GALNT1-Mediated Glycosylation and Activation of Sonic Hedgehog Signaling Maintains the Self-Renewal and Tumor-Initiating Capacity of Bladder Cancer Stem Cells

Published OnlineFirst December 16, 2015; DOI: 10.1158/0008-5472.CAN-15-2309 Cancer Tumor and Stem Cell Biology Research

GALNT1-Mediated Glycosylation and Activation of Sonic Hedgehog Signaling Maintains the Self- Renewal and Tumor-Initiating Capacity of Bladder Cancer Stem Cells Chong Li1, Ying Du1, Zhao Yang1, Luyun He1, Yanying Wang1, Lu Hao1, Mingxia Ding2, Ruping Yan2, Jiansong Wang2, and Zusen Fan1

Abstract

þ þ The existence of bladder cancer stem cells (BCSC) has been was highly expressed in BCMab1 CD44 cells and correlated suggested to underlie bladder tumor initiation and recurrence. with clinicopathologic features of bladder cancers. Mechanis- Sonic Hedgehog (SHH) signaling has been implicated in pro- tically, GALNT1 mediated O-linked glycosylation of SHH moting cancer stem cell (CSC) self-renewal and is activated in to promote its activation, which was essential for the self- bladder cancer, but its impact on BCSC maintenance is unclear. renewal maintenance of BCSCs and bladder tumorigenesis. þ In this study, we generated a mAb (BCMab1) against CD44 Finally, intravesical instillation of GALNT1 siRNA and the SHH human bladder cancer cells that recognizes aberrantly glycosy- inhibitor cyclopamine exerted potent antitumor activity against lated integrin a3b1. The combination of BCMab1 with an anti- bladder tumor growth. Taken together, our ﬁndings identify a þ þ CD44 antibody identiﬁed a BCMab1 CD44 cell subpopula- BCSC subpopulation in human bladder tumors that appears tion as BCSCs with stem cell–like properties. Gene expression to be responsive to the inhibition of GALNT1 and SHH sig- analysis revealed that the hedgehog pathway was activated in naling, and thus highlight a potential strategy for preventing þ þ the BCMab1 CD44 subpopulation and was required for the rapid recurrence typical in patients with bladder cancer. BCSC self-renewal. Furthermore, the glycotransferase GALNT1 Cancer Res; 76(5); 1–11. 2015 AACR.

Introduction differentiate into heterogeneous cell populations, resulting in phenotypic and functional heterogeneity of tumors (8, 9). Qui- Bladder cancer is the second most common urological malig- escent CSCs may resist therapeutic intervention and confer to nancy with over 69,000 new cases and an estimated 150,000 recurrence after initial therapy (7). Chan and colleagues identified deaths worldwide each year (1–3). The big problem of bladder þ þ a subpopulation (Lineage CD44 CK5 CK20 ) from primary cancer is the extremely easy recurrence after treatment and no human bladder cancer as bladder cancer stem cells (BCSC) recognized second-line therapy (4). Thus, for new therapeutic (10), which shared similar expression markers to normal bladder strategies, it is of great importance to delineate the mechanism of basal cells, suggesting the basal-like property of BCSCs. A sub- bladder cancer tumorigenesis. population of primary urothelial cancers, coexpressing the 67- Accumulating evidence from leukemias, germ cell tumors, and kDa laminin receptor (67LR) and the basal cell–specific cytoker- many solid tumors support a cancer stem cell (CSC) model in atin CK17, possesses more carcinogenic ability (11). These subsets which cancers are initiated by a rare subpopulation of self-renew- harbor high tumorigenicity, multipotency, and involvement in ing and differentiating tumor-initiating cells (TIC; refs. 5–7). The bladder tumor progression, suggesting that the CSCs exist in cancer stem cell (CSC) population can self-renew limitlessly and bladder tumors. Bladder cancer arises from the urothelium, a transitional epithelium lining the bladder lumen. This multilayered epithelium is 1CAS Key Laboratory of Infection and Immunity, Institute of Biophys- composed of a luminal layer of fully differentiated umbrella cells ics, Chinese Academy of Sciences, Beijing, China. 2Department of that overlie intermediate cells, and a basal layer of Sonic Hedge- fi Urology, The Second Af liated Hospital of Kunming Medical Univer- hog (SHH)-expressing cells (12). The SHH-expressing basal cells, sity, Kunming, China. capable of regenerating all cell types of the urothelium, have been Note: Supplementary data for this article are available at Cancer Research defined as urothelial stem cells. The same group reported that Online (http://cancerres.aacrjournals.org/). muscle-invasive bladder cancers originate exclusively from the C. Li, Y. Du, Z. Yang contributed equally to this article. SHH-expressing basal stem cells of the basal urothelium in Corresponding Authors: Zusen Fan, Institute of Biophysics, Chinese Academy procarcinogen N-butyl-N-4-hydroxybutyle nitrosamine (BBN)- of Sciences, 15 Datun Road, Room 1611, Beijing 100101, China. Phone: 8610-6488- induced mouse bladder cancer (13). These SHH-expressing basal 8457; Fax: 8610-6487-1293; E-mail: [email protected]; and Jiansong Wang, stem cells act as BCSCs to initiate bladder oncogenesis. Aberrant E-mail: [email protected] activation of the Hedgehog signaling is implicated in many doi: 10.1158/0008-5472.CAN-15-2309 malignancies (14, 15). Emerging data support that Hh signaling 2015 American Association for Cancer Research. plays a pivotal role in the self-renewal and differentiation of CSCs

www.aacrjournals.org OF1

Li et al.

(16, 17). Activation of Hh pathway causes CSC self-renewal and In vitro glycosylation assay expansion, while the SMO antagonist cyclopamine or the ligand- Recombinant human wild-type (WT) SHH and T282A-SHH neutralizing antibody induces terminal differentiation and loss of mutant proteins were expressed in E. coli. WT SHH or T282A-SHH clonogenic growth capacity (16, 18). In this study, we deﬁned that mutant was incubated with GALNT1 (R&D Systems) plus Hh signaling is essential for the self-renewal of BCSCs. GALNT1- BCMab1 CD44 cell lysates at 37C for 4 hours. Reaction pro- mediated glycosylation is required for SHH activation in BCSCs. ducts were analyzed by immunoblotting.

