Detection of metastatic tumors after γ-irradiation using longitudinal molecular imaging and gene expression profiling of metastatic tumor nodules.

A few recent reports have indicated that metastatic growth of several human cancer cells could be promoted by radiotherapy. C6-L cells expressing the firefly luciferase (fLuc) gene were implanted subcutaneously into the right thigh of BALB/c nu/nu mice. C6-L xenograft mice were treated locally with 50-Gy γ-irradiation (γ-IR) in five 10-Gy fractions. Metastatic tumors were evaluated after γ-IR by imaging techniques. Total RNA from non-irradiated primary tumor (NRPT), γ-irradiated primary tumor (RPT), and three metastatic lung nodule was isolated and analyzed by microarray. Metastatic lung nodules were detected by BLI and PET/CT after 6-9 weeks of γ-IR in 6 (17.1%) of the 35 mice. The images clearly demonstrated high [18F]FLT and [18F]FDG uptake into metastatic lung nodules. Whole mRNA expression patterns were analyzed by microarray to elucidate the changes among NRPT, RPT and metastatic lung nodules after γ-IR. In particular, expression changes in the cancer stem cell markers were highly significant in RPT. We observed the metastatic tumors after γ-IR in a tumor-bearing animal model using molecular imaging methods and analyzed the gene expression profile to elucidate genetic changes after γ-IR.

Introduction

Glioma is the most common malignant primary brain tumor; thus, it is treated aggressively with surgery and chemotherapy and radiotherapy. The poor survival of patients with glioma reflects the prevalence of cancer recurrence after surgery, invasion into other sites, and intrinsic or acquired resistance to chemotherapy and radiotherapy (1). Ionizing irradiation (IR) is the most commonly employed treatment modality for various human cancers and regional cancer disease. IR is often used in combined treatments with chemotherapy and surgery (2). Concomitant combination therapy is used to treat patients with non-small lung, head and neck, cervical cancer, and glioblastoma (3–7), ~50% of cancer patients receive radio-therapy (8). The goal of radiotherapy is to provide a suitable dose to the primary tumor, while minimizing IR side-effects to surrounding normal tissues. Notably, one of the problems with lower dose (0.1–0.6 Gy) or high dose (10 Gy and higher) fractionated radiotherapy is radio-resistance and bystander factors released among treated cancer cells and in an animal model (9). Patients undergoing radiotherapy for local tumor control show better survival rates than patients with radiation-induced secondary tumors, which is attributed primarily to the observation that local treatment failure increases the probability of developing metastatic disease at distant organ sites (10,11). The first study on the effect of local tumor IR on metastatic frequency in a transplantable mouse carcinoma model was reported in 1949 by Kaplan and Murphy (12). Since then, various research groups have reported that the incidence of metastasis increases after IR of the primary tumor (13). Radio-therapeutic effects are considered because of the induction of DNA damage causing cell cycle arrest, apoptosis and senescence (14). IR induces modifications in the tumor microenvironment, which have a profound impact on tumor biology (15) and the incurred tumor hypoxic conditions can promote metastasis by recurrence in untreated hypoxic cells (16). It is now becoming evident that IR can also result in cancer cells acquiring a stemness state characterized by increased stemness gene expression and a cancer stem cell-like phenotype (17). Some studies indicate that irradiated tumors contain a stemness population and that growth of distant metastasis is driven by cancer stem-like cells. Furthermore, several studies have shown that the epithelial-mesenchymal transition (EMT) has a crucial role in IR resistance of cancer cells (18). EMT was initially recognized as an important process during the morphogenesis of epithelial tissue in embryonic development, and is now shown to be one of the key steps promoting tumor metastasis. The EMT maintains cancer stemness and is induced by various factors. Several effectors, including transforming growth factor (TGF)-β, fibronectin (Fn1), metalloproteinases (MMPs), vimentin (Vim) and cadherin, mediate the EMT. Acquisition of stemness results in metastasis along with CD24, CD133, β-catenin, Oct-4 and Sox-2 expression in non-small cell lung cancer cells (19). Although the relationships between PI3K/Akt/mTOR signaling, EMT and cancer stem cells are known, the regulation mechanisms of metastasis in irradiated tumor are still unclear.

