DS is caused by an aneuploidy corresponding to the trisomy or partial trisomy of chromosome 21. The publication of the DNA sequence of the human genome revealed the presence of about 350 genes on chromosome 21. Although this number is low as compared to other human autosomes, it is still too large to develop transgenic mouse models for all of them. Moreover transgenic mouse models for single genes are not adequate to address fundamental questions on gene interactions. Thus in order to find the genes that contribute to the DS phenotype and to the phenotypic variability, gene expression profiling has been applied to mouse models with segmental duplications of DNA segments orthologous to human chromosome 21. The studies published so far were conducted in different mouse models, on different tissues and at various developmental stages. In addition the technology and the statistical tools were different thereby reducing the possibility of grouping the analysis and making insights into the genotype-phenotype status of DS. We thus designed a gene expression study in which we could measure the effects of trisomy 21 in the Ts1Cje mouse model, on a large number of samples, in a tissue that is affected in DS where we could quantify the effetcs during postnatal development in order to correlate gene expression changes to the phenotype observed.
DS affects many aspects of brain development and represents the most common cause of inherited human mental retardation. Motor deficits are among the most frequently occurring features of DS. Individuals with DS exhibit hypotonia and motor dysfunction that could be related to cerebellar dysfunction [33
]. DS patients present a consistent reduction of the total brain volume with a disproportionally greater reduction in the cerebellum [35
]. One of the major findings in Ts65Dn and Ts1Cje mouse models of DS carrying segmental duplications, is the significant reduction in cerebellar volume paralleling that observed in DS patients [18
]. This cerebellar hypoplasia likely results from a decrease in the density of granule cells, the most abundant neuronal cell population. Moreover, behavioural studies have shown motor dysfunctions in Ts65Dn that are more severely affected in tasks involving a high degree of balance and coordination [37
]. This hypoplasia occurs during cerebellar postnatal development in the first weeks after birth and does not result from a neurodegenerative process [19
]. More recently, a reduction of about 20% in the number of granule cell precursors has been reported at birth (P0) and 6 days after birth (P6) in the cerebellar external granule cell layer of Ts65Dn mice [20
]. This reduction is much more significant at P0 (p
= 0.001) than at P6 (p
= 0.04). In the present study we have shown a 17% significant decrease of volume in the cerebellum of adult Ts1Cje mice and a 32.6% significant decrease of granule cell precursor proliferation only at birth and not at P3 or P7. We analyzed global gene expression during postnatal development of the cerebellum in Ts1Cje mice at the time when major developmental changes occur: at birth (P0) at P3, P7 and P10. During this period granule cells proliferate, migrate from the external to the internal layer of the cerebellum, and differentiate. Our previous study performed at P0, P15 and P30 showed a clear over expression of triplicated genes in the cerebellum of Ts1Cje mice and a small effect on the expression of euploid genes [8
]. However since the proliferative external granule cell layer disappears by P15 [38
], this first study was not designed to detect gene expression changes between P0 and P15. To increase the statistical power of gene expression analysis, we used a large number of samples: 56 cerebella from Ts1Cje and euploid mice bred on the mixed (B6C3SnF1/Orl) genetic background. We performed an ANOVA with a type I error risk of 5% and we found between 2.4% to 7.5% of genes regulated at each time point including some three-copy genes. Then we applied a false discovery rate (FDR) with a 0.05 threshold (5%) on the gene list and found no significant gene at all time points, not even three-copy genes. By increasing the FDR threshold up to 0.40 we found only 47 significant genes among which 6 are three-copy genes (Additional file 3
, genes in grey boxes). These results suggest a limited secondary effect of trisomy on euploid genes.
Most three-copy genes were not found significantly differentially expressed suggesting that compensatory mechanisms take place during postnatal development. Only 6 three-copy genes were constantly over-expressed in the cerebellum of Ts1Cje mice at P0, P3, P7 and P10. Using qPCR, we confirmed the increased expression of six three-copy genes at P0. Four of these genes (Atp50
) were also found over-expressed in the study of Mao et al. conducted on DS foetal tissues, which included three cerebella [9
We then performed a second transcriptome analysis using dissected cerebellar regions enriched in granule cell precursors from 9 Ts1Cje and nine euploid mice at P0, at the time when we observed a significant decrease of cell proliferation. About 4% of all genes were significantly differentially expressed in Ts1Cje samples as compared to euploid samples. This value was close the ones obtained in the entire cerebellum at P0, P3, P7 or P10. All the expressed three-copy genes (35) were significantly over-expressed in Ts1Cje with 12genes showing expression ratio significantly lower than 1.5 (t-test, α = 5%, data not shown). After multi-test correction using the FDR procedure, 32 genes were still significantly modulated in Ts1Cje mice. Among them 23 are three-copy genes, indicating a major primary effect of trisomy on the three-copy genes.
