EGFRvIII and DNA double-strand break repair: a molecular mechanism for radioresistance in glioblastoma

Abstract

Glioblastoma multiforme (GBM) is the most lethal of brain tumors and is highly resistant to ionizing radiation (IR) and chemotherapy. Here, we report on a molecular mechanism by which a key glioma-specific mutation, epidermal growth factor receptor variant III (EGFRvIII), confers radiation resistance. Using Ink4a/Arf-deficient primary mouse astrocytes, primary astrocytes immortalized by p53/Rb suppression, as well as human U87 glioma cells, we show that EGFRvIII expression enhances clonogenic survival following IR. This enhanced radioresistance is due to accelerated repair of DNA double-strand breaks (DSB), the most lethal lesion inflicted by IR. The EGFR inhibitor gefitinib (Iressa) and the phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002 attenuate the rate of DSB repair. Importantly, expression of constitutively active, myristylated Akt-1 accelerates repair, implicating the PI3K/Akt-1 pathway in radioresistance. Most notably, EGFRvIII-expressing U87 glioma cells show elevated activation of a key DSB repair enzyme, DNA-dependent protein kinase catalytic subunit (DNA-PKcs). Enhanced radioresistance is abrogated by the DNA-PKcs-specific inhibitor NU7026, and EGFRvIII fails to confer radioresistance in DNA-PKcs-deficient cells. In vivo, orthotopic U87-EGFRvIII-derived tumors display faster rates of DSB repair following whole-brain radiotherapy compared with U87-derived tumors. Consequently, EGFRvIII-expressing tumors are radioresistant and continue to grow following whole-brain radiotherapy with little effect on overall survival. These in vitro and in vivo data support our hypothesis that EGFRvIII expression promotes DNA-PKcs activation and DSB repair, perhaps as a consequence of hyperactivated PI3K/Akt-1 signaling. Taken together, our results raise the possibility that EGFR and/or DNA-PKcs inhibition concurrent with radiation may be an effective therapeutic strategy for radiosensitizing high-grade gliomas.

Figure 1. EGFRvIII expression renders astrocytes radioresistant irrespective of tumor-suppressor background

(A) Expression of EGFRvIII (vIII), wild type EGFR (WT), or kinase-dead EGFR (KD) in SV40-LT-expressing mouse astrocytes or in Ink4a/Arf−/− primary mouse astrocytes was assayed by Western blotting with α-EGFR and α-actin (loading control) antibodies. “Par” denotes parental cells. (B) Radiation survival of astrocyte lines was quantified by colony formation assays. The fraction of surviving colonies (y-axis) was plotted against corresponding radiation dose (x-axis). Error bars represent standard error of the mean of experiments performed three or more times. Please note increased survival after EGFRvIII expression (red lines). (C) Radiation survival of astrocytes pre-treated with Gefitinib (Iressa) was quantified as above. Note reduced survival after Iressa pre-treatment (dashed lines).

Figure 2. Increased radioresistance of EGFRvIII-expressing astrocytes correlates with proficient DSB repair

(A) SV40-LT-expressing mouse astrocytes or Ink4a/Arf−/− primary mouse astrocytes were irradiated with 1 Gy of γ rays and immunostained for 53BP1 foci at various time points ranging from 0.5 to 8 h post-irradiation. The pictures depict initial DNA damage (0.5 h) and DNA damage remaining at 4 and 8 h. (B) 53BP1 foci were scored at 0.5, 2, 4, and 8 h post-irradiation. The number of 53BP1 foci was determined for each time point (average of 100 nuclei) and, after subtracting background (number of foci in unirradiated nuclei), the percentage foci remaining was plotted against time to obtain DNA repair kinetics. Error bars represent standard error of the mean. Note nearly complete repair by 4 h in cells expressing EGFRvIII. Pre-treatment of cells with the EGFR inhibitor Iressa (dashed lines) or with the DNA-PKcs inhibitor NU7026 (dotted lines) abrogated DSB repair. (C) Radiation survival of astrocytes pre-treated with NU7026 was quantified by colony formation assays. The fraction of surviving colonies (y-axis) was plotted against corresponding radiation dose (x-axis). Error bars represent standard error of the mean of experiments performed three or more times.