Regular follow-up examinations, such as imaging and tumor marker tests, may be required after radiotherapy to detect metastasis at an early stage. Therefore, many investigators utilize advance imaging techniques to monitor neoplastic progression to metastasis, such as bioluminescence imaging (BLI), computed tomography (CT) and positron emission tomography (PET) (20). These imaging modalities can evaluate progression, such as tumor growth, metabolism, or metastasis in vivo, in a longitudinal manner. A non-invasive imaging method is indispensable for detecting tumor lesions in an internal organ, such as the lungs, liver or brain, and diagnosing and determining the size of tumors in a preclinical animal model (21–24). Other ways to detect cancer involved in distance metastasis include immunohistochemical staining of ex vivo biopsies; however, these often lack reproducibility and accuracy (25).

In the present study, metastatic tumors in C6-L xenografted mice were studied after local treatment with fractionated γ-IR. To accurately detect the metastatic nodules after γ-IR, we observed the effect of γ-IR on distant metastatic tumor growth using different imaging modalities, such as BLI and PET/CT. A non-invasive longitudinal imaging study with repeated measurements of metastatic nodules after γ-IR indicated extensive colonization of C6-L cells in the lungs within 6 weeks after γ-IR. We also identified and described the molecular events occurring after γ-IR through gene expression profiling to elucidate genetic changes. We identified the differentially expressed genes between the γ-IR primary tumors vs. non-γ-IR primary tumors and metastatic lung nodules vs. γ-IR primary tumors using an Agilent Expression microarray contained ~30,003 Entrez Gene RNAs. In particular, we found known cancer stem cell markers and detected EMT among the differentially expressed genes.

Materials and methods

Cell culture

C6 rat glioma cells and C6-L infected cells containing the firefly luciferase (fLuc) gene in lentiviral vectors were used with selected blasticidin treatment (5 mg/ml), as previously described by Park et al (26).

Xenograft model and local Agilent Expression microarray-IR

BALB/c nu/nu mice (females, 5–6 weeks of age) were purchased from Orient Bio, Inc. (Seoul, Korea). C6-L cells (5×105/head) were implanted subcutaneously into the right thigh of mice. When the tumors reached ≥80 mm3 (15 days after inoculation), we randomly assigned them to the C6-L γ-irradiated (γ-IR) and non-IR groups. C6-L tumor-bearing mice were treated locally with 50 Gy γ-IR in five 10-Gy fractions every day using a 60Co γ-IR source (Theratrom 780; AECL, Ltd., Mississauga, ON, Canada; n=35), but not the control group (n=5). The mice were anesthetized with an intraperitoneal injection of a mixture of zolazepam/tiletamine (50 mg/kg; Zoletil 50®; Virbac, Magnyen-Vexin, France) and xylazine (10 mg/kg; Rompun®; Bayer Healthcare, Berlin, Germany) fixed on an acryl plate. All experiments with animals were carried out according to the guidelines for the care and the use of experimental animals and were approved by the Korea Institute of Radiological and Medical Sciences.

BLI acquisition

BLI was performed with a highly sensitive, optical CCD camera mounted in a light-tight specimen chamber (IVIS200; Xenogen, Alameda, CA, USA). Animals were given the firefly substrate D-luciferin potassium salt diluted to 2.5 mg/100 μl in saline. The mice were injected intraperitoneally with 100 μl of this D-luciferin solution and were anesthetized (2% isoflurane) for in vivo imaging. The mice were placed on the stage inside the light-tight camera box with continuous exposure to 0.5% isoflurane. Image acquisition time was 10 min. Bioluminescence signals were expressed in units of photons per cm2 per second per steradian (p/cm2/s/sr). Imaging and signal quantification were controlled by the acquisition and analysis software (Living Image v. 2.50; Xenogen).

