| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Institute of Neurobiology and Molecular Medicine, Consiglio Nazionale Ricerche, Fondazione S. Lucia, Rome, Italy;
Biotechnology Unit, ENEA CR-Casaccia, Rome, Italy; and
Department of Experimental Medicine and Pathology, University La Sapienza, Rome, Italy
2Correspondence: Institute of Neurobiology and Molecular Medicine, Consiglio Nazionale Ricerche, Via Fosso di Fiorano 64, 00143, Rome, Italy. E-mail: tirone{at}inmm.cnr.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
40% and in a marked reduction of preneoplastic abnormalities, such as hyperplastic EGL areas and lesions. Moreover, overexpression of cyclin D1, hyperproliferation, and defective differentiation—observed in Ptc+/– GCPs—are restored to normality in Ptc+/–/TgPC3 mice. The PC3-mediated inhibition of cyclin D1 expression correlates with recruitment of PC3 to the cyclin D1 promoter, which is accompanied by histone deacetylation. Remarkably, down-regulation of PC3 is observed in preneoplastic lesions, as well as in human and murine medulloblastomas. As a whole, this indicates that PC3 may prevent medulloblastoma development by controlling cell cycle and promoting differentiation of GCPs.—Farioli-Vecchioli, S., Tanori, M., Micheli, L., Mancuso, M., Leonardi, L., Saran, A., Ciotti, M. T., Ferretti, E., Gulino, A., Pazzaglia, S., Tirone, F. Inhibition of medulloblastoma tumorigenesis by the antiproliferative and pro-differentiative gene PC3.
Key Words: tumor suppressor gene • neurogenesis • cyclin D1 • Math1
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. GCPs migrate over the surface of the developing cerebellum to form the external granular layer (EGL), where they continue to proliferate perinatally. A failure in the control of proliferation of cerebellar precursors is thought to give rise to medulloblastoma, the most common brain tumor in childhood (2)
.
Sonic hedgehog (Shh) plays a major role in the development of the cerebellum by potently stimulating the proliferation of GCPs in the EGL (3
4
5)
. The crucial role of Shh in GCP proliferation is exerted through N-Myc and cyclin D1, which are among the genes most induced by Shh (6
7
8
9
10)
.
Components of the Shh pathway are frequently altered in medulloblastoma, being mutated or overexpressed. Inherited mutations in the Patched1 gene—encoding the Shh receptor, which inhibits the Shh pathway in the absence of the ligand—lead to the development of Gorlin’s syndrome, characterized by skeletal defects and predisposition to basal cell carcinomas, rhabdomyosarcoma, and a significant incidence of medulloblastoma (11)
. Moreover, the inactivation of Patched1 by deletion or mutation characterizes 10–25% of sporadic medulloblastomas, suggesting that Patched1 functions as a tumor suppressor (12
13
14)
. The subset of medulloblastomas with overactivation of the Shh pathway display induction of the Shh downstream targets Gli1 and N-Myc, along with induction of markers of GCPs (15
, 16)
. Thus, medulloblastomas characterized by a deregulation of the Shh/Patched pathway very likely originate from GCPs.
A key contribution to the understanding of the molecular mechanisms of medulloblastoma tumorigenesis comes from mouse models [for review, see (2)
]. Inactivation of one Patched1 allele leads to the development of spontaneous medulloblastomas, which confirms the key role of the Shh pathway in the etiology of the tumor (17
, 18)
. Patched1 heterozygous mice survive to adulthood, but 8–30% develop medulloblastoma after 2–6 months of age, and more than 50% present ectopic EGL regions, indicative of a preneoplastic condition (17
, 19
, 20)
. The frequency of ectopic EGL and of medulloblastoma is dramatically increased by radiation exposure of Patched1 heterozygous mice immediately after birth, when EGL precursors are actively proliferating (21
, 22)
.
We have recently observed that the PC3 gene plays a role in the control of GCP differentiation. PC3, which we originally isolated as a gene induced at the onset of the neuronal differentiation elicited by nerve growth factor (23)
, possesses antiproliferative properties and is expressed in a restricted number of tissues and in neuronal precursors undergoing a neurogenic asymmetric division (23
24
25
26
27)
. During neurogenesis, PC3 accelerates the shift of neural precursors from proliferation to differentiation, thereby promoting the generation of new neurons (28)
. The marked action exerted by PC3 in GCPs is dual: delay of G1 phase progression through inhibition of cyclin D1 and induction of Math1 (28)
, a gene required for maturation and differentiation of GCPs (29)
. We proposed that these actions of PC3 synergize to effect in GCPs terminal cell cycle exit and differentiation.
