Human Molecular Genetics Advance Access originally published online on January 10, 2006
Human Molecular Genetics 2006 15(4):607-623; doi:10.1093/hmg/ddi477
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Deletion of the triplet repeat encoding polyglutamine within the mouse Huntington's disease gene results in subtle behavioral/motor phenotypes in vivo and elevated levels of ATP with cellular senescence in vitro
Department of Neuroscience, University of Virginia School of Medicine, PO Box 801392, Charlottesville, VA 22908-1392, USA
* To whom correspondence should be addressed at: Department of Neuroscience, University of Virginia School of Medicine, 409 Lane Road, MR4, Room 5022, Charlottesville, VA 22908-1392, USA. Tel: +1 4349245011; Fax: +1 4349824380; Email: soz4n{at}virginia.edu
Received November 8, 2005; Accepted January 4, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Huntingtin (htt), the protein encoded by the Huntington's disease (HD) gene, contains a polymorphic stretch of glutamines (polyQ) near its N-terminus. When the polyQ stretch is expanded beyond 37Q, HD results. However, the role of the normal polyQ stretch in the function of htt is still unknown. To determine the contribution of the polyQ stretch to normal htt function, we have generated mice with a precise deletion of the short CAG triplet repeat encoding 7Q in the mouse HD gene (Hdh
Q). Hdh(Q/Q) mice are born with normal Mendelian frequency and exhibit no gross phenotypic differences in comparison to control littermates, suggesting that the polyQ stretch is not essential for htt's functions during embryonic development. Adult mice, however, commit more errors initially in the Barnes circular maze learning and memory test and perform slightly better than wild-type controls in the accelerating rotarod test for motor coordination. To determine whether these phenotypes may reflect an altered cellular physiology in the HdhQ mice, we characterized the growth and energy status of primary embryonic and adult Hdh(Q/Q) fibroblasts in culture. The HdhQ fibroblasts exhibited elevated levels of ATP, but senesced prematurely in comparison with wild-type fibroblasts. Taken altogether, these results suggest that htt's polyQ stretch is required for modulating longevity in culture and support the hypothesis that the polyQ stretch may also modulate a htt function involved in regulating energy homeostasis. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Huntington's disease (HD) is a fatal autosomal dominant neurodegenerative disorder characterized by midlife onset and a triad of cognitive, motor and emotional disturbances (1
). HD is caused by the expansion of a CAG triplet repeat encoding polyglutamine (polyQ) within huntingtin (htt), the protein product of the IT15 (HD) gene (2). Unaffected individuals have 6–34 CAG repeats while more than 37 repeats result in HD. HD is one of nine polyQ disorders that affect different neuronal populations in the central nervous system (3,4). Although the expanded stretch of polyQ in all of these disorders is thought to confer a deleterious gain-of-function, the normal functions of these proteins likely play important roles in determining neuronal specificity in the different polyQ disorders.
Htt is a large (348 kDa), predominantly cytoplasmic protein that is also found at low steady-state levels in the nucleus (5,6). Htt is essential during early embryonic development (7–9) and is also required during development of the brain (10–12) and in spermatogenesis (10). Htt's normal function(s) may impact many cellular processes including signal transduction, endocytosis, cytoskeletal structure, transcription and axonal transport (13–15). The large number of putative functions for htt is also reflected in the large number of interacting proteins that have been characterized to date (16). Analysis of the htt primary amino acid sequence reveals (sequentially from the N-terminus) a group of three lysine residues within the first 17 amino acids of htt that are a substrate for SUMOylation (17), a polyQ stretch followed closely by a proline-rich domain, a lipid-binding domain (18) located within one of 10 clumped HEAT repeats (an acronym based on members of a protein family containing similar 
38 amino acid repeats: Huntingtin, Elongation Factor 3A, A subunit of Protein Phosphatase 2A and TOR1) (19) and a nuclear export signal near the C-terminus (5). The HEAT repeats favor an alpha helical structure and are implicated in modulating protein–protein interactions (20). This has led to the hypothesis that htt, like other HEAT domain containing proteins, can act as a scaffold mediating various cellular processes. The role of any of these domains in the normal function of htt has yet to be determined in vivo.
The polyQ stretch is conserved among vertebrates, ranging from an average length of approximately 20 glutamines in humans to only four in zebrafish and pufferfish (16,21,22). Interestingly, Drosophila htt does not have a polyQ stretch (23), suggesting the possibility that the polyQ stretch may have a function unique to vertebrates. In lymphoblastoid cell lines derived from HD patients, the length of the polyQ stretch influences cellular energy status, with an inverse relationship between polyQ length and the cell's ATP/ADP ratio. Moreover, this relationship is also valid for short stretches of polyQ (<35Q), suggesting that polyQ length can modulate energy status even in normal cells (24). The impact of deleting polyQ on the normal function of a vertebrate htt, however, has not yet been explored.
