Loss of p53 in quaking viable mice leads to Purkinje cell defects and reduced survival.

The qk(v) mutation is a one megabase deletion resulting in abnormal expression of the qkI gene. qk(v) mice exhibit hypomyelination of the central nervous system and display rapid tremors and seizures as adults. The qkI locus on 6q26-27 has also been implicated as a candidate tumor suppressor gene as the qkI locus maps to a region of genetic instability in Glioblastoma Multiforme (GBM), an aggressive brain tumor of astrocytic lineage. As GBM frequently harbors mutations affecting p53, we crossbred qk(v) and p53 mutant mice to examine whether qk(v) mice on a p53(-/-) background have an increased incidence of GBM. qk(v) (/v); p53(-/-) mice had a reduced survival rate compared to p53(-/-) littermates, and the cause of death of the majority of the mice remains unknown. In addition, immunohistochemistry revealed Purkinje cell degeneration in the cerebellum. These results suggest that p53 and qkI are genetically linked for neuronal maintenance and survival.

The study of neurological diseases has been greatly advanced with the use of mouse models. In particular, spontaneously occurring mutant mice have played an important role in identifying key components required for proper myelination. One such mouse, the autosomal recessive qk mutant, was first described in 1964 by Sidman and colleagues. The qk/ mouse exhibited severe hypomyelination of the central and peripheral nervous system and severe hindlimb shaking1. The mutation responsible for the qk/ phenotype was later identified as a one megabase deletion that results in loss of parkin (park2) and parkin co-regulated (pacrg) function, whereas qkI expression becomes abnormally expressed234. In qk/ mutants, qkI-6/7 expression is preferentially lost in myelin producing cells of the central and peripheral nervous systems, while for the most part expression remains unaffected in all other cell types3. The qkI gene encodes for the alternatively spliced KH domain RNA binding proteins QKI-5, -6, and -75. QKI has been shown to regulate important mRNA targets involved in myelination and oligodendrocyte differentiation67891011. QKI is implicated in myelinogenesis by stabilizing myelin basic protein (MBP) mRNA612, regulating the alternative splicing pattern of myelin associated glycoprotein (MAG)1013, and oligodendrocyte differentiation and Schwann cell differentiation714. The over-expression of QKI-6 in qk mutants has been shown to rescue the myelination defect15.

In addition to myelination defects, qk mice also exhibit ciliopathies including male sterility and mild hydrocephalus1617. Cilia lining the ventricles of the brain are responsible for proper circulation of cerebral spinal fluid (CSF). Ciliogenesis is normal in qk mice, however the cilia are functionally impaired leading to decreased CSF flow and hydrocephalus16. Hydrocephalus in qk mice becomes fatal on a patched1 (ptch1) heterozygous background18.

The long arm of chromosome 6 is known to harbor a common fragile site that is frequently lost or mutated in a variety of cancers1920. The breakpoint includes several genes, one of which has been identified as the qkI gene2122. Chromosomal aberrations at 6q26-27 are common in Glioblastoma Multiforme (GBM)21. In addition to mapping to a region of genomic instability, expression of qkI mRNA transcripts have been shown to be altered specifically in human gliomas compared to other brain cancers23. Thus, it has been proposed that QKI may function as a tumor suppressor and play a role in GBM progression. Evidence for a suppressor role was shown for the C. elegans GLD-1, a QKI homolog24. The GLD-1 C. elegans homolog of QKI is known to associate with the p53 cep-1 mRNA, influencing its activity25. In addition, QKI was discovered to be down-regulated in colorectal cancers26. QKI over-expression in the colon epithelium resulted in increased levels of p27kip1 as well as an increase in membrane bound β-catenin, an indicator of gastric cell differentiation26. These observations along with our data that QKI-6/-7 cause cell cycle arrest7, a property often observed for tumor suppressors, suggest that the absence of QKI proteins may promote tumorigenesis.

Loss of heterozygosity and mutations of the p53 gene is commonly associated with a variety of cancers in multiple organ sites27. Individuals with the Li Fraumeni syndrome, in part caused by mutations in the p53 gene, suffer from early onset of many different types of cancer28. Mutations abrogating p53 function and allelic loss of chromosome 17p were among the first genetic lesions identified in GBM29. p53 lesions are present in all grades of astrocytoma at an average of 30%30, suggesting that the inactivation of p53 is an early event in gliomagenesis. p53 and p53 mice have been found to develop a broad spectrum of tumors, including lymphomas, osteosarcomas, and fibroscarcomas31. p53 and p53 mice have been used as a background strain to breed with other mutant strains to produce entirely novel tumor spectrums that were not found in mice with a mutation in p53 alone32. Although p53 mice showed no incidence of any brain malignancy3133, nf1 mice developed astrocytomas with a penetrance of 92% by 6 months of age34.

