The majority of human cancers show high levels of telomerase activity that is thought to contribute to tumor progression either because telomerase-dependent telomere maintenance provides cells with an extended proliferative potential,1 or because telomerase has undefined growth-promoting functions.2, 3 Although it is widely recognized that the outcome of telomerase inhibition may strongly depend on the cellular context, inhibition of telomerase is considered a promising therapeutic approach for neoplasia.4 In most normal somatic human cells, telomerase is downregulated mainly because of transcriptional silencing of its catalytic subunit, human telomerase reverse transcriptase (hTERT). This results in a progressive telomere shortening with repeated cell duplications. The ensuing telomere dysfunctions lead to cell senescence of normal human fibroblasts.5 In other cellular contexts, telomerase inhibition6, 7 and/or telomere dysfunctions may lead to apoptosis.8
Gliomas, the most common human brain tumors, are classified into 4 grades, according to their aggressiveness. Less-malignant gliomas are defined as grade I and II, whereas grade III (anaplastic) and grade IV (glioblastoma multiforme, GBM) astrocytomas are highly aggressive tumors. With respect to telomerase expression, astrocytomas present considerable variability, with grade I and grade II astrocytoma generally being telomerase negative, and grade III astrocytoma being telomerase positive in about 30% of the cases.9, 10 The percentage of GBM tumors that have been shown to express telomerase enzyme activity varies between 26% and 67%.11, 12, 13, 14, 15 This variability probably reflects the highly heterogeneous nature of GBM. In fact, where sampling of multiple regions from the same tumor or microdissection for selective analysis of tumor cells were used, 100% of GBM were found to show telomerase activity.16, 17 Consistent with these observations, we previously demonstrated, using the in situ hybridization technique, that all GBMs express hTERT mRNA, although with variability among different cells within the same tumor.15 Therefore, it is suspected that telomerase might contribute to the aggressive behavior of GBM tumors. To get insights into the role of telomerase in GBM growth, we used short interfering RNA (siRNA) directed against hTERT to downregulate the expression of telomerase in 2 established cell lines of GBM, TB10 and U87MG.18, 19 Here, we show that telomerase inhibition impaired the growth of GBM tumor cells in vivo both after subcutaneous and after intracranial grafting in nude mice and that this effect was not related to critical telomere shortening. Telomerase downregulation in GBM cells also affected the histological phenotype of the tumor xenograft, as demonstrated both by the absence of massive necrosis and by reduced angiogenesis.
GBM, glioblastoma multiforme; GFAP, glial fibrillary acid protein; hTERT, human telomerase reverse transcriptase; PD, population doubling; RNAi, RNA interference; sh, short hairpin; si, short interference; TRAP, telomeric repeat amplification protocol; TRF, telomeric restriction fragment; VEGF, vascular endothelial growth factor.
Material and methods
Culture of tumor cell lines
All cell lines were grown in DMEM (high glucose, Invitrogen Italia, Milano, Italy) supplemented with 10% heat-inactivated fetal bovine serum. TB10 cell line was established in our laboratory from a human secondary GBM tumor.18 U87MG was derived from a de novo GBM.19 Immunohistochemistry for GFAP was routinely performed to check maintenance of the glial phenotype. The cells were regularly tested for mycoplasma contamination. Growth rate was assessed as follows: 105 TB10 cells were seeded in triplicate in 60-mm dishes, and after 48 hr, cells were collected and counted in duplicate. Convenient dilution was used for successive rounds of seeding. The total cell number as a function of time was calculated relative to the dilution factor.
Production of retroviral stocks and cell infection
The p-RETRO-SUPER (pRS) vector was kindly provided by Dr. Agami (The Netherland Cancer Institute, Amsterdam) and was used as described previously.20 The following 19-bp sequences within the hTERT cDNA were chosen as target for RNA interference (RNAi): si1 5′-GGAGCAGCTGCGGCCCTCC-3′ at position 1013–1031, following the ATG, and si2 5′-GAACGTGCTGGCCTTCGGC-3′ at position 282–300.
