Circulating tumor cells (CTC) are known to be present in the blood of patients with glioblastoma (GBM). Here we report that GBM-derived CTC possess a cancer stem cell (CSC)-like phenotype and contribute to local tumorigenesis and recurrence by the process of self-seeding. Genetic probes showed that mouse GBM-derived CTC exhibited Sox2/ETn transcriptional activation and expressed glioma CSC markers, consistent with robust expression of stemness-associated genes including SOX2, OCT4, and NANOG in human GBM patient-derived samples containing CTC. A transgenic mouse model demonstrated that CTC returned to the primary tumor and generated new tumors with enhanced tumorigenic capacity. These CTCs were resistant to radiotherapy and chemotherapy and to circulation stress-induced cell apoptosis. Single-cell RNA-seq analysis revealed that Wnt activation induced stemness and chemoresistance in CTC. Collectively, these findings identify GBM-derived CTC as CSC-like cells and suggest that targeting Wnt may offer therapeutic opportunities for eliminating these treatment-refractory cells in GBM.
These findings identify CTCs as an alternative source for in situ tumor invasion and recurrence through local micrometastasis, warranting eradication of systemic "out-of-tumor" CTCs as a promising new therapeutic opportunity for GBM.
Circulating tumor cells (CTC), tumor cells that have been shed into the vasculature or lymphatics from a primary tumor and enter the systemic circulation, play a fundamental role in cancer invasion, metastasis, and recurrence (1–4). CTCs can seed, proliferate, and colonize to form secondary tumors in proximal and distal sites. Likewise, as a potential clinical biomarker, the detection of CTCs has correlated with poor prognosis, lack of treatment response, or rapid tumor recurrence in patients with a variety of cancers including glioblastoma (GBM; refs. 5–8). However, the biological mechanism(s) underlying their contribution to tumorigenesis remains largely unknown. Understanding this contribution may serve to uncover new therapeutic targets to prevent cancer progression and recurrence.
GBM, grade IV glioma, is the most common and most aggressive primary brain tumor. GBM is among the most lethal of human malignancies, with a current median overall survival of approximately 16 months (9, 10). Despite aggressive standard-of-care treatments including surgical resection, radiation, and chemotherapy, recurrence of GBM is essentially universal, and recurrent tumors are highly resistant to conventional cytotoxic treatments. It is highly suggested that treatment-resistant glioma cells, particularly cancer stem cells (CSC), that is, tumor-initiating cells or tumor-propagating cells, contribute to the GBM recurrence. Here we show glioma CTCs acquire a CSC-like phenotype: activated in stemness, resistant to genotoxic treatments, and more importantly, able to home to a primary tumor site to repopulate locally and contribute to new tumor formation. Taken together, this suggests a previously unidentified role of CTCs for tumor micrometastasis and local relapse in GBM.
Materials and Methods
Patient-derived samples containing CTCs
Patients with GBM or non–small cell lung carcinoma (NSCLC) were treated at the Department of Radiation Oncology and Department of Neurosurgery of the Hospital of the University of Pennsylvania. Written informed consent was obtained from each patient. The studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki, CIOMS, Belmont Report, and U.S. Common Rule) and approved by the University of Pennsylvania Institutional Review Board. The blood samples were collected under an Institutional Review Board–approved protocol. CTCs were genetically labeled and isolated as previously described. Ten milliliters of blood was diluted with equal volume of PBS, and transferred and separated via Oncoquick tube. After centrifugation at 400 x g for 35 minutes, the layer solution containing CTCs between the Ficoll and the blood was collected. The cells were collected by centrifugation at 400 x g for 35 minutes, and resuspended in 900 μL of DMEM media (Invitrogen) supplemented with 5% FBS (Invitrogen), 1.0% penicillin-streptomycin (Invitrogen). The hTERT promoter activity-based conditional replicated adenovirus (Oncolys BioPharma) was utilized to label the CTCs as previously described (7). Cells were incubated with 2 × 108 viral particles for 24, 48, and 72 hours in chamber slides (BD Biosciences). The cells were harvested by using a capillary-based vacuum-assisted cell acquisition system (Kuiqpick, NeuroInDx). Calibration was initially performed under bright and fluorescence field condition, and then pick up the selected cells (via micro-capillary aspiration) under fluorescence microscopy.
