Cancer stem cells are considered responsible for many important aspects of tumors such as their self-renewal, tumor-initiating, drug-resistance and metastasis. However, the genetic basis and origination of human bladder cancer stem cells (BCSCs) remains unknown. Here, we conducted single-cell sequencing on 59 cells including BCSCs, bladder cancer non-stem cells (BCNSCs), bladder epithelial stem cells (BESCs) and bladder epithelial non-stem cells (BENSCs) from three bladder cancer (BC) specimens. Specifically, BCSCs demonstrate clonal homogeneity and suggest their origin from BESCs or BCNSCs through phylogenetic analysis. Moreover, 21 key altered genes were identified in BCSCs including six genes not previously described in BC (ETS1, GPRC5A, MKL1, PAWR, PITX2 and RGS9BP). Co-mutations of ARID1A, GPRC5A and MLL2 introduced by CRISPR/Cas9 significantly enhance the capabilities of self-renewal and tumor-initiating of BCNSCs. To our knowledge, our study first provides an overview of the genetic basis of human BCSCs with single-cell sequencing and demonstrates the biclonal origin of human BCSCs via evolution analysis.
Human bladder cancer stem cells show the high level of consistency and may derived from bladder epithelial stem cells or bladder cancer non-stem cells. Mutations of ARID1A, GPRC5A and MLL2 grant bladder cancer non-stem cells the capability of self-renewal.
Keywords: bladder cancer, bladder cancer stem cells, genetic alteration, self-renewal, single-cell sequencing, tumor evolution.
With an estimated 430,000 new cases and 165,100 deaths worldwide annually, bladder cancer (BC) is one of the most common urinary malignancies . BCSCs, with typical biomarkers of 67LR, CD44, cytokeratin 5 (CK5) and CD90/CK14, are highly tumorigenic, drug-resistant and metastatic . Although Van Batavia et al. provided the genetic evidence on the progenitors of urothelial carcinoma of mice , the genetic status and origination of human BCSCs are largely unclear.
To further understand the genetic basis and phylogenetic status of human BCSCs, we isolated BCSCs (CD31-CD45-CD44+), BCNSCs (CD31-CD45-CD44−), BESCs (pan-CK+CD44+) and BENSCs (pan-CK+CD44−) from three BC specimens (Supplementary Fig. 1 and Supplementary Table 1) . Notably, BCSCs and BCNSCs could establish tumors in NOD/SCID mice, but BESCs and BENSCs failed in the xenograft assays (Supplementary Fig. 2A). Moreover, BCSCs and BESCs displayed the capabilities of self-renewal and sphere formation, and upregulated expression levels of stem cell-related genes in contrast to BCNSCs and BENSCs (Supplementary Figs. 2B and 2C). Finally, BCSCs outperformed BCNSCs in initiating bladder carcinoma in the limiting dilutions xenograft and serial transplantation assays (Supplementary Figs. 2D and 2E). Taken together, the collection and sorting methods for former four types of cells from bladder carcinoma and normal bladder tissue are applicable and reliable.
In order to illustrate the cell hierarchy in human BC, we isolated 59 cells from three BC specimens, prepared the library by Multiple Displacement Amplification (MDA) (Supplementary Tables 2 and 3) , and performed exome sequencing with a mean depth of 43.6× in exome regions (Supplementary Figs. 3A and 3B). Thereafter, we collected SNPs from individual cells and filtered out the sites that were coincident in all cells or just presented in one single cell. Using ascertained 28649, 29879 and 36313 polymorphic SNP/SNV sites in all individual cells of each patient, we carried out phylogenetic analysis to assess cells clonality based on a modified neighbor joining (NJ) method  (Online Methods). During this process, BESC was selected as out-group to assess the evolutionary relationships of BESC, BENSC, BCSC and BCNSC, and interestingly, the results indicated BCSCs may have originated from the mutations of BESCs (P1 and P3), or the alterations of BCNSCs (P2) (Fig. 1A).
