CAG/polyglutamine expansion has been shown to form the molecular basis of an increasing number of inherited neurodegenerative diseases. The mutation is likely to act by a dominant gain of function but the mechanism by which it leads to neuronal dysfunction and cell death is unknown. The proteins harbouring these polyglutamine tracts are unrelated and without exception are widely expressed with extensively overlapping expression patterns. The factors governing the cell specific nature of the neurodegeneration have yet to be understood. Upon a certain size threshold, expanded CAG repeats become unstable on transmission and a modest degree of somatic mosaicism is apparent. Similarly, the molecular basis of the instability and its tissue specificity has yet to be unravelled. Recent reports describing the first mouse models of CAG/polyglutamine disorders indicate that it will be possible to model both the pathogenic mechanism and the CAG repeat instability in the mouse. This has great potential and promise for uncovering the molecular basis of these diseases and developing therapeutic interventions.
Huntington's disease (HD) (1 ) is one of an increasing number of neurodegenerative disorders caused by a CAG/polyglutamine (polygln) repeat expansion, including spinal and bulbar muscular atrophy (SBMA) (2 ), dentatorubral pallidoluysian atrophy (DRPLA) (3 ,4 ) and spinocerebellar ataxia (SCA) types 1 (5 ), 2 (6 -8 ), 3 (9 ) and 6 (10 ). The inheritance patterns are autosomal dominant (with the exception of X-linked SBMA) and in each case, the proteins can tolerate a large variation in the size of the polygln tracts in the normal range but upon a certain size (~37-40 glutamines) these tracts become pathogenic. It is likely that the novel molecular pathways initiated by this mutation have a common basis (except possibly in the case of SCA6 in which the pathogenic threshold is smaller). The proteins harbouring the polygln stretches are mostly novel and otherwise unrelated. In all cases the proteins are widely or ubiquitously expressed, but despite extensively overlapping expression patterns, the neuronal cell death is relatively specific and can differ markedly (reviewed in 11 ).
The molecular events by which a polygln expansion causes cell death remain to be unravelled (reviewed in 12 ). These mutations are likely to act by a dominant gain of function, this mechanism being supported by the identification of the 1C2 antibody which specifically detects polygln expansions, suggestive of a conformational change at a certain size threshold (13 ). In addition, the factors which convey the specific and differing patterns of cell death between these diseases are not understood. Possible mechanisms include differences in expression levels, subcellular localisation of the mutated protein or cell specific subcellular interactions. A number of proteins have now been reported to interact with huntingtin which include HAP 1 (14 ), HIP-1 (15 ), a specific ubiquitin-conjugating enzyme (16 ) and GAPDH (17 ). It is yet to be established whether any of these proteins play a role in the pathogenic mechanism. Huntingtin has also been shown to be specifically cleaved by apopain, a cysteine protease with a key role in the proteolytic events leading to apoptosis (18 ). Similarly, it is not clear if this participates in the chain of events leading to neurodegeneration.
Expanded triplet repeats are invariably unstable when inherited from one generation to the next and they generally show varying degrees of somatic mosaicism. The intergenerational instability forms the molecular basis of anticipation: the observation that the age of onset of a disease decreases and/or the severity increases as the gene is passed from one generation to the next. Repeat instability on transmission has been described in all of the CAG repeat diseases and, in general, repeats tend to be more unstable on paternal transmission. This may present as larger increases on paternal inheritance as in HD (19 ) (reflected in the paternal sex bias to the anticipation) or as a tendency to increase on male and decrease on female transmission as in SCA1 (20 ). A relatively modest degree of somatic repeat instability has been identified in HD, DRPLA, SCA1 and MJD. In general, expansions have been identified in regions of the CNS, with the exception of the cerebellum which presents a smaller repeat relative to the other brain regions tested (21 -26 ). Of non-CNS tissues, instability has consistently been reported in liver and kidney (21 ,24 -26 ) and also in muscle, lung, testis (21 ), leukocytes (23 ) and colon (24 ,26 ). Studies of DRPLA patients also identified a significant correlation between the range of the expanded allele and the age at death of the patient rather than with the onset of disease (25 ). The molecular events governing triplet repeat instability are not understood and possible mechanisms must address both a CAG repeat size threshold and cell specificity.