Control retrovirus contained the following 19-bp sequence 5′-TGCTAAGCGCCAGCAAGAG-3′, which is not related to hTERT. This sequence is derived from rat VGF, a neuronal-specific mRNA that is not expressed in glial cells.21 Amphotropic retroviruses were generated in Phoenix packaging cells upon transfection of the retroviral vectors by lipofectamine reagent (Invitrogen Italia), according to the manufacturer's instructions. Forty-eight hours post transfection, the viral supernatant was used to infect GBM cells after addition of 4 μg/ml polybrene. GBM cells were infected twice for 6 hr and, after 48 hr, selected with puromycin. PD 0 was conventionally defined as when the selected GBM cells reached confluency.
RT-PCR was performed as previously described.22 Primers for hTERT were as follows: F-TERT/3280 5′-ACCAAGCATTCCTGCTCAAGCTG-3′ and R-TERT/3685 5′-CGGCAGGTGTGCTGGACACTC-3′. Primers for G3PDH, used as an internal control, were: F-G3PDH/4205 5′-ACCACAGTCCATGCCATCAC-3′ and R-G3PDH/4762 5′-TCCACCACCCTGTTGCTGTA-3′. hTERT cDNA was amplified by 35 cycles with the following settings: 94°C for 30 sec; 65°C for 45 sec; 72°C for 30 sec. G3PDH cDNA was amplified by 18 cycles with the following settings: 94°C for 30 sec; 60°C for 45 sec; 72°C for 30 sec.
Telomeric repeat amplification protocol assay
A modified version of telomeric repeat amplification protocol (TRAP) was used to detect telomerase activity.23 Briefly, we used newly designed primers, which allow an increased sensitivity and reproducibility of the assay, by avoiding direct primers interaction. Two negative controls were routinely performed: RNase pretreatment of the protein extract and no extract added to the reaction mix.
Telomere restriction fragments analysis
Three micrograms of genomic DNA was digested to completion with RsaI and HinfI, separated on a 0.7% Tris–Acetate–EDTA agarose gel, transferred to Hybond-N+ membranes (Amersham Bioscience Europe GmbH, Milano, Italy) and hybridized with a 32P-labelled (TTAGGG)3 telomeric probe.
Subcutaneous implantation of TB10 and U87MG glioblastoma tumor cells
Experiments involving animals were approved by the Ethical Committee of the Catholic University School of Medicine, Rome. Nude athymic mice (4–6 weeks of age; HDS-athymic nude mice, Harlan, Udine, Italy) were implanted subcutaneously either with 106 TB10 or with 106 U87MG cells. Tumor cells were grafted at PD 4–5 postinfection. For grafting, cells were resuspended in 0.1 ml of cold phosphate buffered saline and the suspension was mixed with an equal volume of cold Matrigel (BD Bioscience, Bedford, MA). The animals were kept under pathogen-free conditions in positive-pressure cabinets (Tecniplast Gazzada, Varese, Italy) and observed daily for the visual appearance of tumors at injection sites. Tumor growth was measured weekly using calipers. Tumor diameter was calculated as the mean value between the shortest and longest diameter. Mice that did not develop a subcutaneous mass were excluded in calculating tumor diameter. At intervals from 3 to 10 weeks after implantation, the mice were sacrificed with an overdose of barbiturate. Tumors were harvested and processed as described later.
Intracranial implantation of U87MG glioblastoma tumor cells
Nude athymic mice (4–6 weeks of age; HDS-athymic nude mice, Harlan) were implanted intracranially with 0.5 × 106 U87MG cells resuspended in 4 μl of serum-free DMEM. Cells were grafted at PD 4–5 postinfection. For grafting, the mice were anesthetized with intraperitoneal injection of diazepam (2 mg/100 g) followed by intramuscular injection of ketamine (4 mg/100 g). The animal skulls were immobilized in a stereotactic head frame and a burr hole was made 2 mm right of the midline and 1 mm behind the coronal suture. The tip of a 10-μl Hamilton microsyringe was placed at a depth of 3 mm from the dura and the cells were slowly injected. The mice were sacrificed with an overdose of barbiturate either after 8 or after 14 days of survival. The brains were removed, fixed in 4% paraformaldehyde, embedded in paraffin, cut into 5-μm thick sections and stained with H&E. Tumor volumes were calculated on histological section through the tumor epicenter, according to the equation: V = (a2 × b)/2, where a is the shortest diameter and b is the longest diameter of tumors.