Figure 3. A genetic link between EGFRvIII and DNA-PKcs in the repair of radiation-induced DSBs

(A) Ectopic expression of EGFRvIII in DNA-PKcs+/+ and −/− mouse embryonic fibroblasts was assayed by Western blotting with α-EGFR and α-H2AX (loading control) antibodies. (B) Radiation survival was quantified by colony formation assays. The fraction of surviving colonies (y-axis) was plotted against corresponding radiation dose (x-axis). Error bars represent standard error of the mean. Note increased survival of DNA-PKcs+/+ cells expressing EGFRvIII while DNA-PKcs−/− cells expressing EGFRvIII show no enhanced survival compared to parental lines.

Figure 4. Akt-1 activation in mouse astrocytes mimics effects of EGFRvIII expression on DSB repair

(A) Phosphorylation of Akt-1 (at ser 473) was assayed by Western blotting of SV40-LT-expressing mouse astrocytes (parental), SV40-LT-astrocytes with EGFRvIII over-expression, EGFRvIII-expressing astrocytes pre-treated with a PI3K inhibitor (LY294002), or SV40-LT-astrocytes expressing myristylated-Akt-1. (B) Astrocytes were irradiated (1 Gy) and immunostained for 53BP1 foci to obtain DSB repair kinetics. Error bars represent standard error of the mean. Note proficient repair in cells expressing myr-Akt-1 (blue line), while repair was abrogated in EGFRvIII-expressing cells treated with LY294002 (dashed red line).

Fig. 5. EGFRvIII over-expression results in hyperactivation of DNA-PKcs in response to IR

(A) Expression of EGFRvIII (vIII), wild type EGFR (WT), or kinase-dead EGFR (KD) was assessed in human U87 glioma cells by Western blotting with α-EGFR and α-actin (loading control) antibodies. (B) Radiation survival of U87 cells was quantified by colony formation assay. Where indicated, cells were pre-treated with the DNA-PKcs inhibitor, NU7026. The fraction of surviving colonies (y-axis) was plotted against the corresponding radiation dose (x-axis). Error bars represent standard error of the mean of experiments performed three or more times. Note increased survival of EGFRvIII-expressing cells (solid red line) and radiosensitisation with NU7026 (dashed red line). (C) U87-parental or U87-EGFRvIII cells were irradiated (1 Gy) and immunostained for 53BP1 foci at various time points post-IR to obtain DSB repair kinetics. Note proficient repair in cells expressing EGFRvIII (red line). Error bars represent standard error of the mean. (D) Activation of DNA-PKcs in irradiated U87 cells was assayed by Western blotting using α-phospho-DNA-PKcs(pT2647) and α-DNA-PKcs antibodies. Note hyperphosphorylation of DNA-PKcs in cells expressing EGFRvIII.

Fig. 6. EGFRvIII enhances DSB repair in a mouse orthotopic glioma model in vivo

(A) Orthotopic U87-parental and U87-EGFRvIII tumors were generated in nude mice. Once tumors reached 50% maximal tolerated size, mice received cranial irradiation (2 Gy) and were sacrificed at 0.5, 2, 4, and 8 h post-IR to obtain DNA repair kinetics. Intracranial tumors were identified in coronal brain sections by H&E staining. Brain regions occupied by tumor mass were stained with α-53BP1 antibody to visualize radiation-induced DSBs. (B) 53BP1 foci were scored to obtain DSB repair kinetics. Error bars represent standard error of the mean. Note proficient repair in EGFRvIII-expressing tumors (red line). (C) Growth of orthotopic U87-parental and U87-EGFRvIII tumors was monitored by serial luciferase imaging (representative images are shown). Nude mice with intracranial U87-parental or U87-EGFRvIII tumors were either mock-irradiated or irradiated (4Gy) (n=5 per cohort) and luciferase intensities were quantified over a period of 20 days. The plot represents average signal intensity (photons/sec × 10⁴) for each cohort (y axis) plotted versus time post-implantation (x axis). Arrows represent time of radiation. Error bars represent standard error of the mean. Note marked decrease in the rate of U87-parental tumor growth (dashed green line) following IR (p<0.01) while U87-EGFRvIII tumors (dashed red line) show no significant difference in rate of tumor growth (p>0.05). (D) Kaplan-Meier analyses of mice with intracranial U87-parental or U87-EGFRvIII tumors (n=6 per cohort). Note no significant increase in post-radiation (4 Gy) survival of mice harboring U87-EGFRvIII tumors (dashed red line) (p>0.05) in contrast to marked increase in survival of mice bearing U87-parental tumors (dashed green line) (p<0.01).