Purpose: Redundant receptor tyrosine kinase (RTK) signaling is a mechanism for therapeutic resistance to epidermal growth factor receptor (EGFR) inhibition. A strategy to reduce parallel signaling by coexpressed RTKs is inhibition of N-linked glycosylation (NLG), an endoplasmic reticulum (ER) cotranslational protein modification required for receptor maturation and cell surface expression. We therefore investigated the feasibility of blocking NLG in vivo to reduce overexpression of RTKs.

Experimental Design: We developed a model system to dynamically monitor NLG in vitro and in vivo using bioluminescent imaging techniques. Functional imaging of NLG is accomplished with a luciferase reporter (ER-LucT) modified for endoplasmic reticulum translation and glycosylation. After in vitro validation, this reporter was integrated with D54 glioma xenografts to do noninvasive imaging of tumors, and inhibition of NLG was correlated with RTK protein levels and tumor growth.

Results: The ER-LucT reporter shows the ability to sensitively and specifically detect NLG inhibition. Using this molecular imaging approach we carried out serial imaging studies to determine safe and efficacious in vivo dosing of the GlcNAc-1-phosphotransferase inhibitor tunicamycin, which blocks N-glycan precursor biosynthesis. Molecular analyses of tunicamycin-treated tumors showed reduced levels of EGFR and Met, two RTKs overexpressed in gliomas. Furthermore, D54 and U87MG glioma xenograft tumor experiments showed significant reductions in tumor growth following NLG inhibition and radiation therapy, consistent with an enhancement in tumor radiosensitivity.

Conclusions: This study suggests that NLG inhibition is a novel therapeutic strategy for targeting EGFR and RTK signaling in both gliomas and other malignant tumors. Clin Cancer Res; 16(12); 3205–14. ©2010 AACR.

Translational Relevance

Identifying novel targets for cancer therapy requires that preclinical models show the following: (a) that the new target can be blocked in vivo, (b) that target blockade has mechanistic consequences for tumor cells, and (c) that target blockade has efficacy. We investigated a new target for enhancing tumor radiosensitvity, i.e., N-linked glycosylation (NLG), by using molecular imaging to generate a tumor model that permits analysis of the above criteria. We propose that inhibition of NLG is an alternative strategy to reduce receptor tyrosine kinase (RTK) expression in tumors and enhance radiation therapy, and is advantageous by reducing redundant and compensatory RTK survival signals. By using this preclinical model we show for the first time that inhibition of NLG: (a) is feasible in vivo, (b) reduces epidermal growth factor receptor and Met RTKs in tumors, and (c) significantly reduces tumor growth when delivered with radiation therapy. This work provides direct evidence that inhibition of NLG is a promising approach to enhance current radiation therapy treatment regimens.

Previous work has shown that inhibition of N-linked glycosylation (NLG), a sequence-specific, cotranslational modification that occurs in the endoplasmic reticulum (ER), reduces protein levels of overexpressed receptor tyrosine kinases (RTK), namely, epidermal growth factor receptor (EGFR), ErbB2, Erbb3, and insulin-like growth factor I receptor (IGF-1R) in vitro (1). As a result, signaling through both dominant and redundant RTK signaling pathways is reduced, suggesting that inhibition of NLG is an alternative mechanistic approach for targeting RTK signaling in tumors. Like other strategies for targeting multiple RTKs in glioblastoma (25), NLG inhibition produced marked radiosensitization in cancer cell lines but did not radiosensitize nontransformed cells. Although in vitro experiments suggest the potential for a therapeutic ratio, further evaluation of whether this biosynthetic process can be blocked at tolerable levels requires an in vivo experimental model.

A major barrier for evaluating novel molecular targets in cancer therapy has been the inability to measure target activity and to assess inhibition by pharmacologic agents in vivo. Xenograft tumor models, which have been used to measure tumor growth, display heterogeneity and are not optimal for analyses done at the molecular level. To address the deficiencies of the xenograft tumor model, techniques for molecular imaging using engineered reporter genes in small animals have been developed (see ref. 6 for review). The underlying principle of these techniques is based upon unique modifications to genes such as firefly luciferase (Luc), that transform the reporter into a specific molecular sensor that transmits a quantifiable signal to a detection system. In tumor xenografts, Luc was initially employed as a dynamic marker for tumor growth and metastases, but was quickly modified and adapted in approaches to report specific cellular events such as apoptosis (7) or endoplasmic reticulum–induced stress (8, 9). Luc has also been used in vivo to measure changes at the protein level, such as the turnover rate of CDK2-dependent p27 levels (9). Luker et. al. pioneered a split Luc design (or luciferase complementation) to measure in vivo protein-protein interactions for CDC25c and 14-3-3 among others (10), a design recently used to show radiation-induced EGFR signaling (11). Posttranslational modifications, such as site-specific Akt phosphorylation (12), have also been successfully measured using Luc-based reporter vectors, showing the flexibility of Luc as a tool for noninvasive imaging strategies.

We hypothesized that in vivo imaging of protein NLG would be an invaluable model system to investigate the feasibility of blocking this process as a potential cancer therapy. Because our aim was to radiosensitize tumor cells with this maneuver, unlike most cancer therapies that are intended to eliminate all viable tumor cells, we sought to develop a sensitive model system to determine a threshold for NLG inhibition in tumors, to perform serial noninvasive measurements of this biological process, and to guide dose and schedule optimization for in vivo experiments.