Histological and immunohistochemical analysis
The proliferative potential of glioblastoma tumor cells was assessed both by determining the mitotic index (MI) on H&E-stained sections and by using the Ki67 monoclonal antibody, MIB-1 (Ylem, Avezzano, Italy), which is exclusively expressed in the nucleus of proliferating cells. The expression of GFAP and CD31, which are phenotypic markers of astrocytic and endothelial cells, respectively, was assessed with the monoclonal antibodies anti-GFAP (Dako, Glostrup, Denmark) and anti-CD31 (Santa Cruz Biotechnology, Santa Cruz, CA). The expression of vascular endothelial growth factor (VEGF) in the tumor xenografts was analyzed with the polyclonal antibody anti-VEGF (Santa Cruz Biotechnology). Immunohistochemistry was performed on deparaffinized sections, using the avidin–biotin–peroxidase complex method (ABC-Elite kit, Vector Laboratories, Burlingame, CA). Apoptosis was assessed by TUNEL staining, using the In Situ Cell Death Detection Kit, POD (Roche Diagnostics, Monza, Italy). The MI, MIB-1 staining index and apoptotic index were determined as the percentage of mitosis, Ki67-positive and TUNEL-positive cells, respectively, relative to the total number of cells in high-power fields (400×). In each tumor specimen, ∼2,500 tumor cells were counted. The number of neoformed vascular structures was assessed in CD31-immunostained coronal sections of xenografted brains through the tumor epicenter.
Differences in the rate of tumor appearance, in diameter and volume of tumor xenografts, in MI, percentage of Ki67- and TUNEL-positive tumor cells and in the number of neoformed vascular structures between the experimental and control groups were evaluated using the χ2 test and Student's t-test. Statistical significance was assigned to p values <0.05.
siRNA directed against hTERT mRNA inhibit telomerase activity but do not affect TB10 glioblastoma cells growth in vitro
The p-RETRO-SUPER vector allows stable transduction of short hairpin RNAs (shRNAs) that are processed to siRNA in mammalian cells.20 Oligonucleotides encoding for shRNAs complementary to 2 distinct regions of hTERT were cloned in p-RETRO-SUPER and the corresponding retroviruses, named si1-hTERT and si2-hTERT, were used to infect TB10 GBM cells. After selection with puromycin, the surviving polyclonal populations, si1-hTERT TB10 and si2-hTERT TB10, were examined for inhibition of hTERT mRNA expression by RT-PCR. As shown in Figure 1a (lanes 2 and 3), there was a considerable reduction of hTERT mRNA in the TB10 cells expressing either si1-hTERT or si2-hTERT, as compared to the control si-VGF TB10 cells (lane 1). The inhibition of hTERT mRNA expression was stably maintained by the cells, since it was still observed after several cell-division cycles (Fig. 1a, lanes 4 and 5). Some studies raise concern about a nonspecific “interferon response” of cells to siRNA induced by some short duplex RNAs.18, 19, 24, 25 Expression of 2 interferon inducible genes of the oligoadenylate synthase family, OAS1 and OAS2, was thus measured in si-hTERT TB10 cells by RT-PCR. No variation of the mRNA for these genes was observed (data not shown).
Reduction of hTERT mRNA resulted in a decrease in enzymatic activity, as assessed by TRAP (Fig. 1b). As for hTERT downregulation, telomerase inhibition was stably maintained (Fig. 1b). Telomerase downregulation in turn caused shortening of telomeres with increasing PDs, as measured by telomere restriction fragments (TRF) analysis. The mean telomere length was appreciably reduced in si1-hTERT TB10 cells at PD 90 (Fig. 1c, lane 5), while si2-hTERT TB10 displayed more limited telomere attrition (Fig. 1c, lane 7). To assess whether reduction of telomerase activity adversely affected the growth of TB10 cells, we compared the rate of cell division of si-VGF TB10 and of si-hTERT TB10 cells. Both si1-hTERT and si2-hTERT TB10 cells, at early passages postinfection, grew with the same rate as the control si-VGF TB10 cells (not shown). In addition, in spite of measurable shortening of telomeres, late passage si-hTERT TB10 cells displayed the same growth rate as si-VGF TB10 cells (Fig. 1d).