Here we report that qk/ mice on a p53 background had a reduced survival rate compared to qk/; p53and p53 controls, and the cause of death of the majority of the mice remains unknown. qk/; p53 mice also displayed neurological defects including Purkinje neuron degeneration.

Results

The qk mice contain a recessive autosomal mutation of ∼1.1 Mb on chromosome 17 affecting the expression of quaking (qkI), parkin-coregulated gene (pacrg) and parkin (park2)4. The mutation results in aberrant expression of alternatively spliced qkI transcripts, leading to the dysmyelination defects observed in the qk mice4. Deletion of the pacrg gene in qk mice causes ciliopathies including hydrocephalus16 and male sterility35. To examine whether the p53 pathway genetically interacts with the pacrg-parkin-qkI locus in the regulation of the phenotypes of the qk mice, we crossbred the qk mice with p53 null mice. We initially assessed the myelin sheath of resulting qk/; p53 progeny. Post-natal day 30 (P30) coronal sections of the corpus callosum were stained with an antibody specific to myelin basic protein (MBP). As expected, qk/ mice showed reduced thickness of the corpus callosum (98.8 ± 16.4 μm) compared to wild-type (wt) (172.7± 7.03 μm) and qk/+; p53 (191.8 ± 5.99 μm) littermates (Figure 1). However, qk/; p53 mice did not show an increased severity of hypomyelination compared to qk/ mice (86.9 ± 8.89 μm, Figure 1), and corpus callosum thickness was not found to be statistically different between the mice (p = 0.287). Although corpus callosum thickness was reduced in mice with the qk mutation, sparse high intensity MBP staining was observed along the myelin tracts. These MBP-positive “blebs” have previously been observed in spinal cord white matter of qk/ mice, corresponding to redundant loops of uncompacted myelin36. These results suggest that p53 does not cooperate with qkI in the regulation of myelin formation in vivo. The loss of both wt p53 alleles did not affect the onset of hindlimb shaking in mice pups, since both qk/; p53 mice, qk/; p53 mice and qk/ mice displayed tremors by P14 (data not shown). Common husbandry procedures and animal handling are sources of acute stress and elicit tonic clonic seizures in qk mice373839. The onset of stress induced tonic-clonic seizures in qk/; p53 mice was significantly earlier than that of qk/ mice. qk/ mice began demonstrating stress-induced seizures at 12 weeks of age, whereas qk/; p53 mice and qk/; p53 mice were observed to have seizures and ataxic movements as early as 4 weeks of age (data not shown). As expected, mice heterozygous for the qk mutation did not demonstrate tonic-clonic seizures or hindlimb shaking.

Figure 1

qk-associated hypomyelination is not exacerbated on a p53background.

Brains were frozen over acetone-dry ice and cryostat sectioned at a thickness of 10 µm. Coronal sections of age and sex-matched mouse cortexes were stained with anti-MBP antibody. Scale bar represents 100 µm.

We assessed the cellular morphology of Purkinje cells by immunostaining P30 coronal brain sections with an antibody against Calbindin, a Purkinje neuron-specific marker. Immunostaining of the brains of qk/; p53 mice revealed Purkinje cell body loss as well as loss of Calbindin-positive Purkinje cell dendritic arbors, whereas control mice demonstrated normal Purkinje cell appearance (Figure 2A). In order to quantitatively assess differences in Purkinje cell numbers, we analyzed the Purkinje cell linear density of P30 mice. qk/; p53 mice were found to have a significantly lower Purkinje cell density (16.36 ± 1.42 cells/mm) compared with wt (38.77 ± 2.52 cells/mm; p < 0.001), qk/; p53(42.45 ± 6.5 cells/mm; p = 0.006), and qk/ controls (40.16 ± 7.9 cells/mm; p = 0.01). Misplaced Purkinje neurons have occasionally been observed in qk/ mice40. To determine if the Purkinje defects observed were due to impaired migration or other developmental defects, immunostaining was performed on brains of P14 mice. Purkinje cell morphology was normal in qk/; p53 P14 mice, suggesting that Purkinje cell loss was not due to failure in neuronal migration at earlier stages of development (Figure 2B). These results suggest that the Purkinje cell defects observed in qkmice were not due to impaired development, but cell body and dendrite degeneration and defects in neuronal maintenance and survival. Indeed, qkI has been shown play a role in apoptosis in 3T3 mouse fibroblasts, HeLa cells, and primary rat oligodendrocytes4142.