RCAS GBM mouse model
GBM was induced in mice as previously described (11, 12). In brief, chicken DF-1 fibroblasts (ATCC) were transfected with RCAS-PDGF-B and RCAS-Cre plasmids to produce retrovirus, and then orthotopically injected into the brain of Ntv-a;Ink4a-Arf−/−;Ptenfl/fl;LSL-luc mice to induce GBM through RCAS/n-tva-mediated gene transfer. Tumorigenesis in brain was detected by bioluminescence imaging. Tumor growth was monitored by whole-body bioluminescence using an IVIS 200 Spectrum Imaging System (Perkin Elmer) after retro-orbital injection of D-luciferin (150 mg/kg; GoldBio). Tumors were isolated and subjected to mechanical dissociation with a gentleMACS Dissociator (Miltenyi), and enzymatic digestion with collagenase II and dispase II to obtain single cell suspensions. To analyze stemness transcriptional activation in CTCs, the tumor cells were transduced with lentivirus that encodes Sox2/Etn-GFP, followed by orthotopic injection (105 tumor cells/mouse) into wild-type C57BL/6 mice (8-weeks old, half male and half female; The Jackson Laboratory).
Isolation and culture of mouse CTCs
The isolation and labeling CTCs were performed in a protocol similar to isolation of human CTC as described above (7). In brief, 1 mL of blood or tumor cell suspension was collected from each GBM-bearing mice, diluted with equal volume of PBS, and layered over Ficoll solution. After centrifugation, the layer solution between the Ficoll and the blood was collected. The cells were collected by centrifugation, and resuspended in serum-free Neurobasal-A medium (Gibco), and cultured for 3 days in a humidified hypoxic atmosphere with 1% O2 and 5% CO2. Cells were then incubated with 2 × 108 viral particles for 2 days in chamber slides (BD Biosciences), followed by single-cell pickup of mCherry-expressing cells using the Kuiqpick cell acquisition system. The collected mouse CTCs and matched primary tumor cells were maintained in serum-free Neurobasal-A medium (Gibco), supplemented with B-27 Supplement Minus Vitamin A (Gibco), GlutaMax (Gibco), sodium pyruvate (Gibco), fibroblastic growth factor (FGF, 5 ng/mL; R&D Systems), and EGF (20 ng/mL; R&D Systems).
Human GBM CSC culture
Human patient-derived IN528 glioma CSCs were kindly provided by Dr. Jeremy Rich (University of California at San Diego, La Jolla, CA; refs. 13–15). The matched non-CSCs were generated by brief treatment with serum (10% FBS)-containing medium for 24 hours, and cultured back in stem cell medium as described previously (16). CSCs were cultured in serum-free Neurobasal-A medium (Gibco), supplemented with B-27 Supplement Minus Vitamin A (Gibco), GlutaMax (Gibco), sodium pyruvate (Gibco), fibroblastic growth factor (FGF, 5 ng/mL; R&D Systems), and EGF (20 ng/mL; R&D Systems).
Syngeneic glioma model
The CTCs or primary tumor cells (105 cells for each mouse) were subcutaneously injected into the right and left flank sites of C57BL/6 mice (8-weeks old; The Jackson Laboratory half male and half female). In addition, these cells (104 cells for each mouse) were intracranially injected into mouse brains, as previously descried (12). Tumor growth was monitored by whole-body bioluminescence using an IVIS 200 Spectrum Imaging System (Perkin Elmer) after retro-orbital injection of D-luciferin (150 mg/kg; GoldBio). The size of tumors was measured every week by using a caliper and the volume calculated.
