Bladder cancer is the eighth most common type of cancer in the United States, with an estimated 80,000 new cases reported each year [1
]. The stage of disease at the time of diagnosis is critical for determining patient outcome; the survival rate for cancers found at early stages is greater than 85%. However, once the cancer has reached stages three or four and has progressed from the bladder to the abdominal cavity or lymph nodes, the five-year survival rate drops below 50% [2
]. Cancer becomes increasingly challenging to treat as it metastasizes to other tissues. Routine standard of care with first line chemotherapeutics typically result in tumor shrinkage, however, most cases relapse despite initial chemotherapeutic sensitivity and the recurring tumor is usually drug resistant [3
]. For almost all cancer types, the emergence of drug resistance has long been a challenge in achieving a complete cure. This phenomenon is widespread in that almost all cancers can become drug resistant and cancer cells can become resistant to almost all types of therapeutic agents. Chemotherapy resistant cancer cells then progress into more aggressive tumors, capable of both metastasizing to distant sites and withstanding future drug treatments, ultimately contributing to a patient mortality [4
Advanced bladder cancer is commonly treated with nonspecific chemotherapeutics, such as cisplatin and gemcitabine [3
]. Cisplatin belongs to the platinum-based family of chemotherapeutics while gemcitabine is a type of DNA synthesis inhibitor. While having slightly different mechanisms of action, both rely on the rapidly dividing nature of cancer cells. Gemcitabine is an analog of deoxycytidine that once inside a cell it becomes activated via phosphorylation, to be used as a nucleoside during DNA synthesis. Inhibition of DNA synthesis is the most likely mechanism by which gemcitabine causes cell death. Gemcitabine nucleotide incorporation into the elongating DNA strand stalls the DNA polymerase. This action locks the drug onto the DNA rendering proofreading enzymes incapable of removing gemicitabine-DNA adducts. Previous research has shown that high levels of gemcitabine-DNA adducts in bladder cancer patients correlate with drug response [5
]. While gemcitabine is a well-known chemotherapeutic agent, the molecular mechanisms by which cells are rendered chemoresistant post remission have yet to be fully elucidated. Additionally, dormancy-type mechanisms by which cells evade drug treatment can be applicable to other first line nonspecific therapies that do not depend on cancer cell proliferation to render an effect.
Methionine adenosyltransferase 1a (MAT1A
) is an enzyme that helps regulate an important biological molecule, S-adenosylmethionine (SAM) that plays a key role in the methylation cycle [6
]. The importance of S-adenosylmethionine has been a topic of increased interest as it relates to metabolism and dysregulation of cancer cells [7
], however, direct links between this protein and enhanced survival during chemotherapy treatment have yet to be established. Here, we found that upregulation of MAT1A
correlates with increased resistance to gemcitabine treatment. Methyltransferases of various types have been previously implicated in cancer progression, however for the first time, we showed that MAT1A
elevated levels confer a significant survival advantage in bladder cancer cells during drug exposure. Here, we utilized RNA sequencing (RNA-seq) to identify transcripts differentially expressed in drug relapsed patient-derived bladder cancer xenograft (PDX) tumors of bladder cancer when compared to untreated tumors. We determined that transient upregulation of MAT1A
in response to drug exposure allowed cells to adopt a less proliferative state. This work has provided novel insights into the temporal nature of gene regulation in response to nonspecific chemotherapy exposure. Furthermore, MAT1A
may have clinical applications as a biomarker of a drug resistant cell subpopulation.