Materials and Methods Immunohistochemical staining fi Reagents and mice Tumor specimens were xed in 10% buffered formalin and fi Mouse mAb BCMab1 was purified through protein A-Sephar- embedded in paraf n as described (20). The immunosignals were ose from ascites. Corresponding species-specific FITC-conjugated detected using the ABC kit. secondary antibodies were purchased from Sigma. The commer- cial primary antibodies were phycoerythrin (PE)-conjugated anti- Intravesical administration of siRNA/liposomes and CD44 (BD Pharmingen; 550989), goat anti-human C-terminus of cyclopamine for treatment SHH (Santa Cruz Biotechnology, sc-33942), goat anti-human N- Female 6-week-old NOD/SCID mice with a body weight of terminus of SHH (Santa Cruz Biotechnology; sc-1194), rabbit approximately 15 g were used and kept under specific pathogen- þ þ anti-Gli1 (Santa Cruz Biotechnology, sc-20687), goat anti- free conditions. BCMab1 CD44 cells were labeled with lucifer- GALNT1 (Santa Cruz Biotechnology, sc-68491), and mouse ase and transplanted into the murine bladder cavity via 24-gauge anti-b-actin antibody (Sigma, A1978). NOD/SCID mice were angiocatheters (21). The mice were grouped (24 mice per group) obtained from Vital River Laboratory Animal Technology Co. and intravesically applied with 6 mmol/L siCtrl/liposomes or / Ltd.. GALNT1 mice were obtained from Jackson Laboratories. siGALNT1/liposomes with or without 100 mmol/L cyclopamine BBN, benzyl-N-acetyl-D-galactosaminide (BG), and cyclopamine next day. were from Sigma. GALNT1 siRNA (human) and GALNT1 siRNA (mouse) were from Santa Cruz Biotechnology. Statistical analysis The relationship between the expression levels of GALNT1 and Xenograft establishment clinicopathologic features was analyzed using the c2 or the Human bladder cancer specimens were obtained with patient's Kruskal–Wallis test. Kaplan–Meier analysis was used to estimate consent, as approved by the Research Ethics Board at The Second the cumulative cause-specific survival rates, and the log-rank test Affiliated Hospital of Kunming Medical University (Kunming, was used to correlate differences in patient survival with staining China). For the generation of xenografts, cells were injected with intensity of GALNT1. The influence of GALNT1 on the growth of Matrigel subcutaneously into the backs of NOD/SCID mice. bladder cancers was analyzed by the Student t test. Animal work was carried out in compliance with the ethical regulations approved by the Animal Care Committee, Institute Results þ þ of Biophysics, Chinese Academy of Sciences (Beijing, China). A BCMab1 CD44 subpopulation of cancerous cells acts as BCSCs þ Flow cytometry sorting To identify unique biomarkers of BCSCs, we isolated CD44 Primary tumor tissues were digested with 200U type I colla- BCSCs from human bladder cancer resection specimens (Fig. 1A), genase, 20U type IV collagenase (Sigma, C-0130, C-5138), and which were used as immunogen to immunize mice to generate DNase at 37 C for 2 to 6 hours. Then, the cancer cells were stained mAbs. About 458 hybridomas were screened to bind BCSCs. with FITC-conjugated BCMab1 and PE-conjugated anti-CD44 Among these hybridomas, we identified several antibodies, – antibody or the same isotype FITC/PE conjugated antibody for including BCMab1 and BCMab37, which were determined to fl 30 minutes on ice. Labeled cells were analyzed and sorted by ow recognize the BCSC surface antigens, aberrantly glycosylated cytometry (BD FACSAria III). integrin a3b1 and CD47, respectively. We previously showed that the BCMab1 antibody was also generated by T24 cell immu- DNA methylation analysis nization, recognizing the aberrantly glycosylated integrin a3b1 Methylation was assayed by using the procedure as described on bladder cancer cells (20). The aberrantly glycosylated integrin previously (19). Genomic DNA was treated with sodium bisulfate a3b1 is highly expressed in bladder cancer tissues. CD47 was also using the EZ DNA methylation direct kit (Zymo Research). PCR confirmed to be a surface marker of BCSCs (11). products were analyzed with the MassArray EpiTYPER system By using BCMab1 and anti-CD44 antibodies, we distinguished þ þ (Invitrogen). a BCMab1 CD44 subset from the tumor cells of human bladder tumor specimens (Fig. 1A), comprising 3% to 5% of the total cell þ þ Quantitative real-time PCR population. BCMab1 CD44 cells were further confirmed by Total RNA was isolated from primary sorted cells, human immunofluorescence staining (Fig. 1C). We next determined þ þ bladder cancer cell lines, and mouse tissues using the RNA whether BCMab1 CD44 cells were more tumorigenic than their þ þ Isolation Kit (Tiangen Biotech). RNA was then subjected to cDNA BCMab1 CD44 counterparts. BCMab1 CD44 cells freshly synthesis using M-MLV Reverse Transcriptase (Promega). The isolated from bladder cancer specimens were inoculated subcu- þ þ cDNA was then processed for an amplification step with the SYBR taneously into NOD/SCID mice. As few as 10 BCMab1 CD44 Green reaction system (Tiangen Biotech) running on an ABI 7300 cells could form xenograft tumors in vivo (Fig. 1D and E), whereas (Applied Biosystems). Primer oligonucleotides for qPCR are listed at least 1 105 BCMab1 CD44 cells were necessary to establish þ þ in the Supplementary Table S1. xenografts. The xenograft tumors formed with BCMab1 CD44

OF2 Cancer Res; 76(5) March 1, 2016 Cancer Research

GALNT1-Mediated SHH Activation in BCSCs

Figure 1. þ þ A BCMab1 CD44 subpopulation of cancerous cells is defined as BCSCs. A, representative cystoscope of a typical bladder cancer. , bladder tumor. B, a subset of BCMab1þCD44þ cells was fractionated in one human bladder cancer specimen. Single tumor cells were stained with BCMab1 (B) and anti-CD44 (C) antibodies followed by flow cytometry. C, immunofluorescence staining of BCMab1þCD44þ (left) and BCMab1CD44 (right) cells. D, tumor formation frequency of BCSCs from unsorted, BCMab1þCD44þ, and BCMab1CD44 subpopulations derived from bladder cancer patient specimens. CI, confidence interval. Six bladder cancer samples got similar results. E, representative photographs demonstrating xenografted tumors (arrowheads). Bladder tumor cell subpopulations þ þ (BCMab1 CD44 or BCMab1 CD44 ) with limiting dilutions were injected subcutaneously into NOD/SCID mice with the indicated numbers of cells. þ þ F, FACS analysis of BCMab1 and CD44 expression in xenograft tumors generated from 1 105 BCMab1 CD44 or BCMab1 CD44 cells. G, injection of 10, 100, 1,000, 10,000, and 100,000 freshly isolated BCMab1þCD44þ (left) or BCMab1CD44 cells (right) derived from the primary graft tumors. Data represent means SD; n ¼ 6. H, serial tumor formation assay. Tumor volumes were measured at the indicated time points and calculated as means SD in BCMab1þCD44þ (left) or BCMab1CD44 (right) cell transplantations; n ¼ 6. I, oncospheres were counted in five separate fields after two weeks and are shown as means SD.

www.aacrjournals.org Cancer Res; 76(5) March 1, 2016 OF3

Li et al.

