IGFBP2 is a candidate biomarker for Ink4a-Arf status and a therapeutic target for high-grade gliomas Lynette M. Moorea, Kristen M. Holmesa, Sarah M. Smitha, Ying Wua, Elena Tchougounovab, Lene Uhrbomb, Raymond Sawayac, Janet M. Brunera, Gregory N. Fullera,1, and Wei Zhanga,1 Departments of aPathology and cNeurosurgery, University of Texas MD Anderson Cancer Center, Houston, TX 77030; and bDepartment of Genetics and Pathology, Uppsala University, Rudbeck Laboratory, SE-751 85, Uppsala, Sweden Edited by Webster K. Cavenee, University of California at San Diego, La Jolla, CA, and approved August 14, 2009 (received for review January 23, 2009) The levels of insulin-like growth factor-binding protein 2 (IGFBP2) Genetic studies and genomic profiling of human gliomas have are elevated during progression of many human cancers. By using identified a number of common alterations, including frequent a glial-specific transgenic mouse system (RCAS/Ntv-a), we reported deletion of the INK4a-ARF locus, loss of PTEN, and overex- previously that IGFBP2 is an oncogenic factor for glioma progres- pression of IGFBP2, and these comprehensive studies have sion in combination with platelet-derived growth factor-␤ revealed valuable information regarding critical pathways used (PDGFB). Because the INK4a-ARF locus is often deleted in high- by gliomas (9, 10). IGFBP2 overexpression is correlated with grade gliomas (anaplastic oligodendroglioma and glioblastoma), high-grade gliomas and poor prognosis, and experiments from we investigated the effect of the Ink4a-Arf-null background on our laboratory have demonstrated that IGFBP2 is an oncogene IGFBP2-mediated progression of PDGFB-initiated oligodendrogli- that promotes glioma development and progression in a glial- oma. We demonstrate here that homozygous deletion of Ink4a-Arf specific mouse model (11–13). Further, IGFBP2 and PTEN ex- bypasses the requirement of exogenously introduced IGFBP2 for pression levels are negatively correlated in brain and prostate glioma progression. Instead, absence of Ink4a-Arf resulted in cancers, implicating IGFBP2 as a biomarker for PTEN status (14). elevated endogenous tumor cell IGFBP2. An inverse relationship In addition to PTEN, loss of Ink4a-Arf is also a frequent event between p16INK4a and IGFBP2 expression was also observed in in AO (3, 15). This locus encodes the p16INK4a and p14ARF human glioma tissue samples and in 90 different cancer cell lines (p19ARF in mouse) tumor-suppressor proteins, which function as CELL BIOLOGY by using Western blotting and reverse-phase protein lysate arrays. critical regulators of the RB and p53 pathways (16). Loss of When endogenous IGFBP2 expression was attenuated by an RCAS Ink4a-Arf in a mouse model of PDGFB-initiated gliomas led to vector expressing antisense IGFBP2 in our mouse model, a de- an increase in tumor incidence compared with wild-type (17). creased incidence of anaplastic oligodendroglioma as well as Additionally, loss of Ink4a-Arf was found to cooperate with K-ras prolonged survival was observed. Thus, p16INK4a is a negative to induce GBM (18). In the present study, we examined whether regulator of the IGFBP2 oncogene. Loss of Ink4a-Arf results in loss of Ink4a-Arf is directly correlated with increased IGFBP2 increased IGFBP2, which contributes to glioma progression, expression in gliomas, and we functionally interrogated the thereby implicating IGFBP2 as a marker and potential therapeutic significance of Ink4a-Arf loss in IGFBP2-mediated glioma pro- target for Ink4a-Arf-deleted gliomas. gression by using the RCAS/Ntv-a mouse glioma model. We found that the expression of IGFBP2 was inversely liomas are the most common type of human primary brain correlated with both p16INK4a and ARF expression, as well as Gtumor. Surgical resection followed by radiation therapy with PTEN, in 90 different human cancer cell lines. We then (RT) and chemotherapy has been the most common treatment examined the relationship between Ink4a-Arf deletion and IG- regimen. In patients with anaplastic oligodendroglioma (AO), FBP2 expression in an RCAS/Ntv-a glial-specific mouse model. the standard historical treatment was surgery followed by RT, Transgenic mice expressing the avian virus receptor, tv-a, under with chemotherapy being reserved for recurrence (1). Despite the control of the Nestin promoter, which drives expression in treatment, overall survival for patients with AO is 3 to 5 years, glial progenitors, were crossed to Ink4a-Arf-null mice to generate Ϫ Ϫ whereas patients with the highest grade of glioma, glioblastoma the Ntv-a; Ink4a/ARF / mice. Somatic cell gene transfer with (GBM), have a survival rate of less than 1 year (1, 2). Genetic replication-competent avian leukemia virus with splice acceptor signatures have been useful for predicting prognosis and re- (RCAS) vectors allowed the specific delivery of oncogenes to the sponse to therapy. Chromosomal