PET/CT image acquisition

Mice were imaged using a small animal PET/CT system (INVEON™; Siemens Preclinical Solutions, Knoxville, TN, USA). [18F] Fluordeoxyglucose (FDG) (7.4 MBq, 200 μCi) was injected via tail vein 1 h prior to PET/CT scanning. [18F] fluorothymidine (FLT) (same dose) was injected 2 h prior. Mice were anesthetized using 2% isoflurane. PET and CT images were acquired using small animal PET/CT scanner. The mice were moved to the PET scanner on the same bed and scanned for 30 min after CT acquisition. Tissue radioactivity was expressed as the percentage of injected radioactivity dose per gram of tissue (%ID/g). Visualization and analyses of PET images were carried out using AsiPRO™ software (Siemens Preclinical Solutions). Radioactivity concentration in the local region was calculated from the PET images using maximum pixel values.

Evaluation of fLuc expression for reverse transcription-polymerase chain reaction (RT-PCR) analysis with tissue

Total RNA was isolated from metastatic tissue and used as a template to produce cDNA using SuperScript III First-Strand Synthesis for RT-PCR (Invitrogen, Carlsbad, CA, USA). The synthesized cDNA was amplified using Taq DNA polymerase (iNtRON Biotechnology, Inc., Daejeon, Korea) with the fLuc primer: forward, 5′-CGC CTT GAT TGA CAA GGA TGG-3′, and reverse, 5′-GGC CTT TAT GAG GAT CTC TCT-3′. The forward rat GADPH primer was 5′-CAG TGC CAG CCT CGT CTC AT-3′ and the reverse primer was 5′-AGG GGC CAT CCA CAG TCT TC-3′.

Microarray analysis

Total RNA from primary tumors and IR-induced metastatic tissue for each model were used for expression profiling. Total RNA was purified using the Easy-spin Total RNA Extraction kit (iNtRON Biotechnology) according to the manufacturer's recommendations with the Agilent SurePrint G3 Rat Gene Expression 8×60K microarrays (Agilent Technologies, Inc., Santa Clara, CA, USA). The Agilent expression microarray contained ~30,003 Entrez Gene RNAs. The microarray analysis was done by Macrogen (Seoul, Korea). The arrays were scanned using the Agilent Technologies G2600D SG12494263. Array data export processing and analysis were performed using Agilent Feature Extraction software v11.0.1.1.

Results

Detection of metastatic tumors by BLI after γ-IR

The schedule for obtaining the BLI and nuclear medicine images beginning 4 weeks after γ-IR is presented in Fig. 1A. The BLI results of a longitudinal study of primary tumor growth and distant metastasis at a C6-L secondary site are shown in Fig. 1B and C. We detected metastatic tumors 6–9 weeks after γ-IR in the lungs by BLI and confirmed fLuc gene expression in the tissues (Fig. 1B and D). However, no distant metastasis was detected at the secondary site in the non-IR primary tumor (NRPT) model by BLI (Fig. 1C). Light emission of the lungs removed from sacrificed animal was examined at 9 weeks to confirm that the metastatic nodules were from the C6-L primary tumor (Fig. 1B, No. 2). This result suggests that the BLI signal from the lung originated from a γ-IR primary tumor (RPT) and was confirmed by RT-PCR (Fig. 1D). A fLuc-specific RT-PCR DNA band of 399 bp was detected in the metastatic lung nodules after γ-IR. However, survival of γ-IR treated mice was longer than that of the non-IR group because of the relatively low growth rate of the primary tumor mass after γ-IR (26).

Figure 1

Detection of metastatic tumors after γ-IR by bioluminescence imaging (BLI). (A) Schedules of γ-IR treatment and image acquisition by BLI (●) and positron emission tomography/computed tomography with [18F]FDG (■) and [18F]FLT (▲). (B) Image was acquired from 3 weeks after γ-IR, we detected metastatic tumors in chest of No. 4 and 6 mice 6 weeks after γ-IR by BLI. No. 2 mouse was confirmed by ex vivo, hut the spot was not confirm on BLI. (C) BLI images of non-IR C6-L xenografted mice. After 3–7 weeks of xenograft, no distant metastasis was detected at a secondary site in the non-IR primary tumor model. (D) fLuc gene expression in metastatic lung nodules was confirmed by RT-PCR analysis using firefly luciferase and GAPDH specific primers (fLuc, 399 bp and rat GAPDH, 595 bp as an internal control).