Figure 2. þ þ Hh signaling is essential for the self-renewal of BCSCs. A, scatter plot comparing global gene expression profiles between BCMab1 CD44 and BCMab1 CD44 cells. Black lines, 2-fold differences in either direction in gene expression levels. Arrowheads, Hh signaling–related genes. B, real-time PCR analysis of Gli1 mRNA levels in BCMab1þCD44þ cells derived from 72 bladder cancer specimens. , P < 0.01. C, immunohistochemical staining of human bladder cancer tissues and pericarcinoma tissues with ant-SHH and anti-Gli1 antibodies. D, immunofluorescence staining in BCMab1þCD44þ or BCMab1CD44 cells by BCMab1-FITC, anti-CD44-PE and anti-Gli1-PE–labeled antibodies. White arrowheads, nuclear localization of Gli1. E, Western blot analysis of SHH and Gli1 protein þ þ levels in BCMab1 CD44 cells pooled from bladder cancer samples. F and G, cyclopamine significantly inhibits both oncosphere formation and anchorage- þ þ þ þ independent tumor growth of BCMab1 CD44 cells. BCMab1 CD44 cells were treated with cyclopamine for 24 hours prior to oncosphere formation (F) or soft agar colony (G) assays. All data are representative of four independent experiments and are shown as means SD. , P < 0.01. CPM, cyclopamine.

cells and BCMab1 CD44 cells in Fig. 1E were used for further cells freshly isolated from human bladder cancer specimens þ þ analysis. Intriguingly, the tumors induced by BCMab1 CD44 exhibited significantly higher oncosphere-forming efficiency þ þ cells were able to reconstitute a mixture of BCMab1 CD44 cells compared with their respective BCMab1 CD44 counterparts þ þ and differentiated BCMab1 CD44 tumor cells (Fig. 1F), but the (Fig. 1I). Taken together, the BCMab1 CD44 subpopulation is tumors grafted by BCMab1 CD44 cells could not reconstitute highly tumorigenic, with self-renewal and differentiation proper- þ þ þ þ the BCMab1 CD44 cell subpopulation. ties. Thus, we defined this BCMab1 CD44 subset as the BCSC. þ þ In addition, the self-renewal capacity of BCMab1 CD44 cells þ þ was evaluated by anin vivo serial passage assay. BCMab1 CD44 or Hedgehog signaling is essential for the self-renewal of BCSCs BCMab1 CD44 cells were sorted from the primary mouse SHH-expressing basal stem cells, acting as BCSCs, initiate þ þ xenografts induced by the BCMab1 CD44 cells and implanted tumorigenesis in BBN-induced mouse bladder cancer (15). into secondary recipient NOD/SCID mice accordingly. The same These results suggest that Hh signaling may be implicated procedure was used for the third and fourth passages. All the serially in the stemness of BCSCs. To determine the Hh pathway of þ þ þ þ xenografted mice injected with BCMab1 CD44 cells were able BCMab1 CD44 cells, we performed transcriptome microar- þ þ to form tumors (Fig. 1G and H), which grew much faster with serial ray analysis of BCMab1 CD44 and BCMab1 CD44 cells passages. However, xenografted tumors injected with BCMab1 isolated from human bladder tumor specimens. Microarray CD44 cells only formed with high cell number inoculation, while analysis revealed that several Hh signaling molecules, includ- the formed xenografts were smaller even with multiple passages. ing SHH, Smo, Gli1, and Gli2, were highly expressed in þ þ þ þ We next examined the self-renewal property of BCMab1 CD44 BCMab1 CD44 cells compared with their BCMab1 CD44 þ þ þ cells by an in vitro sphere formation assay. BCMab1 CD44 counterparts (Fig. 2A). High Gli1 mRNA levels in BCMab1

OF4 Cancer Res; 76(5) March 1, 2016 Cancer Research

GALNT1-Mediated SHH Activation in BCSCs

þ þ þ þ CD44 cells were verified in BCMab1 CD44 cells derived CD44 cells (data not shown). These results were verified in þ þ from each specimen of 72 human bladder cancer samples BCMab1 CD44 cells isolated from 72 bladder cancer specimens (Fig. 2B). Bladder cancer tissues displayed high expression of by using qPCR (Fig. 3B and C). Moreover, high expression of þ þ SHH and Gli1 by immunohistochemical staining (Fig. 2C). In GALNT1 in pooled BCMab1 CD44 cells from ten bladder þ addition, Gli1 was highly expressed in the nuclei of BCMab1 cancer specimens was further validated by immunoblotting (Fig. þ CD44 cells, but undetectable in BCMab1 CD44 cells (Fig. 3D). High expression of GALNT1 in bladder cancer tissues was 2D). More importantly, SHH was highly expressed in confirmed by immunohistochemical staining (Fig. 3E). In addi- þ þ BCMab1 CD44 cells and apparently hydrolyzed to form its tion, GALNT1-positive staining in bladder cancer tissues was catalyzed band (Fig. 2E), but undetectable in BCMab1 CD44 related to BCMab1 and CD44 staining. þ cells. These data suggest that the Hh signaling pathway is To further verify the high expression of GALNT1 in BCMab1 þ activated in BCSCs. CD44 cells, we examined the 50 CpG sites at the GALNT1 To further confirm the involvement of Hh signaling in BCSCs, promoter of genomic DNAs by bisulfite genomic sequencing þ we used the SMO antagonist cyclopamine to treat BCMab1 analysis. Each group displayed ten independent sequencing þ CD44 cells for an in vitro sphere formation assay. We observed clones. The GALNT1 promoter region was over 90% unmethy- þ þ that cyclopamine treatment dramatically impaired oncosphere lated in BCMab1 CD44 cells, whereas the same cytosine resi- formation (Fig. 2F). In addition, using a soft agar colony forma- dues were about 60% methylated in BCMab1 CD44 cells (Fig. tion assay, cyclopamine treatment also significantly repressed 3F). Similar results were obtained with other three bladder cancer anchorage-independent tumor growth (Fig. 2G). Overall, the Hh specimens. In sum, these results suggest that the GALNT1 is þ þ signaling pathway is essential for the maintenance of self-renewal indeed actively transcribed in BCMab1 CD44 cells. of BCSCs. GALNT1 is required for the maintenance of stemness for BCSCs GALNT1 is highly expressed in BCSCs We next wanted to determine whether GALNT1 takes part in þ þ þ We performed transcriptome microarray of BCMab1 CD44 the regulation of BCSCs. We silenced GALNT1 in BCMab1 þ cells versus BCMab1 CD44 cells derived from bladder cancer CD44 cells derived from one human bladder cancer specimen samples. Microarray analysis displayed that the glycotransferase and established GALNT1-depleted cell lines (Fig. 4A). We þ þ GALNT1 mRNA level in BCMab1 CD44 was over 7-fold higher found that GALNT1 knockdown significantly declined onco- than that in BCMab1 CD44 cells (Fig. 3A). However, other sphere formation and anchorage-independent colony forma- glycotransferases (GALNT4, 7, 8, 9, and 12) that the microarray tion (Fig. 4B and C). BG is a competitive substrate inhibitor for þ þ detected had no significant changes in BCMab1 CD44 cells all glycotransferases (22). BG treatment obtained similar results compared with BCMab1 CD44 cells. As the GALNT family has to GALNT1 depletion (Fig. 4B and C). We detected mRNA levels þ þ 12 members, we then examined other undetected glycotransferase of Gli1 and GALNT1 in BCMab1 CD44 cells derived from 72 members in the microarray, including GALNT2, 3, 5, 6, 10, and bladder cancer samples by qPCR. Interestingly, Gli1 displayed a þ þ 11, in the BCMab1 CD44 cells. We noticed that these unde- good correlation with GALNT1 expression (Pearson correlation þ tected glycotransferases were not expressed in the BCMab1 coefficient, r ¼ 0.945; Fig. 4D).