It has been proposed for many years that the HD mutation most probably acts through a dominant gain of function. Analysis of mice arising from the first transgenic models of HD, SCA1 and MJD in addition to gene targeted knockouts of the mouse HD gene (Hdh) supports this hypothesis.
Three research groups have independently generated knockouts of the mouse HD gene (Hdh) (27 -29 ). In all cases the nullizygous phenotype was embryonic lethal, clearly demonstrating that the HD gene plays an important role in development. In two of these studies, heterozygous mice expressing only one copy of Hdh were phenotypically normal (28 ,29 ). In contrast, Nasir et al. (27 ) reported that heterozygotes showed increased motor activity and cognitive deficits with a significant neuronal loss in the subthalamic nucleus. To explain this discrepancy, it has been suggested that the targeted allele (targeted to replace exon 5) may allow the production of a truncated protein which could conceivably cause a dominant effect in the heterozygous mice, generating a phenotype. Together, these studies demonstrate that HD is not caused by haplo-insufficiency (loss of function of one copy of the gene) or a simple dominant-negative mechanism. In the first case, loss of one allele and in the second case, loss of both alleles, would be expected to generate a model of HD.
A dominant gain of function mechanism would predict that a mouse model of a polygln neurodegenerative disorder could be generated by the introduction into the mouse germ line of a mutant copy of the gene in question, irrespective of the presence of two copies of the endogenous mouse homologue. The first description of HD transgenic mice used a full length cDNA construct under the control of a CMV promoter carrying (CAG)44 (30 ). Of the HD transgenes, 2/6 founders expressed high levels of transgene mRNA but a transgene protein was not detected. Whilst these results could be interpreted as providing evidence that translation of the CAG repeat into a polygln expansion is necessary for pathogenesis, the repeat expansion in this experiment is comparatively modest and it is possible that larger expansions are necessary to generate a phenotype with an age of onset that falls within the lifetime of a mouse.
Genomic clones are frequently more successful at generating transgenic models than cDNAs as they often direct an expression profile that mimics the endogenous gene. The large size of the HD gene (170 kb) necessitates that genomic constructs are prepared and manipulated in the form of yeast artificial chromosomes (YACs). Using YAC technology, Hodgson et al. (31 ) have successfully generated mice that are transgenic for the normal human HD gene. They have crossed the human HD transgene onto an Hdh nullizygous background and shown that the human YAC can rescue the embryonic lethal phenotype. This indicates that the transgene is expressed appropriately and predicts that the introduction of a mutant version of the human YAC would be successful in generating a model of HD.
We have described four lines of mice that are transgenic for exon 1 of the HD gene carrying CAG expansions of 115-156 (R6/1, R6/2, R6/5 and R6/0) and a further two lines transgenic for the same construct carrying 18 repeats (HDex6 and HDex27) (32 ). The transgene is ubiquitously expressed at both the RNA and protein levels in all lines except R6/0, in which no evidence of expression has been detected (32 ,33 ). The transgene protein contains the first 69 amino acids of huntingtin in addition to the number of residues encoded by the CAG repeat (i.e. ~3% of huntingtin).
A progressive neurological phenotype has been observed in three lines: R6/1, (CAG)115; R6/2, (CAG)145; and R6/5, (CAG)130-155. Line R6/2 has an onset of ~2 months, line R6/1 at ~5 months and R6/5 hemizygotes do not show symptoms after >1 year. Lines R6/1 and R6/5 show an earlier age of onset and more rapid progression of the disease when bred to homozygosity. The phenotype includes an irregular gait, resting tremor, stereotypic and abrupt, irregularly timed movements and epileptic seizures. Coincident with the onset of the motor disorder there is a progressive reduction in body weight in the transgenes as compared with their littermate controls. The absence of a phenotype in lines R6/0, HDex6 and HDex27 suggests that expression of the polygln expansion forms the molecular basis of the phenotype rather than the expression of a novel peptide. It is notable that the R6 phenotype does not include an overt cerebellar ataxia as described for the spinocerebellar ataxia lines (34 ,35 ) (see below). Extensive neuropathological analysis has been performed on the brains of R6/2 mice. At 12 weeks, the only difference that could be identified between the transgene and control brains was that the R6/2 brains were ~20% smaller, and that this reduction in brain size occurred across all structures with an apparently normal neuronal density. This is consistent with early changes that occur in the brains of HD patients. More recently, immunocytochemistry with huntingtin N-terminal antibodies has identified the presence of neuronal intranuclear inclusions (NII) in the brains of symptomatic transgenic mice (33 ).