Confirmation of metastatic tumor after γ-IR by nuclear medicine imaging

Metastatic nodules at secondary sites in the non-IR tumor model were monitored for 6 weeks by [18F]FLT-PET (Fig. 2A), however, no metastatic nodules were detected at secondary sites. Mouse (No. 4 of Fig. 1B) underwent PET/CT 6 weeks after γ-IR and the administration of 7.4 MBq [18F]FLT and [18F]FDG to confirm the metastatic lung nodules after γ-IR detected by BLI via fLuc gene expression (Fig. 2B). [18F]FLT and [18F]FDG activity in the four metastatic lung nodules was high (Fig. 2B, white arrows). The activities were re-calculated using a region of interest analysis from a three-dimensional reconstruction encompassing the [18F]FLT and [18F]FDG uptake region. The [18F]FLT and [18F]FDG uptake values into one metastatic lung nodule was 1.5±0.13 and 2.4±0.24 %ID/g, respectively (white arrowhead). Three metastatic lung nodules (blue arrow) and one metastatic spleen nodule (red arrow) were detected by [18F]FLT-PET and [18F]FLT autoradiography in another γ-IR treated C6-L bearing mouse (No. 6 of Figs. 1B and 2C). However, no splenic metastatic nodules were detected by BLI (Fig. 2C, left panel). Metastatic nodules were detected by BLI or nuclear medicine imaging in 6 (17.14%) of the 35 C6-L bearing mice from 6 weeks after γ-IR (Table I). RNA isolated from RPT, 3 meta-static lung nodules of mouse No. 6 of Fig. 1B, and NRPT of a non-irradiated mouse was analyzed by microarray.

Figure 2

Confirmation of metastatic tumors after γ-irradiation (IR) by nuclear medicine imaging. The mice were imaged using positron emission tomography/computed tomography (PET/CT) 6 weeks after γ-IR using uptake of [18F]FDG and [18F]FLT. (A) The [18F]FLT-PET image shows no metastatic nodules from non-IR primary tumor (NRPT) at a secondary site during 6 weeks. Arrows indicate the xenografted C6-L tumor. (B) BLI, [18F]FDG and [18F]FLT PET/CT of C6-L bearing mouse (No. 4 of Fig. 1B) 6 weeks after γ-IR. The images clearly demonstrate higher uptake of [18F]FDG and [18F]FLT in four metastatic lung nodules after γ-IR (white arrow). Ex vivo indication of macroscopic observation of nodules (yellow arrows) in lung. (C) BLI and [18F]FLT PET/CT of C6-L bearing mouse (No. 6 of Fig. 1B) 6 weeks after γ-IR. [18F]FLT PET/CT and autoradiography showed high [18F]FLT uptake in metastatic nodules after γ-IR. Metastatic lung nodules were detected by [18F]FLT and autoradiography (blue arrow), and other tissue metastasis were detected in the spleen (red arrow), but not by bioluminescence.

Table I

Summary of the incidence of metastatic nodules after γ-irradiation (IR).

No.	Images	Detection time (weeks)
1	WB BLI, nuclear imaging (FDG)	8
2	WB BLI, ex vivo BLI	9
3	WB BLI, nuclear imaging (FDG)	9
4	WB BLI, nuclear imaging (FDG, FLT), ex vivo BLI	6
5	WB BLI, nuclear imaging (FLT)	7
6a	WB BLI, nuclear imaging (FLT), autoradiography	6

WB, whole body;

microarray sample mouse.