CTC homing analysis
GBM was induced in Ntv-a;Ink4a-Arf−/−;Ptenfl/fl;LSL-luc mice through RCAS/n-tva-mediated gene transfer. Cultured CTC cells were lentivirally transduced to co-express GFP and rLuc, and prepared as single-cell suspensions in PBS (2 × 106 cells/mL). Mice were anesthetized by intraperitoneal injection of ketamine/xylazine cocktail (ketamine 100 mg/kg, xylazine 10 mg/kg), and injected with 100 μL CTC cell suspension through carotid artery. Ten days after injection, mice were imaged by whole-body bioluminescence using an IVIS 200 Spectrum Imaging System (Perkin Elmer) after retro-orbital injection of coelenterazine (4 mg/kg; Promega), followed by secondary imaging after retro-orbital injection of D-luciferin (150 mg/kg; GoldBio). The tumors were excised and subjected to IHC analysis.
Quantitative real-time PCR analysis of human single cells
mRNA was isolated from harvested single cells including human CTCs, human glioma U251 cells by using an RNeasy Mini Kit (Qiagen). Quantitative real-time PCR was performed in a 20-μL reaction volume using Fast SYBR Green Master Mix (Applied Biosystems) and primers: Oct4 (FP: 5′- GTGGAGGAGCTGACAACAA -3′, RP: 5′- ATTCTCCAGGTTGCCTCTCA -3′), Sox2 (FP: 5′- CCCCCGGCGGCAATAGCA -3′, RP: 5′- TCGGCGCCGGGGAGATACAT -3′), Nanog (FP: 5′- CAAAGGCAAACAACCCACTT-3′, RP: 5′- TCTGCTGGAGGCTGAGGTAT-3′), BTRC (FP: 5′-CTGCAGGGACACTCTGTCTAC -3′, RP: 5′-GAAGTCCCAGATGAGGATTGTG-3′), CCND1 (FP: 5′-TCAAGTGTGACCCGGACTGC-3′, RP: 5′-CAGGTCCACCTCCTCCTCCT-3′), CCND2 (FP: 5′-CCACCGTCGATGATCGCAAC-3′, RP: 5′-GAGGAGCACCGCCTCAATCT-3′), FGF9 (FP: 5′-AGACCACAGCCGATTTGGCA -3′, RP: 5′-CCTTCCAGTGTCCACGTGCT -3′), GAPDH (FP: 5′-GTCTCCTCTGACTTCAACAGCG-3′, RP: 5′-ACCACCCTGTTGCTGTAGCCAA -3′).
GBM was induced in Ntv-a;Ink4a-Arf−/−;Ptenfl/fl;LSL-Luc mice through RCAS/n-tva-mediated gene transfer. Tumors and blood were collected, and single-cell CTCs and matched tumor cells (5–8 cells/mouse) were harvested and pooled. Cells were lyzed in Hank's Balanced Salt Solution buffer (Thermo Fisher) with 0.025% Tween (Thermo Fisher). mRNA was extracted and subjected to aRNA linear amplification by three rounds of reverse transcription, second strand synthesis and in vitro transcription, according to a previously published protocol (17), using SuperScript III First-Strand Synthesis SuperMix (Thermo Fisher), DNA Polymerase I (Thermo Fisher), T4 DNA polymerase (Thermo Fisher), 5x Second Strand buffer (Thermo Fisher), and a MegaScript Kit (Thermo Fisher). DNA library was constructed with a TruSeq mRNA Stranded Kit (Illumina). The RNA from each step and library DNA quality were analyzed with RNA Nano assay chips, RNA Pico assay chips, and DNA Nano assay chips using a 2100 bioanalyer (Agilent). Library was subjected to next-generation sequencing analysis in a high-throughput sequencing center with a HiSeq4000 at the Children's Hospital of Philadelphia/Beijing Genomics Institute core facility (CHOP/BGI). The sequences were aligned to the GRCm38 reference genome using RNA-Star (v2.4.2a; https://github.com/alexdobin/STAR). The gene expression was normalized and calculated as FPKM (Fragments Per Kilobase Million) values by Cufflinks (v2.2.1) (http://cole-trapnell-lab.github.io/cufflinks/releases/v2.2.1/) with Gencode M5 gene annotations (https://www.gencodegenes.org/mouse/release_M5.html). RNA-seq data have been deposited in NCBI's Gene Expression Omnibus under the accession GSE120776.