Single-cell sequencing and phylogenetic analysis of bladder cancer cells.
(A) The neighbor joining phylogenetic tree is constructed by Euclidean distance using filtered polymorphic SNP/SNV sites during BCSCs (orange), BCNSCs (yellow), BESCs (blue) and BENSCs (green). (B) Key genes with somatic mutation. The somatic mutation spectra of single cells is present on the left. The heat map in central shows the distribution of mutations across all 31 single-cells. Unreported key genes are asterisked. (C) Hills representing mutated genes are randomly distributed on the plate, and the heights of hills suggest the relative frequency each somatic mutation in single cells. (D) The key genes selected for functional investigation. The horizontal axis contained genes ranked by alphabet. The vertical axis showed the somatic mutant frequency of genes. The circle size (diameter) indicated the observed number of cells with each gene mutations.
We determined 757, 499 and 166 somatic mutations in P1, P2 and P3, respectively (Online Methods). After filtering those in dbSNP and synonymous mutations, 406 nonsynonymous mutant genes were identified (Supplementary Table 4). In sum, mutations from 200 genes were selected for validation in all single-cell DNA and 97.22% (1188 / 1222) of predicted mutations were confirmed (Online Methods and Supplementary Table 5). To demonstrate the key gene mutations of BCSCs, we screened all 406 genes with nonsynonymous mutations by their function and mutation frequencies (Online Methods), and found 21 key gene mutations mainly distributed in five functional pathways including cell cycle regulation (TP53, STAG2, ATM and CREBBP), transcription regulation (BRF1, PAWR, SIN3A, ERCC2, MKL1, TP53 and ETS1), chromatin remodeling (ARID1A, CREBBP and MLL2), cell differentiation and self-renewal (FAT4 and GPRC5A). Specifically, six genes (ETS1, GPRC5A, MKL1, PAWR, PITX2 and RGS9BP) have not formerly been reported as key altered genes in BC (Fig. 1B). Moreover, BCSCs showed a higher frequency of nonsynonymous mutation than BCNSCs (Fig. 1C), which underscored the critical function of specific mutations on stemness acquiring of BCSCs. However, the spectrum of gene mutations which was dominated by C/G>T/A and A/T>G/A maintained the same in BCNSCs and BCSCs (Supplementary Fig. 3C). We then focused on the altered genes in BCSCs with the function of transcription regulation, chromatin remodeling and self-renewal. Therefore, 19 out of 21 mutated genes were picked (Fig. 1D), from which 15 genes with a mutation rate more than 50% were selected as candidates for further functional investigation (Fig. 1D).
In order to verify the function of BCSCs specific mutations, we introduced the individual mutation into BCNSCs isolated from primary BC samples using CRISPR/Cas9 . Of the 15 mutations, only MLL2 slightly enhanced sphere formation of BCNSCs (Fig. 2A). Previous study indicated BCNSCs may require multiple mutations to dramatically increase the capability to generate spheres . We next co-infected BCNSCs with MLL2 mutation and each of remaining 14 gene mutations. The co-transfection of MLL2 with ARID1A, CREBBP or GPRC5A augmented the sphere-forming abilities of BCNSCs and the combination of MLL2 and ARID1A was the most effective one (Fig. 2A). We thus continued stepwise experiments by adding in a third gene mutation from the rest 13 gene mutations. It turned out that the addition of GPRC5A significantly improved the sphere-forming ability of BCNSCs, comparable to that of isolated CD44+ BCSCs (Fig. 2A). Based on the MLL2+ARID1A+GPRC5A combination, any fourth additional mutation did not remarkably increase the capabilities of sphere formation (Fig. 2A).
ARID1A, GPRC5A and MLL2 mutations promote the self-renewal and tumor-initiating of bladder cancer non-stem cell.