Here, we tested the hypothesis that the antiproliferative/prodifferentiative effects of PC3 in GCPs may counteract the development of medulloblastoma in Patched1 heterozygous mice. In parallel, we investigated the physiological implication of PC3 in murine and human medulloblastoma.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, exploiting the tet-off system to control the activation of the PC3 transgene (30)
, and were maintained in a BDF1 (C57BL/6 x DBA/2) background. In bitransgenic mice the ßACT promoter drives the transcription of the tTA TransActivator (the second transgene), which in turn activates TRE-PC3 by binding the upstream TREs. Doxycycline inactivates tTA, consequently repressing the expression of TRE-PC3.
Mice lacking one Patched1 allele, generated through disruption of exons 6 and 7 in 129/SV ES cells and maintained in a CD1 background (18)
, were crossed for this study to the binary transgenic ßACT-tTA/TRE-PC3. The mouse lines and the F1 progeny resulting from crossings were genotyped using primers specific to the tTA and to the TRE regions for ßACT-tTA/TRE-PC3 mice and to the neo insert and wt regions for heterozygous Patched1 knockout mice, as described (28
, 18)
.
Animal treatment and irradiation
Animals were housed under conventional conditions with a 12 h light–dark schedule. Embryonic day 1 (E1) was considered completed at midnight of the day after mating. Brains collected from postnatal day 14 (P14) mice were fixed and stored as described (28)
. Doxycycline hydrochloride (150 µg/ml; MP Biomedicals, Eschwege, Germany) was supplied to mice in the drinking water supplemented with 5% sucrose, from the day of mating to day 14 of pregnancy, after which it was removed. This allowed the expression of transgenic PC3 after about E18, given the time necessary to metabolize doxycycline.
Mice were irradiated at P1 with with an X-ray dose of 3 Gy using a Gilardoni CHF 320 G X-ray generator (Gilardoni S.p.A., Mandello del Lario, Italy) operated at 250 kVp, 15 mA, with filters of 2.0 mm Al and 0.5 mm Cu (HVL=1.6 mm Cu).
Tumor quantification and histological analysis
Mice were observed daily for their whole life span. On decline of health (i.e., severe weight loss, paralysis, ruffling of fur, or inactivity), they were sacrificed and autopsied. Brains were fixed in 4% buffered formalin. Samples were then embedded in paraffin wax according to standard techniques, sectioned serially and stained with hematoxylin/eosin.
Immunohistochemistry, antibodies, BrdU labeling, and TUNEL analysis
Immunohistochemistry was performed on serial sections with 10 µm thickness of brains from P14 mice using mouse monoclonal antibodies raised against cyclin D1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA; clone 72–13G, 1:75), NeuN (Chemicon International, Temecula, CA, USA; MAB377; 1:100) and phospho-histone H2A.X (Upstate Biotechnology, Lake Placid, NY, USA; clone JBW301; 1:100), or using rabbit polyclonals that recognize cyclin D2 (Santa Cruz, SC-181; 1:100), cyclin B1 (Santa Cruz, SC-752; 1:200), NeuroD1 (R&D Systems, Minneapolis, MN, USA; AF2746; 1:100), and PC3 (A3H, 1:50) (24)
. Antiphospho-histone H2A.X marks DNA repair foci, being histone H2A.X rapidly phosphorylated following DNA double-strand breaks. A3H antibody recognizes both the transgenic PC3 and the endogenous mouse PC3 (i.e., Tis21) proteins. Binding of primary antibodies was revealed using FITC-conjugated goat anti-mouse or TRITC-conjugated goat anti-rabbit secondary antibodies (Jackson ImmunoResearch, West Grove, PA, USA; 1:100 and 1:200, respectively).
Analysis of BrdU incorporation in the EGL was performed using P14 mice injected with BrdU (90 mg/kg, i.p.) 1.5 h before sacrifice. This period of incorporation was judged appropriate to label mainly S phase cells, based on the duration of the S and G2/M phases in GCPs of mice at this age (31)
. Sections were permeabilized and stained with mouse monoclonal anti-BrdU (Amersham Biosciences, Arlington Heights, IL, USA) as described (28)
.