Figure 3. GALNT1 is highly expressed in BCSCs. A, heatmap represents the relative expression of GALNT1, integrin a3b1, þ þ CD44, and Gli1 in BCMab1 CD44 and BCMab1CD44 cells. B, real-time PCR analysis of mRNA levels of GALNTs in BCMab1þCD44þ and BCMab1CD44 cells derived from patient samples. , P < 0.01; n ¼ 6. C and D, real-time PCR (C) and Western blot analysis (D) of GALNT1 expression levels in BCMab1þCD44þ and BCMab1CD44 cells derived from 72 bladder cancer specimens. E, immunohistochemical staining of human bladder cancer tissues with BCMab1, anti-CD44, and anti-GALNT1 antibodies. H&E, hematoxylin and eosin. F, bisulﬁte genomic sequencing analysis of the 50 CpG sites at the GALNT1 promoter. Each group includes ten independent clones. Black dots, methylated CpG; gray dots, unmethylated CpG.

www.aacrjournals.org Cancer Res; 76(5) March 1, 2016 OF5

Li et al.

Figure 4. þ þ GALNT1 is required for maintenance of stemness of BCSCs. A, GALNT1 was silenced in BCMab1 CD44 cells derived from one human bladder cancer specimen. A scrambled sequence (shCtrl) was used as a control. B and C, GALNT1 silencing or BG treatment declines oncosphere formation (B) and anchorage-independent tumor growth (C) of BCMab1þCD44þ cells. Oncospheres or colonies were counted in five separate fields after two weeks. , P < 0.01. D, mRNA levels of GALNT1 and Gli1 were detected by RT-PCR from 72 patient samples. P ¼ 0.001; r ¼ 0.945, Pearson correlation. E, real-time PCR analysis of Gli1 mRNA levels in GALNT1-silenced or shCtrl BCMab1þCD44þ cells. BG-treated BCMab1þCD44þ cells were also analyzed. Data were normalized to b-actin and are shown as means þ þ SD. n ¼ 12; , P < 0.01. F, immunofluorescence staining of GALNT1 and Gli1 in shCtrl or GALNT1-silenced BCMab1 CD44 cells. GALNT1 was labeled with FITC, þ þ and Gli1 was labeled with PE. G, ELISA analysis of SHH-N levels shCtrl or GALNT1-depleted BCMab1 CD44 cells. BG-treated cells were also analyzed; , P < 0.01. H, tumor volumes were measured at the indicated time points and calculated as means SD (left). Representative tumors at week 8 are shown on the right; , P < 0.01. n ¼ 12. I, xenograft BCSCs sorted by FACS and infected with retrovirus expressing either scramble vector (shCtrl) or shGALNT1, followed by transplantation into mice 24 hours after infection (n ¼ 10 for each group).

Importantly, we observed that GALNT1 knockdown or BG GALNT1 depletion impaired SHH processing. Consequently, þ þ þ treatment dramatically reduced Gli1 expression in BCMab1 GALNT1-silenced BCMab1 CD44 cells formed smaller tumor þ CD44 cells (Fig. 4E). These results were verified by immunoflu- burden and displayed much lower xenograft tumor formation orescence staining (Fig. 4F). We next overexpressed Gli1 and Gli2 rates compared with shCtrl-treated cells (Fig. 4H and I). in BCMab1 CD44 cells followed by oncosphere formation We next rescued SHH and/or GALNT1 in GALNT1-silenced assays. We observed that Gli1 but not Gli2 overexpression could BCMab1 CD44 cells. We found that rescue of both SHH and promote oncosphere formation (Supplementary Fig. S1A), sug- GALNT1 was able to process SHH and induce Gli1 expression gesting that Gli1 was activated in BCSCs. In addition, GALNT1 (Supplementary Fig. S1B). In contrast, rescue of SHH in GALNT1- knockdown or BG treatment significantly declined active SHH-N silenced BCMab1 CD44 cells neither hydrolyzed SHH nor þ þ protein levels in BCMab1 CD44 cells (Fig. 4G), suggesting that triggered Gli1 expression (Supplementary Fig. S1B). These data

OF6 Cancer Res; 76(5) March 1, 2016 Cancer Research

GALNT1-Mediated SHH Activation in BCSCs

Figure 5. GALNT1-mediated glycosylation of SHH is needed for SHH activation in BCSCs. A, GALNT1-silenced or BG- treated BCMab1þCD44þ cell lysates were probed with anti-C-terminal SHH, anti-GALNT1, and anti-Gli1 antibodies. IP, immunoprecipitation; IB, immunoblotting. B, immunofluorescence staining with C- or N-terminal SHH antibody in GALNT1-silenced or BG-treated þ þ BCMab1 CD44 cells. The C-terminal of SHH antibody was labeled with FITC, and the N-terminal of SHH antibody was labeled with PE. C, SHH and its three mutations (T282A, S288A, S289A) were constructed into vectors and cotransfected with GALNT1 into BCMab1 CD44 cells for 48 hours, respectively. D, WT SHH or T282A-SHH mutant was incubated with GALNT1 plus BCMab1CD44 cell lysates. Reaction products were analyzed by immunoblotting. E, biotinylated peptides were incubated with GALNT1 and UDP-GalNAc followed by detection of absorbance (450 nmol/L). F and G, cotransfection of SHH and GALNT1 can rescue formation of BCMab1þCD44þ cells with self-renewing ability. BCMab1 CD44 cells were cotransfected with the indicated vectors for 48 hours followed by FACS analysis (F) and a sphere formation assay (G). Oncospheres were counted in five separate fields after two weeks and are shown as means SD; , P < 0.01. H, tumors with the indicated treatment were subcutaneously injected into the backs of NOD/SCID mice (n ¼ 12).