Mice transgenic for both SCA1 (34 ) and MJD (35 ) constructs have also been reported to develop a phenotype. A summary of the main features of these and the R6 transgenes is presented in Table 1 . The SCA1 transgenes were the first demonstration that modelling a polygln repeat disorder would be possible in the mouse. They included mice transgenic for the SCA1 cDNA carrying either a normal interrupted allele of (CAG)12CATCAGCAT(CAG)15 (PS-30) or an expanded uninterrupted allele of (CAG)82 (PS-82) under the control of the pcp2 promoter (Purkinje cell-specific) (34 ). Five of six PS-82 lines showed RNA expression between 10- and 100-fold of endogenous levels. In the original report, transgene protein could not be detected but this has since been shown to be present by immunocytochemistry (H.Orr and H.Zoghbi, personal communication). Mice from all five lines developed ataxia. Onset varied from 12 to 26 weeks and a dosage effect was apparent: in two lines studied, homozygotes were more severely affected than hemizygotes. Neuropathological analysis showed significant loss of the Purkinje cell population, with Bergmann glial proliferation, and shrinkage and gliosis of the molecular layer. Ectopic Purkinje cells were present in the molecular layer and occasionally the granular layer and the dendritic arrays also appeared to be abnormal.
Ikeda et al. (35 ) used expression constructs containing the MJD cDNA carrying 79 CAG repeats (MJD79), the CAG repeat followed by only the C-terminus of the MJD gene with both 79 and 35 CAG repeats (Q79C and Q35C) and a 79 CAG repeat in isolation (Q79) under the control of a Purkinje cell-specific promoter. Ataxia was observed in 3/3 Q79C and 2/6 Q79 transgenic mice, occurring as early as 1 month of age after full activation of the promoter. In contrast, a phenotype was not observed in any of the ten Q35C or four MJD79 transgenic mice as of 7 and 5 months, respectively. Neuropathological analysis of a 2 month old ataxic Q79C mouse showed a very atrophic cerebellum, in which all three layers were affected. The authors suggest that comparison of their data with that of Burright et al. (34 ) indicates that the polygln tracts are more toxic when present in isolation or in the context of a truncated protein. In the absence of any data relating to expression levels it is difficult to come to strong conclusions with regard to the comparative toxicity of the full length and truncated constructs. However, these conclusions were strongly supported by a series of transient transfections of COS cells described in the same paper (35 ).
Repeat stability studies carried out on the first mice transgenic for CAG repeat expansions showed no evidence of instability and suggested that the molecular mechanism underlying triplet repeat instability in humans may not exist in the mouse. These initial studies included (CAG)45 in the androgen receptor cDNA (36 ), (CAG)44 in the HD cDNA (30 ), (CAG)82 in the SCA1 cDNA (34 ) and (CAG)79 in constructs based on the MJD/SCA3 cDNA (35 ).
The R6 lines transgenic for exon 1 of the HD gene carrying (CAG)115-(CAG)155 expansions showed both intergenerational and somatic repeat instability (32 ,37 ). The repeats were clearly unstable on transmission in lines R6/1, R6/2 and R6/5, although this was less clear in line R6/0 as the changes observed in this line could be accounted for by errors in sizing. In line R6/2, the degree of instability increases with the age of the transmitting male (as R6/2 females are sterile it was not possible to look for an age effect on female transmission). R6/5 was the only line in which an extensive comparison of instability on both male and female transmission was conducted and the repeats had a tendency to increase on male transmission and decrease on female transmission (37 ). This trend was supported by the intergenerational instability observed in the other lines. The CAG expansions introduced into these mice are considerably larger than are normally seen in HD patients. The change in size of the repeat on transmission in the mice is smaller than would be expected from comparison with size changes associated with highly expanded CAG repeats seen in humans. The discrepancy in the degree of instability between humans and mice may reflect the difference in their life span, a model supported by the observation that the size of the intergenerational expansion increased with the age of the transmitting male.