Cells positive for BrdU incorporation or for expression of other markers were calculated as percent ratio of the number of labeled cells to the total number of cells (visualized by counterstaining nuclei with Hoechst 33258 (1 µg/ml in PBS, Sigma, St. Louis, MO, USA), for the entire length of the EGL in each photomicrograph field. Nuclei with condensed and fragmented chromatin were considered apoptotic and were not counted.
Apoptotis was measured on sections by TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP-conjugated nick end labeling) (32)
using the in situ cell death detection kit (Roche Products, Hertfordshire, UK), followed by staining with 0.5% DAB, according to the manufacturer’s instruction. Apoptotic nuclei were quantified as percent ratio of the total number of cells in the EGL, visualized by hematoxylin/eosin staining.
Images of the immunostained sections were obtained with an BX51 microscope (Olympus, Tokyo, Japan) connected to a Diagnostic Instruments camera 1.3.0 (Sterling Heights, MI, USA), or by laser scanning confocal microscopy using a CLSM 510 microscope (Zeiss, Jena, Germany). Measurements of positive cells were performed by the I.A.S. software (Delta Systems, Rome, Italy).
In situ hybridization
Preparation of sections and hybridization were performed as previously reported, with some modifications (28)
. Hybridization was performed with digoxigenin labeled specific antisense probes (Transcription kit; Roche). Samples were incubated overnight at 4°C with alkaline phosphatase-conjugated antidigoxigenin antibody (Roche; 1:2000), washed and processed for colometric detection using nitroblue-tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate (NTB/BCIP).
Specific antisense riboprobes were synthesized for: a) Math1, by T7 polymerase from mouse Math1 cDNA (vector CS2-Math1; 33); b) Gli1, by SP6 polymerase from mouse Gli1 cDNA (of which we cloned a 392 nt long region of exon 13 into HindIII5'-EcoRI3' sites of pcDNA3 vector); c) Patched1, by SP6 polymerase from mouse Patched1 cDNA (of which we cloned a 515 nt long region of exon 23 into HindIII5'-EcoRI3' sites of pcDNA3 vector); d)endogenous PC3 (i.e., Tis21), by T3 polymerase from the construct pT7T3D-Tis21 (mouse cDNA Image 836540). No signal was detected using the corresponding sense probes.
RNA extraction, real-time RT-PCR
Total cellular RNA, obtained from tissues and cell lines according to the procedure of Chomczynski and Sacchi (34)
, was reverse-transcribed as described previously (25)
.
Total RNA from murine and human medulloblastomas (snap-frozen in liquid nitrogen), cerebellar granule cells, or PC12 cells was analyzed by real-time RT-PCR amplification, performed with a 7900HT System (Applied Biosystems, Foster City, CA, USA), using either TaqMan probe-based fluorogenic 5' nuclease chemistry (for medulloblastomas) or SYBR GreenI dye chemistry (for cells), in duplicate samples. Relative quantification was performed by the comparative cycle-threshold method (35)
. Control human adult cerebellar RNA was from Clontech (Cambridge, UK) and fetal cerebellar RNA from Biocat (Heidelberg, Germany). The expression values of PC3 in human medulloblastomas were normalized to three different endogenous controls: ßActin, GAPDH, hypoxanthine phosphoribosyltransferase. Endogenous controls used for murine medulloblastomas were ßActin, ß2-microglobulin, hypoxanthine phosphoribosyltransferase, whereas for cerebellar granule cells and PC12 cells TATA-binding protein and/or 18S RNAs were used.
Specific real-time RT-PCR primers for mRNAs of BTG2 (human PC3), Tis21 (mouse PC3), PC3, and rat cyclin D1 and of the endogenous controls were deduced from published murine cDNA sequences and are available on request.
Cell culture and infection by adenoviruses
PC12 cells were grown in Dulbecco’s modified Eagle’s medium with 5% supplemented calf serum and 5% horse serum (HyClone, Logan, UT, USA) in a humidified atmosphere of 12% CO2 at 37°C. Cerebellar granule cultures from Wistar 7-day-old (P7) rats were prepared as described previously (28)
.