þ þ validate that GALNT1-mediated SHH processing can induce Gli1 BCMab1 CD44 cells, suggesting that SHH might not be glyco- expression. Finally, we examined expression levels of SHH and its sylated after depletion of GALNT1. BG treatment showed similar receptor PTCH1 in BCSCs from bladder cancer samples. We results (Fig. 5A). Peanut agglutinin (PNA) lectin can specifically þ þ noticed that PTCH1 was expressed in the BCMab1 CD44 cells bind to the terminal galactose residues of O-glycans (23). To (Supplementary Fig. S1C), and SHH was processed in these examine whether SHH undergoes O-linked glycosylation, we BCSCs. These results suggest that SHH processing in the BCSCs used PNA to probe SHH immunoprecipitates derived from þ þ is directly related to its own Hh signaling in the same type of cell. GALNT1-silenced or shCtrl-treated BCMab1 CD44 cells. Inter- Thus, we conclude that SHH is an autocrine factor in the estingly, we found that PNA could bind to SHH and its hydrolyzed þ þ BCMab1 CD44 cells. Altogether, these data suggest that SHH-C fragment (Fig. 5A), suggesting the SHH underwent O- þ þ GALNT1 contributes to the activation of Hh signaling and tumor- linked glycosylation in BCMab1 CD44 cells. To further confirm igenesis of bladder cancer. O-linked glycans, we used peptide N-glycosidase F and O-glyco- þ þ sidase to treat BCMab1 CD44 cell lysates, respectively. PNA still GALNT1-mediated glycosylation is required for SHH activation bound to the SHH from N-glycosidase–treated cell extracts, but in BCSCs not the SHH from O-glycosidase–treated cell extracts (data not To further test whether GALNT1 influences SHH activation, we shown). These results verify that the O-linked glycosylation takes detected the SHH processing status in GALNT1-silenced or BG- place in SHH of BCSCs. We observed that immunofluorescence þ þ þ þ treated BCMab1 CD44 cells. Interestingly, we found that staining appeared in WT or shCtrl-treated BCMab1 CD44 cells, þ þ GALNT1 knockdown significantly impaired SHH hydrolysis in but not in GALNT1-silenced or BG-treated BCMab1 CD44 cells þ þ BCMab1 CD44 cells derived from bladder cancer specimens (Fig. 5B). These data indicate that GALNT1 participates in the þ þ (Fig. 5A), whereas shCtrl-treated BCMab1 CD44 cells exhibited processing activation of SHH. apparent SHH processing. Moreover, the full-length SHH dis- We next predicted O-linked glycosylation sites of SHH by using played a smaller size on the gel blot in GALNT1-depleted the NetOGlyc 3.1 Server tool. Three potential candidate O-linked

www.aacrjournals.org Cancer Res; 76(5) March 1, 2016 OF7

Li et al.

glycosylation sites of SHH protein, including Thr282, Ser288, and mens. Notably, GALNT1 and SHH cotransfection could induce þ þ Ser289, were selected (Supplementary Fig. S2). To identify the formation of a subpopulation of BCMab1 CD44 cells (Fig. 5F). exact glycosylation sites, we generated three mutation construct However, GALNT1, SHH or T282A-SHH alone, or GALNT1 plus vectors of SHH (T282A-SHH, S288A-SHH, and S289A-SHH). We T282A-SHH–mutant transfection failed to generate this subset. In cotransfected GALNT1 and SHH (WT) or its three respective addition, GALNT1 and SHH cotransfection were able to generate mutants into BCMab1 CD44 cells. We noticed that only oncospheres efficiently (Fig. 5G). GALNT1 and SHH cotransfec- T282A-SHH mutation abolished SHH hydrolysis, but WT SHH tion formed xenograft tumors rapidly after inoculation (Fig. 5H). and other two mutants (S288A-SHH and S289A-SHH) were still In contrast, GALNT1, SHH or T282A-SHH alone, or GALNT1 plus processing to generate its hydrolyzed forms (Fig. 5C). Further- T282A-SHH–mutant transfection had no such capacity. Taken more, T282A-SHH mutant showed a smaller size of SHH on the together, GALNT1-mediated activation of SHH is indispensable blot, suggesting T282A-SHH mutation could not undergo glyco- for the maintenance of self-renewal of BCSCs. sylation. In addition, T282A-SHH mutant could not bind to PNA (Fig. 5C), whereas WT SHH and other two mutants (S288A-SHH GALNT1 deficiency inhibits bladder tumor growth and S289A-SHH) were able to bind to PNA, indicating that the O- We induced bladder cancer in GALNT1-deficient mice or lit- linked glycosylation did not take place in T282A-SHH mutation. termate control mice (WT) by feeding BBN. At 4 months of BBN Overall, Thr282 was defined as the glycosylation site for the O- exposure, 23 WT out of 24 mice displayed robust tumors in linked glycosylation of SHH. bladders, with bigger tumor sizes (Fig. 6A and B). In contrast, To further verify that the SHH hydrolysis is mediated by / only 5 GALNT1-deficient mice of 24 GALNT1 mice showed GALNT1 glycosylation, we generated recombinant GALNT1, WT bladder tumors with smaller sizes. Tumors in WT or GALNT1- SHH, and T282A-SHH mutant proteins for an in vitro glycosyla- deficient mice were confirmed by histology staining (Fig. 6C). We tion assay. As expected, the T282A-SHH mutant incubated with / prepared single tumor cells from WT or GALNT1 mice for an in GALNT1 was not processed, and displayed a smaller size (Fig. / vitro sphere formation assay. We observed that GALNT1 tumor 5D). In contrast, WT SHH incubated with GALNT1 showed cells formed much fewer oncospheres than WT tumor cells (Fig. apparent hydrolysis and displayed a bigger size. Moreover, WT 6D). As expected, mRNA and protein levels of Gli1 were unde- SHH incubated with GALNT1 could bind to PNA, but not for the / tectable in GALNT1 tumors (Fig. 6E and F), whereas Gli1 was T282A-SHH mutant incubated with GALNT1 (data not shown). þ þ highly expressed in WT tumors. In sum, GALNT1 deletion in mice SHH processing in BCMab1 CD44 cells served as a positive suppresses BBN-induced bladder tumors. control. To further validate the glycosylation of SHH and its three mutants, we constructed biotinylated peptides for SHH and its mutants to analyze GALNT1-mediated SHH glycosylation (Sup- GALNT1 silencing or blockade of Hh signaling displays plementary Fig. S3). We found that only the T282A-SHH–mutant antitumor efficacy of bladder cancer in vivo peptide could not be glycosylated by GALNT1 (Fig. 5E), while WT The implanted mice were grouped (24 mice/group) and intra- SHH and the other two mutants (S288A-SHH and S289A-SHH) vesically administrated GALNT1 siRNA/liposomes, or GALNT1 indeed exhibited glycosylation. We conclude that the GALNT1- siRNA/liposomes plus cyclopamine the next day. Intriguingly, mediated O-linked glycosylation of SHH is necessary for SHH both GALNT1 siRNA and cyclopamine could significantly sup- processing. press tumor growth (Fig. 7A and B). Importantly, GALNT1 siRNA We next examined rescue function of GALNT1 and SHH in plus cyclopamine were able to abrogate tumor formation. To cotransfected BCMab1 CD44 cells from bladder cancer speci- detect the delivery efficiency of siRNA with liposome, we analyzed

