Targeting HNRNPU to overcome cisplatin resistance in bladder cancer

Human bladder cancer cells (T24, RT4, HT1197, SW780, RT112, HT1376) and human embryonic kidney HEK293T cells were obtained from ATCC. T24 and HEK293T were maintained in RPMI 1640 medium. The SW780, HT1197, HT1376, and RT112 cells were maintained in EMEM. RT4 cells were maintained in DMEM. All culture media were supplemented with 10% fetal bovine serum (Gibco), 100 mg/mL penicillin, and streptomycin solution (Thermo).

The bladder cancer cell lines were seeded in 96-well plates at 5,000 cells per well and incubated overnight to allow cell attachment. Different concentrations (0–100 μM) of cisplatin, paclitaxel, or doxorubicin were added to the designated wells and the plates were incubated for 3 days. After incubation, 10 µL CCK-8 solution was added to each well and incubated at 37 °C for 2 h, the light absorbance of each sample was then measured at 450 nm using a microplate reader (Biotek). The IC₅₀ values were calculated using GraphPad Prism 8.1.

The sgRNA library was obtained from Addgene. HEK 293 T cells were plated on 10-cm dishes to clone the lentiCRISPR vector and the control sgRNA (sgCtrl), following the protocol of GeCKO v2. Next, T24 cells were infected with slow virus at 0.4 functional multiplicity of infection (MOI) and selected by 2 µg/mL puromycin treatment for 48 h after transfection. Subsequently, 2 × 10⁶ transduced T24 cells were treated with 0.25 µM cisplatin or an equal volume of dimethylformamide (DMF) for 7 or 14 days. Finally, the cells were collected for subsequent genomic DNA analyses.

To identify the effect of mixed sgRNAs, transfected T24 cells were seeded into 96-well plates at a density of 2000 cells per well, and then treated with 0.6 µM cisplatin or an equal volume of DMF for 5 days. To examine the function of single sgRNA of one gene, 1 µM cisplatin or an equal volume of DMF was used to treat cells for 5 days. The cells were then incubated with 1:1000 Annexin V Alexa Fluor 488 to measure the percentage of apoptotic cells, 10 nM TMRE to examine the mitochondrial membrane potential, and 5 µM Draq5 to determine the nuclear morphology according to a previous study [16]. Finally, the cells were evaluated using the opera QEHS high-content screening system (PerkinElmer, USA). The images were collected and analyzed using the Acapella software (version 2.0; PerkinElmer, USA).

Total RNA was isolated with TRIZOL Reagent (Sigma-Aldrich) and reverse transcribed using a PrimeScript RT Master Mix kit (Takara, Japan) according to the manufacturer’s protocol. mRNA expression was determined by RT-PCR. The 2^−ΔΔCT method was used to quantify the mRNA expression. The primer sequences are listed in Supplementary Table S1.

The total protein lysates were prepared, subjected to SDS-PAGE, and transferred to PVDF membranes. To quantify the expression level of HNRNPU, membranes were probed with anti-HNRNPU antibody (catalog# 34,095, Cell Signaling Technology Inc., Danvers, MA). The loading control GAPDH was probed by anti-GAPDH antibody (Thermo Fisher Scientific Inc., Waltham, MA). Rabbit IgG HRP conjugates (Cell Signaling Technology Inc., Danvers, MA) were used as secondary antibody. To visualize the bands, ECL detection reagents were used (Thermo Fisher Scientific Inc., Waltham, MA). The protein band densitometry was measured using ImageJ software (NIH, MD).

T24 cells were seeded in 96-well plates at 5000 cells per well. After treatment with 1 µM cisplatin or DMF for 1, 2, 3, 4, and 5 days, cell proliferation was examined using the Cell Counting kit-8 (CCK-8, Dojindo, Japan) following the manufacturer’s protocol. In brief, cells were collected at different time points and a 10 µL CCK-8 solution was added to each well. After incubation at 37 °C for 2 h, the light absorbance of each sample was measured at 450 nm using a microplate reader (Biotek).

