Class III beta-tubulin isotype (betaIII-tubulin) is widely regarded as a neuronal marker in developmental neurobiology and stem cell research. To test the specificity of this marker protein, we determined its expression and distribution in primary cultures of glial fibrillary acidic protein (GFAP)-expressing astrocytes isolated from the cerebral hemispheres of 2 human fetuses at 18 to 20 weeks of gestation. Cells were maintained as monolayer cultures for 1 to 21 days without differentiation induction. By immunofluorescence microscopy, coexpression of betaIII-tubulin and GFAP was detected in cells at all time points but in spatially distinct patterns. The numbers of GFAP+ cells gradually decreased from Days 1 to 21 in vitro, whereas betaIII-tubulin immunoreactivity was present in 100% of cells at all time points. beta-III-tubulin mRNA and protein expression were demonstrated in cultured cells by reverse-transcriptase-polymerase chain reaction and immunoblotting, respectively. Glial fibrillary acidic protein+/beta-III-tubulin-positive cells coexpressed nestin and vimentin but lacked neurofilament proteins, CD133, and glutamate-aspartate transporter. Weak cytoplasmic staining was detected with antibodies against microtubule-associated protein 2 isoforms. Confocal microscopy, performed on autopsy brain samples of human fetuses at 16 to 20 gestational weeks, revealed widespread colocalization of GFAP and betaIII-tubulin in cells of the ventricular/subventricular zones and the cortical plate. Our results indicate that in the midgestational human brain, betaIII-tubulin is not neuron specific because it is constitutively expressed in GFAP+/nestin+ presumptive fetal astrocytes.

No abstract

Centrosome amplification is a pivotal mechanism underlying tumorigenesis but its role in gliomas is underinvestigated. The present study specifically examines the expression and distribution of the centrosome-associated cytoskeletal protein gamma-tubulin in 56 primary diffuse astrocytic gliomas (grades II-IV) and in 4 human glioblastoma cell lines (U87MG, U118MG, U138MG, and T98G). Monoclonal anti-peptide antibodies recognizing epitopes in C-terminal or N-terminal domains of the gamma-tubulin molecule were used in immunohistochemical, immunofluorescence, and immunoblotting studies. In tumors in adults (n = 46), varying degrees of localization were detected in all tumor grades, but immunoreactivity was significantly increased in high-grade anaplastic astrocytomas and glioblastomas multiforme as compared to low-grade diffuse astrocytomas (p = 0.0001). A similar trend was noted in diffuse gliomas in children but the sample of cases was too small as to be statistically meaningful. Two overlapping patterns of ectopic cellular localization were identified in both primary tumors and glioblastoma cell lines: A punctate pattern, in which gamma-tubulin was partially co-distributed with pericentrin in the pericentriolar region, and a diffuse pattern, independent of pericentrin staining, denoting a soluble pool of gamma-tubulin. Cellular gamma-tubulin was detected in both soluble and insoluble (nocodazole-resistant) fractions of glioblastoma cells. Divergent localizations of gamma-tubulin and pericentrin suggest a differential distribution of these 2 centrosome-associated proteins in glioblastoma cell lines. Our results indicate that overexpression and ectopic cellular distribution of gamma-tubulin in astrocytic gliomas may be significant in the context of centrosome protein amplification and may be linked to tumor progression and anaplastic potential.