We now report on a bioluminescent reporter engineered to exploit the sequence specificity of NLG and the biophysical properties of Luc to measure inhibition of NLG. We have validated this reporter in vitro and used it in a glioma xenograft tumor model to show the feasibility of NLG inhibition in animals. This noninvasive imaging platform became instrumental as a method for testing the hypothesis that inhibition of NLG in vivo reduces RTK protein levels and radiosensitizes tumors. Our results show the power of molecular imaging techniques to efficiently evaluate novel targets for cancer therapy (in this case NLG) as well as the crucial role this knowledge plays in developing new strategies for multimodality therapy.

Reagents

Unless otherwise stated, all reagents were purchased from Invitrogen. Antibodies were purchased from Chemicon (luciferase), Santa Cruz Biotechnology (c-Met), and Invitrogen (EGFR). Tunicamycin and the Concanavalin A–agarose conjugate were purchased from Calbiochem. Luciferin was supplied by Promega.

Vectors

The 24-amino-acid EGFR endoplasmic reticulum translation sequence was added in frame to the NH2-terminus of the luc gene by sequential PCR reactions. The amplification product from primer 1 (5′CTGGCGCTGCTGGCTGCGCTCTGCCCGGCCTCGAGAGCTATGGAAGACGCCAAAAAC 3′) and primer 2 (5′ ACGCGTCGACTTACACGGCGATCTTTCCGCCCTTCTTGGC 3′) was further amplified using primer 3 (CGGATCCACCATGCGACCCTCCGGGACGGCCGGGGCAGCGCTCCTGGCGCTGCTGGCTGC) and primer 2. The resultant PCR product was digested with BamHI and SalI and cloned into BamHI and XhoI sites of pcDNA3. The cleavage site of the endoplasmic reticulum translation sequence was engineered to precede the first Luc methionine so that following cleavage of the leader sequence in the endoplasmic reticulum, the Luc amino acid sequence would be unchanged. The ER-LucT sequence contains three (T) potential glycosylation sites and was amplified from the original pyralis luciferase sequence as found in PUHD10-3. The ER-LucS sequence contains a single (S) potential glycosylation site and was amplified from pGL3 (Promega), which is a modified luciferase with elimination of two of the three potential glycosylation sites inherent in wild-type Luc. Luc is not normally glycosylated because it does not enter the endoplasmic reticulum lumen. The Luc-T was constructed by removal of the endoplasmic reticulum translation sequence through digestion of ER-LucT with ApaI and XhoI (a restriction site we nested in the endoplasmic reticulum translation sequence). Vectors were sequenced to insure that no mutations were introduced during the PCR reaction.

Cell culture

Cell lines were maintained in RPMI 1640 media supplemented with 10% FCS and 100 ug/mL each of penicillin and streptomycin at 37°C with 5% CO2. 293T cells were used for transient transfection experiments. Chinese hamster ovary (CHO) and D54 cell lines with stable expression of the luciferase plasmids were generated through transfection with Lipofectamine and selection with G418 (800 ug/mL and 200 ug/mL), respectively.

Immunoblot

Lysates were prepared using Western lysis buffer [25 mmol/L Tris, 10 mmol/L EDTA, 15% glycerol, 0.1% Triton X-100, 1× protease inhibitor cocktail (Roche), and 1× phosphatase inhibitor cocktails 1 and 2 (Sigma)]. For tumor lysates, 2% SDS was used as a detergent and inhibitors were used at a 2.5× final concentration. Glycosylated proteins were removed from protein lysates by precipitation with Concanavalvin A–agarose for 60 minutes with gentle agitation at 4°C followed by centrifugation at 5,000 rpm. Supernatants from these samples were used for Western blot analysis.

In vitro bioluminescence

Cell lines with stable expression of luciferase plasmids were plated at a density of 40,000 cells/well in a 24-well dish. Inhibitors were added the following day, and imaging was performed 24 hours after drug treatment. Before bioluminescent imaging, the media were aspirated and replaced with fresh media containing 100 ug/mL luciferin. Photon counts were then collected 10 minutes after the addition of luciferin using the Xenogen IVIS charge-coupled device camera system. A signal averaging time of 1 minute was used for all experiments, and signal intensity was quantified as the sum of all detected photons per well. A pseudocolor image representing the detected photons was produced as an overlay on a gray-scale image of the plate. Triplicate wells for each treatment condition were analyzed for each experiment, and untreated wells were used as baseline controls to calculate induction of bioluminescence.

In vivo bioluminescence

All mouse experiments were approved by the University Committee on the Use and Care of Animals of the University of Michigan. The protocol for bioluminescent imaging of mice bearing D54-ERlucT flank tumors has been previously reported (7). Briefly, tumors were grown in athymic nude mice (Charles River) by bilateral s.c. implantation of 1 × 107 cells. Ten days following injection, mice bearing palpable tumors were anesthetized with a 1% isoflurane/air mixture and given a single i.p. dose of 150 mg/kg luciferin in normal saline. Bioluminescent imaging was done from 5 to 20 minutes after luciferin administration, and mice were anesthetized and kept warm with a temperature-controlled bed during image acquisition. Signal intensity was quantified for a region of interest for each tumor over the imaging time period to determine the peak of bioluminescent activity. Tumor bioluminescence before drug treatment was used to establish a baseline of activity and to calculate induction of Luc activity. After obtaining baseline images, mice were treated with i.p. tunicamycin (0.25-1 mg/kg) or 150 mmol/L dextrose as a control and underwent repeated daily imaging. Tunicamycin was prepared by dissolving the compound in DMSO to give a final concentration of 5 mg/mL and then diluted 1:50 in 150 mmol/L dextrose.