Figure 2

qk/; p53 mice display Purkinje cell defects.

(A) qk/; p53 mice display Purkinje cell body and dendritic arbor loss at P30 (arrows), whereas control littermates exhibit normal cerebellar architecture. Sections were stained with anti-Calbindin antibody. Scale bar represents 50 µm. (B) qk/; p53 show normal Purkinje cell morphology of the cerebellum at P14. Coronal sections of the cerebellum were stained with anti-Calbindin antibody. Scale bar represents 50 µm.

p53 mice develop primarily sarcomas, with an incidence of 28% over 17 months. p53 mice have a much accelerated rate of tumorigenesis, with the majority of mice succumbing to lymphomas by six months of age31. qkI has been implicated as a candidate tumor suppressor gene as the qkI locus maps to a region of genetic instability in GBM22, therefore we examined whether qk mice would display a higher incidence of brain tumors and thus may die earlier than mice deficient for p53 alone. Kaplan-Meier curve analysis was performed in order to compare mouse survival rates (Figure 3A). In agreement with previous observations1, none of the qk/ mice cohort succumbed to tumor formation at the end of the 261 day observation period. Overall, qk/; p53 mice were found to have a reduced survival time compared to the other groups, with a median survival time (MST) of 119 days. qk/; p53 mice showed a significantly reduced survival time compared to qk/; p53 mice (MST = 142 days; p = 0.0135) as well as p53 mice (MST = 172 days; p = 0.0443) according to the log rank/Mantel Cox test. There was no significant difference between the survival rates of qk/; p53and p53 mice according to the Mantel Cox test (p = 0.9175).

Figure 3

qk homozygous mice show a reduced survival rate in the absence of p53.

(A) qk/; p53 mice (n = 18) were observed to have a median survival time of 119 days, significantly lower compared to qk/; p53 mice (142 days, n = 18) and p53mice (172 days, n = 11) according to the log rank/Mantel Cox test (p = 0.0135 and p = 0.0443, respectively). The loss of survival was measured by mice that were found dead or that had to be sacrificed due to illness according to the guidelines of the Canadian Animal Care Committee. (B) Low-power magnification showing macroscopic appearance of medulloblastoma occurring in the cerebellum, with an arrow denoting the tumor area. Higher magnification of medulloblastoma, demonstrating the invasion of the molecular layer by the over proliferation of granular like tumor cells. Section was stained with H&E. Immunohistochemistry performed on the medulloblastoma reveals tumor reactivity to GFAP, NFH, and CD31, specific markers for different cell lineages.

Previous studies have shown that the majority of p53 mice develop tumors of the lymphatic system and soft tissue sarcomas, with only one instance of a brain tumor reported in a p53 mouse3133. Similarly, the p53 and qk/; p53 mice we generated developed primarily sarcomas and lymphomas, with no occurrence of any brain malignancies (Table 1). In contrast, none of the qk/; p53 mice demonstrated any occurrence of sarcomas, lymphomas, or tumors previously described in p53mice (Table 1). Several qk/; p53 and qk/; p53 mice displayed neurological symptoms characteristic of hydrocephalus. Brains examined from these mice showed enlarged lateral ventricles with an accumulation of CSF, similar to the hydrocephalic phenotype observed in qk mice18. The brain of one qk/; p53 mouse showed increased vascularization of the cerebellum and abnormal gross cerebellar architecture. Further histological analysis of cerebellar sections revealed granule cell layer invasion of the surrounding parenchyma, consistent with medulloblastoma (Figure 3B). Cytological analysis confirmed the presence of medulloblastoma, as tumor cell nuclei were densely packed and polygonal in shape. Tumor cells were also found to express markers of astrocytic (GFAP), neuronal (NFH) and vascular (CD31) lineage (Fig. 3B). Cells expressing markers of multiple lineages indicate cells of a stem cell precursor nature, consistent with primitive neuroectodermal tumors such as medulloblastoma. Medulloblastoma has not previously been observed to occur spontaneously in either p53 or qk mice131. One study identified only one instance of a brain malignancy out of a cohort of p53+/− mice (n = 40), whereas another study did not observe any tumors of the nervous system in a cohort of p53mice (n = 35) or p53 (n = 96) mice3133. The one p53 brain tumor identified was an ependymoma, a tumor originating from the walls of the ventricular system43. p53 mutations contribute to approximately 10% of human medulloblastoma cases44; in addition, many mouse models of medulloblastoma have been developed by the combination of a specific tumor suppressor mutation and the loss of p5332. Although qkI has been mapped to the common fragile site 6q25-26 in GBM, loss of 6q25-26 in medulloblastoma is rare45, suggesting that somatic qkI mutations in clinical medulloblastoma do not confer any specific tumorigenetic advantage. QKI has been shown to be expressed in post-natal neural progenitor cells of the SVZ46, suggesting that QKI may also be expressed in other populations of multi-potent precursor cells including the rostral rhombic lip progenitor cells that give rise to granule neuronal precursor cells. It has not been determined if the qk mutation affects QKI expression in neural progenitor cells including granule neuronal precursor cells, therefore it remains a possibility that the loss of qkI-6/7 in the absence of p53 could provide sufficient oncogenic stress to induce transformation.