GBM was induced in Ntv-a;Ink4a-Arf−/−;Ptenfl/fl;LSL-luc mice through RCAS/n-tva-mediated gene transfer. Tumors were excised. Single-cell tumor suspension was transduced with lentivirus that encodes Sox2/Etn-GFP or promoter-free sequence. The cells were subjected to flow cytometry analysis with an Accuri C6 flow cytometer (BD Biosciences) by using FlowJo software.
In vitro circulation system
Mouse CTCs and tumor cells were cultured in serum-free Neurobasal-A. A total of 2 × 106 cells in 30 mL medium were transferred into an in vitro circulation system with a fluid circulation rate of 80 mL/min. The circulation system was placed in the cell culture incubator at 37°C. After exposure to circulation for certain time, cells were collected and subjected to cell viability and apoptosis analyses.
Cell viability assay
Mouse CTCs and tumor cells were cultured as neurospheres in serum-free Neurobasal-A medium or as monolayers (for tumor cells only) in DMEM medium supplemented with 10% FBS. Cells were treated with TMZ (100 μmol/L; SelleckChem) for 48 hours, irradiated by 5-Gy X-ray with a dose rate of 2.8 Gy/min using an X-Rad 320ix cabinet system (Precision X-Ray), or placed in the in vitro circulation system. The cell viability was detected using a CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega).
Cell apoptosis assay
Mouse CTCs and tumor cells were cultured in serum-free Neurobasal-A medium, and placed in the in vitro circulation system for 8 hours. Cells were stained with propidium iodide and FITC-conjugated Annexin V, followed by flow cytometry analysis with an Accuri C6 flow cytometer (BD Biosciences) by using FlowJo software.
GBM was induced in Ntv-a;Ink4a-Arf−/−;Ptenfl/fl;LSL-luc mice through RCAS/n-tva-mediated gene transfer. Tumors and blood CTCs were collected and transduced with adenovirus that encode TERT-Cherry. Cells were fixed with 4% paraformaldehyde for 15 minutes in room temperature, and treated with 0.25% Triton X-100 for 10 minutes for permeabilization, followed by incubation with anti-CD133 (1:100, Miltenyi, 130-092-442) and Alexa Fluor 488-conjugated anti-Olig2 antibody (1:100; Millipore, MABN50A4). Cells were stained with Alexa Fluor 568-conjugated secondary IgGs (1:500; Life Technologies) for 1 hour at room temperature, and mounted with the media containing DAPI. Images were acquired with an Axio Imager A1 upright fluorescence microscope (Zeiss) equipped with an Axiocam 506 cooled CCD camera (Zeiss).
Cultured mouse CTCs and tumor cells were lysed with a RIPA buffer containing protease inhibitor cocktail (Roche) on ice for 20 min. A total of 20 μg of protein was resolved by electrophoresis with precasted 4-15% SDS-PAGE gels (Bio-Rad), followed by transfer to PVDF membranes and protein blocking. The following antibodies were employed for immunoblot: anti-GAPDH (Cell Signaling Technology; 5174), anti-Oct4 (Cell Signaling Technology; 2840), and anti-Sox2 (Cell Signaling Technology; 2748S) antibodies at 1:1,000 dilution. Proteins were detected with HRP-conjugated antibodies specific for either rabbit or mouse IgG (Bio-Rad), followed by ECL development (GE Healthcare; RPN2232).
Paraffin tissue sections with mouse tumors were deparaffinized and rehydrated, and subjected to antigen retrieval in Target Retrieve Solution (DAKO; S1699) at 95°C for 20 minutes. For histology study, sections were stained with hematoxylin and eosin dyes. For IHC study, sections were blocked with 5% horse serum for 1 hour at room temperature, and incubated with anti-Ki67 (1:100; Millipore, AB9260) or anti-GFP (1:100; Millipore, 06-896) antibody overnight at 4°C, followed by staining with ImmPACT DAB (Vector Laboratories). Images were acquired with an AxioLab microscope (Zeiss) equipped with an AxioCam HRC CCD camera (Zeiss).
Sphere formation assay
CTCs and primary tumor cells were trypsinized and suspended in serum-free Neurobasal-A medium as single-cell suspensions, followed by seeding into six-well plates at 5,000 cells per well. The cells were treated with or without 1 μmol/L XAV939. After incubation for 8 days, the spheres were imaged and counted under an Axiovert 50 microscope (Zeiss) equipped with an AxioCam HRM CCD camera (Zeiss).