Immunohistochemistry and immunofluorescence

To evaluate EGFR protein expression, immunohistochemistry was done using a standard protocol [DAKO EnVision +System, peroxidase (3,3 '-diaminobenzidine), K4011]. A 5-μm-thick paraffin-embedded section was dewaxed and hydrated in xylene and ethanol, respectively. Antigen retrieval was performed in citrate buffer in microwave at pH 6.0. The primary antibody was a rabbit polyclonal antibody to EGFR (Santa Cruz) which was added to the tumor sections at a 1:70 dilution and incubated for 45 minutes at room temperature. The section was then treated with a horseradish peroxidase–labeled secondary antibody for 30 minutes, followed by peroxide/diaminobenzidine substrate/chromagen. The slides were counterstained with hematoxylin. For quantitation of EGFR distribution in control and treated samples, random fields were imaged in sections from three animals in each group and the number of cells with membrane, cytosolic, or no staining were counted. For Met immunofluorescence staining, deparaffinization, and antigen retrieval were similar as described above. Met antibody was incubated at dilution of 1:100 followed by cy3-coupled secondary antibody (1:200; Jackson Immunoresearch). The slides were then counterstained with 1 μg/mL 4′,6-diamidino-2-phenylindole, mounted, and visualized under a fluorescence Nikon Eclipse TE2000-U microscope (Nikon). All fluorescence images were acquired and quantitated using Metamorph software (Molecular Devices Corporation).

Tumor growth

D54 and U87MG xenograft tumors were generated by inoculation of 1 × 107 cells into the flanks of athymic nude mice. When tumors became palpable, four to six mice were randomized to each of four groups (untreated control; 0.75 mg/kg tunicamycin; radiation; radiation plus 0.75 mg/kg tunicamycin). Median tumor volumes were 477 mm3 for D54 tumors and 359 mm3 for U87MG tumors, with no significant differences in tumor size between groups. Tunicamycin treatment consisted of a single i.p. injection on the day of randomization (day 0). For those animals randomized to receive radiation, therapy began on day 1 and consisted of five daily fractions of 2 Gy, delivered at a dose rate of 2 Gy/minute using a Pantak DXT 300 Orthovoltage Unit producing 300 kV X-rays. Lead shielding was used to reduce animal exposure. The treatment schedule was designed both to optimize NLG inhibition during radiation therapy (RT) and to be similar to clinical delivery of RT with a weekly systemic agent. Tumor dimensions were measured three times weekly, and tumor volume was calculated according to the equation: TV = (π/6) × (L × W2), where L and W are the longer and shorter dimensions of the tumor, respectively. Data are expressed as the ratio of tumor volume at various times after treatment compared with the day of drug administration.

Statistics

Data points are reported as experimental averages and error bars represent SE. Statistical significance was determined using a two-sided Student's t test. A P value <0.05 was considered to be statistically significant.

NLG reporter design

NLG is a cotranslational protein modification that occurs in the endoplasmic reticulum. Because of our interest in disrupting this biological process as a means of sensitizing cancer cells to cytotoxic therapies, we sought to develop a simple method for measuring NLG. We had previously used firefly Luc as a bioluminescent reporter for both in vitro and in vivo cancer models (7), and therefore we chose this gene to develop a way to monitor inhibition of NLG in living cells. We hypothesized that the cotranslational transfer of the 14 sugar glycan precursor from its lipid precursor to the glycosylation sites of Luc in the endoplasmic reticulum would disrupt the ability of this enzyme to use luciferin and ATP as substrates to generate bioluminescence. The design predicts that a Luc enzyme translocated into the endoplasmic reticulum and engineered to contain consensus NLG sites (NXS/T) would have low baseline levels of bioluminescent activity, but that inhibition of NLG (and loss of the added glycan moiety) would therefore enhance bioluminescent activity. Furthermore, the design accounts for the possibility of reduced signal secondary to the cellular effects of NLG inhibition. Although aberrant glycosylation has the potential to reduce protein translation, the “gain of function” premise of this reporter enhances the sensitivity for detecting inhibition of this biosynthetic process.

To test this design, luciferase cDNAs were constructed with an in-frame endoplasmic reticulum translation leader sequence from the EGFR and expressed in 293T cells. The reporters are designated ER-LucT and ER-LucS, and contain three consensus NLG sites or a single consensus NLG site, respectively (Fig. 1A). A control, cytoplasmic expressed luciferase (LucT) was also prepared. The endoplasmic reticulum signal peptide is cleaved just before the initiation methionine of the wild-type Luc, yielding a protein identical to that of a cytoplasmic Luc (Fig. 1B). As predicted, expression of each ER-Luc vector increased the molecular weight of the Luc protein, consistent with its glycosylation (Fig. 1C). The increase in molecular weight for each endoplasmic reticulum–expressed construct was proportional to the number of glycosylation sites in each protein sequence, showing that ER-LucT is more heavily glycosylated than ER-LucS. Pretreatment of CHO cells expressing either the ER-LucT or ER-LucS construct with the GlcNAc-1-phosphotransferase inhibitor tunicamycin reversed this increase in molecular weight, consistent with blockade of the glycosylation of the protein. In comparison, the cytoplasmic-expressed LucT construct maintained identical gel mobility after tunicamycin treatment.

Fig. 1.

Schematic diagram illustrating the NLG reporter design (A, B). The endoplasmic reticulum (ER) translation sequence of EGFR is fused to the NH2-terminus of the Luc enzyme. Constructs contain either three (ER-LucT) sites for NLG or a single (ER-LucS) site. C, constructs were transiently expressed in 293T cells and lysates of samples treated with or without 500 nmol/L tunicamycin (Tn) were analyzed by Western blotting.

Fig. 1.

