Expression of TWIST mRNA and protein in rodent glioma and human glioblastoma cell lines. (A) Northern blot analysis showing the expression of TWIST in rat C6 glioma cells (far left lane), confirming the results of a degenerate RT-PCR screen for class B bHLH genes. TWIST mRNA expression was detected in five of six human glioma cell lines (A172, T98G, SNB19, UW455, UW18, and SF767), as well as in lines derived from a neuroblastoma (SKNMC) and a medulloblastoma (D283). Lines derived from colon (SW480), breast (MDA453), and small cell lung cancer (H82) had no detectable expression. The gel stained with ethidium bromide (lower panel) demonstrates equivalent loading and integrity of total RNA. (B) Southern hybridization of TWIST-specific probes to RT-PCR products from a larger panel of human tumor cell lines. Blots of GAPDH controls show that comparable amounts of total RNA were analyzed for each sample. Differential levels of TWIST expression were detected in 11 of 12 human gliomas and in two neuoblastomas and one medulloblastoma, whereas no detectable expression was noted in non-neural tumors. TWIST mRNA was not detected by RT-PCR or by Northern blot analysis in the glioblastoma line, UW455, or in the non-neural cell lines examined. (C) Western blot analysis of 30 µg of whole cell lysates of UW455 and SF767, transfected either with an empty expression vector (LXSN) or a TWIST overexpression construct (Tw), demonstrates the presence of a band at the expected molecular mass of 26 kDa. UW455, which lacks detectable TWIST mRNA and SF767LXSN with low basal message levels, shows no detectable protein. (D) Nuclear extracts of the SF767LXSN, SNB19LXSN, and T98G cell lines (50 µg each) demonstrate endogenous expression of a 26-kDa TWIST protein comparable to that detected in nuclear extracts from SF767Tw (10 µg) at levels commensurate with mRNA levels (panel A). The nuclear protein, INI1, is used to compare the levels of nuclear-specific protein loaded from each preparation.

TWIST expression in human gliomas. The abundance of mRNA was analyzed by Southern hybridization of TWIST-and GAPDH-specific probes to RT-PCR products. (A) Glial tumors, grouped by grade, include grade II fibrillary astrocytomas (1–3), oligodendroglioma (4–9), and mixed oligoastrocytoma (10); grade III anaplastic mixed oligoastrocytomas (11–13) and anaplastic astrocytoma (14–16); and grade IV glioblastoma (17–26) and gliosarcoma (27 and 28). The nine tumors recurrent after prior treatment are indicated by a caret beneath the corresponding lane. No statistical difference in gene expression levels between primary or recurrent samples was noted (data not shown). Representative RT-PCR controls without RT or cDNA are shown at the far right. (B) Histogram of relative TWIST expression for gliomas, grouped by grade. The signal density of TWIST is expressed as a percentage of the corresponding GAPDH signal. (C) Mean relative TWIST expression with standard errors of the mean by glioma grade. Significant differences (P ≤ .05) indicated by asterisks were demonstrated between grades II and IV (**) and combined grade II plus grade III and IV tumors (*).

TWIST expression patterns in low-grade and high-grade gliomas by in situ hybridization and immunostaining. H&E histology (A–D), TWIST in situ hybridization (E–H), and immunostaining (I–L) are shown from adjacent sections of sample 7, a grade II glioma with low levels of TWIST by RT-PCR (A, E, and I); sample 24, a grade IV glioblastoma (B, F, and J); sample 28, a grade IV gliosarcoma (C, G, and K) with strong TWIST expression; and a rhabdomyosarcoma (D, H, and L), previously reported to express TWIST [10]. H&E and in situ hybridization photomicrographs were taken with x20 objective, whereas iummunostained images were taken with a x40 objective. Sense control hybridization reactions for each sample are shown as insets in the right lower hand corner (E–H). In situ hybridization sections were counterstained with methyl green to facilitate visualization of tissue architecture. Immunostaining demonstrates occasional faint staining in the low-grade sample (I), whereas abundant pronounced nuclear staining is present in all of the high-grade tumors (J–L). Nuclear staining is not apparent in isotype IHC controls (lower right corner insets), indicating the specificity of the antibody. (The 50-µm scale bar in (A) applies to (A)–(H), and the 25-µm scale bar in (I) applies to (I)–(L).)

Increased TWIST protein and mRNA content accompanies malignant progression. Immunostaining revealed increased abundance of TWIST protein accompanying the recurrence of a grade II oligodendroglioma (A) to a grade IV glioblastoma (B). The recurrent tumor demonstrates increased cellularity characteristic of glioblastoma, together with pronounced TWIST immunopositivity in tumor cell nuclei. High levels of TWIST protein (C) were particularly evident in areas of the recurrent tumor with malignant histologic features of increased cellularity, necrosis (asterisk), and microvascular proliferation (arrow). In accord with the elevated level of TWIST protein evident by immunostaining, the recurrent sample demonstrates increased abundance of TWIST mRNA by RT-PCR (D). (The 40-µm scale bar in (A) applies to (A)–(C).)

