Tis21-gene therapy inhibits medulloblastoma growth in a murine allograft model

Medulloblastoma (MB), the tumor of the cerebellum, is the most frequent brain cancer in childhood and a major cause of pediatric mortality. Based on gene profiling, four MB subgroups have been identified, i.e., Wnt or Sonic Hedgehog (Shh) types, and subgroup 3 or 4. The Shh-type MB has been shown to arise from the cerebellar precursors of granule neurons (GCPs), where a hyperactivation of the Shh pathway leads to their neoplastic transformation. We have previously shown that the gene Tis21 (PC3/Btg2) inhibits the proliferation and promotes the differentiation and migration of GCPs. Moreover, the overexpression or the deletion of Tis21 in Patched1 heterozygous mice, a model of spontaneous Shh-type MB, highly reduces or increases, respectively, the frequency of MB. Here we tested whether Tis21 can inhibit MB allografts. Athymic nude mice were subcutaneously grafted with MB cells explanted from Patched1 heterozygous mice. MB allografts were then injected with adeno-associated viruses either carrying Tis21 (AAV-Tis21) or empty (AAV-CBA). We observed that the treatment with AAV-Tis21 significantly inhibited the growth of tumor nodules, as judged by their volume, and reduced the number of proliferating tumor cells (labeled with Ki67 or BrdU), relative to AAV-CBA-treated control mice. In parallel, AAV-Tis21 increased significantly tumor cells labeled with early and late neural differentiation markers. Overall the results suggest that Tis21-gene therapy slows down MB tumor growth through inhibition of proliferation and enhancement of neural differentiation. These results validate Tis21 as a relevant target for MB therapy.


Introduction
Medulloblastoma (MB), a highly malignant cerebellar neoplasm, is the most common brain cancer in infants and children, comprising 15-20% of all pediatric nervous system tumors. Moreover, MB represents the primary cause of pediatric mortality related to cancer. MB is also PLOS

Ethics statement
Animals were subjected to experimental protocols (Authorizations N. 193/2015-PR and N. 872/2015-PR) approved by the Veterinary Department of the Italian Ministry of Health. Experiments were conducted according to the ethical and safety rules and guidelines for the use of animals in biomedical research provided by the relevant Italian law and European Union Directive (Italian Legislative Decree 26/2014 and 2010/63/EU) and the International Guiding Principles for Biomedical Research involving animals (Council for the International Organizations of Medical Sciences, Geneva, CH). All adequate measures were taken to minimize animal pain or discomfort.

Adeno-associated Virus-Tis21 expressing, cloning and infection
Packaging of plasmids in recombinant AAV serotype 9 (hereafter named AAV) is based on methods previously developed and described by Klein et al. [45].
Briefly, 293 helper cells have been transfected with an AAV terminal repeat containing plasmid (either pAAV-CBA-WPRE-empty or pAAV-CBA-WPRE-GFP used as control vectors or pAAV-CBA-WPRE-Tis21, abbreviated in AAV-CBA, AAV-GFP or AAV-Tis21, respectively). R. Klein kindly provided the pGFP plasmid, from which we excised the GFP insert contained in the HindIII site, obtaining AAV-CBA (that carried no insert), and then we cloned in the HindIII site of AAV-CBA the Tis21 (mouse) open reading frame, thus generating AAV-Tis21. Transfection of AAV-CBA, AAV-GFP or AAV-Tis21 occurred in an equimolar ratio with the plasmid pDG, which provides the AAV coat protein genes, and adenovirus 2 genes necessary for helper function in packaging. The plasmids have been packaged in recombinant AAV serotype 9 because this is able to infect CNS cells [46]. Three days after transfection, cells and media have been harvested, pelleted, resuspended in lysis buffer (50 mM Tris, pH 8.5, 150 mM NaCl) and freeze-thawed three times. The samples have been incubated with endonuclease (1500 units for 30 min at 37˚C), centrifuged at 2000 x g and the resulting supernatants applied to a discontinuous gradient of cesium (Sigma-Aldrich), followed by dialysis against 1x PBS. Recombinant AAVs has then been titered for total particles by the described dotblot and quantitative/competitive PCR methods [45] and used for human MB cell lines infections and allograft treatments.