Telomerase inhibition impairs subcutaneous growth of TB10 cells in nude mice
We next assessed the role of telomerase on glioblastoma growth in vivo using a standard assay for tumorigenic potential, in which logarithmically growing tumor cells are injected subcutaneously into athymic nude mice. Three groups of mice were injected with either si1-hTERT TB10 cells (n = 12), or with si2-hTERT TB10 cells (n = 9), or with control si-VGF TB10 cells (n = 15). After 6 weeks, there was a significant difference in the rate of tumor appearance both between si1-hTERT TB10 and siVGF-TB10 groups, and between si2-hTERT TB10 and siVGF-TB10 groups (p comparing si1-hTERT TB10 to si-VGF TB10 <0.0001; p comparing si2-hTERT TB10 to si-VGF TB10 <0.001; χ2 test) (Fig. 2a). Moreover, both si1-hTERT TB10 and si2-hTERT TB10 glioblastomas grew significantly slower than did the control tumors (Fig. 2b). Six weeks after implantation, tumor diameters were 4 ± 1.4 mm (mean ± SD), 7.3 ± 1.7 mm and 14.2 ± 1.8 mm for si1-hTERT TB10, si2-hTER TB10 and control si-VGF TB10, respectively (p comparing si1-hTERT TB10 to siVGF TB10 <0.0001; p comparing si2-hTERT TB10 to si-VGF TB10 <0.0001; t-test). There was no difference in body weight between the different groups. Histological examination revealed that control si-VGF TB10 tumors retained one main feature of human glioblastomas, that is, massive necrosis (Fig. 3a). Conversely, si-hTERT tumors showed either small foci of necrosis or no necrosis at all (Fig. 3f). This finding was not merely related to tumor size because si-hTERT TB10 tumors that grew as long as 10 weeks (n = 8) and that eventually reached the size of control tumors, were still devoid of necrosis (not shown). Cell proliferation was significantly lowered in si-hTERT TB10 tumors (Table I and Figs. 3c and 3h). TUNEL-positive apoptosis was significantly higher in si-hTERT TB10 tumors than in the control si-VGF ones (Table I and Figs. 3d and 3i). In control si-VGF TB10 tumors, immunohistochemistry with anti-VEGF antibody showed diffuse staining by the tumor cells that increased in perinecrotic areas, including the small foci of necrosis (Fig. 3e). In si-hTERT TB10 tumors, the enhancement of VEGF expression by the tumor cells that surrounded the small foci of necrosis was much less remarkable (Fig. 3l). There was no difference in the percentage of tumor cells expressing GFAP among the 3 groups of subcutaneous TB10 tumor xenografts.
Table I. Summary of Results on Tumor Cell Proliferation and Apoptosis in Glioblastoma Xenografts
Values are expressed as means ± SD. p values were calculated using the Student's t-test.
p comparing si1-hTERT TB10 to si-VGF TB10 <0.001.
p comparing si2-hTERT TB10 to si-VGF TB10 <0.001.
p comparing si2-hTERT TB10 to si-VGF TB10 <0.01.
p comparing si1-hTERT U87MG to si-VGF U87MG n.s.
p comparing si1-hTERT U87MG to si-VGF U87MG <0.02.
p comparing si1-hTERT U87MG to si-VGF U87MG <0.001.
p comparing si1-hTERT U87MG to si-VGF U87MG <0.05.
p comparing si1-hTERT U87MG to si-VGF U87MG <0.01.
