OVEREXPRESSION OF PAX5 IS NOT SUFFICIENT FOR NEOPLASTIC TRANSFORMATION OF MOUSE NEUROECTODERM Joachim P. STEINBACH 1 , Zbynek KOZMIK 2 , Peter PFEFFER 2 and Adriano AGUZZI 1 * 1 Institute of Neuropathology, University Hospital, Zurich, Switzerland 2 Research Institute of Molecular Pathology, Dr. Bohr-Gasse 7, Vienna, Austria The developmental control genes of the Pax family are essential for brain development. Several Pax genes are also involved in chromosomal translocations causing malignan- cies in humans, and Pax5 expression is deregulated in medul- loblastomas. We have investigated whether Pax5 can induce tumors in the developing mouse brain. Primary mouse em- bryonic neuroectodermal cells were retrovirally transduced with mouse Pax5 and transplanted into the brain of syngeneic host mice. No tumors developed in 36 transplants after one year, and there were no alterations in the differentiation pattern of the neural transplants. We then generated trans- genic mice expressing human Pax5 under control of the En- grailed-2 promoter, which is expressed in the cerebellar ex- ternal granule cell layer and in medulloblastomas. Sustained expression was achieved in the cerebellum of transgenic an- imals throughout lifetime. Expression levels were similar to those observed in human medulloblastomas. Again, cerebel- lar morphogenesis was undisturbed, and no tumors arose. These results strongly argue against a dominant transform- ing activity of PAX5 in NEC and in cerebellar granule cell precursors of mice, and underline the restricted tissue-spec- ificity of PAX5 related oncogenesis. © 2001 Wiley-Liss, Inc. Key words: Pax5; retrovirus; transgenic mice; CNS; carcinogenesis; medulloblastoma Medulloblastoma is a relatively common, highly malignant pe- diatric tumor that arises in the cerebellum. The molecular patho- genesis of medulloblastoma is poorly understood. Mutations typ- ically encountered in glial neoplasms are rare in medulloblastoma, suggesting a specific pathogenesis of the latter tumor. 1 The histo- genesis of medulloblastoma remains unclear: it was proposed to originate from committed neuronal precursors such as the cells of the external granule cell layer of the cerebellum (EGL), or alter- natively from pluripotent neuroectodermal progenitors of the sub- ependymal germinal layer, from which extracerebellar primitive neuroectodermal tumors (PNETs) may also arise. 2,3 The genetics of medulloblastoma is sketchy. The APC tumor suppressor gene, which is normally expressed in the developing cerebellum, 4 was found to be mutated in patients with medullo- blastoma and with Turcot syndrome. 5 In addition, mutations in the PTCH gene, a gene homologous to the developmental control gene patched in Drosophila, were found in patients with medulloblas- toma and the nevoid basal cell syndrome 6 and in some patients with sporadic medulloblastoma. 7,8 The developmental control genes of the Pax family are tran- scription factors that are expressed in specific regions of the brain and contribute to the patterning of brain structures. 9 It is likely that these genes act by controlling proliferation and migration of neu- roectodermal precursor cells during development. This view is supported by the effect of targeted deletion of Pax5, which leads to hypoplasia of the inferior colliculus and anterior cerebellum in which it is normally expressed. 10,11 Further, paired box-containing developmental control genes seem to act as true oncogenes in vitro and in naturally occurring tumors: (i) ectopic overexpression of Pax genes elicits transformation of mouse 3T3 cells and 208-F rat fibroblasts in vitro, and transformed cells form tumors in nude mice, 12 and (ii) PAX3 and PAX7 genes are consistently activated in alveolar rhabdomyosarcoma, a malignant tumor of childhood char- acterized by specific chromosomal translocations involving either of these two Pax genes. 13,14 Inhibition of tumor cell apoptosis by PAX proteins may contribute to tumorigenesis, 15 and there is a PAX binding site in the promoter of the p53 gene that mediates transcriptional inhibition of p53 by PAX proteins. 16 Pax5 has a unique function in the development of the immune system. Proliferation and differentiation of B-cells is controlled by Pax5, and in Pax5 knockout mice there are no mature B-cells. Pax5 binds specifically to the hypophosphorylated form of the RB protein, suggesting an influence on cell cycle regulation. 17 In addition, there is a wealth of evidence that points specifically to Pax5 as a potential tumor gene. A chromosomal translocation t(9;14)(pl3;q32) involving PAX5 and the potent E enhancer of the IgH gene resulting in overexpression of PAX5 in B-cells was recently reported in cases of B-cell lymphoma. 18 Therefore, at least in lymphoma, PAX5 may act as a bona fide dominant onco- gene. Given the important role of Pax5 in development of both B- lymphocytes and brain, and its likely involvement in B-cell lym- phoma, it seems logical to ask whether Pax5 overexpression may be relevant to brain malignancies. We have previously shown deregulated expression of PAX5 in a large series of medulloblas- tomas, 19 and overexpression of PAX5 was described also in glial brain tumors. 20 Neither study, however, addressed the question whether the observed pattern of enhanced expression was causally involved in neurocarcinogenesis. In the present study, we have therefore investigated whether Pax5 can induce neoplasias in the mouse neuroectoderm. To this end, primary neuroectodermal cells (NECs) were infected in vitro with a retroviral vector encoding Pax5, and transplanted into the caudoputamen of adult recipient mice. We also attempted to achieve gene transfer in vivo by injecting packaging cells that produced the Pax5 retroviral vector directly into the cerebellum of newborn mice. Finally, we produced transgenic mice that express PAX5 under control of the Engrailed-2 promoter (En-2), and highly overexpress PAX5 in the cerebellum throughout life. Abbreviations: EGL, external granule cell layer; NECs, embryonic neu- roectodermal cells; PNETs, primitive neuroectodermal tumors. Grant sponsor: Kanton, Zurich; Grant sponsor: Swiss National Founda- tion; Grant sponsor: Swiss Cancer League. Dr. Steinbach’s current address is: Department of Neurology, University Hospital, Hoppe-Seylerstr. 3, D-72076 T¨ ubingen, Germany. Dr. Kozmik’s current address is: Institute of Molecular Genetics, Fle- mingovo 2, CZ-166 37 Prague, Czechoslovakia. *Correspondence to: Institute of Neuropathology, Schmelzbergstrasse 12, CH-8091 Switzerland. Fax: +41-1-255-4402. E-mail: adriano@pathol.unizh.ch Received 15 January 2001; Revised 7 March 2001; Accepted 9 March 2001 Int. J. Cancer: 93, 459 – 467 (2001) © 2001 Wiley-Liss, Inc. Publication of the International Union Against Cancer