Cell lines AAV infections
DAOY and D283 cells were seeded respectively at 50,000 and 150,000 cells/60 mm plate overnight. They were then infected with AAV (CBA or Tis21) at four different multiplicity of infection (MOI) ranging from 1.3x105 to 6.2x106 viral particles/cell in complete tissue culture medium. Seven days post infection, RNA was harvested using Trizol reagent (Thermo Fisher Scientific) and used to perform quantitative reverse-transcriptase PCR (qRT-PCR).

MB development and isolation of tumor cells for allograft propagation
Ptch1 +/mice were obtained in a CD1 background by gene targeting of exons 6 and 7 [47] and were maintained by continuous inbreeding. The pups were routinely genotyped by PCR, using genomic DNA from tail tips, as previously described [37,47]. It is known that the Ptch1 +/mice on the CD1 background develop spontaneous MB over a period of 10 months with low incidence (about 7%) [48]. Thus, Ptch1 +/mice of either sex were daily observed for tumor formation for 12 months after birth. On the appearance of MB symptoms, such as weight loss, ruffling of fur, lethargy, poor balance, posterior paralysis and impaired coordination, they were euthanized and the MB tumors isolated. The tumor cell suspension was obtained from explanted MB following the protocol described by Sasai et al. [49]. Briefly, MB tumor was reduced in small fragments by mean of a McIlwain tissue chopper and incubated with collagenase type IV (500 μg/ml) and hyaluronidase (500 μg/ml) for 30 minutes at 37˚C, then strained using a cell-strainer (40 μm). A yield of 2-4 x 107 cells per tumor was obtained.
Tumor allografts were generated essentially as previously described [50]. Briefly, 5-10 weeks old mice were injected subcutaneously with MB cells (2.5-3.0 x 106 cells) in 50% matrigel in 200 μl of PBS into the dorsal flank region of each mouse. Tumor volume was calculated by caliper measurements of tumor length (L) and width (W) according to the formula: Engrafted MB cells were allowed to expand before treatment with AAV. When tumors reached an approximate volume of 100-150 mm3 the mice were randomly divided into an experimental and control groups (n = 2-5/group) for AAV or PBS treatment. Six intratumoral injections (T1-T6) of AAV-Tis21 or AAV-CBA or AAV-GFP (50 μl equivalent to 1011 viral particles), or PBS (50 μl), were performed; the T1 and T2, 24 hours apart and T3-T6 each 48-72 hours.
In one of the three independent experiments, mice were also i.p. injected with BrdU (50 mg/Kg) 24 hours before the AAV-treatment T3-T6.
Tumor size and body weight were measured every two days. Twenty-four hours after the last viral particles injection (pT6) all mice previously euthanized with i.p. injections of tiletamine/zolazepam (800 mg/kg) and xylazine (100 mg/kg), were sacrificed and tumors were harvested, measured, photographed, and pathologically examined. Tumors were then divided in two aliquots, one was fresh frozen (for RNA extraction and qRT-PCR) and the other was fixed for immunohistochemistry analyses.