To begin investigating the contribution of the polyQ stretch to normal htt function in vivo, we have generated mice with a precise deletion of the (CAG)2CAA(CAG)4 triplet repeat encoding 7Q within the mouse homolog of the HD gene (HdhQ). We have found that the polyQ stretch is not essential for normal htt function during development, but Hdh(Q/Q) homozygous adult mice exhibit more errors initially in a learning/memory test and show a subtly enhanced motor coordination phenotype in the accelerating rotating rod (rotarod) test. In addition, primary fibroblasts obtained from embryos homozygous for the Q deletion exhibit an elevated level of ATP and senesce more rapidly than wild-type controls. Our results support the hypothesis that the polyQ stretch within htt is required for modulating proliferative life span in primary cell culture and that one of htt's normal functions may also be involved in regulating energy homeostasis in vertebrates.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Derivation of mice carrying the Hdh
Q alleleTo generate a precise deletion of the Hdh (CAG)2CAA(CAG)4 triplet repeat encoding 7Q, a gene-targeting vector was assembled, which contained both the deletion of the (CAG)2CAA(CAG)4 repeat (
Q) and, for the purpose of following expression of Q-htt, a FLAG epitope tag located at the N-terminus of the coding sequence within exon 1 (Fig. 1A). We note that the addition of an N-terminal FLAG tag has the potential to alter normal htt localization in vivo. However, in cell culture, a FLAG tag at the N-terminus of htt does not interfere with its expression or localization (18,25). Moreover, mice carrying a triple FLAG tag (FLAG3X) at the N-terminus of endogenous htt appear phenotypically normal (manuscript in preparation). For positive selection of transfected ES cells, a neomycin phosphotransferase gene cassette flanked by loxP sites (floxed pgkneo) was inserted 1.3 kb upstream of the Hdh transcriptional start site and in the opposite transcriptional orientation with respect to Hdh (Fig. 1B). Although the loxP sites provide a means to remove the pgkneo cassette via Cre-mediated recombination, the location of the pgkneo cassette does not interfere with Hdh expression (26,27) and provides a convenient marker for routine genotyping. Targeted ES cell clones were identified by Southern and PCR analyses (Fig. 2A and B). For PCR detection of the Q mutation, an oligonucleotide primer spanning the site of the (CAG)2CAA(CAG)4 repeat deletion was used to amplify specifically sequence from the mutant allele. In addition, primers flanking the epitope tag insertion site were used to discriminate between the wild-type and HdhQ alleles. Germline transmission of the HdhQ allele (Hdhtm3Szi, GenBank accession no. MGI: 3603606) was obtained from two targeted ES cell lines. All mice analyzed in this study were of a mixed strain background (C57BL6/129Sv).
|
|
-phage DNA (left). On the right, lane 1 represents an Hdh(+/+) mouse, whereas lanes 2 and 3 are Hdh(The polyQ domain is not essential for htt function during embryonic development
Htt is essential for embryonic development, and mice lacking htt expression die between embryonic day 7.5 (E7.5) and E9, shortly after the onset of gastrulation (7
–9). If htt's embryonic function is compromised by the Q mutation, the fraction of Hdh(Q/Q) homozygotes obtained from an Hdh(Q/+) intercross should be lower than the predicted Mendelian distribution of 25%, and any surviving Hdh(Q/Q) neonatal pups may exhibit phenotypic abnormalities. In an Hdh(Q/+) intercross, approximately one-quarter of the progeny were Hdh(Q/Q) (Table 1), and moreover, Hdh(Q/Q) pups were also indistinguishable from Hdh(Q/+) and wild-type littermates at birth (data not shown). These results indicate that lack of the polyQ stretch does not interfere substantially with htt's normal function during development. Genotypes were obtained by both Southern and by PCR analyses to confirm the presence of the FLAG tag, the pgkneo cassette and the Q mutation (Fig. 2A and B). To ensure that the Q mutation and FLAG tag did not affect Q-htt expression, western analysis of whole brain extracts from Hdh(Q/Q) and age-matched wild-type littermate controls (n=3) was performed. An anti-FLAG antibody recognized a htt-sized polypeptide only in the Hdh(Q/Q) brain extracts, whereas an anti-htt antibody (Mab 2166) reacts with both wild-type and Q-htt in both the Hdh(+/+) and Hdh(Q/Q) samples (Fig. 2C). Densitometry of the western exposures, using the level of glyceraldehyde phosphate dehydrogenase (GAPDH) as a loading control, revealed that there was no significant difference in the amounts of htt in the samples (P=0.27, Fig. 2D), thus indicating that the polyQ deletion does not affect htt expression levels.