Table 1

Tumors in qk/; p53 mice a

Genotype	Total no. of mice	Brain tumors	Hydrocephalus	Subcutaneous tumors	Thymic lymphoma	Intraperitoneal tumors	Other
qkv/v; p53−/−	11	1(11%)	2(18%)	0	0	0	0
qkv/v; p53+/−	33	0	5(15%)	0	0	0	0
qkv/+; p53−/−	12	0	0	2(15%)	8(61%)	3(23%)	0
p53−/−	9	0	0	0	5(55%)	3(33%)	1(11%)

a. tumors were classified according to tumor location.

Discussion

Our study indicates that qk; p53 mice do not have decreased corpus callosum thickness or MBP-staining compared to qk littermate controls, suggesting that the qk-associated myelination defect is not further impaired by the loss of p53. These findings indicate that qkI does not cooperate with p53 in vivo to regulate the oligodendrocyte differentiation pathway. qk; p53 mice displayed Purkinje-cell defects in the cerebellum characterized by dendritic arborization defects and cell body loss, suggesting that qkI is required for neuronal cell maintenance in the absence of p53. However, the exact mechanism by which this phenotype occurs has yet to be elucidated. Cerebellar defects have previously been documented in mice homozygous for the qk mutation. Three month old qk mice displayed axonal swellings in both the Purkinje and granular cell layer, indicative of axonal injury40. Axonal swelling is characteristic of inflammatory lesions in Multiple Sclerosis and experimental autoimmune encephalitis, suggesting that axonal pathology is secondary to myelination defects4849. While our work showed that the qk dysmyelinating phenotype was not exacerbated on a p53 background, Purkinje cell degeneration was only observed in qk; p53 mice and not qk controls, suggesting that the neuronal pathology was not due to myelination defects. Interestingly, qk; p53 mice displayed stress-induced tonic-clonic seizures and ataxic movements at about one month of age that was not observed in their qk counterparts. Disruption in normal cerebellar architecture is one cause of ataxia50, thus Purkinje cell degeneration may be related to the ataxic-like phenotype observed in qk; p53 mice. Indeed, 29 QKI-interacting partners were identified in a protein interaction network generated for human inherited cerebellar ataxias51.

qk; p53 mice failed to develop tumors characteristic of p53mice, but demonstrated a reduced survival rate compared to both qk/; p53 and p53mice. One case of medulloblastoma was documented in a cohort of qk; p53mice (n = 11), while 2 other mice demonstrated signs of hydrocephalus. Recently, we have shown that qk/mice on a ptch1background have a reduced survival rate due to the development of fatal hydrocephalus18. A mild hydrocephalic phenotype occurs in qk mice due the loss of PACRG expression in the ciliated ependymal cells lining the ventricular walls, leading to cilia dysfunction16. Similarly, qk/v; ptch1mice showed abnormal cilia function, leading to accumulation of CSF in the ventricles and hydrocephalus18.