Telomerase inhibition impairs subcutaneous growth of U87MG cells in nude mice
To validate these results, we generated a second glioblastoma cell line, U87MG, with reduced telomerase expression, using the si1 retrovirus. The si2 retrovirus was not used because of its lower efficiency in causing telomere attrition. Again, the amount of hTERT mRNA and the level of telomerase activity were significantly reduced by si-hTERT RNA, although not completely turned off (Figs. 4a and 4b). Telomere lengths of si1-hTERT U87MG cells were essentially unaltered, as assessed by TRF analysis (Fig. 4c). As in the case of TB10 cells, si1-hTERT U87MG cells and control si-VGF cells displayed the same growth rate in culture (not shown). In vivo, U87MG cells show high tumorigenic potential, generating subcutaneous tumors in nude mice that grow much faster than the TB10 xenografts. All the mice injected with U87MG cells, both control si-VGF and si1-hTERT, developed subcutaneous tumors (Fig. 5a). After 3 weeks, however, tumor diameters were significantly smaller in mice injected with si1-hTERT U87MG cells (n = 8) as compared with controls (n = 9) (Fig. 5b). Tumor diameters were 12 ± 2.7 mm and 17 ± 1.4 mm in si1-hTERT U87MG and in si-VGF U87MG groups, respectively (p < 0.01, t-test). There was no difference in body weight between the groups. In U87MG subcutaneous xenografts, lowering hTERT by siRNA resulted in much less inhibition of cell proliferation, as compared with si-hTERT TB10-derived tumors (Table I and Figs. 6a and 6c). However, apoptosis was significantly increased by si1-hTERT (Table I and Figs. 6b and 6d). There was no difference in the percentage of tumor cells expressing GFAP between the 2 groups of U87MG-derived tumors.
Telomerase inhibition impairs tumor growth and reduce angiogenesis in orthotopic xenografts of U87MG cells
Since the implantation site may affect the growth features of astrocytoma cells after grafting in nude mice,26 we studied whether the intracerebral growth of GBM might also be impaired by telomerase inhibition. In this experiment, we used U87MG cells which grow aggressively upon intracranial grafting in nude mice, while TB10 cells develop intracranial tumors at a low frequency. Either control si-VGF U87MG or si1-hTERT U87MG cells were implanted into the right striatum of athymic mice. There was no significant difference in brain tumor appearance, since tumors were detected in 8/9 and in 11/11 mice inoculated with si1-hTERT U87MG cells and with si-VGF U87MG cells, respectively. However, telomerase inhibition did affect tumor growth rate, since tumor volumes were significantly smaller in si1-hTERT tumors as compared with control ones (Figs. 7 and 8). Eight days after implantation, tumor volumes were 83.8 ± 30.8 × 106 μm3 and (432.9 ± 212.3) × 106 μm3 in si1-hTERT U87MG and in si-VGF U87MG, respectively (p = 0.012, t-test) (Fig. 8). Fourteen days after implantation, tumor volumes were (165.9 ± 44.6) × 106 μm3 and 866.2 ± 287.3 × 106 μm3 in si1-hTERT U87MG and in si-VGF U87MG, respectively (p = 0.002, t-test) (Fig. 8). Tumor cell proliferation was significantly lowered in si1-hTERT U87MG brain xenografts as compared with control si-VGF U87MG brain tumors (Table I and Figs. 7c and 7g). Apoptosis was significantly higher in the si1-hTERT U87MG brain xenografts (Table I and Figs. 7d and 7h). There was no difference in the percentage of tumor cells expressing GFAP between the 2 groups. We then quantified angiogenesis by counting 2 types of neoformed vascular structures, which can readily be assessed in U87MG brain xenografts, (a) small thin-walled blood vessels, mainly capillaries, that were directed towards the tumor, here named as afferent vessels and (b) convoluted glomerular tufts that were typically seen at the periphery or even beyond the actual confines of the tumor, defined as vascular glomeruli (Figs. 8 and 9). Both types of vascular structures were significantly reduced in si1-hTERT intracranial tumors as compared with controls; however, this effect was much more dramatic for the vascular glomeruli. The number of vascular glomeruli per section was 3.2 ± 1.7 and 15.6 ± 2.1 in si1-hTERT and in control si-VGF tumors, respectively (p < 0.0001, t-test). The number of afferent vessels was 6.2 ± 1.6 and 10.7 ± 3.4 in si1-hTERT and in control si-VGF tumors, respectively (p < 0.01, t-test).
In the present work, telomerase was inhibited in 2 GBM cell lines, TB10 and U87MG, via retroviral transduction of siRNA targeting hTERT mRNA. The transcriptional and functional effects were verified by RT-PCR, TRAP and TRF assays. Although telomerase inhibition had no antiproliferative effects in vitro, GBM cells with impaired telomerase expression exhibited a rapid reduction of their tumorigenic capacity in vivo, as demonstrated by the development of fewer and smaller tumors, following subcutaneous and intracranial xenografting in nude mice. As compared with controls, si-hTERT expressing GBM tumors showed decreased cell proliferation, more apoptosis and reduced angiogenetic phenomena. Telomerase inhibition also affected the histological phenotype because si-hTERT tumors lacked massive necrosis with only small foci of necrosis or no necrosis at all.