The cancer field is rapidly moving towards developing new strategies for more effectively prolonging life for those afflicted by highly aggressive cancer types. These include changing the formulation of chemotherapy by prescribing multi-drug regimens instead of one chemotherapeutic at a time. Gemcitabine has been the backbone of neoadjuvant chemotherapy for many cancer subtypes, and it has been undoubtedly an effective strategy for shrinking tumors prior to surgery, contributing to a prolonged life span [13
]. However, in a recent clinical trial, patients diagnosed with early stage pancreatic ductal carcinoma treated with FOLFIRINOX, a mixture of four chemotherapy drugs (fluorouracil [5-FU], leucovorin, irinotecan, and oxaliplatin) had a substantially increased survival [14
], emphasizing the need to continue to develop multi-drug formulations. Despite these improvements most of these patients continue to succumb to drug relapse [15
]. Therefore, understanding the underlying molecular mechanisms of relapse following drug treatment in cancer model systems will provide valuable information to expand the treatment options for patients with re-emerging tumors [16
Several studies have provided supporting evidence that genes involved in methionine metabolic pathways tend to be altered in a variety of cancer types, hinting that this pathway may be involved in various aspects of tumorigenesis [12
]. Our work strengthens this observation and provides new mechanistic evidence that alterations of this pathway directly affect drug tolerance to gemcitabine and contributes to the emergence of drug relapse in bladder cancer patient derived xenograft models. Here we show that elevated levels of MAT1A
significantly alter chemosensitivity over a wide range of gemcitabine doses, and that this phenotype was directly mediated by MAT1A
). Therefore, we may speculate that inhibiting MAT1A
function in cancer cells might be beneficial in enhancing drug responsiveness. However, future studies will need to address whether attempting to repress MAT1A
in vivo can improve cancer outcomes [17
belongs to a family of methyltransferases that contribute to methylation of a variety of molecules including DNA and histones, via synthesis of SAM [18
]. In cancer research, MAT1A
has been mostly described in the context of hepatocellular carcinoma (HCC), where it was previously shown that a switch in gene expression from MAT1A
(M1-M2 switch) promoted cancer invasion and metastasis. One study showed that enhancing the M1-M2 switch promoted the ability of these cancer cells to metastasize, and this phenomenon was correlated with human data where a balance in favor of M2 (M1 < M2) correlated with increased metastasis and high rate of recurrence in HCC patients. Mechanistically this work highlighted that MAT1A
was essential for methylating the promoters of certain genes such as osteopontin (OPN), and by repressing MAT1A
expression such genes were freed from methylation-dependent repression and corroborated with other MAT1A
activated genes to promote metastasis-driving pathways such as extracellular-signal-regulated kinase (ERK) signaling [19
]. Our study however, finds the expression to be tipped in favor of M1 in the bladder, and the downstream outcomes of M1 overexpression may have broader implications including increased survivability in the presence of chemotherapy. Our findings provide novel mechanistic insights into the emergence of drug resistance in bladder cancer. Natively, MAT1A
has a tissue specific expression pattern with robust expression in liver, pancreas, skin, ovaries, and testis tissues and has been previously characterized in the context of hepatocellular carcinoma [20
]. Our findings further highlight that expression of MAT1A
in extrahepatic tissue associates with poor prognosis. Interestingly, MAT1A
is not expressed in the urinary bladder under normal conditions, further suggesting the importance of this work as a novel biomarker for bladder cancer (Figure S1
). Further research will be needed to evaluate utility of MAT1A
as a biomarker across other cancer types in tissues where MAT1A
is not natively expressed, as an indicator of aggressiveness and/or relapse post treatment.
has been previously shown to repress gene expression via DNA methylation of gene promoters, it is possible that the down-regulation of histone genes observed in cells overexpressing MAT1A
may be caused by direct methylation of these promoters. Alternatively, MAT1A
may methylate histones and contribute to a regulatory feedback. Since the discovery of RNA methyltransferases a few years ago, this area is now receiving increased attention as a new regulatory mechanism of controlling gene expression [21
]. This new layer of regulation is termed “epitranscriptomics.” Similar to MAT1A
, RNA methyltransferases, demethylases, and m6A-binding proteins are frequently upregulated in human cancer tissues from a variety of organ origins. While their expression levels have not yet been examined in the context of drug resistance, elevated levels of these genes are correlated with cancer progression and metastasis. While the biochemistry of MAT1A
has been mostly examined in the context of DNA methylation, addition work is required to determine whether this enzyme also contributes to other types of methylation, including RNA and histone protein methylation. Additionally, ultrasensitive quantification of SAM-derived methyl groups will further elucidate differences in DNA, RNA, and histone protein methylation to provide insight into whether SAM-mediated epigenetic modifications through MAT1A
provide cancer cells with a drug resistant phenotype. DNA methylation, histone modifications, and the chromatin structure are profoundly altered in human cancers and future studies will further determine how MAT1A
is involved in some of these processes to contribute to drug resistance, in bladder, and possible other types of cancer. The work presented here has clinical relevance as well as applications in basic research. Evidence of MAT1A
in bladder tumors may prove as a powerful prognostic tool to evaluate surgical margins and indicate likelihood of relapse in histopathology samples. Lastly, association of MAT1A
with drug resistant cell populations may also lead to new small molecule inhibitors capable of eliminating drug resistant cells that ultimately allow tumor relapse.