aberrations, especially code- glial progenitors of these mice (19). Consistent with the human letion of 1p and 19q in AO, have been associated with both data, we observed increased mouse IGFBP2 in glial progenitor chemotherapeutic response and increased overall and progres- cells from Ink4a/ARFϪ/Ϫ, Ink4aϪ/Ϫ, and ARFϪ/Ϫ mice. Delivery sion-free survival (3). The identification and characterization of and overexpression of PDGFB to the Ntv-a; Ink4a/ARFϪ/Ϫ a number of common genetic alterations in glial tumors have newborn brain led to development of AO without the cointro- incited considerable interest in the development of targeted duction of IGFBP2, which was required in the case of wild-type therapies. A number of these therapies have achieved consid- Ntv-a mice (11). Knockdown of endogenous IGFBP2 by anti- erable success in other cancers; for example, imatinib has been sense IGFBP2 reduced high-grade glioma incidence and pro- used to target the BCR-ABL fusion protein in chronic myelog- longed survival of tumor-bearing mice. These observations enous leukemia and to target KIT and PDGFR in gastrointes- tinal stromal tumors (4). Epidermal growth factor receptor (EGFR) is commonly amplified in a wide spectrum of human Author contributions: R.S., J.M.B., G.N.F., and W.Z. designed research; L.M.M., K.M.H., S.M.S., Y.W., and E.T. performed research; L.U. contributed new reagents/analytic tools; cancers, including GBM, and is another example of targeted L.M.M., K.M.H., S.M.S., E.T., L.U., G.N.F., and W.Z. analyzed data; and L.M.M., J.M.B., G.N.F., drug development (5). However, anti-EGFR treatment has not and W.Z. wrote the paper. had the anticipated success in GBM clinical trials, despite The authors declare no conflict of interest. significant benefit in other neoplastic diseases, such as colon This article is a PNAS Direct Submission. cancer (6–8). The recent progress of targeted therapy in other 1To whom correspondence may be addressed. E-mail: [email protected] or tumor types highlights the need for identification of genetic and [email protected]. molecular alterations to serve as markers for patient stratifica- This article contains supporting information online at www.pnas.org/cgi/content/full/ tion and as targets for glioma therapeutics. 0900807106/DCSupplemental. www.pnas.org͞cgi͞doi͞10.1073͞pnas.0900807106 PNAS Early Edition ͉ 1of5 Downloaded by guest on September 24, 2021 p16Ink4a/ A B WT p16Ink4a-/-ARF-/- ARF-/- p16INK4a XXXXBBBB IGFBP2 IGFBP2 Actin Actin 12345 678 9101112 13 14 15 16 17 1 0.9 21.3 18.7 38 22 29.3 26 C IGFBP2 vs. PTEN D IGFBP2 vs. p16INK4a 26.6% 27.8% 40 10 20 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0.0 Fold Difference Fold Difference 0.0 27.8% 26.6% E IGFBP2 vs. p19ARF F IGFBP2 vs. IGFBP5 18.9% 10 5 3.3% 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 3.3% 0.0 Fold Difference 0.0 Fold Difference 16.6% Fig. 1. IGFBP2 expression is inversely related to p16INK4a expression. (A) Protein lysates from 17 human glioma samples were examined by Western blot analysis for IGFBP2 and p16INK4a protein levels. High-grade samples include grade III oligodendroglioma (lanes 4, 10, and 11) and astrocytomas (lanes 16 and 17) as well as GBM (lanes 1–3, 6–9, and 14–15). Samples 5, 12, and 13 were classified as grade II oligodendroglioma or astrocytoma and are noted in gray. (B) Western blot of Ntv-a glial progenitor cell lysates infected with empty RCAS vector (RCAS-X) or RCAS-PDGFB. IGFBP2 protein levels are elevated in cell lysates from Ink4a/ARF-deleted, Ink4a-deleted, and ARF-deleted mice. The numbers below the figure represent relative IGFBP2 levels after normalization to actin levels. (C–F) Fold difference in protein levels between IGFBP2 and PTEN (C), p16INK4a (D), p19ARF (E), and IGFBP5 (F). Protein levels in 90 cell lines were examined by protein lysate array, and fold differences were plotted. The red areas represent samples where IGFBP2 protein levels were greater than 2-fold higher than the compared protein; green areas represent samples where IGFBP2 protein levels were at least 2-fold lower. The percent of samples with greater than 2-fold differences are listed in each graph. suggest that IGFBP2 may serve as both a marker and therapeutic colorectal carcinoma, ovarian cancer, and acute lymphoblastic target for Ink4a-Arf-deleted gliomas. leukemia (20–23). To further investigate this relationship, we mined data from Results our high-throughput reverse-phase protein lysate array analysis IGFBP2 Expression Is Inversely Correlated with p16INK4a Status. (24). This work examined 90 cell lines comprising 12 tumor types, IGFBP2 is frequently up-regulated in glioma and was identified including GBM, prostate cancer, breast cancer, and sarcoma. We recently as a potential biomarker for PTEN status and PI3K/Akt compared the relative protein levels of IGFBP2 with several pathway activation in both glioma and prostate cancer (14).