Overview of metastatic tumors after γ-IR-related gene expression

The expression patterns of whole mRNAs were analyzed by microarray to elucidate the changes in NRPT, RPT and metastatic lung nodules. A hierarchical clustering analysis of 3,881 genes (≥2-fold change, P-value <0.05) indicated differentially expressed genes between the NRPT, RPT and three metastatic lung nodules (Fig. 3A). RPT and NRPT are closely clustered together and showed a similar heat map pattern of mRNA expression, which is different from that of the meta-static lung nodules. As shown in Fig. 3B, the biological process terms differ between the RPT vs. NRPT and the metastatic lung nodules vs. RPT that reflects their known functions. The RPT enriched genes have linked biological process terms: hypoxia (3%), immune response (4%), inflammatory (3%) and signal transduction (8%). In contrast, transcription factor (4%) and glucose metabolism (2%) are linked by biological process terms for the metastatic lung nodules. Gene Ontology (GO) analysis was performed using DAVID to gain a comprehensive understanding of the gene classes that were differentially regulated in the RPT vs. NRPT and the metastatic lung nodules vs. RPT (Fig. 3C). We found upregulated expression of angiogenesis, migration, and proliferation-related genes in RPT but downregulated expression in glucose metabolism-related genes and apoptosis in the metastatic lung nodules. The molecular mechanisms and therapeutic targets underlying metastatic tumors after γ-IR remain unclear. Identifying metastatic tumors using γ-IR-related molecular target will be helpful to identify useful therapeutic targets, developing novel treatment approaches, and overcome recurrence after γ-IR in patients with glioma. We present novel insight into the EMT and enhanced stemness in RPT based on our total gene expression analysis in γ-IR tumor tissue and metastatic lung nodules of the genes of interest. Our findings are summarized in Fig. 4. In particular, expression changes in RPT with cancer stem cell markers were highly significant. For example, aldehyde dehydrogenases 1A1 and 3A1 (ALDH1A1 and ALDH3A1), which are members of the human aldehyde dehydrogenase superfamily, constitute novel candidate cancer stem cell markers in various solid tumors in the testis, brain, lens, liver, lung and retina (27). These cytoplasmic enzymes act during the oxidative stress response (28), differentiation (29) and drug resistance (30). ALDH1A1 has been reported as a novel marker for glioblastoma cells with stem cell characteristics (31) and showed the highest level change (72.18-fold higher in RPT than in NRPT) in the present study. The cancer stem cell marker CD24 was also differentially expressed (5.5-fold higher in RPT than in NRPT). In contrast to the finding on RPT and NRPT, expression of cancer stem cell markers, such as ALDH1A1, ALDH3A1, CD24, CXCL1 and IL-6 was mostly downregulated in metastatic lung nodules after γ-IR compared to RPT (Fig. 4).

Figure 3

Comparison of primary and metastatic lung nodules. (A) Hierarchical clustering map of 3,881 differentially expressed genes in a non-irradiated (IR) primary tumor (NRPT), γ-IR treated C6-L primary tumor (RPT), and three metastatic lung nodules: red, upregulated; green, downregulated. The significance threshold was 2-fold with a P-value <0.05. (B) Pie chart shows different biological functional terms, (C) differential gene expression patterns based on categorization of the upregulated and downregulated genes. Comparisons between the RPT vs. NRPT and metastatic lung nodules vs. RPT categories were performed.

Figure 4

Diagram showing the model presented between the epithelial mesenchymal transition (EMT) and cancer stem cell marker in γ-irradiated (IR) C6-L primary tumor (RPT) and metastatic lung nodules after 6 weeks of γ-IR. Metastatic lung nodules after γ-IR with the EMT and enhanced cancer stem cell markers, resulting in C6-L metastasis and recurrence. However, the expression of cancer stem cell marker was downregulated in metastatic lung nodules after γ-IR.