Figure 6. GALNT1 deﬁciency inhibits BBN-induced bladder tumor growth. A and B, GALNT1 / mice show a dramatic decrease in tumor formation rates (A) and sizes (B) of BBN-induced bladder cancer. Tumor formation rates of BBN-induced bladder cancer in mice were calculated after four months with BBN exposure. Then, the mice were sacriﬁced for measurement of tumor volumes. (n ¼ 24); , P < 0.01. C, hematoxylin and eosin counterstaining for GALNT1þ/þ and GALNT1/ mice treated with BBN. , bladder tumor. D, tumors induced from GALNT1/ mice suppress oncosphere formation. Sphere formation assays were performed for tumor cells induced from GALNT1þ/þ and GALNT1/ mice. E and F, GALNT1 and Gli1 expression levels in bladder tumors from GALNT1þ/þ or GALNT1/ mice analyzed by PCR (E) and immunoblotting (F).

OF8 Cancer Res; 76(5) March 1, 2016 Cancer Research

GALNT1-Mediated SHH Activation in BCSCs

Figure 7. Intravesical administration of GALNT1 siRNA and cyclopamine eradicates bladder tumor in vivo. A, intravesical instillation of GALNT1 siRNA or cyclopamine signiﬁcantly suppresses bladder tumor growth. Bioluminescence images were obtained by IVIS at 8 weeks after þ þ BCMab1 CD44 cells transplantation with intravesical instillation of the indicated agents. B, photon counts of each mouse are indicated by pseudo- color scales. Growth curves of orthotopical tumors were measured by IVIS (n ¼ 24 per group); , P < 0.01. C, GALNT1 expression levels were analyzed by immunoblotting. D, comparison of survival rates with the indicated treatment. n ¼ 24 per group. E, representative bioluminescence images showing bladder cancer metastasis. Mice were implanted with the indicated human bladder cancer cells via tail vein injection for 5 weeks. F, imaging by stereoscope (middle) and hematoxylin and eosin (H&E) counterstaining (bottom) at 12 weeks after intravesical administration of GALNT1 siRNA with BBN exposure (top). G, tumor formation rates of GALNT1 siRNA or siCtrl-treated mice with exposure of BBN.

GALNT1 expression in bladder cancer tissues after 5 weeks of declined tumor formation rates compared with siCtrl treatment intravesical instillation of GALNT1 siRNA. GALNT1 siRNA was (Fig. 7G). Altogether, GALNT1 is implicated in the tumorigenesis mixed with liposome for intravesical administration as previously of bladder cancer. described (24). GALNT1 was almost undetectable in GALNT1 siRNA/liposome-treated bladder cancer tissues (Fig. 7C), whereas GALNT1 was abundantly expressed in siCtrl-treated control mice. Discussion These data suggest that intravesical instillation of GALNT1 siRNA/ BCSCswereconsideredasTICs(25),whichcouldbe liposome could efficiently deliver into bladder tissues and suc- responsible for bladder cancer initiation and recurrence. In þ þ cessfully deplete GALNT1 expression. Notably, both GALNT1 this study, we identified a BCMab1 CD44 cell population as þ siRNA and cyclopamine treatment were able to prolong survival BCSCs. The BCMab1 antibody, generating against CD44 cells rates of treated mice compared with siCtrl-treated mice. However, from human bladder cancer specimens, was definedtorec- over 60% mice treated with GALNT1 siRNA plus cyclopamine ognize aberrantly glycosylated integrin a3b1inourprevious lived as long as WT mice (Fig. 7D). Therefore, the therapeutic study (20), which is a novel biomarker for BCSCs. We dem- þ þ efficacy is due to the elimination of BCSCs. onstrate that BCMab1 CD44 cells have increased tumorige- We next examined whether GALNT1 affects metastasis of blad- nicity and self-renewal property in vivo and in vitro.Wedefined þ þ der cancer. BCMab1 CD44 cells injected via tail veins induced that the GALNT1-mediated Hh signaling activation is essential severe lung metastasis after 5 weeks, whereas BCMab1 CD44 for the maintenance of self-renewal of BCSCs and bladder cells were unable to form apparent tumors in lungs or other tissues tumorigenesis. þ þ (Fig. 7E). We established GALNT1-silenced BCMab1 CD44 cell The mammalian secreted Hh ligands include SHH, Indian lines sorted from human bladder cancer samples and injected Hedgehog, and Desert Hedgehog, which initiate a signaling them through tail veins. We observed that GALNT1-silenced cascade via engaging with their common receptor Patched1 þ þ BCMab1 CD44 cells significantly inhibited tumor metastasis (PTCH1; ref. 17). Upon Hh engagement, PTCH1 releases its (Fig. 7E), while shCtrl-treated cells still showed robust lung inhibitory role in the positive regulatory transmembrane protein metastasis. SMO, resulting in activation of Gli family transcription factors To further assess GALNT1 function in BBN-induced tumori- (Gli1, Gli2, and Gli3; refs. 26, 27). The Hh pathway is largely genesis of bladder cancer, we induced mouse bladder cancer by inactive in the adult, except for its action in tissue repair and feeding BBN and intravesically applied GALNT1 siRNA every maintenance (14, 28). The role of Hh signaling in the tumori- week. Then, we used ultrasound to observe tumor volumes every genesis of bladder cancer remains controversial. Inconsistently, two weeks. At 12 weeks, siCtrl treated mice displayed robust loss-of-function mutations of PCTH1 were rarely identified in bladder tumors in most mice (Fig. 7F). In contrast, GALNT1 bladder cancer (29, 30). Recently, Robbins and his colleagues siRNA administration effectively suppressed bladder tumor showed that the Hh signaling is highly expressed in bladder growth. In addition, GALNT1 siRNA treatment significantly cancer tissues and implicated in the tumorigenesis of bladder