(A) Data points indicate average number of spheres formed by bladder cancer non-stem cells isolated from primary bladder cancer tissues with distinct mutations in serum-free conditions. Each of the 15 mutations in Fig. 1D was tested separately (first column), or in combination with MLL2 mutation (second column) or in combination with MLL2 and ARID1A alterations (third column). Other mutations were also tested in combination with MLL2, ARID1A, and GPRC5A alterations (fourth column). (B) BCNSCs Mut formed spheres efficiently in contrast to BCNSCs WT. Spheres were counted in six separate fields after 14 days and shown as means ± SD. Scale bar = 100 μm. This assay was repeated four times. (C) The expression levels of stemness-related genes were elevated in BCNSCs Mut. The mRNA expression levels of GLI1, STAT3 and CD44 were increased in BCNSCs Mut. (D) BCNSCs Mut formed bigger tumors than that of BCNSCs WT, n = 5. (E) The percentage of tumor-free immunodeficient mice four months after subcutaneous injection of different dilutions of BCNSCs WT or BCNSCs Mut. (n = 6 grafted tumors per dilution). (F) Serial tumor formation assays of BCNSCs WT and BCNSCs Mut. Tumor volumes were measured and calculated at the indicated time points, n = 3. Data are representative of three independent experiments and shown as means ± SD. *P < 0.05.
BCNSCs with ARID1A/GPRC5A/MLL2 mutations (BCNSCs Mut) showed no significant change in the mRNA expression levels of ARID1A, GPRC5A and MLL2 compared with BCNSCs wild type cells (BCNSCs WT) (Supplementary Figs. 4A, 4B and Supplementary Table 6). Notably, BCNSCs Mut displayed the capabilities of self-renewal, spheres formation, and higher expression (300%) of stemness-related genes compared to BCNSCs WT (Figs. 2B and 2C). Moreover, BCNSCs Mut established larger tumors (595%) than that of BCNSCs WT (Fig. 2D). Additionally, the limiting dilution xenograft and serial transplantation assays are the gold standards to assess cancer stem cells (CSCs) potential . BCNSCs Mut showed significant higher capabilities of initiating bladder carcinoma (Fig. 2E) and serial tumor formation (Fig. 2F) than BCNSCs WT. In summary, the alterations of ARID1A, GPRC5A and MLL2 played a crucial role in the stemness acquiring of BCNSCs.
Altogether, the genetic basis and stemness-related mutations of BCSCs are illustrated in this study with single-cell sequencing. Firstly, the observed origination of BCSCs from BESCs (P1 and P3) or BCNSCs (P2) by accumulation of mutations, has provides the genetic evidence to support the hypothesis of CSCs origin . Additionally, we identified 21 key altered genes in BCSCs, as well as six novel mutated genes. Finally, co-mutations of ARIDA1, GPRC5A and MLL2 grant BCNSCs stemness. Our findings overall indicate the novel possibilities for bladder cancer targeted therapy.
Competing financial interests
The authors declare no competing financial interests.
Author contributions: Song Wu and Zhao Yang had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Zusen Fan, Song Wu and Chong Li.
Acquisition of data: Zhao Yang and Hongjie Liu
Analysis and interpretation of data: Hongjie Liu and Xiaolong Zhang
Drafting of the manuscript: Zhao Yang and Hongjie Liu
Critical revision of the manuscript for important intellectual content: Zhiming Cai and Song Wu
Statistical analysis: Zhao Yang and Chong Li
Obtaining funding: Zusen Fan and Song Wu
Administrative, technical, and material support: Liqin Xu, Jian Luo, Yi Huang, Luyun He, Chunxiao Liu
Supervision: Zusen Fan, Zhiming Cai, Song Wu and Chong Li
Financial disclosures: Song Wu certifies that all conflicts of interest, including specific financial interests, relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (e.g., employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are none.