4. Materials and Methods
4.1. Bladder Cancer Patient Derived Xenografts
Bladder cancer xenografts (BL0293 and BL0440) were previously described with corresponding drug sensitivity data [9
]. Frozen samples from Passage 4 were propagated in NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) immunodeficient mice (The Jackson Laboratory, Bar Harbor, ME, USA) by injecting single cell suspensions (3 × 106
tumor cells administered in 100 µL volume of cell suspension in 1:1 Matrigel (Corning)) subcutaneously (SQ) into the right flank. Tumor burden was assessed by measuring tumor length and width using digital calipers twice per week for up to 10 weeks. Tumor volume was estimated using the formula (length*width2
)/2. Tumors were excised when they reached an estimated volume of 1 cm3
or when other IACUC criteria were met. All animal experiments were approved by the Lawrence Livermore National Laboratory Institutional Animal Care and Use Committee and conform to the Guide for the care and use of laboratory animals. Protocol 168 was approved October 14, 2015 by LLNL IACUC to conduct this work.
4.2. Generation of Drug Relapsed PDX Tumors
Gemcitabine hydrochloride (BioVision, Inc, Milpitas, CA, USA) was dissolved in PBS at a final concentration of 25 mg/mL and administered intraperitoneally (IP) once per week for four consecutive weeks at 150 mg/kg. Concurrently, cisplatin (Spectrum Chemical Mfg Corp, New Brunswick, NJ, USA) was resuspended in PBS at 1 mg/mL and delivered intravenously (IV) at 2 mg/kg per animal. Cisplatin dosing occurred on days one, two, and three and days 15, 16, and 17 of drug treatment. Control animals (n = 3) were treated with an equivalent amount of PBS at the same time as chemotherapy treatment. Tumors were grown in vivo for up to 65 days (n = 5).
4.3. RNA Sequencing
Harvested tumors were minced and manually dissociated using a 40 µm cell strainer. Tumor cells were flash frozen in liquid nitrogen and stored at −80 °C until processed. RNA was isolated from 6 x 106 homogenized cells by using QIAshredder and RNeasy® Mini Qiagen kits (Hilden, Germany). RNA concentrations were quantified using a Qubit® RNA HS Assay Kit (ThermoFisher Scientific, Waltham, MA, USA). Quality was assessed before sequencing using an RNA 6000 Nano kit run on an Agilent 2100 Bioanalyzer (Agilent Technologies Inc., Santa Clara, CA, USA). cDNA libraries were prepared from 100 ng of total isolated RNA per manufacturer specification from BL0293/BL0440 in vivo tumors or in vitro 5637 bladder cancer cells for RNA sequencing using Illumina TruSeq RNA Library Preparation kit version 2.0 (Illumina Inc., San Diego, CA, USA) and run on an Illumina Nextseq 500 using the High Output 75 cycles kit (Illumina Inc., San Diego, CA, USA).
4.4. RNA Sequencing Data Analysis
Read quality from RNA-seq raw data was first assessed using FastQC [22
]. Sequence data was aligned to the human genome (hg19) using TopHat2 and STAR [23
]. Subsequently, gene-wise read counts were calculated using ‘featureCounts’ from Rsubread package. Then, the count data was normalized using TMM normalization method, and differentially expressed genes were identified using ‘limma’ [24
] and ‘voom’ [25
]. Genes that were >1.5 fold up- or down-regulated with an FDR corrected p
-value <0.05 were considered significantly differentially expressed. Pathway ontology analysis was conducted using ToppGene tool using p
-value < 0.05 to denote process significance [26
4.5. Plasmid Transfection Generating 5637MAT1A+ Cells
5637 bladder cancer cell line was obtained from ATCC. MAT1A plasmid (Origene; NM_000429 Human Untagged Clone, Product #SC119881) was isolated and purified using MidiPrep™ kit (Qiagen Inc., Hilden, Germany). Subconfluent 5637 cells were transfected with MAT1A plasmid DNA (2 µg/µL) using a Lipofectamine 3000 kit following manufacturer guidelines (ThermoFisher Scientific, Waltham, MA, USA). Briefly, liposomes containing plasmid DNA were constructed in low serum media and added to cells of interest. Transfection occurred over 48 h, followed by downstream experimental procedure. Control experiments were carried out with an empty expression vector.
4.6. Immunohistochemical Analysis
PDX tumors were prepared for IHC analysis by fixation in 10% neutral buffered formalin for three to four days. Tumors were stored in 70% isopropanol until ready for embedding in Type IV paraffin. Tumors were sectioned using a Leica Biosystem RM2125 microtome to collect 0.2 µm tissue sections. Sections were counterstained with hematoxylin and protein of interest was probed using MAT1A primary antibody (ab174687) and labeled streptavidin biotin (LSAB) secondary antibody.