www.aacrjournals.org Cancer Res; 76(5) March 1, 2016 OF9

Li et al.

cancer (31, 32). However, the role of Hh signaling pathway in therapeutic use of GALNT1 siRNA, the SMO inhibitor, and the BCSCs is largely unknown. BCMab1 antibody will provide a potent antitumor strategy Newly synthesized Hh proteins require a series of posttransla- against bladder cancer. tional processing reactions to generate their respective active forms for signal transduction. Many properties of Hh autoproces- Disclosure of Potential Conflicts of Interest fi sing were de ned from studies of the Drosophila protein (33). It is No potential conflicts of interest were disclosed. widely considered that the Hh mature processing procedure is highly conserved in various species. It has been revealed that the Authors' Contributions secreted Hh signaling domain is covalently modified by choles- Conception and design: C. Li, J. Wang, Z. Fan terol and palmitate (26). The N-terminal cysteine is catalyzed by Development of methodology: C. Li, Z. Yang the acyl-transferase Skinny hedgehog to form the amide linkage of Acquisition of data (provided animals, acquired and managed patients, palmitate (34, 35). Hydrophobic modifications confer mem- provided facilities, etc.): C. Li, Y. Du, Z. Yang, L. He, Y. Wang, R. Yan, J. Wang, brane affinity such that the signaling domain is successfully Z. Fan associated with its membrane receptors, leading to a signaling Analysis and interpretation of data (e.g., statistical analysis, biostatistics, cascade (36, 37). Here we defined a novel glycosylation modifi- computational analysis): C. Li, Z. Yang, L. He, Y. Wang, R. Yan, Z. Fan Writing, review, and/or revision of the manuscript: C. Li, Z. Yang, Z. Fan cation for SHH in BCSCs, which is uniquely mediated by Administrative, technical, or material support (i.e., reporting or organizing GALNT1. data, constructing databases): C. Li, Y. Du, L. Hao, Z. Fan Superficial tumors are usually treated by surgical resection and Study supervision: C. Li, M. Ding, J. Wang, Z. Fan intravesical chemotherapy and/or immunotherapy. The difficulty in managing bladder cancer is unable to predict which tumors will Acknowledgments recur or progress (38). A previous report showed that intravesical The authors thank Drs. Junfeng Hao and Qiang Hou for their technical instillation of polo-like kinase (PLK-1) siRNA can effectively support. prevent bladder cancer growth (24). siRNA-mediated RNAi is fi sequence-speci c gene silencing and widely used to silence gene Grant Support expression (39, 40). Here we showed that intravesical adminis- This work was supported by the National Natural Science Foundation of tration of GALNT1 siRNA can significantly suppress bladder China (81330047, 81472413); the State Projects of Essential Drug Research and tumor growth in orthotopic bladder cancer mouse models and Development (2012ZX09103301-041); 973 Program of the MOST of China BBN-induced bladder tumors. The SMO antagonist cyclopamine (2010CB911902), and the Strategic Priority Research Programs of the Chinese and its derivatives have been reported to be effective agents against Academy of Sciences (XDA01010407). The costs of publication of this article were defrayed in part by the payment of several tumors (18, 41). One of cyclopamine derivative GDC-449 page charges. This article must therefore be hereby marked advertisement in has been approved to treat advanced unresectable basal cell accordance with 18 U.S.C. Section 1734 solely to indicate this fact. carcinoma as a first-line therapy. Of note, intravesical administration of GALNT1 siRNA and cyclopamine can eradicate Received August 24, 2015; revised November 11, 2015; accepted November implanted tumor formation without overt adverse effects. Finally, 30, 2015; published OnlineFirst December 16, 2015.

References 1. Dinney CP, McConkey DJ, Millikan RE, Wu X, Bar-Eli M, Adam L, et al. 12. Shin K, Lee J, Guo N, Kim J, Lim A, Qu L, et al. Hedgehog/Wnt feedback Focus on bladder cancer. Cancer Cell 2004;6:111–6. supports regenerative proliferation of epithelial stem cells in bladder. 2. Wu XR. Urothelial tumorigenesis: a tale of divergent pathways. Nat Rev Nature 2011;472:110–14. Cancer 2005;5:713–25. 13. Shin K, Lim A, Odegaard JI, Honeycutt JD, Kawano S, Hsieh MH, et al. 3. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA: A Cancer J Clin Cellular origin of bladder neoplasia and tissue dynamics of its progression 2010;60:277–300. to invasive carcinoma. Nat Cell Biol 2014;16:469–78. 4. von der Maase H, Sengelov L, Roberts JT, Ricci S, Dogliotti L, Oliver T, et al. 14. Amakye D, Jagani Z, Dorsch M. Unraveling the therapeutic potential of the Long-term survival results of a randomized trial comparing gemcitabine Hedgehog pathway in cancer. Nat Med 2013;19:1410–22. plus cisplatin, with methotrexate, vinblastine, doxorubicin, plus cisplatin 15. Ng JM, Curran T. The Hedgehog's tale: developing strategies for targeting in patients with bladder cancer. J Clin Oncol 2005;23:4602–08. cancer. Nat Rev Cancer 2011;11:493–501. 5. Magee JA, Piskounova E, Morrison SJ. Cancer stem cells: impact, hetero- 16. Merchant AA, Matsui W. Targeting Hedgehog–a cancer stem cell pathway. geneity, and uncertainty. Cancer Cell 2012;21:283–96. Clin Cancer Res 2010;16:3130–40. 6. Rosen JM, Jordan CT. The increasing complexity of the cancer stem cell 17. Teglund S, Toftgard R. Hedgehog beyond medulloblastoma and basal cell paradigm. Science 2009;324:1670–1673. carcinoma. Biochim Biophys Acta 2010;1805:181–208. 7. Visvader JE, Lindeman GJ. Cancer stem cells: current status and evolving 18. Kim J, Aftab BT, Tang JY, Kim D, Lee AH, Rezaee M, et al. Itraconazole and complexities. Cell Stem Cell 2012;10:717–28. arsenic trioxide inhibit Hedgehog pathway activation and tumor growth 8. Reya T, Morrison SJ, Clarke MF, Weissman IL. Stem cells, cancer, and cancer associated with acquired resistance to smoothened antagonists. Cancer stem cells. Nature 2001;414:105–111. Cell 2013;23:23–34. 9. Polyak K, Hahn WC. Roots and stems: stem cells in cancer. Nat Med 19. Siegfried Z, Eden S, Mendelsohn M, Feng X, Tsuberi BZ, Cedar H. 2006;12:296–300. DNA methylation represses transcription in vivo. Nat Genet 1999;22: 10. Chan KS, Espinosa I, Chao M, Wong D, Ailles L, Diehn M, et al. Identi- 203–206. ﬁcation, molecular characterization, clinical prognosis, and therapeutic 20. Li C, Yang Z, Du Y, Tang H, Chen J, Hu D, et al. BCMab1, a monoclonal targeting of human bladder tumor-initiating cells. Proc Natl Acad Sci U S A antibody against aberrantly glycosylated integrin alpha3beta1, has potent 2009;106:14016–021. antitumor activity of bladder cancer in vivo. Clin Cancer Res 11. He X, Marchionni L, Hansel DE, Yu W, Sood A, Yang J, et al. Differentiation 2014;20:4001–13. of a highly tumorigenic basal cell compartment in urothelial carcinoma. 21. Chong L, Ruping Y, Jiancheng B, Guohong Y, Yougang F, Jiansong W, et al. Stem Cells 2009;27:1487–95. Characterization of a novel transplantable orthotopic murine xenograft