Schematic diagram illustrating the NLG reporter design (A, B). The endoplasmic reticulum (ER) translation sequence of EGFR is fused to the NH2-terminus of the Luc enzyme. Constructs contain either three (ER-LucT) sites for NLG or a single (ER-LucS) site. C, constructs were transiently expressed in 293T cells and lysates of samples treated with or without 500 nmol/L tunicamycin (Tn) were analyzed by Western blotting.

Close modal

NLG reporter activity

Functional studies to quantify the effect of glycosylation on Luc activity were also undertaken. To quantitate changes in bioluminescence, we generated pooled clones of CHO cells with stable expression of each vector. Both ER-LucS– and ER-LucT–expressing CHO cells showed increased bioluminescence when treated with tunicamycin (Fig. 2A). Because each N-linked glycan has the potential to disrupt Luc enzymatic activity, we hypothesized that the ER-LucT construct would display greater sensitivity to NLG inhibition with tunicamycin. This was confirmed in the CHO cell clones, where ER-LucT showed a 4.2 ± 0.1–fold induction compared with the 2.6 ± 0.1–fold induction measured for ER-LucS. In contrast, control experiments with CHO cells expressing the cytoplasmic Luc (LucT) did not show an increase in bioluminescence following tunicamycin treatment (Fig. 2B). These results show that endoplasmic reticulum–translated Luc, but not cytoplasmic Luc, can be used to measure inhibition of NLG by tunicamycin. They also suggest that due to the presence of multiple NLG sites, ER-LucT is a superior design for measuring NLG inhibition.

Fig. 2.

Validation of the NLG reporter. A, dose response induction of Luc activity by tunicamycin in pooled CHO clones that have stable expression of either the ER-LucS (top) or ER-LucT (bottom) constructs. B, effect of tunicamycin on CHO cell clones expressing cytoplasmic Luc (LucT). Activity is reported as fold increase of relative light units (RLU) of experimental to control (C) samples. C, induction of Luc activity in an ER-LucT cell line clone with correlation to Luc glycosylation (inset). D, in vitro IVIS imaging of luminescence for 20 μmol/L castanospermine (Cs), 20 μmol/L swainsonine (Sw), and 500 nmol/L tunicamycin (Tn). Results represent the average and SE of three independent experiments performed in triplicate.

Fig. 2.

Validation of the NLG reporter. A, dose response induction of Luc activity by tunicamycin in pooled CHO clones that have stable expression of either the ER-LucS (top) or ER-LucT (bottom) constructs. B, effect of tunicamycin on CHO cell clones expressing cytoplasmic Luc (LucT). Activity is reported as fold increase of relative light units (RLU) of experimental to control (C) samples. C, induction of Luc activity in an ER-LucT cell line clone with correlation to Luc glycosylation (inset). D, in vitro IVIS imaging of luminescence for 20 μmol/L castanospermine (Cs), 20 μmol/L swainsonine (Sw), and 500 nmol/L tunicamycin (Tn). Results represent the average and SE of three independent experiments performed in triplicate.

Close modal

To further investigate the capability of ER-LucT to serve as a sensitive reporter for NLG, we isolated an ER-Luc-T CHO clone by serial dilution. This clone displayed dose-response activity to tunicamycin treatment over several orders of magnitude (10 nmol/L-10 umol/L) and an 11.3 ± 0.4–fold maximal induction of Luc activity (Fig. 2C). Western blot analysis of protein lysates from tunicamycin-treated cultures showed that luminescence measurements directly corresponded to the glycosylation status of the protein (Fig. 2C, inset). These experiments confirm that glycosylation of Luc in the endoplasmic reticulum is a sensitive mechanism to measure the state of this cotranslational event in living cells.

The ER-LucT reporter detects inhibition of enzymatic steps that precede transfer of the glycan chain to the consensus NLG site or those that contribute to glycan synthesis. The goal of imaging this biosynthetic process is based upon previous findings that disrupting core NLG, but not later NLG processing steps, sensitizes tumor cells to other cytotoxic therapies including ionizing radiation (1). To test the specificity of the ER-LucT reporter for discriminating between protein NLG and N-linked glycan processing, we compared the effects of tunicamycin with those of castanospermine and swainsonine, compounds that block processing of N-linked glycans by inhibiting the α-glucosidases and α-mannosidase-II, respectively (Fig. 2D). Neither of these inhibitors enhanced Luc activity, confirming that only inhibition of NLG can induce bioluminescent activity of the reporter.

Imaging NLG in vivo

Tunicamycin is the only known inhibitor of the lipid-linked oligosaccharide precursors of N-linked glycans, and little is known regarding its in vivo pharmacology. Toxicity studies using i.p. administration of tunicamycin in mice have estimated LD1 and LD50 doses to be approximately 1 mg/kg and 1.5 mg/kg respectively (13). However, these studies have not answered the fundamental question of whether a low dose of tunicamycin can disrupt NLG, but remain tolerable for the animal. Furthermore, it is unknown whether systemic delivery of tunicamycin would reach the tumor, as it could be metabolized before entering the arterial blood supply or it could be prevented from perfusing the tumor and reaching the target tumor cells. To answer these questions we integrated the ER-LucT NLG reporter into a glioma xenograft tumor model with the goal of determining whether systemic delivery of tunicamycin could impair NLG in xenograft tumors. D54 glioma cells with stable expression of the ER-LucT reporter were generated for these in vivo experiments. Like the CHO cell line, Luc activity in the D54-ER-LucT cells could be induced ∼10-fold with tunicamycin in vitro (data not shown). After s.c. inoculation, this cell line produced flank tumors with a 100% take rate and a low background level of bioluminescence (Fig. 3A).

Fig. 3.