The generation of recombinant adenovirus expressing PC3 (Adeno-PC3) was described previously (28)
. ß-galactosidase adenovirus was kindly provided by M. Crescenzi. HEK293 cells were used to propagate adenoviruses and to measure the viral titers by plaque formation assays; these were always in the range of 108–109 plaque-forming units (pfu)/ml.
Chromatin immunoprecipitation (ChIP)
Primary cultures of cerebellar granule cells infected with adeno-PC3 or adeno-ß-galactosidase or cultures of PC12 cells induced to differentiate by NGF were fixed with 1% formaldehyde. Then, chromatin was prepared from cell lysates in SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl pH 8) according to standard protocols (Upstate Biotechnology), as described (36)
. Cellular lysates were sonicated to obtain DNA fragments of average size of 500 bp, immunoprecipitated with anti-PC3 A3H or antiacetyl-histone-H4 antibodies, or preimmune serum as control. The presence of the rat cyclin D1 promoter was analyzed by PCR using primers amplifying a region 770 nt before the transcription start: 5'-CCCCAGCGAGGAGGAATAGATG-3', 5'-TGCCAGACGAGCCCTAAGTTC-3'.
Glutathione-S-transferase (GST) fusion proteins
Pull-down assays were performed incubating 10 µl of GST proteins bound to glutathione-Sepharose resin beads with in vitro-programmed nuclease-treated rabbit reticulocyte lysates, as described (37)
. HDAC1 and HDAC4 were transcribed in vitro using the constructs pSCT-HDAC1, generated by subcloning in 5'BamHI-3'XbaI of the pSCT1 vector the coding region of HDAC1 (excised from pcDNA3.HDAC1-Flag obtained from T. Kouzarides, Cambridge, UK), and pcDNA3.1-Myc-HDAC4, provided by T. Kouzarides.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. In TgPC3 mice the expression of transgenic PC3, regulated by the ß-actin (ßACT) promoter, matches the expression of endogenous PC3 in neuronal precursors undergoing neurogenic division, including GCPs (28)
.
To target the effect of PC3 to neuronal precursors proliferating perinatally, i.e., essentially cerebellar GCPs, whose largest wave of maturation occurs from birth to P7, the PC3 transgene was expressed after E18—by withdrawal of doxycycline. Moreover, all the progeny of breedings between Patched1 heterozygous and TgPC3 mice was irradiated at P1, a treatment increasing the frequency of medulloblastomas (22)
.
The F1 progeny of crosses between Patched1 heterozygous and TgPC3 mice (hereafter referred to as Ptc+/–/TgPC3 and Ptc+/+/TgPC3 if both the ßACT-tTA and the TRE-PC3 transgenes were inherited, or Ptc+/– and Ptc+/+ if they were not) did not show early lethality due to irradiation. A high incidence of medulloblastoma was observed in Ptc+/– mice. Of the 64 mice in this group 39 (60.9%) developed medulloblastoma between 12 and 40 wk of age. In contrast, Ptc+/–/TgPC3 showed a highly decreased frequency (37.5%; 9/24 total) with no significant changes in tumor latency (Fig. 1
and Table 1
). Therefore, PC3 appears to possess a strong tumor suppressive activity in vivo. No medulloblastoma developed after irradiation in Ptc+/+/TgPC3 (0/30 total) or in Ptc+/+ mice (0/67 total) (Table 1)
.
|
|
PC3 counteracts the activity of Shh in GCPs by reducing proliferation and enhancing differentiation
To gather information on the mechanisms underlying the tumor suppression by PC3, we sought to analyze the response of GCPs to the actions exerted by the Shh pathway and by PC3, during the initial phases of medulloblastoma tumorigenesis.
It has been previously shown that Patched1 heterozygous mice present thicker EGL regions containing highly proliferating GCPs, defined by Kim et al. (20)
hyperplastic EGL remnants or rests, that persist after P14. Since their incidence is higher than the frequency of medulloblastoma, these hyperplastic EGLs can be considered non committed premalignant lesions. Thus, we decided to analyze the EGL of cerebella at P14, because at this stage hyperplastic EGL regions and other pretumoral abnormalities are already detectable in Patched1 heterozygous mice and coexist with the normal EGL, reduced to a thickness of few layers of GCPs (38
, 20)
.