Cells were plated onto 6-cm dishes at 10⁵ cells per dish. After treatment with 1 µM cisplatin or DMF for 5 days, the cells were collected. For the apoptosis assay, cells were incubated with 10 μg/mL RNase A for 20 min, followed by treatment with 50 μg/mL propidium iodide (PI) solution for 20 min. Subsequently, the cells were stained with 10 μL Annexin V-FITC reaction reagent for 20 min. All the processes were performed at room temperature in the dark. For the cell cycle assay, the cells were fixed in 70% ethanol at -20 °C overnight after cisplatin treatment, followed by incubation with 50 μg/mL propidium iodide (PI) and 10 μg/mL RNase A at 37 °C for 30 min. Finally, the cells were analyzed using a flow cytometer (BD Biosciences), according to the manufacturer’s instructions.

After transfection, the cell culture supernatant of each experimental group was collected in a 5-mL centrifuge tube, washed once with D-Hanks. The cells were then trypsinized, the culture supernatant was terminated, and the cells were collected in the same 5-mL centrifuge tube. Cells were centrifuged at 1500 rpm for 5 min, the supernatant was discarded, and the cells were washed three times in PBS. The final density of the cell suspension was 1 × 10⁶—1 × 10⁷ cell/mL. A 1 × staining buffer was used to resuspend the cell pellet; 5 µL Annexin V-APC was used for staining, and cell pellets were transferred to a flow cytometer for analysis.

A total of 1.5 × 10⁵ cells were plated on the upper chamber of the transwell system, with 200 μL serum-free medium. Culture medium (800 μL) supplemented with 10% FBS was added to the lower chamber. After 12 h, the non-migrating cells were carefully removed using a cotton bud, and the migrated cells were fixed with 4% paraformaldehyde and incubated with 0.1% crystal violet. Cells were counted under a light microscope.

After different treatments, the T24 cells were collected and incubated with 25 \(\mu\) L transposase reaction solution (2 μL Tn5 transposase and 12.5 μL of TD buffer) at 37 °C for 1 h. Subsequently, the reaction was stopped by adding 25 μL EDTA solution. Next, the DNA fragments were enriched by PCR. The library was amplified using the following protocol: 72 °C for 5 min, 98 °C for 1 min, 15 cycles of 98 °C for 10 s, 63 °C for 30 s, and 72 °C for 1 min. Finally, the fragments were sequenced using a HiSeq2500 (Illumina, USA). Peak calling was analyzed using MACS2 [17, 18].

Raw ATAC-seq fastq data were first cleaned using Cutadapt (3.4) to trim low-quality and adapter sequences. The cleaned sequence data was then aligned to the human hg19 genome reference using bowtie2 (2.4.4) in the ‘very-sensitive’ mode. Samtools (0.1.19) was then used to extract uniquely mapped alignments of high quality. To call ATAC-seq peaks, MACS2 (2.0.9) was used with the ‘nomodel’ option. The ‘shift’ and ‘extsize’ parameter of MACS2 was set as 100 and 200, respectively. The identified peaks were then annotated to the nearest genes by the homer (4.8). Pathway enrichment and motif analysis were performed using the online tools Enrichr and MEME-Suite, respectively.

The TCGA bladder cancer cohort was used for analysis. UALCAN, CbioPortal, and GEPIA were used for pan-cancer analysis and expression analysis of the HNRNPU gene [19, 20]. Promoter region methylation analysis was conducted using the UALCAN software. Survival and co-expression analyses were performed using the GEPIA web interface. GEPIA2021, along with the bioinformatics tools CIBERSORT, EPIC, and quanTIseq, was used to generate immune cell expression profiles of HNRNPU using the TCGA bladder cancer cohort.

The control cells (sgCtrl) and sgHNRNPU-transfected cells were treated with 1 µM cisplatin or DMF for 5 days. Total RNA from each group was isolated using the RNeasy Mini Kit (Qiagen), following the manufacturer’s protocol. Next, the samples were labeled and hybridized using the Affymetrix GeneChip Human Genome U133 Plus 2.0 Array, according to the manufacturer’s protocol. The data were analyzed using GeneSpring 12.6 software as previously described [21].