Immunohistochemistry and microscopy analysis
Immunohistochemistry (IHC) was performed essentially as previously described [50]. Briefly, excised tumors were fixed with 3.7% paraformaldehyde (PFA) (wt/vol) in PBS for 20 min at room temperature and then embedded in paraffin. Serial sections 5 μm (NeuN) or 8 μm (Ki67 and NeuroD1) thick were cut from the paraffin embedded tissue blocks and floated onto charged glass slides (Super-Frost Plus, Fisher Scientific, Pittsburgh, PA) and dried overnight at 60˚C. Sections were deparaffinized in changes of xylene and rehydrated in decreasing concentrations of ethanol and rinsed in PBS. For antigen retrieval, samples were boiled for 20 min in 10 mM sodium citrate buffer (pH 6.0) and cooled for 5-10 min in water. Slides were washed in PBS and incubated with blocking buffer (1x PBS, 0.1% Triton X-100, 1% BSA, 4% donkey serum) for 1 hour and then incubated for 16 hours at 4˚C with primary antibodies diluted in blocking buffer. The following day slides were washed three times with washing buffer (1x PBS, 0.1% Triton X-100), incubated with biotinylated anti-Ig secondary antibody for 5 hours followed by streptavidin Alexa 488, and finally washed as before and mounted using Mowiol 4-88 mounting media.
The detection of the BrdU incorporation was performed as previously described [38,43]. Briefly, the tumors of mice i.p. injected with BrdU were fixed by transcardiac perfusion with 4% PFA in PBS and then cryoprotected in 30% sucrose in PBS. The tumor tissues were embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA, USA) and were cut on a rotary microtome in one-in-six series free-floating sections 40 μm thick. BrdU incorporation was detected by treating the sections with 2 N HCl 45 min at 37˚C, followed by 0.1 M sodium borate buffer, pH 8.5, for 10 min. The sections were incubated 16 hours at 4˚C with a rat monoclonal antibody against BrdU together with the primary antibodies against Ki67 and NeuroD1, then with the appropriate diluted secondary antibodies.
Images (20-25 images/each nodule for Ki67, NeuroD1 and BrdU staining) were collected either with an Olympus AX70 fluorescence microscope using 40x lens served by an Olympus XM10 camera and processed using the Olympus CellSens Standard 1.8.1 software or a TCS SP5 laser scanning confocal microscope (Leica Microsystems, Wetzlar, Germany). Labeled cells were counted with the ImageJ software (National Institutes of Health, USA) or with the I. A.S. software (Delta Sistemi, Rome, Italy). The NeuN immunolabelled paraffin sections were analyzed with an Olympus FV1200 laser scanning confocal microscope using a PlanApoN 60x Oil SC (NA = 1.40) Olympus objectives, with optical pinhole at 1AU and a multiline argon (405 nm, 488 nm) laser as excitation source. Confocal Z-stacks were collected at 0.5 μm intervals in a 3.5 μm total optical depth. Approximately 40 representative images from two different paraffin sections, at 30 μm of distance, were acquired for further analysis with ImageJ. Images for direct comparison were collected under same parameters upon setting on negative sample and representative images were chosen.
Images of free-floating sections (40 μm) of nodules injected with AAV-GFP virus stained with Hoechst 33258 were collected with a TCS SP5 laser scanning confocal microscope (Leica Microsystems) using a 63x lens and analyzed for the presence of the GFP fluorescent protein.

RNA extraction, Real time qPCR
Total cellular RNA was extracted from nodules treated with AAV-Tis21 or with AAV-CBA (control), or from DAOY and D283 human cells infected with AAV-Tis21 or with AAV-CBA as reported above, and reverse-transcribed as previously described [51]. Three nodules per group and 3-5 technical replicates for the human cell lines were used for analysis. Real-time qPCR was carried out with a 7900HT System (Applied Biosystems) using SYBR Green I dye chemistry in duplicate samples. Relative quantification was performed by the comparative cycle threshold method [52]. The mRNA expression values were normalized to those of the TATA-binding protein gene, used as endogenous control. Specific qRT-PCR primers used were deduced from published murine and human cDNA sequences and are listed in Table 1.

Statistical analysis
All statistical tests were performed using GraphPad Prism (GraphPad Software, La Jolla, CA, USA) or StatView 5.0 software (SAS Institute, Cary, NC, USA).
For the tumor growth analysis the data are presented as the variance of multiple independent experiments (mean ± SEM). Significance was determined using a two-way ANOVA test

Results and discussion
Our aim was to evaluate the effectiveness of a gene therapy approach using Tis21, by testing whether the AAV-mediated overexpression of Tis21 in neoplastic GCPs is able to suppress proliferation and induce their terminal differentiation. Therefore, we established MB allografts of neoplastic GCPs generated from Ptch1 +/mice, and developed a gene therapy protocol with AAV particles carrying Tis21.