In vivo imaging of NLG in D54 xenograft tumors. A, induction of bioluminescence by i.p. administration of 1 mg/kg tunicamycin. Images are shown for peak induction of bioluminescence at 48 hours for control and experimentally treated animals. B, dose response induction of bioluminescence by tunicamycin for i.p. treatment with 0, 0.25, 0.5, 0.75, and 1 mg/kg measured at 48 hours. Data represent the average and SE of three animals (six tumors) per group. C, time course for tunicamycin-induced luminescence for mice treated with 0.75 mg/kg followed by daily imaging. Data represent the average and SE for four animals (eight tumors) for each group.

Fig. 3.

In vivo imaging of NLG in D54 xenograft tumors. A, induction of bioluminescence by i.p. administration of 1 mg/kg tunicamycin. Images are shown for peak induction of bioluminescence at 48 hours for control and experimentally treated animals. B, dose response induction of bioluminescence by tunicamycin for i.p. treatment with 0, 0.25, 0.5, 0.75, and 1 mg/kg measured at 48 hours. Data represent the average and SE of three animals (six tumors) per group. C, time course for tunicamycin-induced luminescence for mice treated with 0.75 mg/kg followed by daily imaging. Data represent the average and SE for four animals (eight tumors) for each group.

Close modal

To assess the ability of tunicamycin to activate the reporter in vivo, we carried out dose response experiments by delivering single injections of 0, 0.25, 0.5, 0.75, and 1 mg/kg. We found a significant enhancement of bioluminescence for doses of tunicamycin ≥0.5 mg/kg (Fig. 3B), and a 7-fold induction at 0.75 mg/kg. To minimize potential toxicity, we selected this as a standard dose for subsequent experiments. To estimate the duration of NLG inhibition, we also carried out daily imaging on the mice after tunicamycin injection to establish the time course of NLG inhibition in vivo (Fig. 3C). We found that a single dose of 0.75 mg/kg tunicamycin had sustained effects on NLG for several days, with peak luminescence occurring 48 to 72 hours following treatment. In comparison, control-treated animals did not show enhanced luminescence over this time period.

To confirm that in vivo bioluminescence correlated with the glycosylation status of Luc, an analysis of tumor lysates was also done (Fig. 4A). We were unable to resolve the glycosylated and nonglycosylated forms of Luc from tumor samples by Western blot (data not shown). Therefore Concanavalin A–agarose, a lectin that binds to high-mannose and hybrid-type N-linked glycans, was used to separate glycosylated Luc from the sample before Western blotting. We found that tunicamycin-treated animals had tumors with nonglycosylated Luc but control-treated tumors did not. In summary, low, tolerable doses of tunicamycin can reduce NLG in tumor cells and effect the function of target proteins (i.e., luciferase) for up to 96 hours after i.p. administration.

Fig. 4.

Ex vivo analysis of NLG inhibition in D54 tumors. A, inhibition of Luc glycosylation in tumors. Glycosylated forms of Luc were separated by Concanavalin A (ConA)-agaraose precipitation and lysates were analyzed by Western blot for in vitro samples (left) and tumor samples (right). B, EGFR staining by immunohistochemistry in representative control and Tn-treated tumors. Percent membrane staining with the SE is also reported. C, Met fluorescence by immunofluorescence in representative control- (C) and Tn-treated tumors. Fluorescence intensity values and SE for each treatment are also reported.

Fig. 4.

Ex vivo analysis of NLG inhibition in D54 tumors. A, inhibition of Luc glycosylation in tumors. Glycosylated forms of Luc were separated by Concanavalin A (ConA)-agaraose precipitation and lysates were analyzed by Western blot for in vitro samples (left) and tumor samples (right). B, EGFR staining by immunohistochemistry in representative control and Tn-treated tumors. Percent membrane staining with the SE is also reported. C, Met fluorescence by immunofluorescence in representative control- (C) and Tn-treated tumors. Fluorescence intensity values and SE for each treatment are also reported.

Close modal

Inhibition of NLG reduces RTK levels in vivo

Based upon our prior in vitro experiments showing that NLG inhibition reduces RTK protein levels (1), we next investigated whether inhibition of NLG would reduce RTK protein levels in xenograft tumor cells. We carried out immunohistochemistry on tumor specimens treated with tunicamycin to evaluate the effects on EGFR and Met, two receptors overexpressed in wild-type D54 tumors. We found that inhibition of NLG reduced protein levels of both receptors in tumors from tunicamycin-treated animals as compared with controls. For EGFR (Fig. 4B), characteristic membrane staining was apparent in 80 ± 9% of cells from control tumors. In contrast, tumors from tunicamycin-treated animals had a marked decrease of EGFR protein levels, with only 18 ± 9% of cells showing membrane EGFR staining. A similar reduction in receptor protein levels was also observed following immunofluorescence for Met (Fig. 4C). Tunicamycin-treated tumors displayed a 50 ± 2% reduction in fluorescence intensity and membrane staining as compared with control animals. These results confirm that tunicamycin inhibits NLG in vivo, which reduces RTK protein levels in tumor cells.