|
The
Q mutation does not affect htt's subcellular localizationWild-type htt is a predominantly cytoplasmic protein that is localized in the perinuclear region. To determine whether the polyQ deletion affected the subcellular localization of
Q-htt, primary fibroblasts were prepared from embryonic day 13 (E13) Hdh(Q/Q) and wild-type embryos. Subcellular fractionation of passage 4 (P4) cells revealed no difference in the cellular compartmentalization of wild-type and Q-htt, with most htt residing in the cytoplasm (Fig. 3A). Grb2, an SH2/SH3 domain adaptor protein (28), and GAPDH were used as a positive controls for predominantly cytoplasmic proteins, whereas p53 is predominantly nuclear (29), and htt interacting protein-1 (HIP-1) is an example of a protein that partitions both the cytoplasm and the nucleus by virtue of a nuclear localization signal at its C-terminus (30). In our primary mouse embryonic fibroblasts (PMEFs), the majority of HIP-1 is found in the nucleus. To confirm that wild-type htt and Q-htt had similar subcellular localization properties, immunohistochemical analyses were performed on Hdh(Q/Q) and Hdh(+/+) fibroblasts and brain sections using MAb 2166. Htt staining was punctate and perinuclear in both Hdh(Q/Q) and control P4 fibroblasts, and predominantly cytoplasmic in neurons, confirming that there is no significant difference in the subcellular localization of wild-type and Q-htt (Fig. 3B).
|
Although
Q-htt's subcellular localization is indistinguishable from wild-type htt, deletion of the polyQ stretch may affect Q-htt's associations with interacting proteins, especially those known to be sensitive to the length of the polyQ stretch. To examine Q-htt's interactions with protein partners, we performed co-immunoprecipitation analyses with wild-type and Hdh(Q/Q) whole brain extracts using MAb 2166. Both wild-type and Q-htt were immunoprecipitated with similar efficiency, and there were no significant differences observed in the co-precipitation of HIP-1, PSD-95 and GAPDH (Fig. 3C). Although neither HIP-1 nor PSD-95 interacts directly with polyQ, the strength of their interaction with htt is reduced when htt's polyQ stretch is expanded into the disease range (31,32). GAPDH interaction with Q-htt was examined on the basis of previous reports suggesting that GAPDH associates with htt via the polyQ stretch (33,34). However, our results suggest that other regions within htt must be involved, as the interaction of GAPDH with Q-htt is comparable to its interaction with wild-type htt. Thus, although expansion of the polyQ stretch within htt is known to perturb htt's binding to HIP-1, PSD-95 and GAPDH, deletion of the polyQ stretch does not have a similar effect on htt's interaction with these proteins.
Hdh(Q/Q) mice exhibit subtle behavioral and motor phenotypes
There was no significant difference in weights found between Hdh(Q/Q) females and wild-type females or between Hdh(Q/Q) males and wild-type males at any of the ages that were examined (Table 2). Limb clasping, a behavior indicative of neurological dysfunction in rodents, which is seen often in expanded polyQ and HD mouse models (35–37), including a conditional htt loss-of-function model (10), was not observed in Hdh(Q/Q) mice up to the oldest age examined (1 year of age). Gross motor deficits were not observed in Hdh(Q/Q) mice subjected to a cage-top rotation test and a wire-rod hanging test (38). In addition, no obvious anatomical abnormalities were detected in Cresyl violet-stained sections of brains obtained from 1-year-old Hdh(Q/Q) mice and wild-type littermate controls (Fig. 4). Moreover, there was no significant difference in brain weights obtained from female wild-type (+/+) and Hdh(Q/Q) littermates (+/+=0.43±0.03 g; Q/Q=0.45±0.08 g; P=0.70 for brain weight and P=0.59 for brain weight as a percentage of body weight; n=3 for each genotype).
|
|
To determine whether a more subtle phenotype could be detected by behavioral and cognitive testing, Hdh(
Q/+), Hdh(Q/Q) and wild-type control mice were subjected to tests measuring overall anxiety (elevated plus maze, light/dark box, open field and activity cage), a test of spatial learning and memory (Barnes circular maze), tests for olfaction and a test of motor coordination (accelerating rotarod) (38).
Anxiety-like behaviors were measured at 3, 6 and 9 months of age. These tests take advantage of the tension between the natural tendency the mice have to explore a novel environment and their innate preference for the dark and avoidance of open spaces (38). There were no significant differences found between Hdh(Q/Q) mice and wild-type controls at 3 months of age in the elevated plus maze based on the amount of time spent in the open arms (P=0.88) or the total number of arm entries (P=0.67). There were also no differences found at 6 months of age in the light–dark box in the latency to enter the dark box (P=0.99), the amount of time spent in the dark box (P=0.66) or the number of exits from the dark box (P=0.60). Testing in the open field also revealed no significant differences between deletion mutants and controls at 9 months of age based on the amount of time spent in the center squares (P=0.86). Similarly, there were no significant differences in overall horizontal and vertical activity between wild-type littermates and Hdh(Q/Q) mice when measured in an automated activity cage at 9 months of age (P=0.46).