From the euploid genes, 2 map to a chromosomal deletion (see below) and are under-expressed. The other seven are not well characterized and could be new gene candidates for cerebellar DS phenotypes and cerebellum postnatal development. By using an enriched and more homogeneous cell population (the granule cell precursors), we hoped to clarify the secondary effects of trisomy. Surprisingly, not only did we observed a main primary gene dosage effect (23 three-copy genes genes out of 32) but we also identified for the first time a deletion in the Ts1Cje mice (see below). Alternatively it is possible that the secondary effect of trisomy is variable among cells or since we demonstrated a proliferation defect in the granule cell precursors, it is plausible that we failed to identify cell cycle-regulated genes because cells from tissue are not synchronized.
In addition to the three-copy genes, we were able to detect a gene dosage effect on two genes mapping to mouse chromosome 12 (Mmu12), close to the translocation breakpoint of the trisomic segment. A 2 Mb segment of Mmu12 was found to be deleted in Ts1Cje mice. This segment contains 5 genes: Dnahc11
is a member of a family of zinc finger transcription factors and is highly expressed in the developing hippocampus and cerebellum, in particular in granule cells [39
]. Knock-down of Sp4
leads to an increased number of highly branched dendrites during maturation of granule cells in the cerebellar cortex [39
]. A reduced cell proliferation in hippocampus but not cerebellum has also been reported in Sp4
null mutant newborns [40
]. Thus, Sp4
, the only gene found expressed in the external granule cell layer is more likely to affect neuron maturation than proliferation. However it is unlikely that the cerebellum phenotype observed in Ts1Cje mice results from the Mmu12 deletion since a very similar phenotype is present in the Ts65Dn mouse model that do not show any rearrangement on Mmu12. We found a decrease in Cdca7l
gene expression, a transcription factor involved in cell apoptosis [41
] and medulloblastoma transformation [41
]. Since Cdca7l
is located approximately 200 kb proximal to the translocation breakpoint on MMU12 (Figure ) we concluded that the gene dosage effect extends to regions that are close to the chromosomal rearrangement.
We then looked for genes that were regulated during normal postnatal development of the cerebellum and that were also differentially expressed between Ts1Cje and euploid cerebellum at one of the postnatal stages tested. Among the 1187 genes found, only three were triplicated: Olig1
encodes a basic helix-loop-helix transcription factor that is expressed in both the developing and mature vertebrate central nervous system. Olig1
has critical function during the formation of motor neurons and oligodendrocytes of the ventral neural tube (review in [24
encodes an axon guidance molecule that is expressed by neurons in the central nervous system during development and throughout adult life. Its expression in the developing cerebellum is stronger in Purkinje neurons. A Dscam
mutation in mice leads to a subtle defect in the caudal folium of the cerebellum [42
is the only gene over-expressed in Ts1Cje across postnatal development of the cerebellum. It is also over-expressed in the external granule cell layer of Ts1Cje newborns. Girk2
is a member of the G protein-gated inwardly rectifying potassium channel family that regulates cellular excitability and neurotransmission. It is mainly expressed in the cerebellum granule cells [22
]. A missense mutation in Girk2
leads to abnormalities of the cell cycle and apoptosis in the external granule cell layer of the cerebellum in the weaver mice [43
]. The Girk2
mutated gene encodes channels that exhibit loss of potassium selectivity. In contrast, Girk2
knock-out mice are morphologically indistinguishable from wild-type mice, suggesting that the weaver phenotype is likely due to abnormal Girk2
]. Ion channels can regulate cell cycle and implication of transmembrane potassium fluxes via inward rectifier channels in the regulation of cell cycle has been proposed [45
Moreover it has been hypothesized that over-expression of the Girk2
subunit in trisomic mice will likely produce a hyperpolarization of the membrane [46
]. Depolarization enhances calcium entry via voltage-sensitive Ca2+
channels and activates calmodulin kinase and calcineurin phophatase. The activation of calcineurin induces many genes encoding extracellular and intracellular signalling molecules involved in granule cell development. [47
]. Thus, hyperpolarization of the membrane may prevent or reduce calcium entry in cells and decrease cell proliferation.