Funding/Support and role of the sponsor: This project was supported by the National Natural Science Foundation of China (Grant No. 81330047, 81301740 and No. 81472413), 973 Program (No.2015CB755500, No.2014CB745200, and Innovation Program of Shenzhen (Grant No. JCYJ20130401114715714; No.CXZZ20140826163906370; No. JCYJ20130401114928183; No. JSGG20130411091246833; No. CXZZ2013051653248144; No.20120612131638712).
Acknowledgments: We would like to thank all of the patients and the urology surgeons Abai Xu and Youcheng Lin (Zhujiang Hospital of Southern Medical University, Guangzhou, China), and Mingzhu Zhong (Department of Urological Surgery, The Affiliated Luohu Hospital of Shenzhen University). We also thank Drs. Junfeng Hao, Guizhi Shi, Shu Meng and Xiang Shi (Laboratory Animal Center, CAS Key Laboratory of RNA Biology, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China) for their technical supports.
-  Cancer Genome Atlas Research N. Comprehensive molecular characterization of urothelial bladder carcinoma. Nature. 2014;507:315-322
-  P.L. Ho, A. Kurtova, K.S. Chan. Normal and neoplastic urothelial stem cells: getting to the root of the problem. Nature reviews Urology. 2012;9:583-594 Crossref
-  J. Van Batavia, T. Yamany, A. Molotkov, H. Dan, M. Mansukhani, E. Batourina, et al. Bladder cancers arise from distinct urothelial sub-populations. Nature cell biology. 2014;16(982–91):1-5
-  K.S. Chan, I. Espinosa, M. Chao, D. Wong, L. Ailles, M. Diehn, et al. Identification, molecular characterization, clinical prognosis, and therapeutic targeting of human bladder tumor-initiating cells. Proceedings of the National Academy of Sciences of the United States of America. 2009;106:14016-14021 Crossref
-  F.B. Dean, S. Hosono, L. Fang, X. Wu, A.F. Faruqi, P. Bray-Ward, et al. Comprehensive human genome amplification using multiple displacement amplification. Proceedings of the National Academy of Sciences of the United States of America. 2002;99:5261-5266 Crossref
-  N. Saitou, M. Nei. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular biology and evolution. 1987;4:406-425
-  Z. Yang, S. Liu, M. Zhu, H. Zhang, J. Wang, Q. Xu, et al. PS341 inhibits hepatocellular and colorectal cancer cells through the FOXO3/CTNNB1 signaling pathway. Scientific reports.. 2016;6:22090
-  M.L. Suva, E. Rheinbay, S.M. Gillespie, A.P. Patel, H. Wakimoto, S.D. Rabkin, et al. Reconstructing and reprogramming the tumor-propagating potential of glioblastoma stem-like cells. Cell. 2014;157:580-594 Crossref
-  P. Hohenstein. The stem and roots of Wilms’ tumours. EMBO Molecular Medicine. 2013;5:4-6 Crossref
-  R.T. Bryan. Cell adhesion and urothelial bladder cancer: the role of cadherin switching and related phenomena. Philosophical transactions of the Royal Society of London Series B, Biological sciences.. 2015;370:20140042
a CAS Key Laboratory of Infection and Immunity, CAS Center for Excellence in Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China
b BGI Education Centre, University of Chinese Academy of Sciences, Shenzhen 518083, China
c College of Life Science, University of Chinese Academy of Sciences College, Beijing 100049, China
d Department of Urological Surgery, The Affiliated Luohu Hospital of Shenzhen University, Shenzhen University, Shenzhen 518000, China
e Department of Urological Surgery, Shenzhen Second People's Hospital, The First Affiliated Hospital of Shenzhen University, Shenzhen 518000, China
f BGI-Shenzhen, Shenzhen 518000, China
g Department of Urology, Zhujiang Hospital of Southern Medical University, Guangzhou 510000, China
These authors contributed equally to this work.
© 2016 Published by Elsevier B.V.