Establishment of MB allograft
It has been recently shown that grafts of MB cancer cell lines in nude mice are less predictive than MB primary tumors. In fact, the Shh pathway activity, normally very high in tumors derived from Ptch1 +/mice, is suppressed in MB explanted from Ptch1 +/mice and grown in culture and in MB cancer cell lines [49]. Indeed, MB cancer cell lines have been shown to be negative for the Shh effector Gli1 gene [49]. For this reason, we have analyzed the MB suppressor activity of Tis21 using allografts of MB obtained from Ptch1 +/mice. We selected primary tumors with no obvious ulceration or necrosis, because both factors reduce the recovery of viable cells. Single cell suspensions were prepared using collagenase type IV and hyaluronidase from MB tumor with a typical yield of 2-4 x 107 cells per tumor as described in Materials and methods. Next, Foxn1nu/Foxn1+ female mice were subcutaneously grafted with 2.5-3 x 106 cells with matrigel. We achieved 100% allograft growth when we injected a suspension containing at least 1.8 x 106 MB cells. Injection of more than 2.5 x 106 cells accelerated the appearance of detectable tumors of 100-150 mm3 (1 week for 2.5 x 106 vs 2 weeks for 1.8 x 106 cells).

AAV-Tis21 infects MB cell lines and blocks MB allograft growth during treatment
Since the MB primary tumor cells grow well in nude mice, they provide a model for the study of novel gene therapy protocols. As a proof of principle we tested the effects of Tis21 overexpression, known to be an important tumor suppressor gene and potential relevant target gene in MB treatment. To this aim, we have used an AAV vector that has proven to be neural specific [46,53], kindly provided to us by Dr. Klein. The efficiency and specificity of the virus is increased by the presence in the backbone of the AAV of the woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) and a cytomegalovirus/chicken β-actin (CBA) promoter, previously shown to drive more efficiently than other promoters gene expression in different brain structures when stereotaxically injected [53,54]. An important feature of this vector is that the CBA promoter drives gene expression not only in adult neurons, but also in proliferating neural precursors [55] such as neoplastic GCPs. The choice of AAV as carrier was motivated by the fact that this virus is the most used in gene therapy procedures for its high expression and lack of adverse immunological reactions [46].
As a preliminary step, we tested the efficiency of the AAV-Tis21 and of the control AAV-CBA viruses to infect medulloblastoma cell lines and to inhibit a proliferation marker such as cyclin D1, a known target of Tis21 [51]. Two human cell lines, widely used in in vitro studies and in allograft experiments, were selected: DAOY derived from the desmoplastic human medulloblastoma subtype [56] and D283 established from malignant ascites cells and a peritoneal medulloblastoma metastasis [57]. The cells were transduced with increasing amount of viral particles (MOI: 0.13x106, 1.3x106, 6.2x106) and collected after 7 days for RNA extraction and for subsequent analyses. In parallel, cells were transduced with the same amount of the empty virus AAV-CBA, as control (Fig 1).
As the MOI of AAV-Tis21 increased, a parallel increase in levels of Tis21 mRNA expression was observed in both cell lines, with respect to control treated cells, indicating that AAV-Tis21 is suitable to infect cells and overexpress Tis21 (Fig 1A,   , which was set to unit. Human TATA-bindi ng protein mRNA was used as endogenous control for normalization. Two-way ANOVA followed by Fisher's PLSD test was used for the statistical analysis; Ã p < 0.05, ÃÃ p < 0.01, or ÃÃÃÃ p < 0.0001, n.s. p non-significant, PLSD ANOVA test. https://doi.org/10.1371/journal.pone.0194206.g001 noting that the primers used cross-react also with the human Tis21 and thus a basal level of the endogenous gene was detected. As a consequence of the Tis21 overexpression, it was observed a significant dose-dependent decrease of cyclin D1 mRNA levels both in DAOY cells (Fig 1B, MOI 0.13x106: p = 0.8664; MOI 1.3x106: 26% decrease, p = 0.0347; MOI 6.2x106: 42.7% decrease, p = 0.001) and in D283 cells (Fig 1D, MOI 0.13x106: p = 0.4074; MOI 1.3x106: p = 0.8095; MOI 6.2x106: 59% decrease, p = 0.0078). We conclude that AAV-Tis21 infects efficiently human medulloblastoma cell lines and has an inhibitory effect on cyclin D1, and can therefore be used to attempt the inhibition of the growth of tumor allografts.
As a next step, we tested the specificity of infection in vivo by analyzing MB nodules of nude mice injected with AAV-GFP viral particles. As shown in S1 Fig, we observed GFP positive nodule cells indicating that the AAV is able to infect the cells of the MB nodule.
In order to evaluate the effect of Tis21, groups of 2-5 mice were injected in the allograft with AAV-Tis21 and control AAV-CBA viral particles, or with PBS (blank control) as described in Materials and methods. We found that AAV-Tis21 treatment drastically reduces the Ptch1 +/derived MB growth. In particular the tumor growth was completely blocked during the first injections with AAV-Tis21 (T1-T5) and the nodules size was significantly smaller than the control ones 24 hour after the last injection (pT6) (Fig 2). On the contrary, control MB allografts treated with either AAV-CBA or with PBS showed, as expected, an exponential tumor growth (Fig 2; two-way ANOVA, treatment F[2,120] = 27.84, p < 0.0001; single comparisons by Bonferroni's test: AAV-CBA-T5 or PBS-T5 vs AAV-Tis21-T5 p = 0.0007 or 0.0038, respectively; AAV-CBA-T6 or PBS-T6 vs AAV-Tis21-T6 p < 0.0001 or 0.0003, respectively; AAV-CBA-pT6 or PBS-pT6 vs AAV-Tis21-pT6 p < 0.0001 for both). However, a strong growth inhibitory effect of AAV-Tis21 was observed only during treatment, indeed the Tis21treated MB allograft reach a tumor volume similar to the pT6 MB control in 1-2 weeks after the last viral particles injection (AAV-Tis21 653.33 ± 86.7 mm3 vs AAV-CBA 611.00 ± 86.0 mm3, n = 3 per group, p > 0.05).  We also checked whether the effect observed on nodule growth corresponded to an increase of Tis21 expression. To this aim, we tested the expression of Tis21 mRNA by in situ hybridization, using a probe that detects endogenous as well as exogenous Tis21 mRNA (S2 Fig). We observed that in general the expression of Tis21 mRNA is higher in AAV-Tis21 infected nodules; moreover, the expression of Tis21 signal within the nodule is not uniform, likely depending on the sites of injection of the virus.