Xenograft tumor radiosensitization through NLG inhibition

To test the hypothesis that inhibition of NLG in vivo can sensitize tumors to radiation, we designed tumor growth delay experiments based upon data obtained from characterizing the dose response and time course for i.p. administration of tunicamycin. Based on these data, we selected an experimental treatment regimen for D54 or U87MG glioma xenografts of 0.75 mg/kg tunicamycin followed by a 5-day fractionated course of radiation (Fig. 5A and B). We found that tunicamycin treatment alone had only a small effect on tumor volume tripling (TVT) in D54 tumors (8 ± 1 versus 11 ± 1 days) and no measurable effect in U87MG tumors (14 ± 2 versus 15 ± 2 days). Radiation alone, delivered in 2 Gy daily fractions for a total of 10 Gy, reduced tumor growth significantly in both cell lines with TVT of 15 ± 2 days for D54 and 28 ± 2 days for U87MG (P < 0.05 for both). However, the greatest delay in tumor growth was achieved with combined radiation and tunicamycin treatment, which significantly extended TVT time to 24 ± 2 and 37 ± 2 days, respectively (P < 0.05 when compared with RT-only group). We conclude that low doses of tunicamycin rationally selected by using in vivo imaging can radiosensitize glioma tumor xenografts.

Fig. 5.

Tumor growth experiments. Mice bearing wild-type D54 (A) and U87-MG (B) xenograft tumors were randomized to four treatment groups: C, control; Tn, 0.75 mg/kg tunicamycin; RT, radiation, 5 daily fractions of 2 Gy; and RT + Tn, radiation plus tunicamycin, and were followed with tumor measurements. Data points, relative tumor growth compared with the tumor volume on day 0; error bars, SE.

Fig. 5.

Tumor growth experiments. Mice bearing wild-type D54 (A) and U87-MG (B) xenograft tumors were randomized to four treatment groups: C, control; Tn, 0.75 mg/kg tunicamycin; RT, radiation, 5 daily fractions of 2 Gy; and RT + Tn, radiation plus tunicamycin, and were followed with tumor measurements. Data points, relative tumor growth compared with the tumor volume on day 0; error bars, SE.

Close modal

NLG is a complex biosynthetic process that regulates maturation of proteins through the secretory pathway. This cotranslational modification is regulated by a series of enzymatic reactions, culminates in the transfer of a core glycan from the lipid carrier to a protein substrate, and has been identified as a target to enhance cytotoxic cancer therapies (1). To determine NLG activity in vivo, we designed a noninvasive imaging technique with the goal of examining this molecular event in a sensitive and specific manner. To this end we developed and validated the ER-LucT, a bioluminescent NLG reporter, and used it as a surrogate for measuring inhibition of NLG activity. In vitro, induction of reporter activity was achieved with nanomolar concentrations of tunicamycin, was dependent upon endoplasmic reticulum localization and the number of glycosylation sites in the protein, was specific for inhibiting the addition (not processing) of N-linked glycans, and corresponded directly to the glycosylation state of the protein. Using the ER-LucT reporter, we developed a glioma xenograft tumor model to carry out noninvasive, serial imaging of NLG in tumor cells. We found that tolerable doses of tunicamycin (0.5-1 mg/kg) significantly activate the NLG reporter when given by a single i.p. administration and have prolonged inhibitory effects, with quantifiable changes in bioluminescence for up to four days. Analysis of RTK protein levels of EGFR and Met confirmed that inhibition of NLG in vivo is an alternate strategy for reducing overexpressed RTKs in gliomas. Using these in vivo data, we then tested the hypothesis that inhibition of NLG sensitizes tumors to ionizing radiation and found that this treatment significantly delayed tumor growth. In summary, we present evidence that in principle NLG can be targeted in vivo to enhance the radiation sensitivity of tumor cells, and these results have the strength of correlating real-time in vivo monitoring of target inhibition with improvements in tumor control.

N-linked glycosylation, like receptor phosphorylation, plays a major role in modulating protein function. Although it is now understood that RTK phosphorylation is the underlying molecular mechanism for proproliferative and antiapoptotic signal transduction in cancer cells, the contributions of NLG to protein function are just beginning to be understood. NLG alters protein conformation and folding (14), is involved in membrane receptor trafficking (15), modulates receptor function (16), and has been implicated in tumor cell migration and invasion (17). For EGFR, which has been a successful molecular target in cancer therapy (18, 19), elucidating the contribution of this cotranslational modification to the biochemical and biophysical properties of the receptor may provide valuable information for potential therapeutic interventions. Although EGFR has been successfully targeted by small molecule tyrosine kinase inhibitors and receptor-specific antibodies, therapeutic resistance to EGFR-targeted therapies ultimately develops (20), and similar redundant RTK signaling cascades have also been implicated in radioresistance (21, 22). In this study we found that disruption of NLG in vivo is an alternative method for reducing protective RTK signaling and increasing the radiosensitivity of gliomas. In fact, NLG inhibition has the advantage of reducing protein levels of both EGFR and other overexpressed RTKs such as Met, a kinase implicated in resistance to EGFR-targeted therapies (23, 24). This common effect on overexpressed RTKs also suggests that targeting NLG may combine favorably with other classes of EGFR inhibitors to reduce both oncogenic signaling and the mechanisms of therapeutic resistance. Although the concept for NLG inhibition was conceived to reduce redundant RTK signaling, we do not exclude the possibility that NLG inhibition may have other antitumor effects through the functional disruption of other membrane proteins or activation of endoplasmic reticulum signaling.

NLG has been studied in the context of congenital disorders of glycosylation, a group of disorders characterized by neuromuscular abnormalities and developmental delay and defined by reduced activity of enzymes required for glycosylation. Additionally, three mouse models of genes involved in N-linked glycan synthesis – GlcNAc-1-phosphotransferase (GPT/ALG7), phosphomannose isomerase (PMI), and phosphomannomutase (PMM2) – show embryonic lethality following knockout of each respective gene (2527). These clinical and experimental data suggest that NLG is an essential cellular process, an important concept for cancer therapeutics as the viability of targeting essential cellular functions has been questioned (28). However, clinical and animal data for congenital disorders of glycosylation also show that low levels of PMI or PMM2 enzymatic activity (5-25% of normal) are not lethal. In fact, patients or animals with a recessive mutation (and ∼50% activity) have virtually no discernable phenotype. Thus, it seems that mammalian cells have a threshold for loss of enzymatic activity that is required before deficits in NLG biosynthesis become evident. This threshold is also likely to be different between tissue types and to vary with proliferation status, as the requirements for cell surface protein production during cell cycle progression and cell division are increased.