Somatic instability was detected in lines R6/1, R6/2 and R6/5 but not in line R6/0 (Fig. 1 ). In all three lines, onset of instability was at ~6 weeks and the CAG repeat range increased with the age of the mouse. This argues against a pathogenic role for repeat instability as the age of onset of symptoms in these lines differs markedly. The pattern of instability was more widespread in some lines than others although on the whole it was first present and most prominent in brain regions. Peripheral tissues that consistently showed instability included liver and kidney. Overall the somatic instability was comparable with that described in individuals carrying CAG expansions (21 -26 ). The major difference between line R6/0, in which instability was not apparent, and the other lines was the absence of transgene expression. This is probably due to gene silencing by a position effect as the R6/0 transgene has clearly integrated into a region of unusual genomic structure (32 ).
Triplet repeat instability has also been reported in two series of lines transgenic for the myotonic dystrophy (DM) mutation (CTG on the sense strand) (37 ,38 ). The integration fragments used in these lines were a genomic fragment (Dmt162) from the myotonic dystrophy (DM) locus containing a small portion of the coding region and the 3'UTR with (CTG)162 (38 ) and a cosmid (DM55-5) containing the myotonic dystrophy protein kinase gene (DMPK) with (CTG)55 and the flanking DMR-N9 and DMAHP genes (37 ). Intergenerational instability was observed in both of these cases. The DM55-5 transgenes showed intergenerational instability in 6.8% of transmissions, the changes generally being expansions of one repeat unit. A higher frequency of unstable transmissions was observed in the Dmt lines (as in the R6 lines), most likely as a consequence of the larger size of the repeat tracts.
Table 1
It is clear that a polygln expansion can give rise to a progressive neurological phenotype in the mouse. The analysis of existing and further transgenic models of CAG/polygln repeat disease will be informative with respect to uncovering the molecular basis of these disorders. Comparison of transgenes arising from full length and truncated constructs may resolve the speculation that the toxic agent is a truncated version of the proteins in question. The models will be useful in allowing the study of the early disease stages for which patient material is rarely available. Comparison of future models in which the transgenes are under the control of endogenous or ubiquitous promoters may shed light on the factors which determine the differing patterns of neurodegeneration.
Human Molecular Genetics
Pages
Introduction
Transgenic Modelling Of Huntington's Disease
Knockouts of the mouse Hhd gene
Transgenic models of HD
Mice transgenic for a mutant version of exon 1 of the HD gene
Comparison With Other CAG/Polygln Mouse Models
Mouse Models Of CAG/CTG Repeat Stability
Triplet repeat instability in lines transgenic for the HD mutation (R6)
Other mouse models of triplet repeat instability
Conclusion
References
Disease
Construct
Promoter
(CAG)n
Expression
Phenotype
Freq. of lines
RNA
Protein
showing phenotype
HD
exon 1 (genomic)
HD
18
+
+
none
0/2
HD
exon 1 (genomic)
HD
142
-
-
none
0/1
HD
exon 1 (genomic)
HD
115-156
+
+
+
3/3
SCA1
cDNA (full length)
pcp2a
30b
+
+
none
0/7
SCA1
cDNA (full length)
pcp2a
82
+
+
+
5/6
MJD
cDNA (full length)
L7a
79
NR
NR
none
0/4
MJD
cDNA (C-terminus)
L7a
35
NR
NR
none
0/10
MJD
cDNA (C-terminus)
L7a
79
NR
NR
+
3/3
MJD
polyglutamine tract
L7a
79
NR
NR
+
2/6
REFERENCES
This page is maintained by OUP admin. Last updated Fri Sep 12 18:09:23 BST 1997. Part of the OUP Journals World Wide Web service. Copyright Oxford University Press, 1997