Cognitive abilities were assessed using the Barnes circular maze, a measure of spatial learning and memory (39) (Fig. 5A). To avoid a bright light and buzzing sound in this test, mice must find the correct hole in a circular platform leading to a hidden escape tunnel. No significant differences in learning and memory were found at 5 months of age based on a distance score from the target (two-way RM ANOVA for genotype, P=0.16; Fig. 5B) (see Materials and Methods). The distance score relies heavily on spatial cues and proper hippocampal function and appears to be unimpaired. However, when the number of errors was examined, it was apparent that the Hdh(Q/+) and Hdh(Q/Q) mutants would explore many more of the holes in the maze before entering the escape tunnel on the first 2 trial days (Fig. 5C). Besides the significant difference found in the analysis of trial day, which indicates performance improved over the course of the test for all groups [F(8,5)=25.02; P<0.001], an ANOVA revealed a group difference in genotype [F(2,5)=21.1; P<0.001], and a significant interaction was found between trial day and genotype [F(8,5)=4.04; P<0.001]. This indicates that the Hdh(Q/+) and Hdh(Q/Q) mice performed poorly compared with wild-type controls, as an increase in the number of errors is considered a deficit in this type of learning and memory test, even in the absence of a distance score deficit. It is unlikely that this behavior was due to the decreased anxiety because the Hdh(Q/Q) mice behaved similarly to the controls in the elevated maze, light–dark box and open-field tests. Interestingly, the number of errors generated by Hdh(Q) mice and controls were comparable in subsequent days of testing (Fig. 5C).
|
0.05, Student's t-test). Error bars in (B–D) represent the SE.A similar tendency for increased exploration of a novel environment by Hdh(
Q/Q) mice was observed in an olfaction test (38). Olfaction was assessed by a paradigm requiring mice to find food buried in the cage bedding at 7 months of age (n=5) after 24 h of food deprivation. The Hdh(Q/Q) mice were slow to find a chunk of Nutri-grain bar, tending to explore the cage far more than the wild-type controls (+/+=38.4±9.5 s; Q/Q=143±20.7 s; P=0.004). In some instances, the Hdh(Q/Q) mice did not eat the treat, although the controls always did. There was no difference, however, in their latency to dig in the clean bedding (+/+=23.6±4.7 s; Q/Q=33.2±10.3 s; P=0.43). The test was repeated at 8 months of age using a food deprivation period of 36 h and a more palatable treat (chocolate) to ensure that the animals were properly motivated to find the hidden food. As in the previous test, the latency to dig in the bedding did not differ significantly between genotypes (+/+=13.2±3.1 s; Q/Q=24.4±6.5 s; P=0.17), and although not reaching significance, there was a trend toward the Hdh(Q/Q) mice being slower to find the treat and eat it (+/+=21.2±5 s; Q/Q=68.8±29.2 s; P=0.07).
To determine whether the increased latency to find a food treat exhibited by the Hdh(Q/Q) mice reflected a true olfactory deficit or a motivational/exploratory anomaly, olfaction was also assessed using a test that did not require the mice to locate a reward (n=5, mice from the same cohort used in the buried treat test). The mice were placed in a wire cage with clean bedding covering one-half of the cage and their own bedding covering the other half. Typically, mice will prefer to remain on the side with their own bedding, and in this test, there was no difference in the amount of time the Hdh(Q/Q) mice or wild-type controls spent on top of their own bedding (+/+=97±15 s; Q/Q=110±7.4 s; P=0.23). This result suggests that the Hdh(Q/Q) mice have no significant deficit in olfaction and their increased latency to find a treat may represent a difference in motivational/exploratory behavior.
To assess motor coordination in the Hdh(Q/Q) mice, they were tested at 1, 4 and 9 months of age on an accelerating rotarod. The rotarod test assesses motor coordination and balance, as well as motor learning (38). Surprisingly, at some time points, the Hdh(Q/Q) mice performed slightly better on this test than their wild-type littermate controls (Fig. 5D). At 1 and 9 months of age, an overall ANOVA showed differences in rotarod performance. These differences were not reflected in the 4 months data (data not shown).
Both Hdh(Q/Q) and wild-type mice showed an increase in the latency to fall in succeeding trial days at each age examined, indicating that both groups of mice were learning how to stay on the rod longer. Analysis by ANOVA showed a significant effect of trial day at 1 month [F(4,16)=96.9; P<0.001], 4 months [F(4,14)=56.6; P<0.001] and 9 months [F(4,12)=78.22; P<0.001].
The Hdh(Q/Q) mice performed better on the rotarod at 1 and 9 months of age, but curiously not at 4 months of age. There was a significant effect of genotype at 1 month of age [F(1,16)=8.9; P=0.009] and at 9 months of age [F(1, 12)=9.6; P=0.009], but not at 4 months of age [F(1,14)=0.09; P=0.76]. Specifically, the Hdh(Q/Q) mice did better on trial days 1 (P=0.039) and 2 (P=0.05) at 1 month and on trial day 3 (P=0.03) at 9 months. There was also a significant genotype versus trial day interaction at 1 month of age [F(4,16)=7.3; P<0.001], 4 months of age [F(4,14)=3.3; P=0.017] and at 9 months of age [F(4,12)=2.8; P=0.037].