T24 cells (1 × 10⁷ cells) with or without HNRNPU knockdown were suspended in 200 μL PBS and subcutaneously injected into each flank of 4–6-week-old BALB/c nu/nu female mice. The tumor volume was measured every 4 days. The mice were sacrificed after 28 days. Cisplatin was administered by tail injection at a dose of 2 mg/kg every other day. After 4 weeks, the mice were euthanized, the tumor was isolated, and the weight of the tumor was measured. Animal experiments were approved by the Xuzhou Medical University Animal Care and Use Committee.

CCK-8 assay was performed to evaluate the cytotoxicity of cisplatin in six bladder cancer cell lines. As shown in Fig. 1, upon cisplatin treatment, T24 and RT4 had the highest IC₅₀ values, that were 7.637 and 7.426 µM, respectively. HT1197, SW780, and RT112 shared similar IC₅₀ values, while HT1376 was the most sensitive cell line to cisplatin. The cytotoxic profile suggested that T24 is the bladder cancer cell line which confers most resistant to cisplatin. Therefore, T24 cells were selected for further investigation.

To investigate genes that might be associated with cisplatin resistance in BUC, the GeCKO library containing 123,411 sgRNAs targeting 19,050 unique human genes was established and transfected into T24 cells. After puromycin selection, cisplatin (0.25 µM) was used to treat the cells for 7 or 14 days. Finally, the surviving cells were collected to examine the enriched sgRNAs using high-throughput sequencing. Our genome-wide CRISPR screening determined multiple enriched sgRNAs that might contribute to cisplatin response. Among them, 327 genes were significantly changed under cisplatin treatment, with 46 genes altered on day 7, and 284 genes altered on day 14, as compared to the day 0 group. As shown in (Supplementary Fig. 1A), GO analysis of these genes demonstrated that these genes were particularly enriched in the process of tissue development, organonitrogen compound metabolism, and ion transport. Meanwhile, we found that these genes were also related to cancer stem cells, that are closely associated with drug resistance (Supplementary Fig. 1B).

To validate these results, 21 top decreased genes were selected. Six sgRNAs were designed for each gene, and the effects of these candidate targets were detected using HCS analysis. We first examined the cell count and fold change of cells on days 2–5 versus day 1. We found that treatment with many of the sgRNAs (HNRNPU, CCDC146, and more.) decreased the T24 cell proliferative rate compared to the sgCtrl treatment (Fig. 2A, B). The fold change in sensitivity enhancement (FSE) was increased in response to knockout of these genes. Among them, sgRNA treatments targeting HNRNPU, OCLM, PRKCDBP, and CCDC146 significantly decreased cisplatin resistance in T24 cells (FSE = 1.4) (Fig. 3A). We then tested the efficiency of each target sgRNA for four genes (HNRNPU, OCLM, PRKCDBP, and CCDC146) to improve the cisplatin response. The results showed that HNRNPU-04327 and HNRNPU-04325 transfection triggered a higher level of FSE than the other groups. For the CCDC146 gene, CCDC146-04,297 and CCDC146-04,293 were more effective in inhibiting cisplatin resistance. However, the targets for OCLM and PRKCDBP were similar to those of the control group (Fig. 3B). We also sequenced these sgRNA-targeted regions for FSE-related sgRNAs in T24 cells after the indicated treatments. Five of the six HNRNPU-targeted sgRNAs significantly edited the targeted genome region (Fig. 3C). Notably, sgHNRNPU-04325 had a 70.1% editing efficiency and had a significant impact on FSE (Fig. 3B). We then chose HNRNPU as a potential target to improve cisplatin sensitivity. As shown in Fig. 3D, a significant correlation was observed between HNRNPU protein expression levels and IC₅₀ values of cisplatin. T24 cells, that had high HNRNPU expression level, were less sensitive to cisplatin. In contrast, HT1376, with no endogenous HNRNPU expression, was most sensitive to cisplatin. Therefore, we confirmed that HNRNPU could be a potential target to improve cisplatin sensitivity. Subsequently, the T24 cells were selected to establish an HNRNPU knockdown cell model. In addition, we tested the cytotoxicity of paclitaxel and doxorubicin, two chemotherapeutic drugs that are used for bladder cancer treatment (Supplementary Table 3). Unlike cisplatin, no correlation was observed between the cytotoxic effect and HNRNPU expression level in the six bladder cancer cell lines, suggesting that HNRNPU mainly mediates the response of bladder cancer cells to cisplatin.