NLG inhibition may produce systemic side effects, however, in this combined modality approach, the goal of drug therapy is not direct cytotoxicity, but rather the enhancement of the effects of focal radiation therapy. This strategy, which depends upon reaching the threshold for NLG inhibition, would employ lower concentrations of an inhibitor and would be less likely to produce systemic toxicity. But how could these NLG thresholds be determined in human tumors? One possibility is based upon the work of several groups attempting to develop radioactive tracers for EGFR imaging. Using the specificity of EGFR tyrosine kinase inhibitors or EGFR antibodies, these techniques, which are based on positron emission tomography and single-photon emission computed tomography, have shown promise in human clinical trials (29, 30). Because EGFR levels are reduced by NLG inhibition, an EGFR imaging technique could be used as a surrogate for determining dosing and the efficacy of NLG inhibitors and to facilitate phase I clinical investigations.

Tunicamycin itself is not a potential therapeutic agent due to its narrow window of efficacy (13). However, this compound proved to be an excellent experimental tool to inhibit NLG in vitro and in vivo and to test our hypothesis. It allowed us to show three principles crucial for evaluating this potential therapeutic approach: (a) that we could hit the target in vivo and reduce NLG in tumor cells, (b) that hitting the target had consequences for RTK expression levels, and (c) that disrupting NLG enhanced tumor cell radiosensitivity. Although tunicamycin is the only known inhibitor of lipid-linked oligosaccharide precursor synthesis, our data suggest that a compound with similar biological effects, but a broader therapeutic window and potentially a superior toxicity profile, could be useful for radiosensitization. In this context we are currently evaluating the NLG reporter as a tool to identify novel NLG inhibitors. Secondary to its gain of function design, the NLG reporter provides a robust signal over several orders of magnitude and has been optimized for a high-throughput screening format. Paired with the in vivo capabilities shown herein, the NLG reporter has the potential to streamline in vitro/in vivo drug discovery for NLG inhibitors.

In summary, this work describes a novel bioluminescent imaging technique to measure changes in NLG in living cells. This technique provides a real-time method to quantify changes in this cotranslational modification in vitro as well as in xenograft tumors. We have used this model system to investigate the feasibility of targeting NLG in vivo, and we anticipate the ability to image NLG will have applications for identifying novel NLG inhibitors as well as for developing a better understanding of NLG deficiency syndromes (i.e., congenital disorders of glycosylation) and other factors that affect glycan precursor biosynthesis.

No potential conflicts of interest were disclosed.

We thank the members of the University of Michigan Center for Molecular Imaging, University of Michigan Comprehensive Cancer Center Research Histology and Immunoperoxidase Lab, and the University of Michigan Microscopy and Imaging Analysis Laboratory for assistance with experimental procedures.

Grant Support: The RSNA Resident Research Grant, ASTRO Junior Faculty Career Research Training Award (J.N. Contessa), P50 CA93990 (A. Rehemtulla, B. Ross), R01 CA78554 (T.S. Lawrence), R01 DK55615 (H. Freeze).

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.