Hdh(Q/Q) fibroblasts have elevated levels of ATP
The behavioral and motor phenotypes exhibited by the Hdh(Q/Q) mice may reflect specific changes in CNS function or a more global change in cellular metabolism. To begin to address the basis for these phenotypes, we examined the basic metabolic status of Hdh(Q/Q) primary embryonic fibroblasts by measuring ATP levels in Hdh(Q/Q) and wild-type control cell cultures. In initial experiments with wild-type and Hdh(Q/Q) fibroblasts aimed at comparing the subcellular localization of htt and Q-htt, we noticed that the Hdh(Q/Q) fibroblasts were less dense in culture (37% the density of the wild-type cultures) and had an altered morphology at later passages in culture. We measured ATP levels in the cells using a bioluminescent read-out assay at P7 and found that the Hdh(Q/Q) fibroblasts had a significantly increased level of ATP/cell (wild-type=6.4x10–11±6.3x10–12 moles ATP/cell and Q/Q=9.16x10–11±5.3x10–12 moles ATP/cell; P=0.007). To determine whether the change in ATP levels also occurred at earlier passages, we measured ATP levels in wild-type, Hdh(Q/+) and Hdh(Q/Q) embryonic fibroblasts obtained from Hdh(Q/+) intercrosses (embryos from two females, n=3 embryos for each genotype). In early passage fibroblasts (P2–P4), ATP levels were similar in all genotypes (P=0.17–0.41) (Fig. 6A). At P5, however, the ATP level in the Hdh(Q/Q) fibroblasts was increased relative to the levels observed in the Hdh(Q/+) and wild-type control cells (P=0.01), and at P6, the level of ATP in the Hdh(Q/Q) cells was 4-fold higher than the level in the Hdh(Q/+) and wild-type cells (P=0.0006) (Fig. 6A).
|
To determine whether primary cultures of cells derived from adult mice also have increased levels of ATP, fibroblasts from ear biopsies (13 months old, n=1 from each genotype) were obtained and ATP levels were assessed. In P1 cultures, ATP levels in the adult Hdh(
Q/Q) fibroblasts were again elevated 4-fold over wild-type ATP levels (P<0.0001) (Fig. 6B). However, in contrast to the results with primary cells in culture, ATP levels in cortex and skeletal muscle from 14-month-old wild-type and Hdh(Q/Q) mice were not significantly different (P=0.86 and P=0.78, respectively) (Fig. 6C). Therefore, the increase in ATP levels exhibited in primary embryonic and adult fibroblasts may reflect a feature that is elicited by the stress of in vitro cell culture.
Hdh(Q/Q) fibroblasts senesce prematurely
The increase in cellular ATP levels in the Hdh(Q/Q) fibroblasts was accompanied by a decrease in cell proliferation. Although both Hdh(Q/+) and Hdh(Q/Q) primary embryonic fibroblasts grew more slowly than wild-type cells at P2 (P=0.02 and 0.04, respectively), Hdh(Q/+) cell growth was not significantly different from wild-type cells at later passages (P=0.36–1.0, Fig. 6D). In contrast, Hdh(Q/Q) embryonic fibroblasts continued to divide slowly after P2, and at P6, Hdh(Q/Q) cells were 30% the density of the wild-type cultures (P=0.025, Fig. 6D). In addition, many of the Hdh(Q/Q) cells at P4–P6 exhibited an altered morphology and were observed to have double or multiple nuclei (Fig. 7A).
|
To determine whether these phenotypes were related to cellular senescence, Hdh(
Q/Q) and control fibroblast cultures were stained for senescence-associated ß-galactosidase (SA-ßgal) activity (Fig. 7B). Although the role of SA-ßgal expression in cellular senescence is unclear, the presence of SA-ßgal activity correlates well with cellular aging in humans and mice (40). In senescent cells, the fraction of cells staining robustly for pH 6.0 SA-ßgal activity was 2.5-fold higher than in the wild-type control culture [Hdh(+/+)=14±2.9% and Hdh(Q/Q)=37±3.9% cells positive for SA-ßgal staining; P<0.001].
In addition to increased SA-ßgal expression, elevated levels of p21Waf1/Cip1 (41) and p16INK4A (42) are often observed in senescent cells. p21 is a cyclin-dependent kinase inhibitor that interacts with cdk2-associated complexes and prevents the cell cycle from entering into S phase. Depending upon the tissue and cell type, p21 is also involved in differentiation and apoptosis (43). Immunofluorescence staining for p21 expression in wild-type and Hdh(Q/Q) fibroblasts at P6 revealed an increase in nuclear staining for p21 in the mutant fibroblast cultures (Fig. 7C). p21 is also a target gene of p53 and can mediate p53's function as a tumor suppressor (44). The relative levels of p53, however, were not obviously different in wild-type and Hdh(Q/Q) P6 fibroblasts, as evidenced by the similar staining for nuclear p53 in both cultures (Fig. 7D). Although levels of activated p53 can indicate senescence, cells can also undergo p53-independent p21 activation (43).
p16 is also a cyclin-dependent kinase inhibitor that interacts with cdk4-associated complexes. Inhibition of the cdk4 complex kinase activity by p16 results in hypophosphorylation (activation) of the retinoblastoma protein and cell cycle arrest (45). As seen with p21 expression, p16 immunofluorescence staining is also increased in the nuclei of the Hdh(Q/Q) fibroblasts in comparison with wild-type controls (Fig. 7E).