AAV-Tis21 treatment decreases the number of proliferating cells and increases the number of differentiated cells in MB allograft
Immunohistochemical analysis of MB allograft slides with Ki67 antibody confirmed a statistically significant decrease of cell proliferation in the AAV-Tis21 treated-allograft (pT6, 24 hours post-treatment; 32% decrease; p < 0.0001) when compared with control allograft samples (Fig 3A and 3B). The proliferation rate was also examined in MB allograft sections of nude mice treated with four i.p. injection of BrdU, which is incorporated in the DNA of the cells entering S-phase. The percentage of BrdU-positive proliferating cells also was found to be decreased significantly (68% decrease; p < 0.0001) when compared with AAV-CBA-treated control mice (Fig 3C and 3D).
Tis21 has been previously shown to induce cell differentiation in the developing cerebellum [36], as well as in the adult neurogenic niches of the dentate gyrus of the hippocampus [58][59][60] and of the subventricular zone [61]. Thus, immunohistochemical analysis was performed with NeuroD1 and NeuN antibodies as markers of early and late neuronal differentiation, respectively [62,63]. As shown in Fig 4A and 4B, the percentage of the NeuroD1-positive cells in the AAV-Tis21-treated MB allograft increases of about 39% vs control AAV-CBA-treated tumors (i.e., from 16.5% in AAV-CBA to 23% in AAV-Tis21; p < 0.0001). Interestingly, also the number of NeuN positive cells is significantly increased in the experimental allograft when compared with the control samples (about three-fold increase, from 4.3% in AAV-CBA to 13.2% in AAV-Tis21, p < 0.0001; Fig 4C and 4D). Therefore, importantly, Tis21 induces also the differentiation of MB cells in the subcutaneous allograft (Fig 4).
Overall, these results suggest that Tis21 overexpression in MB nodules reduces tumor growth via both antiproliferative and pro-differentiative effects.