The incidence of hyperplastic EGL areas, bromodeoxyuridine (BrdU) incorporation, and expression of cyclin D1, cyclin D2, and cyclin B1 were examined in the cerebellum of Ptc+/– and Ptc+/+ F1 mice lacking the PC3 transgene. At P14, cerebella of Ptc+/– mice presented localized, thick hyperplastic EGL regions containing up to 15–20 layers of highly proliferating GCPs (shown in Fig. 2
A, B), with average frequency of 2.75 ± 0.48 per cerebellum (Table 2
). Throughout the entire EGL length, but particularly in these hyperplastic regions, the frequency of GCPs incorporating BrdU or expressing cyclin D1 resulted highly increased as compared with Ptc+/+ wild-type mice (Fig. 2B
; Table 3
). In contrast, the expression of other cyclins, i.e., cyclin D2 and cyclin B1—controlling G1 and G2/M phases, respectively—was not significantly changed in the EGL of Ptc+/– mice compared with wild-type Ptc+/+ mice (Table 3)
.
|
|
|
A most prominent effect of PC3 overexpression was a drastic reduction, up to four-fold, of the frequency of hyperplastic EGL regions in Ptc+/–/TgPC3 mice, compared with Ptc+/– mice (Table 2)
. Concomitantly, the proliferation of GCPs throughout the EGL and hyperplastic areas, as measured by BrdU incorporation and cyclin D1 expression, was restored to the level of wild-type Ptc+/+ mice (Fig. 2B
; Table 3
). Overall, these data indicate that perinatal up-regulation of PC3 expression may effectively counteract the hyperproliferative stimulus caused by deregulation of the Shh pathway in GCPs.
Next, we examined the expression of neurogenic markers. As shown in Fig. 2B
and Table 3
, a strong decrease of the expression of NeuN, which marks postmitotic differentiated granule neurons (39)
, was observed in the hyperplastic EGL regions and also in the internal granular layer (IGL) of Ptc+/– mice, as compared with wild-type Ptc+/+ mice. The reduced expression of NeuN in the IGL, where postmitotic granule cells migrate, suggests a reduced production of differentiated granule cells by the active GCP pool in the EGL. In Ptc+/– mice we also detected a significant increase of the neurogenic transcription factor NeuroD1, within the whole hyperplastic EGL and in the IGL (Fig. 2B
and Table 3
) and a slight increase of the expression of Math1 mRNA, which is required for the neurogenesis of granule neurons and is expressed in proliferating GCPs of the outer EGL (40)
. The increase of NeuroD1 occurred not only in recently differentiated postmitotic GCPs of the inner EGL, as normally occurs (41)
, but also in proliferating undifferentiated GCPs, which may indicate a failed attempt of GCPs to differentiate.
Remarkably, in Ptc+/–/TgPC3 mice the expression of NeuN was restored to the levels of wild-type Ptc+/+ mice (Fig. 2B
; Table 3
). Furthermore, the EGL of Ptc+/–/TgPC3 mice presented an evident increase in the expression of Math1 and NeuroD1, compared with Ptc+/– mice (Fig. 2B
; Table 3
). In addition, the increase of NeuroD1 positive cells was associated to a decrease of BrdU-incorporating cells (compare Ptc+/–/TgPC3 and Ptc+/– EGLs in Table 3
), strongly suggesting that in fact PC3 induced the expression of NeuroD1 in GCPs that have ceased proliferating and become postmitotic. Such possibility is also supported by NeuroD1/BrdU double-staining analyses (data not shown).
Taken together, the above data indicate that the PC3 gene not only prevents the Patched1-dependent hyperproliferation of GCPs within hyperplastic EGL regions, but also restores their ability to terminally differentiate.
As a further analysis of possible mechanisms of tumor suppression by PC3, we tested whether apoptosis of GCPs was involved. No significant change in the number of cells undergoing apoptosis in the EGL of Ptc+/–/TgPC3 vs. Ptc+/– mice was observed (Table 3)
.