To investigate further the phenotype of the Hdh(Q/Q) fibroblasts, immunofluorescence staining for caveolin-1 and HIP-1 was also performed on P4 cultures (Fig. 7F). Both caveolin-1 and HIP-1 are htt-interacting proteins, and the expression levels of both proteins increases during senescence in culture. Although the role of caveolin-1 in senescence is still unclear, caveolin-1 expression has been observed to increase in senescent human mesenchymal stem cells (46) and murine fibroblasts (47). In addition to its role as a regulator of the androgen receptor, HIP-1 is also involved in the intrinsic apoptotic cell death pathway (30,48). HIP-1 expression is increased in cells derived from progeria patients (a human model of accelerated aging) and is also increased in fibroblasts obtained from aged normal donors (49). As with caveolin-1, the mechanism underlying HIP-1's role in senescence is still unknown. In P4 wild-type primary embryonic fibroblasts, caveolin-1 and HIP-1 immunostaining is weak, but is elevated in the Hdh(Q/Q) fibroblasts (Fig. 7F). Thus, with the exception of p53 levels, the expression of four senescence-associated proteins are increased in Hdh(Q/Q) fibroblasts.
To assess whether premature senescence occurs in vivo, brain sections from 1-year-old control and Hdh(Q/Q) mice (n=2) were examined for the presence of lipofuscin. Lipofuscin pigment accumulates in the cytoplasm of neurons as they age, and lipofuscin-derived autofluorescence has been found to increase with age in the brains of both mice and humans (50,51). In all brain areas examined (thalamus, cortex and hippocampus), a marked increase in the amount of lipofuscin autofluorescence was detected in the Hdh(Q/Q) brains (Supplementary Material, Fig. S1). Despite these results, we have not yet detected any other obvious physical manifestations of premature aging in the oldest of our Hdh(Q/Q) mice.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
The short polyQ stretch within mouse htt is not essential for htt's normal function during embryonic development. Deletion of the polyQ stretch also has little observable effect on the subcellular localization of
Q-htt, on the gross morphology of the adult mouse brain and on the association of Q-htt with several htt-interacting proteins. The lack of an embryonic or more overt phenotype in the Hdh(Q/Q) mice is not unexpected, as the size of htt's polyQ stretch is variable in humans (52), because Drosophila htt functions normally without a polyQ stretch (23), and deletion of polyQ or Q-rich regions in other proteins does not necessarily lead to complete functional inactivation.
For example, deletion of a polyQ stretch is tolerated for dimerization of transcriptional activators (53), formation of protein complexes associated with signaling cascades (54) and regulation of gene expression (53–55). The contribution of a polyQ stretch to normal function may also vary within the same protein for those polypeptides with multiple polyQ or Q-rich domains. Deletion of the polyQ stretch within a regulatory protein of the nitrogen cycle (nit-4) does not affect its function, but co-deletion of a Q-rich domain abolishes its activity (56). Moreover, the DAL81 protein contains two polyQ stretches, one of which is dispensable, whereas the other diminishes DAL81 activity by 50% when deleted (57).
Although the polyQ stretch within mouse htt is not essential for htt's normal function during development, deletion of the polyQ stretch results in more errors in the beginning trials of a learning/memory task, increased exploratory behavior in an olfaction test and subtly enhanced rotarod performance in adult mice. Both heterozygous and homozygous Q-htt mice explore more holes in the Barnes circular maze before entering the escape tunnel on the first 2 days of testing. On subsequent days, they perform similarly to controls. We interpret this behavior to suggest that the Q-htt mice are either slightly impaired in their ability to learn the task or, alternatively, they exhibit more exploratory behavior in a novel environment relative to controls.
The olfaction test involving locating a food treat also revealed a subtle behavioral difference in the Hdh(Q/Q) mice. The Hdh(Q/Q) mice exhibited an increased latency to find the buried treat, even in the presence of food deprivation as a motivational source. We interpret these results as another indication that the Hdh(Q/Q) mice have either a tendency to exhibit more exploratory behavior than controls when placed in a novel environment or they might lack motivation for completing novel tasks. This behavioral phenotype could occur in the absence of hyperactivity, but perhaps is not reflected in the open-field test (which was performed with older mice at 9 months of age).
In addition to deficits in the Barnes circular maze and olfaction test, the Hdh(Q/Q) mice exhibit a slight improvement in rotarod performance. This improvement cannot be attributed to strain differences, as C57BL/6J mice show much longer latencies to fall from the rotarod than 129/SvJ mice (58). In the literature, we are aware of only a few examples of improved rotarod performance that are the result of genetic mutation in the mouse. For example, in a mouse model for Down's syndrome (Ts65Dn) that has three copies of a portion of mouse chromosome 16 (corresponding to the human chromosome 21 trisomy mutation), improved rotarod performance was observed, as well as hyperactivity in the open field (59). Similar phenotypes were also exhibited in mice heterozygous for a mutation in the heregulin gene that encodes a ligand for tyrosine kinase receptor signaling pathways (60). These results were unexpected in both instances, as cerebellar problems were anticipated in both models because Ts65Dn mice have reduced cerebellar volume (61) and heregulin is important for proper cerebellar development (62). In addition, mice deficient for a protein repair methyltransferase enzyme (a mouse model for epilepsy) exhibited enhanced rotarod performance relative to controls on a single day of testing (63).