In order to validate the consequence of the genome editing effect of sgHNRNPU-04325, we performed Western blotting and showed that the protein level of HNRNPU was dramatically decreased compared to the control group. A similar trend was observed in sgRNA-transfected cells after cisplatin administration (Fig. 3E).

Next, we used HNRNPU-04325 sgRNAs for KO HNRNPU in T24 cells. We found that T24 cells were more sensitive to cisplatin in the sgHNRNPU transfection group than the sgCtrl + cisplatin group, with an overall cell count fold change of 7 versus 0.5 (day 5 compared to day 1). Notably, compared with the sgCtrl group, knockout of HNRNPU significantly decreased T24 cell viability, which was similar to the result of the sgCtrl + cisplatin group (Fig. 4A). HNRNPU inhibition leads to cell cycle arrest and cell apoptosis. As demonstrated in Fig. 4B, compared with the sgCtrl group, sgHNRNPU-treated T24 cells exhibited a dramatic increase in S-phase arrest. When cells were treated with cisplatin, this trend increased further. Consistently, the percentage of G1-phase was decreased. In addition, we observed a definite shift in apoptosis, with an approximately 60% increase in sgHNRNPU + cisplatin-treated cells compared to the other groups (Fig. 4C-D). These results indicate that HNRNPU downregulation increased cisplatin sensitivity.

We subsequently validated the effect of HNRNPU knockout on cisplatin sensitivity in vivo. As shown in Fig. 4E-G, cisplatin induced a slight decrease in tumor volume and tumor weight, while HNRNPU knockout significantly decreased the tumor size compared with cisplatin treatment alone, indicating that sgHNRNPU treatment enhanced the cisplatin sensitivity of the tumor.

We examined the overall expression level of HNRNPU in the TCGA bladder cancer cohort (T = 404) and found that HNRNPU was highly expressed in bladder tumors, compared to normal tissues (Fig. 5A). Interestingly, the expression level of HNRNPU was higher in tumor tissues than in normal tissues in almost all cancer types (Fig. 5B). We also observed a more hypomethylated promoter of the HNRNPU gene in bladder cancer tissues from the TCGA dataset, compared to the normal bladder tissues (Fig. 5C). We next divided TCGA bladder cancer patients into two groups, HNRNPU-high (Top 20%) and HNRNPU-low, and found that both the disease-free and overall survival rates of the HNRNPU-high patients were significantly lower than those in the HNRNPU-low group (Fig. 5D-E). These results indicate that high expression of HNRNPU in bladder cancer patients is negatively correlated with their clinical outcome.

We next explored whether genes highly correlated with HNRNPU could affect the survival of bladder cancer patients. Indeed, HNRNPK, an HNRNP family gene, was most significantly correlated with HNRNPU expression (Fig. 5F). However, different expression levels of HNRNPK do not affect the overall survival of patients with bladder cancer. Notably, AHCTF1, which was recently described as a MYC driver [22], is one of the most significantly correlated genes (Pearson's r = 0.78) (Fig. 5G). The overall survival rate of patients with high AHCTF1 was also significantly lower than that of patients with low AHCTF1 (Fig. 5H). Finally, we analyzed the expression of HNRNPU in the immune cells of bladder tumor tissues. Using the GEPIA2021 deconvolution tool for the TCGA bladder cancer cohort, we found that HNRNPU was significantly decreased in B cells, CD4⁺ T cells, and CD8⁺ T cells (Fig. 5I). Compared to its upregulation in tumor cells, the downregulation of HNRNPU in immune cells may also be associated with cisplatin resistance.