Discussion

In the present study, we observed distant metastasis after local γ-IR using BLI of fLuc gene expressing rat glioma and [18F]FLT and [18F]FDG-PET. Next, we observed that γ-IR, particularly fractionated local γ-IR, increased stem cell marker expression in γ-IR primary tumors by microarray. We used the γ-IR dose and schedule for C6-L tumor-bearing mice as described by Camphausen et al (32), who used Lewis lung carcinoma cells to confirm that γ-IR promotes metastasis in a mouse model.

BLI has relatively low cost and high throughput capability, but the depth dependence of the signal is a major disadvantage in small animals. The other major limitation is that BLI does not provide anatomical information. Therefore, metastatic tumors in lung and spleen after γ-IR were confirmed by small-animal PET/CT. [18F]FDG is the most widely used PET tracer and is indispensable for diagnosing and staging PET tracer for a variety of cancers. Several research groups have suggested that [18F]FLT is useful as a PET tracer to monitor proliferation and other biological response of tumors to chemotherapy and radiotherapy (33–36).

In the present study, we evaluated [18F]FDG and [18F] FLT-PET as a potential diagnostic tool for monitoring the response to metastatic tumors after γ-IR in a tumor-bearing mouse model. We found high uptake of [18F]FLT and [18F] FDG in metastatic lung nodules after γ-IR. The four nodules that were discriminated by [18F]FLT and [18F]FDG-PET were detected as one spot on BLI (Fig. 2B). The other limitation of BLI is that it does not discriminate focal signals due to spill-over. Three metastatic lung nodules and one metastatic splenic nodule in another γ-IR C6-L bearing mouse were detected by [18F]FLT-PET and autoradiography, but the splenic nodule was not detected by BLI due to a light penetration problem into deep tissue (Fig. 2C). Therefore, BLI and nuclear medicine imaging may be suitable for metastatic tumor screening after γ-IR and to more precisely locate metastatic tumors, respectively.

Recent studies have shown that a decreased cellular proliferation capacity is an early event in response to 20-Gy IR (37). We wondered whether proliferation had recovered in IR primary tumor lesions at 6 weeks γ-IR. Cancer stem cell markers were upregulated in γ-IR primary tumor lesions compared to that in non-IR primary tumors. We also found downregulation in a proportion of cancer stem cell markers in metastatic lung nodules. The proportion of cancer cell markers, particularly the ALDH family and CD24, increasing in γ-IR primary tumors may be important for distant metastasis in glioma. In particular, we revealed that upregulation of ALDH1A1 in γ-IR C6-L primary tumors may be a cancer stemness property. ALDH1A1 is a predominant isoform of the ALD family located in the cytoplasm (38) and has gained attention as a putative cancer stem cell and progenitor cell marker (39). Our data show that the small number of C6-L cells that survived in γ-IR C6-L primary tumors may have high ALDH1A1 expression, suggesting that cells surviving γ-IR are a source for distant metastasis. CD44 and CD90 have also been proposed as cancer stem-like cell markers in esophageal squamous cell carcinoma but cell heterogeneity limits their application (40,41).

We evaluated the metastatic tumors after γ-IR and found invasive/migration ability after local treatment of C6-L xeno-graft mice, suggesting that the small number of C6-L cells that survived in locally γ-IR treated tumors have more potential to metastasize, which is the main reason for recurrence of glioma after radiotherapy. The microarray study revealed more surviving cancer cells with cancer stem cell markers in the γ-IR primary tumors compared with those in the non-IR primary tumors. After formation of metastatic lung nodules in our experiments, expression of cancer stem cell markers may be downregulated as shown in our microarray data (Fig. 4). Recent studies in patients with glioma observed that the EMT may affect the ability of biomarkers to predict radio-resistant glioma (42). We observed downregulation of TGF-β and Smad5 and upregulation of MMPs, Fn1 and Snail2 in RPT compared with NRPT.

In summary, metastatic tumors were detected after fractionated γ-IR with 60Co by non-invasive longitudinal imaging and repeated measurements of the metastatic tumors after γ-IR. We demonstrated that metastatic tumors after γ-IR are associated with several genes, including the EMT and enhanced cancer stem cell markers which result in cancer cell growth, survival, invasion and proliferation.