Although the Hdh(Q/Q) mice appear to exhibit subtly enhanced motor coordination, their motor learning rate for the rotarod test does not differ from wild-type controls. Also, in contrast to previous observations of enhanced rotarod performance in mutant mice, hyperactivity was not observed in the Hdh(Q/Q) mice when tested at 9 months of age. In an attempt to determine whether the Hdh(Q/Q) mutant's enhanced rotarod performance may be due, in part, to an altered cellular physiology, we obtained fibroblasts from embryos and adult ear tissue in order to characterize ATP levels in culture. Mutant htt with an expanded polyQ stretch, for example, can elicit phenotypes in somatic cells in addition to neurons. Lymphoblasts obtained from HD patients, as well as Hdh(Q111) striatal neurons in culture, exhibit reduced ATP levels or reduced ATP/ADP ratios in vitro (24). Interestingly, we observed elevated ATP levels in the Hdh(Q/Q) fibroblasts, but not in the brain or skeletal muscle from 14-month-old mutant mice. Perhaps, the increase in ATP is a property unique to fibroblasts or it is possible that the elevated ATP levels observed in the mutant fibroblasts are a consequence of their response to in vitro culture conditions. If the latter interpretation is correct, the absence of any significant differences in ATP levels measured in the tissues of the control and mutant mice may be due to homeostatic processes in the context of the whole animal, which are able to control ATP levels more effectively in response to stress.
The elevated ATP level observed in the Hdh(Q/Q) fibroblast cultures was accompanied by a reduction in cell proliferation. It is likely, based on the observed correlation between reduced embryonic fibroblast proliferation and elevated ATP levels, that the increase in cellular ATP is related directly to reduced growth in culture. Both p21 (an inhibitor of cdk2/cyclin E complex activity) and p16 (an inhibitor of cdk4,6/cyclin D1 complex activity) were elevated in the mutant fibroblasts, along with HIP-1, caveolin-1 and SA-ßgal expression, suggesting that the reduced proliferation likely represents senescence. These results are in contrast to those obtained with skin fibroblast cultures obtained from 12-week-old R6/2 transgenic HD model mice. The R6/2 fibroblast cultures exhibited a reduced mitotic index and disorganized centrosomes, therefore making it difficult to maintain the cells in culture, but there was no evidence of premature senescence (64). These phenotypes in the R6/2 fibroblasts obtained from the 12-week-old mice were not as apparent as in the fibroblasts obtained from younger animals, suggesting that they reflected a progressive change in the skin of the R6/2 mice. In our model, we were able to obtain mutant fibroblasts from adult mice, although it is noteworthy that ATP levels were already elevated at P1 in the adult mutant fibroblasts. Further studies are needed to determine whether this represents a progressive phenotype in our Hdh(Q/Q) mice.
Our observation that the Hdh(Q/Q) fibroblasts were senescing prematurely and had higher levels of ATP was somewhat surprising, because there is a large body of data correlating cellular senescence with lower levels of ATP (65). However, the results from recent investigations using antagonists of the different cdk/cyclin complexes may resolve this apparent discrepancy. Roscovitine, an inhibitor of the cdk2 complex causes senescence in cell culture that is not accompanied by any change in ATP concentration, but flavopiridol, an inhibitor of the cdk4 complex, increases cellular ATP content in Hep G2 cells (66). Similarly, in leukemia cells, cell cycle arrest can be uncoupled from energy status by exogenous expression of p16, resulting in elevated levels of ATP (67). In both of these examples, experiments were performed with immortalized cells, and the mechanism linking enhanced p16 expression and/or inhibition of cdk4 complexes to elevated ATP levels is unknown.
On the basis of the concept of the ‘Hayflick limit,’ it has been proposed that a correlation exists between the proliferative life span of in vitro fibroblasts and the longevity of the donor animal (68). Although we have detected increased lipofuscin pigment in the brains of the Hdh(Q/Q) mice, we have not yet detected any other obvious signs of aging in the oldest of these mice (now 14 months old), and further observations are needed in order to determine whether additional signs of premature senescence occurs in vivo. Nonetheless, the elevation in ATP levels exhibited by the embryonic fibroblasts that occurs in parallel with cellular senescence suggests that the polyQ stretch contributes to a normal htt function involved in modulating longevity of primary cells in culture and, potentially, energy status. The Hdh(Q/Q) fibroblasts also provide another data point to compare with the energy level status of lymphoblastoid cells from HD patients exhibiting ATP/ADP ratios that correlate inversely with htt polyQ length. In our Hdh(Q/Q) primary cell cultures, energy status appears to be uncoupled from cell proliferation with high levels of ATP correlating with cellular senescence.