The majority of p53 and qk; p53mice in our study succumbed to thymic lymphoma, demonstrating clear symptoms including difficulty breathing due to tumor-induced compression of the lungs, enlarged ribcages, and lethargy due to tumor burden. Necropsy of these mice revealed highly vascularized, enlarged thymuses that completely encompassed the heart and the majority of the lungs, and in some instances, extended outside the ribcage into the subcutaneous layers. These symptoms were not observed in any of the qk; p53 mice before death, and necropsy revealed normal thymic morphology. In addition, necropsy of qk; p53 mice did not reveal the presence of any overt cancers. Thus, the cause of death of the majority of the mice remains undetermined. Several possible explanations exist for the reduced survival rate observed in qk/; p53 mice. The qk/ mutation results in brain-specific loss of qkI isoforms -6/7, resulting in preferential sensitization of the nervous system to tumorigenesis in the absence of p53. As mentioned earlier, our post-mortem pathological observation methods were limited, thus it is possible that qk/ mice developed additional brain tumors that remained undetected, including medulloblastomas. Although we suspect that qkI and p53 participate in different signaling pathways since double mutant qk; p53 mice displayed a novel phenotype not observed in either parent strain, a bioinformatics analysis has identified p53 as a putative target of QKI in mice. The p53 mRNA harbors a putative QRE (QUAKING Response Element) within its 3'UTR (UACUAACnnnnGAAG)47. These findings indicate that QKI may regulate p53 mRNA levels in vivo. Consistent with these findings, the C. elegans homolog of QKI, GLD-1, is known to associate with the p53 cep-1 mRNA and regulate its activity25.

In conclusion, QKI proteins are highly conserved among species and have been implicated in multiple diverse biological pathways, highlighting the pivotal role of RNA binding proteins in cellular function. Our data suggests that qkI interacts with p53 in vivo to regulate neuronal maintenance and survival.

Methods

Antibodies

Monoclonal MBP and NFH antibodies were obtained from Sternberger Monoclonals (Baltimore, MD). Monoclonal Calbindin-D-28K and GFAP antibodies were purchased from Sigma (St. Louis, MO). Anti-CD31 was purchased from Millipore (Billerica, MA). Antibodies against QKI-5,-6, and -7 were generated as described previously742.

Animals

Mice were monitored daily and were handled and sacrificed in accordance with a protocol approved by the Animal Care Committee at McGill University. The mice were housed in ventilated cages with a 12/12 hour light/dark cycle. p53 mice (lab stock # 002101) and qk mice (lab stock # 000506) were obtained from The Jackson Laboratory (Bar Harbor, ME). Mice colonies were maintained on a C57BL/6 background. Mice homozygous for the p53 mutation were crossbred to heterozygous qk mice and subsequent trans-heterozygous crosses were used to generate qk males. These males were subsequently crossbred to trans-heterozygous females to generate double mutant mice. Mice were screened for the p53 wild-type allele via genomic PCR using oligonucleotides (5′-ATA GGT CGG CGG TTC AT-3′) and (5′-CCC GAG TAT CTG GAA GAC AG-3′). The mutant p53 allele was identified using the following oligonucleotides: (5′-CTT GGG TGG AGA GGC TAT TCG-3′) and (5′-GTC GGG CAT GCG CGC CTT GAG-3′). The qk mutation was amplified using primers directed against the breakpoint (5′-TCT AAA GAG CAT TTT CGA AGT-3′) and (5′-TTG CTA ACT GAA TAT TAC T-3′).

Immunohistochemistry

Mice were anaesthetized with isoflurane and perfused with ice-cold phosphate buffered saline followed by 4% paraformaldehyde. Brains were cryoprotected in 30% sucrose overnight and embedded in OCT compound (Tissue-Tek, Markham, ON) over dry ice in acetone. Tissues were cryostat sectioned at a thickness of 10 µm and collected on +/+ slides (Fisher, Ottawa, ON). Tissue sections were blocked in 10% goat serum in Tris-buffered saline + 0.5 % Triton X-100 for 1 hr followed by incubation with primary antibodies overnight at room temperature. Slides were incubated with Alexa-flour 488 or 546 Immunoglobulin G (Invitrogen, Carlsbad, CA) at a dilution of 1:400 for 4 hr. Corpus callosum thickness was visualized by MBP staining and quantified for 4 different areas using Axiovision software (Zeiss). P14 and P30 mice cerebellar coronal sections were evaluated at bregma ∼−5.8 mm, and representative images were taken of the simple lobule adjacent to the primary fissure. Purkinje cell linear density was quantified for 3 different areas using Axiovision software. Purkinje cell linear density was quantified as Calbindin-positive cell bodies/mm Purkinje cell layer. Statistical significance was calculated according to a two-sample two-tailed paired student t-test.

Author Contributions

The project and all experiments were designed by C.G. and S.R. All experiments and statistical analysis were performed by C.G. The manuscript was written by C.G. and S.R.