It has generally been acknowledged that telomerase inhibition may hinder the growth of cancer cells. More specifically, downregulation of hTERT has been shown to induce both a decline in cell proliferation and apoptotic cell death, in several human cancers.27, 28, 29 Since GBMs express hTERT and exhibit telomerase activity,10, 15, 16, 17 it is reasonable to expect that GBMs behave similar as other malignancies. Therefore, the present in vivo results are in some way expected and confirm previously published data showing that telomerase inhibition exerts antitumor effects on human GBM xenografts.30 However, our observations that hTERT downregulation by RNAi differentially affected the growth of glioblastoma cells in vitro with respect to the in vivo condition, and that the inhibition of tumor growth in vivo occurred even before any detectable telomere shortening represent novel findings that might be important for understanding mechanisms of telomerase inhibition-mediated glioma therapy.
In our experiment, the population of retrovirally infected cells was polyclonal and, therefore, heterogeneous with respect to the degree of telomerase inhibition. The presence of a mixed cell population with some telomerase-positive cells might explain why siRNA did not reduce telomere length appreciably and did not affect cell proliferation in vitro. Nevertheless, the level of telomerase inhibition, as assessed by RT-PCR and TRAP, did not change between early passage and late passage si-hTERT cells, suggesting that in vitro there was no selection for a subpopulation of cells. In fact, where the group of si-hTERT cells with higher residual telomerase activity had any growth advantage in vitro, we would expect a progressive increase of mRNA and enzyme activity with repeated cell divisions. On the other hand, it is clear that our siRNA knocked down the expression of telomerase, though it failed to completely abrogate the activity of the enzyme. The remaining basal activity might be sufficient to preserve telomere lengths, thus allowing the unrestrained growth of tumor cells in vitro.
Other in vitro studies on the telomerase inhibitors found no immediate antiproliferative effect on the growth of cancer cells, consistent with the hypothesis that telomeres must shorten significantly to cause reduced cell growth.31, 32 However, Li et al. reported rapid cell growth impairment both in vitro and in a mouse xenograft tumor model after telomerase inhibition in several telomerase-positive human cancer cell lines.33, 34 This effect was noted at early passages when no bulk telomere shortening had occurred. The authors postulate a mechanism of telomere uncapping that would induce DNA damage independently of telomere shortening.33 They also found that telomerase inhibition by siRNA rapidly led to suppression of specific genes involved in angiogenesis and metastasis.34 The concept of telomere dysfunction that would result in an antiproliferative effect prior to telomere shortening may well apply to our in vivo results, particularly to the brain xenograft paradigm where the inhibition of tumor growth was observed within such a short time that any substantial telomere shortening seems highly unlikely. One obvious experiment to definitively demonstrate that the in vivo effects on tumor growth are independent of telomere shortening would be to assess telomere length directly in the tumor xenograft. Such an experiment, however, is made extremely difficult because of the presence of recruited normal murine cells, like fibroblasts and endothelial cells, that have long telomeres in the tumor xenograft.