OF10 Cancer Res; 76(5) March 1, 2016 Cancer Research

GALNT1-Mediated SHH Activation in BCSCs

model of a human bladder transitional cell tumor (BIU-87). Cancer Biol. 31. Fei DL, Li H, Kozul CD, Black KE, Singh S, Gosse JA, et al. Activation of Ther 2006;5:394–98. Hedgehog signaling by the environmental toxicant arsenic may con- 22. Ulloa F, Real FX. Benzyl-N-acetyl-alpha-D-galactosaminide induces a tribute to the etiology of arsenic-induced tumors. Cancer Res 2010; storage disease-like phenotype by perturbing the endocytic pathway. 70:1981–88. J Biol Chem 2003;278:12374–83. 32. Fei DL, Sanchez-Mejias A, Wang Z, Flaveny C, Long J, Singh S, et al. 23.FoulquierF,VasileE,SchollenE,CallewaertN,RaemaekersT, Hedgehog signaling regulates bladder cancer growth and tumorigenicity. Quelhas D, et al. Conserved oligomeric Golgi complex subunit 1 Cancer Res 2012;72:4449–58. deficiency reveals a previously uncharacterized congenital disorder 33. Mann RK, Beachy PA. Novel lipid modifications of secreted protein signals. of glycosylation type II. Proc Natl Acad Sci U S A 2006;103: Annu Rev Biochem 2004;73:891–923. 3764–69. 34. Chamoun Z, Mann RK, Nellen D, von Kessler DP, Bellotto M, Beachy PA, 24. Nogawa M, Yuasa T, Kimura S, Tanaka M, Kuroda J, Sato K, et al. et al. Skinny hedgehog, an acyltransferase required for palmitoylation and Intravesical administration of small interfering RNA targeting PLK-1 activity of the hedgehog signal. Science 2001;293:2080–84. successfully prevents the growth of bladder cancer. J Clin Invest 35. Dawson PE, Muir TW, Clark-Lewis I, Kent SB. Synthesis of proteins by 2005;115:978–85. native chemical ligation. Science 1994;266:776–79. 25. Ho PL, Kurtova A, Chan KS. Normal and neoplastic urothelial 36. Feng J, White B, Tyurina OV, Guner B, Larson T, Lee HY, et al. Synergistic stem cells: getting to the root of the problem. Nat Rev Urol and antagonistic roles of the Sonic hedgehog N- and C-terminal lipids. 2012;9:583–94. Development 2004;131:4357–70. 26. Robbins DJ, Fei DL, Riobo NA. The Hedgehog signal transduction network. 37. Grover VK, Valadez JG, Bowman AB, Cooper MK. Lipid modifications of Sci Signal 2012;5:re6. Sonic hedgehog ligand dictate cellular reception and signal response. PLoS 27. Varjosalo M, Taipale J. Hedgehog: functions and mechanisms. Genes Dev One 2011;6:e21353. 2008;22:2454–72. 38. Kaufman DS, Shipley WU, Feldman AS. Bladder cancer. Lancet 2009;374: 28. van den Brink GR. Hedgehog signaling in development and homeostasis of 239–49. the gastrointestinal tract. Physiol Rev 2007;87:1343–75. 39. Dorsett Y, Tuschl T. siRNAs: applications in functional genomics and 29. Aboulkassim TO, LaRue H, Lemieux P, Rousseau F, Fradet Y. Alteration of potential as therapeutics. Nat Rev Drug Discov 2004;3:318–29. the PATCHED locus in superficial bladder cancer. Oncogene 2003;22: 40. Fellmann C, Lowe SW. Stable RNA interference rules for silencing. Nat Cell 2967–71. Biol 2014;16:10–18. 30. Xie J, Johnson RL, Zhang X, Bare JW, Waldman FM, Cogen PH, et al. 41. Tremblay MR, Lescarbeau A, Grogan MJ, Tan E, Lin G, Austad BC, et al. Mutations of the PATCHED gene in several types of sporadic extracuta- Discovery of a potent and orally active hedgehog pathway antagonist neous tumors. Cancer Res 1997;57:2369–72. (IPI-926). J Med Chem 2009;52:4400–18.

www.aacrjournals.org Cancer Res; 76(5) March 1, 2016 OF11

GALNT1-Mediated Glycosylation and Activation of Sonic Hedgehog Signaling Maintains the Self-Renewal and Tumor-Initiating Capacity of Bladder Cancer Stem Cells

Chong Li, Ying Du, Zhao Yang, et al.

Cancer Res Published OnlineFirst December 16, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-15-2309

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2015/12/15/0008-5472.CAN-15-2309.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2016/02/14/0008-5472.CAN-15-2309. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.