To investigate the mechanism by which HNRNPU knockout sensitizes BUC to cisplatin at the transcriptional level, we performed RNA-seq on both sgCtrl and sgHNRNPU groups, and identified 672 downregulated genes and 842 upregulated genes (Fig. 6A). The downregulated genes were enriched in several cancer-related pathways, such as “TGF-beta regulation of extracellular matrix,” “Oncostatin M,” “Focal adhesion” and “Pathways in cancer” (Fig. 6B). The upregulated genes were enriched in cell cycle-related pathways such as “Meiosis,” “Meiotic recombination,” “Cell cycle” as well as DNA repair pathways such as “double-strand break repair” and “ATM pathway.” These results demonstrate that HNRNPU knockout could rewire the transcriptome of bladder cancer cells, resulting in increased vulnerability to cisplatin treatment. Genes such as BRCA2, BRCA1, and RIF1, that are highly involved in DNA double-strand break repair pathways, were significantly upregulated in the HNRNPU knockout cells (Fig. 6C). Interestingly, AHCTF1 expression was significantly upregulated in the HNRNPU knockout group, indicating a potential interaction between AHCTF1 and HNRNPU (Supplementary Fig. 3, Fig. 5G).

A previous study reported that HNRNPU is an essential gene which regulates the interphase chromosome structure [23]. We assessed the alteration of interphase chromosome structure by ATAC-sequencing. Volcano plots showed the estimated log2(treat)-log2(control) vs. − log10 (false discovery rate, FDR) for each gene in the sgCtrl vs. the sgHNRNPU group. We identified 23 overrepresented genes and 12,833 downregulated genes in HNRNPU-deficient cells (FDR < 0.05) (Fig. 7A). Pathway analysis of the downregulated genes revealed that focal adhesion, the Rap1 signaling pathway, and pathways in cancers might be regulated under these conditions (Fig. 7B). Focal adhesion and pathways in cancers were also enriched in the downregulated genes, suggesting that the expression of these genes is regulated by chromatin openness. In contrast, the increased expression of genes in HNRNPU KO T24 cells was not accompanied by increased chromatin openness. It is speculated that transcription of these genes is likely to be regulated post-transcriptionally, possibly by regulating mRNA stability. To further investigate the transcription factors involved in regulating the downregulated genes, we performed motif analysis on the regions with decreased ATAC signal. Among the top enriched transcription factor motifs, we identified JUN, ETV5, and IRF3, that are well-known to be involved in cancer development. More interestingly, we also observed enrichment of the FOXG1 motif, which plays a crucial role in the oncogenesis of bladder cancer (Fig. 7C) [24, 25].

We subsequently investigated the mechanism by which HNRNPU knockout enhances cisplatin sensitivity. We first selected the downstream gene of HNRNPU, from which the qPCR results showed the same trend as ATAC sequencing and microarray assay. Among the 25 downstream genes of HNRNPU verified by qRT-PCR (Supplementary Fig. 5), we chose nine genes that showed the same trend in qPCR and microarray to perform further analysis. Among the nine genes, only overexpression of Erb-B2 receptor tyrosine kinase 3 (ERBB3) and knockdown of neurofibromin 1 (NF1) exhibited a reverse regulation of HNRNPU knockout under cisplatin treatment, indicating that knockout of HNRNPU regulates the sensitivity of cisplatin treatment through ERBB3 or NF1 (Supplementary Fig. 5).

We chose NF1 as the downstream target for validation, as NF1 has previously been associated with lower urinary tract dysfunction and resistance to EGFR inhibition in lung cancer [26, 27]. Using the NF1 knockdown lentivirus, we found that knockdown of NF1 abolished the effect of knockout of HNRNPU. Knockout of HNRNPU significantly increased the effect of cisplatin on the inhibition of cell proliferation, invasion, and migration, while knockdown of NF1 abrogated the effect of HNRNPU knockout, as indicated by the CCK-8 assay, transwell assay, and wound healing assay (Fig. 8A-C, Supplementary Fig. 6A-B). We further found that HNRNPU knockout enhanced cisplatin-induced cell cycle arrest in the S and G2 phases, which was reversed by NF1 knockdown (Fig. 8D, Supplementary Fig. 6C). Flow cytometry showed that knockout of HNRNPU increased cisplatin-induced apoptosis, while knockdown of NF1 reversed the effect of HNRNPU knockout on apoptosis (Fig. 8E, Supplementary Fig. 6D). These results suggest that knockout of HNRNPU increases cisplatin sensitivity by regulating NF1 related signaling pathway.