.doi.org/10.1038/s12276-018-0128-8] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30559383]

15. Zhong, C et al. CRISPR screens reveal convergent targeting strategies against evolutionarily distinct chemoresistance in cancer. Nat. Commun.; 2024; 15, 5502.1:CAS:528:DC%2BB2cXhsFSmsLbP [DOI: https://dx.doi.org/10.1038/s41467-024-49673-4] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38951519][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11217446]

16. Bueno, R et al. Multi-institutional prospective validation of prognostic mRNA Signatures in early stage squamous lung cancer (alliance). J. Thorac. Oncol.; 2020; 15, pp. 1748-1757.1:CAS:528:DC%2BB3cXhsFCitrnE [DOI: https://dx.doi.org/10.1016/j.jtho.2020.07.005] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32717408][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8278226]

17. Rousseaux, S et al. Ectopic activation of germline and placental genes identifies aggressive metastasis-prone lung cancers. Sci. Transl. Med; 2013; 5, 186ra166.1:CAS:528:DC%2BC3sXpvFGru7Y%3D [DOI: https://dx.doi.org/10.1126/scitranslmed.3005723]

18. Yamauchi, M et al. Epidermal growth factor receptor tyrosine kinase defines critical prognostic genes of stage I lung adenocarcinoma. PLoS One; 2012; 7, e43923.2012PLoSO..743923Y1:CAS:528:DC%2BC38XhsVaktrbL [DOI: https://dx.doi.org/10.1371/journal.pone.0043923] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23028479][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3446964]

19. Okayama, H et al. Identification of genes upregulated in ALK-positive and EGFR/KRAS/ALK-negative lung adenocarcinomas. Cancer Res; 2012; 72, pp. 100-111.1:CAS:528:DC%2BC38XhvVCmtg%3D%3D [DOI: https://dx.doi.org/10.1158/0008-5472.CAN-11-1403] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22080568]

20. Anastas, JN; Moon, RT. WNT signalling pathways as therapeutic targets in cancer. Nat. Rev. Cancer; 2013; 13, pp. 11-26.1:CAS:528:DC%2BC38XhvVKnsrjL [DOI: https://dx.doi.org/10.1038/nrc3419] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23258168]

21. Singh, S; Upadhyay, AK; Ajay, AK; Bhat, MK. p53 regulates ERK activation in carboplatin induced apoptosis in cervical carcinoma: A novel target of p53 in apoptosis. FEBS Lett.; 2007; 581, pp. 289-295.1:CAS:528:DC%2BD2sXnslyqug%3D%3D [DOI: https://dx.doi.org/10.1016/j.febslet.2006.12.035] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17208232]

22. Tang, D et al. ERK activation mediates cell cycle arrest and apoptosis after DNA damage independently of p53. J. Biol. Chem.; 2002; 277, pp. 12710-12717.1:CAS:528:DC%2BD38XivFSjsrw%3D [DOI: https://dx.doi.org/10.1074/jbc.M111598200] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/11821415]

23. Tentner, AR et al. Combined experimental and computational analysis of DNA damage signaling reveals context-dependent roles for Erk in apoptosis and G1/S arrest after genotoxic stress. Mol. Syst. Biol.; 2012; 8, 568.1:CAS:528:DC%2BC38XkvFequ7Y%3D [DOI: https://dx.doi.org/10.1038/msb.2012.1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22294094][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3296916]

24. Tian, J et al. 5-Fluorouracil efficacy requires anti-tumor immunity triggered by cancer-cell-intrinsic STING. EMBO J.; 2021; 40, 1:CAS:528:DC%2BB3MXksFelsLY%3D [DOI: https://dx.doi.org/10.15252/embj.2020106065] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33615517][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8013832]e106065.

25. Kondo, M et al. Combination of IFN-alpha and 5-fluorouracil induces apoptosis through IFN-alpha/beta receptor in human hepatocellular carcinoma cells. Clin. Cancer Res.; 2005; 11, pp. 1277-1286.1:CAS:528:DC%2BD2MXht1yntb0%3D [DOI: https://dx.doi.org/10.1158/1078-0432.1277.11.3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15709199]

26. Eguchi, H et al. Augmentation of antitumor activity of 5-fluorouracil by interferon alpha is associated with up-regulation of p27Kip1 in human hepatocellular carcinoma cells. Clin. Cancer Res.; 2000; 6, pp. 2881-2890.1:CAS:528:DC%2BD3cXlslentbc%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10914738]

27. Zhao, X et al. Blocking the WNT/beta-catenin pathway in cancer treatment:pharmacological targets and drug therapeutic potential. Heliyon; 2024; 10, e35989.1:CAS:528:DC%2BB2cXhslyrt77J [DOI: https://dx.doi.org/10.1016/j.heliyon.2024.e35989] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39253139][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11381626]

28. Yu, F et al. Wnt/beta-catenin signaling in cancers and targeted therapies. Signal Transduct. Target Ther.; 2021; 6, 307.1:CAS:528:DC%2BB3MXit1elurbK [DOI: https://dx.doi.org/10.1038/s41392-021-00701-5] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34456337][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8403677]