To further identify gene candidates for the DS cerebellar phenotype, we selected gene candidates involved in genetic regulation of the proliferation of the granule cell precursors [26
] and looked at their expression in the cerebellum of Ts1Cje and euploid mice at birth. Sonic hedgehog (Shh) is secreted by Purkinje cells and regulates proliferation of the granule cell progenitors [38
]. The mitotic response of cultured granule cell precursors to Shh was shown to be reduced in Ts65Dn mice as compared to euploids [20
]. Systemic treatment of newborn Ts65Dn mice with an agonist of the Sonic Hedgehog pathway increased the mitotic index of trisomic granule cell precursors and even restored the granule cell population in one week.
However, to date, genes involved in the Shh pathway have not been assigned to human chromosome 21 nor have they been shown to be differentially expressed in trisomic mice or in people with DS. Thus the positive effect of treatment with agonists of the Shh pathway could result from an over-stimulation of Shh receptors that are present on granule cell precursors. It remains to be shown whether this positive effect will normalize the cerebellar volume and improve the cerebellar-related behavioural deficits observed in Ts65Dn mice [37
Other genes have been reported to regulate the proliferation of granule cell progenitors. Some interact with the Shh pathway, such as Numb
which is a suppressor of Shh signaling [28
also regulates Notch1 pathway [28
]. Notch signalling is another crucial development-regulated pathway that appears to maintain cells in an undifferentiated state in the early mammalian central nervous system [49
]. In addition, we tested other genes related to granule cell proliferation or neural proliferation including IGF-I
]. We were not able to show any consistent difference of expression for any of these genes except a slight but significant over-expression of Shh
. Functional profiling of the differentially expressed genes from microarray experiments didn't reveal any enrichment in gene ontology categories related to cell proliferation.
Two genes over-expressed in the external granule cell layer of Ts1Cje mice at birth, S100a6
, encode calcium ion binding proteins. S100a6
belongs to the S100 family proteins that play an important role in cell growth, differentiation, and motility through calcium-dependent signalling pathways [53
encodes the calcyclin protein whose expression is associated with neuronal differentiation [54
]. This gene is thus a candidate for the cerebellar hypoplasia. Dlk1
(delta-like) is a transmembrane and secreted protein from the epidermal growth factor-like homeotic family. Dlk1
expression is increased in gliomas and over-expression in transfected cells promotes cell proliferation by inducing the expression of cyclin D1, CDK2, and E2F4 [56
]. However, we could not detect any change in expression of any of these genes suggesting that either Dlk1
has no effect on granule cell proliferation or that the effect is not visible because of lack of cell synchronization in tissues.
Recently, the generation of a collection of haploid yeast strains that each bear an extra copy of one or more of almost all of the yeast chromosomes has been published [57
]. Their characterization revealed that aneuploid strains share a number of phenotypes, including defects in cell cycle progression, that are independent of the identity of the individual extra chromosomes. It was thus proposed that disruption of cell homeostasy in DS could be due to the additional chromosomal material rather than the gene content [58
]. However aneuploidy is a condition frequently found in tumor cells which often display high rate of proliferation, suggesting that yeast and mammalian cells can respond differently to aneuploidy. Moreover in DS, reduced proliferation rates have not been observed in all proliferative cells across development and in adulthood. We show here a reduced proliferation of granule cell precursors only at birth.
Several studies have reported evidence of neurogenesis impairment in the developing neocortex, in the dentate gyrus of DS foetuses and in mouse models [59
]. In the dentate gyrus, cell proliferation is decreased in aged but not young adult Ts65Dn mice [62
] suggesting a neurodegenerative related process occurring in trisomic mice. Another study has reported a lower cell proliferation due to cell cycle alteration in the dentate gyrus of foetuses with DS and in mouse models. Using different markers of proliferation the authors showed that both DS foetuses and P2 Ts65Dn mice have a higher number of proliferating cells in G2 and a smaller number of cells in the M phase of the cell cycle [59
]. A similar result was reported in the forebrain of Ts65Dn embryos [60
] where delayed expansion of neocortical layers and reduced growth of the hippocampus were found to be correlated with a slower cell cycle. Interestingly, their results suggest that the cell cycle abnormality, occurring in the trisomic ventricular zone of the neocortex but not in the hippocampus, might be developmentally regulated or compensated during neurogenesis. Thus, neurogenesis impairment occurs in different areas of the developing brain in trisomy 21 but not continuously, suggesting that genes involved might be different depending on the brain structure and the developmental stage.