AAV-Tis21 treatment modulates the expression of Tis21 target genes in MB allograft tissues
It is known that Tis21 acts as transcriptional co-regulator and that it controls transcriptional expression of several target genes, involved both in cell cycle and in neuronal differentiation. In particular, cyclin D1 mRNA levels decrease subsequent to Tis21 overexpression in different cell lines and tissues [51]. A similar effect is exerted by Tis21 on the Id3 gene, inhibitor of neuronal differentiation [59]. Thus, we thought to analyze by qRT-PCR the mRNA levels of some known Tis21 target genes and those of cyclins and NeuroD1 to investigate whether: i) the effect on proliferation exerted by Tis21 require the inhibition of cyclins (in particular of cyclin D1); and ii) the effect on neuronal differentiation occurs via Id3 regulation.
https://doi.org/10.1371/journal.pone.0194206.g003 AAV-Tis21 nodules relative to the AAV-CBA nodules (Fig 5). The weak increase, albeit significant, of Tis21 expression could be explained by the fact that the Tis21 signal within the nodule is not uniform, as observed by in situ hybridization (S2 Fig), thus yielding a low level of Tis21 mRNA detected by qRT-PCR within the whole nodule. Moreover, we cannot exclude a low mRNA stability in highly proliferating cells. However, the increased expression of Tis21 obtained with this gene therapy protocol was sufficient to cause a statistically significant decrease of cyclin D1 (15.4%, p = 0.00024) and cyclin E (16.7%, p = 0.049) mRNA levels with respect to control (Fig 5). No significant differences were observed for cyclin A2 and cyclin B1 expression levels (p = 0.85 and p = 0.061, respectively), suggesting that Tis21 influences the G1/S transition phase of cell cycle in this model system. As a further attempt to test the contribution of Tis21 on the G1/S transition we analyzed the expression of cdc6. This is a constituent of the replication initiation machinery and represents a marker of cell proliferation [64], overexpressed in different types of human cancers including medulloblastoma [65]. We observed that cdc6 mRNA levels are significantly reduced by Tis21 (31% decrease, p = 0.046; Fig 5). Finally, we recorded a statistically significant reduction of Id3 (10.8%, p = 0.044) and a significant increase of NeuroD1 (366%, p = 0.00022) mRNA levels.
It is worth noting that even a small change in cyclin D1 levels is sufficient to cause great changes in proliferation. In fact, our previous data in cerebellum showed that a limited but selective increase of cyclin D1 levels is associated to changes in GCPs proliferation [44]. Overall, these results demonstrate that Tis21 has a double action able to reduce proliferation and induce differentiation of primary MB cells by regulating key target genes.

Conclusion
Our study demonstrated, in a model of MB allograft in nude mice, that overexpression of Tis21 by AAV strongly impairs tumor growth under treatment pressure, and delays tumor growth in the short post-treatment time. The effect on tumor cell growth is likely to be due to both a decrease in cell proliferation (Ki67 and BrdU markers) and an increase of cell differentiation (NeuroD1 and NeuN markers). Moreover, the antiproliferative action exerted by Tis21 overexpression could be due to an effect on key cell cycle genes. In particular, as expected, Tis21 acts by inhibiting the cyclins that control the G1/S transition phase of the cell cycle. Moreover, Tis21 overexpression clearly induces an increase of NeuroD1 mRNA (about fourfold increase) and, in parallel, of NeuroD1-positive and NeuN-positive cells. The therapeutic relevance of this effect lies in the fact that tumor cells that have differentiated leave permanently the tumor program.
Our data suggest the possibility that the pro-differentiative action of Tis21, documented by the increase of NeuroD1 and NeuN expression in allografts, could be mediated by down-regulation of Id3 mRNA levels, as we previously reported in the analysis of dentate gyrus [59,60]. Of course it seems as well possible that the cyclin-dependent antiproliferative action of Tis21 contributes to the exit from cell cycle of tumor cells and ultimately to their differentiation.
Overall the results confirm the role of Tis21 as a MB suppressor gene and validate Tis21 as a potential relevant target for gene therapy in brain tumors. Furthermore, we observe that even with a limited expression of AAV in tumor cells in vivo, a large inhibitory effect on tumor growth is obtained. Further studies may aim to improve gene therapy protocols in terms of efficiency and duration of Tis21 expression.