Limited dilution assay
In vitro limiting dilution assay was performed as previously described (18). Briefly, CTCs and primary tumor cells were trypsinized and suspended in serum-free Neurobasal-A medium as single-cell suspensions, followed by seeding into 96-well plates at 1, 5, 10, 20, 40, and 80 cells/well. The cells were treated with or without 10 μmol/L XAV939. After incubation for 10 days, the occurrence of spheres was observed under an Axiovert 50 microscope (Zeiss) and the wells with at least one sphere was recorded as positive.
LEF1 reporter assay
Renilla luciferase cDNA was subcloned into a pTopflash plasmid vector to replace original firefly luciferase. Mouse CTCs and tumor cells were cotransfected with the generated pLEF1-Renilla and a plasmid encoding β-gal by using Lipofectamine 2000 (Invitrogen). The activity of Renilla luciferase was detected 24-hour posttransfection with a Dual-Luciferase Reporter Assay System (Promega), and β-gal was used as an internal control.
All experiments with mice were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania and with the Guide for the Care and Use of Laboratory Animals (8th ed., The National Academies Press, 2011). The collection of human tissues in compliance with the tissue banking protocol was approved by the University of Pennsylvania Institutional Review Board, and written informed consent was obtained from each participant.
Comparison between groups were performed by Student t (unpaired two-tailed analysis) and log-rank (Mantel–Cox) tests using Prism software, and P < 0.05 was considered statistically significant.
GBM CTCs are CSC-like cells
We initially investigated the potential egress of CSCs into circulation as CTCs in mouse gliomas. We took advantage of an orthotopic, genetically engineered murine GBM model with a native microenvironment, induced by the RCAS/N-tva-mediated somatic pdgfb gene transfer in Ink4a-Arf−/−;Pten−/− neural stem/progenitor cells (11, 12). Before tumor transplantation into recipient mice, the tumor cells were transduced to express a genetic probe to trace cells with stemness activation (Fig. 1A), in which GFP expression is driven by an early transposon promoter (Etn) with Sox2-binding motifs (19). CTCs harvested from the blood of the recipient mice were analyzed by a telomerase promoter-driven mCherry expression probe that is part of an adenoviral CTC-tagging system developed to detect human CTCs (7, 20). Notably, all mCherry+ CTCs appeared to express GFP (Fig. 1B), whereas only 2% to 5% primary tumor cells express GFP (Supplementary Fig. S1), suggesting stemness transcriptional activation in glioma CTC cells. Moreover, immunofluorescence analysis showed that glioma CTCs, but not control primary tumor cells, robustly expressed stemness-associated transcriptional factor Olig2 (almost all cells; Fig. 1C) and a subpopulation of CTCs express glioma stem cell marker CD133 (about 40% cells), verifying that glioma CTCs are CSC-like. Furthermore, we investigated expression of stemness-associated transcriptional factors in human astrocytes or CTCs from patients with GBM or early stage NSCLC, related to the expression level of human U251 GBM cells, by using a single-cell microcapillary isolation system (Fig. 1D, left). Real-time RT-PCR analysis of pooled single cells revealed that GBM CTCs, and to a lesser extent, NSCLC CTCs, robustly expressed SOX2, OCT4, and NANOG, in contrast to a much lower expression of these genes in normal astrocytes (Fig. 1D, right), suggesting that human CTCs are also CSC-like.
We next established an optimized protocol to harvest and propagate CTCs in vitro (Fig. 2A). Cultured CTCs were able to form distinct neuropheres in serum-free stem cell medium in vitro (Fig. 2B). Interestingly, half of the spheres maintained long-term (4 weeks) Sox2/Etn-driven GFP expression after the probe transduction (Fig. 2B). Real-time RT-PCR analysis showed that these cultured CTCs consistently expressed Sox2, Oct4, and Nanog mRNA at high levels (Fig. 2C). The increased protein expression of Sox2 and Oct4 in CTCs were validated by immunoblot (Fig. 2D). In addition, we compared expression of these genes in mouse CTCs and human GBM CSCs. Our data show that GBM CTCs expressed SOX2 and OCT4 at a similar or higher level to CSCs, which was higher than the serum-treated, differentiated non-CSCs (Supplementary Fig. S2). Furthermore, as expected, treatment of these CTCs with 10% serum-supplemented medium altered their morphology from a neurosphere population to that of adherent and elongated cells (Supplementary Fig. S3); concomitantly, there was also an induced loss of Sox2/Etn-driven GFP expression, suggesting serum-induced differentiation in these cells.