Moreover, lack of difference in acute cellular response to DNA damage between Ptc+/–/TgPC3 and Ptc+/– or Ptc+/+ mice at P1, as evaluated by measuring the percentage of GCPs expressing phosphorylated histone H2A.X or undergoing apoptosis 3 h postirradiation, clearly indicates that the tumor suppressor action of PC3 does not involve protection from DNA damage (Ptc+/–/TgPC3: 13.8±1.1% phospho-H2A.X-positive GCPs and 9.1±1.0% TUNEL-positive GCPs; Ptc+/–: 14.5±0.8% and 8.2±1.4%, respectively; Ptc+/+: 12.7±1.2% and 11.0±1.4%, respectively. Fourteen sections analyzed per group).
The expression of endogenous PC3 is down-regulated in preneoplastic lesions
In addition to hyperplastic EGL areas, Ptc+/– mice presented cerebellar lesions, consisting in EGLs abnormally expanded as nodular formations inside the cerebellar lobules. These early lesions were formed by GCPs with an even higher rate of BrdU incorporation and cyclin D1 expression, and with lower differentiation capacity, as indicated by the greatly reduced expression of NeuN (see a representative image in Fig. 3
A, B; Table 2
and 3
). These formations also showed reduced expression of Patched1 (Fig. 3B
) and, having lost the organization in layers typical of the EGL, represented preneoplastic lesions at a more advanced stage than hyperplastic EGLs. As shown in Table 2
, the frequency of such lesions in cerebella of Ptc+/–/TgPC3 mice resulted 6-fold lower than in Ptc+/– mice.
|
We have previously observed that the physiological expression of PC3 occurs mainly in the outer EGL, in the pool of proliferating GCPs, and that overexpression of PC3 stimulates these GCPs to generate differentiated granule cells, through induction of a neurogenic division (28)
. Remarkably, in Ptc+/– cerebella we found that endogenous PC3 was significantly expressed only in the superficial layers of the lesions, then decreased internally and was totally absent in deeper regions, where GCPs still continued to actively proliferate and to express cyclin D1 (compare Fig. 3B
with C). Math1 and NeuroD1 were expressed with a distribution similar to PC3, whereas Gli1—a direct downstream target of Shh—was up-regulated in the deeper part of the lesion. Thus, GCPs of the internal part of the lesion showed the highest activation of the Shh pathway and a concomitant down-regulation of PC3, which strongly suggests that the decrease of endogenous PC3 correlates with the severity of the lesion.
PC3 mRNA is down-regulated in medulloblastoma
In view of the above data, we sought to measure the endogenous levels of PC3 mRNA in medulloblastomas from Ptc+/– mice. We screened a panel of 15 medulloblastomas developed in Ptc+/– mice of different backgrounds (i.e., the mixed CD1/BDF1 background of Ptc+/– mice crossed with TgPC3 or the CD1 background of parental Patched1 heterozygous mice). Cerebella from wild-type P5 mice of the corresponding background were used as appropriate control (15)
. As shown in Fig. 4
A, 50% of tumors (7 out of 15) from Ptc+/– mice displayed reduced expression of endogenous PC3 mRNA, irrespective of the genetic background.
|
Next, we examined the expression of PC3 in human medulloblastomas of different histotypes (Fig. 4B
). Endogenous PC3 was decreased in all human medulloblastomas of the desmoplastic histotype analyzed (n=4), in 8 out of 13 of the classic subtype and in 1 out of 4 of the anaplastic subtype. A pool of cerebellar RNAs from adult male and female Caucasians was used as control. Expression of PC3 in this pool resulted nearly identical to levels measured in fetal human cerebella (data not shown).
Thus, PC3 mRNA is down-regulated in mouse and human medulloblastomas, consistently with the reduced expression of PC3 observed in preneoplastic lesions of Ptc+/– cerebella.
PC3 interacts with the cyclin D1 promoter region
The above data point to a central role of cyclin D1 in the PC3-mediated inhibition of proliferation of normal and preneoplastic GCPs in vivo, as cyclin D1, among the tested cyclins, is selectively down-regulated by PC3 (see Table 3
). Thus, we sought to define whether PC3 transcriptionally controls cyclin D1 in GCPs and, more importantly, whether this control is direct. Therefore, we determined cyclin D1 mRNA levels in primary cultures of granule cells infected with PC3-expressing adenovirus, as well as in the neuronal PC12 cell line treated with nerve growth factor (NGF), a stimulus known to highly induce PC3 expression (23)
. As shown in Fig. 5
A, B, the expression of ectopic PC3 in cerebellar granule cells, or the induction of endogenous PC3 in PC12 cells, was accompanied by a corresponding decrease of the levels of cyclin D1 mRNA.