4.7. Human Patient Tissue Microarray Analysis (TMA)
Human bladder cancer tissue microarray (TMA) sections were obtained from the Biorepository at UC Davis Comprehensive Cancer Center. Patient characteristics are described in Table S3
. TMAs were stained for MAT1A
expression using previously described methods in ‘Immunohistochemical Analysis’ section. Each patient tissue was used in triplicate in the TMA, and were scored individually in the nucleus and the cytoplasm, as 0, 0.25, 0.5, 0.75, 1.0, or 1.5, representing increasing fraction of cells staining for MAT1A
4.8. The Human Protein Atlas: MAT1A
4.9. Western Blotting and Protein Analysis
Proteins were isolated from flash frozen tumor cells or 5637 cells in RIPA lysis buffer (Sigma Aldrich Corporation, St. Louis, MO, USA) followed by centrifugation at 14,000 rcf for 5 min. The supernatants were collected and analyzed using the Jess automated Western blotting system (ProteinSimple, San Jose, CA, USA). Jess reagents (biotinylated molecular weight marker, streptavidin-HRP fluorescent standards, sample buffer, DTT, stacking matrix, separation matrix, running buffer, wash buffer, matrix removal buffer, fluorescent labeled secondary antibodies, antibody diluent, and capillaries) were used according to the manufacturer’s standard protocol. Antibodies were diluted with ProteinSimple antibody diluent as follows: MAT1A primary antibody (undiluted, Abcam, Cambridge, United Kingdom, ab174687), beta-Tubulin (1:100, LI-COR Biosciences, Lincoln, ME, Catalog no. 926-42211). Protein concentration was quantified using Compass for SW 4.0 software. Target protein abundance is normalized to the expression of beta-tubulin.
4.10. In Vitro Toxicity
5637MAT1A+ and 5637 cells were dosed with a range of gemcitabine concentrations from 0 mM to 30 mM for 48 h post plasmid transfection. Cell viability was assessed at 48 h post drug treatment using an ATP-based cell viability assay following manufacturer guidelines (CellTiterGlo 2.0, Promega Corporation, Madison, WI, USA). IC50 values were calculated using Prism 8 (GraphPad Software Inc., La Jolla, CA, USA).
4.11. Quantitative PCR (qPCR)
Total RNA was isolated using a commercially available RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA was reverse transcribed to generate cDNA using a SuperScript IV First Strand Synthesis Kit (ThermoFisher Scientific). Cycling was performed using a Fast Real-Time System on an Applied Biosystems 7900HT instrument. Cycling conditions using SYBR as a DNA dye are as follows: 50 °C for 2 min, 95 °C for 2 min followed by 40 cycles of 95 °C for 3 s then 60 °C for 40 s. Data was normalized to Ct values of GAPDH (control gene) and reported as fold change, calculated using the standard ddCt method. GAPDH primer sequences: Forward, GTCTCCTCTGACTTCAACAGCG; reverse, ACCACCCTGTTGCTGTAGCCAA. MAT1A primer sequences: Forward, TCATGTTCACATCGGAGTCTGT; reverse, CATGCCGGTCTTGCACACT.
4.12. Cell Proliferation
5637 cells transfected with MAT1A plasmid or a control empty vector were seeded at 20% confluency (n = 6) and dyed using the CellTrace™ CFSE Cell Proliferation Kit, to quantify cell proliferation. All cells were collected three days post-transfection and immediately analyzed for fluorescence using a FACSMelody™ flow cytometer (Becton, Dickinson & Co, Franklin Lakes, NJ). Expansion indices as well as proliferation/division indices were determined using cell proliferation modeling generated by FlowJo flow cytometry software (FlowJo, LLC, Ashland, OR, USA). Statistics were performed using a Student’s t-test including n = 6 biological replicates from each 5637 and 5637MAT1A+ cells.
4.13. Statistical Analysis
In vitro analysis: Data is presented as averages +/− standard deviation. Student’s t-test was used to calculate statistical significance in Microsoft Excel. TMA analysis of human tissues: Median of three values from each patient were first calculated and then mean and standard deviation derived in each group. MAT1A expression was compared between patients with and without prior treatment or statin or between those with chemotherapy or BCG using Mann-Whitney non-parametric tests. Analyses were conducted using GraphPad Prism 8.2. p values of less than 0.05 were deemed statistically significant.