CTCs are highly tumorigenic and able to home to the primary tumor site
Next, we sought to investigate the tumorigenicity of CTCs by utilizing a subcutaneous implantation model (Fig. 3A) and an intracranial model (Supplementary Fig. S4A). Whole-body bioluminescence imaging of side-by-side subcutaneous tumors showed that implantation of CTCs developed significantly larger tumors than those derived from matched primary tumor cells (Fig. 3B). Moreover, the CTC-derived tumors grew faster than the tumor cell-derived tumors (Fig. 3C). Furthermore, mice with intracranial implantation of CTCs exhibited shorter survival than mice received tumor cell injection (Supplementary Fig. S4B), collectively suggesting a robust tumorigenic characteristic of these CTCs. Interestingly, CTC-derived tumors exhibited a highly invasive phenotype, as evidenced by invasion to adjacent muscle and blood vessel tissues (Fig. 3D). In addition, approximately 40% of the cells in CTC-derived tumors expressed proliferation marker Ki-67, indicating a highly proliferative nature, in comparison to a Ki-67 expression of about 20% of the cells in primary tumor cell-derived tumors (Supplementary Fig. S5).
Stem cells are known to be able to home to inflammatory, injured, or cancerous sites, when circulating in blood. Thus, we postulated that CSC-like glioma CTCs might be capable of seeding and colonizing a primary tumor, which could ultimately contribute to local GBM recurrence. To test this hypothesis, we generated genetically labeled CTCs that co-express GFP and Renilla luciferase (rLuc) under EF1 and CMV promoter, respectively, and via intravascular injection, administered these CTCs into a genetically engineered GBM mouse model in which tumor cells express firefly luciferase (fLuc; Fig. 4A). Whole-body bioluminescence imaging identified a distinct rLuc signal colocalizing with the fLuc-expressing primary tumor (Fig. 4B), indicating that CTCs had in fact homed to the intracranial tumor site. No other appreciable rLuc bioluminescence was observed in the body. Furthermore, IHC analysis of these tumors verified that these GFP+ CTCs were localized in the border area of the primary tumor (Fig. 4C). Importantly, these experiments demonstrated that the CTCs colonized and formed a secondary tumor that invaded to the primary tumor. Together, these results suggest that CTCs are highly tumorigenic and capable of homing to, and generating an additional tumor at the primary tumor site.
CTCs are resistant to cytotoxic treatments and circulation-induced stress
We next investigated the responses of CTCs to cytotoxic treatments. At the basal level without treatments, CTCs and control tumor cells exhibited a similar proliferative capacity when cultured in the same serum-free stem cell medium (Fig. 5A). Interestingly, CTCs were highly resistant to radiotherapy, as indicated by the doubling of viable CTCs 10 days after 5-Gy irradiation, compared to a more than 50% loss of viable tumor cells over the same period of time when cultured in the same serum-free stem cell medium or serum-containing DMEM medium (Fig. 5A). Similarly, CTCs were more resistant to treatment with temozolomide (TMZ), the agent for standard GBM chemotherapy, in comparison to tumor cells (Fig. 5B). These results suggest that CTCs are more refractory to cytotoxic treatments than their tumor cell counterparts. In addition, we compared the effects of circulation-induced stress on the viability of CTCs and tumor cells. To that end, we customized and built an in vitro circulation system by using 3-D printing materials (Fig. 5C). This system was operated with a fluid circulation rate of 80 mL/min, to model the blood flow in the middle arteries within normal human anatomy. Cell viability analysis showed that CTCs were capable of improved survival when compared to tumor cells in circulating medium (Fig. 5D). Consistent with these findings, circulation induced less cell apoptosis in CTCs than in tumor cells (Fig. 5E), suggesting that CTCs have higher tolerance to circulation-induced stress than tumor cells.