|
Next, we asked whether PC3 is recruited to the cyclin D1 promoter. To this aim, we performed chromatin immunoprecipitation (ChIP) experiments on primary cultures of granule cells and in PC12 cells, treated as above, using the anti-PC3 antibody A3H (24)
. Amplification by PCR of a fragment of the cyclin D1 promoter region showed that PC3 resided on the cyclin D1 promoter both in cerebellar granule and PC12 cells and that the levels of PC3 bound to the promoter were higher in cerebellar granule cells infected with PC3-expressing adenovirus and in PC12 cells after induction of endogenous PC3 by NGF (Fig. 5C, D
). Furthermore, the increased binding of PC3 to the cyclin D1 promoter was accompanied by decreased levels of acetylated histone H4 associated to the promoter (Fig. 5C, D
, panel anti-acetyl-H4), suggesting that PC3 binding may inhibit cyclin D1 transcription by promoting histone deacetylation at the cyclin D1 promoter. An implication of this hypothesis is that PC3 might recruit histone deacetylases (HDACs) to the cyclin D1 promoter. To test this possibility, we performed a GST pull-down assay to analyze the ability of PC3 to bind HDAC1 and HDAC4. Both in vitro-translated histone deacetylases resulted specifically able to associate with GST-PC3 (Fig. 5E
).
As a whole, these data strongly suggest that PC3 negatively controls the transcription of cyclin D1 by directly associating to the promoter, possibly recruiting to it HDAC1 and HDAC4.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Important issues raised by these observations concern the molecular mechanism by which the PC3-mediated tumor suppression occurs, and the possible role of endogenous PC3 down-regulation in the process of medulloblastoma tumorigenesis.
PC3 restores to normality the proliferation of GCPs in Ptc+/– mice
The capability to undergo massive expansion during postnatal development before terminal differentiation is a peculiar feature of GCPs of the EGL. Thus, a tight control of the transition from a mitotically active, undifferentiated state to postmitotic terminal differentiation is critical for these neural precursor cells. Cyclin D1, cyclin D2, and N-Myc are essential for cerebellar organogenesis (42
, 9)
and lead to GCP proliferation in response to Shh signaling (6
, 8)
. Accordingly, persistent expression of such genes is recurrent in childhood medulloblastomas.
Our analysis of cerebella of Ptc+/– mice at P14 evidenced several abnormalities preceding medulloblastoma appearance. In first place, localized hyperplastic EGL regions, where GCPs showed an increased rate of proliferation, as measured by BrdU incorporation and cyclin D1 expression. More advanced alterations consisted in focal lesions, in which GCPs displayed a further increase in BrdU incorporation and very high expression of downstream targets of the Shh/Patched1 pathway, i.e., cyclin D1 and Gli1 (6
, 8
, 43)
. Thus, the hyperproliferating EGL regions and the focal lesions observed at P14 appear preneoplastic formations at different levels of commitment, triggered by a strong proliferative stimulus originating from deregulated Shh pathway activation.
A most evident effect of PC3 overexpression in Ptc+/– cerebella was the reversal of Shh-dependent hyperproliferation, throughout the entire EGL and particularly in the hyperplastic regions. This effect is linked to the ability of PC3 to delay cell cycle progression from G1 to S phase, as indicated by the reduced incorporation of BrdU in the EGL of Ptc+/–/TgPC3 mice at P14. We show that this delay occurs through specific down-regulation by PC3 of the levels of cyclin D1 - well known to trigger with the partner cyclin-dependent kinase 4/6 the entry into S phase—as no significant change was observed in the expression of other cyclins.
Down-regulation of cyclin D1 levels suggests that a specific interaction of PC3 with cyclin D1 is involved in the mechanism of tumor suppression. In this regard, results from previous studies have shown no interference of PC3 with N-Myc (28)
, a direct target of Shh responsible for the up-regulation of cyclin D1 in GCPs (7
, 10)
.