TWIST mRNA abundance is elevated by p53 deficiency in mouse astrocytes. (A) Comparison of TWIST mRNA detected by Southern hybridization of RT-PCR reaction products from cultured p53 wild-type (+/+) and hemizygous (+/-) astrocyte cultures derived from neonatal mice. Also shown are levels in normal adult mouse brain (NAB) and a control lacking RT. (B) Density of TWIST mRNA signal normalized to that of GAPDH of samples shown in (A). Values are the mean ± SD of three independent determinations. The appearance of features accompanying malignant transformation at specific passage numbers is indicated as follows: DNA ploidy abnormalities (*), serum-independent growth, and tumor formation in nude mice (^). The marked increase of TWIST expression in p53^+/- astrocytes relative to normal brain and wild-type astrocytes at passages 1 and 4 is maintained through serial passages coinciding with progressive acquisition of malignant phenotype in vitro, including complete malignant transformation (growth of solid tumors in vivo) by passage 50.

Overexpression of TWIST promotes invasion but not cell proliferation in SF767 glioma cells. (A) The impact of TWIST overexpression on cell growth rate in culture was examined by plating 1 x 10⁵ SF767Tw or SF767 LXSN vector control cells in 60-mm plates in triplicate. Cells were trypsinized and counted each day for 4 days using a Coulter counter, and the mean of three wells from each time point is shown. (B) TWIST effects on cell saturation density were assayed by plating 2 x 10⁵ SF767Tw or SF767 LXSN vector control cells in six-well plates in triplicate. Triplicate wells were trypsinized and counted every second day for 7 days using a Coulter counter and plotted. Calculated standard errors were small and cannot be visualized at this scale. (C, D) Invasion through two extracellular matrices is increased by overexpression of TWIST in SF767 assayed by two separate invasion assays (C, Chemicon; D, BD Biosciences). Data were analyzed after 70 hours for BD Biosciences and 48 hours for the Chemicon assay. Photomicrographs of stained cells on the undersurface of matrix-coated 8-µm pore size filter demonstrate differential invasion of SF767 LXSN and SF767 TW cells. Cells from triplicate assays were stained and quantified by colorimetric reading at OD of 560 nm. Invasion by SF767TW was calculated as a percentage of the SF767 LXSN vector control values. Data shown are mean ± SE.

Expression of TWIST in human developing and adult brain. (A) Southern hybridization of RT-PCR reactions from embryonic, fetal, and adult tissues including control bone tissue (79 days) known to express TWIST and a panel of embryonic and fetal brain (Br) samples from 52/3 to 132 days showing TWIST expression throughout this developmental time frame. TWIST expression in temporalis muscle (Ad Muscle) and rhabdomyosarcoma (Rhabdo) is consistent with the known expression of TWIST in adult muscle progenitor cells [36] and the mesenchymal-derived tumor [10]. TWIST is differentially expressed in adult brain, with expression detected in gray matter (GM) but at low or undetectable levels in white matter (WM). (B and F) H&E histologic sections (x20) of normal gray matter (B) and white matter (F) are compared with ISH (x20) for TWIST expression from adjacent sections of gray (C) and white matter (G). Insets in the right lower corner (C and G) are sense controls. TWIST IHC (x20) for normal gray matter (D) and white (H) matter corresponds to expression patterns by ISH. Inset: x40 images in (D) and (H) demonstrate nuclear staining pattern in cells with neuronal morphology. (E and I) The demarcation of TWIST protein expression between normal gray and white matter is demonstrated in low-power (x10) photomicrographs of a normal brain sample with a diagonally oriented junction between gray (upper right) and white matter (lower left) by H&E histology (E) and TWIST IHC (I). These data indicate that TWIST message and protein are almost exclusively detected in gray matter. (The 50-µm scale bar in (B) applies to (B)–(D) and (F)–(H); the 25-µm scale bar in the (D) inset applies to insets of (D) and (H); and the 100-µm scale bar in (E) applies to (E) and (I).)

TWIST is expressed almost exclusively in neurons of the adult brain. (A–C) Photomicrographs of indirect immunofluorescent staining with TWIST antibody (Cy3) of normal human gray matter (A), the neuronal marker NeuN (FITC) (B), and the merged image (C) show colocalization of TWIST with NeuN in almost every TWIST immunoreactive cell (arrow). Sections were also stained with DAPI and indicate that NeuN-negative cells do not express TWIST. (D–F) A section of normal human brain containing gray and white matter stained for TWIST (Cy3) (D) and GFAP (FITC) (E), and the merged image with DAPI nuclear stain (F) demonstrate that TWIST is found exclusively in gray matter and that TWIST-expressing nuclei (arrowhead) do not colocalize with GFAP-immunopositive cells (arrow). The junction between gray and white matter is indicated by the dotted line. These results demonstrate that TWIST is almost exclusively found in neurons. (The 50-µm scale bar in (A) applies to (A)–(F).)