Single-cell RNA-seq analysis reveals Wnt activation in CTCs, inducing stemness and chemoresistance
We finally sought to explore the mechanistic underpinnings by which CTCs acquire stemness and develop treatment resistance by utilizing a CTC RNA-seq transcriptome analysis approach. Single CTCs were labeled by using the TERT-mCherry system and harvested (Fig. 6A). We succeeded in labeling and harvesting five to eight CTCs from each mouse. These cells were then pooled, and the extracted RNA subjected to linear amplification and deep sequencing (Fig. 6A). RNA-seq transcriptome analysis showed a change in global gene expression (Fig. 6B) and revealed that over 25% of total mapped genes had an over 60% change in expression levels (Fig. 6C). Moreover, CTCs and matched tumor cells from the same mouse exhibited distinct global expression profiles (Fig. 6D). Strikingly, computational bioinformatics analysis of the top 1,000 upregulated genes' promoter sequences in CTCs identified consensus DNA motifs that are known to be recognized by several transcriptional factors, including SP1 and LEF1 (Fig. 6E). Notably, LEF1 is the master transcriptional factor that controls Wnt signaling activation, which has a well-established role of Wnt in stemness and tumor treatment resistance. TOP-flash assays confirmed greater constitutive LEF1 activation in CTCs, compared to the matched tumor cells (Fig. 6F).
Interestingly, treatment of CTCs, but not tumor cells, with XAV939, a selective pharmacological inhibitor of Wnt pathway, remarkedly inhibited sphere formation (Fig. 7A). Likewise, limited dilution analysis showed that XAV939 treatment abrogated the stem cell frequency in CTCs but not in tumor cells (Fig. 7B), confirming that Wnt is critical for stemness activation in CTCs. Furthermore, XAV939 treatment sensitized CTCs, but not tumor cells, to TMZ chemotherapy (Fig. 7C), suggesting a requisite role of Wnt for chemoresistance in CTCs. In addition, XAV939 also inhibited neurosphere formation in human GBM CSCs (Supplementary Fig. S6), suggesting a critical role for Wnt in stemness in both GBM CSCs and CTCs. Finally, we tested possible Wnt activation in GBM patient-derived CTCs. RT-PCR analysis demonstrated a robust expression of Wnt target genes including CCND1, CCND2, BTRC, and FGF9 in CTCs, compared with those in U251 glioma cells and human astrocytes (Fig. 7D), verifying Wnt activation in GBM patient-derived CTCs.
CSCs play a critical role in treatment resistance and cancer recurrence. Here our study suggests that CSCs could egress from the tumor bulk into circulation as CTCs, survive in circulation during focal and systemic therapy, and re-seed the primary tumor bed to generate new local tumors, providing an alternative source for in situ tumor invasion. Maximal safe resection of the bulk tumor remains the standard of care in GBM and contributes to improved survival, but we speculate that CTCs may potentially return to the tumor bed via this reseeding mechanism, repopulate, and thus contribute to the universal recurrence rate seen in this disease.
Extracranial metastasis of GBM (EMG) is extremely rare, occurring in less than 2% of patients. This finding lends support to classical theories that patients have insufficient survival time for distant metastasis to manifest before they succumb to relapsed intracranial disease or that the presence of the blood–brain barrier (BBB) tends to limit hematogenous and lymphatic spread of CTCs (21, 22). Even so, the presence of glioma-derived CTCs has now been demonstrated by us and other independent groups (7, 23, 24). For example, a recent study showed that glioma CTCs occur much more frequently, as evidenced by detection of GFAP+ CTCs in over 20% of patients with GBM (24). In accord, our previous work suggests that the sequential enumeration of telomerase transcription-tagged CTCs may correlate with treatment responses in patients with GBM (7). It is also noteworthy that patients with EMG or spinal metastasis show shorten survival time and poor response to radiotherapy (21, 25), collectively suggesting a potential pathologic role of CTCs in GBM tumor progression and recurrence. Although, we did not observe CTC seeding to other sites of the body in our mouse experiments, this may be influenced by the much shorter survival of mice compared to patients. We speculate that metastases to extracranial sites might have eventually occurred over a longer time period than what was tested in our model.