The issue that telomerase inhibition differentially affected GBM growth in vitro versus in vivo may be explained invoking noncanonical roles for telomerase.35 Increasing pieces of evidence suggest that telomerase promotes cancer growth through mechanisms which are independent of telomere maintenance and that are typical of the in vivo condition.36, 37 Angiogenesis, which plays a major role in glioblastoma tumor growth, might be one of such mechanisms. Previously, we hypothesized a role for telomerase in the angiogenesis of malignant astrocytic tumors.22, 38 Although later reports have noted that telomerase-deficient mice develop tumor xenografts with reduced angiogenesis39 and that lowering telomerase levels in tumors led to diminished neovessel formation,33 to date, this issue has not specifically been addressed. To quantify angiogenesis in our orthotopic xenograft model, we counted the vessels afferent to the tumor and the vascular glomeruli at the periphery of tumor. Though both types of vascular structures were decreased in si-hTERT tumors, this phenomenon was much more remarkable for the vascular glomeruli. According to recent views, functional angiogenesis results from a balance between the guided migration of endothelial cells at the tip of vascular sprouts (tip cells) and the proliferating pressure of endothelial cells at the sprout stalks (stalk cells).40 In tumor angiogenesis, this coordinated mechanism is disrupted with formation of chaotic and dysfunctional networks.41, 42 The convoluted vascular glomeruli, which are typical of GBM, might nicely represent such a dysfunction, where the proliferating stalk cells overwhelm any oriented growth of the sprouting vessel. Consistent with this concept, the reduction in vascular glomeruli that we observed in si-hTERT tumors would result from the decreased proliferation of the endothelial stalk cells, and this might be related to the lowered telomerase levels by the tumor cells. Although we have no direct evidence of any bystander effect from the si-hTERT human tumor cells to the murine endothelial cells that provide the vascular support to the xenograft, the hypothesis that telomerase depletion in the glial tumor cells might in turn affect the angiogenetic behavior of the endothelial cells is suggested by several lines of evidence. For example, exposure of human endothelial cells to GBM tumor cells in co-culture systems induces an activated phenotype of the endothelial cells with upregulation of genes related to cell proliferation, angiogenesis, protection from apoptosis and telomerase expression.22, 43, 44 In subcutaneous murine grafts, human endothelial cells transduced to overexpress telomerase participate in murine angiogenesis, while their wild-type counterparts do not survive.45 In human astrocytic tumors, a relationship has been demonstrated between the level of hTERT expression by the glial tumor cells and the proliferation rate of the endothelial cells of the tumor vasculature.38
A mechanism that might nicely explain why telomerase inhibition in the tumor cells would affect new-vessel structure formation is “vasculogenic mimicry.” This term refers to the phenomenon whereby highly angiogenetic malignancies, like glioblastomas and melanomas, generate nonendothelial vascular channels lined by tumor cells that inappropriately express endothelial antigenic markers.46, 47 Interestingly, vascular mimicry has also been observed in network structures composed by several back-to-back loops that emulate the vascular glomeruli of GBM.47
There may be other explanations of why telomerase inhibition differentially affected tumor growth in vitro versus in vivo. For example, telomeres might shorten with an increased rate in vivo because cells are subjected to oxidative stress, and cells with impaired telomerase functioning are more sensitive to this condition.48 It is known that reactive oxygen species are elevated within cells stimulated to divide in nutrient-poor conditions, like the subcutaneous tissue.26, 49
Another major result of the present work is that telomerase inhibition modified the histological phenotype of xenografted glioblastomas. In fact, control tumors showed extensive areas of necrosis while their interfered counterparts had only small foci of necrosis or no necrosis at all. The smaller size of si-hTERT tumors cannot account for such a difference in necrosis, because in si-hTERT tumors which grew longer and eventually reached the size of the control tumors, still necrosis was negligible. As expected, the amount of necrosis affected the degree of VEGF expression in the tumor xenografts. Necrosis is one main histological landmark for GBM, which distinguishes this neoplasm from anaplastic astrocytoma. Tumorigenesis of human brain gliomas is thought to be a multistep process where several genetic changes have been recognized that include impairment of p53 tumor suppressor pathway, activation of tyrosine kinase receptor signaling cascades and up regulation of ras-mediated functions.50, 51, 52 Eventually, conversion of low-grade to high-grade astrocytomas requires inactivation of the retinoblastoma tumor suppressor pathway that coincides with strong increase in mitotic activity and reactivation of telomerase.12, 53, 54In vitro studies on human astrocytic cells have shown that the aforementioned events are all needed so as to obtain cells that are able to form tumor xenografts with areas of necrosis that closely resemble human GBM.55 Should the sequence of genetic changes be incomplete, then these cells form tumors devoid of necrosis reminiscent of anaplastic astrocytoma.56 Here, we found a specular phenomenon, that is, inhibiting telomerase in fully transformed GBM cells converts these cells, which are per se able to form GBM tumors with areas of necrosis in nude mice, into cells that form tumor xenografts resembling anaplastic astrocytoma where necrosis was absent.
In conclusion, our study suggests that focusing solely on tumor cell proliferation in vitro may underestimate the value of telomerase inhibition, because significant reduction in tumor growth does occur in vivo through mechanisms that are unique of this condition.