It is important to note, however, that the polyQ deletion results in only a subtle phenotype in vivo, and thus it is likely that the polyQ stretch is not required for an essential function of htt, but instead, may modulate a normal function of htt. Deletion of the polyQ stretch, for example, may influence htt's ability to act as a scaffold for diverse signaling pathways in the cell, thus impacting both cell cycle progression and energy status. In vertebrates, a short polyQ stretch in htt may be required for optimal normal function. Expansion of the polyQ stretch beyond this short optimum length could then result in altered htt function leading to the dominant energy phenotype observed in lymphoblastoid cells from both normal and HD patients, whereas deletion of the polyQ stretch results in a different set of phenotypes that are characterized by premature senescence in vitro and subtle behavioral abnormalities in vivo.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Generation of Hdh(
Q/+) miceIntroduction of the
Q mutation and N-terminal FLAG epitope modifications of Hdh exon 1 were performed in two sequential steps by replacing first an endogenous Hdh exon 1 XmnI–NarI restriction fragment containing the polyQ stretch of the htt coding sequence, with a synthetic XmnI–NarI fragment that was generated by annealing two complementary oligonucleotides carrying the Q mutation (deltaQ-1: XmnI site double underlined, NarI site single underline; 5'-GATGAAGGCTTTCGAGTCGCTCAAGTCGTTT><CC ACCGCCGCAGGCGCCGCCG-3', deltaQ-2: 5'-CGGCGGC GCCTGCGGCGGTGG><AAACGACTTGAGCGACTCGA AAGCCTTCATC-3'). Following annealing, the DNA was digested with XmnI and NarI and then cloned into an XmnI–NarI vector prepared from a 1.6 kb PstI Hdh genomic fragment subclone containing the Hdh promoter, exon 1, and a portion of intron 1. To introduce the FLAG epitope modification, an AlwNI–XmnI fragment containing the N-terminus of htt was replaced with a synthetic AlwNI–XmnI fragment that was generated by annealing two partially complementary oligonucleotides containing the FLAG epitope-modified N-terminal sequence and appropriate restriction enzyme sites (FLAG forward: AlwNI underlined, complementary bases in italic, 5'-GTCTTCAGGGTCTGTCCCATCGGGCAGGAAGCCGTC ATGGACTACAAGGACGACGATGACAAG-3', FLAG reverse: XmnI double underlined, 5'-CGACTCGAAAGCCTTCAT CAGCTTTTCCAGGGTTGCCTTGTCATCGTCGTCCTTGTA GTC-3'), and then extending the annealed partial duplex with Klenow fragment of DNA Polymerase I enzyme and deoxynucleotide triphosphates. The synthetic DNA fragment was then digested with AlwNI and XmnI restriction enzymes and cloned into the Q-modified 1.6 kb PstI genomic fragment. The Q and FLAG modifications were verified by DNA sequencing, and a gene targeting vector was assembled using 7.1 kb of 5'-flanking homology, a floxed pgkneo cassette inserted within a HindIII restriction site located 1.3 kb upstream of exon 1 for positive selection of transfected ES cells and 2.8 kb of 3'-flanking homology. The gene-targeting vector was linearized at a NotI site located adjacent to the 5'-homology region and electroporated into W9.5 ES cells (129Sv background). ES clones were screened by Southern blotting of NcoI digested genomic DNA with a 3'-flanking probe located outside the targeting vector 3'-homology (150 bp XhoI–HindIII fragment) (7). The 3'-flank probe recognizes a 10 kb wild-type fragment and a 12 kb fragment from the targeted allele (2 kb larger in size because of the addition of the floxed pgkneo cassette). To determine whether the targeted ES clones identified by Southern analysis also contained the FLAG and Q modifications, PCR amplification of the epitope tag and Q regions of exon 1 was performed. For amplification of sequence containing the Q mutation, the 5' (forward) oligonucleotide is located near the 5' end of exon 1, while the 3' (reverse) oligonucleotide spans the CAG repeat deletion site, and is thus able to anneal to the mutant, but not to the wild-type allele, to generate a 226 bp product. Q forward: 5'-GACGGGCCCAAGATGG-3'; Q reverse: 5'-GGCGGTGGAAACGACTT-3'. Cycle conditions: 3 min 30 s initial denaturation at 94°C, followed by 30 cycles consisting of 20 s at 94°C, 30 s at 60.8°C and 55 s at 72°C with a 3 s/cycle elongation, followed by a final 5 min extension at 72°C. Although the Q PCR product also contains the FLAG sequence, for simultaneous detection of both the epitope tag and the wild-type sequence, oligonucleotide primers flanking the site of the FLAG insertion were used to generate either a 112 bp product (wild-type allele) or a 136 bp product (targeted allele). The 5'-oligonucleotide is located 27 bp upstream of the AlwNI restriction site, whereas the 3'-oligonucleotide is located just 5' of the Q deletion site. Epi-forward: 5'-GCGTAGTGC CAGTAGGCTCCAAG-3'; Epi-reverse: 5'-CTGAAACGACT TGAGCGACTCGAAAG-3'. Cycle conditions: 2 min