Indeed, as the present chromatin immunoprecipitation studies in GCPs reveal, PC3 is recruited to the cyclin D1 promoter, in correlation with reduced histone acetylation at the promoter. We also observed that PC3 can directly interact with HDAC1 and HDAC4. Taken together, these observations strongly suggest that PC3 negatively controls cyclin D1 transcription by participating to deacetylase-containing repressor complexes on the cyclin D1 promoter.
Therefore, the functional antagonism to Shh exerted by PC3 in GCPs occurs through direct inhibition of the Shh downstream target cyclin D1. In line with our results, cyclin D1–/–/Ptc+/– double knockout mice develop medulloblastomas with significantly lower incidence than Ptc+/– mice (44)
.
Growth arrest by PC3 is associated to a neurogenic effect in preneoplastic GCPs
Another prominent feature of hyperplastic EGLs and preneoplastic lesions of Ptc+/– mice was the presence of marked differentiation defects, as indicated by reduced expression of NeuN and also by a paradoxical increase of NeuroD1 (i.e., not restricted to the postmitotic GCPs where it is normally expressed). A similar inhibition of differentiation has also been previously observed in cultures of GCPs kept under the proliferative stimulus of Shh (3)
.
Notably, PC3 overexpression rescued GCPs differentiation, as indicated by restoration of normal expression levels of the terminal differentiation marker NeuN in Ptc+/–/TgPC3 mice. Furthermore, our results show a detectable increase in the expression of Math1 and NeuroD1 in the EGL of Ptc+/–/TgPC3 mice at P14 and also suggest that PC3 rescues at least in part the mis-expression of NeuroD1 in proliferating GCPs of Ptc+/– mice, reestablishing its postmitotic expression. Consistent with our previous data showing that PC3 induces transcription of Math1 in GCPs (28)
and with the observation that Math1 induces NeuroD1 (41)
, the prodifferentiative effects of PC3 might result not only from its ability to delay G1-S progression in GCPs but also from a direct effect on proneural gene(s). PC3, whose expression is known to be associated with the induction of an asymmetric neurogenic division [(26
27
28)
; for review, see (45)
], may, therefore, have the ability to restore the neurogenic state in Shh-stimulated GCPs.
Altogether, the combined abilities of PC3 to inhibit the Shh-dependent hyperproliferation of GCPs and to rescue their terminal differentiation can fully account for the tumor suppressive effect of PC3.
Possible physiological role of PC3 as tumor suppressor
While current knowledge indicates PC3 as a regulator of neural progenitor development, we have identified a tumor suppressor role for PC3 in the cerebellum. In support of this view is the observation that endogenous PC3 was down-regulated in preneoplastic lesions of Ptc+/– cerebella at P14, as well as in murine and human medulloblastomas. Moreover, our experiments in the Patched1-deficient mouse model of medulloblastoma demonstrated tumor suppression following up-regulation of PC3 in vivo in proliferating GCPs, where it is normally expressed. Altogether, this suggests that PC3 may behave physiologically as tumor suppressor.
Although our results indicate that PC3 plays a key role in preventing medulloblastoma development from early stage, the recently generated PC3-null mice did not show overt alterations in the cerebellum (46)
. Analysis, however, was limited to the gross cerebellar morphology. Similarly, PC3-null mice have not been reported to develop spontaneous tumors in the early stages of their lives. A possibility is that in PC3-null mice the function of PC3 is vicariated by other genes of the family to which PC3 belongs, such as BTG1, TOB, or PC3B (47
, 48)
. A second possibility, indicated by the finding that the functional ablation of PC3 by RNA interference predisposes to tumorigenesis by allowing Ras-dependent cellular transformation (49)
, is that the inactivation of PC3 cooperates with other oncogenic events to induce neoplastic transformation.
In agreement with our results, additional evidence suggests that PC3 down-regulation promotes the tumorigenic process, since decreased levels of PC3 were observed in prostate, renal, and breast cancers (50
51
52)
.
Thus, a comprehensive hypothesis, accounting for the down-regulation of PC3 observed in medulloblastomas and preneoplastic lesions, implies the inactivation of PC3 as a critical step for the onset of medulloblastoma. Our results, highlighting the tumor suppressive effect of PC3 overexpression in vivo, qualify this gene as a functional antagonist of the Shh pathway in medulloblastoma pathogenesis and as a potential new target for therapy.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication December 16, 2006. Accepted for publication February 8, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| |||||||
2 test; see also Table 1