As a classic biological feature, stem cells can egress from bone marrow into circulation and home to injured or inflammatory sites (26, 27). Consistent with a recent study showing that ex vivo cultured colorectal CTCs express stem cell-associated markers (28), our studies characterize stemness in glioma CTCs, and more importantly, we reveal that glioma CTCs are able to home to the primary tumor site and contribute to tumor growth. Growing evidence suggests that CSCs are tumorigenic and refractory to conventional genotoxic treatments, significantly contributing to tumor recurrence after treatment (29–31). Thus, highly tumorigenic CTCs may play an unexpected role in GBM recurrence as a CSC reservoir, that is, forming a TMZ-resistant circulating subpopulation, evading local radiotherapy and ultimately re-seeding and colonizing the primary tumor site to form new tumors. Likewise, clinical and surgical pathology evidence suggests that extension of GBM throughout the brain is surprisingly pervasive, which might not be sufficiently explained by just direct high invasion of the tumor cells. It is tempting to speculate that a continual seeding of CTCs into the primary tumor bed may contribute to local micro-metastasis and CSC colony evolution-mediated GBM heterogeneity.
CSCs have been well characterized in GBM, whose stemness is subjected to regulation by developmental signaling including Wnt, Notch, and hedgehog (32–35). A recent study shows that noncanonical Wnt signaling in CTCs predicts antiandrogen resistance in prostate cancer cells (36). Here we identify that Wnt activation induces stemness and renders CTCs resistant to conventional treatment. Eradication of CTCs by targeting Wnt may therefore represent a paradigm-shifting therapeutic strategy for GBM, one of the deadliest human malignancies, and possibly other solid tumors. We are in the process of investigating this therapeutic approach.
Disclosure of Potential Conflicts of Interest
M. Alonso-Basanta has received speakers bureau honoraria from Varian. D.M. O'Rourke reports receiving a commercial research grant from Novartis; has ownership interest (including stock, patents, etc.) in Isoma Diagnostics; is a consultant/advisory board member of Isoma Diagnostics; and has provided expert testimony for clinical expert testimony. G.D. Kao has ownership interest (including stock, patents, etc.) in Liquid Biotech USA. No potential conflicts of interest were disclosed by the other authors. J.F. Dorsey ownership interest (including stock, patents, etc.) in Liquid Biotech USA, Inc. No potential conflicts of interest were disclosed by the other authors.
Conception and design: T. Liu, H. Xu, DM. O'Rourke, Y. Gong, G.D. Kao, J.F. Dorsey, Y. Fan
Development of methodology: T. Liu, M. Huang, W. Ma, D. Saxena, G.D. Kao, J.F. Dorsey
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Xu, M. Huang, W. Ma, D. Saxena, R.A. Lustig, DM. O'Rourke, J.F. Dorsey
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Liu, H. Xu, Z. Zhang, DM. O'Rourke, L. Zhang, G.D. Kao, J.F. Dorsey
Writing, review, and/or revision of the manuscript: T. Liu, M. Alonso-Basanta, Z. Zhang, DM. O'Rourke, Y. Gong, G.D. Kao, J.F. Dorsey, Y. Fan
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Liu, J.F. Dorsey
Study supervision: T. Liu, G.D. Kao, J.F. Dorsey, Y. Fan
We are grateful to Eric Holland for providing RCAS-PDGF GBM model, to James Eberwine for single cell transcriptome analysis, and to Celeste Simon for helpful discussions. We acknowledge the Radiation Oncology Clinical Research Coordinators at the University of Pennsylvania for their contribution. This work was supported in part by National Institutes of Health grants R01NS094533 and R01NS106108 (to Y. Fan), R01CA201071 (to G. Kao and J. Dorsey), K08NS076548 and R01CA181429 (to J. Dorsey), and R01CA190415 (to L. Zhang), AACR Judah Folkman award (to Y. Fan), and B*Cured Foundation Brain Cancer Investigator Award (to Y. Fan).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.