1
Contessa
JN
,
Bhojani
MS
,
Freeze
HH
,
Rehemtulla
A
,
Lawrence
TS
. 
Inhibition of N-linked glycosylation disrupts receptor tyrosine kinase signaling in tumor cells
.
Cancer Res
2008
;
68
:
3803
9
.
2
Damiano
V
,
Melisi
D
,
Bianco
C
, et al
. 
Cooperative antitumor effect of multitargeted kinase inhibitor ZD6474 and ionizing radiation in glioblastoma
.
Clin Cancer Res
2005
;
11
:
5639
44
.
3
Lammering
G
,
Valerie
K
,
Lin
PS
, et al
. 
Radiosensitization of malignant glioma cells through overexpression of dominant-negative epidermal growth factor receptor
.
Clin Cancer Res
2001
;
7
:
682
90
.
4
Chakravarti
A
,
Loeffler
JS
,
Dyson
NJ
. 
Insulin-like growth factor receptor I mediates resistance to anti-epidermal growth factor receptor therapy in primary human glioblastoma cells through continued activation of phosphoinositide 3-kinase signaling
.
Cancer Res
2002
;
62
:
200
7
.
5
Dote
H
,
Cerna
D
,
Burgan
WE
,
Camphausen
K
,
Tofilon
PJ
. 
ErbB3 expression predicts tumor cell radiosensitization induced by Hsp90 inhibition
.
Cancer Res
2005
;
65
:
6967
75
.
6
Gross
S
,
Piwnica-Worms
D
. 
Spying on cancer: molecular imaging in vivo with genetically encoded reporters
.
Cancer Cell
2005
;
7
:
5
15
.
7
Laxman
B
,
Hall
DE
,
Bhojani
MS
, et al
. 
Noninvasive real-time imaging of apoptosis
.
Proc Natl Acad Sci U S A
2002
;
99
:
16551
5
.
8
Feldman
DE
,
Chauhan
V
,
Koong
AC
. 
The unfolded protein response: a novel component of the hypoxic stress response in tumors
.
Mol Cancer Res
2005
;
3
:
597
605
.
9
Zhang
GJ
,
Safran
M
,
Wei
WY
, et al
. 
Bioluminescent imaging of Cdk2 inhibition in vivo
.
Nat Med
2004
;
10
:
643
8
.
10
Luker
KE
,
Smith
MCP
,
Luker
GD
,
Gammon
ST
,
Piwnica-Worms
H
,
Piwnica-Worms
D
. 
Kinetics of regulated protein-protein interactions revealed with firefly luciferase complementation imaging in cells and living animals
.
Proc Natl Acad Sci U S A
2004
;
101
:
12288
93
.
11
Li
WR
,
Li
F
,
Huang
Q
,
Frederick
B
,
Bao
S
,
Li
CY
. 
Noninvasive imaging and quantification of epidermal growth factor receptor kinase activation in vivo
.
Cancer Res
2008
;
68
:
4990
7
.
12
Zhang
LM
,
Lee
KC
,
Bhojani
MS
, et al
. 
Molecular imaging of Akt kinase activity
.
Nat Med
2007
;
13
:
1114
9
.
13
Morin
MJ
,
Bernacki
RJ
. 
Biochemical effects and therapeutic potential of tunicamycin in murine L1210 leukemia
.
Cancer Res
1983
;
43
:
1669
74
.
14
Hanson
SR
,
Culyba
EK
,
Hsu
TL
,
Wong
CH
,
Kelly
JW
,
Powers
ET
. 
The core trisaccharide of an N-linked glycoprotein intrinsically accelerates folding and enhances stability
.
Proc Natl Acad Sci U S A
2009
;
106
:
3131
6
.
15
Lau
KS
,
Partridge
EA
,
Grigorian
A
, et al
. 
Complex N-glycan number and degree of branching cooperate to regulate cell proliferation and differentiation
.
Cell
2007
;
129
:
123
34
.
16
Wang
XC
,
Gu
JG
,
Ihara
H
,
Miyoshi
E
,
Honke
K
,
Taniguchi
N
. 
Core fucosylation regulates epidermal growth factor receptor-mediated intracellular signaling
.
J Biol Chem
2006
;
281
:
2572
7
.
17
Zavareh
RB
,
Lau
KS
,
Hurren
R
, et al
. 
Inhibition of the sodium/potassium ATPase impairs N-glycan expression and function
.
Cancer Res
2008
;
68
:
6688
97
.
18
Harari
PM
,
Allen
GW
,
Bonner
JA
. 
Biology of interactions: antiepidermal growth factor receptor agents
.
J Clin Oncol
2007
;
25
:
4057
65
.
19
Bonner
JA
,
Harari
PM
,
Giralt
J
, et al
. 
Radiotherapy plus cetuximab for locoregionally advanced head and neck cancer: 5-year survival data from a phase 3 randomised trial, and relation between cetuximab-induced rash and survival
.
Lancet Oncol
2010
;
11
:
21
8
.
20
Engelman
JA
,
Janne
PA
. 
Mechanisms of acquired resistance to epidermal growth factor receptor tyrosine kinase inhibitors in non-small cell lung cancer
.
Clin Cancer Res
2008
;
14
:
2895
9
.
21
Wheeler
DL
,
Huang
S
,
Kruser
TJ
, et al
. 
Mechanisms of acquired resistance to cetuximab: role of HER (ErbB) family members
.
Oncogene
2008
;
27
:
3944
56
.
22
Contessa
JN
,
Abell
A
,
Mikkelsen
RB
,
Valerie
K
,
Schmidt-Ullrich
RK
. 
Compensatory ErbB3/c-Src signaling enhances carcinoma cell survival to ionizing radiation
.
Breast Cancer Res Treat
2006
;
95
:
17
27
.
23
Engelman
JA
,
Zejnullahu
K
,
Mitsudomi
T
, et al
. 
MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling
.
Science
2007
;
316
:
1039
43
.
24
Mueller
KL
,
Hunter
LA
,
Ethier
SP
,
Boerner
JL
. 
Met and c-Src cooperate to compensate for loss of epidermal growth factor receptor kinase activity in breast cancer cells
.
Cancer Res
2008
;
68
:
3314
22
.
25
DeRossi
C
,
Bode
L
,
Eklund
EA
, et al
. 
Ablation of mouse phosphomannose isomerase (Mpi) causes mannose 6-phosphate accumulation, toxicity, and embryonic lethality
.
J Biol Chem
2006
;
281
:
5916
27
.
26
Thiel
C
,
Lubke
T
,
Matthijs
G
,
Von Figura
K
,
Körner
C
. 
Targeted disruption of the mouse phosphomannomutase 2 gene causes early embryonic lethality
.
Mol Cell Biol
2006
;
26
:
5615
20
,
PMCID: 1592760
.
27
Marek
KW
,
Vijay
IK
,
Marth
JD
. 
A recessive deletion in the GlcNAc-1-phosphotransferase gene results in peri-implantation embryonic lethality
.
Glycobiology
1999
;
9
:
1263
71
.
28
Kamb
A
,
Wee
S
,
Lengauer
C
. 
Why is cancer drug discovery so difficult?
Nat Rev Drug Discov
2007
;
6
:
115
20
.
29
Scott
AM
,
Lee
FT
,
Tebbutt
N
, et al
. 
A phase I clinical trial with monoclonal antibody ch806 targeting transitional state and mutant epidermal growth factor receptors
.
Proc Natl Acad Sci U S A
2007
;
104
:
4071
6
,
PMCID: 1805701
.
30
Liu
N
,
Li
M
,
Li
X
, et al
. 
PET-based biodistribution and radiation dosimetry of epidermal growth factor receptor-selective tracer 11C-PD153035 in humans
.
J Nucl Med
2009
;
50
:
303
8
.